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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0dfa9145-5428-4c50-8b3b-97c3b66cd000 | imagebind-one-embedding-space-to-bind-them | 2305.05665 | null | https://arxiv.org/abs/2305.05665v2 | https://arxiv.org/pdf/2305.05665v2.pdf | ImageBind: One Embedding Space To Bind Them All | We present ImageBind, an approach to learn a joint embedding across six different modalities - images, text, audio, depth, thermal, and IMU data. We show that all combinations of paired data are not necessary to train such a joint embedding, and only image-paired data is sufficient to bind the modalities together. ImageBind can leverage recent large scale vision-language models, and extends their zero-shot capabilities to new modalities just by using their natural pairing with images. It enables novel emergent applications 'out-of-the-box' including cross-modal retrieval, composing modalities with arithmetic, cross-modal detection and generation. The emergent capabilities improve with the strength of the image encoder and we set a new state-of-the-art on emergent zero-shot recognition tasks across modalities, outperforming specialist supervised models. Finally, we show strong few-shot recognition results outperforming prior work, and that ImageBind serves as a new way to evaluate vision models for visual and non-visual tasks. | ['Ishan Misra', 'Armand Joulin', 'Kalyan Vasudev Alwala', 'Mannat Singh', 'Zhuang Liu', 'Alaaeldin El-Nouby', 'Rohit Girdhar'] | 2023-05-09 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Girdhar_ImageBind_One_Embedding_Space_To_Bind_Them_All_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Girdhar_ImageBind_One_Embedding_Space_To_Bind_Them_All_CVPR_2023_paper.pdf | cvpr-2023-1 | ['cross-modal-retrieval'] | ['miscellaneous'] | [ 3.19850415e-01 -1.63084075e-01 -2.36943707e-01 -7.83872157e-02
-8.06407809e-01 -5.03262818e-01 1.13717186e+00 -8.68060067e-02
-3.39792997e-01 2.24490941e-01 6.44832611e-01 2.74246875e-02
4.91142012e-02 -7.11349726e-01 -8.57312024e-01 -5.91432810e-01
-3.39966901e-02 1.67920932e-01 2.40389153e-01 -3.91492158e-01
1.45260349e-01 -3.56277712e-02 -2.15457106e+00 6.94956839e-01
1.48394346e-01 1.15152943e+00 1.69673011e-01 9.52772319e-01
-4.05391492e-02 1.21881998e+00 -1.37920752e-01 -8.67143124e-02
1.95096910e-01 -2.80910969e-01 -7.42551625e-01 1.26899436e-01
7.98922181e-01 -6.62528276e-01 -8.08344483e-01 5.82885683e-01
6.89911783e-01 1.11448988e-01 7.59180009e-01 -1.31563711e+00
-1.54777861e+00 2.18395159e-01 -3.77758861e-01 5.94727434e-02
7.19578743e-01 7.19970107e-01 1.19652653e+00 -1.01952922e+00
9.13377702e-01 1.21968448e+00 5.12000084e-01 8.31538558e-01
-1.18048656e+00 -2.35265195e-01 -1.80455983e-01 3.76817763e-01
-1.13838756e+00 -9.03211594e-01 1.85443416e-01 -6.54318929e-01
1.57747293e+00 2.55901571e-02 7.57584214e-01 1.47308230e+00
-2.34205261e-01 9.38835800e-01 9.88521516e-01 -5.83471596e-01
1.85425952e-01 -1.80857480e-01 -6.87952712e-02 9.80893672e-01
-3.11067134e-01 2.32418522e-01 -1.33843458e+00 6.41280264e-02
7.20585048e-01 1.83355644e-01 -3.13050866e-01 -3.57357353e-01
-1.71243811e+00 4.35890168e-01 4.34711367e-01 2.33613029e-01
-6.24840520e-02 5.89893639e-01 5.11792302e-01 5.17373443e-01
2.73674369e-01 3.31862062e-01 -4.92298789e-02 -3.69234920e-01
-8.07837665e-01 -1.49880007e-01 7.45072246e-01 1.00616455e+00
9.05076027e-01 9.65743661e-02 -3.94769222e-01 7.51029253e-01
9.65972170e-02 8.14014137e-01 4.45483983e-01 -1.11248159e+00
1.09466441e-01 4.35818255e-01 -1.61201179e-01 -4.40272212e-01
2.42920853e-02 2.48358831e-01 -7.84979403e-01 1.52825028e-01
1.43749416e-02 2.44851738e-01 -1.27681494e+00 1.78822124e+00
-1.06303968e-01 4.71968502e-01 3.01038027e-01 8.74343038e-01
1.30842686e+00 8.24419320e-01 -1.11637034e-01 1.58899143e-01
1.43345678e+00 -1.23423409e+00 -4.50075984e-01 -7.33878836e-02
4.19588119e-01 -4.97784346e-01 1.27325177e+00 2.92610228e-01
-1.14953923e+00 -6.07239723e-01 -1.00210559e+00 -3.52096498e-01
-7.80097187e-01 -2.21743137e-01 6.82865858e-01 1.51980594e-01
-1.41722310e+00 2.98596352e-01 -6.28060043e-01 -8.33059967e-01
4.67400312e-01 -5.52949216e-03 -8.68785262e-01 -3.22834879e-01
-9.17464793e-01 9.84316707e-01 1.24556959e-01 -5.31082451e-01
-1.53690279e+00 -7.71499395e-01 -1.09514654e+00 -4.51642945e-02
7.08774403e-02 -1.06712687e+00 1.11564481e+00 -6.68616295e-01
-1.33448362e+00 1.35302114e+00 -9.92952138e-02 -4.10566598e-01
1.08528398e-01 -4.03687172e-02 -4.85575169e-01 5.69891989e-01
7.66298920e-02 1.29292226e+00 1.22683644e+00 -1.12059593e+00
-3.10265929e-01 -2.25431487e-01 3.72882962e-01 2.09886953e-01
-7.92415798e-01 -1.69342399e-01 -4.55248713e-01 -2.46823002e-02
-3.89799982e-01 -7.16263473e-01 3.39323908e-01 4.66714710e-01
-2.06662551e-01 -1.18928149e-01 9.34480965e-01 -1.63353503e-01
7.39106238e-01 -2.33208370e+00 4.79719013e-01 -4.37375695e-01
3.37778091e-01 -7.38248229e-02 -5.79233825e-01 8.04273963e-01
2.24588498e-01 1.65471795e-03 -8.45151916e-02 -7.60399103e-01
4.32505012e-01 3.37949514e-01 -4.78584021e-01 3.98635119e-01
3.37318689e-01 1.33831489e+00 -8.75982583e-01 -4.77144420e-01
4.66190904e-01 6.35972321e-01 -4.77958411e-01 3.31360459e-01
-2.60012209e-01 2.53614988e-02 3.52151275e-01 9.75160003e-01
9.88125876e-02 -5.91343582e-01 -3.76359582e-01 -2.36887306e-01
-7.55275041e-02 -8.42540264e-02 -5.93407989e-01 2.43372202e+00
-4.83165473e-01 8.82966995e-01 3.45088691e-02 -7.62108207e-01
4.04506117e-01 3.87439966e-01 4.27679509e-01 -1.01139903e+00
-9.02648866e-02 -3.88167463e-02 -3.75698715e-01 -8.25448811e-01
5.64144969e-01 -1.73450008e-01 -1.91487178e-01 5.94926417e-01
7.20595181e-01 -3.62882555e-01 1.06415793e-01 5.32568038e-01
1.38940930e+00 1.52508035e-01 -1.75439902e-02 2.44996138e-03
-3.23974416e-02 -1.16481870e-01 -1.98597938e-01 1.05766213e+00
-1.52741969e-01 8.17723811e-01 1.06535777e-01 -2.22979844e-01
-1.26397681e+00 -1.54927886e+00 6.44445652e-03 1.43727374e+00
1.71863392e-01 -4.87718523e-01 -3.88717204e-01 -1.98646322e-01
4.41713147e-02 2.56565064e-01 -8.80109251e-01 -3.66960555e-01
3.60185653e-01 -2.35085040e-01 8.79154027e-01 5.75010180e-01
4.89759713e-01 -9.50592458e-01 -8.82722855e-01 -3.12618017e-01
-2.11607561e-01 -1.38148785e+00 -3.88277292e-01 2.67745465e-01
-2.89481670e-01 -8.52000237e-01 -8.69005740e-01 -8.25859129e-01
2.40492105e-01 6.81225657e-01 1.19995165e+00 -5.38593233e-02
-7.45668411e-01 1.38897908e+00 -5.07258117e-01 -1.86106756e-01
-9.97944698e-02 -2.67468512e-01 2.79309243e-01 -2.30334327e-01
2.58811891e-01 -7.36159146e-01 -7.03243852e-01 -1.17058322e-01
-1.14510119e+00 1.12926304e-01 4.74948853e-01 1.08939719e+00
1.76050246e-01 -6.22917593e-01 2.66887188e-01 -5.80207109e-02
4.60003018e-01 -4.88489747e-01 7.77535886e-02 3.93796235e-01
-3.05647478e-02 7.49596953e-02 3.46523374e-01 -6.00734890e-01
-7.43347704e-01 -2.05912203e-01 2.92093813e-01 -8.72456312e-01
-2.81634629e-01 3.90133619e-01 2.64428079e-01 -1.58062235e-01
8.99392486e-01 4.73475456e-01 2.12829411e-01 -1.36741862e-01
1.00466454e+00 6.22267663e-01 8.95238280e-01 -5.45458853e-01
6.19214833e-01 9.49876308e-01 -5.66777103e-02 -1.25816441e+00
-5.64019382e-01 -6.36277258e-01 -5.15634656e-01 -2.38457561e-01
1.19367957e+00 -1.36967444e+00 -8.22890520e-01 5.28408647e-01
-1.21451867e+00 -5.02543390e-01 -5.70542276e-01 1.88975170e-01
-6.44334137e-01 4.30171251e-01 -8.73215377e-01 -7.10978150e-01
-3.97318840e-01 -9.75445271e-01 1.59007454e+00 9.63535011e-02
-1.49816841e-01 -8.91863465e-01 1.63281098e-01 3.46615672e-01
5.21747470e-01 -4.60520126e-02 7.69081235e-01 -1.74291223e-01
-7.10367441e-01 7.05756620e-02 -5.88422775e-01 3.19912344e-01
-1.11392334e-01 2.94139213e-03 -1.30164611e+00 -3.36209893e-01
-5.08607924e-01 -1.23710549e+00 1.35682237e+00 -3.73443589e-02
7.88616955e-01 -6.60017878e-02 -9.34555233e-02 7.03108132e-01
1.46896195e+00 -3.69532675e-01 6.94924235e-01 6.76465183e-02
7.11400568e-01 2.61529088e-01 -1.21582046e-01 7.05173194e-01
6.49102807e-01 3.19286615e-01 6.14840686e-01 -1.19581446e-01
-3.61761749e-01 -2.46427700e-01 7.59813726e-01 1.03743982e+00
-2.02926733e-02 -2.03029394e-01 -8.57807457e-01 8.58586729e-01
-1.84630609e+00 -1.24580181e+00 3.53386641e-01 1.85898137e+00
6.11126542e-01 -2.12097079e-01 9.48466584e-02 -2.93483883e-01
1.92368820e-01 3.70174736e-01 -5.39181054e-01 -3.59644234e-01
-3.83692592e-01 3.83640796e-01 6.70338050e-02 2.86642730e-01
-9.30787921e-01 8.42801452e-01 7.34682894e+00 6.40874028e-01
-1.05240846e+00 2.81461358e-01 1.18971698e-01 -3.77906859e-01
-5.74096739e-01 1.76496103e-01 -3.65094483e-01 2.98060924e-01
8.84661853e-01 2.50499785e-01 7.66570330e-01 5.23601532e-01
-4.84107941e-01 -2.06779554e-01 -1.42857504e+00 1.47530782e+00
5.72718561e-01 -1.64612615e+00 2.65999287e-01 4.49051894e-02
7.52635062e-01 4.86542642e-01 2.62757361e-01 4.29212719e-01
2.93547481e-01 -1.28928864e+00 7.83593535e-01 8.77645850e-01
1.25497091e+00 -2.39712093e-02 1.42625406e-01 1.81452006e-01
-1.12926531e+00 -9.85141471e-02 -2.41964921e-01 -3.05695623e-01
3.50757092e-02 1.04466721e-01 -3.82763147e-01 3.48073363e-01
8.73195767e-01 1.03483355e+00 -5.39148569e-01 5.37907422e-01
2.32265949e-01 5.73952831e-02 -3.45400453e-01 6.58666268e-02
1.93098873e-01 2.93720484e-01 4.60808575e-01 1.33770919e+00
4.41341788e-01 -2.14511808e-02 1.46357223e-01 1.00038302e+00
-1.74433753e-01 -3.73357773e-01 -1.21200573e+00 -4.81727242e-01
4.22587186e-01 1.02526224e+00 -1.27787068e-01 -5.68883061e-01
-7.29164243e-01 1.48390603e+00 2.26823166e-01 3.52132797e-01
-6.93186998e-01 -2.96933591e-01 6.96228325e-01 -3.03853989e-01
3.76539439e-01 -3.72605175e-01 1.00218087e-01 -1.57059073e+00
-2.51214921e-01 -7.90949225e-01 1.59142241e-01 -1.33929694e+00
-1.69134259e+00 1.99083984e-01 -4.01209779e-02 -1.30916166e+00
-3.05495381e-01 -8.79994929e-01 -4.93267536e-01 3.71334583e-01
-1.28681791e+00 -1.67404413e+00 -5.96338809e-01 9.52401221e-01
3.11991394e-01 -3.46599668e-01 1.30566239e+00 1.58298567e-01
-1.79980248e-01 4.01751399e-01 1.11927548e-02 6.29151016e-02
7.89847374e-01 -1.09497452e+00 2.67762125e-01 5.40401161e-01
4.69397873e-01 5.27970076e-01 4.70878661e-01 -1.37050441e-02
-2.14797902e+00 -4.48880702e-01 2.37610385e-01 -6.85861945e-01
9.94642675e-01 -4.86551136e-01 -4.49412733e-01 4.94666755e-01
9.12568450e-01 3.06491882e-01 8.29277396e-01 1.30396262e-01
-1.13067830e+00 -4.44821119e-02 -6.91934049e-01 6.55952573e-01
1.36101878e+00 -1.56310511e+00 -6.37198687e-01 2.09330603e-01
1.06528747e+00 -1.96545757e-02 -9.54616129e-01 3.29334825e-01
8.56197596e-01 -1.05227387e+00 1.17646742e+00 -6.39678717e-01
1.04009902e+00 -8.92975628e-02 -7.05609441e-01 -1.19360471e+00
-2.41323948e-01 -5.14641523e-01 -5.32157063e-01 1.10160840e+00
1.46200389e-01 -2.45919570e-01 3.98623288e-01 2.43272960e-01
-1.05030835e-01 -5.07207870e-01 -9.04537618e-01 -8.34885895e-01
-2.90593207e-01 -5.58138192e-01 2.65186369e-01 9.30180609e-01
5.24093330e-01 7.51071215e-01 -5.90529680e-01 -2.49645248e-01
6.87085092e-01 1.69950247e-01 8.89393866e-01 -7.35635102e-01
-7.34861255e-01 -4.32329357e-01 -8.54368210e-01 -1.15180278e+00
1.32747531e-01 -9.40513074e-01 3.68670039e-02 -1.67197669e+00
5.89825749e-01 2.00729132e-01 -5.05455911e-01 8.05938661e-01
4.09311086e-01 6.74951375e-01 6.04389906e-01 2.87225515e-01
-1.14787900e+00 8.09190035e-01 1.02272749e+00 -6.76180780e-01
2.66437054e-01 -1.02565134e+00 -5.90555310e-01 2.55681366e-01
2.73354501e-01 2.35109612e-01 -4.71632093e-01 -7.71648109e-01
2.40263745e-01 -3.15318182e-02 8.44701469e-01 -1.19790387e+00
4.73405719e-01 1.23818561e-01 2.75667787e-01 -1.71788439e-01
1.02567911e+00 -5.54471731e-01 -1.70012534e-01 7.83315524e-02
-2.69650370e-01 -1.19543783e-01 2.83540934e-01 6.66160643e-01
-1.88806340e-01 1.75467536e-01 4.94953990e-01 -4.60356504e-01
-1.25937569e+00 2.66257882e-01 -4.04921770e-01 8.30496177e-02
7.80089796e-01 -2.34839872e-01 -1.03463316e+00 -6.60853088e-01
-6.92766368e-01 1.69703186e-01 7.49156296e-01 6.07519209e-01
9.38327968e-01 -1.36656773e+00 -4.34859276e-01 1.65893629e-01
7.96352684e-01 -3.97057980e-01 3.95858556e-01 8.70788455e-01
-1.62160456e-01 2.04841226e-01 -4.09284890e-01 -9.60807383e-01
-1.16119051e+00 7.75742233e-01 1.62904281e-02 3.62700015e-01
-4.90937799e-01 9.25769806e-01 1.45834446e-01 -3.17580849e-01
3.47792089e-01 -1.72429487e-01 3.13608766e-01 1.99326485e-01
5.83042741e-01 1.15097061e-01 -1.90682679e-01 -5.67298234e-01
-3.41889471e-01 6.38797879e-01 2.19510406e-01 -5.52506447e-01
1.35598826e+00 -1.64631844e-01 -3.47290933e-01 1.00203979e+00
1.34987259e+00 -5.26618063e-01 -1.12823224e+00 -3.91951442e-01
-6.75483942e-01 -1.81092709e-01 -3.88118178e-02 -6.98338270e-01
-5.09294033e-01 1.40238249e+00 7.03052223e-01 2.66542703e-01
1.10075366e+00 2.47630090e-01 8.11150014e-01 6.16576374e-01
5.00188589e-01 -1.06669331e+00 8.43103051e-01 7.85634041e-01
6.20066702e-01 -1.50512564e+00 -2.44063720e-01 2.19819456e-01
-7.35614181e-01 9.44593132e-01 5.47809899e-01 3.92518081e-02
6.09719038e-01 2.69091040e-01 -2.43889958e-01 -2.81109393e-01
-1.30342722e+00 -7.99976647e-01 2.93222725e-01 7.14699924e-01
2.81390578e-01 1.10248057e-02 5.39021552e-01 1.81838926e-02
2.27943778e-01 3.20450626e-02 2.68787861e-01 1.04082453e+00
-3.74630362e-01 -7.35094488e-01 -1.53164044e-01 3.56564999e-01
4.46026884e-02 -4.83251691e-01 -6.05936050e-01 5.67373574e-01
1.67536408e-01 8.69828820e-01 2.56360024e-01 -8.44887316e-01
8.54017586e-03 2.71864325e-01 1.02691329e+00 -5.12213588e-01
-2.82400280e-01 -3.19322228e-01 -1.28371771e-02 -6.65622294e-01
-6.15555525e-01 -2.38264322e-01 -9.62807238e-01 -4.24263567e-01
-6.03146255e-02 -4.49755400e-01 4.82838035e-01 8.81736517e-01
7.54795432e-01 3.01181614e-01 2.24494115e-01 -1.28143907e+00
-3.57259572e-01 -8.55174124e-01 -3.91790777e-01 7.15437710e-01
7.32827067e-01 -5.45411408e-01 -3.89193356e-01 2.96759218e-01] | [10.4949369430542, 1.4077117443084717] |
22b116df-159b-4317-a742-66b44e615f72 | av-data2vec-self-supervised-learning-of-audio | 2302.06419 | null | https://arxiv.org/abs/2302.06419v1 | https://arxiv.org/pdf/2302.06419v1.pdf | AV-data2vec: Self-supervised Learning of Audio-Visual Speech Representations with Contextualized Target Representations | Self-supervision has shown great potential for audio-visual speech recognition by vastly reducing the amount of labeled data required to build good systems. However, existing methods are either not entirely end-to-end or do not train joint representations of both modalities. In this paper, we introduce AV-data2vec which addresses these challenges and builds audio-visual representations based on predicting contextualized representations which has been successful in the uni-modal case. The model uses a shared transformer encoder for both audio and video and can combine both modalities to improve speech recognition. Results on LRS3 show that AV-data2vec consistently outperforms existing methods under most settings. | ['Michael Auli', 'Wei-Ning Hsu', 'Alexei Baevski', 'Jiachen Lian'] | 2023-02-10 | null | null | null | null | ['audio-visual-speech-recognition'] | ['speech'] | [ 1.60824969e-01 1.07996345e-01 -3.55866700e-01 -6.03517354e-01
-1.35909486e+00 -3.14269364e-01 8.60084116e-01 -4.88346636e-01
-2.00567320e-01 5.30318737e-01 8.58746648e-01 -4.33375657e-01
4.71035928e-01 -1.58024266e-01 -7.04028487e-01 -4.70437706e-01
2.86680698e-01 4.23406720e-01 -5.45449220e-02 -6.73338026e-02
-1.66236699e-01 1.00824609e-01 -1.72663915e+00 8.43148112e-01
3.58458497e-02 1.02943897e+00 5.48881963e-02 9.99280214e-01
-1.39880985e-01 1.03816283e+00 -4.19067562e-01 -1.88931599e-01
7.67139765e-03 -3.37350488e-01 -9.21511114e-01 4.07409996e-01
9.15481329e-01 -4.41771328e-01 -8.32816303e-01 4.59388107e-01
6.72577679e-01 2.81922579e-01 6.00134969e-01 -1.31269360e+00
-1.06180334e+00 4.20409322e-01 -3.88989836e-01 2.78579324e-01
4.67219055e-01 4.77980748e-02 1.29051149e+00 -1.23082161e+00
5.74836612e-01 1.39351869e+00 3.54577720e-01 8.98639739e-01
-1.37059093e+00 -6.06646895e-01 8.73554051e-02 6.42981112e-01
-1.16768992e+00 -1.07332087e+00 5.59091210e-01 -3.10745150e-01
1.58586586e+00 8.76028687e-02 2.73197234e-01 1.69201541e+00
-5.16464531e-01 1.24788439e+00 1.08953631e+00 -5.08440077e-01
-1.91306204e-01 3.58523399e-01 -1.02763459e-01 2.52944738e-01
-6.77832305e-01 3.22581053e-01 -1.01878142e+00 1.35856748e-01
5.02411604e-01 -6.04316294e-02 -2.61367023e-01 -3.38312566e-01
-1.03771424e+00 8.64973187e-01 3.20988923e-01 3.90974820e-01
-6.90191463e-02 1.50331691e-01 6.70709610e-01 6.12530529e-01
4.99662638e-01 8.54822050e-04 -4.75355744e-01 -2.91465461e-01
-9.82166469e-01 -1.46694988e-01 3.36744457e-01 7.71362424e-01
4.70575899e-01 6.78411603e-01 -2.36782312e-01 1.26805747e+00
3.56504291e-01 5.67826450e-01 6.45200610e-01 -7.80178189e-01
5.73644280e-01 2.00851038e-01 -2.78831571e-01 -4.04146761e-01
-7.06982389e-02 -1.17259905e-01 -5.36531329e-01 2.43598133e-01
6.21326119e-02 1.03131615e-01 -1.39093113e+00 1.76671886e+00
3.56309824e-02 5.07841229e-01 4.10306901e-01 7.15751112e-01
1.26512682e+00 8.79670262e-01 3.07096839e-01 -1.97213255e-02
9.90129292e-01 -1.32435906e+00 -9.02789295e-01 -4.39758539e-01
4.80596572e-01 -6.74600899e-01 1.05092585e+00 2.05484375e-01
-1.13957298e+00 -6.79760814e-01 -9.57842052e-01 -3.01653445e-01
-3.10689747e-01 9.50944573e-02 1.81342766e-01 5.66903412e-01
-1.43984914e+00 6.95880800e-02 -7.62977540e-01 -2.99609482e-01
5.31759322e-01 1.85302794e-01 -8.96103501e-01 -3.58774453e-01
-9.71757829e-01 1.08301997e+00 8.16854276e-03 -3.12948465e-01
-1.36045992e+00 -5.27420402e-01 -1.15413535e+00 9.14878771e-02
1.36441991e-01 -3.88390154e-01 1.39998019e+00 -1.07580674e+00
-1.42089272e+00 7.96136975e-01 -4.39326018e-01 -5.96297443e-01
6.35143220e-02 -7.40404949e-02 -7.30468690e-01 2.35332578e-01
-8.97073671e-02 1.04414904e+00 1.11947262e+00 -1.19509113e+00
-5.40074110e-01 -2.87894994e-01 -1.64439231e-01 3.98981035e-01
-6.43649817e-01 1.21456578e-01 -4.27937686e-01 -5.43036163e-01
-1.73870772e-01 -6.92894459e-01 1.08488269e-01 -3.22566092e-01
-1.45521179e-01 -4.22663152e-01 1.15858591e+00 -8.12726140e-01
6.65990293e-01 -2.50670099e+00 2.15732142e-01 -1.69189170e-01
-4.95756976e-02 6.33873284e-01 -5.26986420e-01 5.51617205e-01
-3.92829835e-01 3.49005871e-02 1.60415873e-01 -9.54871595e-01
-1.13664396e-01 4.34450924e-01 -4.77358043e-01 2.66793638e-01
3.20844769e-01 7.20337272e-01 -6.21162355e-01 -4.58422065e-01
5.40358722e-01 1.18520021e+00 -3.79981339e-01 4.40771878e-01
4.57894895e-03 3.66335481e-01 2.53913492e-01 6.52786076e-01
3.49186182e-01 -1.61827922e-01 1.86328769e-01 -3.94282565e-02
2.56420732e-01 7.28931427e-01 -8.40486705e-01 1.76198816e+00
-7.83578157e-01 1.12333477e+00 2.23183766e-01 -1.11934257e+00
7.82961071e-01 8.88323665e-01 2.99330533e-01 -9.26972032e-01
2.71877134e-03 -6.38070107e-02 -4.51478392e-01 -4.55573738e-01
2.51292765e-01 -4.96991605e-01 3.56889963e-01 2.92335600e-01
6.76649988e-01 -2.31191814e-02 -2.06664145e-01 3.75369579e-01
1.05370295e+00 -2.80158315e-03 3.76132280e-02 5.89370847e-01
4.40635800e-01 -1.37760147e-01 2.87493795e-01 4.26689446e-01
-2.60821581e-01 1.02090478e+00 7.08260536e-02 -1.18810020e-01
-1.07567894e+00 -1.21005678e+00 -1.25272050e-01 1.32226312e+00
-4.23147798e-01 -6.40580297e-01 -3.00902307e-01 -8.18921089e-01
5.29819615e-02 7.71690547e-01 -6.04693234e-01 -1.14955297e-02
-2.84546055e-02 1.24708503e-01 6.42275751e-01 1.01414335e+00
-3.80759314e-03 -1.11452663e+00 9.40793008e-02 4.40271199e-02
-3.12003613e-01 -1.47184420e+00 -2.88310230e-01 2.07885191e-01
-6.17862344e-01 -7.45282531e-01 -7.12685347e-01 -9.02827621e-01
2.37415254e-01 6.81882739e-01 1.10695124e+00 -8.61238465e-02
-2.10881948e-01 7.89711297e-01 -5.37156999e-01 -3.35144013e-01
-3.60477746e-01 -2.57757425e-01 1.68467432e-01 1.94826171e-01
4.53818530e-01 -5.59052944e-01 -2.25199133e-01 2.12962955e-01
-5.89144051e-01 -1.34876713e-01 3.52519453e-01 1.20272160e+00
4.59449232e-01 -4.15284872e-01 7.13213503e-01 -4.99997526e-01
2.04278097e-01 -5.29972076e-01 6.83487132e-02 2.13334933e-01
-3.91804874e-01 -2.13100508e-01 3.22837740e-01 -3.55866045e-01
-1.03209805e+00 1.32759213e-01 -4.25378591e-01 -1.02033138e+00
-4.63598818e-01 4.53897476e-01 -7.25191785e-03 2.50592440e-01
4.01577950e-01 2.08030224e-01 3.00679833e-01 -5.93124926e-01
5.72742105e-01 1.17402387e+00 6.05208993e-01 5.49477078e-02
4.04601991e-01 4.15902197e-01 -5.49399912e-01 -1.34239042e+00
-6.91065073e-01 -6.59174562e-01 -3.67595941e-01 -6.94506317e-02
8.68843734e-01 -1.46834266e+00 -3.28039020e-01 1.74504057e-01
-9.04940486e-01 -3.43567997e-01 -3.98460716e-01 6.21307194e-01
-6.10171080e-01 3.23825121e-01 -6.10370517e-01 -9.69428599e-01
-8.41450393e-02 -1.22320604e+00 1.05436218e+00 -1.42906785e-01
-1.79297969e-01 -8.69777441e-01 1.73332006e-01 9.19302523e-01
5.19007921e-01 -3.65480363e-01 4.30095285e-01 -8.73512447e-01
-3.29468369e-01 -2.91513950e-01 -3.08631450e-01 7.48979449e-01
5.08894660e-02 -3.91020402e-02 -1.68801582e+00 -3.76765639e-01
-4.50656921e-01 -1.10249913e+00 1.14442515e+00 2.80659884e-01
9.51343954e-01 -4.10238951e-02 -2.12101154e-02 4.24366832e-01
9.26639616e-01 4.28732149e-02 6.57800198e-01 6.52566776e-02
8.05149138e-01 6.23590469e-01 2.70540178e-01 2.26901621e-01
4.73388046e-01 9.59992051e-01 4.37143236e-01 -1.27583995e-01
-6.89776838e-01 -4.48390573e-01 7.03047276e-01 9.03659105e-01
2.65340596e-01 -2.83592403e-01 -7.86907554e-01 9.78975654e-01
-1.57508934e+00 -1.19005227e+00 1.76722273e-01 1.82324421e+00
6.87717199e-01 -2.84119487e-01 3.31645310e-01 3.02187234e-01
4.69238967e-01 3.83763522e-01 -2.02804044e-01 -5.14463365e-01
-2.91359633e-01 3.68404657e-01 -3.39324810e-02 6.47403121e-01
-1.08504176e+00 9.94426966e-01 7.42882490e+00 6.74261987e-01
-1.06948900e+00 5.28922081e-01 2.47820914e-01 -4.53975052e-01
-3.39247972e-01 -1.03459924e-01 -6.56936526e-01 1.94102544e-02
1.37296629e+00 2.33391643e-01 3.19292158e-01 8.44720244e-01
-1.49006754e-01 4.22485739e-01 -1.18250275e+00 1.28082979e+00
4.37925667e-01 -1.28497636e+00 1.21511787e-01 -1.01005975e-02
5.21342576e-01 5.47318757e-01 3.54752898e-01 4.65280294e-01
3.85850310e-01 -1.32114220e+00 5.18588066e-01 1.27219468e-01
1.13702846e+00 -6.01523995e-01 4.81719166e-01 -1.87620111e-02
-1.01460850e+00 -9.24031287e-02 -1.32438257e-01 6.89263940e-02
3.46471131e-01 -1.19190380e-01 -1.22430265e+00 2.18783274e-01
8.65832627e-01 8.90965283e-01 -4.47786033e-01 6.59601510e-01
-2.32173786e-01 8.82052302e-01 -3.03798139e-01 4.21582401e-01
3.67256522e-01 4.38571870e-01 3.58117908e-01 1.40664196e+00
1.07083477e-01 -2.45301157e-01 2.13585094e-01 3.02878290e-01
-1.14464819e-01 1.77958995e-01 -1.04114890e+00 -3.95246804e-01
4.30497408e-01 8.53126705e-01 -6.00492954e-02 -5.45765042e-01
-9.40211594e-01 7.61229157e-01 4.97610092e-01 3.88150215e-01
-4.71937418e-01 5.91918193e-02 8.91066015e-01 -4.14218195e-02
5.73083997e-01 -1.85638726e-01 -1.52503982e-01 -1.21947539e+00
-1.57036975e-01 -1.01209486e+00 4.67408627e-01 -1.06769168e+00
-1.17259026e+00 7.67366588e-01 -2.96851128e-01 -1.18929505e+00
-6.06996715e-01 -6.62728786e-01 -3.83137643e-01 7.43198395e-01
-1.46905971e+00 -1.55030465e+00 -6.30247733e-03 9.75333333e-01
9.54186678e-01 -6.28735244e-01 1.12420559e+00 4.40525800e-01
-3.96964729e-01 8.73864472e-01 6.33987710e-02 8.87270197e-02
1.03039896e+00 -1.19978249e+00 4.33562338e-01 6.30324721e-01
8.47114682e-01 2.60052562e-01 6.59448862e-01 -4.19678330e-01
-1.35784364e+00 -7.04100013e-01 1.01830757e+00 -5.47184467e-01
6.16373420e-01 -3.17734629e-01 -9.90332365e-01 9.84669149e-01
7.40326524e-01 2.76645750e-01 1.05186272e+00 4.41497713e-01
-1.09217155e+00 -3.86253707e-02 -8.25435996e-01 3.17756146e-01
8.12886298e-01 -1.36190271e+00 -7.28073716e-01 3.16832155e-01
7.83594370e-01 -1.19870476e-01 -7.23667920e-01 2.50098050e-01
4.46145386e-01 -8.67686987e-01 1.21702039e+00 -1.05600321e+00
3.91251981e-01 -1.12496829e-02 -6.29861474e-01 -1.42899406e+00
-1.95330113e-01 -1.89120844e-01 -2.08008796e-01 1.44288945e+00
5.36660433e-01 -3.21169853e-01 6.57807767e-01 3.00166011e-01
-3.24918687e-01 -4.16346848e-01 -1.23196578e+00 -7.92848706e-01
-1.58686101e-01 -9.19435024e-01 3.12910855e-01 9.72944081e-01
2.55223662e-01 7.72414386e-01 -7.60586441e-01 6.72465889e-04
4.25378084e-01 -2.46758074e-01 8.35352480e-01 -7.08626091e-01
-3.34284902e-01 -3.46540660e-01 -7.67001390e-01 -1.04584455e+00
5.46272933e-01 -9.23914075e-01 -5.12077063e-02 -1.60337293e+00
6.12307377e-02 -8.62482935e-03 -4.81080174e-01 8.63190532e-01
1.49098471e-01 5.91110826e-01 2.45837390e-01 8.38750228e-02
-5.28719008e-01 7.31808960e-01 7.81262457e-01 -4.71869200e-01
2.36832500e-02 -2.76987731e-01 -5.17899692e-01 3.73882622e-01
5.67371607e-01 -8.99523571e-02 -5.72060585e-01 -6.70203030e-01
-2.84119189e-01 2.23461598e-01 4.15068775e-01 -8.97033930e-01
8.37692842e-02 1.90365791e-01 3.46928567e-01 -7.21397817e-01
9.68926787e-01 -8.31861198e-01 -2.47314632e-01 -2.22970724e-01
-6.17758453e-01 -1.39447376e-01 2.35290751e-01 7.13963091e-01
-7.75213718e-01 8.36118311e-02 8.25396001e-01 1.38683096e-01
-8.85863066e-01 1.74669281e-01 -4.44292396e-01 5.99494204e-02
7.92715251e-01 -7.23691210e-02 -2.42814660e-01 -8.67166519e-01
-1.21081686e+00 1.83790103e-01 2.03034177e-01 8.01842153e-01
9.47589755e-01 -1.74276280e+00 -7.37956285e-01 3.69848967e-01
3.90134960e-01 -5.21926105e-01 6.05893850e-01 6.84383512e-01
9.86317992e-02 5.30041099e-01 -1.64800286e-01 -7.74967372e-01
-1.86048603e+00 6.71975195e-01 1.31802201e-01 2.32432649e-01
-6.32699251e-01 1.10982656e+00 6.07961081e-02 -3.62292588e-01
7.69151449e-01 2.72504419e-01 -4.51965362e-01 3.28257978e-01
8.68663788e-01 1.28550962e-01 1.40946150e-01 -1.14906704e+00
-4.50857580e-01 1.94185942e-01 -2.93262988e-01 -5.78685462e-01
1.51723123e+00 -2.08718523e-01 4.41876411e-01 7.05131114e-01
1.64029348e+00 -3.75436306e-01 -1.31878483e+00 -5.07164240e-01
-3.88888925e-01 -5.21588743e-01 4.17081505e-01 -7.67285764e-01
-1.37115467e+00 1.50544453e+00 6.76764250e-01 1.06958985e-01
8.77951384e-01 2.88115323e-01 7.56552219e-01 2.12304726e-01
-5.72311319e-02 -1.03357947e+00 2.23744676e-01 5.56395113e-01
1.00136244e+00 -1.45262206e+00 -2.35179856e-01 5.17543554e-02
-1.27094960e+00 8.57634246e-01 6.22491717e-01 1.69079080e-01
6.46031499e-01 1.41984478e-01 4.45993602e-01 -5.23308106e-02
-1.12482214e+00 -5.51456451e-01 4.34093565e-01 9.29276526e-01
6.64553702e-01 1.20530926e-01 5.62918961e-01 2.76639432e-01
7.60435835e-02 -1.17972732e-01 2.28108928e-01 9.04717028e-01
-1.99541017e-01 -1.22025454e+00 -3.66253227e-01 1.72877774e-01
-4.56886649e-01 -5.85047603e-02 -6.03449404e-01 4.76013601e-01
-2.64771581e-01 1.32600164e+00 1.14189453e-01 -8.48373055e-01
3.64320666e-01 5.47791541e-01 3.96905571e-01 -6.53879344e-01
-3.82555276e-01 4.81534004e-01 4.61797833e-01 -6.11005604e-01
-5.31939983e-01 -8.32700670e-01 -9.07053828e-01 -4.76612933e-02
-2.52533346e-01 9.14783776e-03 7.57444561e-01 9.31855619e-01
4.86440569e-01 4.70953226e-01 7.44418323e-01 -1.11499250e+00
-4.89250243e-01 -1.21389103e+00 -4.41713899e-01 5.13109446e-01
7.51087606e-01 -8.51663113e-01 -5.76270401e-01 6.43182918e-02] | [14.382980346679688, 5.125270366668701] |
94e85590-c310-42ce-822b-c404304570a1 | a-general-gaussian-heatmap-labeling-for | 2109.12848 | null | https://arxiv.org/abs/2109.12848v4 | https://arxiv.org/pdf/2109.12848v4.pdf | A General Gaussian Heatmap Label Assignment for Arbitrary-Oriented Object Detection | Recently, many arbitrary-oriented object detection (AOOD) methods have been proposed and attracted widespread attention in many fields. However, most of them are based on anchor-boxes or standard Gaussian heatmaps. Such label assignment strategy may not only fail to reflect the shape and direction characteristics of arbitrary-oriented objects, but also have high parameter-tuning efforts. In this paper, a novel AOOD method called General Gaussian Heatmap Label Assignment (GGHL) is proposed. Specifically, an anchor-free object-adaptation label assignment (OLA) strategy is presented to define the positive candidates based on two-dimensional (2-D) oriented Gaussian heatmaps, which reflect the shape and direction features of arbitrary-oriented objects. Based on OLA, an oriented-bounding-box (OBB) representation component (ORC) is developed to indicate OBBs and adjust the Gaussian center prior weights to fit the characteristics of different objects adaptively through neural network learning. Moreover, a joint-optimization loss (JOL) with area normalization and dynamic confidence weighting is designed to refine the misalign optimal results of different subtasks. Extensive experiments on public datasets demonstrate that the proposed GGHL improves the AOOD performance with low parameter-tuning and time costs. Furthermore, it is generally applicable to most AOOD methods to improve their performance including lightweight models on embedded platforms. | ['Ran Tao', 'Xiang-Gen Xia', 'Wei Li', 'Zhanchao Huang'] | 2021-09-27 | null | null | null | null | ['object-detection-in-aerial-images'] | ['computer-vision'] | [-2.65377343e-01 -2.28808165e-01 -1.33447841e-01 -4.63814408e-01
-5.00072896e-01 -1.87583014e-01 2.28379279e-01 2.06289470e-01
-2.59402186e-01 8.51401985e-02 -1.97470441e-01 -7.61767477e-02
-1.93102330e-01 -5.45750439e-01 -2.27300107e-01 -9.63468015e-01
-3.19522321e-02 3.32848310e-01 8.32336605e-01 1.68633521e-01
4.22633082e-01 7.01331913e-01 -1.33087409e+00 -7.39406422e-02
1.03239965e+00 1.35683894e+00 3.66400212e-01 2.47408926e-01
-1.73297554e-01 2.51390725e-01 -5.19510686e-01 -4.25383598e-02
1.46310538e-01 1.88342407e-01 -2.55579650e-01 2.66543925e-01
3.41267705e-01 -1.10076152e-01 -1.56960618e-02 1.09424198e+00
7.96568155e-01 1.88612550e-01 8.05942774e-01 -1.57253981e+00
-7.04427660e-01 8.78928006e-02 -7.68897295e-01 9.49959010e-02
-2.44856119e-01 1.14195913e-01 8.19288552e-01 -1.32000244e+00
8.59688148e-02 1.45634449e+00 6.83747530e-01 2.66271144e-01
-1.14144790e+00 -7.43109345e-01 5.52976727e-01 3.42735529e-01
-1.73339260e+00 2.09175497e-01 9.44586754e-01 -6.30067706e-01
5.19020855e-01 3.55168134e-01 5.58570802e-01 4.58007812e-01
3.03041160e-01 1.06188643e+00 9.66118932e-01 -3.46530437e-01
1.83035046e-01 3.65582436e-01 2.35479176e-01 7.22379804e-01
4.21080351e-01 -2.55324244e-01 -1.77722156e-01 -1.93061665e-01
5.53592980e-01 2.96259701e-01 -1.46381274e-01 -1.04321575e+00
-1.24678123e+00 6.97568357e-01 7.94367075e-01 5.99304074e-03
-2.47625366e-01 -4.26899418e-02 3.62303227e-01 -4.48668867e-01
4.82380152e-01 1.68885976e-01 -1.63646370e-01 3.37247372e-01
-3.72672200e-01 4.29192156e-01 2.41623312e-01 1.24193108e+00
8.95199180e-01 -2.45437726e-01 -7.13140428e-01 9.84312057e-01
7.56155193e-01 5.08092642e-01 5.28916597e-01 -5.13969362e-01
5.50328970e-01 1.07215130e+00 2.18026325e-01 -1.57717633e+00
-7.46333122e-01 -3.18312764e-01 -7.54443824e-01 2.13168725e-01
1.95316419e-01 6.04557544e-02 -9.25302207e-01 1.22083187e+00
8.28346074e-01 -1.29144236e-01 -3.77618819e-01 1.23275435e+00
6.19308114e-01 6.46792948e-01 2.43646443e-01 4.13880013e-02
1.69482803e+00 -9.32938039e-01 -6.54531896e-01 -3.65093738e-01
6.06930137e-01 -8.53287280e-01 1.15081465e+00 2.88209140e-01
-6.98090494e-01 -7.68408477e-01 -1.15731609e+00 1.88699678e-01
-3.70707303e-01 4.43133444e-01 6.60489440e-01 7.26841807e-01
-5.95744312e-01 1.26773473e-02 -8.22357535e-01 -1.27006844e-01
6.27751291e-01 2.56785721e-01 2.42610067e-01 -4.47655432e-02
-7.93468118e-01 4.55280632e-01 7.68749177e-01 2.95198411e-01
-8.27007830e-01 -4.95337427e-01 -6.89554870e-01 -1.14495754e-01
4.56707805e-01 -1.84671357e-01 9.23755169e-01 -3.92383933e-01
-1.25798154e+00 5.25868535e-01 1.80555239e-01 1.06957726e-01
4.66478080e-01 -1.40280113e-01 -5.23940384e-01 5.60244173e-03
5.27540110e-02 8.27288330e-01 9.61925447e-01 -1.17917752e+00
-1.11378956e+00 -5.92476606e-01 -2.71682769e-01 4.57317889e-01
-7.24300504e-01 1.29153892e-01 -8.27387989e-01 -8.81934702e-01
7.00815439e-01 -1.05710006e+00 -2.01307833e-01 3.58582228e-01
-6.63718224e-01 -7.14621425e-01 9.94399011e-01 -4.23709124e-01
1.68752539e+00 -2.38909364e+00 -6.99751079e-02 4.05509442e-01
2.56111026e-01 2.16818988e-01 1.77422851e-01 -2.55387962e-01
2.22649559e-01 -7.14592114e-02 -1.73512682e-01 -3.06564253e-02
2.70039469e-01 -3.10060769e-01 5.54160737e-02 4.96294439e-01
3.89272213e-01 5.48671842e-01 -7.15938270e-01 -8.44086409e-01
4.31155376e-02 4.39266741e-01 -4.25564021e-01 3.51715535e-01
-1.08326443e-01 -9.71577168e-02 -8.09902072e-01 9.05847132e-01
9.67538774e-01 -3.01973999e-01 -1.30539358e-01 -5.43979645e-01
-8.82372558e-02 -1.24640696e-01 -1.61825812e+00 1.06945956e+00
-9.94647220e-02 3.15486908e-01 -1.83963090e-01 -7.44402349e-01
1.49337387e+00 -9.61214900e-02 3.41425955e-01 -3.89197648e-01
1.09538153e-01 3.33855510e-01 -5.60760610e-02 -5.56171715e-01
4.68455076e-01 1.82716370e-01 -5.00583798e-02 1.30585745e-01
-4.33578491e-01 7.07067689e-03 -3.63280736e-02 -1.66344836e-01
5.04026890e-01 8.50322098e-02 2.88394868e-01 -3.01923811e-01
7.67150342e-01 -1.15946658e-01 9.08039629e-01 5.40346920e-01
-5.05302906e-01 8.61380100e-01 2.26874962e-01 -5.04217207e-01
-8.99504066e-01 -8.57207596e-01 -6.12131774e-01 1.25776470e+00
6.71802044e-01 -2.29345292e-01 -8.20822835e-01 -9.61133540e-01
1.68768972e-01 5.33558846e-01 -5.88753045e-01 -2.84061223e-01
-5.91909468e-01 -1.05857480e+00 -7.01455772e-03 7.61391103e-01
5.58884382e-01 -9.71321046e-01 -5.09482324e-01 3.00075442e-01
1.33365408e-01 -7.84214079e-01 -9.39503253e-01 -1.47750359e-02
-7.45452225e-01 -9.55648065e-01 -9.92893398e-01 -9.72049892e-01
1.00157619e+00 5.91612518e-01 5.40911376e-01 -7.09618777e-02
-2.99204171e-01 2.17507973e-01 -3.54453802e-01 -6.10812426e-01
5.91526292e-02 -1.18578160e-02 1.28737822e-01 5.41358113e-01
6.85264766e-01 7.70041645e-02 -7.89644182e-01 1.07423556e+00
-7.60119200e-01 -9.66380313e-02 7.78681695e-01 6.05240643e-01
9.30674195e-01 4.18423396e-03 3.77018899e-01 -5.56050837e-01
5.43579757e-01 -4.96999353e-01 -9.35848594e-01 4.24572945e-01
-1.08497453e+00 -2.11051151e-01 3.58697742e-01 -8.24172854e-01
-9.22584593e-01 -7.94377327e-02 5.14541209e-01 -6.93120360e-01
-1.02044322e-01 3.36167783e-01 -4.57889408e-01 -1.02105513e-01
5.53901196e-01 1.82352975e-01 -2.25127283e-02 -6.02733970e-01
1.18234925e-01 1.03909171e+00 2.96231300e-01 -5.59186101e-01
7.83218861e-01 3.97480011e-01 -6.86146021e-02 -2.83012331e-01
-7.14264572e-01 -6.82962656e-01 -5.39748371e-01 -2.04347923e-01
9.85548675e-01 -7.17698157e-01 -6.11387610e-01 6.21065795e-01
-9.67208087e-01 -4.97405976e-02 1.99480787e-01 7.10733771e-01
-3.38526309e-01 3.23045433e-01 -1.66468292e-01 -8.67170870e-01
-4.13294941e-01 -1.42694163e+00 1.10466599e+00 6.74582183e-01
2.41219625e-01 -5.75143516e-01 -2.52998352e-01 1.13847941e-01
1.91791922e-01 -1.11518957e-01 8.34206581e-01 -6.56980515e-01
-6.90992594e-01 -5.69811404e-01 -6.19008422e-01 2.41557434e-01
9.13385525e-02 5.99243082e-02 -7.17144251e-01 -4.69267249e-01
-3.28695178e-01 -1.10048339e-01 3.29385847e-01 4.25018519e-01
1.52200520e+00 -1.98861852e-01 -6.72292590e-01 6.28639460e-01
1.10434949e+00 3.92052323e-01 3.24168175e-01 6.84685826e-01
9.36752737e-01 5.97865283e-01 1.25969481e+00 5.91044247e-01
4.28463072e-01 1.07725775e+00 4.28105265e-01 5.33046462e-02
5.38829230e-02 -1.21070750e-01 1.94764122e-01 5.83574176e-01
1.08214997e-01 2.86083817e-02 -8.42503667e-01 4.10071254e-01
-1.87159848e+00 -3.95827979e-01 -2.36818492e-01 2.14178443e+00
7.11153746e-01 4.15894061e-01 4.32057232e-01 -1.09086402e-01
1.16673338e+00 4.73269150e-02 -7.43004858e-01 4.74324860e-02
7.43214637e-02 -7.17863202e-01 3.72358441e-01 4.68291854e-03
-1.47958100e+00 5.14251232e-01 4.93845654e+00 1.30737376e+00
-1.08935905e+00 8.31549913e-02 6.03065193e-01 3.23795289e-01
-5.73531762e-02 -2.10358977e-01 -1.53839266e+00 8.01076293e-01
1.01288661e-01 -3.84489670e-02 1.75285116e-01 1.46512532e+00
1.23768829e-01 1.59237102e-01 -5.60349703e-01 9.96722281e-01
9.30887461e-02 -8.11833382e-01 -1.10905595e-01 -1.19718343e-01
5.82509339e-01 -2.12722272e-01 1.96794391e-01 3.74864966e-01
3.79941203e-02 -4.79398906e-01 9.02171195e-01 4.16775674e-01
6.34434044e-01 -5.86741507e-01 7.32694387e-01 2.13164926e-01
-1.53988218e+00 -5.06320059e-01 -7.66996264e-01 6.09785855e-01
-5.49455471e-02 4.81502324e-01 -1.01576316e+00 2.90730596e-01
1.16877389e+00 4.50445175e-01 -9.63105381e-01 1.59168077e+00
-1.41831050e-02 4.76226598e-01 -4.08238918e-01 -4.74220097e-01
3.76245201e-01 -2.32478738e-01 6.52387917e-01 1.21376717e+00
3.31546038e-01 -6.61861748e-02 3.62888128e-01 9.52584803e-01
4.05962706e-01 4.62557763e-01 2.32097767e-02 2.83446968e-01
7.42596030e-01 1.57891917e+00 -8.75579417e-01 -2.00320393e-01
-3.29505682e-01 3.22222233e-01 3.28830630e-01 3.14057320e-01
-8.96372497e-01 -7.60597289e-01 3.83667558e-01 1.92572609e-01
4.72595990e-01 -1.90811485e-01 -2.49414548e-01 -6.42002940e-01
1.46212503e-01 -5.49720824e-01 6.90540075e-01 -7.47162223e-01
-1.44916952e+00 4.08595830e-01 1.69429839e-01 -1.78315568e+00
4.90663618e-01 -7.64086008e-01 -7.03766465e-01 8.12386036e-01
-1.34351325e+00 -1.03461003e+00 -7.96354949e-01 1.40898660e-01
4.73498613e-01 -2.23466352e-01 2.88130909e-01 2.39998117e-01
-9.70483959e-01 6.62655771e-01 1.97572783e-01 7.99940601e-02
7.11497724e-01 -1.20548069e+00 5.51942224e-03 5.75870395e-01
-2.90201753e-01 5.45612395e-01 4.08681244e-01 -6.40542984e-01
-1.15408921e+00 -1.41436303e+00 4.30100322e-01 -2.49146849e-01
5.12358189e-01 -4.10608500e-01 -1.26725829e+00 4.38144594e-01
-4.55529988e-01 4.40392524e-01 3.56646746e-01 -7.50517622e-02
-2.66507655e-01 -5.95737517e-01 -9.69130039e-01 6.12964928e-01
7.30498791e-01 2.83160675e-02 -2.73365945e-01 5.06236315e-01
8.10232282e-01 -4.05656189e-01 -8.24071527e-01 7.42146432e-01
3.12783659e-01 -6.80170119e-01 1.10943663e+00 -4.63180453e-01
-1.93691388e-01 -9.81995940e-01 -8.20300281e-02 -9.33098793e-01
-7.59974897e-01 -3.69625837e-01 -2.95441747e-01 1.26265502e+00
4.90237810e-02 -6.37548923e-01 6.86314046e-01 5.92920780e-01
-4.22577858e-01 -1.20037889e+00 -7.32295692e-01 -8.92943799e-01
-5.23855269e-01 -3.74606967e-01 7.45392263e-01 7.67017126e-01
-1.54060110e-01 1.13741927e-01 1.67312577e-01 3.98111403e-01
7.11754918e-01 1.88123375e-01 7.62047529e-01 -1.18401670e+00
4.50378135e-02 -6.80301309e-01 -6.93963349e-01 -1.10448122e+00
-2.97349572e-01 -8.56097579e-01 3.05406123e-01 -1.22440851e+00
1.85589701e-01 -1.16868401e+00 -6.39515460e-01 5.45748830e-01
-6.99595749e-01 3.10100704e-01 1.61049925e-02 4.87104535e-01
-9.32568312e-01 7.79489338e-01 1.21683788e+00 -3.15767795e-01
-2.97349751e-01 3.17113340e-01 -4.39652115e-01 7.51774490e-01
6.28569961e-01 -4.90962178e-01 -2.53622234e-01 -1.18374981e-01
-5.42532466e-02 -5.86842597e-01 3.32457095e-01 -9.60830331e-01
2.70489156e-01 -2.68104434e-01 3.57219398e-01 -9.80016530e-01
3.82596254e-02 -9.73934174e-01 -1.21980794e-01 3.73125166e-01
-1.21492431e-01 -1.09596150e-02 7.74790579e-03 8.05323243e-01
3.39416154e-02 -2.95500427e-01 8.96793127e-01 3.36637139e-01
-8.67319345e-01 5.15491545e-01 -2.90436875e-02 -7.02464953e-02
1.40207183e+00 -4.36885506e-01 -1.75311029e-01 1.64567128e-01
-4.81233239e-01 5.20407379e-01 2.75186121e-01 5.23361564e-01
7.34533489e-01 -1.78833473e+00 -5.96808970e-01 2.28342220e-01
6.48781240e-01 4.36868638e-01 3.09489667e-01 1.04400790e+00
-5.42597950e-01 5.07349551e-01 1.25715688e-01 -9.31543589e-01
-1.04305744e+00 7.99541593e-01 4.10336733e-01 -1.46439178e-02
-6.26541793e-01 8.96469772e-01 5.23972273e-01 -5.32378495e-01
5.96206903e-01 -9.46104974e-02 -5.50413251e-01 7.90764168e-02
5.90793490e-01 5.27098298e-01 -4.61231694e-02 -5.41142702e-01
-4.29523081e-01 7.02068985e-01 -2.65313238e-01 4.67088312e-01
9.47078943e-01 -2.05494374e-01 9.07658339e-02 2.65378982e-01
1.03185928e+00 -2.63328850e-01 -1.58120012e+00 -4.31714237e-01
1.65325955e-01 -6.06232822e-01 1.34723276e-01 -4.28308725e-01
-9.75491464e-01 8.54686975e-01 1.07684243e+00 3.13371241e-01
9.17036414e-01 -4.82513346e-02 5.15758991e-01 2.97366172e-01
2.05052912e-01 -1.28632390e+00 3.20914090e-01 2.41814610e-02
8.93127561e-01 -1.13171971e+00 1.00045130e-01 -5.86547911e-01
-6.40892744e-01 1.19078636e+00 1.18135345e+00 1.77062452e-01
7.71093845e-01 -9.47031155e-02 1.05672464e-01 -1.39833599e-01
-1.11147054e-01 6.30004406e-02 7.71745205e-01 6.14013314e-01
7.66360834e-02 1.07928574e-01 -4.40757453e-01 7.42409289e-01
4.28348064e-01 -3.08059871e-01 -1.33953601e-01 8.53304803e-01
-7.12362289e-01 -5.67692935e-01 -8.35404634e-01 5.19545615e-01
2.14527398e-02 1.75266996e-01 1.22408956e-01 6.43177092e-01
1.84671164e-01 6.50153220e-01 3.14010620e-01 -5.18169641e-01
3.70422900e-01 -1.73807681e-01 7.70595798e-04 -5.10038316e-01
-1.07198566e-01 3.20831597e-01 -4.60991710e-01 -3.66649538e-01
-7.06706271e-02 -6.16584599e-01 -1.44942260e+00 9.69092920e-02
-8.46396208e-01 3.88995260e-02 6.24574721e-01 4.88045365e-01
5.60762227e-01 3.95766735e-01 9.39789474e-01 -9.66530263e-01
-6.99436486e-01 -1.04649961e+00 -7.18638957e-01 3.18680882e-01
-1.75456569e-01 -1.08836722e+00 -2.91642904e-01 -2.35617980e-01] | [8.758581161499023, -0.6998236179351807] |
efd32cdd-b91a-4481-8c85-ee5c209ad729 | real-time-scene-text-detection-based-on | 2203.05251 | null | https://arxiv.org/abs/2203.05251v1 | https://arxiv.org/pdf/2203.05251v1.pdf | Real-time Scene Text Detection Based on Global Level and Word Level Features | It is an extremely challenging task to detect arbitrary shape text in natural scenes on high accuracy and efficiency. In this paper, we propose a scene text detection framework, namely GWNet, which mainly includes two modules: Global module and RCNN module. Specifically, Global module improves the adaptive performance of the DB (Differentiable Binarization) module by adding k submodule and shift submodule. Two submodules enhance the adaptability of amplifying factor k, accelerate the convergence of models and help to produce more accurate detection results. RCNN module fuses global-level and word-level features. The word-level label is generated by obtaining the minimum axis-aligned rectangle boxes of the shrunk polygon. In the inference period, GWNet only uses global-level features to output simple polygon detections. Experiments on four benchmark datasets, including the MSRA-TD500, Total-Text, ICDAR2015 and CTW-1500, demonstrate that our GWNet outperforms the state-of-the-art detectors. Specifically, with a backbone of ResNet-50, we achieve an F-measure of 88.6% on MSRA- TD500, 87.9% on Total-Text, 89.2% on ICDAR2015 and 87.5% on CTW-1500. | ['Xue Xu', 'Wenming Song', 'Enjun Xing', 'Jionghua Yu', 'Fuqiang Zhao'] | 2022-03-10 | null | null | null | null | ['scene-text-detection'] | ['computer-vision'] | [-4.81044054e-02 -4.21884239e-01 9.41021964e-02 5.86660299e-03
-4.93225783e-01 -2.72620231e-01 6.30664885e-01 -5.10165747e-03
-5.15167952e-01 1.15472339e-01 1.50107250e-01 -2.03908235e-01
3.01049531e-01 -1.21005225e+00 -4.65213805e-01 -6.53062582e-01
1.85318008e-01 1.79111272e-01 7.40269959e-01 -1.51478440e-01
4.02330399e-01 3.11079770e-01 -1.36939883e+00 4.70128536e-01
8.64398777e-01 1.17502677e+00 4.61886317e-01 7.45816112e-01
-3.49565506e-01 4.45898533e-01 -6.73838019e-01 -3.27403843e-01
2.96515375e-01 1.90030411e-01 -2.21249878e-01 1.50109261e-01
3.14908773e-01 -4.51157302e-01 -6.23471558e-01 1.13596773e+00
7.55988777e-01 -3.46889719e-02 7.21738458e-01 -9.28146541e-01
-5.60434222e-01 6.54382229e-01 -1.03260398e+00 3.73687863e-01
1.30121976e-01 2.67175049e-01 1.02730715e+00 -1.30865407e+00
4.06771332e-01 1.56710041e+00 8.13850582e-01 4.55279872e-02
-6.00962639e-01 -7.27877200e-01 3.07225764e-01 1.26604795e-01
-1.67828298e+00 1.91215295e-02 4.33536977e-01 -2.85680592e-01
8.87133539e-01 3.00444275e-01 4.46345657e-01 8.24613392e-01
2.50630856e-01 1.27040625e+00 5.53863287e-01 -2.23778337e-01
-8.02678242e-02 -3.02816957e-01 1.64077759e-01 9.19531107e-01
5.72839260e-01 -4.38797504e-01 -5.23002505e-01 3.75642851e-02
8.10731709e-01 8.53046775e-02 -1.83156341e-01 5.66788279e-02
-1.06190193e+00 8.40408027e-01 4.49002296e-01 2.31527284e-01
-9.53825563e-02 1.04309581e-02 6.11396313e-01 -3.02937150e-01
3.79815936e-01 -1.59478541e-02 -2.26804927e-01 1.10660613e-01
-9.03586209e-01 1.93558306e-01 4.13586527e-01 1.20510066e+00
4.85148996e-01 1.81956187e-01 -6.77222133e-01 1.05245340e+00
2.84503996e-01 1.00314593e+00 6.47023916e-01 3.97979990e-02
9.37725246e-01 1.02862167e+00 -1.58773139e-01 -1.32801890e+00
-6.38641000e-01 -8.41279149e-01 -1.13620532e+00 -4.09766495e-01
-1.40652591e-02 -2.86299624e-02 -1.34121919e+00 9.20305908e-01
2.46781915e-01 -2.93784589e-03 -2.45747462e-01 7.54445374e-01
1.29923761e+00 1.02305269e+00 -6.08175108e-03 1.60986468e-01
1.76178336e+00 -1.25543404e+00 -4.85600203e-01 -4.61610436e-01
6.28886759e-01 -1.08503759e+00 1.07497561e+00 5.97448170e-01
-8.21996093e-01 -7.61087894e-01 -1.17072129e+00 -1.39107272e-01
-4.75345343e-01 1.01381004e+00 2.06433654e-01 3.96271735e-01
-6.00179017e-01 6.51793107e-02 -4.64608639e-01 -3.89098555e-01
5.02412140e-01 1.03311963e-01 6.87652305e-02 -4.83003296e-02
-9.38560188e-01 5.35166144e-01 6.51478708e-01 1.82716683e-01
-6.09849811e-01 -3.85029912e-01 -7.85615265e-01 2.65699506e-01
6.16309047e-01 -3.90248358e-01 1.10032356e+00 -2.69562066e-01
-1.04443824e+00 8.11651766e-01 5.75692430e-02 -3.84511739e-01
7.79095650e-01 -1.09074444e-01 -5.14743447e-01 2.03344986e-01
2.26353660e-01 5.91899276e-01 1.00162089e+00 -8.18619192e-01
-1.14587569e+00 -4.52601671e-01 -5.09532332e-01 2.11804137e-01
-5.24040878e-01 -1.48146674e-01 -1.03298819e+00 -9.53329444e-01
3.55452895e-01 -5.86225867e-01 4.62507084e-02 5.23065887e-02
-6.91692173e-01 -5.61853051e-01 1.18421280e+00 -6.34111822e-01
1.54199004e+00 -2.16361618e+00 -4.67211217e-01 1.73462138e-01
2.98253655e-01 5.43833613e-01 -1.53462768e-01 2.79177099e-01
2.47593194e-01 1.28458336e-01 1.83925591e-02 -2.69040197e-01
1.19754812e-02 -1.57268330e-01 -4.93915707e-01 5.68575323e-01
7.32933730e-02 9.32580233e-01 -6.30135059e-01 -7.34028339e-01
5.51155925e-01 1.80689275e-01 -2.43937075e-01 -1.42696172e-01
-1.85471922e-01 -5.49121439e-01 -6.60672247e-01 8.03483248e-01
1.02658510e+00 -2.96713889e-01 -2.04973847e-01 -2.78954804e-01
-3.27411950e-01 8.61092657e-02 -1.51439083e+00 1.24611497e+00
-1.24115035e-01 7.35900283e-01 -2.71627903e-02 -7.66083479e-01
1.24224114e+00 -2.37451613e-01 1.23861633e-01 -7.96293199e-01
4.39143181e-01 -4.30953354e-02 -2.93346196e-01 -4.99847353e-01
9.56748307e-01 7.32426047e-01 -1.74526900e-01 5.13433069e-02
-3.25333804e-01 3.17843072e-02 2.24727198e-01 2.08974242e-01
1.04489136e+00 -2.23002702e-01 2.66358018e-01 -3.70778233e-01
6.38641000e-01 -8.41252878e-02 4.86703545e-01 1.01954389e+00
-5.55354543e-02 7.41968811e-01 3.68887216e-01 -5.59723258e-01
-9.49677110e-01 -9.73704875e-01 -3.19242418e-01 1.23015261e+00
3.22338939e-01 -6.86601996e-01 -7.02074766e-01 -6.24302268e-01
1.08649261e-01 6.17731631e-01 -5.92663825e-01 -1.72732454e-02
-6.83371007e-01 -1.08409214e+00 7.14716196e-01 7.57193327e-01
1.19078481e+00 -9.18824494e-01 -4.35556620e-01 8.84010419e-02
-2.36047119e-01 -1.38162100e+00 -7.72524953e-01 -1.38659596e-01
-6.76077247e-01 -9.67876971e-01 -6.39479280e-01 -9.76317227e-01
6.90549254e-01 6.72287166e-01 8.11554611e-01 2.05969840e-01
-6.83427453e-01 -1.34896502e-01 -5.19894123e-01 -4.31372494e-01
-6.92690164e-03 1.53922558e-01 -4.16463673e-01 -1.08035490e-01
2.80404717e-01 8.07424411e-02 -6.84359312e-01 7.30521560e-01
-8.66704285e-01 3.05920660e-01 7.42409289e-01 6.78072035e-01
4.53376502e-01 1.52970418e-01 1.74601927e-01 -3.32078278e-01
4.99276787e-01 -7.01133162e-02 -7.86033392e-01 1.52048990e-01
-4.17485356e-01 -2.97188193e-01 8.61703694e-01 -3.94987345e-01
-7.73098111e-01 7.45139048e-02 -2.13166773e-01 -2.35351652e-01
2.32119396e-01 6.69744983e-02 -1.99967757e-01 1.76638022e-01
6.25680268e-01 7.10651517e-01 -6.35422528e-01 -4.17636812e-01
1.23009145e-01 1.07749927e+00 7.66412139e-01 -2.98016459e-01
9.04027641e-01 6.18360996e-01 -1.88077763e-01 -1.18821335e+00
-6.57683134e-01 -8.73259664e-01 -2.36405417e-01 -1.02214441e-01
8.73619080e-01 -1.06069434e+00 -7.93549478e-01 9.01349843e-01
-1.18424976e+00 -1.13003135e-01 1.87094793e-01 7.77204409e-02
8.03483576e-02 4.94611740e-01 -6.04401529e-01 -7.13764608e-01
-8.89382899e-01 -9.63397264e-01 1.64963984e+00 3.24630052e-01
2.75766373e-01 -5.61716139e-01 -4.86999631e-01 7.76441023e-02
1.11516871e-01 -1.31073922e-01 6.64304912e-01 -5.86014330e-01
-5.17447233e-01 -2.60680735e-01 -8.20133328e-01 1.82536885e-01
-1.40889123e-01 3.24967980e-01 -9.37776327e-01 -1.55638084e-01
-4.28433746e-01 8.15285072e-02 1.27440810e+00 3.64832520e-01
1.59608436e+00 -1.20722048e-01 -5.29926360e-01 6.00912511e-01
1.30101657e+00 1.30309209e-01 8.49446476e-01 4.12957519e-01
7.20261931e-01 4.13905345e-02 8.13991487e-01 6.74632549e-01
4.95985508e-01 5.71191788e-01 5.15843809e-01 -1.33817703e-01
-1.73148781e-01 -2.92370468e-01 2.09681109e-01 6.71534181e-01
2.79946506e-01 -6.33601367e-01 -1.09173357e+00 4.91885364e-01
-1.80328393e+00 -8.74623299e-01 -6.04834676e-01 1.90937066e+00
4.43669975e-01 6.26008868e-01 1.01355910e-01 1.89185932e-01
1.10986662e+00 4.25411284e-01 -4.19057071e-01 3.56031470e-02
-3.62873524e-01 -1.22875936e-01 7.66894460e-01 1.38115749e-01
-1.55083537e+00 1.24983644e+00 5.24812984e+00 1.47956133e+00
-9.27475154e-01 -1.45313904e-01 7.45620072e-01 1.54222861e-01
3.06989372e-01 -5.71007013e-01 -1.39961278e+00 6.57404721e-01
1.12126783e-01 5.00111282e-02 1.18961439e-01 1.06534517e+00
3.14810425e-01 -2.57694781e-01 -5.16471863e-01 1.29083800e+00
2.90792316e-01 -1.36427200e+00 1.41042531e-01 -1.98779821e-01
6.41654551e-01 2.20393032e-01 2.56696474e-02 4.71281409e-01
1.53661132e-01 -8.68170559e-01 7.59566069e-01 1.37628764e-01
7.49479473e-01 -8.11225772e-01 7.15817213e-01 4.04128790e-01
-1.79234576e+00 -1.42901540e-01 -8.41729403e-01 3.76157284e-01
-1.75991163e-01 8.66604090e-01 -8.25724602e-01 5.04640043e-01
9.09773409e-01 6.91942036e-01 -7.84437895e-01 1.18282604e+00
-3.48183542e-01 5.49475908e-01 -4.83160406e-01 -5.53513408e-01
4.77909774e-01 1.86684111e-03 5.27668893e-01 1.75276017e+00
2.71664381e-01 -1.28484562e-01 3.68525267e-01 5.37029207e-01
-2.86196947e-01 3.85496259e-01 -1.80411085e-01 4.88012522e-01
6.11055553e-01 1.62881386e+00 -1.09053195e+00 -6.22012734e-01
-1.93365470e-01 8.42178702e-01 -8.03367794e-02 1.01644181e-01
-1.12449467e+00 -7.14460433e-01 1.54735193e-01 -6.80277050e-02
7.99736023e-01 -1.08767413e-01 -5.16137362e-01 -1.07724762e+00
1.62804231e-01 -8.58985245e-01 4.30841178e-01 -9.89139974e-01
-1.07743835e+00 4.33988392e-01 -3.92963499e-01 -1.20713198e+00
3.84420365e-01 -8.90525401e-01 -8.83793175e-01 5.29317021e-01
-1.18595016e+00 -1.21909368e+00 -8.20276260e-01 5.13187945e-01
1.03121710e+00 -1.01840511e-01 2.23791763e-01 2.90064335e-01
-1.07205975e+00 7.28786469e-01 3.87849003e-01 6.52593434e-01
6.68234110e-01 -9.51240242e-01 9.31072533e-01 8.81367087e-01
-5.25474213e-02 2.57051647e-01 5.43197095e-01 -8.63268375e-01
-1.40250587e+00 -1.60509861e+00 3.58360976e-01 9.60022863e-03
5.36769271e-01 -7.26500630e-01 -7.46863365e-01 3.77978772e-01
-2.36383155e-01 5.86019307e-02 -3.02773684e-01 -2.90447235e-01
-3.67736250e-01 -2.00053141e-01 -1.06948245e+00 7.36073971e-01
1.10120475e+00 -1.33040428e-01 -5.26364088e-01 4.62480098e-01
6.41607523e-01 -6.44624114e-01 -4.72498596e-01 2.96591908e-01
4.30670917e-01 -7.06798136e-01 1.00621021e+00 -1.43810511e-02
4.33062583e-01 -3.96494061e-01 -4.98512648e-02 -8.57046902e-01
-4.20195013e-01 -3.49506587e-01 -2.09738361e-03 1.10732353e+00
2.29349539e-01 -7.40875483e-01 8.10605049e-01 -3.16291869e-01
-3.59562904e-01 -8.17990243e-01 -8.76065195e-01 -8.06184709e-01
-1.22320816e-01 -5.57232559e-01 6.00741863e-01 6.04871273e-01
-5.22861898e-01 2.56465912e-01 -1.71942487e-01 2.73879260e-01
3.80983293e-01 4.98013534e-02 9.02092278e-01 -1.14594388e+00
-1.24993445e-02 -6.46731734e-01 -4.32699531e-01 -1.58589983e+00
-4.51766044e-01 -5.96821129e-01 1.32103652e-01 -1.63244450e+00
3.32608432e-01 -1.42226890e-01 1.31666318e-01 5.72983801e-01
-3.56184393e-01 2.70506054e-01 2.17288643e-01 5.68215102e-02
-8.34190071e-01 5.73468149e-01 1.33185387e+00 -3.53043497e-01
-1.34046853e-01 -8.81202742e-02 -3.98238689e-01 1.08887637e+00
8.60498130e-01 -4.13644493e-01 -9.56573524e-03 -3.74248177e-01
1.99334249e-01 -5.39418280e-01 2.95842022e-01 -1.11382091e+00
3.55835795e-01 6.03061952e-02 8.00933659e-01 -1.49962354e+00
5.26032224e-02 -4.03884083e-01 -5.75001061e-01 5.87096393e-01
8.12959820e-02 4.07579094e-02 4.34862435e-01 5.59144974e-01
2.07217187e-01 -3.72186184e-01 8.45463216e-01 1.59436926e-01
-7.99494207e-01 2.35273778e-01 -4.34924096e-01 1.56258270e-01
9.56820190e-01 -2.95764625e-01 -7.90131152e-01 -2.96776388e-02
-5.36758788e-02 4.82586622e-01 1.92357361e-01 6.39881611e-01
7.57571995e-01 -1.11440754e+00 -7.72242129e-01 1.69372901e-01
2.18506783e-01 3.65350962e-01 2.12637886e-01 5.87869525e-01
-7.34553277e-01 5.43635011e-01 2.75088847e-01 -8.98251534e-01
-1.36866069e+00 3.01818550e-01 2.85778344e-01 -4.94939953e-01
-8.67695689e-01 7.40804195e-01 4.28438187e-01 -2.93218017e-01
4.34013784e-01 -5.70245624e-01 -1.37116820e-01 8.69672559e-03
6.42561674e-01 6.75200522e-01 2.05160499e-01 -4.66436297e-01
-4.25516456e-01 7.92070091e-01 -2.23951444e-01 2.98185825e-01
1.03027987e+00 1.35589704e-01 2.77269017e-02 -1.70709252e-01
9.50181484e-01 1.35292634e-01 -1.00540519e+00 -1.14137352e-01
-1.81866035e-01 -3.16313386e-01 2.48802096e-01 -7.10895598e-01
-9.33589518e-01 7.80792594e-01 5.78371882e-01 3.36738050e-01
1.03919446e+00 -1.72922909e-01 9.15508926e-01 5.97166598e-01
-1.72184836e-02 -1.43869138e+00 4.64544415e-01 7.91835129e-01
9.33714807e-01 -9.10823405e-01 3.20594251e-01 -6.24187052e-01
-3.54883671e-01 1.23304129e+00 8.94163966e-01 -2.20342815e-01
2.18704849e-01 4.14293438e-01 -4.09138590e-01 -9.83663723e-02
-4.24213797e-01 -1.66286513e-01 4.37950045e-01 7.41621628e-02
-1.34397894e-02 1.80249158e-02 -2.27232814e-01 5.68566263e-01
-2.40580365e-01 -4.37976271e-01 4.93975013e-01 7.74500966e-01
-1.03855264e+00 -4.75773931e-01 -7.79755592e-01 9.06298995e-01
-3.73907179e-01 -3.91805738e-01 -2.76669145e-01 1.04208314e+00
1.41901150e-01 8.57734680e-01 3.70753497e-01 -4.62649256e-01
3.77947599e-01 -3.69079202e-01 -3.54835764e-02 -4.64288324e-01
-4.39231753e-01 3.98447901e-01 3.97829637e-02 -2.95303643e-01
2.19896451e-01 -4.40585673e-01 -1.43401361e+00 -4.55706000e-01
-7.21845448e-01 -3.20145965e-01 6.84848905e-01 7.00652182e-01
2.74000853e-01 6.84972942e-01 5.59004188e-01 -5.64690650e-01
-4.87827808e-01 -1.12256694e+00 -4.73303735e-01 -1.09080106e-01
3.31548825e-02 -4.26539540e-01 -3.14952075e-01 -2.91175209e-02] | [12.07543659210205, 2.279555082321167] |
13ef5168-5e16-48b8-b18f-7a771db877e0 | model-based-uncertainty-in-value-functions | 2302.12526 | null | https://arxiv.org/abs/2302.12526v2 | https://arxiv.org/pdf/2302.12526v2.pdf | Model-Based Uncertainty in Value Functions | We consider the problem of quantifying uncertainty over expected cumulative rewards in model-based reinforcement learning. In particular, we focus on characterizing the variance over values induced by a distribution over MDPs. Previous work upper bounds the posterior variance over values by solving a so-called uncertainty Bellman equation, but the over-approximation may result in inefficient exploration. We propose a new uncertainty Bellman equation whose solution converges to the true posterior variance over values and explicitly characterizes the gap in previous work. Moreover, our uncertainty quantification technique is easily integrated into common exploration strategies and scales naturally beyond the tabular setting by using standard deep reinforcement learning architectures. Experiments in difficult exploration tasks, both in tabular and continuous control settings, show that our sharper uncertainty estimates improve sample-efficiency. | ['Jan Peters', 'Felix Berkenkamp', 'Julia Vinogradska', 'Alessandro G. Bottero', 'Carlos E. Luis'] | 2023-02-24 | null | null | null | null | ['continuous-control'] | ['playing-games'] | [-2.54344791e-01 4.08598840e-01 -6.69549942e-01 -1.23392515e-01
-1.31753862e+00 -7.06732631e-01 2.65732467e-01 3.31475258e-01
-6.78068995e-01 1.54515100e+00 -2.46039834e-02 -4.60934460e-01
-5.09791672e-01 -7.61973679e-01 -9.50625658e-01 -6.40026152e-01
-4.38883275e-01 7.33448863e-01 -1.65653482e-01 -1.68187413e-02
3.33570838e-01 7.50539154e-02 -9.44897294e-01 -3.95487726e-01
1.05950499e+00 1.35369098e+00 4.45552953e-02 5.25575936e-01
-1.50299489e-01 6.59218192e-01 -8.84065926e-01 -1.55966908e-01
3.59733075e-01 -2.52363920e-01 -7.59286404e-01 -6.83924481e-02
3.91928591e-02 -8.88650537e-01 -4.50247861e-02 1.35121703e+00
1.76223859e-01 3.18381727e-01 4.97738451e-01 -1.41410005e+00
-4.92649168e-01 1.23692989e+00 -7.40328074e-01 1.01435319e-01
1.11135796e-01 1.35758787e-01 1.03075600e+00 2.13986933e-02
2.14747652e-01 1.50657618e+00 4.44705546e-01 6.37880504e-01
-1.56072545e+00 -4.87225443e-01 4.38702255e-01 -1.56737357e-01
-9.16123807e-01 7.85524026e-02 2.15454176e-01 -2.71219701e-01
7.64849424e-01 -8.05614293e-02 5.78252196e-01 1.00247192e+00
5.10768950e-01 9.73867178e-01 1.35436785e+00 -1.47071049e-01
8.50284159e-01 -1.40218616e-01 -1.45871744e-01 4.87070620e-01
6.67545319e-01 7.45098412e-01 -2.26088598e-01 -2.49631837e-01
1.03652036e+00 -1.43360212e-01 -1.55388445e-01 -9.12287772e-01
-9.99458849e-01 1.13305163e+00 2.37120464e-01 -3.89280587e-01
-3.01397741e-01 8.91781688e-01 3.74020904e-01 4.34405953e-01
7.61563927e-02 7.99596071e-01 -5.15508890e-01 -6.39848351e-01
-6.02247059e-01 7.68537819e-01 9.10960853e-01 9.34485257e-01
4.79662240e-01 2.18287021e-01 -4.80382979e-01 3.33683550e-01
3.39698255e-01 4.81706113e-01 1.55312285e-01 -1.69253218e+00
7.41092682e-01 -5.80827184e-02 1.01684976e+00 -2.61310607e-01
-3.53879303e-01 -4.52042431e-01 -2.65222996e-01 6.27263069e-01
8.43850255e-01 -8.28450501e-01 -8.39174330e-01 2.20923591e+00
9.82651934e-02 -3.18397582e-01 1.26163229e-01 7.67983258e-01
-2.58278787e-01 4.19949740e-01 -6.37578368e-02 -4.54473853e-01
1.02891147e+00 -8.33713233e-01 -8.73616457e-01 -2.82073885e-01
2.84783661e-01 -7.91181158e-03 1.03693211e+00 5.71519554e-01
-1.33266652e+00 -3.58345769e-02 -1.16600847e+00 4.63070929e-01
-2.94100158e-02 -5.52924037e-01 6.47662520e-01 8.05717051e-01
-9.58999753e-01 1.11990011e+00 -1.00060642e+00 2.49448895e-01
5.57573199e-01 3.77241254e-01 2.91337967e-01 3.73258948e-01
-1.15266550e+00 1.01446271e+00 7.69927204e-01 -1.47619098e-01
-1.42905974e+00 -7.70167232e-01 -8.11920345e-01 3.38730931e-01
1.18300426e+00 -4.60372925e-01 1.95104539e+00 -3.48221213e-01
-1.97737050e+00 -1.01227090e-01 4.61324960e-01 -1.00858986e+00
7.76663601e-01 -3.64338487e-01 2.46720687e-01 -1.50217786e-01
-6.86319470e-02 6.51287019e-01 7.48204350e-01 -1.21122801e+00
-5.73901713e-01 -1.88175261e-01 5.57169139e-01 2.43851602e-01
6.78705201e-02 -6.02496505e-01 1.27377674e-01 -2.99885780e-01
-3.41255844e-01 -8.38874161e-01 -7.31487393e-01 -1.65149719e-01
-3.38294625e-01 -1.52853057e-01 1.52556285e-01 -1.29888862e-01
1.15894532e+00 -1.53639591e+00 8.28595236e-02 2.98054814e-01
1.24613740e-01 -3.41102809e-01 -1.36505533e-02 5.37973464e-01
2.75657564e-01 2.19831884e-01 -4.52757150e-01 -7.16224387e-02
7.60169327e-01 4.45220053e-01 -6.51560783e-01 3.04147393e-01
4.11831960e-02 9.38641191e-01 -1.15284574e+00 -2.97337711e-01
4.66283634e-02 -2.08157673e-01 -5.95053077e-01 1.04463242e-01
-8.44768465e-01 1.74499452e-01 -6.18537784e-01 4.13627982e-01
5.80053747e-01 -2.08104327e-01 3.99413943e-01 4.16403860e-01
1.50008639e-02 1.53152049e-01 -1.32979846e+00 1.65125644e+00
-5.22750616e-01 1.93015441e-01 1.43337980e-01 -9.17273402e-01
6.42660081e-01 -1.51177436e-01 3.96930605e-01 -4.40797508e-01
3.00261199e-01 1.47574365e-01 8.74001812e-03 3.82525511e-02
7.50031054e-01 -3.08744490e-01 -4.83801037e-01 6.64238095e-01
-8.16075951e-02 -4.58802879e-01 4.35622126e-01 2.42752749e-02
7.21456170e-01 4.04678434e-01 3.31306756e-01 -5.75512469e-01
-1.48342744e-01 -1.04646370e-01 4.46024477e-01 1.30535436e+00
-4.62097764e-01 -7.04064146e-02 1.36991537e+00 -2.18646377e-02
-8.39547932e-01 -1.53011930e+00 -4.59872276e-01 8.51906657e-01
1.68196708e-01 -2.36565337e-01 -7.72100210e-01 -9.09318686e-01
6.12882376e-01 1.00628090e+00 -8.90218616e-01 -1.75058827e-01
-6.92467541e-02 -5.95300019e-01 2.20589012e-01 8.46026540e-01
3.81693274e-01 -7.03837276e-01 -1.02028549e+00 3.82426500e-01
2.82551944e-01 -6.25667810e-01 -5.23194253e-01 5.77120721e-01
-9.68637705e-01 -9.02259767e-01 -7.39488900e-01 1.23967737e-01
2.30092138e-01 -5.27359784e-01 1.08823574e+00 -7.42845356e-01
1.39336944e-01 6.97726369e-01 2.17932105e-01 -4.83651757e-01
-2.37447649e-01 -2.01174498e-01 1.40102357e-01 -6.84967279e-01
1.20238759e-01 -2.47456402e-01 -5.35721004e-01 2.80637797e-02
-7.02856302e-01 -5.25930047e-01 3.63451004e-01 1.12920749e+00
6.75887287e-01 1.15517534e-01 8.48803937e-01 -5.42254806e-01
1.20953453e+00 -5.71968257e-01 -1.54746222e+00 1.51351929e-01
-9.31280911e-01 9.90936279e-01 3.92408997e-01 -4.23082680e-01
-1.10207796e+00 -2.81816870e-01 3.63280833e-01 -4.29301739e-01
2.95346379e-01 6.12641990e-01 2.41836801e-01 1.76798031e-01
3.92725319e-01 -1.26843974e-01 2.25207359e-01 -2.94259578e-01
4.54960495e-01 1.11477450e-01 5.18237352e-01 -1.33370268e+00
3.46163541e-01 9.44710597e-02 2.77422786e-01 -8.56691897e-02
-1.03568697e+00 3.83250356e-01 7.34214410e-02 5.07517681e-02
5.06196678e-01 -6.47539258e-01 -1.56136227e+00 -1.53506488e-01
-7.15940654e-01 -7.31271207e-01 -8.61358643e-01 6.10338807e-01
-1.29464185e+00 3.57079983e-01 -5.98045886e-01 -1.39149749e+00
-4.62833093e-03 -1.30595791e+00 9.01507735e-01 4.73924726e-01
-1.17959466e-03 -9.98357415e-01 4.91251230e-01 -2.54760712e-01
4.51840162e-01 3.58565748e-01 9.03213024e-01 -2.89321870e-01
-6.86889589e-01 2.30882630e-01 -1.49386838e-01 1.88050568e-01
-2.40097046e-01 -2.67949969e-01 -5.16566753e-01 -3.72570723e-01
-1.44086853e-01 -8.14843774e-01 1.05799973e+00 7.92075336e-01
1.43208385e+00 -5.10658503e-01 -5.51381819e-02 3.47526848e-01
1.40267074e+00 4.28249598e-01 2.61963427e-01 5.21809816e-01
3.05471886e-02 4.18424219e-01 1.03055847e+00 1.10297167e+00
2.30581790e-01 3.82169306e-01 8.06652367e-01 7.83702552e-01
7.39645958e-01 -4.72788304e-01 4.57642287e-01 -1.84268355e-01
2.48833805e-01 -1.59521595e-01 -6.41127944e-01 4.63410676e-01
-2.09249544e+00 -9.27303910e-01 6.30890310e-01 2.53446651e+00
1.24837160e+00 4.24268335e-01 5.14083266e-01 -3.41228724e-01
4.20325786e-01 3.11401989e-02 -1.32482195e+00 -7.57473409e-01
3.78506035e-01 1.95202142e-01 9.28467810e-01 8.41002822e-01
-9.02750671e-01 6.55077040e-01 7.60801649e+00 9.24965858e-01
-5.46408892e-01 -7.39467591e-02 6.90999806e-01 -6.40873849e-01
-4.37618077e-01 -1.92660093e-01 -8.52463663e-01 4.90822881e-01
9.94269967e-01 -7.44384885e-01 7.02445805e-01 1.24809670e+00
-9.58701223e-02 -4.73797083e-01 -1.34223390e+00 7.89303064e-01
-7.23953247e-01 -1.23745453e+00 -5.49058557e-01 3.74812543e-01
9.90910172e-01 -2.72017680e-02 4.57314283e-01 7.00228870e-01
1.30033123e+00 -1.19722521e+00 8.08803976e-01 3.58692229e-01
7.07029223e-01 -1.34122229e+00 5.21819949e-01 1.39648333e-01
-7.54480362e-01 -5.05392313e-01 -5.33885479e-01 -1.24483809e-01
2.02944443e-01 3.89186621e-01 -7.03495979e-01 8.30485076e-02
4.74836856e-01 2.12872282e-01 1.04174063e-01 1.03137970e+00
-1.33838892e-01 2.55824268e-01 -5.88459611e-01 -4.41636831e-01
7.15448380e-01 -3.45945060e-01 3.87522787e-01 7.99091160e-01
2.94289231e-01 -2.53643930e-01 2.33693838e-01 1.42804098e+00
-9.63144824e-02 -3.98172826e-01 -4.04291630e-01 -4.81111526e-01
5.96508443e-01 8.23968947e-01 -5.14992833e-01 -2.30851665e-01
2.19084799e-01 3.96168590e-01 5.12063622e-01 3.79679382e-01
-1.06205475e+00 -4.00632471e-01 8.95383477e-01 -5.99735320e-01
4.68000919e-01 -2.82541633e-01 -3.57765406e-01 -9.01300967e-01
2.69986391e-02 -6.24330401e-01 4.50996906e-01 -2.24694967e-01
-1.21749425e+00 1.74281418e-01 6.03188515e-01 -8.38112056e-01
-8.66478205e-01 -8.07791889e-01 -4.02434856e-01 8.51780891e-01
-1.32485688e+00 -5.45132346e-02 3.51525158e-01 2.73381889e-01
3.30856264e-01 1.65207177e-01 5.63722610e-01 -3.92715961e-01
-6.58148646e-01 5.75469732e-01 6.61697209e-01 -5.51729560e-01
2.84091383e-01 -1.95543516e+00 1.43291518e-01 4.36978102e-01
-2.66681641e-01 3.79269153e-01 9.95190561e-01 -7.47017086e-01
-1.46307552e+00 -6.03754163e-01 -2.94779897e-01 -4.00665790e-01
1.16562545e+00 -9.58753154e-02 -5.90177953e-01 6.96718991e-01
3.40931654e-01 -2.55330354e-01 1.59962893e-01 3.32459271e-01
-2.37216696e-01 1.13039054e-01 -1.33706605e+00 6.55469179e-01
7.42043316e-01 -1.63010582e-01 -5.45676112e-01 1.42429426e-01
9.84428644e-01 -7.29876637e-01 -1.00828612e+00 2.20819309e-01
6.35444522e-01 -7.07141578e-01 7.91251481e-01 -9.50984597e-01
5.42881489e-01 4.14656214e-02 -2.49003336e-01 -1.71674883e+00
1.20249018e-01 -1.08094776e+00 -7.31641412e-01 6.71701789e-01
3.10219646e-01 -6.93887830e-01 9.11551297e-01 9.22204912e-01
8.58216733e-02 -1.06968415e+00 -1.06124794e+00 -1.32544458e+00
6.41149580e-01 -3.67294699e-01 7.84271717e-01 4.02646780e-01
5.47809184e-01 -2.82738268e-01 -3.69564503e-01 -4.39904854e-02
1.21523404e+00 2.35944316e-01 2.21186772e-01 -8.64524305e-01
-7.60353565e-01 -8.02873313e-01 2.75654525e-01 -1.22123146e+00
1.89692929e-01 -4.20810521e-01 2.99725682e-01 -1.27795160e+00
1.68896586e-01 -4.74250555e-01 -4.71443683e-01 3.19438756e-01
3.83095481e-02 -5.60844243e-01 3.65880728e-01 -4.02411968e-01
-8.55022371e-01 1.12548566e+00 1.40438557e+00 -1.09169006e-01
-1.95209876e-01 6.75249696e-02 -8.56508553e-01 6.34844363e-01
1.01211214e+00 -4.57130194e-01 -7.10244238e-01 -1.45948604e-01
4.99191135e-01 7.47949660e-01 1.56086355e-01 -6.92105472e-01
-8.41689482e-02 -7.45654523e-01 2.60292858e-01 -6.78387642e-01
2.74426490e-01 -6.55770540e-01 -2.17407525e-01 6.48772180e-01
-7.44293630e-01 4.80013341e-02 5.32868922e-01 9.74356174e-01
1.79647565e-01 -5.64207196e-01 7.57054925e-01 -7.28243887e-02
-3.75639260e-01 2.99360484e-01 -4.20992345e-01 4.63928908e-01
9.91814792e-01 2.20772743e-01 -4.87293810e-01 -7.21238554e-01
-8.15638602e-01 8.23510289e-01 2.52824128e-01 5.74656092e-02
4.82274532e-01 -1.28941762e+00 -2.29960367e-01 -2.19452724e-01
-1.25923291e-01 -1.03983171e-01 1.25577495e-01 5.32164514e-01
-2.43653785e-02 5.39758027e-01 -3.31871301e-01 -3.49190444e-01
-3.46281558e-01 6.84816718e-01 6.12175643e-01 -7.25771546e-01
-9.86248255e-02 7.03702867e-01 6.88477457e-02 -2.59691030e-01
6.36296511e-01 -7.33694375e-01 2.08902478e-01 -8.13255012e-02
4.89668459e-01 5.95930994e-01 -5.68465531e-01 4.67793971e-01
1.73217489e-03 6.91620260e-02 -1.69259310e-01 -6.76778913e-01
1.03270626e+00 -2.03522146e-01 3.59989971e-01 5.86278141e-01
7.47353315e-01 -4.03744429e-01 -2.01382208e+00 -1.31737022e-02
1.92274317e-01 -5.64731300e-01 1.84504583e-01 -1.14648890e+00
-7.65230060e-01 7.57180393e-01 5.76739669e-01 3.60404849e-01
6.44380152e-01 -2.07957357e-01 4.22537267e-01 9.37448740e-01
6.56079650e-01 -1.35244071e+00 -7.32331276e-02 5.36660373e-01
8.88417482e-01 -1.36326528e+00 2.57264767e-02 4.12907243e-01
-8.76450300e-01 1.11612535e+00 8.17425370e-01 -2.13472798e-01
4.39913630e-01 6.20410144e-01 -5.07057667e-01 2.21002996e-01
-9.72559392e-01 -2.42893413e-01 -1.18924387e-01 5.34482062e-01
-5.22020049e-02 3.00125062e-01 -2.27138147e-01 7.81760991e-01
-2.65106171e-01 1.14791624e-01 7.32725978e-01 9.64501858e-01
-6.96591914e-01 -1.16229975e+00 -3.88664633e-01 5.12244403e-01
-5.00051737e-01 1.32665247e-01 9.00823027e-02 8.49804640e-01
-6.33169651e-01 8.45010579e-01 2.80386239e-01 -2.91053820e-02
-1.16078695e-02 -2.22463816e-01 9.38685775e-01 -2.37391397e-01
-1.16706312e-01 -6.50233328e-02 6.59707338e-02 -9.52264309e-01
1.09717585e-01 -5.62979341e-01 -1.28615451e+00 -4.42080230e-01
-8.81626084e-02 4.52712417e-01 5.83156228e-01 8.53254914e-01
1.22726239e-01 4.68211204e-01 4.07624036e-01 -6.91156685e-01
-1.76637816e+00 -7.97716200e-01 -8.40008318e-01 -3.14775765e-01
6.49745405e-01 -1.06195593e+00 -3.93966824e-01 -8.26378942e-01] | [4.1533942222595215, 2.487584114074707] |
016523db-3ecd-4a7c-bc10-d7daeb06a81f | explainable-ai-algorithms-for-vibration-data | 2207.10732 | null | https://arxiv.org/abs/2207.10732v1 | https://arxiv.org/pdf/2207.10732v1.pdf | Explainable AI Algorithms for Vibration Data-based Fault Detection: Use Case-adadpted Methods and Critical Evaluation | Analyzing vibration data using deep neural network algorithms is an effective way to detect damages in rotating machinery at an early stage. However, the black-box approach of these methods often does not provide a satisfactory solution because the cause of classifications is not comprehensible to humans. Therefore, this work investigates the application of explainable AI (XAI) algorithms to convolutional neural networks for vibration-based condition monitoring. For this, various XAI algorithms are applied to classifications based on the Fourier transform as well as the order analysis of the vibration signal. The results are visualized as a function of the revolutions per minute (RPM), in the shape of frequency-RPM maps and order-RPM maps. This allows to assess the saliency given to features which depend on the rotation speed and those with constant frequency. To compare the explanatory power of the XAI methods, investigations are first carried out with a synthetic data set with known class-specific characteristics. Then a real-world data set for vibration-based imbalance classification on an electric motor, which runs at a broad range of rotation speeds, is used. A special focus is put on the consistency for variable periodicity of the data, which translates to a varying rotation speed of a real-world machine. This work aims to show the different strengths and weaknesses of the methods for this use case: GradCAM, LRP and LIME with a new perturbation strategy. | ['Deniz Neufeld', 'Oliver Mey'] | 2022-07-21 | null | null | null | null | ['fault-detection'] | ['miscellaneous'] | [ 1.48617933e-02 -2.14244485e-01 8.13053697e-02 6.85792491e-02
8.32724944e-02 -2.38102362e-01 5.38958490e-01 1.12927519e-01
6.77293306e-03 4.54193294e-01 -1.54495239e-01 -2.86659390e-01
-7.85295129e-01 -7.43620396e-01 -5.97480834e-01 -8.92495215e-01
-2.75674105e-01 2.43238688e-01 -2.41798833e-01 -7.38976061e-01
3.03864926e-01 9.77933407e-01 -2.14536357e+00 1.85621336e-01
4.43582416e-01 7.97907472e-01 6.39940277e-02 4.65228915e-01
1.58867642e-01 3.14534664e-01 -1.20144236e+00 3.70243073e-01
5.10333255e-02 -4.93014425e-01 -6.77134097e-01 -8.67612567e-03
-1.02105856e-01 5.54785468e-02 -8.43887776e-02 7.44008839e-01
5.08506238e-01 1.58690691e-01 8.70314240e-01 -1.29989922e+00
-1.84637785e-01 4.16288853e-01 -3.60461384e-01 4.44126815e-01
4.00592923e-01 -1.67936683e-02 5.74478149e-01 -4.87002790e-01
5.44682443e-01 1.07060850e+00 5.14293253e-01 2.01693937e-01
-1.05482113e+00 -1.92530423e-01 -2.32636094e-01 8.39186311e-01
-9.79491949e-01 2.92610824e-01 1.39770722e+00 -4.98658746e-01
1.00314152e+00 2.96460479e-01 7.07660258e-01 1.02713454e+00
5.88167727e-01 2.31355339e-01 9.64734733e-01 -6.21245623e-01
2.25279942e-01 -9.01945531e-02 2.70013332e-01 -8.36066529e-02
2.77798653e-01 2.41259009e-01 -3.97853225e-01 5.15364707e-01
3.88961047e-01 -1.84183046e-01 -3.97020131e-01 -1.50637820e-01
-9.63888645e-01 6.79580867e-01 5.67814171e-01 9.10024345e-01
-5.99188805e-01 1.00749917e-02 6.21492445e-01 5.99759459e-01
3.93244505e-01 1.03385270e+00 -6.67189121e-01 -1.79204941e-01
-6.50435328e-01 3.19689244e-01 4.70672816e-01 -1.04335070e-01
3.32962453e-01 6.47198319e-01 9.55570266e-02 3.69414836e-01
-8.61169677e-03 4.68500346e-01 7.86361873e-01 -4.07008350e-01
2.57875994e-02 5.34833968e-01 -2.15510055e-02 -1.45748484e+00
-1.02394056e+00 -6.39784455e-01 -8.37192714e-01 7.94494212e-01
5.06081045e-01 -1.18659414e-01 -5.38954079e-01 1.41960073e+00
2.74605393e-01 -2.56065607e-01 -2.88084447e-02 1.12833059e+00
5.57255328e-01 5.93424201e-01 -4.12314564e-01 -1.40615299e-01
1.59477186e+00 -3.64673316e-01 -1.04594696e+00 1.63636476e-01
6.06940269e-01 -5.17136693e-01 1.23030913e+00 8.21481884e-01
-5.64012229e-01 -9.18613553e-01 -1.34772336e+00 4.27671224e-01
-7.96705842e-01 3.67614925e-01 1.53676927e-01 5.88543177e-01
-6.20526075e-01 9.51605022e-01 -7.67311931e-01 -8.18487182e-02
-1.46910409e-02 3.61404009e-02 -3.84251952e-01 2.53351420e-01
-1.40052903e+00 1.32262886e+00 3.82347047e-01 4.61212486e-01
-3.44872236e-01 -5.71403801e-01 -6.28106058e-01 3.41028571e-01
-7.78978541e-02 -1.17230415e-01 7.11134434e-01 -9.60835934e-01
-1.45249999e+00 4.58987385e-01 1.45926252e-01 -5.75876594e-01
4.42198366e-01 -4.90050286e-01 -5.16020298e-01 2.78431416e-01
-2.65005916e-01 -1.22068048e-01 1.07180572e+00 -1.14573884e+00
-7.43525997e-02 -2.38449365e-01 -5.58948005e-03 -2.52172798e-01
-1.94748387e-01 -1.01778686e-01 5.57571828e-01 -5.19818723e-01
1.74146876e-01 -7.54239976e-01 4.36327219e-01 -6.10788047e-01
-4.62799519e-01 -3.40183675e-01 1.01489079e+00 -6.25637352e-01
9.58884776e-01 -2.16839170e+00 2.90041625e-01 3.52183044e-01
1.72114652e-02 2.66345888e-01 8.30473304e-02 5.98431706e-01
-9.48392391e-01 -1.43123940e-01 -5.88212162e-02 2.91431367e-01
-2.85543557e-02 -1.06094003e-01 -1.89195648e-01 5.39101601e-01
6.70878828e-01 3.97547364e-01 -5.33697486e-01 1.25055835e-01
6.46900535e-01 5.52124798e-01 -1.52645871e-01 -5.01893573e-02
-1.48550915e-02 5.29691279e-01 -7.63208643e-02 3.09500337e-01
4.56247419e-01 2.83639371e-01 -9.77830067e-02 -5.57737827e-01
-3.34478259e-01 6.51572123e-02 -1.07417202e+00 1.03750753e+00
-3.98085803e-01 1.06366920e+00 -2.56253868e-01 -1.64517915e+00
1.33514690e+00 4.71327305e-01 6.27108335e-01 -8.09743941e-01
3.70512903e-01 2.96742857e-01 4.65708882e-01 -7.65848458e-01
2.22086787e-01 2.42217854e-02 1.96229726e-01 2.08954930e-01
-5.78658916e-02 -1.54367387e-01 3.24101269e-01 -5.47658265e-01
8.37394297e-01 2.11148486e-01 -2.64953952e-02 -3.81098300e-01
5.80907166e-01 -5.28766885e-02 5.23871146e-02 2.38961622e-01
1.34526342e-01 6.08336866e-01 9.39566135e-01 -7.97358930e-01
-9.91899967e-01 -5.58136046e-01 -3.90166909e-01 5.89775383e-01
3.21889110e-02 1.95765927e-01 -7.11444378e-01 -7.69300386e-02
-7.91207403e-02 9.50728297e-01 -8.71362984e-01 -6.18948996e-01
-6.72963321e-01 -8.82488549e-01 2.15799138e-01 2.63603747e-01
3.10653031e-01 -1.39814806e+00 -1.45347941e+00 1.14126720e-01
-1.12848923e-01 -7.23472476e-01 6.22128308e-01 5.29867291e-01
-7.10057914e-01 -1.20648110e+00 -4.99262333e-01 -3.73485148e-01
3.73179734e-01 -5.56290112e-02 1.00556386e+00 1.11797288e-01
-5.99919379e-01 1.15882710e-01 -6.26456976e-01 -7.10011303e-01
-5.23254097e-01 1.06399193e-01 1.60808697e-01 -8.95069540e-02
3.20869803e-01 -5.88603139e-01 -3.91349882e-01 2.91467667e-01
-9.76716399e-01 -2.79675633e-01 4.97461200e-01 8.19304585e-01
1.49832591e-01 6.09052360e-01 6.82935238e-01 -2.63648570e-01
8.17859948e-01 -5.02007902e-01 -4.10747528e-01 -1.69763461e-01
-6.82763040e-01 3.55197303e-02 7.66033411e-01 -4.74965900e-01
-6.00428641e-01 -2.46436834e-01 -1.89392164e-01 -6.30401492e-01
-5.03253520e-01 7.76228368e-01 -9.13715735e-02 3.54950905e-01
1.00201738e+00 2.60798335e-02 2.43475839e-01 -6.20223761e-01
9.65464041e-02 4.73359346e-01 6.77493095e-01 -2.48256281e-01
8.17981601e-01 3.85592170e-02 3.77331555e-01 -9.56061423e-01
-2.16231927e-01 -9.50953886e-02 -6.48607850e-01 -7.22469807e-01
7.63774514e-01 -2.49393135e-01 -8.56287599e-01 6.91621304e-01
-1.23628426e+00 1.49729764e-02 -6.18147492e-01 6.37584984e-01
-4.85434055e-01 1.66163906e-01 -3.97635669e-01 -7.88555622e-01
-3.40699941e-01 -1.23605824e+00 7.41835594e-01 1.44348308e-01
-3.47909540e-01 -7.16004372e-01 -1.55664101e-01 2.16475930e-02
5.50051093e-01 9.07398224e-01 1.16615546e+00 -5.96949637e-01
1.47549555e-01 -5.39091527e-01 2.59965420e-01 4.35624540e-01
1.68666601e-01 3.66981119e-01 -9.53458309e-01 -8.02223161e-02
3.12888831e-01 -6.67827949e-02 5.12160838e-01 6.52226686e-01
1.08010471e+00 8.00513849e-02 8.20068941e-02 1.07018583e-01
1.36502314e+00 3.06894362e-01 7.68577218e-01 7.31765330e-01
4.59449440e-01 8.81808162e-01 6.34711206e-01 3.70434374e-01
-3.70214462e-01 9.03390706e-01 9.10335243e-01 -2.64635026e-01
2.87668668e-02 4.91485387e-01 1.99408233e-01 5.69934428e-01
-4.78610188e-01 -4.64483090e-02 -8.84369969e-01 4.31954861e-01
-1.37524223e+00 -1.04264879e+00 -4.73471165e-01 1.97835982e+00
2.48011202e-01 3.34203780e-01 1.60099357e-01 1.42235744e+00
7.46086180e-01 -1.35493446e-02 -3.54356527e-01 -7.66677737e-01
-2.60255963e-01 2.88899302e-01 6.86310530e-02 6.42932719e-03
-1.02929640e+00 -9.44586284e-03 5.52655268e+00 5.43271005e-01
-1.70168471e+00 -3.49988520e-01 3.22290152e-01 9.38470960e-02
1.00848950e-01 -5.65298140e-01 -4.52419072e-02 6.28666282e-01
1.09833288e+00 6.53154403e-02 4.37208056e-01 8.71209145e-01
3.66074443e-01 -7.27205649e-02 -9.58429933e-01 7.96495736e-01
-4.38143946e-02 -1.03451979e+00 -3.43124509e-01 -1.80575281e-01
2.72255689e-01 -1.27785414e-01 2.26985719e-02 -7.24215358e-02
-8.29171360e-01 -1.13742244e+00 8.00045788e-01 7.46374071e-01
5.99905312e-01 -8.82101476e-01 1.31301510e+00 9.68803093e-02
-7.56677330e-01 -2.98465818e-01 -1.74089789e-01 -3.50830257e-01
2.09890325e-02 6.35919213e-01 -6.90920889e-01 9.70201552e-01
9.13246155e-01 5.36769629e-01 -3.43387157e-01 6.70124769e-01
-1.46829501e-01 7.40477979e-01 -3.92232418e-01 -2.05116197e-01
-5.39319869e-03 -1.67989805e-01 7.24798381e-01 9.80025589e-01
4.45173413e-01 -4.28370982e-01 -5.35057545e-01 9.72645879e-01
5.90378582e-01 -8.89274031e-02 -6.94152653e-01 1.63684338e-02
3.29746574e-01 1.21290410e+00 -8.52186680e-01 7.79674426e-02
1.27538014e-03 3.74739051e-01 -3.97121906e-01 3.11337620e-01
-6.67555213e-01 -7.90167689e-01 5.32407165e-01 3.18671972e-01
2.08740786e-01 -1.07330486e-01 -3.03388804e-01 -4.01877284e-01
1.60189494e-01 -8.33661914e-01 -8.89130235e-02 -1.17216706e+00
-1.00545239e+00 6.00220680e-01 4.25833344e-01 -1.53315985e+00
-6.06521845e-01 -9.72384453e-01 -9.02535439e-01 1.06207514e+00
-1.29883564e+00 -6.11993432e-01 -4.47210521e-01 2.32464492e-01
4.52381462e-01 -2.45360404e-01 7.73534656e-01 2.01494083e-01
-6.18522167e-01 2.10395440e-01 1.02561004e-01 -5.98901836e-03
8.86279643e-02 -1.26492643e+00 1.03735231e-01 6.78094387e-01
-8.88097659e-02 2.95729250e-01 1.51460314e+00 -1.02999441e-01
-1.21694922e+00 -5.25383055e-01 5.59033096e-01 -3.49423438e-01
3.99461895e-01 5.47910444e-02 -1.17884374e+00 1.71042271e-02
2.20132098e-01 -1.37424499e-01 1.27898455e-01 -6.62811622e-02
1.61667243e-01 -2.16025621e-01 -1.06151652e+00 1.87667727e-01
2.82700956e-01 -2.09533423e-01 -8.60541463e-01 2.90472716e-01
4.50799912e-01 -3.05767715e-01 -9.84923244e-01 5.70463359e-01
4.57115442e-01 -1.03797841e+00 9.18533623e-01 -5.11207163e-01
7.60231078e-01 -4.70195979e-01 2.77744353e-01 -1.70209289e+00
-1.89446628e-01 -2.25062832e-01 -8.32298491e-03 8.61030877e-01
1.16468348e-01 -8.55407178e-01 3.13790679e-01 -2.56644338e-01
-2.04489648e-01 -1.00846386e+00 -1.09740508e+00 -7.02019274e-01
4.65676449e-02 -5.59003055e-01 7.12359011e-01 1.01704848e+00
8.89725834e-02 1.22696102e-01 -8.39541033e-02 2.42361233e-01
5.94719872e-02 1.85213313e-02 5.43566585e-01 -1.54846501e+00
-7.03486130e-02 -5.96823037e-01 -9.71301794e-01 2.06088454e-01
-5.53316474e-02 -4.08175170e-01 -1.34547472e-01 -1.33131731e+00
-6.20424211e-01 2.08061021e-02 -3.40831697e-01 3.39270771e-01
8.54341872e-03 5.04294131e-03 1.29587963e-01 8.37232694e-02
4.47139770e-01 4.58079040e-01 1.02772462e+00 -2.49598905e-01
-1.03945009e-01 1.70885652e-01 -5.68341538e-02 5.15518725e-01
1.19425392e+00 -3.43897253e-01 -4.73378539e-01 -3.68733630e-02
5.03722072e-01 5.94173484e-02 5.07831693e-01 -1.49463892e+00
-4.75862771e-01 3.12900662e-01 4.21028346e-01 -7.18008339e-01
6.05468415e-02 -8.08145583e-01 3.69201332e-01 8.81704867e-01
-1.40758052e-01 5.03810704e-01 4.50418204e-01 1.72846168e-01
-4.58608240e-01 -2.82794952e-01 5.65770447e-01 2.51979440e-01
-3.93417567e-01 -5.98630309e-01 -5.80423295e-01 -4.96457070e-01
9.07972634e-01 -4.10582960e-01 -2.13238090e-01 -3.02879900e-01
-6.07117116e-01 -2.11974829e-01 1.65400431e-01 6.26844287e-01
3.48431438e-01 -1.10054505e+00 -7.23375678e-01 2.27894574e-01
-5.20578958e-02 -1.63168520e-01 4.88056064e-01 9.94144022e-01
-7.15571404e-01 4.37747300e-01 -6.96507454e-01 -7.24662006e-01
-1.31604981e+00 7.37729669e-01 6.98731661e-01 -3.42260674e-02
-6.04383528e-01 3.07716519e-01 -4.45214123e-01 -1.70899272e-01
-7.58891925e-02 -6.98300719e-01 -9.86506999e-01 3.97057176e-01
2.11045086e-01 5.51275194e-01 5.79372942e-01 -6.85495555e-01
-3.38241726e-01 7.64514685e-01 6.12573445e-01 1.29593983e-01
1.23517144e+00 2.54280090e-01 1.48141524e-02 7.32418597e-01
1.01668215e+00 -2.26820350e-01 -8.08830380e-01 3.64269137e-01
-4.57572676e-02 -2.19905358e-02 6.67515323e-02 -8.04939926e-01
-1.19927657e+00 1.17399263e+00 1.05044925e+00 8.81077230e-01
1.39205325e+00 -3.64994138e-01 1.34853199e-01 3.01877230e-01
1.06123969e-01 -1.09222853e+00 1.72313657e-02 2.07652107e-01
1.19966769e+00 -7.35996366e-01 8.62367302e-02 -5.48568740e-02
-3.20920557e-01 1.64171052e+00 2.56984830e-01 -1.52278379e-01
5.85168242e-01 -6.25916943e-02 2.29646623e-01 -4.32621956e-01
-3.45072716e-01 -7.25378282e-04 4.00579125e-01 6.58983290e-01
5.08620679e-01 1.25819266e-01 -6.40144885e-01 4.56859529e-01
-5.09894252e-01 -1.35714933e-01 6.65506840e-01 6.81315601e-01
-2.33120233e-01 -6.58585787e-01 -8.96678150e-01 3.49170297e-01
-5.02637804e-01 5.33839524e-01 -1.30120844e-01 1.14125383e+00
3.96231234e-01 1.00325716e+00 2.37094820e-01 -5.59624851e-01
6.39678836e-01 2.62021542e-01 2.26697654e-01 -1.20731667e-01
-4.63304043e-01 -2.88160920e-01 -8.06602910e-02 -3.96553963e-01
-4.00289983e-01 -6.96580350e-01 -1.27449119e+00 3.16987634e-02
-4.77212757e-01 1.24068625e-01 1.17816794e+00 1.09422505e+00
7.80214593e-02 1.32224119e+00 8.09116364e-01 -1.22646618e+00
-4.46578979e-01 -1.39873743e+00 -6.41471684e-01 7.73863018e-01
4.91319120e-01 -1.07711363e+00 -7.52719998e-01 -1.59354374e-01] | [6.790188312530518, 2.3957502841949463] |
e3745a8e-fd0a-4972-a5a4-43445c924d8e | towards-more-accurate-automatic-sleep-staging | 1907.13177 | null | https://arxiv.org/abs/1907.13177v3 | https://arxiv.org/pdf/1907.13177v3.pdf | Towards More Accurate Automatic Sleep Staging via Deep Transfer Learning | Background: Despite recent significant progress in the development of automatic sleep staging methods, building a good model still remains a big challenge for sleep studies with a small cohort due to the data-variability and data-inefficiency issues. This work presents a deep transfer learning approach to overcome these issues and enable transferring knowledge from a large dataset to a small cohort for automatic sleep staging. Methods: We start from a generic end-to-end deep learning framework for sequence-to-sequence sleep staging and derive two networks as the means for transfer learning. The networks are first trained in the source domain (i.e. the large database). The pretrained networks are then finetuned in the target domain (i.e. the small cohort) to complete knowledge transfer. We employ the Montreal Archive of Sleep Studies (MASS) database consisting of 200 subjects as the source domain and study deep transfer learning on three different target domains: the Sleep Cassette subset and the Sleep Telemetry subset of the Sleep-EDF Expanded database, and the Surrey-cEEGrid database. The target domains are purposely adopted to cover different degrees of data mismatch to the source domains. Results: Our experimental results show significant performance improvement on automatic sleep staging on the target domains achieved with the proposed deep transfer learning approach. Conclusions: These results suggest the efficacy of the proposed approach in addressing the above-mentioned data-variability and data-inefficiency issues. Significance: As a consequence, it would enable one to improve the quality of automatic sleep staging models when the amount of data is relatively small. The source code and the pretrained models are available at http://github.com/pquochuy/sleep_transfer_learning. | ['Oliver Y. Chén', 'Alfred Mertins', 'Philipp Koch', 'Zongqing Lu', 'Huy Phan', 'Maarten De Vos', 'Ian McLoughlin'] | 2019-07-30 | null | null | null | null | ['sleep-stage-detection', 'multimodal-sleep-stage-detection', 'sleep-staging', 'automatic-sleep-stage-classification'] | ['medical', 'medical', 'medical', 'medical'] | [ 3.87622528e-02 -9.19565633e-02 -2.94980496e-01 -6.67124450e-01
-8.31846893e-01 -2.13400811e-01 1.36552334e-01 -3.11461121e-01
-7.88769960e-01 1.05172741e+00 1.83755845e-01 -7.84975290e-02
2.10103001e-02 -4.34216797e-01 -3.12610894e-01 -6.64413512e-01
9.15939882e-02 7.79043078e-01 4.12365347e-01 -2.35007912e-01
1.19412556e-01 5.78943128e-03 -1.53666222e+00 1.48957342e-01
1.13067067e+00 9.70009208e-01 4.22627568e-01 5.03369093e-01
1.30147398e-01 3.94698203e-01 -6.24205351e-01 -8.20461214e-02
2.98426986e-01 -7.13619292e-01 -9.68167365e-01 -2.28157371e-01
2.32290089e-01 -1.29120916e-01 5.99730480e-03 9.17425931e-01
8.86417091e-01 3.74559909e-01 4.52229083e-01 -1.23248565e+00
-3.99760842e-01 -5.81613705e-02 -2.03667045e-01 8.19739342e-01
7.31394738e-02 7.86993802e-02 7.12820590e-01 -6.02035165e-01
3.40912908e-01 5.10660946e-01 9.40773010e-01 1.00866175e+00
-1.10649061e+00 -8.98204446e-01 -5.33384144e-01 5.13596654e-01
-1.39689505e+00 -7.88879395e-01 4.61306930e-01 -3.65998864e-01
9.40875471e-01 8.71204436e-02 7.83925056e-01 1.10987163e+00
3.88196886e-01 2.98294872e-01 1.17592072e+00 -3.55114281e-01
4.36019093e-01 3.49863380e-01 1.85605496e-01 7.35249937e-01
5.00009535e-03 1.49714053e-01 -7.12917507e-01 1.03356548e-01
4.78624672e-01 3.90560552e-02 1.03745878e-01 -1.15361303e-01
-9.28188562e-01 8.21920693e-01 4.85856742e-01 5.20410955e-01
-1.40883669e-01 -3.52957875e-01 7.26756752e-01 6.06063366e-01
8.27010334e-01 2.24840194e-01 -6.89923882e-01 -3.83417219e-01
-1.53205562e+00 1.99373253e-03 8.16161871e-01 8.55969548e-01
8.05222929e-01 -2.00616837e-01 -5.11450879e-02 1.06305265e+00
2.69846201e-01 3.54057521e-01 1.11449444e+00 -6.25687301e-01
5.48676789e-01 5.38669109e-01 -3.13259214e-02 -4.85242516e-01
-8.21302712e-01 -3.50077569e-01 -8.35832655e-01 5.83455190e-02
3.90506893e-01 -1.98070258e-01 -8.97754312e-01 1.97206521e+00
2.19721138e-01 1.03527606e-01 2.14135703e-02 8.48198533e-01
9.48093891e-01 4.17702615e-01 2.02265978e-01 -1.08947933e-01
1.50457478e+00 -1.12916481e+00 -4.86981571e-01 -4.67860252e-01
5.62926471e-01 -5.11474609e-01 1.31697857e+00 1.65912718e-01
-1.14192736e+00 -8.91834080e-01 -1.03385317e+00 -4.26676899e-01
-4.27327633e-01 4.25830990e-01 1.26552522e-01 6.29931509e-01
-1.38033974e+00 4.53418791e-01 -1.11562562e+00 -9.85047817e-01
4.31184262e-01 8.82717192e-01 -2.68490404e-01 1.99869975e-01
-1.24918687e+00 9.95785058e-01 3.73709112e-01 1.16973490e-01
-9.13604975e-01 -7.92344868e-01 -5.09107947e-01 1.11468665e-01
-3.28228362e-02 -9.11202669e-01 1.22000206e+00 -1.18266797e+00
-1.32428896e+00 1.29160440e+00 -1.16394110e-01 -4.09796536e-01
4.63438362e-01 -1.10443510e-01 -5.49581647e-01 -8.67282897e-02
4.10354048e-01 5.31155646e-01 6.25842988e-01 -5.56935906e-01
-6.30219281e-01 -4.22537506e-01 -1.97906435e-01 2.16271132e-01
-3.66993755e-01 4.66272198e-02 -3.29267770e-01 -4.44517910e-01
-4.84790951e-01 -1.21903956e+00 1.50922850e-01 -1.10891379e-01
-1.23463780e-01 -2.16898754e-01 3.04475635e-01 -7.67987192e-01
1.26300800e+00 -2.20054293e+00 3.00495148e-01 -2.04251096e-01
2.58076400e-01 3.92058969e-01 1.15441671e-02 4.06563103e-01
-1.90665096e-01 -1.69731125e-01 -3.27257276e-01 -7.31355548e-01
-2.09469467e-01 3.54801089e-01 4.10012811e-01 5.97258091e-01
-2.10527778e-01 6.40891433e-01 -7.50843585e-01 -5.99137008e-01
-1.41425012e-02 4.87688147e-02 -6.58746600e-01 5.46169519e-01
1.40595093e-01 7.59157062e-01 -3.52059752e-01 2.62572676e-01
3.27215880e-01 -2.11239308e-01 -1.18182138e-01 -1.66033581e-01
-1.41359538e-01 4.77798492e-01 -5.52582920e-01 2.11116481e+00
-6.64164841e-01 5.39979279e-01 -1.43200591e-01 -9.90899622e-01
8.82174551e-01 1.35422066e-01 4.86209214e-01 -8.83542299e-01
2.31214061e-01 2.16014460e-01 2.87377089e-01 -5.71265042e-01
4.16223049e-01 -8.76679003e-01 -2.20013276e-01 5.69164634e-01
3.99926871e-01 1.51443794e-01 2.50918925e-01 -1.99606955e-01
1.19927037e+00 -1.26150370e-01 4.57394034e-01 -5.46814024e-01
7.05330253e-01 -1.71217006e-02 8.99282157e-01 3.36245477e-01
-5.97907424e-01 6.12184644e-01 3.01729858e-01 -4.55963939e-01
-1.04973221e+00 -1.07723951e+00 -1.25154123e-01 1.39954233e+00
-7.89742544e-02 -5.10802567e-01 -7.89767981e-01 -7.44659662e-01
-4.05385405e-01 5.14161050e-01 -9.35799837e-01 -5.47319651e-01
-3.41630995e-01 -7.77220964e-01 7.03379989e-01 5.96246183e-01
8.54452252e-01 -1.29918683e+00 -5.25689363e-01 -1.54530302e-01
-3.95069629e-01 -7.60660052e-01 -6.77927136e-01 3.25835198e-01
-1.00725329e+00 -1.00328195e+00 -8.03424478e-01 -8.62521887e-01
4.64001834e-01 -1.06337719e-01 1.19879568e+00 1.92330614e-01
-1.77404702e-01 5.38654067e-02 -3.19393456e-01 -3.45583230e-01
-5.79149783e-01 6.40321076e-01 3.41103971e-01 -1.32182643e-01
6.93659782e-01 -8.07616115e-01 -9.13413167e-01 5.56086481e-01
-6.78024173e-01 2.46237312e-02 6.19340062e-01 8.69638383e-01
3.85453552e-01 -1.18678398e-01 8.32777023e-01 -7.87223160e-01
6.02183878e-01 -9.02599871e-01 -4.05237913e-01 6.89491555e-02
-8.11840117e-01 -3.19963098e-02 6.84426606e-01 -2.13785395e-01
-1.01191616e+00 -1.93526447e-01 -5.18623173e-01 -2.03196362e-01
-2.86583811e-01 2.82710642e-01 -1.22467324e-01 2.86152601e-01
7.76182234e-01 3.66262168e-01 2.57306010e-01 -6.74137533e-01
-1.00513361e-01 7.74168015e-01 2.92306155e-01 -3.03802848e-01
4.96395975e-01 1.78972080e-01 -2.62336671e-01 -8.60036373e-01
-8.72830689e-01 -7.67888248e-01 -1.02473450e+00 1.40700042e-01
1.25593793e+00 -8.88656557e-01 -2.37794995e-01 4.91023213e-01
-4.69451815e-01 -8.63751352e-01 -4.73933756e-01 3.69297177e-01
-7.95626223e-01 1.59950346e-01 -3.86830896e-01 -1.04891531e-01
-8.02399993e-01 -1.04877961e+00 1.06056726e+00 3.61016631e-01
-4.63838875e-01 -1.28651094e+00 7.03557074e-01 4.97850001e-01
4.29665685e-01 -2.41555005e-01 9.39288497e-01 -8.92924011e-01
-2.28191540e-02 1.05333380e-01 -5.00206649e-02 6.01287007e-01
3.68831098e-01 -5.15035689e-01 -9.01986003e-01 -4.07611758e-01
1.41483992e-01 -4.95677680e-01 6.05744123e-01 2.99993038e-01
7.83352435e-01 1.05360754e-01 -2.06304342e-01 6.84531748e-01
1.20040309e+00 1.14810601e-01 6.33658111e-01 4.34592664e-01
3.25626969e-01 2.88891435e-01 4.88356054e-01 2.81561732e-01
5.99063814e-01 8.13545704e-01 -1.62809953e-01 -1.96484923e-01
-3.34078133e-01 4.12166938e-02 3.87972444e-01 1.14352798e+00
-1.63227186e-01 -3.08824833e-02 -8.64430904e-01 6.07719600e-01
-1.59250641e+00 -8.76266599e-01 2.47826830e-01 2.15609717e+00
9.72366214e-01 9.23617464e-03 6.56566858e-01 -9.43496898e-02
4.12028998e-01 -4.32249252e-03 -6.71477616e-01 -5.33146799e-01
5.23451567e-01 4.71136004e-01 1.79102734e-01 1.58896416e-01
-9.13240612e-01 7.71882772e-01 5.68176270e+00 6.66884780e-01
-1.06326783e+00 7.05138743e-01 2.61749834e-01 -4.69966799e-01
4.14918214e-01 -4.31505978e-01 -8.50880742e-01 9.67142582e-01
1.77112305e+00 -1.83538750e-01 4.76829499e-01 6.96842253e-01
4.44134444e-01 -2.76538998e-01 -1.18396699e+00 9.98916686e-01
3.71347852e-02 -9.90230143e-01 -6.43051744e-01 8.52146000e-02
5.10443807e-01 4.73639756e-01 -6.53168634e-02 7.49641538e-01
-1.85195699e-01 -7.99881220e-01 4.71098810e-01 5.80409229e-01
1.07228184e+00 -5.62390029e-01 1.03914165e+00 5.09029746e-01
-1.10232198e+00 -3.08334138e-02 -4.95638639e-01 2.90000811e-02
5.80003578e-03 2.17866927e-01 -9.85252380e-01 3.97157133e-01
9.69371438e-01 8.17289412e-01 -9.00151074e-01 1.04496264e+00
1.15769267e-01 9.22293782e-01 -1.99975178e-01 6.77105933e-02
-1.52392730e-01 -3.39187086e-01 -3.50346416e-02 1.12033212e+00
3.56743783e-01 -9.89377871e-02 -6.73331097e-02 8.99866462e-01
-1.76285490e-01 -1.35003582e-01 -4.18930620e-01 2.02163681e-01
1.66463137e-01 1.27566767e+00 -8.05990517e-01 -2.16218576e-01
-6.32027864e-01 1.19044399e+00 5.23807824e-01 1.81799512e-02
-8.66003215e-01 -2.08392322e-01 7.02977896e-01 2.06017047e-01
2.04059795e-01 3.06978356e-02 -2.56879658e-01 -1.17946029e+00
-2.64769226e-01 -8.03903878e-01 5.53052068e-01 -6.02506518e-01
-1.51727664e+00 7.00719535e-01 2.03717306e-01 -1.23945129e+00
-2.04318896e-01 -1.89716443e-01 -7.84081817e-01 1.00893414e+00
-1.19818699e+00 -1.12891030e+00 -4.83071446e-01 7.87231684e-01
9.04454827e-01 -3.83274227e-01 8.19696844e-01 8.53896081e-01
-7.15520263e-01 8.12589705e-01 2.37666368e-01 -7.77683631e-02
9.43851769e-01 -1.34611809e+00 2.53891647e-01 5.44687212e-01
-2.02541575e-01 7.70969987e-01 4.52033222e-01 -6.04717493e-01
-8.42211604e-01 -1.05821705e+00 9.25425947e-01 -6.21152699e-01
5.21921098e-01 -6.56838655e-01 -7.84546316e-01 7.16527820e-01
1.63380742e-01 -2.03430742e-01 1.18554950e+00 3.35001349e-01
3.65932256e-01 -3.91420096e-01 -1.17131448e+00 2.27700219e-01
9.46190536e-01 -5.87758362e-01 -9.81711149e-01 2.53342062e-01
1.70891032e-01 -3.51897061e-01 -9.36488152e-01 -6.21867813e-02
6.11298442e-01 -1.09827185e+00 6.46170139e-01 -3.96332711e-01
3.22236210e-01 -7.92956874e-02 1.76785484e-01 -1.43683565e+00
-2.61728168e-01 -3.91415387e-01 2.14663506e-01 1.22371209e+00
3.06254178e-01 -5.13272822e-01 9.46296096e-01 6.17533982e-01
-5.11306822e-01 -6.16989911e-01 -1.34570861e+00 -7.18371272e-01
2.11721972e-01 -4.58756275e-02 3.67571652e-01 5.46397924e-01
-1.90579444e-01 7.93109357e-01 -4.03901815e-01 -1.68667689e-01
2.39118412e-01 2.33927593e-02 6.96476340e-01 -1.34319806e+00
-1.59656033e-01 -1.27679370e-02 -2.59984761e-01 -6.68628991e-01
2.14974225e-01 -1.07712007e+00 2.83575833e-01 -1.35187483e+00
4.34296101e-01 -5.68069398e-01 -5.22604942e-01 5.45423627e-01
-3.48405316e-02 3.81916314e-01 6.59336615e-03 3.36229414e-01
-6.44218802e-01 6.77541137e-01 1.07336295e+00 2.69248545e-01
-3.84370208e-01 5.23597896e-01 -6.59350574e-01 5.59850276e-01
1.10065591e+00 -7.11903274e-01 -7.68899798e-01 -8.19881335e-02
-1.90159883e-02 -5.97131588e-02 1.70433998e-01 -1.20397758e+00
1.07745692e-01 2.23331645e-01 3.95599484e-01 -3.70186388e-01
3.72183949e-01 -7.75140524e-01 1.01185858e-01 5.20982444e-01
-2.27317169e-01 2.43231609e-01 3.53444576e-01 4.32697833e-01
1.65001601e-01 -3.31092209e-01 9.73105907e-01 -7.68562332e-02
-8.49019229e-01 1.80612594e-01 -4.82173592e-01 3.86261702e-01
7.36174405e-01 -2.95256346e-01 -2.13385016e-01 -1.76388726e-01
-1.02871943e+00 1.33785650e-01 5.38534284e-01 5.32219708e-01
2.90956438e-01 -1.05015659e+00 -2.98970938e-01 6.50850832e-01
1.32277668e-01 -2.42589697e-01 3.74425560e-01 1.24352503e+00
-4.40050900e-01 3.09873700e-01 -8.46821964e-01 -4.41214263e-01
-1.33141732e+00 3.66813660e-01 4.92887199e-01 -4.31520402e-01
-5.54550827e-01 7.26676464e-01 2.69436359e-01 -5.31492233e-01
6.78717112e-03 -6.45615518e-01 -1.16477869e-01 7.06790388e-02
3.29611242e-01 5.48457861e-01 4.21784163e-01 -6.11369967e-01
-4.96548623e-01 4.01771992e-01 -3.16639282e-02 1.73974648e-01
1.44383979e+00 -4.25696641e-01 7.51456246e-02 6.09096587e-01
1.30523360e+00 -1.30224898e-01 -9.00188327e-01 -7.89408572e-03
1.03716012e-02 -5.96263558e-02 -3.94145325e-02 -9.76060331e-01
-1.01474631e+00 9.50427830e-01 1.15809631e+00 -6.00271709e-02
1.41308272e+00 1.03848964e-01 1.04304969e+00 2.15361148e-01
2.21206844e-01 -9.91334260e-01 -2.74389368e-02 3.50161552e-01
6.75480127e-01 -1.32015872e+00 3.05054188e-02 3.29913229e-01
-7.56236851e-01 8.50712419e-01 6.92829192e-01 -3.08969896e-02
6.57374144e-01 -1.23032354e-01 6.41694143e-02 -3.61839592e-01
-3.96055102e-01 -3.22227031e-01 4.28653061e-01 6.97107434e-01
3.39790881e-01 -1.54044136e-01 -3.80657941e-01 9.29522693e-01
-4.05923456e-01 5.81876695e-01 2.36591652e-01 6.98793769e-01
-3.38417858e-01 -1.31487513e+00 -2.13354789e-02 7.09313393e-01
-4.54662740e-01 -1.80497333e-01 -2.54133165e-01 9.25242901e-01
5.09368479e-01 9.31767106e-01 3.00656408e-02 -4.33986783e-01
3.82825881e-01 2.69201100e-01 3.13345730e-01 -1.03820086e+00
-7.78331935e-01 -2.20575556e-01 2.22743690e-01 -4.69151467e-01
-6.41394913e-01 -5.87997437e-01 -1.13613343e+00 -3.14143121e-01
-1.10375695e-01 3.90311778e-01 2.13405505e-01 1.03600824e+00
5.33284843e-01 6.00091696e-01 2.60448128e-01 -9.97181475e-01
-4.88587737e-01 -1.30989861e+00 -7.90742040e-01 3.90682518e-01
4.90565538e-01 -8.38257909e-01 -1.12853333e-01 2.38199785e-01] | [13.472515106201172, 3.5094892978668213] |
5344f93c-822a-476f-a05c-5120c08558b3 | 3d-instance-segmentation-via-multi-task | 1906.08650 | null | https://arxiv.org/abs/1906.08650v2 | https://arxiv.org/pdf/1906.08650v2.pdf | 3D Instance Segmentation via Multi-Task Metric Learning | We propose a novel method for instance label segmentation of dense 3D voxel grids. We target volumetric scene representations, which have been acquired with depth sensors or multi-view stereo methods and which have been processed with semantic 3D reconstruction or scene completion methods. The main task is to learn shape information about individual object instances in order to accurately separate them, including connected and incompletely scanned objects. We solve the 3D instance-labeling problem with a multi-task learning strategy. The first goal is to learn an abstract feature embedding, which groups voxels with the same instance label close to each other while separating clusters with different instance labels from each other. The second goal is to learn instance information by densely estimating directional information of the instance's center of mass for each voxel. This is particularly useful to find instance boundaries in the clustering post-processing step, as well as, for scoring the segmentation quality for the first goal. Both synthetic and real-world experiments demonstrate the viability and merits of our approach. In fact, it achieves state-of-the-art performance on the ScanNet 3D instance segmentation benchmark. | ['Martin R. Oswald', 'Bernard Ghanem', 'Marc Pollefeys', 'Jean Lahoud'] | 2019-06-20 | 3d-instance-segmentation-via-multi-task-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Lahoud_3D_Instance_Segmentation_via_Multi-Task_Metric_Learning_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Lahoud_3D_Instance_Segmentation_via_Multi-Task_Metric_Learning_ICCV_2019_paper.pdf | iccv-2019-10 | ['3d-instance-segmentation-1', '3d-semantic-instance-segmentation'] | ['computer-vision', 'computer-vision'] | [ 2.80170888e-01 4.89574879e-01 5.56460805e-02 -5.06572306e-01
-9.48808610e-01 -3.28373671e-01 3.98964435e-01 6.57364547e-01
-2.26476535e-01 3.11029911e-01 -1.35914251e-01 3.01186293e-01
-2.30136842e-01 -8.69194448e-01 -7.50196040e-01 -9.25007403e-01
-1.79228544e-01 1.38031089e+00 3.37175161e-01 4.33766574e-01
3.48464042e-01 8.97839725e-01 -1.60866141e+00 1.71169326e-01
5.44840753e-01 1.24478257e+00 4.24250126e-01 3.55611354e-01
-4.48446304e-01 4.55300003e-01 -1.14370085e-01 9.26219150e-02
2.51308858e-01 1.14580154e-01 -1.07989907e+00 6.81548119e-01
5.08162975e-01 -1.70773163e-01 2.17685059e-01 1.08202946e+00
2.96559095e-01 -8.26970115e-02 1.12789369e+00 -9.98828769e-01
1.48057550e-01 1.40735164e-01 -8.45478356e-01 -1.03470385e-01
1.05941877e-01 -1.62521675e-01 9.39366221e-01 -9.41820145e-01
6.31597161e-01 1.11693645e+00 3.61852914e-01 4.85849231e-01
-1.32262111e+00 -2.03914955e-01 1.29317582e-01 3.27768810e-02
-1.38065040e+00 -1.17181189e-01 9.76218700e-01 -1.12030661e+00
5.06992042e-01 1.82099745e-01 7.35431731e-01 4.50199068e-01
-1.53393978e-02 8.93654168e-01 1.14275801e+00 -2.13460196e-02
6.28283381e-01 2.42369965e-01 5.01656830e-01 9.00097549e-01
1.81547388e-01 -4.43165213e-01 -1.33441105e-01 -2.34987605e-02
8.95153344e-01 2.91219980e-01 -9.99031886e-02 -9.31088805e-01
-1.23213661e+00 9.29813981e-01 5.94001234e-01 3.15180153e-01
-6.67704165e-01 8.95198807e-02 2.62437552e-01 -2.73093790e-01
8.33386004e-01 2.30764836e-01 -3.27724904e-01 4.07356948e-01
-9.83538985e-01 1.52286604e-01 7.11505055e-01 8.70928764e-01
1.17295539e+00 -4.29561704e-01 -8.77460018e-02 7.99425602e-01
3.65446150e-01 8.55140239e-02 1.71624795e-01 -1.00482404e+00
3.81078660e-01 1.00561488e+00 -6.64993748e-02 -6.68892801e-01
-5.73340595e-01 -3.50594044e-01 -8.55139732e-01 3.63910377e-01
4.73218381e-01 3.21440369e-01 -1.14022851e+00 1.16713059e+00
9.16748166e-01 4.09003675e-01 -2.18903884e-01 1.12233019e+00
1.07315600e+00 5.33077121e-01 -1.66510582e-01 -7.67437965e-02
1.55822122e+00 -7.82477379e-01 -2.25152522e-01 -2.64732838e-01
5.39871633e-01 -3.90043288e-01 7.12860227e-01 2.79608309e-01
-1.15200973e+00 -4.46139574e-01 -8.84952247e-01 -2.09826436e-02
-1.50660113e-01 -1.97257679e-02 6.35793686e-01 4.19215769e-01
-8.00474823e-01 5.63087642e-01 -1.07864380e+00 3.33266035e-02
1.01965916e+00 3.36089760e-01 -4.10326272e-01 -3.55687201e-01
-4.72101420e-01 4.89488602e-01 4.15229052e-01 -2.82625139e-01
-1.13297451e+00 -9.44326341e-01 -1.05957997e+00 -4.43055704e-02
4.00674850e-01 -6.70281529e-01 7.52718508e-01 -5.60879230e-01
-1.00847685e+00 1.55998802e+00 -2.36265585e-01 -2.18493775e-01
5.77096641e-01 1.10067062e-01 2.60033160e-01 4.70317900e-01
5.00630856e-01 6.18963718e-01 8.72423351e-01 -1.77937067e+00
-5.81302643e-01 -1.18243957e+00 -1.56043783e-01 3.04331928e-01
1.18027568e-01 -5.47188759e-01 -6.20774150e-01 -5.60925640e-02
8.78608823e-01 -6.26221955e-01 -5.73998928e-01 7.11683482e-02
-6.75270081e-01 -4.09734875e-01 8.61539364e-01 -4.86501008e-01
4.96212304e-01 -2.02015901e+00 4.91799772e-01 4.49710637e-01
6.78997219e-01 -3.70496303e-01 1.82990447e-01 -7.36713484e-02
9.20220613e-02 -6.93377107e-02 -6.41041875e-01 -5.50035179e-01
-1.73118189e-01 2.17748955e-01 1.83599979e-01 8.83657813e-01
2.09459573e-01 7.64622986e-01 -8.63810182e-01 -7.71959305e-01
5.97688377e-01 4.27963972e-01 -4.32779521e-01 4.67530221e-01
-4.29572791e-01 8.32307935e-01 -8.10649335e-01 6.92673802e-01
6.77648008e-01 -5.05375803e-01 -1.88169032e-01 -4.36961979e-01
-2.49131508e-02 8.47040489e-02 -1.61967397e+00 1.93281806e+00
-4.65553612e-01 1.25934869e-01 3.35206568e-01 -1.33243048e+00
6.95894659e-01 1.67144299e-01 1.23068583e+00 -4.73792255e-01
1.20952301e-01 2.08567545e-01 -6.73275173e-01 -5.43395460e-01
2.17680290e-01 -2.24897355e-01 -2.19095927e-02 5.20487845e-01
6.41722903e-02 -8.67180288e-01 -1.05372583e-02 1.50243357e-01
9.23664510e-01 -2.15942279e-01 1.80366829e-01 -3.38243276e-01
5.59345424e-01 -1.27840349e-02 4.10064608e-01 3.18268359e-01
9.35285613e-02 8.66802096e-01 4.33710307e-01 -4.02087301e-01
-8.82240236e-01 -1.40210176e+00 -4.64562148e-01 4.18947726e-01
6.11371458e-01 5.66109829e-02 -7.61127472e-01 -7.52326310e-01
1.65318370e-01 6.00462794e-01 -6.41366780e-01 2.88211435e-01
-5.17307878e-01 -5.00562906e-01 -1.85097411e-01 2.76291519e-01
2.70736217e-01 -7.64063537e-01 -9.44185257e-01 3.41933668e-01
-2.22789764e-01 -1.19663262e+00 -1.04882546e-01 4.76635873e-01
-1.03130126e+00 -1.41683805e+00 -7.80234337e-01 -8.73609424e-01
9.58543420e-01 1.60703495e-01 1.23422015e+00 4.48063351e-02
-6.35042250e-01 7.86004841e-01 -1.24846116e-01 -8.96380767e-02
-8.03568885e-02 9.21685342e-03 -2.60673523e-01 2.63359964e-01
-4.71812412e-02 -4.93012488e-01 -6.79084659e-01 3.62569958e-01
-1.01508045e+00 2.79787984e-02 4.24785197e-01 5.87938488e-01
1.40129828e+00 2.51812965e-01 4.69346456e-02 -1.10643804e+00
1.69558991e-02 -5.51117361e-01 -6.26930892e-01 1.01299644e-01
-2.64212281e-01 -3.00321877e-02 1.36297762e-01 7.04617705e-03
-6.07741058e-01 4.26654011e-01 -1.21705256e-01 -6.34222627e-01
-6.13742948e-01 2.01501742e-01 -3.04468662e-01 1.05236299e-01
1.62388906e-01 2.75308132e-01 1.23891979e-03 -4.77354050e-01
3.93195570e-01 4.84471381e-01 3.17486346e-01 -6.32974982e-01
5.71223915e-01 9.67542231e-01 2.65255958e-01 -1.17909682e+00
-9.28551733e-01 -1.06549132e+00 -1.15825808e+00 -4.74712610e-01
1.29011858e+00 -8.68205070e-01 -7.08576262e-01 3.14082861e-01
-1.12463641e+00 -4.80463326e-01 -5.93371987e-01 4.78409439e-01
-9.64574814e-01 2.94551551e-01 -5.51646650e-01 -5.63823402e-01
-1.06806152e-01 -1.42107117e+00 1.68524766e+00 6.60503134e-02
9.16101262e-02 -1.04591548e+00 9.92178321e-02 6.80548370e-01
-1.67085931e-01 4.96687680e-01 1.20674455e+00 -5.74440598e-01
-8.46798599e-01 2.98373792e-02 -2.48233721e-01 1.64147496e-01
1.17572889e-01 -3.21954310e-01 -1.01524210e+00 -2.27192536e-01
2.33013123e-01 -3.70434910e-01 9.24726963e-01 8.66098285e-01
1.58010781e+00 1.12833641e-01 -6.62942231e-01 7.10128427e-01
1.58916581e+00 -1.89033106e-01 4.23307747e-01 -4.20313738e-02
9.76154447e-01 8.89120281e-01 6.57928169e-01 3.82911652e-01
3.55175555e-01 7.62289166e-01 9.46548045e-01 -1.30034998e-01
-1.77621648e-01 6.80170506e-02 -2.85094917e-01 5.50048232e-01
2.05699652e-01 -8.11687950e-03 -1.03867078e+00 6.82281196e-01
-1.49452150e+00 -6.90496445e-01 -4.46962416e-01 2.14971209e+00
4.17447627e-01 6.77147955e-02 8.63846578e-03 2.75261462e-01
7.15105832e-01 2.35535964e-01 -6.92707241e-01 8.15550517e-03
2.70628721e-01 2.47633025e-01 4.48481411e-01 5.42517304e-01
-1.25855088e+00 8.58957112e-01 4.46594572e+00 8.62289011e-01
-7.89580762e-01 1.52380303e-01 9.72466648e-01 9.27019641e-02
-3.09765279e-01 -3.08448106e-01 -8.19228709e-01 2.26205036e-01
3.02468032e-01 2.69739896e-01 1.63776040e-01 7.94066548e-01
3.90160754e-02 -4.22911376e-01 -1.41980743e+00 1.21894181e+00
1.18132629e-01 -1.37148511e+00 2.94958055e-03 2.61470139e-01
6.79882705e-01 8.66943821e-02 -2.10263699e-01 -6.33037016e-02
-3.56007703e-02 -1.02530015e+00 7.70080090e-01 4.18046921e-01
7.66682744e-01 -6.28335536e-01 3.70015472e-01 4.82444197e-01
-1.38885689e+00 2.61468172e-01 -4.11626279e-01 2.97996700e-01
4.02145445e-01 1.37022173e+00 -8.66860867e-01 5.71198106e-01
5.33388555e-01 9.03380394e-01 -3.27217907e-01 1.18936539e+00
-1.24865107e-01 2.94829130e-01 -3.48677725e-01 2.47819424e-01
3.16532612e-01 -5.86104393e-01 6.01174355e-01 9.26266551e-01
8.84320587e-02 3.06467056e-01 5.50355077e-01 1.18382478e+00
-8.14230964e-02 5.32862209e-02 -6.23990834e-01 2.22097233e-01
1.90601751e-01 1.42992103e+00 -1.35618424e+00 -4.35245454e-01
-8.17308202e-02 1.04803658e+00 3.37620020e-01 8.92058238e-02
-6.29423022e-01 1.24539137e-01 6.67360842e-01 4.06675607e-01
2.71765471e-01 -3.84642035e-01 -6.32990956e-01 -7.60467052e-01
1.12035759e-01 -1.71890572e-01 3.61128718e-01 -4.55979824e-01
-1.30959833e+00 3.77102792e-01 -1.17537551e-01 -1.14206970e+00
6.31187335e-02 -5.87680876e-01 -4.13277537e-01 6.81218207e-01
-1.48854291e+00 -9.40974355e-01 -5.18994153e-01 7.36222267e-01
8.24831307e-01 3.41099590e-01 5.61986983e-01 2.97082365e-01
-2.58392036e-01 -1.24259040e-01 -6.33961614e-03 6.86392933e-02
-9.18432400e-02 -1.44054985e+00 -1.87560782e-01 2.08626494e-01
2.58518904e-01 -1.37942359e-01 3.02517682e-01 -5.99734426e-01
-1.25215149e+00 -1.38123059e+00 3.53595316e-01 -3.87968391e-01
2.28902057e-01 -4.54198599e-01 -9.65977311e-01 3.75928611e-01
-4.94779795e-01 2.44251594e-01 4.94064033e-01 -9.03826132e-02
-1.38943298e-02 3.27071967e-03 -1.44127250e+00 8.48865975e-03
9.23747182e-01 -3.87188375e-01 -5.59224367e-01 7.17951238e-01
5.38085639e-01 -6.28986239e-01 -1.02761436e+00 3.30473989e-01
5.15736528e-02 -1.02780151e+00 1.26331413e+00 -4.07478541e-01
6.41404569e-01 -2.36676380e-01 -3.26347142e-01 -1.25820458e+00
-2.00967208e-01 2.89521843e-01 1.21728815e-01 1.00767863e+00
2.44779319e-01 -3.50158840e-01 1.12448537e+00 4.77801591e-01
-4.20391172e-01 -9.92432594e-01 -1.23366225e+00 -5.68162382e-01
-5.00436127e-03 -6.36429131e-01 4.36342508e-01 9.38505590e-01
-6.72954381e-01 2.42414102e-01 3.01067412e-01 5.73909700e-01
1.20096409e+00 6.14678264e-01 6.03635907e-01 -1.56321418e+00
-1.42598106e-02 -3.60777348e-01 -6.89482749e-01 -1.12727094e+00
3.24528217e-01 -1.21390426e+00 7.77799562e-02 -1.97948587e+00
3.00234646e-01 -7.84655750e-01 -1.60931498e-01 1.09598383e-01
1.52424738e-01 2.02481896e-01 -2.53725857e-01 1.50502548e-01
-6.85717463e-01 6.50555551e-01 1.37070489e+00 -5.45715690e-01
-2.21712828e-01 2.74086624e-01 -1.86558113e-01 7.46758759e-01
4.59798485e-01 -5.82033575e-01 -3.12086046e-01 -5.28211296e-01
-1.45933732e-01 2.77385443e-01 7.50288010e-01 -9.59386289e-01
4.81017642e-02 1.03995819e-02 3.30727845e-01 -9.86864746e-01
6.15756035e-01 -1.08783412e+00 -3.23921517e-02 4.35202599e-01
-1.32236227e-01 -4.34682161e-01 -1.51921004e-01 6.75291181e-01
-1.14950813e-01 -2.46908635e-01 1.10579371e+00 -5.11284053e-01
-6.75505936e-01 7.07271576e-01 -6.57493714e-03 2.17159510e-01
1.46932852e+00 -3.51020783e-01 3.42762947e-01 1.73918396e-01
-1.03311336e+00 4.37097341e-01 4.63218302e-01 1.17381275e-01
9.83929753e-01 -1.10249770e+00 -7.26668477e-01 3.46622050e-01
2.68180430e-01 7.50880778e-01 3.23438108e-01 8.12661707e-01
-4.60323483e-01 1.56255171e-01 4.30235527e-02 -1.45474875e+00
-1.13913381e+00 2.98439920e-01 4.94927198e-01 -1.78535253e-01
-9.14329648e-01 1.01439393e+00 4.33343470e-01 -5.50299704e-01
2.66881794e-01 -4.62717593e-01 -3.17928135e-01 3.03469598e-01
2.17411861e-01 3.57714474e-01 3.39965820e-01 -6.07249141e-01
-3.99863243e-01 8.81341815e-01 5.78945540e-02 7.26567879e-02
1.69063091e+00 -3.09925564e-02 -2.97831833e-01 7.35656321e-01
1.48438764e+00 -4.28739488e-01 -1.22956681e+00 -2.57587433e-01
-2.98637357e-02 -6.11059308e-01 2.69602001e-01 -3.58979940e-01
-1.43025005e+00 1.13736761e+00 7.52259791e-01 1.72550052e-01
6.43822134e-01 7.18464911e-01 6.23162866e-01 -1.93005353e-02
7.66437411e-01 -1.02565312e+00 2.25703493e-01 3.01466882e-01
6.71930254e-01 -1.40416610e+00 7.22791106e-02 -6.35262668e-01
-3.45376670e-01 1.02070796e+00 3.45916808e-01 -2.02756986e-01
8.06801140e-01 -7.45404512e-02 -2.16658905e-01 -1.00280261e+00
-1.58082873e-01 -3.36783975e-01 3.37942749e-01 4.57812786e-01
2.21785475e-02 3.03309858e-01 3.79259177e-02 2.86117971e-01
1.09280281e-01 -4.88762110e-01 1.97596237e-01 5.70611179e-01
-5.85859597e-01 -6.93932533e-01 -4.60912287e-01 8.82295787e-01
-4.98020984e-02 4.46316302e-01 -2.48720005e-01 6.34187818e-01
1.47085667e-01 6.12970948e-01 3.26777399e-01 -3.20964046e-02
3.34256381e-01 -7.55101889e-02 6.64304852e-01 -1.06883550e+00
-2.12126881e-01 9.59565267e-02 -3.04782122e-01 -7.92069435e-01
-3.54266256e-01 -8.20253849e-01 -1.61795533e+00 3.07325631e-01
-2.62933015e-03 1.23557083e-01 9.61588502e-01 9.48724806e-01
-1.57126397e-01 4.76501465e-01 9.25695956e-01 -1.43869758e+00
-1.68717280e-01 -4.37303215e-01 -9.93419945e-01 6.35113597e-01
2.13483363e-01 -8.55674684e-01 -4.07610893e-01 -5.44629879e-02] | [8.071526527404785, -3.0769882202148438] |
1c128806-09cb-4b6a-963d-6e9547b19adc | streaming-voice-query-recognition-using | 1812.07754 | null | http://arxiv.org/abs/1812.07754v1 | http://arxiv.org/pdf/1812.07754v1.pdf | Streaming Voice Query Recognition using Causal Convolutional Recurrent Neural Networks | Voice-enabled commercial products are ubiquitous, typically enabled by
lightweight on-device keyword spotting (KWS) and full automatic speech
recognition (ASR) in the cloud. ASR systems require significant computational
resources in training and for inference, not to mention copious amounts of
annotated speech data. KWS systems, on the other hand, are less
resource-intensive but have limited capabilities. On the Comcast Xfinity X1
entertainment platform, we explore a middle ground between ASR and KWS: We
introduce a novel, resource-efficient neural network for voice query
recognition that is much more accurate than state-of-the-art CNNs for KWS, yet
can be easily trained and deployed with limited resources. On an evaluation
dataset representing the top 200 voice queries, we achieve a low false alarm
rate of 1% and a query error rate of 6%. Our model performs inference 8.24x
faster than the current ASR system. | ['Yajie Mao', 'Gefei Yang', 'Raphael Tang', 'Ferhan Ture', 'Jimmy Lin', 'Hong Wei'] | 2018-12-19 | null | null | null | null | ['voice-query-recognition'] | ['speech'] | [-2.26343468e-01 -2.04874307e-01 -4.22509164e-01 -4.06578660e-01
-1.24923253e+00 -6.10945642e-01 3.00340444e-01 -2.47225553e-01
-4.66440231e-01 2.95821458e-01 -6.74334243e-02 -9.54515219e-01
1.71981141e-01 -4.79714036e-01 -6.09399617e-01 -2.10336968e-02
3.57762873e-01 5.74987173e-01 2.98315912e-01 -1.32496804e-01
-3.40783387e-01 5.87573409e-01 -1.58175838e+00 1.33348912e-01
3.55785877e-01 1.48544812e+00 1.41599432e-01 1.21253610e+00
-2.79403090e-01 8.25414181e-01 -1.11437404e+00 -3.40490550e-01
-8.80382732e-02 4.48049396e-01 -6.53717101e-01 -4.68129069e-01
4.49086547e-01 -7.35837817e-01 -7.43987560e-01 7.34334528e-01
1.04399872e+00 7.80466245e-03 -2.88001359e-01 -1.26242888e+00
-6.30751371e-01 7.25611687e-01 2.40299106e-01 4.88100290e-01
4.60777760e-01 3.04044902e-01 1.03053212e+00 -1.00555444e+00
-1.15036648e-02 1.03506577e+00 5.06382287e-01 7.41181374e-01
-8.05402398e-01 -8.51526260e-01 -1.24702930e-01 8.05070698e-02
-1.85262930e+00 -1.33241987e+00 3.18870723e-01 2.54207104e-01
1.76079845e+00 8.26497853e-01 3.30601394e-01 1.23118627e+00
-3.89319927e-01 1.01042795e+00 3.32974672e-01 -5.27809113e-02
3.89960051e-01 2.53170639e-01 -1.81225073e-02 7.13824809e-01
-1.46680400e-01 -2.61279970e-01 -1.09366739e+00 -3.63062024e-01
5.17418683e-01 1.87789910e-02 -2.23430976e-01 7.33059287e-01
-9.94827211e-01 3.90816867e-01 -1.50777504e-01 2.14983210e-01
-5.28598428e-01 6.48227096e-01 4.60422903e-01 3.43513221e-01
6.83456957e-01 3.73485744e-01 -8.69618773e-01 -1.01693821e+00
-1.33734047e+00 7.75699317e-02 1.30420637e+00 1.45090604e+00
3.85895878e-01 6.20870233e-01 -2.56251216e-01 1.08166635e+00
2.28340283e-01 9.33169007e-01 5.43603778e-01 -9.37833309e-01
4.12353605e-01 1.40699372e-01 -6.54397607e-02 -4.49590772e-01
8.20012763e-02 -4.49619353e-01 -6.42816365e-01 -5.86523354e-01
-3.21168005e-02 -2.61365771e-01 -8.96988153e-01 1.09069717e+00
3.09284814e-02 4.69364733e-01 -1.20748319e-01 7.61895239e-01
9.86811996e-01 7.00607836e-01 -2.46438116e-01 -1.53725138e-02
1.41538429e+00 -1.11101818e+00 -9.81180310e-01 -2.29972854e-01
4.33432460e-01 -7.71746874e-01 1.46382403e+00 4.71118659e-01
-1.12834954e+00 -2.39156440e-01 -8.91424239e-01 -1.34535432e-01
-5.75328588e-01 3.09724867e-01 1.07059193e+00 9.60942090e-01
-1.46391118e+00 2.72231311e-01 -9.91929114e-01 -2.07664981e-01
3.20270032e-01 4.97514009e-01 3.01555432e-02 3.24291021e-01
-1.20905769e+00 4.30331230e-01 -2.79829770e-01 9.42475200e-02
-1.02176619e+00 -1.14173043e+00 -5.66119134e-01 3.98564160e-01
8.47348392e-01 -2.68105477e-01 2.06600142e+00 -3.25820297e-01
-2.14862323e+00 4.67160016e-01 -5.10098934e-01 -7.44878948e-01
2.13309929e-01 -4.15696204e-01 -1.13181603e+00 1.06932104e-01
-2.88007736e-01 1.32292360e-01 9.14681077e-01 -5.18071234e-01
-6.88648522e-01 -1.07982941e-01 6.08402956e-03 -1.59887344e-01
-6.38045073e-01 5.27469635e-01 -9.02651429e-01 -3.73035252e-01
-3.67090344e-01 -6.35549724e-01 5.45826852e-02 -3.88738024e-03
-5.46059728e-01 -6.11873388e-01 9.96131241e-01 -8.42147291e-01
1.52453268e+00 -2.12302971e+00 -8.25026751e-01 2.93615758e-02
3.36577386e-01 7.32558310e-01 7.05292448e-03 2.44192526e-01
3.29262584e-01 4.57881510e-01 3.65435660e-01 -5.49595177e-01
2.91924387e-01 3.29620302e-01 -7.29953408e-01 3.43647152e-02
7.29433596e-02 1.06918252e+00 -7.10762680e-01 -2.79000491e-01
1.70620933e-01 6.28598630e-01 -5.65059900e-01 4.16201353e-01
-3.49211067e-01 -3.55328113e-01 -3.39366376e-01 1.14993966e+00
1.15469247e-01 -3.40610951e-01 -1.16777971e-01 -1.00167282e-01
9.65121388e-02 1.06301022e+00 -9.08372581e-01 1.48473871e+00
-1.09447157e+00 8.87687743e-01 5.34114122e-01 -5.55761516e-01
7.36023366e-01 8.23909760e-01 2.06215575e-01 -7.06301570e-01
-9.22332108e-02 4.38209563e-01 -4.98041034e-01 -4.74451244e-01
8.44336212e-01 6.48360431e-01 2.06667215e-01 2.26064727e-01
2.35432982e-01 -1.63135737e-01 -3.01454574e-01 1.29270434e-01
1.31937313e+00 -6.93304062e-01 -2.30588987e-01 2.93127090e-01
2.50427574e-01 -4.02463853e-01 2.94491857e-01 1.20540643e+00
5.84149137e-02 3.48921299e-01 -8.00498426e-02 -1.84256822e-01
-7.36834884e-01 -8.47206533e-01 3.74432728e-02 1.36588252e+00
-4.26454306e-01 -7.45072603e-01 -7.56276190e-01 -4.22302246e-01
7.54276440e-02 8.84670138e-01 3.04173023e-01 3.29042003e-02
-4.05449599e-01 -3.80728364e-01 1.32974827e+00 6.80093646e-01
5.28674543e-01 -8.39694798e-01 -2.82464564e-01 1.63213521e-01
-2.41204016e-02 -1.83678508e+00 -7.81284094e-01 1.09450147e-01
-4.10482705e-01 -2.90833443e-01 -6.30505025e-01 -2.54932255e-01
-2.43700873e-02 3.20506155e-01 1.47747409e+00 8.31619203e-02
-3.14089060e-01 7.14684665e-01 -1.50651559e-01 -3.69602829e-01
-4.16132390e-01 4.66839701e-01 4.59864885e-01 -1.69807777e-01
4.97576714e-01 -6.37686431e-01 -4.55326140e-01 3.33644092e-01
-5.47837198e-01 -6.04986489e-01 5.24824739e-01 5.85322618e-01
4.67066467e-01 -1.13986820e-01 6.51297271e-01 -5.19080937e-01
8.54465783e-01 -3.94667506e-01 -7.04046369e-01 2.85655975e-01
-8.58976245e-01 -3.00129265e-01 6.06330752e-01 -6.95334494e-01
-5.81074655e-01 -2.02237040e-01 -8.01709592e-01 -9.09619570e-01
-1.30862445e-01 4.07492846e-01 -8.89417622e-03 -8.75902399e-02
3.58967990e-01 6.03023410e-01 -1.90806448e-01 -8.17414761e-01
3.60370219e-01 1.62445223e+00 6.45206571e-01 -3.09326053e-01
4.87231016e-01 1.02625094e-01 -7.57548749e-01 -1.49448848e+00
-4.47227269e-01 -6.71559572e-01 1.47582904e-01 1.89571306e-02
3.62348408e-01 -1.12767208e+00 -1.40577400e+00 4.40402895e-01
-1.16965127e+00 -4.83927280e-01 -2.23122150e-01 3.29294622e-01
-3.17249686e-01 4.26652953e-02 -7.44424939e-01 -1.23921132e+00
-1.01365793e+00 -1.03148067e+00 1.55668855e+00 7.32328072e-02
-1.85738474e-01 -4.78535861e-01 -6.77067578e-01 5.79147339e-01
1.13163531e+00 -9.49654043e-01 5.25654793e-01 -1.24084127e+00
-7.03622162e-01 -5.74346423e-01 -1.90750122e-01 5.30511379e-01
3.57461870e-02 -1.30397640e-02 -1.27908695e+00 -1.40760973e-01
-1.54602140e-01 -2.94061303e-01 4.61643755e-01 3.08051035e-02
1.84949660e+00 -7.30669081e-01 -1.94024235e-01 5.31899869e-01
8.50301147e-01 2.46090591e-01 3.13212723e-01 -3.40023249e-01
4.94886577e-01 -2.35586852e-01 2.98876852e-01 4.08770978e-01
2.02159286e-01 9.70991552e-01 1.44239351e-01 6.17794283e-02
-1.60653278e-01 -4.33629245e-01 3.79620880e-01 1.17118216e+00
3.03387374e-01 -5.52211285e-01 -9.11231101e-01 5.93926489e-01
-1.42177296e+00 -5.87515295e-01 3.92514288e-01 2.07269835e+00
1.03953874e+00 1.66953027e-01 6.26506731e-02 1.64698839e-01
4.91265327e-01 3.30654114e-01 -7.34030545e-01 -3.56136113e-01
2.16676623e-01 6.51795626e-01 6.88144803e-01 5.71243823e-01
-6.11053109e-01 1.24414754e+00 6.88714504e+00 1.14005768e+00
-1.43716609e+00 4.83199269e-01 4.61153507e-01 -6.03695631e-01
-1.72181189e-01 -5.84462523e-01 -1.09878051e+00 6.08801305e-01
1.78179145e+00 -1.76712155e-01 1.20042431e+00 1.48542023e+00
1.17697164e-01 3.11836451e-01 -9.23591077e-01 1.50307786e+00
-3.64806801e-02 -1.59562278e+00 -1.41637966e-01 -4.63331863e-02
-2.86105305e-01 5.96763432e-01 1.27268106e-01 6.28272414e-01
3.47765356e-01 -1.11441517e+00 5.71020424e-01 1.67458355e-01
1.37286651e+00 -7.78618038e-01 6.23837233e-01 2.94084340e-01
-1.06467760e+00 -1.09653184e-02 -1.43582135e-01 1.40126720e-01
8.73605907e-02 6.01998031e-01 -1.27651799e+00 4.22582291e-02
9.45360482e-01 2.21471880e-02 -2.58645236e-01 4.62195605e-01
1.06303222e-01 1.14432442e+00 -8.27381372e-01 -6.17665052e-01
5.93104400e-02 5.49949348e-01 5.18418312e-01 1.17837775e+00
3.79485369e-01 1.31947943e-03 -1.08868949e-01 7.64663458e-01
-5.79561710e-01 -1.62472174e-01 -2.95623094e-01 -6.06308579e-01
1.25538838e+00 1.33883572e+00 -1.83132231e-01 -6.08798981e-01
-3.48188370e-01 1.10449982e+00 -1.67815566e-01 4.77588028e-01
-8.64226341e-01 -5.52472293e-01 1.18354785e+00 1.92039207e-01
1.97807625e-01 -4.13930476e-01 2.86100298e-01 -1.10162997e+00
3.14140826e-01 -1.19552600e+00 -2.09826618e-01 -8.28162491e-01
-9.56334233e-01 6.56760037e-01 -5.67339599e-01 -5.20791411e-01
-5.54937363e-01 -5.19386649e-01 -3.17613602e-01 9.62504566e-01
-1.52073252e+00 -7.96521902e-01 -2.28026569e-01 6.60762727e-01
8.58608902e-01 -4.96600091e-01 1.19090474e+00 9.42946196e-01
-5.34414649e-01 1.15444112e+00 -9.16411132e-02 3.98646683e-01
2.30313271e-01 -1.10204577e+00 1.13038898e+00 4.96582270e-01
4.07949746e-01 9.02625859e-01 4.07251179e-01 -3.19259197e-01
-2.28920603e+00 -1.16388822e+00 1.02481961e+00 -3.77955973e-01
8.80203068e-01 -8.20321023e-01 -6.61977351e-01 5.65705597e-01
-1.78964641e-02 5.24663568e-01 6.90992653e-01 4.77252573e-01
-5.23358762e-01 -5.26803613e-01 -8.86520863e-01 6.55598879e-01
9.96725261e-01 -1.36265767e+00 7.76646435e-02 6.41369343e-01
1.56769812e+00 -5.78058064e-01 -9.44061577e-01 1.91742539e-01
5.29996395e-01 -3.63739103e-01 1.02668762e+00 -7.08881140e-01
-2.27360576e-01 -2.72615999e-02 -4.00129974e-01 -8.13634455e-01
3.63030076e-01 -1.34772420e+00 -7.31751740e-01 1.42060649e+00
6.89132273e-01 -8.56046975e-01 9.58470821e-01 7.71948636e-01
-1.71801686e-01 -8.75826895e-01 -1.31839859e+00 -8.56826842e-01
-7.53401577e-01 -1.17713130e+00 8.35286915e-01 5.26721120e-01
-3.06097686e-01 4.54843223e-01 -3.04774046e-01 3.88154298e-01
4.55839075e-02 -3.03338915e-01 6.91739678e-01 -7.11187482e-01
-4.78790194e-01 -1.25801027e-01 -1.56435177e-01 -1.44352376e+00
1.22175545e-01 -5.69805562e-01 7.87348300e-02 -9.84471440e-01
-6.07592106e-01 -3.86889160e-01 -1.89759135e-01 7.47796416e-01
3.98090154e-01 1.16323799e-01 1.17033850e-02 -5.92717081e-02
-8.56826246e-01 4.10721034e-01 4.96018916e-01 -2.38479242e-01
-1.69666111e-01 4.36219841e-01 -6.43413305e-01 4.44098592e-01
6.23280466e-01 -2.25366935e-01 -3.44452530e-01 -6.22313976e-01
3.05620790e-01 3.61007810e-01 2.52413571e-01 -9.36072409e-01
7.33719110e-01 2.50343382e-01 -7.55457729e-02 -6.16732478e-01
9.15659606e-01 -7.33024955e-01 -2.58024037e-01 1.18970452e-02
-3.97712022e-01 1.18005149e-01 1.88079208e-01 4.66117471e-01
-9.50573236e-02 2.21564025e-01 1.84659764e-01 4.41716798e-03
-4.98316854e-01 4.43711996e-01 -6.79667592e-01 1.31778374e-01
5.98454289e-02 2.76307493e-01 -3.22890371e-01 -9.32725132e-01
-4.22701895e-01 6.33104295e-02 -3.76368940e-01 8.50672722e-01
7.66595244e-01 -9.33657944e-01 -2.35043675e-01 2.95160770e-01
-1.00864932e-01 -1.20751120e-01 -5.52724451e-02 5.49982250e-01
-2.47745693e-01 9.29430127e-01 8.81663740e-01 -6.08972132e-01
-1.38807297e+00 2.59649158e-01 5.46831369e-01 2.09120855e-01
-2.96850324e-01 9.53882337e-01 -8.81227612e-01 -5.78506708e-01
8.70980203e-01 -5.92854142e-01 3.60540330e-01 -3.19331795e-01
8.89944255e-01 4.70564365e-01 7.46322453e-01 7.67946616e-03
-4.90554243e-01 -2.96332479e-01 -1.36918975e-02 -2.76403725e-01
1.16209257e+00 9.41796750e-02 4.73151840e-02 5.12731433e-01
1.19032061e+00 1.45767316e-01 -4.85024482e-01 -4.30172294e-01
-2.15449035e-02 -3.93031776e-01 6.38280749e-01 -1.00620914e+00
-1.16994464e+00 7.41750717e-01 5.79288602e-01 5.05802929e-01
8.98760200e-01 1.29787862e-01 1.38798261e+00 1.06847799e+00
2.63849735e-01 -1.19784844e+00 -2.04683289e-01 6.11988068e-01
6.62926376e-01 -1.02174270e+00 -5.36647081e-01 -3.71426255e-01
-1.65912524e-01 7.99068630e-01 3.37464839e-01 3.49413544e-01
8.04489672e-01 1.05488610e+00 2.89697703e-02 -9.70272124e-02
-1.26398802e+00 -2.25195184e-01 1.19266637e-01 2.77696341e-01
3.28273684e-01 1.22332070e-02 4.00463045e-01 8.50468636e-01
-4.31390792e-01 2.12358430e-01 -1.58322249e-02 5.36027670e-01
-6.59909695e-02 -5.50757945e-01 6.96882159e-02 7.18023717e-01
-9.04511452e-01 -6.90127373e-01 -3.69112521e-01 2.48520821e-01
-6.56324565e-01 1.33336139e+00 4.26397264e-01 -8.24929059e-01
3.96466076e-01 1.83049768e-01 -1.40743539e-01 -5.26681483e-01
-6.77105308e-01 -4.14447449e-02 5.87307870e-01 -1.06830812e+00
2.38526613e-01 -5.76379672e-02 -7.72042572e-01 -6.54219806e-01
-6.00009322e-01 1.73498958e-01 1.27170825e+00 7.64657557e-01
1.04559243e+00 7.34721899e-01 6.20636404e-01 -4.01785731e-01
-9.77031827e-01 -9.94465709e-01 -4.46854740e-01 -5.65743625e-01
4.62160826e-01 -3.73644046e-02 -4.57264841e-01 -4.49083686e-01] | [14.306556701660156, 6.275557041168213] |
6aaca035-3833-45dc-b740-38e73a6e90fd | learning-guarantees-for-graph-convolutional | null | null | https://openreview.net/forum?id=dpXL6lz4mOQ | https://openreview.net/pdf?id=dpXL6lz4mOQ | LEARNING GUARANTEES FOR GRAPH CONVOLUTIONAL NETWORKS ON THE STOCHASTIC BLOCK MODEL | An abundance of neural network models and algorithms for diverse tasks on graphs have been developed in the past five years. However, very few provable guarantees have been available for the performance of graph neural network models. This state of affairs is in contrast with the steady progress on the theoretical underpinnings of traditional dense and convolutional neural networks. In this paper we present the first provable guarantees for one of the best-studied families of graph neural network models, Graph Convolutional Networks (GCNs), for semi- supervised community detection tasks. We show that with high probability over the initialization and training data, a GCN will efficiently learn to detect communities on graphs drawn from a stochastic block model. Our proof relies on a fine-grained analysis of the training dynamics in order to overcome the complexity of a non-convex optimization landscape with many poorly-performing local minima. | ['Wei Lu'] | 2021-09-29 | null | null | null | iclr-2022-4 | ['stochastic-block-model'] | ['graphs'] | [ 3.63440245e-01 4.77865756e-01 -9.07390714e-02 -5.88489249e-02
-1.54780835e-01 -5.66036642e-01 5.55867016e-01 3.28547627e-01
-2.41134822e-01 6.40055776e-01 -6.68435320e-02 -6.11208022e-01
-5.07265031e-01 -8.37258637e-01 -9.81630564e-01 -7.76446998e-01
-8.77125323e-01 6.64345264e-01 -8.86000786e-03 2.62467097e-02
-1.13937728e-01 4.40770447e-01 -1.03401303e+00 1.07372813e-02
6.73866689e-01 5.46644688e-01 -9.06193554e-02 1.11871481e+00
6.41819835e-02 9.20949161e-01 -1.01646997e-01 -4.41621184e-01
2.45869055e-01 -4.55617726e-01 -8.21740925e-01 1.64018780e-01
5.73169589e-01 -4.15227860e-02 -8.08553159e-01 1.41803122e+00
2.07555920e-01 -6.74513653e-02 6.51363909e-01 -1.46075869e+00
-8.38769078e-01 8.79099071e-01 -6.23509228e-01 2.59076774e-01
-1.73871681e-01 5.38895978e-03 1.30824113e+00 -4.60952222e-01
7.80599236e-01 1.08389199e+00 1.11805928e+00 3.97521257e-01
-1.32320487e+00 -5.37617087e-01 3.31814885e-01 -1.20966263e-01
-1.28829217e+00 -1.06368631e-01 6.78241193e-01 -5.91887653e-01
9.08661127e-01 -1.17074423e-01 9.67687607e-01 7.95406282e-01
-1.89588293e-01 7.19530165e-01 8.09089243e-01 -4.51995850e-01
1.05794936e-01 -3.18840206e-01 4.00046468e-01 1.14994788e+00
9.10429060e-01 4.81165089e-02 -3.82821888e-01 -2.63733715e-01
1.07058465e+00 -2.24456228e-02 -2.85114259e-01 -1.01113760e+00
-7.65529931e-01 1.01464450e+00 8.35453093e-01 3.39259714e-01
-7.39557445e-02 7.22079694e-01 4.31414604e-01 4.04069066e-01
5.34445882e-01 7.94643089e-02 -1.67305306e-01 1.92413315e-01
-8.29229832e-01 1.06749319e-01 1.14017415e+00 8.42852831e-01
7.65627146e-01 1.30169749e-01 2.60178745e-01 1.53774753e-01
3.86184871e-01 2.22317249e-01 -2.82768101e-01 -5.71760118e-01
5.04401028e-01 6.60292327e-01 -1.68526992e-01 -1.28312039e+00
-3.91441554e-01 -7.74355471e-01 -1.41433239e+00 8.69928375e-02
8.09187353e-01 -2.59682864e-01 -9.18515444e-01 1.61155570e+00
1.39886647e-01 3.15164685e-01 -1.75754324e-01 6.02983713e-01
5.27867794e-01 4.18698519e-01 -2.84349829e-01 -5.27492631e-03
6.78209007e-01 -8.08367789e-01 -3.69669229e-01 -2.06587777e-01
6.74234509e-01 -7.86054879e-02 5.96198082e-01 2.72669554e-01
-1.09610248e+00 -1.06074162e-01 -1.07754028e+00 4.67097759e-02
-3.19000304e-01 -2.25111634e-01 1.14523542e+00 7.78186262e-01
-1.47497153e+00 8.65595818e-01 -1.09706247e+00 -4.83485788e-01
8.36066723e-01 6.63343370e-01 -5.08484066e-01 -2.51356304e-01
-7.63897657e-01 4.93196994e-01 4.71362740e-01 5.37993133e-01
-1.19838881e+00 -3.31609994e-01 -7.90500641e-01 2.12100834e-01
4.44805175e-01 -5.81744671e-01 8.21967840e-01 -1.28808916e+00
-8.30453575e-01 1.00074196e+00 2.36792609e-01 -9.13016319e-01
5.01963019e-01 1.48304254e-01 1.33295506e-01 2.73368955e-01
-2.43269637e-01 3.75369340e-01 6.23419523e-01 -1.04282165e+00
-3.78949910e-01 -2.70409882e-01 1.21412285e-01 -9.84161943e-02
-4.31762040e-01 3.89414169e-02 -3.18706483e-01 -3.12917590e-01
-2.32383944e-02 -9.77287591e-01 -6.05985940e-01 1.27388388e-01
-5.44094801e-01 -2.25524619e-01 3.69983047e-01 -3.25654566e-01
1.02983141e+00 -1.85487604e+00 3.06380570e-01 5.24407089e-01
8.63390684e-01 2.24799559e-01 -1.83494344e-01 6.19278312e-01
-1.57336190e-01 3.70775878e-01 -3.62508506e-01 -4.17856216e-01
1.34394869e-01 2.74904013e-01 -1.45041049e-01 9.17327881e-01
2.31202841e-01 1.12482965e+00 -1.05338144e+00 -1.76248759e-01
-2.10231747e-02 5.18275559e-01 -6.48891032e-01 -8.63712281e-02
-2.89294153e-01 -7.59672746e-02 -2.64065176e-01 2.42226899e-01
5.15840352e-01 -1.03008127e+00 7.61233568e-01 4.69285280e-01
3.24978948e-01 1.31848261e-01 -1.17818606e+00 1.51964056e+00
1.43738315e-01 9.46899593e-01 6.55845404e-01 -1.40692282e+00
5.32393873e-01 2.81764835e-01 4.17310953e-01 -1.12741239e-01
2.30717018e-01 1.76201582e-01 1.64797813e-01 -1.03335418e-01
1.79822713e-01 5.89488931e-02 2.57015973e-01 6.69957221e-01
1.66142076e-01 3.50257218e-01 3.36394548e-01 5.86758912e-01
1.31983483e+00 -9.27711204e-02 5.57500198e-02 -5.66180468e-01
1.10043563e-01 5.33376150e-02 1.69956043e-01 1.27289009e+00
-2.30489075e-01 4.86626565e-01 8.99464309e-01 -7.18132138e-01
-1.26304388e+00 -8.07420135e-01 2.72817075e-01 9.09850717e-01
-1.95875928e-01 -4.16271359e-01 -9.64411557e-01 -4.24968570e-01
-3.87910893e-03 -3.12425464e-01 -9.48854744e-01 8.90635327e-03
-5.85050285e-01 -8.49991083e-01 6.18020356e-01 5.53495526e-01
1.96267009e-01 -9.50060129e-01 -2.09744155e-01 3.19308817e-01
3.49917829e-01 -1.13913608e+00 -2.19422266e-01 5.45539439e-01
-9.92626190e-01 -1.51174128e+00 -6.04706883e-01 -1.13823152e+00
8.60480309e-01 4.14882690e-01 1.44053447e+00 6.34405077e-01
-1.91676542e-01 3.08476567e-01 3.93284000e-02 -1.93351805e-01
-4.40968037e-01 3.92420202e-01 -1.77997380e-01 -8.83351341e-02
3.27493072e-01 -8.88111711e-01 -4.20998186e-01 -3.96322250e-01
-7.39684999e-01 1.25180840e-01 5.38578689e-01 8.38713527e-01
2.51427770e-01 1.52526245e-01 2.66192794e-01 -1.03530657e+00
6.99793100e-01 -5.77865541e-01 -9.35276806e-01 1.95845962e-01
-7.15457737e-01 2.27359086e-01 6.04007363e-01 -3.55922759e-01
-1.75040171e-01 8.63718614e-02 2.71602482e-01 -5.26478171e-01
-2.76487172e-02 8.76500309e-01 2.86172181e-01 -3.30681652e-01
6.56636298e-01 1.21024668e-01 2.46601850e-01 -2.75483161e-01
4.86242175e-01 1.81588948e-01 6.20727122e-01 -6.61108971e-01
1.10241210e+00 7.31387556e-01 2.56124675e-01 -7.52569318e-01
-7.91257560e-01 -4.10473287e-01 -8.50553453e-01 -2.68263638e-01
7.74624944e-01 -8.86787891e-01 -7.98367620e-01 4.73798633e-01
-1.09573698e+00 -8.21526587e-01 -5.71332052e-02 1.35341495e-01
-5.40130973e-01 5.50562978e-01 -8.14670801e-01 -9.98377383e-01
-2.93370008e-01 -6.62254214e-01 6.32936180e-01 7.64565542e-02
3.54915529e-01 -1.47709155e+00 2.25933015e-01 -2.39094030e-02
3.57524335e-01 6.12838209e-01 9.27502871e-01 -6.20323777e-01
-1.00939429e+00 -3.52057159e-01 -5.77761769e-01 2.89630741e-01
-2.14481726e-01 2.05465853e-01 -7.83195376e-01 -7.12929010e-01
-5.24353027e-01 -3.76517057e-01 1.15711761e+00 6.78492546e-01
9.39666927e-01 -4.36911613e-01 -5.27043402e-01 6.87851310e-01
1.64219129e+00 -4.00721371e-01 4.20737833e-01 7.95060545e-02
1.08880162e+00 3.54235470e-01 -3.60308379e-01 -5.36907762e-02
2.17568606e-01 -4.07417417e-02 7.12722421e-01 -3.38679880e-01
6.42284304e-02 -3.12085927e-01 1.34555832e-01 7.49038875e-01
-2.72453308e-01 -4.66754287e-01 -1.19004393e+00 8.50196958e-01
-2.12151456e+00 -9.50997651e-01 -3.62926066e-01 2.02474475e+00
5.67504287e-01 4.38927978e-01 4.01111901e-01 6.23839833e-02
1.00033116e+00 2.91809708e-01 -3.92323136e-01 -1.63560420e-01
-2.09200978e-01 4.56885308e-01 7.49066651e-01 6.15582168e-01
-1.24846613e+00 8.32643747e-01 7.37463570e+00 4.72745121e-01
-7.07035065e-01 -6.88175038e-02 5.61732531e-01 1.06238261e-01
-2.07765043e-01 2.51253784e-01 -4.17089462e-01 -8.64175111e-02
9.52010155e-01 -1.67559788e-01 7.89355218e-01 9.02092755e-01
-7.91866407e-02 3.82575512e-01 -1.12732613e+00 7.64601707e-01
-4.20810096e-02 -1.83077121e+00 -3.30231428e-01 5.03819108e-01
1.04329467e+00 6.60992742e-01 -1.01838075e-01 7.24496469e-02
1.20336473e+00 -1.47075236e+00 4.53300089e-01 1.01349287e-01
6.46799803e-01 -7.54097402e-01 4.55109358e-01 3.71652871e-01
-1.28287148e+00 -1.00850649e-01 -4.50150937e-01 -4.21646655e-01
-2.73397356e-01 6.70444369e-01 -8.54793668e-01 4.57395077e-01
4.44249064e-01 8.67170393e-01 -5.12034178e-01 1.05507445e+00
-1.29197672e-01 9.05976772e-01 -5.40459454e-01 -2.08091289e-01
6.14114106e-01 -3.31326783e-01 4.35858756e-01 1.30529130e+00
-3.05991143e-01 -3.87300879e-01 3.40373576e-01 9.20986652e-01
-4.36808646e-01 -1.97424412e-01 -8.79786313e-01 -6.08463466e-01
1.27418026e-01 1.17701364e+00 -1.12984467e+00 -1.52204812e-01
-4.12098080e-01 6.15750074e-01 9.67401028e-01 4.47600007e-01
-6.03022397e-01 -1.88809276e-01 3.73074710e-01 1.64876103e-01
5.70953965e-01 -5.89056790e-01 -1.64827079e-01 -1.14140034e+00
2.04117950e-02 -8.38060677e-01 6.35779440e-01 -4.21145767e-01
-1.46055293e+00 4.42048877e-01 -2.19285175e-01 -4.21882838e-01
-1.09497458e-01 -8.54657233e-01 -8.44305336e-01 6.60960972e-01
-1.46612835e+00 -1.19841635e+00 -1.76432982e-01 5.89280784e-01
-1.14965171e-01 1.07802749e-01 7.30244696e-01 1.47217259e-01
-5.77637613e-01 3.24306875e-01 3.64709944e-01 8.42652619e-01
2.56669596e-02 -1.70167887e+00 8.10217559e-01 1.37152755e+00
4.17222142e-01 5.55809975e-01 6.34368420e-01 -6.47163093e-01
-1.47752333e+00 -1.06538415e+00 7.99085855e-01 -3.01168144e-01
1.13977587e+00 -7.87664950e-01 -8.47197413e-01 1.01812387e+00
2.03413069e-01 3.12921822e-01 2.08390191e-01 6.67421579e-01
-6.06275320e-01 2.46259898e-01 -5.76649547e-01 3.78751636e-01
1.30172074e+00 -7.93769956e-01 -6.03734665e-02 7.34126687e-01
5.51287770e-01 -2.05974340e-01 -6.28064990e-01 1.09415784e-01
4.31093723e-01 -8.60399127e-01 8.75200987e-01 -1.21206927e+00
3.74878645e-01 -1.89677417e-01 2.15225276e-02 -9.09366846e-01
-4.58668798e-01 -1.03523386e+00 -4.32759821e-01 6.25792146e-01
2.95269907e-01 -5.16429365e-01 1.31871319e+00 2.68074661e-01
2.22480427e-02 -7.19754696e-01 -7.85056055e-01 -8.23420942e-01
4.02262062e-01 -1.94307491e-01 1.17809683e-01 9.62882519e-01
-1.79051876e-01 3.64549905e-01 -2.96044409e-01 3.04606318e-01
9.73319173e-01 1.73064351e-01 7.76808202e-01 -1.59047830e+00
-4.15654004e-01 -7.68720925e-01 -6.07201576e-01 -9.43736613e-01
2.06565574e-01 -1.19217086e+00 -8.59882012e-02 -1.86584830e+00
6.08595729e-01 -4.06919301e-01 -2.75427192e-01 3.73122185e-01
-4.11333367e-02 2.85403907e-01 -2.40001176e-02 7.71641508e-02
-8.81457448e-01 7.14579374e-02 9.87671077e-01 -2.64498204e-01
3.28954011e-02 -2.98090745e-02 -7.44917691e-01 6.67492747e-01
7.12813258e-01 -6.06025279e-01 -3.87790442e-01 -5.82196891e-01
8.24462295e-01 -1.70704618e-01 7.16376483e-01 -8.95028532e-01
4.98014688e-01 4.49197367e-02 1.71562642e-01 -5.42988658e-01
-2.04781279e-01 -4.78492081e-01 1.53297618e-01 6.99869394e-01
-4.63097513e-01 1.65991619e-01 -1.27222925e-01 1.11923909e+00
1.03242524e-01 -7.99901783e-02 7.09725857e-01 -3.47899646e-01
-2.22075924e-01 7.90391326e-01 -2.19673112e-01 3.38982940e-01
6.83051825e-01 -2.82460451e-01 -3.40873450e-01 -4.79376554e-01
-8.02746534e-01 2.97220737e-01 4.89804715e-01 3.41361165e-02
1.87230825e-01 -9.15209472e-01 -9.33437526e-01 -1.26509920e-01
-2.42918119e-01 3.72092724e-01 6.58760453e-03 7.35695004e-01
-8.21992755e-01 3.26993436e-01 -4.17805389e-02 -6.29905820e-01
-1.04487371e+00 8.00295413e-01 7.69868553e-01 -7.04073727e-01
-8.03170979e-01 9.67609227e-01 7.18703791e-02 -3.75264466e-01
5.27743578e-01 -2.99922407e-01 2.18418434e-01 -2.83300668e-01
3.02641898e-01 6.66360557e-02 -5.54058887e-03 -3.02914739e-01
-1.76850662e-01 5.52040488e-02 -1.26503780e-01 1.69391811e-01
1.63309741e+00 2.46060655e-01 -2.65014201e-01 2.16985509e-01
1.06821561e+00 -3.11586738e-01 -1.34044898e+00 -2.62834370e-01
2.07662597e-01 9.17563885e-02 5.38087152e-02 -2.97051609e-01
-1.04482365e+00 1.08005452e+00 2.28706762e-01 7.91400909e-01
7.33238935e-01 8.50935373e-03 3.48218232e-01 6.96336746e-01
2.33183622e-01 -9.34744954e-01 6.29234537e-02 6.49306297e-01
3.90017152e-01 -1.01254261e+00 3.62889916e-02 -4.24676299e-01
5.96399792e-03 1.22233784e+00 1.96283236e-01 -7.19011843e-01
8.86152029e-01 4.07905012e-01 -4.86213297e-01 -7.29235768e-01
-8.65713775e-01 -3.05361241e-01 2.80551240e-02 9.33928728e-01
2.25812793e-01 8.88556987e-02 3.51265341e-01 1.19020492e-01
2.94976365e-02 -1.47729397e-01 6.09720588e-01 6.38727427e-01
-6.53637230e-01 -7.07754433e-01 1.61824301e-01 4.46850419e-01
-5.23754060e-01 -3.83075595e-01 -7.29828656e-01 9.18434083e-01
-3.20468277e-01 7.05825448e-01 1.52274042e-01 -6.27696216e-02
-3.05005699e-01 -1.21243998e-01 7.38007426e-01 -6.75556898e-01
-5.32241583e-01 -2.46359110e-01 1.18045874e-01 -2.95368820e-01
-6.32648230e-01 -4.89456177e-01 -1.00587523e+00 -7.53615260e-01
-6.64094031e-01 3.67797203e-02 4.09065634e-01 9.02871728e-01
1.80404261e-01 3.56818616e-01 3.76299918e-01 -7.20456839e-01
-5.38012147e-01 -7.43796289e-01 -6.23401523e-01 1.61686406e-01
5.15618801e-01 -2.92002648e-01 -6.10346735e-01 -9.84338671e-02] | [6.866573333740234, 6.046314716339111] |
432262ad-993b-47e5-9cd6-bc4889da86b8 | simplex-pb-2-0-a-reliable-dataset-for-lexical | null | null | https://aclanthology.org/2020.winlp-1.6 | https://aclanthology.org/2020.winlp-1.6.pdf | SIMPLEX-PB 2.0: A Reliable Dataset for Lexical Simplification in Brazilian Portuguese | Most research on Lexical Simplification (LS) addresses non-native speakers of English, since they are numerous and easy to recruit. This makes it difficult to create LS solutions for other languages and target audiences. This paper presents SIMPLEX-PB 2.0, a dataset for LS in Brazilian Portuguese that, unlike its predecessor SIMPLEX-PB, accurately captures the needs of Brazilian underprivileged children. To create SIMPLEX-PB 2.0, we addressed all limitations of the old SIMPLEX-PB through multiple rounds of manual annotation. As a result, SIMPLEX-PB 2.0 features much more reliable and numerous candidate substitutions to complex words, as well as word complexity rankings produced by a group underprivileged children. | ['ra', "S Alu{\\'\\i}sio", 'Nathan Hartmann', 'Gustavo Henrique Paetzold'] | 2020-07-01 | null | null | null | ws-2020-7 | ['lexical-simplification'] | ['natural-language-processing'] | [-3.92185777e-01 3.95328790e-01 -4.69535500e-01 -1.91055700e-01
-8.75134587e-01 -4.97769713e-01 1.89528629e-01 5.91620684e-01
-7.16646612e-01 1.16785336e+00 5.80283046e-01 -4.45685774e-01
1.29925892e-01 -7.22985804e-01 -4.77377892e-01 1.53501436e-01
3.23686481e-01 8.53058338e-01 1.38769239e-01 -6.53273404e-01
4.05764580e-02 4.73883212e-01 -1.76847124e+00 8.45739394e-02
1.43162978e+00 -2.61990756e-01 4.22219694e-01 2.94649959e-01
-2.86338270e-01 4.79107171e-01 -7.97773480e-01 -1.04875255e+00
-3.65755372e-02 1.35792568e-01 -6.83098793e-01 -7.14340210e-01
9.43446517e-01 -3.71475279e-01 1.77453812e-02 8.71687055e-01
8.57505441e-01 9.44692940e-02 5.41790187e-01 -9.44734871e-01
-8.06451321e-01 1.22281945e+00 -3.07124764e-01 5.10301590e-01
9.53047037e-01 -6.49280921e-02 1.15021300e+00 -1.21113336e+00
9.60612595e-01 1.84434116e+00 9.65275407e-01 6.91005945e-01
-1.34850943e+00 -1.04268742e+00 1.41033545e-01 -9.82586145e-02
-1.63640893e+00 -7.38882184e-01 4.81466532e-01 -3.00834894e-01
1.60278189e+00 4.47737038e-01 1.15334320e+00 1.04769814e+00
-3.13446909e-01 7.94975996e-01 9.30342972e-01 -7.12950826e-01
-3.27785224e-01 1.74229994e-01 5.59737124e-02 5.05010605e-01
7.94281125e-01 -2.26654261e-01 -8.89328122e-01 -1.56343520e-01
2.51304150e-01 -6.04706705e-01 1.92134024e-03 3.19943912e-02
-8.75139117e-01 7.72797346e-01 -2.83084154e-01 2.19692230e-01
5.57255223e-02 -1.92418769e-01 2.70838380e-01 3.00301909e-01
4.21537429e-01 7.10028589e-01 -6.38620973e-01 -6.13899648e-01
-9.91196096e-01 7.11557806e-01 1.06509364e+00 1.33882558e+00
4.69067931e-01 1.69088960e-01 4.81586903e-03 1.33213389e+00
1.09585516e-01 7.72367001e-01 5.24101138e-01 -8.81566107e-01
8.65764022e-01 3.38346779e-01 -6.80063367e-02 -7.95501053e-01
-4.89400893e-01 -3.53413731e-01 -3.54175091e-01 -1.82104960e-01
4.03571457e-01 8.41853097e-02 -5.51160991e-01 2.06972837e+00
3.45566124e-01 -6.46481991e-01 -2.36484289e-01 -1.72773506e-02
1.06556928e+00 3.48245710e-01 5.61705410e-01 -2.14642331e-01
1.12901580e+00 -8.41351211e-01 -8.94485176e-01 -3.83625984e-01
7.03856766e-01 -1.16750467e+00 1.68722749e+00 5.14040470e-01
-1.61296189e+00 -3.05927485e-01 -8.67259860e-01 -5.97044826e-01
-7.12744772e-01 -2.30181575e-01 7.52224982e-01 1.32563400e+00
-1.08672619e+00 4.78488296e-01 -7.85933077e-01 -4.71348494e-01
4.93775666e-01 3.96294296e-01 -4.06296641e-01 -6.69387653e-02
-1.20945644e+00 1.51742506e+00 3.57185155e-01 -6.57014310e-01
-4.29085612e-01 -1.10231376e+00 -1.12707436e+00 9.34747793e-03
1.66301727e-01 -4.94035900e-01 1.59428799e+00 -6.27767265e-01
-1.27060068e+00 1.21533096e+00 -2.39277124e-01 9.84456241e-02
4.17561293e-01 -5.16013384e-01 -4.92309630e-01 -2.76580691e-01
6.95052147e-01 6.92790806e-01 3.63480777e-01 -6.92051291e-01
-8.26578856e-01 -1.78768739e-01 -7.40959495e-02 5.22510409e-01
-4.17999715e-01 5.66553116e-01 -4.32912707e-01 -1.29251719e+00
-1.30498469e-01 -5.00985980e-01 -7.33426660e-02 -3.02617013e-01
-1.31685928e-01 -5.62437057e-01 4.94247451e-02 -1.03182423e+00
2.01447916e+00 -1.95328033e+00 5.81217855e-02 8.90419856e-02
3.82811546e-01 6.47260249e-01 -9.07385871e-02 5.87704301e-01
-1.24261566e-01 7.80485570e-01 5.96108772e-02 -6.19135022e-01
1.08247235e-01 3.36663902e-01 1.80826053e-01 2.79213071e-01
1.34952605e-01 8.47218633e-01 -1.16099727e+00 -6.34814560e-01
-2.35913396e-02 1.64258152e-01 -1.05266142e+00 -2.63913780e-01
1.08167075e-01 2.69809812e-02 5.46506792e-02 9.96647894e-01
6.89849675e-01 7.37874150e-01 1.10309146e-01 5.01697600e-01
-7.08194613e-01 9.34763789e-01 -1.13100767e+00 1.41828859e+00
-5.21106482e-01 6.45098150e-01 2.46186793e-01 -3.28405976e-01
5.31500578e-01 -2.30170111e-03 1.88518420e-01 -6.79537535e-01
5.55663742e-03 8.39409590e-01 -9.83728562e-03 -4.23205107e-01
8.23296249e-01 9.28332359e-02 -3.46398115e-01 2.10096627e-01
2.55145244e-02 -7.37715602e-01 7.59730041e-01 1.15713067e-01
7.77433038e-01 -1.89792633e-01 8.48670304e-01 -7.08709896e-01
3.98724347e-01 -6.41841218e-02 9.47448432e-01 8.13282490e-01
-1.65533751e-01 4.91506606e-01 1.37848184e-01 -2.76297539e-01
-9.94859159e-01 -1.45367301e+00 -2.21220776e-01 1.38010025e+00
-4.14066732e-01 -8.28642666e-01 -7.87269652e-01 -3.78593594e-01
1.18331671e-01 1.09729195e+00 -1.39374048e-01 9.60879400e-02
-1.04204297e+00 -3.17460418e-01 9.52488601e-01 6.08380064e-02
-2.03174964e-01 -1.03762186e+00 -3.60578090e-01 5.28636992e-01
-2.45739609e-01 -7.36535311e-01 -4.31759655e-01 -2.53869176e-01
-3.42632145e-01 -9.19885635e-01 -4.75983560e-01 -1.16476905e+00
2.43822724e-01 5.81366904e-02 1.66646063e+00 2.11036026e-01
-1.59652784e-01 1.77622288e-01 -2.78181344e-01 -1.19546986e+00
-4.55376297e-01 5.91810226e-01 2.11302757e-01 -1.12030184e+00
6.28916919e-01 -7.25256979e-01 1.95859462e-01 -4.62538719e-01
-5.31550467e-01 -3.71241122e-02 3.07254463e-01 6.45583153e-01
2.46408328e-01 -1.96368158e-01 4.58454281e-01 -1.25454247e+00
8.26640308e-01 -4.69444215e-01 -3.57665062e-01 3.29342216e-01
-5.26534319e-01 -4.20137942e-01 3.43871802e-01 -6.66920304e-01
-9.93954778e-01 -2.58194715e-01 -6.19318366e-01 3.70135695e-01
-2.02191267e-02 4.86713558e-01 -4.49679345e-01 4.67565283e-02
7.09120929e-01 -3.44442040e-01 -3.73132288e-01 -8.72860909e-01
4.41729724e-01 6.75597310e-01 6.35631382e-01 -8.80185068e-01
7.92986512e-01 -1.31857201e-01 -4.33130473e-01 -1.31143773e+00
-6.43754184e-01 -1.42036274e-01 -5.40196061e-01 2.06854358e-01
4.51336890e-01 -1.09139001e+00 -3.18098962e-01 6.01251185e-01
-1.30647743e+00 -1.90521672e-01 -5.03270388e-01 5.02008915e-01
-1.34154670e-02 2.54639864e-01 -4.45627451e-01 -5.00784874e-01
-4.34163898e-01 -1.14133608e+00 7.99159825e-01 1.96202904e-01
-9.90170598e-01 -8.92463982e-01 2.26568565e-01 3.23066980e-01
3.99119198e-01 3.75459827e-02 1.26624644e+00 -7.65640914e-01
4.87835258e-02 1.81557983e-01 1.03482910e-01 2.97541499e-01
-3.36447805e-02 4.43598390e-01 -4.80723083e-01 -2.32913807e-01
-3.10534388e-01 -2.81055361e-01 2.16882169e-01 2.39231810e-01
8.10894489e-01 -4.44373310e-01 -5.51608764e-02 7.17978537e-01
9.28487897e-01 -7.82530680e-02 1.97620556e-01 3.83425921e-01
7.83159673e-01 6.90971613e-01 4.81891900e-01 2.91276455e-01
1.09956443e+00 5.11791050e-01 -1.50114328e-01 -1.92123633e-02
-5.05586207e-01 -6.78858161e-01 4.38048691e-01 1.61400223e+00
2.58365363e-01 -1.02642857e-01 -1.41981089e+00 1.05783176e+00
-1.14212990e+00 -6.40374482e-01 -3.34097713e-01 2.13964844e+00
1.45844483e+00 1.17481627e-01 6.96589500e-02 1.92887247e-01
6.71254218e-01 -6.50827438e-02 -4.95291688e-02 -9.35580969e-01
-5.18188357e-01 1.03233695e+00 2.69199222e-01 8.54631424e-01
-7.39769816e-01 1.55030191e+00 7.19732428e+00 1.33545578e+00
-8.12259078e-01 4.49622989e-01 2.53680348e-01 -4.02920514e-01
-7.81105638e-01 -4.84269597e-02 -1.10833216e+00 3.65797222e-01
5.78296125e-01 -5.17477512e-01 7.74478972e-01 5.81928968e-01
1.02740772e-01 -1.51326120e-01 -1.10477185e+00 9.49543655e-01
4.96546105e-02 -1.06416154e+00 1.78054318e-01 -3.27321410e-01
9.09930587e-01 1.87498987e-01 6.52071014e-02 5.97137928e-01
4.99478489e-01 -1.21883368e+00 1.25518215e+00 1.91552415e-01
1.12287140e+00 -8.41633737e-01 2.49837503e-01 5.59945405e-01
-1.01028228e+00 -2.66735423e-02 -2.52015799e-01 -7.10242271e-01
1.31984159e-01 5.52298248e-01 -6.77164495e-01 3.49782268e-03
7.24339247e-01 5.37506044e-01 -1.02350163e+00 8.83328676e-01
-6.58520877e-01 6.47839367e-01 -7.14223087e-01 -2.38093600e-01
-1.44617453e-01 2.64908075e-02 7.45312512e-01 1.53489149e+00
4.44202274e-01 -2.85466411e-03 2.29749650e-01 2.78761208e-01
-1.58068284e-01 7.93318748e-01 -7.22481549e-01 1.40809063e-02
1.42477751e+00 6.98129416e-01 -3.48922104e-01 -3.36056948e-01
-6.05799019e-01 3.86492878e-01 1.09164643e+00 -1.57069834e-03
-3.55132729e-01 -4.69104975e-01 9.23337042e-01 2.25058094e-01
-1.26350984e-01 -3.01832527e-01 -6.72067285e-01 -1.16851282e+00
-8.52131024e-02 -1.47242200e+00 4.04692978e-01 -4.45659578e-01
-1.11049139e+00 1.21595576e-01 3.31813037e-01 -4.56492424e-01
-9.82547551e-02 -3.57462347e-01 -3.63648236e-01 9.44503784e-01
-1.23102450e+00 -1.34080970e+00 2.29351774e-01 3.55131447e-01
6.77958369e-01 -2.38381382e-02 9.93879139e-01 7.53358483e-01
-8.50392580e-01 1.15003407e+00 -1.62203804e-01 -3.38584781e-01
7.50477374e-01 -1.24465477e+00 7.58617818e-01 1.04764676e+00
-4.32692561e-03 9.31352794e-01 7.88292766e-01 -9.29284275e-01
-7.90473700e-01 -8.49471152e-01 1.89926863e+00 -6.29199922e-01
6.78257883e-01 -5.04536271e-01 -6.41538858e-01 7.12652385e-01
2.47678995e-01 -7.27764428e-01 6.67125642e-01 1.60909593e-01
-1.44006103e-01 -1.56550054e-02 -1.26728439e+00 1.01186311e+00
1.58121204e+00 -6.11079097e-01 -1.04001069e+00 2.99642086e-01
8.61870050e-01 -7.04517126e-01 -8.92361164e-01 2.40493774e-01
4.97526497e-01 -5.65978050e-01 1.08709216e+00 -5.83475471e-01
1.51865274e-01 2.40686342e-01 8.20430890e-02 -1.40474963e+00
-3.49350929e-01 -1.14979076e+00 1.65111095e-01 1.65363634e+00
6.78061485e-01 -6.99565291e-01 2.80807763e-01 5.91983974e-01
-5.48001766e-01 -5.53161323e-01 -1.32227612e+00 -5.07737994e-01
7.13474929e-01 -6.12993002e-01 1.11336267e+00 1.23670185e+00
-1.49163827e-01 8.30117762e-02 4.55243438e-02 -3.40673923e-02
3.09756130e-01 -6.63339794e-01 7.91274607e-01 -1.26529717e+00
4.67450581e-02 -8.27612400e-01 -5.54725118e-02 -5.72522163e-01
4.30495888e-01 -1.04703033e+00 -3.36355865e-01 -1.26341629e+00
-2.42738903e-01 -6.70368254e-01 3.41193795e-01 6.94150746e-01
-5.04186213e-01 2.51081586e-01 3.11292470e-01 -2.25497574e-01
-5.51227294e-02 3.52118760e-01 9.87757206e-01 1.22327074e-01
-5.00728786e-01 -3.78179252e-02 -9.81159151e-01 1.28203070e+00
1.00668204e+00 -8.02732527e-01 -1.52606159e-01 -1.10275567e+00
8.16199422e-01 -5.07316053e-01 -4.06381279e-01 -7.57220030e-01
8.75600055e-03 -4.59892154e-01 -4.76494245e-02 -6.83026195e-01
6.02794029e-02 -3.81870985e-01 2.76170135e-01 3.55139971e-01
1.79969948e-02 6.73673153e-01 3.21096003e-01 -4.62799907e-01
1.19300000e-03 -7.21329570e-01 5.06226718e-01 -2.10551098e-01
-4.21667904e-01 4.18892168e-02 -8.36041570e-01 8.24573398e-01
7.91564047e-01 -1.55470744e-01 -3.75389546e-01 -5.48663288e-02
-3.25949997e-01 3.25145721e-01 6.87850058e-01 3.30515265e-01
3.88955534e-01 -1.19702935e+00 -8.34929466e-01 5.24148107e-01
1.02449939e-01 2.05305770e-01 -3.70055914e-01 5.12243390e-01
-1.20902312e+00 2.39411339e-01 -2.04508722e-01 1.53465241e-01
-1.33356524e+00 1.86989367e-01 -2.34657466e-01 -1.67273283e-01
-2.62089938e-01 1.20442951e+00 -4.07312185e-01 -9.21637952e-01
2.91219682e-01 -4.79028612e-01 -3.76254767e-01 4.14055377e-01
3.81976396e-01 9.67707396e-01 9.36034992e-02 -9.39038217e-01
-2.48626232e-01 2.45361120e-01 -2.01103874e-02 -2.60296673e-01
1.36159039e+00 -2.06070364e-01 -6.80219471e-01 1.97683185e-01
8.13671470e-01 1.21750295e+00 -1.24943540e-01 -2.30384525e-03
3.01572829e-01 -5.95009863e-01 -3.46789122e-01 -6.04645431e-01
-6.44032598e-01 5.96546948e-01 -1.50226146e-01 -1.95617720e-01
9.88446712e-01 -9.83497798e-02 7.78497696e-01 3.62651408e-01
2.41895497e-01 -1.37209916e+00 -5.93663990e-01 1.00030386e+00
8.51094604e-01 -6.95733666e-01 3.69421512e-01 -8.35832000e-01
-2.31574833e-01 6.55362070e-01 7.56666601e-01 4.09963697e-01
5.85646093e-01 1.67510748e-01 1.04953103e-01 -9.40346718e-02
-5.73889554e-01 -1.46315381e-01 1.30166024e-01 9.36011195e-01
6.33179247e-01 5.17573535e-01 -1.07932127e+00 1.07395625e+00
-1.35122418e+00 -4.49710160e-01 6.10916018e-01 1.03505492e+00
-1.91511080e-01 -1.64006269e+00 -4.35540229e-01 4.59584683e-01
-8.05586874e-01 -6.38701260e-01 -5.05548596e-01 1.13406754e+00
6.38449192e-01 9.25541580e-01 -1.62646562e-01 7.51193464e-02
7.06549346e-01 -1.28045585e-02 5.57348490e-01 -1.30503464e+00
-1.06275427e+00 -2.04357639e-01 7.57845223e-01 -4.29990500e-01
4.05020118e-02 -1.05859470e+00 -9.72721875e-01 -7.49145269e-01
-2.40774319e-01 -1.76869094e-01 6.00117922e-01 6.10882461e-01
-9.48535055e-02 5.60790300e-02 1.44247651e-01 -8.66541386e-01
-4.41771150e-01 -1.03082466e+00 -2.60614783e-01 2.01124787e-01
1.65774122e-01 -6.38223112e-01 -1.14803128e-01 -4.30393219e-01] | [10.924373626708984, 10.378673553466797] |
efee2f47-9879-468f-acc5-f429d9af4b1a | learning-inr-for-event-guided-rolling-shutter | 2305.15078 | null | https://arxiv.org/abs/2305.15078v1 | https://arxiv.org/pdf/2305.15078v1.pdf | Learning INR for Event-guided Rolling Shutter Frame Correction, Deblur, and Interpolation | Images captured by rolling shutter (RS) cameras under fast camera motion often contain obvious image distortions and blur, which can be modeled as a row-wise combination of a sequence of global shutter (GS) frames within the exposure time naturally, recovering high-frame-rate GS sharp frames from an RS blur image needs to simultaneously consider RS correction, deblur, and frame interpolation Taking this task is nontrivial, and to our knowledge, no feasible solutions exist by far. A naive way is to decompose the complete process into separate tasks and simply cascade existing methods; however, this results in cumulative errors and noticeable artifacts. Event cameras enjoy many advantages, e.g., high temporal resolution, making them potential for our problem. To this end, we make the first attempt to recover high-frame-rate sharp GS frames from an RS blur image and paired event data. Our key idea is to learn an implicit neural representation (INR) to directly map the position and time coordinates to RGB values to address the interlocking degradations in the image restoration process. Specifically, we introduce spatial-temporal implicit encoding (STE) to convert an RS blur image and events into a spatial-temporal representation (STR). To query a specific sharp frame (GS or RS), we embed the exposure time into STR and decode the embedded features to recover a sharp frame. Moreover, we propose an RS blur image-guided integral loss to better train the network. Our method is relatively lightweight as it contains only 0.379M parameters and demonstrates high efficiency as the STE is called only once for any number of interpolation frames. Extensive experiments show that our method significantly outperforms prior methods addressing only one or two of the tasks. | ['Lin Wang', 'Guoqiang Liang', 'Yunfan Lu'] | 2023-05-24 | null | null | null | null | ['image-restoration'] | ['computer-vision'] | [ 4.97675717e-01 -5.67845464e-01 1.50494441e-01 -4.39292133e-01
-7.82830000e-01 -4.93876845e-01 3.32963765e-01 -6.50873244e-01
-4.49528605e-01 5.95965445e-01 1.26147568e-01 -1.86448246e-01
-8.57641548e-02 -4.11081582e-01 -9.99875367e-01 -7.06142783e-01
3.21598977e-01 -4.07771587e-01 3.61238033e-01 -1.14712575e-02
3.56386900e-01 4.70131248e-01 -1.24267673e+00 1.50055811e-01
8.90079677e-01 1.12813115e+00 6.33669317e-01 7.42903233e-01
3.47982734e-01 1.24677289e+00 -4.78694409e-01 -1.78117409e-01
3.37349653e-01 -5.93252003e-01 -6.94347799e-01 2.21960500e-01
4.29480344e-01 -1.04969907e+00 -8.24111283e-01 1.13696074e+00
2.64230520e-01 3.56691658e-01 1.18086539e-01 -8.51286888e-01
-9.59394813e-01 1.71228305e-01 -8.90945673e-01 2.26250187e-01
4.65552896e-01 3.45255077e-01 4.59072798e-01 -1.01763177e+00
3.10576379e-01 1.06398976e+00 6.96541131e-01 3.97067249e-01
-1.10447919e+00 -5.12903392e-01 -5.24408408e-02 3.79708558e-01
-1.22437239e+00 -6.65657282e-01 7.25572944e-01 -1.68537363e-01
6.99119627e-01 2.95976073e-01 6.13780260e-01 9.07215834e-01
1.09584510e-01 6.11313224e-01 1.10181248e+00 -7.27223381e-02
1.69670805e-01 -5.28530180e-01 -7.07351649e-03 4.50319558e-01
2.47561559e-03 1.81760862e-01 -7.06324995e-01 2.17994839e-01
1.36454475e+00 5.37317336e-01 -1.00321388e+00 4.80219759e-02
-1.39576662e+00 3.31170171e-01 6.23937070e-01 2.12080907e-02
-2.67157704e-01 5.30743897e-01 1.48033919e-02 2.54030496e-01
2.67594159e-01 3.60566765e-01 -3.95623237e-01 -2.55378813e-01
-1.06865954e+00 5.92164285e-02 3.93915951e-01 8.50818396e-01
9.50680077e-01 -5.86863905e-02 -1.61961347e-01 7.31146693e-01
1.27671316e-01 3.85631144e-01 3.95283133e-01 -1.18421829e+00
3.41348678e-01 -1.66204032e-02 4.73459691e-01 -9.26183045e-01
-7.06527615e-03 7.13332668e-02 -1.03492534e+00 7.93970153e-02
4.71998453e-01 3.81257455e-03 -9.24695790e-01 1.58702624e+00
2.60465536e-02 6.44870043e-01 -2.42154151e-01 1.40022314e+00
4.12840694e-01 8.50275815e-01 -3.01607847e-01 -4.85347658e-01
1.34556091e+00 -9.14581895e-01 -9.84843135e-01 -4.05772597e-01
-1.75675541e-01 -9.42047536e-01 9.93471384e-01 2.94283748e-01
-1.51201010e+00 -6.76720798e-01 -9.65614021e-01 -6.74783707e-01
6.34697452e-02 1.61202148e-01 4.27764982e-01 1.07645005e-01
-1.11693704e+00 7.30831325e-01 -1.08229136e+00 5.53050265e-02
2.26215482e-01 1.73210874e-01 -1.57172337e-01 -4.69689578e-01
-1.22362554e+00 7.58774757e-01 6.91310018e-02 4.44139838e-01
-8.07728469e-01 -7.05654442e-01 -9.82759595e-01 1.81725901e-02
4.22086149e-01 -7.80039728e-01 1.11494422e+00 -1.07364976e+00
-1.65736389e+00 5.41052699e-01 -4.00299966e-01 -2.98376024e-01
5.70018828e-01 -5.34726560e-01 -1.92733139e-01 2.91162193e-01
-3.36542539e-02 3.46470743e-01 1.29643345e+00 -1.09255743e+00
-4.03133899e-01 -7.74728786e-03 2.29428276e-01 1.29636511e-01
-1.44605890e-01 2.97285378e-01 -8.07304502e-01 -9.68934774e-01
3.16176176e-01 -7.83016920e-01 -2.83657275e-02 3.05736989e-01
-1.39123484e-01 3.05666864e-01 6.92027330e-01 -9.47615921e-01
1.17041135e+00 -2.33899450e+00 1.93011880e-01 -5.25566757e-01
1.67400226e-01 2.07991078e-01 3.42054665e-02 8.40316117e-02
-2.43026823e-01 -2.03235835e-01 -5.86410761e-01 -4.88342673e-01
-3.95845592e-01 2.02909723e-01 -7.24438310e-01 6.00848317e-01
2.50113875e-01 1.05405784e+00 -1.16989696e+00 -7.34781325e-02
5.66115916e-01 7.52070427e-01 -4.93819982e-01 5.41491687e-01
7.23873824e-02 6.01104558e-01 -1.42640904e-01 4.93423551e-01
1.01526034e+00 -3.74631435e-01 -1.72029138e-01 -7.18030035e-01
-3.47744077e-01 2.93934345e-01 -1.01563025e+00 1.94817615e+00
-4.16693121e-01 8.51840496e-01 -1.84635185e-02 -6.16659522e-01
5.44338226e-01 1.02323316e-01 3.74073237e-01 -5.04574478e-01
-3.84218730e-02 2.35135555e-01 -5.20700097e-01 -6.58618331e-01
7.36965895e-01 -5.50318509e-02 2.18242481e-01 3.46006006e-01
-1.95735812e-01 -1.37007982e-01 -1.36020377e-01 -4.44585271e-02
1.14849734e+00 3.19178849e-01 8.80727097e-02 2.01582253e-01
4.06515241e-01 -5.63292444e-01 6.24472558e-01 6.76189721e-01
-2.13145733e-01 1.38440228e+00 2.24161938e-01 -5.55416346e-01
-9.35989082e-01 -1.19409645e+00 4.00904100e-03 6.46898091e-01
6.84967816e-01 -1.88019156e-01 -7.08208323e-01 -3.56394023e-01
-3.98839533e-01 4.90938395e-01 -3.24671924e-01 -2.62910575e-01
-9.29925978e-01 -5.98183870e-01 2.59994537e-01 4.43415284e-01
8.83720517e-01 -8.61988187e-01 -7.45527089e-01 9.77675319e-02
-5.37330329e-01 -1.26564956e+00 -1.19852626e+00 -1.89187229e-02
-7.38187730e-01 -1.15435064e+00 -9.88820374e-01 -7.17900276e-01
7.41138995e-01 9.25332963e-01 8.14917684e-01 -5.56617640e-02
-1.42489344e-01 1.10943154e-01 -2.88277775e-01 3.27619791e-01
-5.96450046e-02 -5.34701943e-01 -4.29030545e-02 3.10697764e-01
-7.38547072e-02 -5.22583961e-01 -1.10082459e+00 3.92608106e-01
-1.32447672e+00 5.24231911e-01 5.66123903e-01 9.73079979e-01
4.31042671e-01 -4.18352969e-02 1.54122993e-01 -3.88908446e-01
3.38716328e-01 -1.81023911e-01 -6.47106409e-01 1.82493374e-01
-3.48118782e-01 -9.33951586e-02 6.10638618e-01 -5.34626901e-01
-1.22871327e+00 7.30613023e-02 -5.78361005e-02 -9.90308225e-01
-3.02091688e-02 2.68172145e-01 6.13690019e-02 -2.00155854e-01
4.53862101e-01 5.54710805e-01 -1.32746756e-01 -4.68998343e-01
3.74241233e-01 6.76702023e-01 9.35825169e-01 -3.42716336e-01
7.86162019e-01 5.91027141e-01 -2.54957646e-01 -6.59431219e-01
-9.39739883e-01 -2.72372425e-01 -3.15451860e-01 -1.64499179e-01
8.16838801e-01 -1.15915740e+00 -7.59823978e-01 9.30589557e-01
-1.45983982e+00 -5.04816532e-01 -2.03089461e-01 4.63678062e-01
-5.28616965e-01 8.28727484e-01 -1.09068048e+00 -4.51841980e-01
-1.53078631e-01 -1.31374896e+00 1.15596271e+00 3.59941661e-01
1.88614622e-01 -5.37698746e-01 -3.13940436e-01 2.96160489e-01
6.69524431e-01 -4.44729887e-02 3.35397571e-01 5.13514817e-01
-1.20478892e+00 4.40677442e-02 -6.92757666e-01 5.18048108e-01
6.01892829e-01 -1.75761163e-01 -1.04702222e+00 -4.23872530e-01
6.68225765e-01 -1.23111956e-01 1.05807304e+00 5.52518427e-01
1.67565954e+00 -3.34092259e-01 1.32147297e-01 1.20129943e+00
1.46349323e+00 1.50955394e-01 9.72754061e-01 3.03111136e-01
9.90623295e-01 1.22594520e-01 4.92053568e-01 4.07023430e-01
4.10884589e-01 7.38425434e-01 3.46722007e-01 -1.31969213e-01
-4.24084395e-01 -1.68233499e-01 7.58768022e-01 6.31930590e-01
-1.83050558e-01 -1.37291253e-01 -3.65829557e-01 3.45398962e-01
-1.74024308e+00 -1.02452052e+00 -5.63022457e-02 2.30743146e+00
1.25468552e+00 -2.56868094e-01 -4.18967694e-01 -1.24804966e-01
7.82081306e-01 5.27756333e-01 -6.44409001e-01 2.63607115e-01
-2.19633862e-01 6.13050982e-02 5.62715054e-01 6.45425081e-01
-9.12656546e-01 7.59464324e-01 5.80777884e+00 6.48144841e-01
-1.32750213e+00 1.31871388e-03 7.88210511e-01 -3.31715584e-01
-1.34593993e-01 2.25884303e-01 -3.70053172e-01 9.42602634e-01
6.69510722e-01 3.76614630e-02 1.24495137e+00 3.70408595e-01
5.30388415e-01 -1.19805142e-01 -1.04552495e+00 1.48774433e+00
1.08385146e-01 -1.31226802e+00 -1.39540628e-01 -2.42085978e-01
7.97163904e-01 -1.32183200e-02 1.04662143e-01 -1.78479165e-01
-2.45115645e-02 -8.12110424e-01 9.37244892e-01 7.32452273e-01
1.20348823e+00 -3.73506904e-01 3.75301272e-01 1.37327552e-01
-1.08507144e+00 -6.03496004e-03 -3.89591068e-01 -6.06340170e-02
5.25637269e-01 7.13369012e-01 -4.64172028e-02 5.44892013e-01
9.18493927e-01 1.08284295e+00 -3.25483888e-01 9.74684656e-01
-3.06089729e-01 3.53075385e-01 -1.74693078e-01 7.53661335e-01
-1.43926188e-01 -3.62407088e-01 3.97428840e-01 9.51683044e-01
7.47874320e-01 3.73875260e-01 -2.28288665e-01 9.86618340e-01
-1.74807549e-01 -7.78283715e-01 -2.34630644e-01 3.45387787e-01
2.80198544e-01 1.14399552e+00 -2.72492349e-01 -2.70613551e-01
-6.18455291e-01 1.66907370e+00 1.39050603e-01 6.88155890e-01
-1.13478613e+00 -3.98971766e-01 8.66434634e-01 -6.01451565e-03
3.66480917e-01 -3.03127497e-01 -3.29377472e-01 -1.76475000e+00
1.85374826e-01 -7.16135025e-01 3.48406807e-02 -1.35604119e+00
-1.22411442e+00 5.31800151e-01 -2.88128167e-01 -1.43626201e+00
-7.42310807e-02 -4.45316821e-01 -4.26821053e-01 1.00733602e+00
-1.91922128e+00 -9.59196746e-01 -8.20460021e-01 7.88976669e-01
8.00557554e-01 5.53196192e-01 2.09244266e-01 5.18373966e-01
-6.95410371e-01 2.79765666e-01 7.65300244e-02 7.45353568e-03
1.16578364e+00 -1.13123000e+00 4.80212122e-01 1.34039462e+00
-2.26267114e-01 7.35519946e-01 7.26616442e-01 -5.02520978e-01
-1.62724578e+00 -1.17435074e+00 6.73142791e-01 -2.44368315e-01
5.73119700e-01 -5.01573905e-02 -1.11803460e+00 7.24155843e-01
2.67888248e-01 5.04239440e-01 -7.76933432e-02 -7.62115896e-01
-2.01905265e-01 -3.65628093e-01 -8.71363521e-01 5.54988205e-01
9.99673605e-01 -8.95121276e-01 -4.78285670e-01 1.39946386e-01
9.31890428e-01 -9.10558462e-01 -6.21704161e-01 3.08933675e-01
3.32883030e-01 -1.26615798e+00 1.24561596e+00 1.84611142e-01
7.42182612e-01 -8.83202970e-01 1.03485798e-02 -1.13263118e+00
-2.59946257e-01 -1.07751513e+00 -4.92435187e-01 1.16974354e+00
-3.04973513e-01 -7.36915886e-01 3.50516707e-01 8.33804250e-01
-2.40160167e-01 -6.25895023e-01 -7.46746898e-01 -4.68625665e-01
-4.85557318e-01 -2.82121867e-01 6.01576149e-01 9.66964245e-01
-3.98701459e-01 -4.01786044e-02 -8.41578662e-01 3.56275022e-01
7.43552566e-01 2.98971593e-01 5.35671830e-01 -4.72590625e-01
-3.79066765e-01 -6.55961260e-02 7.01168999e-02 -1.87021911e+00
-2.28820935e-01 -2.26125672e-01 4.43101794e-01 -1.19048178e+00
1.68934256e-01 -3.64721954e-01 -4.07512546e-01 2.23231927e-01
-5.73310733e-01 2.62255162e-01 1.76898673e-01 5.30453563e-01
-4.11628008e-01 6.34149075e-01 1.34623885e+00 7.77853802e-02
-4.05461676e-02 -2.62029290e-01 -5.81615269e-01 6.88371718e-01
3.54373604e-01 -1.78388476e-01 -2.85127908e-01 -9.35978115e-01
1.31061658e-01 5.02474129e-01 8.14006269e-01 -8.79587889e-01
5.39336741e-01 -3.72763991e-01 5.63286364e-01 -4.57372516e-01
4.99632537e-01 -6.63072944e-01 4.15494174e-01 6.83559403e-02
-2.06810996e-01 5.79883680e-02 -4.34628911e-02 6.79548621e-01
-3.92149329e-01 -6.51326105e-02 9.70808923e-01 -6.96577951e-02
-7.54724801e-01 4.69012707e-01 -2.72020847e-02 -1.85974076e-01
7.36603975e-01 -2.21537665e-01 -5.12733817e-01 -4.15134937e-01
-2.02132165e-01 -1.57525927e-01 9.84676361e-01 3.05074245e-01
8.75893056e-01 -1.16135180e+00 -2.94149369e-01 3.53750527e-01
-2.60966152e-01 4.52313632e-01 5.69262505e-01 9.05530930e-01
-8.40135098e-01 3.16972993e-02 -6.64522275e-02 -5.23922682e-01
-7.79503882e-01 6.49845481e-01 4.34646904e-01 1.60263941e-01
-7.71202207e-01 8.74948084e-01 5.51286399e-01 2.72618800e-01
-5.29877190e-03 -6.12985432e-01 2.79238969e-01 -3.39555830e-01
8.64872515e-01 3.45228463e-01 -1.27969474e-01 -4.23644930e-01
-1.17091194e-01 7.27188289e-01 3.13632265e-02 1.27548799e-02
1.31585562e+00 -6.56623602e-01 -3.14692408e-01 2.74869531e-01
1.27776432e+00 -1.19173683e-01 -2.06164503e+00 -1.69453710e-01
-4.62428629e-01 -1.02489531e+00 2.05784500e-01 -5.55072546e-01
-1.26227081e+00 7.86315560e-01 5.76862693e-01 -4.57942709e-02
1.69139194e+00 -1.75461307e-01 1.23374331e+00 -3.86191830e-02
2.74354428e-01 -6.87840819e-01 1.95530862e-01 4.54981983e-01
9.19596910e-01 -1.20466340e+00 2.33471394e-02 -4.07459468e-01
-3.83860797e-01 1.37420177e+00 4.19436723e-01 -1.97761908e-01
1.65537268e-01 6.34194240e-02 -1.14209495e-01 1.43417060e-01
-5.92751503e-01 8.20734799e-02 1.73820496e-01 2.56020457e-01
2.39466235e-01 -3.39771956e-01 9.43180993e-02 2.33694166e-01
1.46795020e-01 4.37551796e-01 7.24551082e-01 6.88225627e-01
-1.38344988e-01 -7.15762615e-01 -4.46422368e-01 9.94133577e-02
-4.82370645e-01 -2.80829608e-01 2.92821437e-01 2.08527833e-01
-2.67803185e-02 8.90109122e-01 3.71433794e-02 -2.09835619e-01
2.17849061e-01 -5.05716920e-01 4.95845437e-01 -8.45965296e-02
-2.29665235e-01 1.16859123e-01 -5.06362557e-01 -8.89581978e-01
-4.61096942e-01 -6.06088281e-01 -9.95401502e-01 -4.53146398e-01
-2.64630824e-01 -4.59613532e-01 4.35290575e-01 8.30742002e-01
3.05155694e-01 5.00700355e-01 8.17231894e-01 -1.32653582e+00
-3.64727378e-01 -6.24648213e-01 -4.25549895e-01 5.86349785e-01
1.01130664e+00 -2.90235490e-01 -6.81532621e-01 5.29975176e-01] | [11.246898651123047, -2.294271945953369] |
f67b5cb0-5ebd-4b60-b87c-121ea9281187 | monocular-3d-object-detection-for-autonomous | null | null | http://openaccess.thecvf.com/content_cvpr_2016/html/Chen_Monocular_3D_Object_CVPR_2016_paper.html | http://openaccess.thecvf.com/content_cvpr_2016/papers/Chen_Monocular_3D_Object_CVPR_2016_paper.pdf | Monocular 3D Object Detection for Autonomous Driving | The goal of this paper is to perform 3D object detection in single monocular images in the domain of autonomous driving. Our method first aims to generate a set of candidate class-specific object proposals, which are then run through a standard CNN pipeline to obtain high-quality object detections. The focus of this paper is on proposal generation. In particular, we propose a probabilistic model that places object candidates in 3D using a prior on ground-plane. We then score each candidate box projected to the image plane via several intuitive potentials such as semantic segmentation, contextual information, size and location priors and typical object shape. The weights in our model are trained with S-SVM. Experiments show that our object proposal generation approach significantly outperforms all monocular baselines, and achieves the best detection performance on the challenging KITTI benchmark, among the published monocular competitors. | ['Sanja Fidler', 'Raquel Urtasun', 'Ziyu Zhang', 'Xiaozhi Chen', 'Huimin Ma', 'Kaustav Kundu'] | 2016-06-01 | null | null | null | cvpr-2016-6 | ['object-proposal-generation', 'vehicle-pose-estimation'] | ['computer-vision', 'computer-vision'] | [ 1.53443977e-01 1.23448551e-01 -2.18416169e-01 -5.21946728e-01
-6.54861033e-01 -5.65095723e-01 8.30145359e-01 -1.34278670e-01
-6.64638281e-01 3.27674508e-01 -2.96099007e-01 -1.16638817e-01
4.62340236e-01 -5.38828254e-01 -9.94115949e-01 -3.17419946e-01
2.59951532e-01 7.36649156e-01 9.95315373e-01 6.61822930e-02
5.25957823e-01 7.58945227e-01 -1.81483531e+00 -6.85613230e-02
5.40099025e-01 1.21118045e+00 5.05758762e-01 7.05930769e-01
1.98823258e-01 2.06422314e-01 -6.08287215e-01 -2.19258040e-01
6.93614006e-01 2.26510316e-01 -3.73615980e-01 3.37031543e-01
8.69606912e-01 -5.81861854e-01 -2.61650383e-01 1.17549860e+00
3.25251609e-01 -7.43205175e-02 8.22505176e-01 -1.32619667e+00
-8.29528198e-02 2.77504444e-01 -7.84772635e-01 3.24999481e-01
1.62471429e-01 5.66878140e-01 9.45540965e-01 -1.51172900e+00
7.39610016e-01 1.43692517e+00 4.94244695e-01 3.99551362e-01
-1.06265914e+00 -6.65936410e-01 4.79918629e-01 4.69536223e-02
-1.25671124e+00 -2.30852544e-01 5.71291268e-01 -5.69930434e-01
9.59122300e-01 -2.25057468e-01 6.53182566e-01 7.72632420e-01
1.90623730e-01 1.09298277e+00 9.30295169e-01 -1.26364514e-01
3.03996712e-01 1.75234571e-01 1.53103232e-01 6.29402518e-01
6.01074994e-01 5.18066347e-01 -6.14421725e-01 -7.90672563e-03
6.66343451e-01 -2.57851481e-01 1.04418375e-01 -1.02628827e+00
-1.21470499e+00 8.04879665e-01 6.94694221e-01 -3.38118672e-01
-1.80798739e-01 3.60632211e-01 -1.62085332e-02 -4.24129218e-01
3.52842599e-01 4.81470317e-01 -3.52172613e-01 3.93228978e-01
-7.54510462e-01 7.38504231e-01 1.80056423e-01 1.38872170e+00
8.23771000e-01 -1.61342189e-01 -4.46894795e-01 5.32658875e-01
5.69972873e-01 7.50581861e-01 1.11502744e-02 -1.10857689e+00
5.45268595e-01 6.27496481e-01 5.31582654e-01 -5.94200373e-01
-5.78042388e-01 -6.73756242e-01 7.81010389e-02 6.12081289e-01
3.65726978e-01 4.91114892e-02 -1.30789018e+00 1.31698251e+00
6.77327812e-01 8.10935795e-02 -1.37272149e-01 1.25297761e+00
1.09245706e+00 4.59921986e-01 2.97350939e-02 5.04773676e-01
1.27815759e+00 -1.07605922e+00 -1.01581894e-01 -1.01579988e+00
2.34097540e-01 -8.81344914e-01 5.78356922e-01 1.27862440e-02
-1.14434743e+00 -8.59140813e-01 -1.10959506e+00 -1.76690102e-01
-1.45772368e-01 8.08339119e-01 3.21846843e-01 4.96564120e-01
-9.17704105e-01 2.53458302e-02 -8.23515952e-01 -3.28428030e-01
6.36454582e-01 2.32590914e-01 -1.21004783e-01 -2.54661925e-02
-5.71106553e-01 1.17802036e+00 7.25024581e-01 -1.01485774e-01
-1.16452181e+00 -5.91044962e-01 -1.11048639e+00 -1.92852870e-01
3.96241874e-01 -9.03002262e-01 1.41867459e+00 -3.44755620e-01
-1.25338042e+00 1.28265524e+00 -2.65178978e-01 -8.45955610e-01
6.62681818e-01 -3.78868461e-01 2.72136152e-01 9.27539542e-02
4.78595465e-01 1.67766726e+00 9.98307705e-01 -1.49796808e+00
-1.38525605e+00 -3.74038041e-01 -2.33150702e-02 4.09514308e-01
5.00410020e-01 4.04445641e-02 -8.86753500e-01 -1.63338527e-01
5.81074893e-01 -1.12234867e+00 -4.72815037e-01 2.27675676e-01
-5.84035099e-01 -5.11100709e-01 7.77700365e-01 -6.18558601e-02
2.98178107e-01 -1.96398115e+00 -1.36502152e-02 5.40864207e-02
1.70952559e-01 4.40559648e-02 6.63244128e-02 -1.82385221e-01
3.17329586e-01 -3.81025523e-01 -2.25873098e-01 -5.12333095e-01
1.66577712e-01 -1.76139817e-01 -4.62063998e-01 6.06456161e-01
7.30251551e-01 1.18811476e+00 -8.99905562e-01 -3.97680163e-01
5.43192923e-01 1.87595278e-01 -5.98092794e-01 1.19805425e-01
-5.90736389e-01 1.70273930e-01 -4.20018226e-01 7.49793231e-01
9.98193085e-01 -7.82774910e-02 -3.30163270e-01 -8.33502337e-02
-3.03396434e-01 4.88760471e-01 -1.06404483e+00 1.46904397e+00
7.47504309e-02 6.39242589e-01 -1.14865370e-01 -6.04093850e-01
1.15151131e+00 -4.10215348e-01 6.40799031e-02 -3.07189673e-01
2.44407207e-01 3.53695810e-01 -6.62193522e-02 2.54911575e-02
8.83913279e-01 2.64826626e-01 -8.18939954e-02 1.33227497e-01
1.22294175e-02 -7.73559630e-01 2.02577814e-01 1.80978909e-01
7.51744092e-01 5.80729723e-01 2.07947627e-01 -1.21583968e-01
3.15906227e-01 4.57748145e-01 4.70236838e-01 9.75688934e-01
-5.36726475e-01 9.12682831e-01 2.29717806e-01 -3.60190094e-01
-1.26533496e+00 -1.14196491e+00 -3.55440378e-01 6.85214996e-01
8.50233912e-01 9.93208215e-02 -5.40616632e-01 -7.79761374e-01
3.26696455e-01 6.64490223e-01 -5.04319191e-01 -7.77525306e-02
-6.39170885e-01 -5.37968934e-01 1.02405427e-02 7.61921644e-01
4.03193235e-01 -9.81151760e-01 -1.25091112e+00 1.50424108e-01
3.69400084e-02 -1.52404571e+00 -3.57605398e-01 3.10517281e-01
-7.44401693e-01 -9.84610260e-01 -5.48437893e-01 -8.10890615e-01
6.69885218e-01 7.98657000e-01 1.00827801e+00 -3.46606374e-01
-4.85649228e-01 1.76318362e-01 -4.32681739e-02 -8.62309992e-01
-5.17487489e-02 9.25000235e-02 -2.61214282e-02 -9.95343104e-02
5.78606725e-01 1.92366228e-01 -8.13266814e-01 4.37928259e-01
-2.51720816e-01 9.98172536e-02 7.65847564e-01 5.99047542e-01
8.05717111e-01 -4.90100324e-01 1.54330388e-01 -3.39040667e-01
-1.51248887e-01 -1.97821990e-01 -1.33392012e+00 -3.39293063e-01
-1.52549461e-01 -1.56650871e-01 -1.57307144e-02 -9.49173272e-02
-8.81585598e-01 7.83647656e-01 1.81966826e-01 -5.35919249e-01
-4.35308754e-01 -1.17995419e-01 -9.53814108e-03 -2.13039801e-01
8.69453251e-01 2.34437659e-01 -2.34431684e-01 -2.20389381e-01
5.84879518e-01 2.50614673e-01 7.40146399e-01 -2.92549968e-01
1.28570700e+00 8.35636556e-01 -2.20113583e-02 -6.44031107e-01
-1.05649936e+00 -8.30963850e-01 -8.87286603e-01 -3.07107657e-01
9.85301912e-01 -1.31972003e+00 -3.68613005e-01 3.10072660e-01
-1.38033986e+00 -2.49885157e-01 -2.17087641e-01 5.84179044e-01
-6.25390887e-01 1.25429202e-02 -2.18842924e-01 -9.29646134e-01
-2.63745002e-02 -1.27554095e+00 1.66424274e+00 3.71398360e-01
4.08900976e-02 -2.57283777e-01 -2.09489018e-01 4.42771703e-01
-3.83294225e-02 -9.74603891e-02 2.20862031e-01 -3.52496773e-01
-1.46678925e+00 -3.79479527e-01 -7.41582572e-01 2.98636425e-02
-4.63385522e-01 2.88430098e-02 -1.18979180e+00 -1.24384567e-01
-4.08660710e-01 -3.43165278e-01 1.40944195e+00 7.98917711e-01
1.00980580e+00 4.31744337e-01 -8.82374048e-01 6.56653762e-01
1.08636868e+00 9.37344506e-02 1.33630797e-01 3.89185429e-01
5.22600472e-01 5.85233390e-01 1.06073713e+00 2.83883721e-01
4.97450441e-01 8.74087512e-01 7.83014715e-01 4.34721671e-02
-1.92840755e-01 -3.74711812e-01 1.51477665e-01 -3.34088027e-01
3.88765812e-01 -1.42398467e-02 -9.01556313e-01 7.84435868e-01
-1.87990451e+00 -7.83084452e-01 -2.61931300e-01 1.95573318e+00
3.23805481e-01 7.65043318e-01 3.95872027e-01 -2.32829645e-01
7.09171772e-01 -8.48787576e-02 -7.80673742e-01 2.33754575e-01
-2.40837276e-01 4.53916145e-03 1.02712476e+00 4.63295102e-01
-1.68573642e+00 1.31209683e+00 6.69026279e+00 3.61205131e-01
-9.34164107e-01 -2.74538603e-02 5.59542060e-01 -1.50463715e-01
1.61671281e-01 -2.13918742e-02 -1.75797009e+00 9.12243798e-02
2.32374460e-01 4.61725406e-02 -2.88732708e-01 1.27181184e+00
6.38877079e-02 -4.14277583e-01 -1.06262422e+00 9.16945517e-01
1.73586100e-01 -1.33068275e+00 -2.14453340e-01 -2.15161853e-02
1.04584002e+00 7.08654821e-01 2.67900646e-01 5.23773357e-02
5.82164824e-01 -8.09832394e-01 1.43207431e+00 1.07567787e-01
3.11609834e-01 -5.38477898e-01 5.73240340e-01 3.96767825e-01
-1.17027795e+00 -1.72088087e-01 -7.53055036e-01 -2.20619421e-02
1.39405206e-01 4.83202249e-01 -1.38550305e+00 1.80707164e-02
7.47385800e-01 7.46445775e-01 -6.51004970e-01 1.80792809e+00
-5.21563470e-01 3.69787425e-01 -5.03370464e-01 -1.88308939e-01
5.06421804e-01 1.12139456e-01 8.44960570e-01 1.07657468e+00
2.17692271e-01 -4.07673195e-02 5.75363100e-01 1.36083722e+00
-4.61759083e-02 -1.81370988e-01 -5.48597097e-01 5.28448820e-01
6.29064620e-01 1.31468999e+00 -9.15681720e-01 -3.99156481e-01
-3.53558332e-01 4.64002222e-01 2.41623923e-01 1.96759954e-01
-7.89622545e-01 -1.64952591e-01 6.34138942e-01 1.41428247e-01
8.46696913e-01 -3.46948057e-01 -6.58598900e-01 -8.52512419e-01
2.53244579e-01 -2.43573651e-01 -1.29787729e-03 -1.07723820e+00
-1.00680661e+00 4.41033334e-01 1.44306898e-01 -1.24748564e+00
1.28052354e-01 -9.75426257e-01 -5.72812259e-01 8.34806204e-01
-1.89167821e+00 -1.09396803e+00 -5.12186289e-01 -1.85385589e-02
7.46278822e-01 -9.43858922e-02 -6.29459396e-02 6.29802374e-03
-2.93896317e-01 2.62618899e-01 -3.74024868e-01 -4.05761935e-02
5.14974296e-01 -1.30168378e+00 1.11952090e+00 9.88374949e-01
1.78027779e-01 1.82751149e-01 7.43831217e-01 -8.14361930e-01
-9.56543505e-01 -1.52037084e+00 7.21741676e-01 -1.04733539e+00
3.70974749e-01 -4.55900162e-01 -6.16231978e-01 6.23085141e-01
-2.23255828e-01 2.95077652e-01 -2.53457725e-01 -3.84704322e-01
-2.83062190e-01 6.06368519e-02 -8.09755027e-01 4.64199781e-01
1.14833748e+00 -1.08519912e-01 -5.98130882e-01 5.40983438e-01
5.97824037e-01 -8.69256318e-01 -3.11453268e-02 5.68255961e-01
3.66609484e-01 -8.75016928e-01 1.32675159e+00 -3.61113846e-01
2.12444887e-01 -8.57606292e-01 7.07205683e-02 -9.72514391e-01
-2.38319129e-01 -3.79466832e-01 1.72856048e-01 4.49105024e-01
7.01166928e-01 -3.24837685e-01 1.16111493e+00 2.06926391e-01
-6.02566719e-01 -4.23540950e-01 -9.72400606e-01 -8.29859018e-01
-2.51147687e-01 -6.21813476e-01 3.94768834e-01 1.01390436e-01
-4.70322907e-01 3.33923131e-01 -2.61618085e-02 5.71294844e-01
1.02584076e+00 4.62530702e-01 1.24026680e+00 -1.29729915e+00
-6.90584928e-02 -7.02872634e-01 -6.50973618e-01 -1.72027874e+00
2.79583018e-02 -9.15258884e-01 7.10019708e-01 -1.58136642e+00
2.75801331e-01 -4.89755005e-01 1.37595847e-01 2.42299944e-01
-4.32658881e-01 5.98625541e-01 1.81677520e-01 1.01254717e-01
-7.72233665e-01 6.79814398e-01 1.26015425e+00 -2.19705179e-01
-2.68893182e-01 2.71285087e-01 -3.07034194e-01 8.33342254e-01
6.01989150e-01 -4.48095798e-01 -1.00705847e-01 -2.89790928e-01
-8.75407606e-02 -3.31648707e-01 8.50862980e-01 -1.23640811e+00
1.87312528e-01 -1.73782364e-01 6.25068188e-01 -1.50582421e+00
7.60667741e-01 -3.82841498e-01 -5.59548557e-01 5.15326917e-01
-7.51573965e-02 -3.79723668e-01 2.35701367e-01 5.91389596e-01
1.01581119e-01 -1.66832939e-01 1.07864892e+00 -7.56936446e-02
-8.79840434e-01 4.15147871e-01 -2.16961324e-01 1.00245632e-01
1.19398499e+00 -4.13776100e-01 -2.73011625e-01 6.62728548e-02
-4.84615505e-01 5.27008772e-01 5.18526852e-01 6.88391566e-01
8.09027314e-01 -1.23623872e+00 -7.79887438e-01 2.29508579e-01
6.17776215e-01 3.72442275e-01 -3.03657621e-01 5.69728076e-01
-5.17720938e-01 8.58940303e-01 -1.42216012e-01 -1.23221362e+00
-1.05783260e+00 3.30425292e-01 5.79619050e-01 4.59570497e-01
-8.34257245e-01 1.33790982e+00 6.86036110e-01 -4.78626251e-01
3.79795432e-01 -7.00792253e-01 -2.24508457e-02 -4.50590760e-01
3.62992764e-01 -2.41529159e-02 -7.07344040e-02 -8.05822253e-01
-4.23846364e-01 8.03823769e-01 2.60024294e-02 -2.15681091e-01
9.69277143e-01 -6.78875446e-02 4.66370434e-01 1.12850033e-01
5.83363414e-01 -3.46077710e-01 -1.94053686e+00 -4.29751724e-01
4.95818034e-02 -6.52058423e-01 1.37824351e-02 -5.39161146e-01
-6.61493719e-01 8.83071363e-01 5.57977378e-01 -3.21821094e-01
2.52946854e-01 4.80462104e-01 4.82193381e-01 3.99326414e-01
4.52659696e-01 -9.24337268e-01 3.01394314e-01 7.38464892e-01
7.47001469e-01 -1.61242425e+00 -3.20631117e-02 -7.00173140e-01
-3.57144386e-01 8.75356019e-01 1.00175583e+00 -3.99831116e-01
4.05767143e-01 1.28735846e-03 6.39487579e-02 -2.87203521e-01
-5.78301251e-01 -6.84003711e-01 7.86621690e-01 7.19351649e-01
-4.46020477e-02 2.05611363e-02 -3.45534794e-02 2.70581931e-01
-3.96536529e-01 -4.49679792e-01 5.01405478e-01 7.23579347e-01
-1.17992806e+00 -6.81565285e-01 -6.25817776e-01 3.14074457e-01
6.70797527e-02 2.18375057e-01 -6.68804944e-01 6.21940970e-01
4.04551923e-01 7.91170239e-01 2.20526978e-01 2.04509366e-02
5.19139886e-01 -5.68393618e-02 5.24642766e-01 -1.01533735e+00
-2.38175727e-02 2.33568281e-01 -8.27512220e-02 -5.17270088e-01
-1.74559399e-01 -9.35169160e-01 -1.24417746e+00 4.44486737e-01
-5.45759141e-01 -3.16033840e-01 9.39231634e-01 8.93702328e-01
2.38720447e-01 2.28452995e-01 2.65289009e-01 -1.47337115e+00
-4.71974134e-01 -8.04656386e-01 -6.87818229e-02 4.69545424e-02
2.16387436e-01 -1.06932330e+00 -2.04371899e-01 -1.52716920e-01] | [7.702524185180664, -2.581723690032959] |
29d054ac-ae1c-49ad-8283-f9fc26b7ebfe | can-humans-do-less-than-one-shot-learning | 2202.04670 | null | https://arxiv.org/abs/2202.04670v1 | https://arxiv.org/pdf/2202.04670v1.pdf | Can Humans Do Less-Than-One-Shot Learning? | Being able to learn from small amounts of data is a key characteristic of human intelligence, but exactly {\em how} small? In this paper, we introduce a novel experimental paradigm that allows us to examine classification in an extremely data-scarce setting, asking whether humans can learn more categories than they have exemplars (i.e., can humans do "less-than-one shot" learning?). An experiment conducted using this paradigm reveals that people are capable of learning in such settings, and provides several insights into underlying mechanisms. First, people can accurately infer and represent high-dimensional feature spaces from very little data. Second, having inferred the relevant spaces, people use a form of prototype-based categorization (as opposed to exemplar-based) to make categorical inferences. Finally, systematic, machine-learnable patterns in responses indicate that people may have efficient inductive biases for dealing with this class of data-scarce problems. | ['Thomas L. Griffiths', 'Kerem Oktar', 'Ilia Sucholutsky', 'Maya Malaviya'] | 2022-02-09 | null | null | null | null | ['one-shot-learning'] | ['methodology'] | [ 1.41978592e-01 2.24632904e-01 -9.83593762e-02 -8.08103025e-01
-3.00303370e-01 -5.77534258e-01 7.23618388e-01 5.30282557e-01
-8.06444466e-01 8.43439341e-01 2.00045153e-01 -3.10887903e-01
-3.38243425e-01 -9.46061254e-01 -4.94556487e-01 -4.32831585e-01
8.18317160e-02 7.64368594e-01 -2.37145454e-01 -2.64509439e-01
8.23200107e-01 2.46340632e-01 -1.74694502e+00 3.52817684e-01
8.65209162e-01 6.30472422e-01 -3.81467715e-02 4.74030107e-01
-1.91896006e-01 8.98657560e-01 -5.26205301e-01 -5.45015574e-01
2.79122442e-01 -3.16115558e-01 -9.93376315e-01 -1.14871003e-01
6.40111804e-01 -3.76450151e-01 -5.27240075e-02 9.95899498e-01
1.78114548e-01 5.01503229e-01 1.00730681e+00 -1.11656654e+00
-1.16311562e+00 6.78024054e-01 -1.25886068e-01 4.65748936e-01
7.11518466e-01 4.28267658e-01 9.27082181e-01 -7.49995172e-01
4.71384674e-01 1.73727500e+00 6.19186342e-01 7.35159576e-01
-1.87148285e+00 -6.66907370e-01 2.09068567e-01 1.64672598e-01
-1.30950046e+00 -6.27685726e-01 6.39461637e-01 -7.55679488e-01
6.76535249e-01 3.76493067e-01 6.76919460e-01 1.18389344e+00
7.41573200e-02 3.40244204e-01 1.81839132e+00 -4.67417240e-01
6.24588311e-01 6.99288368e-01 6.49966419e-01 3.78896266e-01
8.31293106e-01 4.07441169e-01 -5.99433780e-01 -3.03463489e-01
6.11904383e-01 2.47283548e-01 -3.81888531e-04 -2.77674615e-01
-1.40450287e+00 1.02730262e+00 7.60363519e-01 3.58992249e-01
-4.86490399e-01 -1.08514011e-01 2.19027892e-01 7.73091078e-01
2.44344041e-01 1.26173985e+00 -2.95724303e-01 5.79765067e-02
-5.85217118e-01 4.49042290e-01 9.67009366e-01 7.99816310e-01
9.21744704e-01 -1.00006526e-02 -1.65624674e-02 7.06615806e-01
-8.29709917e-02 4.47031260e-01 6.70157433e-01 -9.56531167e-01
2.28869095e-01 5.46340227e-01 3.54332685e-01 -1.03314590e+00
-4.97979313e-01 -1.22199766e-01 -7.39324152e-01 2.39204451e-01
7.94547021e-01 -2.54919380e-01 -5.98223984e-01 1.99777496e+00
1.70181677e-01 -4.56654817e-01 2.14144029e-02 9.04899776e-01
1.67387158e-01 1.71917185e-01 4.66660380e-01 -3.40678096e-01
1.35431302e+00 -1.11779653e-01 -3.62590641e-01 -5.30168295e-01
5.88107347e-01 -3.68575007e-02 1.50125289e+00 4.38556939e-01
-8.91511559e-01 -7.82313406e-01 -9.44695055e-01 -1.20947629e-01
-6.21729314e-01 -6.27380967e-01 1.24819851e+00 6.53815150e-01
-8.44101012e-01 5.78048289e-01 -2.34537497e-01 -6.86236024e-01
4.68271405e-01 1.14342466e-01 -2.12276623e-01 -3.02308917e-01
-1.10727930e+00 1.01325738e+00 5.48199534e-01 -2.60548174e-01
-8.07047307e-01 -5.38966358e-01 -7.51380205e-01 -3.56179737e-02
3.64126295e-01 -7.03212500e-01 1.08190870e+00 -1.21808600e+00
-7.45025396e-01 1.09938705e+00 -3.03265125e-01 -3.03009152e-01
1.91241518e-01 1.30865434e-02 -3.74353468e-01 2.15286493e-01
2.17332706e-01 6.24350011e-01 8.06586504e-01 -1.39487684e+00
-4.76062745e-01 -7.94662654e-01 2.13969961e-01 8.38624686e-02
-3.20931852e-01 -1.52111411e-01 9.85047698e-01 -5.36827147e-01
2.37494662e-01 -6.84817374e-01 -1.33352771e-01 1.04408236e-02
-5.96710993e-03 -6.43990815e-01 -5.79497442e-02 -1.72644094e-01
1.00718105e+00 -2.01917124e+00 -1.30121976e-01 2.72720456e-01
5.53917944e-01 -1.38496116e-01 3.12113408e-02 3.99700582e-01
1.29141212e-01 5.88328540e-01 -1.38807684e-01 2.71268159e-01
3.51847500e-01 1.71889976e-01 -4.65792954e-01 2.49385759e-01
7.70093780e-03 9.05628085e-01 -1.05673575e+00 -4.51497287e-01
-2.19871439e-02 -1.84333563e-01 -5.07126451e-01 2.39036232e-01
-1.78310573e-02 1.49902239e-01 -3.27383786e-01 6.22721016e-01
3.71509612e-01 -2.34781012e-01 2.49024630e-01 2.35423416e-01
-7.55912662e-02 3.08346689e-01 -8.58750105e-01 1.12955523e+00
-3.51578355e-01 7.92328537e-01 -3.06305200e-01 -1.34587669e+00
8.22108507e-01 1.13132345e-02 -3.86649877e-01 -7.86670029e-01
1.33296415e-01 1.60093814e-01 4.70570236e-01 -5.63883483e-01
1.87860861e-01 -7.51715660e-01 -1.94677398e-01 7.64257669e-01
-7.35865161e-02 -3.79025251e-01 8.81503746e-02 1.61214411e-01
8.03863466e-01 -6.06345713e-01 6.54122412e-01 -6.20576918e-01
1.96212158e-01 1.87400296e-01 6.01094544e-01 1.39085591e+00
-1.67437792e-01 -2.41960436e-02 3.96502286e-01 -8.67028773e-01
-1.10065305e+00 -1.26455057e+00 -4.77121353e-01 1.48562956e+00
1.50988419e-02 1.40892819e-01 -5.12368083e-01 -3.92163604e-01
4.17952031e-01 1.40715146e+00 -9.85851705e-01 -3.85593772e-01
-1.32001325e-01 -6.03877604e-01 1.37352750e-01 6.38899028e-01
2.95457721e-01 -1.13012981e+00 -7.52324045e-01 2.51566321e-02
3.86584993e-03 -4.40313518e-01 3.51609066e-02 1.26109034e-01
-9.82721388e-01 -9.34318602e-01 -2.08819106e-01 -4.98369098e-01
9.13128078e-01 3.99962842e-01 1.55309820e+00 3.74058545e-01
-3.43223780e-01 5.16655147e-01 -2.25674063e-01 -7.92732418e-01
-8.59160572e-02 -3.29491764e-01 5.04966438e-01 -2.65173972e-01
1.30006897e+00 -6.48221314e-01 -4.41134900e-01 2.51647890e-01
-5.54964185e-01 -1.84196189e-01 8.28434229e-01 7.77461112e-01
1.35835052e-01 -3.55453085e-04 9.88713562e-01 -1.08523393e+00
1.09111166e+00 -9.27629232e-01 -3.11701208e-01 2.02098235e-01
-7.69621015e-01 7.30480850e-02 8.12371314e-01 -8.14808905e-01
-9.98052359e-01 -5.28962255e-01 4.10995722e-01 1.38124257e-01
-5.94302535e-01 3.71733963e-01 4.54715863e-02 1.14999920e-01
1.27065372e+00 2.13218689e-01 -1.09161884e-01 -4.26748842e-01
3.65635276e-01 9.62872148e-01 2.37467527e-01 -1.05436695e+00
8.82406354e-01 4.28306937e-01 -3.15848887e-01 -9.16906595e-01
-1.27245831e+00 -1.77842900e-01 -8.06893349e-01 -6.39143661e-02
7.58975744e-01 -8.07035029e-01 -1.09191716e+00 3.16719823e-02
-5.41802406e-01 -4.40294087e-01 -5.54657280e-01 7.43129492e-01
-6.48015678e-01 -3.04302692e-01 -3.42214078e-01 -1.05591488e+00
1.18385233e-01 -5.17794788e-01 2.63916939e-01 3.62157673e-01
-7.64883637e-01 -9.34442818e-01 -1.17250502e-01 3.13922495e-01
5.05340397e-01 4.50694785e-02 1.24171329e+00 -1.33265829e+00
-4.16387379e-01 -1.78858668e-01 -1.87385231e-01 6.17820248e-02
-9.32738930e-02 -5.75706363e-01 -1.05596793e+00 -4.74761516e-01
4.19855356e-01 -1.04702187e+00 8.04856002e-01 -4.42404523e-02
1.44029117e+00 -7.32932746e-01 -2.10305110e-01 3.60796899e-01
1.22219682e+00 1.75115719e-01 2.21768066e-01 5.26012434e-03
2.10974976e-01 1.00032365e+00 5.37352145e-01 3.97702426e-01
5.58927476e-01 1.50220484e-01 -4.16018218e-01 4.95222777e-01
3.18226755e-01 -4.60736930e-01 -4.28078920e-02 4.44407523e-01
-6.02474399e-02 5.15661985e-02 -9.41896319e-01 5.72118759e-01
-1.57605338e+00 -1.32232225e+00 1.77629948e-01 2.42251921e+00
1.02657437e+00 3.01065058e-01 1.29546702e-01 8.60318989e-02
6.11689985e-01 -1.18311867e-01 -1.02701819e+00 -6.13472462e-01
5.06064855e-02 1.58754826e-01 -3.66299860e-02 3.18860888e-01
-6.91323280e-01 6.48955107e-01 7.63488913e+00 3.33692551e-01
-7.62520134e-01 -1.85950533e-01 7.64276385e-01 -5.36194667e-02
-6.23065948e-01 1.11098997e-01 -3.81266564e-01 4.21337992e-01
1.07174993e+00 -5.72305202e-01 6.43200874e-01 1.04914474e+00
-2.72878319e-01 -3.04151833e-01 -1.76517868e+00 8.36286247e-01
1.58454657e-01 -1.07074440e+00 3.39890182e-01 1.91063136e-01
4.47472721e-01 -6.50899708e-01 1.19153500e-01 6.92731261e-01
7.08827078e-01 -1.45809722e+00 6.60291433e-01 8.05427611e-01
6.55709267e-01 -5.61563671e-01 5.69675565e-01 8.57730508e-01
-4.93949294e-01 -6.64625466e-01 -9.72978234e-01 -8.23438525e-01
-4.14487392e-01 4.60915148e-01 -8.91299069e-01 -2.99705714e-01
4.34444606e-01 2.26335600e-01 -9.71319556e-01 6.75403476e-01
-1.80808574e-01 5.69069862e-01 -3.48618813e-02 -4.73813385e-01
-9.81497318e-02 5.11395074e-02 1.64058387e-01 9.88116384e-01
1.31463483e-01 7.44200230e-01 1.10743374e-01 1.36309147e+00
-4.59321626e-02 -1.02543741e-01 -8.80841792e-01 -1.44193292e-01
1.05610919e+00 1.04687393e+00 -6.05849564e-01 -6.42655730e-01
-3.09708357e-01 5.17489493e-01 7.06680834e-01 2.92590708e-01
4.98075485e-02 -3.05657774e-01 5.57706952e-01 2.18626678e-01
-3.20350081e-01 -1.39848664e-01 -4.96274263e-01 -1.25323415e+00
-3.15155596e-01 -1.00368130e+00 4.51867610e-01 -7.68986940e-01
-1.94303405e+00 -7.90104941e-02 2.39842996e-01 -5.62773585e-01
-4.11389828e-01 -6.47235513e-01 -5.33284783e-01 7.92090416e-01
-8.94913673e-01 -7.03290403e-01 -3.93959522e-01 6.01212323e-01
6.71996057e-01 -1.69790804e-01 8.74418318e-01 -5.04141986e-01
-2.64840603e-01 5.48500776e-01 -1.53260045e-02 2.57837534e-01
7.95819461e-01 -1.54729676e+00 2.72298783e-01 3.33389640e-01
1.52736053e-01 1.27007413e+00 5.98845363e-01 -4.66301858e-01
-1.38902199e+00 -5.91697276e-01 7.11006582e-01 -1.00002098e+00
4.17344838e-01 -6.28904581e-01 -1.29775035e+00 8.33026946e-01
-1.32519290e-01 -1.47926003e-01 1.14003384e+00 6.54929519e-01
-8.17471623e-01 -9.71767902e-02 -1.55712020e+00 6.31176829e-01
1.37241697e+00 -8.12060237e-01 -1.58752167e+00 2.21491665e-01
5.87908208e-01 4.36123759e-01 -8.90611291e-01 -9.79940221e-02
6.10965788e-01 -1.11708653e+00 1.02539599e+00 -1.12950277e+00
3.68449867e-01 7.20487013e-02 -3.44106644e-01 -1.66466606e+00
-8.47443938e-01 -2.24719778e-01 1.35440007e-01 1.09888387e+00
2.11940080e-01 -9.47253168e-01 2.36462817e-01 1.15646756e+00
3.57504636e-01 -2.47153506e-01 -4.37575012e-01 -5.57436407e-01
5.43955326e-01 -1.53488159e-01 7.13163853e-01 1.37229991e+00
3.26043844e-01 3.66639644e-01 1.56839222e-01 -2.94779211e-01
1.16567075e+00 2.82282352e-01 7.05242157e-01 -1.85309327e+00
6.18209653e-02 -2.22209081e-01 -3.97238642e-01 -7.25214183e-01
2.57647723e-01 -8.66406262e-01 -6.05724379e-02 -1.00888014e+00
4.54530090e-01 -6.59994841e-01 -4.04029369e-01 1.90924361e-01
-3.83849949e-01 -4.00472395e-02 3.20519269e-01 4.04758662e-01
-5.10856450e-01 -7.16857836e-02 9.51154292e-01 1.65046994e-02
8.44738856e-02 -1.40205160e-01 -1.52579975e+00 1.08849585e+00
8.57259929e-01 -2.57360578e-01 -5.16870916e-01 -2.16879338e-01
5.66750824e-01 -1.25097066e-01 6.80968702e-01 -9.16417837e-01
3.35554510e-01 -7.54149318e-01 1.11444497e+00 -2.46356860e-01
1.25409320e-01 -5.60646772e-01 -1.68358713e-01 4.40677941e-01
-9.19328153e-01 -3.69839574e-04 -1.71620488e-01 5.44026852e-01
2.33279794e-01 -3.11281592e-01 9.13287759e-01 -7.14766920e-01
-9.39359307e-01 -1.47820428e-01 -4.75747645e-01 7.26440430e-01
9.00928140e-01 -2.82094240e-01 -6.10859156e-01 -2.31871083e-01
-5.59230745e-01 8.41991007e-02 7.30051041e-01 2.72712469e-01
6.50209725e-01 -1.24323249e+00 -7.57350385e-01 3.18128467e-01
4.51475680e-01 -2.73973078e-01 4.93803434e-02 5.17223954e-01
-6.01978898e-02 5.71710110e-01 -5.22372365e-01 -2.92710006e-01
-5.82567096e-01 1.10591340e+00 8.86500105e-02 4.77873176e-01
-4.50857252e-01 9.46646929e-01 6.12596035e-01 -3.26611787e-01
-5.40684955e-03 -1.48723066e-01 -6.70992136e-02 4.18147504e-01
9.07962680e-01 3.96784961e-01 -4.64134753e-01 -2.14307994e-01
-2.88938791e-01 1.62104368e-01 -3.29594135e-01 -1.36830643e-01
1.20464170e+00 -2.48327181e-01 -1.71028718e-01 1.11007249e+00
8.53227258e-01 -2.88928688e-01 -9.61184323e-01 -4.07659769e-01
1.52216911e-01 -1.03893054e+00 -4.14951801e-01 -9.67864811e-01
-8.25463235e-02 1.06073165e+00 3.04458767e-01 5.26988745e-01
5.99875391e-01 4.81788181e-02 8.15065280e-02 1.12527740e+00
7.56592333e-01 -1.41549277e+00 2.20534608e-01 9.59294140e-02
9.02282298e-01 -1.57857728e+00 9.79693383e-02 1.80974245e-01
-6.42264485e-01 8.04114103e-01 8.08340549e-01 -1.50868878e-01
3.44498843e-01 -2.89063036e-01 -2.58151948e-01 -2.50911981e-01
-1.29497051e+00 -1.52332932e-01 1.20089896e-01 1.04861259e+00
5.23468435e-01 3.75511020e-01 -2.94055253e-01 6.34336412e-01
-7.03771532e-01 -1.46294519e-01 7.22493470e-01 6.61764801e-01
-1.04551864e+00 -3.12846601e-01 -6.20194376e-01 9.74404395e-01
-8.41033682e-02 -3.85857075e-02 -7.87719250e-01 9.03016925e-01
1.37914971e-01 1.18646181e+00 5.06863415e-01 -3.40306222e-01
1.46089762e-01 4.68233287e-01 5.19506454e-01 -8.76353323e-01
-3.71692359e-01 -7.94388235e-01 -1.15483664e-01 -5.48459470e-01
-2.22355232e-01 -7.97678947e-01 -7.89886773e-01 -6.27542675e-01
3.92067730e-02 3.17932725e-01 1.57418683e-01 8.80139410e-01
-1.31860496e-02 -5.05248941e-02 7.40593493e-01 -5.98301709e-01
-9.69989777e-01 -1.10975730e+00 -7.08870053e-01 9.89669561e-01
3.99925679e-01 -7.85190225e-01 -9.49425578e-01 -2.19804853e-01] | [9.480953216552734, 6.551426887512207] |
20925217-6244-4fa6-b7c2-2c31d16d03b7 | optimal-algorithms-for-latent-bandits-with | 2301.07040 | null | https://arxiv.org/abs/2301.07040v3 | https://arxiv.org/pdf/2301.07040v3.pdf | Optimal Algorithms for Latent Bandits with Cluster Structure | We consider the problem of latent bandits with cluster structure where there are multiple users, each with an associated multi-armed bandit problem. These users are grouped into \emph{latent} clusters such that the mean reward vectors of users within the same cluster are identical. At each round, a user, selected uniformly at random, pulls an arm and observes a corresponding noisy reward. The goal of the users is to maximize their cumulative rewards. This problem is central to practical recommendation systems and has received wide attention of late \cite{gentile2014online, maillard2014latent}. Now, if each user acts independently, then they would have to explore each arm independently and a regret of $\Omega(\sqrt{\mathsf{MNT}})$ is unavoidable, where $\mathsf{M}, \mathsf{N}$ are the number of arms and users, respectively. Instead, we propose LATTICE (Latent bAndiTs via maTrIx ComplEtion) which allows exploitation of the latent cluster structure to provide the minimax optimal regret of $\widetilde{O}(\sqrt{(\mathsf{M}+\mathsf{N})\mathsf{T}})$, when the number of clusters is $\widetilde{O}(1)$. This is the first algorithm to guarantee such strong regret bound. LATTICE is based on a careful exploitation of arm information within a cluster while simultaneously clustering users. Furthermore, it is computationally efficient and requires only $O(\log{\mathsf{T}})$ calls to an offline matrix completion oracle across all $\mathsf{T}$ rounds. | ['Prateek Jain', 'Karthikeyan Shanmugam', 'Arun Sai Suggala', 'Soumyabrata Pal'] | 2023-01-17 | null | null | null | null | ['matrix-completion'] | ['methodology'] | [ 1.82333589e-03 3.61840129e-01 -4.67773616e-01 -1.76000252e-01
-1.12321770e+00 -1.08266830e+00 -1.29252285e-01 1.30375043e-01
-5.37696421e-01 6.43478930e-01 -2.65844047e-01 -7.07579374e-01
-8.24987471e-01 -8.69816542e-01 -1.02902353e+00 -1.00944972e+00
-5.68948030e-01 8.77454758e-01 -4.06288922e-01 1.30730286e-01
6.90755993e-02 1.89354196e-01 -1.05066597e+00 2.15122122e-02
9.05173838e-01 1.32851422e+00 2.62062430e-01 6.73773527e-01
-2.16277167e-02 6.83845580e-01 -2.49010995e-01 -5.83384752e-01
6.51751280e-01 -4.82451886e-01 -6.81233168e-01 5.08310258e-01
-1.20026022e-01 -2.71661460e-01 -5.74623346e-01 1.35073900e+00
6.86850119e-03 4.15289104e-01 4.68234330e-01 -9.03744936e-01
-4.81935054e-01 1.17156267e+00 -1.24190831e+00 9.05057341e-02
-6.84864819e-02 -1.43997997e-01 1.32057774e+00 -3.09257686e-01
3.13641690e-02 8.33787978e-01 8.19150135e-02 1.55098572e-01
-1.35654759e+00 -1.02212059e+00 6.00321770e-01 -3.40929687e-01
-1.14685321e+00 -3.22542787e-01 3.97800565e-01 -3.39943290e-01
1.73065484e-01 7.35246301e-01 4.79779363e-01 2.62298375e-01
-7.34953642e-01 1.18362427e+00 8.39521945e-01 -5.67278326e-01
6.14063025e-01 1.38742980e-02 3.64521623e-01 5.48064053e-01
3.09530348e-01 -2.37850770e-01 -1.68633878e-01 -2.37021074e-01
8.88457239e-01 5.79947352e-01 -1.99577555e-01 -1.99040428e-01
-9.42148805e-01 1.09730899e+00 2.75731862e-01 9.07892287e-02
-3.60384494e-01 6.25970364e-01 -9.97705907e-02 5.17507792e-01
4.83506024e-01 3.66051756e-02 -2.41254196e-01 -4.13146354e-02
-1.08831823e+00 1.72161683e-01 5.69103301e-01 1.38398314e+00
8.29431534e-01 -2.07580864e-01 -3.07261437e-01 8.07920992e-01
2.23319814e-01 9.06414986e-01 -2.94352978e-01 -1.42633450e+00
1.18694544e+00 3.97731215e-01 7.78539896e-01 -7.32106447e-01
-1.41969040e-01 -8.20301950e-01 -1.12565780e+00 -2.84949467e-02
6.67516649e-01 -5.04149258e-01 -6.17913365e-01 1.79639995e+00
9.55031365e-02 -1.59462348e-01 -3.38933796e-01 8.89610291e-01
-2.73404628e-01 7.43697584e-01 -5.02628803e-01 -8.73854458e-01
1.00340104e+00 -8.93601179e-01 -3.41331065e-01 -3.68403077e-01
6.13061965e-01 -6.98935509e-01 8.03593397e-01 6.60133958e-01
-1.72759259e+00 1.15515403e-01 -7.39491224e-01 6.78953528e-01
3.62271428e-01 5.72487339e-02 1.04786980e+00 1.19181681e+00
-7.47648060e-01 4.61673558e-01 -9.21198905e-01 3.09999853e-01
4.78260517e-01 7.90712714e-01 8.27462375e-02 -4.16754037e-01
-3.87406737e-01 -1.27148524e-01 1.42553806e-01 2.93796331e-01
-8.01649630e-01 -2.94994950e-01 -2.81067371e-01 2.20608935e-01
7.31083512e-01 -2.99340189e-01 1.11767220e+00 -7.36942887e-01
-1.15974975e+00 4.87360358e-01 -1.96137607e-01 -3.99992853e-01
5.72270811e-01 1.01044506e-01 3.95164825e-02 7.28260055e-02
2.26899758e-01 -7.51696676e-02 6.97895050e-01 -1.11629939e+00
-9.70634818e-01 -8.04135323e-01 2.89591908e-01 2.98490301e-02
-4.74202067e-01 1.91912241e-02 -7.95413256e-01 -5.65527380e-01
6.78145170e-01 -1.37350452e+00 -5.17273664e-01 -5.75673521e-01
-4.86338437e-01 -7.36978371e-03 -7.65531957e-02 -2.03898892e-01
1.37123561e+00 -2.22069454e+00 3.26469600e-01 9.19832826e-01
4.24893826e-01 -1.56744197e-01 1.41079754e-01 3.82695794e-01
1.63904503e-01 2.80338198e-01 -6.51013181e-02 -4.99504775e-01
1.66881561e-01 1.30464643e-01 -3.62163574e-01 7.15937078e-01
-1.04838002e+00 5.20747900e-01 -6.04821444e-01 2.02503428e-01
-9.96101298e-04 -3.51221323e-01 -8.16157937e-01 1.83963496e-02
-3.72772127e-01 3.43278676e-01 -9.12145436e-01 4.12163824e-01
9.69365895e-01 -7.30438590e-01 4.79666203e-01 4.21349734e-01
2.04860279e-03 -2.00947195e-01 -1.75820506e+00 1.33302605e+00
-2.95824766e-01 -1.45047143e-01 5.93956172e-01 -1.29502439e+00
2.16572672e-01 3.11724633e-01 7.64495075e-01 -4.73256975e-01
1.76383436e-01 3.40042621e-01 -3.34776580e-01 -9.94611457e-02
2.23026443e-02 -1.93696275e-01 -1.61036059e-01 1.11568403e+00
-4.38583195e-01 4.83561009e-01 3.06605130e-01 5.93533576e-01
1.33166778e+00 -5.16052663e-01 -3.40683192e-01 -9.52556059e-02
6.80853128e-02 -2.87172705e-01 4.89250690e-01 1.41211712e+00
1.33287966e-01 3.06057066e-01 7.17503309e-01 -2.58529872e-01
-1.04515278e+00 -1.14731205e+00 2.52666205e-01 1.47375119e+00
2.78137296e-01 -1.35061085e-01 -7.62712538e-01 -4.92624879e-01
1.15597315e-01 7.31098354e-01 -8.04368615e-01 3.08500737e-01
-2.86629796e-01 -1.03335249e+00 2.42455229e-01 2.22916469e-01
2.43592128e-01 -6.88210726e-01 -1.78052321e-01 3.61858249e-01
-2.51903236e-01 -6.33836508e-01 -7.88418531e-01 5.20024300e-01
-8.69579077e-01 -7.31916606e-01 -7.87559152e-01 -3.09121639e-01
9.30993855e-01 6.97023034e-01 5.94475806e-01 -5.60268275e-02
5.84645048e-02 2.50416607e-01 -5.06396413e-01 -2.01561883e-01
1.88088879e-01 5.74909598e-02 7.31372684e-02 2.81348854e-01
6.55828267e-02 -7.42058516e-01 -9.20156837e-01 3.82244200e-01
-9.27903950e-01 -6.56617358e-02 5.18753469e-01 5.42248547e-01
6.35662973e-01 8.34927857e-02 2.11637691e-01 -1.01918423e+00
3.62211585e-01 -4.94507194e-01 -1.07473946e+00 2.45735422e-01
-4.38313991e-01 -2.11666082e-03 5.29855371e-01 -3.06115836e-01
-6.61524177e-01 8.03729072e-02 3.32546443e-01 -5.51022768e-01
1.67655319e-01 4.65918332e-01 1.79060116e-01 3.11300367e-01
4.14952278e-01 3.17923039e-01 -1.26175433e-01 -7.10188329e-01
4.35875535e-01 5.90519249e-01 2.16926530e-01 -7.09666848e-01
7.22029805e-01 5.81494570e-01 -3.96759771e-02 -4.00451750e-01
-1.29271019e+00 -4.27358329e-01 -3.89437191e-02 1.89260729e-02
3.60929012e-01 -9.06939447e-01 -1.73610067e+00 -1.20245628e-01
-6.04029298e-01 -4.89223987e-01 -3.10684651e-01 6.77732229e-01
-7.41675436e-01 3.88735652e-01 -5.02882183e-01 -1.66525578e+00
-1.77578598e-01 -1.00944984e+00 4.56397563e-01 8.36131070e-03
1.50678590e-01 -5.11254549e-01 -5.12034774e-01 9.80215669e-01
1.57272235e-01 -1.15427129e-01 8.37110519e-01 -4.74359691e-01
-1.07388341e+00 -5.24896681e-01 -2.46771559e-01 2.45438561e-01
9.73665267e-02 -7.15574265e-01 -4.00192618e-01 -7.83889055e-01
-5.02334498e-02 -2.67653733e-01 6.88316464e-01 7.28893876e-01
1.58335268e+00 -1.01218498e+00 -4.05175030e-01 4.00745988e-01
1.26052928e+00 3.69380504e-01 3.57413620e-01 7.93202315e-03
3.24491322e-01 9.19429064e-02 4.24244791e-01 9.77421820e-01
8.66019875e-02 5.14984071e-01 6.54523611e-01 1.52672887e-01
6.75979972e-01 1.85082823e-01 2.78846025e-01 4.28411216e-01
-2.61164635e-01 -3.86089146e-01 -5.51326334e-01 5.92499554e-01
-2.16762519e+00 -9.75725710e-01 -1.42740503e-01 2.83077836e+00
7.73328781e-01 2.42254511e-01 3.23866934e-01 5.53003997e-02
7.33483911e-01 -1.39290154e-01 -6.41655266e-01 -5.44920415e-02
3.47624114e-03 2.48517379e-01 1.02928686e+00 5.21996260e-01
-8.18211615e-01 7.21937597e-01 4.20134687e+00 1.21160936e+00
-5.08392870e-01 1.80361688e-01 9.87811089e-01 -1.06836379e+00
-3.06585640e-01 2.41055503e-01 -6.90843761e-01 7.15748429e-01
7.76239455e-01 4.87058386e-02 1.30702960e+00 7.73962259e-01
3.86052132e-01 -3.49123210e-01 -1.07350707e+00 9.70224559e-01
-3.06492835e-01 -1.38364553e+00 -4.62798893e-01 6.08565390e-01
1.06068611e+00 2.87333895e-02 4.88916278e-01 6.40934408e-02
1.07000458e+00 -9.73646879e-01 7.61659622e-01 1.38391495e-01
7.67248333e-01 -1.15989065e+00 2.10117593e-01 8.48022461e-01
-1.03323698e+00 -7.41371393e-01 -3.75700861e-01 -1.27398625e-01
1.30716071e-01 9.34105694e-01 -1.86827600e-01 5.26303470e-01
9.01982367e-01 -1.00992564e-02 3.27306658e-01 9.08968389e-01
1.80864558e-01 6.76105857e-01 -7.58420944e-01 6.72011748e-02
5.27348995e-01 -8.39298964e-01 4.07415122e-01 7.09322989e-01
5.10815561e-01 7.02893674e-01 6.61042929e-01 5.29908240e-01
-3.89615655e-01 1.31511822e-01 -3.92169086e-03 -9.18230936e-02
6.41282141e-01 1.02412200e+00 -9.26064849e-01 -2.62632966e-01
-1.04054764e-01 8.95114720e-01 3.93172473e-01 5.88337660e-01
-8.36231232e-01 -2.54639059e-01 5.35935760e-01 2.83648700e-01
6.21358335e-01 -1.83167681e-01 -6.16346896e-01 -9.12970543e-01
1.98171616e-01 -4.29032713e-01 6.31311834e-01 -2.15962067e-01
-1.17461812e+00 7.36560300e-02 -2.78559536e-01 -1.04434979e+00
1.31699726e-01 -7.48370141e-02 -9.52753797e-02 8.05954397e-01
-8.67867053e-01 -8.39568734e-01 1.70171008e-01 7.65314043e-01
1.86557338e-01 -2.86681324e-01 6.39369607e-01 3.35795283e-01
-6.36714160e-01 9.67690825e-01 1.09794140e+00 1.29405474e-02
1.54017672e-01 -9.89164531e-01 -3.70877445e-01 4.92038429e-01
1.01997398e-01 7.90313542e-01 6.90503478e-01 -3.98564130e-01
-1.54394603e+00 -1.07832956e+00 3.83309871e-01 -1.49089128e-01
5.53744256e-01 -4.24794674e-01 -1.86342627e-01 1.05813980e+00
-9.40238908e-02 -4.87180129e-02 9.45560992e-01 2.51964599e-01
-2.44607314e-01 -3.25656235e-01 -1.11974525e+00 6.47075117e-01
1.09713638e+00 -5.23751117e-02 3.06232810e-01 7.24071264e-01
5.04476488e-01 -3.14871490e-01 -9.03531075e-01 1.51836732e-02
4.16539669e-01 -9.91057992e-01 7.17496932e-01 -5.97149849e-01
-1.42910779e-02 -1.27072796e-01 -4.92190719e-01 -7.45568991e-01
-5.15884936e-01 -1.12130070e+00 -4.33153152e-01 9.00053978e-01
8.76728952e-01 -4.58782881e-01 1.31901729e+00 8.89522135e-01
1.61739931e-01 -6.04331017e-01 -1.15574121e+00 -6.37125969e-01
2.53906697e-01 -5.56996107e-01 4.07103181e-01 7.48628795e-01
1.88541993e-01 3.63285169e-02 -7.15180278e-01 3.06831658e-01
9.33522642e-01 4.38333482e-01 6.59038305e-01 -1.02836108e+00
-1.04220533e+00 -3.70329738e-01 4.69724685e-01 -1.44339812e+00
-4.44387943e-01 -8.57222617e-01 -3.02549869e-01 -1.33201170e+00
6.62822664e-01 -1.21820843e+00 -5.59246421e-01 6.87619925e-01
1.40708908e-01 4.08952881e-04 4.17938352e-01 4.40245539e-01
-1.10303390e+00 2.51484841e-01 9.71895039e-01 -9.98727530e-02
-4.90070224e-01 7.72571206e-01 -1.01046646e+00 5.14655888e-01
5.97539961e-01 -6.72225416e-01 -4.31488425e-01 -6.13104880e-01
6.63603425e-01 9.03156698e-01 2.24697907e-02 -5.39954126e-01
2.76386738e-01 -4.18244451e-01 2.21980855e-01 -7.20092535e-01
3.96718204e-01 -9.13983107e-01 2.62253612e-01 4.20416534e-01
-4.64140803e-01 -3.27124089e-01 -2.82737613e-01 1.02480948e+00
3.90955240e-01 -2.98000038e-01 7.46787310e-01 -3.36255074e-01
5.66147447e-01 5.18016636e-01 -4.45018113e-01 -1.32023752e-01
1.27763700e+00 -1.04151182e-01 9.79508460e-02 -7.84004390e-01
-9.92644727e-01 6.34025991e-01 1.41011730e-01 -1.95314974e-01
1.06206588e-01 -1.11315060e+00 -5.00406504e-01 -1.37740552e-01
-2.81057477e-01 2.47707590e-01 6.74145460e-01 9.67389107e-01
1.79713033e-02 5.15452862e-01 4.55616385e-01 -4.74119365e-01
-8.57807517e-01 8.08134973e-01 7.93439746e-02 -4.98685628e-01
-1.86152205e-01 1.25911450e+00 6.81671277e-02 -1.23358898e-01
4.77733642e-01 9.72656682e-02 4.68016624e-01 3.85901779e-02
4.33348060e-01 7.18684793e-01 -1.24489211e-01 9.35999677e-03
1.13772891e-01 5.29704541e-02 -5.56527913e-01 -5.14968276e-01
1.50107956e+00 -4.17137116e-01 -3.09715688e-01 1.23962685e-01
1.04758310e+00 2.37777576e-01 -1.29079616e+00 -4.55206394e-01
-9.44904834e-02 -6.42300546e-01 -1.57845810e-01 -6.55520380e-01
-1.27389944e+00 6.93701029e-01 5.43744981e-01 6.79016769e-01
9.81758654e-01 1.82483181e-01 6.90311372e-01 3.45539272e-01
8.14533591e-01 -1.21392024e+00 1.71945378e-01 3.12851071e-01
3.75649810e-01 -9.43065405e-01 -8.97543654e-02 -2.56234139e-01
-5.09102404e-01 6.26166940e-01 5.43279015e-02 -1.26841724e-01
5.51531672e-01 -1.13083385e-01 -6.59925640e-01 -4.41738591e-02
-3.09851974e-01 -5.81672527e-02 -1.83076128e-01 -1.72648475e-01
2.05178931e-01 4.99098212e-01 -1.78060755e-01 1.03758311e+00
-8.05654377e-02 -8.93014148e-02 3.38940233e-01 9.34456229e-01
-7.32101619e-01 -1.34500992e+00 -6.17818117e-01 9.78928924e-01
-7.80872345e-01 4.93428595e-02 2.32382730e-01 4.84178513e-02
-8.50708708e-02 1.36901045e+00 -8.16223174e-02 1.91333089e-02
6.15244061e-02 -1.95591494e-01 4.55559641e-01 -5.72916925e-01
-2.71374613e-01 7.59840548e-01 -1.30698785e-01 -2.20428422e-01
7.07334056e-02 -6.34162307e-01 -1.25169837e+00 -6.87949955e-01
-2.83441722e-01 6.07845843e-01 3.93252343e-01 1.02259874e+00
2.01106414e-01 1.24753453e-01 1.25773752e+00 -5.36789715e-01
-7.80405402e-01 -7.72061169e-01 -1.20325768e+00 1.58753887e-01
9.10797119e-02 -2.88933277e-01 -2.31068477e-01 -2.40758002e-01] | [4.726862907409668, 3.490882396697998] |
f40cec07-51da-4ebc-a604-c10d8dc7a2e0 | conceptual-edits-as-counterfactual | null | null | http://ceur-ws.org/Vol-3121/paper6.pdf | http://ceur-ws.org/Vol-3121/paper6.pdf | Conceptual Edits as Counterfactual Explanations | We propose a framework for generating counterfactual explanations of black-box classifiers, which answer the question “What has to change for this to be classified as X instead of Y?” in terms of given domain knowledge. Specifically, we identify minimal and meaningful “concept edits” which, when applied, change the prediction of a black-box classifier to a desired class. Furthermore, by accumulating multiple counterfactual explanations from interesting regions of a dataset, we propose a method to estimate a "global" counterfactual explanation for that region and a desired target class. We implement algorithms and show results from preliminary experiments employing CLEVR-Hans3 and COCO as datasets. The resulting explanations were useful, and even managed to unintendedly reveal a bias in the classifier’s training set, which was unknown to us. | ['Giorgos Stamou1', 'Edmund Dervakos1', 'Konstantinos Thomas', 'Giorgos Filandrianos'] | 2022-03-23 | null | null | null | aaai-make-2022-3 | ['counterfactual-explanation'] | ['miscellaneous'] | [ 6.04743361e-01 1.00617468e+00 -3.42307448e-01 -6.84990227e-01
-6.38150334e-01 -6.31566942e-01 8.13472211e-01 2.25541070e-01
-1.36411622e-01 1.54732227e+00 2.51841873e-01 -7.44059682e-01
-2.00628847e-01 -8.58515263e-01 -1.11154616e+00 -5.97725868e-01
1.20084308e-01 4.39937502e-01 -1.57853767e-01 2.34189600e-01
4.47531104e-01 3.12717915e-01 -1.72561574e+00 6.97843432e-01
9.44555581e-01 5.29882908e-01 -1.80473715e-01 4.19230521e-01
1.20857758e-02 6.38075054e-01 -8.32750857e-01 -8.82224560e-01
1.87067956e-01 -7.18884647e-01 -9.35743034e-01 2.02157021e-01
5.81740916e-01 5.19956425e-02 2.70112813e-01 1.07536697e+00
-9.07912850e-02 1.47663265e-01 9.57543790e-01 -1.50577724e+00
-6.53827906e-01 1.15153265e+00 -2.55752236e-01 -1.54140979e-01
3.23133379e-01 3.18262100e-01 1.13406336e+00 -5.61326683e-01
8.73909831e-01 1.42790616e+00 4.83083665e-01 9.41959083e-01
-1.56568587e+00 -8.24897945e-01 4.93519366e-01 2.36927032e-01
-7.92683542e-01 -3.17834139e-01 6.08296692e-01 -2.28528544e-01
7.77081609e-01 9.06025827e-01 6.14892483e-01 1.24465621e+00
4.58862484e-01 5.69584012e-01 1.24756753e+00 -5.05141914e-01
7.35876203e-01 6.07019722e-01 -1.53235510e-01 3.77564132e-01
7.36281455e-01 5.90595782e-01 -5.60598195e-01 -4.65256095e-01
-7.19655529e-02 -1.20600894e-01 -3.05952996e-01 -6.45741403e-01
-1.18834889e+00 1.11291361e+00 4.84744847e-01 -2.53724549e-02
-3.74028444e-01 1.49002954e-01 7.58634657e-02 3.00788343e-01
5.73515594e-01 1.13461459e+00 -1.09367108e+00 2.15677261e-01
-6.40916884e-01 6.67873681e-01 7.87358522e-01 7.10296392e-01
6.58163369e-01 -3.04831564e-01 -1.58862039e-01 2.10354298e-01
-1.71101298e-02 3.22252721e-01 5.25382698e-01 -9.58828151e-01
3.48302782e-01 7.78922498e-01 3.38065684e-01 -5.91452360e-01
-1.38673097e-01 -2.47449964e-01 -4.65439469e-01 4.21498477e-01
5.81366777e-01 -2.97706366e-01 -8.51264596e-01 1.82483530e+00
4.27358717e-01 -3.67249921e-02 3.55767071e-01 7.67979741e-01
2.93502629e-01 2.06950232e-01 1.34080127e-01 -5.18290401e-01
9.14369702e-01 -5.08615792e-01 -5.09790957e-01 -3.41874450e-01
9.83187020e-01 -3.37248147e-01 1.01799834e+00 3.41654450e-01
-7.19493330e-01 -4.11248803e-01 -1.07778835e+00 4.30503398e-01
-4.21696812e-01 -1.28664553e-01 1.03150940e+00 7.40905881e-01
-4.43967998e-01 8.11958790e-01 -2.99561799e-01 -1.25833377e-01
5.79864800e-01 1.14429861e-01 -2.24029407e-01 1.96084250e-02
-1.18609858e+00 8.78623068e-01 7.92537153e-01 -3.51164669e-01
-1.05444133e+00 -1.15732503e+00 -6.32327914e-01 3.25934291e-01
7.73425579e-01 -8.67353261e-01 1.30137634e+00 -1.50337470e+00
-9.04772460e-01 7.26418614e-01 -3.26279491e-01 -1.04212153e+00
8.53413701e-01 7.73768649e-02 -4.01102692e-01 -3.10259104e-01
3.71915132e-01 8.34338784e-01 7.52244234e-01 -1.32576883e+00
-1.03441024e+00 -5.75983047e-01 1.82525963e-01 1.11401945e-01
3.35728854e-01 -3.00604790e-01 6.23838842e-01 -8.05472076e-01
-1.76619794e-02 -9.60515440e-01 -5.11682808e-01 -1.36371449e-01
-8.14083993e-01 -9.67801288e-02 8.23415637e-01 -2.25420222e-01
7.46088862e-01 -1.81971920e+00 -3.33038330e-01 2.29969889e-01
1.08799547e-01 -2.28147671e-01 3.45534831e-01 -2.57302612e-01
-7.14056194e-01 7.48305202e-01 -4.87925678e-01 2.99778074e-01
-2.47329459e-01 1.58156753e-01 -8.48610222e-01 1.78399697e-01
2.42010206e-01 5.83243728e-01 -9.56389666e-01 -6.19354919e-02
2.84615383e-02 -4.11423951e-01 -7.78920770e-01 8.80957544e-02
-6.62814796e-01 2.28931382e-01 -1.70296937e-01 1.99534714e-01
7.33178020e-01 2.33150661e-01 4.66069639e-01 3.81301343e-01
3.17228585e-02 5.38253844e-01 -1.10134828e+00 1.03016722e+00
-3.29854429e-01 5.50342083e-01 -5.38936317e-01 -9.69431818e-01
8.86445642e-01 1.60711080e-01 -4.08653095e-02 -1.49921298e-01
-2.88400389e-02 2.36772895e-01 1.00159466e-01 -2.90320635e-01
3.42067450e-01 -6.47932053e-01 -5.06771803e-02 5.40260434e-01
-1.55706167e-01 -1.86588466e-01 -1.28672048e-01 3.09022404e-02
1.00113332e+00 -8.84629339e-02 6.92291319e-01 -3.79263818e-01
2.94228166e-01 5.76274574e-01 7.98066914e-01 1.08840346e+00
2.31397226e-02 5.38662493e-01 9.05815125e-01 -5.68574071e-01
-8.80702794e-01 -9.20919120e-01 -4.16289046e-02 5.46099126e-01
-1.88863784e-01 1.45160645e-01 -5.76634824e-01 -1.61808431e+00
4.12176281e-01 1.84866536e+00 -1.13905883e+00 -5.34884691e-01
-1.11096255e-01 -7.17715919e-01 1.33943960e-01 2.53373772e-01
2.96632349e-01 -9.34180498e-01 -8.49563956e-01 1.73312090e-02
-4.60570529e-02 -3.70319992e-01 -2.31943488e-01 3.69763136e-01
-1.14427471e+00 -1.73321021e+00 -3.08429807e-01 -1.00704003e-02
9.39566672e-01 -2.72812899e-02 1.04992568e+00 4.85924929e-02
-1.09547772e-01 -3.06642592e-01 -1.77605852e-01 -9.14997995e-01
-7.28616059e-01 -3.06072563e-01 1.54826837e-02 -1.11125432e-01
4.64856416e-01 -2.06377715e-01 -2.70583838e-01 2.79327542e-01
-5.34127533e-01 1.36259526e-01 4.64129716e-01 1.10072458e+00
3.48460823e-01 2.30691046e-01 7.93725669e-01 -1.76078069e+00
5.59496045e-01 -5.51613927e-01 -6.66599393e-01 6.17167890e-01
-1.07697928e+00 4.00124878e-01 7.61859834e-01 -4.80198324e-01
-1.68652618e+00 1.36460245e-01 4.56338882e-01 -1.71178371e-01
-4.47257131e-01 2.73442447e-01 -4.98055518e-01 6.17693484e-01
1.08502829e+00 -1.24042235e-01 -1.90673575e-01 -2.99271643e-01
7.48547196e-01 5.02590060e-01 5.94042122e-01 -5.93087554e-01
5.97441435e-01 6.18717730e-01 -1.10851064e-01 -3.30761746e-02
-1.03312171e+00 4.82582711e-02 -6.53223217e-01 1.27338141e-01
4.79589462e-01 -5.54773033e-01 -7.06962526e-01 -1.59828797e-01
-1.16230989e+00 -1.56687334e-01 -8.20995331e-01 5.21624506e-01
-5.59809566e-01 -2.28567600e-01 6.33144677e-01 -9.93268490e-01
2.07383901e-01 -8.34437430e-01 4.79423761e-01 2.47543812e-01
-6.67945027e-01 -8.91270220e-01 -2.80146837e-01 3.97496074e-01
-1.15034774e-01 3.87310386e-01 1.16808009e+00 -1.14311695e+00
-4.99259770e-01 -2.49357268e-01 -1.57458726e-02 8.25076774e-02
1.30964831e-01 1.62378430e-01 -1.14682531e+00 2.42552022e-03
5.21190725e-02 -5.62501177e-02 9.32965517e-01 4.65552002e-01
1.50823438e+00 -9.09204364e-01 -7.54052460e-01 2.13374391e-01
9.83927965e-01 5.37212908e-01 3.82008880e-01 1.81508467e-01
3.78066972e-02 8.40891182e-01 9.95692730e-01 3.47264707e-01
1.53080255e-01 6.03149414e-01 4.49386150e-01 4.91210297e-02
1.13876566e-01 -6.28149927e-01 2.03307226e-01 -5.12739301e-01
2.30861098e-01 5.33828847e-02 -5.43603003e-01 1.00603104e+00
-1.83308864e+00 -1.26735997e+00 -1.64298445e-01 2.36018109e+00
8.67347538e-01 2.83958077e-01 -7.69283473e-02 2.10581031e-02
7.49398112e-01 -2.53821313e-01 -9.75606918e-01 -7.58820057e-01
-9.42122042e-02 -7.95451272e-03 6.59707725e-01 5.25385380e-01
-7.30980098e-01 7.23160803e-01 6.54250574e+00 5.46779513e-01
-8.82074177e-01 -3.78077291e-02 1.08346415e+00 -3.02921355e-01
-1.01757407e+00 6.72038376e-01 -6.88012362e-01 3.62896502e-01
9.51108217e-01 -8.43858242e-01 2.44519919e-01 1.10797870e+00
3.31983387e-01 -4.19850409e-01 -1.70286489e+00 1.79031909e-01
-1.56041384e-01 -1.84379542e+00 3.66147071e-01 9.33573395e-02
1.04528940e+00 -6.34960175e-01 -2.15929896e-01 3.09507430e-01
9.37098444e-01 -1.10040736e+00 1.00147247e+00 4.87382263e-01
6.21498168e-01 -9.77717340e-01 8.26459169e-01 5.89482784e-01
-4.10652101e-01 -5.44365048e-01 -4.78640497e-01 -4.06767517e-01
-4.98411179e-01 8.09967220e-01 -1.48520589e+00 5.05462050e-01
5.37662804e-01 3.25782180e-01 -6.73189521e-01 7.39623070e-01
-7.81135678e-01 6.33795261e-01 1.91235289e-01 -2.10362837e-01
-1.04860485e-01 2.75206536e-01 6.03904605e-01 8.31775188e-01
3.24513912e-01 1.16113462e-01 -3.36984962e-01 1.46209788e+00
-2.61666894e-01 -3.02666426e-01 -1.03751206e+00 2.91817307e-01
7.25640416e-01 7.37453282e-01 -5.03450871e-01 -6.74364924e-01
6.47939295e-02 6.87333047e-01 5.22427000e-02 2.57904649e-01
-7.31599331e-01 -3.03976297e-01 8.61591876e-01 -7.24567324e-02
2.08709985e-02 8.24788570e-01 -8.86704743e-01 -1.29491663e+00
-5.55149131e-02 -9.04997706e-01 7.85041809e-01 -8.91127825e-01
-1.03134823e+00 -2.68384330e-02 2.06840694e-01 -8.96276295e-01
-4.11037594e-01 -4.17394459e-01 -9.65736270e-01 7.73284793e-01
-1.06610000e+00 -6.28049910e-01 1.36625797e-01 1.52312830e-01
5.72753251e-01 -2.07496867e-01 8.51133466e-01 -5.90946496e-01
-1.37175426e-01 5.78447819e-01 -8.51658881e-02 -2.71895498e-01
5.75922132e-01 -1.55661571e+00 3.88044119e-01 9.30597961e-01
2.59755373e-01 7.09947109e-01 1.14334512e+00 -8.83934915e-01
-6.24045193e-01 -1.22218382e+00 1.30551851e+00 -5.96359134e-01
3.64040643e-01 -1.58867195e-01 -7.43294477e-01 8.83667290e-01
-6.85464516e-02 -4.00632322e-01 5.84751427e-01 4.23694164e-01
-4.55363363e-01 -2.23400630e-02 -1.63627183e+00 7.39322364e-01
1.00905097e+00 -1.14552774e-01 -1.20297527e+00 3.30131322e-01
8.39247406e-01 -2.35931903e-01 -2.08526522e-01 8.92530084e-02
4.23776776e-01 -1.00764441e+00 6.76933587e-01 -1.38854074e+00
6.89600646e-01 -3.04884762e-01 -8.03750828e-02 -1.98577428e+00
1.43809482e-01 -4.84984905e-01 2.06895784e-01 1.05311334e+00
1.00364614e+00 -8.32178593e-01 8.13713253e-01 1.09728158e+00
1.32105008e-01 -4.73155081e-01 -1.10706234e+00 -6.63709164e-01
1.91729635e-01 -4.94700581e-01 1.17293513e+00 1.20767725e+00
3.13177615e-01 5.99878170e-02 -1.45000055e-01 1.61163986e-01
7.18072236e-01 5.45371592e-01 9.18041229e-01 -1.03821981e+00
-3.59936431e-02 -2.66618282e-01 -6.69222549e-02 -2.98803329e-01
3.84926945e-01 -9.92173135e-01 2.56657004e-02 -9.91981626e-01
4.84528303e-01 -2.94383556e-01 4.29022089e-02 6.86375439e-01
-3.81892055e-01 -3.88615936e-01 2.25217730e-01 -1.66723490e-01
2.90551083e-03 4.16042238e-01 9.44975913e-01 -2.20231578e-01
-1.28708959e-01 3.39322269e-01 -1.54101968e+00 8.82305980e-01
7.48700917e-01 -9.56964850e-01 -3.47623199e-01 2.96370555e-02
1.84524760e-01 1.13736331e-01 7.83588111e-01 -3.95661861e-01
-1.25803053e-01 -7.54596591e-01 6.47074044e-01 -3.01549584e-01
-1.31361246e-01 -9.29639161e-01 3.71599108e-01 8.57230961e-01
-1.12739885e+00 -3.11698198e-01 9.33905914e-02 7.72469282e-01
-1.46354325e-02 -4.01441127e-01 7.31927752e-01 -2.67166078e-01
-4.35803592e-01 -3.60744417e-01 -2.73970813e-01 -1.79919749e-02
1.33031368e+00 -2.05733720e-02 -6.23761058e-01 -3.26286376e-01
-7.32399821e-01 3.13749164e-01 4.73644972e-01 3.64608139e-01
5.71089447e-01 -1.18380833e+00 -8.36862445e-01 1.88548923e-01
2.56749630e-01 -1.21055670e-01 5.39787002e-02 1.74157396e-01
9.69921276e-02 6.58010781e-01 -1.30242661e-01 -5.39506376e-02
-1.24802125e+00 6.37805402e-01 5.61206937e-01 -2.13970095e-01
-3.79641414e-01 7.98272371e-01 2.96242207e-01 -7.38800347e-01
-1.93504244e-01 -5.09038985e-01 4.67393957e-02 3.99703495e-02
4.47900176e-01 1.91756606e-01 -1.73037909e-02 4.23820093e-02
-3.25153410e-01 -3.88065845e-01 -3.23820300e-02 -5.32996878e-02
1.30804992e+00 2.84608714e-02 1.51610896e-01 3.67878139e-01
6.89822853e-01 5.48389591e-02 -1.17725444e+00 9.18232575e-02
4.24564034e-01 -1.01407588e+00 -3.10138375e-01 -1.40850616e+00
-6.88371837e-01 7.15357661e-01 2.40875810e-01 2.15651289e-01
8.29439282e-01 1.33488551e-01 -1.14068598e-01 5.26363075e-01
1.75511867e-01 -9.86768186e-01 -4.95428145e-01 -2.30164733e-02
1.24356711e+00 -1.30867624e+00 2.25416586e-01 -2.71458387e-01
-9.47764635e-01 1.03539348e+00 6.62476599e-01 2.57446647e-01
2.97064155e-01 -2.75094926e-01 -7.07243830e-02 -1.61894247e-01
-1.43868458e+00 1.48552030e-01 1.88997030e-01 7.19953537e-01
3.27324122e-01 5.83344936e-01 -3.02286863e-01 6.66329145e-01
-6.28414452e-01 1.34492517e-01 8.57709229e-01 3.45924288e-01
-2.86626220e-01 -7.15176284e-01 -5.91420591e-01 8.24076593e-01
-3.15679222e-01 5.93274608e-02 -8.57281744e-01 1.08698559e+00
3.78326118e-01 9.56951261e-01 6.06903955e-02 -2.38690823e-01
3.09006751e-01 3.23966295e-01 1.47234306e-01 -8.30422878e-01
-4.81007069e-01 -4.86435533e-01 4.47750181e-01 -3.96719128e-01
-2.29820609e-01 -9.98222053e-01 -1.22357070e+00 -2.31281728e-01
-4.55994546e-01 4.67165053e-01 5.66480875e-01 1.25288284e+00
2.32173562e-01 4.20819044e-01 7.23246932e-01 -1.83905244e-01
-8.30940247e-01 -9.55442786e-01 -4.03927565e-01 5.33552825e-01
2.57639498e-01 -5.88165164e-01 -5.89566708e-01 1.30944282e-01] | [8.67680835723877, 5.603013515472412] |
d298b05e-16cb-4428-98a5-e01855fc6b1a | object-adaptive-lstm-network-for-real-time | 2002.02598 | null | https://arxiv.org/abs/2002.02598v1 | https://arxiv.org/pdf/2002.02598v1.pdf | Object-Adaptive LSTM Network for Real-time Visual Tracking with Adversarial Data Augmentation | In recent years, deep learning based visual tracking methods have obtained great success owing to the powerful feature representation ability of Convolutional Neural Networks (CNNs). Among these methods, classification-based tracking methods exhibit excellent performance while their speeds are heavily limited by the expensive computation for massive proposal feature extraction. In contrast, matching-based tracking methods (such as Siamese networks) possess remarkable speed superiority. However, the absence of online updating renders these methods unadaptable to significant object appearance variations. In this paper, we propose a novel real-time visual tracking method, which adopts an object-adaptive LSTM network to effectively capture the video sequential dependencies and adaptively learn the object appearance variations. For high computational efficiency, we also present a fast proposal selection strategy, which utilizes the matching-based tracking method to pre-estimate dense proposals and selects high-quality ones to feed to the LSTM network for classification. This strategy efficiently filters out some irrelevant proposals and avoids the redundant computation for feature extraction, which enables our method to operate faster than conventional classification-based tracking methods. In addition, to handle the problems of sample inadequacy and class imbalance during online tracking, we adopt a data augmentation technique based on the Generative Adversarial Network (GAN) to facilitate the training of the LSTM network. Extensive experiments on four visual tracking benchmarks demonstrate the state-of-the-art performance of our method in terms of both tracking accuracy and speed, which exhibits great potentials of recurrent structures for visual tracking. | ['Yang Hua', 'Yan Yan', 'Si Chen', 'Yihan Du'] | 2020-02-07 | null | null | null | null | ['real-time-visual-tracking'] | ['computer-vision'] | [-0.15929487 -0.76066136 -0.45058805 -0.04235404 -0.29333937 -0.30972752
0.38818398 -0.44217598 -0.45766595 0.49329245 -0.26862985 -0.03073559
0.04021067 -0.7599547 -0.68382406 -0.9243842 0.20299062 0.14105883
0.60464317 -0.03690377 -0.04143897 0.6779602 -1.5493743 -0.21455427
0.8419731 1.3638592 0.12261043 0.17796558 -0.30877268 0.6979043
-0.5744615 -0.29067537 0.29339457 -0.22314037 0.04752588 0.18252303
0.73779285 -0.39419416 -0.79440325 1.2575809 0.57413435 0.16161494
0.25880212 -1.4292203 -0.59883726 0.17177503 -0.6308755 0.34658098
-0.08542781 0.4797443 0.6186104 -0.9543212 0.5367285 1.1179857
0.93644017 0.7771083 -0.9426059 -1.2265853 0.45130792 0.20957229
-1.2761008 -0.43167493 0.9424328 -0.37934288 0.49939057 -0.06013916
1.0202267 1.0533631 0.25634363 1.1274886 0.63649184 -0.06838635
-0.09771087 -0.16172743 -0.12705502 1.1046991 0.32753628 0.5435083
-0.4576368 0.04166682 1.1169537 0.6189018 -0.3025776 -0.52432835
-1.2728947 0.6953873 0.8692025 0.27079672 -0.36327 0.45517805
0.6668405 0.23343077 0.34500661 -0.1681963 -0.31242877 0.16396558
-1.0521727 0.16137132 0.07980796 0.98297876 0.34058535 0.7509897
-0.59507406 0.6066545 0.5591147 0.8491382 0.7315487 -0.47537574
0.41285768 0.7385316 0.10836586 -1.1340905 -0.3517415 -0.85447985
-1.1400754 0.34132743 0.4218147 -0.00678603 -1.2250079 1.8627956
0.52797294 0.26979482 -0.27380922 0.9421208 0.80078554 0.47431156
0.22130713 -0.30104098 1.1373764 -1.0135033 -0.90652376 -0.01045271
0.21047123 -0.623486 0.67124283 0.02923048 -0.8305364 -0.9917267
-0.9909762 0.39781198 -0.16530614 0.4416881 0.80196065 0.5569888
-0.70336765 0.48693296 -1.1746732 -0.0547629 0.9552349 0.6346572
-0.02031286 0.15709418 -0.9926623 0.49791053 0.44635734 0.44595256
-0.8208173 -0.5924496 -0.827127 0.04090089 0.43644866 -0.76126504
1.0141759 -1.111905 -1.5987418 0.38055345 -0.16632451 -0.42887622
0.63901955 -0.10610505 -0.4030721 -0.21117699 -0.06230294 0.72356534
1.3466542 -0.7900817 -0.76048934 -0.27407342 -0.3001041 -0.13190535
-0.5966185 0.02766289 -0.6600122 -1.105973 0.22518732 -0.94131124
-0.13914768 0.62083787 -0.21621293 -0.36504498 1.2022908 -0.28372103
1.1886239 -1.9970531 0.02786111 -0.02349311 0.25441918 0.73264694
-0.08211596 -0.15917343 0.30071005 -0.3380857 0.16192268 -0.2577277
-0.02011605 0.13230172 -0.30029148 0.6303904 0.20635392 1.2688879
-0.7862251 -0.7675108 0.41111007 0.5191292 -0.34832436 0.1607694
-0.37888965 0.42599648 -0.81043774 0.9287399 0.62527514 -0.48853442
-0.10367098 -0.3887105 -0.18405782 -0.24480441 -1.0244651 1.5051271
-0.24264777 0.5334493 -0.14397852 -0.8078178 0.9731923 0.14901511
0.5062233 -0.8795056 0.33747914 0.1597996 0.0842085 -0.13958396
0.34726933 -0.11920834 0.18819565 0.12237506 0.02036875 0.6132087
0.07771237 -0.07451863 0.7987318 0.45185736 0.01569585 0.10471575
0.6495973 -0.09252559 1.0367076 0.54933846 -0.44107577 0.12683159
-0.25427192 -0.7846588 -0.75333756 -0.84175235 -0.04055528 0.8347553
0.41409218 -0.03444802 -0.340493 -0.796475 0.14135025 0.00707281
-0.5958944 -0.3594378 -0.8060323 -0.5523903 0.44913444 0.9592908
0.7072044 -1.2876129 -0.68267375 0.40538907 0.17934887 -0.9474516
-0.69522625 -0.25155082 -1.1640563 -0.9550816 -0.9252493 -0.85083956
0.7945677 0.32569718 0.5727215 0.42301178 -0.22218226 0.08077317
-0.14653781 -0.27722883 -0.1127554 0.04476854 0.18766119 0.07530329
0.21555303 -0.1322084 -0.6777956 0.4516573 -0.76034963 -0.12223472
0.7099423 1.2025561 0.78222644 -0.01407357 0.5646353 -0.5459737
0.08018749 -0.09191306 -1.098741 0.29525316 -0.49478957 0.04099854
0.9404089 -0.894788 -0.9754018 0.31816715 -0.06602708 -1.1825508
0.32564083 0.21768408 0.02800927 -0.669284 0.12598026 0.5759329
0.08540054 -0.3375206 0.07212441 0.09877941 0.4983307 -0.2394973
1.3151962 0.45472243 0.08046602 -0.39103362 -0.86941946 -0.1584959
-0.3642725 -0.5277601 0.63333285 -0.9229159 -1.0798976 0.79299814
-0.99585176 -0.12218212 -0.11947128 0.569726 -0.25965592 0.2894865
-0.70253044 -0.8766596 -0.6899628 -1.0937155 0.88824046 0.74569994
0.3853275 -0.94687676 -0.21958663 -0.05377176 0.7564465 0.29835165
0.4421056 -0.3968858 -1.0958072 -0.3835345 -0.3272136 0.02034971
0.17315924 0.10851234 -0.6698344 -0.7004774 -0.20870274 -0.26700458
0.9079108 0.65116155 1.1790406 -0.14678071 -0.6956541 0.8086361
1.4894001 0.34269375 0.36338994 0.40942863 0.99633706 -0.11110672
0.95766693 0.42495388 0.03576636 0.95058787 0.46840635 -0.24546865
-0.23097071 -0.3545748 0.34112614 0.9284868 -0.11564457 -0.11454219
-0.40183666 0.33404088 -1.9359727 -1.1147912 0.09042356 2.3384876
0.5952954 0.3773627 0.24417801 -0.11520465 0.8083764 0.2790137
-0.8571733 0.36507836 0.12599118 0.07841547 0.58425575 -0.13063036
-1.2949089 1.0516262 5.473166 1.0790899 -1.3773777 0.14496802
0.2203534 -0.10458273 0.07876894 -0.39004153 -1.1685226 0.86203766
0.38699037 0.04901442 0.1546223 0.95042753 -0.0120672 0.40248582
-0.6687379 0.99160326 -0.0610638 -1.549653 0.1305388 -0.10454548
0.63444215 0.09125645 0.24060825 0.4788203 0.06118801 -0.58891994
0.758322 0.72343665 0.7542044 -0.7902869 0.65286875 0.23673894
-1.8170298 -0.18874577 -0.565417 0.3140064 0.14295721 0.26017034
-0.14346191 0.6037557 0.72940224 0.9675531 -0.4862901 1.4288994
0.00930361 0.39810398 -0.3540741 -0.21437363 0.29685986 -0.01378852
0.5005573 0.859336 0.34478918 -0.15879229 0.51632637 0.8331797
-0.20777507 -0.1425591 -0.4136536 -0.04423109 0.6713358 1.2286394
-0.88552123 -0.40907833 -0.5004994 0.68740577 0.31241724 0.23221953
-1.2095466 -0.46638232 0.45394555 -0.09748087 0.822918 -0.14970952
0.2233185 -1.247185 0.20573615 -0.796877 0.3556142 -0.45437402
-1.2712874 0.6172185 -0.5045375 -1.8755528 -0.0904304 -0.5186434
-0.66552365 0.501745 -1.531635 -1.4085842 -0.4061554 0.6633697
0.6287762 -0.33387166 0.25785843 0.6853909 -0.9015682 0.92043984
0.08082704 0.46759558 0.5972307 -0.7224401 0.3516202 0.9421297
0.13910455 0.63610715 0.27468184 -0.7285703 -1.6447713 -1.5055283
0.40727004 -0.05280301 0.6382256 -0.07320318 -0.9923524 0.70184344
-0.22100312 0.5761362 0.20041287 -0.26960772 -0.19049534 -0.35879862
-0.8644838 0.5201198 1.2090946 -0.22590633 -0.43808886 0.21490207
0.5999825 -0.69299763 -0.75885063 0.6290012 0.7510935 -0.73688036
1.0355223 -0.45811987 -0.10097819 -0.63939905 0.31361902 -0.82413995
-0.5260739 -0.52088106 -0.4763378 1.2071574 -0.10321368 -0.783387
1.1337067 0.2552983 -0.13576356 -0.90926003 -0.9661302 -1.0970846
-0.35948923 0.03681673 0.4259062 0.82784915 -0.71991944 0.04985807
-0.58580106 0.04585056 0.8550561 0.5690475 0.75924784 -1.2843758
-0.1866366 -0.48464793 -0.67940235 -1.3056555 0.18734312 -0.7689019
0.11283118 -1.1716821 0.1988542 -0.6288789 -0.60440725 0.6205712
-0.35693678 0.3730284 0.35133654 0.4916296 -0.9671899 0.9493494
1.4536531 -0.24715738 -0.19457859 0.31920812 -0.15095674 0.6746321
0.45201182 -0.72806513 -0.09246035 -0.49232146 -0.19007461 0.00965122
0.5026839 -1.2464689 0.61332697 0.10002683 0.93433404 -0.98118454
0.29562953 -0.91710025 0.21962506 1.001099 -0.02007145 0.09540431
0.46399334 0.8455085 -0.28591892 0.06393918 0.88627464 0.16500023
-0.7864389 0.97986656 -0.17393863 -0.05425769 0.9051478 -0.34248018
-0.16902226 -0.05428426 -0.31017205 0.3212936 0.46245104 0.40985224
0.63994676 -1.8700372 -0.3047252 0.27067327 -0.1115504 -0.18169193
0.34982184 0.96432966 -0.20159002 0.3870164 -0.4455841 -0.81444544
-1.2747751 0.7383467 0.36433387 -0.27994296 -0.82200587 0.8175111
0.12355826 -0.02719079 0.44915235 -0.25635058 -0.10233382 -0.17437686
0.53538555 0.11480711 -0.26811326 -0.6069634 -0.5172813 0.87376785
-0.18319051 0.5266258 1.028739 0.15272556 0.34679085 0.05223797
0.9214654 -0.19815014 -1.6280067 -0.5885271 -0.11122351 -0.6742907
0.17339917 -0.4075931 -1.8366313 0.8031602 0.9621615 0.0229259
1.1562912 -0.4208032 1.1930722 0.41112515 0.40339783 -0.8549663
0.17230044 0.3438599 0.43419 -1.3078859 0.15200564 -0.12719813
-0.29879108 1.2183865 0.9158268 -0.23270138 0.38317612 0.22671735
0.03958927 -0.13335432 -0.59145486 -0.25535804 0.4255249 0.41214812
0.13953906 -0.32045147 -0.03163893 0.27739513 0.41549304 0.27709258
-0.17613076 0.86334944 -0.37277424 -0.93298465 -0.30165625 0.52961826
-0.4237837 0.11230259 0.08028841 1.0239221 0.10331701 0.31117526
0.12846446 -0.41245484 0.29388708 -0.2593967 0.60945207 -0.1320818
-0.6912361 0.24621472 -0.4726197 -0.6250658 -0.69732255 -0.55858546
-1.2466986 -0.1905816 -0.773544 -0.07364281 0.36885887 0.9683854
0.3243748 0.8074041 0.6406153 -1.0356895 -0.67848116 -0.638192
-0.23073152 0.20305265 0.35621536 -1.1811719 -0.03860571 -0.11977784] | [6.273575782775879, -2.165163516998291] |
c6584835-2551-46af-bed6-1b7c2c5c28e6 | revisiting-point-cloud-classification-a-new | 1908.04616 | null | https://arxiv.org/abs/1908.04616v2 | https://arxiv.org/pdf/1908.04616v2.pdf | Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data | Deep learning techniques for point cloud data have demonstrated great potentials in solving classical problems in 3D computer vision such as 3D object classification and segmentation. Several recent 3D object classification methods have reported state-of-the-art performance on CAD model datasets such as ModelNet40 with high accuracy (~92%). Despite such impressive results, in this paper, we argue that object classification is still a challenging task when objects are framed with real-world settings. To prove this, we introduce ScanObjectNN, a new real-world point cloud object dataset based on scanned indoor scene data. From our comprehensive benchmark, we show that our dataset poses great challenges to existing point cloud classification techniques as objects from real-world scans are often cluttered with background and/or are partial due to occlusions. We identify three key open problems for point cloud object classification, and propose new point cloud classification neural networks that achieve state-of-the-art performance on classifying objects with cluttered background. Our dataset and code are publicly available in our project page https://hkust-vgd.github.io/scanobjectnn/. | ['Sai-Kit Yeung', 'Binh-Son Hua', 'Quang-Hieu Pham', 'Mikaela Angelina Uy', 'Duc Thanh Nguyen'] | 2019-08-13 | revisiting-point-cloud-classification-a-new-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Uy_Revisiting_Point_Cloud_Classification_A_New_Benchmark_Dataset_and_Classification_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Uy_Revisiting_Point_Cloud_Classification_A_New_Benchmark_Dataset_and_Classification_ICCV_2019_paper.pdf | iccv-2019-10 | ['3d-object-classification'] | ['computer-vision'] | [-3.70542333e-02 -4.52334821e-01 -1.75726414e-01 -5.68578780e-01
-6.01107895e-01 -5.64132929e-01 3.65846187e-01 -7.05518723e-02
-8.60679522e-02 4.47840504e-02 -7.23661482e-01 -5.87774813e-01
1.02559179e-01 -8.09814334e-01 -1.14610422e+00 -3.71797055e-01
-2.82645345e-01 9.68098044e-01 4.69257653e-01 -2.84110382e-03
1.66269124e-01 1.14446020e+00 -1.61667430e+00 1.65605143e-01
5.93044460e-01 1.30700529e+00 2.06686661e-01 5.57137132e-01
-7.27706671e-01 2.77765095e-01 -6.01501048e-01 -2.19544336e-01
7.66003847e-01 4.45940942e-01 -8.14954579e-01 1.87697262e-01
1.09682727e+00 -4.66213673e-01 -4.69010204e-01 1.14555943e+00
4.50633228e-01 -1.91148445e-01 5.64927340e-01 -1.54211271e+00
-7.38467336e-01 4.57301177e-02 -7.22584903e-01 2.41396815e-01
-1.41087770e-01 2.20503747e-01 6.88971937e-01 -1.09708178e+00
5.07146776e-01 1.39475441e+00 8.11566591e-01 4.66894358e-01
-1.10589838e+00 -1.02132297e+00 4.50644791e-01 2.29588613e-01
-1.36788523e+00 -5.42923361e-02 9.82333720e-01 -6.07205093e-01
9.37796772e-01 3.89183164e-01 9.59824562e-01 1.02121449e+00
-6.30849041e-03 1.05349326e+00 8.11211109e-01 -9.08237472e-02
2.91776866e-01 -2.93855280e-01 5.52118301e-01 4.78963584e-01
3.95426959e-01 5.39483838e-02 -8.62706304e-02 -9.54523906e-02
1.03152895e+00 3.36261272e-01 -1.09377250e-01 -8.35150778e-01
-1.17845583e+00 8.26345444e-01 9.90874290e-01 6.35749325e-02
-1.61738798e-01 4.33673978e-01 2.01002359e-01 2.35508144e-01
7.63093650e-01 8.89609307e-02 -6.34188414e-01 2.29777470e-01
-6.98247552e-01 5.70407033e-01 5.96824050e-01 1.48468792e+00
6.97529078e-01 -5.33569604e-02 4.22613174e-01 9.35849428e-01
4.13926452e-01 8.84263933e-01 -1.97809950e-01 -9.24888492e-01
4.27385956e-01 6.88688219e-01 4.55137901e-02 -8.96178901e-01
-3.45863461e-01 -6.67389810e-01 -9.14716065e-01 6.15743160e-01
2.58242816e-01 5.14774084e-01 -1.37400365e+00 1.08658862e+00
5.67750454e-01 4.73713279e-01 -3.27580899e-01 9.79422569e-01
1.43254209e+00 5.73894083e-01 -1.78518698e-01 4.04662549e-01
1.10237026e+00 -6.94055259e-01 -1.60740882e-01 -2.96551704e-01
3.33921701e-01 -6.09196484e-01 9.81957853e-01 3.13100427e-01
-8.11193109e-01 -6.00097716e-01 -1.00739634e+00 -1.41347677e-01
-3.33340704e-01 -2.39756212e-01 8.58744562e-01 3.03545177e-01
-7.59894729e-01 5.08384407e-01 -1.08103395e+00 -4.23016518e-01
1.25821972e+00 4.58566010e-01 -2.53804952e-01 -2.97766328e-01
-3.84077728e-01 7.26149142e-01 1.43644899e-01 1.84702784e-01
-8.96610141e-01 -1.00088680e+00 -5.84949195e-01 -1.73153803e-01
2.86357284e-01 -7.10650444e-01 1.43816054e+00 -4.52632993e-01
-9.94693995e-01 1.38371241e+00 5.31335659e-02 -3.28293622e-01
6.16063833e-01 -3.99876773e-01 -6.00463003e-02 -5.63549474e-02
2.10556328e-01 8.07725310e-01 7.42627740e-01 -1.68950915e+00
-5.73455632e-01 -7.44119227e-01 7.62064382e-02 -4.30492647e-02
3.75633001e-01 -7.85925090e-02 -7.21893072e-01 -2.02324927e-01
7.28854537e-01 -1.07472932e+00 -3.67255032e-01 6.44092858e-01
-4.36181068e-01 -4.83248055e-01 1.18240869e+00 3.33151296e-02
2.08437443e-01 -1.97490239e+00 -2.73064107e-01 5.76735996e-02
4.76650566e-01 3.92978907e-01 -2.00515106e-01 -1.12019740e-01
-1.43905878e-01 2.18634680e-01 -3.00669104e-01 -2.97396809e-01
2.30505038e-02 3.59118015e-01 -4.39185262e-01 7.38569319e-01
3.72512102e-01 1.02203262e+00 -5.56640387e-01 -3.85267317e-01
6.03211820e-01 4.64460403e-01 -3.59846175e-01 1.14033431e-01
-5.66185534e-01 5.19906700e-01 -5.66810906e-01 1.20484400e+00
1.33622646e+00 -4.62017655e-01 -5.84688127e-01 -1.26493782e-01
3.86447087e-02 -1.12120574e-02 -1.22169352e+00 1.80570054e+00
-1.43670827e-01 8.04006934e-01 1.63295776e-01 -1.06127977e+00
9.92201984e-01 -2.26953581e-01 7.15546727e-01 -4.41849202e-01
1.87528580e-01 2.29336604e-01 -9.01585221e-02 -4.29554224e-01
2.22112447e-01 7.25824460e-02 2.55916595e-01 1.47721507e-02
-5.46644963e-02 -9.28139806e-01 -3.51977885e-01 -1.01220891e-01
1.10643589e+00 -2.95267962e-02 -1.85299739e-01 -2.09502250e-01
1.16623104e-01 4.47380602e-01 3.21981847e-01 9.86420333e-01
-3.21326107e-01 8.78011286e-01 1.52662724e-01 -9.09673095e-01
-1.03015983e+00 -1.26271081e+00 -6.43894553e-01 5.21575272e-01
5.32444894e-01 1.70207508e-02 -2.69164503e-01 -5.36709726e-01
4.33167666e-01 3.35475862e-01 -4.71892774e-01 2.64155775e-01
-7.73606598e-01 -6.63936794e-01 2.81127006e-01 6.01302385e-01
4.83381659e-01 -8.32762241e-01 -5.78219056e-01 5.25952391e-02
2.22182289e-01 -1.42187798e+00 1.25511378e-01 3.31539303e-01
-1.17976737e+00 -1.21607339e+00 -6.96381152e-01 -8.48114789e-01
6.07320666e-01 8.10597718e-01 1.64739251e+00 3.25373858e-01
-4.90456551e-01 4.27978784e-01 -3.07858109e-01 -1.16775537e+00
-6.62679598e-02 1.36785164e-01 -1.11852013e-01 -4.88875806e-01
6.35208130e-01 -5.97443819e-01 -5.99887788e-01 4.41955119e-01
-7.30253816e-01 6.02533557e-02 3.01647842e-01 4.39307004e-01
1.03722155e+00 -1.23711184e-01 1.72345899e-02 -5.70720434e-01
6.88082119e-03 -3.87405306e-01 -1.02953780e+00 -1.57801270e-01
-3.41007188e-02 -5.78523159e-01 -8.05838872e-03 -3.57090056e-01
-3.57499242e-01 1.57060400e-01 -4.73518968e-01 -1.15775847e+00
-6.37560248e-01 3.40265222e-02 -5.41803800e-02 -4.56109226e-01
5.32929301e-01 2.49194633e-02 -1.69453233e-01 -7.05595255e-01
2.39523143e-01 5.94254315e-01 3.83721977e-01 -6.65184975e-01
1.14693296e+00 9.40427423e-01 1.59536675e-01 -9.11047637e-01
-8.91809165e-01 -6.63122654e-01 -9.91186798e-01 -2.10479409e-01
8.74683738e-01 -1.11715746e+00 -7.50038981e-01 6.33914888e-01
-1.54118037e+00 -4.24652159e-01 -2.30871916e-01 2.73370862e-01
-4.83462811e-01 8.72993618e-02 -3.81112188e-01 -7.45305240e-01
-3.19105893e-01 -1.32687902e+00 1.54164648e+00 -5.47876619e-02
2.98612148e-01 -5.37734210e-01 -2.72205502e-01 3.03926855e-01
2.40874782e-01 5.17415166e-01 8.90590191e-01 -5.37660778e-01
-1.33412814e+00 -2.33473927e-01 -5.12065351e-01 3.55451316e-01
-4.87493202e-02 1.61279872e-01 -1.01985383e+00 -1.56184465e-01
9.24411863e-02 -1.56280115e-01 7.70280778e-01 5.59826016e-01
1.79857326e+00 3.27289999e-01 -5.98172307e-01 1.05743682e+00
1.55853868e+00 1.84389189e-01 3.54160607e-01 3.83819699e-01
1.06024945e+00 3.01086664e-01 6.10263228e-01 7.96973556e-02
2.77408689e-01 6.31029129e-01 1.05183244e+00 -3.50140631e-01
-8.78255069e-02 8.51702914e-02 -4.75740016e-01 5.32115459e-01
-6.49382221e-03 -1.65146530e-01 -1.50220954e+00 4.25969899e-01
-1.67188799e+00 -6.37184381e-01 -7.58088589e-01 1.77461565e+00
2.64800817e-01 3.73698831e-01 -5.07460965e-04 6.43154755e-02
6.25474334e-01 7.78217614e-02 -8.15844834e-01 3.31833154e-01
-1.38334945e-01 4.27812636e-01 5.96563995e-01 7.08265007e-02
-1.47946620e+00 9.90423262e-01 5.82990599e+00 7.23485947e-01
-1.41778982e+00 1.22390196e-01 5.15419364e-01 -1.83451980e-01
1.06201239e-01 -4.63502675e-01 -8.60382199e-01 3.54795158e-01
2.13023424e-01 2.67489761e-01 -8.87763426e-02 1.28749776e+00
-1.67806864e-01 2.30454668e-01 -1.32128179e+00 1.41485286e+00
-1.88799694e-01 -1.59644008e+00 2.42468324e-02 1.62625208e-01
5.72668374e-01 9.44533646e-01 1.13275737e-01 2.01802969e-01
2.75037050e-01 -1.05248404e+00 8.76919687e-01 3.99666019e-02
7.22711682e-01 -4.33597863e-01 6.04475737e-01 4.23144341e-01
-1.03125024e+00 1.88962981e-01 -8.83621335e-01 7.38042817e-02
6.24949709e-02 7.87253857e-01 -7.04038322e-01 2.54073560e-01
1.30206382e+00 9.25384760e-01 -5.55865765e-01 1.51530707e+00
1.35699570e-01 4.20749515e-01 -7.14743793e-01 1.27751082e-01
3.89837086e-01 -1.78808942e-01 7.21505702e-01 1.03802085e+00
2.71989286e-01 2.32712910e-01 3.23481530e-01 1.23817980e+00
-2.18425453e-01 -1.96410254e-01 -9.39636111e-01 2.50982583e-01
3.31697375e-01 9.63895500e-01 -1.00946987e+00 -1.75256655e-01
-4.62537438e-01 4.94042277e-01 2.16597110e-01 2.59157479e-01
-8.57255876e-01 -1.94753846e-03 1.15289366e+00 1.45017028e-01
5.64126849e-01 -7.37626195e-01 -6.12712979e-01 -1.20409453e+00
2.12078199e-01 -5.14283180e-01 -8.78940001e-02 -8.61468315e-01
-1.60055101e+00 5.41715324e-01 1.95652470e-01 -1.56646097e+00
5.60402155e-01 -1.15383756e+00 -6.05117679e-01 4.80730951e-01
-1.70525622e+00 -1.30596459e+00 -8.33552122e-01 5.46602249e-01
8.70587230e-01 -9.12676454e-02 5.20245731e-01 4.83864993e-01
-2.55084455e-01 -8.12165730e-04 8.85139182e-02 4.17842954e-01
1.47459596e-01 -1.20221329e+00 1.03441262e+00 4.32974726e-01
2.93196827e-01 3.08766752e-01 3.40656132e-01 -5.41377187e-01
-1.68302631e+00 -1.33645916e+00 -2.65327073e-03 -9.06913698e-01
3.58857930e-01 -8.48295152e-01 -1.13157487e+00 6.98985517e-01
-3.98713589e-01 5.03513455e-01 4.45202678e-01 -3.12520750e-02
-3.21106970e-01 -2.03584321e-02 -1.15045345e+00 2.75781542e-01
1.64432549e+00 -2.18474016e-01 -5.22128880e-01 7.75049448e-01
1.08245325e+00 -1.01023471e+00 -6.81141376e-01 7.89345503e-01
3.30675364e-01 -8.46873641e-01 1.52655602e+00 -6.98044538e-01
3.62112254e-01 -3.44756126e-01 -4.80654776e-01 -1.01396453e+00
-2.44806215e-01 1.13203518e-01 -8.69056135e-02 7.36962497e-01
4.54449560e-03 -4.98567313e-01 1.15078270e+00 3.85276020e-01
-4.74148750e-01 -8.01812470e-01 -1.19564545e+00 -1.11212993e+00
4.37422782e-01 -1.03722608e+00 9.03124332e-01 9.87617195e-01
-1.06905997e+00 1.52515069e-01 2.38123059e-01 4.95879501e-01
9.54313934e-01 4.97107208e-01 1.22384906e+00 -1.79185677e+00
2.71769285e-01 -5.35442472e-01 -7.97076702e-01 -1.24743187e+00
3.40002328e-01 -1.00440240e+00 -1.28808226e-02 -1.68668127e+00
-7.42600933e-02 -1.19551396e+00 7.26348385e-02 3.04727137e-01
2.26864696e-01 4.05481726e-01 5.02908051e-01 5.50093949e-01
-5.31248569e-01 4.69840527e-01 1.38136995e+00 -7.76672602e-01
-6.20505102e-02 3.42077821e-01 -2.11749926e-01 7.98993886e-01
6.99211419e-01 -5.78826249e-01 -1.27445869e-02 -1.14947867e+00
-1.81011394e-01 -2.69021004e-01 8.46052527e-01 -1.20891404e+00
5.36441058e-02 -1.76389128e-01 6.33572578e-01 -1.37282670e+00
7.29915261e-01 -1.28450656e+00 2.32264087e-01 4.52779353e-01
1.75103232e-01 -1.55449510e-01 5.04745603e-01 5.15541315e-01
-7.50253126e-02 -3.49589139e-02 9.02565718e-01 -4.14637893e-01
-8.02906871e-01 9.45735157e-01 1.66684791e-01 -2.03460902e-01
9.62612450e-01 -4.35356587e-01 -4.07580674e-01 1.65282879e-02
-6.57679617e-01 3.04367781e-01 5.47524810e-01 6.84063613e-01
7.14247584e-01 -1.35323679e+00 -8.05222452e-01 2.93140620e-01
2.93892264e-01 1.01593053e+00 9.94270220e-02 3.70703727e-01
-9.88442898e-01 4.41614538e-01 -1.77964702e-01 -1.65333247e+00
-1.24690700e+00 4.60960180e-01 5.55137873e-01 4.18011844e-01
-1.00556564e+00 1.12725401e+00 3.44404250e-01 -8.08944166e-01
4.85283494e-01 -8.44468296e-01 3.71725619e-01 -4.56812501e-01
8.29432383e-02 8.23642015e-02 3.96346450e-01 -4.84882593e-01
-5.78068733e-01 1.08013701e+00 7.20967352e-02 5.37229955e-01
1.69443822e+00 3.48621458e-01 1.86201520e-02 4.12824005e-01
1.32126570e+00 -5.92877865e-01 -1.00870478e+00 -3.77238125e-01
-2.18329012e-01 -9.40192640e-01 1.16135247e-01 -4.53047991e-01
-1.21536303e+00 1.14556670e+00 1.06609523e+00 2.71568835e-01
6.70331538e-01 5.69055796e-01 6.43066823e-01 7.80398011e-01
6.91638291e-01 -5.41547596e-01 -8.35457370e-02 8.05872738e-01
1.15197313e+00 -1.78766680e+00 1.00101940e-01 -8.27493608e-01
-7.31002763e-02 1.03488028e+00 8.48915815e-01 -3.99657279e-01
1.06351018e+00 2.15414509e-01 2.18270779e-01 -7.36974776e-01
-3.68108124e-01 -3.24214041e-01 1.80904031e-01 9.25730109e-01
5.18136565e-03 1.61579788e-01 5.80113709e-01 1.83684126e-01
-4.20232147e-01 -2.26234436e-01 2.07711533e-01 1.04490411e+00
-3.96261871e-01 -7.79244423e-01 -6.52045906e-01 6.44452274e-01
-1.62061170e-01 2.76763707e-01 -3.57996553e-01 1.01122856e+00
2.03597337e-01 4.43658859e-01 6.30878389e-01 -1.76108167e-01
5.72029769e-01 -2.19859168e-01 5.65109015e-01 -6.39968216e-01
-2.77539045e-01 -8.78800973e-02 -3.44831854e-01 -6.16458476e-01
-4.40874994e-01 -6.32883072e-01 -1.06240273e+00 -3.04398477e-01
-4.66928869e-01 -3.97949070e-01 1.22914255e+00 6.22156382e-01
3.93651128e-01 4.43950653e-01 3.86332840e-01 -1.49227464e+00
-2.56048620e-01 -6.87210500e-01 -4.00980115e-01 3.10411572e-01
5.84022939e-01 -8.70488942e-01 -2.08136082e-01 -2.04905480e-01] | [7.941969871520996, -3.3452587127685547] |
09d4ec4a-496a-4cc8-b700-ea56f6d691e7 | iss-image-as-stepping-stone-for-text-guided | 2303.15181 | null | https://arxiv.org/abs/2303.15181v1 | https://arxiv.org/pdf/2303.15181v1.pdf | ISS++: Image as Stepping Stone for Text-Guided 3D Shape Generation | In this paper, we present a new text-guided 3D shape generation approach (ISS++) that uses images as a stepping stone to bridge the gap between text and shape modalities for generating 3D shapes without requiring paired text and 3D data. The core of our approach is a two-stage feature-space alignment strategy that leverages a pre-trained single-view reconstruction (SVR) model to map CLIP features to shapes: to begin with, map the CLIP image feature to the detail-rich 3D shape space of the SVR model, then map the CLIP text feature to the 3D shape space through encouraging the CLIP-consistency between rendered images and the input text. Besides, to extend beyond the generative capability of the SVR model, we design a text-guided 3D shape stylization module that can enhance the output shapes with novel structures and textures. Further, we exploit pre-trained text-to-image diffusion models to enhance the generative diversity, fidelity, and stylization capability. Our approach is generic, flexible, and scalable, and it can be easily integrated with various SVR models to expand the generative space and improve the generative fidelity. Extensive experimental results demonstrate that our approach outperforms the state-of-the-art methods in terms of generative quality and consistency with the input text. Codes and models are released at https://github.com/liuzhengzhe/ISS-Image-as-Stepping-Stone-for-Text-Guided-3D-Shape-Generation. | ['Chi-Wing Fu', 'Xiaojuan Qi', 'Ruihui Li', 'Peng Dai', 'Zhengzhe Liu'] | 2023-03-24 | null | null | null | null | ['3d-shape-generation'] | ['computer-vision'] | [ 2.23702654e-01 -4.18428406e-02 1.74852923e-01 -1.05373777e-01
-6.55963898e-01 -7.77296543e-01 8.68504763e-01 -5.02880812e-01
4.92969722e-01 1.15499385e-01 3.92264932e-01 -2.04979151e-01
1.63701415e-01 -1.02967703e+00 -6.85346365e-01 -6.76043391e-01
5.95736325e-01 4.67227578e-01 1.01439111e-01 -3.85743856e-01
2.54703552e-01 7.10794210e-01 -1.29770446e+00 2.26924941e-01
1.06296527e+00 9.41459894e-01 3.45980525e-01 5.77717841e-01
-3.78953218e-01 1.33463770e-01 -4.59355749e-02 -4.29288149e-01
5.35730183e-01 -5.53167462e-01 -2.37092689e-01 5.45105100e-01
3.88469070e-01 -4.37090278e-01 -4.01105523e-01 6.56611323e-01
7.12601721e-01 -1.74451843e-01 9.23334539e-01 -9.48475122e-01
-1.31231713e+00 1.85499042e-01 -8.63422215e-01 -4.51498032e-01
4.01788861e-01 3.54865521e-01 6.45045877e-01 -1.40752125e+00
9.60643351e-01 1.34805894e+00 5.04375935e-01 5.78315020e-01
-1.31939936e+00 -5.49839199e-01 -5.05789593e-02 -5.47072947e-01
-1.14594173e+00 -3.83945972e-01 1.17290390e+00 -6.27320468e-01
7.66694784e-01 3.75710160e-01 8.24811876e-01 1.00245023e+00
1.87234417e-01 7.86761284e-01 9.23819065e-01 -3.87692273e-01
3.50784473e-02 1.42028883e-01 -6.77914202e-01 7.86846697e-01
-7.66632408e-02 3.69898528e-01 -4.76911128e-01 -8.13030303e-02
1.50918221e+00 2.87351996e-01 -1.78052694e-01 -7.38400459e-01
-1.15977061e+00 7.79407203e-01 5.40981352e-01 3.20938349e-01
-3.76309305e-01 2.04743311e-01 -5.82132675e-03 -1.76694114e-02
8.77716660e-01 4.73064214e-01 -1.01486176e-01 3.71274091e-02
-1.06955504e+00 2.95957029e-01 3.52090687e-01 1.24530053e+00
7.22490132e-01 4.55015600e-01 -5.34494042e-01 9.02257919e-01
3.89710218e-01 1.00070846e+00 2.37069502e-01 -9.81338382e-01
4.25302029e-01 8.20943356e-01 -8.99964273e-02 -9.79582846e-01
1.18593521e-01 -3.94848585e-01 -8.89592707e-01 2.61029631e-01
-7.45047256e-02 1.44936204e-01 -1.03297698e+00 1.25616646e+00
6.03638411e-01 -5.72391748e-02 -2.84505755e-01 8.65480304e-01
9.86762047e-01 6.56617761e-01 -4.20712471e-01 7.55240396e-02
1.06992733e+00 -1.02373672e+00 -4.45503592e-01 5.00452034e-02
3.07251185e-01 -1.20417678e+00 1.30525076e+00 -2.20126435e-01
-1.49020803e+00 -5.64988852e-01 -7.68253982e-01 -1.17785878e-01
1.13697648e-02 1.95002794e-01 3.36633503e-01 4.95008558e-01
-1.13261664e+00 4.17066842e-01 -6.09475493e-01 -2.32457876e-01
5.98430336e-01 -8.08399469e-02 -8.28453675e-02 -1.54901534e-01
-6.25930786e-01 4.57048833e-01 -6.41243011e-02 -4.06580418e-01
-6.60551727e-01 -9.94837403e-01 -8.00936818e-01 -1.89075291e-01
1.42176270e-01 -1.31135321e+00 9.34247196e-01 -6.43821895e-01
-1.79072249e+00 8.84924710e-01 -3.18722337e-01 3.39870930e-01
6.31605446e-01 2.16328520e-02 1.37383759e-01 1.36681557e-01
6.43571243e-02 7.89112687e-01 1.18451619e+00 -1.92389512e+00
-8.98538679e-02 -1.91540062e-01 -3.94933760e-01 3.93680781e-01
-2.50218362e-01 -2.72860914e-01 -8.34520578e-01 -1.23735368e+00
3.07812363e-01 -8.65092039e-01 -2.53263563e-01 3.35762918e-01
-5.97206056e-01 2.21203804e-01 1.01238024e+00 -5.01437783e-01
1.09981024e+00 -2.22311902e+00 3.06606472e-01 3.69878054e-01
3.25424641e-01 3.28033231e-02 -4.42443311e-01 7.28626013e-01
4.88287807e-02 3.18823159e-01 -3.34813058e-01 -7.52358437e-01
-5.85008934e-02 -7.94927329e-02 -6.54573679e-01 7.13032782e-02
4.24548030e-01 1.44424510e+00 -7.40185022e-01 -3.55611920e-01
3.82534444e-01 7.77888656e-01 -7.27951646e-01 3.43110114e-01
-3.40355545e-01 9.12914634e-01 -6.92788899e-01 8.39695394e-01
9.30951297e-01 -3.63159865e-01 -1.72575355e-01 -2.66611218e-01
-1.12780735e-01 -9.51147974e-02 -1.00035918e+00 1.83778429e+00
-5.00044048e-01 1.54831648e-01 -9.87230465e-02 -3.93144310e-01
1.43032932e+00 1.17866822e-01 5.75647831e-01 -6.70357406e-01
1.53795734e-01 4.52967942e-01 -5.43331087e-01 -3.01474601e-01
6.65190279e-01 -1.00209683e-01 1.38531327e-01 6.90070987e-01
-1.48318291e-01 -9.32415009e-01 -1.65432230e-01 3.19112897e-01
4.97093946e-01 6.25217199e-01 -1.59098104e-01 -2.45238617e-02
3.26301664e-01 -3.08203250e-01 1.17679581e-01 4.19824064e-01
4.07284588e-01 1.11972523e+00 1.23818390e-01 -2.12358564e-01
-1.62012649e+00 -1.24583185e+00 -1.04418010e-01 6.87697351e-01
1.93647847e-01 -2.38964468e-01 -6.62328243e-01 -5.84035814e-01
8.05559456e-02 6.85670674e-01 -6.66285753e-01 -2.39188001e-02
-4.69438523e-01 -2.31308237e-01 2.68952638e-01 5.34953833e-01
4.13246989e-01 -8.15120876e-01 -2.53503531e-01 1.26809597e-01
-3.04606743e-02 -9.13721859e-01 -1.11712050e+00 -5.09110272e-01
-1.16476560e+00 -5.44746518e-01 -1.11967611e+00 -7.84176707e-01
1.10286534e+00 5.59147418e-01 9.74186778e-01 3.36869925e-01
-1.92910939e-01 5.99105418e-01 -5.05595863e-01 -1.28090665e-01
-6.62265480e-01 -2.61751920e-01 -2.10849077e-01 1.31310672e-01
-4.42528129e-01 -9.09370363e-01 -7.69535959e-01 3.77211183e-01
-1.15445304e+00 6.41583323e-01 4.28775817e-01 8.16271126e-01
8.76123905e-01 -4.52406853e-01 1.64932117e-01 -5.31995535e-01
5.72230995e-01 -1.05138056e-01 -3.55458856e-01 1.58591196e-01
-4.19067979e-01 7.97100589e-02 5.41611791e-01 -6.06403887e-01
-1.03133488e+00 1.56255692e-01 -1.79209486e-01 -7.52446890e-01
1.14703760e-01 3.02150577e-01 -1.58588037e-01 -1.59207538e-01
5.53733408e-01 6.87981546e-01 2.09155172e-01 -5.52370727e-01
8.31804216e-01 4.93149996e-01 3.32596928e-01 -6.52588367e-01
1.41129053e+00 6.80783272e-01 2.20573638e-02 -6.32706404e-01
-4.65639949e-01 1.22309029e-02 -7.04015613e-01 -2.98861504e-01
6.01413190e-01 -8.32535565e-01 -1.83984622e-01 6.41282916e-01
-1.08760667e+00 -5.44825375e-01 -5.96903801e-01 4.65510637e-02
-8.79659712e-01 4.84524786e-01 -5.68203807e-01 -7.30535507e-01
-6.74711406e-01 -1.16406333e+00 1.69869602e+00 3.14465523e-01
5.32453060e-02 -9.73289907e-01 3.08556929e-02 3.34381521e-01
7.95066178e-01 3.02745700e-01 9.67126548e-01 -3.07100397e-02
-8.93624306e-01 -4.65958677e-02 -2.87505478e-01 2.86052804e-02
3.15294921e-01 1.96320146e-01 -7.46259332e-01 -3.17097694e-01
-1.20690979e-01 -1.79869011e-01 6.78748071e-01 6.10649645e-01
1.00996041e+00 -2.58575082e-01 -3.09771657e-01 1.01085615e+00
1.26225209e+00 -2.89872102e-02 8.05920780e-01 7.42807314e-02
8.68555665e-01 4.10153061e-01 3.36590469e-01 6.34230554e-01
4.93782043e-01 8.95514309e-01 2.47874111e-01 -5.25329709e-01
-6.54459238e-01 -8.61257732e-01 2.49508917e-01 1.03972399e+00
-2.05243170e-01 -2.52712309e-01 -4.57242161e-01 3.09779048e-01
-1.58785224e+00 -8.91347587e-01 -1.31707326e-01 2.00022316e+00
7.76267111e-01 -2.83988327e-01 -1.75419748e-02 -1.63722396e-01
6.49273455e-01 1.80423394e-01 -6.30569935e-01 -2.50505179e-01
-2.14078590e-01 4.65120934e-02 -5.20610809e-02 5.71892738e-01
-6.11003995e-01 1.17356277e+00 5.52218390e+00 1.20201743e+00
-1.34840763e+00 -1.33342713e-01 7.80916095e-01 -2.03327686e-02
-1.08211327e+00 -4.55312245e-02 -6.33585870e-01 3.54562074e-01
7.97054023e-02 -1.57785445e-01 6.33125544e-01 5.84142923e-01
3.15549493e-01 2.29203552e-01 -7.68300593e-01 1.04383755e+00
1.20725334e-01 -1.79040372e+00 4.27043974e-01 2.36351758e-01
1.16652000e+00 -3.44263107e-01 4.60156649e-01 -1.38228089e-01
2.05088764e-01 -9.24875379e-01 9.31086481e-01 7.78040409e-01
1.44567108e+00 -4.97185677e-01 1.23932697e-01 2.19128400e-01
-1.50767362e+00 2.63045758e-01 -2.23384812e-01 4.71133411e-01
5.30364573e-01 6.88341200e-01 -5.87631404e-01 7.47615993e-01
5.35623193e-01 8.01254451e-01 -4.28489536e-01 7.53702521e-01
-2.36784831e-01 2.09617719e-01 -1.19770959e-01 6.13862127e-02
-2.07576696e-02 -4.61290658e-01 8.32809865e-01 9.29516613e-01
8.98900568e-01 1.29601091e-01 2.11100474e-01 1.52322221e+00
-5.84421158e-02 1.91524044e-01 -9.37213838e-01 -2.11633928e-02
6.72417939e-01 1.31223154e+00 -7.69068301e-01 -2.97889709e-01
-1.24309555e-01 1.13790536e+00 3.70665155e-02 4.13611293e-01
-6.87635422e-01 -1.40356943e-01 2.24465847e-01 4.79120672e-01
5.79943478e-01 -4.08556700e-01 -8.43203127e-01 -1.03128719e+00
-1.67530123e-03 -7.11046875e-01 -3.55475038e-01 -1.16661036e+00
-1.56696677e+00 6.47188902e-01 -2.58339554e-01 -1.50933695e+00
-1.36256367e-01 -1.87669784e-01 -7.49858439e-01 1.15237498e+00
-1.33364105e+00 -1.84431469e+00 -3.41486990e-01 6.24471724e-01
6.08514607e-01 -2.20194563e-01 6.36206567e-01 -8.48087966e-02
-1.09782174e-01 6.01761520e-01 2.20876232e-01 -1.96516201e-01
6.43266201e-01 -7.78290212e-01 8.85100842e-01 6.65604353e-01
2.89571602e-02 4.72286433e-01 2.70825058e-01 -9.78285074e-01
-1.79640305e+00 -1.23517275e+00 5.94213784e-01 -6.69523895e-01
3.28807652e-01 -4.70824808e-01 -7.37376273e-01 2.24169970e-01
2.32536778e-01 -2.76620328e-01 4.73814040e-01 -4.56760675e-01
-4.20212030e-01 1.79450005e-01 -1.05918896e+00 7.98723817e-01
1.35441208e+00 -5.29547036e-01 -2.59167820e-01 2.82969158e-02
8.50535929e-01 -7.87424266e-01 -8.64219308e-01 3.62055868e-01
5.42619109e-01 -9.80573118e-01 1.19331610e+00 -1.23098411e-01
9.20503378e-01 -4.36459452e-01 -2.81099528e-01 -1.41766500e+00
-5.64030290e-01 -8.23229671e-01 -1.63271263e-01 1.41212547e+00
4.61124450e-01 -4.08962816e-01 5.91849148e-01 4.42597568e-01
-3.91871601e-01 -1.03769863e+00 -6.91696644e-01 -6.91498280e-01
1.71031073e-01 -3.39412451e-01 9.07506704e-01 9.10472453e-01
-3.13525498e-01 2.32975677e-01 -4.17238086e-01 -1.70992672e-01
4.52665359e-01 6.26274824e-01 1.07583916e+00 -8.15260828e-01
-3.95720601e-01 -6.38487339e-01 6.79624379e-02 -1.48289287e+00
-3.97609025e-01 -1.18577945e+00 -1.80168942e-01 -1.73759055e+00
4.26448882e-01 -6.63558364e-01 5.13313055e-01 3.55274618e-01
-2.27913007e-01 4.60108101e-01 6.25266671e-01 4.18168694e-01
-6.45761797e-03 1.11422372e+00 2.08783793e+00 2.36419775e-02
-5.81516743e-01 -1.38800144e-01 -9.11548316e-01 2.92156011e-01
4.39023912e-01 -1.76414937e-01 -3.61057103e-01 -6.30301178e-01
9.10553113e-02 9.16354284e-02 3.95863086e-01 -5.26352406e-01
2.56821662e-02 -2.31808811e-01 5.11430800e-01 -8.67357790e-01
4.16043282e-01 -6.71725392e-01 3.55597526e-01 3.16379368e-01
-3.47438604e-02 -6.16815388e-02 1.44495964e-01 4.74281549e-01
2.26941019e-01 1.53731748e-01 7.57132947e-01 8.22003037e-02
4.19462062e-02 6.47492707e-01 3.64997685e-02 4.93757650e-02
8.75708282e-01 -5.80697238e-01 -2.83379793e-01 -5.36112309e-01
-4.03374225e-01 6.01654537e-02 1.08583379e+00 5.53065240e-01
1.00289547e+00 -1.79945874e+00 -9.06474173e-01 6.31447971e-01
2.33281311e-02 7.19463378e-02 4.28085536e-01 7.61717975e-01
-3.25973749e-01 1.42130792e-01 -4.32705842e-02 -8.19905043e-01
-9.54818249e-01 4.30417359e-01 1.32181063e-01 -2.39479899e-01
-8.01967978e-01 7.30951846e-01 4.97243971e-01 -6.91314876e-01
-1.94429353e-01 -1.15725957e-01 2.65687674e-01 -3.36581171e-01
3.18876684e-01 6.35110512e-02 -2.37859055e-01 -6.43413723e-01
-2.54263496e-03 1.25881255e+00 2.55610079e-01 -3.32547277e-01
1.60141361e+00 -2.13209391e-01 5.64562157e-02 9.77766067e-02
9.34737444e-01 3.80502403e-01 -1.63561714e+00 -2.34490275e-01
-7.37973690e-01 -8.49384248e-01 -7.30539784e-02 -7.15386152e-01
-1.35554135e+00 8.78547907e-01 4.08737272e-01 6.27184063e-02
1.08783925e+00 2.36914456e-01 1.01417887e+00 -3.37594420e-01
2.00585276e-01 -6.63311839e-01 6.89357758e-01 5.50855577e-01
1.48365307e+00 -8.15528631e-01 -7.86191300e-02 -6.96499527e-01
-9.14028049e-01 1.10875916e+00 4.03902799e-01 -1.90111041e-01
6.38756335e-01 3.92389387e-01 -2.31678169e-02 -2.44053185e-01
-5.40816367e-01 -5.58862500e-02 6.58507645e-01 9.04373646e-01
4.58537549e-01 -8.20923038e-03 -2.96307039e-02 4.90194321e-01
-2.16906756e-01 -3.50859463e-02 2.83312112e-01 7.22719789e-01
-2.00164616e-01 -1.37510455e+00 -4.72778320e-01 3.51220816e-01
2.23833188e-01 -3.83771300e-01 -3.93342286e-01 3.11806858e-01
-9.41292942e-02 6.72882318e-01 -9.08484235e-02 -5.76091528e-01
4.54633206e-01 -1.26249000e-01 5.91081798e-01 -4.83588547e-01
-4.21272129e-01 6.80388212e-01 -2.74536908e-01 -5.09236872e-01
-2.86614567e-01 -5.18563449e-01 -1.02914286e+00 -4.63359028e-01
-3.42807591e-01 -4.61032003e-01 5.78290761e-01 6.14650309e-01
9.40899849e-01 3.17384630e-01 1.18269134e+00 -1.50579238e+00
-2.16345727e-01 -6.70485258e-01 -4.48993057e-01 3.60920310e-01
-3.40704992e-03 -5.31277776e-01 -1.72755882e-01 2.64757454e-01] | [9.079986572265625, -3.520026206970215] |
bd42418d-e2e3-4e1d-83f3-2b69110709d0 | multi-view-subspace-clustering | null | null | http://openaccess.thecvf.com/content_iccv_2015/html/Gao_Multi-View_Subspace_Clustering_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Gao_Multi-View_Subspace_Clustering_ICCV_2015_paper.pdf | Multi-View Subspace Clustering | For many computer vision applications, the data sets distribute on certain low-dimensional subspaces. Subspace clustering is to find such underlying subspaces and cluster the data points correctly. In this paper, we propose a novel multi-view subspace clustering method. The proposed method performs clustering on the subspace representation of each view simultaneously. Meanwhile, we propose to use a common cluster structure to guarantee the consistence among different views. In addition, an efficient algorithm is proposed to solve the problem. Experiments on four benchmark data sets have been performed to validate our proposed method. The promising results demonstrate the effectiveness of our method. | ['Xuelong. Li', 'Heng Huang', 'Feiping Nie', 'Hongchang Gao'] | 2015-12-01 | null | null | null | iccv-2015-12 | ['multi-view-subspace-clustering'] | ['computer-vision'] | [-4.31049645e-01 -6.67203486e-01 -2.24680707e-01 -1.96409300e-01
-3.98955613e-01 -7.63650298e-01 3.00467283e-01 -3.69471371e-01
-3.32682282e-02 1.08827122e-01 2.83568323e-01 9.06329229e-02
-2.54381150e-01 -3.25474381e-01 -1.44798785e-01 -1.00893283e+00
3.66497725e-01 3.51342499e-01 2.84616530e-01 4.15778607e-01
2.78502524e-01 4.06580597e-01 -1.32699800e+00 1.83412254e-01
8.39012444e-01 4.94910777e-01 2.82848179e-01 1.68017164e-01
1.07761107e-01 2.38001838e-01 -2.88298637e-01 2.64285386e-01
4.69921917e-01 -3.67418796e-01 -3.22209477e-01 7.69241333e-01
1.69191971e-01 -1.70331344e-01 -1.28774822e-01 1.30373693e+00
3.58495772e-01 2.19084516e-01 5.63108325e-01 -1.36250603e+00
-2.18505576e-01 1.41715661e-01 -1.17860305e+00 2.17773058e-02
1.30766153e-01 -2.14167207e-01 7.17336297e-01 -1.10151088e+00
5.63934326e-01 1.18869293e+00 1.60425767e-01 4.39705104e-01
-1.16015184e+00 -6.37984693e-01 4.03673798e-01 3.03923458e-01
-1.69962835e+00 -3.89253765e-01 1.06301332e+00 -4.64740902e-01
1.95990652e-01 2.03885809e-01 6.90248609e-01 5.05641222e-01
-1.23792395e-01 7.56937027e-01 1.29744577e+00 5.73634589e-03
3.56454492e-01 2.91971892e-01 3.20127368e-01 3.75748456e-01
4.76490587e-01 -4.06122088e-01 4.60537337e-02 -4.76413459e-01
6.99814081e-01 6.61543906e-01 -5.94203949e-01 -1.08817208e+00
-1.45981002e+00 6.93513989e-01 9.10324231e-02 4.08303916e-01
-1.68437675e-01 -3.39848369e-01 5.96065164e-01 -9.27809328e-02
2.31061369e-01 -4.23112959e-01 2.12874766e-02 3.20056796e-01
-7.65163541e-01 -1.29291281e-01 6.57474518e-01 1.22789633e+00
4.82643396e-01 -2.24882156e-01 2.58499652e-01 8.98566484e-01
5.26504099e-01 4.59398896e-01 5.18146455e-01 -9.70336914e-01
4.82003570e-01 8.10594618e-01 1.00843556e-01 -1.44922280e+00
-1.73445374e-01 -2.91467577e-01 -1.27992237e+00 -5.60728321e-03
9.50308666e-02 1.71475351e-01 -4.93530750e-01 1.35057700e+00
7.21523285e-01 4.73683089e-01 3.83681357e-01 1.23906159e+00
6.49083197e-01 8.88544440e-01 -3.22445035e-01 -6.61449969e-01
9.88020837e-01 -7.13357210e-01 -7.50160456e-01 2.63158441e-01
2.02035189e-01 -7.07268715e-01 8.05245996e-01 6.33594751e-01
-8.48818660e-01 -6.90480411e-01 -9.48620856e-01 4.69668448e-01
2.32631817e-01 4.81067926e-01 4.07790512e-01 4.69570667e-01
-7.97852397e-01 7.83714727e-02 -8.85141134e-01 -5.40835440e-01
1.12614378e-01 4.00553793e-01 -6.26501501e-01 -2.18076095e-01
-4.85001624e-01 8.31840187e-02 8.77690732e-01 2.67624795e-01
-5.72980642e-01 -2.17493013e-01 -6.18197620e-01 -6.19957782e-02
2.29568928e-01 -5.78919291e-01 6.97359860e-01 -7.19329715e-01
-9.57955301e-01 6.87320054e-01 -5.14882028e-01 1.66403979e-01
2.50999749e-01 -4.25458439e-02 -5.59594572e-01 2.65961558e-01
2.05719382e-01 2.80009583e-02 7.22755969e-01 -1.81607592e+00
-7.59508491e-01 -8.24827492e-01 -5.94377756e-01 4.80583698e-01
-7.25908995e-01 -4.91021909e-02 -1.06627977e+00 -2.41426736e-01
9.77301896e-01 -1.04590595e+00 -4.95004237e-01 -4.67574954e-01
-5.24025083e-01 -3.39827985e-01 1.41495538e+00 -4.28768635e-01
1.24866533e+00 -2.50537658e+00 6.60643995e-01 6.61188364e-01
5.46766639e-01 -4.66947295e-02 2.76797354e-01 4.12609607e-01
-1.91347629e-01 -2.45860860e-01 -1.65467039e-01 -1.04334258e-01
-4.45154577e-01 1.64718568e-01 -3.31311166e-01 9.80602860e-01
-6.69277847e-01 4.37345728e-02 -5.58438003e-01 -6.24861598e-01
3.98713738e-01 1.71091542e-01 -4.26713347e-01 3.01440060e-01
4.19198155e-01 4.98310924e-01 -6.03648961e-01 4.33217615e-01
1.13206708e+00 -2.61352450e-01 5.07588625e-01 -4.95087147e-01
-5.01933172e-02 -6.22449756e-01 -1.87096381e+00 1.72830808e+00
1.14139095e-01 -6.49128631e-02 4.05234069e-01 -1.03259158e+00
8.21392715e-01 2.98527718e-01 8.49713743e-01 1.27096968e-02
1.65465653e-01 2.75560558e-01 -2.89343279e-02 -3.90877724e-01
1.16552673e-01 -1.86047643e-01 1.28617465e-01 4.53623265e-01
-1.29976243e-01 3.61692518e-01 1.87196642e-01 3.95120621e-01
4.47554111e-01 -1.83918327e-01 4.91053581e-01 -4.75364298e-01
1.06951416e+00 9.89528000e-02 9.70963538e-01 1.55056566e-01
-2.99415886e-01 6.05920196e-01 9.98295620e-02 -3.01352590e-01
-1.00089407e+00 -1.01802886e+00 3.70823219e-02 4.11910564e-01
6.98234916e-01 -5.47685742e-01 -8.78612220e-01 -7.73858488e-01
-2.24949867e-01 3.37424904e-01 -1.71452209e-01 -2.82244105e-02
-4.37540323e-01 -6.13391995e-01 -8.92425999e-02 5.03666341e-01
5.68417847e-01 -3.83120984e-01 -3.36324722e-01 4.55983616e-02
-2.30542526e-01 -1.11024702e+00 -6.84184611e-01 -4.79706109e-01
-1.26997709e+00 -1.21287847e+00 -8.02291095e-01 -1.08027351e+00
9.38427687e-01 1.32164216e+00 4.95042473e-01 1.79511040e-01
1.68576464e-01 6.01304650e-01 -2.70484298e-01 1.30280420e-01
-1.17932938e-01 -2.70387083e-01 6.96285605e-01 4.23898667e-01
6.46172285e-01 -7.09595621e-01 -6.27338409e-01 7.87258923e-01
-8.96827877e-01 3.74378301e-02 4.22094792e-01 6.87946200e-01
9.88014877e-01 6.54366910e-01 3.53785127e-01 -7.14902341e-01
4.23198134e-01 -5.07852972e-01 -6.92641914e-01 2.60106027e-01
-5.81118405e-01 -3.53155971e-01 8.73595417e-01 -1.80499107e-01
-1.03478873e+00 4.55208033e-01 4.64398652e-01 -1.12453997e+00
-4.87488717e-01 3.12978745e-01 -7.34590113e-01 3.33519757e-01
9.42189246e-02 7.36787140e-01 1.95859522e-01 -6.46670401e-01
4.96480376e-01 8.03359866e-01 6.45234525e-01 -4.84003752e-01
1.01039088e+00 1.00094032e+00 -7.14151710e-02 -7.30251849e-01
-4.05579239e-01 -1.18404686e+00 -8.62726986e-01 -3.40477496e-01
8.49236369e-01 -1.23231173e+00 -6.97246194e-01 2.67389685e-01
-9.51931059e-01 5.98049402e-01 3.27294469e-01 8.16943586e-01
-5.69616973e-01 8.57814908e-01 -2.88690239e-01 -7.75182188e-01
-2.11697638e-01 -1.07589734e+00 8.13796759e-01 2.95146257e-01
2.65443474e-01 -1.06375897e+00 2.62808412e-01 3.83678377e-01
-2.80220985e-01 1.60782248e-01 5.17207205e-01 -7.04486728e-01
-4.38054860e-01 6.37202486e-02 -2.58930147e-01 7.04843104e-02
4.81247336e-01 1.10967480e-01 -6.62805676e-01 -7.86302924e-01
5.40599942e-01 2.01152802e-01 7.03711212e-01 3.08392555e-01
1.29566264e+00 -1.29650697e-01 -7.00037003e-01 7.20338583e-01
1.65492356e+00 5.36213994e-01 2.53985912e-01 1.19598016e-01
1.01150155e+00 8.12610388e-01 6.88405752e-01 5.13674200e-01
1.09071210e-01 4.85491484e-01 1.86761230e-01 -1.26286909e-01
4.27967846e-01 -1.45861939e-01 2.87644655e-01 1.41816127e+00
-2.18199976e-02 -2.20903773e-02 -8.57449114e-01 4.87600058e-01
-2.16918898e+00 -1.20657098e+00 -6.36795938e-01 2.20268035e+00
1.80841580e-01 -1.40759155e-01 3.17961872e-01 2.51589566e-01
1.11302817e+00 6.37254491e-02 -5.12531221e-01 1.33920476e-01
-9.35894623e-02 -6.63816333e-01 2.33578295e-01 1.11208566e-01
-1.31152534e+00 7.42464423e-01 5.86179638e+00 7.62477160e-01
-8.77782285e-01 1.14324372e-02 4.28551376e-01 -5.11616748e-03
-3.33979040e-01 6.58516437e-02 -6.61454916e-01 3.67926151e-01
3.43385667e-01 -4.16090280e-01 3.99786770e-01 1.00236392e+00
3.79906267e-01 7.46989623e-02 -8.39940131e-01 1.51388478e+00
2.58392543e-01 -1.06049287e+00 2.49603450e-01 1.64205804e-01
8.28272402e-01 -3.44270378e-01 1.01727629e-02 -2.17436925e-01
-1.96865313e-02 -4.83509481e-01 2.58744895e-01 3.22355151e-01
6.95260227e-01 -1.13070369e+00 3.93840730e-01 7.14153290e-01
-1.47631001e+00 -1.13417402e-01 -6.74326718e-01 1.79144904e-01
-1.13243265e-02 5.89446366e-01 -5.87534845e-01 1.02898943e+00
7.37734199e-01 1.23469388e+00 -5.59785783e-01 1.03762794e+00
1.49744719e-01 4.70214963e-01 -1.68739125e-01 3.04601550e-01
1.86389282e-01 -1.01648295e+00 7.31906116e-01 8.52450430e-01
3.42686802e-01 4.94426578e-01 4.87096101e-01 7.29774237e-01
4.34786528e-01 4.27716225e-01 -1.02296972e+00 2.08963737e-01
5.44997096e-01 1.54689538e+00 -1.06135821e+00 -3.40462714e-01
-7.10182965e-01 1.16172421e+00 1.35889024e-01 4.49961036e-01
-5.70853293e-01 -1.04395375e-01 3.95449460e-01 -1.59865856e-01
2.54699737e-01 -4.74682242e-01 -2.46299520e-01 -1.47275662e+00
1.47362217e-01 -8.75212848e-01 6.72198892e-01 -4.72816080e-01
-1.35990369e+00 2.79861391e-01 -2.06592283e-03 -1.94475508e+00
1.70383036e-01 -3.32836211e-01 -8.26690435e-01 6.51025593e-01
-8.21587384e-01 -8.99600744e-01 -4.31028813e-01 1.22409976e+00
4.53833520e-01 -5.83805978e-01 5.66326559e-01 1.54952824e-01
-8.41099560e-01 5.96706532e-02 8.20558906e-01 1.35391966e-01
7.31064379e-01 -1.18734610e+00 -2.79046565e-01 1.00701058e+00
1.86888888e-01 1.12322342e+00 4.34861183e-01 -6.48858368e-01
-1.65132558e+00 -9.94746387e-01 1.16933458e-01 -1.02001190e-01
3.68465424e-01 -2.97213882e-01 -1.00425112e+00 6.76993549e-01
3.46941620e-01 -1.31401137e-01 1.17887497e+00 1.02351360e-01
-3.12060207e-01 -2.32778981e-01 -1.01161718e+00 4.40661341e-01
7.36949563e-01 -1.73107833e-01 -6.28584683e-01 4.08833444e-01
5.11919558e-01 -2.10012808e-01 -8.29456985e-01 3.24293703e-01
3.50383788e-01 -1.26165617e+00 8.78766239e-01 -4.73452985e-01
9.45264176e-02 -1.03336513e+00 -4.71901387e-01 -1.39353800e+00
-5.74077427e-01 -2.02779323e-01 -1.19825520e-01 1.47325742e+00
-2.51010627e-01 -5.94332755e-01 9.07040417e-01 2.47823760e-01
1.64740607e-01 -4.60356802e-01 -8.17105830e-01 -1.03641927e+00
-2.66983896e-01 -7.90476426e-03 4.47070003e-01 1.13144028e+00
-5.35775833e-02 4.66073871e-01 -3.45161885e-01 5.41420102e-01
1.19967318e+00 6.74681067e-01 9.16206896e-01 -1.34680724e+00
-1.98927373e-01 -2.35552371e-01 -5.02007425e-01 -8.16660285e-01
2.18653053e-01 -7.96661019e-01 -2.21260220e-01 -1.57079554e+00
9.51009512e-01 -3.58394176e-01 -4.72714424e-01 -1.77617520e-01
-4.73798931e-01 -1.93587780e-01 2.16882572e-01 8.95427167e-01
-9.52024579e-01 5.91202676e-01 1.01886046e+00 -1.47751840e-02
-3.14793050e-01 7.46690705e-02 -6.56316817e-01 7.82450855e-01
8.75578821e-01 -1.92438170e-01 -7.55189002e-01 -2.84505069e-01
-5.69906116e-01 1.26253128e-01 2.25944772e-01 -1.09772289e+00
4.23480064e-01 -2.86965102e-01 5.36067188e-01 -1.06696856e+00
1.55033082e-01 -1.45861495e+00 4.01497036e-01 6.19248569e-01
2.50910819e-01 -6.90389797e-02 -6.64822683e-02 1.07037807e+00
-4.94171083e-01 2.21040443e-01 9.16229308e-01 6.77946284e-02
-9.26154673e-01 4.34655607e-01 -2.33319141e-02 -2.16117397e-01
1.57089162e+00 -2.63990998e-01 1.03477068e-01 -2.95259386e-01
-7.10935414e-01 5.81718147e-01 8.05026293e-01 3.47850800e-01
9.24748838e-01 -1.84272230e+00 -5.97190320e-01 2.91174889e-01
3.44787121e-01 -2.06022877e-02 4.70639050e-01 7.85272241e-01
-2.93137878e-01 4.79403228e-01 -2.40412965e-01 -1.13113713e+00
-1.63497233e+00 1.19628513e+00 1.59158841e-01 2.68189609e-01
-7.46335208e-01 2.52969444e-01 4.63505268e-01 -4.93101180e-01
2.82290336e-02 3.88590485e-01 -5.44294119e-01 -7.46769756e-02
6.54401898e-01 5.98410547e-01 -4.83716577e-01 -9.13149476e-01
-4.49871510e-01 9.21769857e-01 -5.65959699e-02 -1.88537329e-01
1.20320261e+00 -4.62312788e-01 -5.36535561e-01 5.39178848e-01
1.39257181e+00 1.18919365e-01 -9.93948996e-01 -2.49491215e-01
-3.50583754e-02 -6.78442121e-01 -1.62966862e-01 1.10094927e-01
-1.30192292e+00 7.96343386e-01 5.89192927e-01 8.91928896e-02
1.43587220e+00 -3.84767540e-02 6.10080123e-01 2.85117656e-01
5.48105121e-01 -9.78128552e-01 -2.29448169e-01 -9.13292021e-02
6.58555865e-01 -1.10324824e+00 1.61857858e-01 -7.04145133e-01
-6.94459021e-01 1.05064607e+00 9.85021830e-01 -2.22102031e-01
5.50150037e-01 -2.11200267e-01 6.14510197e-03 -3.72414142e-01
-5.06949663e-01 1.39405429e-01 9.01431069e-02 3.49264175e-01
1.48706883e-01 1.22853413e-01 -5.58789611e-01 5.83382845e-01
3.37725461e-01 -2.86623865e-01 4.29830402e-01 7.57535517e-01
-5.71321368e-01 -1.09923041e+00 -9.44255948e-01 2.06827626e-01
-1.31222680e-01 5.13089001e-01 -5.53833663e-01 6.02466226e-01
-1.36021301e-01 1.15645278e+00 -7.68268779e-02 -5.98954141e-01
2.45703712e-01 -1.69809714e-01 1.09675661e-01 -5.98926008e-01
-2.99112108e-02 8.78218591e-01 -5.72459280e-01 -6.36215746e-01
-7.49237180e-01 -8.93005073e-01 -1.37339795e+00 -1.51682988e-01
-1.60681024e-01 5.92253089e-01 2.68109143e-01 4.55951691e-01
3.82242650e-01 1.19376138e-01 1.29473567e+00 -3.75242084e-01
-4.99064624e-01 -4.77072299e-01 -1.06218159e+00 6.38725042e-01
-8.59636292e-02 -5.24241745e-01 -3.20827216e-01 1.31205633e-01] | [8.155718803405762, 4.625516414642334] |
d8b8a4ef-67eb-47aa-b1bf-2c41a1cffaa5 | 190909803 | 1909.09803 | null | https://arxiv.org/abs/1909.09803v4 | https://arxiv.org/pdf/1909.09803v4.pdf | Visual Odometry Revisited: What Should Be Learnt? | In this work we present a monocular visual odometry (VO) algorithm which leverages geometry-based methods and deep learning. Most existing VO/SLAM systems with superior performance are based on geometry and have to be carefully designed for different application scenarios. Moreover, most monocular systems suffer from scale-drift issue.Some recent deep learning works learn VO in an end-to-end manner but the performance of these deep systems is still not comparable to geometry-based methods. In this work, we revisit the basics of VO and explore the right way for integrating deep learning with epipolar geometry and Perspective-n-Point (PnP) method. Specifically, we train two convolutional neural networks (CNNs) for estimating single-view depths and two-view optical flows as intermediate outputs. With the deep predictions, we design a simple but robust frame-to-frame VO algorithm (DF-VO) which outperforms pure deep learning-based and geometry-based methods. More importantly, our system does not suffer from the scale-drift issue being aided by a scale consistent single-view depth CNN. Extensive experiments on KITTI dataset shows the robustness of our system and a detailed ablation study shows the effect of different factors in our system. | ['Jia-Wang Bian', 'Huangying Zhan', 'Chamara Saroj Weerasekera', 'Ian Reid'] | 2019-09-21 | null | null | null | null | ['monocular-visual-odometry'] | ['robots'] | [-4.99496371e-01 -1.61804512e-01 -8.13626796e-02 -4.69382107e-01
-2.01991096e-01 -4.41374272e-01 5.83815575e-01 -6.37686729e-01
-2.67998517e-01 5.58219790e-01 1.39299721e-01 -2.55361527e-01
3.27948362e-01 -7.13312864e-01 -9.01644051e-01 -4.50304598e-01
1.88842744e-01 4.06337917e-01 3.94426793e-01 -3.06220531e-01
3.73194546e-01 6.62154794e-01 -1.27674794e+00 -1.75318256e-01
6.42751515e-01 9.65040565e-01 1.81945860e-01 9.93503869e-01
1.27910614e-01 1.00080276e+00 -3.57192993e-01 -9.92238447e-02
8.21190774e-01 -1.77253485e-01 -6.40441179e-01 -5.82441911e-02
1.00265384e+00 -7.84591734e-01 -8.25281501e-01 7.77523696e-01
8.13359618e-01 -1.40583187e-01 3.24489295e-01 -1.31477690e+00
-4.60730702e-01 -3.17309350e-01 -4.25212502e-01 8.52072909e-02
5.66695929e-01 4.69859838e-01 8.25575590e-01 -1.19881201e+00
8.31628561e-01 1.31487668e+00 9.29649472e-01 2.78934360e-01
-8.44231725e-01 -4.60392594e-01 9.55867469e-02 1.51416123e-01
-1.26358986e+00 -4.78605419e-01 7.79642224e-01 -5.39237738e-01
1.25860584e+00 -3.97356361e-01 8.01481903e-01 8.23793530e-01
5.24981558e-01 8.19813013e-01 7.83287883e-01 -9.03966203e-02
-4.07952350e-03 -1.37332126e-01 -6.22106075e-01 9.50032532e-01
3.59313399e-01 4.21146244e-01 -6.19734287e-01 3.77796799e-01
1.16935754e+00 1.38592392e-01 -3.56231689e-01 -1.12103558e+00
-1.45341933e+00 7.45461762e-01 1.01276803e+00 -2.07903296e-01
-1.55477747e-01 5.03252089e-01 2.05989987e-01 3.80682141e-01
3.61300021e-01 2.63815373e-01 -6.52873218e-01 4.47925776e-02
-8.01159918e-01 2.68423527e-01 5.75454354e-01 1.27939177e+00
1.10527086e+00 1.91446826e-01 3.89895022e-01 3.04526120e-01
5.50077617e-01 5.22259176e-01 3.05547982e-01 -1.20973778e+00
5.90661764e-01 6.05333388e-01 1.24328882e-01 -1.06734920e+00
-8.47759545e-01 -3.49835336e-01 -7.35542595e-01 5.61677575e-01
3.15190256e-01 -3.28145832e-01 -9.90750909e-01 1.44621551e+00
4.21278894e-01 2.05208004e-01 1.48576096e-01 1.16169131e+00
9.52237427e-01 2.06138879e-01 -5.21947086e-01 3.18569422e-01
7.62843311e-01 -1.36370504e+00 -5.79098105e-01 -5.49276292e-01
8.76512945e-01 -7.89305806e-01 7.75580645e-01 3.04892987e-01
-9.11468267e-01 -6.73092544e-01 -1.37107849e+00 -6.76501453e-01
-1.77362025e-01 1.38808385e-01 6.79913819e-01 4.11437154e-01
-1.33423841e+00 5.02153337e-01 -8.81382167e-01 -6.13911629e-01
2.05641225e-01 4.52092797e-01 -5.55761755e-01 -1.71432376e-01
-9.33754385e-01 1.01477456e+00 1.62390366e-01 1.53155461e-01
-8.54809701e-01 -6.59756601e-01 -1.26053751e+00 -3.54006201e-01
1.07423365e-01 -1.36886179e+00 1.26432228e+00 -6.41648650e-01
-1.97518921e+00 8.95114362e-01 -4.12104338e-01 -5.15382469e-01
9.23063755e-01 -4.77821350e-01 7.97913596e-02 2.34953642e-01
1.05137631e-01 1.00261188e+00 4.35056537e-01 -1.23086894e+00
-7.83976614e-01 -3.35773617e-01 3.77741158e-01 5.27431369e-01
4.65523690e-01 -6.00127459e-01 -4.93472636e-01 -2.58933697e-02
7.98039019e-01 -1.07758737e+00 -1.95547149e-01 5.32461822e-01
-3.24521095e-01 2.51978308e-01 7.81518638e-01 -2.35579550e-01
6.77701414e-01 -1.81068635e+00 3.93144824e-02 -4.89040017e-01
3.35571170e-01 1.51041389e-01 1.69377223e-01 4.00972158e-01
2.98333853e-01 -2.55397260e-01 2.56079495e-01 -7.77309239e-01
-1.11163795e-01 3.25145364e-01 -1.78561494e-01 9.97349679e-01
1.19018890e-01 1.05741465e+00 -1.05536330e+00 -3.15662563e-01
8.32228720e-01 5.25965095e-01 -7.74933398e-01 3.99719834e-01
-8.04436356e-02 8.80999267e-01 -3.13113369e-02 8.94817412e-01
8.70299280e-01 -2.07021654e-01 -9.41219181e-02 -3.58433574e-01
-3.33335161e-01 5.53354800e-01 -1.08840919e+00 2.17839861e+00
-7.06393301e-01 9.87518191e-01 -1.35526583e-01 -5.56404114e-01
9.27506864e-01 4.01968732e-02 4.26139832e-01 -7.53966868e-01
1.37381330e-01 4.61198449e-01 -2.33295411e-01 -4.60654974e-01
7.34665275e-01 -1.95774168e-01 1.86551899e-01 -1.76658243e-01
1.67955756e-01 -4.73587155e-01 -3.97478700e-01 9.49397385e-02
7.90215373e-01 7.96838403e-01 5.49577773e-01 -1.01552578e-02
5.37395656e-01 8.10714811e-02 7.75929034e-01 4.93941784e-01
-7.08312273e-01 1.08630180e+00 2.09712297e-01 -7.91511297e-01
-1.23888874e+00 -9.61156070e-01 -7.04312921e-02 4.05283630e-01
7.58932710e-01 -1.65238589e-01 -1.52572662e-01 -4.24057841e-01
1.67383730e-01 1.03819042e-01 -3.21523756e-01 1.77357756e-02
-6.21191382e-01 -2.77681619e-01 4.19512242e-01 6.50671721e-01
8.63136113e-01 -5.63741922e-01 -8.11043024e-01 1.78389847e-01
-1.05220646e-01 -1.60790324e+00 -2.60198444e-01 -1.33426324e-01
-1.02523553e+00 -1.17512119e+00 -6.52440488e-01 -7.01077282e-01
3.22462201e-01 8.72909844e-01 1.05516839e+00 -1.89580619e-01
1.66954532e-01 1.95197940e-01 -1.57966316e-01 -3.13489407e-01
6.41459078e-02 1.86388507e-01 3.26024294e-01 -2.64187515e-01
3.19817990e-01 -6.99591398e-01 -1.13675916e+00 4.32394952e-01
-3.72378647e-01 2.55826563e-01 5.98698974e-01 6.77178621e-01
4.39264148e-01 -6.71612084e-01 6.65892959e-02 -3.98995608e-01
-2.29569152e-01 -3.84331733e-01 -8.32245708e-01 -2.95769542e-01
-5.75777829e-01 -3.73020731e-02 4.63774770e-01 5.14153093e-02
-7.74739683e-01 5.00705957e-01 -2.27861673e-01 -6.54122472e-01
-5.31772748e-02 1.03267640e-01 -1.45634368e-01 -5.34526646e-01
7.05316067e-01 1.34270579e-01 -8.70471522e-02 -2.46601135e-01
4.85323101e-01 5.11176407e-01 6.21270478e-01 -6.55688345e-02
8.62591445e-01 1.26771128e+00 2.32380196e-01 -6.34590387e-01
-9.40761209e-01 -6.94950223e-01 -1.20332491e+00 -1.11450821e-01
9.11897182e-01 -1.67624784e+00 -8.42562377e-01 6.56826079e-01
-1.30186224e+00 -3.62627447e-01 3.96907702e-02 8.35102141e-01
-8.99432361e-01 5.05677283e-01 -4.50319499e-01 -5.82460165e-01
-2.21459344e-01 -1.33034992e+00 1.45878208e+00 3.08024764e-01
2.12232023e-01 -1.14508951e+00 2.79728651e-01 1.90623984e-01
2.40226507e-01 4.45314854e-01 1.03962764e-01 -1.20704383e-01
-1.14530540e+00 -2.46051140e-02 -3.97084653e-01 1.08162776e-01
-5.41487103e-03 -1.05968758e-01 -1.20884907e+00 -5.27259409e-01
-1.08651422e-01 -3.70016426e-01 7.71026790e-01 5.38672566e-01
4.45319831e-01 1.65725797e-01 -2.80831456e-01 1.60714865e+00
1.75006497e+00 -5.50936200e-02 6.69809282e-01 8.15269947e-01
1.29418123e+00 2.53008664e-01 5.56306303e-01 2.81658888e-01
1.03085232e+00 7.38714576e-01 9.09574807e-01 -2.25058123e-01
-2.18859926e-01 -5.14155090e-01 4.29108411e-01 7.35059977e-01
3.15666236e-02 -7.78116211e-02 -7.87580967e-01 7.20126212e-01
-1.87516022e+00 -6.76564455e-01 -2.71513134e-01 1.95580137e+00
2.92108059e-01 4.43042107e-02 -1.19647250e-01 -1.15502499e-01
1.37603864e-01 4.12483126e-01 -6.63118660e-01 -3.64999771e-01
-2.81301349e-01 -2.86350965e-01 9.20938134e-01 7.43895948e-01
-1.08055913e+00 1.32547247e+00 6.11165190e+00 -1.65845945e-01
-1.47591889e+00 -2.15918534e-02 3.35017405e-02 -1.42321944e-01
-3.82322371e-02 3.05231839e-01 -1.02827775e+00 -2.51231864e-02
5.72972178e-01 1.64884627e-01 1.27643526e-01 1.02076209e+00
2.93314427e-01 -2.94881731e-01 -1.14039147e+00 1.31846261e+00
9.16602388e-02 -1.39359605e+00 -1.76669627e-01 2.24723667e-01
1.10010326e+00 8.42712402e-01 -2.30731651e-01 2.29479879e-01
4.26583946e-01 -8.06695521e-01 8.65877092e-01 3.43119264e-01
7.88688302e-01 -3.85626346e-01 8.68325353e-01 3.15148741e-01
-1.32405138e+00 -7.36074448e-02 -6.67018950e-01 -4.83524352e-01
4.58274961e-01 4.24968123e-01 -9.17580485e-01 8.25969279e-01
7.00362146e-01 1.30266392e+00 -3.82472128e-01 1.15822470e+00
-4.51328218e-01 -2.17150941e-01 -3.27250391e-01 3.85475427e-01
4.54871237e-01 3.70031260e-02 5.08997202e-01 9.77759838e-01
4.21567202e-01 -1.53486803e-01 1.64770558e-01 7.47889757e-01
1.67704765e-02 -2.99457222e-01 -1.04115486e+00 6.63552701e-01
3.59335750e-01 1.02252197e+00 -1.98604465e-01 -2.18805134e-01
-7.62116075e-01 1.09689772e+00 3.80338550e-01 2.74927348e-01
-6.06723309e-01 -1.73255980e-01 1.17328215e+00 2.20783964e-01
4.94723976e-01 -7.68199027e-01 -5.54583848e-01 -1.65656734e+00
4.40918244e-02 -3.87355328e-01 1.38333635e-02 -1.03774273e+00
-1.00804424e+00 4.40120816e-01 -3.62075686e-01 -1.62918937e+00
-2.83048660e-01 -7.50121951e-01 -4.28748637e-01 8.66103411e-01
-2.33820987e+00 -1.11508965e+00 -8.56841326e-01 6.36365235e-01
5.72830558e-01 8.23591053e-02 3.69685560e-01 2.88367331e-01
-2.25620359e-01 3.49531442e-01 -5.51110283e-02 3.17878544e-01
9.88458157e-01 -1.27669930e+00 9.25108671e-01 1.08774436e+00
-4.54968889e-04 4.88238364e-01 5.71690142e-01 -4.56113189e-01
-1.61245489e+00 -1.05938053e+00 9.48410869e-01 -7.83716321e-01
3.81394863e-01 -2.46068135e-01 -4.85598296e-01 1.06691372e+00
6.22796007e-02 4.83782321e-01 -2.82609239e-02 -2.51249880e-01
-2.71730572e-01 -1.34190485e-01 -1.01790369e+00 5.43220162e-01
1.36968493e+00 -6.57659471e-01 -4.41157401e-01 1.77871123e-01
9.04232562e-01 -1.11511302e+00 -6.21575475e-01 5.88242769e-01
6.63672686e-01 -1.60118318e+00 1.08687449e+00 -2.36902103e-01
3.02748054e-01 -7.23574519e-01 -2.93051422e-01 -1.28258443e+00
-2.65283078e-01 -6.86328173e-01 -2.34483868e-01 5.37594855e-01
-1.15836866e-01 -6.90418005e-01 8.71841908e-01 1.25152066e-01
-3.19929689e-01 -6.01587832e-01 -9.70958889e-01 -7.54519284e-01
-1.64140463e-02 -3.36085916e-01 5.50024986e-01 8.59925210e-01
-3.48861963e-01 3.77298206e-01 -6.62708461e-01 6.01233602e-01
6.46024644e-01 5.84331192e-02 1.52160597e+00 -9.89529908e-01
-1.45204559e-01 -1.35120094e-01 -9.42641258e-01 -1.82157028e+00
-2.09065333e-01 -5.50067961e-01 8.22357163e-02 -1.76940858e+00
-3.60449463e-01 -1.90735191e-01 1.60937086e-01 -1.63137726e-03
4.21291552e-02 2.92777717e-01 2.21005693e-01 4.61702228e-01
-4.92332876e-01 6.31595552e-01 1.36955428e+00 2.73164034e-01
-3.16573739e-01 -7.74559528e-02 -2.70012885e-01 8.43603373e-01
6.39494359e-01 -2.33583063e-01 -4.23785329e-01 -8.50759804e-01
3.79644394e-01 8.73798355e-02 5.50200880e-01 -1.22668183e+00
4.65354145e-01 -5.21123670e-02 4.84113425e-01 -9.17518020e-01
4.05333370e-01 -6.35785639e-01 -1.49804518e-01 5.69507241e-01
3.27533633e-01 2.76249975e-01 -2.84656975e-02 5.96630275e-01
-1.72039375e-01 4.42173064e-01 9.56250608e-01 -3.14459622e-01
-1.10575342e+00 6.32996738e-01 6.81193322e-02 8.48186202e-03
7.99962103e-01 -5.87470055e-01 -4.25428778e-01 -6.22969747e-01
-3.53557706e-01 2.83837169e-01 9.69929874e-01 5.08641660e-01
7.17480838e-01 -1.30381632e+00 -3.81276906e-01 3.44344467e-01
3.89877319e-01 7.40576088e-01 5.40325157e-02 1.01687038e+00
-1.36854696e+00 7.19651222e-01 -2.82810897e-01 -1.08645046e+00
-6.93705499e-01 4.34262991e-01 8.09990168e-01 6.64586723e-02
-8.91051233e-01 8.55693936e-01 4.22364742e-01 -8.18918824e-01
2.01212689e-01 -6.70316637e-01 1.45917326e-01 -4.29486722e-01
3.13219041e-01 2.50742972e-01 -6.03819638e-03 -7.61264145e-01
-4.85031545e-01 9.64130342e-01 2.64719486e-01 -7.58603634e-03
1.20580351e+00 -6.76458478e-01 2.49873146e-01 4.71731991e-01
1.44970250e+00 -1.93188652e-01 -1.92337322e+00 -3.81843954e-01
-3.14337909e-01 -7.39097059e-01 9.09461305e-02 -3.01120162e-01
-1.17884517e+00 1.27533686e+00 5.33096492e-01 -4.81764913e-01
8.10090065e-01 -3.44106823e-01 9.29418683e-01 3.51949543e-01
6.19136810e-01 -8.31415713e-01 1.53403267e-01 8.87472451e-01
5.33888876e-01 -1.74666119e+00 2.85241395e-01 -2.49907523e-01
-5.25334537e-01 1.22544932e+00 8.92780781e-01 -4.31419909e-01
6.43013716e-01 -7.24417483e-03 5.91345131e-01 -2.93278784e-01
-5.70186615e-01 -3.67315143e-01 1.75598741e-01 6.91430569e-01
2.66660243e-01 -2.99984992e-01 1.47160918e-01 -3.14161539e-01
-3.09630871e-01 1.50832176e-01 7.43643582e-01 9.83546078e-01
-5.07523537e-01 -7.57905602e-01 -1.25172317e-01 -4.22056049e-01
-1.47360181e-02 5.53338304e-02 -2.64665633e-01 1.02959776e+00
1.41888261e-01 6.73337996e-01 7.42319375e-02 -6.18213236e-01
4.03073579e-01 -2.95753241e-01 5.66821575e-01 -4.57853705e-01
-2.86324024e-01 -1.61717728e-01 -1.73399433e-01 -1.05471265e+00
-6.06336713e-01 -4.39083546e-01 -1.28944612e+00 -7.12433934e-01
-8.89405534e-02 -7.35301614e-01 7.96725750e-01 1.02890408e+00
5.89215040e-01 1.84193745e-01 7.00414956e-01 -1.39390111e+00
-2.48291776e-01 -6.89361811e-01 -2.90995449e-01 1.50774911e-01
8.99325073e-01 -7.26212680e-01 -3.81799489e-01 -2.70117998e-01] | [8.136763572692871, -2.2861247062683105] |
03e91c99-c83a-4aa1-b877-b20bd09fc8e1 | object-centric-learning-for-real-world-videos | 2306.04829 | null | https://arxiv.org/abs/2306.04829v1 | https://arxiv.org/pdf/2306.04829v1.pdf | Object-Centric Learning for Real-World Videos by Predicting Temporal Feature Similarities | Unsupervised video-based object-centric learning is a promising avenue to learn structured representations from large, unlabeled video collections, but previous approaches have only managed to scale to real-world datasets in restricted domains. Recently, it was shown that the reconstruction of pre-trained self-supervised features leads to object-centric representations on unconstrained real-world image datasets. Building on this approach, we propose a novel way to use such pre-trained features in the form of a temporal feature similarity loss. This loss encodes temporal correlations between image patches and is a natural way to introduce a motion bias for object discovery. We demonstrate that this loss leads to state-of-the-art performance on the challenging synthetic MOVi datasets. When used in combination with the feature reconstruction loss, our model is the first object-centric video model that scales to unconstrained video datasets such as YouTube-VIS. | ['Georg Martius', 'Maximilian Seitzer', 'Andrii Zadaianchuk'] | 2023-06-07 | null | null | null | null | ['object-discovery'] | ['computer-vision'] | [ 3.51128936e-01 -2.18837738e-01 -6.02126598e-01 -4.12112981e-01
-5.92297375e-01 -4.20702457e-01 7.17421949e-01 -3.84439044e-02
-5.06167710e-01 5.97515285e-01 2.74963766e-01 3.31565827e-01
-2.03898072e-01 -4.24168050e-01 -1.15995157e+00 -6.02319002e-01
-6.40177488e-01 3.66130561e-01 4.72466648e-01 1.13344088e-01
2.83949971e-01 3.16428244e-01 -1.87281346e+00 4.75588232e-01
2.47741565e-01 1.06181753e+00 2.83146650e-01 5.70621252e-01
1.50657266e-01 1.10069287e+00 -2.67819297e-02 2.16994822e-01
5.17080605e-01 -3.74995500e-01 -1.04369998e+00 5.67377925e-01
9.24183965e-01 -4.13506657e-01 -6.58521593e-01 7.49114573e-01
1.22573748e-01 2.81199157e-01 6.33089900e-01 -1.23156869e+00
-4.28895146e-01 3.16543579e-01 -3.52471411e-01 5.63862264e-01
4.49172378e-01 3.10744941e-01 1.18563282e+00 -9.63105202e-01
1.25181472e+00 1.24235928e+00 5.37685990e-01 4.85472262e-01
-1.41445172e+00 -1.87076867e-01 4.16461438e-01 5.61866224e-01
-1.26979625e+00 -4.41603124e-01 8.10038090e-01 -6.04695201e-01
8.47238302e-01 2.00181622e-02 7.82979310e-01 1.23731709e+00
1.41963791e-02 1.34065461e+00 8.00288498e-01 -3.39509249e-01
1.91558450e-01 -5.34900501e-02 4.18497771e-02 8.12316358e-01
2.38468483e-01 2.95699447e-01 -8.57088745e-01 1.70572065e-02
8.13095093e-01 2.90875494e-01 -3.37533265e-01 -1.12653542e+00
-1.48592293e+00 8.61790478e-01 5.30940473e-01 3.97610486e-01
-2.42658168e-01 3.67189199e-01 5.04600048e-01 3.74837667e-01
5.87245882e-01 4.19137836e-01 -4.35501784e-01 -1.95588976e-01
-1.16866744e+00 3.67990702e-01 4.86801267e-01 9.71991241e-01
9.75503564e-01 1.28264636e-01 -1.49236336e-01 5.23798287e-01
1.67829216e-01 2.79974788e-01 6.88328922e-01 -1.18683898e+00
2.68228054e-01 4.45701063e-01 6.16735369e-02 -9.30359066e-01
-1.97604597e-01 -1.31449968e-01 -4.32342649e-01 2.33666614e-01
4.52260286e-01 1.87722355e-01 -9.71895933e-01 1.81077504e+00
1.57001078e-01 4.25317466e-01 -6.48876801e-02 9.07712221e-01
5.14839232e-01 4.60794568e-01 -1.76789612e-01 -1.89691886e-01
8.16820264e-01 -1.20052767e+00 -4.04794425e-01 4.95450646e-02
6.48473680e-01 -2.93752789e-01 8.82125497e-01 4.07759517e-01
-1.07716012e+00 -6.10146582e-01 -9.46822464e-01 8.54672566e-02
-6.12480156e-02 -1.35906860e-01 7.60537207e-01 3.00723702e-01
-1.03990018e+00 9.14533734e-01 -1.10438371e+00 -8.18785429e-01
7.56818473e-01 3.23972106e-01 -6.94833398e-01 -2.36251414e-01
-5.45506597e-01 5.04025698e-01 4.16524082e-01 -2.90346026e-01
-1.46557605e+00 -7.37629533e-01 -9.70794499e-01 -1.33661777e-01
4.92035389e-01 -5.49623609e-01 9.45100546e-01 -1.48876429e+00
-1.22798264e+00 9.03341532e-01 -2.15243608e-01 -8.21797431e-01
4.18680906e-01 -3.20832789e-01 -1.34387553e-01 7.29567647e-01
1.34445131e-01 9.59991634e-01 1.22917736e+00 -1.26224804e+00
-5.61522782e-01 -2.58403003e-01 1.46567002e-01 -5.54897413e-02
-6.26406014e-01 -1.31736338e-01 -4.69576776e-01 -8.25421333e-01
-1.45296842e-01 -1.08580053e+00 -2.88899243e-01 3.99005771e-01
-1.83954649e-02 -1.22535884e-01 1.15509343e+00 -2.77736992e-01
8.64767969e-01 -2.14105964e+00 3.98399353e-01 -7.37556741e-02
4.95612249e-02 2.73888856e-01 -4.50389028e-01 3.63988161e-01
-1.14479020e-01 -4.58340533e-02 -3.02446336e-01 -2.48299539e-01
-4.19643581e-01 2.86598623e-01 -1.47643670e-01 6.53596401e-01
5.73442519e-01 9.92216647e-01 -1.22829485e+00 -4.67109680e-01
2.58312196e-01 3.91617030e-01 -8.97687674e-01 2.73866922e-01
-4.06534523e-01 5.98128319e-01 -3.98801506e-01 5.90410471e-01
3.43889862e-01 -4.37087804e-01 1.23296969e-01 -3.95359807e-02
1.52276367e-01 -2.74188649e-02 -9.32331145e-01 2.19429755e+00
-2.14172423e-01 9.74103987e-01 -1.16980948e-01 -1.47807264e+00
4.56756800e-01 3.06927353e-01 9.98225212e-01 -3.82379562e-01
-8.12889189e-02 9.37011540e-02 -1.97125241e-01 -6.77633226e-01
3.11427772e-01 7.11353049e-02 2.76463479e-01 2.95735896e-01
6.83015645e-01 2.38250103e-02 5.25628984e-01 4.23454881e-01
1.32049346e+00 4.68559533e-01 1.52933329e-01 -3.79886419e-01
4.11857009e-01 9.94758308e-02 4.93283302e-01 6.54500663e-01
-1.37526527e-01 9.83124614e-01 5.71576282e-02 -7.90554702e-01
-1.05501759e+00 -1.02088451e+00 -9.93162245e-02 8.44468594e-01
1.10290736e-01 -4.31592315e-01 -3.89332205e-01 -1.05644357e+00
7.16158673e-02 1.39791831e-01 -7.30363190e-01 -1.97640166e-01
-6.36664689e-01 -1.80311248e-01 4.76711653e-02 6.26150489e-01
2.43797943e-01 -1.14434338e+00 -6.20986581e-01 3.47130358e-01
-5.68562075e-02 -1.21120048e+00 -4.79364514e-01 3.01598042e-01
-1.32810056e+00 -1.20971799e+00 -7.98178852e-01 -7.96524227e-01
7.72604227e-01 5.51821768e-01 1.21146345e+00 1.44346505e-01
-6.65708005e-01 9.94094193e-01 -6.28772736e-01 -4.13325801e-02
-5.90274744e-02 -8.75914916e-02 2.66683966e-01 3.55576396e-01
2.01568216e-01 -6.50652766e-01 -7.25251198e-01 3.12057465e-01
-1.20726502e+00 -3.33039522e-01 2.95674801e-01 1.07741022e+00
6.56428218e-01 -3.74038100e-01 2.86793172e-01 -8.07322443e-01
-2.19080180e-01 -5.81859469e-01 -4.38764453e-01 4.95772399e-02
-4.34003443e-01 2.44570926e-01 6.22625113e-01 -7.51398563e-01
-6.95266783e-01 4.84794915e-01 3.20288092e-01 -8.78547668e-01
-1.79660380e-01 4.15024817e-01 6.44202232e-02 -2.32821971e-01
6.02433085e-01 3.65536034e-01 1.11825898e-01 -4.17875588e-01
4.33348000e-01 1.12387374e-01 4.05064642e-01 -6.01185977e-01
9.35523391e-01 9.81859863e-01 -3.17379646e-02 -1.01445568e+00
-8.25179994e-01 -8.47649992e-01 -8.26964796e-01 -1.90946534e-01
7.67441452e-01 -1.20782447e+00 -3.46794099e-01 1.48175180e-01
-7.65644670e-01 -4.02460396e-01 -7.89344609e-01 6.51092947e-01
-1.23123276e+00 5.23229539e-01 -3.27278465e-01 -6.19129241e-01
9.86904576e-02 -8.79811764e-01 1.13169873e+00 -5.07280454e-02
-1.15904361e-01 -9.30142701e-01 1.74033105e-01 2.66988665e-01
2.42735669e-01 2.44660035e-01 5.39053679e-01 -4.88098264e-01
-1.03008068e+00 -5.66069819e-02 -4.59884927e-02 6.02374196e-01
1.18268974e-01 -6.04852028e-02 -7.87239492e-01 -6.99978828e-01
-7.88523853e-02 -8.09219718e-01 1.32552266e+00 3.61899316e-01
1.27587199e+00 -2.39936873e-01 -2.83340126e-01 7.90726781e-01
1.60807765e+00 -9.97024328e-02 5.18748999e-01 2.42013872e-01
7.25071609e-01 4.03854281e-01 7.87680566e-01 4.31819290e-01
1.94757402e-01 8.46589208e-01 5.44850767e-01 1.46562308e-01
-2.40391687e-01 -3.40581775e-01 5.19720793e-01 5.75544655e-01
-3.65289986e-01 -1.52717948e-01 -5.84100068e-01 1.04502821e+00
-2.06214547e+00 -1.34181058e+00 3.33910912e-01 2.28158355e+00
5.15416801e-01 1.01737380e-01 3.68936002e-01 3.18785310e-02
3.46500367e-01 4.48399484e-01 -5.01005709e-01 1.15402430e-01
-2.28580311e-01 9.70396698e-02 3.87421191e-01 1.47777498e-01
-1.41734362e+00 9.23458874e-01 6.85863590e+00 5.67788839e-01
-1.23073912e+00 2.60874778e-01 3.86305481e-01 -3.81937653e-01
-1.40891537e-01 1.23139337e-01 -5.70045471e-01 3.73537898e-01
8.68110836e-01 -1.39457896e-01 2.65021950e-01 1.00043857e+00
1.45171449e-01 -1.09754093e-01 -1.55229664e+00 1.01005197e+00
2.48899758e-01 -1.67594349e+00 2.67429858e-01 1.83364615e-01
1.02968347e+00 2.49366611e-01 1.25965597e-02 1.24891236e-01
1.06225662e-01 -8.72835815e-01 8.14067423e-01 4.40299392e-01
6.49594128e-01 -4.20481205e-01 3.05238038e-01 1.07762218e-01
-1.25328827e+00 -2.62400180e-01 -4.29703087e-01 -1.86232533e-02
2.34148070e-01 3.19256425e-01 -6.51303172e-01 3.65630299e-01
7.84801424e-01 1.47085094e+00 -4.64610875e-01 1.24913144e+00
1.31842852e-01 7.97704816e-01 -3.45134169e-01 2.90164083e-01
4.06985462e-01 2.37294331e-01 7.74167478e-01 1.05834413e+00
8.14952925e-02 -2.41578549e-01 4.81779784e-01 4.75905091e-01
-1.99158132e-01 2.09577680e-02 -9.61790144e-01 -3.36040169e-01
-1.68853506e-01 1.00040138e+00 -6.50656700e-01 -3.40361238e-01
-5.95433533e-01 1.19473839e+00 5.96698999e-01 4.85239834e-01
-4.34329182e-01 1.02885991e-01 7.69288123e-01 5.67524016e-01
9.03329134e-01 -4.42174524e-01 4.05835629e-01 -1.56696367e+00
1.42917946e-01 -8.26542437e-01 3.40887994e-01 -3.51297259e-01
-1.31150043e+00 3.85842741e-01 8.26946273e-02 -1.66320395e+00
-3.89650345e-01 -7.32365668e-01 -5.08372307e-01 -2.72581782e-02
-1.64667034e+00 -1.04542267e+00 -2.15763509e-01 9.29735482e-01
9.64243174e-01 -3.99047732e-01 6.75037444e-01 2.09462762e-01
-8.24367702e-02 3.87493134e-01 3.97971272e-01 1.65875316e-01
6.89556062e-01 -1.01412332e+00 9.18498039e-02 7.37824321e-01
8.04560840e-01 4.69560444e-01 4.87811267e-01 -3.86420518e-01
-1.63776040e+00 -1.12421739e+00 3.96124482e-01 -6.67250931e-01
5.70694804e-01 -4.22391355e-01 -7.78067350e-01 8.53033781e-01
8.10003504e-02 8.12088370e-01 4.95821804e-01 -8.87184516e-02
-4.43061501e-01 -1.69848919e-01 -9.53932166e-01 2.25001246e-01
1.43243301e+00 -4.78781670e-01 -4.36486304e-01 5.64959466e-01
6.91402197e-01 8.54853392e-02 -7.61436701e-01 3.72555107e-01
5.07500768e-01 -7.92404115e-01 1.09551358e+00 -9.97067273e-01
5.87563336e-01 -3.13965380e-01 -3.00102085e-01 -1.03250110e+00
-3.65776688e-01 -7.30388701e-01 -4.34371322e-01 8.53974640e-01
2.12184250e-01 -1.04929730e-01 1.02643061e+00 9.59848166e-02
1.54729754e-01 -7.17706323e-01 -8.84612203e-01 -1.29481840e+00
-1.59925878e-01 -3.22515100e-01 -1.35720968e-01 6.82347894e-01
-5.21579534e-02 -2.37314962e-02 -6.22208714e-01 -1.81090504e-01
7.98885405e-01 1.98798880e-01 7.86504030e-01 -1.26958156e+00
-3.05217117e-01 -8.00945237e-02 -1.20541596e+00 -1.28780997e+00
2.98407882e-01 -8.96302819e-01 3.60114835e-02 -1.17815757e+00
4.18162078e-01 -2.81910539e-01 -5.08841276e-01 1.85608849e-01
-5.99148916e-03 4.49276745e-01 4.02591825e-01 3.33076686e-01
-1.17672420e+00 7.14146733e-01 1.01023364e+00 -4.00333911e-01
-8.71252269e-02 -1.95366591e-01 -2.86187083e-01 7.00376570e-01
3.16329420e-01 -6.22396171e-01 -5.52857995e-01 -4.95740801e-01
-2.52366424e-01 -2.08867937e-01 4.95586097e-01 -1.16245663e+00
1.39009386e-01 -2.24813849e-01 4.02743578e-01 -4.15847510e-01
4.02471840e-01 -9.07744586e-01 -1.94547743e-01 3.88655931e-01
-4.14526820e-01 -3.11231315e-01 -7.60375410e-02 1.05066085e+00
-5.81991076e-01 -1.16349816e-01 7.83582211e-01 -3.87381077e-01
-1.15325272e+00 6.84041321e-01 -2.89492548e-01 3.63879561e-01
1.17659056e+00 -3.04257512e-01 2.03974009e-01 -4.42935437e-01
-7.09548295e-01 1.45892814e-01 5.95596254e-01 6.97033882e-01
7.74506807e-01 -1.38940156e+00 -6.98773324e-01 1.67068928e-01
4.12269413e-01 -1.55773118e-01 1.14098042e-01 7.07726121e-01
-4.30218339e-01 4.87517715e-01 -4.94885504e-01 -1.08844829e+00
-1.18064654e+00 9.44139004e-01 -8.11209679e-02 -2.97872961e-01
-9.16186452e-01 8.66575360e-01 2.56627172e-01 -4.11340445e-02
3.85135978e-01 -2.69973010e-01 -4.40231934e-02 -3.06143593e-02
6.68360114e-01 5.67426207e-03 -2.12498069e-01 -7.73729622e-01
-4.62508321e-01 6.23690367e-01 -1.81284159e-01 -1.48900047e-01
1.70555151e+00 -1.73176557e-01 3.46071929e-01 5.58924615e-01
1.61245930e+00 -3.36060077e-01 -1.75423837e+00 -4.63509917e-01
1.89371347e-01 -8.51950049e-01 -1.22706518e-01 -1.08844288e-01
-1.22422743e+00 8.34517360e-01 6.41478539e-01 -1.53146207e-01
9.79294896e-01 7.11394995e-02 6.76496089e-01 6.27724051e-01
8.69786382e-01 -1.03102839e+00 7.35847652e-01 4.29361731e-01
7.75490820e-01 -1.59983659e+00 1.88819245e-02 -2.25717396e-01
-5.90900481e-01 1.09036851e+00 4.01328027e-01 -5.88327169e-01
7.92523444e-01 -9.24191996e-02 -2.59757012e-01 -1.45147666e-01
-8.95249188e-01 -3.91093433e-01 3.76337141e-01 7.75966048e-01
2.81019211e-01 -2.89942861e-01 -2.06501767e-01 1.14845619e-01
4.69998419e-01 3.01049888e-01 4.80506808e-01 1.21993518e+00
-3.48904490e-01 -1.08370912e+00 4.21852320e-02 3.79548758e-01
-4.43170935e-01 1.26047611e-01 -8.79600719e-02 7.06039786e-01
-1.11923620e-01 6.93550527e-01 8.62301141e-02 -3.05208087e-01
6.65257648e-02 4.02193656e-03 7.72213399e-01 -8.08375895e-01
-3.37949932e-01 -5.98864295e-02 -1.72920316e-01 -1.08245039e+00
-9.87779081e-01 -7.19233811e-01 -9.53588367e-01 2.81304210e-01
-7.61987492e-02 -5.99213876e-02 3.92192245e-01 8.39189470e-01
3.48804981e-01 1.12513267e-01 8.12359631e-01 -1.35584462e+00
-3.45541120e-01 -5.69589257e-01 -6.12418354e-01 1.01844347e+00
6.02424383e-01 -9.40545678e-01 -3.88607234e-01 5.36941469e-01] | [8.760191917419434, 0.7470486164093018] |
ec40caa9-8b9d-4f71-8f61-ec18df58f3a2 | hyperbolic-entailment-cones-for-learning | 1804.01882 | null | http://arxiv.org/abs/1804.01882v3 | http://arxiv.org/pdf/1804.01882v3.pdf | Hyperbolic Entailment Cones for Learning Hierarchical Embeddings | Learning graph representations via low-dimensional embeddings that preserve
relevant network properties is an important class of problems in machine
learning. We here present a novel method to embed directed acyclic graphs.
Following prior work, we first advocate for using hyperbolic spaces which
provably model tree-like structures better than Euclidean geometry. Second, we
view hierarchical relations as partial orders defined using a family of nested
geodesically convex cones. We prove that these entailment cones admit an
optimal shape with a closed form expression both in the Euclidean and
hyperbolic spaces, and they canonically define the embedding learning process.
Experiments show significant improvements of our method over strong recent
baselines both in terms of representational capacity and generalization. | ['Gary Bécigneul', 'Octavian-Eugen Ganea', 'Thomas Hofmann'] | 2018-04-03 | hyperbolic-entailment-cones-for-learning-1 | https://icml.cc/Conferences/2018/Schedule?showEvent=2487 | http://proceedings.mlr.press/v80/ganea18a/ganea18a.pdf | icml-2018-7 | ['hypernym-discovery'] | ['natural-language-processing'] | [-1.22805655e-01 6.88229859e-01 -3.67340714e-01 -3.08935165e-01
-9.36785191e-02 -1.04654729e+00 8.37471366e-01 3.24519515e-01
-2.44328186e-01 2.05744997e-01 7.58773327e-01 -5.42010069e-01
-6.47730947e-01 -1.14455068e+00 -5.29949188e-01 -5.16991496e-01
-6.00028694e-01 6.60137892e-01 1.67590320e-01 -2.72607744e-01
1.21762216e-01 8.42784822e-01 -7.24796832e-01 6.82819784e-02
6.34717643e-01 4.59698945e-01 -6.88479483e-01 8.63419890e-01
4.10480559e-01 7.40391970e-01 1.18727081e-01 -8.50624859e-01
4.52006102e-01 -1.71011701e-01 -1.18853867e+00 -4.37646061e-02
5.93576133e-01 -2.69945502e-01 -1.30287886e+00 8.95258904e-01
-1.57560371e-02 3.24368268e-01 9.78141725e-01 -1.47617579e+00
-1.40697086e+00 4.32197511e-01 -2.71825492e-01 3.65950465e-01
2.32019708e-01 -5.00712872e-01 1.91332090e+00 -8.18354785e-01
8.52002442e-01 1.31323767e+00 6.41427577e-01 4.68090206e-01
-1.82935083e+00 -1.72848031e-01 -3.77375670e-02 3.48404944e-01
-1.37226081e+00 -5.74528426e-02 6.58195794e-01 -3.93378377e-01
8.06861699e-01 3.70294154e-01 6.77631795e-01 9.02816176e-01
3.82058889e-01 3.42017025e-01 7.03677177e-01 -3.94826531e-01
1.60872042e-01 -2.27122426e-01 6.01538956e-01 1.12029791e+00
6.31322682e-01 -1.84205905e-01 -1.57417357e-01 -1.83925301e-01
8.51181567e-01 3.42799388e-02 -2.49688700e-01 -1.10249150e+00
-1.00334775e+00 1.43508959e+00 9.48016644e-01 3.71579558e-01
1.07158497e-01 4.51107025e-01 2.35229179e-01 3.02983463e-01
4.02159184e-01 5.84528267e-01 1.26548097e-01 3.83482128e-01
-1.42736495e-01 -4.55245562e-02 9.96957064e-01 1.08081353e+00
5.54195762e-01 -3.06406558e-01 2.88509518e-01 6.17911339e-01
3.24989945e-01 -6.32983297e-02 5.55144437e-03 -1.13179564e+00
4.07225490e-01 8.04291189e-01 -3.89963239e-01 -1.59562898e+00
-5.87209821e-01 -3.17801476e-01 -8.45681489e-01 -2.27446660e-01
3.26216340e-01 2.90343791e-01 -1.84036583e-01 1.63931119e+00
6.03331067e-02 6.47093207e-02 9.61423293e-02 5.59134841e-01
4.21486199e-01 6.87369168e-01 -2.18253329e-01 1.44003585e-01
1.19607460e+00 -9.36694205e-01 -5.02485037e-01 8.83805901e-02
1.07546139e+00 -1.20994985e-01 8.73641968e-01 -1.13180690e-01
-9.59476113e-01 6.49470240e-02 -1.26444793e+00 -6.23085618e-01
-4.57589239e-01 -3.29232663e-01 9.30719554e-01 7.50456631e-01
-1.46967828e+00 7.28858113e-01 -7.32601702e-01 -5.19575894e-01
5.21201730e-01 3.62404585e-01 -8.73337150e-01 -2.31056228e-01
-9.15239573e-01 7.08856821e-01 2.43805885e-01 5.90081885e-02
-5.74612379e-01 -5.64520717e-01 -1.09926832e+00 1.27139315e-01
1.62640959e-01 -5.67517936e-01 6.78130865e-01 -2.04414073e-02
-8.65743279e-01 9.85611498e-01 1.50397509e-01 -5.41111231e-01
1.40545815e-01 -7.25567639e-02 -3.13653529e-01 5.94387949e-01
-2.13040322e-01 2.72388756e-01 3.19268614e-01 -1.00700963e+00
-1.43870533e-01 -7.05890954e-01 8.64215076e-01 3.04811269e-01
-9.35222566e-01 -1.55529767e-01 -2.40950003e-01 -4.50363904e-01
4.54490483e-01 -1.37308836e+00 -3.90845478e-01 4.49950665e-01
-3.92258584e-01 -6.40800893e-01 7.28718638e-01 -3.91090840e-01
1.29491854e+00 -1.89427912e+00 4.18864429e-01 5.41902840e-01
8.57974708e-01 -2.07184315e-01 -1.59513161e-01 8.19142222e-01
-2.04908475e-01 6.48915410e-01 -2.58538455e-01 -5.86252846e-03
2.53116012e-01 5.17459154e-01 -2.40248471e-01 1.08253825e+00
1.39171839e-01 1.06135881e+00 -1.05556178e+00 -4.84996378e-01
-1.90495774e-02 4.34587926e-01 -9.66705561e-01 5.51469214e-02
1.26996577e-01 -1.17535472e-01 -3.82190794e-01 9.83676016e-02
4.93334353e-01 -3.16602230e-01 6.11488104e-01 -1.64519727e-01
4.39940959e-01 4.54168081e-01 -1.03154957e+00 1.71793592e+00
-4.86469626e-01 8.65812063e-01 -1.17480122e-01 -1.19891953e+00
7.57365167e-01 8.02716017e-02 3.95014793e-01 -2.22001478e-01
-3.57017666e-02 -7.28080496e-02 1.56859994e-01 -2.26531237e-01
4.37458664e-01 7.53197074e-02 -1.53250592e-02 6.30040169e-01
1.06651559e-01 5.67992218e-02 1.92397743e-01 9.22404170e-01
1.37253857e+00 -2.18780205e-01 2.15841427e-01 -8.10861766e-01
3.23236525e-01 -3.90242457e-01 3.90989423e-01 2.37094790e-01
-1.92924589e-01 5.06687999e-01 1.26363337e+00 -5.02964914e-01
-1.22379279e+00 -1.50383878e+00 -4.44454253e-01 1.15494871e+00
1.86404899e-01 -1.01716244e+00 -5.44108391e-01 -8.42732906e-01
-1.10617541e-01 5.62006354e-01 -1.04690337e+00 -5.22277296e-01
-7.23912776e-01 -4.69236374e-01 6.08028531e-01 8.41729283e-01
1.25857145e-01 -4.73138750e-01 -1.07657254e-01 -1.75639406e-01
2.42253438e-01 -1.22297609e+00 -8.26103628e-01 6.67622164e-02
-1.05902386e+00 -1.46616578e+00 -2.13237822e-01 -1.10767007e+00
8.22883427e-01 3.91268581e-01 1.05785465e+00 2.22329080e-01
-2.73589373e-01 7.00238585e-01 -2.70273149e-01 3.83418083e-01
-2.40631878e-01 3.21613163e-01 4.92352061e-02 7.89167136e-02
2.79099524e-01 -8.74105394e-01 -6.18917227e-01 1.39042765e-01
-1.01344550e+00 -2.65767545e-01 8.20984840e-02 5.74066162e-01
2.43879497e-01 -4.58890386e-02 1.45537272e-01 -1.09066927e+00
8.69890630e-01 -6.06445849e-01 -5.04213095e-01 3.00575674e-01
-8.04942191e-01 5.16844034e-01 6.92885578e-01 2.25434572e-01
-4.99588460e-01 -2.98092395e-01 2.78819919e-01 -2.25963309e-01
2.97299325e-01 4.36692983e-01 -7.84231722e-02 -2.07837403e-01
9.04768765e-01 -1.83777019e-01 -1.04943156e-01 -1.85235962e-01
9.57258701e-01 3.34730566e-01 3.43573689e-01 -8.41338992e-01
1.06202447e+00 7.13177979e-01 6.23331547e-01 -9.38571990e-01
-7.49031007e-01 -4.42657411e-01 -1.17665899e+00 2.40603894e-01
9.65670407e-01 -4.68858480e-01 -7.73460031e-01 -5.62928200e-01
-1.24529362e+00 1.20571695e-01 -2.70326585e-01 4.78256553e-01
-5.47667563e-01 6.13887429e-01 -9.58568931e-01 -4.64826256e-01
-6.26522899e-02 -6.51177824e-01 7.72919834e-01 -2.37417743e-01
-2.60266960e-01 -1.65853417e+00 4.38535631e-01 1.53651163e-01
8.02749768e-02 3.65870684e-01 1.55052817e+00 -7.78245330e-01
-6.19957149e-01 -1.69785887e-01 -4.47513789e-01 3.15628618e-01
-1.42328218e-01 1.73850268e-01 -6.23313427e-01 -3.21877301e-01
-3.65798801e-01 -1.89487204e-01 7.97823012e-01 -1.23915732e-01
1.14888930e+00 -6.48826897e-01 -3.84341121e-01 8.97663176e-01
1.62617457e+00 -2.51884520e-01 5.67710161e-01 9.39934328e-02
1.01282871e+00 8.49948764e-01 -2.96082675e-01 -2.34465320e-02
4.36389327e-01 4.21683997e-01 5.18901587e-01 3.63522202e-01
5.39010726e-02 -5.74150503e-01 2.07922161e-01 1.14640725e+00
-2.37742066e-01 1.02633685e-02 -8.84148002e-01 6.13360167e-01
-1.87872183e+00 -1.06029844e+00 -3.67853433e-01 2.06646252e+00
5.46558619e-01 3.17237303e-02 2.27497920e-01 1.65478081e-01
5.51204622e-01 4.55948770e-01 -1.50171399e-01 -8.96555543e-01
-1.81650788e-01 4.67240214e-01 6.57369852e-01 7.85316944e-01
-9.79476690e-01 7.02273190e-01 7.02334547e+00 1.91521645e-01
-3.77224445e-01 6.75697327e-02 2.26621196e-01 1.29103228e-01
-8.06043029e-01 1.79267287e-01 -4.22728658e-01 -1.52456433e-01
1.07572663e+00 -4.68846560e-01 5.23533046e-01 7.44585693e-01
-3.60290885e-01 7.35382617e-01 -1.66302383e+00 7.58113682e-01
1.20161779e-01 -1.61274552e+00 5.07812947e-02 5.12059748e-01
9.15201128e-01 -1.02836601e-01 1.70197323e-01 -2.14568973e-02
7.93178141e-01 -1.37464988e+00 3.31260085e-01 2.60895714e-02
6.18995607e-01 -7.77918518e-01 2.33700886e-01 -6.45402893e-02
-1.56718278e+00 -1.10090591e-01 -7.66047120e-01 -4.53549549e-02
-2.60425866e-01 2.56018102e-01 -7.83218861e-01 6.17920578e-01
2.30912209e-01 1.02772737e+00 -7.37174153e-01 8.07459950e-01
-5.19717336e-01 6.22903109e-01 -3.00981700e-01 -1.89498551e-02
3.90488893e-01 -8.21740568e-01 3.80989641e-01 1.18347013e+00
-1.11198813e-01 3.15188348e-01 8.56076255e-02 9.31410849e-01
-6.82238162e-01 1.19185455e-01 -1.36840045e+00 -2.68424064e-01
4.97582555e-01 1.36814630e+00 -8.71241212e-01 2.76838481e-01
-4.77000654e-01 9.02733564e-01 7.87057161e-01 4.62382644e-01
-7.02756345e-01 -4.33445454e-01 7.69824564e-01 2.42106974e-01
1.28359020e-01 -8.41874838e-01 -2.24603668e-01 -1.34126341e+00
1.16211228e-01 -2.85186291e-01 7.31265247e-01 -3.25502992e-01
-1.37748325e+00 6.05306804e-01 2.69579375e-03 -6.38539135e-01
-2.66901031e-02 -1.08339250e+00 -6.45610154e-01 4.04490173e-01
-9.72335696e-01 -1.11643863e+00 7.03452751e-02 7.32719481e-01
1.84892807e-02 2.31886581e-01 1.17021894e+00 -1.90243928e-03
-5.51295400e-01 5.96886098e-01 4.88302529e-01 4.55913335e-01
-7.47482898e-03 -1.73617768e+00 4.50604975e-01 7.33491302e-01
9.63828623e-01 8.25842083e-01 2.70036280e-01 -2.01550588e-01
-1.74917376e+00 -1.15097845e+00 8.21502566e-01 -7.38873899e-01
1.15408123e+00 -7.88100660e-01 -7.43062615e-01 1.36112881e+00
3.62167299e-01 4.93138224e-01 9.45805788e-01 6.22202098e-01
-1.10895467e+00 -9.13750660e-03 -8.89533520e-01 8.65895271e-01
1.67232537e+00 -1.04812443e+00 -7.12695360e-01 7.15351105e-01
9.06445682e-01 1.92874104e-01 -1.20598900e+00 1.19055592e-01
4.67942506e-01 -7.69081056e-01 9.51401114e-01 -1.24892843e+00
4.48765993e-01 5.22490144e-02 -5.11295378e-01 -1.33981991e+00
-7.67106831e-01 -8.83963048e-01 -2.87716061e-01 6.96095467e-01
3.89083356e-01 -7.75617599e-01 9.47519779e-01 4.63300824e-01
3.82893011e-02 -9.65251267e-01 -9.15728688e-01 -1.14160168e+00
7.06402183e-01 -1.09241307e-01 1.84277311e-01 9.33295786e-01
5.39872110e-01 8.01071465e-01 1.52307212e-01 2.51571000e-01
7.75487542e-01 4.10126615e-03 3.88408750e-01 -1.61898124e+00
6.16983091e-03 -4.73973453e-01 -9.95243728e-01 -1.02419102e+00
5.51195800e-01 -1.71222186e+00 -4.98742312e-01 -1.89275205e+00
2.51331121e-01 -3.24727297e-01 -2.02958271e-01 7.22428486e-02
4.57963705e-01 1.59252793e-01 8.66442397e-02 1.70412108e-01
-7.23545372e-01 7.45154977e-01 9.42144573e-01 1.30655849e-02
2.15170726e-01 -3.43432635e-01 -8.76583934e-01 6.39209628e-01
6.95861816e-01 -4.08881545e-01 -7.62252510e-01 -6.87754214e-01
5.84439099e-01 -3.25000793e-01 3.06293637e-01 -6.03115082e-01
2.64054060e-01 -2.77027451e-02 -3.77034172e-02 8.92107934e-02
2.20934927e-01 -8.62261891e-01 -2.66398907e-01 3.87548417e-01
-8.66524458e-01 3.95718396e-01 -3.37552786e-01 1.04642081e+00
2.02433914e-01 -2.76121259e-01 8.23507190e-01 3.01072180e-01
-3.22522730e-01 7.61532068e-01 -2.87037287e-02 4.85801190e-01
1.24792397e+00 -6.06691539e-02 -4.60314900e-01 -5.07939994e-01
-7.32889473e-01 1.05016656e-01 4.89208847e-01 3.73623878e-01
7.59015560e-01 -1.69466913e+00 -6.83913529e-01 -1.60172626e-01
2.94301510e-01 -1.44836932e-01 -3.89120400e-01 5.20543694e-01
-8.50738168e-01 5.78328848e-01 -1.21851787e-01 -1.95044741e-01
-9.85591888e-01 7.17770219e-01 2.92368799e-01 -2.81623662e-01
-8.99008155e-01 8.10630500e-01 5.11745274e-01 -6.01564825e-01
1.17552623e-01 -2.79174149e-01 -7.65229836e-02 -4.78683375e-02
2.01630339e-01 6.52426064e-01 -8.96000490e-02 -7.82154739e-01
-3.87872487e-01 4.32153255e-01 -3.33373286e-02 -1.47540286e-01
1.50714386e+00 1.01860188e-01 -5.71117759e-01 2.32799768e-01
2.00115252e+00 5.05794361e-02 -9.17863607e-01 -2.82633036e-01
4.15937185e-01 -5.42168081e-01 -1.85130864e-01 1.31836504e-01
-9.61976886e-01 1.01521730e+00 1.45749059e-02 8.25270712e-01
4.45820242e-01 4.51126605e-01 3.67072076e-01 7.01673567e-01
1.00186065e-01 -8.32093894e-01 4.02157694e-01 5.74393153e-01
1.12579298e+00 -6.62159801e-01 2.16142256e-02 -8.31307590e-01
-1.54520243e-01 1.46776164e+00 3.19041789e-01 -8.07681501e-01
9.88104582e-01 -1.75664440e-01 -7.17821538e-01 -4.69319016e-01
-7.97343433e-01 -8.20769146e-02 4.96517420e-01 9.44244623e-01
5.03838599e-01 2.27630734e-01 -3.52763981e-01 3.39162618e-01
-4.61512834e-01 -7.86476314e-01 8.58330727e-01 6.26014829e-01
-3.64267170e-01 -9.64213550e-01 1.14123203e-01 3.79140198e-01
-1.92005053e-01 -1.52293727e-01 -6.20220780e-01 8.24845791e-01
-3.97012323e-01 9.65104580e-01 2.22496361e-01 -3.47402394e-01
3.08923423e-01 -2.10994691e-01 7.90354013e-01 -9.02853847e-01
1.64124966e-01 -6.40233636e-01 1.65540591e-01 -5.64119160e-01
4.35939506e-02 -4.29745913e-01 -1.31443179e+00 -7.13349700e-01
-1.71676889e-01 1.43531099e-01 3.31446767e-01 5.32505572e-01
2.58508891e-01 2.10853130e-01 9.46066558e-01 -2.04227358e-01
-8.76210928e-01 -5.09013951e-01 -8.35726023e-01 4.59155351e-01
1.16507225e-01 -4.80926394e-01 -5.29098749e-01 -1.29565939e-01] | [7.139491558074951, 6.040533065795898] |
9621f48e-d1ae-4094-84bb-48a52288e594 | modeling-scale-free-graphs-for-knowledge | 2108.06468 | null | https://arxiv.org/abs/2108.06468v3 | https://arxiv.org/pdf/2108.06468v3.pdf | Modeling Scale-free Graphs with Hyperbolic Geometry for Knowledge-aware Recommendation | Aiming to alleviate data sparsity and cold-start problems of traditional recommender systems, incorporating knowledge graphs (KGs) to supplement auxiliary information has recently gained considerable attention. Via unifying the KG with user-item interactions into a tripartite graph, recent works explore the graph topologies to learn the low-dimensional representations of users and items with rich semantics. However, these real-world tripartite graphs are usually scale-free, the intrinsic hierarchical graph structures of which are underemphasized in existing works, consequently, leading to suboptimal recommendation performance. To address this issue and provide more accurate recommendation, we propose a knowledge-aware recommendation method with the hyperbolic geometry, namely Lorentzian Knowledge-enhanced Graph convolutional networks for Recommendation (LKGR). LKGR facilitates better modeling of scale-free tripartite graphs after the data unification. Specifically, we employ different information propagation strategies in the hyperbolic space to explicitly encode heterogeneous information from historical interactions and KGs. Our proposed knowledge-aware attention mechanism enables the model to automatically measure the information contribution, producing the coherent information aggregation in the hyperbolic space. Extensive experiments on three real-world benchmarks demonstrate that LKGR outperforms state-of-the-art methods by 3.6-15.3% of Recall@20 on Top-K recommendation. | ['Jianye Hao', 'Irwin King', 'Ziqiao Meng', 'Mengchen Zhao', 'Yingxue Zhang', 'Menglin Yang', 'Yankai Chen'] | 2021-08-14 | null | null | null | null | ['knowledge-aware-recommendation'] | ['miscellaneous'] | [-6.32448018e-01 6.75325319e-02 -4.54342186e-01 -1.60292789e-01
-8.54454637e-02 -6.20110452e-01 3.31241131e-01 2.10616469e-01
1.69570655e-01 3.75374317e-01 6.66594982e-01 -2.94762313e-01
-7.57245719e-01 -1.12726367e+00 -6.47675991e-01 -7.63745666e-01
-2.45228186e-01 4.07522619e-01 1.57612950e-01 -5.58034897e-01
1.65627033e-01 2.57933110e-01 -1.07008266e+00 1.54420540e-01
1.13930213e+00 9.28308070e-01 1.65041666e-02 2.73360103e-01
-2.09666222e-01 5.53377748e-01 -4.03415179e-03 -8.63692045e-01
3.02088499e-01 -2.50468198e-02 -7.53678918e-01 -8.31182897e-02
2.66254812e-01 -1.19853653e-01 -9.64586616e-01 9.65200901e-01
3.97212893e-01 5.19125462e-01 5.64953506e-01 -1.01400530e+00
-1.63436222e+00 1.27422762e+00 -5.06351411e-01 2.16730893e-01
1.60445690e-01 -1.03035688e-01 1.47049153e+00 -8.01627874e-01
3.72193426e-01 1.03311789e+00 7.10384071e-01 -6.80636764e-02
-9.99159157e-01 -5.55194497e-01 6.40496910e-01 3.66246641e-01
-1.56269884e+00 2.39035845e-01 9.10123289e-01 -1.39640674e-01
7.72634327e-01 1.48133263e-01 9.15264964e-01 1.00959742e+00
2.96657793e-02 6.94753528e-01 5.46927273e-01 3.38591993e-01
-5.01809567e-02 -7.14599863e-02 6.35385573e-01 7.88644254e-01
7.16151476e-01 -8.69372413e-02 -5.86551249e-01 -8.62372667e-02
1.07879055e+00 3.57283145e-01 -3.86224508e-01 -5.55244923e-01
-1.09258449e+00 8.02360892e-01 1.16318679e+00 1.38220772e-01
-4.19870496e-01 -6.13467656e-02 3.71980608e-01 2.58425921e-01
5.98727286e-01 5.73729634e-01 -3.21359694e-01 4.25917804e-01
-3.45066935e-01 1.05783686e-01 7.17131376e-01 1.24933910e+00
7.47194052e-01 -5.44547569e-03 -2.23483428e-01 6.33216858e-01
5.12318492e-01 4.14841771e-01 1.33216709e-01 -3.55051339e-01
5.34227908e-01 9.97616589e-01 3.47326584e-02 -1.56660235e+00
-7.51515090e-01 -1.32835519e+00 -1.23714721e+00 -8.95504832e-01
4.16750908e-01 8.47198218e-02 -5.80914438e-01 1.40360808e+00
4.55414534e-01 5.69651902e-01 -2.37180274e-02 1.29993737e+00
1.22233582e+00 6.50975525e-01 -1.42564133e-01 -1.61755562e-01
1.19269276e+00 -1.20248663e+00 -5.23555458e-01 3.08676332e-01
8.87006581e-01 -3.25935453e-01 1.11542606e+00 4.07928526e-01
-8.04252505e-01 -5.18200338e-01 -9.63541627e-01 -1.04516454e-01
-4.98904318e-01 -1.08945452e-01 1.13911283e+00 4.30808365e-01
-9.54958200e-01 5.87666273e-01 -4.22678351e-01 -2.01871544e-01
3.10206115e-01 3.42446923e-01 -1.63412899e-01 -3.00082415e-01
-1.56974101e+00 3.63494992e-01 2.84827620e-01 2.52754480e-01
-2.36612901e-01 -9.82687652e-01 -6.33776009e-01 2.38394082e-01
5.91753185e-01 -8.39494705e-01 6.60496414e-01 -4.53839242e-01
-1.31102931e+00 5.86558953e-02 4.88456011e-01 -2.78885543e-01
-1.33386478e-01 -1.24300085e-01 -7.44746149e-01 -7.42804632e-02
-2.71337658e-01 -1.31118745e-01 5.89253247e-01 -1.22396624e+00
-4.00394022e-01 -5.28287649e-01 6.72384739e-01 5.44555783e-01
-6.51280701e-01 -5.43623626e-01 -7.95010865e-01 -9.17727888e-01
2.10386470e-01 -9.64121580e-01 -1.82270691e-01 -6.89899862e-01
-2.99702138e-01 -3.28217655e-01 4.51034129e-01 -4.61076528e-01
1.66241753e+00 -1.78800964e+00 4.50773090e-01 3.76919359e-01
6.20712399e-01 2.05830127e-01 -3.32433015e-01 4.49301004e-01
4.37145233e-01 -1.10649124e-01 3.93067181e-01 -7.68717080e-02
7.91992098e-02 3.76674235e-01 -2.97336400e-01 4.35818583e-01
-4.48869675e-01 1.26181185e+00 -1.29553974e+00 -1.29966810e-01
9.46418867e-02 6.75837398e-01 -7.78436661e-01 2.17831880e-02
-2.49249190e-01 3.92891079e-01 -7.41538405e-01 4.10072356e-01
9.08486009e-01 -8.29448879e-01 3.91292274e-01 -5.99886894e-01
1.85084686e-01 4.09001946e-01 -1.16584480e+00 1.90004444e+00
-3.97042811e-01 -1.87786415e-01 -1.43339649e-01 -1.01808059e+00
8.76844466e-01 -1.67225212e-01 4.64657605e-01 -8.55269969e-01
-2.83200592e-02 -1.67453796e-01 -1.43800512e-01 -1.63118228e-01
8.87216389e-01 3.38978410e-01 -6.58448637e-02 4.25322145e-01
2.53438652e-01 4.61220056e-01 -1.16442114e-01 7.50443637e-01
9.94297802e-01 -1.71684716e-02 -9.04931650e-02 -5.08168340e-01
3.98506910e-01 -5.26046336e-01 4.80137438e-01 7.45216072e-01
3.78302306e-01 4.12633955e-01 1.76024541e-01 -5.87155521e-01
-5.20626485e-01 -1.11668277e+00 1.19341798e-01 1.37225509e+00
6.68719947e-01 -8.91607285e-01 -2.49826118e-01 -1.01296294e+00
1.58355325e-01 4.68342721e-01 -6.31297052e-01 -4.56093282e-01
-4.02406842e-01 -8.71633470e-01 1.47322521e-01 5.20821095e-01
3.05136770e-01 -6.44616723e-01 5.07017851e-01 2.64121085e-01
-1.93186015e-01 -8.53464127e-01 -9.60391104e-01 -1.59936413e-01
-8.26256931e-01 -1.08711672e+00 -5.74311972e-01 -5.91797352e-01
6.59810305e-01 9.21415329e-01 1.34112120e+00 3.11435193e-01
2.67996639e-01 5.58561981e-01 -8.86484027e-01 2.49806672e-01
3.57685268e-01 4.36415851e-01 2.51815647e-01 2.56231040e-01
3.35581779e-01 -8.60100567e-01 -1.15945113e+00 6.11072183e-01
-7.99856424e-01 2.90123433e-01 5.34811914e-01 6.36158943e-01
5.78118384e-01 8.48376453e-02 8.30361724e-01 -1.07013297e+00
6.37708008e-01 -1.01550722e+00 -3.81807625e-01 3.92797589e-01
-9.41792011e-01 2.01479197e-01 1.06505144e+00 -4.49486256e-01
-1.02628553e+00 -5.36717951e-01 1.94867048e-02 -3.79231453e-01
2.47563958e-01 9.82454181e-01 -9.63960141e-02 -2.40754530e-01
5.70152700e-01 1.13770746e-01 -5.82089782e-01 -6.57465279e-01
1.03297174e+00 3.95308167e-01 3.12079936e-01 -6.01975083e-01
6.86042786e-01 3.60896945e-01 -2.94512715e-02 -5.48967540e-01
-1.31769133e+00 -6.86965406e-01 -5.50922453e-01 -7.03515485e-02
3.30175102e-01 -9.06159699e-01 -7.28963494e-01 1.91435039e-01
-6.79503620e-01 -7.26402029e-02 -1.85549244e-01 7.09685504e-01
-1.41605869e-01 6.04855835e-01 -7.33476818e-01 -4.29684490e-01
-7.14106619e-01 -7.19716728e-01 7.91514277e-01 2.98027217e-01
4.99253094e-01 -1.25923073e+00 1.51895449e-01 5.04130125e-01
4.87267911e-01 -2.89016396e-01 9.31319654e-01 -7.56273985e-01
-8.15639436e-01 7.37235472e-02 -6.03139997e-01 -1.10676743e-01
7.94273391e-02 -4.73035425e-01 -5.03331304e-01 -3.97316039e-01
-4.86530423e-01 -1.50012737e-02 1.00915754e+00 2.59315938e-01
1.12655783e+00 -6.25263155e-01 -2.87222177e-01 9.28725779e-01
1.18940401e+00 -3.33908588e-01 4.49181199e-01 -1.68826744e-01
1.47651148e+00 2.74003416e-01 4.62944537e-01 6.04283869e-01
1.13150692e+00 6.95505798e-01 5.29113173e-01 1.73289537e-01
-1.45438015e-01 -5.32810152e-01 1.02878854e-01 1.69289935e+00
-5.30484676e-01 -4.69149560e-01 -4.51734692e-01 4.03027266e-01
-2.26761389e+00 -8.14057887e-01 -3.79342109e-01 2.21173763e+00
4.68664169e-01 -5.50037287e-02 3.41274351e-01 -4.29212600e-01
6.64122939e-01 2.76679486e-01 -6.75347626e-01 9.92475450e-02
-1.98222667e-01 -1.81111693e-01 6.65795863e-01 3.25990945e-01
-9.70471680e-01 8.31856608e-01 4.80208349e+00 9.13696527e-01
-7.80571342e-01 -5.90764917e-02 1.86574027e-01 -6.43054917e-02
-8.14837158e-01 -1.14943326e-01 -7.72261918e-01 4.46394980e-01
8.07391226e-01 -2.15774968e-01 8.34806919e-01 6.21151745e-01
-1.50337353e-01 5.27419269e-01 -7.04042017e-01 1.05356300e+00
1.56357288e-01 -1.54796660e+00 4.28383231e-01 1.76485136e-01
1.03908610e+00 2.54182190e-01 2.26257816e-01 7.21116841e-01
6.31070852e-01 -6.47003770e-01 1.73058793e-01 9.25269008e-01
4.19928581e-01 -9.78278697e-01 7.39591062e-01 1.89796716e-01
-1.76715302e+00 -7.78589174e-02 -7.50385582e-01 -3.15859355e-02
6.89680949e-02 7.27387786e-01 -4.05304641e-01 1.40894127e+00
6.22167528e-01 1.29272127e+00 -5.34080505e-01 1.11805844e+00
-3.92340943e-02 7.58946836e-01 -2.41341203e-01 -1.24104120e-01
2.72285014e-01 -8.26185763e-01 5.00519872e-01 1.08679366e+00
4.64486659e-01 4.20705676e-01 3.63910079e-01 6.30618691e-01
-5.06078422e-01 5.36493838e-01 -4.66942817e-01 -2.20758975e-01
2.25708008e-01 1.53081727e+00 -8.20436656e-01 -1.96345374e-02
-8.21422100e-01 8.01222384e-01 7.90772140e-01 5.41072488e-01
-7.75263429e-01 -1.48041576e-01 5.23349464e-01 3.47374707e-01
5.21481931e-01 -2.64643252e-01 2.88522001e-02 -1.47398818e+00
-2.43149221e-01 -4.69243377e-01 7.87280262e-01 -5.59757411e-01
-1.91163850e+00 3.87178272e-01 -1.71853110e-01 -1.00938094e+00
4.26831812e-01 -3.68722171e-01 -4.90965605e-01 5.54342687e-01
-1.35949075e+00 -1.46292305e+00 -4.14767951e-01 8.05062413e-01
5.61326481e-02 -1.43982902e-01 6.84084177e-01 5.48371196e-01
-4.99369383e-01 8.41419816e-01 4.83040482e-01 -6.64259717e-02
5.07796288e-01 -1.38382423e+00 4.95841116e-01 4.11558807e-01
4.85234022e-01 9.75083470e-01 2.18318671e-01 -7.90280938e-01
-2.17603064e+00 -1.40959239e+00 2.20812172e-01 -4.71439391e-01
9.55522001e-01 -3.48598510e-01 -1.13450825e+00 6.32444024e-01
-1.27616763e-01 2.40897939e-01 8.46513152e-01 8.40908289e-01
-8.82227361e-01 -3.44796509e-01 -6.84118092e-01 5.98941982e-01
1.67814648e+00 -4.46264565e-01 -3.51340681e-01 4.29477036e-01
1.14810514e+00 -1.89673439e-01 -1.28870249e+00 3.34807009e-01
5.13757229e-01 -6.92848146e-01 1.19814813e+00 -7.24554479e-01
-1.56270564e-02 -3.95623058e-01 5.33245541e-02 -1.57085741e+00
-1.05388951e+00 -6.64491653e-01 -7.85167575e-01 9.82462764e-01
2.13770360e-01 -5.07748961e-01 6.78942442e-01 2.04640076e-01
-3.55886459e-01 -9.08498824e-01 -3.87761563e-01 -5.95287979e-01
-2.25002710e-02 -2.94172883e-01 8.79824340e-01 1.21324396e+00
3.75402629e-01 7.59243846e-01 -7.21567035e-01 3.40891868e-01
6.61026061e-01 6.19636476e-01 5.89343369e-01 -1.54885018e+00
-4.45020169e-01 -3.85347724e-01 -3.86524677e-01 -1.41883254e+00
-1.67771038e-02 -1.47147155e+00 -5.70762038e-01 -1.92516267e+00
2.65823871e-01 -7.79998064e-01 -7.91312099e-01 1.43526137e-01
-1.89088240e-01 2.69016862e-01 -9.91195589e-02 1.81144297e-01
-1.28634155e+00 7.60788560e-01 1.64541185e+00 -5.69522642e-02
-2.46101677e-01 1.75107755e-02 -1.22658134e+00 4.54243749e-01
6.35024846e-01 -9.94258001e-02 -8.95399153e-01 -4.91091669e-01
9.97750998e-01 -4.45295945e-02 4.92138006e-02 -6.16365910e-01
4.85808969e-01 -1.42576262e-01 2.57601440e-02 -6.23688459e-01
6.19465150e-02 -7.19152272e-01 3.29546034e-01 -2.54942030e-01
-2.09127113e-01 -2.75584161e-01 -1.53616384e-01 1.23209953e+00
2.05841064e-01 3.03236991e-01 3.81253362e-01 8.37652385e-02
-4.99402255e-01 9.68547285e-01 2.66892999e-01 2.01184712e-02
5.74330866e-01 2.55036414e-01 -6.81232810e-01 -3.80285203e-01
-6.85251236e-01 4.09755707e-01 3.06580901e-01 5.77131689e-01
5.71712673e-01 -1.65183711e+00 -5.72890341e-01 -1.90942250e-02
3.71409535e-01 -9.91551951e-02 7.94226468e-01 9.59001601e-01
-5.71573414e-02 4.77872878e-01 8.82169902e-02 -1.96320727e-01
-6.01782739e-01 1.07737017e+00 7.97877386e-02 -5.40043056e-01
-1.02625310e+00 7.00845361e-01 4.66668665e-01 -5.20679593e-01
2.54633784e-01 -3.27816606e-01 -5.44682384e-01 1.03786506e-01
4.74523544e-01 4.61655915e-01 2.66668033e-02 -7.07237124e-01
-5.39491810e-02 4.72979754e-01 -3.95513445e-01 6.66195095e-01
1.27542222e+00 -4.64133292e-01 -3.56242061e-02 1.82305306e-01
9.94376540e-01 1.06100447e-01 -8.86623085e-01 -8.65542591e-01
-3.23695064e-01 -3.95568132e-01 2.77232111e-01 -6.19223654e-01
-1.41680884e+00 4.79568124e-01 1.45674571e-02 6.13389492e-01
8.41655374e-01 5.83230071e-02 9.93818462e-01 6.28156006e-01
6.99207187e-01 -9.01465237e-01 1.00292535e-02 6.26840234e-01
8.40482950e-01 -1.04598141e+00 1.12325065e-01 -4.80024248e-01
-6.36340618e-01 8.93608749e-01 6.80901468e-01 -3.56386393e-01
1.21449697e+00 -3.44223946e-01 -5.24215758e-01 -4.73708183e-01
-5.81836879e-01 -2.92965889e-01 9.42827880e-01 4.65429366e-01
2.29019254e-01 3.33160013e-01 -2.27863207e-01 1.38235700e+00
-4.95039858e-02 -2.52623200e-01 2.00918809e-01 1.87946230e-01
-2.81975538e-01 -7.29212940e-01 2.04648122e-01 8.84309292e-01
-1.98695228e-01 -3.02002132e-01 -2.33090997e-01 3.77461970e-01
-1.25325561e-01 9.09520328e-01 -1.49691254e-01 -8.72791886e-01
2.87258595e-01 -4.67132449e-01 4.04275209e-01 -4.93487030e-01
-6.18471861e-01 7.80720497e-03 -4.29765992e-02 -4.84138578e-01
-2.83318520e-01 -1.01946354e-01 -1.16768706e+00 -6.28633082e-01
-6.00766480e-01 5.52508712e-01 1.37034625e-01 8.22611809e-01
8.63563418e-01 5.75169444e-01 6.33720040e-01 -8.26358557e-01
-3.13271224e-01 -8.47303987e-01 -1.06979144e+00 5.26205420e-01
-3.10181361e-02 -9.36349750e-01 -1.88894242e-01 -5.82916379e-01] | [10.237386703491211, 5.638768672943115] |
b90fd7e6-1d7a-421a-b573-14127eb84160 | an-unsupervised-approach-for-aspect-category | 1812.03361 | null | https://arxiv.org/abs/1812.03361v2 | https://arxiv.org/pdf/1812.03361v2.pdf | An Unsupervised Approach for Aspect Category Detection Using Soft Cosine Similarity Measure | Aspect category detection is one of the important and challenging subtasks of aspect-based sentiment analysis. Given a set of pre-defined categories, this task aims to detect categories which are indicated implicitly or explicitly in a given review sentence. Supervised machine learning approaches perform well to accomplish this subtask. Note that, the performance of these methods depends on the availability of labeled train data, which is often difficult and costly to obtain. Besides, most of these supervised methods require feature engineering to perform well. In this paper, we propose an unsupervised method to address aspect category detection task without the need for any feature engineering. Our method utilizes clusters of unlabeled reviews and soft cosine similarity measure to accomplish aspect category detection task. Experimental results on SemEval-2014 restaurant dataset shows that proposed unsupervised approach outperforms several baselines by a substantial margin. | ['Heshaam Faili', 'Sajad Movahedi', 'Erfan Ghadery', 'Azadeh Shakery'] | 2018-12-08 | null | null | null | null | ['aspect-category-detection'] | ['natural-language-processing'] | [ 2.04908952e-01 -1.56763151e-01 -3.60194385e-01 -7.30701089e-01
-7.95515954e-01 -7.74583459e-01 7.64705539e-01 6.51900530e-01
-3.51608008e-01 3.69273365e-01 9.02414545e-02 -3.31701756e-01
2.52026111e-01 -6.26078427e-01 -2.94547915e-01 -4.49242592e-01
4.24095243e-01 3.02262336e-01 8.92201141e-02 -2.51605093e-01
8.55468452e-01 -1.71961188e-01 -1.53610301e+00 2.52140373e-01
7.07849324e-01 8.91560614e-01 -1.59655195e-02 4.33014244e-01
-4.24806386e-01 4.63352948e-01 -5.05808473e-01 -3.59556288e-01
1.07631110e-01 -4.66449082e-01 -6.11124754e-01 5.57340920e-01
1.84202164e-01 1.64696157e-01 7.68985212e-01 1.13666999e+00
5.34259304e-02 2.60566801e-01 1.12054539e+00 -1.01188087e+00
-5.03293395e-01 4.27213609e-01 -8.33910942e-01 4.42045704e-02
3.67474616e-01 -2.51891226e-01 1.46823227e+00 -1.19023383e+00
2.16181219e-01 7.51705050e-01 5.11717916e-01 9.15783197e-02
-6.68397844e-01 -4.74531710e-01 5.19208789e-01 -1.37211189e-01
-9.75415349e-01 -1.71876028e-01 8.97389650e-01 -4.31832403e-01
9.41415668e-01 5.32143116e-02 5.40134370e-01 2.78665185e-01
-8.86705820e-04 7.46049047e-01 1.32617366e+00 -5.53201079e-01
6.75616264e-01 7.30136216e-01 1.01117706e+00 3.75008881e-01
4.11918312e-01 -5.54151893e-01 -4.24504250e-01 -2.89921552e-01
8.07572752e-02 1.47498503e-01 7.28567392e-02 -3.16183269e-01
-8.75694394e-01 1.04146957e+00 -6.18412122e-02 3.43224674e-01
-2.71772087e-01 -3.00765246e-01 5.59352517e-01 4.33584481e-01
6.92869842e-01 5.79599500e-01 -7.97602236e-01 -1.93269327e-01
-7.16579080e-01 -1.43906787e-01 8.74619544e-01 1.08671629e+00
9.66892421e-01 -7.43583143e-02 1.61843330e-01 8.89574170e-01
4.71985310e-01 3.65690857e-01 5.18584490e-01 -2.79798985e-01
4.27829206e-01 1.12619472e+00 1.48334980e-01 -1.00062823e+00
-7.38100559e-02 -4.76504803e-01 -5.49144804e-01 -3.03878095e-02
1.23593077e-01 -5.86717352e-02 -1.03493404e+00 1.04770732e+00
5.61460555e-01 -3.00316364e-01 1.09578803e-01 8.16267848e-01
7.99004853e-01 5.58102071e-01 1.97987542e-01 -3.43677461e-01
1.67428386e+00 -1.02744532e+00 -7.11833179e-01 -4.06557500e-01
5.53405702e-01 -1.08486640e+00 1.34807992e+00 4.34736967e-01
-5.56755126e-01 -3.66466850e-01 -1.25591993e+00 3.46121311e-01
-6.00285411e-01 3.39407414e-01 9.48543131e-01 9.06714439e-01
-5.42296886e-01 8.84737894e-02 -6.83560610e-01 -5.16265869e-01
1.22989051e-01 4.42124426e-01 -3.08838516e-01 7.31170103e-02
-7.99236178e-01 5.13929904e-01 1.45482346e-01 8.12294185e-02
-5.78795552e-01 -2.37160876e-01 -1.06086254e+00 1.72633808e-02
3.36224049e-01 -2.92171180e-01 1.27376282e+00 -1.28520298e+00
-1.16932666e+00 7.12204874e-01 -4.14150387e-01 -2.52712131e-01
-1.72321707e-01 -4.07514930e-01 -3.23748142e-01 2.14710459e-01
2.22859412e-01 1.88183978e-01 1.00108397e+00 -1.20867205e+00
-8.63006711e-01 -4.69206899e-01 2.73690820e-01 6.03145778e-01
-6.37448668e-01 2.32996747e-01 -3.60422552e-01 -5.11115134e-01
1.87964737e-01 -8.79808545e-01 -3.96484375e-01 -4.38525826e-01
-3.88541669e-01 -4.74553078e-01 8.97733748e-01 -2.56946236e-01
1.08900762e+00 -1.93974149e+00 -4.38813776e-01 2.53373235e-01
1.67855799e-01 1.82654724e-01 1.91024229e-01 2.56194562e-01
7.80027434e-02 8.49230811e-02 -5.17305911e-01 -2.54103661e-01
-6.78372011e-02 -1.48814842e-01 -1.88917652e-01 3.69389325e-01
2.71787345e-01 4.86916095e-01 -8.29918563e-01 -5.09393752e-01
1.09277278e-01 2.26720273e-01 -5.35730720e-01 1.85080141e-01
-2.15267867e-01 3.08239013e-01 -5.70036352e-01 7.66960442e-01
5.69370806e-01 -1.87409297e-01 -1.29328500e-02 -2.13441193e-01
-2.81578060e-02 6.04320347e-01 -1.21401525e+00 1.37451506e+00
-5.15138984e-01 3.23948741e-01 -3.14363092e-01 -1.44343615e+00
1.06750643e+00 3.32059979e-01 3.55726540e-01 -1.66844502e-01
2.96938986e-01 1.12023607e-01 2.46123923e-03 -3.20196629e-01
7.91090548e-01 -5.63256085e-01 -3.55911493e-01 8.17053139e-01
1.84798725e-02 -2.69676715e-01 3.77591759e-01 1.55779317e-01
9.17530358e-01 -1.15588784e-01 7.67412543e-01 -3.03782791e-01
7.46826649e-01 3.83244008e-01 7.03878343e-01 5.12037456e-01
-3.90205979e-01 7.54206359e-01 3.65612924e-01 -3.09760094e-01
-7.25316644e-01 -6.60737634e-01 1.08419294e-02 1.03553152e+00
1.60093784e-01 -5.39401472e-01 -5.51098406e-01 -1.22916138e+00
-4.17350024e-01 6.15309954e-01 -7.70945728e-01 1.61288492e-02
-2.72160798e-01 -9.81810331e-01 -1.85869381e-01 3.70946646e-01
4.22293842e-01 -9.14663792e-01 -4.67086792e-01 2.18270406e-01
5.69576062e-02 -1.00807214e+00 -6.00639641e-01 3.96583050e-01
-8.50554943e-01 -1.25450933e+00 -3.56970727e-01 -1.12693191e+00
1.10809076e+00 7.63058484e-01 1.30245829e+00 3.78165469e-02
-6.60936534e-02 3.72952610e-01 -8.31688881e-01 -6.42811477e-01
-1.13530487e-01 2.48428598e-01 -1.78616343e-03 1.32051155e-01
1.25524604e+00 -2.83140987e-01 -7.65718937e-01 3.95554960e-01
-8.88749719e-01 -3.09737116e-01 7.51034498e-01 7.27136195e-01
6.94925487e-01 2.32930437e-01 7.65321076e-01 -1.35660362e+00
6.32091165e-01 -3.99010301e-01 -3.86325955e-01 1.92871094e-01
-9.42048788e-01 -1.28878400e-01 7.11158097e-01 -2.64297873e-01
-9.68534708e-01 2.03026712e-01 5.76244965e-02 3.82838756e-01
-3.11151117e-01 8.03788900e-01 -2.24084541e-01 3.45950842e-01
3.93956304e-01 2.73413301e-01 -1.94335088e-01 -3.51589829e-01
1.41019195e-01 1.03619254e+00 -1.11674830e-01 -2.53857672e-01
7.28206992e-01 4.22903359e-01 -3.50292653e-01 -7.40814507e-01
-1.28241849e+00 -1.21882176e+00 -7.67146468e-01 -1.82413980e-02
8.01304460e-01 -9.87926900e-01 -1.40606925e-01 1.38185784e-01
-7.56989717e-01 2.13500336e-01 -2.05757767e-02 4.93683100e-01
-2.04655170e-01 3.41255724e-01 -2.30007470e-01 -8.44527006e-01
-7.20472515e-01 -9.69838381e-01 1.04833460e+00 2.71680385e-01
-3.57980847e-01 -9.99458730e-01 6.61005229e-02 5.58304906e-01
3.23175699e-01 2.62338407e-02 8.04969907e-01 -1.11442840e+00
-2.21270710e-01 -5.94236195e-01 -1.16450833e-02 5.91744363e-01
7.24157989e-01 -8.27472359e-02 -8.07005763e-01 -7.49831051e-02
3.95883650e-01 -2.49822885e-01 7.56988525e-01 2.37351909e-01
5.64800501e-01 -2.32423618e-01 -1.35499537e-01 -8.01132172e-02
1.40363407e+00 2.84989625e-01 2.03487322e-01 4.00394976e-01
6.60141528e-01 7.23375738e-01 1.33186495e+00 3.44149113e-01
6.18883491e-01 2.84185767e-01 7.77964368e-02 4.61084694e-02
2.37248659e-01 4.24639583e-02 3.27229917e-01 1.46906555e+00
3.29166830e-01 1.55464217e-01 -7.43917882e-01 9.23720658e-01
-1.69773960e+00 -7.02600241e-01 -2.54359543e-01 2.14141917e+00
9.64046061e-01 4.03928012e-01 9.82104540e-02 5.04210889e-01
7.02730477e-01 1.48093570e-02 -3.07539076e-01 -7.40018368e-01
1.54756129e-01 9.71891806e-02 -3.43605913e-02 2.53280759e-01
-1.44664085e+00 9.04852927e-01 5.52147436e+00 5.68550408e-01
-7.62433290e-01 1.43863946e-01 7.15296686e-01 3.36464465e-01
-3.50956410e-01 2.83191353e-01 -9.18959558e-01 2.35796288e-01
7.09614575e-01 -1.85963273e-01 -1.76003188e-01 1.07003105e+00
1.99636996e-01 -4.84393537e-01 -9.65877891e-01 6.55432761e-01
3.87429833e-01 -7.46747613e-01 -4.52800356e-02 -2.70189822e-01
8.80452693e-01 -3.53191435e-01 8.78003687e-02 5.82358658e-01
2.25142151e-01 -8.01768243e-01 1.27091095e-01 -7.24656209e-02
3.39323819e-01 -9.26792979e-01 9.61889207e-01 2.39498705e-01
-1.40189159e+00 2.79715806e-01 -5.71512938e-01 -2.07889974e-01
1.65415760e-02 9.85239387e-01 -7.54174531e-01 2.52790213e-01
5.64139724e-01 1.02714622e+00 -5.15019953e-01 9.73429799e-01
-3.99690270e-01 9.86924410e-01 -8.38314518e-02 -3.69874567e-01
3.67387563e-01 -3.99988592e-01 3.23975146e-01 1.12128234e+00
-2.88787317e-02 2.43740659e-02 2.51868069e-01 2.80960351e-01
1.10413469e-01 5.86926818e-01 -6.46778762e-01 -7.62133449e-02
6.14988357e-02 1.63435936e+00 -1.08229911e+00 -4.05856341e-01
-8.22567046e-01 6.44970894e-01 2.27205791e-02 1.10455953e-01
-5.14715672e-01 -5.61374307e-01 4.58513409e-01 -7.15606436e-02
4.79717404e-01 -1.62923634e-01 -6.41152382e-01 -1.21952283e+00
7.48421550e-02 -9.93502855e-01 3.71978372e-01 -2.93686748e-01
-1.24813724e+00 7.44888127e-01 -2.53042996e-01 -1.74270356e+00
-1.99147508e-01 -4.14417297e-01 -8.06301355e-01 4.05388027e-01
-1.78665066e+00 -1.05452490e+00 -2.48027384e-01 3.28203112e-01
1.10211110e+00 -2.69714713e-01 7.99210250e-01 1.16062105e-01
-3.13538611e-01 5.15686154e-01 -1.53984889e-01 2.80670114e-02
8.78196299e-01 -1.57104003e+00 2.14486808e-01 9.02489185e-01
2.31930420e-01 9.50010538e-01 8.92158389e-01 -7.12336302e-01
-1.27313995e+00 -1.12890899e+00 9.87962484e-01 -5.09526610e-01
5.99557102e-01 -2.44983196e-01 -7.02095926e-01 5.43462038e-01
4.24661040e-01 -2.25486383e-01 1.33850312e+00 3.92360806e-01
-4.52905625e-01 -1.51043504e-01 -9.96315479e-01 2.88170278e-01
4.01207566e-01 -4.07175064e-01 -9.65377927e-01 3.47480744e-01
4.51823056e-01 -4.26042452e-03 -6.45865977e-01 3.81140828e-01
4.26846176e-01 -7.22551107e-01 5.44586718e-01 -4.84977305e-01
5.78388333e-01 -6.88287318e-01 -1.37529984e-01 -1.36999965e+00
1.70210853e-01 -2.26765156e-01 1.07174695e-01 1.57236135e+00
7.51997650e-01 -4.79863048e-01 8.35818350e-01 6.82030201e-01
4.08446565e-02 -7.04471231e-01 -3.55258375e-01 -5.04963875e-01
-6.94905641e-03 -3.16287458e-01 4.02221441e-01 8.97131383e-01
2.12225631e-01 8.23738515e-01 -1.74768418e-01 5.19709662e-04
5.86463809e-01 4.60882306e-01 8.90712798e-01 -1.03854597e+00
-3.17573994e-01 -1.30560204e-01 -3.04986149e-01 -7.87438333e-01
-3.11556235e-02 -6.23582125e-01 3.02951187e-01 -1.70474148e+00
4.58827525e-01 -5.05732119e-01 -5.23513377e-01 4.94411260e-01
-7.77401507e-01 4.06659156e-01 -1.45143732e-01 2.50155568e-01
-1.08393908e+00 6.73025250e-01 8.55020642e-01 -3.45037758e-01
-2.83730388e-01 3.80285442e-01 -1.14929509e+00 8.67709219e-01
1.08346462e+00 -6.27777755e-01 -6.80476010e-01 -1.72127232e-01
2.97712594e-01 -6.00807309e-01 -2.99962848e-01 -6.21793091e-01
1.23067640e-01 -1.13990456e-01 1.33746341e-02 -8.52664411e-01
1.10183015e-01 -9.65965927e-01 -4.73257154e-01 6.30623177e-02
-2.25916952e-01 2.71341503e-01 -7.42110759e-02 6.76298261e-01
-7.33095646e-01 -3.51059526e-01 5.74708521e-01 -1.29536226e-01
-7.33109355e-01 -5.64511456e-02 -5.30836284e-01 2.33213365e-01
1.02504134e+00 -1.49799898e-01 -8.82442519e-02 -3.28868985e-01
-4.52258319e-01 1.32625952e-01 3.64573359e-01 4.03865963e-01
5.82849264e-01 -1.07305706e+00 -3.66093308e-01 -5.29454164e-02
5.96873581e-01 -2.37391535e-02 -2.50919580e-01 9.36226487e-01
-2.23534733e-01 3.40506285e-01 1.81965783e-01 -5.81537843e-01
-1.35043669e+00 5.00067592e-01 -1.10255376e-01 -4.26027387e-01
-2.45808080e-01 6.13145530e-01 2.46086165e-01 -7.04567611e-01
-4.09742445e-03 -1.92807481e-01 -8.59086156e-01 2.85298288e-01
6.14231765e-01 -1.92004949e-01 2.75487632e-01 -6.32965565e-01
-4.84434605e-01 7.90749192e-01 -4.78700519e-01 3.35197560e-02
1.37967718e+00 -3.34140062e-01 -1.51793644e-01 5.92084229e-01
1.16440153e+00 1.94650646e-02 -7.28304386e-01 -2.21629888e-01
3.49165440e-01 -1.90549970e-01 -9.52630043e-02 -6.87762976e-01
-9.20785844e-01 7.57035911e-01 2.71315604e-01 3.40956688e-01
1.17592800e+00 -1.66664645e-01 7.25417435e-01 4.03077036e-01
3.11477125e-01 -1.29792380e+00 1.39024541e-01 5.33320963e-01
3.87190551e-01 -1.82395649e+00 2.38155141e-01 -6.30507946e-01
-8.94228816e-01 8.17205012e-01 7.97429323e-01 -3.12894851e-01
1.05205131e+00 1.07191682e-01 4.60529804e-01 -3.70090753e-01
-7.43502438e-01 -4.12252456e-01 5.18091619e-01 4.04520333e-01
7.63147473e-01 -2.90876757e-02 -8.54660153e-01 8.10965002e-01
-1.35961264e-01 -4.85224962e-01 5.19396007e-01 1.34568906e+00
-7.35265136e-01 -1.21831381e+00 -1.49008512e-01 7.91130006e-01
-8.37772012e-01 -3.04291964e-01 -3.82259846e-01 5.39930940e-01
-2.22283512e-01 1.43025267e+00 -3.34202558e-01 -1.65607333e-01
2.42947742e-01 -4.23015617e-02 -1.72471777e-01 -1.16314602e+00
-9.30302083e-01 2.35355303e-01 2.46978581e-01 -2.24515833e-02
-7.83160925e-01 -7.29728281e-01 -1.16699183e+00 3.60297710e-01
-6.29942954e-01 6.61853254e-01 9.09809589e-01 1.25060523e+00
1.58096537e-01 1.32804945e-01 9.68806267e-01 -1.81651413e-01
-4.53016728e-01 -1.08553231e+00 -5.79829991e-01 5.98745883e-01
1.18568763e-01 -5.61914206e-01 -6.36190832e-01 2.41809428e-01] | [11.329687118530273, 6.678438663482666] |
a0560673-623d-4a47-9c0d-2b720714aa31 | welfare-and-fairness-in-multi-objective | 2212.01382 | null | https://arxiv.org/abs/2212.01382v3 | https://arxiv.org/pdf/2212.01382v3.pdf | Welfare and Fairness in Multi-objective Reinforcement Learning | We study fair multi-objective reinforcement learning in which an agent must learn a policy that simultaneously achieves high reward on multiple dimensions of a vector-valued reward. Motivated by the fair resource allocation literature, we model this as an expected welfare maximization problem, for some non-linear fair welfare function of the vector of long-term cumulative rewards. One canonical example of such a function is the Nash Social Welfare, or geometric mean, the log transform of which is also known as the Proportional Fairness objective. We show that even approximately optimal optimization of the expected Nash Social Welfare is computationally intractable even in the tabular case. Nevertheless, we provide a novel adaptation of Q-learning that combines non-linear scalarized learning updates and non-stationary action selection to learn effective policies for optimizing nonlinear welfare functions. We show that our algorithm is provably convergent, and we demonstrate experimentally that our approach outperforms techniques based on linear scalarization, mixtures of optimal linear scalarizations, or stationary action selection for the Nash Social Welfare Objective. | ['Brandon Fain', 'Muhang Tian', 'Nianli Peng', 'Zimeng Fan'] | 2022-11-30 | null | null | null | null | ['multi-objective-reinforcement-learning'] | ['methodology'] | [-2.28494912e-01 1.98940799e-01 -6.53031707e-01 -2.32236236e-01
-1.23245656e+00 -5.57249367e-01 1.87669694e-01 1.27681494e-01
-1.10746014e+00 1.52415538e+00 3.74975324e-01 -3.43256742e-01
-6.32115126e-01 -6.09416902e-01 -4.97045875e-01 -7.97157109e-01
-5.23283005e-01 4.56914157e-01 -5.00274241e-01 -5.03836721e-02
4.11427557e-01 3.26665074e-01 -1.20759475e+00 -5.53826988e-01
9.79954004e-01 1.20521748e+00 -1.14764065e-01 8.13414335e-01
-1.09321915e-01 8.98425698e-01 -6.78410470e-01 -4.69869256e-01
6.45529568e-01 -5.83236933e-01 -8.79020333e-01 2.78434195e-02
1.81788042e-01 -8.02422523e-01 2.40454644e-01 1.27831256e+00
5.60409606e-01 3.10541749e-01 6.87308908e-01 -1.70248222e+00
-4.43658650e-01 8.09039891e-01 -6.27804756e-01 -2.71234453e-01
1.77767053e-02 2.91576713e-01 1.68317199e+00 -1.53321372e-02
4.07911092e-01 1.50779045e+00 2.02844575e-01 8.21838319e-01
-1.62814152e+00 -2.81400204e-01 2.18627378e-02 -2.09456831e-01
-4.88579929e-01 -2.74427474e-01 2.01773316e-01 -1.28771901e-01
6.60749674e-01 3.75048339e-01 6.38663709e-01 3.85790199e-01
1.17185935e-01 9.54158366e-01 1.26393759e+00 -3.07331741e-01
6.80294216e-01 6.40752241e-02 -2.43315682e-01 5.82143307e-01
3.38269889e-01 4.96591389e-01 -1.16950631e-01 -5.98646164e-01
7.12079465e-01 -5.37348092e-02 1.18590117e-01 -8.44377100e-01
-9.24174070e-01 1.15983474e+00 1.57502607e-01 -1.95041358e-01
-8.62229586e-01 8.57273459e-01 5.71412861e-01 6.83644891e-01
4.31976557e-01 7.82579839e-01 -5.88408887e-01 -3.16603661e-01
-7.92436659e-01 6.77514195e-01 1.02805817e+00 5.96863449e-01
7.11953044e-01 3.95965040e-01 -5.88456392e-01 6.17646337e-01
3.88388067e-01 8.91498208e-01 -1.86389774e-01 -1.91065824e+00
3.33094686e-01 -3.37789096e-02 9.55900908e-01 -1.68752536e-01
-4.94532734e-01 -5.42655945e-01 -3.92867863e-01 7.65597522e-01
7.76610017e-01 -1.10052538e+00 -2.42628530e-02 2.27875423e+00
1.20727472e-01 -3.88274312e-01 1.96582600e-01 8.50779235e-01
-2.06494957e-01 5.63454151e-01 3.73825163e-01 -1.11954916e+00
6.99230254e-01 -6.70540988e-01 -8.52722466e-01 1.15467332e-01
4.61188823e-01 -4.62638170e-01 8.72346461e-01 7.34610483e-02
-1.53760278e+00 2.44008437e-01 -5.56806564e-01 4.18041825e-01
2.55209267e-01 -2.91142166e-01 7.00637519e-01 7.58495152e-01
-1.10944211e+00 9.75301445e-01 -2.16547772e-01 -5.45940781e-03
6.37832463e-01 4.72201556e-01 2.88949937e-01 4.15586084e-01
-1.04838240e+00 9.72311974e-01 1.94547638e-01 -4.92137730e-01
-1.13429320e+00 -8.79432619e-01 -4.31513906e-01 3.66535157e-01
9.64532971e-01 -8.02367687e-01 1.70656383e+00 -1.26682246e+00
-1.96856105e+00 6.74723983e-01 4.17923450e-01 -6.25328422e-01
9.11157429e-01 -1.50958687e-01 2.92502403e-01 -7.54246265e-02
2.28672028e-01 3.37518245e-01 9.40277934e-01 -1.13417614e+00
-7.36790061e-01 -2.23304525e-01 5.50527573e-01 6.06428087e-01
-4.23871756e-01 3.50463331e-01 6.71447277e-01 -1.37227222e-01
-9.50014174e-01 -6.79629385e-01 -8.18542242e-01 2.71212250e-01
-2.60060161e-01 -2.79999256e-01 1.22948155e-01 -3.60395223e-01
1.00247777e+00 -1.69064569e+00 8.57495219e-02 2.64724225e-01
1.51335821e-01 -2.24806458e-01 -4.48378623e-01 3.56252849e-01
2.51911193e-01 8.02496150e-02 -2.33056813e-01 -9.89461392e-02
7.31065929e-01 2.11598575e-01 -1.02102108e-01 7.99168825e-01
-1.60080522e-01 9.01461899e-01 -1.13243198e+00 -4.12399888e-01
-5.96308149e-02 -5.30943453e-01 -8.05393040e-01 4.46126819e-01
-4.17680204e-01 -1.75600752e-01 -6.04360521e-01 1.82828173e-01
5.33702850e-01 -9.96033698e-02 2.91387141e-01 3.40476185e-01
-2.34346062e-01 -2.35050395e-01 -1.34631693e+00 1.21680689e+00
-3.57646823e-01 9.62404311e-02 5.87809205e-01 -1.19347823e+00
5.10942101e-01 1.84132800e-01 1.31739128e+00 -3.62667203e-01
3.49440157e-01 2.00851098e-01 -2.88865685e-01 -2.62554586e-01
6.88067019e-01 -8.11080039e-01 -3.98155242e-01 1.10073042e+00
3.50630842e-03 -1.12991802e-01 1.91582069e-01 2.92410195e-01
9.47304428e-01 1.18921310e-01 6.76743448e-01 -7.13318348e-01
2.69749850e-01 -1.05074748e-01 4.80505526e-01 8.70179534e-01
-6.47953391e-01 -2.00810730e-01 1.27916932e+00 3.84961553e-02
-1.21467590e+00 -1.15412104e+00 1.47775695e-01 1.60500681e+00
6.37063757e-02 1.53152555e-01 -6.63425624e-01 -8.22635531e-01
7.14825928e-01 7.85572648e-01 -4.69196290e-01 1.86951347e-02
-2.44152844e-01 -8.96160722e-01 3.83783914e-02 1.96237266e-02
1.99881151e-01 -9.65813458e-01 -8.54777694e-01 4.02951837e-01
1.77609041e-01 -5.49822032e-01 -9.11276698e-01 1.47734389e-01
-6.57312453e-01 -9.15758669e-01 -1.16196537e+00 2.10048947e-02
3.76669168e-01 -9.31707025e-02 1.12614155e+00 -4.32698339e-01
3.41532007e-02 8.68913412e-01 2.59378552e-01 -4.32529837e-01
-3.79190803e-01 -3.56133491e-01 2.25298479e-01 8.33997801e-02
-1.54254690e-01 -1.98124066e-01 -5.95935762e-01 6.10672869e-02
-8.01649690e-01 -4.41022575e-01 1.88512221e-01 9.33386624e-01
4.74150091e-01 -3.40920538e-01 1.30336690e+00 -7.53235161e-01
1.49721062e+00 -5.58288515e-01 -1.09996915e+00 4.54297423e-01
-8.25492561e-01 6.93300366e-01 8.09688389e-01 -2.94491291e-01
-9.37577367e-01 -2.07563490e-01 1.93167523e-01 -3.08484975e-02
4.46972787e-01 1.94632381e-01 -8.27274006e-03 -1.48679093e-01
5.49985826e-01 -1.53499410e-01 4.02155131e-01 -9.20070037e-02
7.89214730e-01 5.14405787e-01 2.68143192e-02 -1.08713496e+00
4.70020473e-01 8.03009644e-02 4.55662251e-01 -2.88805753e-01
-7.83149064e-01 -7.12172762e-02 2.40355596e-01 -4.66140926e-01
4.22806531e-01 -5.48730016e-01 -1.68049622e+00 2.20516026e-02
-8.24388564e-01 -5.33652604e-01 -1.16945601e+00 5.60663223e-01
-1.59313977e+00 1.36514008e-01 -2.74463475e-01 -1.53895354e+00
-4.59986180e-01 -7.21259296e-01 6.81507528e-01 3.11131895e-01
2.39613160e-01 -9.87471581e-01 3.00734103e-01 -2.01124460e-01
7.01416671e-01 3.32438171e-01 7.82999933e-01 -2.23709404e-01
-3.75857830e-01 1.97702274e-02 -2.67931521e-01 5.21471918e-01
-1.37337744e-01 -1.45809308e-01 -4.81502533e-01 -5.70978343e-01
-2.04491764e-01 -9.48732257e-01 5.49911857e-01 8.84191334e-01
9.90946651e-01 -7.90582061e-01 3.29696327e-01 2.55894452e-01
1.87123239e+00 5.83816953e-02 4.16359566e-02 2.18776971e-01
6.03937581e-02 5.98402143e-01 7.37538576e-01 1.29336667e+00
3.07580531e-01 4.47255045e-01 8.01487565e-01 1.65553004e-01
7.17611015e-01 5.14914431e-02 5.44977069e-01 -2.33562523e-03
-4.08464491e-01 5.81009351e-02 -2.82945186e-01 3.75663072e-01
-2.19569802e+00 -1.33367038e+00 5.57573676e-01 2.58136249e+00
9.14976358e-01 -1.97676077e-01 9.43788111e-01 -3.15538138e-01
5.54967880e-01 1.60372064e-01 -1.17504859e+00 -9.33251023e-01
-4.40408587e-02 1.47184096e-02 1.21153569e+00 7.55248070e-01
-8.65794778e-01 6.12324834e-01 7.22564697e+00 9.03704047e-01
-6.40212238e-01 1.70654297e-01 6.77575231e-01 -6.13301635e-01
-5.45796216e-01 -2.20259607e-01 -3.74607831e-01 2.15279281e-01
8.89034927e-01 -1.15450084e+00 1.04427493e+00 1.01748526e+00
6.37412488e-01 -1.80702209e-01 -8.49199116e-01 7.31109977e-01
-6.55821621e-01 -1.13185644e+00 -5.96137047e-01 4.65055496e-01
1.14862704e+00 -2.30461285e-01 4.83831875e-02 5.01069963e-01
1.08187950e+00 -7.88954496e-01 6.17396295e-01 6.94153309e-01
9.10021007e-01 -1.36554050e+00 3.30320597e-01 3.25210720e-01
-6.74408019e-01 -3.93789679e-01 -4.21230495e-01 -8.32423791e-02
3.72512907e-01 3.01217139e-01 -1.78045034e-01 2.69858897e-01
-5.04606850e-02 3.55955869e-01 3.64568383e-01 1.25286996e+00
1.05352648e-01 2.07602531e-01 -2.35524073e-01 -5.52338481e-01
6.84776545e-01 -3.87725562e-01 5.59960783e-01 8.61844063e-01
2.52778500e-01 -1.50032550e-01 4.37090367e-01 8.44441831e-01
-3.93369138e-01 4.94840592e-01 -4.28547323e-01 -1.54063836e-01
-2.41546350e-05 1.26159120e+00 -2.11492896e-01 -2.07976431e-01
-1.74529195e-01 6.45116508e-01 4.56311971e-01 5.38207948e-01
-8.20854366e-01 -2.20077381e-01 1.41045821e+00 -4.18467939e-01
-1.23472951e-01 4.18442301e-02 -1.08317517e-01 -1.05756938e+00
-2.33248681e-01 -6.49827123e-01 4.69331354e-01 -2.05739632e-01
-1.51819479e+00 -5.68139963e-02 -7.44235143e-03 -1.06428492e+00
-7.05391169e-01 -4.75121766e-01 -4.73899662e-01 9.28841949e-01
-1.64030266e+00 -3.48077446e-01 6.92857265e-01 5.85129619e-01
1.41545879e-02 -2.88280606e-01 8.31573308e-01 -2.14687269e-02
-3.45766664e-01 4.65108752e-01 8.50609183e-01 -5.67122877e-01
5.93165576e-01 -1.79497600e+00 -5.38817585e-01 3.64720374e-01
-5.53526163e-01 -7.79467970e-02 1.03490555e+00 -2.29028225e-01
-1.43973112e+00 -8.88927817e-01 4.75728929e-01 2.53194004e-01
1.01229620e+00 1.91816777e-01 -1.61294460e-01 2.77254552e-01
3.27845663e-01 1.12635694e-01 4.36093509e-01 2.26551667e-02
2.96030492e-01 -4.53645766e-01 -1.64460886e+00 7.10482180e-01
7.63726234e-01 -1.10015437e-01 -4.25289944e-03 5.89622319e-01
6.92222536e-01 1.85355637e-02 -9.09346044e-01 -2.40158245e-01
6.91456497e-01 -6.65279567e-01 8.86529744e-01 -1.33984768e+00
4.86341685e-01 3.24612379e-01 -2.73498774e-01 -1.78723562e+00
-3.34173888e-01 -1.19220293e+00 -5.72795160e-02 6.81622624e-01
2.12298289e-01 -6.10908270e-01 8.90715182e-01 9.18355644e-01
3.45471442e-01 -8.04940701e-01 -1.11873329e+00 -1.17225325e+00
5.77195704e-01 -8.40689391e-02 5.25325894e-01 5.97242594e-01
3.24451923e-01 8.94723535e-02 -1.05496562e+00 -5.24575293e-01
1.42273951e+00 3.22989166e-01 5.12325346e-01 -8.59268963e-01
-4.54907864e-01 -1.01174569e+00 1.03435934e-01 -9.12153065e-01
5.44756889e-01 -7.45728314e-01 7.14016035e-02 -1.40040469e+00
3.66591454e-01 -3.50480407e-01 -4.70840275e-01 2.56138474e-01
-8.78223851e-02 -4.99252021e-01 7.00206041e-01 -2.56864935e-01
-1.12320828e+00 9.41372573e-01 1.61313927e+00 -2.33575385e-02
-1.33475199e-01 3.92534286e-01 -1.05607605e+00 3.98440182e-01
9.40357685e-01 -7.51862943e-01 -4.01718736e-01 1.15262434e-01
2.37490773e-01 8.29004884e-01 8.36299639e-03 -2.45434254e-01
-7.43763298e-02 -1.25692904e+00 -2.79482186e-01 -2.72792876e-02
7.55164698e-02 -8.49636137e-01 -2.70032585e-01 7.04501510e-01
-9.10497546e-01 -1.49892613e-01 -2.55511135e-01 5.43603063e-01
2.12660238e-01 -5.66359580e-01 1.13739336e+00 -2.55094916e-01
-2.62260824e-01 5.67456007e-01 -4.14065510e-01 6.83369279e-01
1.33212376e+00 4.47979987e-01 -3.13156039e-01 -9.83325779e-01
-4.53090280e-01 7.46298254e-01 7.84796253e-02 -1.87979743e-01
4.19765472e-01 -1.45968640e+00 -1.04038119e+00 -5.71962893e-01
-2.67515570e-01 -9.94202137e-01 1.04437605e-03 5.16976535e-01
-6.68428689e-02 2.78876781e-01 -5.15935063e-01 1.72947288e-01
-9.56810296e-01 3.32315654e-01 7.02683449e-01 -7.31449246e-01
1.07904918e-01 4.60362583e-01 -1.48940891e-01 -3.27419698e-01
2.84742326e-01 4.19045165e-02 -1.06944712e-02 2.71351695e-01
5.31799436e-01 8.06026936e-01 -7.83743739e-01 -2.54174948e-01
-3.54536138e-02 1.96687132e-01 3.87252778e-01 -6.84261918e-01
1.45058417e+00 -3.34363580e-01 -1.61095083e-01 1.54808491e-01
1.17499733e+00 -3.01966578e-01 -1.65653825e+00 -4.97776985e-01
-1.52360471e-02 -6.30533397e-01 6.33689836e-02 -8.60891342e-01
-1.04875267e+00 5.11670530e-01 5.90982676e-01 5.70705652e-01
9.28090632e-01 -4.52587545e-01 4.23509806e-01 5.61711133e-01
6.52702510e-01 -1.55901825e+00 -1.30857736e-01 3.90215099e-01
9.41394746e-01 -1.29664028e+00 1.42221615e-01 4.93754745e-01
-9.72731411e-01 1.07995248e+00 2.61161000e-01 -5.42059183e-01
7.02759326e-01 3.04933518e-01 -1.49135143e-01 3.29126745e-01
-8.60196531e-01 -6.78672969e-01 -1.94552153e-01 3.67616117e-01
1.70873970e-01 7.58866787e-01 -8.95452678e-01 4.01452571e-01
6.06647357e-02 1.19339518e-01 6.18818700e-01 8.90236378e-01
-7.07179785e-01 -1.20551598e+00 -2.38411993e-01 9.49208379e-01
-7.67866373e-01 1.54072821e-01 1.09626204e-01 2.77372062e-01
-4.59584415e-01 8.63392830e-01 -1.54479221e-01 2.55320340e-01
3.22200209e-01 -1.95049852e-01 6.88446581e-01 -2.91561216e-01
-6.78781331e-01 1.83526635e-01 2.11546227e-01 -7.71804273e-01
-5.40164173e-01 -7.71594822e-01 -1.03638399e+00 -7.87532210e-01
6.59805015e-02 3.81455600e-01 5.90494752e-01 9.76429343e-01
-3.05392832e-01 4.64464486e-01 1.14909899e+00 -5.27926445e-01
-1.86041808e+00 -5.08705914e-01 -1.26359808e+00 3.73928636e-01
6.55940413e-01 -4.45738435e-01 -3.84580642e-01 -5.90522170e-01] | [4.238687038421631, 2.637141466140747] |
88699d38-205e-4097-aea9-9fe4be7f0269 | an-additive-latent-feature-model-for | null | null | http://papers.nips.cc/paper/3808-an-additive-latent-feature-model-for-transparent-object-recognition | http://papers.nips.cc/paper/3808-an-additive-latent-feature-model-for-transparent-object-recognition.pdf | An Additive Latent Feature Model for Transparent Object Recognition | Existing methods for recognition of object instances and categories based on quantized local features can perform poorly when local features exist on transparent surfaces, such as glass or plastic objects. There are characteristic patterns to the local appearance of transparent objects, but they may not be well captured by distances to individual examples or by a local pattern codebook obtained by vector quantization. The appearance of a transparent patch is determined in part by the refraction of a background pattern through a transparent medium: the energy from the background usually dominates the patch appearance. We model transparent local patch appearance using an additive model of latent factors: background factors due to scene content, and factors which capture a local edge energy distribution characteristic of the refraction. We implement our method using a novel LDA-SIFT formulation which performs LDA prior to any vector quantization step; we discover latent topics which are characteristic of particular transparent patches and quantize the SIFT space into transparent visual words according to the latent topic dimensions. No knowledge of the background scene is required at test time; we show examples recognizing transparent glasses in a domestic environment. | ['Sergey Karayev', 'Michael J. Black', 'Gary Bradski', 'Trevor Darrell', 'Mario Fritz'] | 2009-12-01 | null | null | null | neurips-2009-12 | ['transparent-objects'] | ['computer-vision'] | [ 2.43962258e-01 -2.96849579e-01 -1.34307235e-01 -4.37188983e-01
-3.73700023e-01 -4.67533022e-01 6.19647026e-01 -1.67765424e-01
3.94763499e-01 5.73309958e-02 3.92735809e-01 4.47662354e-01
1.99227668e-02 -8.82825196e-01 -7.91042507e-01 -1.05677402e+00
-1.06616624e-01 4.23625827e-01 5.22668183e-01 -1.81137715e-02
2.66873270e-01 6.13769293e-01 -1.89299929e+00 9.17111933e-01
3.30228537e-01 1.08724201e+00 1.40712932e-01 7.56242394e-01
-2.72644401e-01 7.64750481e-01 -5.35608053e-01 -1.69774652e-01
5.49662232e-01 -2.60997862e-01 -4.50057536e-01 9.02483702e-01
1.05255711e+00 -2.68455863e-01 -1.12347327e-01 1.26552701e+00
-1.40450969e-01 -1.11542000e-02 1.34019768e+00 -7.73938775e-01
-9.25865054e-01 -8.67957249e-02 -6.81950092e-01 -2.01046690e-02
4.35109615e-01 -2.23794073e-01 9.92083848e-01 -8.48031521e-01
9.38104808e-01 1.52458465e+00 4.86656070e-01 1.22214392e-01
-1.41762650e+00 -5.31635843e-02 3.50373656e-01 2.34916657e-01
-1.43812549e+00 -3.92347813e-01 1.03670406e+00 -6.96097970e-01
5.64041793e-01 6.62577808e-01 7.66932726e-01 7.24653184e-01
7.55981743e-01 7.84134150e-01 8.98218155e-01 -5.78315914e-01
3.37305337e-01 5.98192334e-01 4.00925688e-02 8.82839084e-01
6.44178316e-02 -3.24994624e-01 -5.53209901e-01 -5.19220412e-01
6.33308649e-01 8.44166502e-02 -4.03091937e-01 -8.48847330e-01
-8.05030882e-01 7.61145711e-01 4.02176023e-01 2.17459723e-01
-3.51545483e-01 1.02209993e-01 -1.50145367e-01 3.99072349e-01
8.20911527e-01 4.95619886e-02 -3.84925574e-01 3.35405022e-01
-8.88060987e-01 -2.88465340e-02 8.30842495e-01 9.27604139e-01
1.25507092e+00 -7.47828707e-02 6.50347918e-02 7.98745096e-01
5.43738067e-01 6.46091282e-01 2.44768351e-01 -7.59759784e-01
-2.27441173e-02 7.01504529e-01 1.61306635e-01 -1.12194252e+00
1.03195021e-02 -3.06487698e-02 -1.51943356e-01 5.11884272e-01
1.90443635e-01 4.78347331e-01 -1.00088656e+00 1.25518858e+00
5.17909467e-01 -2.74047643e-01 -2.32413158e-01 7.25220442e-01
6.91931129e-01 8.50027263e-01 -4.00920361e-01 -1.80686623e-01
1.56552279e+00 -6.84531689e-01 -7.85437703e-01 -1.16511010e-01
1.94394320e-01 -1.23314381e+00 1.08506095e+00 5.32247305e-01
-1.01104844e+00 -4.76652086e-01 -9.48563576e-01 1.87443383e-02
-2.75843620e-01 4.42748338e-01 6.30098701e-01 7.44407296e-01
-1.13143480e+00 2.51647472e-01 -8.98440003e-01 -4.48119968e-01
1.73064172e-02 3.34535152e-01 -2.87886143e-01 -1.93722248e-01
-3.93277049e-01 5.35560012e-01 -4.22970206e-01 8.81761238e-02
-6.99506104e-01 -4.70648080e-01 -7.20316768e-01 -1.18960656e-01
5.27281798e-02 -3.00924301e-01 7.41148889e-01 -1.34886503e+00
-1.30132556e+00 9.25912380e-01 -6.60567999e-01 3.04152846e-01
3.11315298e-01 -1.26845360e-01 -4.88763958e-01 5.50754249e-01
-4.82843220e-02 6.59267232e-02 1.58773887e+00 -1.83508503e+00
-4.06562053e-02 -3.33227545e-01 -2.53202263e-02 2.92932749e-01
-4.28141922e-01 2.48814046e-01 -5.50199926e-01 -5.48752189e-01
4.65198696e-01 -1.06512880e+00 8.53679851e-02 3.59278947e-01
-3.50487679e-01 -6.18794225e-02 1.14971364e+00 -5.48700392e-01
7.34707355e-01 -2.50463867e+00 -2.73652494e-01 5.10047793e-01
3.21910799e-01 -4.17306155e-01 2.27451995e-02 5.05810440e-01
-2.17581987e-02 -1.17695019e-01 7.87954181e-02 -1.55177429e-01
7.22313970e-02 -4.33304906e-03 -2.90894181e-01 9.13539767e-01
-9.41315200e-03 3.90604258e-01 -5.34604609e-01 -4.52723563e-01
1.45331100e-01 6.99446261e-01 -5.23145139e-01 -4.17102389e-02
-2.11154632e-02 -2.56351113e-01 -6.01870716e-01 6.70498073e-01
8.91642272e-01 -1.19232893e-01 -9.83532518e-02 -6.15521431e-01
-1.80540636e-01 2.43216738e-01 -1.23213625e+00 1.25757933e+00
-1.89583495e-01 8.03857803e-01 2.03849941e-01 -3.68871003e-01
8.96917462e-01 2.18856797e-01 4.27675903e-01 -2.14171857e-01
-4.18760553e-02 5.94295561e-03 -2.15093389e-01 -5.48588932e-01
5.15096843e-01 -6.04713187e-02 3.09347004e-01 3.53949815e-01
-1.38636589e-01 -3.19421768e-01 -2.83928692e-01 1.56126291e-01
1.12522614e+00 8.11696947e-02 -1.34063020e-01 -5.99259794e-01
3.83728258e-02 -9.86577794e-02 3.25089216e-01 5.63402236e-01
2.56255239e-01 8.61313403e-01 -6.03765510e-02 -7.93058455e-01
-7.04489112e-01 -1.36859953e+00 -5.92553914e-01 9.14782345e-01
4.81642812e-01 -5.16331494e-01 -5.49886584e-01 -2.79603153e-01
2.78581139e-02 2.47316271e-01 -8.46102059e-01 -4.51548606e-01
-1.38254628e-01 -8.76010358e-01 -4.43842620e-01 7.37875700e-02
2.85411239e-01 -8.55134964e-01 -5.60053945e-01 8.92166421e-02
-1.43183698e-03 -1.01479542e+00 -6.37982249e-01 2.20869496e-01
-7.01808453e-01 -7.83280969e-01 -4.07923788e-01 -7.57359684e-01
1.14804935e+00 5.80943227e-01 1.12073541e+00 -1.80045679e-01
-8.85983527e-01 8.40640128e-01 -2.66386330e-01 -1.84098944e-01
-2.78362900e-01 -8.07869017e-01 2.34667465e-01 7.18317151e-01
4.27650750e-01 -3.41386646e-01 -7.32431948e-01 4.04639721e-01
-6.22197628e-01 -1.70687228e-01 4.67911929e-01 3.61112684e-01
5.65604687e-01 4.50733095e-01 -8.81949127e-01 -7.63051569e-01
1.28440276e-01 -3.18480432e-01 -4.31374222e-01 3.81957412e-01
2.49564424e-01 -4.46982030e-03 6.91817030e-02 -7.05477774e-01
-1.15714610e+00 6.65577948e-02 6.60222888e-01 -6.33232653e-01
-2.35619977e-01 -5.06755002e-02 -3.77858281e-01 -5.95333397e-01
6.95723236e-01 2.77435988e-01 -4.81952161e-01 -5.64634323e-01
1.97069198e-01 4.40110534e-01 -3.25848125e-02 -5.15158176e-01
1.05181122e+00 9.00106311e-01 -6.05713651e-02 -1.45938945e+00
-5.04202485e-01 -6.51943862e-01 -5.85910916e-01 -2.75188625e-01
9.35850441e-01 -8.80433202e-01 -5.88930130e-01 3.03627551e-01
-9.03419375e-01 -2.50047773e-01 -4.29293483e-01 5.42540729e-01
-4.30333257e-01 3.75351399e-01 -6.89144790e-01 -9.50516999e-01
2.26983830e-01 -1.00784838e+00 1.40263784e+00 -2.65648156e-01
-2.35653996e-01 -9.96433675e-01 4.59226370e-02 7.83384219e-03
3.21646035e-01 -1.69745870e-02 1.03262222e+00 2.03603640e-01
-1.02418149e+00 -2.91616410e-01 4.28418256e-02 3.36109459e-01
7.38879383e-01 4.20204699e-01 -1.35392225e+00 -2.06528217e-01
5.71111381e-01 -5.87657019e-02 8.77303839e-01 7.27413416e-01
9.21247959e-01 -6.29109204e-01 -4.43808228e-01 6.39492869e-01
1.64496553e+00 1.93415970e-01 3.00626963e-01 8.93501937e-02
1.00718594e+00 9.12494421e-01 2.41555855e-01 2.71015435e-01
6.62031248e-02 9.84552562e-01 2.22027466e-01 -2.97386169e-01
-3.94118458e-01 1.46710381e-01 4.71231490e-01 7.56821752e-01
-2.49269366e-01 -7.49983341e-02 -6.43011928e-01 4.64102536e-01
-1.10311437e+00 -9.10950482e-01 -3.66749674e-01 2.43771219e+00
4.78829563e-01 2.37184041e-03 -2.60097861e-01 -4.74174991e-02
5.24452150e-01 -5.86376525e-02 -3.65153432e-01 -4.31069613e-01
-3.24814796e-01 -2.40649775e-01 5.42488396e-01 5.88225543e-01
-1.01778483e+00 7.28880942e-01 7.12105751e+00 7.05454290e-01
-1.18419576e+00 -1.71921737e-02 3.96199316e-01 -5.34508713e-02
-7.84793377e-01 5.24627790e-02 -7.64753878e-01 2.66332477e-01
4.49523866e-01 2.06576705e-01 2.20601588e-01 7.95016766e-01
-5.59399426e-02 -1.31470598e-02 -1.04645932e+00 9.13373888e-01
3.59197259e-01 -8.20561886e-01 3.56418401e-01 5.00537574e-01
9.62382495e-01 4.46963161e-02 5.64135611e-01 -5.84124088e-01
3.69455367e-01 -4.00305688e-01 9.59510326e-01 5.80881238e-01
4.36364204e-01 -3.30475509e-01 2.31461659e-01 -1.88977182e-01
-1.34260345e+00 1.07116930e-01 -8.90154839e-01 2.63897240e-01
-3.52908343e-01 7.93777227e-01 -6.41012847e-01 5.23375794e-02
9.92270231e-01 5.85405171e-01 -5.86764932e-01 1.09199309e+00
7.41286427e-02 5.47112048e-01 -5.27082741e-01 4.38554920e-02
2.44313385e-03 -4.76438463e-01 7.34001935e-01 1.02344060e+00
2.30310500e-01 -1.51395157e-01 3.33069563e-01 9.78747129e-01
1.72120586e-01 2.15126440e-01 -6.28740966e-01 1.64802253e-01
-9.54018831e-02 1.21442175e+00 -1.05123067e+00 -4.17351753e-01
-6.65869296e-01 1.18230903e+00 -1.86183468e-01 6.28399730e-01
-2.78538793e-01 -1.27560005e-01 7.57418096e-01 3.47909510e-01
6.24400020e-01 -1.89129278e-01 6.91394806e-02 -1.52652311e+00
5.21494448e-01 -6.14830732e-01 -1.71533719e-01 -1.02921224e+00
-1.48425603e+00 7.33353913e-01 -4.85894792e-02 -1.62985706e+00
1.61255971e-01 -6.90173864e-01 -6.16150677e-01 7.12511778e-01
-1.11161661e+00 -1.23446727e+00 -2.97229677e-01 8.82064164e-01
5.23642778e-01 3.52899656e-02 8.54147673e-01 -2.02687338e-01
1.81470349e-01 1.63164899e-01 6.63512468e-01 -4.37993467e-01
6.03749692e-01 -1.40577257e+00 2.18873769e-01 6.61124647e-01
4.60679561e-01 8.41742694e-01 1.14534497e+00 -5.72111011e-01
-1.59055781e+00 -7.96783864e-01 6.16422534e-01 -8.11972857e-01
7.16508269e-01 -9.14309442e-01 -9.42545533e-01 4.22673494e-01
-7.21489778e-03 2.02140957e-01 6.83601797e-01 1.65308297e-01
-6.83978498e-01 -1.76475093e-01 -9.54171598e-01 4.01467651e-01
7.45172977e-01 -8.14032018e-01 -4.25338209e-01 8.04431677e-01
3.60986590e-01 -4.17431444e-02 -5.90197980e-01 9.50156990e-03
6.47194624e-01 -9.89496768e-01 9.52277303e-01 -2.78387200e-02
-4.44305465e-02 -2.68285990e-01 -4.16490883e-01 -9.59836841e-01
-6.30716741e-01 -4.21644509e-01 1.28795519e-01 1.10228026e+00
2.11064994e-01 -4.84261751e-01 8.85324121e-01 6.67057037e-01
-1.54485419e-01 -1.78617626e-01 -7.77628362e-01 -6.75858974e-01
-5.02745271e-01 -2.15386897e-01 3.52048188e-01 7.90794730e-01
-2.29614183e-01 -4.37749699e-02 -2.65715808e-01 4.62463111e-01
9.44045365e-01 6.42268121e-01 5.52742958e-01 -1.36301899e+00
-4.19355989e-01 -1.43633977e-01 -9.48911965e-01 -9.21110451e-01
-2.03406706e-01 -4.90877330e-01 3.15415353e-01 -1.43512928e+00
4.43689823e-01 -2.71396369e-01 -1.57775477e-01 2.67193675e-01
1.43748239e-01 5.25373161e-01 -1.00982711e-01 5.97449660e-01
-4.56008852e-01 6.48747027e-01 1.21990442e+00 -5.23038566e-01
-2.40472034e-01 -1.24231977e-02 -3.33472133e-01 9.47170258e-01
3.57284397e-01 -4.63616103e-01 -4.02733386e-01 -5.42778432e-01
3.05693567e-01 -4.20847118e-01 2.42573351e-01 -1.14001119e+00
-1.63484886e-01 -4.85297218e-02 6.06573939e-01 -2.19355062e-01
7.30369925e-01 -1.09677601e+00 1.96997732e-01 6.39932379e-02
-3.26489583e-02 -3.74034613e-01 -4.26842570e-02 7.75741518e-01
-2.02854797e-01 -6.34322017e-02 7.23563015e-01 -1.05345182e-01
-4.01359171e-01 2.72189170e-01 -8.21789622e-01 -7.57009387e-01
6.28625154e-01 -6.36049569e-01 -2.81112731e-01 -4.06514794e-01
-7.38384366e-01 -6.04265332e-01 9.20671523e-01 4.14555341e-01
5.71191549e-01 -1.28394866e+00 -3.39949608e-01 5.12528718e-01
2.26747155e-01 -4.91338909e-01 2.47990653e-01 5.86766422e-01
-4.88914460e-01 8.79983380e-02 -4.97168712e-02 -7.02451587e-01
-1.66405559e+00 5.57085693e-01 1.54139653e-01 4.74502534e-01
-9.52067196e-01 1.10944974e+00 1.23022461e+00 9.26764011e-02
-2.89488826e-02 -5.79504728e-01 -1.37223616e-01 -1.89519614e-01
6.37351751e-01 6.00141436e-02 2.36924171e-01 -1.01430166e+00
-2.90322661e-01 1.21759379e+00 -1.45311594e-01 1.15646966e-01
1.17372143e+00 -2.74939239e-01 -3.21184307e-01 9.50306416e-01
1.51798451e+00 6.24538124e-01 -1.23212898e+00 -2.56171733e-01
-4.23529238e-01 -7.55611002e-01 3.31905007e-01 -2.92768091e-01
-1.05640173e+00 8.98003519e-01 9.20884132e-01 5.25410891e-01
1.06513548e+00 3.39565486e-01 1.95838183e-01 3.36739659e-01
5.60505092e-01 -7.81788826e-01 1.48287699e-01 4.91671711e-02
1.01241148e+00 -8.03274393e-01 2.17814863e-01 -9.09156680e-01
-3.66028637e-01 1.14918315e+00 9.76564437e-02 -4.16318834e-01
9.95014727e-01 1.36415243e-01 3.26899618e-01 -6.30619347e-01
-7.18907893e-01 2.67783087e-03 9.87023830e-01 3.84492606e-01
2.11818159e-01 -3.59820426e-02 2.99307287e-01 -1.08051993e-01
-2.86841959e-01 -6.87515378e-01 3.58605564e-01 9.50558603e-01
-6.42007947e-01 -7.42243886e-01 -6.97278738e-01 4.93873268e-01
-5.22394061e-01 -2.32426021e-02 -5.84774017e-01 5.01457989e-01
2.84827858e-01 8.41499746e-01 3.48499924e-01 -3.15453649e-01
-4.36120434e-03 -7.89840743e-02 7.64915347e-01 -8.34686875e-01
-5.65522350e-03 7.29096532e-01 -1.59845635e-01 -7.30701327e-01
-6.21417820e-01 -1.03396153e+00 -6.49986088e-01 2.06281886e-01
-4.97080088e-01 3.00914466e-01 8.63080740e-01 6.90227330e-01
-9.76162404e-02 1.87879845e-01 6.42294943e-01 -1.16670346e+00
1.63014159e-01 -8.05783331e-01 -1.18849862e+00 4.03252304e-01
6.35612607e-01 -7.36651778e-01 -9.39790547e-01 6.43888175e-01] | [9.823302268981934, -2.84889817237854] |
de866465-8058-4953-871b-ede0ea08e4d7 | memorization-and-generalization-in-neural | 2106.08704 | null | https://arxiv.org/abs/2106.08704v3 | https://arxiv.org/pdf/2106.08704v3.pdf | Memorization and Generalization in Neural Code Intelligence Models | Deep Neural Networks (DNNs) are increasingly being used in software engineering and code intelligence tasks. These are powerful tools that are capable of learning highly generalizable patterns from large datasets through millions of parameters. At the same time, their large capacity can render them prone to memorizing data points. Recent work suggests that the memorization risk manifests especially strongly when the training dataset is noisy, involving many ambiguous or questionable samples, and memorization is the only recourse. The goal of this paper is to evaluate and compare the extent of memorization and generalization in neural code intelligence models. It aims to provide insights on how memorization may impact the learning behavior of neural models in code intelligence systems. To observe the extent of memorization in models, we add random noise to the original training dataset and use various metrics to quantify the impact of noise on various aspects of training and testing. We evaluate several state-of-the-art neural code intelligence models and benchmarks based on Java, Python, and Ruby codebases. Our results highlight important risks: millions of trainable parameters allow the neural networks to memorize anything, including noisy data, and provide a false sense of generalization. We observed all models manifest some forms of memorization. This can be potentially troublesome in most code intelligence tasks where they rely on rather noise-prone and repetitive data sources, such as code from GitHub. To the best of our knowledge, we provide the first study to quantify memorization effects in the domain of software engineering and code intelligence systems. This work raises awareness and provides new insights into important issues of training neural models in code intelligence systems that are usually overlooked by software engineering researchers. | ['Vincent J. Hellendoorn', 'Mohammad Amin Alipour', 'Aftab Hussain', 'Md Rafiqul Islam Rabin'] | 2021-06-16 | null | null | null | null | ['variable-misuse', 'code-documentation-generation', 'code-search', 'code-search', 'method-name-prediction', 'code-documentation-generation'] | ['computer-code', 'computer-code', 'computer-code', 'computer-vision', 'natural-language-processing', 'natural-language-processing'] | [-1.84898302e-01 -7.45259821e-02 1.34700313e-01 -3.11801940e-01
-9.72455963e-02 -6.78309619e-01 1.81529924e-01 3.51763040e-01
-4.68628556e-01 4.36193913e-01 -1.28037766e-01 -7.01908231e-01
-1.50983721e-01 -9.61445987e-01 -1.09096253e+00 -2.59250373e-01
-8.57550576e-02 2.23030478e-01 -9.66683328e-02 -3.24589431e-01
4.41236466e-01 2.06991449e-01 -1.79762006e+00 4.06696349e-01
1.13301921e+00 7.06454515e-01 1.76324829e-01 5.74971914e-01
-4.29174691e-01 1.21744943e+00 -1.06643367e+00 -3.93903613e-01
1.37447044e-01 1.25337780e-01 -9.07764852e-01 -4.68729138e-01
5.22824705e-01 -1.53357610e-01 8.83721858e-02 1.11541545e+00
4.42088276e-01 -1.13688987e-02 1.96379095e-01 -1.13038850e+00
-1.13385737e+00 9.38856006e-01 -3.06313038e-01 4.60720986e-01
2.82514989e-01 5.28940618e-01 4.87353683e-01 -8.39381158e-01
4.34363574e-01 9.54243839e-01 1.10777509e+00 5.76392233e-01
-1.21371973e+00 -7.88687348e-01 -7.31671751e-02 1.28632903e-01
-1.25636792e+00 -2.68138140e-01 5.31822920e-01 -6.56160176e-01
1.45152545e+00 2.20759273e-01 5.62750816e-01 1.10427952e+00
4.32939202e-01 4.36448544e-01 7.28527009e-01 -3.88188571e-01
3.69393557e-01 2.33030051e-01 5.71200550e-01 7.54608274e-01
6.45353496e-01 9.28866342e-02 -4.78357792e-01 -3.18674505e-01
2.77640194e-01 2.12028727e-01 -1.88147321e-01 4.14028354e-02
-8.02659392e-01 6.82573020e-01 5.28208852e-01 4.86023843e-01
-1.61375746e-01 3.09738398e-01 5.93886077e-01 6.20934725e-01
1.47727370e-01 1.04190481e+00 -6.47768080e-01 -6.35066569e-01
-8.03403199e-01 1.54207513e-01 1.16511440e+00 8.63434732e-01
1.00023413e+00 2.81242222e-01 2.20740840e-01 9.02764142e-01
-6.43930137e-02 2.44127586e-01 1.02422845e+00 -7.64131606e-01
5.14160275e-01 1.06881070e+00 -2.99556881e-01 -1.14376521e+00
-4.19316053e-01 -7.14325428e-01 -6.40731514e-01 3.44929993e-01
2.30486587e-01 -2.01753765e-01 -8.07261646e-01 1.65997183e+00
-4.58745569e-01 -1.01665273e-01 -3.37271169e-02 5.69813311e-01
9.15960371e-01 3.27921689e-01 2.49822870e-01 3.48985314e-01
9.77525294e-01 -7.86215603e-01 -2.70802379e-01 -6.68089449e-01
8.51035655e-01 -5.78619182e-01 1.60519147e+00 4.63612318e-01
-1.06145906e+00 -5.68403661e-01 -1.18866622e+00 7.87246451e-02
-6.62014842e-01 -1.55526891e-01 7.28280067e-01 8.40349078e-01
-1.13666058e+00 7.14919508e-01 -8.44640553e-01 -3.26638877e-01
4.23235536e-01 2.56732792e-01 -2.29067788e-01 -1.93336830e-01
-9.75668609e-01 1.00705349e+00 4.00350451e-01 -1.43755227e-01
-7.31848001e-01 -1.02181327e+00 -7.34084308e-01 3.22189271e-01
2.05662280e-01 -5.00793934e-01 1.32572889e+00 -1.25017738e+00
-8.13522756e-01 6.05406284e-01 8.06438625e-02 -4.44571286e-01
1.64041474e-01 -2.87869632e-01 -3.78803313e-01 -6.24362767e-01
-8.61307830e-02 2.19867736e-01 5.27581692e-01 -1.14774752e+00
-2.92824898e-02 -3.81251246e-01 8.76114517e-02 -3.30496281e-01
-6.63487315e-01 -1.81925818e-01 -1.60931364e-01 -5.49349844e-01
-1.39604285e-01 -7.25933015e-01 -3.33025306e-02 -4.09791023e-01
-1.72200307e-01 -9.45255235e-02 5.58268189e-01 -5.23669302e-01
1.50849760e+00 -2.16192245e+00 -2.40830168e-01 1.93418413e-01
4.86467749e-01 3.96159917e-01 -3.90128374e-01 2.19878808e-01
-5.50172664e-02 7.08731413e-01 -1.93251461e-01 -4.10527699e-02
-1.18296504e-01 4.07822192e-01 -2.46449232e-01 -6.54130653e-02
3.45523179e-01 1.06471777e+00 -7.47367620e-01 2.85617951e-02
-2.34801814e-01 3.81416649e-01 -8.60864818e-01 1.61717489e-01
-2.91763812e-01 -2.09259570e-01 -1.38077796e-01 6.37411416e-01
5.88032067e-01 -4.27950710e-01 -4.45176587e-02 2.10792899e-01
-8.97459220e-03 4.48624551e-01 -8.94037008e-01 1.48769915e+00
-7.73082793e-01 7.60921061e-01 -4.19139951e-01 -8.17562997e-01
9.31738257e-01 -3.30253802e-02 -2.26382524e-01 -8.22739244e-01
-6.75474778e-02 1.24258228e-01 3.06514353e-01 -7.21998096e-01
6.64514899e-01 1.33972332e-01 -6.41863197e-02 8.15036774e-01
1.48185387e-01 -1.56460300e-01 2.27708086e-01 6.17621392e-02
1.49423695e+00 -4.05443788e-01 8.35451111e-02 -3.60687315e-01
5.86455129e-02 -2.61098426e-02 5.56763351e-01 1.25754392e+00
1.41209379e-01 4.99560982e-01 6.26622021e-01 -8.98153305e-01
-9.47431803e-01 -8.70501399e-01 4.71470281e-02 1.33969128e+00
-3.48057300e-01 -5.70698678e-01 -7.31360853e-01 -4.83477235e-01
1.56613350e-01 7.94922411e-01 -8.26866090e-01 -6.22045457e-01
-4.29015964e-01 -8.99609506e-01 9.55222309e-01 7.20743597e-01
4.79737788e-01 -1.20765829e+00 -8.61328661e-01 3.32461625e-01
-3.57420859e-03 -6.52399898e-01 2.75909305e-02 5.25207579e-01
-1.08460546e+00 -1.14956164e+00 -1.00940719e-01 -5.49767315e-01
7.85503983e-01 3.53183411e-02 1.93699956e+00 9.71487880e-01
-4.84257549e-01 2.88346142e-01 -2.31914535e-01 -7.16681421e-01
-7.86531627e-01 3.03682715e-01 -5.47550879e-02 -8.09646308e-01
8.23367715e-01 -8.09387088e-01 -2.87901551e-01 2.13102534e-01
-1.14826238e+00 -2.69236982e-01 5.89968383e-01 8.83939564e-01
-1.45062253e-01 9.52463448e-02 5.46609759e-01 -1.13123715e+00
1.31640005e+00 -9.54148650e-01 -4.71456587e-01 4.26198900e-01
-9.02774274e-01 3.36492658e-01 8.79943728e-01 -6.12547219e-01
-8.37528467e-01 -4.03931290e-01 -1.75924495e-01 -4.55341451e-02
-2.17851907e-01 9.62141812e-01 2.32714355e-01 -3.52419406e-01
1.48492038e+00 1.95620477e-01 -1.50411110e-02 -6.01520300e-01
1.09904282e-01 6.90380633e-01 3.50380659e-01 -1.02652299e+00
4.67176497e-01 -4.80794385e-02 -7.13875353e-01 -5.44146359e-01
-3.72030020e-01 -6.10426851e-02 -1.78761765e-01 1.72367655e-02
1.78893819e-01 -6.87945724e-01 -7.19044924e-01 5.99726439e-01
-1.19162369e+00 -5.99162102e-01 -2.46368185e-01 -4.41109948e-02
-1.67369366e-01 1.85367569e-01 -7.95920551e-01 -4.01663750e-01
-4.82004941e-01 -1.36837304e+00 2.76755363e-01 2.82679886e-01
-6.30474985e-01 -9.51501310e-01 7.65544102e-02 2.22048745e-01
1.14692390e+00 3.89734004e-03 1.22248590e+00 -6.34743989e-01
-5.06379724e-01 -1.56207025e-01 -7.13370293e-02 5.34205675e-01
-5.05604781e-02 1.86005637e-01 -1.14702451e+00 -1.39212817e-01
7.08355457e-02 -5.36313415e-01 8.06279659e-01 -1.18199833e-01
1.42627013e+00 -5.21337867e-01 4.96863797e-02 5.93612373e-01
1.56991911e+00 2.26103783e-01 6.47830307e-01 5.15186429e-01
5.04427135e-01 2.21479788e-01 -3.62324603e-02 4.51786578e-01
3.50655168e-01 1.58720791e-01 5.09305239e-01 1.71377093e-01
2.77546495e-01 -6.15186282e-02 2.57713735e-01 8.42819631e-01
-8.38845968e-02 3.28846052e-02 -1.67636478e+00 6.29192889e-01
-1.51530790e+00 -7.07005143e-01 1.61333844e-01 2.05130315e+00
1.19407618e+00 1.90203995e-01 -3.20368201e-01 1.13824964e-01
2.42597789e-01 -3.35155666e-01 -6.18280292e-01 -8.22701633e-01
-2.73860879e-02 4.33880270e-01 2.61559874e-01 2.06382230e-01
-5.76705158e-01 6.90547884e-01 6.55555534e+00 4.39462185e-01
-1.31733263e+00 7.51349702e-02 6.31488264e-01 -2.78765678e-01
-4.48099196e-01 -1.75201029e-01 -4.31322128e-01 5.99814713e-01
1.07888949e+00 -2.54637629e-01 8.57700348e-01 1.27899134e+00
-1.62647277e-01 -2.72401303e-01 -1.20933735e+00 6.60329580e-01
1.09533034e-03 -1.48014939e+00 3.66322473e-02 -3.33599091e-01
7.19795048e-01 4.00062442e-01 1.91142499e-01 6.48207247e-01
5.32875955e-01 -1.50525463e+00 6.10001981e-01 5.30488610e-01
4.17179823e-01 -6.63187385e-01 8.99514019e-01 5.45886934e-01
-5.33942580e-01 -4.72236782e-01 -5.69914758e-01 -5.53966224e-01
-5.89603305e-01 8.50285828e-01 -8.80267382e-01 -1.39868051e-01
1.13047075e+00 3.73305380e-01 -1.24138737e+00 1.08582294e+00
-6.17630854e-02 7.58464992e-01 -3.13992202e-01 -2.32538700e-01
-7.08096400e-02 4.03371066e-01 1.30558252e-01 1.31383395e+00
3.19029123e-01 -3.00008833e-01 -1.81199640e-01 1.43564034e+00
-3.09865743e-01 -1.51037827e-01 -7.16613054e-01 -2.24342272e-01
6.19513690e-01 9.05638754e-01 -3.67249995e-01 -1.84270486e-01
-4.21927929e-01 4.38550234e-01 8.08468044e-01 4.71926719e-01
-4.54761982e-01 -4.08789694e-01 7.61938691e-01 7.67178386e-02
-2.28927433e-01 -2.02558190e-01 -8.56383383e-01 -1.04075432e+00
3.93655568e-01 -1.30659890e+00 2.64830172e-01 -6.39362693e-01
-1.21757150e+00 5.91707170e-01 -2.93588668e-01 -6.16259754e-01
-2.68070281e-01 -5.38646698e-01 -8.69896710e-01 8.47871065e-01
-1.21657968e+00 -5.55995166e-01 -6.67689025e-01 2.38378167e-01
2.44774163e-01 -3.95530820e-01 9.57791388e-01 3.25549752e-01
-4.55330998e-01 9.65330422e-01 1.20767958e-01 3.05079758e-01
4.74972188e-01 -1.09866583e+00 7.18154132e-01 7.24436581e-01
-7.01284632e-02 1.49846268e+00 7.16914594e-01 -5.14531910e-01
-1.41519356e+00 -8.92096460e-01 6.55887604e-01 -7.66265690e-01
6.28832102e-01 -2.76912659e-01 -1.46756673e+00 7.24128187e-01
-5.74570931e-02 -4.03396338e-02 8.06858480e-01 4.69824284e-01
-7.29064643e-01 -2.82863714e-02 -1.06812024e+00 4.95058686e-01
9.06979680e-01 -6.71691060e-01 -6.50710106e-01 1.93927754e-02
6.91794634e-01 -3.98346573e-01 -9.54733133e-01 2.79675096e-01
4.96280968e-01 -1.28796136e+00 8.33658099e-01 -7.21662939e-01
8.49035084e-01 -3.52845713e-02 1.92417726e-01 -1.43065774e+00
-1.70768231e-01 -2.52728641e-01 -1.24299094e-01 1.08210039e+00
6.90382481e-01 -7.10660100e-01 7.28530765e-01 1.07937229e+00
3.97343002e-02 -7.40343750e-01 -6.60061181e-01 -7.55929530e-01
4.69629765e-01 -7.22746789e-01 9.47545826e-01 1.18697274e+00
2.36178294e-01 -9.20264870e-02 5.88848218e-02 -1.44012466e-01
2.56411374e-01 -1.60723954e-01 7.29443491e-01 -1.22184908e+00
-4.57259983e-01 -5.21596849e-01 -4.82160568e-01 -5.41623950e-01
2.38708019e-01 -7.93010294e-01 -3.90417292e-03 -1.09720218e+00
2.45834708e-01 -6.36990488e-01 -3.21211547e-01 8.90429020e-01
-2.13800654e-01 -3.02919373e-02 2.21309379e-01 1.60850897e-01
-4.74023640e-01 5.45030236e-02 6.90491259e-01 -2.95749396e-01
-1.83684900e-01 -6.04908653e-02 -1.12834752e+00 6.57277346e-01
1.13326621e+00 -6.13960326e-01 -3.85369778e-01 -9.82587218e-01
8.95433843e-01 -4.00664836e-01 4.04688030e-01 -1.21053743e+00
4.94314253e-01 5.39552271e-02 3.38504910e-01 9.83690545e-02
-2.13423908e-01 -7.44258106e-01 2.05047831e-01 4.90275592e-01
-3.25832844e-01 5.63274801e-01 7.38772988e-01 2.39063114e-01
-1.42186135e-01 -5.39644241e-01 6.10336781e-01 -5.09285569e-01
-8.64681900e-01 -1.61765620e-01 -6.06024384e-01 4.43759263e-01
6.41714752e-01 -1.66253507e-01 -7.72495806e-01 -2.05418125e-01
-3.50770503e-01 3.29909381e-03 7.61775613e-01 7.90063441e-01
6.25363111e-01 -9.57899690e-01 -3.58859777e-01 5.36455095e-01
3.33673954e-01 -3.91409397e-02 1.31498445e-02 3.80842805e-01
-7.10172951e-01 1.48474976e-01 -4.10429597e-01 -4.78235722e-01
-1.00667286e+00 4.90713090e-01 5.75739443e-01 -8.47875401e-02
-3.50275576e-01 8.92122626e-01 -2.09458262e-01 -8.08296382e-01
3.68854612e-01 -5.82231581e-01 1.34116970e-02 -2.68118739e-01
6.75652802e-01 3.40865374e-01 5.02062678e-01 1.13205068e-01
-3.18294406e-01 2.39910007e-01 -2.96032786e-01 4.66124177e-01
1.41447091e+00 4.04735625e-01 -4.87242430e-01 3.99743319e-01
1.05138195e+00 -2.60099143e-01 -8.02906275e-01 -8.59391093e-02
3.00010473e-01 -2.65941560e-01 -2.16493234e-01 -1.13913643e+00
-1.08374822e+00 1.01859927e+00 5.32886744e-01 3.50850821e-01
1.00283837e+00 -2.58674294e-01 4.52950507e-01 9.13747013e-01
3.60934079e-01 -1.10153937e+00 1.49142757e-01 9.44393635e-01
8.05348396e-01 -1.24739516e+00 -2.55871296e-01 1.40174732e-01
-2.13591173e-01 1.24202764e+00 1.06250417e+00 6.61389995e-03
5.72538376e-01 8.70442092e-01 3.16824794e-01 -3.10480475e-01
-9.00607765e-01 4.62014139e-01 4.87037515e-03 5.93331695e-01
7.65016258e-01 -6.62597641e-02 -3.87242846e-02 8.72112334e-01
-4.68869299e-01 3.35806012e-01 7.90488958e-01 1.10707188e+00
-3.61101151e-01 -8.45920026e-01 -3.44785750e-01 6.77210927e-01
-5.40358543e-01 -5.65711021e-01 -4.06164348e-01 5.78520119e-01
8.32930431e-02 8.58647823e-01 3.77081260e-02 -6.44418895e-01
2.94884115e-01 3.52060199e-01 1.01678550e-01 -6.65345073e-01
-1.26800203e+00 -1.13917470e+00 -1.71335548e-01 -5.73611975e-01
1.57907009e-01 -8.08145031e-02 -1.18961465e+00 -7.77891695e-01
-2.49397710e-01 1.66556433e-01 7.18905151e-01 7.61898160e-01
7.24259675e-01 6.38731360e-01 -8.42427984e-02 -6.27792478e-01
-7.16928661e-01 -9.08722401e-01 -1.33641586e-01 6.61602914e-01
3.13139498e-01 -3.49389583e-01 -5.93553483e-01 -8.46148208e-02] | [7.689231872558594, 7.724452018737793] |
2f2b79ea-4471-4429-a922-7a1d19a0ad85 | a-unified-framework-for-task-driven-data | 2106.05484 | null | https://arxiv.org/abs/2106.05484v1 | https://arxiv.org/pdf/2106.05484v1.pdf | A Unified Framework for Task-Driven Data Quality Management | High-quality data is critical to train performant Machine Learning (ML) models, highlighting the importance of Data Quality Management (DQM). Existing DQM schemes often cannot satisfactorily improve ML performance because, by design, they are oblivious to downstream ML tasks. Besides, they cannot handle various data quality issues (especially those caused by adversarial attacks) and have limited applications to only certain types of ML models. Recently, data valuation approaches (e.g., based on the Shapley value) have been leveraged to perform DQM; yet, empirical studies have observed that their performance varies considerably based on the underlying data and training process. In this paper, we propose a task-driven, multi-purpose, model-agnostic DQM framework, DataSifter, which is optimized towards a given downstream ML task, capable of effectively removing data points with various defects, and applicable to diverse models. Specifically, we formulate DQM as an optimization problem and devise a scalable algorithm to solve it. Furthermore, we propose a theoretical framework for comparing the worst-case performance of different DQM strategies. Remarkably, our results show that the popular strategy based on the Shapley value may end up choosing the worst data subset in certain practical scenarios. Our evaluation shows that DataSifter achieves and most often significantly improves the state-of-the-art performance over a wide range of DQM tasks, including backdoor, poison, noisy/mislabel data detection, data summarization, and data debiasing. | ['Ruoxi Jia', 'Ming Jin', 'Yi Zeng', 'Tianhao Wang'] | 2021-06-10 | null | null | null | null | ['data-summarization'] | ['miscellaneous'] | [ 1.96569756e-01 -2.26837873e-01 -4.56751674e-01 -1.48845568e-01
-1.43889213e+00 -8.64108980e-01 5.16185880e-01 4.88609850e-01
-3.48055869e-01 5.23231089e-01 1.28324568e-01 -4.61585402e-01
-3.62431973e-01 -6.85152709e-01 -1.00211537e+00 -8.75274479e-01
2.45663568e-01 3.15001220e-01 -1.99069932e-01 -1.83402315e-01
5.70623755e-01 3.72376084e-01 -1.22076476e+00 1.73923060e-01
1.16951120e+00 9.88553762e-01 -2.25967556e-01 4.51885104e-01
1.15822181e-01 7.41509855e-01 -1.02743435e+00 -1.00512493e+00
4.86030132e-01 -7.51825124e-02 -7.56484866e-01 -5.59648946e-02
6.23633742e-01 -2.72602797e-01 -3.02651674e-01 1.10204601e+00
7.13502765e-01 -2.05891266e-01 5.90807438e-01 -1.86892247e+00
-7.81220019e-01 7.71051407e-01 -4.73195910e-01 1.25904992e-01
-1.59722731e-01 5.63359141e-01 1.26715291e+00 -7.89948404e-01
4.34659123e-01 1.19034362e+00 6.44872427e-01 6.62927687e-01
-1.60861897e+00 -7.52123833e-01 -2.65456699e-02 3.99852023e-02
-1.13300216e+00 -4.71923321e-01 6.31656885e-01 -2.20708609e-01
4.51003194e-01 4.89470899e-01 -1.20456824e-02 1.28690422e+00
3.42289329e-01 1.18294358e+00 1.15941107e+00 -1.23454845e-02
6.01605475e-01 9.43762735e-02 1.79821774e-01 3.98724079e-01
5.45940518e-01 2.19738081e-01 -8.77327800e-01 -6.17790282e-01
1.93629395e-02 2.49716975e-02 -2.16112003e-01 -4.65015441e-01
-8.74652684e-01 8.97171319e-01 3.08209240e-01 -5.61125726e-02
-6.93048164e-02 2.31944770e-01 6.67571843e-01 5.29290676e-01
4.59153414e-01 7.41222918e-01 -7.01571643e-01 7.34304115e-02
-8.17322254e-01 4.84863430e-01 6.16261959e-01 9.12747562e-01
5.17788231e-01 -2.42506087e-01 -5.53565443e-01 3.97794813e-01
1.62081674e-01 4.82538640e-01 2.40508839e-01 -1.03623235e+00
9.75143313e-01 6.80120170e-01 -8.60136971e-02 -8.36998343e-01
-2.74509221e-01 -5.08501053e-01 -1.08844388e+00 1.72450677e-01
3.87409300e-01 8.54656249e-02 -6.48380995e-01 1.83346808e+00
4.49504822e-01 8.89316387e-03 2.86945522e-01 6.00503027e-01
5.39329410e-01 3.75164360e-01 2.48672232e-01 -2.68484503e-01
1.18746209e+00 -5.87652624e-01 -6.02587521e-01 -6.86357096e-02
9.91330445e-01 -4.32178169e-01 1.45762110e+00 7.60918379e-01
-9.90459025e-01 -2.37470865e-01 -9.98020470e-01 -2.31220767e-01
-2.98177212e-01 -4.24772203e-01 4.07371014e-01 9.11021590e-01
-8.42341363e-01 7.59736180e-01 -5.57111502e-01 8.67563859e-02
1.04432428e+00 2.53416210e-01 -3.74939054e-01 -2.86214113e-01
-1.10442936e+00 6.93540394e-01 2.81593561e-01 -2.48558298e-01
-1.38588059e+00 -1.18967462e+00 -4.56743002e-01 8.18985794e-03
6.62848115e-01 -9.39057112e-01 1.11629260e+00 -4.19280261e-01
-9.49813724e-01 8.26380372e-01 1.48518637e-01 -7.52802372e-01
7.87336588e-01 -4.22221720e-01 -2.54062146e-01 -1.78773895e-01
7.57129723e-03 2.45177656e-01 1.05450559e+00 -1.45590603e+00
-7.31674969e-01 -5.39653957e-01 1.30574092e-01 -6.42536730e-02
-6.14279091e-01 -1.07563563e-01 -2.65962452e-01 -7.74228811e-01
-3.18813235e-01 -7.13836610e-01 -3.40395778e-01 2.10494772e-02
-8.71418417e-01 -2.40790829e-01 7.67680228e-01 -2.74569571e-01
1.50392950e+00 -2.41489816e+00 5.63780069e-02 -6.15613163e-03
5.65736890e-01 4.86104369e-01 -2.73772836e-01 4.91572708e-01
3.55028987e-01 6.66244090e-01 -4.87282008e-01 -6.59041107e-01
1.69209123e-01 2.42415518e-01 -5.31791031e-01 5.64048231e-01
2.90118277e-01 1.03755188e+00 -7.98418641e-01 -3.72007400e-01
-4.65635136e-02 -6.63792994e-03 -7.37282991e-01 3.37734371e-01
-3.96410227e-01 3.36788982e-01 -3.76855850e-01 7.49718726e-01
9.21766400e-01 -2.70358294e-01 5.48836626e-02 -8.53176937e-02
2.30475366e-01 1.02783002e-01 -1.15302217e+00 1.62622881e+00
-4.11708295e-01 2.06405461e-01 8.51392224e-02 -7.68444121e-01
7.06547976e-01 7.43522421e-02 3.33284169e-01 -6.72824264e-01
1.54181374e-02 2.52541810e-01 -2.83696830e-01 -3.82954717e-01
5.09784818e-01 -1.20478058e-02 -5.68131268e-01 4.73617941e-01
-1.61555797e-01 3.33764218e-02 -5.52493446e-02 5.11331022e-01
1.51067412e+00 -4.86137897e-01 3.61564159e-01 -1.59845605e-01
2.47259319e-01 4.08672616e-02 7.43898332e-01 1.10618222e+00
-3.34702313e-01 4.77576852e-01 7.29337811e-01 -1.22644931e-01
-9.24759388e-01 -1.15825355e+00 -2.35508278e-01 1.17526054e+00
2.88121790e-01 -5.78271866e-01 -7.62080252e-01 -1.10839379e+00
5.22942901e-01 8.59100521e-01 -3.71633291e-01 -7.55327523e-01
-4.31144357e-01 -1.07032967e+00 8.86193871e-01 2.54384428e-01
3.88064235e-01 -7.88173914e-01 -2.98924565e-01 2.22951829e-01
-1.06022887e-01 -8.80677640e-01 -4.92583543e-01 3.07761788e-01
-9.25100386e-01 -1.19142461e+00 -8.25430229e-02 -3.98373604e-02
4.41974849e-01 3.69971633e-01 1.25304234e+00 1.09831758e-01
-3.52837816e-02 1.22427419e-01 -3.14235508e-01 -4.86549020e-01
-8.46324682e-01 2.74490237e-01 3.55160423e-02 1.09669305e-02
2.56432533e-01 -2.74852514e-01 -5.40964425e-01 1.96465373e-01
-1.40962994e+00 -4.82124865e-01 7.47749209e-01 7.76407242e-01
6.85248613e-01 2.04488665e-01 9.64880764e-01 -1.38333011e+00
8.05045187e-01 -6.58413231e-01 -5.12582064e-01 4.46115077e-01
-1.17477059e+00 2.78924167e-01 9.43949342e-01 -3.52891028e-01
-6.76314712e-01 -3.24590534e-01 -1.41041413e-01 -5.13128042e-01
-1.03002414e-01 2.03219414e-01 -8.07026803e-01 3.24535593e-02
8.27199161e-01 1.43978417e-01 -4.51823324e-02 -6.91735208e-01
7.25966990e-01 5.61948538e-01 6.02889895e-01 -5.75123250e-01
1.08592129e+00 4.85738546e-01 2.48116374e-01 4.45743836e-03
-1.17535377e+00 -2.91652620e-01 -2.42560476e-01 2.22679064e-01
3.93443048e-01 -8.38368058e-01 -9.55981553e-01 7.07498074e-01
-8.51968288e-01 -8.91143829e-02 -4.72759664e-01 -1.94388315e-01
-4.82592136e-01 4.27716762e-01 -5.64461648e-01 -7.13875353e-01
-6.49721205e-01 -1.16359329e+00 1.15908909e+00 -2.64175832e-01
1.07377455e-01 -8.34220052e-01 -1.83728918e-01 7.30660319e-01
3.80036533e-01 4.03615147e-01 1.34089255e+00 -8.57925117e-01
-7.60392725e-01 -8.20157230e-02 -4.69622319e-04 5.89469552e-01
-1.00594558e-01 -5.57224452e-02 -1.15326440e+00 -5.81218481e-01
1.55666456e-01 -5.58932602e-01 1.05939507e+00 1.95893168e-01
1.52860463e+00 -7.62617290e-01 -1.60012379e-01 6.85563922e-01
1.49009347e+00 -2.33730063e-01 5.70675433e-01 3.26512456e-01
8.09309781e-01 4.25316989e-01 9.22099769e-01 7.36939430e-01
3.52232069e-01 4.93261337e-01 9.42572534e-01 3.84777901e-03
1.64401144e-01 -5.14378726e-01 5.02026677e-01 2.80764312e-01
7.23645627e-01 -6.32954121e-01 -5.95355034e-01 3.26769769e-01
-1.81136358e+00 -7.31574535e-01 -3.26911837e-01 2.45071411e+00
1.08880758e+00 3.39266479e-01 1.60362244e-01 4.56013143e-01
4.05024529e-01 1.94295451e-01 -9.82665598e-01 -2.63520360e-01
-2.69019216e-01 -3.22426483e-02 8.54204357e-01 -6.72139451e-02
-1.12348461e+00 8.86216462e-01 5.98469734e+00 1.30391300e+00
-8.71974826e-01 3.57323289e-01 8.28384876e-01 -2.42935523e-01
-5.91319323e-01 -2.37522833e-02 -9.06548679e-01 6.45758629e-01
8.84294093e-01 -4.32273120e-01 3.55255485e-01 8.95030141e-01
1.84155688e-01 1.29322693e-01 -1.37817955e+00 9.39827800e-01
-2.49733239e-01 -1.40875387e+00 9.20670554e-02 3.22916329e-01
8.51792872e-01 -5.74563183e-02 4.84195441e-01 4.11282063e-01
4.96721685e-01 -1.09019172e+00 6.56473517e-01 2.81136900e-01
6.61631405e-01 -9.45108533e-01 6.22880936e-01 6.84583187e-01
-5.34767568e-01 -4.54630852e-01 -3.46569061e-01 4.25000161e-01
-3.45636755e-02 1.09634411e+00 -5.96179307e-01 7.67117321e-01
6.30513072e-01 3.94363254e-01 -8.77381742e-01 8.78770173e-01
3.52003053e-02 7.77340233e-01 -9.68741104e-02 3.87582064e-01
-9.91341192e-03 1.33418158e-01 5.63244820e-01 8.01994681e-01
1.06808618e-01 -4.36291307e-01 -1.53133161e-02 7.73862898e-01
-7.52004981e-01 6.75014034e-02 -7.15063155e-01 1.58425897e-01
9.55437541e-01 1.02055478e+00 1.25410333e-01 -1.33412704e-01
-1.71193525e-01 6.82235897e-01 3.93776834e-01 1.40061006e-02
-7.19523251e-01 3.83470654e-02 1.01696146e+00 9.73731354e-02
-1.45068929e-01 1.99042112e-01 -6.72145724e-01 -1.01124132e+00
2.75609046e-01 -1.34135640e+00 8.58971357e-01 -7.36795589e-02
-1.82469523e+00 1.68248460e-01 -2.20031694e-01 -1.27561867e+00
1.42335191e-01 -2.65227348e-01 -4.71196532e-01 5.02374411e-01
-1.65813112e+00 -8.66499066e-01 -5.91383427e-02 6.40781105e-01
1.97474480e-01 -9.13503990e-02 5.12268960e-01 3.70815694e-01
-7.53866851e-01 1.09597456e+00 4.52649355e-01 -2.20022440e-01
9.45475519e-01 -1.42341566e+00 5.40806115e-01 9.74964738e-01
1.16881564e-01 4.33408618e-01 6.44556463e-01 -4.95781720e-01
-1.85285997e+00 -1.60564256e+00 6.96498156e-01 -7.82964647e-01
4.30519253e-01 -5.21096230e-01 -1.15119493e+00 2.89261669e-01
-3.46114486e-01 1.68890327e-01 6.09317482e-01 8.90023410e-02
-6.84854507e-01 -5.71521938e-01 -1.57613587e+00 5.02602935e-01
1.23967791e+00 -5.05531967e-01 -2.21284285e-01 4.45389330e-01
1.01496863e+00 -2.20034376e-01 -1.00052869e+00 5.40501118e-01
-5.63960001e-02 -1.01290989e+00 9.63483632e-01 -9.33297753e-01
3.83801848e-01 -1.58569723e-01 -2.77910978e-01 -1.25609565e+00
-1.53980374e-01 -8.22672188e-01 -4.02489930e-01 1.65901113e+00
3.24687392e-01 -5.95471799e-01 7.99629033e-01 7.71411598e-01
1.36297976e-03 -7.55612850e-01 -1.13125610e+00 -1.02152765e+00
3.72445077e-01 -6.12486780e-01 9.15543675e-01 9.33840036e-01
-4.89530444e-01 1.69142157e-01 -6.59310222e-01 3.39459389e-01
8.79790246e-01 8.33411142e-02 1.16988134e+00 -1.26320720e+00
-2.80520201e-01 -5.49603641e-01 -1.78153127e-01 -8.99435937e-01
1.50633246e-01 -1.16319859e+00 -1.46925807e-01 -1.17548728e+00
2.53840148e-01 -6.77996397e-01 -4.37392980e-01 5.22704422e-01
-5.77339292e-01 -7.70015642e-02 3.28185946e-01 4.36018527e-01
-6.60965085e-01 5.31981230e-01 1.10171509e+00 -1.67206511e-01
-3.32872234e-02 2.48188630e-01 -1.30781472e+00 1.75789192e-01
7.65413404e-01 -1.01026714e+00 -3.19085062e-01 -3.06548864e-01
4.60053086e-01 -5.07050194e-02 6.38391197e-01 -6.66304886e-01
7.33873174e-02 -3.60066384e-01 -1.54628113e-01 -4.93932515e-01
-1.02290563e-01 -8.63202691e-01 6.64312690e-02 5.41965127e-01
-6.93559051e-01 -1.16782798e-03 -1.76944420e-01 9.18635011e-01
-7.03317001e-02 -1.88568085e-01 9.66954172e-01 1.03702545e-01
-4.24863100e-01 6.38083696e-01 1.90010145e-01 5.37625253e-01
1.17729330e+00 3.48828763e-01 -9.45430040e-01 3.40651534e-02
-1.52200982e-01 4.83573973e-01 6.14463210e-01 3.42507005e-01
4.07603025e-01 -1.27550232e+00 -9.41539943e-01 1.02655433e-01
4.95201170e-01 2.74053395e-01 2.66504586e-01 6.75093830e-01
6.50235489e-02 5.03539369e-02 1.63167641e-01 -5.25596261e-01
-1.12188745e+00 1.03523362e+00 1.11694731e-01 -6.59383893e-01
-4.75586683e-01 7.51181841e-01 2.66731113e-01 -5.02620101e-01
3.89640808e-01 -2.62943953e-01 4.05646265e-01 2.24869940e-02
4.45818692e-01 5.13996840e-01 4.31685567e-01 -1.96969882e-01
-2.87292510e-01 6.58639800e-03 -1.06385246e-01 3.67024660e-01
1.08863568e+00 5.06700128e-02 8.95117372e-02 2.09072575e-01
1.40105128e+00 1.47311702e-01 -1.36929572e+00 -5.10477245e-01
2.56501317e-01 -6.65155470e-01 -6.91032931e-02 -8.69889379e-01
-1.12090015e+00 8.75791907e-01 3.78643900e-01 3.63310546e-01
1.41506267e+00 4.59526703e-02 1.03025651e+00 2.19864354e-01
4.90712613e-01 -1.04027331e+00 2.28134841e-01 3.24611925e-02
6.97317243e-01 -1.39798737e+00 -1.06633238e-01 -1.58828512e-01
-9.44188774e-01 5.69016516e-01 4.66565818e-01 1.91286638e-01
3.20291758e-01 3.14382434e-01 -1.86042171e-02 -9.28785577e-02
-1.20799887e+00 1.30173564e-01 7.36227185e-02 5.39236009e-01
-2.76457995e-01 1.82779908e-01 -1.54211149e-01 6.13780677e-01
-3.58982123e-02 -1.65462628e-01 4.52072650e-01 8.17243636e-01
-2.33922005e-01 -1.30399942e+00 -4.08101320e-01 7.03947186e-01
-7.24741936e-01 -1.34050652e-01 -3.19969147e-01 6.80628896e-01
1.37923285e-01 1.16610932e+00 -3.56075555e-01 -5.47727764e-01
5.78050554e-01 -5.21392450e-02 -1.19810123e-02 -3.46616685e-01
-1.03096020e+00 -3.57213974e-01 -1.61485493e-01 -6.12762749e-01
-3.16784158e-02 -5.94094753e-01 -1.01516294e+00 -7.70909250e-01
-2.57860750e-01 -1.61527231e-01 6.16025269e-01 9.04430687e-01
7.92092144e-01 1.22380875e-01 1.15438998e+00 -4.21724059e-02
-1.49495769e+00 -5.97973764e-01 -6.19186163e-01 8.37571561e-01
4.26682711e-01 -3.75488192e-01 -5.55490434e-01 -2.86004931e-01] | [8.901933670043945, 4.205723762512207] |
36f590d7-bec7-4b43-8a13-c38ddc9830f7 | local-stochastic-factored-gradient-descent | 2203.11579 | null | https://arxiv.org/abs/2203.11579v2 | https://arxiv.org/pdf/2203.11579v2.pdf | Local Stochastic Factored Gradient Descent for Distributed Quantum State Tomography | We propose a distributed Quantum State Tomography (QST) protocol, named Local Stochastic Factored Gradient Descent (Local SFGD), to learn the low-rank factor of a density matrix over a set of local machines. QST is the canonical procedure to characterize the state of a quantum system, which we formulate as a stochastic nonconvex smooth optimization problem. Physically, the estimation of a low-rank density matrix helps characterizing the amount of noise introduced by quantum computation. Theoretically, we prove the local convergence of Local SFGD for a general class of restricted strongly convex/smooth loss functions, i.e., Local SFGD converges locally to a small neighborhood of the global optimum at a linear rate with a constant step size, while it locally converges exactly at a sub-linear rate with diminishing step sizes. With a proper initialization, local convergence results imply global convergence. We validate our theoretical findings with numerical simulations of QST on the Greenberger-Horne-Zeilinger (GHZ) state. | ['Anastasios Kyrillidis', 'César A. Uribe', 'Mohammad Taha Toghani', 'Junhyung Lyle Kim'] | 2022-03-22 | null | null | null | null | ['quantum-state-tomography'] | ['medical'] | [-2.35299632e-01 2.56733373e-02 -3.30646435e-05 -3.37185562e-01
-1.31534088e+00 -2.19095722e-01 3.43186706e-01 -1.99667603e-01
-6.27748966e-01 9.09803927e-01 1.79203972e-02 -3.15134674e-01
-3.30879927e-01 -7.84839451e-01 -9.16079283e-01 -1.22809875e+00
-2.64662027e-01 7.50912368e-01 -3.83675486e-01 -1.10593721e-01
1.62238374e-01 3.62122715e-01 -6.13340795e-01 -3.38446736e-01
8.94996107e-01 6.64505184e-01 2.81180330e-02 8.02177966e-01
4.21655118e-01 8.02674234e-01 -1.79733962e-01 -2.34490842e-01
3.89750063e-01 -7.69649088e-01 -1.12026083e+00 -1.93414629e-01
5.18501997e-02 -2.31096029e-01 -7.91885316e-01 1.85497808e+00
3.56964946e-01 2.53264427e-01 4.23240930e-01 -7.46172309e-01
-1.67400628e-01 6.91338122e-01 -2.44831711e-01 1.96335748e-01
5.61342910e-02 1.50745735e-01 1.20547950e+00 -7.34189630e-01
6.74342215e-01 1.09537113e+00 4.31194395e-01 3.90571415e-01
-1.67020655e+00 -6.13988459e-01 -4.75798398e-01 3.95899564e-02
-1.83567131e+00 -3.04459006e-01 5.46650529e-01 -1.55699715e-01
5.98774910e-01 1.11319393e-01 3.64849657e-01 3.97329420e-01
3.87195021e-01 5.06381333e-01 1.49742138e+00 -4.14845049e-01
5.55848777e-01 1.04291015e-03 1.60234287e-01 1.21278286e+00
1.81589782e-01 1.49732426e-01 -6.29989445e-01 -6.04334295e-01
6.79051697e-01 -3.37270290e-01 -3.38607550e-01 -2.93990701e-01
-1.28366148e+00 9.67833877e-01 5.90362370e-01 3.61140817e-01
-4.76628184e-01 4.65641350e-01 2.73165442e-02 6.36704087e-01
3.67949694e-01 2.72051871e-01 -1.30924135e-01 -2.57929593e-01
-8.46570134e-01 5.92208616e-02 1.09631813e+00 6.19818389e-01
1.42843914e+00 -1.23603009e-01 -1.87328786e-01 1.35165662e-01
3.52499932e-01 1.15232670e+00 -3.81165802e-01 -8.96434188e-01
5.16159713e-01 -2.05429509e-01 4.40320998e-01 -6.35055065e-01
-3.58164340e-01 -5.24936199e-01 -1.21027589e+00 -1.89841047e-01
6.27160192e-01 -6.51086032e-01 -5.24447501e-01 2.10076928e+00
3.78156602e-01 2.30399877e-01 -4.55653779e-02 1.17557847e+00
9.65928733e-02 1.05723000e+00 -3.33845079e-01 -4.57445323e-01
8.41953099e-01 -4.27208543e-01 -6.44842863e-01 -1.79379031e-01
8.57357204e-01 -5.02309203e-01 6.96332395e-01 2.72570699e-01
-8.91235828e-01 2.00424399e-02 -9.07663345e-01 1.29710510e-01
3.31293970e-01 -6.91368580e-02 6.81756794e-01 9.45017159e-01
-1.07834804e+00 1.02293062e+00 -1.09011221e+00 5.50288297e-02
1.78414375e-01 5.12579143e-01 -4.12966222e-01 -2.98913866e-01
-1.01963913e+00 6.47774100e-01 3.40203911e-01 3.04413855e-01
-1.11171353e+00 -6.63840652e-01 -4.51637805e-01 -2.72744503e-02
3.37040275e-01 -7.53629923e-01 8.82171571e-01 -4.61194962e-01
-1.92761219e+00 5.92419446e-01 -4.94721293e-01 -4.85040128e-01
1.62137821e-01 1.06110781e-01 -8.28633830e-02 7.82460794e-02
1.32645160e-01 -2.55537391e-01 6.74868226e-01 -7.94089913e-01
-9.99674946e-02 -6.07027531e-01 1.26124769e-01 4.53585163e-02
2.11196110e-01 -1.09030433e-01 2.84140352e-02 3.35090488e-01
5.12890160e-01 -1.21719921e+00 -5.87622523e-01 -5.33523083e-01
-4.63493943e-01 7.98596591e-02 2.86880165e-01 -4.11292553e-01
1.00545144e+00 -1.90293169e+00 4.46320683e-01 6.40353143e-01
3.51761729e-01 -2.23808780e-01 -7.42537668e-03 4.89990830e-01
2.56661922e-01 -3.57340798e-02 -2.96352834e-01 -4.56275135e-01
1.57710269e-01 1.61933824e-01 -4.92805243e-02 1.27606988e+00
-1.15873054e-01 8.78577650e-01 -1.13910866e+00 -3.98910716e-02
2.17724666e-02 1.03844181e-01 -8.39816868e-01 1.35784686e-01
2.20685378e-01 7.75961161e-01 -7.64098823e-01 3.15782517e-01
1.05786467e+00 -6.85245097e-01 1.86167091e-01 -1.90574035e-01
-1.41652301e-01 4.23352540e-01 -1.36886549e+00 1.81303060e+00
-5.37111461e-01 2.64266402e-01 6.49061084e-01 -1.24678433e+00
5.84534168e-01 1.54419348e-01 4.49434876e-01 -4.90796149e-01
4.01718885e-01 5.56774318e-01 2.33057942e-02 -7.73544088e-02
2.93154061e-01 -8.81374776e-01 -3.15006852e-01 6.56399131e-01
4.05750841e-01 -1.39170066e-01 -9.48733538e-02 4.81558293e-01
1.24620342e+00 -4.86385912e-01 2.05668300e-01 -8.34613740e-01
4.90405858e-01 -4.45226550e-01 5.66234946e-01 1.22027099e+00
-2.60883301e-01 1.91209853e-01 5.89671433e-01 -1.98392317e-01
-1.22767830e+00 -1.11550593e+00 -3.64962369e-01 7.35971153e-01
5.41145384e-01 -5.86614728e-01 -7.59890497e-01 -2.94015408e-01
-2.59986222e-01 4.49410319e-01 -4.01547849e-01 -2.08373755e-01
-3.15741211e-01 -1.38054276e+00 2.79095560e-01 -3.47446978e-01
9.05218840e-01 -5.82918406e-01 1.80197418e-01 9.18987840e-02
-4.58628573e-02 -1.16264069e+00 -5.13949394e-01 3.97657573e-01
-7.09587038e-01 -6.09521747e-01 -5.80911160e-01 -3.44150275e-01
7.23045826e-01 5.19372104e-03 8.58866155e-01 -4.45482612e-01
2.23304734e-01 2.25492641e-01 1.29950583e-01 3.05476606e-01
-5.04548013e-01 1.41897321e-01 3.50346446e-01 3.13494325e-01
5.84309734e-02 -5.03937244e-01 -6.28786743e-01 1.66836545e-01
-6.82293713e-01 -2.65890770e-02 3.81549269e-01 1.06449330e+00
6.41466200e-01 2.43308410e-01 -6.21658983e-03 -9.64458287e-01
5.84484041e-01 -5.07588148e-01 -1.08280778e+00 1.80785567e-01
-4.28118765e-01 7.03686714e-01 7.04834640e-01 -5.30669373e-03
-9.40942049e-01 8.62246566e-03 2.69209556e-02 -2.56497532e-01
3.16385955e-01 7.35717297e-01 4.10321131e-02 -6.73366427e-01
6.85619473e-01 4.13469702e-01 -1.25736848e-01 -2.58960694e-01
6.18269145e-01 3.58654678e-01 3.29758674e-01 -1.03112638e+00
1.27298260e+00 7.65424907e-01 5.36594033e-01 -7.98523545e-01
-1.05984938e+00 -5.56368232e-01 -4.34450209e-01 2.48304568e-02
5.25722623e-01 -8.00504029e-01 -1.16767490e+00 4.31446552e-01
-7.87167549e-01 -4.86847669e-01 -2.72802889e-01 9.12664056e-01
-5.75448871e-01 4.77302402e-01 -9.01834607e-01 -1.22199547e+00
-3.92703533e-01 -8.99242997e-01 1.01064265e+00 2.37007976e-01
5.15727639e-01 -1.03337705e+00 3.66084725e-01 2.69037068e-01
4.66981113e-01 -3.97481918e-02 6.94354117e-01 -5.66708297e-02
-9.95490670e-01 -2.43285298e-01 -1.30355880e-01 3.94806355e-01
-2.71065444e-01 -5.73937416e-01 -6.51398838e-01 -7.27822661e-01
4.34939384e-01 -3.64353746e-01 7.62350857e-01 4.00815845e-01
7.87932396e-01 -3.92665386e-01 -4.17866819e-02 1.09079790e+00
1.46494448e+00 -3.37492853e-01 5.09936869e-01 -1.56376541e-01
5.71657300e-01 -8.55558589e-02 3.18955779e-01 6.42678142e-01
1.15231588e-01 2.96837598e-01 2.50917405e-01 3.39066088e-01
5.00614047e-01 -3.14889789e-01 4.63964343e-01 1.32202601e+00
-2.02677399e-01 2.01164365e-01 -7.06700325e-01 1.83357328e-01
-1.74486780e+00 -8.37182760e-01 -3.14609334e-02 2.61916113e+00
8.17074478e-01 -1.57409295e-01 -1.49294019e-01 -3.11853200e-01
7.62289345e-01 1.98523834e-01 -5.45256674e-01 -2.38546953e-01
-2.82342196e-01 6.88892007e-01 1.09692478e+00 8.81978095e-01
-8.72768641e-01 1.02500069e+00 6.15314007e+00 1.15501714e+00
-1.13595128e+00 5.24949491e-01 5.81004143e-01 7.32061863e-02
-4.38712656e-01 4.99516755e-01 -5.22706389e-01 4.77467299e-01
1.15260136e+00 -4.50204670e-01 1.07446575e+00 6.56706572e-01
1.71851277e-01 -1.64932832e-01 -6.80877507e-01 1.25551033e+00
-6.50295079e-01 -1.51839066e+00 -4.62585151e-01 4.34570342e-01
1.17658234e+00 6.44054472e-01 1.57396477e-02 2.94101745e-01
4.71662164e-01 -8.95869613e-01 3.91057909e-01 4.62580532e-01
7.65763521e-01 -9.66457427e-01 7.81563163e-01 5.75916350e-01
-8.58126938e-01 5.04791662e-02 -6.72960997e-01 -3.53561677e-02
4.13662612e-01 1.09164929e+00 -4.82606351e-01 4.70911086e-01
2.11687177e-01 3.36398900e-01 1.46972209e-01 8.42078209e-01
-2.61240959e-01 9.71069515e-01 -9.46396768e-01 -1.77572116e-01
4.32431072e-01 -1.11706018e+00 7.76182890e-01 8.16289783e-01
4.02422249e-01 3.14723790e-01 1.25801846e-01 1.21284473e+00
-4.17606682e-01 1.08625673e-01 -1.91588461e-01 -2.78500944e-01
3.66846412e-01 1.22010887e+00 -2.75903076e-01 4.77524400e-02
-4.35150824e-02 1.03195453e+00 4.19076681e-01 4.36136067e-01
-5.84743559e-01 -3.66090566e-01 6.70006454e-01 -1.17611818e-01
2.42488101e-01 -4.74804491e-01 -8.45255854e-04 -1.80290329e+00
1.04010522e-01 -5.11690259e-01 6.25557527e-02 -1.49478212e-01
-1.36878455e+00 3.04662019e-01 -4.54963088e-01 -7.03817546e-01
-8.78056437e-02 -2.88253188e-01 -6.04991198e-01 9.47402477e-01
-1.19474268e+00 -5.97474217e-01 1.14100292e-01 7.47991145e-01
-6.58913553e-01 6.15497231e-02 8.43778074e-01 2.03120083e-01
-6.57564998e-01 4.83334452e-01 8.88992310e-01 1.89832404e-01
9.26049501e-02 -1.42031705e+00 1.06719188e-01 1.02873433e+00
1.44573495e-01 9.01100397e-01 8.75513494e-01 -5.06855071e-01
-2.05849242e+00 -7.90365219e-01 6.96597934e-01 1.45313824e-02
1.06967485e+00 -4.99848098e-01 -6.83018684e-01 5.79011202e-01
-1.64886758e-01 5.76150775e-01 3.42923671e-01 3.60768020e-01
-1.66488022e-01 -1.32998377e-01 -1.41992033e+00 2.17140719e-01
7.97828019e-01 -1.25948894e+00 1.15919329e-01 8.33191454e-01
2.10329145e-01 -4.82479960e-01 -1.18094480e+00 2.40103770e-02
1.14210382e-01 -7.23229885e-01 5.98462045e-01 -4.82248217e-01
-5.22148572e-02 -4.08307523e-01 -3.37134659e-01 -1.26248598e+00
-4.56368536e-01 -1.52541387e+00 -1.38866410e-01 4.27048385e-01
3.30090970e-01 -9.08639491e-01 1.01390028e+00 4.33172464e-01
1.93259969e-01 -4.79755074e-01 -1.56729615e+00 -7.80978084e-01
3.32241327e-01 -3.54503065e-01 3.62087190e-02 7.26444960e-01
1.85280174e-01 4.40149337e-01 -6.46039903e-01 5.61209440e-01
1.22129750e+00 1.74486503e-01 7.38273382e-01 -6.73441947e-01
-5.96379042e-01 -1.56882241e-01 -6.13152742e-01 -1.52086663e+00
2.61255473e-01 -1.15754259e+00 2.00647026e-01 -9.25333023e-01
6.01308942e-01 -6.17474556e-01 -4.15396661e-01 -3.05223852e-01
-1.19071245e-01 -2.70468108e-02 5.72755896e-02 2.44632885e-01
-9.26162720e-01 9.29057717e-01 1.31728613e+00 -4.69463170e-02
-9.65780467e-02 3.09422702e-01 -3.78326327e-01 3.31617415e-01
5.43140948e-01 -8.52634311e-01 1.14657693e-01 -1.36709273e-01
5.83979607e-01 3.85962009e-01 2.26931393e-01 -9.48487520e-01
4.01185900e-01 -1.87884077e-01 -3.71911228e-01 -2.54423440e-01
4.77631748e-01 -8.25179145e-02 8.73381943e-02 5.47664165e-01
-1.95204824e-01 -7.59144843e-01 -4.51717556e-01 7.14916348e-01
-1.24267444e-01 -3.31883341e-01 1.14716864e+00 9.56324413e-02
-9.90483537e-02 6.46701574e-01 -2.63143003e-01 1.76417932e-01
5.13957024e-01 6.10486746e-01 -1.49348766e-01 -7.98765421e-01
-7.68403590e-01 1.01563605e-02 2.97611147e-01 -5.82149982e-01
2.91965783e-01 -1.42696786e+00 -7.88384795e-01 1.83046013e-01
-1.62441596e-01 -5.95051311e-02 3.99928153e-01 1.29590726e+00
-5.31182110e-01 5.90056539e-01 3.78594995e-01 -6.01633251e-01
-5.92594326e-01 6.35184944e-02 7.52727628e-01 -3.28155607e-01
-4.70624417e-01 1.10352397e+00 1.25900820e-01 -6.21909082e-01
-4.40366954e-01 -2.37291902e-01 7.55507767e-01 -6.97974980e-01
4.56928402e-01 3.84319127e-01 4.31784503e-02 -8.19202065e-01
-2.15598509e-01 3.97220582e-01 9.35014188e-02 -2.37314776e-01
1.18190730e+00 -9.89281684e-02 -6.10589564e-01 3.28177005e-01
1.81363857e+00 2.34042645e-01 -1.05866706e+00 -7.67886996e-01
-1.76102400e-01 -3.39534283e-01 5.82804799e-01 -1.11498758e-01
-1.02957559e+00 6.08710468e-01 5.44164956e-01 1.54143587e-01
6.15711808e-01 3.15832555e-01 7.86752105e-01 9.98269260e-01
1.00615764e+00 -9.03519213e-01 -3.46550524e-01 8.12648296e-01
1.60221085e-01 -1.21519339e+00 -1.20421894e-01 -2.58588884e-02
-2.93470889e-01 9.71669734e-01 -2.04392731e-01 -4.18493032e-01
8.49364638e-01 8.33514333e-03 -5.56730509e-01 -2.85198659e-01
-4.82106119e-01 -1.88386723e-01 1.81745857e-01 5.96002638e-02
-1.20513495e-02 5.84159315e-01 -1.04862094e-01 1.74815759e-01
-4.50986564e-01 -2.00475290e-01 5.75152099e-01 6.16034508e-01
-5.52921355e-01 -9.95844543e-01 -1.75083503e-01 4.34772760e-01
-5.68843484e-01 -1.64477661e-01 1.76274270e-01 3.83829206e-01
-3.52871031e-01 8.22997749e-01 -2.53847241e-01 -3.45336914e-01
-1.82428226e-01 -1.50341213e-01 8.88588607e-01 -5.09537518e-01
-6.51322305e-02 1.02320224e-01 -1.62707090e-01 -7.87352145e-01
-3.57571751e-01 -7.18633950e-01 -1.11940300e+00 -9.01034594e-01
-5.81610620e-01 6.40323758e-01 6.51784539e-01 1.39223337e+00
9.48813036e-02 -1.81044728e-01 1.07562613e+00 -5.05587399e-01
-1.13430870e+00 -9.26761806e-01 -1.13698840e+00 1.97333604e-01
4.76121575e-01 -8.68054479e-02 -9.36183453e-01 -6.41406953e-01] | [5.99885368347168, 4.780940532684326] |
889a780b-e7c8-435d-9bc5-ea5ed13b7953 | improving-the-harmony-of-the-composite-image | 1907.06406 | null | https://arxiv.org/abs/1907.06406v3 | https://arxiv.org/pdf/1907.06406v3.pdf | Improving the Harmony of the Composite Image by Spatial-Separated Attention Module | Image composition is one of the most important applications in image processing. However, the inharmonious appearance between the spliced region and background degrade the quality of the image. Thus, we address the problem of Image Harmonization: Given a spliced image and the mask of the spliced region, we try to harmonize the "style" of the pasted region with the background (non-spliced region). Previous approaches have been focusing on learning directly by the neural network. In this work, we start from an empirical observation: the differences can only be found in the spliced region between the spliced image and the harmonized result while they share the same semantic information and the appearance in the non-spliced region. Thus, in order to learn the feature map in the masked region and the others individually, we propose a novel attention module named Spatial-Separated Attention Module (S2AM). Furthermore, we design a novel image harmonization framework by inserting the S2AM in the coarser low-level features of the Unet structure in two different ways. Besides image harmonization, we make a big step for harmonizing the composite image without the specific mask under previous observation. The experiments show that the proposed S2AM performs better than other state-of-the-art attention modules in our task. Moreover, we demonstrate the advantages of our model against other state-of-the-art image harmonization methods via criteria from multiple points of view. Code is available at https://github.com/vinthony/s2am | ['Chi-Man Pun', 'Xiaodong Cun'] | 2019-07-15 | null | null | null | null | ['image-harmonization'] | ['computer-vision'] | [ 2.41086230e-01 -7.97984377e-02 1.37246817e-01 -1.60431534e-01
-3.34489971e-01 -1.37967810e-01 3.25308591e-01 -1.65150940e-01
-3.48886639e-01 4.31833655e-01 5.71838915e-02 4.91169915e-02
1.48280129e-01 -7.87770212e-01 -9.47567403e-01 -8.44246447e-01
6.05286002e-01 -1.56619728e-01 3.67864698e-01 -4.22447890e-01
1.79897964e-01 2.73056358e-01 -1.52860606e+00 4.21373039e-01
1.17125559e+00 1.06158614e+00 5.94507873e-01 2.03349918e-01
-2.64454007e-01 3.52588981e-01 -5.82130969e-01 -4.30357933e-01
5.94892383e-01 -8.27619612e-01 -5.81083000e-01 3.19304287e-01
6.01531208e-01 -1.07697532e-01 -2.45413467e-01 1.52772284e+00
5.25373757e-01 1.10415019e-01 3.10612321e-01 -1.14835179e+00
-9.32873905e-01 4.63227540e-01 -1.06372249e+00 2.06868947e-02
6.51824921e-02 1.56958088e-01 7.93948948e-01 -9.03021216e-01
4.70626593e-01 1.24236047e+00 5.13622224e-01 3.11117500e-01
-1.27321541e+00 -8.15455914e-01 2.36322567e-01 4.59481865e-01
-1.52338064e+00 -3.31572503e-01 9.91888523e-01 -2.66392946e-01
3.35303366e-01 3.18740994e-01 7.32123554e-01 7.67965913e-01
3.45500946e-01 6.73146665e-01 1.34371233e+00 -5.25138378e-01
-1.11550421e-01 1.83919385e-01 -2.13099092e-01 6.38894260e-01
5.00017256e-02 -8.01028982e-02 -2.61229098e-01 4.36929047e-01
7.59670258e-01 7.44699594e-03 -5.94037473e-01 -2.56624967e-01
-1.04823399e+00 5.17914414e-01 7.28516638e-01 6.92254305e-01
-3.60847294e-01 -1.11767955e-01 -1.29608614e-02 8.96444619e-02
4.11566913e-01 3.59759182e-01 -1.72417477e-01 4.33301121e-01
-9.48381007e-01 1.13980800e-01 2.49815941e-01 6.96681082e-01
9.93285060e-01 -9.05180350e-02 -3.27651381e-01 8.97967458e-01
3.71754058e-02 1.89684436e-01 6.26158059e-01 -6.36836529e-01
3.99039000e-01 6.78020358e-01 1.66285217e-01 -1.16322625e+00
-1.99478820e-01 -6.07903779e-01 -9.51541305e-01 4.55499470e-01
2.47889012e-01 -1.76628288e-02 -9.20066595e-01 1.77034676e+00
4.54545736e-01 2.59393394e-01 -4.76963893e-02 1.03539729e+00
9.30506170e-01 7.37259567e-01 -1.77389637e-01 -1.80708587e-01
1.43777156e+00 -1.44542754e+00 -8.41784120e-01 -2.34743819e-01
5.16502708e-02 -1.05753982e+00 1.16561472e+00 8.24538097e-02
-1.24481034e+00 -1.09021485e+00 -1.16669476e+00 -1.87318817e-01
-2.38770887e-01 1.73316658e-01 1.17771693e-01 3.71011734e-01
-1.12158978e+00 5.40145397e-01 -4.28036600e-01 -4.39253867e-01
1.91579610e-01 7.53909722e-02 -4.71463740e-01 7.28103593e-02
-1.05336964e+00 9.38880026e-01 4.75967973e-01 3.74933869e-01
-4.66096193e-01 -5.74676394e-01 -8.39764297e-01 1.93268061e-01
4.15670127e-01 -7.44276583e-01 9.42586184e-01 -1.56654131e+00
-1.45987749e+00 1.05834579e+00 -1.61614940e-01 -1.75414369e-01
5.77757239e-01 -1.09542653e-01 -3.07948679e-01 -8.03070813e-02
2.56400526e-01 6.16117954e-01 9.93084013e-01 -1.53761744e+00
-6.70262516e-01 -3.80203664e-01 -1.68593273e-01 4.55010295e-01
-1.96545526e-01 -5.23640029e-02 -8.72597277e-01 -9.99521792e-01
1.47218704e-01 -6.73265398e-01 -5.17728217e-02 3.38330418e-02
-5.47832072e-01 1.25535831e-01 9.53314841e-01 -8.32973599e-01
1.13331532e+00 -2.40758896e+00 2.18985125e-01 -7.67523609e-03
7.44085386e-02 4.43861485e-01 -3.84922147e-01 1.48545787e-01
-3.89197618e-01 1.72283024e-01 -3.78110260e-01 -4.04960185e-01
-2.05921799e-01 -5.10678887e-02 -1.08292505e-01 3.38553280e-01
2.37827882e-01 9.16030169e-01 -6.88787818e-01 -6.01395667e-01
1.70712054e-01 4.52405691e-01 -4.32967335e-01 4.70083177e-01
3.94256599e-02 7.68066049e-01 -1.66926801e-01 3.75800192e-01
1.00610411e+00 4.88757342e-02 7.20576197e-02 -7.60189533e-01
-2.80944258e-01 -1.66285083e-01 -1.26464534e+00 1.65860391e+00
-4.99121368e-01 3.63976300e-01 1.88993767e-01 -9.14065540e-01
9.67801392e-01 -2.32154373e-02 3.71147633e-01 -1.10644305e+00
2.14971527e-01 2.74751514e-01 1.87955126e-01 -6.19272113e-01
4.59452063e-01 -2.39628538e-01 1.85196653e-01 3.28174114e-01
-2.10212432e-02 -1.93917871e-01 6.80444166e-02 -2.32690290e-01
5.19934535e-01 1.00138664e-01 3.76029372e-01 -1.53647348e-01
7.08274662e-01 -3.84856820e-01 6.77869499e-01 6.81550145e-01
-2.27237612e-01 9.29389834e-01 4.50321853e-01 -3.47316146e-01
-1.07815945e+00 -9.79514062e-01 -6.20172918e-02 8.01232457e-01
6.94591701e-01 -9.19383466e-02 -1.08555090e+00 -5.35532355e-01
-1.68168634e-01 4.88059491e-01 -7.84704566e-01 -1.98058963e-01
-6.73735797e-01 -6.58213198e-01 1.36864483e-01 2.73405164e-01
1.10743022e+00 -1.05177832e+00 -4.07708198e-01 1.25378789e-02
-4.08342689e-01 -1.04029632e+00 -8.91681790e-01 -9.16296467e-02
-4.78112817e-01 -9.55165207e-01 -8.24007213e-01 -9.38194036e-01
7.19840825e-01 5.17737746e-01 9.54277813e-01 2.71384954e-01
-1.63008809e-01 -1.41082391e-01 -4.21743602e-01 -2.67274320e-01
-3.15220028e-01 -1.58343047e-01 -3.19417596e-01 7.28328645e-01
-9.34139043e-02 -5.49193561e-01 -8.65737855e-01 5.20086944e-01
-1.14652991e+00 4.36503083e-01 7.33083725e-01 8.87263477e-01
6.67669773e-01 5.16216338e-01 2.65218824e-01 -6.57902062e-01
5.34310102e-01 -1.65769055e-01 -5.45688331e-01 2.82456964e-01
-4.09723908e-01 3.30470614e-02 6.04611039e-01 -3.64070088e-01
-1.15660155e+00 -9.42557002e-04 -1.65514842e-01 -6.06653273e-01
-7.50422850e-02 1.50742605e-01 -6.85875654e-01 -7.14047924e-02
2.15615109e-01 2.80570954e-01 1.23084165e-01 -5.58726728e-01
3.20015609e-01 6.20430470e-01 7.42608070e-01 -3.03893566e-01
8.20484400e-01 6.48205161e-01 -2.90273756e-01 -5.27291656e-01
-7.94774413e-01 -1.01807475e-01 -5.81499636e-01 -6.10033609e-02
1.06413543e+00 -5.91015697e-01 -3.50240827e-01 8.23997259e-01
-1.30636489e+00 -4.09728110e-01 -3.31979364e-01 2.05383822e-01
-3.77405465e-01 5.09999216e-01 -3.66566360e-01 -5.47195852e-01
-3.08194906e-01 -1.43537223e+00 1.09904408e+00 5.54100990e-01
5.93694299e-02 -5.97841144e-01 -1.48490928e-02 3.91915590e-01
4.12038535e-01 1.69389755e-01 1.07956338e+00 -4.03888345e-01
-4.65833724e-01 1.65312335e-01 -4.87788200e-01 4.87510383e-01
3.82622868e-01 3.36499549e-02 -9.13118243e-01 -2.00827435e-01
2.31472522e-01 1.84871286e-01 1.07475758e+00 4.32530493e-01
1.10599113e+00 -2.48387918e-01 -3.72461155e-02 7.94991791e-01
1.46101582e+00 4.45053637e-01 9.90082562e-01 5.09511769e-01
6.94841683e-01 7.66152024e-01 5.67378938e-01 1.26572803e-01
1.19249433e-01 1.07451868e+00 4.77393687e-01 -6.71238005e-01
-5.56304336e-01 -2.74462789e-01 2.31073856e-01 5.33012390e-01
1.95859820e-02 -2.66654938e-01 -5.35445333e-01 4.71813440e-01
-1.90821600e+00 -8.86185050e-01 -5.77147231e-02 2.30481100e+00
7.99412251e-01 -9.32621136e-02 -4.02951166e-02 9.51030175e-04
1.15536284e+00 3.05417806e-01 -4.15167302e-01 -3.10230762e-01
-3.85266930e-01 3.18931133e-01 1.88997984e-01 5.78830004e-01
-1.14143538e+00 9.43102121e-01 5.22124290e+00 1.17165542e+00
-1.29972255e+00 1.40902489e-01 8.62575829e-01 6.57370985e-02
-3.33019495e-01 -5.37730604e-02 -5.36636353e-01 8.24691057e-01
1.66469201e-01 -1.17069017e-03 4.09430683e-01 3.65501434e-01
4.23440248e-01 -2.49583453e-01 -8.58608365e-01 9.93169606e-01
3.30079466e-01 -1.03450191e+00 2.10878536e-01 -3.49873424e-01
8.52853179e-01 -4.62239206e-01 2.81366467e-01 2.97248289e-02
-2.19246835e-01 -9.81232345e-01 9.33934867e-01 3.79967839e-01
7.21554935e-01 -5.65688372e-01 8.74535680e-01 -3.57080251e-02
-1.30027711e+00 -8.57423469e-02 -2.23149210e-01 2.12036520e-01
6.27214983e-02 5.49573779e-01 -2.75416672e-01 7.68648028e-01
9.80479598e-01 6.29472673e-01 -7.44086385e-01 1.04617941e+00
-3.45920354e-01 -2.07609795e-02 6.47187606e-02 5.58127820e-01
4.83308658e-02 -5.35141230e-01 6.63844407e-01 9.42369521e-01
4.12329227e-01 -1.09100096e-01 -4.63682823e-02 1.21973348e+00
-1.62547618e-01 3.12849760e-01 -4.84914899e-01 3.94337595e-01
2.96514213e-01 1.28980505e+00 -7.36914992e-01 -3.33342046e-01
-4.60006654e-01 1.35639691e+00 2.51531929e-01 4.60402638e-01
-1.01077795e+00 -4.13472354e-01 5.74168086e-01 2.12042913e-01
5.15616655e-01 2.59206474e-01 -3.64586562e-01 -1.11308813e+00
2.75360376e-01 -1.06385434e+00 2.04452559e-01 -1.10570896e+00
-1.27829981e+00 8.09634566e-01 -1.73292801e-01 -1.28846824e+00
2.90270984e-01 -3.19196284e-01 -9.57694292e-01 1.01220822e+00
-1.48658097e+00 -1.22214663e+00 -5.96998155e-01 6.89374864e-01
5.27112246e-01 6.76832572e-02 4.26994205e-01 4.56941515e-01
-8.25479209e-01 6.23605609e-01 1.25302330e-01 1.70496345e-01
8.96254659e-01 -8.98447037e-01 2.75016844e-01 1.10324526e+00
1.10218547e-01 4.28281784e-01 7.02487528e-01 -6.29770637e-01
-7.91936457e-01 -1.12824070e+00 6.23258293e-01 1.76590502e-01
3.58585358e-01 -1.98295206e-01 -1.04902625e+00 4.24143702e-01
6.50432706e-01 -1.35958269e-01 1.49822608e-01 -4.48374808e-01
-4.12672579e-01 -4.32169914e-01 -1.01735854e+00 7.99503624e-01
7.95820475e-01 -3.05145085e-01 -4.69035506e-01 -6.81304857e-02
7.02382326e-01 -3.74984980e-01 -4.91153032e-01 5.79395413e-01
5.78270257e-01 -1.36345267e+00 8.76941919e-01 -2.51860112e-01
6.08435392e-01 -7.57679403e-01 -9.59499180e-02 -1.40448534e+00
-3.52136433e-01 -5.95776021e-01 5.32898784e-01 1.34118474e+00
2.51756608e-01 -6.76387668e-01 4.38835561e-01 3.29205066e-01
-2.24526241e-01 -7.43774533e-01 -7.88464069e-01 -5.63824534e-01
-2.97066662e-02 1.83600653e-02 6.64548695e-01 8.53752732e-01
-4.66871023e-01 2.77191103e-01 -4.80764836e-01 1.57250047e-01
4.04456437e-01 6.06536508e-01 7.94809878e-01 -6.92818284e-01
-4.56952900e-01 -6.06721163e-01 -2.95992494e-01 -7.54122198e-01
1.01694092e-01 -6.45170689e-01 1.01743653e-01 -1.35433853e+00
4.88494962e-01 -2.07535908e-01 -4.13567841e-01 5.25592625e-01
-6.19481921e-01 5.94090283e-01 6.10688746e-01 1.31800219e-01
-3.01961124e-01 6.51493430e-01 1.58939362e+00 -3.12871009e-01
-2.96899408e-01 -2.55818307e-01 -8.19825351e-01 7.38689482e-01
8.64271283e-01 -2.88430601e-01 -1.47503123e-01 -5.42299032e-01
-2.37431511e-01 -2.76104420e-01 4.32680786e-01 -9.56166863e-01
1.30440801e-01 -1.10085033e-01 3.66916656e-01 -4.22372609e-01
2.52698988e-01 -8.84775519e-01 3.77642423e-01 4.93932754e-01
-2.00727522e-01 1.33138105e-01 1.31223753e-01 3.85026544e-01
-4.31167752e-01 -2.43027762e-01 1.29341388e+00 -2.45131731e-01
-6.71854913e-01 2.06992343e-01 -7.98641220e-02 -6.61340654e-02
1.20714402e+00 -4.04057682e-01 -3.33123982e-01 -3.37272346e-01
-5.15796959e-01 -4.60816287e-02 7.43069351e-01 5.11347771e-01
6.24340713e-01 -1.26585126e+00 -7.82000005e-01 5.32939911e-01
-4.28008437e-02 -6.85239360e-02 6.19125485e-01 1.03805184e+00
-4.16585654e-01 -7.80516937e-02 -6.14928544e-01 -4.33291703e-01
-1.28673863e+00 8.07734132e-01 6.27645075e-01 -2.12867409e-01
-5.08460999e-01 6.62439644e-01 8.19904447e-01 -1.34576261e-01
6.89741597e-02 -1.49584785e-01 -2.29453012e-01 4.59428616e-02
4.66651499e-01 3.78675833e-02 1.98822934e-02 -8.60643446e-01
-1.00057632e-01 9.63194489e-01 -9.68258083e-02 3.33297625e-02
1.20314217e+00 -3.37862104e-01 -4.81782705e-01 2.05391780e-01
1.28969288e+00 1.52786374e-01 -1.11539614e+00 -2.24821433e-01
-4.60984260e-01 -7.89437175e-01 -1.74616799e-02 -5.06405532e-01
-1.52462506e+00 7.55218148e-01 8.08164954e-01 4.58454899e-02
1.54619491e+00 -1.57256290e-01 8.01351011e-01 -3.01116824e-01
2.51624826e-02 -9.17130828e-01 2.28931755e-01 6.21652938e-02
1.17006385e+00 -1.14932847e+00 -9.92672071e-02 -4.43182170e-01
-6.21462464e-01 8.03127229e-01 8.45930338e-01 -1.43245086e-01
4.40355241e-01 8.10797215e-02 2.69942462e-01 -5.60286455e-02
-2.73757309e-01 -4.28004265e-01 3.71160716e-01 4.92010921e-01
2.60453463e-01 -2.82014787e-01 -4.37963188e-01 5.48873305e-01
-9.34265554e-02 -3.32674772e-01 4.01389807e-01 5.48018336e-01
-2.81389892e-01 -1.10797465e+00 -6.12302363e-01 2.07123347e-02
-3.73504251e-01 -2.98641324e-01 -4.59575713e-01 7.66551554e-01
5.24855554e-01 8.98280144e-01 1.62080973e-01 -5.19651413e-01
4.91362154e-01 -1.93818480e-01 3.64120543e-01 -4.19414580e-01
-6.37329280e-01 3.67198914e-01 -2.93001026e-01 -7.03036368e-01
-4.93246526e-01 -2.25879699e-01 -9.18494701e-01 -2.20698923e-01
-2.33267024e-01 -2.72539202e-02 4.49523032e-01 6.59418821e-01
3.94773811e-01 5.70691347e-01 7.64502406e-01 -9.59981918e-01
-1.24029569e-01 -8.13583255e-01 -7.79443681e-01 8.20701838e-01
3.07543337e-01 -6.24668658e-01 -1.98958203e-01 1.33413792e-01] | [11.246770858764648, -1.205302357673645] |
9866b16f-1341-4558-bba4-14b8108dab87 | dynamic-neural-program-embeddings-for-program | null | null | https://openreview.net/forum?id=BJuWrGW0Z | https://openreview.net/pdf?id=BJuWrGW0Z | Dynamic Neural Program Embeddings for Program Repair | Neural program embeddings have shown much promise recently for a variety of program analysis tasks, including program synthesis, program repair, code completion, and fault localization. However, most existing program embeddings are based on syntactic features of programs, such as token sequences or abstract syntax trees. Unlike images and text, a program has well-defined semantics that can be difficult to capture by only considering its syntax (i.e. syntactically similar programs can exhibit vastly different run-time behavior), which makes syntax-based program embeddings fundamentally limited. We propose a novel semantic program embedding that is learned from program execution traces. Our key insight is that program states expressed as sequential tuples of live variable values not only capture program semantics more precisely, but also offer a more natural fit for Recurrent Neural Networks to model. We evaluate different syntactic and semantic program embeddings on the task of classifying the types of errors that students make in their submissions to an introductory programming class and on the CodeHunt education platform. Our evaluation results show that the semantic program embeddings significantly outperform the syntactic program embeddings based on token sequences and abstract syntax trees. In addition, we augment a search-based program repair system with predictions made from our semantic embedding and demonstrate significantly improved search efficiency.
| ['Zhendong Su', 'Ke Wang', 'Rishabh Singh'] | 2018-01-01 | null | null | null | iclr-2018-1 | ['fault-localization', 'program-repair', 'program-repair'] | ['computer-code', 'computer-code', 'reasoning'] | [-6.75141811e-02 -3.17949653e-01 -6.67672694e-01 -5.85740685e-01
-3.65708888e-01 -5.56471646e-01 -1.87413692e-02 7.19291747e-01
-5.46511225e-02 -1.32258788e-01 3.16304922e-01 -6.94135487e-01
3.82584244e-01 -1.00099003e+00 -1.04846370e+00 -1.74308255e-01
-9.93278697e-02 -8.27102512e-02 2.30711624e-01 -2.18525365e-01
3.72401595e-01 7.31525719e-02 -1.80591333e+00 4.37001437e-01
7.93735266e-01 4.48655397e-01 4.17895645e-01 8.26497257e-01
-6.23159766e-01 1.29553580e+00 -7.18698084e-01 -2.73868829e-01
-5.30304074e-01 -1.32871166e-01 -9.39101458e-01 -4.93911684e-01
4.36076790e-01 -2.01423392e-01 -6.34253263e-01 1.47154713e+00
7.57645145e-02 1.27889425e-01 -7.61715695e-02 -1.24711096e+00
-1.43547702e+00 9.19510484e-01 -5.62698543e-02 2.02549472e-01
6.85940623e-01 1.56628251e-01 1.40200508e+00 -5.50210595e-01
5.69247484e-01 1.22800279e+00 1.09942532e+00 6.69204354e-01
-1.59806085e+00 -3.20726514e-01 1.55591384e-01 2.24535421e-01
-8.95577013e-01 8.93882886e-02 6.33016646e-01 -6.86500251e-01
1.43150222e+00 2.46277258e-01 5.54825723e-01 9.14993346e-01
6.34571671e-01 1.09718215e+00 4.23361689e-01 -3.59547466e-01
1.15997262e-01 3.01499218e-01 7.20363736e-01 1.25835145e+00
-1.10099211e-01 1.16120584e-01 -2.49428809e-01 -3.80568773e-01
1.64853364e-01 5.80435336e-01 -3.60296547e-01 -3.69456023e-01
-1.13697958e+00 9.69762325e-01 6.61700130e-01 2.49381632e-01
2.74607748e-01 1.01017714e+00 1.05794156e+00 6.70654058e-01
1.77357183e-03 6.94775701e-01 -4.86770511e-01 -4.52743858e-01
-7.53817022e-01 2.93821454e-01 6.88199401e-01 1.02853131e+00
8.79594803e-01 6.08470559e-01 -5.95657267e-02 6.76334441e-01
5.49887896e-01 3.95801783e-01 9.94733870e-01 -5.04884243e-01
1.61563098e-01 1.24918509e+00 -5.67058980e-01 -1.18450761e+00
-1.50435418e-01 -2.80659553e-02 -3.40008467e-01 1.82140678e-01
1.74070150e-02 2.99560428e-01 -8.70650530e-01 1.73661029e+00
-3.01138431e-01 3.43025550e-02 1.73354089e-01 4.68968719e-01
1.10330379e+00 7.74529755e-01 -1.17273331e-01 5.32226622e-01
1.34317350e+00 -1.12250221e+00 -4.91277963e-01 -3.87249321e-01
1.43968129e+00 -3.90634894e-01 1.38538921e+00 9.17421561e-03
-7.81596839e-01 -5.55676758e-01 -1.28988898e+00 -1.54236913e-01
-4.28788841e-01 -1.34150773e-01 9.87511039e-01 6.31760359e-01
-1.21124673e+00 8.44663739e-01 -8.94872725e-01 -2.73815766e-02
1.96544304e-01 3.03058773e-01 -2.38793418e-01 -1.48037955e-01
-7.30359018e-01 4.96502072e-01 1.31269202e-01 -3.16207796e-01
-1.14041841e+00 -1.04714274e+00 -1.54381764e+00 5.11012197e-01
2.07537219e-01 -3.53450358e-01 1.54671371e+00 -1.18624854e+00
-9.91779208e-01 9.78491127e-01 -4.59737420e-01 -5.10881007e-01
-3.20454091e-01 9.82780308e-02 -2.73803711e-01 -3.25077474e-01
-1.87496338e-02 -8.91835801e-03 3.01144630e-01 -8.64671528e-01
-1.62451625e-01 -2.99473733e-01 3.17718327e-01 -3.82759959e-01
-4.58203256e-01 5.63126862e-01 6.41874373e-02 -6.83895350e-01
-2.05128770e-02 -9.90111113e-01 -2.83318907e-01 -8.12195837e-02
-2.83562690e-01 -2.91845143e-01 9.21028614e-01 -5.91240406e-01
1.66691315e+00 -2.64196730e+00 2.62672663e-01 -6.87215989e-03
3.71274441e-01 1.63685590e-01 -1.42525584e-01 4.71777827e-01
-3.13401610e-01 6.80671394e-01 -3.94451439e-01 8.88199732e-02
5.14802396e-01 1.35578305e-01 -8.45617592e-01 3.46248120e-01
3.22162919e-02 1.18547225e+00 -1.01907527e+00 -1.02074720e-01
4.82749566e-02 1.41423196e-01 -1.00476968e+00 3.51820290e-01
-7.08910465e-01 -3.63296449e-01 -4.56020147e-01 7.25845635e-01
1.09207198e-01 -3.01876336e-01 6.45298213e-02 4.91125911e-01
-2.91526299e-02 5.26865304e-01 -6.51074648e-01 1.72794282e+00
-7.85067141e-01 1.13744450e+00 -2.26156369e-01 -1.12063432e+00
9.04232860e-01 4.40023124e-01 1.58689946e-01 -4.64847207e-01
-2.17018574e-01 9.12977979e-02 -1.87125966e-01 -6.65376961e-01
7.13361919e-01 1.44719213e-01 -6.21583521e-01 8.20741773e-01
1.15610272e-01 -2.34402955e-01 -1.88013256e-01 1.68158337e-01
1.59556925e+00 -1.07982054e-01 2.76285615e-02 -2.36638322e-01
4.92899299e-01 2.72580773e-01 5.63865960e-01 7.59444892e-01
-3.12287927e-01 3.86753142e-01 1.09494197e+00 -7.48640597e-01
-8.32930505e-01 -8.78707409e-01 1.32433116e-01 1.48454511e+00
-1.35595202e-01 -1.10463154e+00 -6.65735841e-01 -7.47329712e-01
1.22032613e-01 6.40144229e-01 -6.33404613e-01 -6.78572357e-01
-8.08640718e-01 -3.46934289e-01 6.70785010e-01 1.01829171e+00
2.44956389e-02 -1.07617950e+00 -6.69732034e-01 1.46242514e-01
3.76505405e-01 -4.73460734e-01 -8.82073283e-01 4.41335738e-01
-9.07043457e-01 -1.03055871e+00 2.15990506e-02 -1.41874015e+00
6.12370729e-01 1.31236702e-01 1.40236199e+00 8.55858088e-01
-3.03565562e-01 4.94855076e-01 -2.92572737e-01 -9.38980058e-02
-7.94289231e-01 -3.02479893e-01 -1.91365033e-01 -7.49670208e-01
5.21586001e-01 -5.42877078e-01 -1.68930858e-01 -4.19596583e-02
-1.06021512e+00 -2.64961541e-01 1.30171686e-01 1.25394785e+00
3.96140426e-01 3.16010378e-02 2.29484409e-01 -1.08314097e+00
7.72199452e-01 -3.35539728e-01 -6.26667857e-01 3.96420300e-01
-7.53885210e-01 5.48056960e-01 1.05428445e+00 -3.76617074e-01
-7.11885870e-01 -2.69486219e-01 -2.91491508e-01 -6.88966572e-01
2.21977588e-02 8.78130317e-01 9.56846103e-02 -2.57577986e-01
7.17978179e-01 4.68195468e-01 -3.64135504e-02 -9.59237739e-02
1.93178058e-01 3.06910127e-01 6.19690180e-01 -1.07925737e+00
4.96808320e-01 -1.03057429e-01 -4.44296271e-01 -5.88222861e-01
-1.46441132e-01 -1.32190570e-01 -1.82728916e-01 3.56236815e-01
8.10313046e-01 -7.39349604e-01 -7.02184081e-01 2.12315902e-01
-1.35222542e+00 -4.17159528e-01 -4.81153399e-01 2.42196977e-01
-4.72456723e-01 4.12969649e-01 -1.07559490e+00 -2.63225257e-01
-1.07266597e-01 -2.01579499e+00 8.19773018e-01 9.37846005e-02
-5.06418526e-01 -1.25145090e+00 2.31960118e-01 -2.24472389e-01
5.30664146e-01 1.79519102e-01 1.73945415e+00 -6.92730308e-01
-7.79757857e-01 -4.94391054e-01 -1.06713120e-02 3.97899806e-01
-2.89006680e-01 3.88672799e-01 -8.11060488e-01 -2.21634224e-01
-7.67587358e-03 -1.76172152e-01 7.18938172e-01 3.11141670e-01
1.49241519e+00 -3.57599318e-01 -3.75549316e-01 1.00813913e+00
1.68095863e+00 5.46057262e-02 3.48169148e-01 2.62861013e-01
9.25791919e-01 4.52205420e-01 -3.20595279e-02 2.61479169e-01
3.05877328e-01 4.75529283e-01 4.46306735e-01 5.75748920e-01
1.02743238e-01 -5.23977518e-01 8.97246480e-01 1.05574310e+00
5.86189866e-01 2.77653188e-01 -1.26984704e+00 7.75165319e-01
-1.83836353e+00 -1.04628205e+00 -1.59803763e-01 2.18255949e+00
7.61677444e-01 -1.05957538e-01 -2.43622079e-01 -6.79769218e-02
5.49038947e-01 3.30695927e-01 -2.79375851e-01 -1.24439192e+00
4.27845627e-01 5.76629937e-01 3.86120141e-01 2.83408105e-01
-8.41326237e-01 8.72548759e-01 6.25064564e+00 4.23640132e-01
-1.19294322e+00 5.73798083e-02 3.13282460e-02 1.97462857e-01
-9.79814768e-01 2.57676929e-01 -6.02171481e-01 4.44726586e-01
1.41331482e+00 -9.32666779e-01 3.97634774e-01 1.48920238e+00
-3.61976326e-01 2.73949206e-01 -1.80303478e+00 7.31598258e-01
8.89911652e-02 -1.65749013e+00 -1.21998645e-01 -3.00653964e-01
6.00025117e-01 -6.11231700e-02 2.14205682e-01 9.62261677e-01
6.27271235e-01 -1.33438528e+00 9.09627378e-01 4.05972004e-02
5.65822005e-01 -8.52404118e-01 5.99834561e-01 3.39942984e-02
-1.34911180e+00 -3.61228555e-01 -2.61496097e-01 -9.72080976e-02
-3.81719530e-01 3.05492818e-01 -6.48325026e-01 -5.73813505e-02
7.57749975e-01 9.93208587e-01 -8.47469330e-01 7.63680339e-01
-3.29962015e-01 5.40190399e-01 3.98075253e-01 -3.84919971e-01
2.06039861e-01 2.53090948e-01 3.20405394e-01 1.26786602e+00
2.55628467e-01 -3.74044776e-01 3.11439276e-01 1.73366332e+00
-2.47771323e-01 -2.43058190e-01 -9.60911393e-01 -4.15897340e-01
4.68894243e-01 6.84875190e-01 -2.41019696e-01 -3.65910530e-01
-8.87499392e-01 7.54419982e-01 3.39560419e-01 2.67618686e-01
-8.78981888e-01 -7.58918226e-01 1.13279188e+00 -2.90402979e-01
1.03191547e-01 -4.19063210e-01 -3.55939567e-01 -1.27081347e+00
2.24129871e-01 -9.17801380e-01 1.23539664e-01 -5.94202936e-01
-6.44590676e-01 4.01255965e-01 -2.42635369e-01 -6.49940133e-01
-2.76878595e-01 -7.43645728e-01 -1.09633541e+00 8.02794814e-01
-1.24394226e+00 -5.58661699e-01 -1.57065578e-02 8.61211866e-02
5.76834917e-01 -7.58806989e-02 1.14954066e+00 2.23235682e-01
-8.52934718e-01 9.72600996e-01 2.20265388e-01 4.87832099e-01
3.87764543e-01 -1.63310087e+00 7.90039539e-01 9.13870931e-01
6.60866275e-02 1.06268561e+00 6.41645372e-01 -4.24589097e-01
-2.24430203e+00 -1.10266173e+00 8.16878557e-01 -6.02284610e-01
8.42213154e-01 -3.18498909e-01 -1.24246442e+00 1.24111867e+00
-5.79729192e-02 2.78850615e-01 7.59987235e-01 3.19978029e-01
-9.11924064e-01 1.71562552e-01 -8.48163724e-01 5.80957890e-01
6.95982635e-01 -1.25793707e+00 -7.91377842e-01 3.87729526e-01
1.36907160e+00 -4.90846068e-01 -9.66776490e-01 1.08663358e-01
4.22858268e-01 -8.05020630e-01 7.75856733e-01 -8.94281089e-01
9.85835552e-01 -1.44852754e-02 -4.33748305e-01 -1.25674582e+00
-1.36578009e-01 -2.56434768e-01 -2.96268433e-01 9.22069192e-01
2.29366168e-01 -5.59098363e-01 8.58148277e-01 8.24573994e-01
-6.65906250e-01 -7.72178531e-01 -4.20591146e-01 -7.65404820e-01
2.75421888e-01 -6.83494568e-01 8.47792804e-01 1.03596783e+00
5.32046795e-01 -8.60117450e-02 2.72942573e-01 1.62735328e-01
2.19531208e-01 6.67313337e-01 5.73946893e-01 -1.00772941e+00
-5.26051939e-01 -8.98579240e-01 -7.08521128e-01 -8.43384266e-01
8.83572042e-01 -1.53573096e+00 1.86967626e-01 -1.00572002e+00
3.36036354e-01 -3.44536990e-01 -1.00946814e-01 5.60036302e-01
4.18283977e-02 -4.82352167e-01 -1.40663669e-01 -1.06115015e-02
-3.51545066e-01 4.87499535e-01 6.11216366e-01 -5.63069105e-01
-1.01470882e-02 -2.76214302e-01 -6.04021430e-01 6.21979058e-01
4.57575291e-01 -4.98869389e-01 -2.35606804e-01 -8.85539651e-01
5.66440523e-01 3.70457977e-01 4.63470578e-01 -9.34419513e-01
2.96813756e-01 -1.81980520e-01 -2.91932911e-01 3.55524644e-02
-3.76829207e-01 -5.17225683e-01 -1.52602598e-01 6.03782654e-01
-6.71625853e-01 5.32066762e-01 5.59453905e-01 7.65190125e-01
-4.50356513e-01 -7.44248629e-01 4.65058178e-01 -3.76882046e-01
-1.29632401e+00 1.94013923e-01 -6.10397935e-01 7.27795660e-02
9.30462658e-01 -1.77614495e-01 -4.64178443e-01 1.16036639e-01
-4.41678703e-01 3.85011546e-02 9.41148698e-01 8.89360249e-01
8.48764896e-01 -1.35464728e+00 -1.15024298e-01 4.50022370e-01
4.80968356e-01 -1.28968179e-01 2.18560711e-01 5.28676450e-01
-7.33277202e-01 4.46721286e-01 2.42212825e-02 -5.72737515e-01
-1.45093560e+00 9.49826360e-01 3.15647721e-01 -1.65422812e-01
-7.31417775e-01 9.25401628e-01 2.79155552e-01 -9.91759777e-01
3.73169005e-01 -1.10487568e+00 3.59969586e-01 -6.47894084e-01
6.88944519e-01 -4.02025729e-02 -6.82419464e-02 -3.17317277e-01
-4.81954068e-01 3.29634845e-01 -1.55722618e-01 4.72098947e-01
1.31823945e+00 6.33226812e-01 -5.97037613e-01 5.30389547e-01
1.63239145e+00 -1.69207126e-01 -6.07381642e-01 -2.94629782e-01
3.96874011e-01 -4.22970742e-01 9.74551663e-02 -3.34138453e-01
-9.16998565e-01 1.18470693e+00 3.50394368e-01 6.33970648e-02
6.00107312e-01 -2.26135273e-02 8.52387547e-01 5.20827174e-01
3.58204454e-01 -7.12161660e-01 3.02888662e-01 7.98374355e-01
4.19716984e-01 -1.17764461e+00 -3.16977471e-01 -4.95627485e-02
-8.05245563e-02 1.42170870e+00 8.76778901e-01 -2.25381255e-01
4.78956461e-01 5.80530226e-01 -5.31055093e-01 -4.59802628e-01
-9.64514554e-01 3.39685768e-01 4.44466732e-02 3.37266624e-01
9.81878757e-01 1.70993954e-01 3.49445134e-01 8.81176114e-01
-1.56791970e-01 -2.56788015e-01 9.78376925e-01 1.42579758e+00
-6.35649502e-01 -1.10005832e+00 -4.86037284e-02 5.47285318e-01
-3.82959932e-01 -2.43752107e-01 -6.31409287e-02 3.68252784e-01
-2.60968745e-01 5.83565593e-01 -1.21976726e-01 -9.26942170e-01
3.34218621e-01 3.08214366e-01 4.02549744e-01 -1.28678524e+00
-9.50373232e-01 -1.02938128e+00 -4.22197163e-01 -8.66222501e-01
2.64175534e-01 -4.97486621e-01 -1.57235706e+00 -7.69583523e-01
-1.46854460e-01 1.85746685e-01 6.39839053e-01 4.48181212e-01
3.82512808e-01 1.07996392e+00 3.47582608e-01 -3.85747313e-01
-9.03841376e-01 -2.60698020e-01 -2.96489298e-01 3.08656871e-01
7.19793499e-01 -2.03531563e-01 -3.85639876e-01 7.33654248e-03] | [7.524619102478027, 7.811716556549072] |
94fc8256-8a2e-4322-8973-8a6181f071d7 | text-only-image-captioning-with-multi-context | 2305.18072 | null | https://arxiv.org/abs/2305.18072v1 | https://arxiv.org/pdf/2305.18072v1.pdf | Text-Only Image Captioning with Multi-Context Data Generation | Text-only Image Captioning (TIC) is an approach that aims to construct a model solely based on text that can accurately describe images. Recently, diffusion models have demonstrated remarkable capabilities in generating high-quality images that are semantically coherent with given texts. This presents an opportunity to generate synthetic training images for TIC. However, we have identified a challenge that the images generated from simple descriptions typically exhibit a single perspective with one or limited contexts, which is not aligned with the complexity of real-world scenes in the image domain. In this paper, we propose a novel framework that addresses this issue by introducing multi-context data generation. Starting with an initial text corpus, our framework employs a large language model to select multiple sentences that describe the same scene from various perspectives. These sentences are then summarized into a single sentence with multiple contexts. We generate simple images using the straightforward sentences and complex images using the summarized sentences through diffusion models. Finally, we train the model exclusively using the synthetic image-text pairs obtained from this process. Experimental results demonstrate that our proposed framework effectively tackles the central challenge we have identified, achieving the state-of-the-art performance on popular datasets such as MSCOCO, Flickr30k, and SS1M. | ['Xiaoyan Sun', 'Yueyi Zhang', 'Fengyun Rao', 'Yizhou Zhou', 'Feipeng Ma'] | 2023-05-29 | null | null | null | null | ['image-captioning'] | ['computer-vision'] | [ 5.80664158e-01 -2.08353940e-02 3.42518508e-01 -3.59869182e-01
-7.80426562e-01 -4.91250277e-01 1.09236634e+00 -3.61193895e-01
-2.16494694e-01 8.33362520e-01 2.71148980e-01 -1.04328024e-03
2.84028322e-01 -7.73579776e-01 -9.02601779e-01 -6.29097342e-01
5.44989347e-01 3.13515782e-01 2.54940122e-01 -1.77972287e-01
1.62977427e-01 1.23308539e-01 -1.63647556e+00 6.37887478e-01
9.87735689e-01 7.34495878e-01 8.16550493e-01 6.24138296e-01
-4.49701309e-01 1.06713653e+00 -7.97928631e-01 -5.72539389e-01
8.00259560e-02 -7.64641881e-01 -5.27445257e-01 6.64365232e-01
5.38267314e-01 -5.39043069e-01 -2.51449764e-01 1.06944203e+00
5.11771202e-01 -1.07011914e-01 7.81476617e-01 -1.23398983e+00
-9.50964093e-01 4.29176211e-01 -6.09716892e-01 -1.94635257e-01
4.96486396e-01 3.73679757e-01 6.49539888e-01 -1.03368056e+00
1.07626522e+00 1.30391741e+00 2.41620541e-01 6.90026879e-01
-1.22051394e+00 -4.28765982e-01 1.24353185e-01 2.50291377e-02
-1.51346886e+00 -5.16851664e-01 8.72865736e-01 -3.51278961e-01
5.90921283e-01 2.92031676e-01 5.35796344e-01 1.48505664e+00
9.52354819e-02 9.60425436e-01 1.29072428e+00 -4.14633632e-01
1.16999306e-01 4.69575524e-01 -3.80494058e-01 3.35533440e-01
1.10542603e-01 -2.85653353e-01 -3.95412683e-01 1.93434685e-01
5.25514185e-01 -1.52471811e-01 -3.91297609e-01 -3.52588743e-01
-1.59718013e+00 7.55312264e-01 1.49147704e-01 3.99319917e-01
-4.90486026e-01 4.78260703e-02 3.17122906e-01 1.96563117e-02
4.00883257e-01 1.86305225e-01 6.92365542e-02 2.62377441e-01
-1.07358456e+00 3.88445497e-01 5.66904128e-01 1.36404192e+00
6.83066070e-01 1.21070497e-01 -5.04882634e-01 8.34024966e-01
9.30927023e-02 9.65141892e-01 4.16277528e-01 -8.70419204e-01
7.36535907e-01 3.15979600e-01 3.89737308e-01 -1.16801083e+00
1.01282142e-01 -4.37662989e-01 -1.12202537e+00 -1.79407984e-01
1.38150886e-01 -6.61732033e-02 -8.50826085e-01 1.66986668e+00
2.31436312e-01 1.68503836e-01 5.03444672e-01 7.32191563e-01
9.25702453e-01 1.05130088e+00 4.79548834e-02 -3.65998954e-01
1.27610421e+00 -1.28938651e+00 -9.67824101e-01 -2.63111919e-01
2.77785689e-01 -1.04621744e+00 1.22741783e+00 1.70606762e-01
-1.07933545e+00 -6.72064424e-01 -9.51847672e-01 6.11030199e-02
-2.58677214e-01 3.73682320e-01 2.71559060e-02 4.62100059e-01
-1.25584841e+00 5.27212769e-02 -9.25071687e-02 -4.40280080e-01
2.24804536e-01 -3.33036065e-01 -3.44037533e-01 -3.53633523e-01
-1.16648710e+00 7.39980459e-01 5.00765741e-01 1.70420259e-02
-1.07558656e+00 -3.39147091e-01 -8.71137202e-01 -1.30312592e-01
5.22006571e-01 -9.34450686e-01 1.16301012e+00 -1.17601097e+00
-1.31812739e+00 7.76461422e-01 -2.27605343e-01 -5.00040650e-01
9.18380558e-01 1.24490686e-01 -3.88591886e-01 5.13569117e-01
5.30030012e-01 1.02564454e+00 1.09238350e+00 -1.88659739e+00
-4.27512914e-01 3.79477106e-02 3.15394066e-02 3.71747941e-01
-3.16855311e-01 8.41367915e-02 -7.67142236e-01 -7.60564864e-01
-2.64999449e-01 -8.98998618e-01 -2.82565981e-01 -3.56054306e-02
-6.79451108e-01 1.47912279e-01 9.16627705e-01 -4.74394113e-01
1.14947581e+00 -1.91577268e+00 4.78214845e-02 -1.77635342e-01
-1.45640709e-02 2.54479051e-01 -4.28925246e-01 7.08166063e-01
1.57428637e-01 1.63269073e-01 -5.24263084e-01 -7.30211139e-01
-7.31751770e-02 1.73784077e-01 -6.01808488e-01 -1.05942123e-01
2.83977807e-01 1.08838058e+00 -1.01955140e+00 -9.48380053e-01
5.11458457e-01 4.54407871e-01 -2.34452829e-01 3.67881984e-01
-6.08391881e-01 6.31533086e-01 -4.55325246e-01 1.97713971e-01
9.28545833e-01 -4.08767819e-01 7.82503486e-02 -3.91356468e-01
-3.18528182e-04 -4.45276767e-01 -1.06318545e+00 1.85524917e+00
-5.91282308e-01 5.18461108e-01 -2.66847491e-01 -8.27375233e-01
8.40360940e-01 2.49637470e-01 3.66980404e-01 -6.46259308e-01
-1.16437949e-01 1.75193995e-01 -4.90932584e-01 -8.48537087e-01
6.62676930e-01 -1.10756829e-01 -2.48102751e-02 4.91825670e-01
-6.43488988e-02 -6.40226305e-01 5.97552598e-01 4.14630473e-01
5.72640836e-01 2.08490998e-01 -4.54269499e-02 1.84341758e-01
7.85164237e-01 1.07912183e-01 1.00329064e-01 9.85221446e-01
1.28336726e-02 1.28777862e+00 4.01293874e-01 -2.84220785e-01
-1.48677480e+00 -1.04507172e+00 1.70477748e-01 1.28277838e-01
3.23922038e-01 -5.08406520e-01 -8.68590653e-01 -6.53790951e-01
-5.79621553e-01 1.00730634e+00 -5.71925700e-01 2.26684362e-01
-3.04943681e-01 -6.61774814e-01 4.76564646e-01 -5.35749160e-02
1.01671815e+00 -1.22904694e+00 -3.36393803e-01 2.39437222e-01
-7.40931630e-01 -1.78927779e+00 -7.59649277e-01 -6.14285409e-01
-3.83278280e-01 -7.06327140e-01 -1.31808996e+00 -9.67342436e-01
9.49946284e-01 6.74878836e-01 1.16083658e+00 -1.19876206e-01
-1.83395565e-01 3.62841278e-01 -4.74154472e-01 -2.16559887e-01
-8.67470622e-01 -3.25336844e-01 -2.48993218e-01 5.35660744e-01
-1.46936104e-01 -2.69074768e-01 -7.59028316e-01 2.29412705e-01
-1.60794580e+00 7.86480427e-01 8.62224936e-01 7.81787515e-01
6.94907844e-01 2.03892037e-01 7.12385595e-01 -7.68255711e-01
8.55139792e-01 -4.42275345e-01 -1.46336913e-01 6.67760372e-01
-2.68041372e-01 9.12724249e-03 9.77301598e-01 -5.78413546e-01
-1.44434774e+00 2.28655085e-01 7.92681938e-04 -5.86410165e-01
-1.85053363e-01 4.05200928e-01 -4.63881418e-02 2.24722594e-01
5.45574546e-01 1.13427603e+00 4.18158621e-02 -6.22802638e-02
6.39409065e-01 8.60645771e-01 5.52398562e-01 -5.62910855e-01
9.73377943e-01 7.15227485e-01 -1.55294403e-01 -9.43160832e-01
-8.06185305e-01 -1.05639823e-01 -4.72542375e-01 -4.72803622e-01
1.12435341e+00 -1.04307866e+00 -2.79604346e-02 6.74000978e-01
-1.47759748e+00 9.79425851e-04 -1.77568138e-01 1.49171114e-01
-6.54083431e-01 8.23740304e-01 -2.76334912e-01 -7.27407336e-01
-4.66975480e-01 -1.39106846e+00 1.37104559e+00 1.89814225e-01
4.38229740e-02 -7.28152514e-01 -2.38516644e-01 4.75566953e-01
5.26515901e-01 3.61028969e-01 5.99333763e-01 -1.64297029e-01
-8.77631664e-01 2.59897411e-02 -3.90815765e-01 5.17504811e-01
2.12960705e-01 7.19437823e-02 -7.66505718e-01 -1.34676293e-01
8.73108655e-02 -4.08713579e-01 6.63871586e-01 2.45682791e-01
1.02409601e+00 -2.31766298e-01 -1.02321975e-01 2.88113952e-01
1.60549426e+00 6.88054934e-02 7.38574684e-01 2.39058882e-01
5.79468012e-01 7.97514200e-01 7.73804545e-01 2.42689520e-01
5.32774568e-01 6.63613319e-01 1.78564891e-01 -2.81130910e-01
-3.60051304e-01 -6.98222876e-01 3.52138340e-01 7.64784694e-01
4.54408079e-01 -5.59247613e-01 -7.67463982e-01 7.79560447e-01
-1.92840612e+00 -1.16848624e+00 -3.23155373e-01 1.89140487e+00
7.37391353e-01 7.28180334e-02 -2.16867462e-01 -2.36099869e-01
8.74594569e-01 3.39540809e-01 -4.69246030e-01 -3.33294063e-03
-5.72169185e-01 -3.73561293e-01 2.35425830e-01 2.30218738e-01
-8.82318497e-01 1.05121398e+00 5.71187210e+00 1.21810424e+00
-1.04251277e+00 -4.20705378e-02 8.75946820e-01 9.48909670e-02
-6.09892130e-01 -4.75378297e-02 -5.86366236e-01 6.72595978e-01
7.00891078e-01 -5.01234114e-01 1.51147246e-01 5.95622420e-01
6.68881834e-01 -1.83256403e-01 -7.52636313e-01 1.06977773e+00
5.49896955e-01 -1.51502359e+00 8.98095012e-01 -1.51211768e-01
1.30238533e+00 -2.77071655e-01 3.46449763e-01 5.00274897e-02
1.34701356e-02 -8.26428413e-01 9.69187737e-01 6.27265811e-01
9.64010060e-01 -3.55258495e-01 6.04205489e-01 6.27865076e-01
-1.02829719e+00 1.41245067e-01 -3.86074126e-01 4.01115775e-01
4.49703336e-01 5.98548651e-01 -8.86987984e-01 8.69626701e-01
4.04261738e-01 7.12606668e-01 -7.17037976e-01 8.94651234e-01
-1.23641603e-01 1.49013072e-01 3.93226594e-02 -1.51913494e-01
3.09805632e-01 -3.91686797e-01 5.27249873e-01 1.08010268e+00
6.98351026e-01 -4.32691313e-02 5.98567650e-02 1.12830770e+00
-7.71136209e-02 3.76384765e-01 -1.10864472e+00 -4.77880016e-02
2.71948576e-01 1.18887925e+00 -7.73661673e-01 -7.82532334e-01
-5.79265058e-01 1.28034079e+00 -1.24800503e-02 4.70784277e-01
-9.87223983e-01 -2.14040294e-01 2.68505584e-03 3.41921882e-03
2.63982147e-01 -1.18327171e-01 -3.39483581e-02 -1.53427172e+00
3.79697263e-01 -1.06254745e+00 -1.53068215e-01 -1.55043936e+00
-1.24403954e+00 1.00550103e+00 1.40648082e-01 -1.61513984e+00
-3.47253084e-01 -1.24285117e-01 -5.26040733e-01 8.10287058e-01
-1.71521986e+00 -1.49483275e+00 -5.73802829e-01 5.95849514e-01
1.07538199e+00 -3.51828299e-02 5.86147785e-01 2.45522127e-01
-3.89884919e-01 1.40940055e-01 2.39247829e-01 -1.61117673e-01
8.51517141e-01 -1.04568005e+00 5.73172033e-01 9.17449534e-01
1.81565523e-01 3.90900761e-01 8.26965153e-01 -6.90862477e-01
-1.13358676e+00 -1.38856018e+00 9.49057162e-01 -3.04173738e-01
4.69965547e-01 -3.53089184e-01 -6.80337012e-01 2.87861913e-01
6.05007827e-01 -3.03480893e-01 2.42095619e-01 -1.05629289e+00
4.36241180e-02 -9.17677507e-02 -9.76989150e-01 1.00103629e+00
9.66957033e-01 -3.80565673e-01 -3.81811202e-01 5.37806451e-01
9.76992846e-01 -2.80782580e-01 -5.28635502e-01 1.68980777e-01
2.43176892e-01 -1.07602310e+00 9.57768202e-01 -8.65600854e-02
9.67450500e-01 -5.31367421e-01 -5.50036356e-02 -1.23563778e+00
2.22723633e-01 -7.02583671e-01 3.66977215e-01 1.38341653e+00
4.03334200e-01 -2.96656162e-01 5.47786295e-01 4.05152708e-01
9.25677121e-02 -4.85083550e-01 -4.74736810e-01 -7.11294115e-01
-6.41931444e-02 -3.20976198e-01 6.72788560e-01 8.30853879e-01
-6.19462550e-01 4.75819379e-01 -7.58857548e-01 -4.95183505e-02
8.31884146e-01 2.95350850e-01 1.03046632e+00 -5.94950259e-01
-5.88056035e-02 -1.57370016e-01 -8.55042264e-02 -1.35145032e+00
1.51739836e-01 -6.19699955e-01 2.72248946e-02 -1.83006990e+00
5.45573294e-01 -1.36012495e-01 2.99512714e-01 -1.34318620e-01
-2.91197121e-01 3.81101340e-01 3.79130810e-01 3.40490401e-01
-7.05365300e-01 7.74658084e-01 1.87343764e+00 -4.03944463e-01
1.04120888e-01 -2.90001303e-01 -7.03508735e-01 4.89510834e-01
6.27859771e-01 -2.36604884e-01 -8.20812345e-01 -6.20868385e-01
2.44128108e-02 3.40213299e-01 2.92645782e-01 -1.00810361e+00
2.37855852e-01 -2.40272060e-01 3.35397780e-01 -8.07266295e-01
4.35303897e-01 -8.89884293e-01 3.92395854e-01 3.06086093e-01
-5.39054453e-01 2.00552195e-02 -8.99472982e-02 7.36929238e-01
-4.87787664e-01 -2.52969533e-01 6.17158592e-01 -3.28273863e-01
-7.45655954e-01 2.20379859e-01 -2.49191090e-01 7.64320120e-02
1.34602368e+00 -2.10214466e-01 -3.88856977e-01 -8.65380108e-01
-5.98279893e-01 3.64849359e-01 6.17633700e-01 7.27588534e-01
9.40953612e-01 -1.41086972e+00 -1.16157126e+00 1.76368743e-01
3.87561589e-01 6.04018867e-02 6.35054231e-01 4.59710300e-01
-6.90967143e-01 5.53930938e-01 -1.33814499e-01 -7.92417586e-01
-1.24541652e+00 7.54550636e-01 1.86337382e-01 -3.49033862e-01
-5.89649081e-01 3.82265121e-01 6.42375052e-01 -3.24062258e-01
-1.31660640e-01 -2.29547828e-01 -2.23186791e-01 -1.59635127e-01
5.17767668e-01 -2.83307076e-01 -3.52713943e-01 -1.06447554e+00
1.21128812e-01 7.25165844e-01 -6.04976043e-02 -3.92623246e-01
1.14163733e+00 -5.66814542e-01 1.00197554e-01 2.44632468e-01
1.24552572e+00 -1.33239508e-01 -1.26667917e+00 -3.70405465e-01
-3.88174683e-01 -5.51705658e-01 -3.48662198e-01 -6.80760145e-01
-8.57529700e-01 7.88121998e-01 3.51617277e-01 2.15324059e-01
1.20618379e+00 -1.21078402e-01 9.75172162e-01 2.21389502e-01
2.93501854e-01 -9.19507027e-01 5.17000377e-01 1.32709831e-01
1.26051128e+00 -1.35499620e+00 -2.39043698e-01 -5.01755893e-01
-1.09730148e+00 9.90499198e-01 6.36978686e-01 1.31477758e-01
1.40841022e-01 -3.67261246e-02 1.26948431e-01 3.08127934e-03
-9.20233905e-01 -2.24194452e-01 1.09154530e-01 6.90317690e-01
1.90247204e-02 -1.71900198e-01 -3.72599930e-01 2.78779894e-01
-6.37988970e-02 3.74165662e-02 9.55563247e-01 6.68011308e-01
-3.49315584e-01 -1.11186385e+00 -4.37186986e-01 2.30189085e-01
-2.65136540e-01 -2.53939986e-01 -4.70830023e-01 6.00787222e-01
-6.15162142e-02 1.20540571e+00 -1.47345439e-01 -2.67035514e-01
3.24773639e-01 -1.19013019e-01 3.21879923e-01 -5.13994217e-01
-1.51857346e-01 1.24898162e-02 7.46100098e-02 -1.70770809e-01
-6.59287393e-01 -4.79983538e-01 -8.21994543e-01 1.26590929e-03
-9.19276252e-02 1.32510290e-01 7.94295192e-01 7.58856714e-01
4.14760202e-01 5.16598403e-01 8.20892751e-01 -7.53747523e-01
-6.07732177e-01 -8.54874671e-01 -4.16000396e-01 7.87507176e-01
2.56607711e-01 -2.21475512e-01 -3.25421810e-01 5.33376634e-01] | [11.177918434143066, 0.6114633083343506] |
1ba06dd4-b3c7-4d0b-8b5d-c5f0479ad96c | deep-temporal-contrastive-clustering | 2212.14366 | null | https://arxiv.org/abs/2212.14366v1 | https://arxiv.org/pdf/2212.14366v1.pdf | Deep Temporal Contrastive Clustering | Recently the deep learning has shown its advantage in representation learning and clustering for time series data. Despite the considerable progress, the existing deep time series clustering approaches mostly seek to train the deep neural network by some instance reconstruction based or cluster distribution based objective, which, however, lack the ability to exploit the sample-wise (or augmentation-wise) contrastive information or even the higher-level (e.g., cluster-level) contrastiveness for learning discriminative and clustering-friendly representations. In light of this, this paper presents a deep temporal contrastive clustering (DTCC) approach, which for the first time, to our knowledge, incorporates the contrastive learning paradigm into the deep time series clustering research. Specifically, with two parallel views generated from the original time series and their augmentations, we utilize two identical auto-encoders to learn the corresponding representations, and in the meantime perform the cluster distribution learning by incorporating a k-means objective. Further, two levels of contrastive learning are simultaneously enforced to capture the instance-level and cluster-level contrastive information, respectively. With the reconstruction loss of the auto-encoder, the cluster distribution loss, and the two levels of contrastive losses jointly optimized, the network architecture is trained in a self-supervised manner and the clustering result can thereby be obtained. Experiments on a variety of time series datasets demonstrate the superiority of our DTCC approach over the state-of-the-art. | ['Chang-Dong Wang', 'Dong Huang', 'Ying Zhong'] | 2022-12-29 | null | null | null | null | ['time-series-clustering'] | ['time-series'] | [-1.93020925e-01 -4.16807830e-01 1.25061601e-01 -4.79001760e-01
-7.90855527e-01 -3.93862635e-01 7.60855258e-01 2.02521265e-01
-2.68192291e-01 2.01997980e-01 -5.17511591e-02 1.06980562e-01
-4.99669224e-01 -5.94458044e-01 -6.65765524e-01 -1.16653132e+00
-6.34379864e-01 4.38398272e-01 -3.30077022e-01 8.83991942e-02
-6.10658750e-02 4.74321455e-01 -1.79204881e+00 2.33312368e-01
8.31368148e-01 1.32998025e+00 1.79964870e-01 3.41604888e-01
5.40354699e-02 9.09892261e-01 -4.66699153e-01 -4.12873514e-02
3.60258281e-01 -5.17893016e-01 -3.99111748e-01 3.37036014e-01
-6.80690631e-02 2.48043835e-02 -6.52486384e-01 8.87587667e-01
4.60654885e-01 4.75881189e-01 4.37490404e-01 -1.28158271e+00
-6.54334247e-01 4.82413679e-01 -8.01085234e-01 2.02546909e-01
-2.68147886e-01 1.07323729e-01 9.42285419e-01 -7.22325087e-01
3.36753577e-01 1.06706595e+00 7.16622055e-01 2.19200760e-01
-1.28239274e+00 -5.33619940e-01 3.39230120e-01 4.24636841e-01
-1.48674953e+00 1.22829564e-02 1.40877831e+00 -5.40564418e-01
7.65269399e-01 1.28839053e-02 6.34592175e-01 1.07345092e+00
-9.98013616e-02 8.63929868e-01 1.19382441e+00 -2.53093749e-01
5.20273149e-01 -1.83331653e-01 -2.88592130e-02 3.26350629e-01
-4.48491663e-01 1.90867066e-01 7.78894275e-02 1.26108244e-01
5.64652860e-01 5.98971307e-01 -1.83082253e-01 -4.76281881e-01
-1.07638085e+00 8.49611998e-01 8.07154953e-01 7.82646418e-01
-4.70566750e-01 1.47119567e-01 8.24435771e-01 3.42886537e-01
8.06332231e-01 3.04158628e-01 -4.83955115e-01 -4.09519672e-02
-1.24138319e+00 -2.34153811e-02 4.63485986e-01 6.33267105e-01
7.80079722e-01 2.82969922e-01 -2.15341434e-01 8.20701182e-01
-9.29413661e-02 3.86966057e-02 8.92021477e-01 -7.53830612e-01
4.17147428e-01 6.31349742e-01 -3.45956117e-01 -1.07200408e+00
-3.94723445e-01 -9.33596075e-01 -1.35030520e+00 1.79339871e-01
1.85859352e-01 -5.59370369e-02 -8.62658620e-01 2.15655088e+00
1.43226892e-01 6.32047355e-01 -6.23414740e-02 7.32293546e-01
1.75009161e-01 8.73510778e-01 -1.27045378e-01 -6.31035805e-01
9.62679505e-01 -8.65182579e-01 -9.36239183e-01 3.68674606e-01
3.48367184e-01 -5.21449983e-01 9.79866445e-01 4.12533075e-01
-1.12674499e+00 -8.53447735e-01 -1.09651816e+00 1.95664212e-01
-5.45919120e-01 2.04414681e-01 4.96461987e-01 1.90542817e-01
-9.95675504e-01 8.95908713e-01 -1.45658267e+00 -2.22058043e-01
5.24410427e-01 2.03404620e-01 -1.22564264e-01 -2.81825941e-02
-1.07006168e+00 5.38927257e-01 4.08606619e-01 2.18323648e-01
-9.03482199e-01 -8.65711987e-01 -6.80826724e-01 2.50850886e-01
1.55884996e-01 -3.99129540e-01 8.25123072e-01 -1.33337545e+00
-1.40798736e+00 7.26358235e-01 1.29827008e-01 -5.92283607e-01
6.67560160e-01 8.19291174e-02 -4.84713525e-01 4.17482734e-01
1.60487220e-01 4.76850480e-01 8.96977901e-01 -1.35529578e+00
-4.29306388e-01 -5.69191396e-01 -1.85214162e-01 1.29546538e-01
-6.34764969e-01 -1.72080845e-01 -4.33932185e-01 -9.93481159e-01
-6.40576184e-02 -7.26292372e-01 -3.26337606e-01 -4.70103957e-02
-1.71057150e-01 -3.02790910e-01 1.07504559e+00 -7.48300910e-01
1.30901468e+00 -2.48602009e+00 4.03556764e-01 2.56802142e-01
1.98782548e-01 2.80823912e-02 -1.31638438e-01 5.72102904e-01
-3.95285189e-01 -1.38111964e-01 -6.93001270e-01 -8.63084495e-01
7.86433592e-02 3.02779734e-01 -1.84795931e-01 7.26074874e-01
4.30097729e-01 6.67684674e-01 -9.85453546e-01 -2.30494723e-01
6.54404819e-01 8.30653429e-01 -5.11195481e-01 3.52829039e-01
-4.80138091e-03 7.68482685e-01 -2.02550486e-01 1.96531475e-01
6.40735388e-01 -2.19759122e-01 1.77961737e-01 -2.38560304e-01
-3.29233021e-01 -4.91369283e-04 -8.49085152e-01 1.96516550e+00
-4.21122521e-01 4.53449667e-01 -5.88006787e-02 -1.70836139e+00
8.28383029e-01 4.97404218e-01 1.25148904e+00 -8.18503022e-01
-7.73761049e-02 3.11371446e-01 -3.48228998e-02 -1.94856480e-01
1.00577064e-01 -3.45188767e-01 2.16146484e-02 2.90540665e-01
1.43329352e-01 2.28593796e-01 5.38736768e-02 -1.39110178e-01
1.00713074e+00 1.80649906e-01 -1.08856119e-01 -2.14135692e-01
4.69186127e-01 -6.61967993e-02 4.99694288e-01 3.54084343e-01
-1.29445165e-01 8.29916358e-01 3.43780309e-01 -3.51704329e-01
-1.29370642e+00 -1.09156418e+00 -2.10597411e-01 8.07647049e-01
-2.39248648e-01 -1.50036216e-01 -6.62931681e-01 -4.47405368e-01
-1.81064159e-01 5.14683783e-01 -8.20946515e-01 -4.09157544e-01
-7.96497941e-01 -8.71191263e-01 3.00829619e-01 6.37245893e-01
4.95053083e-01 -9.58709955e-01 -4.79851544e-01 4.96431261e-01
-7.88168609e-02 -8.41333330e-01 -3.48012745e-01 5.77724397e-01
-1.01069546e+00 -7.64905632e-01 -8.21625233e-01 -8.61099660e-01
4.91035521e-01 1.60240903e-01 9.32317734e-01 -1.19692229e-01
-3.55758041e-01 3.79936963e-01 -5.64533174e-01 -6.82628602e-02
2.70332228e-02 5.24882749e-02 -1.98419183e-01 4.61690366e-01
2.43729532e-01 -1.26092482e+00 -7.73794830e-01 1.22314198e-02
-1.17413545e+00 -2.20725954e-01 2.84776926e-01 1.00301743e+00
7.06453323e-01 7.28813112e-01 7.25672483e-01 -4.65800703e-01
3.75258833e-01 -7.54571497e-01 -4.48862672e-01 -3.83912660e-02
-5.77151120e-01 -1.15624428e-01 1.21728253e+00 -5.11444747e-01
-8.40708971e-01 -1.45150097e-02 -8.88209939e-02 -1.29937577e+00
-7.99434185e-02 8.89980674e-01 -1.23975620e-01 5.21672606e-01
3.18839401e-01 5.08100331e-01 1.98236570e-01 -5.58896899e-01
4.94403213e-01 1.94741115e-01 6.29852712e-01 -6.35271549e-01
8.11545670e-01 8.81135285e-01 -1.71128541e-01 -4.15600300e-01
-7.77256608e-01 -6.52441144e-01 -7.66175628e-01 -2.29662076e-01
9.89382625e-01 -9.89039779e-01 -4.25704747e-01 5.24090469e-01
-7.24906921e-01 -3.01552594e-01 -7.15510070e-01 5.72922409e-01
-7.78214157e-01 4.15805757e-01 -6.51016474e-01 -8.87740433e-01
-2.88521759e-02 -9.15037513e-01 1.05021644e+00 -1.60692528e-01
2.66400367e-01 -1.21473777e+00 1.10452734e-01 -1.61805615e-01
3.30765545e-01 8.28852117e-01 9.47456896e-01 -7.00875580e-01
-3.13197225e-01 -2.39774972e-01 -2.57272795e-02 6.22249842e-01
2.52803534e-01 5.05283996e-02 -1.02504683e+00 -5.59936702e-01
4.39436197e-01 -2.03327656e-01 8.90096784e-01 6.04077280e-01
1.82236552e+00 -3.28986168e-01 -1.58240810e-01 7.05974221e-01
1.59909058e+00 3.58067334e-01 6.42508566e-01 1.69920340e-01
8.41673970e-01 7.67357945e-01 2.80531764e-01 7.91242003e-01
4.72172201e-01 7.03743875e-01 5.21044970e-01 -1.95015773e-01
-2.48468146e-02 -1.16844147e-01 2.33440876e-01 1.44571745e+00
-1.89152554e-01 -4.84056659e-02 -7.96518266e-01 7.56070197e-01
-2.07325435e+00 -1.39799881e+00 1.57775849e-01 2.12591648e+00
6.98803484e-01 -1.87704399e-01 2.23444983e-01 6.40697181e-01
8.62567902e-01 4.05678302e-01 -7.16267228e-01 -6.79640891e-03
-2.02039972e-01 1.36911973e-01 -3.68246585e-02 1.72641482e-02
-1.38157070e+00 3.92880946e-01 4.84177876e+00 9.28464830e-01
-1.35146177e+00 2.52580464e-01 7.62914360e-01 -1.40969485e-01
-2.46612608e-01 -1.83266237e-01 8.65595341e-02 7.11430311e-01
9.60603714e-01 -4.78797480e-02 7.52158701e-01 6.34736359e-01
3.31738889e-01 6.65746212e-01 -1.27366531e+00 1.13177395e+00
-1.46156371e-01 -1.08916306e+00 -2.06425086e-01 1.82464465e-01
8.36328685e-01 -6.53176103e-03 3.87400389e-01 6.41211748e-01
8.38054344e-02 -9.11275566e-01 9.57498133e-01 6.79124773e-01
5.64519584e-01 -1.07034838e+00 6.52592123e-01 3.16196114e-01
-1.56241786e+00 -4.64458197e-01 -1.40215993e-01 -8.75963196e-02
1.01732947e-01 8.02408218e-01 -7.97493979e-02 1.06688869e+00
8.31637025e-01 1.40294218e+00 -3.07082891e-01 7.87665606e-01
3.69241834e-01 6.93961561e-01 -1.98193595e-01 4.55419362e-01
6.52983308e-01 -5.49721897e-01 4.11202073e-01 1.07938874e+00
4.76840705e-01 -1.56121522e-01 4.64065343e-01 9.89616513e-01
-4.82147224e-02 -1.50426388e-01 -3.04313123e-01 -8.08278844e-02
4.44819570e-01 1.15096211e+00 -6.33681417e-01 -3.62779021e-01
-3.40419620e-01 9.48736191e-01 4.48224217e-01 4.22015190e-01
-8.96423340e-01 -2.47622475e-01 6.04117692e-01 -1.79161679e-03
6.66995227e-01 -2.73779035e-01 -2.56592304e-01 -1.24096930e+00
3.26197177e-01 -8.01618695e-01 6.04761183e-01 -4.36516345e-01
-1.93248343e+00 6.98296070e-01 -2.75952257e-02 -1.63509548e+00
-3.34311336e-01 -3.14840317e-01 -7.50359774e-01 6.17237449e-01
-1.63103306e+00 -1.17085099e+00 -1.60690099e-01 9.17274356e-01
5.80939770e-01 -1.11702830e-01 5.24387479e-01 7.33072877e-01
-8.01263988e-01 4.84384835e-01 6.68559909e-01 1.90998361e-01
3.54645073e-01 -1.44888616e+00 -1.15226172e-01 7.19142318e-01
-1.46033540e-01 5.50769150e-01 3.96572590e-01 -1.09265178e-01
-1.19680476e+00 -1.63273811e+00 3.71245712e-01 -1.42185926e-01
8.66536498e-01 -4.90486562e-01 -1.16613245e+00 4.76956934e-01
2.65722573e-01 2.30093136e-01 6.73524320e-01 -2.10631751e-02
-5.02673149e-01 -4.91613418e-01 -1.04463673e+00 3.60960126e-01
9.03006375e-01 -8.03901672e-01 -5.41218042e-01 3.40653300e-01
7.30122328e-01 1.21289060e-01 -1.20787811e+00 5.11013269e-01
2.72168785e-01 -9.83803511e-01 9.63073313e-01 -4.60245222e-01
5.82397103e-01 -4.20964330e-01 -2.86043674e-01 -1.28243804e+00
-4.35203224e-01 -4.66196030e-01 -2.75817841e-01 1.52865684e+00
-7.22272396e-02 -4.09503281e-01 4.69848424e-01 1.71906278e-02
-6.74605072e-01 -8.63932490e-01 -1.13358533e+00 -9.92736101e-01
3.25740576e-01 -4.21791762e-01 6.46083534e-01 1.45266640e+00
-1.32671535e-01 1.10946655e-01 -3.65433007e-01 1.82189643e-01
6.64452970e-01 3.48666579e-01 2.69921750e-01 -1.33835614e+00
-2.85921335e-01 -8.02221715e-01 -3.61418307e-01 -7.59800971e-01
2.79919505e-01 -1.03051126e+00 -1.77175254e-02 -1.17783689e+00
7.95697123e-02 -4.52455550e-01 -8.28524232e-01 1.63616687e-01
-4.95007858e-02 -2.12093458e-01 2.31549278e-01 5.44775426e-01
-5.93741357e-01 1.03244495e+00 1.00504315e+00 -2.23714247e-01
-1.32850364e-01 -1.13844208e-01 -3.95789295e-01 3.71916890e-01
7.33627975e-01 -3.91142190e-01 -6.33261383e-01 -2.94250131e-01
-1.60059303e-01 1.52424827e-01 4.68570083e-01 -1.15143001e+00
3.17238390e-01 2.11513788e-01 3.88976276e-01 -8.01080048e-01
4.22398865e-01 -1.06408310e+00 1.87336564e-01 4.40883070e-01
-4.60926592e-01 2.17047989e-01 7.04982281e-02 8.81972432e-01
-7.45898426e-01 3.11668545e-01 8.67058814e-01 -2.43582636e-01
-4.24969494e-01 7.61930048e-01 -2.79488564e-01 -6.49538562e-02
8.98955584e-01 -2.95943260e-01 1.29549205e-01 -2.67001331e-01
-7.58744001e-01 3.13509315e-01 3.29241782e-01 3.25736880e-01
3.99145007e-01 -1.72268736e+00 -6.67608559e-01 1.84024453e-01
1.54603794e-01 1.00102834e-01 5.87472856e-01 1.12919843e+00
-8.53153393e-02 1.71491001e-02 -3.72261256e-01 -9.23627257e-01
-3.37958604e-01 1.15810847e+00 4.26538795e-01 -7.07420588e-01
-8.20578158e-01 4.33219790e-01 4.20274436e-01 -5.30254900e-01
3.80012184e-01 -2.54287452e-01 -2.35480472e-01 2.47710168e-01
3.42670500e-01 4.18672085e-01 1.70274571e-01 -3.73877764e-01
-3.29213798e-01 5.99963784e-01 4.47515771e-02 -1.07749857e-01
1.82142448e+00 -2.01750815e-01 -8.70756805e-02 9.15992618e-01
1.78766704e+00 -5.81070602e-01 -1.71010888e+00 -2.59250283e-01
4.15806584e-02 -2.70999372e-01 -9.19383690e-02 -4.89692360e-01
-1.45063317e+00 1.24834609e+00 7.28066087e-01 5.33593833e-01
1.68836915e+00 -1.62369639e-01 7.42388368e-01 -9.79531705e-02
1.95643961e-01 -1.16050172e+00 3.37909222e-01 2.31581852e-01
7.96247900e-01 -1.14369702e+00 -4.12199557e-01 1.41198486e-01
-4.76979047e-01 9.48834062e-01 3.72480780e-01 -5.53382337e-01
8.55103016e-01 1.79957777e-01 -1.82108238e-01 -2.57719904e-01
-8.21196258e-01 -2.31683075e-01 1.95700675e-01 5.84712088e-01
4.46233988e-01 1.26321197e-01 -1.08444445e-01 5.08831143e-01
1.01275310e-01 -3.43940645e-01 7.74887055e-02 6.75238490e-01
1.27183162e-02 -7.77513266e-01 -1.63228616e-01 3.32632929e-01
-4.62051123e-01 1.55653730e-01 -6.53295219e-02 8.24573338e-01
1.51635736e-01 9.07681823e-01 4.10703540e-01 -3.60273659e-01
3.11896712e-01 -1.41649902e-01 1.97521344e-01 -2.71053761e-01
-8.87440205e-01 4.13464636e-01 -5.95004261e-01 -4.88070875e-01
-8.36557627e-01 -7.68761218e-01 -1.20124769e+00 -2.75539935e-01
-9.46405232e-02 1.79735765e-01 4.58020031e-01 7.14976609e-01
3.99684638e-01 8.00440669e-01 1.32266104e+00 -1.02372789e+00
-5.51849365e-01 -8.72114003e-01 -7.91846573e-01 8.51567924e-01
4.96174812e-01 -5.56514978e-01 -5.12855053e-01 1.19481437e-01] | [7.447592258453369, 3.0917491912841797] |
94c67a23-b136-401a-aa71-bcc6b3082152 | pseudo-relevance-feedback-for-multiple | 2106.11251 | null | https://arxiv.org/abs/2106.11251v2 | https://arxiv.org/pdf/2106.11251v2.pdf | Pseudo-Relevance Feedback for Multiple Representation Dense Retrieval | Pseudo-relevance feedback mechanisms, from Rocchio to the relevance models, have shown the usefulness of expanding and reweighting the users' initial queries using information occurring in an initial set of retrieved documents, known as the pseudo-relevant set. Recently, dense retrieval -- through the use of neural contextual language models such as BERT for analysing the documents' and queries' contents and computing their relevance scores -- has shown a promising performance on several information retrieval tasks still relying on the traditional inverted index for identifying documents relevant to a query. Two different dense retrieval families have emerged: the use of single embedded representations for each passage and query (e.g. using BERT's [CLS] token), or via multiple representations (e.g. using an embedding for each token of the query and document). In this work, we conduct the first study into the potential for multiple representation dense retrieval to be enhanced using pseudo-relevance feedback. In particular, based on the pseudo-relevant set of documents identified using a first-pass dense retrieval, we extract representative feedback embeddings (using KMeans clustering) -- while ensuring that these embeddings discriminate among passages (based on IDF) -- which are then added to the query representation. These additional feedback embeddings are shown to both enhance the effectiveness of a reranking as well as an additional dense retrieval operation. Indeed, experiments on the MSMARCO passage ranking dataset show that MAP can be improved by upto 26% on the TREC 2019 query set and 10% on the TREC 2020 query set by the application of our proposed ColBERT-PRF method on a ColBERT dense retrieval approach. | ['Iadh Ounis', 'Nicola Tonellotto', 'Craig Macdonald', 'Xiao Wang'] | 2021-06-21 | null | null | null | null | ['passage-ranking'] | ['natural-language-processing'] | [ 2.06199646e-01 -7.35885426e-02 -6.04116693e-02 -3.90427411e-02
-1.25061524e+00 -6.77387774e-01 1.04998565e+00 8.19250405e-01
-9.84878778e-01 4.79568422e-01 7.20126212e-01 -1.99448206e-02
-9.10233021e-01 -7.05689490e-01 -4.75180656e-01 -5.56810915e-01
-2.69379884e-01 6.92696929e-01 5.66880405e-01 -6.92627668e-01
7.53119826e-01 5.06204009e-01 -1.91658294e+00 5.37794948e-01
5.94118237e-01 8.11841011e-01 2.18182102e-01 8.51581156e-01
-1.96012422e-01 2.43353888e-01 -5.92381895e-01 4.34562080e-02
2.63842732e-01 -8.35301578e-02 -1.01095080e+00 -4.76452827e-01
2.92844772e-01 -3.59436870e-01 -3.65838021e-01 6.28491640e-01
5.39827704e-01 4.70961928e-01 7.51202345e-01 -5.52735925e-01
-7.49771237e-01 4.38937008e-01 -1.38150826e-01 5.53512692e-01
7.16052353e-01 -5.94880998e-01 1.35420012e+00 -1.18890250e+00
7.66510129e-01 1.16859686e+00 1.74051434e-01 2.38963708e-01
-1.09596717e+00 -1.98376849e-01 -1.84690446e-01 2.42501542e-01
-1.53185940e+00 -2.90671229e-01 5.04243016e-01 -2.14051068e-01
1.19269907e+00 5.98319888e-01 3.84570688e-01 5.76271355e-01
-6.35660216e-02 6.46910906e-01 7.12262034e-01 -1.01106346e+00
2.51038373e-01 3.80470872e-01 5.54514587e-01 9.74866971e-02
2.00015754e-01 3.29884529e-01 -3.04572344e-01 -5.10548711e-01
7.55318403e-02 1.92683488e-02 -3.37046057e-01 -1.92536145e-01
-7.49944985e-01 1.00186074e+00 7.22228169e-01 9.08464193e-01
-5.09592712e-01 -1.22720916e-02 5.02643168e-01 5.67687511e-01
4.37795252e-01 7.93656886e-01 -1.64177284e-01 1.20813489e-01
-1.03255010e+00 5.27011275e-01 4.99381393e-01 5.81383228e-01
1.00177288e+00 -5.77752709e-01 -5.92225492e-01 9.29125547e-01
5.76479256e-01 3.14921647e-01 9.65875685e-01 -3.09917331e-01
4.61205959e-01 8.29870999e-01 2.56140083e-01 -1.23547339e+00
-1.85434848e-01 -4.64776754e-01 -3.97963464e-01 -2.96710938e-01
-1.45339504e-01 3.61761779e-01 -9.69071925e-01 1.43187916e+00
2.16578424e-01 -3.45027834e-01 8.19512978e-02 7.71533310e-01
4.60877001e-01 8.00576806e-01 4.00193259e-02 -1.28572792e-01
1.27171552e+00 -6.13224685e-01 -4.67525095e-01 1.45290211e-01
9.56722617e-01 -9.23777819e-01 9.34373021e-01 7.20565245e-02
-8.44599545e-01 -6.78203583e-01 -1.07285500e+00 -2.03073800e-01
-9.58779395e-01 -5.35566173e-02 1.60347953e-01 6.78318381e-01
-1.58286667e+00 4.17378634e-01 -2.24839672e-01 -5.12676358e-01
-2.75635034e-01 5.47635198e-01 -3.38141412e-01 -3.78658473e-01
-1.63508153e+00 8.58219147e-01 6.47748649e-01 -1.43777117e-01
-6.85054243e-01 -4.91131634e-01 -6.99122906e-01 3.57819051e-01
-1.92154963e-02 -4.51619118e-01 8.51268053e-01 -6.48194015e-01
-8.51831317e-01 4.95004505e-01 -1.51473165e-01 -5.08173764e-01
5.93155064e-02 -5.34789026e-01 -3.15482348e-01 7.55031884e-01
1.93962187e-01 8.39370251e-01 6.89994991e-01 -1.20431757e+00
-7.34019876e-01 -4.51456726e-01 3.49208504e-01 5.24838746e-01
-8.44264150e-01 2.51971483e-01 -7.48266220e-01 -5.44757426e-01
-7.75539577e-02 -1.01105917e+00 -1.16965629e-01 -5.36711454e-01
6.43130913e-02 -6.05556428e-01 7.97881603e-01 -5.77031314e-01
1.78386593e+00 -1.97361207e+00 8.75910837e-03 8.39883745e-01
9.88013893e-02 4.90424961e-01 -4.58364457e-01 8.74218583e-01
-1.37920575e-02 4.24314708e-01 1.51297569e-01 -1.02335349e-01
-4.58481759e-02 3.77179049e-02 -3.40828478e-01 9.80396196e-02
4.80845273e-02 8.79139304e-01 -9.28096414e-01 -4.68468696e-01
1.05438039e-01 7.14466691e-01 -5.69762349e-01 2.51086503e-02
9.71872360e-02 -2.37002358e-01 -4.63590384e-01 1.61943465e-01
2.30610952e-01 -1.09391578e-01 -7.41965994e-02 -1.70126781e-02
2.63598450e-02 3.07682127e-01 -9.10011649e-01 1.40603423e+00
-4.64860320e-01 5.32547057e-01 -4.31355685e-01 -7.36788988e-01
8.91724527e-01 6.23736858e-01 5.44549048e-01 -1.16257524e+00
-1.62119836e-01 4.75903243e-01 -2.24596351e-01 -3.13159615e-01
1.30879056e+00 8.07432756e-02 -1.82660982e-01 6.39570296e-01
2.13133749e-02 2.72536933e-01 7.20387399e-01 7.10073829e-01
1.16228771e+00 -1.78650573e-01 1.00851417e-01 -3.18082541e-01
7.21523464e-01 -2.72015650e-02 -2.58299857e-01 9.90470767e-01
2.21592784e-01 8.38006616e-01 -3.05792689e-02 8.88645183e-03
-1.00216734e+00 -7.16722548e-01 -2.06987619e-01 1.30478644e+00
6.35617375e-02 -5.42779028e-01 -4.63796407e-01 -6.46572769e-01
1.60401955e-01 5.93298495e-01 -8.60858083e-01 -5.78752935e-01
-5.53963184e-01 -6.46833479e-01 4.35037911e-01 3.73696446e-01
-2.72466452e-03 -1.04033220e+00 -4.27153528e-01 2.77324796e-01
-2.17249632e-01 -4.45759326e-01 -4.46398884e-01 3.44533026e-01
-8.70194435e-01 -7.67260015e-01 -1.10307002e+00 -7.08036482e-01
6.48204088e-01 4.51272815e-01 9.32788730e-01 3.37330133e-01
-2.00954109e-01 1.03298867e+00 -9.68243659e-01 1.14273444e-01
-2.85554200e-01 3.41145575e-01 -3.48524004e-02 -2.15612445e-02
4.15732682e-01 -8.90005231e-02 -8.25819314e-01 1.76541135e-01
-1.56894183e+00 -7.06749976e-01 6.35037124e-01 8.31097126e-01
3.38010758e-01 -1.44086719e-01 8.59579265e-01 -6.34829104e-01
1.27725303e+00 -5.77857912e-01 -2.26339057e-01 5.59249461e-01
-1.06319511e+00 4.24729437e-01 2.10658655e-01 -1.82374090e-01
-9.38923001e-01 -4.85086203e-01 -1.86554134e-01 -2.38616467e-01
2.00281307e-01 8.66233885e-01 5.11113882e-01 -1.16165899e-01
1.04816532e+00 2.32076600e-01 -3.05461228e-01 -5.25513530e-01
4.07311320e-01 9.06098425e-01 9.58025735e-03 -4.97321934e-01
7.77572215e-01 1.25730515e-01 -2.62739509e-01 -7.83838630e-01
-3.38217109e-01 -1.50571871e+00 -6.18338943e-01 -1.77619502e-01
4.85609382e-01 -6.49692178e-01 -1.64946303e-01 -2.01659888e-01
-9.38616335e-01 3.96561682e-01 -4.64653403e-01 5.68615556e-01
-1.51609704e-01 6.44888461e-01 -5.71276903e-01 -7.75920093e-01
-6.04126334e-01 -9.48424578e-01 1.21718347e+00 2.19288975e-01
-2.83763468e-01 -1.01192534e+00 6.17776096e-01 1.44202322e-01
7.06585586e-01 -4.16184783e-01 1.21704245e+00 -1.17721510e+00
-3.09335202e-01 -8.72927964e-01 -2.45402306e-01 3.06372702e-01
8.08739290e-02 -3.27624768e-01 -9.92850125e-01 -5.77655554e-01
-2.70650566e-01 -3.38335156e-01 1.15019846e+00 -2.59286258e-02
4.07404393e-01 -1.95812732e-01 -2.76178509e-01 -2.52315760e-01
1.67678118e+00 2.37583175e-01 8.43987286e-01 5.45186222e-01
1.49712652e-01 5.98147869e-01 6.27022088e-01 2.96106696e-01
-1.27375275e-01 6.74747825e-01 2.04354122e-01 2.49574691e-01
-5.36539406e-02 -1.83759108e-01 2.25430712e-01 9.32605863e-01
2.29031257e-02 -3.29377770e-01 -6.86442733e-01 8.55837047e-01
-1.57051075e+00 -8.35105956e-01 2.83792287e-01 2.68777251e+00
6.94872200e-01 5.73940910e-02 -6.87969476e-02 4.51506466e-01
5.71510971e-01 -2.22101100e-02 1.56944782e-01 -4.50251609e-01
1.06226407e-01 4.41275626e-01 1.95388794e-01 7.25690901e-01
-8.83407116e-01 5.64532459e-01 5.68964815e+00 8.96452308e-01
-7.95463204e-01 -2.39816606e-01 2.64221340e-01 -1.21199630e-01
-6.45280182e-01 1.21390887e-01 -1.03570855e+00 2.14604035e-01
1.32890499e+00 -3.38637799e-01 1.23684235e-01 4.92181480e-01
2.97741382e-04 -3.37152362e-01 -8.89112055e-01 5.62883735e-01
3.29633236e-01 -1.12091994e+00 5.93244672e-01 2.12477267e-01
5.51230729e-01 -1.18642643e-01 2.55623125e-02 8.14939380e-01
-9.09442529e-02 -5.93421221e-01 3.92409146e-01 4.96007353e-01
5.34085512e-01 -6.01143479e-01 1.02568340e+00 7.91739300e-02
-1.13198376e+00 -1.95873275e-01 -4.55246925e-01 4.08495098e-01
-1.42728895e-01 1.64614692e-01 -9.56090569e-01 8.50630820e-01
5.58794558e-01 2.92771786e-01 -9.48247313e-01 1.36466062e+00
2.20725581e-01 3.56649727e-01 -3.77723455e-01 -2.88415432e-01
5.77584267e-01 1.88751966e-01 5.08626044e-01 1.38620627e+00
3.34182352e-01 -1.96196079e-01 -5.20892367e-02 2.92698413e-01
-6.04937300e-02 5.17646492e-01 -5.29329956e-01 -1.07028514e-01
3.80476415e-01 1.10523355e+00 -5.66509128e-01 -4.61591929e-01
-1.32505864e-01 8.31649303e-01 1.27744690e-01 5.44817507e-01
-2.19344363e-01 -8.34069431e-01 -2.91265771e-02 1.83968037e-01
3.21436733e-01 1.54659390e-01 3.72429609e-01 -6.40154541e-01
1.22944094e-01 -6.27646446e-01 5.66563368e-01 -7.09259391e-01
-9.49735522e-01 7.50613928e-01 4.34529871e-01 -1.22813129e+00
-6.70651138e-01 -2.70149976e-01 -1.23361625e-01 1.09249783e+00
-1.69644248e+00 -6.99451804e-01 1.21335864e-01 5.41512191e-01
3.96123171e-01 -6.75527453e-02 9.70824301e-01 5.44644237e-01
1.53383374e-01 5.19981980e-01 5.38366139e-01 -1.08736260e-02
6.39441073e-01 -1.23168159e+00 -2.29237363e-01 4.91627753e-01
4.48369354e-01 1.14328909e+00 3.29899043e-01 -5.88342428e-01
-1.06605923e+00 -6.77347422e-01 1.43692648e+00 -2.39257470e-01
4.47332412e-01 7.48480260e-02 -1.14812350e+00 1.95749149e-01
3.43699217e-01 -3.34484518e-01 7.25033045e-01 2.54404873e-01
-3.57594490e-01 -2.45711923e-01 -9.74800289e-01 4.87089068e-01
3.77045900e-01 -8.62184286e-01 -1.09870148e+00 2.42468446e-01
7.61560261e-01 2.37518489e-01 -7.64320135e-01 4.36141014e-01
6.28271103e-01 -7.03356683e-01 1.15757370e+00 -5.85862696e-01
1.20851934e-01 -2.08686098e-01 -4.10649091e-01 -1.12127447e+00
-4.39227104e-01 -2.18096972e-01 1.82735007e-02 1.13576138e+00
4.56952691e-01 -3.14323097e-01 4.64880198e-01 4.50281769e-01
-3.57078910e-02 -6.96205080e-01 -9.39249277e-01 -6.10526323e-01
-3.01723089e-02 -2.38750264e-01 2.38185093e-01 4.04370844e-01
4.69720699e-02 4.20630693e-01 7.38229603e-02 -6.32738695e-02
1.05213508e-01 -2.05074847e-01 3.41329008e-01 -1.32013273e+00
-1.26867920e-01 -3.74745518e-01 -3.90745759e-01 -1.06544089e+00
-2.67466903e-01 -1.03383183e+00 1.25332147e-01 -1.57853532e+00
1.93994522e-01 -4.65612888e-01 -9.45621789e-01 1.37783945e-01
-2.57664472e-01 2.54628092e-01 9.74795520e-02 5.49113154e-01
-8.28136206e-01 6.06759787e-01 8.10809612e-01 -3.00553322e-01
-3.33187670e-01 -1.71730042e-01 -7.00609565e-01 -5.20363487e-02
4.75324124e-01 -6.38544321e-01 -5.71221709e-01 -2.39414766e-01
6.55147612e-01 -5.29286638e-02 2.17469439e-01 -9.42568719e-01
3.46266210e-01 5.05937815e-01 2.63811618e-01 -5.92551172e-01
2.99426764e-01 -8.06666315e-01 -2.69933850e-01 4.12760079e-01
-8.77435565e-01 3.01879019e-01 3.22534949e-01 8.82921994e-01
-6.20062649e-01 -8.37111771e-01 1.98366597e-01 -8.61127153e-02
-7.95212448e-01 -1.87774912e-01 -4.62597460e-01 -6.65306002e-02
4.96150583e-01 -2.21840903e-01 -1.69530541e-01 -3.77017409e-01
-5.99070489e-01 1.37840092e-01 1.35278836e-01 6.31228328e-01
8.38076830e-01 -1.36241984e+00 -6.11892462e-01 7.69647658e-02
4.85322773e-01 -4.38942254e-01 1.68499395e-01 5.87135732e-01
-3.05345893e-01 1.15632415e+00 1.33982524e-01 -3.91094387e-01
-1.30563688e+00 5.33535659e-01 -2.73906644e-02 -9.10797656e-01
-2.81378686e-01 6.16717279e-01 -1.06219510e-02 -2.64968008e-01
1.84394985e-01 -1.67727530e-01 -7.32971847e-01 5.11751115e-01
6.87006176e-01 2.32323602e-01 6.10523164e-01 -5.97415030e-01
-2.85078645e-01 5.74286878e-01 -6.17978573e-01 -5.29939413e-01
1.03539288e+00 -2.25638777e-01 -8.36570486e-02 2.24796996e-01
1.72270763e+00 1.90598398e-01 -3.01575154e-01 -4.68064755e-01
5.42721450e-01 -4.07904178e-01 2.05011487e-01 -8.51317108e-01
-4.65285897e-01 7.90356636e-01 8.94921124e-01 5.70230305e-01
1.09651411e+00 -1.15322806e-01 4.46184933e-01 7.42579758e-01
3.41582417e-01 -1.12926316e+00 1.95219710e-01 5.47699273e-01
1.08736467e+00 -9.05004323e-01 -2.02089511e-02 3.16822410e-01
-2.83705801e-01 1.05640054e+00 9.29008424e-03 -1.00443527e-01
7.35798836e-01 -4.97511029e-01 -4.73472364e-02 -2.47141272e-01
-6.88041568e-01 -3.71413082e-01 7.65763998e-01 2.09137648e-01
5.04478693e-01 -2.94334710e-01 -6.30132258e-01 -1.02953124e-03
1.59469426e-01 -5.02499007e-02 1.00707360e-01 1.10659873e+00
-7.61293888e-01 -1.25093269e+00 -5.00948668e-01 5.97464323e-01
-4.04371470e-01 -3.07404459e-01 -4.86752093e-01 7.85024226e-01
-2.97595263e-01 1.13283813e+00 -5.46629168e-02 -4.22699153e-01
1.94293439e-01 3.60073894e-01 1.04443133e-01 -7.88245738e-01
-1.04860759e+00 2.40493819e-01 -2.84417961e-02 -1.53095707e-01
-4.86524165e-01 -4.84331280e-01 -8.16306293e-01 3.37390810e-01
-6.37426019e-01 8.91458511e-01 6.11898780e-01 9.51491416e-01
4.60918188e-01 2.70173341e-01 6.94735467e-01 -7.77087152e-01
-7.94194937e-01 -1.19251215e+00 -5.12418509e-01 6.49048924e-01
2.40706533e-01 -5.28455973e-01 -6.16965294e-01 -3.14389080e-01] | [11.439048767089844, 7.596195220947266] |
aef3ed15-6dbf-4b14-a1ec-78a96d0be537 | data-driven-covariance-steering-control | 2303.17675 | null | https://arxiv.org/abs/2303.17675v1 | https://arxiv.org/pdf/2303.17675v1.pdf | Data-Driven Covariance Steering Control Design | This paper studies the problem of steering the distribution of a linear time-invariant system from an initial normal distribution to a terminal normal distribution under no knowledge of the system dynamics. This data-driven control framework uses data collected from the input and the state and utilizes the seminal work by Willems et al. to construct a data-based parametrization of the mean and the covariance control problems. These problems are then solved to optimality as convex programs using standard techniques from the covariance control literature. We also discuss the equivalence of indirect and direct data-driven covariance steering designs, as well as a regularized version of the problem that provides a balance between the two. We illustrate the proposed framework through a set of randomized trials on a double integrator system and show that the results match up almost exactly with the corresponding model-based method in the noiseless case. We then analyze the robustness properties of the data-free and data-driven covariance steering methods and demonstrate the trade-offs between performance and optimality among these methods in the presence of data corrupted with exogenous noise. | ['Panagiotis Tsiotras', 'Joshua Pilipovsky'] | 2023-03-30 | null | null | null | null | ['steering-control'] | ['computer-vision'] | [ 2.79938281e-01 2.15961829e-01 -1.31408378e-01 1.20142251e-01
-7.36401439e-01 -6.97452784e-01 7.30806172e-01 -1.27746388e-01
-5.16814053e-01 8.77345562e-01 7.98601005e-03 -4.54716653e-01
-8.84825349e-01 -3.00167114e-01 -5.33723533e-01 -1.17654324e+00
-1.49739340e-01 3.60397696e-01 -7.86328763e-02 -2.92691559e-01
1.80339262e-01 2.82353133e-01 -1.19851220e+00 -6.66283727e-01
7.85606682e-01 9.78495717e-01 -6.57780617e-02 7.62527168e-01
6.96952343e-01 1.54222801e-01 -1.99989319e-01 3.52985501e-01
7.17009187e-01 -2.51098931e-01 -1.26452327e-01 2.88342267e-01
7.99524486e-02 -5.53620569e-02 -3.25633705e-01 1.13710928e+00
6.87715948e-01 4.37531322e-01 6.72595859e-01 -1.38406670e+00
-3.83858770e-01 -6.43128753e-02 -4.51382697e-01 8.98484048e-03
-4.45446596e-02 3.95727009e-01 5.49752653e-01 -6.03298068e-01
5.85252881e-01 1.16827655e+00 7.55416870e-01 4.85058248e-01
-1.77383506e+00 -2.58855581e-01 2.42923304e-01 -4.45112705e-01
-1.26064837e+00 -5.20428002e-01 2.80250520e-01 -1.04729080e+00
5.16568840e-01 1.75096184e-01 3.97942871e-01 9.09489453e-01
3.75268817e-01 1.82160214e-01 1.35439086e+00 -2.03060552e-01
6.05955303e-01 1.86728105e-01 2.66420424e-01 2.77313620e-01
8.05589795e-01 7.51814961e-01 5.51689789e-02 -4.95577246e-01
6.89168870e-01 -2.95171797e-01 -5.00317335e-01 -1.14244425e+00
-1.12887108e+00 8.45699251e-01 7.57464534e-03 -1.28981948e-01
-6.01735532e-01 6.63776919e-02 3.18856329e-01 4.16334480e-01
3.81111264e-01 5.37733912e-01 -5.74867547e-01 3.59365419e-02
-4.85616863e-01 7.36813545e-01 9.81510401e-01 1.14474761e+00
1.99730322e-01 6.28935039e-01 -2.61417806e-01 3.12358290e-01
5.80549896e-01 1.07356489e+00 2.17913449e-01 -1.08924723e+00
6.65024221e-01 1.20776452e-01 7.33942807e-01 -8.65342617e-01
-3.78417313e-01 -3.55004638e-01 -8.09351802e-01 5.73524475e-01
7.04619288e-01 -9.44165289e-01 -7.56284595e-01 2.04088855e+00
4.78992105e-01 -1.60848871e-01 5.38416356e-02 9.91875052e-01
-3.78723860e-01 5.45512319e-01 -4.71917421e-01 -8.15146089e-01
7.35153675e-01 -2.92062700e-01 -1.17586720e+00 -2.90005617e-02
3.01381767e-01 -5.58145583e-01 9.00383830e-01 4.05030191e-01
-9.67862129e-01 -2.22472385e-01 -1.15932214e+00 6.15101099e-01
-1.36243567e-01 -8.78310278e-02 -2.81743169e-01 7.06256926e-01
-9.98751640e-01 7.33687758e-01 -9.16088462e-01 -3.38105470e-01
-1.85625359e-01 3.51771206e-01 -5.76786548e-02 5.53623438e-01
-9.73899186e-01 1.04544485e+00 9.99412388e-02 1.80771723e-01
-9.13606524e-01 -9.46941733e-01 -6.34515166e-01 -2.55635500e-01
7.42274284e-01 -7.52969563e-01 1.35973549e+00 -6.52808726e-01
-1.74776852e+00 1.78883061e-01 -7.79536366e-02 -2.86206782e-01
7.40624070e-01 -2.67376512e-01 7.96361566e-02 -3.83764237e-01
2.06965413e-02 -2.51483887e-01 9.72162485e-01 -1.06617856e+00
-4.39588130e-01 -3.22790295e-01 -2.99202263e-01 8.99565294e-02
1.58219691e-02 -4.30227399e-01 2.53674723e-02 -5.32180130e-01
-1.53470665e-01 -1.37428117e+00 -5.28086960e-01 1.12856331e-03
-7.46749759e-01 2.95068681e-01 8.35671186e-01 -3.47587377e-01
1.30733609e+00 -1.90159285e+00 5.11756480e-01 5.79523921e-01
-1.31842002e-01 1.43574238e-01 5.27955778e-02 8.07227612e-01
-4.48071331e-01 -7.19199982e-03 -4.11156625e-01 -1.20086059e-01
2.31700957e-01 1.25128552e-01 -5.35668671e-01 9.59560633e-01
2.00008661e-01 2.12945536e-01 -7.93536782e-01 1.02680482e-01
1.98509574e-01 3.27814430e-01 -5.33209085e-01 2.54008800e-01
6.97048083e-02 5.87176025e-01 -7.01241136e-01 -5.36473095e-02
4.06528980e-01 1.54900134e-01 2.96895415e-01 -8.83572623e-02
-4.99049127e-01 -3.50452125e-01 -1.71202230e+00 9.55730498e-01
-3.90613019e-01 5.27888536e-01 8.58785689e-01 -1.15006876e+00
6.95209742e-01 4.38696533e-01 5.23010671e-01 -3.26502740e-01
5.02892613e-01 1.07407443e-01 1.76498011e-01 -4.07927126e-01
1.83380157e-01 -4.37188894e-01 -1.12397440e-01 4.30496633e-01
-1.50105715e-01 -5.78425467e-01 1.20691590e-01 -9.71325263e-02
9.03883934e-01 3.50997746e-02 5.85832357e-01 -9.84187126e-01
5.44124126e-01 3.95289771e-02 7.02925861e-01 6.78976953e-01
-3.33111465e-01 3.57389927e-01 8.46517086e-01 2.51726329e-01
-1.35846841e+00 -8.83619726e-01 -3.04979146e-01 4.56097484e-01
-1.25444338e-01 -1.23323947e-01 -7.51667261e-01 9.97503921e-02
3.61428112e-01 6.32460713e-01 -8.68810534e-01 -3.77523363e-01
-3.00539821e-01 -7.62956500e-01 -8.10634792e-02 2.18481645e-01
-7.33871153e-03 -1.17686920e-01 -4.82970327e-01 1.53368637e-01
4.13582593e-01 -8.16078305e-01 -5.81602275e-01 2.25832343e-01
-8.91372561e-01 -1.23105121e+00 -6.54789567e-01 -2.74689764e-01
8.33887875e-01 -3.74665350e-01 2.76273370e-01 -5.76629519e-01
-2.02168271e-01 8.28631282e-01 3.56151819e-01 -5.81783712e-01
-4.88061666e-01 -2.25890860e-01 8.04778993e-01 2.95897365e-01
-5.58303356e-01 -2.08020523e-01 -2.42138267e-01 3.95482838e-01
-7.73383975e-01 -4.78071898e-01 6.94576129e-02 9.00936186e-01
5.91031075e-01 -1.32987410e-01 6.34245217e-01 -7.87979603e-01
1.21118248e+00 -5.36317229e-01 -1.45031250e+00 -1.99760064e-01
-9.81595039e-01 3.30990314e-01 7.33092606e-01 -6.31277382e-01
-9.31002319e-01 3.61018300e-01 5.03550053e-01 -4.58090454e-01
3.12007487e-01 3.60875010e-01 -1.20876849e-01 -6.06753305e-02
5.43711543e-01 -1.11344814e-01 6.74794316e-01 -3.42299849e-01
5.60271323e-01 5.91744423e-01 3.63902330e-01 -7.36468196e-01
1.04485404e+00 4.20650363e-01 3.96395773e-01 -8.70618999e-01
-4.79260772e-01 -3.21027577e-01 -5.83159328e-01 -4.78240065e-02
6.41321480e-01 -6.11747265e-01 -9.70113814e-01 5.11708260e-01
-7.72448182e-01 -5.34785748e-01 -7.18026638e-01 4.65772957e-01
-1.25819910e+00 4.10383530e-02 -2.38764167e-01 -1.37593138e+00
6.48818761e-02 -1.03166604e+00 8.67482662e-01 4.02532518e-02
-6.77116960e-02 -1.20536160e+00 6.28779411e-01 -5.26458502e-01
6.60941362e-01 6.43769324e-01 7.10530281e-01 -5.49527943e-01
-2.05001727e-01 -5.00126541e-01 1.82892486e-01 4.44333851e-01
7.06541017e-02 -1.12537167e-03 -6.23087466e-01 -7.45052576e-01
4.08145338e-01 -2.61878613e-02 2.73107231e-01 8.01619887e-01
3.10466200e-01 -4.29100841e-01 -3.36268991e-01 3.81527156e-01
1.61406493e+00 4.56855625e-01 -1.41034294e-02 3.82774919e-01
4.23034608e-01 6.35170400e-01 6.56371057e-01 6.44651830e-01
-2.57395655e-02 7.06796467e-01 2.76500493e-01 6.62366524e-02
5.85356474e-01 -3.64598678e-03 2.80574679e-01 5.07254243e-01
2.11900249e-01 -7.84971118e-02 -7.58726418e-01 6.40115499e-01
-2.04987526e+00 -7.54636526e-01 -1.92281038e-01 2.70981932e+00
7.06444800e-01 -1.56772956e-01 2.79949576e-01 1.52022362e-01
9.15528655e-01 -1.75124839e-01 -7.18233645e-01 -5.89970827e-01
1.29208982e-01 -1.54733017e-01 9.83982921e-01 8.76430809e-01
-1.22592795e+00 4.57990766e-02 7.50698471e+00 3.38395000e-01
-8.91758323e-01 -1.93222165e-01 3.06778252e-01 -1.06765009e-01
3.12789351e-01 9.85284150e-03 -7.87206531e-01 4.33246970e-01
1.32759786e+00 -1.06823385e+00 5.00053525e-01 7.90974796e-01
1.03892219e+00 -1.11483626e-01 -1.03185475e+00 4.87156808e-01
-2.88138777e-01 -8.24766219e-01 -6.68150187e-01 3.74244869e-01
1.09203279e+00 -1.94554180e-01 3.07298183e-01 1.51047945e-01
4.34539080e-01 -5.78279316e-01 6.56756878e-01 8.15977693e-01
5.04212618e-01 -4.06719208e-01 5.45091152e-01 4.01049495e-01
-9.31984544e-01 -5.27405262e-01 1.30355343e-01 -2.24587694e-01
5.98837852e-01 4.94499862e-01 -3.71009767e-01 3.52840126e-01
1.23123251e-01 6.15530074e-01 5.61385527e-02 1.20308697e+00
1.63955972e-01 5.05960941e-01 -5.37766814e-01 -5.02664670e-02
1.69202656e-01 -5.76798320e-01 1.11780131e+00 8.48383605e-01
3.27814996e-01 -8.87654349e-02 3.13270271e-01 7.59859741e-01
7.47437119e-01 -2.43242234e-01 -9.04316068e-01 3.49174850e-02
2.12746456e-01 9.62120771e-01 -1.59480810e-01 -2.03804344e-01
-1.43141031e-01 1.96734190e-01 -1.52506247e-01 6.34337425e-01
-5.44831455e-01 -6.27047837e-01 9.49354470e-01 1.91850156e-01
3.33411664e-01 -4.86473292e-01 -1.87161446e-01 -7.41339028e-01
4.82688397e-02 -8.22044849e-01 1.77948162e-01 -4.05872285e-01
-1.33833742e+00 2.43959457e-01 4.97840732e-01 -1.37933123e+00
-6.53479040e-01 -6.90456867e-01 -4.28813905e-01 1.11866403e+00
-9.30607796e-01 -1.01939991e-01 2.29873583e-01 4.07338917e-01
8.83221477e-02 -2.10262947e-02 4.09727722e-01 2.24475175e-01
-8.30363929e-01 -7.15235323e-02 1.00212467e+00 -2.60973752e-01
6.35522246e-01 -1.45246971e+00 4.16071452e-02 8.85165870e-01
-9.15277362e-01 8.05370986e-01 1.27069640e+00 -6.96395278e-01
-1.75203562e+00 -1.11276984e+00 2.99179912e-01 -4.23956156e-01
1.32919705e+00 -2.47115284e-01 -6.55832946e-01 6.06816769e-01
6.17051497e-02 6.08115047e-02 -1.18553579e-01 -2.16423392e-01
2.52143383e-01 -1.97399318e-01 -1.04364705e+00 5.77816665e-01
5.89473367e-01 -2.21865118e-01 -5.78758895e-01 2.08807021e-01
5.05304098e-01 -3.66375476e-01 -8.15833628e-01 4.22845721e-01
3.56192559e-01 -4.20467168e-01 7.65068114e-01 -8.39398205e-01
-1.83690533e-01 -5.09882927e-01 -3.21631372e-01 -1.93808186e+00
5.46281934e-02 -1.22817039e+00 5.40905185e-02 1.02930617e+00
3.48912984e-01 -1.02355826e+00 4.20683593e-01 6.44646704e-01
-3.44221331e-02 -5.60855091e-01 -1.12518847e+00 -1.32233238e+00
4.37650114e-01 -2.96681881e-01 -2.28267342e-01 6.21795177e-01
1.44765168e-01 2.57246912e-01 -3.16491544e-01 4.58910793e-01
9.01147783e-01 -3.07180524e-01 7.68725336e-01 -8.18177640e-01
-4.04601544e-01 -3.73387367e-01 -2.76466668e-01 -6.63716495e-01
-8.06997940e-02 -4.68023300e-01 5.81334293e-01 -1.17821932e+00
-1.24371402e-01 -3.54392827e-01 8.29040036e-02 3.46439518e-02
-1.06445976e-01 -4.29440767e-01 2.83200681e-01 -1.04581080e-01
-1.01824179e-01 7.08083391e-01 1.06907439e+00 1.24991901e-01
-4.61279929e-01 3.93936515e-01 -4.91201431e-01 5.16409755e-01
5.57778299e-01 -3.61977249e-01 -9.37472939e-01 -1.10172108e-01
-9.54573303e-02 4.05359954e-01 1.98621422e-01 -8.32476318e-01
1.99890465e-01 -4.58198607e-01 -3.36407334e-01 -2.61053503e-01
1.76425293e-01 -1.04536116e+00 2.71518469e-01 6.65858150e-01
-4.68127906e-01 3.02179188e-01 2.49342307e-01 9.93774533e-01
2.67799869e-02 -1.38441958e-02 1.06084788e+00 4.08182025e-01
-4.31661978e-02 2.75060445e-01 -6.73334599e-01 4.97390389e-01
1.17716336e+00 -3.51280527e-04 -3.48845512e-01 -7.21562803e-01
-1.05767775e+00 4.03383434e-01 1.40658453e-01 2.72977561e-01
-2.18358450e-02 -1.11601663e+00 -3.60537261e-01 3.00637960e-01
-2.63028085e-01 -3.67649376e-01 -9.76440385e-02 1.22821736e+00
1.55633241e-01 6.74809873e-01 1.25521049e-01 -7.78185010e-01
-9.00162578e-01 6.32620454e-01 6.68267310e-01 2.98261028e-02
-1.76525950e-01 1.01292506e-01 2.50495583e-01 -5.49039364e-01
2.28405088e-01 -5.84616125e-01 3.14504266e-01 3.15174647e-02
3.51541907e-01 5.94168961e-01 1.23855127e-02 -4.00172174e-01
-2.27056965e-01 7.93362021e-01 4.46731865e-01 -4.72491145e-01
1.13746226e+00 -3.95468503e-01 3.65033567e-01 8.02221239e-01
1.10502911e+00 -3.08650851e-01 -1.74146318e+00 -1.11092985e-01
1.26076683e-01 -2.90374935e-01 1.49414048e-01 -5.18732429e-01
-9.48681295e-01 5.04917383e-01 7.69758403e-01 4.72652376e-01
8.67573023e-01 -6.50947154e-01 -1.24491639e-01 3.68356586e-01
1.50985643e-01 -1.19928968e+00 -1.65042982e-01 7.45229125e-01
9.21808660e-01 -7.65349329e-01 5.48036061e-02 -2.02758685e-01
-6.39328122e-01 7.81388581e-01 4.07380193e-01 -7.39611149e-01
1.05693614e+00 7.49292970e-01 -1.01380453e-01 2.75715351e-01
-1.05259848e+00 -1.81302503e-01 1.69826269e-01 9.67327476e-01
1.15006678e-01 -1.61575943e-01 -5.41132629e-01 3.65741372e-01
3.42977762e-01 1.27558475e-02 7.50770688e-01 1.20121741e+00
-3.82678926e-01 -1.02191424e+00 -7.24383354e-01 2.99228638e-01
-2.01848662e-03 2.82351196e-01 -1.22443728e-01 1.27802420e+00
-5.48058629e-01 9.99987841e-01 5.30530699e-03 1.01665027e-01
1.06685233e+00 -3.33914207e-03 1.35178253e-01 -4.27982986e-01
-2.79300302e-01 3.12607348e-01 1.21702157e-01 -6.66179895e-01
-1.76131681e-01 -9.03029799e-01 -7.79428005e-01 -2.35822245e-01
-4.67607319e-01 7.35468119e-02 6.15798116e-01 8.52556705e-01
4.33725238e-01 6.13883972e-01 7.73978591e-01 -9.36968029e-01
-1.48083448e+00 -6.78691566e-01 -6.51243329e-01 7.08554611e-02
8.32744598e-01 -8.35521460e-01 -7.95950234e-01 -1.33037139e-02] | [5.087167739868164, 2.524186372756958] |
575de333-d0f0-4c5a-8cdf-49a5afdab96a | multi-domain-stain-normalization-for-digital | 2301.09431 | null | https://arxiv.org/abs/2301.09431v1 | https://arxiv.org/pdf/2301.09431v1.pdf | Multi-domain stain normalization for digital pathology: A cycle-consistent adversarial network for whole slide images | The variation in histologic staining between different medical centers is one of the most profound challenges in the field of computer-aided diagnosis. The appearance disparity of pathological whole slide images causes algorithms to become less reliable, which in turn impedes the wide-spread applicability of downstream tasks like cancer diagnosis. Furthermore, different stainings lead to biases in the training which in case of domain shifts negatively affect the test performance. Therefore, in this paper we propose MultiStain-CycleGAN, a multi-domain approach to stain normalization based on CycleGAN. Our modifications to CycleGAN allow us to normalize images of different origins without retraining or using different models. We perform an extensive evaluation of our method using various metrics and compare it to commonly used methods that are multi-domain capable. First, we evaluate how well our method fools a domain classifier that tries to assign a medical center to an image. Then, we test our normalization on the tumor classification performance of a downstream classifier. Furthermore, we evaluate the image quality of the normalized images using the Structural similarity index and the ability to reduce the domain shift using the Fr\'echet inception distance. We show that our method proves to be multi-domain capable, provides the highest image quality among the compared methods, and can most reliably fool the domain classifier while keeping the tumor classifier performance high. By reducing the domain influence, biases in the data can be removed on the one hand and the origin of the whole slide image can be disguised on the other, thus enhancing patient data privacy. | ['Titus J. Brinker', 'Tabea-Clara Bucher', 'Martin J. Hetz'] | 2023-01-23 | null | null | null | null | ['whole-slide-images'] | ['computer-vision'] | [ 2.23051414e-01 -2.63523050e-02 -2.07480595e-01 -2.65439063e-01
-6.65108323e-01 -7.97249794e-01 4.36414897e-01 2.58227736e-01
-7.82064974e-01 8.01228166e-01 -1.28557131e-01 -4.39956576e-01
-1.73838571e-01 -7.78534412e-01 -4.85430568e-01 -1.26031590e+00
5.03946960e-01 4.20878798e-01 3.02647501e-01 -1.36855736e-01
1.01831809e-01 6.66172206e-01 -1.13516331e+00 3.57027352e-01
6.67506456e-01 7.24394381e-01 -1.84691861e-01 6.27799630e-01
1.59840137e-02 4.49465632e-01 -7.76867032e-01 -5.82587361e-01
3.93670291e-01 -6.13253176e-01 -7.71219075e-01 -2.78907456e-02
3.86649191e-01 1.23476144e-03 -5.44532109e-03 1.34426463e+00
5.38329363e-01 -3.66937459e-01 8.39953542e-01 -1.13972616e+00
-1.81281880e-01 2.81664878e-01 -5.30043304e-01 1.64875403e-01
-7.32577741e-02 2.33912870e-01 5.64107776e-01 -2.81153440e-01
1.07781959e+00 8.25582325e-01 6.94262505e-01 4.23127413e-01
-1.51269650e+00 -6.86650217e-01 -2.73160577e-01 1.55641615e-01
-1.40414560e+00 -2.15787143e-01 5.94244599e-01 -5.99296391e-01
1.20905980e-01 4.45504963e-01 4.58814025e-01 7.33380675e-01
5.56505203e-01 1.20655097e-01 1.30890536e+00 -5.44687390e-01
1.87256455e-01 4.51756418e-01 -1.32921949e-01 7.41829216e-01
5.40859580e-01 -1.75772056e-01 -3.69130485e-02 -1.99016020e-01
4.86424685e-01 -4.87677008e-02 -4.82688874e-01 -5.24946511e-01
-1.14254951e+00 7.74279773e-01 4.20176625e-01 7.85909474e-01
4.85029556e-02 -3.12389791e-01 3.70142937e-01 4.15095627e-01
3.00562322e-01 5.44292629e-01 5.26798852e-02 4.43999469e-01
-1.05068529e+00 -1.50751755e-01 5.94405413e-01 3.93655330e-01
5.35775065e-01 -5.00226796e-01 -1.35175318e-01 5.77382743e-01
-5.70090562e-02 3.50366592e-01 8.30100298e-01 -6.62126184e-01
-7.38428310e-02 7.50335395e-01 -2.63156623e-01 -1.00252271e+00
-4.57269013e-01 -4.60013986e-01 -1.01962924e+00 4.70143646e-01
9.53419566e-01 1.44976005e-02 -9.35416579e-01 1.67558789e+00
2.72801131e-01 -1.34138186e-02 9.79487672e-02 6.94125295e-01
5.91416478e-01 3.89374979e-02 8.36862549e-02 -4.63191532e-02
1.50364006e+00 -4.98881459e-01 -6.60827219e-01 1.30762964e-01
1.08855128e+00 -8.54841888e-01 7.13753939e-01 3.13402504e-01
-5.53324103e-01 -2.25599185e-01 -1.20783508e+00 5.67965060e-02
-5.78246295e-01 4.83761132e-02 1.89771146e-01 8.15247655e-01
-1.06656110e+00 5.21301270e-01 -7.75281370e-01 -5.23170054e-01
5.86883426e-01 6.68753386e-01 -8.19903433e-01 -2.80734479e-01
-9.45639670e-01 9.79845941e-01 4.50262934e-01 -3.83661151e-01
-3.22971702e-01 -7.66671300e-01 -4.93768245e-01 -1.30459994e-01
7.21491203e-02 -5.83795488e-01 7.70785630e-01 -1.29377198e+00
-1.10679770e+00 1.30522406e+00 3.54556516e-02 -3.43973190e-01
8.69852841e-01 7.50023246e-01 -4.89915997e-01 2.00279519e-01
2.38796204e-01 5.61691523e-01 6.26877129e-01 -9.94537234e-01
-7.65956104e-01 -4.71758962e-01 -3.28282386e-01 -1.14662267e-01
-4.45924431e-01 -4.57580268e-01 -5.11224151e-01 -5.34923673e-01
-2.22845271e-01 -1.14719546e+00 -2.32246011e-01 1.52057678e-01
-3.12897235e-01 2.22749472e-01 8.21895719e-01 -4.32444006e-01
1.08200693e+00 -2.77167010e+00 -1.36486605e-01 6.40629292e-01
3.73240858e-01 2.80003726e-01 -1.41523927e-01 -1.03960499e-01
-3.11838865e-01 3.10154617e-01 -1.35541931e-01 -1.57370474e-02
-4.80340660e-01 2.28711858e-01 3.25891912e-01 9.64282095e-01
5.80025278e-03 5.94889164e-01 -7.18518198e-01 -8.35208237e-01
9.56498906e-02 3.29561144e-01 -5.14097095e-01 3.11483331e-02
2.17037871e-01 5.00819802e-01 -9.09566507e-02 3.50974619e-01
7.61208177e-01 -2.48941645e-01 5.11577785e-01 -2.56388187e-01
1.05096012e-01 -2.05892831e-01 -8.72440755e-01 1.36551178e+00
-1.83742240e-01 7.98944712e-01 3.30586769e-02 -8.76569450e-01
7.99346268e-01 2.91332096e-01 5.59457541e-01 -6.27408922e-01
3.32051992e-01 4.33333665e-01 4.16906029e-01 -3.32014948e-01
1.43219531e-01 -3.77073646e-01 1.08092599e-01 1.77083224e-01
-4.73617911e-02 -7.64370486e-02 1.38210282e-01 2.79439613e-02
1.07039940e+00 -6.91328108e-01 4.28553849e-01 -5.03827274e-01
7.53789961e-01 2.64269799e-01 3.71774942e-01 3.95297647e-01
-4.50959921e-01 6.38643980e-01 1.00170851e+00 -1.39909700e-01
-9.44083154e-01 -9.04342592e-01 -2.71969408e-01 6.54759943e-01
5.68008870e-02 3.07645947e-01 -7.82677293e-01 -9.50867653e-01
1.93484515e-01 3.43653142e-01 -9.77514446e-01 -2.96276897e-01
-3.16547543e-01 -1.00483215e+00 8.66933167e-01 1.03173740e-01
2.45442137e-01 -4.65868294e-01 -5.08534908e-01 -8.08251053e-02
-1.14387475e-01 -9.13891435e-01 -4.54351604e-01 3.39289218e-01
-8.36007357e-01 -1.37689364e+00 -8.74948382e-01 -7.61134803e-01
1.06814623e+00 3.64971757e-02 9.37854052e-01 2.34038502e-01
-4.29483443e-01 5.69294021e-02 -2.38418385e-01 -5.55678010e-01
-8.50873351e-01 3.19011033e-01 -2.39590123e-01 1.63567051e-01
5.27032137e-01 -1.81494787e-01 -6.40250325e-01 4.15915638e-01
-1.18259835e+00 -3.43979716e-01 6.42215252e-01 9.17011619e-01
7.24183619e-01 1.33176923e-01 3.31400990e-01 -1.41173923e+00
3.71036559e-01 -2.54093528e-01 -5.70922613e-01 2.83713400e-01
-8.23665679e-01 1.97769210e-01 5.55951715e-01 -3.56254250e-01
-7.61628270e-01 2.19908834e-01 6.47525070e-03 -2.67233968e-01
-8.78369957e-02 2.41793647e-01 -1.44124970e-01 -4.61912006e-01
1.00179064e+00 -1.14775486e-01 6.52824461e-01 -2.17225477e-02
-3.34275179e-02 6.01865113e-01 5.42274296e-01 8.62569064e-02
7.60426521e-01 8.83384407e-01 3.32257271e-01 -7.77595222e-01
-4.06396508e-01 -6.00070000e-01 -5.20662010e-01 -1.53099790e-01
8.96984994e-01 -6.23720109e-01 -4.72867608e-01 4.79080558e-01
-6.92556918e-01 -9.19096246e-02 -2.19521821e-01 4.06564176e-01
-1.87654138e-01 3.18786621e-01 -3.44271332e-01 -2.03565270e-01
-7.25719407e-02 -1.25899887e+00 5.89808345e-01 2.25510523e-01
-3.10003847e-01 -1.28785658e+00 1.98603153e-01 2.10623249e-01
3.20964068e-01 6.21530414e-01 1.19220769e+00 -8.93568218e-01
-2.14396581e-01 -2.78224289e-01 -1.97176948e-01 3.52817506e-01
3.58499497e-01 2.45341852e-01 -9.65759158e-01 -4.03137684e-01
-2.60318890e-02 2.46049866e-01 7.25086033e-01 4.37635422e-01
9.68263805e-01 2.62204967e-02 -6.67507231e-01 7.03796089e-01
1.61150301e+00 2.90070236e-01 8.68482471e-01 5.43399215e-01
4.32639599e-01 7.89913952e-01 6.16256595e-01 -1.03631187e-02
-4.35932130e-02 4.31337506e-01 3.24068308e-01 -6.54970646e-01
-9.99025702e-02 4.89846319e-02 -2.02603769e-02 2.87958384e-01
7.96491355e-02 -3.13114822e-01 -8.94671202e-01 7.01317191e-01
-1.44852734e+00 -7.40693867e-01 -2.59618819e-01 2.08440709e+00
8.42443526e-01 4.86378185e-03 4.50446084e-02 2.45686516e-01
8.11373234e-01 -2.18622759e-01 -3.86474907e-01 -4.64983493e-01
-2.93996990e-01 2.59087354e-01 9.69407439e-01 1.96305916e-01
-9.31642771e-01 4.30544496e-01 6.15307713e+00 8.03875327e-01
-1.57628369e+00 3.63765359e-02 9.36910033e-01 -1.27195761e-01
-4.08722386e-02 -3.76249790e-01 -4.68625963e-01 4.97356415e-01
7.45435655e-01 -1.84735268e-01 -3.28957215e-02 7.47830629e-01
6.94064051e-02 -3.53078216e-01 -1.22458744e+00 8.94072771e-01
7.77206421e-02 -1.34272063e+00 -8.04262459e-02 5.28607607e-01
6.23706460e-01 -2.92986691e-01 2.52889335e-01 -2.16477469e-01
1.01621710e-01 -1.05335438e+00 2.31893659e-01 3.79821181e-01
1.05747712e+00 -8.20406199e-01 1.25689387e+00 -1.28011615e-03
-6.12619400e-01 1.25861883e-01 -1.50850773e-01 4.50893283e-01
-4.29569989e-01 7.00992763e-01 -1.32274783e+00 4.80207831e-01
4.61474359e-01 2.78249651e-01 -7.70304322e-01 8.88039947e-01
1.30041480e-01 2.80256301e-01 -1.65364459e-01 -1.45680190e-03
3.76775861e-02 2.81261001e-02 2.50298828e-01 1.24998474e+00
4.18638855e-01 -1.02669902e-01 -2.25428149e-01 2.96037525e-01
-1.02531642e-01 3.85098755e-01 -5.93056798e-01 2.28619128e-02
2.80238152e-01 1.20671642e+00 -9.43167627e-01 -2.64694214e-01
-2.56991744e-01 8.65470946e-01 -1.03169553e-01 -1.27785997e-02
-6.44835413e-01 -3.96704197e-01 6.83598697e-01 4.07402486e-01
2.04177454e-01 4.12339777e-01 -4.28650886e-01 -9.05866921e-01
-3.75163525e-01 -1.03015375e+00 8.49044561e-01 -4.35808867e-01
-1.18209815e+00 5.20280480e-01 -3.01924229e-01 -1.45891643e+00
-1.02143399e-01 -8.24434459e-01 -2.36180589e-01 8.25504780e-01
-1.61745882e+00 -8.68212938e-01 -2.10420266e-01 7.07734168e-01
-2.03630105e-01 -1.98588356e-01 7.93281198e-01 4.10834908e-01
-2.68172622e-01 1.02488172e+00 4.45706576e-01 4.31011736e-01
1.36254215e+00 -1.15312207e+00 -5.28053582e-01 8.22819650e-01
-2.12466612e-01 3.37528169e-01 5.74841380e-01 -3.51488948e-01
-8.26763213e-01 -1.10323381e+00 6.86634481e-01 -4.57686931e-01
5.91298759e-01 2.00353656e-02 -8.36515367e-01 4.05412912e-01
4.31644619e-02 -2.32387595e-02 1.10336435e+00 -2.85907179e-01
-3.21538150e-01 -4.00710285e-01 -1.55573153e+00 5.44427216e-01
3.41101795e-01 -3.08207035e-01 -1.77000746e-01 1.03952281e-01
9.96363014e-02 -3.44576567e-01 -1.05832922e+00 1.31274089e-01
6.15221202e-01 -1.08709669e+00 6.63211286e-01 -2.41542459e-01
4.05956209e-01 -3.62037688e-01 9.13459063e-02 -1.29367995e+00
-3.61738265e-01 5.46650961e-02 8.70055556e-01 1.17543137e+00
6.44820333e-01 -9.19406116e-01 9.32027519e-01 4.93352950e-01
3.19738656e-01 -4.47174966e-01 -1.08107495e+00 -6.81460738e-01
2.98356295e-01 1.52493864e-01 6.36742771e-01 1.17624152e+00
-6.06976487e-02 -1.60149604e-01 1.29080445e-01 9.99859795e-02
4.40247267e-01 6.86131185e-03 6.82582080e-01 -1.26616311e+00
-1.89005539e-01 -6.56227112e-01 -8.59444082e-01 -4.83190641e-02
-9.19700116e-02 -1.05824816e+00 -5.16325608e-02 -8.95299852e-01
2.15242147e-01 -4.60529774e-01 -4.12388116e-01 5.99367321e-01
6.90072309e-03 6.79661214e-01 6.52954206e-02 2.25563422e-01
-1.77289978e-01 -3.79974574e-01 1.32334793e+00 -3.20454597e-01
-1.03866123e-02 -1.86484084e-01 -8.77605259e-01 5.25465667e-01
7.28206217e-01 -8.55510771e-01 -1.48925215e-01 -2.29776185e-02
-2.79084593e-02 -3.09550643e-01 2.88774520e-01 -9.94584858e-01
1.44287035e-01 -2.90379487e-02 5.90495229e-01 -5.33618815e-02
-1.11627072e-01 -1.16289949e+00 4.68054473e-01 1.00647688e+00
-6.35156557e-02 -1.59056008e-01 1.96036980e-01 3.20991099e-01
-3.42987418e-01 -2.83342510e-01 1.28730941e+00 -8.65051597e-02
-3.38737875e-01 5.47178574e-02 -3.34501207e-01 -1.86527669e-01
1.30844820e+00 -4.23774362e-01 -4.61339444e-01 -3.40758488e-02
-5.01985371e-01 4.66064140e-02 1.02648413e+00 2.10325886e-02
7.55115896e-02 -9.55379367e-01 -5.94126225e-01 2.54931539e-01
2.92720675e-01 -1.60568431e-01 2.60016531e-01 1.12766898e+00
-8.46439481e-01 2.67776430e-01 -3.51346672e-01 -7.58568764e-01
-1.75487006e+00 5.68841279e-01 5.56342363e-01 -6.47483826e-01
-1.12693213e-01 7.19681561e-01 2.52331674e-01 -7.87830502e-02
-1.59866825e-01 -3.49271804e-01 -2.18043581e-01 3.95046413e-01
3.49655986e-01 2.08911538e-01 4.04789031e-01 -5.51762342e-01
-5.34423709e-01 4.02035087e-01 -3.18183124e-01 -5.74522372e-03
1.16418326e+00 2.11944729e-01 -2.99149543e-01 1.29175037e-01
1.35857296e+00 1.79255918e-01 -7.67608821e-01 1.89326316e-01
-2.24814832e-01 -4.82173771e-01 4.21249829e-02 -8.58469307e-01
-1.14901292e+00 6.77019775e-01 1.04017699e+00 2.88828790e-01
1.48335516e+00 -2.14608341e-01 3.64117146e-01 -9.50898156e-02
7.16579854e-02 -9.58911777e-01 -2.05300584e-01 4.09016795e-02
2.18310922e-01 -1.25632226e+00 1.74250931e-01 -4.47356552e-01
-5.28426170e-01 1.06355941e+00 2.70376951e-01 7.89643452e-02
5.76671064e-01 5.17621875e-01 6.38989866e-01 -1.12270424e-02
-3.69777858e-01 5.88684641e-02 5.54833189e-02 7.14773893e-01
2.71906257e-01 1.26110807e-01 -5.76437294e-01 2.22571000e-01
-2.53421694e-01 2.94889987e-01 7.57945955e-01 6.93876088e-01
-7.27146417e-02 -1.39632392e+00 -6.99699283e-01 5.25631011e-01
-7.43017972e-01 2.89943665e-01 -5.21759808e-01 1.01934910e+00
3.88942510e-01 6.25601411e-01 1.41404375e-01 -2.59630233e-01
2.78549135e-01 6.45266846e-02 2.49838948e-01 -4.30378616e-01
-7.34435081e-01 -8.50078091e-02 -2.88245469e-01 -1.36364281e-01
-5.52023768e-01 -7.29464531e-01 -9.75755215e-01 -5.57786167e-01
-3.17700326e-01 2.07412407e-01 6.86675906e-01 7.05763817e-01
3.92420322e-01 5.62512457e-01 5.76731980e-01 -1.69838399e-01
-2.11645886e-01 -5.34587741e-01 -6.89034522e-01 5.84557235e-01
4.09922689e-01 -4.87113148e-01 -5.01924217e-01 2.42093369e-01] | [15.016923904418945, -3.0938961505889893] |
881c2f4b-853a-4457-8e11-a60d1fab5e84 | non-log-concave-and-nonsmooth-sampling-via | 2305.15988 | null | https://arxiv.org/abs/2305.15988v1 | https://arxiv.org/pdf/2305.15988v1.pdf | Non-Log-Concave and Nonsmooth Sampling via Langevin Monte Carlo Algorithms | We study the problem of approximate sampling from non-log-concave distributions, e.g., Gaussian mixtures, which is often challenging even in low dimensions due to their multimodality. We focus on performing this task via Markov chain Monte Carlo (MCMC) methods derived from discretizations of the overdamped Langevin diffusions, which are commonly known as Langevin Monte Carlo algorithms. Furthermore, we are also interested in two nonsmooth cases for which a large class of proximal MCMC methods have been developed: (i) a nonsmooth prior is considered with a Gaussian mixture likelihood; (ii) a Laplacian mixture distribution. Such nonsmooth and non-log-concave sampling tasks arise from a wide range of applications to Bayesian inference and imaging inverse problems such as image deconvolution. We perform numerical simulations to compare the performance of most commonly used Langevin Monte Carlo algorithms. | ['Thomas Pock', 'Han Liu', 'Tim Tsz-Kit Lau'] | 2023-05-25 | null | null | null | null | ['image-deconvolution', 'bayesian-inference'] | ['computer-vision', 'methodology'] | [ 2.54249096e-01 -8.16475824e-02 3.73334914e-01 4.96549299e-03
-9.22240198e-01 -3.32382411e-01 8.24480951e-01 -2.99581826e-01
-5.22713959e-01 1.03166020e+00 -6.06545992e-03 -1.61786258e-01
-1.86278850e-01 -4.52891678e-01 -6.67805493e-01 -1.07582045e+00
6.44269586e-02 1.12395537e+00 6.79632723e-02 3.80605996e-01
2.15278178e-01 5.84011734e-01 -1.10935366e+00 -3.97266954e-01
1.13336825e+00 4.78058040e-01 6.00676313e-02 7.50649035e-01
2.32918169e-02 3.33900988e-01 -6.08156510e-02 -1.95185617e-01
-1.73943341e-01 -6.74894392e-01 -5.09707034e-01 3.64269823e-01
-2.09081583e-02 -2.67146796e-01 2.27460340e-01 1.48091674e+00
4.40646678e-01 5.13532579e-01 1.60436738e+00 -7.67458439e-01
-2.12394327e-01 2.93562442e-01 -7.90904999e-01 -1.39939502e-01
1.79826230e-01 2.63310168e-02 1.90176740e-01 -1.08618081e+00
6.79863274e-01 1.49055898e+00 6.79914415e-01 3.51761490e-01
-1.75843883e+00 -1.69598550e-01 -4.67903763e-01 -1.60706744e-01
-1.60172069e+00 -5.37259936e-01 6.65447414e-01 -9.74389851e-01
2.38103554e-01 -8.08102041e-02 3.85720164e-01 1.12297904e+00
2.44566888e-01 6.07490659e-01 1.45501006e+00 -4.72122699e-01
9.13123608e-01 3.01508978e-02 -1.67540818e-01 5.33549666e-01
3.33637595e-01 -1.41900554e-01 -5.04266098e-02 -8.97970378e-01
1.05676103e+00 -6.34573177e-02 -2.15233490e-01 -6.26665890e-01
-1.13580596e+00 1.16885090e+00 -1.87963307e-01 7.97215700e-02
-7.39830494e-01 3.42126936e-01 2.08602231e-02 -2.20668346e-01
4.83099490e-01 -2.04001144e-02 2.00237706e-01 -1.42794773e-01
-1.23451102e+00 6.11232460e-01 1.10040128e+00 9.57800329e-01
6.89280272e-01 1.06430694e-01 -2.85693761e-02 6.49624467e-01
7.84683764e-01 9.23706412e-01 -1.23787671e-01 -1.21123123e+00
1.47664353e-01 -5.13401926e-01 9.38674390e-01 -2.71246046e-01
-1.33199647e-01 -2.40455389e-01 -1.20143318e+00 4.25137281e-01
8.83778691e-01 -2.67175227e-01 -7.65735209e-01 1.61431408e+00
5.81383824e-01 2.11732179e-01 -1.22379646e-01 7.57747829e-01
5.88288717e-02 7.28271484e-01 2.37331819e-02 -8.21511984e-01
1.20358109e+00 -5.04954696e-01 -7.66107142e-01 1.29861668e-01
1.76901966e-01 -1.07346129e+00 6.62703216e-01 5.77082276e-01
-1.41768527e+00 -8.58825520e-02 -5.97476542e-01 2.20666379e-01
1.63207278e-01 -7.17070773e-02 2.28800938e-01 6.03579938e-01
-7.63351083e-01 8.62167120e-01 -1.34525323e+00 -2.20196128e-01
4.46756035e-01 -9.09348354e-02 2.46472448e-01 -3.14343870e-01
-7.04640746e-01 7.67704785e-01 -4.52814102e-02 1.08120538e-01
-1.08376551e+00 -3.87777746e-01 -4.77295995e-01 -1.32833764e-01
2.33809620e-01 -8.30135167e-01 1.17544663e+00 -5.94025791e-01
-1.73911870e+00 6.34057820e-01 -4.30188894e-01 -2.43500650e-01
1.05609798e+00 -5.34270518e-02 3.28421928e-02 1.50370553e-01
-1.70144532e-02 2.35638067e-01 1.34867644e+00 -1.17863452e+00
3.11668128e-01 -2.32046396e-01 -5.92802823e-01 1.29231036e-01
3.73082280e-01 -5.35485148e-02 -8.43841508e-02 -6.00370228e-01
2.72257537e-01 -1.20554197e+00 -5.58306456e-01 9.56498459e-02
-5.25645256e-01 -2.97941342e-02 3.94092143e-01 -5.07788360e-01
8.11116815e-01 -1.83513546e+00 4.03691769e-01 2.08494648e-01
8.83013606e-02 -3.82281989e-02 2.10098654e-01 6.98622048e-01
2.67263263e-01 -1.18121564e-01 -8.94312143e-01 -4.86868918e-01
2.04485625e-01 1.21861354e-01 -1.67927623e-01 1.03541338e+00
-6.05313107e-02 4.58078921e-01 -9.50728714e-01 -5.95189273e-01
2.31106319e-02 6.88031673e-01 -6.05445206e-01 -4.53199148e-02
-3.89273793e-01 9.77740407e-01 -6.94916129e-01 2.55387366e-01
9.45357084e-01 -5.13374209e-01 1.81215145e-02 -4.52317782e-02
-2.23101318e-01 -3.93283963e-01 -1.57995296e+00 1.56964219e+00
-2.71037638e-01 3.21044028e-01 5.05064905e-01 -6.95280194e-01
3.82492989e-01 4.08091843e-01 2.71268666e-01 3.70925307e-01
1.77742064e-01 4.89722878e-01 -1.09940737e-01 -4.68938828e-01
3.69341969e-01 -9.41912115e-01 2.43066818e-01 5.03293455e-01
-1.59423694e-01 -5.12037992e-01 2.11907163e-01 2.38776699e-01
8.90391767e-01 2.92675197e-01 5.52649021e-01 -7.45999992e-01
1.90298542e-01 -6.65296465e-02 2.84848422e-01 1.09008777e+00
-4.86390628e-02 8.70698214e-01 4.13758337e-01 1.97716296e-01
-1.26225710e+00 -1.35709405e+00 -7.89504468e-01 1.19593486e-01
-9.14017856e-02 9.52037722e-02 -9.49259341e-01 2.17956245e-01
-2.08919138e-01 6.58683717e-01 -1.91024914e-01 1.82032406e-01
-3.97899359e-01 -1.33617687e+00 1.30486488e-01 1.73041001e-02
3.31196666e-01 -5.63630879e-01 -3.85714412e-01 5.38934827e-01
-8.22086409e-02 -9.40959036e-01 -2.94765502e-01 -9.28567424e-02
-1.19309950e+00 -7.25977778e-01 -1.42872512e+00 -2.44749606e-01
6.17516696e-01 -2.72522748e-01 8.95220518e-01 -7.60625899e-01
-9.85202566e-02 5.72015941e-01 -3.67396371e-03 -1.58582646e-02
-8.85440826e-01 -3.24706018e-01 2.74087846e-01 2.88024813e-01
4.43100221e-02 -6.88333452e-01 -6.00112081e-01 3.86697441e-01
-1.10723579e+00 1.30254333e-03 5.21718144e-01 7.78485537e-01
8.21538687e-01 7.49054253e-02 4.51194674e-01 -1.05789423e+00
7.71542370e-01 -7.23201573e-01 -8.81781399e-01 -5.12437709e-02
-3.48735869e-01 1.70842126e-01 4.36543882e-01 -7.56718755e-01
-1.25472701e+00 7.17539340e-02 -1.30857497e-01 -4.02381778e-01
-3.33114207e-01 4.59910721e-01 4.94034663e-02 -1.49139255e-01
6.26546562e-01 1.48232788e-01 2.49207690e-01 -6.30561531e-01
4.76979017e-01 5.82677066e-01 4.66075957e-01 -7.95736909e-01
6.65706456e-01 9.12864625e-01 4.80431229e-01 -1.39432943e+00
-4.54145432e-01 -1.72331825e-01 -2.45983526e-01 -1.01957053e-01
1.23447847e+00 -5.80779612e-01 -7.67710090e-01 7.51942396e-01
-1.20743024e+00 -3.89705181e-01 -3.61857742e-01 8.91665220e-01
-1.03327954e+00 6.78719044e-01 -7.91574299e-01 -1.44332206e+00
1.27050564e-01 -1.35714436e+00 9.07003880e-01 2.90380657e-01
-1.69047073e-01 -1.24075639e+00 2.83365637e-01 -7.68945962e-02
4.39385533e-01 4.06915843e-01 9.46136534e-01 -1.58219904e-01
-7.12164521e-01 -2.54833788e-01 2.24978644e-02 1.49168596e-01
-3.07514668e-01 -4.61140834e-02 -8.24073672e-01 -3.57049435e-01
4.70868766e-01 -1.10916674e-01 7.19790161e-01 1.10519993e+00
5.82300603e-01 -1.31571621e-01 -4.55348849e-01 3.98174852e-01
1.51665032e+00 9.64261889e-02 5.19811094e-01 -5.30278385e-01
4.62185234e-01 4.28458691e-01 2.95680106e-01 7.81639814e-01
-9.44978371e-02 5.82696676e-01 1.73621014e-01 2.42903218e-01
1.64157376e-01 -1.01532012e-01 1.89555939e-02 8.16847384e-01
-7.97469094e-02 -3.09427410e-01 -9.45869923e-01 4.28468317e-01
-1.87242079e+00 -9.73778248e-01 -6.99846089e-01 2.45798659e+00
9.63401496e-01 -2.20920518e-02 2.74150595e-02 -2.78019756e-01
9.11909759e-01 -1.01025723e-01 -6.90941513e-01 1.13354564e-01
9.17672515e-02 1.22279875e-01 4.24274117e-01 7.42548943e-01
-9.04711843e-01 5.10348439e-01 6.60737419e+00 1.26270998e+00
-3.93636942e-01 5.01845658e-01 4.32698429e-01 2.18553334e-01
-3.10763091e-01 2.41896555e-01 -9.23700929e-01 7.14489460e-01
8.07288170e-01 1.72027290e-01 5.16661346e-01 4.49245691e-01
4.67748046e-01 -8.43527675e-01 -8.59590232e-01 1.10591066e+00
-2.55939931e-01 -1.12335074e+00 -2.63589352e-01 3.10074180e-01
1.23123133e+00 2.23420616e-02 -1.19582087e-01 -1.37208462e-01
5.88095486e-01 -8.15725327e-01 5.82606494e-01 1.04080498e+00
7.48904943e-01 -5.41044414e-01 2.98459858e-01 6.69410169e-01
-6.32747173e-01 4.41251874e-01 -5.05680382e-01 1.01678804e-01
9.91837800e-01 1.50344360e+00 -3.10177058e-01 1.92471836e-02
2.72470325e-01 3.48472923e-01 3.13853204e-01 1.44767809e+00
-3.41258124e-02 8.00077140e-01 -7.71981835e-01 -4.26754914e-02
1.62948102e-01 -1.01222754e+00 1.04375780e+00 9.23647821e-01
6.03097320e-01 8.96945074e-02 -1.27598092e-01 1.54637837e+00
1.87190652e-01 -2.44558319e-01 -4.78465408e-01 -3.75819542e-02
3.74866456e-01 1.00463617e+00 -1.16243410e+00 -4.77479488e-01
-2.70062953e-01 8.34312677e-01 -1.31999731e-01 9.01391029e-01
-7.05783665e-01 -1.17064543e-01 4.05060410e-01 2.09466577e-01
4.52015579e-01 -5.56431890e-01 3.16330492e-02 -1.28710890e+00
-4.17058349e-01 -3.48801076e-01 -1.20279507e-03 -8.25594962e-01
-1.58347142e+00 3.05068731e-01 5.29050171e-01 -1.04385984e+00
-5.10407627e-01 -3.59238952e-01 -6.18822396e-01 1.21338916e+00
-1.30397022e+00 -8.00191164e-01 1.39798611e-01 4.30023104e-01
1.89749941e-01 1.31965995e-01 6.12417579e-01 1.51618078e-01
-2.54421175e-01 -3.50707620e-01 1.04663682e+00 -4.84586328e-01
3.13736767e-01 -1.17081439e+00 1.19436689e-01 6.91650093e-01
-3.21672738e-01 7.11391568e-01 1.17379951e+00 -8.79908323e-01
-1.34638965e+00 -6.69685066e-01 4.09993440e-01 2.28195358e-03
5.99715054e-01 -3.44014674e-01 -9.56740737e-01 4.52528238e-01
-6.82966262e-02 1.06823266e-01 3.62365335e-01 -4.61942375e-01
3.36899877e-01 5.43845415e-01 -1.29282236e+00 5.02381027e-01
6.73507154e-01 -2.57252067e-01 -1.89320043e-01 5.92826903e-01
-6.77198246e-02 -1.39678568e-01 -8.05519342e-01 2.89343111e-02
4.58825052e-01 -7.93599665e-01 1.01259506e+00 -2.35054106e-01
1.54323906e-01 -4.30104166e-01 -1.44340977e-01 -1.32937348e+00
-1.20367602e-01 -1.12131250e+00 -1.93376422e-01 1.02597809e+00
-5.26306592e-02 -7.67622769e-01 6.02573931e-01 4.70042765e-01
1.57110959e-01 -4.19215113e-01 -1.14643526e+00 -1.00782919e+00
3.33884090e-01 -3.25766325e-01 1.35754079e-01 6.82920218e-01
-4.87466902e-01 9.98888165e-02 -4.87621337e-01 2.46624686e-02
1.58155727e+00 -5.37854694e-02 4.53665197e-01 -1.26320791e+00
-6.84153616e-01 -2.65838593e-01 9.33186188e-02 -1.42427397e+00
1.91628020e-02 -7.70948112e-01 4.65184569e-01 -1.38051414e+00
4.17233080e-01 -4.35798258e-01 6.37275577e-01 -5.50392210e-01
8.73930380e-02 1.38005823e-01 -4.10305351e-01 5.39051950e-01
-2.40140393e-01 7.80174553e-01 1.27147830e+00 2.51836479e-01
1.01387138e-02 3.46924216e-01 -2.03819033e-02 1.15194631e+00
2.89194584e-01 -9.06813025e-01 -3.32616925e-01 -9.51705277e-02
4.64007378e-01 7.70407557e-01 5.23266673e-01 -8.90760601e-01
3.14362913e-01 -2.72287250e-01 6.22617314e-03 -7.24057734e-01
5.40425658e-01 -4.88882184e-01 6.37051284e-01 4.12702322e-01
2.88151409e-02 -4.84353423e-01 -3.53368998e-01 9.56647933e-01
5.96526414e-02 -9.62532401e-01 1.39399886e+00 -4.22246665e-01
1.47539750e-01 1.95501462e-01 -1.07126927e+00 2.15493098e-01
6.22078598e-01 -1.56072434e-02 1.06091484e-01 -6.91760063e-01
-1.29162705e+00 -3.79520178e-01 6.17030025e-01 -7.76620507e-01
4.46241826e-01 -1.43452466e+00 -6.27668619e-01 -8.45752507e-02
-3.45738351e-01 1.20498575e-01 4.08034503e-01 1.44458377e+00
-6.78844750e-01 -3.80647555e-02 3.02163184e-01 -7.52554178e-01
-5.59336841e-01 3.25739890e-01 3.56081247e-01 -1.62321657e-01
-5.70566356e-01 6.58068180e-01 3.17647636e-01 -1.90434933e-01
-1.37599215e-01 -2.16592923e-01 2.37717777e-01 2.16176920e-02
4.61514860e-01 9.42911088e-01 -3.60539943e-01 -5.05834341e-01
-5.43502495e-02 6.71890616e-01 3.18681031e-01 -6.69448316e-01
1.14567363e+00 -3.80214661e-01 -3.02216887e-01 6.44228458e-01
1.02622116e+00 -2.90567465e-02 -1.60175347e+00 -1.67093366e-01
-1.79802135e-01 -2.59960830e-01 -5.22717368e-03 -2.02113211e-01
-3.95314336e-01 1.05334246e+00 5.02807081e-01 2.80155152e-01
5.54012418e-01 9.79032218e-02 5.34970641e-01 1.58888593e-01
3.90280694e-01 -9.64657605e-01 -1.94077000e-01 3.54794383e-01
6.86506331e-01 -1.07185292e+00 1.48078173e-01 -3.62269133e-01
-1.99118346e-01 1.13760936e+00 -3.48065227e-01 -4.27577019e-01
1.08076191e+00 3.70700777e-01 -5.90155602e-01 -6.91335425e-02
-2.49413297e-01 -1.46318078e-01 8.91133100e-02 5.64061821e-01
1.87398478e-01 1.93402678e-01 -5.47086060e-01 3.12851250e-01
4.19308186e-01 2.09770992e-01 9.42257941e-01 9.19489026e-01
-3.34304035e-01 -7.78362453e-01 -6.16413295e-01 4.87266868e-01
-3.71808171e-01 -1.62029773e-01 3.68208051e-01 3.68577421e-01
-2.44585544e-01 7.07329750e-01 -1.58796996e-01 6.52660251e-01
-2.01889634e-01 2.09570140e-01 7.35445738e-01 -4.35887843e-01
2.19601169e-01 8.74088585e-01 -1.87911928e-01 -1.11790553e-01
-5.47343314e-01 -1.28973436e+00 -1.07538295e+00 -2.34092116e-01
-4.11766171e-01 2.96026111e-01 9.81276095e-01 1.05916846e+00
-9.01889801e-02 7.38507584e-02 2.76267789e-02 -1.15947664e+00
-9.79959428e-01 -1.17657077e+00 -1.31736887e+00 2.76797324e-01
1.99951857e-01 -7.18171358e-01 -6.58325672e-01 8.24600384e-02] | [6.864263534545898, 3.84503173828125] |
c6f7e261-233f-482b-8e68-37c70f63b246 | skin-lesion-synthesis-with-generative | 1902.03253 | null | http://arxiv.org/abs/1902.03253v1 | http://arxiv.org/pdf/1902.03253v1.pdf | Skin Lesion Synthesis with Generative Adversarial Networks | Skin cancer is by far the most common type of cancer. Early detection is the
key to increase the chances for successful treatment significantly. Currently,
Deep Neural Networks are the state-of-the-art results on automated skin cancer
classification. To push the results further, we need to address the lack of
annotated data, which is expensive and require much effort from specialists. To
bypass this problem, we propose using Generative Adversarial Networks for
generating realistic synthetic skin lesion images. To the best of our
knowledge, our results are the first to show visually-appealing synthetic
images that comprise clinically-meaningful information. | ['Fábio Perez', 'Alceu Bissoto', 'Sandra Avila', 'Eduardo Valle'] | 2019-02-08 | null | null | null | null | ['skin-cancer-classification', 'medical-image-generation'] | ['medical', 'medical'] | [ 4.83838379e-01 3.12594771e-01 -2.51874626e-01 -7.28598684e-02
-9.46779132e-01 -3.44446391e-01 4.72731918e-01 2.34503329e-01
-3.38203251e-01 8.55201423e-01 -5.79053313e-02 -3.14637184e-01
3.27210218e-01 -7.25288987e-01 -5.12648880e-01 -5.66446245e-01
2.21492708e-01 8.42237175e-02 3.14878911e-01 -1.70981452e-01
-4.42453288e-02 6.23063505e-01 -1.15386713e+00 3.99906665e-01
1.11588717e+00 8.42689812e-01 -1.64482012e-01 6.31966829e-01
-6.79510534e-02 4.98573780e-01 -4.17857021e-01 -5.96217752e-01
-1.30548924e-02 -7.95803607e-01 -8.39839220e-01 -6.24856073e-03
5.19361556e-01 -5.74834526e-01 -2.26450488e-01 8.99258375e-01
6.29469037e-01 -4.11793858e-01 6.85376823e-01 -1.01747811e+00
-5.36273181e-01 2.44197533e-01 -5.04631698e-01 -1.20261580e-01
2.84710795e-01 2.12115750e-01 3.47406954e-01 -7.27747619e-01
8.95905674e-01 6.32168412e-01 7.59900451e-01 1.13753569e+00
-1.00948882e+00 -4.05037969e-01 -6.21368513e-02 1.20880820e-01
-1.15566897e+00 -4.00298029e-01 7.46847093e-01 -2.95088589e-01
3.77197772e-01 4.54011232e-01 1.02453649e+00 1.32739329e+00
1.06324978e-01 8.19664419e-01 1.23410165e+00 -5.77939630e-01
2.93593884e-01 1.63034290e-01 -3.51108074e-01 8.05140853e-01
4.79010284e-01 -1.03207923e-01 -3.34981009e-02 8.00826028e-02
7.86087096e-01 8.93109143e-02 -4.27344143e-01 -2.98152208e-01
-6.52620316e-01 7.41377890e-01 5.60180485e-01 3.36068034e-01
-3.89925212e-01 3.29591990e-01 2.92624801e-01 -2.25622848e-01
4.42444026e-01 3.86802495e-01 1.13579623e-01 -5.63238785e-02
-9.83351409e-01 -7.36467093e-02 6.90022707e-01 2.34723970e-01
-4.63202074e-02 -1.75833777e-02 1.44269586e-01 6.27727151e-01
1.51438653e-01 2.73559570e-01 3.88395816e-01 -7.36404538e-01
-1.04400128e-01 6.41093373e-01 2.79483646e-02 -5.81749916e-01
-3.16053838e-01 -4.64438319e-01 -8.34488928e-01 4.71093953e-01
8.44437361e-01 -1.34195253e-01 -1.34319246e+00 1.32987952e+00
2.22498015e-01 8.67030025e-02 1.90474913e-02 9.03197944e-01
8.74664664e-01 1.99311584e-01 4.44486588e-01 -1.83250494e-02
1.24651611e+00 -8.23829710e-01 -6.31359279e-01 -2.86195248e-01
3.16358596e-01 -7.12153256e-01 8.28134954e-01 4.03387904e-01
-1.00994778e+00 -7.62966555e-03 -1.08241892e+00 2.18046993e-01
-2.99781173e-01 7.08070174e-02 1.03291488e+00 9.85194683e-01
-8.86447728e-01 4.77169633e-01 -1.11157155e+00 -9.56211984e-01
8.88778389e-01 3.77746951e-03 -6.73905432e-01 -3.90302628e-01
-1.09712481e+00 9.64035451e-01 6.60000890e-02 1.88062906e-01
-8.17747235e-01 -7.03000724e-01 -6.68557644e-01 -4.00396675e-01
3.56252432e-01 -7.53264904e-01 1.34019029e+00 -1.20386565e+00
-1.29994273e+00 1.03113616e+00 -1.11963272e-01 -2.91536182e-01
7.78706074e-01 1.17751122e-01 -3.17180097e-01 4.12220031e-01
-2.96271712e-01 6.00390017e-01 4.71280307e-01 -1.53263485e+00
-2.06573233e-01 -2.55713969e-01 -1.74668938e-01 -5.22515588e-02
-3.97015899e-01 4.67113778e-03 -5.38041055e-01 -4.91016090e-01
-2.09528014e-01 -1.02921009e+00 -5.48534155e-01 6.73233271e-01
-4.29783404e-01 1.72472864e-01 5.64195216e-01 -1.09623897e+00
9.10552144e-01 -1.95697320e+00 -2.22180933e-01 -3.59794572e-02
2.20990881e-01 6.62944973e-01 8.06226395e-03 4.64583784e-01
2.48381998e-02 5.67005634e-01 -2.30369732e-01 -2.57716507e-01
-3.71392041e-01 -4.29976992e-02 2.72888765e-02 4.75012183e-01
4.37807739e-01 1.12044990e+00 -1.06987143e+00 -6.12682760e-01
2.69487739e-01 8.14710081e-01 -4.87308651e-02 1.04798740e-02
-1.72197446e-01 3.86976629e-01 -3.51061910e-01 8.89628232e-01
4.65958774e-01 -1.92247286e-01 2.10844666e-01 -2.27127075e-01
1.55037373e-01 -1.90952465e-01 -8.15384150e-01 1.60983515e+00
-2.02296555e-01 6.43537939e-01 2.48265136e-02 -6.56829715e-01
4.05554116e-01 3.84043187e-01 4.51668888e-01 -5.34629643e-01
1.28503248e-01 2.72846371e-01 6.82056174e-02 -5.10981798e-01
4.44631875e-02 -3.97958636e-01 1.26656815e-01 -4.89526913e-02
-3.63436311e-01 -4.30883318e-01 1.39370069e-01 4.69722599e-02
1.04585052e+00 -6.22651428e-02 4.47138429e-01 1.97680950e-01
2.14039192e-01 2.89582431e-01 4.24394995e-01 5.00855982e-01
-3.82599384e-01 9.12615299e-01 5.36266387e-01 -4.44774717e-01
-1.06446397e+00 -1.29366767e+00 -1.21368110e-01 1.95273161e-01
-1.41443297e-01 3.52928266e-02 -1.10995734e+00 -7.22033024e-01
-1.28568977e-01 5.64690053e-01 -8.16437781e-01 1.13613851e-01
-2.75898844e-01 -6.72563612e-01 6.85220063e-01 6.26691699e-01
4.02562529e-01 -8.39448392e-01 -2.85536677e-01 9.01151672e-02
4.94672544e-03 -9.95898187e-01 -1.83940858e-01 -1.70687541e-01
-6.36953056e-01 -1.30553603e+00 -1.33186078e+00 -7.70608187e-01
1.26027584e+00 -2.00992361e-01 8.79387796e-01 1.75773993e-01
-1.15844738e+00 1.31382227e-01 -2.42497757e-01 -6.33189261e-01
-6.19137466e-01 -2.22960055e-01 -3.74226213e-01 -2.08167732e-01
-1.83739886e-01 -2.00859845e-01 -9.71537769e-01 4.26809378e-02
-9.22591925e-01 3.09152246e-01 8.41924369e-01 8.39097440e-01
8.22158039e-01 1.95727453e-01 2.60269314e-01 -9.67876613e-01
6.68281794e-01 -1.40298530e-01 -1.95786119e-01 2.86810428e-01
-2.62939155e-01 -3.28473181e-01 7.57128358e-01 -4.67877775e-01
-1.04175842e+00 3.59652877e-01 -1.99298576e-01 -6.63905367e-02
-5.08308113e-01 3.90917063e-01 2.58221686e-01 -4.12434906e-01
8.56824875e-01 1.39541969e-01 6.08014911e-02 -2.32160717e-01
2.23992378e-01 4.60738122e-01 5.12519062e-01 -2.89946538e-03
5.25313437e-01 7.07794666e-01 3.59534532e-01 -8.43309641e-01
-8.10210705e-01 -3.45074646e-02 -3.33233833e-01 -3.57940495e-01
1.04232705e+00 -6.55970991e-01 -3.67130131e-01 7.41127133e-01
-8.87025177e-01 -5.37296236e-01 -1.35029256e-01 1.15695059e-01
-2.49825254e-01 5.43805063e-01 -7.68986583e-01 -7.57060766e-01
-5.76954603e-01 -1.00909150e+00 9.19535637e-01 4.48496610e-01
-2.51113206e-01 -9.98961747e-01 1.29012177e-02 4.11261648e-01
5.28312981e-01 9.62054670e-01 4.09134746e-01 -1.55915886e-01
-4.67979431e-01 -6.37714028e-01 -2.63817132e-01 3.25607866e-01
3.21521819e-01 4.74860311e-01 -1.04625201e+00 -2.74212599e-01
-4.19350266e-01 -4.02345061e-01 8.91327500e-01 5.04700363e-01
1.36880112e+00 3.57364900e-02 -6.64441168e-01 4.55270559e-01
1.52505243e+00 2.23408803e-01 9.03688967e-01 -2.35413294e-02
5.34272611e-01 4.26019222e-01 5.41089356e-01 5.91933429e-02
2.91592419e-01 3.03286791e-01 4.79538083e-01 -8.24715495e-01
-6.28572464e-01 -4.30654556e-01 -1.46769568e-01 1.93461388e-01
-2.41574004e-01 -4.53581244e-01 -1.12081969e+00 6.43250525e-01
-1.34738159e+00 -8.95586729e-01 -5.98063283e-02 2.06343412e+00
7.04851866e-01 1.49687111e-01 -5.67592755e-02 1.45215243e-01
6.34320974e-01 -2.04854980e-01 -4.68454033e-01 -2.46569395e-01
1.97164714e-01 4.34960425e-01 5.07680297e-01 2.78017253e-01
-1.13997805e+00 7.83110559e-01 7.00391817e+00 7.62050390e-01
-1.50507808e+00 -1.20518915e-01 9.12890971e-01 6.71180561e-02
-2.49112099e-01 -6.53987899e-02 -2.33422607e-01 5.01581848e-01
5.89287400e-01 1.72359079e-01 1.90654695e-01 7.86373436e-01
1.50210664e-01 -4.20736402e-01 -9.60866272e-01 7.11892128e-01
2.57285893e-01 -1.30873728e+00 -2.53965676e-01 1.43692464e-01
6.87974751e-01 -2.29748160e-01 1.43378228e-01 -2.44503811e-01
1.81982026e-01 -1.44478917e+00 2.38746881e-01 8.97014260e-01
1.17128360e+00 -6.11533344e-01 8.61609936e-01 1.36331350e-01
-8.10607791e-01 3.28038812e-01 8.71464759e-02 4.11157727e-01
3.26230913e-01 6.18048310e-01 -1.09120524e+00 3.62474680e-01
3.14586133e-01 2.73888469e-01 -6.73698366e-01 1.34618866e+00
-4.58119124e-01 7.43602037e-01 -2.27484405e-01 -4.04146314e-01
-9.75899324e-02 3.95967335e-01 1.17490828e-01 9.29700255e-01
3.02028894e-01 -4.03318778e-02 5.50946034e-02 6.38120353e-01
5.66444397e-02 9.85884741e-02 -6.28442526e-01 -4.81661111e-01
2.04695478e-01 1.34816229e+00 -9.48751986e-01 -4.02369909e-02
-3.96162421e-01 1.23832440e+00 2.54260182e-01 1.28857777e-01
-6.17863834e-01 -3.64076585e-01 3.99957955e-01 3.70430619e-01
-1.84863031e-01 -1.45330038e-02 -3.02411467e-01 -8.63218188e-01
-1.05135990e-02 -7.72744298e-01 3.76718879e-01 -7.00750947e-01
-1.35750735e+00 4.92354155e-01 -4.71881866e-01 -1.02479756e+00
-2.26310849e-01 -6.92332029e-01 -7.86450863e-01 8.05830240e-01
-1.49466074e+00 -1.50955570e+00 -7.47297883e-01 2.02103153e-01
2.89636791e-01 1.75109297e-01 1.31671441e+00 3.27816792e-02
-6.39969349e-01 7.51279950e-01 -5.44892065e-02 1.56394303e-01
7.55219400e-01 -1.19705653e+00 3.76259297e-01 7.55607605e-01
-3.58543396e-01 3.78116757e-01 6.64211094e-01 -7.25427151e-01
-1.30252779e+00 -7.22969055e-01 4.66855377e-01 -2.94543207e-01
6.26758397e-01 -1.92158908e-01 -6.22620523e-01 1.11374155e-01
1.85950756e-01 1.52835399e-01 8.84552002e-01 -3.35144371e-01
-3.67848529e-03 -7.91860595e-02 -1.50627851e+00 8.77075315e-01
5.87791264e-01 -3.41034889e-01 3.09581216e-03 3.82862687e-01
3.57273042e-01 -6.11017764e-01 -1.02503729e+00 5.88015497e-01
7.14349210e-01 -8.90598893e-01 9.59719956e-01 -1.61902621e-01
6.83500826e-01 -1.17148250e-01 3.14267427e-01 -1.45466077e+00
1.11526079e-01 -3.98988575e-01 2.20671013e-01 1.04193199e+00
4.72127557e-01 -5.48588276e-01 1.36420596e+00 7.99295366e-01
1.24543883e-01 -1.14032388e+00 -6.51737869e-01 -5.01770496e-01
1.92866147e-01 -1.97369590e-01 3.29953641e-01 8.51262152e-01
2.02388670e-02 -3.70526284e-01 -2.23134726e-01 -1.19917534e-01
7.09397018e-01 -2.62751430e-01 6.00125015e-01 -8.72597039e-01
-1.64085954e-01 -5.06630182e-01 -4.45335627e-01 -1.02375790e-01
-1.22741446e-01 -5.18534422e-01 -1.50653183e-01 -2.07404304e+00
3.46735090e-01 -2.63106644e-01 -1.31063044e-01 7.23858058e-01
-2.50231624e-01 7.13261366e-01 8.86608213e-02 -3.34040105e-01
-1.65856078e-01 6.40976578e-02 1.69392991e+00 -6.62183762e-02
1.25460923e-01 -1.36517048e-01 -1.01015854e+00 6.91488445e-01
1.29735541e+00 -6.36438504e-02 -4.06874210e-01 -1.29503697e-01
-1.77171543e-01 1.19008422e-01 4.35716242e-01 -9.63425159e-01
1.57252952e-01 -4.36754256e-01 7.84601212e-01 -2.15658799e-01
5.17151713e-01 -6.76360071e-01 3.91619205e-01 9.84759390e-01
-2.62094021e-01 -6.04034662e-01 3.27736825e-01 3.13283533e-01
-1.76133320e-01 -1.88275397e-01 9.62906361e-01 -2.84021109e-01
-5.63179731e-01 2.73836762e-01 -4.19551998e-01 -1.90684512e-01
1.47401893e+00 -2.31367528e-01 -4.48836327e-01 -5.55811703e-01
-5.74203610e-01 -3.35313082e-02 8.03028226e-01 1.43180341e-01
6.98606014e-01 -1.08581471e+00 -7.77389765e-01 -9.29123685e-02
1.23711139e-01 5.44595495e-02 5.69009244e-01 6.17087126e-01
-9.58590627e-01 1.80943042e-01 -2.28705063e-01 -2.11367711e-01
-1.44786346e+00 4.55010086e-01 4.66976523e-01 -3.85967880e-01
-5.27144730e-01 8.42904270e-01 -1.41510099e-01 -4.56202850e-02
7.06458688e-02 4.72867824e-02 -6.89219609e-02 -3.52092773e-01
5.20101964e-01 2.50637919e-01 6.97275922e-02 -1.90382272e-01
-3.80353004e-01 4.08986211e-01 -2.15876967e-01 -8.28316733e-02
1.14571261e+00 4.17350024e-01 -3.64630520e-02 -1.27386481e-01
7.07686424e-01 8.98518190e-02 -1.16768622e+00 4.21645701e-01
-4.99512732e-01 -5.72400868e-01 1.14968926e-01 -1.27215445e+00
-1.06864369e+00 8.25585127e-01 8.88123035e-01 1.56202968e-02
1.32599688e+00 -1.18445471e-01 9.13829982e-01 -6.27250522e-02
3.56645823e-01 -9.77410495e-01 1.93818703e-01 -2.55656809e-01
9.30183470e-01 -1.27891076e+00 1.76107779e-01 -9.27612007e-01
-6.51858747e-01 9.74914730e-01 7.22033024e-01 -8.18130523e-02
3.39095116e-01 5.10382652e-01 4.53599125e-01 -2.58628633e-02
-4.43843812e-01 -2.60660827e-01 3.57944548e-01 7.49067307e-01
6.18673205e-01 2.94997394e-01 -5.89977980e-01 2.76186854e-01
1.81199580e-01 1.76166266e-01 3.75500888e-01 9.40354228e-01
-1.45215839e-01 -1.47719371e+00 -2.97003955e-01 5.09886861e-01
-7.48894393e-01 -6.80161491e-02 -8.35813046e-01 9.77159619e-01
-1.70849320e-02 7.25632727e-01 -3.46255153e-01 -4.11034971e-02
2.07389921e-01 -4.62206230e-02 8.36868048e-01 -3.44126731e-01
-2.01230004e-01 -6.43542483e-02 2.76209295e-01 -2.96152353e-01
-1.96785733e-01 -4.41545814e-01 -1.15396357e+00 -1.52842343e-01
-1.12897716e-01 -4.49657738e-01 1.11467779e+00 5.92628956e-01
1.40830576e-01 5.78058422e-01 3.12390774e-01 -6.54500782e-01
-2.73201495e-01 -6.62676454e-01 -3.85828257e-01 5.38644433e-01
2.37868112e-02 -3.28758746e-01 -1.43514529e-01 4.18354850e-03] | [15.377365112304688, -2.7944066524505615] |
d8b0233f-7898-416f-9ee7-1c9431f0f3a6 | learning-gait-representation-from-massive | 2206.13964 | null | https://arxiv.org/abs/2206.13964v1 | https://arxiv.org/pdf/2206.13964v1.pdf | Learning Gait Representation from Massive Unlabelled Walking Videos: A Benchmark | Gait depicts individuals' unique and distinguishing walking patterns and has become one of the most promising biometric features for human identification. As a fine-grained recognition task, gait recognition is easily affected by many factors and usually requires a large amount of completely annotated data that is costly and insatiable. This paper proposes a large-scale self-supervised benchmark for gait recognition with contrastive learning, aiming to learn the general gait representation from massive unlabelled walking videos for practical applications via offering informative walking priors and diverse real-world variations. Specifically, we collect a large-scale unlabelled gait dataset GaitLU-1M consisting of 1.02M walking sequences and propose a conceptually simple yet empirically powerful baseline model GaitSSB. Experimentally, we evaluate the pre-trained model on four widely-used gait benchmarks, CASIA-B, OU-MVLP, GREW and Gait3D with or without transfer learning. The unsupervised results are comparable to or even better than the early model-based and GEI-based methods. After transfer learning, our method outperforms existing methods by a large margin in most cases. Theoretically, we discuss the critical issues for gait-specific contrastive framework and present some insights for further study. As far as we know, GaitLU-1M is the first large-scale unlabelled gait dataset, and GaitSSB is the first method that achieves remarkable unsupervised results on the aforementioned benchmarks. The source code of GaitSSB will be integrated into OpenGait which is available at https://github.com/ShiqiYu/OpenGait. | ['Shiqi Yu', 'Yongzhen Huang', 'Jilong Wang', 'Saihui Hou', 'Chao Fan'] | 2022-06-28 | null | null | null | null | ['gait-recognition'] | ['computer-vision'] | [-8.97716880e-02 -5.36168218e-01 -4.75381255e-01 -1.80836752e-01
-7.26915956e-01 -1.34256184e-01 2.93759406e-01 -4.11846548e-01
-3.20831776e-01 9.16283906e-01 2.65336633e-01 2.03154370e-01
-7.04511581e-03 -6.93509519e-01 -3.73661071e-01 -9.14059162e-01
-4.86938149e-01 7.06534266e-01 1.76624507e-01 -2.75709361e-01
-2.78716572e-02 1.53836042e-01 -1.66226566e+00 -2.17221051e-01
8.08829904e-01 8.90999019e-01 -5.62560819e-02 5.67843735e-01
4.51801538e-01 3.57874781e-01 -4.03372884e-01 -6.29095137e-01
1.37881696e-01 -4.38164264e-01 -7.59516895e-01 2.21220404e-01
3.70166361e-01 -2.16200978e-01 -4.42230940e-01 8.46892178e-01
8.40670288e-01 4.74117249e-02 8.58013391e-01 -1.50824809e+00
-6.09588027e-01 2.28021875e-01 -7.27924109e-01 1.40674815e-01
4.50843513e-01 6.81098461e-01 8.65058959e-01 -7.28763461e-01
4.29339409e-01 1.05162752e+00 1.09264469e+00 7.39404559e-01
-1.01699805e+00 -6.25451684e-01 -2.28113949e-01 6.50012612e-01
-1.48119867e+00 -3.66316974e-01 9.13493812e-01 -3.59748960e-01
6.16018116e-01 2.49983534e-01 6.76265419e-01 1.67038369e+00
-9.69521105e-02 9.05814588e-01 9.71334875e-01 -2.65185356e-01
2.43434254e-02 -5.87851644e-01 3.10902834e-01 9.29472268e-01
6.16204858e-01 1.25674471e-01 -4.46876019e-01 6.34850562e-02
6.93383276e-01 -2.40577370e-01 -3.55693579e-01 -3.04001629e-01
-1.21866035e+00 5.12160480e-01 2.34795049e-01 1.73479557e-01
-1.84999868e-01 1.51664808e-01 6.28604531e-01 3.37794513e-01
1.78065598e-01 -2.33256832e-01 -1.82219625e-01 -7.64076591e-01
-9.17761087e-01 1.16610289e-01 6.37322605e-01 5.93494058e-01
7.22936869e-01 2.46201128e-01 4.89574000e-02 1.00041640e+00
2.98158497e-01 8.98196042e-01 7.66438067e-01 -7.28150368e-01
5.91544926e-01 2.35072762e-01 3.24043855e-02 -1.14546859e+00
-3.71623516e-01 -2.43782729e-01 -1.10073638e+00 -2.72763789e-01
6.68170393e-01 -1.17422394e-01 -7.79648900e-01 1.66086471e+00
2.82243397e-02 6.68637097e-01 -2.28167668e-01 9.58348811e-01
8.88127327e-01 3.51998508e-01 2.01058984e-01 1.22579582e-01
1.27298462e+00 -9.21037078e-01 -2.61532992e-01 -3.12486917e-01
5.68122029e-01 -4.02444720e-01 1.26196194e+00 2.41690755e-01
-6.05022132e-01 -8.30036461e-01 -1.02289581e+00 3.69804263e-01
-1.20805636e-01 4.76019591e-01 6.44769788e-01 9.25991714e-01
-8.67032170e-01 7.13703096e-01 -1.14158690e+00 -8.77493680e-01
5.50981879e-01 2.28888229e-01 -6.18446767e-01 -1.93506420e-01
-1.18643463e+00 6.02034450e-01 4.50014502e-01 2.47483149e-01
-6.20576084e-01 -2.08658502e-01 -1.04605961e+00 -1.76622033e-01
1.47753447e-01 -8.91126573e-01 7.32205212e-01 -2.33794495e-01
-1.35376990e+00 1.20815229e+00 -8.70829225e-02 -5.04635096e-01
8.20785761e-01 -2.82730043e-01 -5.18101990e-01 1.44445941e-01
3.94377202e-01 4.43719029e-01 1.00345814e+00 -8.71487319e-01
-2.74589717e-01 -3.11145067e-01 -6.22950673e-01 -9.32513401e-02
-4.37799037e-01 -3.52174163e-01 -6.89361989e-01 -8.67375374e-01
-3.33536536e-01 -9.80794668e-01 4.81637381e-02 -3.46062958e-01
-3.13571632e-01 -1.60266727e-01 8.25767636e-01 -8.20616901e-01
1.28779399e+00 -2.03977513e+00 7.61345327e-02 3.33346203e-02
-5.24383560e-02 5.56543887e-01 -3.16399276e-01 4.82665628e-01
-6.00912869e-02 -9.19899046e-02 -4.62165892e-01 -3.95170540e-01
3.76155585e-01 3.70799631e-01 1.19664200e-01 5.26364148e-01
1.76424608e-01 1.18828595e+00 -1.04961741e+00 -6.20013297e-01
4.17651296e-01 2.59296983e-01 -4.58275735e-01 1.18678153e-01
3.86982054e-01 6.74922705e-01 -2.75824189e-01 1.16404903e+00
6.65625155e-01 -3.88635993e-01 8.34191591e-02 -3.43189329e-01
4.23240006e-01 -3.18182558e-01 -1.12880206e+00 1.61190200e+00
-3.71726640e-02 3.71113837e-01 -4.27192867e-01 -1.56784606e+00
1.07677734e+00 6.13945946e-02 7.39268482e-01 -5.65033197e-01
1.59564808e-01 3.56853634e-01 -1.63636208e-01 -6.85263932e-01
2.51214057e-01 7.49779046e-02 -4.99176592e-01 2.28541955e-01
4.15192693e-01 2.54891127e-01 5.78607976e-01 -2.66618550e-01
1.10326612e+00 3.94776851e-01 5.81804693e-01 -8.66204426e-02
6.76907778e-01 -3.43308628e-01 8.70249927e-01 6.98455691e-01
-8.81458879e-01 7.83091724e-01 8.44842270e-02 -4.35086727e-01
-7.88632035e-01 -1.43471253e+00 -3.21544968e-02 7.23498404e-01
1.49311617e-01 -5.98173201e-01 -6.90829217e-01 -5.57498932e-01
2.43865699e-02 -1.00740328e-01 -5.95385671e-01 -2.50086427e-01
-7.52602935e-01 -1.20409667e+00 9.96678889e-01 8.80566955e-01
1.22901118e+00 -1.07289755e+00 -4.58834350e-01 1.77659169e-01
-6.82764530e-01 -1.17595398e+00 -7.12430954e-01 -2.17652366e-01
-6.84058666e-01 -1.29982996e+00 -1.15067351e+00 -8.77566695e-01
2.60334373e-01 -6.23939373e-02 1.15590966e+00 2.65790612e-01
-6.34251893e-01 4.94817108e-01 -4.81434196e-01 2.42688715e-01
3.28204781e-02 2.49815062e-01 3.80349487e-01 1.41193777e-01
6.71064138e-01 -7.72397697e-01 -6.34753287e-01 5.87535441e-01
-1.63609877e-01 -1.75268665e-01 4.88878012e-01 1.31503916e+00
2.57610440e-01 2.51436196e-02 7.46395409e-01 -3.47476363e-01
3.39007914e-01 -4.04419482e-01 -1.34170175e-01 1.47650555e-01
-1.13034397e-01 -1.74784333e-01 4.43977416e-01 -3.19517761e-01
-7.52946615e-01 -1.33086396e-02 -4.50107425e-01 -2.74209529e-01
-5.51429272e-01 4.24254328e-01 -4.58744884e-01 -8.04735161e-03
5.89320898e-01 5.52291989e-01 -1.06620312e-01 -4.38899815e-01
1.78347513e-01 5.33286273e-01 9.80119348e-01 -1.17037451e+00
1.09898138e+00 3.66713375e-01 3.23627852e-02 -1.16132045e+00
-3.01095158e-01 -5.79211652e-01 -9.99991655e-01 -3.76732230e-01
6.48188114e-01 -8.70777667e-01 -7.44898140e-01 1.17594862e+00
-4.78115141e-01 -6.26363397e-01 -2.38034874e-01 3.92868817e-01
-9.17832255e-01 9.88179147e-01 -9.47538197e-01 -5.25287509e-01
-3.68462235e-01 -9.35046613e-01 1.10615349e+00 1.82583660e-01
-4.80090648e-01 -1.12396908e+00 1.46640673e-01 5.73617578e-01
1.79693282e-01 5.34551203e-01 4.59549189e-01 -3.20476443e-01
-1.37830153e-01 -2.01389775e-01 -1.48520797e-01 2.37433001e-01
5.45008183e-01 1.02578814e-03 -7.68823206e-01 -4.31068391e-01
-5.95974147e-01 -5.08965492e-01 1.10858166e+00 4.14829016e-01
8.59741807e-01 -1.67707186e-02 -3.30045283e-01 6.79280102e-01
1.22112966e+00 2.15017609e-02 8.40683401e-01 4.99699801e-01
8.67269278e-01 3.05215091e-01 7.26381660e-01 5.66546142e-01
5.67707598e-01 6.99389815e-01 1.50256380e-02 1.36706978e-01
-2.08648324e-01 -3.62686485e-01 4.21658576e-01 1.12211668e+00
-7.28677154e-01 -1.65103495e-01 -1.09825170e+00 6.25590384e-01
-2.07806134e+00 -1.20512998e+00 1.41557619e-01 2.21173429e+00
4.68841165e-01 1.11484133e-01 8.09099495e-01 5.68737626e-01
8.09432626e-01 2.03541979e-01 -4.47635174e-01 1.39009923e-01
-1.40209526e-01 4.35016900e-01 3.18824708e-01 5.31944670e-02
-1.65953159e+00 8.98335755e-01 5.55443954e+00 8.88322234e-01
-9.71565425e-01 -1.14503406e-01 4.13525552e-01 3.35719138e-01
5.76545835e-01 -5.28690636e-01 -6.41894042e-01 9.71718848e-01
9.83579993e-01 -1.61813289e-01 3.64713441e-03 7.58311749e-01
1.69994101e-01 1.94356233e-01 -9.25803304e-01 1.44887674e+00
-2.76508480e-02 -9.70383584e-01 -4.98288348e-02 1.41985878e-01
4.03252631e-01 -1.00638747e-01 1.33362338e-01 4.07904744e-01
2.07492530e-01 -8.82502377e-01 2.20553443e-01 3.29573095e-01
9.41644967e-01 -7.30401039e-01 8.72325778e-01 7.76782408e-02
-1.90504181e+00 -1.87817514e-02 -2.59699166e-01 -9.42308530e-02
3.73226166e-01 3.69947284e-01 -2.42200091e-01 9.97936606e-01
9.26075518e-01 1.48250902e+00 -7.67349124e-01 1.21886456e+00
-2.03920782e-01 1.03298390e+00 -2.09775597e-01 2.18529776e-01
8.88163894e-02 -2.36764207e-01 4.11958694e-01 1.41762877e+00
5.97964466e-01 -2.16714948e-01 4.05897290e-01 2.84457624e-01
-6.24929108e-02 -2.05381494e-03 -4.78585511e-01 -1.18835814e-01
3.80996853e-01 1.02380216e+00 -7.07995534e-01 -4.29546893e-01
-3.99676323e-01 1.15212989e+00 8.46073776e-02 1.87310115e-01
-9.72583413e-01 -1.93789825e-01 8.07197452e-01 2.75722910e-02
3.80434424e-01 -2.09618464e-01 5.02832532e-02 -1.78262246e+00
4.35807444e-02 -8.70698273e-01 7.29656577e-01 -3.53006124e-01
-1.80481410e+00 3.54028374e-01 -1.98972803e-02 -1.61353695e+00
-5.35472095e-01 -8.45927298e-01 -6.99154556e-01 4.77550089e-01
-1.24436724e+00 -1.43170142e+00 -6.89966679e-01 7.43794262e-01
3.59054387e-01 -3.85534912e-01 9.32814360e-01 8.08367670e-01
-8.07761908e-01 9.74348128e-01 -3.07406522e-02 7.82852530e-01
9.50811267e-01 -1.11414337e+00 7.11563826e-01 9.64903057e-01
4.38661240e-02 4.02967215e-01 5.69795966e-01 -6.67849004e-01
-1.24281573e+00 -1.16112459e+00 4.58997935e-01 -3.98980260e-01
7.76663542e-01 -1.08757101e-01 -1.01480532e+00 7.09339917e-01
-1.58835694e-01 2.32148707e-01 7.52905786e-01 -6.77407235e-02
-2.31268033e-01 -1.01681910e-01 -1.13070571e+00 5.49127102e-01
1.54744732e+00 -2.93575257e-01 -7.22123504e-01 -3.75348218e-02
-4.38029543e-02 -2.59289652e-01 -1.03986371e+00 5.49455523e-01
8.12559128e-01 -8.55001092e-01 1.28282189e+00 -3.90689224e-01
1.71053126e-01 -2.81371951e-01 -3.06225598e-01 -1.35979462e+00
-3.39307278e-01 -3.93063784e-01 -3.49744588e-01 1.32424617e+00
-3.08036298e-01 -8.34395170e-01 1.28902090e+00 9.09503028e-02
1.33968405e-02 -4.67988491e-01 -9.21801627e-01 -1.35511017e+00
1.13391958e-01 -3.76975268e-01 6.10246956e-01 9.71326411e-01
4.70231995e-02 -7.16465642e-04 -9.65625584e-01 -6.49065673e-02
1.25360072e+00 7.25784451e-02 1.09955370e+00 -1.32738388e+00
-4.02033716e-01 -3.65365565e-01 -1.13034451e+00 -9.84530270e-01
2.63874710e-01 -7.96806574e-01 -3.40882316e-02 -1.24749553e+00
2.01766729e-01 -1.37544408e-01 -2.84792691e-01 6.53086543e-01
-3.85122120e-01 7.23446310e-01 2.08705038e-01 2.65829325e-01
-6.04321420e-01 5.68164885e-01 1.13868153e+00 -4.02149528e-01
1.13302618e-01 -8.12018290e-03 -3.77541721e-01 7.22196937e-01
1.00920284e+00 1.18131302e-01 -2.61750817e-01 1.68842718e-01
-7.05985546e-01 -1.50393873e-01 5.79013348e-01 -1.44609308e+00
-1.66994967e-02 2.21201461e-02 4.60600942e-01 -6.10274732e-01
4.26171690e-01 -3.42319936e-01 2.22589374e-01 8.19437683e-01
3.53058994e-01 -9.78050232e-02 8.04131031e-02 3.53785157e-01
-3.77660722e-01 1.81578636e-01 7.01116383e-01 -1.34504020e-01
-1.43268764e+00 6.83895528e-01 -3.85520905e-01 3.87434036e-01
7.55636990e-01 -5.12694478e-01 -3.51905584e-01 -3.44841510e-01
-9.24375236e-01 1.77047059e-01 5.37691355e-01 3.79052430e-01
6.56259537e-01 -1.63581252e+00 -8.15218270e-01 2.49668539e-01
2.83910900e-01 -5.37865281e-01 4.68674034e-01 7.35941947e-01
-4.50907201e-01 3.87112439e-01 -7.17786610e-01 -8.27204823e-01
-1.30598855e+00 3.22224468e-01 2.22706735e-01 -4.96955037e-01
-7.93243289e-01 4.20751035e-01 -1.79141492e-01 -4.94818300e-01
-9.67245996e-02 -6.99938834e-02 -9.70603824e-02 -5.46780303e-02
4.95134592e-01 6.46128416e-01 -1.52089939e-01 -1.05816019e+00
-6.23699605e-01 1.07150257e+00 3.01391155e-01 3.34232152e-01
1.26231003e+00 -1.51807874e-01 2.86967188e-01 2.62262791e-01
1.14762533e+00 -5.16410112e-01 -1.42005599e+00 -3.35168615e-02
2.25837469e-01 -5.12121439e-01 -8.86401415e-01 -3.60535473e-01
-1.14637637e+00 1.05366647e+00 7.96293616e-01 -3.12753260e-01
1.23146534e+00 -2.95400143e-01 1.04987752e+00 3.24487954e-01
9.01391983e-01 -8.98716986e-01 5.16067982e-01 3.54991794e-01
5.53631186e-01 -1.44435477e+00 -2.42885411e-01 -3.97761792e-01
-5.81054628e-01 9.31727588e-01 7.16032028e-01 -2.93754995e-01
4.80693817e-01 1.78983569e-01 -1.72345806e-02 1.66141570e-01
-2.72331446e-01 -3.75570178e-01 3.47724140e-01 1.27101421e+00
2.89449662e-01 2.97118276e-01 -3.00212055e-01 7.05017626e-01
-3.47993284e-01 3.43580306e-01 8.13547075e-02 7.70050645e-01
-1.81710377e-01 -1.26861584e+00 -4.68633264e-01 4.43648905e-01
-1.34411469e-01 2.57940501e-01 6.01881333e-02 9.12166536e-01
3.77889834e-02 6.93839312e-01 -2.42677167e-01 -7.13607550e-01
2.24573672e-01 1.12836443e-01 7.31463492e-01 -2.32668281e-01
8.37686807e-02 -2.47355357e-01 1.91965282e-01 -5.38165629e-01
-6.20260954e-01 -8.32408488e-01 -1.00463939e+00 -6.29654288e-01
3.31548005e-01 3.20473611e-02 -2.48612925e-01 8.65796804e-01
1.84376732e-01 2.11993262e-01 3.59812766e-01 -1.21814179e+00
-3.58509779e-01 -7.31717885e-01 -7.55642056e-01 7.62549937e-01
4.13207300e-02 -1.13079071e+00 -1.68363973e-01 3.14904839e-01] | [14.30085277557373, 1.4248937368392944] |
c580219e-0fa4-4f95-a229-1bcbefcbd63f | impact-of-microphone-position-measurement | null | null | https://aclanthology.org/2021.icon-main.23 | https://aclanthology.org/2021.icon-main.23.pdf | Impact of Microphone position Measurement Error on Multi Channel Distant Speech Recognition & Intelligibility | It was shown in (Raikar et al., 2020) that the measurement error in the microphone position affected the room impulse response (RIR) which in turn affected the single channel speech recognition. In this paper, we ex-tend this to study the more complex and realistic scenario of multi channel distant speech recognition. Specifically we simulate m speakers in a given room with n microphones speaking without overlap. Then channel audio is beamformed and passed through a speech to text (s2t) engine. We compare the s2t accuracy when the microphone locations are known exactly (ground truth) with the s2t accuracy when there is a measurement error in the location of the microphone. We report the performance of an end-to-end s2t on beamformed input in terms of character error rate (CER) and and also speech intelligibility and quality in terms of STOI and PESQ respectively. | ['Sunil Kumar Kopparapu', 'Karan Nathwani'] | null | null | null | null | icon-2021-12 | ['room-impulse-response', 'distant-speech-recognition'] | ['audio', 'speech'] | [ 2.15672582e-01 -4.16007340e-02 9.46898520e-01 -2.52841949e-01
-1.29580748e+00 -6.55441165e-01 3.36874187e-01 -2.03464687e-01
-3.44223708e-01 5.45791686e-01 4.75795567e-01 -4.98676389e-01
-8.08012262e-02 -3.32263201e-01 -7.76081741e-01 -6.73074186e-01
5.28601594e-02 1.39663190e-01 1.57176495e-01 5.34113497e-02
6.15501171e-03 3.88022602e-01 -1.39027905e+00 1.81865349e-01
3.33909452e-01 9.95277941e-01 3.15326750e-01 1.60483623e+00
1.46754161e-01 5.60908914e-01 -1.23292065e+00 -1.29250079e-01
2.97301501e-01 -2.31488153e-01 -3.73435646e-01 -6.62617981e-02
2.70181775e-01 -2.13785410e-01 -3.85241419e-01 9.86406446e-01
1.17614388e+00 2.18078613e-01 4.77864236e-01 -6.78944647e-01
1.07980922e-01 6.06844366e-01 -1.91614568e-01 1.50558427e-01
6.79846883e-01 -1.36247218e-01 5.07740676e-01 -9.24130261e-01
-6.48510233e-02 1.31418073e+00 8.48349154e-01 2.73494899e-01
-8.78683925e-01 -7.45389044e-01 -2.65968353e-01 2.05421932e-02
-1.57097960e+00 -1.08478248e+00 4.01170224e-01 -2.55146325e-01
9.07937586e-01 7.00085998e-01 -3.06450296e-02 1.00452065e+00
-4.87321876e-02 3.59069258e-01 1.17259002e+00 -7.64303386e-01
2.62260050e-01 2.60390550e-01 -5.01452805e-03 2.78890193e-01
-4.12416011e-01 4.50569123e-01 -5.85665286e-01 -2.26565474e-03
6.15136445e-01 -6.67834222e-01 -5.22792220e-01 6.55204177e-01
-1.25665379e+00 3.09756070e-01 -9.99798253e-03 3.85196626e-01
-3.06146920e-01 1.39532471e-02 1.48216844e-01 4.84841079e-01
6.94827437e-02 1.47357926e-01 -1.57867476e-01 -4.17798340e-01
-9.39762533e-01 -1.62254378e-01 1.19040322e+00 1.01348221e+00
-1.37462514e-02 3.61401737e-01 -2.09350750e-01 9.88178909e-01
5.55817902e-01 1.26599920e+00 3.42472345e-01 -6.51353776e-01
8.66987884e-01 -6.59762561e-01 2.67611921e-01 -8.28647852e-01
-3.03230226e-01 -5.04035771e-01 -8.36582124e-01 7.56176040e-02
5.30669212e-01 -5.85783541e-01 -7.09325135e-01 1.27782190e+00
1.40186355e-01 2.96787202e-01 4.21781600e-01 8.89417827e-01
5.16512692e-01 1.05083907e+00 -5.77069938e-01 -3.57464701e-01
9.93512154e-01 -9.22653913e-01 -1.11127985e+00 -1.09840028e-01
1.92644820e-01 -1.54139745e+00 7.81694531e-01 7.65170515e-01
-1.08258176e+00 -7.41626501e-01 -1.04938197e+00 5.87253869e-01
-1.25010103e-01 1.23608023e-01 -3.72049928e-01 1.47714293e+00
-1.11486244e+00 3.79928574e-02 -5.58958411e-01 -1.50026791e-02
-3.71564090e-01 1.99705154e-01 -1.23247810e-01 1.02230623e-01
-1.06093800e+00 6.90318882e-01 -4.19803143e-01 5.43714464e-01
-9.25631166e-01 -2.98113108e-01 -4.16100413e-01 -1.61566753e-02
2.10769445e-01 -2.43656486e-01 1.77194881e+00 -4.85092908e-01
-2.03273201e+00 6.87958002e-02 -4.26118821e-01 -2.05998003e-01
5.29362381e-01 -2.83552498e-01 -1.04602718e+00 -1.18052498e-01
-2.14919344e-01 -1.93952993e-01 8.17471445e-01 -1.27193415e+00
-4.22887862e-01 -3.04638714e-01 -5.25352359e-01 2.61663109e-01
3.67734097e-02 1.51514620e-01 -1.13661751e-01 -5.97373068e-01
3.92126083e-01 -7.71352172e-01 9.61908326e-02 -4.75461841e-01
-1.06748283e+00 4.06225681e-01 5.05507350e-01 -1.10733283e+00
1.20052671e+00 -2.28734875e+00 -2.52259195e-01 5.69481969e-01
-3.07730377e-01 3.10701370e-01 8.85145217e-02 6.45613015e-01
-1.15444712e-01 6.77621812e-02 7.07725389e-03 -7.26180553e-01
-5.71692623e-02 -1.20429642e-01 -3.74770492e-01 5.25645792e-01
-6.46405697e-01 1.90312594e-01 -2.51788318e-01 -9.07219648e-02
3.79979223e-01 6.76485419e-01 -3.29657108e-01 5.38049221e-01
4.93542016e-01 4.60999101e-01 -2.31574848e-01 3.56789827e-01
6.58439338e-01 5.68701029e-01 -2.85248220e-01 8.87978598e-02
-4.54318821e-01 4.74568158e-01 -1.69961894e+00 1.20205557e+00
-1.06272829e+00 9.93942022e-01 6.07487142e-01 -4.34906483e-01
1.06319404e+00 1.05417967e+00 -2.13129416e-01 -5.58589280e-01
1.02718383e-01 4.13310826e-01 3.30486931e-02 -6.24542713e-01
3.53674173e-01 -1.65702090e-01 1.56940505e-01 2.09046245e-01
-3.81092906e-01 -2.71435469e-01 -7.17228174e-01 -2.45928958e-01
9.18226898e-01 -8.15835297e-01 -2.15954706e-03 -2.39027679e-01
5.99839091e-01 -5.68277597e-01 -1.47407681e-01 8.58771980e-01
-4.86070141e-02 8.54801118e-01 -9.89322364e-02 4.12851870e-01
-1.19925392e+00 -1.22454786e+00 1.05905263e-02 6.04435384e-01
-3.22735459e-01 -5.06764315e-02 -9.71640706e-01 2.78070688e-01
-3.49414855e-01 9.92553711e-01 -1.20168053e-01 1.39145941e-01
-4.86433387e-01 -1.82383314e-01 1.13805282e+00 3.54090333e-01
6.58597767e-01 -5.04305363e-01 -2.32067287e-01 2.45275319e-01
-2.11698353e-01 -1.18080163e+00 -7.83901095e-01 3.66528034e-01
-3.85089815e-01 -4.09934908e-01 -7.88591623e-01 -4.83978271e-01
3.86750907e-01 1.11664779e-01 4.33949441e-01 -5.20881593e-01
-7.14464784e-02 6.14772797e-01 -2.83521593e-01 -6.32642031e-01
-7.91547000e-01 -4.12034720e-01 2.85200804e-01 2.15021953e-01
-3.10880512e-01 -3.03140491e-01 -3.85319144e-01 5.87434828e-01
-5.17727673e-01 -3.94410729e-01 2.61484563e-01 5.09724736e-01
2.02548783e-02 4.19826031e-01 3.33027482e-01 -2.51939952e-01
8.60304177e-01 -4.45295013e-02 -6.45106196e-01 1.82507351e-01
-3.51681620e-01 -1.37557641e-01 6.82053208e-01 -3.23963225e-01
-1.18166935e+00 -1.48459941e-01 -7.49832630e-01 -2.32685149e-01
-3.43463987e-01 1.49095505e-01 -5.31525970e-01 6.92639053e-02
6.26094580e-01 4.04845953e-01 -1.76340610e-01 -6.85548425e-01
2.91853454e-02 1.45007718e+00 6.95529222e-01 -2.69253671e-01
8.07082236e-01 -7.84133822e-02 -1.57295108e-01 -1.32261384e+00
-6.62134439e-02 -6.27369165e-01 -2.01883823e-01 -2.30032369e-01
7.10331857e-01 -8.27701986e-01 -9.11116719e-01 8.02188993e-01
-1.48434591e+00 -1.92038223e-01 1.54309019e-01 1.00622451e+00
-3.68057251e-01 2.18485698e-01 -4.12095100e-01 -1.68480134e+00
-1.38564378e-01 -1.24296236e+00 9.81637001e-01 -9.52423438e-02
-5.70619963e-02 -9.01275814e-01 5.53207397e-02 5.25991857e-01
7.49768853e-01 -4.24192131e-01 5.11100948e-01 -5.58251858e-01
-2.92143643e-01 -4.33361113e-01 2.91485816e-01 6.44317448e-01
-1.86441895e-02 -3.65860015e-01 -1.62442601e+00 -2.07540870e-01
5.18467009e-01 2.53028840e-01 1.83633447e-01 6.58989489e-01
7.83752680e-01 -5.83559334e-01 3.03566009e-02 2.56955445e-01
1.35659432e+00 6.30101442e-01 8.48585308e-01 -1.39745235e-01
3.75393927e-01 4.55326676e-01 6.77557960e-02 3.88347089e-01
-1.70161054e-01 8.79543543e-01 2.36210823e-02 2.62232088e-02
-3.49300921e-01 -1.76437423e-01 5.08495390e-01 1.26315057e+00
9.78574678e-02 -9.52598631e-01 -9.67077494e-01 2.47438610e-01
-1.05700719e+00 -9.72818077e-01 -2.65711784e-01 2.59493995e+00
5.29779792e-01 1.28522068e-02 -6.15185536e-02 6.66521788e-01
9.40562189e-01 -1.49053976e-01 -1.01523168e-01 -7.88380504e-01
-4.95806895e-02 7.16846585e-02 9.77056980e-01 1.39642107e+00
-5.45043170e-01 1.12541728e-01 6.61059952e+00 7.28463709e-01
-1.09145224e+00 1.88156754e-01 6.10014260e-01 3.92254516e-02
5.78569844e-02 -6.14685655e-01 -8.00360203e-01 3.37244868e-01
1.49269032e+00 1.90799251e-01 7.91029155e-01 3.65888864e-01
5.71278811e-01 -2.13787794e-01 -1.02875173e+00 1.18590629e+00
1.36841327e-01 -5.51539302e-01 -5.08880734e-01 -1.07884500e-02
2.09793627e-01 -1.55909315e-01 3.59233379e-01 -1.12774998e-01
-6.10375069e-02 -1.30104935e+00 1.00593376e+00 5.21568477e-01
8.40618253e-01 -4.07036841e-01 6.33190870e-01 5.04259229e-01
-1.18785644e+00 -6.41912296e-02 1.10128544e-01 -1.05511481e-02
3.32651645e-01 6.22537732e-01 -1.54806221e+00 3.56975704e-01
4.09426600e-01 -4.37312394e-01 8.17456171e-02 1.21194577e+00
-5.65818734e-02 1.07509637e+00 -5.78500092e-01 -2.47691661e-01
1.22059055e-01 1.85283512e-01 8.67254913e-01 1.39404786e+00
7.47688353e-01 2.61287451e-01 -2.37572417e-01 4.34183627e-01
2.28099927e-01 -9.50154811e-02 -3.49771291e-01 4.57581699e-01
8.25824916e-01 6.00040436e-01 -3.66283894e-01 -4.86037806e-02
-1.45087406e-01 1.05823874e+00 -7.64985919e-01 7.39620447e-01
-6.25173867e-01 -6.09391093e-01 4.70195085e-01 6.88725114e-02
2.21149176e-01 -3.56406063e-01 -3.09444904e-01 -4.65694666e-01
2.01470137e-01 -7.89332271e-01 -3.67162317e-01 -9.07351434e-01
-8.15959096e-01 7.70942211e-01 -2.08618283e-01 -1.06757212e+00
-2.55593717e-01 -6.03766024e-01 -4.08797652e-01 1.53990209e+00
-1.06964767e+00 -5.23203731e-01 3.15254182e-02 6.46524668e-01
6.40163779e-01 -7.82751590e-02 9.91332710e-01 7.02027917e-01
-4.64184999e-01 9.65545893e-01 8.35839272e-01 3.15212965e-01
5.20704627e-01 -9.99230921e-01 5.31551600e-01 9.07786667e-01
-1.20915502e-01 5.50853729e-01 1.21621537e+00 -2.80675620e-01
-1.42056894e+00 -9.49050903e-01 1.00694573e+00 -3.62090617e-01
2.68668652e-01 -6.52649105e-01 -5.44316769e-01 3.30660135e-01
1.52245849e-01 -3.33057642e-01 5.93713105e-01 -1.96379766e-01
-1.79913849e-01 -1.44070208e-01 -1.14246702e+00 2.71106660e-01
4.99629378e-01 -9.39511359e-01 -2.60328054e-01 2.38607645e-01
5.43410778e-01 -5.47944844e-01 -6.22544646e-01 -1.59536660e-01
5.84776223e-01 -9.24100935e-01 9.52830911e-01 1.60120904e-01
-2.69946605e-01 -2.14651674e-01 -9.62955236e-01 -1.47267878e+00
2.58479983e-01 -1.01699388e+00 6.29626334e-01 1.30572152e+00
8.58425856e-01 -7.47691929e-01 2.39158139e-01 5.27507007e-01
-3.80622834e-01 -1.60259798e-01 -1.42807639e+00 -1.08082926e+00
-1.69711206e-02 -8.60191464e-01 4.39682990e-01 2.28062883e-01
3.22313979e-02 3.38970631e-01 -6.35469317e-01 8.43852758e-01
4.85666871e-01 -5.88451385e-01 5.27942240e-01 -8.45980644e-01
-5.62719464e-01 -1.01930894e-01 -1.56973109e-01 -1.54580677e+00
-4.45100188e-01 -2.20733687e-01 6.77820802e-01 -1.27021050e+00
-5.75681865e-01 -3.33029062e-01 1.60387963e-01 -2.53390074e-01
1.38180211e-01 -4.54934925e-01 1.75761402e-01 -1.10421911e-01
-2.78409198e-02 2.77300209e-01 1.00195277e+00 1.71617549e-02
-2.23304465e-01 6.13277853e-01 -7.52186105e-02 3.57399970e-01
6.16367996e-01 -3.83947611e-01 -3.48296642e-01 -5.91121912e-01
-1.93503127e-01 7.54954517e-01 3.64599138e-01 -1.49350870e+00
5.86560786e-01 4.29105967e-01 1.14083834e-01 -5.48815429e-01
8.25116336e-01 -1.18820941e+00 3.72908324e-01 3.58347148e-01
-5.20716608e-01 4.48225588e-02 3.20183545e-01 5.66813111e-01
-1.99486017e-01 -8.16866010e-02 6.19628072e-01 7.02984259e-02
1.11124888e-01 -3.51290137e-01 -7.32856154e-01 -3.21620196e-01
5.59678257e-01 -1.96413279e-01 -2.46648997e-01 -8.30077648e-01
-7.70637631e-01 -3.30048710e-01 -1.57851741e-01 1.06175438e-01
6.69481695e-01 -9.38450038e-01 -6.99242115e-01 3.41900468e-01
-4.61330891e-01 -1.84840605e-01 1.94068789e-01 6.92252755e-01
-3.57465535e-01 9.02272105e-01 5.11812270e-01 -5.44772387e-01
-1.61374927e+00 2.25135192e-01 7.82353282e-01 4.12053645e-01
-2.14936242e-01 1.07565176e+00 -1.84447795e-01 -2.42261246e-01
7.74328232e-01 -4.67871219e-01 1.73253044e-01 -5.46153665e-01
6.89953506e-01 5.67099512e-01 3.86467755e-01 -6.16869092e-01
-2.39905924e-01 5.86607635e-01 4.85402316e-01 -1.02237785e+00
6.98144734e-01 -4.59157884e-01 2.42440373e-01 8.48896682e-01
1.43367040e+00 6.18623435e-01 -6.37594461e-01 -1.36077508e-01
-4.63404864e-01 -5.59449553e-01 2.91188329e-01 -1.13635314e+00
-6.04693055e-01 1.13604140e+00 1.17342818e+00 3.43917429e-01
1.05393267e+00 -2.67755151e-01 7.77992785e-01 5.30633092e-01
2.48240963e-01 -9.69448566e-01 -1.70974568e-01 4.97032523e-01
7.92538702e-01 -8.13149214e-01 -6.93526626e-01 -4.34674442e-01
-3.27396750e-01 1.14000571e+00 5.48924990e-02 5.22502482e-01
9.32080626e-01 7.34939516e-01 5.08212566e-01 3.37999523e-01
-2.99294055e-01 2.35614926e-01 -6.36604503e-02 7.31869638e-01
5.61961234e-01 3.47805262e-01 3.76938730e-01 4.50526953e-01
-6.72412932e-01 -4.38933134e-01 4.45346236e-01 5.39938450e-01
-7.45553076e-01 -7.37068892e-01 -1.30563843e+00 1.04207486e-01
-4.14527804e-01 -3.83064687e-01 -2.45711491e-01 1.14736468e-01
-2.41148308e-01 1.84679425e+00 -1.72429129e-01 -6.29844606e-01
8.19161832e-01 2.28389561e-01 5.83840683e-02 -2.32091829e-01
-6.07410431e-01 4.17399675e-01 4.71168935e-01 -1.65837765e-01
-4.90206704e-02 -7.32827425e-01 -1.07553160e+00 -5.42442203e-01
-4.94830608e-01 3.64352942e-01 1.34224856e+00 7.41649210e-01
5.42724617e-02 9.76099372e-01 9.00902569e-01 -4.77851301e-01
-8.59253883e-01 -1.21303844e+00 -7.50392318e-01 -1.34134859e-01
8.83555472e-01 6.24998473e-03 -6.84420645e-01 -2.77791452e-02] | [15.014361381530762, 5.87323522567749] |
abdb7d2d-3957-4ced-8baf-df0543b6ad6a | dont-classify-translate-multi-level-e | 1812.05774 | null | http://arxiv.org/abs/1812.05774v1 | http://arxiv.org/pdf/1812.05774v1.pdf | Don't Classify, Translate: Multi-Level E-Commerce Product Categorization Via Machine Translation | E-commerce platforms categorize their products into a multi-level taxonomy
tree with thousands of leaf categories. Conventional methods for product
categorization are typically based on machine learning classification
algorithms. These algorithms take product information as input (e.g., titles
and descriptions) to classify a product into a leaf category. In this paper, we
propose a new paradigm based on machine translation. In our approach, we
translate a product's natural language description into a sequence of tokens
representing a root-to-leaf path in a product taxonomy. In our experiments on
two large real-world datasets, we show that our approach achieves better
predictive accuracy than a state-of-the-art classification system for product
categorization. In addition, we demonstrate that our machine translation models
can propose meaningful new paths between previously unconnected nodes in a
taxonomy tree, thereby transforming the taxonomy into a directed acyclic graph
(DAG). We discuss how the resultant taxonomy DAG promotes user-friendly
navigation, and how it is more adaptable to new products. | ['Maggie Yundi Li', 'Liling Tan', 'Stanley Kok'] | 2018-12-14 | null | null | null | null | ['product-categorization'] | ['miscellaneous'] | [ 9.16685387e-02 1.28668651e-01 -8.72222483e-01 -7.04300225e-01
-1.71415538e-01 -1.15160358e+00 3.72138768e-01 7.20380783e-01
2.01510563e-01 1.76120475e-01 9.76002067e-02 -1.00310183e+00
8.55666306e-03 -1.36155760e+00 -2.81362206e-01 -4.96894792e-02
-6.16677217e-02 5.72374165e-01 1.95822626e-01 -2.44116738e-01
3.01024824e-01 1.71990126e-01 -1.50373435e+00 5.70814550e-01
1.01892662e+00 8.80424142e-01 -1.02169454e-01 1.38930783e-01
-8.71811330e-01 3.50216389e-01 -2.51875907e-01 -1.05198050e+00
2.99414098e-01 -3.62804443e-01 -1.14658010e+00 2.33982012e-01
2.33813763e-01 1.17449872e-01 1.54858723e-01 1.43518186e+00
-5.96039653e-01 -1.70935050e-01 8.85555387e-01 -1.65054989e+00
-1.06543720e+00 1.07769549e+00 -6.44278347e-01 -1.93160713e-01
5.88359594e-01 -4.66929615e-01 1.41911077e+00 -7.71602154e-01
7.89160848e-01 1.32463014e+00 7.14566290e-01 2.22266704e-01
-1.50521970e+00 -5.79369307e-01 6.18396938e-01 2.90381789e-01
-1.25803590e+00 2.84231067e-01 8.41050386e-01 -7.93187499e-01
6.39686525e-01 1.72599122e-01 6.26927495e-01 5.73213160e-01
5.14443755e-01 4.26277667e-01 1.01998842e+00 -6.06461763e-01
3.80087584e-01 7.10352600e-01 8.38375509e-01 6.37750566e-01
6.60857260e-01 -1.21003717e-01 -1.63149163e-01 -2.59406030e-01
5.09312570e-01 1.38887271e-01 3.81229669e-01 -7.10574269e-01
-8.91544998e-01 1.16377878e+00 7.83519566e-01 3.82624239e-01
-2.03031853e-01 -2.96679586e-01 4.36170161e-01 5.78217804e-01
5.09766281e-01 5.94961762e-01 -7.72662878e-01 4.30051118e-01
-3.03563118e-01 -1.94140617e-02 1.15740252e+00 1.31439948e+00
8.93259168e-01 -4.78668153e-01 3.60715479e-01 6.80173397e-01
6.70809090e-01 -1.11037180e-01 5.69299400e-01 -5.37109852e-01
2.16509402e-01 1.18734264e+00 -2.45497916e-02 -1.15791893e+00
-3.51438463e-01 -7.89690912e-01 -5.84942341e-01 9.30868555e-03
1.91881925e-01 2.77203768e-01 -7.53075838e-01 1.35010207e+00
4.41697866e-01 -3.17712903e-01 1.11635320e-01 6.23504102e-01
6.43728435e-01 4.89832431e-01 4.33898747e-01 -7.57874176e-02
1.72021842e+00 -1.25198102e+00 -3.86349648e-01 -4.39763069e-02
1.13566029e+00 -7.66846418e-01 1.14024496e+00 4.86459970e-01
-4.44992542e-01 -7.48665273e-01 -1.01906443e+00 5.87637909e-02
-9.23963726e-01 -2.48106077e-01 1.11267102e+00 8.65509450e-01
-7.25035131e-01 5.16332805e-01 -3.85204166e-01 -6.63055837e-01
1.96169596e-02 1.28434658e-01 -2.46823579e-01 -1.74137712e-01
-1.22602069e+00 7.09572196e-01 5.84116518e-01 -4.29457217e-01
-3.38111579e-01 -6.46937609e-01 -9.80583727e-01 9.07967389e-02
2.94553280e-01 -8.07865620e-01 1.56969309e+00 -1.01902568e+00
-1.25200558e+00 7.38146901e-01 -1.16077274e-01 -2.77429163e-01
-1.29361168e-01 3.61808777e-01 -9.08311725e-01 -9.84066129e-02
4.74880397e-01 7.05332458e-01 5.34252346e-01 -1.42600048e+00
-1.36967063e+00 -5.12751043e-01 5.82656205e-01 -7.50174075e-02
-6.54680133e-01 1.34643584e-01 -1.16069332e-01 -6.19162440e-01
5.87839127e-01 -9.55774426e-01 -3.96248311e-01 -1.11523055e-01
-3.27636451e-01 -7.40760505e-01 3.43174428e-01 -2.17593685e-01
1.38810718e+00 -1.90322506e+00 -1.55711040e-01 3.04284364e-01
4.10717428e-01 -4.03878123e-01 -1.17269211e-01 4.66555119e-01
-4.14739363e-02 6.42863452e-01 1.42665729e-01 4.84016091e-02
2.20982045e-01 1.82634443e-01 -3.10849249e-01 6.57034013e-03
-2.78938234e-01 9.36883867e-01 -1.08741128e+00 -4.56395447e-01
-1.94519043e-01 -4.38666679e-02 -6.00118339e-01 -1.99994966e-01
-2.78148621e-01 9.04667154e-02 -5.41202128e-01 7.28793561e-01
6.56759560e-01 -2.51398146e-01 5.69641411e-01 -2.89928406e-01
-9.59110335e-02 7.28956819e-01 -7.75972664e-01 1.52362752e+00
-7.65391946e-01 1.85103938e-01 -3.33233356e-01 -8.94204736e-01
1.18042827e+00 -8.24916214e-02 2.39701360e-01 -5.29449046e-01
7.76668936e-02 2.60599732e-01 2.34556213e-01 8.21414217e-03
5.52820086e-01 -2.63686310e-02 -4.02068138e-01 5.96328437e-01
-6.97954297e-02 1.76018432e-01 3.80884379e-01 2.04985946e-01
5.84329188e-01 1.49744342e-03 4.72952366e-01 -5.65645754e-01
4.81421828e-01 5.88147461e-01 4.30981606e-01 5.42990386e-01
1.06153093e-01 -1.36592224e-01 5.55944026e-01 -6.09900177e-01
-1.00101483e+00 -9.78428245e-01 -2.60769993e-01 1.38404608e+00
2.37828836e-01 -8.22388947e-01 -5.98549902e-01 -9.34864581e-01
2.39735737e-01 7.93954730e-01 -5.92444658e-01 -2.60664552e-01
-1.73537493e-01 -3.15730274e-01 -3.04765143e-02 5.39709270e-01
3.63046139e-01 -6.09606028e-01 -2.17739552e-01 4.42029893e-01
1.65808480e-02 -9.13906038e-01 -5.18522322e-01 4.89788428e-02
-1.15392554e+00 -8.70012462e-01 -1.39791310e-01 -1.34514952e+00
8.64316702e-01 7.33635068e-01 1.22263646e+00 -4.75784279e-02
5.45271263e-02 -8.95032436e-02 -7.57516086e-01 -1.75412074e-01
-7.34122396e-01 6.74263656e-01 2.47641623e-01 -3.60245928e-02
9.84859586e-01 -5.72384238e-01 -4.18783396e-01 7.76060760e-01
-7.01403916e-01 1.92445591e-01 4.29724604e-01 5.87984204e-01
4.73523289e-01 5.65692842e-01 5.77872515e-01 -1.24512029e+00
7.05880165e-01 -7.73285925e-01 -5.98291218e-01 3.25530291e-01
-1.36533952e+00 7.84847289e-02 7.78973639e-01 -6.74576700e-01
-1.03663039e+00 2.14807078e-01 1.43783484e-02 1.18796721e-01
-1.64930031e-01 1.11111093e+00 -6.68356940e-02 2.36126073e-02
6.97501481e-01 -2.72875428e-02 -4.26263392e-01 -8.23875964e-01
9.30856943e-01 1.09515023e+00 4.03435767e-01 -4.13901538e-01
8.94939661e-01 1.24429055e-01 -1.29492894e-01 -2.79399697e-02
-8.52962434e-01 -7.01114595e-01 -1.28066027e+00 1.30261049e-01
4.29939866e-01 -6.00747943e-01 -7.23737478e-01 -7.75940195e-02
-9.31548238e-01 1.03047587e-01 -8.21327269e-02 3.04918796e-01
-1.59499452e-01 1.83221355e-01 -7.73810327e-01 -3.56753290e-01
-2.67191499e-01 -7.11969554e-01 6.16540134e-01 1.97048739e-01
-5.28539240e-01 -1.25850964e+00 -4.83806618e-02 2.25825354e-01
2.31805772e-01 -2.41021961e-01 1.66698205e+00 -8.68523598e-01
-3.34636718e-01 -3.27319294e-01 -2.40985647e-01 -8.25626850e-02
5.73169231e-01 -5.75198233e-02 -4.04425770e-01 -1.91354662e-01
-2.40197554e-01 2.42226914e-01 5.67940712e-01 -1.64431900e-01
8.51092458e-01 -6.72315001e-01 -7.60618806e-01 3.77169549e-01
1.39724958e+00 6.99555874e-01 2.64267892e-01 6.55062616e-01
6.41260743e-01 8.61482143e-01 1.02704000e+00 -6.35713860e-02
6.67089641e-01 8.33417058e-01 1.56299964e-01 -1.84626788e-01
1.09057799e-01 -7.74001658e-01 8.40763971e-02 8.65231276e-01
4.15902257e-01 -2.61811744e-02 -7.75682032e-01 5.16387880e-01
-1.78163373e+00 -3.64635736e-01 -4.88672644e-01 2.02152848e+00
8.06429148e-01 4.12624538e-01 3.29494804e-01 6.11024275e-02
5.90356708e-01 -4.84471887e-01 -5.90081990e-01 -8.51784408e-01
4.36480731e-01 -7.39882737e-02 5.50314903e-01 3.98557901e-01
-1.00014997e+00 1.15161872e+00 6.65740061e+00 3.59662533e-01
-8.74602556e-01 2.51844287e-01 3.47676218e-01 6.66162968e-01
-6.58309340e-01 1.08672298e-01 -9.04499412e-01 2.74261057e-01
8.98155153e-01 -6.88703835e-01 7.14695174e-03 1.27922571e+00
-3.18705261e-01 2.90919542e-01 -1.33304667e+00 6.20374084e-01
-1.93317845e-01 -1.25125062e+00 4.95480806e-01 3.22789282e-01
5.58682263e-01 -6.09991610e-01 1.24784268e-01 4.89493757e-01
9.10993695e-01 -6.54808044e-01 7.36664712e-01 -2.85656601e-01
7.80696988e-01 -6.84699953e-01 6.49784803e-01 1.02511227e-01
-1.64013112e+00 -2.24638537e-01 -4.96781647e-01 -1.79746196e-01
-1.54942907e-02 4.56436515e-01 -9.45150077e-01 8.12766194e-01
5.69181800e-01 8.82447481e-01 -7.29720771e-01 9.77466226e-01
-3.03666264e-01 5.92545152e-01 1.52232334e-01 -2.19582275e-01
2.63001233e-01 -4.93967891e-01 3.39711681e-02 1.12868178e+00
2.56338030e-01 2.06770808e-01 5.43498755e-01 8.13369751e-01
-1.80831090e-01 3.66881520e-01 -5.94263077e-01 -2.69706063e-02
4.99754936e-01 1.44685316e+00 -1.17007911e+00 -5.41636944e-01
-6.15704834e-01 8.16892803e-01 1.23450823e-01 1.61012188e-01
-5.07501006e-01 -6.48633301e-01 9.69676197e-01 1.68096751e-01
4.91066054e-02 -2.11201191e-01 -6.95806861e-01 -1.19044328e+00
-1.16406403e-01 -6.24019384e-01 6.50472939e-01 -6.09014511e-01
-1.57646489e+00 8.69511068e-01 -3.82637745e-03 -1.48595691e+00
-2.65213311e-01 -6.14045978e-01 -2.20198184e-01 5.99362075e-01
-1.26493859e+00 -1.19440901e+00 -6.90033957e-02 2.44839832e-01
6.12703681e-01 -1.03157707e-01 9.99817729e-01 4.42665130e-01
-2.45807022e-01 7.12197900e-01 3.67287137e-02 1.88225225e-01
3.65415841e-01 -1.24614513e+00 1.00878716e+00 6.43372536e-01
3.36277038e-01 9.96933222e-01 6.11388922e-01 -7.72715032e-01
-1.16254401e+00 -1.22057486e+00 1.16813672e+00 -2.98497409e-01
1.17882359e+00 -6.87730491e-01 -9.34119463e-01 1.00596356e+00
5.14747761e-02 -6.19729221e-01 9.93743896e-01 5.59943199e-01
-1.05479503e+00 -1.49878070e-01 -1.18531299e+00 5.99541664e-01
1.38255966e+00 -3.62197131e-01 -7.02142656e-01 5.81985056e-01
1.15784264e+00 9.03834030e-02 -1.14771998e+00 7.79170245e-02
7.40782797e-01 -2.95914739e-01 7.90899932e-01 -1.08957362e+00
4.50309545e-01 -3.85608554e-01 -3.51007581e-02 -1.56683087e+00
-9.99964297e-01 -3.01000029e-01 2.06528574e-01 1.28286207e+00
8.15301001e-01 -9.48064566e-01 8.00173938e-01 3.63947451e-01
-6.55290186e-02 -5.88624179e-01 -3.82812977e-01 -1.05688500e+00
2.82921225e-01 -1.42420530e-01 1.08355045e+00 1.17838669e+00
7.67669082e-01 7.29350269e-01 2.41945118e-01 5.01611270e-04
7.23563313e-01 7.94101775e-01 5.43073952e-01 -1.87428772e+00
-7.72004807e-03 -5.82066298e-01 -4.86691564e-01 -1.19161582e+00
1.44506350e-01 -1.39304149e+00 -2.88014650e-01 -1.78274298e+00
3.37895602e-01 -9.02214170e-01 -2.06498116e-01 6.68249846e-01
2.89671481e-01 2.94597447e-01 1.13385044e-01 4.84999329e-01
-3.22323769e-01 -1.18552350e-01 1.03306961e+00 -4.84554350e-01
-4.20430243e-01 2.45537192e-01 -1.31453109e+00 6.35475516e-01
9.11962986e-01 -8.14361036e-01 -4.82898980e-01 -1.05686873e-01
3.81491184e-01 -2.45514378e-01 -2.02981263e-01 -3.41423094e-01
2.26354659e-01 -2.58717865e-01 1.25792325e-02 -4.76387233e-01
-2.48735636e-01 -8.92147601e-01 3.79719526e-01 7.51996815e-01
-5.36459565e-01 1.41092658e-01 -5.63237816e-02 4.29924160e-01
-2.09811464e-01 -4.01540697e-01 4.18625623e-01 -6.51199445e-02
-8.80607247e-01 1.25179693e-01 -4.92081583e-01 -7.28492081e-01
1.00781631e+00 -3.21374983e-01 -4.51181144e-01 -9.82336029e-02
-8.03807557e-01 2.19821140e-01 5.69606960e-01 1.00138080e+00
1.50868148e-01 -1.43694055e+00 -4.12966728e-01 8.20776075e-02
6.07315719e-01 -6.38859570e-01 -4.00935501e-01 1.89001754e-01
-2.80230433e-01 8.37588847e-01 -4.05080616e-01 -4.64918792e-01
-1.36233151e+00 1.18783152e+00 1.60691887e-01 -3.28263372e-01
-5.04379570e-01 4.40456212e-01 4.13863838e-01 -6.86866701e-01
4.86614555e-02 -6.48466885e-01 -4.45746303e-01 1.77203014e-01
4.15462643e-01 4.67111766e-02 1.63163900e-01 -5.51649868e-01
-3.34214509e-01 6.52627468e-01 -4.99347568e-01 3.45994532e-02
8.28384995e-01 -3.82925749e-01 -3.83724540e-01 3.71583670e-01
1.28200817e+00 -2.24019468e-01 -4.54627395e-01 -5.11529922e-01
5.26698411e-01 -6.08075500e-01 -1.74365491e-01 -9.04043019e-01
-8.66876721e-01 4.83457327e-01 4.73692507e-01 8.13800693e-01
9.87414658e-01 2.92643875e-01 7.46483803e-01 3.30290675e-01
1.03036392e+00 -5.81026554e-01 -2.13727742e-01 2.53774107e-01
4.93134290e-01 -8.76266599e-01 -2.05230266e-01 -1.33903491e+00
-4.45059538e-01 1.21292377e+00 4.04692501e-01 2.83878773e-01
8.27954471e-01 -5.80769032e-02 2.45012924e-01 -1.20124407e-01
-6.40618801e-01 -1.62763059e-01 1.84624091e-01 8.09033751e-01
4.31441605e-01 4.61760253e-01 -5.72052419e-01 8.42944026e-01
-6.50590181e-01 2.85217464e-02 6.58633053e-01 7.35855341e-01
-5.11746407e-01 -1.58000970e+00 1.39561351e-02 4.80682492e-01
-1.92275524e-01 -1.41497329e-01 -4.47738051e-01 5.63745141e-01
1.03099443e-01 1.27782190e+00 2.36861661e-01 -8.04382026e-01
4.02647227e-01 4.93452102e-02 2.09305674e-01 -1.11923897e+00
-6.12191498e-01 -2.00126618e-01 1.00032270e-01 -1.82506457e-01
-1.05279125e-01 -6.05065107e-01 -1.22806180e+00 -5.49011469e-01
-3.37852627e-01 4.87871289e-01 8.35709631e-01 6.04105830e-01
4.86593217e-01 4.00410473e-01 9.47275579e-01 -1.86428145e-01
-4.09945369e-01 -8.06485355e-01 -7.00617313e-01 3.68872732e-01
-1.46449670e-01 -7.72611916e-01 -9.39728171e-02 4.42074120e-01] | [9.88339900970459, 6.276978015899658] |
a6a3aa46-c8a0-44d5-af1a-385887f70b6f | a-review-and-evaluation-of-elastic-distance | 2205.15181 | null | https://arxiv.org/abs/2205.15181v2 | https://arxiv.org/pdf/2205.15181v2.pdf | A Review and Evaluation of Elastic Distance Functions for Time Series Clustering | Time series clustering is the act of grouping time series data without recourse to a label. Algorithms that cluster time series can be classified into two groups: those that employ a time series specific distance measure; and those that derive features from time series. Both approaches usually rely on traditional clustering algorithms such as $k$-means. Our focus is on distance based time series that employ elastic distance measures, i.e. distances that perform some kind of realignment whilst measuring distance. We describe nine commonly used elastic distance measures and compare their performance with k-means and k-medoids clustering. Our findings are surprising. The most popular technique, dynamic time warping (DTW), performs worse than Euclidean distance with k-means, and even when tuned, is no better. Using k-medoids rather than k-means improved the clusterings for all nine distance measures. DTW is not significantly better than Euclidean distance with k-medoids. Generally, distance measures that employ editing in conjunction with warping perform better, and one distance measure, the move-split-merge (MSM) method, is the best performing measure of this study. We also compare to clustering with DTW using barycentre averaging (DBA). We find that DBA does improve DTW k-means, but that the standard DBA is still worse than using MSM. Our conclusion is to recommend MSM with k-medoids as the benchmark algorithm for clustering time series with elastic distance measures. We provide implementations in the aeon toolkit, results and guidance on reproducing results on the associated GitHub repository. | ['Anthony Bagnall', 'Matthew Middlehurst', 'Chris Holder'] | 2022-05-30 | null | null | null | null | ['time-series-clustering'] | ['time-series'] | [-3.82277071e-01 -6.44195020e-01 1.00434877e-01 -2.32255965e-01
-6.87397540e-01 -1.07203650e+00 6.47652924e-01 5.54855406e-01
-6.36337280e-01 2.88426373e-02 4.39862132e-01 -4.60926145e-01
-8.88466179e-01 -7.62199640e-01 2.72437707e-02 -8.97291601e-01
-8.31350267e-01 3.67585272e-01 2.66968697e-01 -2.63961554e-01
4.66598004e-01 5.78870416e-01 -1.39997172e+00 5.23039959e-02
6.05779946e-01 7.56976426e-01 -3.14594984e-01 6.07747376e-01
-3.22087370e-02 3.02580893e-01 -7.07357705e-01 2.06075534e-01
4.86968160e-01 -5.16281843e-01 -6.83629572e-01 -2.91852266e-01
-4.12291676e-01 2.59389222e-01 -4.75174822e-02 4.47611570e-01
6.00964248e-01 7.39418507e-01 6.62524402e-01 -1.61330855e+00
-3.42292160e-01 6.81110799e-01 -6.37365937e-01 5.86657345e-01
4.65729982e-01 -3.69985998e-02 6.57916903e-01 -4.56177503e-01
6.21542633e-01 1.02740872e+00 1.30835342e+00 -2.54693627e-02
-1.60423970e+00 -6.27554893e-01 -1.79532707e-01 2.51682997e-01
-1.67489529e+00 -1.62655890e-01 6.98229074e-01 -5.64601302e-01
1.49781358e+00 7.19476104e-01 6.24049723e-01 3.65446061e-01
1.71775445e-01 1.65434435e-01 1.19103587e+00 -3.41775477e-01
4.14918184e-01 -5.42048395e-01 2.19080210e-01 -1.67718962e-01
-1.10348135e-01 4.33903784e-02 1.32185161e-01 -6.51245952e-01
1.56892180e-01 3.36212099e-01 -9.18073505e-02 -1.40593514e-01
-1.66497052e+00 7.51340330e-01 1.52744979e-01 1.03892624e+00
-3.40560555e-01 -4.47394103e-02 7.88597703e-01 7.54721463e-01
6.34489477e-01 6.42759204e-01 -5.87041140e-01 -6.33494318e-01
-1.47860920e+00 3.74697626e-01 6.96152925e-01 4.21179414e-01
4.10782039e-01 3.28431912e-02 1.37596771e-01 7.87676156e-01
-1.41007513e-01 2.56866992e-01 1.01618099e+00 -1.13963985e+00
-6.37605488e-02 5.53905845e-01 -2.37701446e-01 -1.32484508e+00
-7.14812756e-01 1.55586988e-01 -6.67551756e-01 8.42596665e-02
5.03302753e-01 -2.08148882e-01 -7.27411985e-01 1.56573725e+00
2.65722871e-01 3.58366966e-01 -1.07997514e-01 4.37436789e-01
3.51224273e-01 8.77685189e-01 -8.22100788e-02 -8.52019906e-01
8.28127623e-01 -3.98303807e-01 -6.47868216e-01 6.18642211e-01
1.03035939e+00 -1.14209270e+00 8.21055830e-01 3.89311552e-01
-9.34781551e-01 -4.62862462e-01 -8.19882154e-01 5.23287654e-01
-8.26475382e-01 -8.21804285e-01 2.83040762e-01 5.86932778e-01
-1.47730303e+00 1.05362999e+00 -1.19421077e+00 -8.34674954e-01
-3.27402145e-01 3.12944949e-01 -2.59572148e-01 4.07117128e-01
-9.21264529e-01 8.33455026e-01 5.37250102e-01 -5.77897251e-01
-9.66397598e-02 -6.98887527e-01 -5.07448792e-01 -2.93563128e-01
-6.41797855e-02 1.03866152e-01 8.21428180e-01 -6.23759687e-01
-1.11972308e+00 6.98104501e-01 3.60976309e-02 -4.60443467e-01
2.53287494e-01 4.60455805e-01 -9.72645342e-01 1.38567418e-01
1.23792931e-01 3.58571261e-01 5.80422103e-01 -1.00159597e+00
-4.49604422e-01 -2.86147028e-01 -4.51667607e-01 -2.35933721e-01
-3.07586581e-01 4.28327173e-01 1.43830493e-01 -9.74479556e-01
2.92541325e-01 -1.18615508e+00 -2.55137593e-01 -6.13069117e-01
4.86681759e-02 -6.51135862e-01 1.21690822e+00 -6.63806796e-01
1.94336355e+00 -2.21030951e+00 -4.39753644e-02 7.48161435e-01
8.94383937e-02 1.50233820e-01 9.80562568e-02 9.74679291e-01
-8.19056213e-01 2.65946388e-01 -3.97519439e-01 2.62088012e-02
2.05197647e-01 2.02228189e-01 -3.33113104e-01 7.59767115e-01
-2.78427124e-01 5.00061929e-01 -9.49029267e-01 -4.65874195e-01
4.52975184e-01 3.89794588e-01 -4.49214876e-01 -3.31529170e-01
4.77180660e-01 1.01060987e-01 2.85975337e-01 2.49455437e-01
4.43919420e-01 5.42927906e-02 -2.57504862e-02 -7.33904392e-02
-7.06088781e-01 -1.09823875e-01 -1.33829474e+00 1.42927575e+00
2.65909210e-02 7.32850552e-01 -4.08900410e-01 -1.18460619e+00
1.04801452e+00 5.13154745e-01 1.36174846e+00 -3.27327788e-01
-3.30561362e-02 2.80413359e-01 4.64370549e-01 -4.03900892e-01
4.40255821e-01 -1.74432516e-01 -2.41806790e-01 7.92297721e-01
-1.79902643e-01 -5.16118050e-01 3.57153445e-01 1.51934430e-01
1.37467659e+00 -5.51261716e-02 2.43535265e-01 -7.34139264e-01
1.30480736e-01 2.51688033e-01 5.14302611e-01 1.64807379e-01
-3.39570284e-01 5.18801033e-01 2.22885996e-01 -3.59851182e-01
-1.13947177e+00 -1.14310169e+00 -3.47088397e-01 9.85593140e-01
-3.16244572e-01 -8.36193860e-01 -6.49418771e-01 -2.12802932e-01
1.02054901e-01 8.14146459e-01 -6.84832454e-01 -8.17646533e-02
-4.89713699e-01 -9.04676080e-01 6.87363565e-01 2.81258404e-01
1.17945755e-02 -9.76523876e-01 -7.35394597e-01 3.69681090e-01
-2.47656733e-01 -2.53735691e-01 -7.22516298e-01 3.08977991e-01
-1.12202573e+00 -7.52652228e-01 -7.72160351e-01 -4.90366429e-01
3.87193412e-01 3.89574140e-01 8.67815733e-01 -1.12214550e-01
-2.79701978e-01 6.89909935e-01 -1.01458669e+00 -4.83432382e-01
-1.97790742e-01 -1.52328968e-01 4.25414771e-01 -1.93648979e-01
9.05231595e-01 -9.38611269e-01 -7.00743020e-01 6.19000137e-01
-1.01304436e+00 -7.52663910e-01 -2.33518973e-01 2.89824814e-01
5.00008583e-01 7.12890744e-01 4.49875921e-01 -1.57105789e-01
9.73462343e-01 -7.08596170e-01 -8.97691250e-02 -1.31023958e-01
-8.73815060e-01 -3.35532010e-01 6.49060071e-01 -6.85065210e-01
-2.25795493e-01 -2.76927084e-01 1.33699194e-01 -6.49211586e-01
-9.51368585e-02 6.40908122e-01 6.44963145e-01 5.17119877e-02
8.80774498e-01 1.27891496e-01 5.26867136e-02 -3.49641740e-01
1.39765993e-01 7.69327044e-01 2.87405968e-01 -4.99484301e-01
6.94304347e-01 8.61663699e-01 -3.76057804e-01 -8.89159322e-01
1.83133289e-01 -9.89915371e-01 -7.66336083e-01 -3.99642110e-01
9.49804902e-01 -3.50732535e-01 -7.21336722e-01 3.40346038e-01
-4.59263980e-01 -4.57728148e-01 -5.63444912e-01 7.64848053e-01
-6.33588791e-01 4.91561055e-01 -4.05889332e-01 -7.43090808e-01
-5.45275547e-02 -7.37365961e-01 7.21128702e-01 -2.05577478e-01
-1.02692950e+00 -1.34581459e+00 5.12987614e-01 -2.48107299e-01
5.58441162e-01 1.07212830e+00 6.98196709e-01 -9.99355733e-01
5.63558817e-01 -3.32778066e-01 3.37600648e-01 2.93082129e-02
5.13971269e-01 6.01398170e-01 -5.30770123e-01 -6.13629162e-01
2.31406406e-01 3.45998496e-01 3.78798991e-01 7.55333841e-01
1.04423881e+00 -3.95347178e-02 -4.32350785e-01 4.61728603e-01
1.43514276e+00 1.00820100e+00 5.88059247e-01 6.23241544e-01
3.93749744e-01 8.78122687e-01 6.24572873e-01 7.18678594e-01
5.74851990e-01 5.71762860e-01 -1.18397452e-01 -5.72964251e-02
5.05186796e-01 4.32968885e-01 3.82305652e-01 1.58501446e+00
-5.98085523e-01 2.65116654e-02 -1.47813618e+00 7.68852651e-01
-1.85387468e+00 -1.54499972e+00 -4.94688898e-01 2.13662148e+00
6.07044458e-01 1.08583048e-01 8.14706981e-01 9.14209545e-01
6.52893066e-01 1.50604665e-01 -2.38653347e-01 -6.96494579e-01
-1.31225465e-02 3.00829440e-01 4.78988558e-01 3.72856647e-01
-1.03382099e+00 3.87066513e-01 6.72862720e+00 6.74261689e-01
-1.10177362e+00 4.60238196e-02 1.52473226e-01 -2.44396254e-01
-1.07442960e-01 3.85354124e-02 9.99223217e-02 8.86936545e-01
1.44303191e+00 -9.14146125e-01 4.37700212e-01 4.23978448e-01
7.34559119e-01 -2.93275714e-02 -9.34694052e-01 1.13697362e+00
-1.85398161e-01 -8.47615540e-01 -5.00814259e-01 1.94153920e-01
6.51831448e-01 1.28391728e-01 -6.27173409e-02 2.43280202e-01
6.13081217e-01 -7.45419323e-01 4.57946599e-01 5.53576887e-01
4.62487251e-01 -9.91372406e-01 4.47663486e-01 4.15260270e-02
-1.57290030e+00 2.31720563e-02 2.82441755e-03 -1.54857606e-01
2.60096371e-01 5.89065433e-01 -2.80757844e-01 7.26951897e-01
1.26786530e+00 1.00254726e+00 -3.63453239e-01 1.13113880e+00
7.91195452e-01 7.07574606e-01 -5.67963421e-01 4.00665462e-01
4.97898191e-01 -4.51132655e-01 5.33826590e-01 1.45012045e+00
7.89524376e-01 2.67349899e-01 1.98582381e-01 1.48640096e-01
9.10769045e-01 1.87227443e-01 -6.63931727e-01 -1.43493414e-01
7.85483181e-01 9.76777613e-01 -1.40904939e+00 -4.82012630e-01
-3.55918854e-01 6.86447918e-01 -5.34332454e-01 1.76399007e-01
-9.26700652e-01 -9.02406573e-01 7.94827878e-01 2.35821694e-01
1.67289928e-01 -6.47547662e-01 -5.82264848e-02 -7.00074017e-01
-2.75638700e-01 -9.34687376e-01 9.40523684e-01 -7.77521610e-01
-1.44354951e+00 5.97073376e-01 5.39294183e-01 -1.87349629e+00
-5.13988853e-01 -8.89789611e-02 -5.42579114e-01 5.94564915e-01
-6.40865624e-01 -2.76562780e-01 -1.31335586e-01 1.11128080e+00
3.38535160e-01 2.64709532e-01 8.72457802e-01 4.17434752e-01
-2.04305619e-01 2.51781791e-01 5.99246085e-01 7.99391195e-02
1.06312788e+00 -1.36344528e+00 3.78026783e-01 7.26423383e-01
-1.36975437e-01 7.53585339e-01 9.58705008e-01 -6.60690784e-01
-9.87012863e-01 -9.20699418e-01 8.84159446e-01 -5.15321314e-01
1.07790184e+00 1.25708326e-03 -1.10802269e+00 6.66839719e-01
2.52487898e-01 -3.58338803e-01 1.28208959e+00 -2.78790295e-02
-1.87968180e-01 -1.14322165e-02 -1.21303487e+00 4.25952077e-01
9.95213151e-01 -5.59499681e-01 -9.85973597e-01 3.69047552e-01
6.11509144e-01 1.55116469e-01 -1.84039783e+00 2.65355676e-01
4.31759775e-01 -1.09400558e+00 1.16117704e+00 -3.02638739e-01
-4.50447649e-02 -7.11405337e-01 -2.96959102e-01 -1.67260838e+00
-6.70058072e-01 -7.72173584e-01 4.62343276e-01 1.34029889e+00
1.74639508e-01 -9.00954843e-01 3.22006166e-01 5.04044712e-01
-1.56433761e-01 -2.36937866e-01 -8.94428849e-01 -1.39252782e+00
2.85158873e-01 -6.97098494e-01 7.45391726e-01 1.99337995e+00
6.90030992e-01 -4.65036929e-01 1.29825041e-01 -3.50677013e-01
6.30311310e-01 5.35543300e-02 4.89954054e-01 -1.39989686e+00
-2.02282090e-02 -8.92432690e-01 -7.30475068e-01 2.94784531e-02
-4.33863550e-02 -9.99800503e-01 -2.75045812e-01 -1.41707313e+00
-4.97825712e-01 -4.83302921e-01 -3.48006874e-01 3.62886071e-01
1.13257527e-01 2.80671299e-01 1.50139078e-01 7.53733695e-01
-4.03553009e-01 -1.96322035e-02 6.58062220e-01 2.12856695e-01
-8.02773654e-01 -2.76540130e-01 -2.19814137e-01 5.77875793e-01
9.09062326e-01 -4.91514146e-01 -5.03757596e-01 -4.21552593e-03
-2.09466904e-01 6.48210049e-02 -1.04249224e-01 -1.06841791e+00
3.97660196e-01 -1.96734130e-01 1.82121336e-01 -6.14767253e-01
-1.28817074e-02 -1.02565420e+00 8.53969038e-01 6.95934415e-01
-1.11248016e-01 9.37522590e-01 3.96008521e-01 1.80708259e-01
-3.03037256e-01 2.75370508e-01 7.03840733e-01 7.33860210e-03
-7.07461536e-01 5.74964769e-02 -8.14393222e-01 -3.41803655e-02
1.32095242e+00 -8.58415961e-01 -3.02621964e-02 -4.67061043e-01
-1.12928128e+00 2.78786957e-01 6.24607146e-01 4.18532282e-01
2.04071075e-01 -1.69585752e+00 -6.35681987e-01 -7.86650404e-02
1.61114767e-01 -5.24160802e-01 1.13995709e-01 1.33113945e+00
-5.38578272e-01 1.85284644e-01 -3.36369067e-01 -6.98270559e-01
-1.45066607e+00 9.62949038e-01 1.82273045e-01 2.10094266e-02
-6.55282497e-01 2.90277660e-01 -4.87356722e-01 -5.14357448e-01
4.68357876e-02 -5.83555162e-01 -1.17169887e-01 7.46544480e-01
5.41417718e-01 9.89296615e-01 1.48779020e-01 -8.57301116e-01
-7.14602351e-01 9.69710767e-01 3.66466045e-01 -3.24041486e-01
1.70894670e+00 -1.98245585e-01 -4.17093903e-01 1.00695086e+00
1.39372647e+00 -2.34687269e-01 -6.96315229e-01 2.32523791e-02
5.11277854e-01 -4.91391510e-01 -1.94285408e-01 -2.63836533e-01
-7.63343275e-01 3.92232686e-01 6.56995595e-01 1.27156675e+00
1.51670992e+00 -1.64333984e-01 5.43623388e-01 3.48852463e-02
3.66823584e-01 -1.43793845e+00 -2.57060856e-01 3.35487515e-01
6.92436337e-01 -7.86572218e-01 -4.94745225e-02 1.58226714e-01
-4.49104369e-01 1.08474743e+00 3.11286002e-02 -5.04267454e-01
1.27598572e+00 5.38430154e-01 1.64674163e-01 -2.14579999e-01
-5.57503998e-01 -3.27331811e-01 3.18475179e-02 6.73671007e-01
7.33804226e-01 3.05421650e-01 -6.10158026e-01 9.85317901e-02
-6.84801161e-01 -2.18549535e-01 3.04078460e-01 1.04431772e+00
-3.69728029e-01 -9.78588343e-01 -7.53961742e-01 6.55980349e-01
-3.40451300e-01 1.91339403e-01 -4.45816696e-01 9.55485940e-01
1.85912907e-01 1.39353621e+00 6.18458807e-01 -9.26419675e-01
5.48317432e-01 3.25610936e-01 8.38923901e-02 -1.59869775e-01
-1.08115304e+00 3.28538030e-01 -2.44274989e-01 -6.05138779e-01
-9.68302310e-01 -1.19310558e+00 -1.34805453e+00 -1.17595291e+00
-1.46738186e-01 4.47094947e-01 4.99169141e-01 5.98858237e-01
2.57188976e-01 3.73572886e-01 8.39341462e-01 -9.54356551e-01
1.68754123e-02 -7.87485361e-01 -7.97732532e-01 6.79267526e-01
1.13109872e-01 -4.54404205e-01 -6.96393251e-01 2.81070620e-01] | [7.247849941253662, 3.3556313514709473] |
10387f47-3cad-4026-9494-c54b31f8f8e2 | multi-stage-clarification-in-conversational | 2110.15235 | null | https://arxiv.org/abs/2110.15235v1 | https://arxiv.org/pdf/2110.15235v1.pdf | Multi-stage Clarification in Conversational AI: The case of Question-Answering Dialogue Systems | Clarification resolution plays an important role in various information retrieval tasks such as interactive question answering and conversational search. In such context, the user often formulates their information needs as short and ambiguous queries, some popular search interfaces then prompt the user to confirm her intent (e.g. "Did you mean ... ?") or to rephrase if needed. When it comes to dialogue systems, having fluid user-bot exchanges is key to good user experience. In the absence of such clarification mechanism, one of the following responses is given to the user: 1) A direct answer, which can potentially be non-relevant if the intent was not clear, 2) a generic fallback message informing the user that the retrieval tool is incapable of handling the query. Both scenarios might raise frustration and degrade the user experience. To this end, we propose a multi-stage clarification mechanism for prompting clarification and query selection in the context of a question answering dialogue system. We show that our proposed mechanism improves the overall user experience and outperforms competitive baselines with two datasets, namely the public in-scope out-of-scope dataset and a commercial dataset based on real user logs. | ['Eric Charton', 'Marc Queudot', 'Louis Marceau', 'Nada Naji', 'Hadrien Lautraite'] | 2021-10-28 | null | null | null | null | ['conversational-search'] | ['natural-language-processing'] | [ 2.79295206e-01 1.72910362e-01 5.41388579e-02 -3.88411790e-01
-9.24768686e-01 -9.79056299e-01 6.68422401e-01 5.65478086e-01
-7.33679235e-01 6.28922343e-01 3.02243441e-01 -8.22464347e-01
-1.51305303e-01 -3.72120231e-01 1.24540709e-01 -2.23081559e-01
6.02898598e-01 5.61870694e-01 4.33129221e-01 -7.73325264e-01
5.32478392e-01 6.19264692e-02 -1.18303514e+00 3.41929466e-01
1.06855667e+00 8.56659949e-01 6.25287473e-01 8.68067384e-01
-5.15047848e-01 9.04328763e-01 -8.14027965e-01 -2.86619544e-01
-2.91372359e-01 -2.89051592e-01 -1.48314893e+00 -4.17914428e-02
4.65933159e-02 -5.54164767e-01 3.15230973e-02 9.14189875e-01
5.21886885e-01 3.20806026e-01 2.50838637e-01 -1.01351404e+00
-1.63359672e-01 4.96173948e-01 -1.32342145e-01 4.25391793e-01
9.46077287e-01 2.96336472e-01 1.21250629e+00 -6.66318595e-01
3.11471969e-01 1.10384405e+00 -6.06979467e-02 6.74816728e-01
-1.08527112e+00 -2.70980626e-01 1.60468146e-01 -8.51889956e-04
-8.73561442e-01 -7.21430957e-01 7.29774892e-01 -2.46173128e-01
7.12782621e-01 9.32359397e-01 -2.96718087e-02 1.03618932e+00
-2.07678765e-01 8.41074705e-01 7.48466790e-01 -3.18422198e-01
1.99352205e-01 4.68460381e-01 5.82626045e-01 6.69215769e-02
-4.18780386e-01 -4.66280520e-01 -4.67404097e-01 -7.31380403e-01
2.57781237e-01 -1.04647934e-01 -5.21284819e-01 2.73286521e-01
-1.01151633e+00 7.48557568e-01 2.37114981e-01 3.53298962e-01
-6.12920642e-01 -5.13152540e-01 4.89333063e-01 6.85294390e-01
8.57309178e-02 9.62692380e-01 -4.12360966e-01 -6.07099831e-01
-3.61376077e-01 4.55984265e-01 1.33198249e+00 6.27984047e-01
5.65885961e-01 -7.90243030e-01 -7.38424540e-01 1.11974669e+00
-2.42011994e-03 2.45997146e-01 2.91581690e-01 -9.81522620e-01
5.32917738e-01 8.11579883e-01 9.43636119e-01 -9.10550237e-01
-2.42885038e-01 -3.89173366e-02 -5.26768863e-01 -4.16217744e-01
5.81929147e-01 -3.41291785e-01 -9.19888169e-02 1.50430095e+00
3.89099061e-01 -4.60488409e-01 4.43034843e-02 1.15456557e+00
1.08217716e+00 5.81458807e-01 1.49974450e-02 -4.20638472e-01
1.87355745e+00 -7.69705355e-01 -1.08784521e+00 -2.91909635e-01
5.34972906e-01 -1.16575885e+00 1.51128507e+00 1.09821685e-01
-8.53385806e-01 -3.68595093e-01 -5.49386501e-01 -1.63429394e-01
-1.91876269e-03 7.16977492e-02 1.72228217e-01 1.93661630e-01
-7.51234770e-01 7.92876557e-02 -3.10660034e-01 -5.23508847e-01
-4.72269744e-01 1.25329688e-01 3.14092226e-02 8.19051787e-02
-1.55763888e+00 5.32972515e-01 -1.86307251e-01 3.18876326e-01
-2.32672364e-01 -2.86607623e-01 -4.17829901e-01 3.10185015e-01
8.05549681e-01 -5.20741165e-01 1.95213974e+00 -5.88515639e-01
-1.60764050e+00 5.89344561e-01 -3.19835812e-01 -7.40332454e-02
5.53209782e-01 -5.04535973e-01 -1.70881152e-01 1.62039459e-01
1.36999235e-01 2.44872302e-01 8.57248306e-01 -8.91492665e-01
-7.18790591e-01 -2.91290522e-01 7.16316700e-01 7.09744155e-01
-4.59718592e-02 2.57679015e-01 -4.65381533e-01 -4.02903482e-02
-3.78585607e-02 -9.21496570e-01 -3.25276665e-02 -3.84367824e-01
-6.78204000e-01 -7.75334120e-01 9.33131516e-01 -6.93178236e-01
1.60460114e+00 -2.01994157e+00 -6.48585781e-02 -5.88635653e-02
3.13896984e-01 4.94703531e-01 3.85273178e-03 9.41092849e-01
3.12598258e-01 3.14413548e-01 1.23064116e-01 -1.58557091e-02
-7.80379623e-02 -1.04824811e-01 -4.25052166e-01 -1.64683461e-01
-5.69689535e-02 6.50030613e-01 -1.04609120e+00 -3.61954361e-01
-9.39072222e-02 -3.36580835e-02 -3.47761035e-01 1.08488059e+00
-4.45719481e-01 6.31755471e-01 -8.04356754e-01 3.70264351e-01
3.44156772e-01 -4.36316997e-01 -1.34372443e-01 -1.22463137e-01
-1.01450771e-01 8.90691102e-01 -8.46742034e-01 1.28606701e+00
-6.89638197e-01 3.76507878e-01 5.36250532e-01 -5.06731033e-01
5.17085195e-01 4.66119021e-01 -6.26947880e-02 -7.56985188e-01
-1.26266122e-01 2.17705593e-02 1.11508556e-01 -7.59427428e-01
6.93395138e-01 5.34906268e-01 -1.18875675e-01 9.68048811e-01
-5.91902673e-01 -4.28094417e-02 1.26490936e-01 6.90500379e-01
1.29053199e+00 -4.13232505e-01 2.88645148e-01 -1.01621740e-01
8.66118133e-01 -1.73283175e-01 -1.07631236e-02 1.24163198e+00
-2.12846965e-01 1.79618284e-01 7.89193332e-01 3.78607251e-02
-4.14566934e-01 -4.49637324e-01 3.46441627e-01 1.69080925e+00
3.70523036e-01 -6.69552088e-01 -6.95089519e-01 -8.30269814e-01
-2.21063957e-01 8.12109232e-01 -1.68717965e-01 -8.84656385e-02
-4.76787090e-01 -3.40096876e-02 2.54908949e-01 -2.10862800e-01
7.04256654e-01 -1.29479921e+00 -9.79059815e-01 3.31746608e-01
-9.83465672e-01 -1.03083682e+00 -1.00583971e+00 -2.02363431e-01
-5.23248196e-01 -1.09701931e+00 -4.94689435e-01 -5.27476788e-01
3.92562926e-01 7.12418735e-01 1.15572298e+00 8.66493583e-01
1.58035651e-01 5.48334122e-01 -5.71415365e-01 1.96220413e-01
-4.56883430e-01 3.29360306e-01 -3.91548514e-01 1.04811847e-01
2.87026763e-01 -1.86521322e-01 -1.19627023e+00 8.17351937e-01
-8.23479712e-01 2.42912889e-01 1.30340874e-01 8.33155394e-01
-1.32952586e-01 -3.82535517e-01 8.03818643e-01 -1.04261386e+00
1.73864472e+00 -5.76771736e-01 -2.79508740e-01 3.75766218e-01
-4.90169674e-01 9.75577310e-02 5.46063721e-01 -5.40757596e-01
-1.13442707e+00 -3.00198793e-01 -4.83706415e-01 2.78928369e-01
-2.45428607e-01 6.33488894e-01 5.97890019e-02 2.70123422e-01
8.11770797e-01 2.76147157e-01 -4.20312360e-02 -6.30078197e-01
1.30205631e-01 1.22429419e+00 3.03113431e-01 -6.64931834e-01
4.92607236e-01 -1.95217967e-01 -8.97774398e-01 -9.50600505e-01
-6.35836124e-01 -1.14807296e+00 -5.17013296e-02 -1.92046151e-01
5.53509355e-01 -4.06569719e-01 -1.26145303e+00 1.66048840e-01
-1.47931135e+00 -1.25235811e-01 4.02796060e-01 -1.56405747e-01
-1.57095581e-01 4.72192675e-01 -5.28392017e-01 -1.23236489e+00
-6.12879574e-01 -1.26858509e+00 9.56729531e-01 5.38269222e-01
-7.66612470e-01 -6.03340209e-01 -2.82143086e-01 7.66155481e-01
8.13371062e-01 -4.56846386e-01 1.12794805e+00 -1.17384112e+00
-2.90167600e-01 -2.03591183e-01 -1.48249120e-01 -1.50083527e-01
4.93816286e-01 -3.17409486e-01 -7.44332254e-01 -1.59374252e-01
1.40804008e-01 -4.73057449e-01 1.52864560e-01 -3.77419829e-01
9.53430116e-01 -8.97861362e-01 -2.06365213e-01 -2.59375870e-01
6.50507748e-01 4.75507587e-01 3.64740074e-01 7.20753223e-02
1.36018410e-01 8.98141861e-01 6.89669311e-01 4.05878097e-01
5.13932407e-01 8.95275712e-01 2.25900203e-01 -3.06950957e-02
3.31286967e-01 -3.42655092e-01 1.72922704e-02 3.90249759e-01
5.28438270e-01 -4.45673138e-01 -7.05036581e-01 2.87477523e-01
-1.99735641e+00 -7.55433619e-01 -1.22382276e-01 2.21566534e+00
1.15529454e+00 1.22346051e-01 1.81302249e-01 -7.34493583e-02
6.27274871e-01 2.77130544e-01 -8.34938884e-01 -3.06856543e-01
4.16979909e-01 -2.64859587e-01 -3.30916911e-01 9.30403709e-01
-5.88738024e-01 8.73277605e-01 4.34380007e+00 3.82311195e-01
-1.02831662e+00 -7.29062334e-02 8.83077741e-01 3.93746346e-01
-3.58645290e-01 1.98480427e-01 -6.58622086e-01 5.50181329e-01
6.32727683e-01 -2.71098197e-01 5.32266736e-01 4.91256028e-01
4.29112941e-01 -6.14048183e-01 -1.26277375e+00 9.21101272e-01
-2.68141896e-01 -8.62233579e-01 -1.54460967e-01 -3.35412383e-01
-3.42355251e-01 -4.31346476e-01 -1.64257213e-01 5.58252752e-01
-6.23683631e-02 -7.09175646e-01 9.18565467e-02 2.99545318e-01
5.73603332e-01 -3.05349946e-01 5.09176314e-01 1.21807468e+00
-7.78095782e-01 1.29379675e-01 1.70193002e-01 -1.86733007e-01
2.46761739e-01 -2.51510483e-03 -1.33755481e+00 7.66431540e-02
4.19086158e-01 -2.53844202e-01 -4.66906071e-01 7.18760967e-01
-2.72273839e-01 6.43956840e-01 -5.13421535e-01 -5.59636950e-01
9.47284549e-02 -2.46013049e-02 7.53714025e-01 9.42374229e-01
-7.08592311e-02 7.23309815e-01 1.09187618e-01 6.44387782e-01
-1.55626044e-01 4.49770004e-01 -3.05029750e-01 -2.45356277e-01
7.91637003e-01 1.42092192e+00 -3.65564853e-01 -2.78507322e-01
-2.20406190e-01 1.26807606e+00 2.98780769e-01 6.53211594e-01
-3.05044144e-01 -6.28159165e-01 4.06782508e-01 3.46702665e-01
-2.57081717e-01 2.99355406e-02 1.81252792e-01 -1.15459883e+00
2.52039313e-01 -1.29642510e+00 5.21133780e-01 -7.43185878e-01
-8.95918787e-01 8.20223749e-01 -2.81000376e-01 -5.95224380e-01
-6.97445095e-01 1.52230173e-01 -6.85002506e-01 1.24638724e+00
-1.35637140e+00 -2.54219204e-01 -4.22898322e-01 4.23512787e-01
1.02847171e+00 3.60770375e-01 8.85596573e-01 2.97077835e-01
-3.97107184e-01 4.98126447e-01 -4.95151311e-01 -2.13708053e-03
1.02011883e+00 -1.09645605e+00 2.39007980e-01 4.55711126e-01
-1.28448829e-01 1.05582261e+00 1.04680347e+00 -4.63666618e-01
-1.59351695e+00 -3.34671646e-01 1.36132610e+00 -3.85507554e-01
4.13516968e-01 -4.47705775e-01 -1.46108866e+00 1.43615767e-01
3.55872035e-01 -5.80872178e-01 7.15963960e-01 2.49868497e-01
3.15050930e-02 1.68375194e-01 -1.00352001e+00 9.15617228e-01
4.62928414e-01 -7.76527107e-01 -7.74216950e-01 6.12842679e-01
9.06600058e-01 -4.67759103e-01 -3.40457201e-01 1.99400872e-01
3.75853390e-01 -9.21898425e-01 6.93014503e-01 -7.55982876e-01
1.17615692e-01 -1.02807596e-01 5.53331375e-02 -1.00811327e+00
9.94407088e-02 -1.37969542e+00 -8.26429427e-02 1.20628786e+00
4.95571405e-01 -4.84187067e-01 3.57071489e-01 1.51474643e+00
4.31664288e-01 -7.54443765e-01 -6.85683370e-01 1.41968861e-01
-5.70566893e-01 -4.03945893e-03 5.67957103e-01 6.33400142e-01
5.47126114e-01 1.10588777e+00 -3.92981619e-01 4.43801172e-02
-9.46617350e-02 1.98974058e-01 9.39088047e-01 -1.10047591e+00
-5.44265099e-02 -4.17764187e-01 6.20960891e-01 -2.04889250e+00
-1.65891781e-01 -3.81471992e-01 3.00994098e-01 -1.39697063e+00
9.32872295e-02 -4.40468758e-01 6.53023422e-02 3.57722074e-01
-6.69711530e-01 -6.69703841e-01 3.29133831e-02 5.27879536e-01
-8.43749642e-01 4.33030039e-01 1.03808725e+00 -5.26884757e-02
-6.76435351e-01 7.01467395e-01 -9.95256841e-01 4.63152617e-01
5.92261434e-01 -1.29757792e-01 -5.57694018e-01 -1.47507459e-01
5.22193968e-01 1.09775162e+00 -4.39444790e-03 -1.74284294e-01
6.92706883e-01 -3.42245668e-01 -5.06944716e-01 -4.84769762e-01
2.22729906e-01 -8.06764543e-01 -3.77343595e-01 3.62732798e-01
-9.47631240e-01 2.86892086e-01 -8.20261762e-02 4.52790231e-01
-1.75741136e-01 -4.45382059e-01 3.40962499e-01 -5.70376031e-02
-3.57262284e-01 1.48874044e-01 -6.06874108e-01 2.08456680e-01
3.63789618e-01 8.72767568e-02 -4.97223139e-01 -1.16013265e+00
-4.34843212e-01 8.58861387e-01 1.23674951e-01 7.28400171e-01
6.24423504e-01 -8.05179119e-01 -6.55159056e-01 -1.63629949e-01
3.64169657e-01 -5.84880486e-02 6.97561204e-02 7.24069834e-01
-9.88142639e-02 6.51538610e-01 4.87072915e-01 -6.20076895e-01
-1.46523404e+00 9.37264785e-02 3.22194517e-01 -1.69911087e-01
-3.18250299e-01 8.09854925e-01 2.70080686e-01 -3.62151176e-01
7.07870364e-01 -1.93461940e-01 -6.64824724e-01 1.43577516e-01
8.40444982e-01 -3.47418152e-02 2.21369252e-01 -2.75496513e-01
-2.55171627e-01 -2.59615123e-01 -4.59427625e-01 -3.99961740e-01
6.01202965e-01 -7.60558844e-01 -1.59712642e-01 4.88926113e-01
1.03269899e+00 1.49929211e-01 -8.26126456e-01 -6.62727475e-01
1.57792225e-01 -5.01762748e-01 -3.15628290e-01 -1.28179657e+00
-2.80279607e-01 6.69745266e-01 2.42203459e-01 8.79795551e-01
1.04858649e+00 8.49762857e-02 9.49397624e-01 1.04987955e+00
2.12715834e-01 -9.14562702e-01 3.34273010e-01 6.83640122e-01
1.18522906e+00 -1.59335446e+00 -4.30965513e-01 -2.22526953e-01
-8.68980229e-01 1.02271247e+00 9.23274577e-01 6.35761321e-01
2.98884571e-01 -2.93609917e-01 3.95030946e-01 -3.67457747e-01
-1.28832567e+00 6.91821277e-02 2.55193889e-01 -8.97737965e-02
7.14386106e-01 -1.47462443e-01 -7.00484335e-01 5.71162164e-01
-1.51892439e-01 -3.42743695e-01 3.63632858e-01 8.50534320e-01
-5.48297346e-01 -8.43885601e-01 -1.66682839e-01 5.07530868e-01
-5.30820549e-01 -2.21434817e-01 -8.23470533e-01 1.50730029e-01
-8.60998571e-01 1.64880228e+00 -2.47112483e-01 -1.68337494e-01
5.96902609e-01 2.21486673e-01 -3.11942220e-01 -6.74655676e-01
-1.12432039e+00 -2.23694788e-03 5.37456334e-01 -5.34819961e-01
2.53871251e-02 -3.42117220e-01 -1.10907829e+00 2.06824746e-02
-6.76226616e-01 9.38505709e-01 5.36172271e-01 1.01739454e+00
5.10804772e-01 -2.28703544e-02 8.19139540e-01 -1.96717843e-01
-9.44210947e-01 -1.25172853e+00 1.81511551e-01 4.78993922e-01
7.76974559e-01 -2.62758464e-01 -4.86515790e-01 -2.98650593e-01] | [12.244190216064453, 7.857144832611084] |
2c9e10bd-e115-4f6e-b3e8-ab3c59c50754 | gittables-a-large-scale-corpus-of-relational | 2106.07258 | null | https://arxiv.org/abs/2106.07258v5 | https://arxiv.org/pdf/2106.07258v5.pdf | GitTables: A Large-Scale Corpus of Relational Tables | The success of deep learning has sparked interest in improving relational table tasks, like data preparation and search, with table representation models trained on large table corpora. Existing table corpora primarily contain tables extracted from HTML pages, limiting the capability to represent offline database tables. To train and evaluate high-capacity models for applications beyond the Web, we need resources with tables that resemble relational database tables. Here we introduce GitTables, a corpus of 1M relational tables extracted from GitHub. Our continuing curation aims at growing the corpus to at least 10M tables. Analyses of GitTables show that its structure, content, and topical coverage differ significantly from existing table corpora. We annotate table columns in GitTables with semantic types, hierarchical relations and descriptions from Schema.org and DBpedia. The evaluation of our annotation pipeline on the T2Dv2 benchmark illustrates that our approach provides results on par with human annotations. We present three applications of GitTables, demonstrating its value for learned semantic type detection models, schema completion methods, and benchmarks for table-to-KG matching, data search, and preparation. We make the corpus and code available at https://gittables.github.io. | ['Paul Groth', 'Çağatay Demiralp', 'Madelon Hulsebos'] | 2021-06-14 | null | null | null | null | ['table-annotation', 'table-annotation'] | ['knowledge-base', 'natural-language-processing'] | [-4.83296543e-01 8.47756684e-01 -5.05028665e-01 -4.76516545e-01
-1.26043177e+00 -9.11504030e-01 6.33290172e-01 1.02091014e+00
-1.10742353e-01 7.54829049e-01 5.22733986e-01 -2.81338453e-01
-3.05604279e-01 -1.18627930e+00 -1.14641690e+00 3.05441111e-01
-1.60815731e-01 1.33807516e+00 2.86506861e-01 -5.14328361e-01
-1.96605548e-01 1.49205878e-01 -1.39690769e+00 1.11265552e+00
5.70730627e-01 1.36186934e+00 -2.54694670e-01 2.00849816e-01
-6.25485361e-01 1.10242760e+00 -6.12138987e-01 -1.09429038e+00
2.93458939e-01 2.75801748e-01 -1.17944205e+00 -4.33714926e-01
7.59901762e-01 -1.69870168e-01 -5.46611845e-01 6.75020397e-01
2.38045707e-01 -3.32472831e-01 2.04920679e-01 -1.51664698e+00
-5.70620775e-01 1.60643363e+00 -1.43361390e-02 -1.01831127e-02
4.45471138e-01 -5.12573779e-01 1.47798765e+00 -8.07248235e-01
1.29115152e+00 1.20849884e+00 1.07297873e+00 4.00153905e-01
-1.15432334e+00 -3.49200159e-01 -2.76801735e-01 1.35299653e-01
-1.34496844e+00 -7.65162468e-01 1.40078962e-01 -2.07517117e-01
1.48549795e+00 5.89859545e-01 1.88724995e-01 7.57560432e-01
-2.11244062e-01 7.77235925e-01 3.09105515e-01 -3.65391105e-01
-9.38295498e-02 6.27631545e-01 -2.63058543e-02 6.88879728e-01
7.77578950e-01 -8.19409311e-01 -7.43422270e-01 -1.04214817e-01
3.20340663e-01 -5.54916143e-01 1.49253875e-01 -8.44735086e-01
-1.30843508e+00 3.51912230e-01 5.32518506e-01 3.75945605e-02
-2.31837139e-01 1.77876741e-01 9.97225583e-01 1.59768343e-01
2.41656005e-01 6.38993263e-01 -8.35987031e-01 -2.02026710e-01
-4.00803417e-01 6.84211373e-01 1.40956783e+00 2.01713848e+00
6.25569344e-01 -6.00600421e-01 -8.71891007e-02 1.08459353e+00
-3.57757136e-02 3.93855155e-01 -8.38042870e-02 -1.12178588e+00
1.31189299e+00 1.11337018e+00 1.18958160e-01 -9.01224732e-01
-4.28778261e-01 -1.43618640e-02 -6.24018133e-01 -4.69249517e-01
7.02454805e-01 4.60642904e-01 -4.85694379e-01 1.24897277e+00
2.81952709e-01 -1.16470098e+00 2.56763667e-01 3.50002140e-01
1.60416365e+00 4.16960508e-01 8.52871910e-02 3.14402997e-01
1.66353738e+00 -5.24333358e-01 -8.05376053e-01 -2.31202245e-01
1.12719822e+00 -6.67435408e-01 1.09886062e+00 3.13566506e-01
-1.41096509e+00 -3.99097830e-01 -9.65442359e-01 -7.95845926e-01
-1.21847224e+00 8.02618787e-02 9.97897744e-01 4.29888457e-01
-8.62593949e-01 4.78210360e-01 -6.59094214e-01 -9.15499449e-01
6.84326768e-01 1.28956705e-01 -6.93528712e-01 -1.06573351e-01
-1.32774055e+00 9.83371377e-01 7.98690736e-01 -1.16246305e-01
-4.20600533e-01 -1.04700696e+00 -9.27893460e-01 2.06377506e-01
8.59775484e-01 -4.67196941e-01 1.40502167e+00 1.29709885e-01
-2.83433735e-01 1.30693614e+00 9.42378640e-02 -8.46447825e-01
4.30653363e-01 -1.24982670e-01 -4.85561818e-01 -1.17900416e-01
4.68686730e-01 6.57203436e-01 -3.80449146e-01 -1.10749030e+00
-7.05180645e-01 -2.30873019e-01 2.84017533e-01 7.15772808e-02
-1.71188667e-01 7.56461322e-02 -9.60913241e-01 -2.49962702e-01
1.23191796e-01 -4.22259659e-01 2.56460011e-01 -1.53695509e-01
-8.26269507e-01 -5.09446442e-01 1.88574225e-01 -9.34390604e-01
1.41577601e+00 -1.71912634e+00 -3.30153316e-01 1.08766384e-01
4.26665634e-01 -2.42249787e-01 3.56695622e-01 1.04492855e+00
1.09530509e-01 5.18786550e-01 2.11437908e-03 -2.68021058e-02
6.34023428e-01 2.91550457e-01 -3.84711593e-01 -2.32606575e-01
1.30787075e-01 1.13119149e+00 -4.43953723e-01 -8.97469819e-01
-3.37773532e-01 7.95285497e-03 -7.09361196e-01 5.15340753e-02
-7.71559477e-01 -5.90323269e-01 -1.94603071e-01 1.11381710e+00
4.91288275e-01 -5.39798975e-01 7.04048812e-01 -6.74282670e-01
1.04592308e-01 1.29391587e+00 -1.19054818e+00 1.85003877e+00
-2.87954032e-01 4.17304903e-01 1.90188549e-02 -5.97835422e-01
9.34286296e-01 1.63066715e-01 7.63723552e-01 -9.66934144e-01
-2.87778765e-01 4.72087353e-01 -3.30337822e-01 -4.52044189e-01
8.34469914e-01 6.88485682e-01 -5.25830150e-01 1.94947377e-01
2.30990261e-01 -1.97726220e-01 1.00563622e+00 7.83854127e-01
1.21386623e+00 3.31517071e-01 2.31475800e-01 -3.79652560e-01
3.16200286e-01 6.75981045e-01 4.68627185e-01 7.83977807e-01
3.62524360e-01 2.45978132e-01 9.72557008e-01 -7.77565062e-01
-1.39076388e+00 -1.13128316e+00 -4.72087175e-01 1.01936305e+00
-9.42466483e-02 -1.30344450e+00 -6.74309373e-01 -7.31854498e-01
6.80068791e-01 6.04343176e-01 -6.35029614e-01 2.34583020e-01
-8.39137018e-01 -6.60084486e-01 6.50672674e-01 7.51732290e-01
5.32333791e-01 -1.02395892e+00 -1.69049069e-01 3.23593497e-01
-5.20977914e-01 -1.43494952e+00 2.53326744e-02 4.65521187e-01
-4.45584387e-01 -1.40263093e+00 3.16407353e-01 -6.15819871e-01
2.01460481e-01 -3.93167049e-01 2.21193290e+00 -1.27819344e-01
-2.46530473e-01 2.75936782e-01 -7.37797990e-02 -7.11049736e-01
-7.24557579e-01 7.33440816e-01 -3.55164260e-01 -9.82773602e-01
8.44755590e-01 -2.31520414e-01 -1.43235728e-01 2.56733179e-01
-8.23148787e-01 1.44479394e-01 4.07423764e-01 4.09910053e-01
7.53455162e-01 4.31952626e-02 3.66157264e-01 -1.50638521e+00
4.01518673e-01 -5.59687197e-01 -8.27819526e-01 7.20383942e-01
-9.60466743e-01 9.37362835e-02 3.99156332e-01 4.05414104e-01
-9.53539491e-01 -3.30311477e-01 -1.37782544e-01 1.52964100e-01
1.48173705e-01 8.26127946e-01 -6.11375928e-01 5.61033249e-01
8.70644093e-01 -2.57456481e-01 -1.42100349e-01 -8.46820295e-01
6.19785666e-01 5.06005883e-01 7.98595965e-01 -9.82189178e-01
6.68466926e-01 3.92705083e-01 -1.83219835e-01 -6.79197907e-02
-7.31511772e-01 -1.46476910e-01 -8.14727783e-01 1.85633481e-01
5.58220506e-01 -1.17662370e+00 -1.09359038e+00 -1.45148888e-01
-8.31876040e-01 -4.26020682e-01 -4.97979224e-01 -1.48954049e-01
-4.14313704e-01 -2.28791863e-01 -9.21846151e-01 -2.86818832e-01
-4.06572491e-01 -7.17975676e-01 1.05950153e+00 -3.60093206e-01
-4.99206364e-01 -8.96103442e-01 -2.67941535e-01 6.92098439e-01
2.63482273e-01 2.36938491e-01 1.40334785e+00 -1.11833215e+00
-1.13412321e+00 -2.53466696e-01 -4.09828871e-01 -2.51184165e-01
-7.77736604e-02 -1.56058725e-02 -6.52445138e-01 1.82439417e-01
-7.96109855e-01 -7.75514960e-01 5.73977232e-01 -2.12219089e-01
1.19833291e+00 -5.84307313e-01 -5.10463476e-01 6.43474698e-01
1.58019149e+00 2.50708312e-01 5.84332824e-01 1.01291275e+00
8.87628198e-01 9.05794740e-01 6.49973392e-01 3.51072729e-01
9.75915551e-01 6.85237229e-01 3.63779724e-01 1.26528576e-01
-2.19640821e-01 -7.39182591e-01 -3.49785179e-01 8.61107826e-01
5.00724912e-01 -3.95653009e-01 -1.46531594e+00 5.92546642e-01
-1.72806048e+00 -7.31012642e-01 -2.21101835e-01 1.92092168e+00
1.34810281e+00 5.43766320e-01 7.54346373e-03 -2.48596128e-02
2.58852839e-01 -1.88203946e-01 -5.74307680e-01 -2.82637239e-01
-3.65901560e-01 1.36068806e-01 8.03911328e-01 2.02463165e-01
-1.10452211e+00 7.69664049e-01 5.77632761e+00 6.47182465e-01
-3.46835434e-01 -1.32881597e-01 8.55906606e-01 -1.79987386e-01
-7.15253234e-01 7.82576874e-02 -1.34215748e+00 2.17468947e-01
1.05063879e+00 -4.93793190e-01 5.05870223e-01 1.09634078e+00
-4.66036201e-01 8.84707645e-02 -1.68021107e+00 8.33365262e-01
-2.32254073e-01 -2.07058740e+00 3.62039208e-02 7.28514269e-02
1.05295129e-01 1.14654921e-01 -3.05511415e-01 8.57720733e-01
6.54208362e-01 -1.07082355e+00 1.01344156e+00 2.37881154e-01
9.15893137e-01 -5.46195745e-01 7.48652518e-01 -2.49412686e-01
-1.17601669e+00 1.64383948e-01 -3.97306055e-01 4.40847695e-01
-3.81919801e-01 4.04414445e-01 -1.27981484e+00 6.77974105e-01
1.17206109e+00 7.07608461e-01 -1.08898365e+00 4.80372429e-01
2.72353083e-01 8.38484541e-02 -4.92247105e-01 -4.69479226e-02
-2.06956193e-01 2.60132197e-02 2.71074399e-02 1.13886619e+00
-1.23840673e-02 -3.32609534e-01 -2.20232438e-02 1.14505982e+00
-8.46547723e-01 2.11279690e-01 -6.28116250e-01 -2.80443311e-01
1.12636042e+00 1.21968782e+00 -6.40757918e-01 -7.45884717e-01
-6.28859222e-01 2.07306203e-02 5.13046265e-01 -3.81820574e-02
-7.15246260e-01 -6.57401919e-01 6.41560912e-01 3.70899886e-01
2.32644647e-01 1.79165646e-01 -7.65056014e-01 -1.15395200e+00
6.28303289e-01 -1.32438421e+00 1.00961280e+00 -9.88077462e-01
-1.07882130e+00 7.16945171e-01 4.21258539e-01 -8.23879123e-01
-4.16968316e-01 -4.15759653e-01 3.76007706e-01 8.07643652e-01
-8.90349448e-01 -1.15959406e+00 -3.83535028e-01 3.46243948e-01
2.16895506e-01 -1.49879202e-01 8.85410428e-01 8.23291659e-01
-5.34008086e-01 7.78252363e-01 7.91792572e-02 6.89234793e-01
8.94357026e-01 -1.70863223e+00 1.10202110e+00 6.09444320e-01
1.78855404e-01 8.87116015e-01 5.66772997e-01 -7.57800877e-01
-1.87166274e+00 -9.16188478e-01 1.27650404e+00 -1.07033741e+00
8.45075965e-01 -1.18427801e+00 -1.11832142e+00 1.14553344e+00
3.79834563e-01 -7.69067183e-02 3.70624274e-01 6.74493909e-01
-8.01618457e-01 -8.46013010e-01 -1.11674917e+00 4.76826906e-01
1.22938967e+00 -5.92419446e-01 -4.47891623e-01 6.55187964e-01
8.22707951e-01 -9.40395951e-01 -1.69668186e+00 4.21266526e-01
6.40276730e-01 -7.53293514e-01 1.09158647e+00 -6.45905912e-01
6.21177554e-01 1.53816277e-02 -3.70175183e-01 -8.21153700e-01
-2.62734760e-02 -4.88261849e-01 -4.52917367e-01 1.78924239e+00
9.57243323e-01 -3.04888070e-01 1.17576301e+00 1.07565665e+00
-2.09509209e-01 -6.69395268e-01 -5.63323677e-01 -6.67267323e-01
8.94117281e-02 -3.20117533e-01 1.15108883e+00 9.48492944e-01
4.59928721e-01 2.70202011e-01 2.63980389e-01 -7.05550984e-02
5.44422090e-01 2.18374699e-01 1.10467732e+00 -1.19895673e+00
-1.10077681e-02 -2.58685827e-01 2.41606683e-02 -6.43325865e-01
-2.83829600e-01 -1.29963410e+00 -2.51539916e-01 -2.04805613e+00
7.74505734e-02 -1.04261935e+00 1.76875163e-02 9.01156366e-01
2.22819179e-01 -1.63835764e-01 -7.13816732e-02 4.32408243e-01
-6.30446851e-01 2.45895442e-02 7.18153417e-01 -3.14951479e-01
2.38775998e-01 -5.70397854e-01 -1.11012816e+00 1.88361272e-01
7.48095274e-01 -4.69929308e-01 -4.51081336e-01 -6.83815122e-01
8.85523319e-01 1.43207207e-01 -1.19773157e-01 -1.06759191e+00
2.56228566e-01 6.53725639e-02 4.16693896e-01 -1.00218236e+00
3.38261604e-01 -6.66814864e-01 2.98696488e-01 7.69898668e-02
-7.67940819e-01 4.47490960e-01 5.41588783e-01 -6.91232877e-03
-3.64932120e-01 1.11836173e-01 1.78399906e-01 -5.79409957e-01
-6.32656217e-01 3.11510235e-01 -6.75088614e-02 7.21270144e-01
3.02822590e-01 2.23900616e-01 -9.78324413e-01 -7.66601935e-02
-5.29789627e-01 3.48014712e-01 4.88692075e-01 6.16723180e-01
1.40627787e-01 -1.50206828e+00 -5.00973463e-01 1.16783962e-01
7.50128865e-01 4.65234369e-01 -3.28475296e-01 4.47193325e-01
-9.76643205e-01 9.64150488e-01 -3.04968297e-01 -2.68590987e-01
-9.75597680e-01 7.73095787e-01 1.71511903e-01 -5.52472830e-01
-5.27099550e-01 8.14371228e-01 -1.52517930e-01 -9.06473279e-01
5.47129154e-01 -8.58165443e-01 8.78207833e-02 1.69261396e-01
2.59228408e-01 1.07293695e-01 8.86415541e-01 3.84735353e-02
-5.23895144e-01 -3.67237955e-01 -2.58189440e-01 2.01291159e-01
1.47600508e+00 -7.65945688e-02 -5.40646195e-01 4.74755108e-01
1.01592374e+00 2.67709583e-01 -6.22369468e-01 -2.66183585e-01
8.06929350e-01 -2.32749015e-01 -5.60996592e-01 -1.32115293e+00
-7.61198342e-01 3.23620766e-01 -9.93212461e-02 6.15193963e-01
7.72130311e-01 2.72810698e-01 7.94286370e-01 1.00821984e+00
5.30592561e-01 -1.07383716e+00 -4.53275949e-01 4.73922104e-01
9.58230674e-01 -1.37359309e+00 1.98252425e-01 -7.22446442e-01
-4.09690082e-01 1.03685462e+00 9.52197611e-01 5.02840459e-01
5.83438396e-01 6.56286776e-01 1.97697014e-01 -5.88619769e-01
-1.32886279e+00 -1.13063321e-01 2.37782925e-01 6.01863086e-01
7.79662609e-01 -1.24274030e-01 4.17901799e-02 5.77253222e-01
-6.12332463e-01 -1.38983816e-01 5.15624285e-01 1.10715187e+00
-1.16672993e-01 -1.37802529e+00 -1.52949557e-01 7.83505082e-01
-5.60147166e-01 -4.43128407e-01 -5.31902909e-01 1.35136628e+00
7.43822241e-03 5.82875133e-01 4.15527403e-01 -1.91739351e-01
6.86944127e-01 2.43326291e-01 3.83512229e-01 -6.71199024e-01
-7.58841991e-01 -2.75440425e-01 1.08286870e+00 -8.35817933e-01
4.82395664e-02 -6.41411126e-01 -1.04815674e+00 -7.23956048e-01
4.55087215e-01 5.47565520e-01 6.83643699e-01 2.14706197e-01
5.84952414e-01 3.80661070e-01 -3.07557881e-01 2.48941556e-01
-1.44872472e-01 -1.06420076e+00 -3.27142209e-01 6.78574681e-01
-3.21117431e-01 -3.96012127e-01 2.53839403e-01 2.26001635e-01] | [9.56798267364502, 7.960536479949951] |
3233cb65-d4ea-4a15-b516-1d011ba51f6a | opttyper-probabilistic-type-inference-by | 2004.00348 | null | https://arxiv.org/abs/2004.00348v3 | https://arxiv.org/pdf/2004.00348v3.pdf | OptTyper: Probabilistic Type Inference by Optimising Logical and Natural Constraints | We present a new approach to the type inference problem for dynamic languages. Our goal is to combine \emph{logical} constraints, that is, deterministic information from a type system, with \emph{natural} constraints, that is, uncertain statistical information about types learnt from sources like identifier names. To this end, we introduce a framework for probabilistic type inference that combines logic and learning: logical constraints on the types are extracted from the program, and deep learning is applied to predict types from surface-level code properties that are statistically associated. The foremost insight of our method is to constrain the predictions from the learning procedure to respect the logical constraints, which we achieve by relaxing the logical inference problem of type prediction into a continuous optimisation problem. We build a tool called OptTyper to predict missing types for TypeScript files. OptTyper combines a continuous interpretation of logical constraints derived by classical static analysis of TypeScript code, with natural constraints obtained from a deep learning model, which learns naming conventions for types from a large codebase. By evaluating OptTyper, we show that the combination of logical and natural constraints yields a large improvement in performance over either kind of information individually and achieves a 4% improvement over the state-of-the-art. | ['Andrew D. Gordon', 'Earl T. Barr', 'Charles Sutton', 'Irene Vlassi Pandi'] | 2020-04-01 | null | null | null | null | ['type-prediction'] | ['computer-code'] | [-0.04340437 0.36808464 -0.4887595 -0.6591929 -1.029332 -0.7170144
0.5773888 0.33383736 -0.27816105 0.7951315 0.14266889 -0.6729876
0.05952645 -1.0461072 -1.3059711 -0.32822958 -0.38842875 0.38543355
0.28277433 0.09120496 0.24196438 0.09891213 -1.9565369 0.9221579
0.7243959 1.2205026 -0.00906971 0.7116849 -0.6575671 1.126154
-0.42363888 -0.35529843 -0.16535707 0.18198268 -1.0660924 -0.7155239
0.50410426 -0.12505306 0.3497402 1.087632 -0.06310071 -0.40471333
0.5771246 -1.5249977 -0.37648678 1.2196677 -0.2080092 -0.3244628
0.47413796 0.09244663 1.5236696 -0.69560677 0.59678316 1.5097128
0.9393976 0.49816108 -1.6151158 -0.1326739 0.2421307 0.26137784
-1.1766725 -0.2993312 0.4116087 -0.90219647 1.2397792 0.47131145
0.09011617 0.9325518 0.2042792 0.9433012 1.0217777 -0.47628242
0.6018405 0.5397454 0.36833706 0.7279814 0.19976759 0.22173136
-0.32628766 -0.76034826 -0.02422782 -0.6098726 0.16828482 -0.41086677
-1.0637219 0.6607232 0.13462226 -0.01928181 0.10406329 0.53377646
0.7191554 0.12235114 0.13378353 0.43136802 -1.3400621 -0.4313521
-0.7142348 0.49860957 1.2707181 1.2017853 0.9225484 -0.27942806
-0.03110243 0.67494774 0.67687553 0.714004 0.02479429 -0.9307736
0.14465238 0.68224245 -0.01086597 -0.3954392 -0.12082759 0.17681749
-0.21903126 0.5735419 0.69950604 -0.2762657 -0.6002096 1.9615766
0.22064656 0.05658864 -0.04500652 0.34045637 0.5410059 0.43686777
-0.08829309 0.0781009 1.3085922 -0.2290634 -0.28062695 -0.12441145
1.0507553 -0.41293412 0.9311321 0.48936525 -0.8130465 -0.07624456
-0.8016028 -0.12788118 -0.57407516 -0.16078784 0.6518562 0.6794737
-0.9260399 0.5387 -0.9226213 0.3016671 0.29993802 0.33135244
-0.22478701 0.16936317 -0.9422432 0.512766 0.522862 0.05701358
-0.73424536 -1.0595925 -1.0236596 0.09032019 0.82842225 -0.587133
1.5723139 -0.98109746 -1.146482 0.7646986 -0.46780005 -0.37522215
0.30492342 -0.08779463 -0.01133536 -0.611453 -0.16509676 -0.09252463
0.368496 -1.5824151 -0.92807 -0.27058282 0.59083503 -0.6647793
0.09245791 0.03714383 -0.2578086 -0.24468507 -0.24844973 -0.9146927
0.01455804 -0.03477523 -0.6530756 -0.5486785 0.54713947 -0.6592228
1.1653194 -2.035169 0.32667702 0.5161921 0.20647702 -0.17886224
0.23779438 0.0510778 -0.02487211 0.52012086 -0.4865321 -0.11801426
0.8148857 0.50237674 -0.46149758 0.12848318 0.22737648 0.69867545
-0.9918972 -0.34728912 -0.25463015 0.17578329 -1.258282 0.43361345
-1.5338286 -0.3074184 -0.4426621 0.5823634 0.73943096 0.0954662
0.55621576 0.03997495 -0.20736672 0.85383844 -1.3806174 1.5488796
-0.90969646 0.06717173 0.21550074 -0.86256397 0.56960857 0.0750057
0.31488854 -0.12441435 -0.16022669 0.40867448 -0.27628177 -0.7346131
0.3656799 -0.05310426 -0.7014935 0.6241705 0.10583971 -0.17352569
0.01400667 0.06932428 1.2395258 0.73618734 0.08353319 -0.43644473
0.63903844 -0.02999034 1.0128844 1.0817829 0.43951064 -0.01935649
1.4539866 -0.4975604 -0.902069 -1.1875666 -0.0963058 1.1338289
-0.43559465 -0.81528205 -0.6228085 -1.1567593 0.31774855 0.7818337
-0.6238195 0.02163875 -0.74650145 -0.60308635 0.567922 0.6492092
-0.05180442 -0.92938995 -0.36208355 0.03579286 -0.12942284 -0.90316993
0.08539216 0.37783802 -0.37240985 -1.1837944 0.2508543 -0.5199662
0.33759242 -0.95681477 1.5004979 0.3645068 -0.1766916 0.03417232
-0.14699113 -0.41473886 -0.85433394 0.12801763 -0.06203425 -0.2910098
0.34915245 -0.6831352 0.20497333 -0.2549321 -0.8541027 -0.20627773
0.5878996 0.9942809 0.47681463 0.39629292 -0.04977793 -1.4749368
-0.05691438 -0.5970943 -1.2238153 0.3303982 -0.4643858 0.869419
1.0265418 -0.0898786 -1.1452135 0.10173094 -0.39642656 0.06331211
-0.39045873 0.95572895 -0.78334457 0.3594556 0.5941906 0.0690394
-0.17947 -0.59745246 0.5025143 0.69551146 0.8230335 -1.6354889
0.9304919 0.19721895 0.03082306 -0.4563293 -0.8211302 -0.03556218
-0.7259532 0.2356876 0.5373539 -0.48630977 -1.1388348 0.68682384
-1.1373082 -0.8815856 -0.08916846 0.10882539 -0.8414755 0.3278032
-0.5200563 -1.0394799 0.21424069 -1.2411208 1.2803862 -0.3788719
-0.1531392 -1.184272 -0.05081449 -0.07583074 0.21277976 0.299589
1.8176962 -0.69945663 -0.96282804 0.04670333 -0.20347078 0.35724652
-0.18783818 0.33354637 -1.1390902 0.13408896 -0.15773824 -0.2825082
0.59102696 0.07945707 1.5452775 -0.8624385 -0.3362739 0.75761485
1.7341549 -0.21539937 0.51872575 0.3000924 0.77096236 0.48724908
0.38379788 0.71174645 0.67011577 0.7835087 0.5536577 0.4200028
0.2795069 -0.25666305 0.54229623 0.29816556 0.26381823 0.1956757
-1.1526567 0.32381245 -1.8746759 -1.1265486 -0.22281668 2.529973
1.6062869 0.36385813 0.39526534 -0.03243229 0.33085206 -0.18567972
-0.33020806 -0.60183126 0.260404 0.1565187 0.5716483 0.91087073
-0.97873133 0.59418553 5.949174 0.68782145 -0.95660543 -0.22526291
0.13319634 0.23275015 -0.94576716 0.49761313 -0.98259866 0.615023
1.060633 -0.34408766 0.8021376 1.1365557 -0.13796411 -0.26284108
-1.8351102 0.28175312 -0.21411957 -1.3479347 -0.2820314 0.28754526
0.35971773 0.09961114 -0.01812566 0.65772057 1.0132049 -0.9517697
1.0734638 0.7891982 0.6824316 -0.5665093 0.7981144 0.47310615
-0.9347488 -0.3200421 -0.06292915 -0.03295132 -0.2139656 1.1154752
-0.8225501 0.6325951 0.57384473 0.5939589 -0.45502198 0.8725775
-0.3975846 0.52591425 -0.45036376 -0.08814471 -0.25200292 0.39151144
0.48963508 1.4534827 -0.05793373 -0.34140584 0.28591314 1.602206
0.01550248 -0.448616 -0.47347394 0.27991095 0.560175 1.029418
-0.01779749 -0.59651846 -0.512093 0.11058124 0.4032687 0.142858
-0.7625837 -0.4233913 0.8046601 -0.0656843 0.41648844 -0.21132733
-0.6030265 -1.341863 0.5589468 -1.1649305 0.51901156 -0.4699954
-1.3056648 0.13798262 0.4978168 -0.48620936 -0.67723495 -1.0934111
-0.3910109 1.2071611 -1.4749572 -0.9744316 0.14829594 0.11202671
-0.01642526 0.08759014 1.0566741 -0.03981661 -0.4636542 0.88837445
-0.04276701 0.35502186 0.33171237 -1.9624974 0.47541848 0.9545436
-0.25755006 0.89468634 0.78125095 -0.6028168 -2.1720784 -1.1369135
1.0124217 -0.8695059 1.0621209 -0.80546427 -1.0599573 0.8687436
-0.34917378 0.23323148 0.4467504 0.7575685 -1.2407165 0.02568874
-1.0592296 0.26240337 0.73271453 -0.86661273 -0.7463787 0.17234693
0.69085556 -0.45456538 -1.2820424 0.29207528 0.99390435 -0.89366364
0.69758916 -0.68326384 0.56396294 -0.37241736 -0.6372987 -1.1180418
0.0344118 -0.5412781 -0.34961784 1.4661667 0.8122311 -0.5701585
0.71238005 1.1372244 -0.20107795 -0.5606111 -0.75734437 -0.81148887
0.45271775 -0.8566864 0.73514473 0.80764157 0.5323788 -0.0140979
0.12117918 0.5747924 0.6644117 0.2990204 0.8317071 -1.1768193
-0.9998753 -0.6159119 -0.4523051 -0.47728953 0.8519671 -1.1450539
0.5692744 -1.1396596 0.16829312 -1.1032684 -0.11504007 0.99214256
-0.05519029 -0.38100663 -0.08696014 -0.27434707 -0.30756977 0.10644005
0.17181347 -0.3432497 0.0690436 0.01636832 -0.71485984 0.9905969
0.38212472 -0.56508607 -0.00915458 -0.4557986 0.95287263 0.27850726
0.49320307 -0.6384843 0.10137636 -0.3189191 -0.21859284 -0.4021757
-0.17983276 -0.6072536 -0.01064495 0.18424451 -0.60421306 -0.11907296
0.32883185 0.28696722 0.12833592 -0.37774652 0.46444112 -0.10643341
-0.95923465 -0.02899386 -0.20850718 0.4362847 0.309829 0.43580854
-0.5050647 0.22088222 -0.7310038 0.14659716 0.69710106 0.45072168
0.07613073 -1.1665419 -0.6241357 0.3907172 0.31249705 0.16498552
-0.3181117 0.7078591 0.02194924 0.07663189 0.1604551 -0.66891164
-1.003399 0.74426013 0.25660977 -0.11633483 -0.36846823 0.8239571
-0.08977196 -0.91722924 0.31528392 -0.8075029 0.28167754 -0.3408179
0.5601533 -0.11684176 0.30134955 -0.20872378 -0.5990557 0.11476409
-0.0739256 -0.17171355 1.4142494 0.38031718 -0.6896372 0.96163696
1.2433763 0.4276244 -1.1000613 -0.4591938 0.47908303 -0.3828481
-0.23697338 -1.3258129 -0.6654092 0.5068968 0.00685801 0.29960313
0.730187 0.17992674 0.31904712 0.6437512 0.7080911 -0.89148873
-0.49377853 0.8505044 0.53596413 -0.9054409 -0.12038016 -0.34793252
0.03359399 1.1339043 0.4094285 0.083343 0.7234091 1.1852009
-0.20530175 0.09514863 -1.2859814 0.03857574 0.19337326 0.78437006
0.64375246 0.13089383 0.11599567 0.8113897 -0.42873904 0.23636691
0.7593334 0.9712271 -0.21566781 -1.5488073 -0.44603488 0.5845362
-0.63365716 -0.28551847 -0.13463145 0.71030474 0.4175738 0.6967538
-0.1403015 -0.40372854 0.12127201 0.27319306 0.5934286 -0.91451687
-0.30711612 -0.34344772 0.44958314 -0.7763137 -0.02378838 -0.9671509
-1.2267553 -0.27615806 -0.09567753 0.18800631 0.75648755 1.1671834
0.22555122 0.4064686 0.5403375 -0.59866804 -0.9812407 -0.47374365
-0.4470953 0.2155717 0.47538492 -0.6052201 -0.6324149 0.28749824] | [8.027800559997559, 7.50396728515625] |
13e3ffa8-3f49-495a-828f-c4a395b48c36 | thundr-transformer-based-3d-human | 2106.09336 | null | https://arxiv.org/abs/2106.09336v1 | https://arxiv.org/pdf/2106.09336v1.pdf | THUNDR: Transformer-based 3D HUmaN Reconstruction with Markers | We present THUNDR, a transformer-based deep neural network methodology to reconstruct the 3d pose and shape of people, given monocular RGB images. Key to our methodology is an intermediate 3d marker representation, where we aim to combine the predictive power of model-free-output architectures and the regularizing, anthropometrically-preserving properties of a statistical human surface model like GHUM -- a recently introduced, expressive full body statistical 3d human model, trained end-to-end. Our novel transformer-based prediction pipeline can focus on image regions relevant to the task, supports self-supervised regimes, and ensures that solutions are consistent with human anthropometry. We show state-of-the-art results on Human3.6M and 3DPW, for both the fully-supervised and the self-supervised models, for the task of inferring 3d human shape, joint positions, and global translation. Moreover, we observe very solid 3d reconstruction performance for difficult human poses collected in the wild. | ['Cristian Sminchisescu', 'Rahul Sukthankar', 'William T. Freeman', 'Eduard Gabriel Bazavan', 'Andrei Zanfir', 'Mihai Zanfir'] | 2021-06-17 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Zanfir_THUNDR_Transformer-Based_3D_Human_Reconstruction_With_Markers_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Zanfir_THUNDR_Transformer-Based_3D_Human_Reconstruction_With_Markers_ICCV_2021_paper.pdf | iccv-2021-1 | ['3d-human-reconstruction'] | ['computer-vision'] | [-1.19210862e-01 6.60190880e-01 9.74290073e-02 -5.83398938e-01
-5.42464674e-01 1.18299499e-01 4.21942055e-01 -3.51731896e-01
-3.39268386e-01 4.77550834e-01 5.32756150e-01 3.74298215e-01
9.25146490e-02 -4.45700079e-01 -9.38380420e-01 -2.30250493e-01
-1.69348478e-01 1.36991155e+00 -1.12921812e-01 -3.39968383e-01
-5.47227025e-01 5.45619130e-01 -1.44608521e+00 -1.12971999e-01
3.16445202e-01 1.10602534e+00 -3.76589596e-01 6.25794768e-01
6.90820694e-01 4.35957372e-01 -6.60376921e-02 -5.94421387e-01
6.29514754e-01 -2.36621201e-01 -6.09218419e-01 2.37764925e-01
8.66101980e-01 -4.65894938e-01 -6.33027852e-01 3.95122170e-01
8.55992734e-01 -6.17753267e-02 7.04672098e-01 -1.21935332e+00
-3.57097000e-01 2.07085565e-01 -2.69285917e-01 -5.81008971e-01
7.11475790e-01 3.57093036e-01 5.98478198e-01 -7.67338097e-01
8.62823367e-01 1.54993176e+00 1.46232820e+00 7.17552245e-01
-1.24628007e+00 -2.64919907e-01 -2.49162421e-01 -3.33988011e-01
-1.34869683e+00 -5.27544737e-01 8.72401178e-01 -6.57945633e-01
8.10269415e-01 2.04024494e-01 1.28140616e+00 1.36297143e+00
2.77733654e-01 7.41801322e-01 9.35189128e-01 -3.33981037e-01
-1.11549817e-01 -3.41680616e-01 -3.79407853e-01 1.17121506e+00
1.43949494e-01 2.07553431e-01 -7.96640456e-01 -7.95545503e-02
1.11304176e+00 -5.59732392e-02 2.85168849e-02 -1.09595871e+00
-1.37577629e+00 3.65889102e-01 6.86121106e-01 -2.71760225e-01
-5.69222033e-01 4.89687145e-01 2.31694400e-01 -1.55747071e-01
8.38350773e-01 3.63440253e-02 -8.40409875e-01 -1.29095316e-01
-9.98186409e-01 8.43067825e-01 6.54119432e-01 9.52477872e-01
4.89172667e-01 1.87424216e-02 -3.92317139e-02 4.29648191e-01
6.62076771e-01 1.05641544e+00 1.93240508e-01 -1.22918701e+00
3.53650808e-01 7.30454326e-01 4.89198528e-02 -9.16806281e-01
-9.73688722e-01 -4.11117405e-01 -9.34594214e-01 3.04044485e-01
6.20375514e-01 -4.30318639e-02 -1.03641653e+00 1.73938632e+00
6.56752050e-01 -4.27769363e-01 -1.98492765e-01 1.23764384e+00
9.12060022e-01 -1.13372887e-02 1.21750925e-02 4.28556263e-01
1.15936887e+00 -7.25818932e-01 -6.22590072e-02 -4.10282671e-01
3.05755824e-01 -2.50414789e-01 9.29343164e-01 1.53871134e-01
-1.46532166e+00 -7.73114145e-01 -7.20547199e-01 -4.93567109e-01
2.08869278e-02 2.02147737e-01 3.94089699e-01 5.16393960e-01
-9.99035060e-01 9.36443329e-01 -1.08805645e+00 -6.36025608e-01
3.98243636e-01 5.01745582e-01 -8.16130638e-01 1.71724364e-01
-8.49935234e-01 1.20270312e+00 -3.52615081e-02 4.12273437e-01
-7.30678797e-01 -7.34152615e-01 -1.17893815e+00 -5.45884788e-01
5.34661859e-02 -1.42686570e+00 1.12978148e+00 -5.91229975e-01
-1.50538421e+00 1.60689104e+00 4.00668383e-02 -6.31848574e-01
1.26628959e+00 -7.92085171e-01 2.54209042e-01 5.88579699e-02
-4.96206731e-02 1.00038552e+00 6.36539638e-01 -1.20091462e+00
5.47592901e-02 -8.77888024e-01 -4.82458472e-01 3.82525414e-01
3.76654118e-01 -2.88653493e-01 -5.22689104e-01 -5.35368145e-01
2.94134796e-01 -1.10684609e+00 -4.04592216e-01 6.54151857e-01
-6.41531289e-01 1.67862609e-01 2.26500779e-01 -1.32290721e+00
2.52877474e-01 -1.73613250e+00 8.01331699e-01 3.97901356e-01
2.62799084e-01 -9.56645310e-02 2.80246049e-01 1.92243624e-02
1.30557135e-01 -4.58965391e-01 -3.81819040e-01 -1.11316657e+00
4.39169973e-01 2.84812152e-01 2.41089791e-01 9.29279745e-01
2.50833929e-02 1.27773523e+00 -4.97789264e-01 -7.15604782e-01
4.75146413e-01 7.99102187e-01 -5.54192185e-01 5.30299723e-01
-2.36337394e-01 8.17157149e-01 -2.75792293e-02 7.82778859e-01
3.89663875e-01 -1.67812958e-01 1.46020025e-01 -4.57143575e-01
3.34119618e-01 -4.56572846e-02 -9.76672351e-01 2.27860856e+00
-2.78165638e-01 2.27570862e-01 1.38817564e-01 -5.85383356e-01
9.32601333e-01 2.80555993e-01 8.79486144e-01 -5.60584068e-01
3.47043008e-01 2.10080251e-01 -4.01848018e-01 -2.60429353e-01
2.76144922e-01 -2.42063791e-01 -1.77080587e-01 3.16679209e-01
2.43708819e-01 -2.69837864e-02 -5.55243909e-01 -2.52160192e-01
5.53789914e-01 1.28448284e+00 3.15142363e-01 -2.83365637e-01
-2.02126056e-02 -2.17213631e-01 3.41751546e-01 2.29492709e-01
-1.19322747e-01 1.13262451e+00 4.58892249e-03 -9.17447448e-01
-1.47503519e+00 -1.34056556e+00 8.00011829e-02 9.83831048e-01
-1.22223996e-01 -2.43033618e-01 -8.06371152e-01 -4.50083375e-01
4.97662723e-01 1.99887127e-01 -1.03952229e+00 -9.90450606e-02
-9.34892893e-01 -4.80011791e-01 7.41440475e-01 7.43940473e-01
3.33700985e-01 -8.52463126e-01 -1.27792227e+00 5.06885499e-02
-2.67842501e-01 -1.17980874e+00 -2.64540136e-01 2.34873462e-02
-8.15832317e-01 -1.17416561e+00 -1.03708029e+00 -5.04499972e-01
6.62721515e-01 -6.26919508e-01 1.37390876e+00 -1.09114334e-01
-2.71639615e-01 5.68326652e-01 4.30897549e-02 -1.69358552e-01
-2.57406145e-01 -1.16117038e-01 4.99740869e-01 -1.75332680e-01
6.55558482e-02 -7.54299760e-01 -6.87157214e-01 3.21400106e-01
2.76324116e-02 3.82647216e-01 3.18609029e-01 3.64428550e-01
7.38150418e-01 -7.78567135e-01 -2.39043146e-01 -3.35141838e-01
-1.18092790e-01 1.54017925e-01 -9.65550691e-02 7.89372176e-02
-1.59155756e-01 1.46602541e-01 1.87575623e-01 -2.98297644e-01
-6.64942741e-01 6.19118035e-01 -4.37865108e-01 -5.04347920e-01
-3.36485714e-01 -3.92784774e-02 -3.20704013e-01 -6.24351166e-02
9.40233469e-01 1.18459560e-01 5.15553355e-01 -7.61216938e-01
5.30121505e-01 3.23434263e-01 1.03780186e+00 -8.38106096e-01
8.92543435e-01 6.83286071e-01 4.51001883e-01 -7.40234792e-01
-7.30740130e-01 -1.39927328e-01 -1.42614484e+00 -3.61413091e-01
9.87101436e-01 -1.19584692e+00 -8.58127058e-01 7.98201382e-01
-1.08763731e+00 -8.37528050e-01 -5.92570364e-01 2.08820701e-01
-1.13902533e+00 3.86944652e-01 -5.88081598e-01 -9.30912495e-01
-8.18338096e-01 -7.95959234e-01 1.81563306e+00 -2.13398367e-01
-6.71902418e-01 -8.46206725e-01 2.41413265e-01 4.48913515e-01
2.78323274e-02 1.11496079e+00 2.92565882e-01 -3.12621176e-01
-7.27934763e-02 -2.63396800e-01 1.92481831e-01 1.56861618e-01
-2.45232835e-01 -3.85474741e-01 -9.59946632e-01 -2.98998505e-01
-4.72326726e-01 -6.72895730e-01 4.58495319e-01 5.08704722e-01
6.59868062e-01 -2.64063179e-01 -1.62896618e-01 9.86769736e-01
8.15555632e-01 -9.35519934e-01 4.75657403e-01 2.65376389e-01
1.07979488e+00 8.47719610e-01 3.29645783e-01 6.01647437e-01
8.51456940e-01 1.01917064e+00 4.02705967e-01 -3.62466872e-01
-3.35299075e-01 -7.48865366e-01 1.99567914e-01 5.52618027e-01
-6.21681690e-01 4.12181199e-01 -1.07533967e+00 1.98507220e-01
-1.85160840e+00 -4.80664402e-01 -3.76015902e-02 2.18278074e+00
7.31322110e-01 1.44741669e-01 7.80726135e-01 8.22448879e-02
1.62866697e-01 -3.88984978e-02 -6.03898346e-01 2.38331571e-01
-1.38068080e-01 2.61234075e-01 4.69735950e-01 4.15567845e-01
-1.12548053e+00 7.32857585e-01 6.48831892e+00 5.50541840e-02
-9.14382935e-01 1.07288510e-01 4.34041381e-01 -3.50564510e-01
7.84544870e-02 -4.91686910e-01 -5.22666574e-01 1.85584188e-01
7.66931832e-01 4.22833025e-01 1.77769735e-01 8.62345099e-01
2.05964878e-01 2.51069337e-01 -1.43249714e+00 1.21532547e+00
3.37342978e-01 -8.66623938e-01 -2.24578962e-01 2.08637923e-01
3.85050148e-01 9.45257396e-02 -1.15937769e-01 1.80801675e-02
1.88202649e-01 -1.16212642e+00 1.49577701e+00 9.69893157e-01
9.79608655e-01 -5.15623152e-01 4.65305537e-01 5.39193451e-01
-1.03824675e+00 3.79107207e-01 -2.53681540e-01 -3.20942625e-02
3.92321527e-01 3.26637596e-01 -6.63075566e-01 6.86983228e-01
9.44806933e-01 6.96175158e-01 -7.03069031e-01 6.67931020e-01
-1.74276680e-01 7.54218968e-03 -6.45055711e-01 3.63468349e-01
-3.91589791e-01 -1.45077314e-02 4.13078725e-01 1.10579121e+00
1.62020504e-01 -1.13430835e-01 3.48805577e-01 8.36318314e-01
1.66132972e-01 -7.91889131e-02 -4.67654794e-01 5.28143108e-01
-1.90969959e-01 8.49341393e-01 -3.25950980e-01 -2.21108273e-02
2.39970654e-01 1.13202620e+00 3.49465191e-01 2.42649652e-02
-7.56193161e-01 3.47890496e-01 5.72870851e-01 6.13125801e-01
1.00822680e-01 -5.41741252e-01 -5.05730748e-01 -1.19406879e+00
2.97009826e-01 -8.26863647e-01 1.89239979e-01 -1.08942783e+00
-1.15912426e+00 4.25625831e-01 2.11076871e-01 -1.03367388e+00
-6.55311704e-01 -7.83028960e-01 -2.58032829e-02 7.65630066e-01
-8.47958624e-01 -2.06503749e+00 -3.54284257e-01 5.74226916e-01
6.41659275e-02 4.07807343e-02 8.76448929e-01 2.38365177e-02
-9.88398045e-02 5.66050529e-01 -3.64547402e-01 3.40063632e-01
5.78474402e-01 -1.35854554e+00 8.84583235e-01 5.95020592e-01
3.31500731e-02 4.63540733e-01 8.51763189e-01 -8.51694345e-01
-1.62027931e+00 -1.01928437e+00 8.61730814e-01 -1.04046202e+00
-2.15151645e-02 -6.74856186e-01 -3.97560984e-01 9.72013474e-01
-4.16403204e-01 1.78892523e-01 4.68326539e-01 2.93488204e-01
-3.44071031e-01 4.15672548e-02 -1.39761639e+00 5.10008931e-01
1.64962184e+00 -4.00039256e-01 -8.38992596e-01 3.70915413e-01
5.34159720e-01 -1.12066483e+00 -1.13972747e+00 4.26666111e-01
1.27272582e+00 -9.64458466e-01 1.51875806e+00 -7.45792210e-01
4.55933183e-01 2.07762588e-02 -4.51960713e-01 -1.21073771e+00
-1.60859898e-01 -6.92686141e-01 -5.29325545e-01 6.02230072e-01
-6.90463185e-02 -2.68488433e-02 1.30408335e+00 7.86176980e-01
7.39807189e-02 -8.15716565e-01 -1.20685315e+00 -6.31397128e-01
1.21179923e-01 -4.53416854e-01 4.80534613e-01 4.58890676e-01
-2.98773766e-01 6.63311779e-02 -1.08459842e+00 -1.53702060e-02
1.18105257e+00 -8.30614194e-02 1.39437103e+00 -1.56093717e+00
-4.30459291e-01 -5.99201769e-02 -7.05634415e-01 -1.14100718e+00
1.81505829e-01 -6.66040838e-01 1.61766976e-01 -1.53203154e+00
-4.89710970e-03 -1.27269417e-01 3.65719557e-01 6.31411195e-01
1.44238800e-01 6.50513291e-01 2.93148160e-01 1.10741086e-01
-4.91829515e-01 7.65861750e-01 1.20941663e+00 1.75706014e-01
1.66306514e-02 -3.75284664e-02 -2.33280048e-01 9.34020877e-01
3.33115131e-01 -1.30295947e-01 7.47026280e-02 -4.65079784e-01
2.07931221e-01 -1.03698432e-01 1.00763297e+00 -1.16004550e+00
-1.17210127e-01 1.01547234e-01 1.04020607e+00 -7.88648784e-01
8.27897310e-01 -8.41036618e-01 5.30169129e-01 7.80792356e-01
-2.23327339e-01 9.32025239e-02 -1.41960025e-01 2.16070339e-01
6.01079881e-01 6.38330817e-01 7.98001468e-01 -4.55751717e-01
-3.36699098e-01 5.60064316e-01 3.77083600e-01 3.19697976e-01
4.66215044e-01 -4.89229530e-01 2.61782110e-01 -3.74023408e-01
-9.88105237e-01 9.13749114e-02 7.82153189e-01 3.18631321e-01
7.32210398e-01 -1.57065427e+00 -1.05453408e+00 3.53666067e-01
1.36216581e-01 3.68966460e-01 5.08356877e-02 8.19459260e-01
-8.13725233e-01 1.88226312e-01 -5.56997836e-01 -9.47124064e-01
-1.13070011e+00 1.86196476e-01 8.72866273e-01 -9.31963474e-02
-1.20674562e+00 6.71808481e-01 -1.97234333e-01 -1.21578753e+00
4.07552063e-01 -4.13807631e-01 4.33825493e-01 -4.49677855e-01
4.95855100e-02 4.62991178e-01 -2.75486391e-02 -1.35898972e+00
-5.09720862e-01 9.36791122e-01 8.29642594e-01 -2.10938320e-01
1.49216259e+00 -9.20409337e-02 9.45619196e-02 5.52132607e-01
1.04865122e+00 -1.92370787e-01 -1.65307236e+00 -3.58188838e-01
-4.69368279e-01 -3.08466643e-01 -3.93756688e-01 -8.41420233e-01
-8.24683011e-01 8.87199581e-01 7.41277635e-01 -5.31603336e-01
5.80443799e-01 1.96759239e-01 8.04730237e-01 2.66133010e-01
7.43800759e-01 -9.13643718e-01 -2.45337281e-02 3.41075182e-01
1.16745484e+00 -1.00285053e+00 4.41067010e-01 -2.56756604e-01
-4.41229373e-01 8.94328654e-01 5.70725441e-01 -2.72666454e-01
5.93724668e-01 -7.64187146e-03 1.32429376e-01 -2.42115557e-01
-1.58067390e-01 -1.26734242e-01 8.10218036e-01 1.13938236e+00
3.93227339e-01 3.25502574e-01 3.84515822e-01 5.25989830e-01
-8.57315719e-01 3.68992239e-02 -2.67015427e-01 7.93981254e-01
-1.52576819e-01 -7.73209453e-01 -6.11640692e-01 1.00541055e-01
-1.56846624e-02 3.72455657e-01 -5.81921637e-01 9.97762680e-01
3.44643205e-01 1.57434180e-01 4.63651717e-02 -4.56150025e-01
8.14015985e-01 2.07293585e-01 1.12717700e+00 -2.72545934e-01
-5.84972680e-01 1.68990977e-02 2.10153043e-01 -8.61355841e-01
-5.75860858e-01 -8.26934814e-01 -9.94866490e-01 -3.67371529e-01
3.32923412e-01 -4.55088705e-01 5.56578279e-01 1.11716044e+00
1.92197233e-01 1.30454510e-01 -5.56408009e-03 -1.65537441e+00
-6.52239084e-01 -1.04123485e+00 -4.33863580e-01 8.76417935e-01
3.24748486e-01 -8.32814872e-01 1.50078446e-01 1.02884077e-01] | [6.977191925048828, -1.0321201086044312] |
8e364742-1749-4a1d-876a-63182fc0ef96 | microservice-deployment-in-edge-computing | null | null | https://ieeexplore.ieee.org/abstract/document/9712168 | https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9712168 | Microservice Deployment in Edge Computing Based on Deep Q Learning | The microservice deployment strategy is promising in reducing the overall service response time in the microservice-oriented edge computing platform. However, existing works ignore the effect of different interaction frequencies among microservices and the decrease in service execution performance caused by the increased node loads. In this article, we first model the invocation relationships among microservices as an undirected and weighted interaction graph to characterize the communication overhead. Then, we propose a multi-objective microservice deployment problem (MMDP) in edge computing. MMDP aims to minimize the communication overhead while achieving load balance between edge nodes. Without the requirement for domain experts, we propose Reward Sharing Deep Q Learning (RSDQL), a learning-based algorithm, to solve MMDP and obtain the optimal deployment strategy. In addition, to improve the scalability of the services, we propose an Elastic Scaling algorithm (ES) based on heuristics to deal with the dynamic pressure of requests. Finally, we conduct a series of experiments in Kubernetes to evaluate the performance of our approach. Experimental results indicate that, compared with interaction-aware strategy and Kubernetes default strategy, RSDQL has shorter response times, more balanced resource loads, and makes services scale elastically according to the request pressure | ['Pengfei Yang', 'Quan Wang', 'Wenkai Lv'] | 2022-02-11 | null | null | null | ieee-transactions-on-parallel-and-distributed-2 | ['q-learning', 'edge-computing'] | ['methodology', 'time-series'] | [-9.29002583e-01 -2.15370610e-01 3.22357155e-02 -2.47063950e-01
4.48579527e-02 -3.56502146e-01 -3.85549329e-02 -3.76956582e-01
8.57018027e-03 5.25165260e-01 1.08685181e-01 -4.20705944e-01
-5.92214942e-01 -9.41553533e-01 -4.25410241e-01 -9.63700175e-01
-2.19729647e-01 5.21958768e-01 3.55253890e-02 -1.29339635e-01
1.00518577e-01 5.26650786e-01 -1.23505425e+00 -9.35999304e-02
1.11047053e+00 1.18983603e+00 4.51608747e-01 1.59166068e-01
-4.56981510e-01 5.47439039e-01 -5.90241671e-01 6.50962070e-02
2.90081680e-01 -6.26925230e-02 -4.25260097e-01 8.58159140e-02
-8.69677305e-01 -3.55566382e-01 -1.71669737e-01 9.31995869e-01
8.18265676e-01 2.60674924e-01 1.41876921e-01 -1.90797496e+00
-6.53110325e-01 1.03483462e+00 -6.12745702e-01 4.41152543e-01
1.83232963e-01 8.25281590e-02 6.64539635e-01 -2.70102233e-01
5.73984623e-01 9.36756194e-01 1.77052155e-01 3.47747535e-01
-9.65911388e-01 -4.72472966e-01 6.16710842e-01 7.06116319e-01
-1.23417950e+00 -1.08428605e-01 8.61039877e-01 1.00844996e-02
1.30705392e+00 4.99684125e-01 8.56355548e-01 6.71404183e-01
4.69701320e-01 4.38404083e-01 8.89854491e-01 -2.22100526e-01
7.00926006e-01 -3.74124646e-02 -2.78496146e-01 3.50573249e-02
-9.93948653e-02 2.49534905e-01 -1.44076543e-02 -1.95800364e-01
3.37032855e-01 3.39911789e-01 7.22975805e-02 -4.31933850e-01
-6.31897628e-01 2.86944717e-01 3.39719623e-01 2.46467143e-01
-8.86014819e-01 1.06809912e-02 4.93806571e-01 5.38670242e-01
5.62553048e-01 2.17096299e-01 -7.49757171e-01 -2.24933311e-01
-4.35310066e-01 -1.97516307e-01 1.26553202e+00 1.05459201e+00
4.30272669e-01 4.76233453e-01 -7.68694133e-02 6.39146864e-01
2.21635312e-01 4.14334565e-01 3.05793852e-01 -1.04463422e+00
-5.52168414e-02 6.23141944e-01 3.38225752e-01 -7.19182789e-01
-6.36718750e-01 -4.52939302e-01 -6.34973705e-01 1.85723752e-02
-4.08344030e-01 -6.43972039e-01 -2.72977203e-01 1.44938684e+00
8.23977649e-01 3.88981998e-01 -1.89383402e-01 1.26441014e+00
3.45722705e-01 8.85315776e-01 9.36732888e-02 -6.49989605e-01
9.01928365e-01 -1.10287142e+00 -6.16433442e-01 3.66978794e-01
5.32359958e-01 -6.26784921e-01 1.07633197e+00 8.36373791e-02
-8.70622814e-01 7.62464851e-02 -5.87643445e-01 7.40726888e-01
-1.77038833e-01 -3.18737179e-01 7.79427767e-01 4.60350603e-01
-1.17584074e+00 2.42094815e-01 -8.29113781e-01 -3.01812112e-01
-2.87905306e-01 6.51979625e-01 2.37520382e-01 1.91077545e-01
-9.02758896e-01 8.40632498e-01 4.06457037e-02 -1.38077989e-01
-1.02101183e+00 -7.96148241e-01 -3.75957519e-01 6.99416041e-01
7.10738242e-01 -6.45051241e-01 1.19232988e+00 -1.01138473e+00
-1.90236640e+00 7.82807395e-02 3.38065207e-01 9.08745825e-02
3.39645028e-01 3.55690479e-01 -7.65541911e-01 -3.40377033e-01
-3.49342972e-01 -7.61582181e-02 2.85431892e-01 -1.23882520e+00
-6.78688526e-01 -2.24938542e-01 4.99852151e-01 3.78867567e-01
-5.37053168e-01 1.73531115e-01 -1.13489635e-01 9.00149420e-02
-3.81737322e-01 -9.11956489e-01 -5.78217208e-01 -9.00816202e-01
-1.27915278e-01 -6.59584284e-01 6.96356297e-01 -3.68203670e-01
1.16253078e+00 -2.23804331e+00 2.64584571e-01 3.36483240e-01
6.87415618e-03 -1.64991975e-01 -3.71050000e-01 8.65598381e-01
2.38560140e-01 -8.34790096e-02 5.44044375e-01 1.70209944e-01
1.86720520e-01 3.84261161e-01 2.09222749e-01 1.02208786e-01
-5.73476017e-01 5.05682111e-01 -8.78427267e-01 -3.15275133e-01
3.34490299e-01 2.15702027e-01 -4.42058533e-01 5.00732541e-01
-3.57516468e-01 4.22520548e-01 -7.09097326e-01 7.36274004e-01
8.74286711e-01 -4.20345008e-01 6.71269357e-01 1.48278490e-01
-2.63004780e-01 -7.90680498e-02 -1.09656644e+00 1.26321781e+00
-1.10172808e+00 -2.93667555e-01 4.52016085e-01 -1.28691077e+00
8.92306268e-01 2.36125976e-01 1.15394723e+00 -1.12395000e+00
2.59681225e-01 -6.70603886e-02 8.44697207e-02 -9.76919472e-01
3.65585387e-02 1.41750470e-01 7.08715543e-02 7.03037918e-01
-3.82998317e-01 2.81189233e-01 1.73525304e-01 5.61480336e-02
1.17728198e+00 1.17706927e-02 -7.49863759e-02 -5.24318278e-01
3.18234682e-01 -2.32845768e-01 9.70364869e-01 2.03164771e-01
-1.62616208e-01 -5.76230943e-01 6.32244647e-01 -7.07501113e-01
-8.54498863e-01 -9.73919690e-01 2.70796388e-01 1.49883139e+00
6.49178386e-01 -1.00754565e-02 -7.20499098e-01 -4.50917631e-01
1.69284716e-01 9.72776234e-01 -2.07147807e-01 -1.42800108e-01
-2.77923971e-01 -8.70580494e-01 -4.54209447e-01 3.60215694e-01
-3.09729930e-02 -1.09435928e+00 -5.84236622e-01 3.87599945e-01
1.50681054e-02 -9.88667965e-01 -7.19628215e-01 -5.25497533e-02
-5.58640242e-01 -8.77877116e-01 -5.44398800e-02 -5.89997411e-01
6.88211858e-01 6.09747887e-01 1.07205367e+00 7.51522034e-02
-2.24173978e-01 5.77370226e-01 -2.83389390e-01 -6.24949625e-03
-8.21662471e-02 9.68855619e-02 1.75961062e-01 -9.65667590e-02
3.73922497e-01 -1.14840722e+00 -1.02131069e+00 5.22071779e-01
-7.93464363e-01 -1.00300595e-01 4.68974680e-01 5.73703349e-01
5.98266780e-01 8.53612944e-02 1.11567163e+00 -6.73466623e-01
9.34876144e-01 -1.04920673e+00 -8.84406686e-01 4.66664702e-01
-1.26102638e+00 -2.42011458e-01 1.23444915e+00 -5.39293945e-01
-1.04105151e+00 -3.46170157e-01 1.40792578e-01 -4.69306111e-01
2.34308422e-01 4.57379580e-01 -2.67869622e-01 -1.05873026e-01
2.04388529e-01 3.19350660e-01 -1.16323560e-01 -3.10270637e-01
3.15640330e-01 8.53077233e-01 -3.41554701e-01 -6.43115819e-01
1.93445072e-01 1.40700847e-01 -1.94684654e-01 -1.99889112e-02
-2.09351704e-01 -2.58681089e-01 4.87093836e-01 -4.91927356e-01
3.13395232e-01 -5.70228457e-01 -1.63795972e+00 -2.11094514e-01
-7.23839402e-01 -2.35627800e-01 -6.93266690e-02 3.94253016e-01
-6.28161132e-01 7.17096329e-02 -7.00153291e-01 -8.53500783e-01
-6.13470912e-01 -1.21243250e+00 5.13158500e-01 4.25840706e-01
3.96092176e-01 -7.40459263e-01 3.83652076e-02 9.33054239e-02
9.57560837e-01 -2.34439205e-02 8.86645257e-01 -6.33804917e-01
-7.97144234e-01 -8.65337253e-02 -1.57889023e-01 1.44108191e-01
4.51542549e-02 1.16834067e-01 -9.28509906e-02 -6.96995795e-01
-1.18400767e-01 9.04384851e-02 -2.89200544e-01 4.76737857e-01
1.43096721e+00 -5.81190944e-01 -2.23607615e-01 6.17209077e-01
1.71964765e+00 7.49072731e-01 2.89698660e-01 4.96926874e-01
3.44653815e-01 6.61274910e-01 4.60264325e-01 1.04160297e+00
9.77118552e-01 5.36085844e-01 9.69376445e-01 6.24030493e-02
3.33069444e-01 3.41750570e-02 3.38138461e-01 1.36384094e+00
-2.14816630e-01 -2.82681257e-01 -5.74435413e-01 3.23003262e-01
-2.14255500e+00 -5.87644458e-01 1.62367791e-01 2.04124331e+00
3.14539105e-01 -2.42328599e-01 2.24015951e-01 -4.91216630e-01
7.08259702e-01 -1.73068061e-01 -1.04116106e+00 -8.41234803e-01
2.33535782e-01 -4.06899542e-01 2.40887567e-01 1.17361523e-01
-1.23250701e-01 7.10910559e-01 5.82873726e+00 5.93100011e-01
-1.38252437e+00 4.72666502e-01 3.97493869e-01 -4.29360151e-01
-6.62943780e-01 7.66146258e-02 -1.64031014e-01 1.08918142e+00
1.15346408e+00 -9.50002372e-01 1.38679409e+00 1.33969629e+00
7.77559459e-01 1.59672067e-01 -8.08805406e-01 7.15629399e-01
-4.46484387e-01 -1.17455661e+00 -5.45548558e-01 6.99357539e-02
9.55815732e-01 3.89248729e-01 -4.34569538e-01 3.92672211e-01
6.99362278e-01 -2.73272306e-01 3.65313262e-01 5.67950964e-01
2.66944170e-01 -1.10170281e+00 8.09441924e-01 4.96526420e-01
-9.44084644e-01 -5.61786890e-01 -4.68066037e-01 -1.20425120e-01
3.44536453e-01 8.46499920e-01 -4.96987820e-01 7.29844928e-01
9.60242271e-01 -8.28262791e-02 2.95245111e-01 1.14021957e+00
1.87812105e-01 2.47242391e-01 -2.78816849e-01 -4.97685999e-01
-8.98244008e-02 -6.20600402e-01 4.02808964e-01 7.81369507e-01
4.76483107e-01 9.28408206e-02 7.24486470e-01 7.64033020e-01
-1.29329830e-01 6.14877760e-01 -2.42871225e-01 -6.58597127e-02
8.42565238e-01 1.70859635e+00 -6.80398464e-01 -1.89339533e-01
-5.09426117e-01 5.55953026e-01 2.82283634e-01 6.65873945e-01
-8.71216536e-01 -1.37642369e-01 7.73493350e-01 1.34909842e-02
6.18404523e-02 -1.36852443e-01 -1.34893283e-01 -9.82856154e-01
1.26283213e-01 -5.62659502e-01 1.68606058e-01 -6.14893079e-01
-1.26652682e+00 6.29216015e-01 -4.40779507e-01 -1.08469176e+00
-8.99092034e-02 5.42205125e-02 -8.63319993e-01 5.73678017e-01
-1.44058323e+00 -7.76689291e-01 -2.83334225e-01 3.64207238e-01
4.85487372e-01 -3.74714984e-03 6.93161488e-01 5.77226043e-01
-8.01534414e-01 1.85775012e-01 5.06322742e-01 -8.57205510e-01
2.87897050e-01 -9.87663507e-01 -4.38193716e-02 5.18899977e-01
-5.35367668e-01 2.67130166e-01 7.98839271e-01 -3.05121958e-01
-1.93834603e+00 -7.32773244e-01 3.12064081e-01 3.20610374e-01
7.59526670e-01 -1.52214900e-01 -5.14148235e-01 3.83740872e-01
4.53864247e-01 1.25752628e-01 7.76667774e-01 1.54653370e-01
7.09252134e-02 -7.61871457e-01 -1.26324117e+00 4.27629381e-01
1.05810034e+00 4.29394878e-02 2.29468063e-01 7.08432257e-01
1.02087379e+00 -1.12648673e-01 -9.62001979e-01 3.56870443e-01
3.29565495e-01 -5.91217935e-01 4.86640930e-01 -8.63783062e-01
1.48451701e-01 -2.18600258e-01 -1.22000925e-01 -1.85003638e+00
-6.08274698e-01 -4.95616287e-01 -3.93668324e-01 1.12958944e+00
1.11201242e-01 -8.18817675e-01 5.34259617e-01 7.29115427e-01
-3.78518939e-01 -1.02266514e+00 -9.76230443e-01 -1.02998400e+00
-4.26569670e-01 9.10268351e-02 1.33012462e+00 9.85144138e-01
2.93515384e-01 8.95775482e-02 -4.45658356e-01 2.07038432e-01
4.94554609e-01 6.65785432e-01 4.01654929e-01 -9.11119342e-01
-4.83282030e-01 -2.75794983e-01 6.07165657e-02 -5.37830651e-01
3.47704768e-01 -7.38517284e-01 -7.19418898e-02 -1.76057124e+00
2.01784268e-01 -6.81604445e-01 -8.65460753e-01 2.13540792e-01
2.12941796e-01 -7.45998144e-01 2.06670448e-01 1.38257161e-01
-1.11179292e+00 6.33490682e-01 1.20979655e+00 -8.56258944e-02
-3.75166863e-01 2.97494471e-01 -6.18613958e-01 1.79716438e-01
7.44668484e-01 -4.97315556e-01 -5.97230673e-01 -6.83270037e-01
3.59577477e-01 7.35210955e-01 -3.66796106e-01 -2.68913090e-01
3.83644283e-01 -9.31223094e-01 -3.06421459e-01 -3.46465670e-02
-3.88155460e-01 -1.15328455e+00 7.33216524e-01 5.94735205e-01
5.37983552e-02 5.73722422e-01 -2.28474155e-01 4.64839280e-01
2.18479261e-01 5.01287170e-02 3.77486408e-01 7.47126788e-02
-6.86008155e-01 6.18050933e-01 -4.52843219e-01 -3.86854768e-01
1.48386633e+00 2.65828758e-01 -1.99248388e-01 -3.10830474e-01
-7.98829317e-01 9.40756619e-01 4.41706806e-01 7.54400671e-01
3.07365358e-01 -1.11053538e+00 -3.94519806e-01 -3.28417748e-01
-2.28481337e-01 -5.13334751e-01 7.86704063e-01 8.88805926e-01
-4.16262239e-01 -4.57685255e-02 -3.63428444e-01 -3.15829128e-01
-7.59819269e-01 1.04879832e+00 3.87453586e-01 -1.18828036e-01
-4.18134987e-01 3.70790839e-01 -9.87421423e-02 -6.26221597e-01
2.35310718e-01 1.65515304e-01 6.47663698e-02 -2.63230473e-01
1.96219906e-01 8.51542294e-01 2.33742326e-01 2.60430366e-01
-4.46716309e-01 -1.01311266e-01 1.95970312e-01 3.12874436e-01
1.64611399e+00 -4.56512690e-01 -5.61856449e-01 -8.47352743e-02
6.73170507e-01 -2.11965501e-01 -1.16629744e+00 1.20345779e-01
5.91161065e-02 -5.64263940e-01 2.06395522e-01 -9.53885555e-01
-1.58618605e+00 1.72221690e-01 7.24161327e-01 8.89266253e-01
1.40291762e+00 -9.27050505e-03 9.26444769e-01 3.14432755e-02
8.00963640e-01 -1.33112633e+00 -3.36516321e-01 -1.02372929e-01
5.70342541e-01 -7.56542742e-01 -4.70709175e-01 -2.70469904e-01
-4.78900790e-01 8.28822017e-01 1.16740119e+00 -1.09813236e-01
7.74200320e-01 7.79241502e-01 -7.27102533e-02 -7.63764009e-02
-1.18464053e+00 5.16380519e-02 -6.17095709e-01 5.05768418e-01
1.59593467e-02 7.06648469e-01 -9.10644174e-01 6.78835213e-01
2.71129310e-01 1.55664727e-01 4.53897387e-01 6.99833453e-01
-2.84247100e-01 -1.24387228e+00 8.38540569e-02 4.72475737e-01
-2.45266154e-01 2.09195405e-01 1.93309903e-01 1.24536589e-01
6.12805448e-02 1.07968271e+00 1.11903444e-01 -5.32962859e-01
4.98471290e-01 -3.83465707e-01 1.62809789e-01 -1.88937128e-01
-5.51414549e-01 6.06387742e-02 -1.21343270e-01 -6.04898870e-01
2.37627216e-02 -2.65031368e-01 -1.42612231e+00 -7.82232523e-01
-3.80743027e-01 5.23528636e-01 1.08719790e+00 8.08783174e-01
1.09808826e+00 1.00444329e+00 1.58183551e+00 -4.43763882e-01
-8.59452963e-01 -8.24006855e-01 -7.25437760e-01 1.86052948e-01
-3.98583114e-01 -5.46851277e-01 -4.36556727e-01 -6.30613625e-01] | [5.876976013183594, 1.7867498397827148] |
faf8c0a9-2495-4555-9838-f6295f589e56 | clickbait-detection-in-tweets-using-self | 1710.05364 | null | http://arxiv.org/abs/1710.05364v1 | http://arxiv.org/pdf/1710.05364v1.pdf | Clickbait Detection in Tweets Using Self-attentive Network | Clickbait detection in tweets remains an elusive challenge. In this paper, we
describe the solution for the Zingel Clickbait Detector at the Clickbait
Challenge 2017, which is capable of evaluating each tweet's level of click
baiting. We first reformat the regression problem as a multi-classification
problem, based on the annotation scheme. To perform multi-classification, we
apply a token-level, self-attentive mechanism on the hidden states of
bi-directional Gated Recurrent Units (biGRU), which enables the model to
generate tweets' task-specific vector representations by attending to important
tokens. The self-attentive neural network can be trained end-to-end, without
involving any manual feature engineering. Our detector ranked first in the
final evaluation of Clickbait Challenge 2017. | ['Yiwei Zhou'] | 2017-10-15 | null | null | null | null | ['clickbait-detection'] | ['natural-language-processing'] | [ 1.55482953e-02 -1.88030750e-01 -4.11668390e-01 -6.20596230e-01
-1.53786099e+00 -4.28791195e-01 7.98009336e-01 4.00219336e-02
-4.44620967e-01 6.07235730e-01 2.17929170e-01 -5.31810403e-01
9.28748325e-02 -5.55800736e-01 -6.69717669e-01 -3.79532278e-01
1.98451877e-02 3.87778372e-01 4.29791063e-01 -2.28127033e-01
2.53086746e-01 -4.22377288e-01 -1.47493303e+00 7.61213362e-01
5.49336612e-01 1.23019266e+00 -8.98348391e-02 1.22264826e+00
1.29446477e-01 1.04934037e+00 -6.75897121e-01 -2.75680184e-01
-5.69657683e-02 -2.13331968e-01 -7.02005625e-01 -4.90989268e-01
8.26494813e-01 -3.25751871e-01 -5.26490092e-01 8.02335620e-01
3.63253832e-01 -1.98474646e-01 5.68031251e-01 -1.06257963e+00
-6.03593767e-01 8.89219880e-01 -2.88999975e-01 8.85036111e-01
-1.64737374e-01 5.13458960e-02 1.81188500e+00 -8.68186653e-01
4.09592241e-01 9.28141475e-01 4.77238923e-01 5.80568254e-01
-1.40465963e+00 -8.07882428e-01 1.53038114e-01 3.35013837e-01
-1.16342890e+00 -1.59455553e-01 6.97470844e-01 -7.84327030e-01
6.71447575e-01 6.90709829e-01 4.55889285e-01 1.63920772e+00
4.08803448e-02 1.66273606e+00 1.05571067e+00 2.53641475e-02
9.77524966e-02 1.23680726e-01 8.80885720e-01 4.83746290e-01
5.54443598e-02 4.71452028e-02 -4.24400479e-01 -3.19460213e-01
2.04228505e-01 5.42743281e-02 3.18207234e-01 2.78945923e-01
-9.15983081e-01 1.28279090e+00 7.71839678e-01 5.28579593e-01
-4.13235843e-01 7.68822849e-01 3.26198667e-01 1.29720613e-01
1.06354070e+00 3.94015133e-01 -4.02336299e-01 -2.69900411e-02
-1.27307963e+00 3.92672271e-01 5.33629537e-01 6.33933246e-01
8.86583984e-01 -3.29751931e-02 -9.17355418e-01 5.56335747e-01
4.22633648e-01 2.17204660e-01 8.30606520e-01 -5.47945425e-02
5.95530748e-01 4.95411128e-01 1.60812303e-01 -6.19960368e-01
-2.83648401e-01 -7.55348980e-01 -4.44513589e-01 -3.69275898e-01
1.50888488e-01 -2.94808447e-01 -1.15600932e+00 1.39522755e+00
7.91966170e-02 2.24173799e-01 -4.51290667e-01 5.66598475e-01
6.24832511e-01 6.59807503e-01 5.38015485e-01 3.90351355e-01
1.72328615e+00 -1.12463701e+00 -8.28837991e-01 -4.42140430e-01
7.83047020e-01 -4.39046234e-01 9.86687124e-01 -9.39150229e-02
-8.76701653e-01 -4.53043342e-01 -9.80525911e-01 -3.06958616e-01
-9.71163273e-01 1.08115755e-01 5.04079342e-01 5.16605556e-01
-6.04480982e-01 2.62784213e-01 -5.95927179e-01 1.40597761e-01
4.75250423e-01 4.09102067e-02 4.74265814e-01 5.15230834e-01
-1.65436864e+00 8.52908254e-01 4.00025249e-01 1.51233137e-01
-1.11704671e+00 -8.14932525e-01 -4.39210266e-01 3.40345025e-01
2.08454266e-01 -4.42580163e-01 1.83768177e+00 -6.55218601e-01
-1.09950972e+00 7.66488850e-01 -2.15066135e-01 -9.28102136e-01
6.93136871e-01 -3.06141287e-01 -4.29602385e-01 -2.49633968e-01
3.66736263e-01 3.29744309e-01 1.09270918e+00 -9.20847118e-01
-1.09351468e+00 -1.70867547e-01 -3.37788761e-01 -2.69897938e-01
-1.11638688e-01 6.93027228e-02 -2.85810888e-01 -5.47915995e-01
-2.61025995e-01 -7.73403943e-01 -3.27592731e-01 -9.98189449e-01
-1.00709903e+00 -8.18512917e-01 7.57002652e-01 -8.36279929e-01
1.76612997e+00 -2.11993480e+00 -4.07060921e-01 3.53542686e-01
7.49178290e-01 3.57971668e-01 -1.34548187e-01 2.56761551e-01
5.64670637e-02 6.20276809e-01 4.91010457e-01 -3.98498476e-01
4.03391302e-01 -3.66374254e-01 -6.15648568e-01 1.79499984e-01
3.25333744e-01 1.44289458e+00 -8.97959650e-01 -3.24164510e-01
-1.67920254e-02 3.17939013e-01 -7.27362335e-01 2.53302187e-01
-6.56830013e-01 6.42656907e-02 -1.10020339e+00 4.76440161e-01
4.07120585e-01 -5.70073843e-01 -2.27906778e-01 -1.83402479e-01
-5.30633152e-01 9.28747535e-01 -5.16579807e-01 7.66709447e-01
-6.22459292e-01 8.73524070e-01 -2.32671425e-01 -6.59570336e-01
7.36548543e-01 2.32841626e-01 2.11808249e-01 -9.43486929e-01
2.83326477e-01 2.97049969e-01 -1.87441260e-01 -5.81175208e-01
4.90303308e-01 2.98521549e-01 -5.48963010e-01 7.49209702e-01
-1.26977906e-01 3.40135157e-01 2.70181447e-01 3.60983431e-01
1.30098748e+00 -3.88466209e-01 -7.50157461e-02 -3.01559716e-02
2.07710966e-01 -8.28929693e-02 -3.49483676e-02 1.33981824e+00
-1.75468281e-01 5.33045828e-01 4.40426350e-01 -6.08073175e-01
-1.18107963e+00 -7.93251574e-01 2.60681435e-02 1.99291253e+00
-5.45030415e-01 -4.47475553e-01 -5.41875482e-01 -7.47515917e-01
6.37662113e-02 8.90649796e-01 -1.04415393e+00 -1.45173147e-01
-5.82820833e-01 -9.01681960e-01 8.47418606e-01 2.12802038e-01
6.19337678e-01 -8.79357815e-01 -4.10947382e-01 5.55848956e-01
-2.34069079e-01 -9.33170199e-01 -5.49132645e-01 4.45942909e-01
-9.43520963e-02 -7.86820292e-01 -5.91484308e-01 -8.22912216e-01
1.09562270e-01 6.60424726e-03 8.18136871e-01 9.24889930e-03
-4.88688707e-01 9.43693891e-02 -7.09735900e-02 -4.01241332e-01
-2.97450960e-01 8.70348573e-01 -5.41027665e-01 5.03547549e-01
6.14786327e-01 -9.35248509e-02 -7.86773562e-01 3.83755565e-01
-8.90155017e-01 3.14479955e-02 5.03586471e-01 8.22212517e-01
8.97347182e-02 -4.07562912e-01 7.49430180e-01 -9.37512577e-01
1.20156598e+00 -7.40254700e-01 -5.24964869e-01 2.97372162e-01
-2.68718362e-01 2.50924140e-01 2.15145648e-01 -5.64544022e-01
-7.35152721e-01 -5.06510697e-02 -2.99447536e-01 2.67727703e-01
1.77160159e-01 7.15703249e-01 5.49035370e-01 2.66670138e-01
7.03348875e-01 4.40481097e-01 -5.25324464e-01 -5.77214539e-01
5.03934383e-01 9.96346951e-01 1.18699752e-01 -1.16507187e-02
8.80031168e-01 1.48002189e-02 -6.85911715e-01 -7.57319152e-01
-1.69220304e+00 -7.21835971e-01 -2.86016494e-01 -2.63707906e-01
1.03833604e+00 -8.59459043e-01 -1.33263183e+00 2.10516497e-01
-1.05662143e+00 -3.20133567e-01 1.65545166e-01 2.45342001e-01
-1.54353753e-01 -2.23956555e-01 -7.75632858e-01 -9.14119124e-01
-6.26882374e-01 -8.33700478e-01 1.00829315e+00 1.41579658e-01
-1.64956912e-01 -9.88757670e-01 3.36906582e-01 6.40777171e-01
1.01163018e+00 -7.74423853e-02 7.69087553e-01 -1.52951348e+00
-8.04118574e-01 -8.11669290e-01 -4.25863177e-01 3.95381212e-01
-3.13566357e-01 -9.14323181e-02 -1.24887764e+00 -4.39374242e-03
-1.36787370e-01 -6.37679636e-01 1.38778555e+00 1.77123651e-01
1.20640135e+00 -7.86761642e-01 -4.09585029e-01 4.81922105e-02
1.03259957e+00 -4.06268597e-01 3.72153372e-01 4.08196270e-01
5.92841268e-01 1.97690159e-01 4.07090068e-01 3.66706342e-01
5.96100032e-01 6.08206630e-01 4.47015196e-01 -1.70330077e-01
5.73750250e-02 -7.65357375e-01 2.27527916e-01 5.05120039e-01
3.23438227e-01 -1.79122850e-01 -8.92992496e-01 3.63228410e-01
-1.93201721e+00 -1.19373322e+00 -2.90750682e-01 1.79621625e+00
8.76203418e-01 3.88946801e-01 3.13809097e-01 -6.09130263e-02
8.38621855e-01 4.10822541e-01 -4.45629269e-01 -2.24991724e-01
2.47969285e-01 1.84733585e-01 7.65336573e-01 6.90835476e-01
-1.43716300e+00 1.26866734e+00 6.55982494e+00 9.86985445e-01
-1.21117723e+00 4.10066098e-01 7.27589130e-01 -1.62767574e-01
-1.31487936e-01 -2.70655036e-01 -1.40796995e+00 8.12401712e-01
1.22885275e+00 -2.93641210e-01 2.65086651e-01 1.08129752e+00
3.81940305e-01 2.14505762e-01 -9.93756294e-01 5.75964212e-01
-7.99592659e-02 -1.62951660e+00 -8.92434642e-02 5.52651249e-02
3.83901477e-01 7.21647084e-01 5.41240454e-01 1.12306237e+00
6.71290755e-01 -8.06883276e-01 7.30799317e-01 2.54901767e-01
1.11023247e-01 -1.58462226e-01 4.06401008e-01 4.75019097e-01
-9.53568876e-01 -2.86763400e-01 -4.25459296e-02 2.16373518e-01
2.03293428e-01 5.52713275e-01 -1.47969222e+00 -2.05486715e-01
3.50987762e-01 5.14399111e-01 -1.14713788e+00 1.15460670e+00
-2.25989550e-01 1.16427004e+00 -2.97041774e-01 -9.28688347e-01
9.04741883e-01 3.62130255e-01 2.80780435e-01 1.60347247e+00
-9.49611291e-02 -5.57058930e-01 1.77877530e-01 1.00969923e+00
-2.23528042e-01 -1.29443318e-01 -1.60063595e-01 -3.38944882e-01
1.83218688e-01 1.35756493e+00 -2.30184153e-01 -9.05748427e-01
7.25471321e-03 7.38137007e-01 5.38038015e-01 5.96642613e-01
-1.23115885e+00 -3.71030301e-01 3.26460630e-01 1.57791495e-01
6.10189080e-01 6.19719317e-03 -1.09730624e-01 -1.30091166e+00
-1.48363143e-01 -4.47198540e-01 5.34907937e-01 -6.52030408e-01
-1.18800771e+00 6.64198875e-01 -2.54401952e-01 -8.47574472e-01
-2.84258336e-01 -3.69316787e-01 -6.81173027e-01 7.06647396e-01
-1.73878467e+00 -1.36938083e+00 -8.85734856e-02 3.64203215e-01
6.97210252e-01 1.59954652e-01 4.40805435e-01 7.59459913e-01
-9.00318146e-01 6.32740021e-01 1.15009181e-01 4.90893960e-01
3.65853846e-01 -1.17232728e+00 7.42806077e-01 3.55991483e-01
3.67600508e-02 6.24285519e-01 8.56361091e-01 -4.87195015e-01
-1.00273192e+00 -1.42019248e+00 1.62635541e+00 -6.06849432e-01
1.51712537e+00 -9.93102491e-01 -7.36185551e-01 1.13170278e+00
1.12283185e-01 -5.52047864e-02 7.62978137e-01 4.76689577e-01
-5.90613008e-01 3.50320749e-02 -6.91100776e-01 5.19622266e-01
3.51331413e-01 -7.91518152e-01 -7.26877868e-01 7.97150195e-01
5.59534132e-01 5.12476675e-02 -4.72490907e-01 -5.56367487e-02
5.22000372e-01 -5.45764506e-01 8.94044638e-01 -1.30345070e+00
1.81860209e-01 1.48067892e-01 3.24367769e-02 -1.14548159e+00
-6.92427576e-01 -5.20918131e-01 9.19329375e-02 1.22271919e+00
1.04841352e+00 -6.20953560e-01 9.21017349e-01 3.99263740e-01
-3.10060177e-02 -4.33446795e-01 -1.29294193e+00 -5.27558565e-01
-2.18373984e-01 -4.76742268e-01 2.74904668e-01 5.35254836e-01
9.19102505e-03 1.07384908e+00 -7.91851342e-01 -1.01787105e-01
5.13540328e-01 -4.49306443e-02 6.71680450e-01 -1.13808787e+00
-1.62859231e-01 -6.46392941e-01 -5.10751270e-02 -1.47335625e+00
5.70157841e-02 -1.25010538e+00 2.79524475e-01 -1.20763779e+00
9.27564949e-02 -3.73981893e-01 -5.51726580e-01 4.97934788e-01
-2.41995007e-01 6.00729108e-01 8.80888179e-02 1.48651123e-01
-1.29369605e+00 5.22913456e-01 6.84580088e-01 -6.21506810e-01
-1.20792896e-01 5.11138558e-01 -6.11894965e-01 4.05579895e-01
8.48281384e-01 -7.78683960e-01 3.96763563e-01 -3.26985359e-01
5.63524544e-01 -3.74450922e-01 5.46002567e-01 -5.74427545e-01
2.52654046e-01 2.52386898e-01 9.98296663e-02 -8.90371442e-01
4.52681482e-01 -3.82992029e-01 -3.26264679e-01 7.09671974e-01
-1.13589489e+00 -1.97008997e-01 -2.34810337e-01 6.13130510e-01
-1.59574404e-01 -2.25883991e-01 6.71295047e-01 -9.26499590e-02
-2.12713763e-01 2.75549650e-01 -8.98120940e-01 4.66115959e-02
7.30198920e-01 3.01938891e-01 -6.44058406e-01 -2.96083987e-01
-8.48940551e-01 4.70154822e-01 -7.92136550e-01 6.17746174e-01
3.00927818e-01 -1.42403686e+00 -8.69237363e-01 2.87998974e-01
3.18210244e-01 -5.96064091e-01 2.02616841e-01 6.50546253e-01
-1.46049887e-01 7.79037178e-01 3.72393340e-01 -5.29265165e-01
-9.02509212e-01 3.69558454e-01 2.99139559e-01 -7.60351121e-01
-1.42790660e-01 7.90191650e-01 -1.56393617e-01 -3.36938836e-02
2.62492985e-01 -4.40221995e-01 -1.05085582e-01 6.16760969e-01
6.83103740e-01 2.07305521e-01 2.09671617e-01 -1.86307162e-01
-1.20952606e-01 -3.00781056e-02 -7.04468906e-01 -1.28850341e-01
1.34554994e+00 1.09868482e-01 3.36239815e-01 5.24546325e-01
1.57870114e+00 -3.11574608e-01 -7.77267933e-01 -2.47836739e-01
4.87604856e-01 -2.21865803e-01 3.41190815e-01 -8.05710852e-01
-5.98547161e-01 8.16825271e-01 3.34649116e-01 1.01975977e+00
1.00978665e-01 3.24039578e-01 9.94700491e-01 6.58878624e-01
-3.05221796e-01 -1.23156095e+00 4.16749954e-01 7.29247570e-01
8.36347759e-01 -1.31764805e+00 -4.87638474e-01 1.58323824e-01
-5.80196559e-01 7.45142460e-01 4.52248693e-01 -3.57092917e-01
9.13650036e-01 -2.74120748e-01 -1.24959107e-02 -3.78179938e-01
-1.18418384e+00 -3.18296164e-01 3.98848206e-01 -1.53639376e-01
3.86999786e-01 1.18047655e-01 -2.82966673e-01 4.59321648e-01
-2.82701343e-01 -1.93879560e-01 1.65079132e-01 6.82900071e-01
-8.13410521e-01 -4.93783206e-01 8.73747244e-02 8.12532008e-01
-6.13612771e-01 -4.47568119e-01 -3.06864798e-01 5.23783743e-01
-4.00429755e-01 7.63343930e-01 1.50774017e-01 -5.28406560e-01
2.48565599e-01 3.93152952e-01 -3.90545011e-01 -5.88092089e-01
-1.07331491e+00 2.27621883e-01 2.65587270e-01 -3.12852800e-01
1.69511765e-01 -4.12111133e-01 -5.87227821e-01 4.96532805e-02
-5.26390553e-01 5.32976747e-01 1.06036723e+00 7.90317237e-01
4.56215143e-01 5.44757128e-01 9.91393030e-01 -7.59290874e-01
-9.61979806e-01 -1.41699421e+00 -3.10785323e-01 4.53517199e-01
8.92532051e-01 -2.46028140e-01 -8.24411094e-01 -3.51782739e-01] | [7.750486373901367, 9.783860206604004] |
d5070f32-8c86-42cd-91ed-76077808ba68 | multi-behavior-graph-neural-networks-for | 2302.08678 | null | https://arxiv.org/abs/2302.08678v1 | https://arxiv.org/pdf/2302.08678v1.pdf | Multi-Behavior Graph Neural Networks for Recommender System | Recommender systems have been demonstrated to be effective to meet user's personalized interests for many online services (e.g., E-commerce and online advertising platforms). Recent years have witnessed the emerging success of many deep learning-based recommendation models for augmenting collaborative filtering architectures with various neural network architectures, such as multi-layer perceptron and autoencoder. However, the majority of them model the user-item relationship with single type of interaction, while overlooking the diversity of user behaviors on interacting with items, which can be click, add-to-cart, tag-as-favorite and purchase. Such various types of interaction behaviors have great potential in providing rich information for understanding the user preferences. In this paper, we pay special attention on user-item relationships with the exploration of multi-typed user behaviors. Technically, we contribute a new multi-behavior graph neural network (MBRec), which specially accounts for diverse interaction patterns as well as the underlying cross-type behavior inter-dependencies. In the MBRec framework, we develop a graph-structured learning framework to perform expressive modeling of high-order connectivity in behavior-aware user-item interaction graph. After that, a mutual relation encoder is proposed to adaptively uncover complex relational structures and make aggregations across layer-specific behavior representations. Through comprehensive evaluation on real-world datasets, the advantages of our MBRec method have been validated under different experimental settings. Further analysis verifies the positive effects of incorporating the multi-behavioral context into the recommendation paradigm. Additionally, the conducted case studies offer insights into the interpretability of user multi-behavior representations. | ['Liefeng Bo', 'Peng Dai', 'Yong Xu', 'Chao Huang', 'Lianghao Xia'] | 2023-02-17 | null | null | null | null | ['collaborative-filtering'] | ['miscellaneous'] | [-1.36285514e-01 -6.98393807e-02 -5.46681404e-01 -7.72936821e-01
2.14524329e-01 -2.62148470e-01 2.36485884e-01 9.94050279e-02
-3.55237499e-02 1.62742868e-01 4.56071019e-01 -4.99372751e-01
-8.13399911e-01 -1.01185751e+00 -6.20613337e-01 -2.94226915e-01
-6.42563760e-01 2.85787165e-01 -1.23363249e-01 -6.19092822e-01
-1.74058795e-01 2.89871931e-01 -1.39110827e+00 6.37619793e-01
8.81694734e-01 1.16135859e+00 6.32293448e-02 2.63577163e-01
-1.51904866e-01 6.21113539e-01 -1.49957076e-01 -7.91845024e-01
6.00527972e-02 -1.23683013e-01 -3.50423157e-01 1.47514105e-01
1.76852688e-01 -4.43706155e-01 -6.24694049e-01 9.15288448e-01
5.34447432e-02 4.75937724e-01 4.82144117e-01 -1.10482740e+00
-1.37415004e+00 1.24331188e+00 -4.51923251e-01 2.62844294e-01
4.29754019e-01 -2.24941269e-01 1.58206725e+00 -5.79141080e-01
3.24987322e-01 1.17680156e+00 5.88836133e-01 2.13929415e-01
-1.09059262e+00 -6.02359593e-01 8.49405885e-01 4.04266447e-01
-1.06822979e+00 2.02514604e-01 1.08454013e+00 -3.08070928e-01
9.66293097e-01 3.24973404e-01 8.17439914e-01 1.45996809e+00
2.99583167e-01 1.03944576e+00 6.17743254e-01 8.26902092e-02
-1.94962665e-01 2.28634402e-01 7.45399773e-01 6.44869804e-01
3.50349158e-01 3.68909538e-02 -1.80728212e-01 -1.46457389e-01
7.56662428e-01 6.95019066e-01 -5.18274270e-02 -1.76164731e-01
-7.22443461e-01 9.41497147e-01 9.39761877e-01 4.14390206e-01
-4.65717435e-01 -1.26890510e-01 2.79201061e-01 4.83400583e-01
4.56753016e-01 3.99133503e-01 -5.64137459e-01 3.54508698e-01
-2.43430227e-01 -1.22025730e-02 9.12345827e-01 1.01926982e+00
6.70529246e-01 2.13936925e-01 -1.46832705e-01 1.08388197e+00
6.71669602e-01 3.00866757e-02 7.12986052e-01 -1.99794546e-01
3.65770131e-01 1.06246042e+00 -2.09911928e-01 -1.49996662e+00
-7.52283275e-01 -9.45097387e-01 -1.17730868e+00 -6.14858568e-01
-5.22888415e-02 -2.31761903e-01 -5.15385270e-01 1.72095144e+00
1.99964598e-01 2.33860329e-01 -2.02846333e-01 8.88457239e-01
9.67413604e-01 4.73742157e-01 1.82408541e-01 -1.32807717e-01
1.26336849e+00 -9.01721478e-01 -7.11043119e-01 -3.53980251e-02
6.02047324e-01 -1.02345124e-01 1.12602234e+00 3.44308645e-01
-6.95450902e-01 -7.70024955e-01 -8.50111902e-01 1.14979886e-01
-4.79497254e-01 7.26988018e-02 1.38624465e+00 5.84721863e-01
-7.29364634e-01 6.66839063e-01 -6.73362553e-01 -2.41271943e-01
3.35554421e-01 7.09867299e-01 -1.66528344e-01 4.71644215e-02
-1.61038363e+00 4.03960794e-01 2.11740822e-01 4.07473624e-01
-4.44084764e-01 -7.05833673e-01 -6.98389113e-01 4.29660320e-01
5.83491683e-01 -5.79677999e-01 8.48972917e-01 -1.11140096e+00
-1.48974514e+00 1.43353373e-01 2.10846648e-01 -4.02792305e-01
8.29923376e-02 -2.34508008e-01 -1.27629578e+00 -5.54202020e-01
-4.50464815e-01 -7.32859150e-02 3.56505066e-01 -9.99393940e-01
-5.27777195e-01 -5.06038725e-01 6.00838840e-01 2.34792203e-01
-9.06551242e-01 -1.40486017e-01 -5.54917395e-01 -6.58654809e-01
-1.58572271e-01 -8.24807584e-01 -3.09707284e-01 -5.34017980e-01
-4.48916674e-01 -5.45446992e-01 5.51628172e-01 -3.57343763e-01
1.63454270e+00 -2.18673873e+00 2.25951076e-01 4.53173488e-01
4.05016333e-01 6.26468137e-02 -1.75985590e-01 6.14616692e-01
8.98205414e-02 7.84832910e-02 4.21578735e-01 -1.98879868e-01
1.86159804e-01 2.89606005e-01 -1.19650841e-01 1.20459311e-01
-1.18560143e-01 1.19414139e+00 -6.80589855e-01 -3.07269278e-03
4.26112078e-02 5.74893415e-01 -9.21095431e-01 3.73556942e-01
-3.51866186e-01 3.20814490e-01 -9.30555522e-01 4.23953861e-01
4.35967416e-01 -7.98903048e-01 5.82914174e-01 -6.18873835e-01
4.15781379e-01 1.78121552e-01 -1.06872427e+00 1.37738824e+00
-6.74954712e-01 4.91563268e-02 -8.41416325e-03 -9.32286263e-01
8.86913300e-01 -8.56122598e-02 6.28018260e-01 -8.23030829e-01
2.53400534e-01 -2.24262103e-01 4.60708797e-01 -5.88001132e-01
6.38604462e-01 2.61499524e-01 -9.81656015e-02 4.71690446e-01
2.37482682e-01 9.32408333e-01 1.54103085e-01 2.23335564e-01
9.09204006e-01 -6.39619231e-02 1.84502169e-01 -1.43752605e-01
5.25007963e-01 -5.42649209e-01 4.47879076e-01 6.12746775e-01
1.10324405e-01 3.22201066e-02 2.24208191e-01 -5.57959735e-01
-4.08109784e-01 -7.37466097e-01 -3.33061479e-02 1.65648365e+00
3.49992126e-01 -4.75682974e-01 -1.41570404e-01 -7.24424958e-01
2.39935398e-01 7.04922497e-01 -7.57730544e-01 -5.48027635e-01
-4.59563941e-01 -9.88159180e-01 1.58853337e-01 6.09303355e-01
3.60845745e-01 -1.16160798e+00 2.37284362e-01 3.12553912e-01
1.86082572e-01 -1.09713662e+00 -5.14337420e-01 9.36593637e-02
-8.56679857e-01 -9.46858644e-01 -1.47233278e-01 -7.68035889e-01
4.52731967e-01 3.45259726e-01 1.04431975e+00 1.44594088e-01
9.76275131e-02 5.55761695e-01 -5.56299448e-01 8.14453810e-02
-8.07191506e-02 1.54317901e-01 1.96994871e-01 7.22404718e-01
6.56983614e-01 -8.58090639e-01 -8.52817774e-01 5.36663830e-01
-9.15783763e-01 -1.39618054e-01 6.94027841e-01 7.22547531e-01
3.73036414e-01 1.78034112e-01 7.35705733e-01 -1.55875206e+00
1.19778514e+00 -1.20292437e+00 -2.22033978e-01 3.66494954e-01
-9.17826593e-01 -1.99268535e-01 1.12602854e+00 -8.37561488e-01
-1.15064013e+00 -3.85214746e-01 -2.24046767e-01 -3.71394962e-01
-1.29218414e-01 1.09881258e+00 -3.08488667e-01 1.27846628e-01
4.37996924e-01 2.04521254e-01 -4.04254794e-01 -6.54082775e-01
6.11324251e-01 6.51886523e-01 1.01222396e-01 -4.43028837e-01
3.63920718e-01 1.26146287e-01 -3.82739156e-01 -5.85852087e-01
-8.73445928e-01 -5.11000693e-01 -3.37561935e-01 -1.21912986e-01
5.68314373e-01 -7.30504155e-01 -9.98782516e-01 2.14296028e-01
-5.51050127e-01 -1.96952850e-01 7.05467016e-02 6.47109568e-01
-6.22979328e-02 3.01344216e-01 -9.17631567e-01 -6.94840193e-01
-2.72460759e-01 -1.03388536e+00 6.02921426e-01 3.92215937e-01
-3.52764241e-02 -1.47208858e+00 -1.16341226e-01 4.10411179e-01
5.63940167e-01 -2.30331182e-01 1.26335406e+00 -1.25762117e+00
-3.78251463e-01 -3.02464157e-01 -1.77396223e-01 1.59270227e-01
3.08699459e-01 -3.32707852e-01 -4.75379258e-01 -2.87549555e-01
-3.50155473e-01 -9.36702713e-02 5.71301401e-01 3.83246005e-01
1.53324234e+00 -4.92664456e-01 -3.50921899e-01 7.03809798e-01
1.23313558e+00 4.45128083e-01 3.05395514e-01 -1.72752038e-01
1.16648841e+00 5.35165906e-01 1.11584038e-01 4.14709657e-01
8.12393725e-01 6.74076319e-01 5.13832748e-01 2.31561828e-02
2.12175310e-01 -5.21635473e-01 2.38953412e-01 1.10665512e+00
-2.33395010e-01 -7.01748073e-01 -2.75195897e-01 9.38079357e-02
-1.95590520e+00 -7.07962751e-01 -1.83491930e-01 1.89955080e+00
3.71375024e-01 9.71528664e-02 2.01390237e-01 -3.74850690e-01
6.12487018e-01 2.17651576e-01 -9.12639439e-01 -3.40881348e-01
1.08769692e-01 -1.14068769e-01 1.95307523e-01 2.81865805e-01
-1.01491129e+00 9.04579878e-01 5.56423664e+00 5.98114967e-01
-9.78326499e-01 4.44423733e-03 5.52112818e-01 7.64584467e-02
-6.37885630e-01 -5.39180458e-01 -8.40975046e-01 4.35180575e-01
8.78940046e-01 3.69098298e-02 8.27019811e-01 8.13540399e-01
3.48818898e-02 6.16617501e-01 -1.17009664e+00 8.98599565e-01
1.04806989e-01 -1.13125014e+00 2.42430255e-01 2.52044648e-01
6.12814665e-01 -1.34700360e-02 2.42115125e-01 8.36763263e-01
7.33608425e-01 -8.27030897e-01 1.10825621e-01 5.71174204e-01
3.48957568e-01 -5.27767181e-01 6.16469264e-01 2.78644681e-01
-1.28935039e+00 -5.58283925e-01 -3.06984991e-01 -1.12106323e-01
1.24982275e-01 5.36601186e-01 -3.60269815e-01 9.31408048e-01
7.82632947e-01 1.21504974e+00 -5.41511953e-01 7.10167527e-01
6.29501939e-02 8.02518010e-01 -1.77326947e-01 -3.86536926e-01
9.94581357e-02 -6.71336651e-01 2.17429504e-01 1.05210090e+00
2.08114922e-01 2.18180314e-01 3.31816584e-01 7.19738185e-01
-2.68245369e-01 4.42775875e-01 -5.19199073e-01 -3.90782982e-01
1.50102556e-01 1.42993677e+00 -4.72641826e-01 -1.09755516e-01
-8.95376563e-01 8.19405437e-01 7.26290464e-01 6.87892020e-01
-7.60729373e-01 5.70488721e-02 6.90551579e-01 2.22892180e-01
3.69503498e-01 -1.52272448e-01 1.37640163e-01 -1.39158654e+00
-1.32519409e-01 -8.31419408e-01 6.58601284e-01 -4.34886098e-01
-1.84109569e+00 7.23948300e-01 -3.92704178e-03 -1.01729643e+00
-1.19230531e-01 -6.97434783e-01 -6.10952675e-01 4.52468455e-01
-1.03112769e+00 -1.34255898e+00 -2.13000014e-01 8.15341771e-01
3.95454198e-01 -4.17051166e-01 7.10336447e-01 7.54811764e-01
-8.95420849e-01 9.82561886e-01 6.97792247e-02 1.83177620e-01
1.79583862e-01 -1.01320422e+00 1.92071274e-01 3.27215731e-01
5.98917067e-01 1.04293597e+00 2.44142756e-01 -6.50861979e-01
-1.82150412e+00 -1.17309129e+00 3.94801199e-01 -2.68461764e-01
9.84782636e-01 -5.58571398e-01 -9.99128044e-01 1.08797348e+00
5.87814255e-03 -7.11465627e-02 1.13248646e+00 9.63584483e-01
-4.06701565e-01 -2.88000524e-01 -7.00535655e-01 7.47842193e-01
1.40055728e+00 -4.41239476e-01 -3.53834540e-01 3.03433567e-01
7.83969223e-01 -1.11192875e-01 -1.19152689e+00 3.65838945e-01
7.06047058e-01 -8.78778815e-01 1.02524650e+00 -1.11670959e+00
3.77547890e-01 1.00777574e-01 -2.16793522e-01 -1.38596106e+00
-9.76105571e-01 -4.18010592e-01 -5.90204597e-01 9.82700646e-01
5.49050689e-01 -7.07114041e-01 7.26082563e-01 6.16837561e-01
-2.89735764e-01 -1.08683693e+00 -2.99453884e-01 -2.28244156e-01
-4.14851725e-01 -4.41708505e-01 8.34524751e-01 1.11244559e+00
2.27879807e-01 6.61766350e-01 -7.98514605e-01 2.21129730e-01
2.07630947e-01 4.16615456e-01 5.95625758e-01 -1.40081823e+00
-9.03106868e-01 -5.02846062e-01 -3.07553887e-01 -1.53214014e+00
1.55476764e-01 -1.16192722e+00 -6.61958933e-01 -1.52380991e+00
6.28903285e-02 -4.34634358e-01 -9.95244026e-01 2.77082920e-01
-1.83986023e-01 -6.08103052e-02 -1.78092584e-01 1.88970312e-01
-9.23868775e-01 5.83323240e-01 1.50957012e+00 -3.21372867e-01
-4.32066888e-01 4.86683339e-01 -1.17518723e+00 4.92587596e-01
5.29987335e-01 -2.23187879e-02 -9.32735622e-01 -5.00822663e-01
7.32981086e-01 1.30546644e-01 -5.54506034e-02 -5.11860013e-01
2.03056991e-01 -1.22630253e-01 3.26334238e-01 -2.23926857e-01
2.68079460e-01 -1.18724823e+00 2.08346844e-01 5.64799383e-02
-6.55061722e-01 1.30127948e-02 -7.71298110e-02 1.20513487e+00
-1.25971332e-01 2.22925961e-01 8.30445886e-02 1.01239614e-01
-9.22270298e-01 8.59629273e-01 -1.81161523e-01 -3.79502535e-01
8.06478560e-01 -6.60844892e-02 -1.42504990e-01 -5.85664749e-01
-1.07121980e+00 3.56996834e-01 -2.71821886e-01 1.04359698e+00
3.79019946e-01 -1.46719015e+00 -3.12574685e-01 4.78311390e-01
2.75460958e-01 -4.95073348e-01 6.48427367e-01 6.37471855e-01
2.08674029e-01 2.35882804e-01 -1.12562224e-01 -2.79667318e-01
-8.54120553e-01 8.38791907e-01 3.03770125e-01 -5.37452042e-01
-6.44790769e-01 7.06348717e-01 4.18806583e-01 -6.19674146e-01
2.70367444e-01 -4.80963558e-01 -9.27239358e-01 -1.40436646e-02
2.56828904e-01 2.33690470e-01 -1.01235025e-01 -5.62190413e-01
-1.00114912e-01 2.78843373e-01 -5.25134921e-01 6.28253102e-01
1.37076926e+00 -1.97311193e-01 1.07296750e-01 4.54491347e-01
1.23182678e+00 -1.53561141e-02 -9.26071048e-01 -4.64197248e-01
-9.75229740e-02 -1.74261421e-01 9.56817120e-02 -9.01548624e-01
-1.48370361e+00 6.09525084e-01 3.70391458e-01 7.81059980e-01
9.97361779e-01 -3.17334682e-02 8.35609615e-01 6.60039365e-01
4.45062667e-01 -8.58671665e-01 5.23238964e-02 2.85879940e-01
7.47652888e-01 -1.28155410e+00 -2.83114426e-03 -4.29939717e-01
-7.88412035e-01 9.36413944e-01 8.30368638e-01 -3.57100427e-01
1.31296444e+00 -1.53257817e-01 -2.17021197e-01 -4.23945069e-01
-8.18461001e-01 -6.67049724e-04 7.74970710e-01 3.16722214e-01
7.17348814e-01 4.37682718e-01 -2.70260751e-01 1.22082984e+00
-7.57223368e-02 -6.20105676e-02 3.06214914e-02 2.94834495e-01
-1.08738535e-03 -1.05882001e+00 5.12278438e-01 1.00069821e+00
-3.26015800e-01 -1.86213091e-01 -1.95978656e-01 5.82825243e-01
-4.45032492e-02 1.00946569e+00 1.77304685e-01 -1.05467737e+00
4.80296582e-01 -1.99433684e-01 1.63437411e-01 -6.01953447e-01
-9.51238811e-01 7.64010996e-02 1.78605154e-01 -7.01870918e-01
-2.45303065e-01 -1.64234847e-01 -9.42397177e-01 -1.26403809e-01
-4.57285166e-01 1.06630765e-01 3.35565418e-01 8.90261054e-01
7.55834222e-01 8.57073307e-01 6.96610570e-01 -5.12084603e-01
-6.50145531e-01 -1.05852485e+00 -8.53221536e-01 7.76349604e-01
4.37116884e-02 -7.70002425e-01 -1.87636763e-01 -3.60496819e-01] | [10.186408996582031, 5.617628574371338] |
ebcc2cc0-9550-4f4e-98a0-d674a172a5d1 | characterizing-the-efficiency-vs-accuracy | 2204.07288 | null | https://arxiv.org/abs/2204.07288v1 | https://arxiv.org/pdf/2204.07288v1.pdf | Characterizing the Efficiency vs. Accuracy Trade-off for Long-Context NLP Models | With many real-world applications of Natural Language Processing (NLP) comprising of long texts, there has been a rise in NLP benchmarks that measure the accuracy of models that can handle longer input sequences. However, these benchmarks do not consider the trade-offs between accuracy, speed, and power consumption as input sizes or model sizes are varied. In this work, we perform a systematic study of this accuracy vs. efficiency trade-off on two widely used long-sequence models - Longformer-Encoder-Decoder (LED) and Big Bird - during fine-tuning and inference on four datasets from the SCROLLS benchmark. To study how this trade-off differs across hyperparameter settings, we compare the models across four sequence lengths (1024, 2048, 3072, 4096) and two model sizes (base and large) under a fixed resource budget. We find that LED consistently achieves better accuracy at lower energy costs than Big Bird. For summarization, we find that increasing model size is more energy efficient than increasing sequence length for higher accuracy. However, this comes at the cost of a large drop in inference speed. For question answering, we find that smaller models are both more efficient and more accurate due to the larger training batch sizes possible under a fixed resource budget. | ['Lisa Wu Wills', 'Bhuwan Dhingra', 'Phyllis Ang'] | 2022-04-15 | null | https://aclanthology.org/2022.nlppower-1.12 | https://aclanthology.org/2022.nlppower-1.12.pdf | nlppower-acl-2022-5 | ['2048'] | ['playing-games'] | [ 3.01387548e-01 -1.18282005e-01 -3.59828532e-01 -2.51302630e-01
-9.70128238e-01 -7.50886023e-01 4.55549777e-01 5.77332616e-01
-8.56096685e-01 6.57149613e-01 4.50429946e-01 -4.96379554e-01
2.87673273e-03 -8.32092643e-01 -7.10469246e-01 -2.25619614e-01
4.57683811e-03 2.52820164e-01 3.11565965e-01 5.39712049e-02
5.45724273e-01 1.07996024e-01 -1.45069420e+00 2.39750162e-01
8.20687890e-01 5.72419047e-01 2.06969887e-01 1.18491268e+00
-2.30330750e-01 6.02415800e-01 -7.98615277e-01 -3.69261920e-01
1.94119334e-01 -5.28504133e-01 -1.10219121e+00 -4.69959021e-01
3.91168833e-01 -2.78247327e-01 -2.41567627e-01 8.65546346e-01
8.44813526e-01 2.69544989e-01 4.68757063e-01 -1.08152223e+00
-3.08226317e-01 8.87236774e-01 -5.53385377e-01 5.76934457e-01
5.16651511e-01 4.59935665e-01 1.16077244e+00 -2.38014340e-01
5.62972307e-01 1.27382779e+00 6.11399472e-01 3.40454668e-01
-1.24490881e+00 -4.72675949e-01 -1.19757138e-01 3.01404625e-01
-1.28102231e+00 -8.22426319e-01 3.34684458e-03 7.21469298e-02
1.74671829e+00 4.11205500e-01 4.76664245e-01 9.40459490e-01
5.11415839e-01 5.13982236e-01 6.97857082e-01 -6.75773442e-01
5.32641470e-01 -1.31719694e-01 4.76505101e-01 3.91728580e-01
5.38520515e-01 -2.29339123e-01 -6.57723784e-01 -3.57789189e-01
1.69065267e-01 -3.98920536e-01 -3.59663099e-01 3.82309079e-01
-1.24973893e+00 7.60460317e-01 -8.56015384e-02 3.82432610e-01
-1.35623902e-01 5.95009446e-01 6.92080855e-01 4.69691813e-01
3.02403301e-01 9.32242513e-01 -6.07504487e-01 -8.79291892e-01
-1.20458448e+00 3.04697275e-01 1.31908798e+00 8.40391219e-01
4.05184627e-01 -2.75360085e-02 -5.95721364e-01 9.31000710e-01
-4.60896082e-02 4.57140476e-01 8.37557614e-01 -1.00965703e+00
7.27418065e-01 2.84229279e-01 1.21039273e-02 -6.84726059e-01
-5.91788232e-01 -1.61303863e-01 -6.97058737e-01 -3.91263813e-01
6.51775777e-01 -3.31415713e-01 -7.69359469e-01 1.75234568e+00
1.13000451e-02 -8.11669752e-02 7.94929862e-02 6.32659614e-01
5.77235520e-01 1.23369026e+00 1.56870812e-01 -3.63497019e-01
1.67222548e+00 -1.00495589e+00 -6.33221686e-01 -5.51921844e-01
8.86058807e-01 -6.39423430e-01 1.44803333e+00 2.34726802e-01
-1.33307576e+00 -3.18051338e-01 -1.21827424e+00 -4.76251960e-01
-2.05984563e-01 -1.44002676e-01 3.25969785e-01 8.51597369e-01
-9.58190203e-01 7.59893835e-01 -9.15476024e-01 -6.21489942e-01
5.73136471e-02 1.67323798e-01 2.14966998e-01 -3.55213210e-02
-1.10559106e+00 1.18137407e+00 5.18225729e-01 -4.71444517e-01
-2.57368475e-01 -9.36754048e-01 -5.42190015e-01 5.79623640e-01
3.02767068e-01 -6.49054646e-01 1.52303147e+00 -8.50057542e-01
-1.58090770e+00 3.70914131e-01 -2.19695628e-01 -7.12965727e-01
1.87304094e-01 -2.25886315e-01 -3.28824937e-01 -4.86401655e-02
-4.08311456e-01 9.00534689e-01 4.00880009e-01 -5.58843434e-01
-3.93019289e-01 -2.11068913e-01 1.66094676e-01 3.37812513e-01
-5.38648903e-01 -8.23931172e-02 -3.99031639e-01 -5.17447591e-01
-5.36838233e-01 -1.02989101e+00 -4.25389111e-02 -1.73032522e-01
-3.93584132e-01 -1.34419382e-01 3.99862885e-01 -5.86025834e-01
1.65812504e+00 -1.94580019e+00 -1.17847055e-01 2.51783654e-02
-2.93066323e-01 2.65298545e-01 -4.08864081e-01 5.32500088e-01
3.48474503e-01 4.48695362e-01 -2.26759240e-01 9.62247252e-02
-9.14254598e-03 2.99384266e-01 -3.26967426e-02 1.40343592e-01
-5.73254675e-02 9.20233786e-01 -8.10472906e-01 -6.44339800e-01
-1.20371543e-01 2.82916278e-01 -7.97949195e-01 -8.78378004e-02
-5.36471069e-01 -2.59597898e-01 -2.84824431e-01 1.94035545e-01
2.75736302e-01 -4.73371357e-01 3.33602011e-01 -8.66889954e-02
2.72945613e-02 5.77529609e-01 -9.63711679e-01 1.74953127e+00
-7.30728745e-01 9.59120333e-01 -2.15828210e-01 -4.87863719e-01
4.43902433e-01 1.99487016e-01 1.99106168e-02 -1.03485370e+00
2.53987424e-02 2.52146453e-01 3.38448852e-01 -5.69200099e-01
9.77981985e-01 1.53180912e-01 -2.57400377e-03 8.63666177e-01
-2.37588406e-01 -1.04459003e-01 7.46915042e-01 1.59027070e-01
1.43139088e+00 -2.25673020e-01 3.91064793e-01 -4.19042051e-01
5.28612621e-02 6.53176159e-02 1.78784013e-01 8.38311315e-01
3.91337201e-02 2.70066291e-01 5.76342165e-01 -2.34627336e-01
-1.39208841e+00 -5.61387479e-01 -1.19170137e-01 1.41234899e+00
3.44994366e-02 -6.74068153e-01 -9.50740457e-01 -2.69623518e-01
-4.59903367e-02 1.14712560e+00 -1.59141958e-01 -2.72624344e-01
-6.25882447e-01 -9.37060118e-01 1.01042497e+00 4.57526654e-01
5.45231164e-01 -1.02664483e+00 -1.39565134e+00 3.65479738e-02
-3.76128674e-01 -1.22499692e+00 -7.39056587e-01 6.32599965e-02
-9.80897546e-01 -5.73507190e-01 -4.80621547e-01 -3.48290503e-01
3.34149152e-01 -7.97298402e-02 1.54537249e+00 2.99809486e-01
-2.39722267e-01 3.17206651e-01 -5.25114775e-01 -4.24180448e-01
-7.37510741e-01 6.99320674e-01 -1.96987763e-01 -8.00124228e-01
2.84597814e-01 -4.26436841e-01 -7.27102578e-01 8.66032541e-02
-1.09102070e+00 9.04016718e-02 6.89618409e-01 5.25644183e-01
3.28742445e-01 -1.32400423e-01 6.67065918e-01 -6.99050248e-01
9.68988180e-01 -6.02908671e-01 -3.74286771e-01 5.39605319e-01
-8.67215157e-01 7.81408727e-01 8.49361837e-01 -4.40708458e-01
-7.52493858e-01 -3.44307333e-01 -4.48825628e-01 1.91040039e-01
1.19991139e-01 4.43695456e-01 2.01308385e-01 3.11110258e-01
8.66074502e-01 2.62645483e-01 -4.45813797e-02 -2.79666126e-01
3.16653252e-01 7.08764493e-01 2.68507600e-01 -5.92614472e-01
2.43074894e-01 -1.53250843e-01 -2.85800844e-01 -8.89349222e-01
-5.61372519e-01 -1.66240767e-01 -5.88791110e-02 1.62039205e-01
7.23060966e-01 -5.40378392e-01 -6.62357330e-01 8.99168476e-02
-9.99967456e-01 -6.80654287e-01 -4.26932544e-01 3.53103936e-01
-3.54680777e-01 5.36291718e-01 -6.94419801e-01 -5.12156367e-01
-1.04306602e+00 -1.08120072e+00 9.65856135e-01 7.12502748e-02
-6.79307759e-01 -8.48094583e-01 3.48966420e-02 1.23185351e-01
6.68817341e-01 -6.53554574e-02 1.35429156e+00 -6.96197689e-01
-2.83657819e-01 1.75011232e-01 -1.52489677e-01 7.72514790e-02
-1.42297611e-01 4.83309850e-02 -7.45875716e-01 -4.15677309e-01
-2.62292862e-01 -3.57755810e-01 7.34371543e-01 2.79380769e-01
1.37370706e+00 -7.24610806e-01 -2.75930196e-01 3.00081819e-01
1.59634066e+00 4.60704304e-02 7.28300035e-01 1.97613314e-01
2.58206099e-01 2.02463061e-01 4.97445583e-01 4.41960037e-01
2.76845515e-01 7.15167761e-01 1.92204975e-02 3.99778217e-01
-5.37978187e-02 -1.76858157e-01 6.78938091e-01 9.26904619e-01
2.08060950e-01 -8.59042168e-01 -1.17636013e+00 4.73644167e-01
-1.56273210e+00 -8.85647237e-01 6.32833689e-02 2.33858371e+00
1.16280198e+00 3.00847441e-01 1.72340512e-01 3.70034784e-01
3.54797602e-01 1.85982659e-01 -7.27215886e-01 -9.92861807e-01
1.58894330e-01 2.83483148e-01 8.24292302e-01 4.37713146e-01
-4.15037483e-01 3.91766399e-01 6.89096642e+00 1.23787141e+00
-1.05305624e+00 4.58889641e-02 8.16774070e-01 -7.49781311e-01
-3.20756286e-01 -2.73736358e-01 -8.32858384e-01 8.08504760e-01
1.94353676e+00 -4.83564049e-01 6.07606888e-01 5.04084408e-01
1.52613431e-01 -4.96480167e-01 -1.53382945e+00 8.42914701e-01
-8.03352892e-02 -1.49271834e+00 6.56286702e-02 -4.75696661e-02
6.50691092e-01 1.73316598e-01 -4.42335755e-01 3.15008044e-01
2.17264920e-01 -1.03091359e+00 6.76179647e-01 1.85164690e-01
8.55855942e-01 -7.66698897e-01 6.18364811e-01 6.19258285e-01
-8.56904984e-01 -2.35153526e-01 -4.25191551e-01 -6.24424703e-02
2.26101562e-01 7.90704906e-01 -7.31889904e-01 2.56299049e-01
5.97960830e-01 5.20173311e-02 -4.74711210e-01 8.05537522e-01
2.21376821e-01 8.52124095e-01 -6.38079345e-01 -5.13071656e-01
1.65678665e-01 3.18387300e-01 3.39755803e-01 1.61832702e+00
3.82095933e-01 2.36337170e-01 -1.86762035e-01 5.92056811e-01
-3.46791476e-01 8.47758278e-02 -9.83922929e-02 -9.71730798e-02
9.85507190e-01 1.07695067e+00 -7.40082741e-01 -4.74260688e-01
-1.67717755e-01 6.68382227e-01 2.04505593e-01 6.72678649e-02
-1.04042935e+00 -3.59539837e-01 3.88339072e-01 1.33415371e-01
9.41656828e-02 -3.78993809e-01 -4.80409175e-01 -9.11919773e-01
-1.15176579e-02 -1.07956600e+00 4.33875889e-01 -6.45225763e-01
-8.12850654e-01 3.71238410e-01 2.09231257e-01 -5.98389268e-01
-3.24991524e-01 -1.14402249e-01 -5.04257500e-01 5.73227346e-01
-1.32431805e+00 -2.59448707e-01 -2.61987627e-01 -2.74643656e-02
8.86266649e-01 4.50513095e-01 7.95523226e-01 3.20502907e-01
-5.53981245e-01 9.28811193e-01 3.63983274e-01 -1.72939926e-01
4.77013081e-01 -9.81999457e-01 8.11874092e-01 6.24991894e-01
8.06514472e-02 4.70042467e-01 7.67045319e-01 -3.04379761e-01
-1.81503797e+00 -1.07329917e+00 8.98579359e-01 -4.26975220e-01
2.68149406e-01 -2.70385623e-01 -8.91521573e-01 3.78401518e-01
5.12286544e-01 -3.70872408e-01 6.03866220e-01 1.49078006e-02
-2.10512251e-01 -4.97446104e-04 -1.24874890e+00 6.14577413e-01
8.98441672e-01 -2.79615015e-01 -2.38800675e-01 3.17605793e-01
9.38265324e-01 -3.88742149e-01 -1.25851059e+00 2.75679827e-01
6.82717979e-01 -6.95940554e-01 8.33437443e-01 -1.93332538e-01
5.63935935e-01 9.79622826e-02 -5.10099642e-02 -1.36926997e+00
-9.32730138e-02 -4.18962091e-01 -2.82530189e-01 1.07198143e+00
7.49190688e-01 -4.99620765e-01 5.99854231e-01 6.51836991e-01
1.49602860e-01 -1.18194437e+00 -8.40485215e-01 -8.94603252e-01
2.73556054e-01 -3.53063226e-01 7.28573382e-01 6.55334651e-01
2.58319408e-01 6.92595601e-01 2.53133681e-02 -3.64559203e-01
8.71817097e-02 1.76822636e-02 5.08158445e-01 -6.76671684e-01
-4.91039753e-01 -5.91377378e-01 1.60709009e-01 -1.23100996e+00
-1.34390429e-01 -7.30137289e-01 1.63160086e-01 -1.53799951e+00
4.27229583e-01 -3.66297603e-01 2.51173049e-01 5.60700536e-01
-3.55037898e-01 -4.93732467e-02 2.38981694e-01 1.94801003e-01
-6.35684371e-01 2.74253577e-01 7.83397198e-01 -1.71266217e-02
-2.93427736e-01 -3.66339147e-01 -5.56466162e-01 3.79403025e-01
9.01883423e-01 -4.39916670e-01 -5.13916731e-01 -9.24223065e-01
5.75272143e-01 1.37564763e-01 -9.54755545e-02 -1.08307683e+00
3.40140641e-01 1.47477177e-03 1.14558071e-01 -3.14144820e-01
1.97014272e-01 -5.42090595e-01 3.22328322e-02 7.20524430e-01
-7.99314499e-01 4.48509902e-01 4.62346643e-01 4.24829483e-01
2.04436287e-01 -5.04956067e-01 8.12928081e-01 4.58096862e-02
-4.95503068e-01 -1.41113862e-01 -4.19226676e-01 5.74863374e-01
9.28664029e-01 -3.79070848e-01 -5.23656309e-01 -3.97632778e-01
-6.80624172e-02 3.99046957e-01 3.03962857e-01 5.36255896e-01
1.38827324e-01 -8.76544595e-01 -8.38898897e-01 -1.42399997e-01
-1.26314312e-01 -6.46512806e-02 6.95996284e-02 7.26300240e-01
-8.04196477e-01 6.12347960e-01 4.79251109e-02 -5.28041303e-01
-1.31584322e+00 2.81387627e-01 3.16302150e-01 -6.05168462e-01
-2.92688638e-01 6.40365899e-01 -5.29123187e-01 -8.22234824e-02
3.10323566e-01 -7.58748829e-01 2.39130825e-01 -2.52393205e-02
3.90437663e-01 8.25482130e-01 2.49794185e-01 -1.29227145e-02
-3.13252687e-01 3.60373914e-01 5.36857508e-02 -7.10643455e-02
1.08626413e+00 2.43689865e-02 1.73720375e-01 2.77110100e-01
1.12221968e+00 -1.44479260e-01 -8.90245616e-01 -5.13504334e-02
1.44796595e-01 -2.91440487e-01 4.91047837e-02 -7.51426876e-01
-8.61634970e-01 6.76077127e-01 4.66475248e-01 3.16261679e-01
1.04467070e+00 -2.98363984e-01 1.20647895e+00 5.09163499e-01
3.30544978e-01 -1.29426157e+00 -1.01298410e-02 6.85038269e-01
5.40355682e-01 -8.09845448e-01 3.40942889e-01 -9.92537849e-03
-5.05677700e-01 9.57321942e-01 5.65251172e-01 2.46758237e-01
2.62242854e-01 4.98619229e-01 -4.94249016e-01 -3.86856832e-02
-1.23282480e+00 2.50379771e-01 -1.99646708e-02 -6.28748238e-02
6.31109715e-01 -5.23005649e-02 -5.61644316e-01 3.42217982e-01
-5.86711049e-01 9.30848271e-02 6.25400007e-01 1.02144265e+00
-6.37005448e-01 -1.02990091e+00 -2.78474748e-01 9.50879753e-01
-7.30643511e-01 -3.76946658e-01 -1.93697333e-01 5.74432015e-01
-2.88957447e-01 1.15322697e+00 4.63084102e-01 -1.94212362e-01
2.86626041e-01 2.61885256e-01 6.75856948e-01 -5.36549866e-01
-1.02160168e+00 -5.52903950e-01 5.40669918e-01 -4.98535633e-01
-2.26838544e-01 -3.86358440e-01 -1.44368219e+00 -8.54409575e-01
-6.14339948e-01 2.73574054e-01 7.32484102e-01 8.64059269e-01
7.38314569e-01 5.58222234e-01 1.52424663e-01 -4.79063004e-01
-9.40944433e-01 -1.12090504e+00 -9.24029350e-02 7.14650974e-02
1.02066360e-01 1.05932526e-01 -2.26911828e-01 -2.14706194e-02] | [11.360861778259277, 8.555187225341797] |
e5398c74-69fb-4ccf-9a0f-62c9a50c67d1 | molecular-transformer-for-chemical-reaction | 1811.02633 | null | https://arxiv.org/abs/1811.02633v2 | https://arxiv.org/pdf/1811.02633v2.pdf | Molecular Transformer - A Model for Uncertainty-Calibrated Chemical Reaction Prediction | Organic synthesis is one of the key stumbling blocks in medicinal chemistry. A necessary yet unsolved step in planning synthesis is solving the forward problem: given reactants and reagents, predict the products. Similar to other work, we treat reaction prediction as a machine translation problem between SMILES strings of reactants-reagents and the products. We show that a multi-head attention Molecular Transformer model outperforms all algorithms in the literature, achieving a top-1 accuracy above 90% on a common benchmark dataset. Our algorithm requires no handcrafted rules, and accurately predicts subtle chemical transformations. Crucially, our model can accurately estimate its own uncertainty, with an uncertainty score that is 89% accurate in terms of classifying whether a prediction is correct. Furthermore, we show that the model is able to handle inputs without reactant-reagent split and including stereochemistry, which makes our method universally applicable. | ['Alpha A. Lee', 'Théophile Gaudin', 'Costas Bekas', 'Teodoro Laino', 'Philippe Schwaller', 'Peter Bolgar'] | 2018-11-06 | null | null | null | null | ['chemical-reaction-prediction'] | ['medical'] | [ 6.04220688e-01 3.99153233e-01 -6.03332281e-01 -1.28117740e-01
-9.03083801e-01 -9.46438253e-01 8.60102177e-01 5.40143669e-01
-1.45882651e-01 1.23119521e+00 2.12511271e-01 -7.95053124e-01
2.81658798e-01 -6.88718855e-01 -1.16004753e+00 -8.31465483e-01
2.84766167e-01 8.77716601e-01 -6.08400144e-02 -5.72972260e-02
4.92797136e-01 6.13733470e-01 -1.09954548e+00 3.22607279e-01
9.43715394e-01 9.08935487e-01 -8.30546208e-03 6.41915977e-01
-6.87505491e-03 1.07397294e+00 -2.59593308e-01 -6.98136032e-01
1.96625605e-01 -3.99533212e-01 -1.19129658e+00 -5.78803957e-01
3.91555838e-02 2.14832813e-01 2.28909012e-02 9.59179342e-01
4.18777972e-01 1.17753811e-01 1.16592920e+00 -8.61367822e-01
-4.89128917e-01 7.09807336e-01 -1.13251559e-01 -1.39334589e-01
6.48875535e-01 3.45834255e-01 1.38573885e+00 -9.86883700e-01
7.86536574e-01 9.73229527e-01 2.57475227e-01 6.41570151e-01
-1.47265148e+00 -6.94164455e-01 1.99909404e-01 1.63776159e-01
-1.18644094e+00 -5.31076789e-01 2.97380835e-01 -7.27535307e-01
1.41897500e+00 3.47841054e-01 4.35397595e-01 1.03517663e+00
8.35568428e-01 5.12823999e-01 8.13993990e-01 -2.34090388e-01
5.91040075e-01 5.34119084e-02 -1.62853047e-01 6.42582178e-01
1.75948262e-01 2.99902260e-01 -4.18164998e-01 -1.85178444e-01
2.72397608e-01 2.30617434e-01 -2.25001782e-01 -2.65120536e-01
-1.20428002e+00 9.73059952e-01 4.58351552e-01 -2.49338942e-03
-3.84411693e-01 1.26944676e-01 1.47651255e-01 -7.07501359e-03
2.79112190e-01 1.17448175e+00 -8.35957646e-01 -5.62756062e-02
-8.88421059e-01 1.95865527e-01 9.53529894e-01 1.03149974e+00
3.84967417e-01 -2.40408748e-01 -7.56861866e-02 6.59799948e-02
-1.86938010e-02 1.68784559e-01 1.58068866e-01 -7.06103861e-01
3.49077135e-01 2.22032949e-01 3.50107789e-01 -3.57008696e-01
-6.37647867e-01 -4.02866453e-01 -5.85689723e-01 6.92292303e-02
3.98278803e-01 1.42833376e-02 -1.12064850e+00 1.73133516e+00
2.83068001e-01 -2.61851639e-01 1.61343440e-01 5.05524933e-01
6.57718122e-01 1.10332966e+00 3.30548853e-01 -7.98598051e-01
1.16094255e+00 -1.05698287e+00 -6.15769327e-01 -1.61338940e-01
7.44076371e-01 -9.04066324e-01 5.46287119e-01 9.05979216e-01
-1.34944904e+00 -1.05584756e-01 -9.70917702e-01 -4.84240383e-01
-5.67296445e-01 -1.86017692e-01 1.03692377e+00 4.14066821e-01
-4.71989363e-01 1.24369264e+00 -5.93058527e-01 1.24147564e-01
3.17119092e-01 7.78411865e-01 -4.91678357e-01 -6.36792257e-02
-1.05278444e+00 1.39955068e+00 6.59053028e-01 -1.18860625e-01
-1.13158226e+00 -1.27302814e+00 -7.23462582e-01 8.95395949e-02
5.36568224e-01 -9.38709021e-01 1.57916951e+00 -7.02131391e-01
-1.80344713e+00 5.33852696e-01 -3.69424909e-01 -4.02369618e-01
6.31874740e-01 1.65354326e-01 -1.91335976e-01 -3.21653098e-01
5.27951494e-02 6.93582356e-01 3.24469239e-01 -8.33920538e-01
-5.83278120e-01 -2.69231677e-01 -9.77360383e-02 2.75250487e-02
5.06468236e-01 -1.79611072e-01 5.24170399e-02 -2.81416863e-01
-2.01324280e-02 -8.94501090e-01 -6.40617907e-01 -2.72987098e-01
-6.63127780e-01 -2.66588479e-01 -8.19673091e-02 -4.47948128e-01
9.94939506e-01 -1.41931045e+00 6.17257416e-01 3.93451363e-01
2.50738233e-01 6.96958154e-02 1.18545189e-01 5.63508809e-01
-5.95492065e-01 2.22670287e-01 -3.06532711e-01 3.98109620e-03
5.46003282e-02 -1.51835144e-01 -5.78347027e-01 3.59774768e-01
3.21476400e-01 9.36815381e-01 -1.11701953e+00 -4.34052385e-02
2.13851929e-01 2.68047541e-01 -7.48736560e-01 1.43176019e-01
-8.56210053e-01 6.58624291e-01 -4.22175258e-01 6.52981877e-01
4.35583740e-01 -3.72725159e-01 4.97717828e-01 -1.81306973e-01
-3.64568233e-01 8.84822249e-01 -7.32832015e-01 1.71243501e+00
-5.14945388e-01 1.98562946e-02 -6.16130114e-01 -6.56958163e-01
7.42097378e-01 3.79524261e-01 3.62355113e-01 -4.56276268e-01
2.91316301e-01 4.92215335e-01 1.34194836e-01 -8.96684527e-02
3.07369173e-01 -8.01607013e-01 -4.56377119e-02 2.32598916e-01
2.16725588e-01 -5.60922861e-01 1.84389248e-01 7.80908689e-02
8.59276474e-01 2.44412333e-01 9.77071106e-01 -2.03764126e-01
5.34001827e-01 -8.49379599e-02 7.01106787e-01 5.40647328e-01
1.81327224e-01 4.82075274e-01 6.70224190e-01 -7.78302789e-01
-1.26304317e+00 -7.83046186e-01 -1.01026215e-01 9.27646041e-01
-2.33660758e-01 -6.47531807e-01 -5.94444513e-01 -6.23350203e-01
-6.51920661e-02 9.94778693e-01 -7.25256324e-01 -2.50969172e-01
-1.72187015e-01 -6.64994895e-01 1.01050630e-01 6.03347123e-01
-3.61311495e-01 -7.83077300e-01 4.89100814e-02 6.16130412e-01
-6.80127069e-02 -8.81188214e-01 -5.77857733e-01 7.08383501e-01
-6.80925369e-01 -1.33491230e+00 -4.56601769e-01 -4.33854252e-01
5.10742724e-01 -2.12150410e-01 8.85832548e-01 -4.24781293e-01
-4.22376603e-01 -5.72872519e-01 2.57010132e-01 -6.95441604e-01
-7.44890332e-01 1.67464674e-01 -8.92964378e-02 -4.10609722e-01
1.11882336e-01 -5.95304430e-01 -5.67499518e-01 -4.97246310e-02
-4.82107073e-01 3.83143276e-01 5.02772748e-01 7.74074316e-01
8.21956038e-01 -1.79489002e-01 2.80657202e-01 -1.12466919e+00
2.56519437e-01 -4.70956832e-01 -8.48031759e-01 4.74366426e-01
-7.94759095e-01 7.85196602e-01 1.13507009e+00 -9.94402915e-02
-8.22220504e-01 5.76653123e-01 -3.62103611e-01 9.12259817e-02
9.09442082e-02 3.27341467e-01 -4.90347534e-01 4.50265892e-02
6.54072583e-01 2.00395852e-01 -3.47380370e-01 -2.88801700e-01
4.93427306e-01 2.35081866e-01 3.46527338e-01 -8.79575968e-01
3.43664259e-01 1.78166151e-01 6.22933507e-01 -2.50637025e-01
-9.73107338e-01 -2.36915380e-01 -4.80953813e-01 5.01531959e-01
7.30574906e-01 -7.96558797e-01 -1.51338184e+00 -2.02215657e-01
-1.33680844e+00 -8.76780227e-02 -1.96453139e-01 5.38338423e-01
-8.85004342e-01 3.32974792e-01 -5.26325762e-01 -5.60981393e-01
-5.89950502e-01 -1.65296602e+00 9.37181950e-01 -1.17787920e-01
-4.49737936e-01 -8.71331453e-01 5.25729917e-02 4.19358015e-01
7.55578279e-02 4.66900378e-01 1.23473346e+00 -8.84994268e-01
-7.11132109e-01 -3.38159949e-02 1.50210887e-01 -1.82337523e-01
2.79346228e-01 1.26515254e-01 -8.40808511e-01 2.37684455e-02
-3.08090478e-01 -2.71752328e-01 1.17044401e+00 3.75782341e-01
1.18263960e+00 -5.33685744e-01 -4.97032285e-01 4.96383250e-01
9.80169177e-01 5.65458596e-01 5.40966868e-01 8.51825401e-02
5.68243682e-01 3.87538940e-01 5.22174060e-01 3.29617232e-01
1.93021759e-01 5.35437107e-01 4.58740145e-01 1.38881594e-01
2.74804533e-01 -6.82050169e-01 2.78031319e-01 1.53716788e-01
-3.76314938e-01 -3.83825511e-01 -7.72867560e-01 -1.30157666e-02
-1.71729732e+00 -1.06758809e+00 2.45092995e-02 2.31614041e+00
1.34584570e+00 1.87454775e-01 1.95603102e-01 2.65789598e-01
3.08044553e-01 -1.32670641e-01 -8.13105583e-01 -6.55521452e-01
1.77162558e-01 7.87929595e-01 8.17711532e-01 9.38060999e-01
-1.12159133e+00 1.27451694e+00 7.25649405e+00 7.90913284e-01
-1.23161948e+00 -2.85674721e-01 7.77049482e-01 -5.46797924e-02
-4.95195419e-01 1.88653603e-01 -8.82637382e-01 4.36593115e-01
1.09829950e+00 -3.66742104e-01 6.70217872e-01 7.02697456e-01
2.54820377e-01 1.89014271e-01 -1.80460203e+00 5.62747896e-01
-2.49326855e-01 -2.02356124e+00 3.10056806e-01 -8.17736164e-02
7.53108144e-01 -3.46561193e-01 2.49274567e-01 1.76616870e-02
7.86312938e-01 -1.43709016e+00 7.69152641e-01 2.57992566e-01
8.89140487e-01 -9.46250916e-01 2.90484428e-01 2.46472374e-01
-7.07026124e-01 -6.05295375e-02 -2.00797781e-01 -2.66041458e-02
1.32918194e-01 5.10097563e-01 -1.39368117e+00 6.47556722e-01
-1.36566505e-01 7.86139429e-01 1.69648632e-01 8.36864471e-01
-6.05074286e-01 2.25746870e-01 -2.78190732e-01 -1.82799071e-01
2.90035903e-01 -2.60413706e-01 4.87647839e-02 1.15287185e+00
2.73624420e-01 3.98963332e-01 6.70188889e-02 9.84786153e-01
-3.40929031e-01 2.00305700e-01 -5.30925095e-01 -2.97702372e-01
3.40355545e-01 7.50393867e-01 -4.39146578e-01 -3.06463718e-01
-3.33133601e-02 8.85699868e-01 1.27548158e-01 2.21748814e-01
-8.55781376e-01 -9.81554985e-02 7.70972550e-01 -2.68698364e-01
2.46902302e-01 1.14127487e-01 -2.75989741e-01 -1.00782371e+00
-3.49959046e-01 -9.78015721e-01 3.63486916e-01 -6.17621183e-01
-8.56622636e-01 3.37183893e-01 -5.15635252e-01 -7.07496226e-01
-1.40830293e-01 -9.21893299e-01 -3.56860697e-01 7.89149165e-01
-1.52959764e+00 -9.79568899e-01 5.28156221e-01 2.31526822e-01
5.31248093e-01 1.51953533e-01 1.09650230e+00 4.16108109e-02
-6.38329506e-01 5.08802354e-01 1.59482867e-01 -5.29194832e-01
6.96079075e-01 -1.32835555e+00 5.50230622e-01 5.87227166e-01
-1.15992166e-01 8.73616338e-01 1.07388151e+00 -5.99248230e-01
-1.60814774e+00 -1.20068336e+00 1.19821501e+00 -6.38412416e-01
6.56489968e-01 -4.91569787e-01 -4.18804288e-01 7.79934883e-01
-2.39124209e-01 -1.67714626e-01 1.00532222e+00 1.59926206e-01
-7.52651274e-01 2.56062508e-01 -1.08717155e+00 6.62528992e-01
1.04630506e+00 -4.05302137e-01 -3.71446818e-01 9.86177564e-01
9.58712697e-01 -7.89240837e-01 -1.06364441e+00 1.35149062e-01
5.10074377e-01 -4.58604485e-01 1.19647586e+00 -1.32887650e+00
8.43779147e-01 -2.68212587e-01 -1.60850480e-01 -1.28950870e+00
-4.99793410e-01 -1.07349181e+00 -1.50410593e-01 4.76161957e-01
1.09995830e+00 -5.55094421e-01 6.10337973e-01 9.89626467e-01
-2.29540512e-01 -7.33040512e-01 -7.23757625e-01 -6.92303658e-01
4.69429940e-01 -1.87777385e-01 8.11001420e-01 8.17802608e-01
4.56572056e-01 7.33350635e-01 -6.40880346e-01 8.27898085e-03
2.10805640e-01 3.12300950e-01 2.25933880e-01 -1.20533454e+00
-3.62222791e-01 -6.15275562e-01 -7.30594173e-02 -8.60331893e-01
2.96341896e-01 -1.37180007e+00 -2.52722710e-01 -1.42818761e+00
4.23094928e-01 -2.74623692e-01 -2.37789407e-01 6.74543023e-01
-2.13356167e-02 -1.66367173e-01 1.07608095e-01 -1.69744954e-01
-5.05324125e-01 4.85179812e-01 1.21986508e+00 -4.66057837e-01
-1.39436319e-01 2.37179101e-01 -1.18576825e+00 3.75960022e-01
6.20725572e-01 -6.15506887e-01 -1.91556722e-01 2.34882250e-01
6.32811069e-01 2.69156307e-01 -1.56455651e-01 -4.81865853e-01
2.29545295e-01 -5.74848831e-01 2.70732045e-01 -4.15590376e-01
3.60009253e-01 -6.65323317e-01 6.17419660e-01 7.89517522e-01
-6.59794331e-01 -3.24611783e-01 1.42051503e-01 5.02995968e-01
3.15693498e-01 -9.69840586e-02 6.76820815e-01 -2.81580508e-01
-8.63703489e-02 6.23125255e-01 -3.34646732e-01 -1.76094010e-01
1.14427483e+00 1.29074762e-02 -3.23824584e-01 -4.26594689e-02
-7.88464010e-01 1.41097412e-01 4.80473071e-01 1.57070801e-01
2.53340244e-01 -8.51632953e-01 -2.75668681e-01 -1.19212896e-01
9.70936790e-02 1.71262771e-02 -1.72244877e-01 5.54678798e-01
-5.19119263e-01 1.10650682e+00 8.82782266e-02 -1.77278534e-01
-1.11473393e+00 1.11127067e+00 4.40591902e-01 -3.91277581e-01
-1.27902791e-01 8.36658657e-01 3.22655439e-01 -3.55758995e-01
1.61070481e-01 -6.47133231e-01 2.08211899e-01 -1.00450568e-01
6.91379309e-01 8.48310813e-02 4.62693781e-01 -1.74943402e-01
-5.45117140e-01 4.27328438e-01 -5.62591553e-01 2.82793939e-01
1.53245533e+00 7.56389201e-01 1.06964082e-01 1.58115178e-01
1.19683945e+00 -1.74777433e-01 -9.98935461e-01 -1.40071779e-01
-1.36891305e-01 -6.96709603e-02 1.74280014e-02 -1.25851870e+00
-5.48074305e-01 1.00730550e+00 -6.09333105e-02 -4.46159095e-01
6.20072424e-01 -9.68577936e-02 5.53824425e-01 8.05841744e-01
3.49713415e-01 -8.41264009e-01 -2.37952590e-01 3.94465297e-01
7.78176963e-01 -1.14325964e+00 3.13093662e-01 -5.58396935e-01
-6.64404452e-01 1.20254564e+00 -3.16335596e-02 4.71228361e-01
1.48875222e-01 1.79576516e-01 -5.87487876e-01 -1.18053779e-01
-1.08278787e+00 3.82913500e-02 1.43519789e-01 3.09034169e-01
7.81543672e-01 2.41851449e-01 -3.99471790e-01 5.86516976e-01
-3.27229917e-01 1.27832025e-01 4.24228758e-01 8.68129969e-01
-4.56134975e-01 -1.52590024e+00 -1.07762359e-01 6.32498488e-02
-6.47906423e-01 -5.02456307e-01 -6.64686203e-01 4.21612620e-01
1.31617347e-02 8.01980615e-01 -4.18836325e-01 -1.90822005e-01
3.22529435e-01 2.35419542e-01 1.01592207e+00 -6.73662305e-01
-5.92866957e-01 -3.27051133e-01 1.22137919e-01 -5.93236268e-01
-1.51848821e-02 -4.97524261e-01 -1.54128516e+00 -5.60479283e-01
-4.36772317e-01 6.30668581e-01 7.61091113e-01 1.02310944e+00
5.69143116e-01 3.60596478e-01 8.27312410e-01 -6.71164632e-01
-5.54376960e-01 -3.52128625e-01 -1.38806000e-01 3.49864922e-02
3.01399946e-01 -5.87150276e-01 -8.35085958e-02 -1.37675896e-01] | [4.546925067901611, 6.059456825256348] |
6a333742-0c4c-4534-bdcd-783773752a91 | 190107910 | 1901.07910 | null | https://arxiv.org/abs/1901.07910v3 | https://arxiv.org/pdf/1901.07910v3.pdf | NLSC: Unrestricted Natural Language-based Service Composition through Sentence Embeddings | Current approaches for service composition (assemblies of atomic services) require developers to use: (a) domain-specific semantics to formalize services that restrict the vocabulary for their descriptions, and (b) translation mechanisms for service retrieval to convert unstructured user requests to strongly-typed semantic representations. In our work, we argue that effort to developing service descriptions, request translations, and matching mechanisms could be reduced using unrestricted natural language; allowing both: (1) end-users to intuitively express their needs using natural language, and (2) service developers to develop services without relying on syntactic/semantic description languages. Although there are some natural language-based service composition approaches, they restrict service retrieval to syntactic/semantic matching. With recent developments in Machine learning and Natural Language Processing, we motivate the use of Sentence Embeddings by leveraging richer semantic representations of sentences for service description, matching and retrieval. Experimental results show that service composition development effort may be reduced by more than 44\% while keeping a high precision/recall when matching high-level user requests with low-level service method invocations. | ['Sushma A. Akoju', 'Oscar J. Romero', 'Ankit Dangi'] | 2019-01-23 | null | null | null | null | ['service-composition'] | ['miscellaneous'] | [ 2.89835632e-01 3.83673877e-01 -5.89065962e-02 -8.78048241e-01
-7.74725556e-01 -7.54801095e-01 1.02675664e+00 5.20745059e-03
-8.66073593e-02 1.20045006e-01 6.76347435e-01 -6.55057728e-01
1.30589992e-01 -8.44247162e-01 -2.96117485e-01 -6.21966179e-03
-9.78578553e-02 6.28447473e-01 4.30219471e-01 -5.57878435e-01
-5.85269295e-02 4.47379857e-01 -1.77891791e+00 7.96455920e-01
3.88560981e-01 9.34758604e-01 1.76477596e-01 5.24038494e-01
-1.02545202e+00 5.03839254e-01 -3.74735534e-01 -5.63066185e-01
1.72862530e-01 -2.81104594e-01 -9.82999623e-01 2.04083592e-01
-1.22815594e-01 -2.31425554e-01 1.76969647e-01 9.19455171e-01
7.60645643e-02 -9.51715410e-02 2.20095620e-01 -1.70657766e+00
-1.06474924e+00 8.86996150e-01 1.30740851e-01 -3.02750170e-01
1.15134239e+00 2.11625099e-01 1.40531754e+00 -6.89297855e-01
8.65771055e-01 1.20343971e+00 4.40902144e-01 1.22199368e+00
-1.28765070e+00 -4.60559100e-01 4.38982219e-01 -2.11713091e-01
-1.20205343e+00 -7.85101235e-01 3.19134414e-01 -2.99736142e-01
1.85760415e+00 6.09002411e-01 4.93325651e-01 1.02960849e+00
-2.52886236e-01 4.07342315e-01 4.81716841e-01 -5.82932949e-01
3.53697240e-01 5.16624153e-01 2.69616365e-01 4.60011363e-01
1.70378104e-01 -3.62548769e-01 -1.17533125e-01 -7.54631341e-01
1.91828147e-01 2.69559950e-01 -6.95385635e-02 -3.06966543e-01
-8.05760264e-01 7.00170934e-01 -8.67221132e-02 6.71587288e-01
-2.73058176e-01 2.17987239e-01 8.12222183e-01 7.03161418e-01
2.18954161e-01 7.06317961e-01 -8.47934127e-01 -4.52244759e-01
-2.90644884e-01 4.13461685e-01 1.66694593e+00 1.69289625e+00
7.03507185e-01 1.62176803e-01 3.16857547e-01 7.81369984e-01
5.58628440e-01 5.47487020e-01 5.92193961e-01 -9.76619124e-01
5.63165434e-02 9.78571653e-01 3.80106986e-01 -5.33705473e-01
-9.15362779e-03 2.93918371e-01 2.31811658e-01 -3.48037332e-02
2.22845208e-02 2.18518019e-01 -4.49561685e-01 1.61346364e+00
-7.41660371e-02 -2.89791703e-01 4.67681319e-01 8.92415822e-01
5.61218560e-01 5.03091812e-01 1.74258083e-01 -1.28417313e-01
1.61897373e+00 -8.75703871e-01 -4.81668472e-01 -2.70753294e-01
7.18417943e-01 -5.35991907e-01 1.54493535e+00 -2.03903913e-01
-1.26827133e+00 2.12369829e-01 -8.83059323e-01 -1.94877367e-02
-4.52384025e-01 -7.18748391e-01 9.88349676e-01 6.54003918e-01
-1.31878984e+00 2.70593375e-01 -8.34261119e-01 -8.77865016e-01
-8.49214196e-02 4.74415541e-01 -5.25905848e-01 -1.37312353e-01
-9.63272035e-01 6.69556022e-01 -1.71110690e-01 -5.05869985e-01
-1.04799592e+00 -4.33130711e-01 -1.32191741e+00 3.86363834e-01
2.21758664e-01 -8.27334583e-01 1.81308615e+00 -1.52277052e+00
-1.34734738e+00 6.42795682e-01 -3.09349269e-01 -2.73537874e-01
-2.66852289e-01 1.87006325e-01 -8.76902640e-01 -3.20677063e-03
2.28930280e-01 1.48824975e-01 3.72589558e-01 -1.06561220e+00
-9.51552689e-01 -2.67771840e-01 7.23074377e-01 -1.86765473e-02
-4.50681210e-01 9.02962565e-01 -3.82674336e-01 8.24499689e-03
1.88054144e-01 -6.97124362e-01 -2.95388132e-01 9.79410857e-02
3.64493072e-01 -5.68171084e-01 5.46334445e-01 -4.81032729e-01
7.24382401e-01 -2.06208253e+00 -2.07133189e-01 1.76201835e-01
-1.70660261e-02 -1.47098482e-01 -6.11161470e-01 1.00394380e+00
1.89463377e-01 4.59813893e-01 -1.47184417e-01 -2.35495955e-01
7.49685943e-01 6.52053773e-01 -3.80848467e-01 2.04502016e-01
4.44931865e-01 7.28056610e-01 -1.04474270e+00 -9.92929116e-02
-7.42392316e-02 5.38842738e-01 -8.59756887e-01 5.46614766e-01
-6.65402472e-01 -2.09861189e-01 -7.51204669e-01 8.14733267e-01
1.19197227e-01 -1.66661561e-01 6.85956538e-01 3.00769448e-01
-7.32135773e-03 1.13803864e+00 -9.44728434e-01 1.84070921e+00
-1.18633223e+00 2.11771831e-01 4.96965528e-01 -8.37557614e-01
9.23449278e-01 8.86386335e-01 4.91260022e-01 -7.30825543e-01
-4.54302698e-01 6.31269693e-01 -2.44693086e-01 -6.99021876e-01
7.23647699e-02 -4.39767122e-01 -4.89068776e-01 1.10293818e+00
-1.25297546e-01 -2.98475355e-01 1.94401085e-01 9.56571624e-02
1.41085100e+00 4.49880868e-01 -4.05218489e-02 -2.16093108e-01
5.82290530e-01 1.05337888e-01 5.90758145e-01 3.95407468e-01
2.03047603e-01 2.69530952e-01 4.47047532e-01 -7.08015561e-01
-1.19878531e+00 -9.54244316e-01 1.85667858e-01 1.77450347e+00
-1.47078767e-01 -6.68891549e-01 -6.57293797e-01 -6.55583024e-01
5.42521328e-02 8.90878379e-01 1.10055879e-01 -1.41549096e-01
-8.62006962e-01 -9.07118320e-02 8.06998968e-01 6.46320462e-01
-4.95063901e-01 -1.14216030e+00 -3.77042919e-01 5.16470969e-01
1.91022649e-01 -1.45796895e+00 -4.59152281e-01 4.45360057e-02
-5.25313139e-01 -9.89065409e-01 -2.25950420e-01 -8.28349113e-01
8.33721399e-01 1.91597268e-01 1.37210286e+00 4.70724076e-01
1.30424097e-01 6.32343471e-01 -7.63681650e-01 1.61963150e-01
-8.95852268e-01 -1.26380816e-01 1.67445481e-01 -3.76419574e-01
9.50433314e-01 -9.70218480e-01 -3.90069038e-01 3.43424350e-01
-1.16757548e+00 -1.91715341e-02 1.54910907e-01 5.19923389e-01
-3.75008769e-02 -6.09990299e-01 4.98640329e-01 -7.72916615e-01
7.67652988e-01 -3.88664275e-01 -4.49080318e-01 1.29908502e-01
-1.16474152e+00 2.75702626e-01 1.00422180e+00 -3.61434937e-01
-9.89587665e-01 -4.57891434e-01 -6.38035655e-01 3.29985648e-01
-4.41908449e-01 2.88584977e-01 -4.53116417e-01 1.83516815e-02
6.00467503e-01 3.90461624e-01 1.31514072e-01 -7.26864517e-01
5.81545651e-01 1.00589991e+00 3.67564783e-02 -1.14915407e+00
7.91827977e-01 5.50538242e-01 -6.41690016e-01 -3.95437568e-01
-8.86032358e-02 -4.73210692e-01 -4.29103011e-03 6.56616092e-01
5.51300347e-01 -5.92364907e-01 -6.46858335e-01 -4.84791338e-01
-1.10034966e+00 -9.80590805e-02 -4.92633879e-01 1.09620899e-01
-6.19107127e-01 2.34973356e-01 -7.61191010e-01 -6.36725426e-01
-5.69667757e-01 -1.40047073e+00 1.31396985e+00 -2.63645262e-01
-5.95321298e-01 -7.71251917e-01 -1.81605682e-01 1.69889465e-01
1.02303684e+00 -2.79350907e-01 1.22363222e+00 -1.27732921e+00
-3.49018931e-01 -4.73573476e-01 -3.70359421e-01 4.58606780e-01
3.78932983e-01 -9.51405540e-02 -5.33620238e-01 -2.33352274e-01
-2.73505151e-01 9.64066386e-02 -3.38128507e-01 -4.55990732e-01
5.10615349e-01 -7.09525585e-01 -1.11590542e-01 5.48964381e-01
1.40276456e+00 4.04678494e-01 2.57078946e-01 3.91812801e-01
3.41566235e-01 8.10098410e-01 -1.60931610e-02 4.47781414e-01
7.49985516e-01 8.17840576e-01 4.92500626e-02 1.99197337e-01
-3.90025638e-02 -2.46571079e-01 6.68660522e-01 1.00209773e+00
1.90475956e-01 3.82295817e-01 -1.07195234e+00 4.94504154e-01
-1.77015710e+00 -6.21612847e-01 3.90936881e-02 1.75355864e+00
9.36499894e-01 -2.07305804e-01 -4.00387459e-02 -2.68616408e-01
2.44377986e-01 -3.07084862e-02 -2.40868464e-01 -1.03173125e+00
2.26008385e-01 1.90840617e-01 1.05096176e-01 5.49759984e-01
-3.14211667e-01 9.53595817e-01 6.27327776e+00 -6.67641386e-02
-1.04582345e+00 4.68841493e-01 -2.31843621e-01 9.87869278e-02
-1.25961256e+00 5.69893003e-01 -6.73366487e-01 2.11334288e-01
1.43941557e+00 -5.93621612e-01 8.71794403e-01 1.08158517e+00
3.18097085e-01 6.35214686e-01 -1.68195975e+00 4.40421402e-01
1.50264055e-01 -1.40220761e+00 3.28427464e-01 -3.14228743e-01
3.63433897e-01 1.18025661e-01 -7.26386726e-01 5.08428752e-01
5.54366827e-01 -8.14205945e-01 7.50428557e-01 1.60256565e-01
8.11957836e-01 -2.36355722e-01 6.91171408e-01 1.15789443e-01
-1.27183163e+00 -2.03374401e-01 -2.46727496e-01 -3.71581197e-01
2.30705157e-01 -6.06063195e-02 -7.77255833e-01 3.39248776e-01
6.08912706e-01 3.16981733e-01 -2.19911505e-02 2.47292548e-01
-1.01068482e-01 2.40710944e-01 -2.71382928e-01 -3.87710005e-01
3.78366500e-01 -2.25973725e-01 4.24411416e-01 1.62485790e+00
3.07740033e-01 -1.77788258e-01 5.09375215e-01 8.84311855e-01
1.00021027e-01 1.80188015e-01 -6.39795005e-01 -6.25606418e-01
3.89125854e-01 9.27995980e-01 -2.76304096e-01 -5.14833450e-01
-1.08507740e+00 8.91521335e-01 -2.27880523e-01 6.75421536e-01
-3.43883395e-01 -4.12020802e-01 1.57941306e+00 6.11829638e-01
-1.38502363e-02 -2.08839521e-01 1.58569589e-02 -1.38937593e+00
3.59144151e-01 -1.27791893e+00 1.40639782e-01 -7.95289218e-01
-1.27876258e+00 9.30229127e-01 -4.06182818e-02 -6.74028695e-01
-7.61507511e-01 -4.06486630e-01 -2.25526899e-01 9.87374365e-01
-1.40453291e+00 -1.36404204e+00 1.91244408e-01 5.17861366e-01
7.03081667e-01 -4.14936543e-01 1.64205229e+00 4.22323644e-01
1.77093759e-01 2.66421705e-01 -3.62040550e-01 2.39928742e-03
1.14828326e-01 -1.13410842e+00 9.62234437e-01 5.78545094e-01
1.51768774e-01 1.15398180e+00 9.73620951e-01 -1.46621823e-01
-2.16262746e+00 -5.60253441e-01 1.45342648e+00 -4.42767441e-01
1.07888615e+00 -8.09964597e-01 -8.08235109e-01 9.24269438e-01
1.25167653e-01 7.40030706e-02 8.37794542e-01 2.43191630e-01
-8.82309318e-01 -3.31265926e-01 -1.35942340e+00 7.70783961e-01
1.21431661e+00 -1.22615218e+00 -6.93654716e-01 5.33523202e-01
1.33391631e+00 2.46457964e-01 -9.83623743e-01 8.78227130e-02
7.52992153e-01 -4.84234214e-01 6.21943533e-01 -1.15304089e+00
3.88939939e-02 -4.11821395e-01 -6.93769336e-01 -7.42732465e-01
2.23776773e-02 -6.53596997e-01 7.87841007e-02 1.17453158e+00
7.33350158e-01 -7.94843554e-01 5.00170588e-01 1.42579293e+00
-3.28564912e-01 -5.96224427e-01 -5.57787776e-01 -7.47670114e-01
-1.76389143e-01 -9.67210829e-01 1.29105854e+00 7.41626024e-01
5.78850746e-01 1.69975143e-02 3.73427570e-01 1.92517817e-01
-8.17200765e-02 4.58871454e-01 5.47749996e-01 -9.34768736e-01
-5.09310484e-01 -5.00749350e-01 -4.91124928e-01 -1.08077300e+00
5.75112343e-01 -1.06880128e+00 1.79952532e-01 -1.93570650e+00
-3.50024924e-02 -4.63936865e-01 -1.02418050e-01 6.58091903e-01
8.00547838e-01 -1.76853493e-01 -4.53524962e-02 2.07993016e-01
-5.84370553e-01 1.73826069e-01 6.07802987e-01 -5.22635616e-02
2.51107961e-01 3.13007017e-03 -1.14547479e+00 5.28540850e-01
3.16937029e-01 -5.40557384e-01 -4.61777061e-01 -8.90042365e-01
5.93745351e-01 9.07465294e-02 6.24552704e-02 -3.66943300e-01
-3.35707627e-02 -5.19504249e-01 -6.86813772e-01 7.99484015e-01
1.75813466e-01 -1.24457252e+00 2.39127159e-01 2.22602099e-01
-5.56598246e-01 4.03552562e-01 -1.53538615e-01 1.36960922e-02
-1.67319268e-01 -5.63196421e-01 2.04651013e-01 -4.60391074e-01
-8.63984287e-01 1.96480677e-01 -7.74867833e-01 -2.08949715e-01
7.12482393e-01 -6.45538345e-02 -3.11760336e-01 -4.88582730e-01
-7.06665456e-01 -9.55131426e-02 1.19300044e+00 9.20597076e-01
6.90052927e-01 -1.10116327e+00 -3.77211511e-01 5.22965968e-01
5.63818097e-01 -4.98659551e-01 -6.41185522e-01 3.91885966e-01
-6.29125714e-01 3.61538053e-01 4.10016142e-02 -1.60756469e-01
-9.98403192e-01 5.09846568e-01 1.54266059e-01 3.62762034e-01
-1.03101242e+00 6.53611422e-01 -1.17297895e-01 -7.68890023e-01
2.89480090e-01 -5.47416627e-01 3.43943447e-01 -5.80029249e-01
9.24021244e-01 -3.85106742e-01 2.30762005e-01 -6.00538909e-01
-8.72309029e-01 6.54425398e-02 1.54348016e-01 -4.34479773e-01
1.49933124e+00 -7.55637735e-02 -5.99748194e-01 1.89065576e-01
1.27316356e+00 1.71446204e-02 -7.92414308e-01 -2.73733169e-01
7.81121910e-01 -5.12373388e-01 -3.82972479e-01 -6.60203218e-01
-6.04752898e-01 5.31909347e-01 2.56140858e-01 6.71619475e-01
8.44430864e-01 5.35316825e-01 1.12971795e+00 6.23053551e-01
9.69481587e-01 -8.35674524e-01 -3.82501483e-01 4.18689430e-01
7.48525500e-01 -7.24368393e-01 -5.17814159e-01 -2.63550013e-01
-5.29837966e-01 1.17010343e+00 1.37136057e-01 -1.48386344e-01
4.39453930e-01 1.01605952e+00 4.08878893e-01 -3.03985029e-01
-1.45243025e+00 -1.59446031e-01 -1.51404247e-01 9.89215076e-01
7.71458149e-01 1.84815899e-01 -5.21135986e-01 1.00574458e+00
8.41731876e-02 -1.39686629e-01 5.40111601e-01 1.13458002e+00
-4.53480452e-01 -1.78433669e+00 1.63192436e-01 2.58197278e-01
-6.22545421e-01 -4.75158691e-01 -1.74186140e-01 4.61085439e-01
-1.60044938e-01 1.26702273e+00 2.31820554e-01 -1.80650383e-01
5.88203788e-01 5.03822386e-01 2.05638096e-01 -1.26014507e+00
-7.61187553e-01 -2.11636528e-01 7.65426993e-01 -8.63762498e-01
-1.49861231e-01 -6.90985382e-01 -1.68451381e+00 -1.15961477e-01
-4.07996513e-02 7.08620846e-01 1.11061490e+00 1.17584372e+00
4.89574283e-01 2.41731092e-01 4.99263406e-01 -3.35629165e-01
-8.78673375e-01 -6.36808455e-01 -4.23394591e-01 8.20125937e-01
-4.53495681e-02 -1.57676190e-01 -5.48133969e-01 3.05390328e-01] | [9.782556533813477, 7.94589376449585] |
1d22a4dc-d010-4334-bef6-86235def8a55 | rethinking-graph-neural-networks-for-anomaly | 2205.15508 | null | https://arxiv.org/abs/2205.15508v1 | https://arxiv.org/pdf/2205.15508v1.pdf | Rethinking Graph Neural Networks for Anomaly Detection | Graph Neural Networks (GNNs) are widely applied for graph anomaly detection. As one of the key components for GNN design is to select a tailored spectral filter, we take the first step towards analyzing anomalies via the lens of the graph spectrum. Our crucial observation is the existence of anomalies will lead to the `right-shift' phenomenon, that is, the spectral energy distribution concentrates less on low frequencies and more on high frequencies. This fact motivates us to propose the Beta Wavelet Graph Neural Network (BWGNN). Indeed, BWGNN has spectral and spatial localized band-pass filters to better handle the `right-shift' phenomenon in anomalies. We demonstrate the effectiveness of BWGNN on four large-scale anomaly detection datasets. Our code and data are released at https://github.com/squareRoot3/Rethinking-Anomaly-Detection | ['Jia Li', 'Ziqi Gao', 'Jiajin Li', 'Jianheng Tang'] | 2022-05-31 | null | null | null | null | ['graph-anomaly-detection'] | ['graphs'] | [-9.77663975e-03 -1.18806064e-01 1.39770105e-01 -6.21615024e-03
-1.82079505e-02 -3.41424555e-01 3.62354606e-01 4.76510078e-01
2.23347664e-01 2.46110588e-01 3.74366134e-01 -5.09076834e-01
-3.44811201e-01 -1.12602270e+00 -4.81255293e-01 -6.98504150e-01
-5.53225040e-01 -3.04519117e-01 2.02745348e-01 -5.51868141e-01
8.73079225e-02 4.84539866e-01 -1.13635182e+00 -1.25583559e-01
7.70947576e-01 8.83317053e-01 -4.05241400e-01 6.83255732e-01
-1.10027909e-01 4.59066391e-01 -6.00910544e-01 -3.08101624e-01
3.24189931e-01 -5.60480952e-01 -4.87092286e-01 -2.37327628e-02
2.28442699e-01 2.77909227e-02 -6.26848757e-01 1.69746780e+00
4.27812248e-01 3.96957219e-01 4.58841383e-01 -1.48064041e+00
-8.00458789e-01 7.37439632e-01 -8.29683840e-01 7.60441005e-01
1.96671382e-01 -5.12290932e-02 1.16322732e+00 -6.51508629e-01
2.52789706e-01 1.08893943e+00 9.49467480e-01 1.81615904e-01
-1.26942182e+00 -3.72460276e-01 1.80259004e-01 4.35398817e-01
-1.42877126e+00 -2.16721117e-01 1.28461754e+00 -3.35523516e-01
6.90414548e-01 3.70399207e-01 8.30348969e-01 1.33473730e+00
2.04831347e-01 4.05347914e-01 4.61351693e-01 -4.66458142e-01
-9.67750996e-02 -5.13072073e-01 8.76088589e-02 6.58689797e-01
6.21638775e-01 1.18724108e-01 -5.16900122e-01 -3.27158213e-01
7.54512012e-01 2.26285100e-01 -4.25395429e-01 -2.05048360e-03
-8.34870577e-01 9.65136051e-01 5.78423738e-01 5.33503413e-01
-5.07969677e-01 4.06763643e-01 4.67647910e-01 5.83459914e-01
8.38167489e-01 2.56759167e-01 -1.55165210e-01 -1.51227359e-02
-4.82585460e-01 -3.58617418e-02 4.76161093e-01 4.47824180e-01
6.10298276e-01 5.08816957e-01 2.04059631e-02 9.72304046e-01
4.37686592e-01 3.55270982e-01 3.23992014e-01 -4.26323622e-01
1.22452572e-01 6.44014597e-01 -6.07527196e-01 -1.62813902e+00
-7.47832000e-01 -8.96146953e-01 -1.24489021e+00 -1.71238855e-01
5.53075969e-01 -3.15293819e-01 -7.30266213e-01 1.65380979e+00
3.25402617e-01 4.29483771e-01 -3.19280714e-01 6.99959755e-01
7.09585965e-01 5.45836627e-01 -1.65002912e-01 5.20096309e-02
1.28194487e+00 -4.69206452e-01 -7.81628847e-01 -1.08325511e-01
6.71606243e-01 -6.03163838e-01 9.49347317e-01 2.36955777e-01
-5.17883539e-01 -2.11869389e-01 -1.08960521e+00 4.02192444e-01
-5.31891108e-01 -3.86758208e-01 4.96118635e-01 6.27858996e-01
-9.50304568e-01 7.83916116e-01 -7.17288911e-01 -4.16696936e-01
1.92135900e-01 -1.78038612e-01 -2.13220656e-01 2.01113209e-01
-1.32765102e+00 3.38517398e-01 4.21443164e-01 2.07948998e-01
-2.03207776e-01 -6.32055640e-01 -6.82842433e-01 1.48651287e-01
4.12004620e-01 -2.30838761e-01 9.24006701e-01 -9.74403918e-01
-8.62783074e-01 4.94461238e-01 2.42878020e-01 -6.54367030e-01
2.40088776e-01 -4.00227755e-02 -1.08093727e+00 1.57502547e-01
-3.88736576e-02 -3.76778275e-01 1.12478042e+00 -8.55669737e-01
-4.76966441e-01 -3.73609483e-01 -3.79645914e-01 -4.18372542e-01
-6.71533108e-01 -1.46526799e-01 -1.11044254e-02 -1.25374889e+00
5.73933601e-01 -6.14260256e-01 7.96710253e-02 -5.46246946e-01
-7.78965116e-01 -2.32554987e-01 9.20197487e-01 -8.75198781e-01
1.71930766e+00 -2.27583623e+00 -2.06666127e-01 8.33455503e-01
7.12044239e-01 1.74404725e-01 -6.28292263e-02 8.61870766e-01
-5.90450108e-01 1.01970278e-01 -1.72335744e-01 2.14412451e-01
6.58978075e-02 -1.42166391e-01 -5.69593489e-01 7.92799175e-01
2.11085364e-01 6.71829760e-01 -7.95726180e-01 1.19015612e-01
5.26666604e-02 3.50347757e-01 -5.98155439e-01 -2.11159557e-01
-6.35021925e-02 4.47857469e-01 -3.61795574e-01 7.27101147e-01
7.51057029e-01 -4.11304057e-01 -6.89958781e-02 -2.52163291e-01
3.89486626e-02 4.04692478e-02 -1.13981533e+00 1.12968433e+00
1.56693503e-01 8.97951961e-01 3.08613759e-02 -1.35422182e+00
9.35151517e-01 1.32829279e-01 7.81740010e-01 -8.50524724e-01
7.80571327e-02 1.91793963e-01 2.41507784e-01 -4.47019428e-01
1.06494054e-01 2.00873524e-01 1.70796469e-01 4.08460647e-01
1.35576770e-01 4.85497326e-01 2.06790179e-01 2.02624470e-01
1.59459293e+00 -5.38956225e-01 3.01986128e-01 -3.87164116e-01
5.58004498e-01 -2.85594642e-01 5.20642221e-01 6.65057063e-01
-2.65930414e-01 3.36906523e-01 1.00537360e+00 -7.51559436e-01
-8.48813713e-01 -9.67256188e-01 -8.52746516e-02 1.29031658e+00
-1.21768735e-01 -5.46990037e-01 -7.33572245e-01 -6.31083548e-01
2.13171512e-01 6.80817068e-01 -6.26420736e-01 -5.22278368e-01
-6.52991295e-01 -9.26539898e-01 6.38933897e-01 2.63523251e-01
4.19105560e-01 -1.15514863e+00 -2.14532793e-01 2.23754123e-01
-1.25995696e-01 -8.54393303e-01 -5.55472434e-01 7.97488168e-02
-6.81843162e-01 -1.26207733e+00 -2.91149020e-01 -2.43035316e-01
6.43845618e-01 2.19431788e-01 1.08564460e+00 4.77941662e-01
-4.12264228e-01 4.99948084e-01 -7.08519757e-01 -5.38672149e-01
-3.12603921e-01 9.88325924e-02 1.19123153e-01 3.85833412e-01
4.14161980e-01 -1.09786785e+00 -6.88230395e-01 -1.15301479e-02
-9.66712117e-01 -4.18773234e-01 3.57948869e-01 6.24678016e-01
1.59146190e-01 4.70599383e-01 6.18819177e-01 -6.95798218e-01
1.16284144e+00 -5.67678511e-01 -8.28006089e-01 -1.24657467e-01
-4.26044941e-01 -1.79626808e-01 8.33474755e-01 -3.24231863e-01
-3.33092332e-01 -6.14115775e-01 -3.30114007e-01 -3.81052196e-01
-7.81674534e-02 6.97524965e-01 1.86136931e-01 -2.03879133e-01
1.01467729e+00 1.46010056e-01 -1.66396290e-01 -3.63157600e-01
1.30347416e-01 8.17004666e-02 3.75292659e-01 -3.89815420e-01
1.11434948e+00 4.09682125e-01 2.34596804e-01 -1.28563476e+00
-4.28084642e-01 -3.40411425e-01 -2.57814080e-01 -4.14809763e-01
6.31867409e-01 -3.94253641e-01 -7.83785999e-01 5.06803811e-01
-8.07200193e-01 -2.91222036e-01 -2.43494406e-01 2.04537347e-01
-2.87278295e-02 4.53388155e-01 -4.64093924e-01 -8.98894012e-01
-5.74328303e-01 -5.17788708e-01 5.48202872e-01 1.49788067e-01
-1.66916475e-01 -1.23641801e+00 2.34510005e-01 -3.63140285e-01
6.24733865e-01 6.59983218e-01 1.07011509e+00 -1.07433081e+00
-2.70540327e-01 -4.46078807e-01 -3.61295015e-01 2.99493611e-01
1.52494356e-01 2.69264519e-01 -8.67001772e-01 -4.50464427e-01
6.69426173e-02 3.61875832e-01 8.27723861e-01 7.37286031e-01
1.28687477e+00 -3.40319753e-01 -5.84977530e-02 9.90110099e-01
1.40239882e+00 7.03627616e-02 4.04823899e-01 2.87208557e-01
1.07019091e+00 3.52309108e-01 -7.43676424e-02 5.11128962e-01
-9.33922902e-02 3.89172077e-01 5.66210508e-01 -5.95195442e-02
7.46478364e-02 -2.42443874e-01 2.16747880e-01 9.76331413e-01
-2.23930448e-01 -3.69143873e-01 -1.32573640e+00 3.98727626e-01
-1.83239710e+00 -8.71825814e-01 -4.06714380e-01 2.00397420e+00
1.41934246e-01 2.05318049e-01 3.99127424e-01 4.11894888e-01
8.80641937e-01 5.91716826e-01 -2.91120261e-01 -2.92552829e-01
1.55447703e-02 1.87804133e-01 5.18844783e-01 4.14116889e-01
-1.20750570e+00 5.49290240e-01 5.24518728e+00 8.31957102e-01
-1.21419978e+00 -2.07987517e-01 3.24294776e-01 2.05044150e-01
-3.03867817e-01 -2.54105002e-01 -2.18942821e-01 5.62164664e-01
1.00747001e+00 -3.62481624e-01 5.69396615e-01 6.19729042e-01
2.08379984e-01 3.13388556e-01 -4.76368994e-01 7.96189785e-01
-2.59881526e-01 -1.18106627e+00 -7.11480379e-02 2.17128560e-01
2.64028192e-01 1.80228680e-01 -2.11015088e-03 1.36998743e-01
2.80795813e-01 -9.63467419e-01 3.42416644e-01 4.26263481e-01
3.85559171e-01 -9.76727962e-01 6.03304744e-01 -3.91920172e-02
-1.50681341e+00 -1.97036311e-01 -3.45228791e-01 1.71376303e-01
2.33906750e-02 1.18571496e+00 -7.05197155e-01 6.01035118e-01
9.33006048e-01 7.46109247e-01 -5.72508395e-01 1.08758783e+00
-2.14009777e-01 8.35344970e-01 -2.46401101e-01 1.11419447e-01
1.17213801e-01 -5.50516903e-01 9.79762077e-01 9.99119759e-01
6.43984675e-01 -8.82013515e-02 3.82838212e-02 8.04991841e-01
-1.00392312e-01 9.82705653e-02 -7.53041983e-01 -4.84903038e-01
2.94307530e-01 1.19580829e+00 -1.06858873e+00 1.24833062e-01
-5.31661332e-01 5.23361862e-01 1.05720989e-01 6.72522545e-01
-7.66767263e-01 -6.93714023e-01 7.01206565e-01 3.07570547e-01
1.89509839e-01 -1.83034211e-01 -1.44315183e-01 -1.03716445e+00
3.76006551e-02 -9.54117954e-01 7.27884829e-01 -3.43334943e-01
-1.67486179e+00 5.54074228e-01 -1.98839113e-01 -1.10693395e+00
-7.54877168e-04 -7.16184080e-01 -1.06286049e+00 6.84911489e-01
-1.10216534e+00 -9.12713170e-01 -5.27585447e-01 8.15059543e-01
1.34499684e-01 -1.78725556e-01 6.58262372e-01 5.10880470e-01
-8.18097949e-01 4.78867501e-01 -7.93708861e-02 6.30605340e-01
4.93777782e-01 -1.37507010e+00 8.86795402e-01 1.37012970e+00
4.04870301e-01 5.39583921e-01 9.21642601e-01 -9.17106867e-01
-1.13109982e+00 -1.00457060e+00 3.65362585e-01 -9.80509147e-02
1.27208948e+00 -3.85200799e-01 -1.24843240e+00 6.76148593e-01
1.08120188e-01 3.15552145e-01 5.91439009e-01 2.69219965e-01
-2.94777691e-01 -1.20804854e-01 -8.82984340e-01 7.38774896e-01
1.13146532e+00 -5.06604910e-01 -1.26554206e-01 4.14068848e-01
6.68358684e-01 -2.23833218e-01 -5.94329715e-01 6.40797257e-01
1.66904628e-01 -1.21519947e+00 1.09935927e+00 -4.89862442e-01
2.70408839e-01 -1.98095262e-01 -1.29702911e-01 -1.51579452e+00
-5.71887791e-01 -8.35029483e-01 -5.40652335e-01 9.23908949e-01
2.50847280e-01 -1.30936980e+00 5.50795555e-01 -3.10854673e-01
-2.03258827e-01 -6.47833824e-01 -9.87699270e-01 -5.81654370e-01
-3.42721641e-01 -7.32848585e-01 6.81818068e-01 1.15535557e+00
-4.81428429e-02 -1.96699761e-02 -2.99349040e-01 5.46916664e-01
6.25832558e-01 -2.24293292e-01 5.75103104e-01 -1.52314746e+00
-1.67203888e-01 -7.78342903e-01 -6.49465799e-01 -3.52917343e-01
-1.81541443e-01 -8.30225110e-01 -2.78201789e-01 -1.14181626e+00
-4.54465300e-01 1.21173538e-01 -5.29879749e-01 3.01909447e-01
-7.65611976e-02 2.93895215e-01 -2.41047010e-01 -1.29286692e-01
-2.97327489e-01 4.79981571e-01 8.67746592e-01 1.39690310e-01
-1.23837411e-01 1.73294768e-01 -6.68166816e-01 1.00507891e+00
1.14712775e+00 -3.68199617e-01 -4.23881501e-01 -4.21209857e-02
7.10429549e-01 -3.28127950e-01 4.77302253e-01 -1.15779662e+00
1.59695014e-01 7.46706203e-02 3.99298221e-01 -5.26307940e-01
-2.07172751e-01 -7.41065681e-01 3.67388390e-02 4.91694689e-01
1.64810866e-01 5.96401811e-01 3.02009791e-01 8.05528879e-01
-2.21481919e-01 5.05978242e-02 6.54555142e-01 1.89357921e-01
-6.04481936e-01 5.04047394e-01 -4.48183119e-01 2.08119780e-01
6.67163074e-01 5.06375320e-02 -6.55519009e-01 -6.22308731e-01
-7.04977036e-01 8.42009857e-03 9.68148187e-02 3.64853889e-01
6.35949492e-01 -1.52948940e+00 -6.21467888e-01 8.32285285e-01
1.25043735e-01 -3.12666833e-01 4.39625412e-01 1.07699513e+00
-8.08603644e-01 1.13466829e-01 -8.69834870e-02 -4.96569663e-01
-1.09819996e+00 4.08065379e-01 5.66906154e-01 -8.81741494e-02
-1.20160139e+00 9.35124636e-01 -8.66395831e-02 -2.41019860e-01
1.58561647e-01 -1.74081415e-01 -2.46694088e-01 1.92462772e-01
3.45082760e-01 6.34796798e-01 2.08864525e-01 -4.74533647e-01
-3.19783241e-01 4.19652164e-01 2.00679764e-01 2.39664987e-01
1.46912980e+00 6.56284243e-02 -5.35347104e-01 3.50669384e-01
9.62717772e-01 3.29355329e-01 -7.28091955e-01 -3.77077490e-01
3.38274747e-01 -3.94142032e-01 8.04475620e-02 -1.17531843e-01
-1.26208305e+00 5.25075614e-01 5.34016609e-01 1.45147943e+00
1.34768486e+00 -1.14692368e-01 7.58731425e-01 1.52670667e-01
-9.15155783e-02 -1.12476683e+00 2.39458174e-01 5.69958985e-01
9.72473323e-01 -8.97384703e-01 3.58516090e-02 -2.76557028e-01
-1.07047580e-01 1.21995246e+00 4.68010426e-01 -5.32213330e-01
9.85476017e-01 -7.31558055e-02 1.77789778e-01 -7.81295002e-01
-3.39242548e-01 -2.11658031e-01 6.23552978e-01 4.89726275e-01
2.51587063e-01 6.71938881e-02 -7.66447261e-02 3.13132435e-01
-4.03656423e-01 -8.64943206e-01 3.58532488e-01 4.49966043e-01
-4.10144985e-01 -8.38113308e-01 -5.76992035e-01 6.86941803e-01
-5.89258730e-01 -1.76331364e-02 -3.83448243e-01 6.93295479e-01
-9.61315557e-02 8.26472759e-01 6.95154592e-02 -5.57888985e-01
4.60022599e-01 2.69747376e-01 -8.00138991e-03 -2.21039727e-01
-2.31256202e-01 2.79517412e-01 -1.04346141e-01 -8.04954290e-01
1.53960502e-02 -2.68158257e-01 -9.35864151e-01 -7.61496603e-01
-9.65466201e-02 1.65808424e-02 4.43490744e-01 6.68918192e-01
3.29440296e-01 1.21942449e+00 5.24199843e-01 -5.64879835e-01
-1.18477702e-01 -1.07636166e+00 -8.35503280e-01 6.37446523e-01
7.45569646e-01 -5.60318291e-01 -8.17173898e-01 -4.96755332e-01] | [6.643148899078369, 5.812422752380371] |
fad39c75-2716-4d91-ab7b-f758d46134dc | probabilistic-deep-learning-using-random-sum | 1806.01910 | null | http://arxiv.org/abs/1806.01910v2 | http://arxiv.org/pdf/1806.01910v2.pdf | Probabilistic Deep Learning using Random Sum-Product Networks | The need for consistent treatment of uncertainty has recently triggered
increased interest in probabilistic deep learning methods. However, most
current approaches have severe limitations when it comes to inference, since
many of these models do not even permit to evaluate exact data likelihoods.
Sum-product networks (SPNs), on the other hand, are an excellent architecture
in that regard, as they allow to efficiently evaluate likelihoods, as well as
arbitrary marginalization and conditioning tasks. Nevertheless, SPNs have not
been fully explored as serious deep learning models, likely due to their
special structural requirements, which complicate learning. In this paper, we
make a drastic simplification and use random SPN structures which are trained
in a "classical deep learning manner", i.e. employing automatic
differentiation, SGD, and GPU support. The resulting models, called RAT-SPNs,
yield prediction results comparable to deep neural networks, while still being
interpretable as generative model and maintaining well-calibrated
uncertainties. This property makes them highly robust under missing input
features and enables them to naturally detect outliers and peculiar samples. | ['Kristian Kersting', 'Alejandro Molina', 'Martin Trapp', 'Karl Stelzner', 'Robert Peharz', 'Antonio Vergari', 'Zoubin Ghahramani'] | 2018-06-05 | null | null | null | null | ['probabilistic-deep-learning'] | ['computer-vision'] | [-1.23444065e-01 1.32400677e-01 8.74662772e-03 -5.70943773e-01
-7.19594598e-01 -4.02009904e-01 1.01577830e+00 2.13704184e-01
-3.89940977e-01 1.03972232e+00 -4.54098135e-02 -1.53546721e-01
-3.16142827e-01 -9.08553839e-01 -9.52291727e-01 -9.50170875e-01
7.98326433e-02 9.77105677e-01 1.07610993e-01 2.15076655e-01
1.39151722e-01 6.71272695e-01 -1.57483041e+00 -3.36356193e-01
1.22459245e+00 1.08452821e+00 3.20142284e-02 1.95030704e-01
-1.44409329e-01 5.30148745e-01 -4.21982944e-01 -7.02181637e-01
-1.86868191e-01 -2.82137468e-02 -3.54858726e-01 -2.99115747e-01
2.37256840e-01 -4.46919680e-01 -7.45513737e-02 1.07133329e+00
3.97019178e-01 2.27586821e-01 1.11947358e+00 -1.06035614e+00
-5.42679310e-01 7.60070682e-01 -4.11951244e-01 -1.02096379e-01
9.62038562e-02 4.36605141e-03 1.23269570e+00 -8.33972216e-01
2.20131859e-01 1.30334473e+00 8.09245586e-01 3.02264750e-01
-1.77699137e+00 -2.69257933e-01 9.16561484e-02 2.21110713e-02
-1.39108562e+00 -3.69652152e-01 7.12997139e-01 -4.50051516e-01
7.98990488e-01 1.23864271e-01 2.84965664e-01 1.59536827e+00
5.43080211e-01 8.91874075e-01 1.07369542e+00 -1.71861365e-01
7.29130507e-01 2.65079826e-01 -2.61219218e-02 3.24928224e-01
5.11033654e-01 2.68133525e-02 -4.48697239e-01 -1.94082543e-01
7.13499010e-01 3.42738591e-02 -1.35451183e-01 -6.89959347e-01
-1.01208723e+00 9.00347412e-01 4.04304504e-01 1.46186948e-01
-3.53075236e-01 3.46042067e-01 3.15399885e-01 -1.26784191e-01
7.30907559e-01 2.69757211e-01 -3.41221571e-01 -4.44883555e-02
-1.14522874e+00 5.45580804e-01 7.86376953e-01 5.20843565e-01
8.17102730e-01 2.40049466e-01 -8.91162530e-02 7.07916319e-01
4.95650560e-01 5.21752894e-01 1.63037017e-01 -7.99393594e-01
2.25546062e-01 2.66857028e-01 9.97457504e-02 -9.65457618e-01
-4.85694677e-01 -8.44993830e-01 -1.22302270e+00 4.97470289e-01
4.67428893e-01 -8.99731964e-02 -8.88319671e-01 1.85259080e+00
1.71509042e-01 1.09878436e-01 3.92220430e-02 6.11897886e-01
3.79331946e-01 5.80840588e-01 1.52415693e-01 1.39807090e-01
1.14091134e+00 -3.80257785e-01 -5.26160896e-01 -1.40455529e-01
4.43052918e-01 -5.11482000e-01 1.00811446e+00 7.74543226e-01
-8.51898611e-01 -4.18931752e-01 -9.92319047e-01 -2.00136527e-02
-4.51445580e-01 2.01986041e-02 9.09418344e-01 7.84444392e-01
-9.10501838e-01 1.09344411e+00 -1.25610101e+00 8.87615141e-03
4.95312512e-01 4.09563631e-01 -1.37242854e-01 3.04292776e-02
-1.07577407e+00 9.29866910e-01 6.25843406e-01 5.63816011e-01
-1.04522240e+00 -5.98288059e-01 -9.42862034e-01 3.25877577e-01
3.79299313e-01 -7.53618538e-01 1.13434637e+00 -4.75898564e-01
-1.55437410e+00 3.95839065e-01 -3.22993323e-02 -7.13240921e-01
8.37216914e-01 -3.78470272e-01 -3.18866938e-01 -3.42803180e-01
-1.48430571e-01 7.37902164e-01 9.18231666e-01 -1.18765855e+00
-2.36929119e-01 -2.60065228e-01 6.25178218e-02 -1.11178070e-01
-9.34099257e-02 -2.17293456e-01 -8.91806260e-02 -5.42102695e-01
1.91567615e-01 -8.23633254e-01 -4.49916571e-01 -2.93975808e-02
-8.06492269e-01 -2.26958528e-01 4.03831899e-01 -3.79698217e-01
8.61978769e-01 -2.02108979e+00 9.11795497e-02 3.34009916e-01
1.31790727e-01 1.86151996e-01 1.89081475e-01 3.39262843e-01
8.46592709e-02 5.43198772e-02 -5.27872503e-01 -8.24497342e-01
5.27802885e-01 6.10618472e-01 -4.86058921e-01 4.69662517e-01
6.49509549e-01 8.20659757e-01 -6.61603093e-01 -4.44440603e-01
4.34075505e-01 7.29725540e-01 -5.96112669e-01 1.71781313e-02
-7.29199290e-01 5.51521540e-01 -4.72958595e-01 3.50120157e-01
8.15939486e-01 -1.72673151e-01 7.92332739e-02 1.56794526e-02
-4.90164906e-02 2.24209771e-01 -1.37586987e+00 1.55628610e+00
-5.34269989e-01 4.44523424e-01 -1.89119697e-01 -1.08406699e+00
9.66357052e-01 1.32797405e-01 1.25160560e-01 -4.58550900e-01
1.62134618e-01 5.44821560e-01 -1.11496978e-01 -4.08312641e-02
5.55770993e-01 -2.09407732e-01 -6.01138026e-02 1.92173928e-01
2.11438343e-01 -1.72606826e-01 2.62933165e-01 -1.71428174e-02
5.79760671e-01 5.46953082e-01 1.38549402e-01 -3.64473969e-01
4.72279996e-01 -4.84976321e-01 6.46418095e-01 8.84833455e-01
3.21643502e-01 8.32498908e-01 7.72525549e-01 -3.30072850e-01
-8.61449778e-01 -1.47887063e+00 -6.08778238e-01 7.03144789e-01
-1.82175905e-01 -5.09856679e-02 -7.79664755e-01 -4.13491845e-01
2.18394071e-01 1.01442742e+00 -4.57192034e-01 -7.45841637e-02
-2.52944380e-01 -1.09813106e+00 3.65815848e-01 5.60001850e-01
1.98802769e-01 -9.02526438e-01 -3.26549917e-01 4.48420525e-01
1.11012161e-01 -8.68655503e-01 2.96748519e-01 4.64162081e-01
-9.44338918e-01 -6.96337581e-01 -7.56112278e-01 1.74614713e-02
4.61487323e-01 -4.32215065e-01 1.18992007e+00 -4.01229501e-01
-1.96798995e-01 6.11912413e-03 1.09491438e-01 -3.46486032e-01
-4.06832248e-01 1.00773290e-01 2.91961223e-01 1.02222376e-01
2.19392240e-01 -1.06712782e+00 -3.80932063e-01 1.21348970e-01
-1.10338545e+00 -2.44564846e-01 8.45597982e-01 8.76845181e-01
5.70390165e-01 1.88986342e-02 6.39966726e-01 -8.30713391e-01
4.34457928e-01 -4.54769850e-01 -8.91705036e-01 1.53431222e-01
-7.88007617e-01 4.70261008e-01 8.48845959e-01 1.63849201e-02
-1.28935409e+00 -1.08383693e-01 -5.17041326e-01 -1.65724441e-01
-4.55486834e-01 5.71767449e-01 -2.67835259e-01 2.01102525e-01
6.17698371e-01 2.11870119e-01 -1.39613017e-01 -8.02286267e-01
3.07878643e-01 3.20205629e-01 5.90165198e-01 -8.49079132e-01
6.70403063e-01 3.88620645e-01 2.83027917e-01 -7.20120013e-01
-9.13837850e-01 1.04251042e-01 -6.54400289e-01 1.54330790e-01
7.76827514e-01 -8.91575754e-01 -6.31810367e-01 5.24588466e-01
-1.14143157e+00 -5.36696203e-02 -3.32444608e-01 6.63590491e-01
-5.43789446e-01 5.67916751e-01 -6.72153950e-01 -1.07875192e+00
-1.26579210e-01 -1.27702701e+00 1.05320990e+00 2.10346013e-01
-1.52962148e-01 -1.18197429e+00 -1.30563632e-01 1.01324275e-01
5.21339059e-01 4.08366263e-01 1.13357568e+00 -7.61453867e-01
-7.15944827e-01 -2.13048831e-01 -3.52076948e-01 7.23870397e-01
-1.59340292e-01 2.83824295e-01 -1.32471299e+00 -9.70824733e-02
-8.57040584e-02 -2.32583985e-01 1.03814220e+00 6.60056591e-01
1.21982014e+00 2.21869629e-02 -2.70308048e-01 5.29884577e-01
1.29628456e+00 -3.59701008e-01 5.71279109e-01 4.53980938e-02
4.13122177e-01 5.01405120e-01 1.63670957e-01 6.53435469e-01
1.39517501e-01 5.15626431e-01 7.90042043e-01 2.34074622e-01
2.70337671e-01 -2.85490036e-01 2.48967916e-01 5.21935821e-01
-9.32646468e-02 -4.80937153e-01 -8.97432029e-01 3.23767990e-01
-1.84452486e+00 -6.88004076e-01 -3.54018390e-01 2.27050567e+00
7.62368441e-01 5.53004742e-01 -2.91305304e-01 8.16122368e-02
4.90553737e-01 2.79026270e-01 -6.27067804e-01 -1.42603412e-01
-2.24399537e-01 2.93974131e-01 1.55011714e-01 3.65829468e-01
-1.09834421e+00 5.79132080e-01 5.91084576e+00 1.06063676e+00
-8.71406674e-01 -2.00007185e-02 8.20682824e-01 1.01671256e-01
-7.64165103e-01 1.41288087e-01 -8.77790928e-01 6.13763571e-01
1.02671528e+00 2.90023595e-01 1.13064021e-01 9.54012156e-01
2.43448794e-01 -3.84507805e-01 -1.35110784e+00 8.87420416e-01
-2.90408850e-01 -1.14635479e+00 3.25517654e-02 9.50629711e-02
5.74511647e-01 1.48603708e-01 3.03042769e-01 3.49561423e-01
4.13440496e-01 -1.19544590e+00 8.25052738e-01 7.64048517e-01
4.08380061e-01 -9.70762610e-01 8.44624996e-01 4.69446003e-01
-4.61060464e-01 2.53985047e-01 -7.54383385e-01 -3.29665132e-02
4.57701296e-01 1.45382512e+00 -6.19519711e-01 6.62510157e-01
5.71420729e-01 3.28645974e-01 -4.32938606e-01 1.01276350e+00
-2.45802566e-01 6.74866438e-01 -7.72208452e-01 -1.03906155e-01
3.15189719e-01 -4.21678692e-01 4.34654832e-01 1.09693599e+00
5.33984065e-01 -7.71284938e-01 -1.46243647e-01 1.46459734e+00
3.83936688e-02 -1.80307403e-01 -4.06096220e-01 2.06803791e-02
1.50470525e-01 1.14767098e+00 -7.33131528e-01 1.53582925e-02
-2.35665828e-01 6.95330501e-01 3.16944182e-01 3.01398039e-01
-8.83449852e-01 -1.62501171e-01 7.40659416e-01 -1.14085302e-01
2.74881095e-01 -4.16991472e-01 -2.59613812e-01 -1.26757872e+00
1.18660241e-01 -3.30920964e-01 -2.03398596e-02 -5.32325625e-01
-1.49702537e+00 4.65250045e-01 2.85959858e-02 -9.97795105e-01
-5.74814618e-01 -9.32906866e-01 -3.60176295e-01 8.64543378e-01
-1.62017870e+00 -1.05389357e+00 6.77993149e-02 2.84277737e-01
5.18425852e-02 1.25757143e-01 8.56668591e-01 3.06861550e-01
-6.85812235e-01 4.74499375e-01 6.15322411e-01 -2.44256720e-01
5.25342405e-01 -1.46183038e+00 2.97159344e-01 8.18165839e-01
3.03838044e-01 7.53448367e-01 1.05295622e+00 -4.16458845e-01
-1.26537561e+00 -8.82710040e-01 7.13821411e-01 -4.58869666e-01
7.52392948e-01 -6.93196535e-01 -1.00232935e+00 5.19805491e-01
-1.81649532e-02 -1.40834402e-03 5.78925431e-01 5.12626529e-01
-4.36724305e-01 -1.81471542e-01 -1.02371395e+00 5.74839652e-01
6.98025465e-01 -3.78456652e-01 -5.03063679e-01 3.36993665e-01
4.84789997e-01 -3.34169596e-01 -8.12504709e-01 3.94734502e-01
3.70017678e-01 -1.48690176e+00 9.44198489e-01 -3.22852314e-01
5.05920827e-01 -2.30605051e-01 -1.32298306e-01 -1.17208540e+00
-4.70762439e-02 -4.30646151e-01 -2.60090321e-01 1.50622761e+00
5.67027211e-01 -8.93229783e-01 7.71832228e-01 8.16328526e-01
-2.74421960e-01 -6.08573854e-01 -1.14961743e+00 -9.43448663e-01
1.21634655e-01 -8.94153237e-01 6.82924211e-01 6.03060484e-01
-4.54744190e-01 5.37405722e-02 -5.31144083e-01 2.40138441e-01
8.50011349e-01 7.63939247e-02 4.94753003e-01 -1.65237355e+00
-5.21900952e-01 -6.92437351e-01 -3.98347348e-01 -1.01970518e+00
3.81906301e-01 -6.44379199e-01 1.52709767e-01 -1.25144732e+00
3.25847790e-02 -5.76865375e-01 -2.80064166e-01 2.62083471e-01
-1.90366227e-02 2.02845842e-01 -3.13402176e-01 1.32916257e-01
-3.84170145e-01 9.53118384e-01 6.58474028e-01 8.29366744e-02
1.94884166e-01 3.47502261e-01 -5.45219719e-01 8.39960873e-01
6.92596793e-01 -3.47508848e-01 -2.49058813e-01 -5.05680919e-01
5.22181451e-01 -2.48973876e-01 6.95777774e-01 -1.24396181e+00
7.33727589e-02 2.54410505e-02 5.54108858e-01 -7.16658771e-01
4.88344789e-01 -7.51012146e-01 3.67665261e-01 1.51837811e-01
-1.50888517e-01 -1.60935715e-01 1.03685997e-01 8.53596509e-01
-3.90301019e-01 -5.49146712e-01 6.53483689e-01 2.79084034e-03
-4.65762943e-01 2.05857083e-01 -3.73496503e-01 -2.04724848e-01
6.14848256e-01 -4.90778275e-02 -5.93356416e-03 -3.09657037e-01
-7.61403322e-01 5.33388816e-02 4.03807998e-01 3.81600522e-02
2.69880831e-01 -1.05612206e+00 -6.04822040e-01 4.00716923e-02
-1.16542540e-01 5.95946789e-01 5.19171476e-01 7.96201169e-01
-5.35543740e-01 4.31060523e-01 5.27802035e-02 -9.01931286e-01
-3.80778134e-01 4.83611375e-01 1.17209040e-01 -3.69656146e-01
-6.27536654e-01 8.09630990e-01 1.62460774e-01 -6.05897248e-01
2.78441191e-01 -5.52821219e-01 1.81353539e-02 1.14295945e-01
2.47033373e-01 2.01839358e-01 3.14332128e-01 -2.49866873e-01
-1.69031993e-01 1.96921602e-01 -6.58788681e-02 6.03651665e-02
1.50614798e+00 -1.23880366e-02 -2.32104897e-01 4.98620778e-01
8.74876738e-01 -1.00113519e-01 -1.51884937e+00 -2.67226309e-01
3.53858143e-01 -2.32104644e-01 -6.62344918e-02 -6.36173248e-01
-8.91682506e-01 1.32297063e+00 3.09606403e-01 2.80666381e-01
8.15535843e-01 -7.67645910e-02 4.28827733e-01 4.20041025e-01
4.01468903e-01 -1.03572667e+00 -2.81495869e-01 5.61283767e-01
6.66651428e-01 -1.15518844e+00 -7.06923008e-03 -1.32046357e-01
-2.92407632e-01 1.17365277e+00 2.79156595e-01 -2.02483431e-01
4.98847216e-01 2.86689878e-01 -3.34317744e-01 1.44914329e-01
-5.48055589e-01 -4.04984616e-02 1.84866562e-01 6.13378048e-01
4.54889566e-01 1.37107288e-02 -1.21830098e-01 6.06413782e-01
-1.59923747e-01 -1.16781697e-01 4.31936860e-01 4.86613035e-01
-3.14528584e-01 -1.10958791e+00 -2.60114640e-01 5.06233037e-01
-4.98562843e-01 -1.39273912e-01 2.08901092e-01 7.36652255e-01
-5.25920726e-02 6.03710175e-01 -9.43041965e-03 -3.79550792e-02
1.11759059e-01 1.17219560e-01 2.90898085e-01 -4.75031674e-01
-8.75640288e-02 -7.42261261e-02 1.06719974e-02 -3.85069311e-01
-1.23249002e-01 -8.23144794e-01 -8.94300699e-01 -3.54938716e-01
-3.68569940e-01 1.63538530e-01 8.28559339e-01 1.11094117e+00
3.04702491e-01 4.06024307e-01 2.19049037e-01 -1.20381629e+00
-1.17774439e+00 -1.04257154e+00 -9.53727305e-01 2.90390234e-02
2.29138628e-01 -9.19651151e-01 -4.03654277e-01 -4.27248836e-01] | [7.2564239501953125, 3.7992563247680664] |
d40e737a-b607-4d81-bfb8-82a02cd89711 | hallucinated-heartbeats-anomaly-aware-remote | 2303.06452 | null | https://arxiv.org/abs/2303.06452v1 | https://arxiv.org/pdf/2303.06452v1.pdf | Hallucinated Heartbeats: Anomaly-Aware Remote Pulse Estimation | Camera-based physiological monitoring, especially remote photoplethysmography (rPPG), is a promising tool for health diagnostics, and state-of-the-art pulse estimators have shown impressive performance on benchmark datasets. We argue that evaluations of modern solutions may be incomplete, as we uncover failure cases for videos without a live person, or in the presence of severe noise. We demonstrate that spatiotemporal deep learning models trained only with live samples "hallucinate" a genuine-shaped pulse on anomalous and noisy videos, which may have negative consequences when rPPG models are used by medical personnel. To address this, we offer: (a) An anomaly detection model, built on top of the predicted waveforms. We compare models trained in open-set (unknown abnormal predictions) and closed-set (abnormal predictions known when training) settings; (b) An anomaly-aware training regime that penalizes the model for predicting periodic signals from anomalous videos. Extensive experimentation with eight research datasets (rPPG-specific: DDPM, CDDPM, PURE, UBFC, ARPM; deep fakes: DFDC; face presentation attack detection: HKBU-MARs; rPPG outlier: KITTI) show better accuracy of anomaly detection for deep learning models incorporating the proposed training (75.8%), compared to models trained regularly (73.7%) and to hand-crafted rPPG methods (52-62%). | ['Adam Czajka', 'Patrick Flynn', 'Lu Niu', 'Benjamin Sporrer', 'Nathan Vance', 'Jeremy Speth'] | 2023-03-11 | null | null | null | null | ['face-presentation-attack-detection'] | ['computer-vision'] | [ 1.71704501e-01 4.19090651e-02 3.42986852e-01 -2.20963314e-01
-4.70089942e-01 -2.14496881e-01 1.76401302e-01 -3.44692260e-01
-8.33362043e-02 7.84905493e-01 6.14134297e-02 -2.84606099e-01
4.45078835e-02 -2.31203198e-01 -6.20711923e-01 -9.22769785e-01
-7.33930290e-01 1.06050216e-01 -8.69268253e-02 1.03999883e-01
3.29151541e-01 5.51671326e-01 -1.31455517e+00 5.11580348e-01
6.76991701e-01 1.31718874e+00 -7.66000390e-01 9.30240691e-01
3.09394419e-01 8.68228972e-01 -9.34249341e-01 -3.72978747e-01
3.31298977e-01 -6.52490079e-01 -2.95296073e-01 -1.60846874e-01
3.65676314e-01 -6.34940624e-01 -4.91611511e-01 7.82104671e-01
1.06651831e+00 -2.96417356e-01 5.12460470e-01 -1.63772631e+00
-3.95045698e-01 9.55778211e-02 -6.69985652e-01 7.64549911e-01
5.82521081e-01 6.09947026e-01 1.58286229e-01 -6.84609532e-01
4.02355701e-01 7.28181422e-01 1.08658147e+00 7.66701460e-01
-1.21407533e+00 -7.37932384e-01 -4.53307003e-01 5.37622154e-01
-1.25297582e+00 -5.76811254e-01 7.45131671e-01 -2.50107735e-01
8.69885921e-01 3.33923995e-01 6.79534078e-01 1.94933867e+00
4.69934970e-01 4.06000376e-01 1.10996246e+00 8.27071071e-02
1.61707431e-01 1.63590223e-01 6.37783334e-02 3.83775115e-01
6.32203743e-02 3.63547474e-01 -8.17088366e-01 -6.99495733e-01
5.24752200e-01 -3.54705036e-01 -8.10781479e-01 1.70905679e-01
-8.32403898e-01 4.35703069e-01 -2.92908251e-01 2.80436456e-01
-8.66985440e-01 -3.82166617e-02 7.41330266e-01 7.14610159e-01
5.74334979e-01 5.80177963e-01 -5.05808592e-01 -7.96507299e-01
-1.15126026e+00 2.54568327e-02 1.21990561e+00 4.19468373e-01
7.77773112e-02 6.17349744e-01 -3.34399819e-01 6.34443700e-01
-1.44015685e-01 4.22372013e-01 6.91343963e-01 -9.77912128e-01
1.45496158e-02 6.70058047e-03 1.98840097e-01 -1.19857919e+00
-6.55653477e-01 -4.78804588e-01 -8.69810581e-01 9.64632351e-03
3.57927114e-01 -5.55428624e-01 -9.13084984e-01 1.33264053e+00
4.87893894e-02 1.17197299e+00 1.93846926e-01 9.76268053e-01
8.51470172e-01 6.15555942e-01 -1.39900237e-01 -6.62850201e-01
8.88886333e-01 -3.73891324e-01 -8.15035343e-01 6.82535544e-02
5.06659448e-01 -3.89230222e-01 7.73917139e-01 1.00798011e+00
-9.91602480e-01 -3.39663655e-01 -8.84741187e-01 7.42464900e-01
-2.47152112e-02 2.72058398e-02 2.41174310e-01 1.07964301e+00
-1.22722590e+00 1.00892043e+00 -6.20927811e-01 -4.08224016e-01
6.80175602e-01 4.00628120e-01 -3.61024886e-01 1.01226740e-01
-1.23461115e+00 9.41595852e-01 -1.54173791e-01 4.25101608e-01
-1.07828414e+00 -1.00307453e+00 -5.47407806e-01 -2.46920977e-02
-3.03806178e-03 -1.87655792e-01 6.19321883e-01 -1.00711906e+00
-1.44116879e+00 9.43641305e-01 2.41932347e-02 -9.27545369e-01
7.00822532e-01 -4.42834616e-01 -1.03073454e+00 6.92856371e-01
-4.22531009e-01 1.28502786e-01 1.41166711e+00 -9.79028940e-01
5.09099849e-02 -1.99637026e-01 -6.29196346e-01 -2.08574280e-01
-1.70048922e-01 1.73937529e-01 2.16295287e-01 -2.58347571e-01
-1.22192383e-01 -6.69251502e-01 2.73841977e-01 -1.24833494e-01
-5.02274692e-01 9.52159092e-02 1.17998552e+00 -1.14155269e+00
1.12714469e+00 -2.12550354e+00 -4.62397516e-01 4.08900082e-01
1.41905576e-01 7.60209620e-01 -5.21493666e-02 2.62292415e-01
-3.31784904e-01 -4.24329527e-02 -3.00629973e-01 -2.91127741e-01
-3.21003765e-01 2.85746336e-01 -6.09775066e-01 8.65891159e-01
2.86851555e-01 5.38102508e-01 -5.93861103e-01 -2.56108731e-01
3.72223377e-01 4.07817811e-01 -3.13262790e-01 3.68340403e-01
3.35352153e-01 8.94771814e-01 1.37845036e-02 1.05199027e+00
8.08114231e-01 -1.44289210e-02 -2.03386351e-01 -3.55998009e-01
1.49426132e-01 -2.46464595e-01 -9.16631460e-01 1.14967108e+00
-3.29867378e-02 9.62787926e-01 -1.68047413e-01 -1.11606562e+00
9.90310371e-01 8.26583207e-01 8.22659910e-01 -5.44287980e-01
1.76288322e-01 1.22422040e-01 2.18696922e-01 -1.30247021e+00
-1.97993815e-02 5.29199578e-02 4.61184263e-01 3.54428560e-01
3.33741874e-01 5.48576117e-01 -3.44691634e-01 5.29338196e-02
1.60152256e+00 4.72503677e-02 -3.88064831e-02 -8.31918791e-02
3.96994561e-01 -5.17025232e-01 6.50884092e-01 1.07993150e+00
-9.22943115e-01 9.50160921e-01 1.11711049e+00 -7.03166306e-01
-8.29264760e-01 -9.56062078e-01 -3.91598761e-01 4.75403845e-01
-1.63205832e-01 -4.32796478e-02 -5.66436052e-01 -5.50980330e-01
3.62975597e-02 7.61119843e-01 -6.40088081e-01 -4.41966921e-01
-5.42130172e-01 -1.15710723e+00 1.17093241e+00 3.74804258e-01
4.89989609e-01 -1.43980896e+00 -7.49476254e-01 1.20154046e-01
-1.91595808e-01 -1.35023463e+00 1.16160981e-01 -2.63518810e-01
-6.01591945e-01 -1.23551738e+00 -6.01284623e-01 5.19199260e-02
3.23237568e-01 -5.04714131e-01 9.13657546e-01 5.54366820e-02
-6.34912848e-01 8.61244440e-01 -2.72974253e-01 -3.21129858e-01
-4.21084642e-01 -7.63472021e-01 4.67957646e-01 6.03112876e-01
6.68520808e-01 -8.34585488e-01 -8.15450311e-01 2.10019305e-01
-4.69723076e-01 -6.38668895e-01 4.66258883e-01 7.68692255e-01
4.93117571e-02 -5.20025134e-01 9.78799045e-01 -8.51223350e-01
6.40579641e-01 -7.69826829e-01 -2.33454049e-01 2.88713947e-02
-5.67067504e-01 -5.40301204e-01 5.57095528e-01 -7.20674157e-01
-8.48384142e-01 -1.58306524e-01 -3.69158894e-01 -1.18937278e+00
-4.13079530e-01 1.99285150e-01 4.05930102e-01 -3.60187858e-01
1.09633493e+00 6.01614594e-01 3.32336754e-01 -2.15357557e-01
-3.55720282e-01 7.09197640e-01 9.16672051e-01 -4.00962859e-01
4.35218543e-01 3.92822206e-01 2.21187398e-02 -1.17732084e+00
-3.31250131e-01 -3.16735268e-01 -4.12989348e-01 -5.61697483e-01
5.90054631e-01 -7.25373864e-01 -8.39706659e-01 7.34450281e-01
-1.00218236e+00 -1.61151692e-01 -1.29181698e-01 5.75277507e-01
-4.53477532e-01 5.52312076e-01 -5.88582397e-01 -1.21155834e+00
-6.86227798e-01 -6.40126407e-01 7.15058684e-01 2.54541159e-01
-2.98070222e-01 -8.71577263e-01 1.19864367e-01 1.07003629e-01
8.54043663e-01 9.06965792e-01 4.04634297e-01 -1.22615719e+00
7.69342855e-02 -4.57051396e-01 -6.18396811e-02 7.64004707e-01
-1.20166123e-01 2.38189071e-01 -1.71649194e+00 -2.29437366e-01
4.02920902e-01 -3.00400108e-01 5.57861626e-01 6.43130839e-01
1.51652133e+00 -3.95601481e-01 -1.76564872e-01 7.79789567e-01
9.96470928e-01 2.81008601e-01 1.25025165e+00 -4.60207388e-02
3.51639479e-01 3.32709044e-01 1.52658880e-01 8.40725303e-01
-2.48238772e-01 1.89901829e-01 5.65478444e-01 3.76546830e-02
3.31075430e-01 1.86893582e-01 5.98748803e-01 2.91287955e-02
-3.01902443e-01 -1.26987845e-01 -7.70385861e-01 3.87266159e-01
-1.54402888e+00 -1.16582537e+00 -2.65502602e-01 2.36614990e+00
4.04421985e-01 5.95377833e-02 3.16858977e-01 2.99472004e-01
7.48781621e-01 9.70881283e-02 -6.97568595e-01 -5.76038778e-01
-1.71418786e-01 3.50295454e-01 1.80696189e-01 -1.13194041e-01
-1.05913424e+00 2.87871331e-01 6.06038857e+00 2.19586194e-01
-1.52949309e+00 1.32719234e-01 7.61321545e-01 -2.29498073e-01
5.17017305e-01 -3.61323595e-01 -3.05639029e-01 1.08846748e+00
1.51801467e+00 3.90134715e-02 2.25401148e-01 7.20988452e-01
4.80813980e-01 8.81506726e-02 -1.05988383e+00 1.42585719e+00
5.43312550e-01 -1.21928716e+00 -5.50914466e-01 -1.26076434e-02
2.79139906e-01 2.50079874e-02 -4.36762571e-02 4.05234724e-01
-6.04105532e-01 -1.10043645e+00 2.76472652e-03 8.41498435e-01
7.39377081e-01 -4.64120597e-01 1.05261970e+00 2.15708852e-01
-3.81457061e-01 -1.45655960e-01 -4.70872551e-01 1.09735928e-01
1.18359089e-01 6.19017601e-01 -8.33914936e-01 4.29516852e-01
9.16772485e-01 7.61380315e-01 -3.88196141e-01 1.21005201e+00
2.77414061e-02 1.15055776e+00 -4.20780718e-01 3.96475166e-01
-2.05082327e-01 1.09894551e-01 9.98516083e-01 1.18521166e+00
4.72676098e-01 2.49357313e-01 -2.48288304e-01 9.15358663e-01
1.78786874e-01 -3.24246347e-01 -8.65365803e-01 2.16412861e-02
2.79517740e-01 1.30716789e+00 -2.78579026e-01 -1.30215824e-01
-3.93340081e-01 9.45799530e-01 -3.04144979e-01 4.19622421e-01
-1.07219481e+00 -2.49397248e-01 6.41245842e-01 1.60155416e-01
-7.20432475e-02 4.05378670e-01 -1.14098147e-01 -1.23470020e+00
4.66567986e-02 -8.63286078e-01 6.32302761e-01 -1.02603221e+00
-1.45045793e+00 6.51526213e-01 -2.10719720e-01 -1.44478428e+00
-4.36994553e-01 -5.24254441e-01 -1.17902899e+00 6.76935256e-01
-1.19814324e+00 -5.79260945e-01 -4.59942639e-01 7.66037583e-01
1.20740585e-01 -3.06182712e-01 8.78384233e-01 3.99325311e-01
-5.57917178e-01 7.28648484e-01 -3.78753394e-01 2.14477673e-01
8.12959015e-01 -9.78391886e-01 2.84727034e-03 9.84129608e-01
-1.89133570e-01 2.08160728e-01 9.38124895e-01 -4.52222317e-01
-1.31033039e+00 -9.40649211e-01 4.24881488e-01 -4.22083586e-01
4.72925186e-01 -7.64590055e-02 -1.42303360e+00 5.40504873e-01
1.10518150e-01 6.36305034e-01 6.71175539e-01 -1.64168045e-01
-3.94586883e-02 -1.91650599e-01 -1.70138884e+00 1.87965408e-01
4.91970628e-01 -3.17335486e-01 -5.38737357e-01 4.46133882e-01
1.54783532e-01 -6.70357108e-01 -9.55539048e-01 5.18096507e-01
8.08621526e-01 -1.34011257e+00 7.74506688e-01 -6.91093624e-01
3.38830709e-01 6.09284975e-02 2.55000085e-01 -1.34646475e+00
3.04397047e-01 -1.01896977e+00 -8.45586240e-01 8.27669442e-01
1.16474181e-01 -9.91786778e-01 7.60829926e-01 8.20515573e-01
-3.44540834e-01 -8.95831585e-01 -1.09148622e+00 -7.05617368e-01
-4.14414704e-01 -5.52269042e-01 1.61959767e-01 1.15526116e+00
9.43511799e-02 -3.43383282e-01 -1.01452804e+00 5.97444296e-01
4.89271224e-01 -4.07958031e-01 5.01564264e-01 -1.10323453e+00
-2.80795276e-01 -3.62093933e-03 -8.37532163e-01 -1.24519266e-01
-3.64191048e-02 -1.56426787e-01 -2.44790643e-01 -8.28476608e-01
-2.23330528e-01 -1.49579067e-02 -3.97950888e-01 5.75635612e-01
8.95723253e-02 5.22497892e-01 -1.95992023e-01 1.02572694e-01
-8.68792236e-02 1.39822155e-01 6.83172822e-01 4.22369778e-01
-3.32794189e-01 3.50884974e-01 -3.00491273e-01 6.78260505e-01
7.54183948e-01 -4.11488265e-01 -1.54771969e-01 3.33266705e-01
-1.50940632e-02 4.68760639e-01 8.31134379e-01 -1.39197516e+00
2.53637075e-01 2.88063675e-01 7.95291245e-01 -3.68829429e-01
5.58825254e-01 -6.32442474e-01 7.63914138e-02 5.34234822e-01
-6.52850941e-02 -5.06140701e-02 3.55329722e-01 5.16259372e-01
-3.18274125e-02 1.63486358e-02 9.12878633e-01 -4.76340344e-03
-5.00109971e-01 2.70927429e-01 -3.53182465e-01 1.09467685e-01
9.61711168e-01 -4.55522746e-01 -7.25322068e-01 -7.78906465e-01
-1.08846641e+00 2.51036976e-03 -1.74213439e-01 2.42966935e-01
9.27862942e-01 -8.34943175e-01 -8.64701807e-01 6.22493327e-01
-1.13765281e-02 -7.22513735e-01 6.10520542e-01 1.56066239e+00
-5.22905946e-01 -8.66610557e-02 -4.54358965e-01 -8.43403339e-01
-1.18903482e+00 2.85462826e-01 8.36068690e-01 2.18440622e-01
-1.07215202e+00 6.31007493e-01 -2.53835469e-01 1.31397441e-01
3.90421301e-01 -1.01207849e-02 -1.32019103e-01 -4.95113917e-02
6.14340246e-01 6.08853757e-01 3.41066569e-01 -3.86928409e-01
-3.45871389e-01 1.14521302e-01 -5.56113152e-03 3.44428450e-01
1.19010258e+00 2.46148020e-01 -4.88009537e-03 3.33620369e-01
9.99558210e-01 -3.57487321e-01 -1.20432448e+00 1.41080186e-01
-2.02249065e-01 -5.46387374e-01 -1.22846194e-01 -1.07725394e+00
-1.12532437e+00 9.61928904e-01 1.17360210e+00 3.90235782e-01
1.37492752e+00 -4.10514712e-01 6.84749126e-01 2.28690520e-01
1.02016106e-01 -8.54036450e-01 1.67812273e-01 -2.73053218e-02
8.75846922e-01 -1.04287946e+00 -6.17478192e-02 8.03881139e-02
-1.09104383e+00 1.18923664e+00 6.82378531e-01 -3.49982113e-01
7.82844067e-01 2.26577923e-01 4.44491096e-02 -3.35588753e-01
-9.40179229e-01 3.86829764e-01 8.91366377e-02 7.76902139e-01
-2.50498336e-02 -2.65251845e-01 -7.09702596e-02 6.54677451e-01
1.44892171e-01 2.26211175e-01 9.82699871e-01 6.26210272e-01
-1.84910774e-01 -2.73480684e-01 -3.66682589e-01 1.14217997e+00
-8.81559670e-01 1.04963845e-02 -1.60189033e-01 8.56822789e-01
-5.88068226e-03 7.15420485e-01 2.07551047e-01 -3.92707765e-01
2.25569621e-01 6.09802783e-01 1.79587349e-01 -1.01777561e-01
-6.64861202e-01 -9.66775864e-02 1.96907625e-01 -9.34799314e-01
-1.94305047e-01 -6.31413877e-01 -7.52211452e-01 -3.64126891e-01
-1.25426231e-02 -7.42042810e-02 5.04005969e-01 1.04815066e+00
6.43703640e-01 2.79499799e-01 6.85821533e-01 -7.66578913e-01
-5.80563486e-01 -1.18074322e+00 -7.99858510e-01 3.73041153e-01
6.27538204e-01 -5.25686443e-01 -1.11942518e+00 -4.36503589e-02] | [13.810835838317871, 2.72556734085083] |
8a5bd8ee-fff0-4d82-9abc-bf87993a5412 | a-language-independent-and-compositional | 1610.04345 | null | http://arxiv.org/abs/1610.04345v1 | http://arxiv.org/pdf/1610.04345v1.pdf | A Language-independent and Compositional Model for Personality Trait Recognition from Short Texts | Many methods have been used to recognize author personality traits from text,
typically combining linguistic feature engineering with shallow learning
models, e.g. linear regression or Support Vector Machines. This work uses
deep-learning-based models and atomic features of text, the characters, to
build hierarchical, vectorial word and sentence representations for trait
inference. This method, applied to a corpus of tweets, shows state-of-the-art
performance across five traits and three languages (English, Spanish and
Italian) compared with prior work in author profiling. The results, supported
by preliminary visualisation work, are encouraging for the ability to detect
complex human traits. | ['Fei Liu', 'Julien Perez', 'Scott Nowson'] | 2016-10-14 | a-language-independent-and-compositional-1 | https://aclanthology.org/E17-1071 | https://aclanthology.org/E17-1071.pdf | eacl-2017-4 | ['personality-trait-recognition'] | ['computer-vision'] | [-3.16248804e-01 1.17558464e-01 -2.92812377e-01 -7.21114218e-01
-3.98982942e-01 -3.20075423e-01 1.04464912e+00 6.26749694e-01
-4.97184008e-01 5.25348663e-01 7.49818325e-01 -1.58739328e-01
-1.59496129e-01 -6.51298642e-01 2.74989784e-01 -1.09600358e-01
-1.57863319e-01 6.62321091e-01 -3.70054722e-01 -3.39917421e-01
7.07748711e-01 4.44109619e-01 -1.51723099e+00 3.75303686e-01
5.88303208e-01 8.48321736e-01 -5.83540440e-01 1.01781583e+00
-5.81441820e-01 1.01104176e+00 -6.41107380e-01 -7.99936771e-01
-4.35651541e-01 9.39071998e-02 -9.38104272e-01 1.47616297e-01
6.70002460e-01 4.92617674e-02 -2.00602394e-02 6.82982624e-01
6.69849634e-01 -9.07808319e-02 8.92719507e-01 -7.40041733e-01
-1.03427804e+00 7.73733854e-01 -5.24428904e-01 2.16558307e-01
7.95234203e-01 -2.17597142e-01 1.29876959e+00 -7.50756025e-01
2.87619412e-01 1.55010021e+00 1.28705263e+00 3.06402564e-01
-1.46812892e+00 -4.82193768e-01 -4.38481458e-02 -2.10446209e-01
-1.26399064e+00 -5.85717320e-01 7.54794955e-01 -9.79268134e-01
1.22056627e+00 2.18251228e-01 6.16940618e-01 1.30433679e+00
3.16403747e-01 6.56837463e-01 1.62998283e+00 -4.78273213e-01
-1.01483546e-01 9.27169800e-01 6.70663178e-01 1.11228037e+00
2.21395105e-01 -3.00986588e-01 -6.73934937e-01 -7.63262689e-01
4.33693588e-01 -3.92847896e-01 6.34994030e-01 3.21847588e-01
-9.13132906e-01 1.19256377e+00 -3.97097349e-01 4.30816889e-01
-4.49720263e-01 -3.92853737e-01 8.68257642e-01 3.13163698e-01
9.51409101e-01 9.52575743e-01 -5.41015148e-01 -5.94546199e-01
-1.45564151e+00 4.47701752e-01 1.28053212e+00 5.25662899e-01
6.31479979e-01 2.38217399e-01 -3.95273894e-01 1.27121639e+00
3.54693353e-01 6.08640425e-02 9.01356816e-01 -6.25970066e-01
7.16436207e-02 8.18719506e-01 -4.68939878e-02 -1.35053432e+00
-7.93548882e-01 -3.92368257e-01 -5.64419985e-01 -9.48960781e-02
3.54439497e-01 -3.84523720e-01 -3.35176110e-01 1.18005323e+00
-2.15443835e-01 -3.75104010e-01 -9.99277160e-02 2.28863448e-01
1.19605064e+00 2.99141943e-01 3.00094187e-01 4.70387042e-02
1.61777461e+00 -3.69392961e-01 -3.97050679e-01 -2.53191411e-01
7.33329713e-01 -5.10238409e-01 1.07842660e+00 6.97486699e-01
-1.27615952e+00 -7.79248178e-01 -5.36546826e-01 -1.25519946e-01
-7.89424658e-01 2.99329877e-01 8.98400545e-01 1.28370297e+00
-1.10325515e+00 6.32880867e-01 -3.03675234e-01 -5.72070837e-01
7.03317285e-01 3.86345983e-01 -2.08333924e-01 4.51979935e-01
-1.20156908e+00 7.56148338e-01 1.69860542e-01 -6.17621183e-01
3.55198234e-03 -4.88360107e-01 -8.18382442e-01 1.07460767e-01
-4.79186505e-01 -5.11799097e-01 1.03855801e+00 -7.98275352e-01
-1.70732808e+00 1.46251535e+00 -2.45082095e-01 -4.16739434e-01
4.49287772e-01 -2.45103270e-01 -6.84825659e-01 -2.99810290e-01
1.15182832e-01 3.01193178e-01 5.73230505e-01 -7.31918573e-01
-5.46523750e-01 -3.93437117e-01 -4.89402980e-01 -2.25375861e-01
-1.02006626e+00 7.30872631e-01 8.16796124e-02 -3.16346765e-01
-3.86022151e-01 -5.06676972e-01 -1.63686007e-01 -1.04504013e+00
-6.45055592e-01 -1.03826344e+00 2.20822424e-01 -1.01722407e+00
1.39803410e+00 -1.68948054e+00 -1.35059893e-01 3.66812378e-01
5.18838346e-01 -8.96923169e-02 2.88669288e-01 5.61327040e-01
1.84620216e-01 4.21273649e-01 5.10553479e-01 -8.36545825e-01
5.03111959e-01 -1.07316554e-01 4.31986898e-02 4.62288320e-01
2.34426349e-01 9.00054514e-01 -6.58888936e-01 -7.64132857e-01
-5.79561889e-02 3.32072079e-01 -4.19231951e-01 6.13138303e-02
1.96799755e-01 5.14684469e-02 -2.27510631e-01 4.41863626e-01
1.70502692e-01 -2.60549765e-02 1.06430672e-01 3.40869367e-01
-4.84761924e-01 3.89592648e-01 -6.36338353e-01 1.03646541e+00
-4.09939826e-01 9.55269098e-01 -2.28007913e-01 -6.98617339e-01
1.54095590e+00 9.40675661e-02 4.89345253e-01 -3.85683358e-01
4.44277763e-01 -8.81206170e-02 -1.07634291e-01 -7.47012854e-01
7.32141376e-01 -1.01339556e-01 -6.54611945e-01 2.70895958e-01
5.12278378e-02 4.30719778e-02 1.71670273e-01 1.50457606e-01
8.12029660e-01 4.84642014e-03 5.29072285e-01 -5.77385664e-01
7.39071012e-01 -1.05868436e-01 2.98772097e-01 8.72696757e-01
-6.45372570e-02 6.53333962e-02 1.06011677e+00 -5.62627614e-01
-1.10518539e+00 -4.19481456e-01 -3.76458257e-01 1.79890752e+00
-9.99360800e-01 -8.18636775e-01 -7.49787688e-01 -3.47195864e-01
3.40381145e-01 7.60880411e-01 -8.58983040e-01 9.97916907e-02
-1.82564929e-01 -6.68328583e-01 1.05285096e+00 4.47881639e-01
1.12564206e-01 -1.12341535e+00 -4.97213304e-01 3.87650162e-01
5.29530823e-01 -1.11051822e+00 1.51945964e-01 1.44029064e-02
-6.48581564e-01 -5.19631505e-01 -4.00728464e-01 -6.39162064e-01
-3.52607556e-02 -6.40266418e-01 1.43966436e+00 -1.14656046e-01
-3.83804649e-01 4.58355337e-01 -1.92817703e-01 -6.70319080e-01
-4.05806363e-01 4.57263708e-01 3.79587144e-01 2.43461113e-02
1.02774763e+00 -2.67994881e-01 3.30933668e-02 -2.91639119e-01
-1.83202505e-01 -4.40048397e-01 7.39825249e-01 8.09133768e-01
-1.05322860e-01 6.85852766e-03 6.54336452e-01 -1.31948256e+00
1.39528251e+00 -6.53739214e-01 1.79107279e-01 -2.06915632e-01
-9.81145799e-01 -1.29712105e-01 7.21798062e-01 -4.91115630e-01
-8.23505938e-01 -3.05227816e-01 -5.88743746e-01 4.88786399e-01
-7.55983472e-01 7.74387240e-01 1.36824176e-01 1.04076922e-01
7.11613715e-01 2.16857284e-01 9.57178771e-02 -6.40001774e-01
1.65467381e-01 1.18951869e+00 4.62245196e-01 -5.70452929e-01
2.92678475e-01 -1.70348994e-02 -8.76907855e-02 -1.41377008e+00
-1.04668188e+00 -5.95116794e-01 -1.05875468e+00 -4.66674380e-02
7.43869305e-01 -8.02239656e-01 -1.01354504e+00 4.75693554e-01
-7.15087950e-01 -1.10265657e-01 5.68187870e-02 1.74938530e-01
-3.26830894e-01 3.61761630e-01 -8.90485644e-01 -1.06428719e+00
-5.37311673e-01 -5.57585478e-01 1.18459439e+00 2.02049777e-01
-1.20513988e+00 -1.65033078e+00 1.95870548e-01 3.60554397e-01
4.68967617e-01 3.42573702e-01 1.02386081e+00 -1.22395205e+00
7.09560394e-01 -6.85066104e-01 -2.09700733e-01 1.88598722e-01
-1.09808609e-01 1.36308655e-01 -1.02057624e+00 -1.42171636e-01
-4.98616755e-01 -6.28592014e-01 8.09985876e-01 1.99917302e-01
1.23173666e+00 -5.75522065e-01 -1.25718668e-01 3.62235695e-01
1.09384263e+00 -6.31730676e-01 3.48902464e-01 6.01124346e-01
7.22632468e-01 7.70705342e-01 1.15325943e-01 1.07628703e+00
7.72637725e-01 4.40953672e-01 -3.97208124e-01 -1.97760150e-01
4.20055926e-01 -2.27452293e-02 3.72570157e-01 5.99963427e-01
-1.86664402e-01 2.22163618e-01 -1.17374837e+00 2.23785102e-01
-1.52119732e+00 -1.14461017e+00 -5.71154356e-01 1.62692964e+00
8.16557527e-01 4.60986853e-01 9.30015326e-01 2.09366009e-01
1.80561334e-01 3.42044771e-01 -1.51250541e-01 -1.50779128e+00
-1.68082967e-01 7.37463534e-02 4.04431611e-01 3.34430426e-01
-1.09338117e+00 1.17585468e+00 7.60067272e+00 1.71134293e-01
-8.82143736e-01 7.33136712e-03 7.19798684e-01 -1.12862229e-01
-4.73011434e-02 -5.69858372e-01 -1.20837033e+00 3.94359648e-01
1.57509673e+00 -2.15843003e-02 2.67011374e-01 7.86695123e-01
2.74121791e-01 8.50400627e-02 -9.64825034e-01 1.00457215e+00
4.21773255e-01 -1.04435372e+00 -3.55683684e-01 3.32415521e-01
5.06130695e-01 -1.01981089e-01 2.48047933e-01 7.38870502e-01
6.36179328e-01 -1.21669674e+00 5.74239314e-01 8.59615445e-01
7.55284071e-01 -9.52810049e-01 9.16074634e-01 3.07198584e-01
-4.92844075e-01 -4.08727348e-01 -5.16381621e-01 -7.09807992e-01
-2.58725524e-01 7.51315057e-01 -9.44414437e-01 7.20103532e-02
8.70307922e-01 9.33768868e-01 -1.09177101e+00 3.66712511e-01
1.47505730e-01 9.21758473e-01 2.42895007e-01 -6.11644685e-01
3.89647365e-01 -6.63033500e-02 2.55179554e-01 2.22179794e+00
-6.04017451e-02 -2.38872424e-01 3.49003822e-01 7.87414908e-01
1.64604798e-01 6.02614045e-01 -6.84826255e-01 -4.06839877e-01
1.81723267e-01 1.63779724e+00 -4.34794009e-01 -5.77209532e-01
-7.09977031e-01 8.41315269e-01 6.85872972e-01 -9.52176005e-02
-2.37779170e-01 -4.69000161e-01 6.87399089e-01 2.54342079e-01
-2.07151398e-02 -3.73399436e-01 -9.49455857e-01 -8.43284190e-01
-6.99157417e-01 -9.01547074e-01 3.93388033e-01 -2.90497631e-01
-1.74309003e+00 4.25949454e-01 -2.94673175e-01 -2.93184072e-01
-4.34847951e-01 -8.75962853e-01 -8.09786201e-01 1.26528001e+00
-9.42734838e-01 -1.13064528e+00 -1.92525610e-01 1.80288732e-01
3.69025409e-01 -9.06188011e-01 1.00723743e+00 -1.31748855e-01
-7.19801545e-01 7.75701225e-01 7.51901865e-02 4.41414982e-01
7.48726606e-01 -1.71615744e+00 6.11096740e-01 6.00866154e-02
-7.68576413e-02 7.05523133e-01 8.18605959e-01 -7.57745624e-01
-1.15153027e+00 -7.11937010e-01 1.54341960e+00 -1.02837729e+00
9.93040919e-01 -6.82538509e-01 -7.81011224e-01 5.23287058e-01
5.10075927e-01 -5.87256551e-01 1.36423266e+00 8.88558984e-01
-3.61613095e-01 2.76992440e-01 -1.24709272e+00 2.85487175e-01
6.44517183e-01 -5.56255639e-01 -6.02758527e-01 1.82910934e-01
2.31500845e-02 1.70767277e-01 -1.52366745e+00 -1.64278179e-01
1.17573786e+00 -1.00856829e+00 9.14446414e-01 -1.05833411e+00
7.18384266e-01 8.07546318e-01 7.54455924e-02 -1.49546075e+00
-1.23049009e+00 -5.15053391e-01 -2.75936216e-01 1.78138113e+00
3.76854599e-01 -6.54849112e-01 7.92056501e-01 8.45083714e-01
1.48616493e-01 -9.52941060e-01 -3.17467988e-01 -6.47253215e-01
6.01648092e-01 -2.68473476e-01 7.09001362e-01 1.17373800e+00
3.58992755e-01 9.77094531e-01 -4.33588535e-01 -4.21744823e-01
4.77431685e-01 -2.47787163e-01 8.22128057e-01 -2.28137994e+00
-6.02738038e-02 -1.08362460e+00 -6.42323256e-01 -3.05521369e-01
7.80252457e-01 -8.02755594e-01 -4.35207725e-01 -1.71724594e+00
3.42935801e-01 -2.88562089e-01 -2.25433819e-02 3.43281299e-01
-1.37977093e-01 1.87563658e-01 -3.41400325e-01 1.77255739e-02
-5.01074195e-01 8.34040344e-03 7.18943775e-01 -1.05809215e-02
-2.15985000e-01 -2.64375489e-02 -1.33580446e+00 7.93914318e-01
1.17948890e+00 -1.36851713e-01 2.22626358e-01 1.17701344e-01
4.82469529e-01 -4.00261760e-01 5.04111201e-02 -8.73671174e-01
6.48591202e-03 5.67938760e-02 1.11326957e+00 -3.33922088e-01
3.67456943e-01 -1.97240174e-01 -6.76801383e-01 2.37541020e-01
-8.45190644e-01 3.62705737e-01 1.40975296e-01 1.80980340e-01
1.02990404e-01 -4.22686070e-01 6.20207787e-01 -3.71655583e-01
-5.76348543e-01 9.58121195e-02 -7.13835955e-01 8.29237849e-02
4.64260638e-01 -3.83632064e-01 -2.16172710e-01 -4.77807164e-01
-6.28595114e-01 -9.51824188e-02 3.95696104e-01 3.82250875e-01
4.68835235e-01 -1.04783583e+00 -1.24410856e+00 3.32771182e-01
1.27518728e-01 -1.06064939e+00 -1.93403006e-01 8.76567662e-01
-4.09398787e-03 6.70092046e-01 -3.31449568e-01 -1.88374028e-01
-1.46556664e+00 4.52976435e-01 5.20824529e-02 -2.69737154e-01
-5.24698973e-01 8.08385491e-01 -6.05589747e-01 -4.99269664e-01
9.05031562e-02 2.36227736e-01 -9.60475802e-01 6.68056071e-01
6.16872787e-01 6.73350036e-01 -6.73804283e-02 -9.20261800e-01
-2.79753685e-01 2.00762659e-01 -2.43045419e-01 -1.68423459e-01
1.39532447e+00 -5.34481443e-02 -3.74017417e-01 9.83645201e-01
1.24065244e+00 4.24348295e-01 -5.83481252e-01 -3.85506332e-01
6.46736264e-01 -3.39889318e-01 2.16006041e-01 -7.78930843e-01
-2.05125406e-01 9.26425099e-01 3.69333506e-01 9.25769329e-01
2.53938884e-01 1.45615533e-01 4.75703537e-01 2.91999549e-01
-6.56875670e-02 -1.54134214e+00 -3.45195383e-02 8.73135686e-01
3.81029308e-01 -1.26418495e+00 3.73721272e-01 1.64168030e-01
-1.03319192e+00 1.36553669e+00 4.26843941e-01 -1.02880634e-01
6.30969524e-01 1.49734899e-01 -3.83152664e-02 -4.87288296e-01
-7.79775023e-01 -8.17934647e-02 6.32196784e-01 4.76052344e-01
1.42427278e+00 4.96188104e-01 -3.04338217e-01 8.90690804e-01
-8.71473312e-01 -2.71943986e-01 5.16639948e-01 4.84830797e-01
-6.26741111e-01 -9.60735917e-01 -3.64030659e-01 1.31297874e+00
-1.05151367e+00 -2.82475471e-01 -1.22928619e+00 6.32358432e-01
-1.07866839e-01 1.05180526e+00 3.14964622e-01 -5.34782469e-01
1.42414302e-01 6.35541677e-01 3.41766953e-01 -9.07177567e-01
-1.05968595e+00 -1.58311069e-01 7.76326418e-01 -7.35866949e-02
-1.45806789e-01 -1.42823577e+00 -6.89029932e-01 -7.39357591e-01
2.48023257e-01 1.14578322e-01 6.40467882e-01 7.82250583e-01
1.93771869e-01 4.42874849e-01 6.07023001e-01 -8.12738359e-01
-5.15959859e-01 -1.44515395e+00 -7.99947560e-01 4.69223082e-01
3.91586572e-02 -3.78395796e-01 -6.49789199e-02 -1.48914039e-01] | [9.53443717956543, 10.310481071472168] |
3150c44d-b825-4a84-90fe-46ea72465579 | diversity-in-machine-learning | 1807.01477 | null | https://arxiv.org/abs/1807.01477v2 | https://arxiv.org/pdf/1807.01477v2.pdf | Diversity in Machine Learning | Machine learning methods have achieved good performance and been widely applied in various real-world applications. They can learn the model adaptively and be better fit for special requirements of different tasks. Generally, a good machine learning system is composed of plentiful training data, a good model training process, and an accurate inference. Many factors can affect the performance of the machine learning process, among which the diversity of the machine learning process is an important one. The diversity can help each procedure to guarantee a total good machine learning: diversity of the training data ensures that the training data can provide more discriminative information for the model, diversity of the learned model (diversity in parameters of each model or diversity among different base models) makes each parameter/model capture unique or complement information and the diversity in inference can provide multiple choices each of which corresponds to a specific plausible local optimal result. Even though the diversity plays an important role in machine learning process, there is no systematical analysis of the diversification in machine learning system. In this paper, we systematically summarize the methods to make data diversification, model diversification, and inference diversification in the machine learning process, respectively. In addition, the typical applications where the diversity technology improved the machine learning performance have been surveyed, including the remote sensing imaging tasks, machine translation, camera relocalization, image segmentation, object detection, topic modeling, and others. Finally, we discuss some challenges of the diversity technology in machine learning and point out some directions in future work. | ['Weidong Hu', 'Ping Zhong', 'Zhiqiang Gong'] | 2018-07-04 | null | null | null | null | ['camera-relocalization'] | ['computer-vision'] | [ 2.39002593e-02 -4.25991446e-01 -6.49738312e-01 -4.12261486e-01
-3.61256182e-01 -2.40889683e-01 2.71551877e-01 -3.05106729e-01
-2.47578397e-02 7.49959290e-01 -1.73559323e-01 -1.88686877e-01
-4.94149983e-01 -8.94014180e-01 -4.74466950e-01 -1.34208977e+00
2.79515475e-01 7.58715451e-01 2.63994157e-01 -8.14183056e-02
4.57557440e-01 7.42282331e-01 -1.71266413e+00 1.38597533e-01
1.21417344e+00 1.01854169e+00 7.69233704e-01 4.96962845e-01
-3.10823530e-01 3.70663226e-01 -4.33316320e-01 3.08625214e-03
2.73863316e-01 -4.94131237e-01 -6.51194155e-01 1.03346452e-01
3.30506079e-02 2.82821596e-01 1.59455031e-01 1.04038453e+00
5.82934439e-01 1.38925210e-01 1.20759130e+00 -1.21581352e+00
-5.12774825e-01 5.95126808e-01 -6.45850778e-01 2.11614177e-01
-4.25516784e-01 -7.57082039e-03 8.22710931e-01 -7.37112045e-01
1.89089537e-01 1.41210175e+00 5.10829210e-01 5.25466204e-01
-8.16307902e-01 -8.66398752e-01 1.01296179e-01 4.71375346e-01
-1.41775179e+00 4.68548127e-02 8.26177418e-01 -5.41104734e-01
3.91351819e-01 4.90896523e-01 8.25940967e-01 4.93715227e-01
5.68147719e-01 7.66342640e-01 9.43095922e-01 -4.09228772e-01
1.85574263e-01 4.40043837e-01 9.87595990e-02 6.67775452e-01
7.55137056e-02 -8.19926560e-02 -2.39308372e-01 -1.11920968e-01
7.77962565e-01 3.57047349e-01 -1.95283249e-01 -1.45337597e-01
-7.69080102e-01 1.02511573e+00 6.06520295e-01 2.11889401e-01
-2.36316457e-01 -1.16861887e-01 2.84665465e-01 1.30115315e-01
2.04728022e-01 4.58415180e-01 -7.54956663e-01 4.33942854e-01
-7.40058541e-01 7.41947368e-02 4.69671190e-01 5.48483312e-01
1.11784637e+00 2.58800518e-02 -1.06649205e-01 1.07289171e+00
4.19640958e-01 6.93856955e-01 7.14766145e-01 -8.03099155e-01
3.30051929e-01 6.11285925e-01 -2.63694644e-01 -1.07607281e+00
-5.43645501e-01 -7.41103888e-01 -1.28571725e+00 -8.61832500e-02
-2.70421971e-02 -1.60808653e-01 -6.35214329e-01 1.39333558e+00
5.70365191e-01 3.31765451e-02 -3.41224745e-02 6.54067039e-01
8.78170133e-01 1.17030156e+00 -5.58004528e-02 -5.26074588e-01
1.05334556e+00 -1.14742434e+00 -4.97281939e-01 -2.28281423e-01
2.96096593e-01 -8.58939350e-01 1.02853954e+00 6.06178045e-01
-6.04897082e-01 -9.27913487e-01 -7.17154443e-01 2.88049251e-01
-1.06407002e-01 2.03224644e-01 6.94974482e-01 3.86790186e-01
-4.55374271e-01 3.68293524e-01 -5.00405192e-01 -1.00975521e-01
3.59572351e-01 2.79947966e-01 1.19675390e-01 -1.86430171e-01
-1.05315030e+00 9.56792235e-01 7.93131053e-01 2.75601476e-01
-7.16740549e-01 -5.47905505e-01 -4.15551543e-01 6.32613618e-03
4.26104963e-01 -8.71606171e-01 9.57824588e-01 -1.19853270e+00
-1.27644145e+00 6.28501296e-01 -2.26218283e-01 -2.25216672e-01
3.07693511e-01 1.64554808e-02 -4.29110736e-01 -2.35082611e-01
-1.30331501e-01 5.88924229e-01 8.88538718e-01 -1.11781132e+00
-1.01189160e+00 -3.46706420e-01 -4.33887631e-01 2.82539666e-01
-1.00823484e-01 -9.07260478e-02 -4.56778467e-01 -6.64825857e-01
4.69361752e-01 -9.56692040e-01 -3.88731688e-01 2.71434262e-02
-2.06818357e-02 -4.24719006e-01 1.09918022e+00 -5.56294739e-01
1.51512754e+00 -2.01625681e+00 3.58212084e-01 2.71415800e-01
-1.48040786e-01 2.93014884e-01 2.02938929e-01 2.80050308e-01
9.69700962e-02 3.59919846e-01 -3.54946196e-01 2.24172682e-01
-4.73410547e-01 5.23648441e-01 -3.53922486e-01 2.09716842e-01
8.66423920e-03 7.66608655e-01 -6.30318820e-01 -7.38015532e-01
2.83534557e-01 2.70701110e-01 -4.18611020e-01 2.56830543e-01
-4.53725547e-01 6.98287785e-01 -9.25674975e-01 6.62160873e-01
6.96644843e-01 -3.33693683e-01 -5.27062640e-02 -3.71283740e-01
-9.49264783e-03 -2.62614608e-01 -1.50550699e+00 9.96815264e-01
-4.74902749e-01 3.79969597e-01 -1.49604186e-01 -1.27619588e+00
1.29559767e+00 -1.04013667e-01 3.91388208e-01 -5.55726290e-01
-1.83870792e-01 3.63611281e-01 1.11763306e-01 -9.12330627e-01
2.12359309e-01 -1.89256981e-01 2.47852162e-01 4.44510728e-01
-2.94909477e-01 -3.62713009e-01 -1.73713610e-01 -4.16344464e-01
1.04683824e-03 -1.23278841e-01 7.92643428e-01 -4.23151821e-01
6.14100277e-01 5.06576933e-02 1.05358934e+00 6.99392676e-01
9.69015285e-02 4.37251091e-01 -2.21261382e-02 -6.88539982e-01
-9.06210124e-01 -6.02394462e-01 -5.79427540e-01 1.09541154e+00
3.57836276e-01 1.71758890e-01 -4.20403481e-01 -6.04190469e-01
-1.08981747e-02 5.16899645e-01 -4.11744177e-01 -2.43404523e-01
-5.04976630e-01 -1.23081362e+00 2.23766938e-01 2.44166166e-01
1.05639827e+00 -9.75169599e-01 -2.61558205e-01 1.09217763e-01
-1.30294159e-01 -5.25784254e-01 -2.17565745e-01 7.17199296e-02
-1.36565924e+00 -1.00665092e+00 -2.38114581e-01 -9.41055357e-01
4.69235718e-01 5.48853099e-01 1.03708005e+00 4.91726130e-01
-1.45703077e-01 -1.87734455e-01 -3.80075634e-01 -6.83477044e-01
-4.51538086e-01 2.95082629e-01 -1.94379147e-02 -3.07744723e-02
1.81090951e-01 -2.97584802e-01 -3.72071207e-01 6.10663950e-01
-1.00549006e+00 1.48617655e-01 5.30449390e-01 1.12170899e+00
7.67648697e-01 7.52288342e-01 6.75627947e-01 -8.03559542e-01
2.47395322e-01 -7.04019547e-01 -4.89140213e-01 6.87435389e-01
-7.53040731e-01 1.66376919e-01 7.47782826e-01 -3.92670095e-01
-1.22566414e+00 4.51582521e-02 -2.00020045e-01 -2.02812269e-01
8.78903419e-02 6.89666867e-01 -4.81339514e-01 -9.71128568e-02
7.80767083e-01 2.91447997e-01 9.62793753e-02 -6.37389004e-01
9.14766043e-02 8.47056687e-01 -2.00191289e-02 -6.49292827e-01
4.89431888e-01 1.69708505e-01 2.45060697e-01 -9.92444694e-01
-9.43771839e-01 -4.32666451e-01 -7.88230658e-01 -2.96242744e-01
7.15295374e-01 -7.89674520e-01 -2.32799858e-01 7.81187415e-01
-9.71246302e-01 -1.58192188e-01 7.50253163e-03 5.68964243e-01
-1.03170738e-01 1.04531065e-01 -8.09795335e-02 -9.32640016e-01
-3.64814937e-01 -1.41192055e+00 7.81002402e-01 8.84862006e-01
2.73626983e-01 -1.39370644e+00 -1.73562676e-01 4.47362274e-01
2.62581676e-01 -6.87733963e-02 1.27722275e+00 -5.63100100e-01
-6.69722617e-01 2.54039764e-01 3.63779813e-02 3.83368254e-01
3.90681446e-01 5.93631327e-01 -9.66609240e-01 -9.04220790e-02
1.97333857e-01 -9.31304693e-02 1.08281672e+00 8.86134505e-01
1.63808310e+00 -4.36073959e-01 -6.76013708e-01 8.80102992e-01
1.42328072e+00 4.94960994e-01 6.76738083e-01 2.73081452e-01
7.29159176e-01 6.92167044e-01 9.90302801e-01 3.85362715e-01
2.14638680e-01 6.13703430e-01 3.64775926e-01 -1.95140094e-01
1.68239236e-01 -8.77969936e-02 -5.14198393e-02 1.00119960e+00
-8.76993015e-02 -2.20082879e-01 -9.46039677e-01 5.94926125e-04
-1.98337913e+00 -1.15168941e+00 -1.27290934e-01 2.10736227e+00
7.86906540e-01 -2.33011320e-01 -5.27797863e-02 9.24618617e-02
8.94628227e-01 4.54644337e-02 -9.54484642e-01 -2.07821876e-01
-2.98741609e-01 -2.73894042e-01 3.23192239e-01 6.63280964e-01
-9.61528718e-01 8.93463254e-01 6.82966566e+00 1.38026261e+00
-1.61943245e+00 -1.05580442e-01 8.68106842e-01 1.09982744e-01
-5.38850605e-01 -1.36553839e-01 -1.24583888e+00 6.36738896e-01
2.44382203e-01 1.07243787e-02 4.24471229e-01 9.52988744e-01
2.78605700e-01 -2.75416017e-01 -6.93962336e-01 1.20020616e+00
-3.08637265e-02 -1.63673472e+00 4.60129440e-01 -8.34821463e-02
9.95965004e-01 7.95490295e-02 5.03399633e-02 1.74373120e-01
8.55232403e-02 -1.16447186e+00 4.10039663e-01 8.67073834e-01
2.43969440e-01 -7.41070807e-01 6.07112527e-01 1.00390565e+00
-1.17920327e+00 -3.99614841e-01 -6.14321768e-01 2.51479098e-03
-6.95233420e-02 9.98950303e-01 -8.35261226e-01 7.18863606e-01
7.57695377e-01 8.43982399e-01 -6.13627613e-01 1.12006104e+00
-4.60939966e-02 7.48097003e-01 -2.20875487e-01 -2.43991330e-01
6.92433864e-02 -5.50995708e-01 4.91010219e-01 9.72825289e-01
2.21125185e-01 1.36682555e-01 4.67661321e-01 5.41245162e-01
2.99787253e-01 1.65366530e-01 -3.31556678e-01 3.12323928e-01
8.78389835e-01 9.82399046e-01 -5.69767058e-01 -2.97898538e-02
-2.35610120e-02 3.73598039e-01 2.02897564e-01 3.18438858e-01
-7.67302275e-01 1.19260728e-01 5.74205041e-01 -2.17830539e-01
1.56401936e-02 -1.38592556e-01 -7.20341623e-01 -1.04208815e+00
-2.53590308e-02 -1.02800798e+00 7.22141325e-01 -6.62761927e-01
-1.44090796e+00 3.68726313e-01 1.33747503e-01 -1.19153726e+00
-1.98114678e-01 -6.37185156e-01 -7.24376321e-01 9.74800467e-01
-1.57258189e+00 -1.04077435e+00 -3.24618787e-01 3.80109340e-01
8.27155769e-01 -6.83920860e-01 4.22308654e-01 1.19410656e-01
-9.13888931e-01 2.24444330e-01 5.29737175e-01 -3.52140218e-01
6.31267428e-01 -6.58211470e-01 -2.08182588e-01 7.71302581e-01
8.52546543e-02 5.94824672e-01 5.63943386e-01 -6.29130483e-01
-1.22733760e+00 -1.20901012e+00 6.04999185e-01 -2.81202272e-02
1.29849240e-01 3.02766144e-01 -1.19280899e+00 4.25405979e-01
-3.55311394e-01 -2.30161443e-01 6.47484601e-01 2.07637101e-01
5.15070651e-03 -6.41636848e-01 -1.23326695e+00 5.10221362e-01
6.60870671e-01 -1.76978528e-01 -3.51092964e-01 4.29845572e-01
6.44609213e-01 -3.98543119e-01 -6.31688893e-01 6.07269466e-01
6.60160720e-01 -8.54977012e-01 1.00217080e+00 -6.11093640e-01
2.73315132e-01 -4.91229832e-01 -3.04509848e-01 -1.17220974e+00
-5.46396852e-01 1.21576162e-02 -4.31704335e-02 1.20684338e+00
3.82337004e-01 -8.56307507e-01 4.88529801e-01 6.05826437e-01
-1.34062231e-01 -1.18221211e+00 -6.54869020e-01 -6.15884721e-01
2.45271623e-01 -3.23894054e-01 1.02554142e+00 1.02608359e+00
-8.10432613e-01 3.72523040e-01 -5.06818950e-01 2.85084158e-01
4.43380117e-01 3.87244850e-01 7.76997626e-01 -1.55528319e+00
-4.22408074e-01 -7.20185399e-01 -1.24976188e-01 -1.26546311e+00
7.45574310e-02 -8.58942628e-01 -2.39945173e-01 -1.60306060e+00
5.47946155e-01 -1.07011294e+00 1.25389978e-01 3.32785815e-01
-4.65025514e-01 -2.13502139e-01 6.82068542e-02 7.26239681e-01
-8.89267996e-02 4.97195512e-01 1.51472795e+00 -1.64651424e-01
-4.40626472e-01 5.18026114e-01 -7.82129347e-01 7.96094477e-01
9.86197650e-01 -5.30357838e-01 -5.02506435e-01 -6.34524643e-01
3.38980615e-01 -2.14236245e-01 1.94153324e-01 -6.11702621e-01
3.05260122e-01 -8.42279196e-01 4.56286401e-01 -5.40450394e-01
4.71476614e-02 -9.98708189e-01 5.62124848e-01 7.31555521e-01
-4.42556515e-02 -2.09639058e-01 -5.47825582e-02 3.39910001e-01
-2.59023845e-01 -7.33690619e-01 1.24840999e+00 -2.40162194e-01
-9.88477111e-01 5.76299846e-01 2.81533320e-03 6.56631812e-02
1.09160471e+00 -4.11403775e-01 -2.72904597e-02 -4.08595838e-02
-3.85878503e-01 3.87814194e-01 2.24858239e-01 3.75932902e-01
2.40335435e-01 -1.19657409e+00 -8.09511423e-01 2.47584790e-01
-1.91148922e-01 3.32751006e-01 3.44447017e-01 5.22415042e-01
-4.92183387e-01 8.14607590e-02 -1.55000985e-01 -1.01849425e+00
-1.26669049e+00 2.95770943e-01 7.28159368e-01 -1.80049956e-01
-1.28897965e-01 7.92496502e-01 2.73626000e-01 -5.67996442e-01
1.34832636e-01 -2.99688727e-01 -4.79369909e-01 9.17745009e-02
4.55397367e-01 5.96410394e-01 -1.91644952e-02 -4.32298809e-01
-2.67813414e-01 1.15068257e+00 1.44415170e-01 3.05523902e-01
1.18757832e+00 -2.39169106e-01 -3.55388284e-01 5.88362277e-01
1.00855684e+00 -2.10471243e-01 -1.19389737e+00 -5.29516995e-01
-3.96010697e-01 -4.10022736e-01 2.10016161e-01 -7.56667137e-01
-1.17472064e+00 9.06408012e-01 4.98057961e-01 3.39332432e-01
1.32467699e+00 2.34839804e-02 4.59589183e-01 6.01192236e-01
5.04470289e-01 -1.26614189e+00 1.46511793e-02 6.77432716e-01
1.02980459e+00 -1.63771498e+00 3.10945570e-01 -3.14763606e-01
-9.21898007e-01 1.28376305e+00 6.09779537e-01 2.47803912e-01
9.48130786e-01 -5.96987791e-02 2.50909161e-02 7.04803988e-02
-5.14144540e-01 8.79906490e-02 6.61544442e-01 6.08946621e-01
1.87677354e-01 4.62320149e-02 -2.74956018e-01 5.66365898e-01
-2.14540064e-01 -2.61739969e-01 -2.08254039e-01 3.49765956e-01
-1.02255976e+00 -1.08042526e+00 -5.84877372e-01 6.00629568e-01
-7.29365870e-02 6.49924874e-02 -1.41945779e-01 7.21481860e-01
4.54230517e-01 8.38384986e-01 4.30747084e-02 -3.87564659e-01
-4.40902151e-02 -2.09616512e-01 3.44693214e-01 -6.16562545e-01
-5.43469369e-01 -1.45104051e-01 -3.51680249e-01 -6.00206405e-02
-5.39615810e-01 -6.47990406e-01 -1.36675155e+00 -4.03684258e-01
-5.31998932e-01 3.69380713e-01 6.04164839e-01 1.28285480e+00
2.69044191e-01 2.79933184e-01 1.06551480e+00 -6.23302042e-01
-4.57520157e-01 -7.83680081e-01 -6.24334872e-01 -2.06417903e-01
1.70853540e-01 -7.22709894e-01 -4.14698988e-01 1.02553360e-01] | [9.873506546020508, 0.09700153023004532] |
ccbbc147-a3f5-4710-a885-151edb67ba7b | storytrans-non-parallel-story-author-style | 2208.13423 | null | https://arxiv.org/abs/2208.13423v2 | https://arxiv.org/pdf/2208.13423v2.pdf | StoryTrans: Non-Parallel Story Author-Style Transfer with Discourse Representations and Content Enhancing | Non-parallel text style transfer is an important task in natural language generation. However, previous studies concentrate on the token or sentence level, such as sentence sentiment and formality transfer, but neglect long style transfer at the discourse level. Long texts usually involve more complicated author linguistic preferences such as discourse structures than sentences. In this paper, we formulate the task of non-parallel story author-style transfer, which requires transferring an input story into a specified author style while maintaining source semantics. To tackle this problem, we propose a generation model, named StoryTrans, which leverages discourse representations to capture source content information and transfer them to target styles with learnable style embeddings. We use an additional training objective to disentangle stylistic features from the learned discourse representation to prevent the model from degenerating to an auto-encoder. Moreover, to enhance content preservation, we design a mask-and-fill framework to explicitly fuse style-specific keywords of source texts into generation. Furthermore, we constructed new datasets for this task in Chinese and English, respectively. Extensive experiments show that our model outperforms strong baselines in overall performance of style transfer and content preservation. | ['Juan Liu', 'Minlie Huang', 'Jian Guan', 'Xuekai Zhu'] | 2022-08-29 | null | null | null | null | ['text-style-transfoer'] | ['natural-language-processing'] | [ 4.16517258e-01 2.97661722e-01 -2.23638222e-01 -6.26025200e-01
-7.66917467e-01 -7.75043547e-01 9.28513587e-01 -2.08666176e-01
-3.15431476e-01 9.95738268e-01 8.96589935e-01 -1.52148694e-01
5.15031159e-01 -8.95988643e-01 -7.53445685e-01 -4.09102887e-01
7.87569642e-01 3.72384548e-01 -2.28281066e-01 -5.32254398e-01
1.96621925e-01 -1.67149499e-01 -7.37763643e-01 5.97388566e-01
1.16242564e+00 6.44056737e-01 3.11201125e-01 5.04951239e-01
-5.10708272e-01 1.03816676e+00 -9.34791327e-01 -7.19207227e-01
-1.27233222e-01 -8.39645982e-01 -1.02537835e+00 2.64072478e-01
3.18293720e-01 -4.47269499e-01 -2.35700354e-01 8.80928516e-01
5.03121018e-01 9.57964584e-02 9.98021901e-01 -9.83382583e-01
-1.41128993e+00 1.04395521e+00 -4.60721970e-01 -1.60756856e-01
3.39204937e-01 2.41248325e-01 1.14590228e+00 -8.40362430e-01
7.70668149e-01 1.58988690e+00 2.41047412e-01 9.89809752e-01
-1.31798625e+00 -6.26908958e-01 3.65661085e-01 -1.56999230e-01
-7.90565073e-01 -4.06094849e-01 1.18627691e+00 -3.39010358e-01
4.95426953e-01 3.61837931e-02 3.45108569e-01 1.66913950e+00
-1.17600113e-02 1.14805377e+00 1.01830471e+00 -3.04321766e-01
-3.53385955e-02 2.54420847e-01 -1.08343251e-01 3.33009362e-01
-2.37745512e-02 -2.78828830e-01 -5.05784869e-01 2.84217417e-01
8.17864895e-01 -2.54747421e-01 -3.32695574e-01 1.03183843e-01
-1.43028736e+00 1.07835031e+00 3.52362901e-01 2.62723088e-01
-2.70629320e-02 1.44140124e-02 7.46773303e-01 5.10092616e-01
7.21609533e-01 7.72941172e-01 -2.61686891e-01 -5.42399660e-02
-7.53695905e-01 4.85792279e-01 7.04389155e-01 1.44310701e+00
5.70960641e-01 1.80828467e-01 -9.25528765e-01 9.58876252e-01
1.25780359e-01 4.54272389e-01 6.33040786e-01 -6.16325557e-01
9.56673563e-01 6.74581110e-01 -1.14420079e-01 -6.98744535e-01
7.12162033e-02 -3.64099860e-01 -1.04811740e+00 -3.05092365e-01
1.85296178e-01 -5.61318696e-01 -5.60068607e-01 1.99169660e+00
9.29183960e-02 -1.76511407e-01 3.23468834e-01 8.64355862e-01
9.77706969e-01 9.98099327e-01 8.23003724e-02 -6.30644783e-02
1.46310353e+00 -1.33349419e+00 -8.03663313e-01 -3.38491112e-01
6.71060264e-01 -9.00598347e-01 1.64083350e+00 -6.45072386e-02
-1.30673230e+00 -6.19569540e-01 -9.81295347e-01 -6.17313147e-01
-1.71098217e-01 3.84067178e-01 2.05671862e-01 2.31097400e-01
-6.78271353e-01 5.44561207e-01 -4.45953906e-01 -7.92701542e-02
5.01441836e-01 -1.07239909e-01 -5.40038571e-02 -1.77874956e-02
-1.45661962e+00 7.35735834e-01 5.98108351e-01 -2.04757094e-01
-5.88978112e-01 -9.04004931e-01 -1.23082817e+00 1.78597658e-03
2.77896702e-01 -1.08768678e+00 1.41752195e+00 -1.27263236e+00
-1.96789217e+00 8.54105413e-01 -1.84071869e-01 -1.35160208e-01
6.06280446e-01 -5.06808221e-01 -1.61903143e-01 -8.71409774e-02
2.72947878e-01 6.68197691e-01 7.81255603e-01 -1.26250434e+00
-4.72484738e-01 5.28952926e-02 3.67264926e-01 4.94758964e-01
-7.11654603e-01 8.60310346e-02 -1.35014966e-01 -1.38755465e+00
-5.63828051e-01 -6.97034121e-01 7.53764063e-02 -2.76862741e-01
-5.70485175e-01 -5.79372525e-01 6.45321667e-01 -8.17715585e-01
1.25268769e+00 -1.86994016e+00 6.29514873e-01 -3.51860106e-01
-7.86313228e-03 7.25238258e-03 -3.90941352e-01 4.75011200e-01
2.66419470e-01 1.27051502e-01 -3.80941600e-01 -7.17050254e-01
2.03175351e-01 1.80862620e-01 -7.12490141e-01 -1.17047086e-01
7.52010584e-01 1.09053469e+00 -1.04472077e+00 -5.28507113e-01
-2.14915648e-01 2.66895205e-01 -8.06697845e-01 5.86828172e-01
-5.65090954e-01 6.78016424e-01 -5.22546351e-01 8.37394819e-02
4.44706708e-01 -1.30497918e-01 -2.00727526e-02 -1.60249665e-01
4.08838168e-02 8.69545937e-01 -7.17262328e-01 2.07934880e+00
-9.29185331e-01 3.73157918e-01 -1.63516745e-01 -8.34216952e-01
1.00694644e+00 4.26343590e-01 -9.38877240e-02 -4.38764304e-01
2.52223313e-01 3.07974182e-02 -9.95969400e-02 -2.50851512e-01
8.92645180e-01 -4.14752901e-01 -5.76129496e-01 8.23046982e-01
2.85804927e-01 -4.28959966e-01 3.16923529e-01 2.12742582e-01
5.46652615e-01 3.84932131e-01 3.25915515e-02 -4.01645869e-01
5.34195185e-01 -2.84293950e-01 5.28715968e-01 3.61073703e-01
3.05987656e-01 8.06079209e-01 7.27614939e-01 6.71437383e-02
-1.01261389e+00 -9.40485835e-01 3.18597317e-01 1.19647729e+00
9.19714272e-02 -3.82301152e-01 -9.87571537e-01 -1.00840068e+00
-2.04799771e-01 1.09736514e+00 -6.31389081e-01 -3.99574399e-01
-1.00280416e+00 -3.76528621e-01 5.74151039e-01 6.38681769e-01
5.73270500e-01 -1.31916225e+00 -2.01299295e-01 3.20213258e-01
-4.24256951e-01 -1.01914620e+00 -1.27254117e+00 -4.26383436e-01
-6.60897732e-01 -6.78785801e-01 -1.08855271e+00 -9.63629305e-01
6.33920908e-01 -7.51991421e-02 1.31584978e+00 -2.55958170e-01
3.67196411e-01 1.10400738e-02 -4.90864426e-01 -5.49234211e-01
-6.10817015e-01 7.34438419e-01 -2.75125235e-01 7.15992600e-02
1.18831836e-01 -4.66123581e-01 -4.25764263e-01 -1.59670323e-01
-9.97194350e-01 5.48225045e-01 5.18455029e-01 1.17118931e+00
2.32604250e-01 -4.41015750e-01 1.00535715e+00 -1.30375886e+00
1.11779523e+00 -4.56389874e-01 -1.54068917e-01 2.29711294e-01
-2.95159549e-01 3.74160558e-01 9.91787553e-01 -5.24339676e-01
-1.58912468e+00 -3.27256203e-01 -7.49079213e-02 -3.06692809e-01
-1.58665655e-03 5.60167670e-01 -7.73972929e-01 7.80703068e-01
3.79781425e-01 3.39794278e-01 -9.62354764e-02 -5.35171151e-01
7.33465314e-01 6.48434043e-01 4.49662060e-01 -1.01238906e+00
9.84900177e-01 6.04560561e-02 -4.41794187e-01 -6.96343124e-01
-1.27624428e+00 3.64444517e-02 -4.78471041e-01 1.54603094e-01
8.98856163e-01 -9.87290263e-01 -6.81728125e-02 4.59347665e-01
-1.64768958e+00 -3.83746982e-01 -5.15433848e-01 4.29589897e-02
-6.05006933e-01 1.42625183e-01 -8.76799583e-01 -3.94201934e-01
-6.10671222e-01 -8.86297405e-01 1.20631790e+00 3.45332056e-01
-5.54638326e-01 -1.33577847e+00 1.37139484e-01 2.70867556e-01
4.27372128e-01 2.83086985e-01 1.02439106e+00 -5.19254446e-01
-2.65262097e-01 1.85579240e-01 -3.44319731e-01 5.16299367e-01
5.66598892e-01 -2.96631396e-01 -8.77780139e-01 -1.88507795e-01
4.94788438e-02 -5.89622438e-01 9.46796298e-01 -4.56776954e-02
1.15171921e+00 -8.27760398e-01 4.64771166e-02 6.13392591e-01
9.26008046e-01 -2.00481042e-01 5.53604484e-01 1.15188614e-01
1.10027182e+00 9.46664393e-01 4.56951916e-01 2.23191291e-01
7.24923790e-01 5.02472162e-01 -2.26056546e-01 -1.80359915e-01
-3.86972219e-01 -7.21993983e-01 6.64436877e-01 1.12027180e+00
1.39857665e-01 -3.62081617e-01 -3.67559642e-01 5.66881478e-01
-1.63175666e+00 -9.06040490e-01 2.36291438e-01 1.67548478e+00
1.67349470e+00 2.06275314e-01 1.23623841e-01 -2.41757616e-01
5.73089719e-01 4.05059695e-01 -4.65958565e-01 -5.26973367e-01
-2.75448263e-01 8.12834203e-02 -1.86018795e-02 4.29695696e-01
-1.05383694e+00 1.23916328e+00 4.82233715e+00 9.57784653e-01
-1.22381270e+00 1.59816489e-01 6.99256182e-01 -1.50164902e-01
-8.62317145e-01 -8.78853276e-02 -8.81006300e-01 7.18030393e-01
6.41182780e-01 -5.63670456e-01 2.63476998e-01 8.13451052e-01
2.20945090e-01 5.71444929e-01 -1.42227697e+00 6.52725399e-01
2.61319518e-01 -1.30432844e+00 5.27123749e-01 -1.74989343e-01
1.06443834e+00 -5.62002718e-01 3.97918411e-02 6.27017438e-01
4.04833913e-01 -1.05924273e+00 1.05028939e+00 3.62482339e-01
1.04949331e+00 -7.76962638e-01 4.89089638e-01 2.49816120e-01
-9.72083867e-01 3.62242550e-01 -2.38940448e-01 -2.10325405e-01
4.88459200e-01 4.56768870e-01 -5.72313070e-01 6.49270535e-01
3.70743833e-02 9.83386397e-01 -4.29724753e-01 3.14507425e-01
-6.21921241e-01 6.21614397e-01 1.34127885e-01 -1.45231113e-01
2.90096939e-01 -1.32621452e-01 4.00741845e-01 1.39984739e+00
3.47667873e-01 -5.94052412e-02 1.70414075e-01 1.36035979e+00
-5.40336967e-01 1.34809151e-01 -6.33369267e-01 -3.70554477e-01
4.51054335e-01 1.11955321e+00 -1.85883939e-01 -4.49785888e-01
-4.20640945e-01 1.42230880e+00 4.51698869e-01 3.60530823e-01
-6.48428082e-01 -6.51739478e-01 7.37060130e-01 -7.68583491e-02
6.13968372e-02 -4.66548093e-02 -6.18997455e-01 -1.52239525e+00
1.90656349e-01 -9.75480974e-01 1.19924404e-01 -3.93789709e-01
-1.67943621e+00 6.23174191e-01 -1.15421154e-01 -1.15930772e+00
-3.52181584e-01 -4.78288293e-01 -1.22864008e+00 1.22484481e+00
-1.63839555e+00 -1.69619453e+00 7.47906938e-02 3.51674974e-01
1.11743414e+00 -2.80382693e-01 6.29281700e-01 1.24234958e-02
-5.65776289e-01 9.52522337e-01 -3.32542360e-02 3.56882960e-01
1.04660308e+00 -1.47203135e+00 4.93502915e-01 8.00352573e-01
-1.89334854e-01 6.73515439e-01 5.87948203e-01 -7.15016842e-01
-9.90549088e-01 -1.51188552e+00 1.19154048e+00 -4.99556035e-01
6.36479139e-01 -6.74179256e-01 -9.48444188e-01 7.16172576e-01
8.88183057e-01 -6.29828930e-01 7.39324450e-01 9.26198587e-02
-2.83963352e-01 1.83657527e-01 -7.14184642e-01 1.01430488e+00
1.09925246e+00 -5.65222323e-01 -9.54092681e-01 1.77728295e-01
1.19765413e+00 -4.55419004e-01 -7.88516343e-01 1.55817464e-01
1.85045585e-01 -4.06242907e-01 5.60948372e-01 -7.72026360e-01
1.31324625e+00 -3.83122489e-02 2.03525916e-01 -1.82264924e+00
-2.22472414e-01 -7.51167357e-01 -7.64189214e-02 2.02699471e+00
5.26474476e-01 -2.51418382e-01 5.26201606e-01 3.29887629e-01
-3.99440438e-01 -6.37956202e-01 -5.72052658e-01 -6.61482394e-01
7.71302104e-01 2.81923711e-01 8.46187472e-01 1.01437891e+00
2.92970724e-02 1.15554392e+00 -4.53609645e-01 -3.55045676e-01
3.30706805e-01 3.23805451e-01 7.84353733e-01 -8.81917477e-01
-4.22788799e-01 -7.49044299e-01 4.75666314e-01 -1.27678418e+00
6.25997722e-01 -1.05704057e+00 1.39064178e-01 -1.52066195e+00
1.56119600e-01 -3.15562844e-01 7.40855262e-02 1.68600038e-01
-7.02594161e-01 -1.42942250e-01 2.42576927e-01 -1.17699511e-03
-4.39962149e-01 1.26774561e+00 1.75827968e+00 -2.31332183e-01
-8.43712762e-02 -1.42306432e-01 -1.16029572e+00 6.13254011e-01
8.45100045e-01 -1.85062662e-01 -6.84894145e-01 -9.22646642e-01
1.28140986e-01 1.06247999e-02 4.47011627e-02 -4.28111076e-01
-2.16723487e-01 -4.41180140e-01 2.54564047e-01 -2.49128878e-01
2.06039593e-01 -2.99501836e-01 -5.80384016e-01 8.57739970e-02
-8.83885503e-01 5.15867770e-02 6.05396405e-02 2.59283990e-01
-3.53684783e-01 -2.70776600e-01 6.48352265e-01 -1.19461574e-01
-1.16739549e-01 3.35074723e-01 -1.62499502e-01 5.55797577e-01
7.34185934e-01 2.19583288e-01 -4.93122190e-01 -4.11256135e-01
-2.29317352e-01 4.89670932e-01 5.98869383e-01 8.26831758e-01
4.85340536e-01 -1.83549488e+00 -1.17375565e+00 1.09380871e-01
2.34252632e-01 4.38405275e-01 1.57443985e-01 3.84726852e-01
-2.14921534e-01 2.56836325e-01 -1.03938185e-01 -1.10115103e-01
-8.98169518e-01 4.16114122e-01 -6.37682006e-02 -5.00882149e-01
-4.69820350e-01 8.80353808e-01 7.66242206e-01 -6.42877698e-01
-2.44508311e-02 -3.09021533e-01 -3.24259818e-01 1.77609757e-01
6.46764696e-01 1.17232107e-01 -3.68715346e-01 -4.26323116e-01
1.67168617e-01 2.83369690e-01 -4.04823661e-01 -3.03472072e-01
1.11015415e+00 -3.01223576e-01 -1.67378888e-01 6.17038846e-01
1.19464672e+00 8.87734741e-02 -1.46336639e+00 -4.55155283e-01
9.85525250e-02 -1.28882766e-01 -2.15101108e-01 -6.88862979e-01
-8.42264354e-01 1.13578665e+00 -1.81128174e-01 -1.37603596e-01
9.36841607e-01 1.82077661e-02 1.26459265e+00 1.64805099e-01
-2.05786869e-01 -1.21514678e+00 3.82735014e-01 1.00838542e+00
1.41853070e+00 -1.10785115e+00 -5.74360013e-01 -2.90031612e-01
-9.89147365e-01 9.80672598e-01 1.06972802e+00 -2.85947949e-01
2.60924786e-01 4.57766000e-04 8.17304477e-03 2.35310555e-01
-8.15206528e-01 1.49979979e-01 3.84346753e-01 5.25069714e-01
9.23856735e-01 1.62285730e-01 -3.97908330e-01 1.18737447e+00
-6.75183177e-01 -2.21923385e-02 5.17559469e-01 7.04892933e-01
-1.80348381e-01 -1.39360464e+00 -6.85258806e-02 2.44652554e-01
-3.38334590e-01 -4.21511233e-01 -6.19837284e-01 3.37881118e-01
-2.88214050e-02 6.47514522e-01 1.82216525e-01 -2.28586152e-01
4.13009644e-01 1.36149123e-01 5.23958325e-01 -1.02658880e+00
-6.96538210e-01 8.45984071e-02 2.78095454e-01 -2.46451749e-03
-1.70067325e-01 -6.20692074e-01 -1.04376161e+00 -3.73553246e-01
-4.57052402e-02 2.80966491e-01 9.59700495e-02 1.07771838e+00
2.96903282e-01 9.99521732e-01 7.09481359e-01 -7.49798536e-01
-7.38706708e-01 -1.37372410e+00 -4.07248467e-01 5.94905138e-01
2.49937013e-01 -2.98812747e-01 3.06780245e-02 4.08762932e-01] | [11.709510803222656, 9.464496612548828] |
86adcaad-4a0e-4f8f-8e16-f4a89112c16f | extending-rnn-t-based-speech-recognition | 2207.13965 | null | https://arxiv.org/abs/2207.13965v1 | https://arxiv.org/pdf/2207.13965v1.pdf | Extending RNN-T-based speech recognition systems with emotion and language classification | Speech transcription, emotion recognition, and language identification are usually considered to be three different tasks. Each one requires a different model with a different architecture and training process. We propose using a recurrent neural network transducer (RNN-T)-based speech-to-text (STT) system as a common component that can be used for emotion recognition and language identification as well as for speech recognition. Our work extends the STT system for emotion classification through minimal changes, and shows successful results on the IEMOCAP and MELD datasets. In addition, we demonstrate that by adding a lightweight component to the RNN-T module, it can also be used for language identification. In our evaluations, this new classifier demonstrates state-of-the-art accuracy for the NIST-LRE-07 dataset. | ['George Saon', 'Samuel Thomas', 'Hong-Kwang Kuo', 'Matheus Damasceno', 'Edmilson Morais', 'Hagai Aronowitz', 'Zvi Kons'] | 2022-07-28 | null | null | null | null | ['emotion-classification', 'emotion-classification'] | ['computer-vision', 'natural-language-processing'] | [ 1.31978855e-01 -2.94285685e-01 1.43282086e-01 -6.61736071e-01
-9.31977451e-01 -5.04744649e-01 4.88618314e-01 -1.36384934e-01
-4.55594212e-01 1.28019929e-01 2.36333251e-01 -4.45003361e-01
5.47656119e-01 3.94074507e-02 -1.31823555e-01 -5.27415156e-01
3.62035543e-01 4.70209509e-01 -2.34399602e-01 -3.25507849e-01
-2.19928443e-01 5.28390408e-01 -1.51133406e+00 4.90774602e-01
3.83284062e-01 1.44949806e+00 -1.94538295e-01 9.94208992e-01
-4.91637617e-01 1.13225555e+00 -7.82356143e-01 -3.14616382e-01
-5.14919646e-02 -3.23561370e-01 -7.95512497e-01 -1.47800818e-01
1.01037011e-01 2.29521859e-02 -2.65977710e-01 6.65522218e-01
9.29349542e-01 2.74229258e-01 6.28707111e-01 -9.97263789e-01
-3.27402174e-01 7.68692315e-01 7.50022158e-02 -2.43848935e-01
5.07783592e-01 -4.01543558e-01 5.65050542e-01 -1.20899606e+00
8.10333863e-02 1.32568407e+00 7.81532884e-01 1.01800513e+00
-9.87299919e-01 -7.80120194e-01 -1.20294370e-01 1.40804484e-01
-1.29079485e+00 -1.14775062e+00 7.02507615e-01 -1.27364814e-01
1.63121283e+00 5.09345949e-01 3.19408298e-01 1.62975574e+00
-3.10845196e-01 1.00371695e+00 1.11703789e+00 -7.17687607e-01
2.93945551e-01 4.26053196e-01 4.19041038e-01 3.96667212e-01
-6.55090034e-01 -2.08260819e-01 -8.05444241e-01 -8.70051086e-02
8.51201639e-02 -3.54780614e-01 -7.85719976e-02 4.10448045e-01
-7.98882604e-01 6.23844087e-01 -5.32080770e-01 5.98139167e-01
-3.59972715e-01 2.42095347e-02 1.14496255e+00 8.16057146e-01
8.18984866e-01 1.80851985e-02 -5.91828823e-01 -9.13183689e-01
-8.71674895e-01 -5.95928252e-01 1.35894930e+00 7.80532062e-01
1.34311542e-01 7.07992077e-01 -1.61120787e-01 1.35152709e+00
2.33059064e-01 5.35086095e-01 9.92345035e-01 -4.24073964e-01
2.18261957e-01 3.53407979e-01 -1.94517031e-01 -4.55529660e-01
-3.60502243e-01 -3.09401095e-01 -8.70247483e-01 -1.15456097e-01
-5.86571768e-02 -2.47823954e-01 -9.35820401e-01 1.42827916e+00
-8.75511467e-02 3.08571696e-01 5.80153584e-01 5.08135200e-01
1.17103255e+00 8.87838662e-01 1.86005309e-02 -3.34306508e-01
1.28186417e+00 -1.03717327e+00 -1.01713479e+00 -1.81050256e-01
8.63734722e-01 -9.46538091e-01 9.75669920e-01 5.77675402e-01
-8.61820996e-01 -4.93589431e-01 -8.42986584e-01 -6.86310530e-02
-7.00587630e-01 6.01077437e-01 2.35111326e-01 1.35961759e+00
-1.20247960e+00 1.31098673e-01 -5.87239265e-01 -7.21089959e-01
-3.59684139e-01 4.50217605e-01 -4.65530068e-01 5.16649425e-01
-1.28710341e+00 1.02208531e+00 2.36309737e-01 2.81034052e-01
-6.18694663e-01 -1.18336558e-01 -8.79648328e-01 1.23640843e-01
1.26211807e-01 -1.19834043e-01 1.56148601e+00 -8.07731092e-01
-2.56762338e+00 9.29866850e-01 -6.62630677e-01 -6.64320469e-01
-2.90110745e-02 -9.47815701e-02 -1.04669380e+00 -2.46552303e-02
-5.55303037e-01 3.01022530e-01 9.21942234e-01 -7.56274462e-01
-3.67444605e-01 -2.29867786e-01 -8.37885678e-01 1.07767142e-01
-8.25487852e-01 8.82885516e-01 -3.72342616e-01 -5.56963801e-01
7.17083961e-02 -9.28481877e-01 1.78753942e-01 -9.04848814e-01
-3.54583412e-01 -4.50130731e-01 1.07474768e+00 -8.60961258e-01
1.42012393e+00 -2.39123464e+00 -9.91879776e-02 1.15783408e-01
-3.90962422e-01 5.81657529e-01 -2.24524602e-01 4.70460474e-01
-3.39530110e-01 2.52242267e-01 1.87095523e-01 -1.17343760e+00
3.35858196e-01 2.69601941e-01 -4.02954340e-01 1.20577417e-01
4.60960455e-02 8.20809662e-01 -3.16486061e-01 3.07214772e-03
4.55953300e-01 6.51082933e-01 2.31336445e-01 4.51781958e-01
2.07833126e-01 3.36411357e-01 -2.33449414e-02 5.96864700e-01
3.01020145e-01 3.56801808e-01 7.61344237e-03 7.59776160e-02
-2.08053619e-01 7.10999846e-01 -1.04263937e+00 1.39447188e+00
-8.44724476e-01 7.74875462e-01 3.14766645e-01 -1.04075646e+00
1.36827409e+00 1.18833709e+00 1.84703022e-01 -4.19238657e-01
6.79378748e-01 5.50115108e-01 -2.10599214e-01 -6.81032538e-01
6.08250558e-01 -1.44739613e-01 -3.23543400e-01 4.63881463e-01
5.41934133e-01 -1.37091681e-01 -2.52367586e-01 -3.00831646e-01
8.68002653e-01 -4.48342681e-01 1.59518555e-01 1.73684359e-01
7.78868079e-01 -5.16998529e-01 3.77709270e-01 6.48103774e-01
-1.44081995e-01 3.99320632e-01 1.59969732e-01 -7.01449513e-02
-8.64852071e-01 -4.71293777e-01 7.08693191e-02 1.36864841e+00
-7.85941005e-01 -5.16851604e-01 -7.51894236e-01 -5.42542279e-01
-4.96172249e-01 8.84702802e-01 -3.04013997e-01 -5.53450324e-02
-2.11786136e-01 -3.53116959e-01 1.41185892e+00 3.56085569e-01
3.17663282e-01 -1.19041681e+00 -2.09937673e-02 3.15137088e-01
-3.23306650e-01 -1.56574762e+00 -6.25632226e-01 7.66648531e-01
-4.23269212e-01 -2.49593914e-01 -6.77852869e-01 -1.04290545e+00
3.66950477e-03 -5.67945000e-03 7.69776523e-01 -4.27464902e-01
1.86714962e-01 7.49866784e-01 -7.38712966e-01 -6.82414234e-01
-1.03189409e+00 3.56989443e-01 4.15422171e-01 5.31901062e-01
6.84360266e-01 -4.95682508e-01 1.88535854e-01 3.10495913e-01
-5.83919227e-01 -4.27612722e-01 1.96794376e-01 7.91449964e-01
2.10456893e-01 -1.47590399e-01 6.99635804e-01 -3.23547959e-01
1.13656986e+00 -9.49743320e-04 -2.35301286e-01 4.79848385e-01
-4.16501611e-01 -1.62694529e-01 7.14056075e-01 -6.98719442e-01
-1.06865704e+00 2.88884640e-01 -8.11944306e-01 -7.02471197e-01
-5.36790550e-01 5.27663708e-01 -1.44738585e-01 -2.68982530e-01
2.74347425e-01 5.72631955e-01 7.05756918e-02 -7.57336438e-01
1.72159970e-01 1.61888301e+00 4.33038920e-01 -2.28142798e-01
1.39633372e-01 -1.45914122e-01 -4.89637583e-01 -1.18019640e+00
-4.32111681e-01 -8.38441551e-01 -5.18549144e-01 -2.86171854e-01
7.66807556e-01 -1.01563847e+00 -1.18716037e+00 9.75954235e-01
-1.29010558e+00 -2.82335162e-01 -6.98719472e-02 7.29124188e-01
-2.76275575e-01 2.93852061e-01 -8.98388982e-01 -1.61819816e+00
-9.51264262e-01 -1.12894201e+00 1.17556262e+00 8.06176066e-02
-4.19107884e-01 -1.00958657e+00 -5.63041531e-02 2.97405332e-01
6.99338198e-01 -3.90309095e-01 4.77208287e-01 -1.41166735e+00
4.55731452e-01 -5.35330772e-01 3.01260591e-01 1.01962340e+00
-5.24708219e-02 -3.95641066e-02 -1.57096100e+00 -1.41725764e-01
5.22154629e-01 -6.19577587e-01 8.25433433e-01 7.16587454e-02
9.70969379e-01 -1.66752189e-01 1.89522207e-01 4.33120579e-01
7.95629323e-01 5.65086126e-01 5.85048556e-01 -1.05137408e-01
5.43833375e-01 7.73223400e-01 1.79340124e-01 2.21639067e-01
2.14912489e-01 7.70920575e-01 -2.04186022e-01 2.05550957e-02
3.52546759e-02 2.29648873e-02 1.09211910e+00 1.77815783e+00
3.66512030e-01 -4.49649751e-01 -1.04230428e+00 2.44897485e-01
-1.76391757e+00 -9.11549926e-01 -1.81477174e-01 2.07638335e+00
6.50901914e-01 -1.72833726e-01 2.18502343e-01 3.99450988e-01
5.83450437e-01 -2.06755870e-03 -2.96900690e-01 -1.19286954e+00
-2.15175882e-01 4.90195721e-01 2.39116415e-01 4.25076723e-01
-9.80450332e-01 1.15981460e+00 7.12187243e+00 1.16746449e+00
-1.68262243e+00 3.92557383e-01 6.23199821e-01 -5.50404713e-02
2.70802528e-01 -5.23922384e-01 -8.73787344e-01 1.34036481e-01
1.70590568e+00 -3.51692475e-02 6.30320549e-01 8.93750727e-01
2.71748096e-01 2.03795776e-01 -9.60803270e-01 1.42923880e+00
4.74056870e-01 -6.73112929e-01 -1.84962705e-01 -2.77667284e-01
4.55201380e-02 2.59563416e-01 3.68875265e-02 6.90814734e-01
-9.52420160e-02 -1.16414332e+00 6.62075937e-01 1.52763456e-01
9.53476787e-01 -7.91410089e-01 8.97014976e-01 2.28307784e-01
-1.30069423e+00 -1.75920184e-04 1.35463685e-01 -7.65813328e-03
1.56945840e-01 4.45815474e-01 -1.12329304e+00 4.35717493e-01
5.28579772e-01 5.80033481e-01 -1.82606444e-01 4.53163743e-01
-1.41268075e-01 1.12217534e+00 -3.66564572e-01 -5.18761277e-01
5.05167432e-02 -2.66103707e-02 5.47983527e-01 1.74409902e+00
5.44282794e-01 -1.83269143e-01 1.03868239e-01 2.38343969e-01
-2.46463373e-01 5.04337132e-01 -5.62494755e-01 -3.44616890e-01
4.91750479e-01 1.40860152e+00 -4.36535180e-01 -5.26182890e-01
-2.09773898e-01 1.25530195e+00 1.41730472e-01 3.60029668e-01
-6.37791038e-01 -6.00006163e-01 6.84289634e-01 -6.99308455e-01
2.59597301e-01 -3.13046634e-01 -1.80474624e-01 -1.18110061e+00
1.05198231e-02 -1.06778145e+00 6.85316771e-02 -9.38398421e-01
-1.25266671e+00 1.23411918e+00 -5.48361063e-01 -1.11189747e+00
-5.99743247e-01 -7.81686962e-01 -4.74224716e-01 1.01757574e+00
-1.17555118e+00 -1.11298728e+00 1.02786243e-01 6.54402375e-01
5.98984480e-01 -5.87627411e-01 1.46023285e+00 4.89333302e-01
-8.50126088e-01 9.29404616e-01 2.47101068e-01 4.70323443e-01
7.52284586e-01 -1.12387145e+00 4.60033953e-01 6.42573476e-01
3.08877170e-01 4.41056788e-01 4.57268268e-01 -2.10455835e-01
-1.45559323e+00 -9.58286703e-01 1.16703570e+00 -3.72915268e-01
7.17851877e-01 -9.14636433e-01 -8.31967056e-01 6.05589330e-01
5.01537323e-01 -4.77472425e-01 1.04016936e+00 4.25012141e-01
-4.48307753e-01 -6.06302470e-02 -9.63214457e-01 4.98659432e-01
5.78934669e-01 -1.36518121e+00 -3.79863530e-01 2.09568795e-02
8.75556707e-01 -3.44904482e-01 -1.05449939e+00 3.19239765e-01
7.74060547e-01 -6.69919729e-01 6.06052816e-01 -3.84776026e-01
-2.82372892e-01 -1.59043483e-02 -4.12684202e-01 -1.46522057e+00
1.28941268e-01 -8.86172831e-01 -1.25164986e-01 1.65536177e+00
6.86336339e-01 -9.03177261e-01 3.30052793e-01 4.95593578e-01
-3.20885748e-01 -4.32509899e-01 -1.42657769e+00 -1.00502539e+00
-3.26921582e-01 -1.06101930e+00 4.26539242e-01 8.68327320e-01
3.72815460e-01 8.80113661e-01 -7.38652408e-01 4.21401002e-02
-1.59665626e-02 -4.14922327e-01 6.37649417e-01 -1.03946245e+00
-1.19740494e-01 -5.54412127e-01 -3.55852515e-01 -1.02788413e+00
4.77669358e-01 -9.30003464e-01 1.61004841e-01 -9.93509650e-01
-3.21713775e-01 7.44803175e-02 -3.35841000e-01 6.22983813e-01
4.12638038e-01 4.89214323e-02 1.45877272e-01 -4.55913544e-02
-4.54380333e-01 6.86949372e-01 2.87754208e-01 -9.86441746e-02
-3.38578731e-01 4.55560498e-02 -2.78174937e-01 5.46547711e-01
1.02859783e+00 -5.40430844e-01 -6.88570812e-02 -4.04931791e-02
-9.55993757e-02 1.75613374e-01 -1.23530827e-01 -1.11219025e+00
4.70239371e-01 4.79332298e-01 -3.82832587e-02 -7.40777671e-01
8.96359324e-01 -7.93783903e-01 2.29721963e-02 1.05774842e-01
-5.20771325e-01 -1.02227703e-02 4.83836979e-01 -1.43445106e-02
-5.36420703e-01 -3.93306345e-01 4.14473593e-01 1.74815282e-01
-5.23730814e-01 -7.58691356e-02 -1.08938074e+00 -3.30867290e-01
5.48826277e-01 -1.04001589e-01 -1.42507315e-01 -6.92013621e-01
-8.41139793e-01 -6.43887073e-02 -6.72846138e-02 7.31487334e-01
5.94123185e-01 -1.19099438e+00 -6.97899103e-01 4.89718050e-01
2.36834660e-01 -7.13647783e-01 9.21094790e-02 8.68094623e-01
1.62765399e-01 5.89911163e-01 2.95931578e-01 -4.96608883e-01
-1.90775490e+00 2.70978212e-01 6.60952806e-01 -2.91435957e-01
7.45474175e-02 7.76927531e-01 -5.46745062e-01 -8.45033586e-01
7.96679378e-01 -3.71525794e-01 -3.29432994e-01 3.07695180e-01
5.97583532e-01 2.17187420e-01 5.33620894e-01 -8.65069807e-01
-3.27987134e-01 2.83677280e-01 -1.19667687e-02 -5.33389032e-01
1.14299297e+00 -3.24381709e-01 -2.48671368e-01 1.20102942e+00
1.28602755e+00 6.56675473e-02 -7.00790808e-02 -2.13415787e-01
2.18508393e-01 3.05015713e-01 2.23522589e-01 -9.49259460e-01
-6.52145207e-01 1.10233665e+00 7.28477359e-01 6.45486534e-01
1.26110709e+00 -2.36428007e-01 9.30425167e-01 6.77638650e-01
1.94042504e-01 -1.48472786e+00 -1.66484743e-01 1.25339317e+00
8.51629555e-01 -1.10398674e+00 -8.63257587e-01 -2.25256652e-01
-7.82993972e-01 1.30174637e+00 2.58776724e-01 5.55140495e-01
7.00677395e-01 5.24988115e-01 5.15939176e-01 8.48776177e-02
-1.09372592e+00 -1.64033741e-01 3.49890828e-01 3.77190828e-01
8.36802185e-01 9.62310359e-02 9.52725559e-02 8.11526418e-01
-2.08688661e-01 -1.10221185e-01 3.13467145e-01 6.01840258e-01
-2.29568839e-01 -1.39257967e+00 -5.04013360e-01 3.37977886e-01
-6.45559967e-01 -3.90250891e-01 -8.26136649e-01 1.71276540e-01
-4.33060765e-01 1.53322375e+00 -1.95262715e-01 -1.02806771e+00
4.04206783e-01 9.82378364e-01 2.81578712e-02 -5.11193693e-01
-1.17426944e+00 3.49190444e-01 5.90323329e-01 -3.32547724e-01
-4.83066440e-01 -4.90889460e-01 -1.09329474e+00 -1.79536223e-01
-3.71203542e-01 4.16821629e-01 1.13408732e+00 9.19050634e-01
5.54391384e-01 6.17607474e-01 8.93979609e-01 -6.29926443e-01
-4.54489499e-01 -1.49744236e+00 -7.59023786e-01 4.33560237e-02
2.39691511e-01 -2.10482348e-03 -4.88337368e-01 -1.88542485e-01] | [13.899874687194824, 6.207845211029053] |
ddf56dd3-9480-4a9c-bddb-d95cfbb279b7 | native-language-identification-using-large-1 | null | null | https://aclanthology.org/L14-1051 | https://aclanthology.org/L14-1051.pdf | Native Language Identification Using Large, Longitudinal Data | Native Language Identification (NLI) is a task aimed at determining the native language (L1) of learners of second language (L2) on the basis of their written texts. To date, research on NLI has focused on relatively small corpora. We apply NLI to the recently released EFCamDat corpus which is not only multiple times larger than previous L2 corpora but also provides longitudinal data at several proficiency levels. Our investigation using accurate machine learning with a wide range of linguistic features reveals interesting patterns in the longitudinal data which are useful for both further development of NLI and its application to research on L2 acquisition. | ['Dora Alexopoulou', 'Xiao Jiang', 'Yufan Guo', 'Anna Korhonen', 'Lin Sun', 'Jeroen Geertzen'] | 2014-05-01 | null | null | null | lrec-2014-5 | ['native-language-identification'] | ['natural-language-processing'] | [ 1.90509427e-02 -1.62436888e-01 -9.55156326e-01 -3.75836134e-01
-1.00768495e+00 -7.54384339e-01 8.57934356e-01 4.91578609e-01
-8.13639522e-01 5.67845106e-01 5.33854663e-01 -7.64137447e-01
-1.84904858e-01 -2.20432699e-01 -2.88040638e-01 -1.30776212e-01
9.72522795e-02 4.96973962e-01 1.68666020e-01 -1.97815329e-01
4.29123163e-01 6.27343535e-01 -1.47882926e+00 1.88570902e-01
1.02455127e+00 3.14084917e-01 3.73997688e-01 3.85781586e-01
-3.77144724e-01 9.21173394e-01 -4.39616233e-01 -3.92430931e-01
-9.82854515e-02 -4.90829438e-01 -1.09530926e+00 -2.96946198e-01
9.89777923e-01 -1.56816050e-01 -1.47770360e-01 1.13774717e+00
2.05805644e-01 4.53951396e-02 6.56316638e-01 -6.70782149e-01
-6.96094930e-01 1.04687428e+00 -2.74972796e-01 5.99025548e-01
7.26553261e-01 -3.46508361e-02 1.09910285e+00 -8.46899092e-01
6.30449772e-01 1.49930191e+00 6.41742766e-01 9.25086915e-01
-1.39776659e+00 -9.23927546e-01 -4.67746742e-02 -1.89896766e-02
-1.34826541e+00 -9.73476708e-01 5.11517763e-01 -9.70398009e-01
6.64909005e-01 -3.55400443e-02 4.01651323e-01 1.04112160e+00
-1.97912157e-02 8.78653824e-01 1.76596868e+00 -9.13192987e-01
-3.34227562e-01 5.05242646e-01 6.54837608e-01 4.71189588e-01
8.83643106e-02 3.78148884e-01 -9.54637647e-01 4.33192253e-01
5.35326004e-01 -8.61080110e-01 -7.10371584e-02 1.72349349e-01
-1.36066878e+00 9.62822974e-01 -2.58221745e-01 1.11639261e+00
1.23469047e-01 -4.57549870e-01 5.55645585e-01 7.42951095e-01
4.58177060e-01 6.16486788e-01 -4.86784041e-01 -5.77047050e-01
-9.20236349e-01 -9.56308991e-02 6.48983717e-01 7.39128888e-01
3.84603918e-01 6.24545142e-02 3.61250266e-02 1.31511152e+00
3.60348493e-01 4.49022979e-01 1.11103892e+00 -6.47229373e-01
6.18733943e-01 5.57791352e-01 -3.57959509e-01 -4.22426879e-01
-3.33216995e-01 6.27892790e-03 -2.21519932e-01 2.35150412e-01
1.14864683e+00 -3.28930765e-02 -3.36360097e-01 2.09201169e+00
2.86712516e-02 -5.44023514e-01 1.74863622e-01 2.49286249e-01
9.87342775e-01 4.97309178e-01 5.22096455e-01 -4.38838452e-01
1.04281592e+00 -4.65452164e-01 -4.56977248e-01 -2.51238823e-01
1.06553352e+00 -7.49803007e-01 1.21832955e+00 5.02407312e-01
-1.25034690e+00 -8.40103745e-01 -7.51702547e-01 -2.40566745e-01
-4.36640143e-01 2.55240619e-01 9.15177107e-01 1.27100945e+00
-1.30816174e+00 3.81011635e-01 -4.71271664e-01 -6.30203903e-01
-2.04045307e-02 4.08305258e-01 -8.14114928e-01 6.70603141e-02
-1.12930620e+00 9.93269444e-01 4.50688154e-01 -4.26743746e-01
-4.96296614e-01 -8.53840828e-01 -1.08925247e+00 -4.68249708e-01
1.00967705e-01 5.10790110e-01 1.35672951e+00 -1.01125062e+00
-1.44917905e+00 1.61895990e+00 -1.37868300e-01 1.66744292e-02
2.51428872e-01 2.48041898e-02 -6.99566126e-01 -1.93048790e-01
5.41924000e-01 5.14032960e-01 3.02509576e-01 -7.17118740e-01
-1.00573552e+00 -6.96255803e-01 -3.19633543e-01 2.39236966e-01
-5.01892388e-01 7.59918571e-01 9.30587724e-02 -6.32316411e-01
1.57068726e-02 -6.90418780e-01 4.46796417e-01 -4.47326720e-01
8.68400186e-03 -1.04664314e+00 5.74058555e-02 -1.00445616e+00
1.51762795e+00 -2.12332129e+00 -1.96012966e-02 7.06585795e-02
1.52665779e-01 3.70777041e-01 -1.53190359e-01 5.65482020e-01
-1.77841783e-01 3.43606323e-01 2.45417073e-01 -1.61864728e-01
6.37743399e-02 -7.88632780e-02 8.89505595e-02 5.64562380e-01
-1.56742364e-01 8.54407966e-01 -9.74687696e-01 -6.34448409e-01
4.56230529e-02 -9.82661098e-02 -2.69383997e-01 1.97448060e-01
2.93867022e-01 8.14767003e-01 -1.79663748e-01 5.02011418e-01
2.11573876e-02 6.05148971e-01 7.55827203e-02 5.07386684e-01
-1.06600511e+00 8.90909016e-01 -5.79135776e-01 1.55751872e+00
-6.66517258e-01 1.09620309e+00 1.75637469e-01 -1.16298187e+00
7.89456666e-01 3.97165060e-01 1.58736512e-01 -6.37000144e-01
1.70575932e-01 6.66863024e-01 5.91322362e-01 -4.86261249e-01
9.60904509e-02 -2.21648380e-01 -2.34734237e-01 5.02622008e-01
2.17473760e-01 4.58599031e-02 5.35244465e-01 -2.44571328e-01
5.35019755e-01 -1.90316215e-02 5.12987792e-01 -1.05869520e+00
9.82404590e-01 4.58941497e-02 4.22123224e-01 9.10483718e-01
-4.59199071e-01 -1.30243137e-01 2.71925688e-01 1.91446524e-02
-8.01656306e-01 -1.04084122e+00 -7.22517431e-01 1.48601651e+00
-4.90195811e-01 -2.45012287e-02 -5.70778608e-01 -4.28793490e-01
6.23576110e-03 1.20453656e+00 -4.12090331e-01 4.96022999e-02
-7.50404060e-01 -1.29319161e-01 4.52479869e-01 5.08824646e-01
-1.01078348e-02 -1.26973915e+00 1.58035517e-01 1.17155850e-01
2.17195317e-01 -1.11403120e+00 -6.51955307e-01 -8.53320882e-02
-6.29765570e-01 -1.02754188e+00 -4.02405828e-01 -1.44595647e+00
4.17928874e-01 -7.11655989e-02 1.09529901e+00 1.25666052e-01
1.87729970e-01 5.68236232e-01 -3.10580760e-01 -4.95400935e-01
-1.33064640e+00 5.29934168e-01 6.32375777e-01 -3.46064121e-01
9.74441588e-01 -2.34950185e-01 5.13009906e-01 -1.15973279e-01
-3.72645169e-01 3.34126167e-02 3.31883162e-01 6.52735054e-01
1.36066079e-01 1.55786440e-01 6.81042790e-01 -8.40996921e-01
7.05307543e-01 -3.58539879e-01 -6.87974930e-01 3.23112249e-01
-6.93508446e-01 -1.49926126e-01 5.55877447e-01 -8.45705867e-01
-1.07896018e+00 -3.53884131e-01 -4.63258326e-01 3.61962378e-01
-5.64622164e-01 5.41380405e-01 -2.19634660e-02 -3.40099901e-01
3.89766067e-01 2.60952443e-01 3.72468084e-02 -8.94230783e-01
-1.50618732e-01 9.90536809e-01 8.73034120e-01 -7.88367033e-01
7.23321676e-01 -5.05568027e-01 -2.79600471e-01 -1.40573239e+00
-9.25564528e-01 -4.53422725e-01 -1.40525150e+00 -1.87380940e-01
6.63648486e-01 -1.01862526e+00 -8.08318973e-01 7.10409343e-01
-5.75598538e-01 -5.64775586e-01 -4.11464088e-02 1.18630421e+00
-4.30150688e-01 8.62853900e-02 -5.21879435e-01 -8.22325945e-01
-5.11648059e-02 -1.12925684e+00 3.92025918e-01 1.49721995e-01
-7.13438392e-01 -1.35119677e+00 3.62500519e-01 4.95900482e-01
9.33806691e-03 -3.52160722e-01 1.27073729e+00 -1.23074245e+00
1.63612187e-01 -3.84801179e-02 1.35640875e-01 3.63098443e-01
1.27136171e-01 -6.18572533e-02 -6.42104447e-01 -5.48007607e-01
4.00615670e-02 -8.55415046e-01 4.74981457e-01 1.54278055e-01
4.87373590e-01 -2.37921000e-01 -9.60529298e-02 3.58118683e-01
1.22377396e+00 4.14666802e-01 -1.31112173e-01 2.57852793e-01
7.83419609e-01 1.01550519e+00 3.59667510e-01 -1.87256142e-01
4.61820960e-01 6.50434375e-01 -7.28026986e-01 3.37906361e-01
-4.17190343e-01 -5.01614869e-01 7.83158183e-01 1.45209992e+00
2.92618185e-01 2.35979393e-01 -1.35313249e+00 1.14552343e+00
-1.05190456e+00 -9.28139091e-01 -5.74137419e-02 2.45824647e+00
1.42132187e+00 1.18166730e-01 4.04430985e-01 2.21946314e-01
6.25613809e-01 -9.84830484e-02 -2.05909640e-01 -6.66912913e-01
-8.34922567e-02 3.62187117e-01 3.17092389e-01 9.18440819e-01
-8.48833919e-01 1.27566671e+00 7.44597292e+00 9.58930314e-01
-1.16599619e+00 -6.62426054e-02 4.64672655e-01 1.22225665e-01
-3.19109350e-01 -2.18717322e-01 -1.42925751e+00 2.68792391e-01
1.18492877e+00 -4.56748098e-01 6.77058876e-01 4.27195340e-01
8.57245848e-02 -1.75200582e-01 -1.44616044e+00 7.70487666e-01
4.35894206e-02 -6.66595578e-01 -2.22017065e-01 -7.61993751e-02
5.66290200e-01 5.43589368e-02 7.01596439e-02 6.14401221e-01
1.92952931e-01 -1.04727781e+00 1.04418921e+00 5.68723679e-02
1.43901336e+00 -6.48744941e-01 1.68059289e-01 6.81412637e-01
-8.50417376e-01 -1.72964558e-01 -1.04760103e-01 -5.06583452e-01
-2.91480482e-01 -2.62702227e-01 -9.06301200e-01 -2.54839063e-01
4.11420614e-01 6.22071505e-01 -8.61333787e-01 4.30659086e-01
-2.98472285e-01 1.28990793e+00 -1.68440014e-01 -2.85686761e-01
2.43431836e-01 -5.76007180e-02 4.73869741e-01 1.09718168e+00
2.46921808e-01 3.68112847e-02 3.38448286e-01 5.88903129e-01
1.63391709e-01 9.12913561e-01 -8.98084760e-01 -6.26636386e-01
6.28831744e-01 8.05601597e-01 -4.34834987e-01 2.34310254e-02
-1.05057919e+00 4.66925144e-01 8.28144133e-01 -8.49718675e-02
2.77398288e-01 -2.00446695e-01 6.22288704e-01 3.62503052e-01
-3.51319849e-01 -6.35530591e-01 -2.53360987e-01 -1.04224694e+00
-3.93095374e-01 -9.63721991e-01 2.38593847e-01 -4.70061637e-02
-1.46607816e+00 2.06436709e-01 3.09226096e-01 -4.82433885e-01
-6.43854737e-01 -1.03877699e+00 -5.33137262e-01 1.26077068e+00
-1.05536759e+00 -1.04984081e+00 3.62674028e-01 4.13432926e-01
7.53963053e-01 -8.53481710e-01 7.15787888e-01 1.50279701e-01
-8.13937187e-01 1.00633240e+00 2.90151000e-01 4.32565063e-01
6.82532966e-01 -1.32010412e+00 3.68396521e-01 7.95255899e-01
4.40537602e-01 7.06792414e-01 2.56271273e-01 -7.85421908e-01
-1.10434091e+00 -4.59007621e-01 1.73168027e+00 -5.86660326e-01
1.16728604e+00 -4.10025358e-01 -8.02480102e-01 7.75794089e-01
3.10142845e-01 -8.32246184e-01 9.62216556e-01 3.99377108e-01
1.42432526e-01 6.42076731e-02 -8.85674119e-01 6.37368143e-01
1.08017957e+00 -9.29975450e-01 -7.22279727e-01 2.74153292e-01
5.29708266e-01 -9.27022174e-02 -1.41669381e+00 3.58180672e-01
6.51069701e-01 -9.09058630e-01 7.00752616e-01 -5.33849120e-01
5.73207662e-02 2.96930760e-01 1.13303103e-01 -1.18270779e+00
-5.25155127e-01 -5.83718956e-01 6.88756883e-01 1.69548249e+00
4.74581212e-01 -4.91790563e-01 3.77187520e-01 6.34510279e-01
-2.34850459e-02 -4.71856833e-01 -9.37525690e-01 -8.53009760e-01
8.05274069e-01 -6.17624223e-01 2.43259653e-01 1.13253164e+00
2.80994087e-01 4.02318507e-01 1.77303463e-01 -2.67544895e-01
4.15804833e-01 -8.52034837e-02 5.31354904e-01 -1.54552424e+00
1.04894355e-01 -1.09842181e+00 -2.18989268e-01 -7.71242738e-01
9.10380781e-01 -1.38986349e+00 -2.09200252e-02 -6.81415021e-01
1.72549784e-01 -5.63879609e-01 -5.37813678e-02 5.06557703e-01
-2.97439337e-01 1.38984561e-01 1.12836011e-01 1.53962642e-01
-3.34207714e-02 1.90467060e-01 1.00274158e+00 -6.86147343e-03
-4.10569370e-01 2.38496304e-01 -7.84604967e-01 8.17857742e-01
7.70465910e-01 -3.00968677e-01 -4.42093551e-01 -3.58743727e-01
-1.33565962e-01 -3.31322895e-03 -3.94333661e-01 -9.41602468e-01
1.09220400e-01 -4.27231491e-01 2.32895628e-01 -4.04978007e-01
-2.88004577e-01 -3.05996865e-01 -4.06044781e-01 3.08955073e-01
-7.05646694e-01 1.79331094e-01 3.52115542e-01 -5.70978105e-01
-2.02519327e-01 -9.14367735e-01 9.15826976e-01 -1.89310089e-01
-9.69934762e-01 2.76244700e-01 -8.00716519e-01 5.75500846e-01
8.31859171e-01 -1.61638811e-01 2.73666620e-01 -1.53330445e-01
-4.53670382e-01 8.15913379e-02 6.69996858e-01 6.90524817e-01
7.44312257e-02 -1.51485467e+00 -9.60894406e-01 5.98035455e-01
1.81728810e-01 -6.22505248e-01 -3.56790811e-01 6.80382431e-01
-4.88155037e-01 9.37340856e-01 -2.85418898e-01 -1.32574484e-01
-1.33776081e+00 5.63650310e-01 -6.53135031e-03 -5.70500679e-02
-2.62082338e-01 9.20106530e-01 2.81461000e-01 -6.38197660e-01
3.60150963e-01 2.85716385e-01 -8.23636889e-01 2.01245710e-01
6.49834931e-01 5.06939173e-01 -5.54819167e-01 -1.55072939e+00
-2.67247349e-01 5.10142624e-01 -1.82913423e-01 -4.89631444e-01
9.59617555e-01 -2.87364155e-01 -3.27646047e-01 1.32902467e+00
1.27102256e+00 7.63902664e-01 -5.42817831e-01 -4.45807397e-01
4.99719769e-01 -3.90391499e-01 6.11476414e-02 -5.61691642e-01
-5.55055559e-01 8.05063307e-01 2.45173723e-01 3.87641713e-02
7.22591817e-01 1.08812630e-01 4.13542986e-01 1.86021611e-01
1.91248596e-01 -1.29748762e+00 -4.03263897e-01 7.40882695e-01
7.58782804e-01 -1.50185561e+00 -1.49184734e-01 -5.81471138e-02
-4.09871489e-01 1.00591540e+00 7.96678185e-01 3.74669403e-01
6.25419796e-01 -6.99728280e-02 3.29465479e-01 7.49209151e-02
-4.10371393e-01 -2.37586796e-01 6.99590027e-01 5.75863302e-01
1.13685846e+00 2.05332056e-01 -1.09351802e+00 7.54217386e-01
-8.21214914e-01 -4.17206943e-01 2.21875206e-01 4.57241923e-01
-4.46106136e-01 -1.57601810e+00 -4.00374472e-01 4.52697694e-01
-8.09578717e-01 -2.93250322e-01 -4.00622994e-01 1.05769336e+00
3.01394731e-01 1.01969087e+00 1.89267453e-02 -1.38675869e-01
1.69143379e-02 3.80396783e-01 7.55939186e-01 -9.38156545e-01
-6.82889283e-01 -3.75645906e-01 1.85564503e-01 -2.18494609e-01
-4.62864131e-01 -1.39888310e+00 -8.95558774e-01 -2.84873039e-01
-7.92630538e-02 1.96514934e-01 5.80158889e-01 1.11743379e+00
-7.30383098e-01 -1.33215755e-01 5.42755365e-01 -4.41380888e-01
-5.37230849e-01 -1.20502782e+00 -9.80223417e-01 2.82637149e-01
2.45420411e-01 -4.95355517e-01 -2.41107970e-01 -2.39457309e-01] | [10.664031982421875, 10.17415714263916] |
73c02674-8398-434a-964e-60607f1b3ebf | disentangling-propagation-and-generation-for | 1812.00452 | null | https://arxiv.org/abs/1812.00452v2 | https://arxiv.org/pdf/1812.00452v2.pdf | Disentangling Propagation and Generation for Video Prediction | A dynamic scene has two types of elements: those that move fluidly and can be predicted from previous frames, and those which are disoccluded (exposed) and cannot be extrapolated. Prior approaches to video prediction typically learn either to warp or to hallucinate future pixels, but not both. In this paper, we describe a computational model for high-fidelity video prediction which disentangles motion-specific propagation from motion-agnostic generation. We introduce a confidence-aware warping operator which gates the output of pixel predictions from a flow predictor for non-occluded regions and from a context encoder for occluded regions. Moreover, in contrast to prior works where confidence is jointly learned with flow and appearance using a single network, we compute confidence after a warping step, and employ a separate network to inpaint exposed regions. Empirical results on both synthetic and real datasets show that our disentangling approach provides better occlusion maps and produces both sharper and more realistic predictions compared to strong baselines. | ['Fisher Yu', 'Ruth Wang', 'Qi-Zhi Cai', 'Hang Gao', 'Huazhe Xu', 'Trevor Darrell'] | 2018-12-02 | disentangling-propagation-and-generation-for-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Gao_Disentangling_Propagation_and_Generation_for_Video_Prediction_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Gao_Disentangling_Propagation_and_Generation_for_Video_Prediction_ICCV_2019_paper.pdf | iccv-2019-10 | ['predict-future-video-frames'] | ['computer-vision'] | [ 5.59560478e-01 4.34476286e-01 -2.83502787e-01 -4.60801780e-01
-5.28896332e-01 -4.16527778e-01 9.76644218e-01 -2.50291109e-01
-1.32304104e-03 1.09032834e+00 8.47412407e-01 3.71798202e-02
5.78125596e-01 -7.58448422e-01 -1.13314426e+00 -4.98788118e-01
-1.29118934e-01 2.72694677e-01 4.74625796e-01 -5.70533648e-02
3.73506956e-02 3.67785662e-01 -1.36599267e+00 7.59763598e-01
7.47676373e-01 8.78262520e-01 1.17510945e-01 9.28292990e-01
2.86238104e-01 1.51298070e+00 -2.37427056e-01 -3.70199114e-01
4.42854345e-01 -5.74468851e-01 -6.82236969e-01 3.48275334e-01
9.40641820e-01 -8.33457291e-01 -7.44898140e-01 7.12478876e-01
-1.12089276e-01 1.44198418e-01 4.75914001e-01 -1.14603305e+00
-6.92945361e-01 3.65077734e-01 -5.62240124e-01 2.18270868e-01
7.24092662e-01 6.04458988e-01 8.19022655e-01 -8.40995908e-01
1.29425740e+00 1.34380531e+00 5.99562645e-01 6.57627225e-01
-1.70220435e+00 -5.35199642e-01 4.31025386e-01 1.43542618e-01
-1.07855380e+00 -6.74577832e-01 8.12916279e-01 -6.89107835e-01
8.56718183e-01 1.04549766e-01 9.54383790e-01 1.52115631e+00
6.01643622e-01 8.30233395e-01 8.96969020e-01 -2.92177685e-02
7.01317936e-02 2.15547122e-02 -6.75865412e-01 6.49020493e-01
-1.19811542e-01 7.05206156e-01 -7.56225765e-01 2.85074413e-02
1.07043934e+00 -9.71440077e-02 -8.51579726e-01 -5.01298726e-01
-1.27430701e+00 6.58483505e-01 3.28942358e-01 -2.49308094e-01
-6.57418489e-01 4.25133616e-01 1.33846439e-02 1.38321295e-01
6.23055875e-01 3.15693676e-01 -2.93172717e-01 -6.16982616e-02
-1.49069917e+00 3.18178147e-01 7.42270172e-01 8.28736126e-01
9.11817133e-01 3.90868992e-01 -5.79936206e-01 2.30364308e-01
1.94573432e-01 3.06164116e-01 9.96782258e-02 -1.51446652e+00
4.25055295e-01 -1.24002576e-01 5.78635931e-01 -1.14979196e+00
1.17423974e-01 -3.01556468e-01 -6.50747538e-01 7.00425923e-01
3.55894834e-01 -2.32395336e-01 -1.16742575e+00 2.04934859e+00
1.02626875e-01 1.02544701e+00 -2.00624410e-02 1.07373047e+00
3.91719908e-01 1.00714004e+00 2.82007426e-01 -1.51087821e-01
6.68844342e-01 -1.41709578e+00 -5.63908517e-01 -5.18065274e-01
1.38736755e-01 -6.82994664e-01 5.82364738e-01 3.79372656e-01
-1.43561149e+00 -7.82186210e-01 -1.13476157e+00 -4.11766469e-01
1.25175834e-01 -4.33988839e-01 4.76624280e-01 9.75857750e-02
-1.41553879e+00 9.43665326e-01 -1.02415764e+00 -7.04527125e-02
5.98626196e-01 -8.51752385e-02 -5.98600864e-01 -2.72198975e-01
-1.04139411e+00 1.09051824e+00 2.59553313e-01 -1.48373634e-01
-1.26154709e+00 -9.95410621e-01 -1.05236578e+00 -5.51385470e-02
6.94296276e-03 -1.12212026e+00 9.59887445e-01 -1.55080843e+00
-1.49864554e+00 4.21789736e-01 -3.23192805e-01 -8.14139366e-01
1.03383231e+00 -5.03463447e-01 -2.54697472e-01 2.50418305e-01
5.46053350e-02 1.48217523e+00 1.05932462e+00 -1.52973104e+00
-7.40608156e-01 3.68747741e-01 1.94352582e-01 2.92782158e-01
2.85412610e-01 -5.39766908e-01 -3.85607779e-01 -8.42598975e-01
-1.92461401e-01 -6.73428714e-01 -3.24929535e-01 7.30922759e-01
-1.87430516e-01 5.18128633e-01 1.10139775e+00 -1.16592920e+00
9.20259833e-01 -1.77772284e+00 5.12092352e-01 -1.61715806e-01
2.19581455e-01 3.91025878e-02 -4.14794177e-01 2.28242964e-01
-1.87555730e-01 -4.41309772e-02 -4.17020589e-01 -6.15800738e-01
-3.48233461e-01 5.31720579e-01 -8.56089473e-01 4.17643964e-01
6.72442853e-01 9.13150907e-01 -1.25679982e+00 -3.88717651e-01
5.17445624e-01 1.02737832e+00 -9.87356544e-01 2.78519154e-01
-5.65712929e-01 8.71630728e-01 -7.34184086e-02 3.92743558e-01
5.72919250e-01 -1.23979993e-01 -8.31739139e-03 -2.70699829e-01
-1.84816301e-01 1.94091141e-01 -9.11434710e-01 2.13631368e+00
-4.41003442e-01 1.00407481e+00 -5.96646294e-02 -5.89179933e-01
6.72302544e-01 5.08621752e-01 3.82147312e-01 -3.30419630e-01
-1.54266655e-01 -1.94705620e-01 -3.84794593e-01 -4.71978843e-01
8.58569086e-01 -1.52090013e-01 3.97476226e-01 2.34205291e-01
2.41542831e-01 -2.97303706e-01 7.45547935e-02 2.95700699e-01
1.16423273e+00 1.16214681e+00 -2.21178681e-02 4.36496362e-02
3.43993694e-01 -2.16599833e-02 8.95772398e-01 6.70306146e-01
-2.63842851e-01 1.15701950e+00 5.68268180e-01 -7.29305148e-01
-1.30527830e+00 -1.25076640e+00 1.81444660e-01 6.05372906e-01
3.46739620e-01 -3.11960548e-01 -4.15003061e-01 -5.72303951e-01
-1.25631720e-01 1.06036532e+00 -7.12018669e-01 -8.98386538e-02
-9.15506840e-01 -3.08498740e-01 -1.22088043e-03 6.50626361e-01
3.59868616e-01 -1.06469154e+00 -6.65848732e-01 4.97241706e-01
-4.33625966e-01 -1.24974477e+00 -5.94072640e-01 -1.83346152e-01
-7.81820655e-01 -7.10514843e-01 -6.75220370e-01 -5.94308734e-01
5.71103930e-01 1.57359596e-02 1.45851779e+00 4.65514734e-02
-3.62821147e-02 2.50502378e-01 -1.43452570e-01 5.85672930e-02
-5.29240966e-01 -4.51229334e-01 -3.74632210e-01 2.13892028e-01
-2.16432393e-01 -8.69531929e-01 -8.97754252e-01 -3.68402898e-02
-7.99201071e-01 7.73817241e-01 3.67171228e-01 9.00401294e-01
5.46224415e-01 -4.62170899e-01 1.73238143e-01 -8.81812811e-01
-3.09573729e-02 -6.52534068e-01 -4.31950808e-01 1.48332745e-01
-2.92069018e-01 1.62599653e-01 5.60226917e-01 -4.52302665e-01
-1.46890306e+00 4.43868518e-01 -6.49461150e-02 -7.94008136e-01
-1.62496969e-01 -5.18092550e-02 1.08053111e-01 4.19676267e-02
6.19488657e-01 1.27123475e-01 -1.92022920e-01 -6.56410009e-02
6.87201917e-01 -1.98282212e-01 9.49753642e-01 -5.73496342e-01
8.47657084e-01 9.24391508e-01 -7.46683404e-02 -3.93025577e-01
-6.83164477e-01 2.06348211e-01 -6.33149743e-01 -4.85351175e-01
9.65966821e-01 -1.17208242e+00 -4.46722172e-02 7.19563514e-02
-1.51267648e+00 -6.01831555e-01 -7.72235751e-01 4.50589418e-01
-7.82486200e-01 3.74047279e-01 -8.89747381e-01 -5.25867641e-01
-7.74576738e-02 -1.13758433e+00 9.86096621e-01 2.26786926e-01
-4.87154871e-01 -1.12638354e+00 1.32513151e-01 7.36799138e-03
4.10775483e-01 7.05032706e-01 4.73656029e-01 1.68563619e-01
-1.09369111e+00 -2.10709777e-02 -1.29810825e-01 2.21222445e-01
-1.17921025e-01 2.22446606e-01 -1.02234840e+00 -2.20554307e-01
-9.28301811e-02 -2.17095956e-01 1.16099381e+00 4.33317900e-01
9.66192186e-01 -6.02363229e-01 -4.12723660e-01 9.50510919e-01
1.45945525e+00 6.64511323e-02 1.01294196e+00 6.06073178e-02
7.86405683e-01 6.26429796e-01 4.51174885e-01 4.18630064e-01
2.82571316e-01 5.22306740e-01 5.99889040e-01 -1.29477873e-01
-6.15038395e-01 -7.54098535e-01 5.22140026e-01 3.52255493e-01
-3.48480791e-01 -6.89119995e-01 -4.61318672e-01 5.94986320e-01
-2.02391791e+00 -1.30107856e+00 5.34300208e-02 1.98317516e+00
9.11487281e-01 1.73267409e-01 -3.25879812e-01 -1.95991516e-01
5.39231122e-01 6.67906344e-01 -5.66296339e-01 -2.58466929e-01
-3.08973938e-01 3.26643705e-01 2.69702464e-01 1.22714758e+00
-1.03341186e+00 1.17994392e+00 6.56521177e+00 3.79150003e-01
-1.10055602e+00 1.71070263e-01 1.05044377e+00 -2.73710340e-01
-6.75304174e-01 4.38472927e-01 -3.46324831e-01 3.23877990e-01
6.87111855e-01 1.32538363e-01 3.14344555e-01 5.11246800e-01
3.27672541e-01 -1.80448338e-01 -1.38179898e+00 5.89202285e-01
9.32425708e-02 -1.75959802e+00 2.63239115e-01 -2.12237209e-01
1.16872585e+00 3.34588764e-03 1.08955793e-01 -4.02299277e-02
5.89810610e-01 -9.72558022e-01 1.19813490e+00 1.00999641e+00
6.47543252e-01 -2.80284345e-01 3.86064023e-01 5.00078201e-02
-1.07498908e+00 1.86405912e-01 -2.14903355e-01 -2.86712885e-01
8.96966338e-01 4.02368784e-01 -4.72470522e-01 3.43894482e-01
3.46486211e-01 1.23943853e+00 -1.22315884e-01 8.98403764e-01
-6.93014383e-01 5.05519271e-01 -2.01712213e-02 7.64625490e-01
1.41036630e-01 -1.48806348e-01 6.73962355e-01 1.15363550e+00
4.05862778e-01 -3.16224550e-03 2.17992380e-01 1.13252664e+00
1.26860961e-01 -4.41347927e-01 -4.97951388e-01 4.42649066e-01
2.58045554e-01 1.03929102e+00 -4.44664896e-01 -7.57651508e-01
-5.82691789e-01 1.43399501e+00 2.27150306e-01 8.94946754e-01
-1.04052353e+00 2.31995180e-01 8.87113810e-01 4.11227256e-01
5.03518045e-01 -1.84144288e-01 -2.23723307e-01 -1.41071522e+00
-2.62799054e-01 -3.05357993e-01 4.20810096e-02 -1.25362766e+00
-1.02495515e+00 7.28323042e-01 -1.01884477e-01 -1.33565152e+00
-6.80240452e-01 -2.98251301e-01 -6.48041844e-01 1.03482926e+00
-1.78165317e+00 -1.14859426e+00 -2.62459755e-01 3.34133655e-01
7.44227648e-01 4.07141954e-01 6.18504584e-01 9.63647589e-02
-1.64230838e-01 4.46312055e-02 -3.36510658e-01 -9.11744125e-03
7.20466197e-01 -1.02084589e+00 4.92933214e-01 1.35623324e+00
2.08167046e-01 2.79921037e-03 1.08184600e+00 -7.97464073e-01
-8.79351079e-01 -1.43905389e+00 8.84930611e-01 -5.97417474e-01
3.45581949e-01 -8.22417140e-02 -9.49639380e-01 1.00341177e+00
5.56804776e-01 5.58356285e-01 4.80184704e-02 -6.13048494e-01
-5.30941427e-01 7.39818886e-02 -1.07983065e+00 7.55755961e-01
1.17350900e+00 -4.75252211e-01 -3.50390434e-01 1.87594935e-01
8.11174512e-01 -4.87041265e-01 -7.67744243e-01 3.38714749e-01
8.26107025e-01 -1.44757235e+00 1.19062364e+00 -4.94212598e-01
1.15316784e+00 -3.75758588e-01 -1.21210709e-01 -1.29196179e+00
-4.30157900e-01 -7.52462924e-01 -5.07087231e-01 8.34642529e-01
3.19212765e-01 -2.87779361e-01 1.03792393e+00 7.86026895e-01
-2.93521255e-01 -6.05356812e-01 -6.39236689e-01 -4.82427925e-01
1.09703340e-01 -3.82153153e-01 4.60338950e-01 8.25139403e-01
-2.95830458e-01 7.70303011e-02 -1.01255524e+00 1.30460501e-01
5.28109312e-01 8.47587362e-02 6.50710762e-01 -7.47909307e-01
-5.64735472e-01 -8.36969391e-02 -4.64755565e-01 -1.39506423e+00
1.03144780e-01 -6.33175850e-01 3.74166787e-01 -1.54261756e+00
3.87324318e-02 -6.13772050e-02 -9.63751003e-02 3.69854331e-01
-1.13686398e-01 4.65625674e-01 3.51105005e-01 2.18653977e-01
-3.35132241e-01 6.09045744e-01 1.56112027e+00 -1.35362735e-02
-1.57884076e-01 -5.02729774e-01 -2.82741904e-01 8.77093256e-01
4.21083748e-01 -3.98756117e-01 -6.17802620e-01 -7.56503761e-01
-1.52509671e-03 7.12645292e-01 6.15143776e-01 -1.19818985e+00
6.10185079e-02 -3.61157447e-01 8.69398475e-01 -3.80986482e-01
7.48769224e-01 -5.46819985e-01 5.26752591e-01 5.36852956e-01
-3.90642852e-01 1.15379263e-02 3.77586624e-03 9.22148228e-01
-2.41847709e-01 2.45530114e-01 7.15044975e-01 -2.45416686e-01
-9.41244721e-01 6.60212576e-01 -3.71413350e-01 -8.26825760e-03
8.82577777e-01 -4.42753166e-01 -3.74949366e-01 -8.55638266e-01
-8.25899601e-01 -2.60404442e-02 7.72276998e-01 5.86218894e-01
7.71371484e-01 -1.30262911e+00 -8.77600610e-01 2.67326683e-01
-3.45040768e-01 -2.81854961e-02 3.07749391e-01 6.00914836e-01
-7.39486098e-01 4.45848852e-02 -4.49282348e-01 -6.59443915e-01
-6.74500287e-01 5.36587954e-01 1.99055567e-01 -1.59021825e-01
-8.44839454e-01 9.37944472e-01 6.15870655e-01 1.40146881e-01
8.53743777e-02 -2.91490525e-01 6.86476147e-03 -3.06802332e-01
6.03391886e-01 8.10825527e-02 -5.89788198e-01 -1.06666780e+00
9.45768803e-02 4.67949957e-01 5.47869354e-02 -4.36129838e-01
1.30265594e+00 -2.92136312e-01 2.61501819e-01 3.26077908e-01
9.64139163e-01 -1.94604471e-01 -2.45513082e+00 -1.43568024e-01
-4.37248528e-01 -8.74156713e-01 -2.74976585e-02 -8.44051123e-01
-1.27535212e+00 8.86443913e-01 3.64236772e-01 -2.11529255e-01
1.07693291e+00 -2.34994709e-01 8.48876655e-01 -3.46958905e-01
3.30226988e-01 -7.35888481e-01 9.38662887e-02 2.19990835e-01
1.13972723e+00 -1.17271352e+00 8.21907893e-02 -5.70781589e-01
-6.55330718e-01 1.09032154e+00 7.70680785e-01 -4.33325410e-01
5.69730878e-01 3.35087150e-01 -1.79970086e-01 3.75122160e-01
-1.28024685e+00 2.39899591e-01 3.35760921e-01 8.67300451e-01
3.95102918e-01 -1.19482517e-01 3.10769342e-02 1.04760535e-01
-5.22968918e-02 2.46855110e-01 6.74725890e-01 8.72839987e-01
-3.13071012e-01 -9.37051833e-01 -2.64814824e-01 1.91130504e-01
-1.95586935e-01 -1.95124999e-01 9.90514681e-02 6.18463576e-01
4.69490319e-01 8.26920688e-01 3.10028195e-01 -4.53624606e-01
-2.77954787e-01 -7.34235272e-02 5.83483756e-01 -4.53258574e-01
-2.09478781e-01 6.63601011e-02 2.08979338e-01 -1.08804619e+00
-6.89468920e-01 -7.04238057e-01 -9.31937099e-01 -2.81109184e-01
2.15566549e-02 -2.35415280e-01 2.11864337e-01 8.34931195e-01
4.17649746e-01 6.41049743e-01 3.00784826e-01 -1.40860736e+00
1.28134981e-01 -6.14747822e-01 -6.47602752e-02 4.74813610e-01
6.07462108e-01 -4.54454005e-01 -3.56563836e-01 6.75546706e-01] | [10.755776405334473, -1.0561455488204956] |
ea7cbaa0-8047-4346-b56d-f8385fe9ef43 | a-study-of-autoregressive-decoders-for-multi | 2303.17376 | null | https://arxiv.org/abs/2303.17376v1 | https://arxiv.org/pdf/2303.17376v1.pdf | A Study of Autoregressive Decoders for Multi-Tasking in Computer Vision | There has been a recent explosion of computer vision models which perform many tasks and are composed of an image encoder (usually a ViT) and an autoregressive decoder (usually a Transformer). However, most of this work simply presents one system and its results, leaving many questions regarding design decisions and trade-offs of such systems unanswered. In this work, we aim to provide such answers. We take a close look at autoregressive decoders for multi-task learning in multimodal computer vision, including classification, captioning, visual question answering, and optical character recognition. Through extensive systematic experiments, we study the effects of task and data mixture, training and regularization hyperparameters, conditioning type and specificity, modality combination, and more. Importantly, we compare these to well-tuned single-task baselines to highlight the cost incurred by multi-tasking. A key finding is that a small decoder learned on top of a frozen pretrained encoder works surprisingly well. We call this setup locked-image tuning with decoder (LiT-decoder). It can be seen as teaching a decoder to interact with a pretrained vision model via natural language. | ['Xiaohua Zhai', 'Liang-Chieh Chen', 'Qihang Yu', 'Xiao Wang', 'Emanuele Bugliarello', 'André Susano Pinto', 'Alexander Kolesnikov', 'Andreas Steiner', 'Filip Pavetic', 'Gagan Madan', 'Bo Wan', 'Lucas Beyer'] | 2023-03-30 | null | null | null | null | ['specificity'] | ['natural-language-processing'] | [ 4.76486236e-01 1.86224639e-01 1.49894908e-01 -5.40380061e-01
-1.08139515e+00 -7.10180223e-01 9.77674365e-01 -2.78449237e-01
-5.57597101e-01 1.85967252e-01 4.12021041e-01 -4.06485736e-01
2.93141276e-01 -1.14739232e-01 -1.19536889e+00 -6.68749571e-01
5.85384786e-01 5.74181616e-01 1.91920549e-01 6.34363713e-03
8.85203183e-02 -7.72528127e-02 -1.27642322e+00 6.26460493e-01
6.31880879e-01 7.98023403e-01 4.77393717e-01 8.94969940e-01
6.06289543e-02 1.27738452e+00 -4.29760158e-01 -8.35295558e-01
-1.84243411e-01 -4.20192450e-01 -9.41120267e-01 3.73052925e-01
8.33360195e-01 -4.34989721e-01 -3.72621477e-01 7.08796561e-01
5.88678002e-01 -5.34049049e-02 8.56515288e-01 -1.16952622e+00
-9.87243116e-01 3.85602504e-01 -6.28970325e-01 7.04027712e-02
2.18316749e-01 3.88418108e-01 1.07167578e+00 -8.22701216e-01
3.89395297e-01 1.41823208e+00 5.19404113e-01 6.99512661e-01
-1.33539963e+00 -1.98196352e-01 2.27682307e-01 1.71229243e-01
-8.36713731e-01 -8.45133841e-01 4.15776789e-01 -7.02716470e-01
1.06389320e+00 -1.20771855e-01 3.33603770e-02 1.62747562e+00
2.97443986e-01 9.46248353e-01 9.63913620e-01 -5.39036274e-01
5.02484106e-03 4.64427620e-01 2.56954521e-01 7.17807114e-01
-2.54625738e-01 -3.23333219e-02 -5.09239078e-01 -2.89941058e-02
6.62244320e-01 -1.93952858e-01 -3.42163622e-01 -4.21821982e-01
-1.09314072e+00 7.94341207e-01 2.09921971e-01 -6.55525029e-02
-1.83705077e-01 3.48616511e-01 4.71048534e-01 3.85183781e-01
1.24214999e-01 2.94927537e-01 -4.73346204e-01 1.28010154e-01
-7.08060384e-01 -1.43104032e-01 7.11881816e-01 8.94541323e-01
6.61175668e-01 1.13271154e-01 -4.33707803e-01 1.14587486e+00
5.01540601e-01 4.67954278e-01 5.56837618e-01 -9.75221753e-01
4.33114946e-01 1.90633833e-01 -2.56017476e-01 -3.85504454e-01
-2.92582422e-01 -1.88587964e-01 -7.03444660e-01 -3.19808014e-02
5.20744383e-01 -3.28521609e-01 -1.12726676e+00 1.83911824e+00
-1.82279170e-01 -2.15541292e-02 1.51198963e-02 8.82546604e-01
8.77327859e-01 6.44821286e-01 2.51695931e-01 1.70117319e-01
1.74931419e+00 -1.35745764e+00 -5.74722946e-01 -7.49057770e-01
4.13032979e-01 -8.73729706e-01 1.26292396e+00 3.28217924e-01
-1.11187649e+00 -6.53826773e-01 -6.53054595e-01 -3.73611540e-01
-1.72812596e-01 2.56253421e-01 4.14628446e-01 6.05744064e-01
-1.26639605e+00 -1.70313078e-03 -6.56893969e-01 -5.54041505e-01
1.44324690e-01 1.97855487e-01 -4.34544772e-01 -2.99608678e-01
-8.75785589e-01 1.24262714e+00 1.20985642e-01 -2.42640334e-03
-1.36472082e+00 -4.04714614e-01 -9.26341593e-01 -5.70465699e-02
4.32392329e-01 -1.13011885e+00 1.60140705e+00 -1.20002711e+00
-1.47534442e+00 1.21032321e+00 -3.65352899e-01 -3.88198853e-01
2.81707376e-01 -3.32529128e-01 -9.09407213e-02 1.51539862e-01
-1.14550412e-01 9.06575978e-01 1.23957014e+00 -1.37521350e+00
-2.01098859e-01 -3.54484051e-01 1.69955194e-01 2.01166943e-01
-7.71219432e-02 2.35943466e-01 -8.81109774e-01 -2.99691647e-01
-3.27881455e-01 -1.03985953e+00 -1.30617423e-02 -2.45365933e-01
-3.72669935e-01 -1.24589808e-01 5.51478684e-01 -6.72737300e-01
7.37009525e-01 -2.26129746e+00 2.88855672e-01 -3.27290922e-01
5.57136685e-02 1.41421482e-01 -5.68773508e-01 5.15877604e-01
-2.27013841e-01 -6.63567409e-02 -1.44194946e-01 -7.12285817e-01
4.78400569e-03 2.26475492e-01 -5.16562819e-01 3.52384448e-01
4.01023537e-01 1.20709872e+00 -4.71688211e-01 -4.60829020e-01
2.46548489e-01 7.00539529e-01 -6.36981070e-01 3.88340443e-01
-5.44286609e-01 4.28346187e-01 -2.37781048e-01 5.51327229e-01
2.86466599e-01 -7.06987202e-01 7.33030513e-02 -4.17532027e-01
7.93555528e-02 1.60933539e-01 -6.57310605e-01 1.73526680e+00
-3.81224006e-01 9.37441945e-01 2.32030362e-01 -1.14408982e+00
5.44613242e-01 3.46506536e-01 2.02810429e-02 -9.49793220e-01
1.58120841e-01 -9.86955613e-02 -1.25921413e-01 -7.03799546e-01
3.92526627e-01 -1.73458949e-01 6.29554316e-02 3.59404922e-01
5.22801936e-01 -1.27161846e-01 -8.21131021e-02 3.52212906e-01
1.04795551e+00 2.31544301e-01 1.11358995e-02 1.19619735e-01
3.55862945e-01 -7.26451948e-02 5.70438569e-04 8.99578452e-01
-1.93914279e-01 9.32979703e-01 5.99574387e-01 -6.56613186e-02
-1.14050305e+00 -9.41779673e-01 7.19056800e-02 1.59520805e+00
-1.55654222e-01 -3.54132533e-01 -7.18656540e-01 -3.35786372e-01
-1.57237470e-01 7.37489164e-01 -6.40090048e-01 -3.12310249e-01
-4.70093697e-01 -7.75098383e-01 6.91279650e-01 4.77271318e-01
4.41721946e-01 -9.78985846e-01 -5.90714574e-01 -6.08211495e-02
-2.93626219e-01 -1.48776829e+00 -4.95028228e-01 2.99265355e-01
-6.90982640e-01 -8.77573073e-01 -1.01297271e+00 -9.18246329e-01
3.86397868e-01 3.21973145e-01 1.46799302e+00 -2.86096752e-01
1.62596568e-01 9.14144635e-01 -2.17739955e-01 -3.65163893e-01
-4.28266615e-01 3.07356138e-02 -3.90468210e-01 2.53449798e-01
1.48342699e-01 -2.86689848e-01 -5.23290038e-01 3.61956686e-01
-1.03596854e+00 1.86637983e-01 1.09558356e+00 8.60219359e-01
4.03251708e-01 -7.56455898e-01 1.99528217e-01 -9.81353998e-01
6.45871580e-01 -3.69879931e-01 -4.12658870e-01 6.26619101e-01
-2.23496169e-01 4.06165034e-01 3.57224554e-01 -5.01772285e-01
-1.14046586e+00 1.86393350e-01 -1.41333282e-01 -5.30005574e-01
-2.04614908e-01 3.76111776e-01 -1.25094518e-01 6.64101262e-03
6.66669130e-01 3.76510561e-01 6.28448054e-02 -5.18775225e-01
6.84136093e-01 6.56904161e-01 6.71921968e-01 -5.16014576e-01
3.36595744e-01 3.27280104e-01 -4.16098088e-01 -9.31911469e-01
-1.02848327e+00 -3.12172264e-01 -4.77616727e-01 -2.59569675e-01
1.07288611e+00 -1.28969502e+00 -7.68401086e-01 6.65453970e-01
-1.30711567e+00 -7.37415969e-01 1.17581449e-01 2.44746745e-01
-7.68570840e-01 2.58259684e-01 -6.91497386e-01 -5.93125939e-01
-1.14957660e-01 -1.44127536e+00 1.38632882e+00 1.58885151e-01
-2.81076226e-02 -9.88863051e-01 1.77751243e-01 8.78609240e-01
5.63642859e-01 -2.37677708e-01 8.95290613e-01 -6.46563828e-01
-7.75023401e-01 2.57405490e-01 -5.10868132e-01 3.51730764e-01
-3.62637937e-01 -1.27840430e-01 -1.30454159e+00 -2.60291964e-01
1.46959171e-01 -9.47205067e-01 1.32727575e+00 6.73737884e-01
1.04877508e+00 -8.98812786e-02 -2.18624011e-01 5.90676248e-01
1.38103461e+00 -8.22337642e-02 7.65182853e-01 2.15593517e-01
8.28766942e-01 6.43484592e-01 -1.17334828e-03 6.12002574e-02
8.19946826e-01 6.44344747e-01 4.44525689e-01 1.87294502e-02
-2.32608914e-01 -1.22446924e-01 6.95543289e-01 6.56027496e-01
1.32434815e-01 -4.49788868e-01 -1.13275611e+00 3.42652380e-01
-1.72167623e+00 -7.40914941e-01 -8.48740488e-02 1.98296237e+00
8.42186034e-01 -3.31032909e-02 6.85733631e-02 -4.83766824e-01
5.89269698e-01 3.03762853e-01 -5.95261931e-01 -3.88858438e-01
-2.69176662e-01 -2.51914442e-01 4.03925389e-01 5.00716388e-01
-1.15456736e+00 9.55247521e-01 6.95645571e+00 5.80832243e-01
-1.05166459e+00 1.98733047e-01 6.67102456e-01 -9.89394961e-04
-2.65548974e-01 9.69417319e-02 -7.55545855e-01 2.10365355e-01
1.19066858e+00 3.60756785e-01 4.06809121e-01 6.01694703e-01
-2.53969734e-03 -3.34575266e-01 -1.37444282e+00 1.12202692e+00
5.47322094e-01 -1.02005160e+00 2.18334645e-01 -2.60766828e-03
4.48351234e-01 5.11357605e-01 1.60441339e-01 4.29266095e-01
3.43863666e-01 -1.09666479e+00 7.39507496e-01 6.82774901e-01
6.29028797e-01 -3.63444500e-02 3.07335675e-01 3.86672646e-01
-6.69800043e-01 -1.79423034e-01 -1.01160243e-01 1.90130398e-01
2.07761109e-01 3.23582023e-01 -4.46234912e-01 2.13589177e-01
5.36753356e-01 4.94877249e-01 -7.02044010e-01 9.47252214e-01
-4.21116352e-02 6.78052008e-01 -4.18747589e-02 1.93190157e-01
3.25392962e-01 -1.29889116e-01 2.21480370e-01 1.38920212e+00
-1.93048537e-01 -1.32752836e-01 -2.83637904e-02 6.54181540e-01
-1.50971264e-01 -1.94392890e-01 -5.80360949e-01 -1.38428882e-01
2.36901995e-02 1.16059721e+00 -2.38051414e-01 -3.18872422e-01
-8.59935403e-01 1.22272182e+00 4.49952155e-01 6.77952588e-01
-8.35859478e-01 1.74751550e-01 3.95139486e-01 -1.02810323e-01
3.26487303e-01 -3.53036791e-01 -1.87338799e-01 -1.42703545e+00
-8.95946920e-02 -1.05920887e+00 3.44017029e-01 -1.21319044e+00
-1.30474949e+00 5.21893442e-01 -1.59035742e-01 -6.63078368e-01
-3.47509682e-01 -8.50310087e-01 -4.45883542e-01 6.67795062e-01
-1.59618032e+00 -1.37212026e+00 -2.92586833e-01 7.05894291e-01
7.99402058e-01 -2.10659713e-01 6.03121877e-01 3.46165627e-01
-7.09199190e-01 6.04593754e-01 1.45616114e-01 3.07065457e-01
1.05361152e+00 -1.26894879e+00 2.12034181e-01 6.49191916e-01
2.96152562e-01 3.71529073e-01 7.04462230e-01 -1.82193011e-01
-1.63112187e+00 -9.08916712e-01 7.88841009e-01 -7.78254330e-01
7.14624584e-01 -5.53556621e-01 -9.72911298e-01 1.06890333e+00
8.35999548e-01 -3.37862611e-01 5.91724634e-01 1.97435364e-01
-7.16501355e-01 6.00563316e-03 -5.07062316e-01 4.75541443e-01
6.06851757e-01 -9.22579706e-01 -5.71860731e-01 4.25237030e-01
6.43573999e-01 -3.47193360e-01 -5.32324851e-01 1.82148725e-01
5.70993483e-01 -1.07020450e+00 1.01256609e+00 -6.79475844e-01
6.73068762e-01 1.13307968e-01 -2.24071175e-01 -1.21673715e+00
-2.42763162e-01 -4.63857591e-01 -2.78687645e-02 1.15762484e+00
4.93939400e-01 -4.66626227e-01 3.66377532e-01 7.93521881e-01
-2.44238466e-01 -4.58758742e-01 -5.96003234e-01 -4.00842786e-01
1.71378627e-01 -3.54481727e-01 -2.42147014e-01 6.88358307e-01
-3.85113150e-01 1.16160691e+00 -5.79143107e-01 3.72678973e-02
5.66307962e-01 -6.29751012e-02 7.93749988e-01 -8.49904120e-01
-5.48715293e-01 -3.61207277e-01 1.74672976e-01 -1.41816688e+00
1.03523761e-01 -6.49105489e-01 4.47271429e-02 -1.51832187e+00
6.23339713e-01 1.43955514e-01 2.18395554e-02 5.04637122e-01
-2.20797971e-01 8.22677463e-02 3.58697891e-01 3.85137111e-01
-8.48344386e-01 5.22291005e-01 1.19212699e+00 -1.43423244e-01
1.43985391e-01 -1.53581589e-01 -7.91678071e-01 6.99047208e-01
5.23090959e-01 -1.80545017e-01 -3.87100756e-01 -1.11398423e+00
3.05667549e-01 2.36525863e-01 7.14551330e-01 -6.80705249e-01
5.00836194e-01 1.61000833e-01 2.24296927e-01 -2.85842985e-01
6.66716754e-01 -6.09421790e-01 -1.75595015e-01 -1.52044874e-02
-6.73796415e-01 1.43284187e-01 2.56240308e-01 5.36097825e-01
-2.79261410e-01 -2.62353629e-01 7.74915636e-01 -2.20360219e-01
-9.15531397e-01 2.33710725e-02 -5.65661728e-01 3.17743391e-01
6.73504114e-01 3.70549820e-02 -6.03617430e-01 -6.09891415e-01
-7.50631928e-01 3.90309632e-01 2.64282584e-01 5.50755739e-01
4.63757873e-01 -8.79961073e-01 -8.36420655e-01 4.75170910e-02
1.86691061e-01 -2.49026924e-01 2.43673846e-01 9.06921208e-01
-3.37950736e-02 5.36653101e-01 -2.76810043e-02 -7.76536167e-01
-1.17581284e+00 4.65926081e-01 5.54187357e-01 -1.26498014e-01
-2.78129280e-01 8.50252748e-01 6.10113740e-01 -4.53776449e-01
3.87149066e-01 -5.16649857e-02 -1.69966251e-01 2.09671408e-01
3.67486417e-01 -1.30282596e-01 -6.93629906e-02 -5.22758067e-01
-3.38060588e-01 5.62383711e-01 -1.70696229e-01 -3.48085016e-01
1.08548510e+00 -3.09651673e-01 8.20250157e-03 6.12986863e-01
1.21293616e+00 -4.59320277e-01 -1.36341727e+00 -4.13687050e-01
-8.09209272e-02 2.27576829e-02 4.02016379e-03 -9.59369779e-01
-8.74058902e-01 1.21291137e+00 4.10491288e-01 2.08443671e-01
1.15090811e+00 4.98386085e-01 4.30416346e-01 5.43828368e-01
-1.07386164e-01 -9.33450222e-01 4.70710546e-01 8.37135434e-01
9.64900076e-01 -1.52135289e+00 -4.25854474e-01 -2.09093876e-02
-1.27716470e+00 8.14098537e-01 6.37435198e-01 -3.89848812e-03
3.92694920e-01 1.92623764e-01 2.92289704e-01 -1.94762498e-01
-1.20146585e+00 -3.82617593e-01 3.47449929e-01 5.28562963e-01
5.67094624e-01 -2.80823946e-01 4.41444486e-01 3.83984357e-01
1.34036187e-02 -3.17657925e-02 4.17674631e-01 5.45461833e-01
-4.56173480e-01 -9.86747086e-01 -3.32435846e-01 4.39570546e-01
-3.68644923e-01 -3.02338660e-01 -4.25618380e-01 5.11615932e-01
-2.33679354e-01 1.02467000e+00 1.33384168e-01 -2.01924577e-01
3.79975080e-01 3.48062187e-01 6.54967368e-01 -5.99045277e-01
-6.11135125e-01 1.48123011e-01 1.33864969e-01 -4.86243665e-01
-5.02899349e-01 -6.27819717e-01 -5.80783069e-01 1.79450169e-01
-3.02125484e-01 -1.62924439e-01 7.45420039e-01 1.22709179e+00
4.61647809e-01 5.57954907e-01 2.25447178e-01 -7.08322644e-01
-7.87085593e-01 -1.03229809e+00 -1.59610555e-01 2.49395534e-01
6.09042764e-01 -3.58489960e-01 -1.34167671e-01 5.55656016e-01] | [10.939334869384766, 1.61617910861969] |
91c989a8-7693-48b1-ad77-2edf77a8a432 | hybridnets-end-to-end-perception-network-1 | 2203.09035 | null | https://arxiv.org/abs/2203.09035v1 | https://arxiv.org/pdf/2203.09035v1.pdf | HybridNets: End-to-End Perception Network | End-to-end Network has become increasingly important in multi-tasking. One prominent example of this is the growing significance of a driving perception system in autonomous driving. This paper systematically studies an end-to-end perception network for multi-tasking and proposes several key optimizations to improve accuracy. First, the paper proposes efficient segmentation head and box/class prediction networks based on weighted bidirectional feature network. Second, the paper proposes automatically customized anchor for each level in the weighted bidirectional feature network. Third, the paper proposes an efficient training loss function and training strategy to balance and optimize network. Based on these optimizations, we have developed an end-to-end perception network to perform multi-tasking, including traffic object detection, drivable area segmentation and lane detection simultaneously, called HybridNets, which achieves better accuracy than prior art. In particular, HybridNets achieves 77.3 mean Average Precision on Berkeley DeepDrive Dataset, outperforms lane detection with 31.6 mean Intersection Over Union with 12.83 million parameters and 15.6 billion floating-point operations. In addition, it can perform visual perception tasks in real-time and thus is a practical and accurate solution to the multi-tasking problem. Code is available at https://github.com/datvuthanh/HybridNets. | ['Hung Phan', 'Bao Ngo', 'Dat Vu'] | 2022-03-17 | hybridnets-end-to-end-perception-network | https://arxiv.org/abs/2203.09035 | https://arxiv.org/ftp/arxiv/papers/2203/2203.09035.pdf | null | ['lane-detection', 'drivable-area-detection'] | ['computer-vision', 'computer-vision'] | [-1.87321469e-01 -1.23026185e-01 -4.07413661e-01 -6.54460907e-01
-8.25524390e-01 -4.01003271e-01 1.22286521e-01 -2.06414163e-01
-6.75487041e-01 5.27841866e-01 -2.56892473e-01 -5.82851708e-01
5.60383797e-02 -7.69627512e-01 -8.96033347e-01 -3.75882626e-01
1.88513830e-01 3.20132643e-01 7.31604695e-01 -3.56077552e-01
3.04845065e-01 4.77662653e-01 -1.72679412e+00 1.25032365e-01
9.80774403e-01 1.42445719e+00 3.98914754e-01 1.04930747e+00
2.46603400e-01 5.44495881e-01 -4.48666602e-01 -5.32548785e-01
4.22285616e-01 5.70132852e-01 -2.96457976e-01 -2.50871152e-01
1.11761677e+00 -3.56247842e-01 -4.72580791e-01 9.39138114e-01
7.92253256e-01 2.71153152e-01 3.67924571e-01 -1.79824269e+00
-6.53723031e-02 1.75891086e-01 -8.07705998e-01 3.77380222e-01
-4.41543221e-01 4.67389315e-01 9.91987109e-01 -8.43520164e-01
7.66159073e-02 1.48056173e+00 6.89253867e-01 2.08490208e-01
-8.98324132e-01 -9.58560467e-01 1.69705331e-01 6.63003087e-01
-1.40138757e+00 -5.58944762e-01 5.36974609e-01 -4.12511319e-01
9.74793971e-01 1.06433742e-01 4.84468102e-01 6.34322643e-01
6.50623083e-01 9.36666787e-01 7.49605775e-01 9.71019194e-02
-2.21097037e-01 6.90748841e-02 2.75150925e-01 9.45314288e-01
3.05531204e-01 2.78158128e-01 -4.34666544e-01 3.88378084e-01
4.49687362e-01 -1.51684791e-01 3.21014345e-01 -2.63425797e-01
-1.11800742e+00 7.44688451e-01 5.81504226e-01 -4.20423746e-01
-6.22581579e-02 6.04825616e-01 7.29423285e-01 -1.71326790e-02
3.43579561e-01 -1.50745645e-01 -3.88265669e-01 -2.70489126e-01
-6.69411063e-01 4.69273299e-01 4.11620736e-01 1.10047388e+00
1.11716902e+00 1.48490906e-01 -2.62302727e-01 1.03468561e+00
4.05657023e-01 8.78912747e-01 3.20107304e-02 -1.37109697e+00
7.30463743e-01 4.13184911e-01 -2.28504702e-01 -1.02075374e+00
-6.87350631e-01 -5.11355639e-01 -7.47491956e-01 5.51735878e-01
3.59266818e-01 -4.49936956e-01 -8.91796529e-01 1.46039367e+00
3.97359937e-01 2.42028460e-01 1.50310586e-03 9.41853940e-01
9.01525617e-01 5.90085268e-01 1.12752989e-01 4.99267697e-01
1.50037766e+00 -1.53134263e+00 -3.47500026e-01 -6.89342856e-01
6.18699908e-01 -7.60665834e-01 8.83151054e-01 5.17602682e-01
-8.97406399e-01 -1.05893505e+00 -1.14121938e+00 -5.20694911e-01
-5.68008900e-01 2.72867888e-01 4.95195538e-01 6.71419024e-01
-1.01851177e+00 8.52416381e-02 -4.53821182e-01 -1.27797276e-01
6.08408868e-01 3.08222055e-01 3.25854085e-02 -3.36768143e-02
-1.08042145e+00 9.36770320e-01 3.97958428e-01 1.40079647e-01
-8.57649446e-01 -7.67086089e-01 -8.45794797e-01 -2.97999159e-02
5.94110966e-01 -6.95079446e-01 1.42597461e+00 -4.92123485e-01
-1.29207969e+00 6.85400605e-01 -3.84310961e-01 -7.84058630e-01
6.27485037e-01 -3.06948304e-01 -5.40200412e-01 -5.57954535e-02
3.76164585e-01 1.17308342e+00 6.36322618e-01 -1.17338002e+00
-1.45772183e+00 -3.70303780e-01 -7.12112710e-03 2.36719951e-01
3.32209542e-02 -3.78850251e-01 -7.74090707e-01 -2.76129961e-01
-2.09762946e-01 -8.18477809e-01 -3.79918456e-01 2.05991998e-01
-6.07693970e-01 -3.63215715e-01 1.15036571e+00 -3.73396605e-01
9.63223934e-01 -2.31808043e+00 -4.62345868e-01 1.86316133e-01
5.85616887e-01 2.94554800e-01 -1.80161044e-01 -2.05264941e-01
3.26862425e-01 -1.19481348e-01 9.80951861e-02 -5.52455425e-01
1.98492363e-01 2.37895638e-01 -2.74167866e-01 5.01082420e-01
-1.78908154e-01 1.06205332e+00 -6.28515780e-01 -7.07186162e-01
6.70812130e-01 3.95644337e-01 -3.76385182e-01 -1.55869231e-01
-5.47951609e-02 4.65395600e-02 -3.41264963e-01 7.60212183e-01
9.01821494e-01 3.46059469e-03 -4.32446361e-01 -3.61534774e-01
-4.26089257e-01 -6.02097511e-02 -1.06120014e+00 1.47388923e+00
-5.83830059e-01 1.18710470e+00 1.62513331e-01 -9.24068511e-01
1.07447684e+00 -2.93262571e-01 2.65178710e-01 -1.12556040e+00
2.53253400e-01 1.41819447e-01 -7.50884116e-02 -3.18253577e-01
9.41908836e-01 3.52973849e-01 -2.58055508e-01 -2.49759689e-01
-3.33540469e-01 -1.72201663e-01 3.32255334e-01 1.09461375e-01
8.18092585e-01 -9.45657417e-02 2.20558159e-02 -2.42713898e-01
5.49153328e-01 1.67234153e-01 8.27756166e-01 7.98879325e-01
-7.51569867e-01 2.89128035e-01 4.42623466e-01 -5.54737926e-01
-9.83171105e-01 -1.16789126e+00 -1.11431465e-01 1.46572244e+00
6.47495687e-01 -5.36759458e-02 -5.20606339e-01 -4.13677216e-01
5.10611117e-01 7.27927089e-01 -3.07983667e-01 1.88245941e-02
-7.05713391e-01 -4.02248263e-01 7.53100812e-01 6.76248014e-01
1.03661382e+00 -6.56495154e-01 -7.89409518e-01 1.85916603e-01
-2.84202904e-01 -1.45237958e+00 -6.05809689e-01 1.74572784e-02
-5.86474478e-01 -9.97779548e-01 -2.41870090e-01 -7.62797713e-01
1.01578474e-01 8.02248955e-01 9.95309770e-01 -1.95715427e-01
-3.57913822e-01 6.08561076e-02 1.33281738e-01 -6.37607992e-01
4.21413369e-02 2.65563160e-01 -9.03609470e-02 -7.55015388e-02
4.51023877e-01 -2.85048932e-01 -7.60679841e-01 7.44983554e-01
-2.61741728e-01 2.72933960e-01 7.61687517e-01 4.23370302e-01
8.45188558e-01 -1.81998894e-01 4.93509769e-01 -5.11295974e-01
3.14244717e-01 -3.56473416e-01 -1.07145154e+00 2.67642867e-02
-5.90543628e-01 -2.36782119e-01 5.98426640e-01 2.53149662e-02
-8.25921535e-01 7.25653544e-02 -2.99937755e-01 -5.27912855e-01
-3.23658675e-01 -1.09083943e-01 -5.77885360e-02 -2.77554572e-01
5.40294647e-01 2.14285050e-02 7.55523592e-02 -5.91279827e-02
5.18715978e-01 7.93621123e-01 7.27416694e-01 -2.91505098e-01
5.44931531e-01 6.24320865e-01 1.91910267e-01 -1.05020952e+00
-8.90563250e-01 -6.79605067e-01 -4.22839403e-01 -6.03956699e-01
1.06331277e+00 -1.18094528e+00 -1.21747112e+00 5.19785464e-01
-1.07639265e+00 -6.68281674e-01 9.07369852e-02 4.48869795e-01
-7.00312257e-01 2.28581578e-01 -3.34197611e-01 -6.79153264e-01
-3.83729279e-01 -1.46220374e+00 1.16252625e+00 3.43274862e-01
2.62046695e-01 -7.04935014e-01 -3.61415535e-01 7.69198298e-01
4.49889928e-01 1.67238638e-01 3.53672147e-01 -1.71182036e-01
-9.03129637e-01 1.15606636e-02 -8.31449807e-01 4.35492128e-01
-3.90836209e-01 1.41706318e-01 -1.01461482e+00 -1.29316598e-01
-6.35364890e-01 -1.23581760e-01 1.25832844e+00 8.01968575e-01
1.54082310e+00 1.79793850e-01 -6.20019555e-01 8.84169102e-01
1.30010486e+00 2.78164327e-01 5.06988525e-01 4.25990224e-01
9.55264509e-01 4.81404632e-01 8.80057871e-01 2.72109061e-01
1.16699684e+00 7.15548873e-01 7.06511319e-01 -2.37619773e-01
-3.27034682e-01 -3.26580070e-02 2.26619080e-01 3.26845050e-01
1.83353469e-01 -3.62375945e-01 -1.07293618e+00 5.11696100e-01
-2.05920434e+00 -8.24912131e-01 -5.15576661e-01 1.92242634e+00
8.81470963e-02 6.64709210e-01 4.42284107e-01 -1.76717758e-01
7.87476063e-01 2.56733567e-01 -9.18017447e-01 -5.65506518e-01
-2.93489676e-02 -3.29509676e-01 1.32084441e+00 6.66506648e-01
-1.26096117e+00 1.21506846e+00 5.61048031e+00 1.24289703e+00
-9.36286867e-01 2.19093129e-01 8.88105750e-01 -1.60801724e-01
1.79081574e-01 -3.38754356e-01 -1.48942304e+00 4.88952518e-01
8.83827567e-01 -2.15757396e-02 1.61357090e-01 1.10834706e+00
5.25336444e-01 -4.07251567e-01 -6.88745677e-01 1.13904929e+00
-4.98462170e-02 -1.32078207e+00 -3.03008914e-01 1.60720646e-01
3.62833261e-01 5.72060883e-01 2.51563549e-01 4.22327518e-01
3.90204042e-01 -8.31496298e-01 9.67152596e-01 3.88572961e-01
8.53268623e-01 -1.07965016e+00 6.25137150e-01 3.06438088e-01
-1.57687569e+00 -3.69497508e-01 -3.84443641e-01 -1.39029566e-02
4.57605898e-01 6.90513372e-01 -6.74027085e-01 2.06757799e-01
8.42116594e-01 7.08326042e-01 -6.91683710e-01 1.23445892e+00
1.64720751e-02 3.75137061e-01 -3.42894018e-01 -1.68844476e-01
4.57888931e-01 -4.32707295e-02 4.93719906e-01 1.46618927e+00
1.99053422e-01 -2.92299956e-01 4.36628819e-01 4.96663570e-01
2.41625328e-02 -2.42102951e-01 -5.35535634e-01 6.93056107e-01
5.56892633e-01 1.62932754e+00 -5.66265225e-01 -4.86262232e-01
-4.60107982e-01 4.33171451e-01 2.98586220e-01 3.81303698e-01
-1.39001918e+00 -7.78456092e-01 1.26821291e+00 -1.69599371e-03
4.79567111e-01 -4.28171873e-01 -7.20203876e-01 -5.45250595e-01
6.54588267e-02 -3.35857332e-01 1.57222793e-01 -6.71970010e-01
-9.51658249e-01 3.60358030e-01 -3.30655947e-02 -1.12848079e+00
2.93330669e-01 -8.64793837e-01 -5.47739804e-01 6.59230709e-01
-1.95663762e+00 -1.09442556e+00 -6.39242768e-01 4.57107365e-01
8.14376354e-01 -3.05025637e-01 -8.09784681e-02 7.26005077e-01
-9.11814034e-01 8.51042926e-01 -1.82200655e-01 -4.75586951e-02
7.83699811e-01 -1.13352633e+00 7.84163713e-01 8.34164023e-01
-4.20894325e-01 -1.64000124e-01 5.32737911e-01 -3.20262372e-01
-1.25740874e+00 -1.55071616e+00 6.93490267e-01 -3.48083794e-01
6.62759542e-01 -4.20793802e-01 -5.47441125e-01 5.34207165e-01
1.16021417e-01 8.36515129e-02 2.52710164e-01 -1.22804731e-01
-2.72660136e-01 -7.98290670e-01 -1.10131979e+00 6.40818417e-01
1.14893639e+00 -1.48352772e-01 2.48353947e-02 3.47693503e-01
8.83933783e-01 -8.09756577e-01 -5.11484802e-01 3.44758451e-01
6.35392606e-01 -1.06818151e+00 1.10464776e+00 2.18206514e-02
2.41013259e-01 -5.00415862e-01 -2.45847180e-01 -1.01120651e+00
-3.88158858e-01 -4.80686724e-01 -8.04882720e-02 7.22363532e-01
5.96573412e-01 -1.03725410e+00 8.52739155e-01 2.43775383e-01
-8.11556935e-01 -9.83663380e-01 -1.13566971e+00 -7.75409102e-01
-2.99012829e-02 -8.59569490e-01 5.31738877e-01 2.24191964e-01
-5.98726213e-01 4.14568901e-01 -2.89300174e-01 3.05595040e-01
9.83557284e-01 5.39953262e-03 1.20214331e+00 -1.13972497e+00
1.38809413e-01 -7.76284456e-01 -5.68655729e-01 -1.69460428e+00
1.77963823e-01 -7.46835887e-01 2.65227228e-01 -1.61320031e+00
-2.08093643e-01 -6.07209027e-01 -4.59300280e-02 5.15116990e-01
1.02174707e-01 3.54294211e-01 2.53166556e-01 -2.35420659e-01
-9.82205391e-01 4.09742206e-01 1.48941481e+00 -1.45886049e-01
2.22046711e-02 1.47825032e-01 -7.56779611e-01 6.51922822e-01
1.37213016e+00 -8.36999640e-02 -1.78096101e-01 -6.06291950e-01
1.00900598e-01 -1.18412107e-01 7.19035566e-01 -1.30348063e+00
5.83364546e-01 -1.25537142e-01 1.79504707e-01 -1.26369524e+00
5.96512794e-01 -6.27342522e-01 -4.32147980e-01 4.19657290e-01
2.33433824e-02 3.01108032e-01 6.37824476e-01 5.47828496e-01
-9.97448489e-02 1.83297828e-01 9.81024146e-01 2.22877562e-01
-1.62153947e+00 4.05943453e-01 -4.18247283e-01 2.04242244e-01
1.36376810e+00 -6.86767399e-01 -7.20377862e-01 -4.40471917e-02
-3.42860460e-01 9.97372746e-01 -6.51721284e-02 7.15620875e-01
7.05396056e-01 -1.28372419e+00 -6.65003061e-01 2.22842693e-01
1.21704504e-01 1.83915347e-01 5.36108017e-01 8.72162759e-01
-5.28738618e-01 5.71851790e-01 -2.28138000e-01 -9.07468677e-01
-1.31184256e+00 9.65885371e-02 4.51710731e-01 1.29452229e-01
-4.30022359e-01 8.62928391e-01 2.07057059e-01 -4.47150767e-01
3.24585080e-01 -4.52680558e-01 -7.49896765e-02 -4.97795716e-02
3.96294355e-01 1.02399313e+00 -3.89694832e-02 -8.31825733e-01
-3.35996419e-01 8.12008083e-01 -5.74409217e-02 3.65342498e-02
8.30669999e-01 -3.84864151e-01 1.59132496e-01 2.84225971e-01
1.26849091e+00 -3.31719548e-01 -1.66428339e+00 -7.61034191e-02
-4.29540455e-01 -5.80891907e-01 4.06957626e-01 -6.44388437e-01
-1.24313319e+00 9.59262550e-01 8.38414669e-01 1.99722797e-02
8.49164665e-01 -1.79116100e-01 1.15440083e+00 5.41931510e-01
2.56087810e-01 -1.45098972e+00 -1.17173329e-01 9.10748780e-01
5.49136043e-01 -1.71005654e+00 -1.74278691e-01 -5.13028026e-01
-6.48005247e-01 9.04760480e-01 1.16608357e+00 -2.21564732e-02
6.53636575e-01 3.53009015e-01 1.39971733e-01 -1.66323036e-01
-6.94663763e-01 -4.06190395e-01 3.06624144e-01 5.26391447e-01
-3.49909887e-02 3.68639052e-01 1.28881648e-01 5.67634739e-02
-4.27124143e-01 -2.21028790e-01 2.31138214e-01 4.46731836e-01
-9.52061594e-01 -6.10843360e-01 -2.25473046e-01 7.12068439e-01
-5.26561998e-02 8.24843273e-02 2.07820654e-01 6.98139310e-01
4.91673946e-01 1.16108119e+00 2.71775603e-01 -6.34450018e-01
6.89123690e-01 -4.71128643e-01 -2.84199212e-02 -2.43846759e-01
-3.82189542e-01 -2.06228063e-01 2.18777210e-01 -9.11531091e-01
1.13285296e-01 -3.80408496e-01 -1.28859639e+00 -7.03554213e-01
-1.05965227e-01 -3.67065191e-01 8.07699978e-01 5.90291679e-01
7.55064607e-01 7.80469358e-01 5.83558381e-01 -9.85509276e-01
-2.32970461e-01 -6.37848020e-01 -3.32358956e-01 -2.33539417e-01
5.72290778e-01 -8.09093118e-01 -2.19746798e-01 -3.29834312e-01] | [8.053057670593262, -1.3039683103561401] |
06091c2c-7149-4bbb-b3bc-e46fd8ddc7f5 | distributed-feature-selection-for-high | 2205.07932 | null | https://arxiv.org/abs/2205.07932v2 | https://arxiv.org/pdf/2205.07932v2.pdf | DDAC-SpAM: A Distributed Algorithm for Fitting High-dimensional Sparse Additive Models with Feature Division and Decorrelation | Distributed statistical learning has become a popular technique for large-scale data analysis. Most existing work in this area focuses on dividing the observations, but we propose a new algorithm, DDAC-SpAM, which divides the features under a high-dimensional sparse additive model. Our approach involves three steps: divide, decorrelate, and conquer. The decorrelation operation enables each local estimator to recover the sparsity pattern for each additive component without imposing strict constraints on the correlation structure among variables. The effectiveness and efficiency of the proposed algorithm are demonstrated through theoretical analysis and empirical results on both synthetic and real data. The theoretical results include both the consistent sparsity pattern recovery as well as statistical inference for each additive functional component. Our approach provides a practical solution for fitting sparse additive models, with promising applications in a wide range of domains. | ['Ruiyang Wu', 'Yang Feng', 'Yong Zhou', 'Yifan He'] | 2022-05-16 | null | null | null | null | ['additive-models'] | ['methodology'] | [ 1.42216519e-01 -4.93908584e-01 -2.25503474e-01 -3.80296290e-01
-1.01805913e+00 -2.33179182e-01 1.74206078e-01 4.57328819e-02
1.78786114e-01 7.72485912e-01 2.50032425e-01 2.37312183e-01
-5.64291954e-01 -5.09385467e-01 -5.69202542e-01 -1.16436803e+00
-3.30250800e-01 2.49144137e-01 -4.76827584e-02 3.30767661e-01
6.95934966e-02 1.64853886e-01 -1.20789158e+00 1.03355199e-01
1.22770917e+00 1.25189626e+00 1.77716866e-01 2.54608721e-01
1.46647021e-01 7.39050984e-01 -3.64226550e-01 -2.00347528e-01
5.34888923e-01 -5.10134459e-01 8.03215355e-02 3.70228231e-01
3.74778748e-01 -1.66662484e-01 -1.70440122e-01 1.29201245e+00
4.11337495e-01 5.40532880e-02 6.67121172e-01 -1.24270725e+00
-4.13635314e-01 7.41516829e-01 -1.23789239e+00 -9.33117196e-02
-4.64963727e-02 -3.73143226e-01 9.82359350e-01 -1.01053214e+00
5.61735556e-02 1.17581618e+00 9.14819121e-01 -3.79951417e-01
-1.21297133e+00 -7.99034834e-01 1.16602732e-02 2.39145100e-01
-1.82568252e+00 -5.24283230e-01 9.84889328e-01 -4.58311975e-01
1.09225355e-01 1.73596457e-01 1.87382147e-01 4.85861063e-01
3.94566894e-01 8.18237662e-01 1.31173098e+00 -2.42854282e-01
6.65193737e-01 -4.79003005e-02 3.32156986e-01 5.48363805e-01
8.37664545e-01 -4.71129306e-02 -7.52746880e-01 -8.48658562e-01
7.61971116e-01 3.79071921e-01 -1.00102670e-01 -6.47331893e-01
-9.27704692e-01 9.46871340e-01 1.58133671e-01 1.50960952e-01
-8.88236880e-01 5.78857325e-02 3.01361263e-01 1.86527014e-01
8.42376530e-01 -1.76676854e-01 -4.09241378e-01 3.13116670e-01
-1.25760794e+00 1.99289769e-01 6.70140505e-01 9.00712550e-01
8.18119347e-01 3.27268839e-01 1.03820637e-01 1.04670608e+00
4.61753279e-01 8.51137221e-01 2.76658654e-01 -1.16753924e+00
4.07633901e-01 1.72424227e-01 7.36658499e-02 -1.67716157e+00
-2.51662940e-01 -7.45107591e-01 -1.47332835e+00 -1.54902086e-01
2.05730647e-01 -4.37024713e-01 -1.21710733e-01 1.58067131e+00
5.52777708e-01 7.75169611e-01 -7.75390342e-02 8.52844238e-01
2.82509923e-01 5.88157535e-01 -3.11759353e-01 -7.40667284e-01
1.06142950e+00 -6.72726154e-01 -9.44724083e-01 -1.45976633e-01
1.58563524e-01 -7.39364147e-01 3.03178281e-01 6.60028756e-01
-9.30079579e-01 -4.72410440e-01 -7.98044920e-01 3.62086028e-01
4.22020435e-01 4.22638506e-01 9.10151958e-01 4.15173501e-01
-8.07941079e-01 1.54883996e-01 -1.12278390e+00 6.29499480e-02
3.36675406e-01 3.06306511e-01 -3.17866355e-01 -3.52829069e-01
-7.65089154e-01 2.18722507e-01 8.29013716e-03 2.50393778e-01
-5.48386872e-01 -8.05021703e-01 -8.22285891e-01 3.07071835e-01
4.78086650e-01 -6.90484107e-01 9.13424790e-01 -1.04426169e+00
-1.11998987e+00 1.49102896e-01 -6.96567714e-01 -5.54526627e-01
1.77842602e-01 -3.83219868e-01 -2.73975015e-01 -5.37936799e-02
2.82714784e-01 -3.16564709e-01 1.20298767e+00 -1.21125126e+00
-4.71154749e-01 -6.45351589e-01 -7.60363519e-01 3.12800743e-02
-4.62049842e-01 -1.86536741e-02 -3.13841522e-01 -1.06523323e+00
5.22108138e-01 -6.72729969e-01 -5.27188599e-01 -7.20116720e-02
-2.79239714e-01 4.97388840e-03 6.92237675e-01 -6.66957736e-01
1.48322225e+00 -2.25139666e+00 1.19720951e-01 5.17269313e-01
4.33147609e-01 -3.30284208e-01 -4.86870259e-02 4.91441786e-01
-3.44340689e-02 -4.73121047e-01 -5.04992008e-01 -3.86903167e-01
-2.60145009e-01 -1.73246235e-01 -4.13316309e-01 9.34896708e-01
-2.38478690e-01 3.34706247e-01 -6.53450549e-01 -4.05637980e-01
5.61740734e-02 3.79886806e-01 -7.08305478e-01 1.82059929e-01
2.84555763e-01 1.52739286e-01 -5.78214407e-01 6.26084745e-01
1.27418721e+00 -3.69015425e-01 7.00906992e-01 -1.61574379e-01
2.36441046e-02 -2.68720269e-01 -1.88840854e+00 1.35226381e+00
-2.91544497e-01 2.29350016e-01 7.21731246e-01 -1.42512369e+00
9.21699345e-01 1.36489213e-01 9.57432747e-01 -3.25029045e-01
1.20551519e-01 3.60961109e-01 7.40683824e-02 -2.43649080e-01
2.24422902e-01 -2.09507108e-01 -2.05286160e-01 5.16874015e-01
-6.95744753e-02 2.70629317e-01 3.30360234e-02 2.37217113e-01
1.04086149e+00 -4.12232369e-01 7.81488717e-01 -5.95942795e-01
4.28540200e-01 -2.56144375e-01 9.16621864e-01 7.08529472e-01
5.50429001e-02 3.24222922e-01 3.33464354e-01 -1.81344494e-01
-6.40716016e-01 -9.40465808e-01 -9.89216715e-02 9.82407808e-01
2.02799171e-01 -3.63377452e-01 -5.60015142e-01 -4.16073442e-01
4.35652941e-01 3.31986219e-01 -4.45918947e-01 7.42796883e-02
-3.53716075e-01 -1.01763368e+00 -5.93267418e-02 4.55435038e-01
3.44209850e-01 -3.24163675e-01 2.76252329e-02 1.08643927e-01
-1.12020582e-01 -9.00201023e-01 -6.58318281e-01 1.53920427e-01
-9.90387142e-01 -1.13598597e+00 -5.22586167e-01 -5.69804847e-01
7.76616037e-01 9.69395876e-01 5.58136284e-01 -2.85219431e-01
-1.41830996e-01 1.98252559e-01 -7.99855590e-02 -1.45278797e-01
8.00913945e-02 -3.35367799e-01 3.64398479e-01 6.42096519e-01
1.75510183e-01 -8.05635512e-01 -4.53161567e-01 5.39602995e-01
-7.35163510e-01 -3.17805767e-01 8.99768174e-01 8.93304527e-01
8.36882174e-01 5.21720171e-01 7.51852572e-01 -1.02628124e+00
6.61272109e-01 -1.03664398e+00 -7.87574410e-01 6.60033822e-02
-4.85755324e-01 -2.54250262e-02 8.39871407e-01 -2.69233048e-01
-1.15886188e+00 2.26390511e-01 4.27088201e-01 -6.99025154e-01
2.18784288e-01 1.00122499e+00 -3.91517580e-01 7.65245184e-02
4.18786198e-01 4.12729114e-01 1.47082493e-01 -6.78759694e-01
3.51313472e-01 6.03084385e-01 5.10072708e-01 -7.26215482e-01
8.35935056e-01 5.50256312e-01 2.12291002e-01 -1.03372359e+00
-7.49194622e-01 -8.46610069e-01 -5.09086549e-01 2.66192824e-01
1.76572934e-01 -1.29239893e+00 -5.02288461e-01 6.14714324e-01
-8.34473372e-01 8.85945633e-02 -6.29267097e-02 7.50913560e-01
-2.57620960e-01 6.00081563e-01 -2.89803565e-01 -9.59064186e-01
-3.02447528e-01 -7.35690117e-01 9.37156856e-01 -9.62085873e-02
-4.45513092e-02 -1.07688665e+00 3.84251416e-01 2.41691411e-01
3.72079760e-01 -3.87583557e-03 6.04857743e-01 -7.02479839e-01
-4.79738742e-01 -4.43094015e-01 -2.70528734e-01 3.14341187e-01
3.64002883e-01 -1.95351511e-01 -4.91237909e-01 -4.06530887e-01
3.72688949e-01 -2.05965817e-01 7.77591228e-01 8.56245279e-01
1.22699881e+00 -5.30925751e-01 -3.70623112e-01 5.42528391e-01
1.37408638e+00 -4.86419909e-02 3.31541896e-01 -3.64446789e-01
5.42250216e-01 4.23339367e-01 6.46222293e-01 9.87729490e-01
2.30974019e-01 5.45842290e-01 1.97672844e-02 3.67015461e-03
3.01913917e-02 -1.86254770e-01 3.56712341e-01 1.33483994e+00
2.41705686e-01 -2.68186629e-01 -5.40321112e-01 5.65157175e-01
-2.19306850e+00 -1.13425839e+00 -4.47182178e-01 2.34767628e+00
6.75632954e-01 -4.98089105e-01 5.63286990e-03 -1.16374582e-01
8.90442848e-01 2.45788440e-01 -6.72381163e-01 1.71116501e-01
-3.26333374e-01 1.74158871e-01 5.00011146e-01 3.81789923e-01
-1.15965641e+00 4.54843462e-01 7.05514765e+00 1.11388159e+00
-8.89383852e-01 1.95357993e-01 5.85133910e-01 -3.24007459e-02
-1.55284882e-01 -3.48343775e-02 -6.28825903e-01 6.37237430e-01
7.10604250e-01 -2.91886747e-01 3.85794282e-01 1.07117069e+00
6.09389901e-01 -2.94204980e-01 -6.17024839e-01 1.20719373e+00
3.03328961e-01 -1.14141059e+00 -3.07075500e-01 1.00191981e-01
1.12708521e+00 -9.00113210e-02 -3.23692709e-02 -7.96379447e-02
4.54671860e-01 -7.11585879e-01 4.62390333e-01 6.58340156e-01
5.95390677e-01 -6.11881971e-01 6.96259081e-01 5.95604122e-01
-1.30675101e+00 -1.06153443e-01 -5.63211322e-01 -2.88139611e-01
4.16796245e-02 1.28543365e+00 -3.43353927e-01 7.14518368e-01
5.08294284e-01 1.00554407e+00 -2.38009959e-01 1.22062898e+00
3.57847959e-02 1.00965083e+00 -5.76121628e-01 1.73271403e-01
-3.85354072e-01 -7.22590387e-01 3.92640650e-01 8.82843733e-01
5.95103025e-01 9.36269909e-02 5.25164545e-01 4.08857852e-01
-1.84120107e-02 3.25216860e-01 -4.12920594e-01 1.26808032e-01
8.54025662e-01 1.20651436e+00 -5.24963498e-01 -2.98380733e-01
-6.13942146e-01 5.91250479e-01 1.75429523e-01 2.75955886e-01
-7.39908636e-01 -2.10886091e-01 6.01943254e-01 8.73747095e-02
6.79421365e-01 -3.83877397e-01 -7.82240868e-01 -1.41176558e+00
7.10629746e-02 -1.17011821e+00 4.97359395e-01 -3.11725885e-01
-1.68756998e+00 4.63948958e-02 -1.30383819e-01 -1.32164955e+00
-5.60002327e-02 -1.45479545e-01 -6.31636143e-01 7.83011854e-01
-8.94611180e-01 -9.09615874e-01 -3.25278908e-01 6.86378300e-01
4.90693152e-01 -3.71679962e-01 5.39623260e-01 3.08092684e-01
-7.58671105e-01 4.90840256e-01 9.78721440e-01 -2.63015538e-01
8.76056969e-01 -8.39754045e-01 -4.08067107e-01 8.35222363e-01
9.53901932e-02 9.20201004e-01 7.15945899e-01 -7.35758364e-01
-1.59529877e+00 -1.04029775e+00 5.29357016e-01 2.00519770e-01
9.28976655e-01 -3.10618609e-01 -8.95438194e-01 5.32303333e-01
1.71815693e-01 2.47977182e-01 1.13193369e+00 3.81198227e-01
-3.34343523e-01 -5.55431366e-01 -1.10293698e+00 2.71507557e-02
6.83783114e-01 -1.38025895e-01 -2.58018345e-01 3.57014209e-01
2.85810232e-01 -3.18265706e-02 -8.52558851e-01 2.09288895e-01
6.20657265e-01 -7.97476411e-01 8.69312286e-01 -3.94348323e-01
3.78895164e-01 -4.81856793e-01 -5.56510925e-01 -1.25629830e+00
-6.05295122e-01 -6.85226083e-01 -4.63638902e-01 1.11864734e+00
1.47799820e-01 -7.75166154e-01 6.16197467e-01 4.08334136e-01
-1.67583134e-02 -5.90727746e-01 -8.01492393e-01 -7.83934593e-01
-2.92937815e-01 -5.14417589e-01 2.73858100e-01 1.15370047e+00
-4.32992578e-02 2.61764288e-01 -8.76986265e-01 4.90785539e-01
1.09872007e+00 4.99603450e-01 1.05703330e+00 -1.18490040e+00
-7.25954473e-01 2.53278017e-03 -2.85840839e-01 -1.13073361e+00
1.36394590e-01 -7.24052727e-01 8.93444940e-02 -1.09668267e+00
6.57894015e-01 -5.49650252e-01 -3.57528090e-01 3.04020822e-01
-3.26777399e-01 3.33431736e-02 -5.68635650e-02 5.60968935e-01
-8.06055665e-01 6.81772828e-01 6.97626352e-01 3.24008763e-02
-1.31334633e-01 2.53746182e-01 -1.10694540e+00 8.47974539e-01
4.75370735e-01 -4.96258050e-01 -5.17267168e-01 -2.98049450e-01
-8.81254524e-02 2.74009764e-01 1.30952597e-01 -8.28494549e-01
3.62229794e-01 -4.90751684e-01 3.47835392e-01 -5.72629809e-01
1.09044701e-01 -9.81871843e-01 4.15527701e-01 4.18051928e-01
-1.16131306e-01 -2.33812287e-01 -4.73454446e-02 8.80505562e-01
-4.78070706e-01 2.29779258e-02 7.66717613e-01 3.84610415e-01
-3.16605955e-01 3.26746464e-01 -1.00405261e-01 -1.05194397e-01
1.08442628e+00 3.52197528e-01 -2.89486349e-01 -6.28971815e-01
-5.04208624e-01 2.91521132e-01 1.64719701e-01 -1.10960893e-01
6.10138297e-01 -1.37911844e+00 -9.65555310e-01 4.28598911e-01
-2.88471639e-01 -1.86291471e-01 7.00637698e-01 1.21461558e+00
-1.72104165e-01 3.50279123e-01 1.87504888e-01 -5.11230648e-01
-1.18479764e+00 6.67060137e-01 -1.53954878e-01 -2.35239804e-01
-2.13575497e-01 5.26622176e-01 5.25841534e-01 -8.56941752e-03
-5.13669997e-02 4.38497178e-02 1.64527372e-01 2.55528651e-02
7.66471982e-01 7.16966212e-01 -1.59392774e-01 -5.49982667e-01
-3.16405773e-01 5.70752561e-01 6.65386440e-03 1.30399242e-01
1.62915111e+00 -3.23946357e-01 -5.22436917e-01 5.18290520e-01
1.35199463e+00 6.66841388e-01 -1.11654568e+00 -6.42799973e-01
-2.48499736e-01 -7.20636606e-01 1.70968205e-01 -3.52159649e-01
-1.06065929e+00 5.77116251e-01 3.41297299e-01 4.03881878e-01
1.13394964e+00 8.03264081e-02 5.89558125e-01 3.35538715e-01
5.44350386e-01 -7.95329511e-01 -3.13322246e-02 2.94142246e-01
9.05321538e-01 -1.18562126e+00 4.94644076e-01 -7.01182008e-01
-5.51756561e-01 6.84677005e-01 2.79333830e-01 -6.04936063e-01
9.31360960e-01 3.44503880e-01 -2.73292035e-01 -4.99376021e-02
-6.57232583e-01 1.82012215e-01 3.56786311e-01 3.89467806e-01
2.81144530e-01 2.63310075e-01 -6.39106750e-01 1.17327082e+00
6.39739931e-02 -7.65557885e-02 2.17929229e-01 7.12763250e-01
-5.45615852e-01 -1.00149775e+00 -8.50760698e-01 6.53714180e-01
-2.64172345e-01 -6.13165312e-02 -2.93563902e-01 4.57911938e-01
-9.05367360e-02 8.68187487e-01 -1.70164555e-02 -2.63028949e-01
-1.04734212e-01 -2.35767663e-01 1.12266131e-01 -6.11280203e-01
5.10439239e-02 8.64435315e-01 -2.56878376e-01 -4.82173115e-01
-4.25191879e-01 -1.16785872e+00 -9.24493730e-01 -3.37127179e-01
-4.53363061e-01 3.78190398e-01 4.28630650e-01 7.16982186e-01
7.01142788e-01 3.07128966e-01 1.18471169e+00 -5.99703550e-01
-8.17786455e-01 -1.03492248e+00 -1.18453050e+00 1.51214629e-01
1.67100564e-01 -7.06055939e-01 -4.28588599e-01 2.00990438e-01] | [7.089601039886475, 4.513828754425049] |
426ed98f-5a65-479f-a698-de3376ce18c7 | non-deterministic-behavior-of-ranking-based | 1806.07171 | null | http://arxiv.org/abs/1806.07171v2 | http://arxiv.org/pdf/1806.07171v2.pdf | Non-deterministic Behavior of Ranking-based Metrics when Evaluating Embeddings | Embedding data into vector spaces is a very popular strategy of pattern
recognition methods. When distances between embeddings are quantized,
performance metrics become ambiguous. In this paper, we present an analysis of
the ambiguity quantized distances introduce and provide bounds on the effect.
We demonstrate that it can have a measurable effect in empirical data in
state-of-the-art systems. We also approach the phenomenon from a computer
security perspective and demonstrate how someone being evaluated by a third
party can exploit this ambiguity and greatly outperform a random predictor
without even access to the input data. We also suggest a simple solution making
the performance metrics, which rely on ranking, totally deterministic and
impervious to such exploits. | ['Vincent Christlein', 'Anguelos Nicolaou', 'Sounak Dey', 'Andreas Maier', 'Dimosthenis Karatzas'] | 2018-06-19 | null | null | null | null | ['computer-security'] | ['miscellaneous'] | [ 2.08875701e-01 6.63690343e-02 -1.45238414e-01 -3.54070485e-01
-8.24348927e-01 -9.19994235e-01 7.76496470e-01 3.94103765e-01
-8.28015387e-01 5.44228196e-01 -9.01806802e-02 -6.12200022e-01
-1.05785877e-01 -7.74170041e-01 -5.95656574e-01 -7.16400743e-01
-4.10116732e-01 2.99485773e-01 3.84772629e-01 -3.56363684e-01
6.16659522e-01 6.44056618e-01 -1.26099992e+00 3.03106140e-02
2.53717452e-01 8.83867562e-01 -5.90206504e-01 9.81197774e-01
2.18726724e-01 1.61995247e-01 -8.31518233e-01 -7.31290162e-01
7.37922847e-01 2.18814284e-01 -6.42681301e-01 -3.00789148e-01
3.89510781e-01 -4.47846681e-01 -3.85769606e-01 1.23663628e+00
5.04059434e-01 -2.25971103e-01 7.78712213e-01 -1.29591036e+00
-8.97573709e-01 5.52539408e-01 -1.15436025e-01 2.88026989e-01
4.08597648e-01 -2.06262097e-02 1.39322448e+00 -1.07585919e+00
5.44722617e-01 9.26171660e-01 7.05720127e-01 4.83425409e-01
-1.42178345e+00 -2.54569381e-01 -2.25803211e-01 2.92387158e-01
-1.40741622e+00 -3.21140379e-01 7.36772418e-01 -5.14313817e-01
8.90370786e-01 7.32463777e-01 3.65352035e-02 1.02311110e+00
2.51429230e-01 5.06046891e-01 9.93215382e-01 -6.74416602e-01
4.36001629e-01 5.84442854e-01 6.38129890e-01 6.28405035e-01
5.27357757e-01 5.79734087e-01 -2.95754403e-01 -6.47714436e-01
2.34154582e-01 1.47663876e-01 -2.45860517e-01 -7.77351677e-01
-1.06033468e+00 1.02612185e+00 3.39030564e-01 3.34532320e-01
-9.85676497e-02 1.81742355e-01 3.74525756e-01 6.55711472e-01
1.26061160e-02 8.58885050e-01 -5.05388796e-01 -2.23540887e-01
-5.45613348e-01 -3.51054668e-02 9.41765845e-01 6.17768228e-01
5.34684598e-01 -2.82048643e-01 1.33682087e-01 4.08645630e-01
6.14789538e-02 3.52787912e-01 4.99255836e-01 -5.66050231e-01
2.97176212e-01 3.87692004e-01 1.86605200e-01 -1.29943144e+00
-1.17014386e-01 8.06390122e-02 -5.94145358e-01 4.21730280e-01
3.39863747e-01 6.78019747e-02 -4.74114656e-01 1.48589754e+00
8.67176205e-02 -3.90160792e-02 -3.99799161e-02 6.74303770e-01
1.41791748e-02 2.96880543e-01 -2.84061372e-01 -8.58620405e-02
1.21483397e+00 -5.96691310e-01 -4.93459731e-01 2.21627563e-01
6.48411751e-01 -6.48425043e-01 1.12202096e+00 6.19433045e-01
-6.00808263e-01 -3.84677082e-01 -1.60198593e+00 1.74441710e-01
-8.23783576e-01 -3.20731074e-01 5.90503454e-01 1.31131637e+00
-1.13068902e+00 9.78462636e-01 -8.63361597e-01 -2.49278516e-01
1.33046135e-01 7.13883817e-01 -5.13297260e-01 3.17624807e-01
-1.38236058e+00 1.03808594e+00 2.27664649e-01 -1.19070888e-01
-6.08236372e-01 -2.55032390e-01 -6.44716859e-01 -9.29512009e-02
1.02262251e-01 -3.53953755e-03 9.61056650e-01 -2.92386353e-01
-1.20917988e+00 7.95286179e-01 1.69480160e-01 -7.08178580e-01
5.49242020e-01 -2.58108497e-01 -5.71452200e-01 4.56162216e-03
-2.79424846e-01 2.18310207e-01 1.11409104e+00 -8.68255734e-01
-3.53082567e-01 -3.70919973e-01 4.75325957e-02 -4.65004265e-01
-1.00835669e+00 -6.42993972e-02 7.30916187e-02 -5.71225405e-01
-1.07264638e-01 -1.10728037e+00 -3.96432787e-01 2.55127311e-01
-4.89788860e-01 -1.72573850e-01 7.50964046e-01 -1.61447346e-01
1.52742195e+00 -2.36669374e+00 1.79464325e-01 5.32335877e-01
3.29926372e-01 5.03180027e-01 -8.60911384e-02 6.80766106e-01
-1.08555153e-01 7.22252548e-01 -2.12338358e-01 -1.48502767e-01
3.73141080e-01 2.55919337e-01 -8.03097308e-01 8.09987783e-01
1.98872223e-01 5.76433659e-01 -9.06624138e-01 -2.05023199e-01
1.78025991e-01 5.24489343e-01 -6.64272606e-01 6.30635992e-02
4.52926636e-01 -3.32725078e-01 -4.40757513e-01 5.91755092e-01
6.14960372e-01 -1.51605802e-02 2.43105039e-01 -1.29987285e-01
6.34002537e-02 2.32845306e-01 -1.32172716e+00 1.00899136e+00
-2.45941505e-01 8.13934445e-01 -3.99476200e-01 -9.54796791e-01
9.36544001e-01 1.56715721e-01 9.51225832e-02 -1.26672745e-01
3.19233872e-02 3.21271539e-01 -9.22662579e-03 -1.20460417e-03
6.69456065e-01 5.16881607e-03 -3.78376931e-01 5.25696278e-01
-1.55894622e-01 1.73407212e-01 -2.21069306e-01 -1.64257716e-02
1.58751690e+00 -6.44307375e-01 5.18270671e-01 -2.74485677e-01
6.56773746e-01 -3.40546966e-01 2.15821266e-01 1.02381682e+00
-4.30549175e-01 4.46103126e-01 6.13613665e-01 -6.87342644e-01
-1.16103780e+00 -1.27682710e+00 -4.24239159e-01 9.92040575e-01
1.28122851e-01 -8.08936119e-01 -5.28455377e-01 -9.85138953e-01
4.48268533e-01 4.99493420e-01 -8.49501431e-01 -4.04159069e-01
-4.06961501e-01 -6.80499613e-01 7.98362315e-01 4.92512733e-01
1.09452493e-01 -6.38065815e-01 -6.56115174e-01 -5.20239882e-02
4.73125249e-01 -8.91470015e-01 -4.24010426e-01 4.52286899e-01
-7.66890466e-01 -9.87067997e-01 -2.56884038e-01 -2.93089896e-01
6.89397275e-01 -2.27312874e-02 8.32161784e-01 3.15230340e-01
-4.08056825e-01 3.17477256e-01 -3.07721585e-01 -2.11611956e-01
-4.31280166e-01 3.77877839e-02 6.06644511e-01 -1.18081652e-01
8.17229807e-01 -5.55106103e-01 -4.03236032e-01 5.27647197e-01
-1.08373916e+00 -9.29857433e-01 5.84210634e-01 1.12353003e+00
3.54940444e-01 -6.69951886e-02 1.23534620e-01 -9.62621152e-01
9.10898268e-01 -2.15028808e-01 -6.87809944e-01 1.91297352e-01
-1.02557015e+00 4.98661846e-01 5.70954978e-01 -6.28866136e-01
1.52355164e-01 1.93803757e-02 8.73559341e-02 -4.89496768e-01
1.34731606e-01 1.18459873e-01 -1.63327813e-01 -4.73866701e-01
1.01591325e+00 9.56364498e-02 -1.88623860e-01 -3.25770855e-01
6.11067116e-01 9.78651285e-01 3.70240748e-01 -4.79597420e-01
1.17985332e+00 3.84247690e-01 3.17390040e-02 -9.40639675e-01
-1.96008548e-01 -4.21607912e-01 -7.18648612e-01 3.03879052e-01
4.63768601e-01 -3.34030390e-01 -6.56873822e-01 -3.47635634e-02
-1.20092225e+00 3.40645343e-01 -4.43637371e-01 2.38099441e-01
-3.20875436e-01 5.55775523e-01 -4.59306270e-01 -6.82562053e-01
-9.19330120e-02 -1.23568153e+00 7.31726885e-01 -2.91228801e-01
-4.60972548e-01 -1.08425760e+00 3.09866846e-01 -1.54613525e-01
3.72564256e-01 2.51242787e-01 8.05148244e-01 -1.22666585e+00
-4.15352136e-01 -6.61984980e-01 -5.03111891e-02 7.11551309e-01
2.68897742e-01 9.30635259e-02 -9.27511036e-01 -4.00454134e-01
7.71531340e-05 -3.44796032e-02 6.49871349e-01 -3.89528960e-01
1.12064576e+00 -5.90588689e-01 -2.81583220e-01 5.17781794e-01
1.48706853e+00 -3.24617177e-02 6.08088613e-01 5.71841776e-01
4.75705355e-01 4.18337345e-01 4.32745874e-01 5.09696364e-01
-1.56185895e-01 8.24932873e-01 3.09200495e-01 3.44583988e-01
4.12741274e-01 -2.77433157e-01 4.29508537e-01 9.18780446e-01
2.25297764e-01 -1.56393256e-02 -8.94244909e-01 4.90989566e-01
-1.41385901e+00 -7.79938281e-01 2.73345113e-01 2.43661284e+00
7.31317818e-01 5.64777911e-01 -3.22807543e-02 5.77102125e-01
6.01019502e-01 2.34293386e-01 -4.00923550e-01 -1.02662623e+00
-1.20041985e-03 3.56123090e-01 1.08296144e+00 7.79549479e-01
-1.17352664e+00 7.22623229e-01 7.96727180e+00 6.26438677e-01
-9.55690503e-01 7.70076215e-02 3.63732934e-01 1.28081635e-01
-4.43845838e-01 -3.79884243e-02 -8.16373467e-01 4.45106447e-01
9.95631814e-01 -3.93059015e-01 3.34572464e-01 1.09964514e+00
-5.97791672e-01 3.89530957e-01 -1.67350829e+00 1.00958788e+00
-5.35397464e-03 -1.08280778e+00 1.45012587e-01 5.79721153e-01
2.59517282e-01 -7.13624880e-02 4.76471215e-01 2.86251068e-01
1.95804879e-01 -1.16755843e+00 2.62321800e-01 2.47040585e-01
7.87606001e-01 -7.34046221e-01 7.47074366e-01 2.04840481e-01
-8.09994221e-01 -2.32635021e-01 -7.31854498e-01 -2.05444500e-01
-2.49098375e-01 5.32893360e-01 -9.74789560e-01 1.79905653e-01
1.65793687e-01 2.65614331e-01 -8.93612802e-01 7.24097311e-01
-2.00694054e-01 6.08211517e-01 -5.28037071e-01 -3.45090866e-01
1.20554179e-01 8.20862278e-02 7.01457143e-01 1.16432965e+00
2.81901389e-01 -1.69544727e-01 -1.41997978e-01 4.33330268e-01
-2.68077962e-02 -4.32801321e-02 -1.02639842e+00 -2.11810470e-01
6.88410103e-01 9.54224765e-01 -4.32384193e-01 -1.16599768e-01
-2.16548383e-01 1.12192142e+00 3.73160303e-01 3.48657258e-02
-7.51697719e-01 -8.09958875e-01 9.53947842e-01 9.74978507e-03
4.92864877e-01 -2.99051881e-01 -2.21422285e-01 -1.07386112e+00
3.72525305e-01 -8.67652595e-01 2.80897528e-01 -6.47134557e-02
-1.43556976e+00 6.94980443e-01 -2.54305393e-01 -1.34082162e+00
-4.72509205e-01 -1.10229719e+00 -3.70593309e-01 5.56819081e-01
-1.18280983e+00 -4.53467131e-01 3.42020601e-01 3.61928374e-01
-7.46739283e-02 -4.88437444e-01 1.25085604e+00 2.91236430e-01
-3.07404339e-01 1.18869913e+00 4.07310277e-01 3.21327090e-01
6.19084060e-01 -1.45727885e+00 6.05626941e-01 7.96950758e-01
7.85714984e-01 9.32004035e-01 9.25261199e-01 -1.21827729e-01
-1.62312627e+00 -6.69722736e-01 8.89605820e-01 -1.06121647e+00
9.93868351e-01 -4.80727255e-01 -8.22002828e-01 6.26937568e-01
-2.55886465e-01 3.35910201e-01 1.01540208e+00 3.23544323e-01
-8.57343197e-01 -1.06983237e-01 -1.25860429e+00 5.66422760e-01
8.48520219e-01 -1.08093965e+00 -9.11122322e-01 1.59615546e-01
7.58404315e-01 -1.30254626e-02 -8.59825075e-01 2.54009664e-01
7.16752052e-01 -1.03869331e+00 1.22903013e+00 -9.03216779e-01
7.64488503e-02 -2.66674519e-01 -6.40971541e-01 -1.08175635e+00
-1.86976090e-01 -8.31551015e-01 -3.13826114e-01 8.45783472e-01
5.25507271e-01 -1.02662551e+00 9.33150232e-01 7.70455241e-01
4.75194186e-01 -6.89067185e-01 -1.25583005e+00 -1.10633469e+00
2.48452425e-01 -4.64104712e-01 6.05623603e-01 8.52787077e-01
4.32099193e-01 -3.58985141e-02 -3.71733427e-01 5.26942432e-01
6.57737911e-01 -2.89467752e-01 6.50041938e-01 -1.13050783e+00
-3.64806384e-01 -4.82410342e-01 -1.39746201e+00 -1.00487149e+00
1.36273932e-02 -8.29437613e-01 -3.55787218e-01 -4.91095871e-01
-2.39609331e-01 -4.00532514e-01 -1.01789117e+00 2.83590138e-01
5.38345650e-02 2.89004982e-01 1.57232732e-01 2.03779995e-01
-6.57300055e-01 2.16129690e-01 4.03280616e-01 -2.49498770e-01
1.22274034e-01 -1.41462147e-01 -8.07206869e-01 6.30350113e-01
7.49128401e-01 -8.16046953e-01 -3.00391704e-01 -2.16536313e-01
2.11521387e-01 -1.93807170e-01 3.82462680e-01 -9.09970462e-01
4.18113083e-01 1.36872604e-01 2.47906953e-01 -2.42674083e-01
2.97705352e-01 -1.06538069e+00 -1.50182411e-01 7.02114403e-01
-5.58515668e-01 3.54294509e-01 -3.58900763e-02 7.28613436e-01
-1.17455475e-01 -3.17802310e-01 4.60406035e-01 4.65531766e-01
-4.53849286e-01 1.40144065e-01 -8.66476297e-02 -1.41272992e-01
1.19057012e+00 -1.52585313e-01 -2.95612067e-01 -1.56671673e-01
-7.65415788e-01 -2.80771926e-02 6.14386380e-01 3.78523320e-01
8.20274115e-01 -1.58616173e+00 -4.51539338e-01 4.46796507e-01
2.90418416e-01 -9.52229798e-01 -4.50962424e-01 4.01261896e-01
-5.80277324e-01 5.44519067e-01 -1.60107061e-01 -5.41995585e-01
-1.45701265e+00 9.67227161e-01 1.13468647e-01 -2.81105936e-01
-3.11916232e-01 7.69282699e-01 -3.19762766e-01 -5.24078012e-01
4.52779979e-01 -4.31665421e-01 1.15156651e-01 2.08083969e-02
7.44284153e-01 2.03920916e-01 6.73790872e-02 -2.92612076e-01
-6.92046642e-01 4.48785931e-01 -2.13182256e-01 -2.86009789e-01
9.95045781e-01 8.82046893e-02 6.12392277e-02 5.32467961e-01
1.62988615e+00 2.70768940e-01 -7.07056880e-01 -1.41159028e-01
3.23299676e-01 -7.60422528e-01 -2.15577140e-01 -3.38057518e-01
-5.93982041e-01 1.14076781e+00 8.84277582e-01 8.88569832e-01
6.59308553e-01 -3.05467337e-01 6.08234942e-01 1.07731438e+00
6.16279006e-01 -8.67475927e-01 -3.75969484e-02 1.95683256e-01
6.78823411e-01 -1.25083840e+00 1.63569748e-01 -2.20103115e-01
-3.44391495e-01 1.08681226e+00 8.67008120e-02 -4.77364659e-01
1.16544986e+00 3.03141326e-01 3.18227429e-03 1.75433427e-01
-7.51567781e-01 2.76562661e-01 2.22455978e-01 8.00634325e-01
8.59805942e-02 2.24415049e-01 -4.03761744e-01 5.01201928e-01
-4.79987025e-01 -3.78716499e-01 6.11202657e-01 9.77398098e-01
-4.28423345e-01 -1.51123679e+00 -3.36663485e-01 1.78292751e-01
-6.02992117e-01 -1.96272969e-01 -5.08763850e-01 8.09321463e-01
-1.58134431e-01 6.84562981e-01 -9.06494856e-02 -9.19697881e-01
4.38511997e-01 4.53893803e-02 9.36162621e-02 -4.01574999e-01
-3.50729764e-01 -6.02193177e-01 -3.88371646e-02 -6.06564641e-01
1.69759467e-01 -5.24912000e-01 -7.85490513e-01 -4.98888880e-01
-3.59144747e-01 3.77818048e-01 6.94725156e-01 4.33177054e-01
4.54223305e-01 1.13740385e-01 1.06856370e+00 -4.97656822e-01
-1.43448949e+00 -4.88770545e-01 -6.26399219e-01 2.84060061e-01
6.36563540e-01 -5.19486487e-01 -9.67891455e-01 -3.24406177e-01] | [8.227996826171875, 4.149080276489258] |
4cef6651-abac-4cc9-9f6e-478ab188b339 | mcmc-guided-cnn-training-and-segmentation-for | 2003.03938 | null | https://arxiv.org/abs/2003.03938v1 | https://arxiv.org/pdf/2003.03938v1.pdf | MCMC Guided CNN Training and Segmentation for Pancreas Extraction | Efficient organ segmentation is the precondition of various quantitative analysis. Segmenting the pancreas from abdominal CT images is a challenging task because of its high anatomical variability in shape, size and location. What's more, the pancreas only occupies a small portion in abdomen, and the organ border is very fuzzy. All these factors make the segmentation methods of other organs less suitable for the pancreas segmentation. In this report, we propose a Markov Chain Monte Carlo (MCMC) sampling guided convolutional neural network (CNN) approach, in order to handle such difficulties in morphological and photometric variabilities. Specifically, the proposed method mainly contains three steps: First, registration is carried out to mitigate the body weight and location variability. Then, an MCMC sampling is employed to guide the sampling of 3D patches, which are fed to the CNN for training. At the same time, the pancreas distribution is also learned for the subsequent segmentation. Third, sampled from the learned distribution, an MCMC process guides the segmentation process. Lastly, the patches based segmentation is fused using a Bayesian voting scheme. This method is evaluated on the NIH pancreatic datasets which contains 82 abdominal contrast-enhanced CT volumes. Finally, we achieved a competitive result of 78.13% Dice Similarity Coefficient value and 82.65% Recall value in testing data. | ['Chudong Cai', 'Yi Gao', 'Xiaxia Yu', 'Jinchan He'] | 2020-03-09 | null | null | null | null | ['pancreas-segmentation'] | ['medical'] | [-1.22145779e-01 -1.54531002e-01 -1.60302848e-01 -4.35908943e-01
-5.62079668e-01 -3.76962721e-01 1.74327597e-01 4.09067005e-01
-4.07816678e-01 4.86561030e-01 -6.44192174e-02 -4.23448347e-03
4.34417129e-02 -7.56232381e-01 -5.08675992e-01 -1.12899888e+00
-7.69025609e-02 6.87505662e-01 2.80855715e-01 3.85729790e-01
7.23300949e-02 5.53339481e-01 -7.01801777e-01 -1.46303728e-01
1.17550433e+00 1.09810996e+00 2.41351590e-01 2.81843811e-01
-3.10413063e-01 4.78356868e-01 -1.62853613e-01 -2.44621828e-01
1.89334601e-01 -6.37420535e-01 -4.67516750e-01 3.39575589e-01
-3.53022702e-02 -4.29969966e-01 -1.00071706e-01 1.30397189e+00
3.97813529e-01 1.36371449e-01 1.03588080e+00 -6.98531687e-01
-2.09193990e-01 6.75198019e-01 -6.51398599e-01 5.57498485e-02
-1.58971250e-01 2.92486727e-01 3.54324311e-01 -6.30439758e-01
2.49087572e-01 9.59729254e-01 6.82133913e-01 2.85297602e-01
-1.05113339e+00 -6.87112272e-01 -1.47702709e-01 -3.79009634e-01
-1.35654128e+00 2.02291608e-01 7.77987421e-01 -6.52268589e-01
3.91458087e-02 -1.34915188e-02 1.15772247e+00 5.60808063e-01
5.72773457e-01 6.70449674e-01 9.56385791e-01 -9.94658843e-02
2.86810607e-01 -1.24042057e-01 -7.88594559e-02 9.14667726e-01
5.59965491e-01 -7.15658218e-02 3.11063915e-01 -1.88091993e-01
1.01346314e+00 4.78254229e-01 -2.24433139e-01 -5.95004201e-01
-1.31495118e+00 9.25967455e-01 7.64002085e-01 1.41626224e-01
-7.42144406e-01 -1.13730364e-01 3.78545344e-01 -3.42461556e-01
2.00661898e-01 5.35465628e-02 -1.50586128e-01 4.41664070e-01
-1.08526707e+00 -5.95583729e-02 9.91074741e-01 8.37173522e-01
5.00846207e-01 -1.50168255e-01 6.52089424e-04 6.66599274e-01
8.99721265e-01 5.27563930e-01 5.21571159e-01 -7.17802048e-01
1.74274117e-01 7.68725336e-01 -1.89726219e-01 -8.42670679e-01
-6.14317179e-01 -5.03691316e-01 -1.44142497e+00 4.57992256e-02
6.47782385e-01 -3.14699769e-01 -1.15479994e+00 1.24493790e+00
7.66805589e-01 5.84297217e-02 -1.63002446e-01 1.22604847e+00
9.24714446e-01 5.77297688e-01 6.01725519e-01 -3.86987865e-01
1.75805247e+00 -7.31590390e-01 -4.43251938e-01 1.62882939e-01
3.60836655e-01 -8.45042825e-01 4.22813624e-01 3.40977818e-01
-8.97970080e-01 -2.86699772e-01 -9.36907470e-01 4.27321285e-01
1.76682055e-01 3.22760761e-01 6.84966028e-01 4.97955084e-01
-5.09114802e-01 3.67157251e-01 -1.24278653e+00 -3.16806257e-01
7.44397998e-01 2.48101071e-01 -4.77344915e-02 -9.29972008e-02
-8.46376061e-01 7.24657416e-01 7.03364372e-01 3.26778710e-01
-7.83865035e-01 -5.82002461e-01 -8.10122788e-01 8.44773948e-02
1.21946737e-01 -7.85828590e-01 1.09860313e+00 -8.80069435e-01
-1.65091240e+00 7.36774921e-01 2.98803806e-01 -3.12970221e-01
7.54088342e-01 4.19301331e-01 -8.47590417e-02 2.91502565e-01
-3.53867449e-02 6.00642323e-01 5.78396797e-01 -1.18385267e+00
-4.84162658e-01 -4.41217244e-01 -4.02669251e-01 3.39875519e-01
3.31150740e-01 -2.12899610e-01 -5.86788237e-01 -6.59097314e-01
7.55872786e-01 -1.04270864e+00 -6.07069850e-01 2.04364970e-01
-2.91674584e-01 5.38539235e-03 1.75778642e-01 -9.03638363e-01
8.29598606e-01 -2.14376187e+00 -1.49975149e-02 5.41881740e-01
1.59702376e-01 -1.84682570e-02 4.48321402e-01 -1.85327604e-01
2.42671549e-01 -1.52636960e-01 -6.30190790e-01 -7.12492615e-02
-1.63806126e-01 2.19034776e-02 2.88440794e-01 7.94474959e-01
-8.94571766e-02 7.33038962e-01 -7.42584705e-01 -1.21193111e+00
3.63003343e-01 3.51376921e-01 -5.62364578e-01 1.83736965e-01
-1.52411923e-01 9.05036330e-01 -7.40835130e-01 8.82653475e-01
1.08033752e+00 -2.99556851e-01 3.37964416e-01 -3.87073308e-01
-1.39293060e-01 -3.34025174e-01 -1.20399070e+00 1.82473898e+00
-2.21631721e-01 -1.92206115e-01 1.61300480e-01 -8.74404848e-01
1.03179371e+00 3.44817191e-01 7.95850515e-01 -4.72905725e-01
5.33340752e-01 5.01742780e-01 2.62083232e-01 -5.41707039e-01
-8.84109885e-02 -4.25763935e-01 4.17764746e-02 2.79104412e-01
7.47986138e-03 -4.25780386e-01 1.37573227e-01 -1.24418996e-01
5.49369633e-01 1.38168886e-01 4.80579138e-01 -5.31340957e-01
5.54653823e-01 3.75702232e-01 8.06844056e-01 3.07016999e-01
-4.25784856e-01 9.26254809e-01 3.33995819e-01 -6.72559738e-01
-8.52899671e-01 -1.16547167e+00 -4.63477343e-01 2.31935173e-01
5.59528708e-01 3.13404471e-01 -7.44895697e-01 -7.96067953e-01
-4.17881683e-02 3.72685492e-01 -3.65794510e-01 -3.10439039e-02
-6.80436552e-01 -1.06814563e+00 2.15368688e-01 4.16229874e-01
7.38497198e-01 -1.16556382e+00 -6.76617205e-01 3.96281391e-01
-1.86757982e-01 -5.12318969e-01 -4.94873643e-01 9.00137499e-02
-1.40438747e+00 -1.19806314e+00 -1.12081289e+00 -1.04685605e+00
9.66650963e-01 2.12729480e-02 9.39282238e-01 1.45241767e-01
-2.78632969e-01 -2.05402419e-01 -1.70560956e-01 -1.64150447e-01
-3.53403717e-01 7.60252029e-02 -4.20473307e-01 -1.13069460e-01
1.61527500e-01 -4.24589127e-01 -1.10975385e+00 3.68820697e-01
-8.81163001e-01 2.62111425e-02 9.65571582e-01 8.47889304e-01
9.67571437e-01 -2.23065708e-02 1.41492561e-01 -6.87512100e-01
1.24945924e-01 -6.40599728e-01 -8.31678629e-01 2.90272087e-01
-2.81313866e-01 -2.94874579e-01 5.99322021e-01 -6.67341173e-01
-1.13759398e+00 5.13750792e-01 -1.95858702e-01 -1.13967419e-01
-3.61169159e-01 4.66331959e-01 -2.68365704e-02 -5.88189140e-02
3.32858384e-01 3.11339557e-01 3.43776256e-01 -3.52748781e-01
-6.68954402e-02 3.87128741e-01 1.68733418e-01 -4.04655069e-01
3.50353777e-01 3.99803519e-01 1.04384467e-01 -4.62390184e-01
-3.71739537e-01 -3.16333920e-01 -6.53917909e-01 -2.02438816e-01
1.22105372e+00 -8.10804248e-01 -6.62586093e-01 5.41180372e-01
-9.24758255e-01 -2.01416999e-01 -9.27897245e-02 1.32154191e+00
-3.59532952e-01 4.22797382e-01 -9.26957965e-01 -4.61376786e-01
-6.65811598e-01 -1.56318450e+00 6.42633379e-01 7.86851466e-01
1.35642976e-01 -9.56482887e-01 1.64610054e-02 9.46507901e-02
4.26351309e-01 4.38294381e-01 9.15177763e-01 -7.29001462e-01
-5.95532537e-01 -2.84837931e-01 -3.35782766e-01 1.72494709e-01
2.45964341e-02 1.15685612e-02 -3.07516515e-01 -1.75658077e-01
1.67037874e-01 5.84270172e-02 6.51370823e-01 1.04928446e+00
1.06136477e+00 -3.06592509e-02 -3.24644893e-01 9.15671825e-01
1.52799642e+00 6.51372194e-01 2.47886330e-01 1.02469027e-01
6.21063530e-01 3.38801086e-01 7.01240301e-01 6.22209430e-01
3.55277121e-01 1.41770095e-01 5.97134471e-01 -2.64936477e-01
-1.04516573e-01 -9.10057575e-02 -2.57389337e-01 1.00292397e+00
1.04751058e-01 3.57958786e-02 -9.27372932e-01 3.89253855e-01
-1.53151238e+00 -4.39519525e-01 -3.00875783e-01 2.02176690e+00
8.39178205e-01 -1.16487563e-01 -3.21103968e-02 -4.57958430e-01
9.33893383e-01 -1.44102842e-01 -5.96429706e-01 7.59096742e-02
3.01993310e-01 -4.17014882e-02 6.06377244e-01 2.07018152e-01
-1.19296372e+00 4.24722105e-01 4.93131113e+00 6.70150340e-01
-1.09263730e+00 8.87577459e-02 8.22664320e-01 3.19599479e-01
-6.71608895e-02 1.11991078e-01 -5.43356061e-01 7.76798785e-01
1.85422406e-01 2.71391243e-01 2.47894689e-01 6.55660570e-01
5.28801940e-02 -4.92908418e-01 -6.90257490e-01 8.51512194e-01
-1.24844268e-01 -9.25589502e-01 -5.49297407e-02 4.61052805e-02
7.07607090e-01 -2.28531979e-04 -2.80101061e-01 1.24768749e-01
1.85267866e-01 -7.86485612e-01 5.37000656e-01 7.87487209e-01
5.73721468e-01 -9.13061976e-01 1.05936944e+00 4.48449284e-01
-1.06324327e+00 2.70758092e-01 -5.11693239e-01 6.18524790e-01
2.28501961e-01 9.54776466e-01 -7.88337111e-01 5.11879146e-01
5.71093678e-01 2.12380797e-01 -2.20512182e-01 1.65572679e+00
-1.85522586e-01 5.29100299e-01 -5.29668331e-01 -1.49314389e-01
1.39418870e-01 -7.30312884e-01 3.81268799e-01 1.13891995e+00
5.76719463e-01 2.47096092e-01 4.62460071e-01 9.24219251e-01
-6.58372277e-03 5.13606846e-01 -1.02915883e-01 4.38209206e-01
2.81113952e-01 1.54537392e+00 -1.36273789e+00 -4.71057892e-01
-2.47164547e-01 6.88244998e-01 -8.53238031e-02 1.27778247e-01
-1.03320277e+00 -1.23735756e-01 -2.03909308e-01 -1.81255907e-01
1.15880392e-01 -2.55561870e-04 -2.75033891e-01 -1.29650593e+00
-2.08620846e-01 -6.29771054e-01 6.29146278e-01 -3.86710465e-01
-1.36114514e+00 3.87613147e-01 1.54505121e-02 -1.27967203e+00
8.29605460e-02 -2.79758900e-01 -6.97724938e-01 9.59736824e-01
-1.32621586e+00 -9.95344043e-01 -6.37578011e-01 3.18164945e-01
4.63962555e-01 1.85065106e-01 6.62031174e-01 3.65927160e-01
-5.61777830e-01 2.00521603e-01 2.23366112e-01 4.47446615e-01
5.45590401e-01 -1.25452614e+00 -2.99593329e-01 6.47079229e-01
-4.61273789e-01 5.74877977e-01 2.70934016e-01 -8.28714371e-01
-1.23216987e+00 -1.07387495e+00 4.28527713e-01 2.33118951e-01
2.53081977e-01 9.52958688e-02 -7.81827033e-01 5.33463001e-01
-2.21358575e-02 3.82510841e-01 5.60029984e-01 -4.56789166e-01
3.00396532e-01 -1.01005383e-01 -1.57764065e+00 4.82209712e-01
2.44093284e-01 3.14516485e-01 -5.58758020e-01 2.09385365e-01
3.18717003e-01 -8.42843056e-01 -1.28056014e+00 4.80071455e-01
7.22081959e-01 -7.11243689e-01 8.72543097e-01 -7.12188287e-03
3.47711980e-01 -4.84415978e-01 9.28126425e-02 -1.22930241e+00
-2.40414828e-01 -1.37142479e-01 2.23114222e-01 1.13670743e+00
2.03101933e-01 -7.28239775e-01 8.11240733e-01 5.85580528e-01
-2.79254287e-01 -7.34412193e-01 -8.28475237e-01 -2.40652815e-01
2.60959893e-01 7.55497143e-02 5.37483215e-01 9.33221102e-01
-3.19004089e-01 -2.77566552e-01 2.63345420e-01 2.09510744e-01
1.05157006e+00 4.90180224e-01 4.57956672e-01 -1.20921481e+00
-7.75203481e-02 -3.98893625e-01 -4.54413593e-02 -8.86387825e-01
-3.72287959e-01 -1.02867186e+00 2.46548831e-01 -1.64100170e+00
4.93896544e-01 -6.54661775e-01 -2.05856547e-01 2.43721455e-02
-2.08069548e-01 1.69483930e-01 8.85290466e-03 4.59985673e-01
-4.86665159e-01 3.84233564e-01 1.77427042e+00 -5.74940927e-02
-2.09423229e-01 3.87173414e-01 -2.65761763e-01 8.72965515e-01
1.00321531e+00 -6.72194660e-01 -4.84803021e-02 -8.11005607e-02
-2.79792994e-01 4.92004275e-01 3.98983628e-01 -1.00848591e+00
3.60188693e-01 -1.17799155e-01 8.72695565e-01 -9.13298011e-01
-1.84150368e-01 -1.08201766e+00 3.35832298e-01 1.10259068e+00
-2.12731715e-02 -2.39581779e-01 -3.41483653e-02 3.35265994e-01
-1.94382146e-01 -5.18841386e-01 1.26228952e+00 -5.17806351e-01
-1.43501356e-01 6.75306499e-01 -3.78207266e-01 5.39052784e-02
1.08597183e+00 -1.20169982e-01 3.08574080e-01 1.84807390e-01
-7.81174123e-01 3.08051169e-01 3.11711460e-01 -3.85824502e-01
5.14950216e-01 -1.39579475e+00 -7.14635134e-01 2.59333014e-01
-1.05364569e-01 6.97659671e-01 5.86674392e-01 1.57037401e+00
-1.22716749e+00 2.46763080e-02 -1.52770475e-01 -9.51990426e-01
-8.20495725e-01 3.66662562e-01 5.29403389e-01 -3.49940985e-01
-7.38862395e-01 5.38574100e-01 1.64594069e-01 -5.86437702e-01
1.01420067e-01 -7.12312222e-01 -2.02333152e-01 1.44713596e-01
4.84759547e-02 2.43466139e-01 -1.78532019e-01 -5.22770584e-01
-3.75277698e-01 6.86345994e-01 1.11378811e-01 5.93266524e-02
1.21063495e+00 -7.36380070e-02 -1.93308756e-01 -3.60613912e-02
1.01665914e+00 -1.13491006e-01 -1.38686740e+00 -2.44891346e-01
-2.40412325e-01 -2.44973928e-01 -3.01654804e-02 -8.53403628e-01
-1.40378892e+00 8.73910844e-01 7.21011341e-01 1.33933388e-02
9.93477404e-01 -1.48081690e-01 8.81727874e-01 -1.46025121e-01
1.89669564e-01 -7.23138213e-01 -4.11164224e-01 2.79695213e-01
5.02799094e-01 -1.35913324e+00 2.07426548e-01 -3.23094964e-01
-7.04832435e-01 1.32516623e+00 5.09640574e-01 -4.87349480e-01
7.37089515e-01 1.60417661e-01 1.28656626e-01 -1.94515243e-01
4.90835048e-02 8.25205445e-02 3.26673955e-01 2.82241106e-01
3.93981189e-01 2.45005399e-01 -6.36685550e-01 7.30117857e-01
1.86903119e-01 -8.15074239e-03 6.05948232e-02 8.17688048e-01
-5.34552217e-01 -5.75806081e-01 -5.36543012e-01 5.08531451e-01
-6.65187240e-01 -3.32010053e-02 2.22130194e-01 9.50408936e-01
3.06977034e-01 4.92199063e-01 -3.08450498e-03 2.36665443e-01
6.00820817e-02 -1.17407486e-01 4.33074653e-01 -4.63368446e-01
-7.68274307e-01 5.96029699e-01 -4.82124001e-01 -3.96747172e-01
-2.82058179e-01 -8.03216875e-01 -1.76197731e+00 5.17822336e-03
-4.01033491e-01 1.91476926e-01 8.41296911e-01 7.44991124e-01
-2.13409364e-01 5.32319486e-01 6.90945923e-01 -8.20864856e-01
-6.53957963e-01 -9.76710677e-01 -6.67105258e-01 4.14913654e-01
-3.68873514e-02 -5.41790485e-01 -4.03438538e-01 -5.70721477e-02] | [14.521695137023926, -2.6666057109832764] |
4c87a80e-ee2c-47b5-9841-f5184360e2ca | ensemble-conformalized-quantile-regression | 2202.08756 | null | https://arxiv.org/abs/2202.08756v2 | https://arxiv.org/pdf/2202.08756v2.pdf | Ensemble Conformalized Quantile Regression for Probabilistic Time Series Forecasting | This paper presents a novel probabilistic forecasting method called ensemble conformalized quantile regression (EnCQR). EnCQR constructs distribution-free and approximately marginally valid prediction intervals (PIs), which are suitable for nonstationary and heteroscedastic time series data. EnCQR can be applied on top of a generic forecasting model, including deep learning architectures. EnCQR exploits a bootstrap ensemble estimator, which enables the use of conformal predictors for time series by removing the requirement of data exchangeability. The ensemble learners are implemented as generic machine learning algorithms performing quantile regression, which allow the length of the PIs to adapt to local variability in the data. In the experiments, we predict time series characterized by a different amount of heteroscedasticity. The results demonstrate that EnCQR outperforms models based only on quantile regression or conformal prediction, and it provides sharper, more informative, and valid PIs. | ['Stian Norman Anfinsen', 'Filippo Maria Bianchi', 'Vilde Jensen'] | 2022-02-17 | null | null | null | null | ['prediction-intervals', 'probabilistic-time-series-forecasting'] | ['miscellaneous', 'time-series'] | [-5.23264766e-01 -2.79084742e-01 -2.38016799e-01 -8.22171450e-01
-8.44660819e-01 -5.41551530e-01 5.54574847e-01 -1.64654836e-01
-3.80701050e-02 9.93751585e-01 1.98094249e-01 -6.00299358e-01
-4.71420288e-01 -1.04578245e+00 -7.91255593e-01 -8.30584168e-01
-4.32657987e-01 5.64590394e-01 -2.35463604e-01 -8.15472230e-02
1.28789634e-01 5.83682716e-01 -1.69959056e+00 2.93448836e-01
1.12628293e+00 1.29135990e+00 -4.21584308e-01 3.92257303e-01
-2.64688432e-02 4.54109401e-01 -7.02732086e-01 -3.17358464e-01
-9.32948757e-03 -2.17532236e-02 3.49266976e-02 -7.66274691e-01
1.26908615e-01 -3.48644823e-01 -3.70226311e-03 3.50517452e-01
3.54889929e-01 2.06213027e-01 1.19405663e+00 -1.52777851e+00
-7.29978144e-01 9.02779102e-01 -3.76615524e-01 1.75187111e-01
-2.48875469e-01 -5.20872593e-01 6.10694349e-01 -8.52212965e-01
-1.54634342e-01 1.11505401e+00 1.14578366e+00 1.23682030e-01
-1.30437255e+00 -1.03782082e+00 9.42223221e-02 2.78518014e-02
-1.32928443e+00 -5.95584437e-02 7.44929373e-01 -6.73246622e-01
7.54687488e-01 1.96253002e-01 3.26681584e-01 1.43470252e+00
1.12042248e+00 3.57660949e-01 1.03026724e+00 -9.28343311e-02
6.21871233e-01 5.14659919e-02 2.23062783e-01 -9.05905515e-02
-3.01809385e-02 7.18613803e-01 -2.80068040e-01 -4.75983441e-01
6.59458518e-01 4.09342051e-01 3.42536788e-03 -1.75546482e-01
-9.17654872e-01 1.07876432e+00 1.80570096e-01 7.93090388e-02
-5.60337484e-01 -4.54786718e-02 5.94429195e-01 4.90965456e-01
1.08342123e+00 9.60366894e-03 -9.49384093e-01 -9.50329676e-02
-1.12817121e+00 1.89760834e-01 6.76838875e-01 8.59774709e-01
4.91714269e-01 4.60438073e-01 -3.90078574e-01 4.73927796e-01
9.74908564e-03 9.46905255e-01 5.10898948e-01 -6.82686746e-01
1.64591566e-01 1.72181919e-01 4.16619778e-01 -8.75174761e-01
-6.15108788e-01 -8.19574654e-01 -1.14649570e+00 1.81967467e-02
1.57844529e-01 -5.28242886e-01 -7.56921291e-01 1.71477640e+00
1.20223239e-01 4.24139529e-01 1.32481039e-01 3.54191780e-01
4.49417025e-01 8.92743289e-01 2.82154888e-01 -3.47422332e-01
9.34849858e-01 -3.55555385e-01 -6.02608323e-01 4.94403899e-01
4.68512237e-01 -3.22594911e-01 8.23931873e-01 5.58231533e-01
-5.72239816e-01 -7.94821084e-01 -7.28303194e-01 3.53815526e-01
-5.45294642e-01 1.98168918e-01 4.38287318e-01 7.60684669e-01
-7.92957842e-01 8.54337275e-01 -9.05903995e-01 4.11540627e-01
-7.33567635e-04 6.53378218e-02 -2.20688254e-01 3.23187888e-01
-1.49324811e+00 6.48646474e-01 5.51942527e-01 1.43218562e-01
-6.56260312e-01 -1.00901222e+00 -7.51529515e-01 4.27796572e-01
-4.34181601e-01 -4.60612923e-01 1.00683212e+00 -1.30996931e+00
-1.58983397e+00 3.71406563e-02 -1.27712741e-01 -5.38310170e-01
5.26124120e-01 -3.33722651e-01 -8.58456969e-01 -4.10410374e-01
-3.15735303e-02 2.80641206e-02 1.00857627e+00 -7.68728971e-01
-5.98934710e-01 -4.17400777e-01 -6.23461664e-01 -3.97082746e-01
-5.88071235e-02 -5.88303320e-02 4.11140054e-01 -8.95046294e-01
-2.07215533e-01 -7.24025249e-01 -6.03324082e-03 -7.86169171e-01
-1.28628209e-01 -6.32595003e-01 6.43028975e-01 -8.76282632e-01
1.33468485e+00 -2.17288113e+00 -8.96467268e-02 5.73119283e-01
-3.26559693e-01 -4.89182264e-01 -6.05758540e-02 7.80844033e-01
-4.09482419e-01 -3.57495323e-02 -1.59254983e-01 -2.65986145e-01
6.44095317e-02 1.53389812e-01 -1.21480227e+00 3.44816774e-01
8.03077295e-02 4.54099238e-01 -3.70529920e-01 -5.82446083e-02
5.39349578e-02 5.27502477e-01 -1.55814439e-01 2.15632766e-01
-2.59831727e-01 6.12898648e-01 -3.51152748e-01 4.10670161e-01
1.00987947e+00 -1.50034949e-01 -3.56611758e-02 6.37051463e-02
-4.06887770e-01 -7.65721351e-02 -8.21375728e-01 8.79996836e-01
-5.80596983e-01 3.22305113e-01 -7.11559415e-01 -1.06009543e+00
1.58694112e+00 4.04906631e-01 4.57651794e-01 -4.15446520e-01
6.87413141e-02 2.77123272e-01 -4.52438951e-01 7.95245618e-02
4.22620207e-01 -8.19055960e-02 -3.00908446e-01 3.38851541e-01
-3.02196387e-02 3.13041002e-01 -1.91621944e-01 -3.35193038e-01
5.44361770e-01 3.98265630e-01 1.15828782e-01 -6.04669690e-01
4.28439796e-01 -2.15819940e-01 7.42167473e-01 7.61497438e-01
3.57760459e-01 3.17347109e-01 5.32048166e-01 -7.64354408e-01
-1.10014379e+00 -1.40383196e+00 -7.90859461e-01 1.43491244e+00
-2.71259516e-01 4.90043089e-02 -3.48598838e-01 -5.41866183e-01
3.08370054e-01 1.25607002e+00 -9.47370887e-01 -4.84638810e-02
-2.81829059e-01 -7.96663642e-01 2.48634294e-01 1.10265446e+00
7.61945918e-02 -9.15179431e-01 -1.41665652e-01 4.62141603e-01
1.50932558e-02 -3.91613066e-01 -2.94093847e-01 3.28713894e-01
-1.13668776e+00 -5.79864740e-01 -6.28428996e-01 -1.68038651e-01
1.38956144e-01 -2.00109050e-01 1.28888357e+00 -6.12899065e-01
5.60998023e-01 3.26395512e-01 -3.32229972e-01 -7.90874720e-01
-2.34693706e-01 2.00568646e-01 1.69689134e-01 1.33147866e-01
3.92091811e-01 -8.79892290e-01 -6.43807352e-01 6.42482221e-01
-7.28400111e-01 -3.23544353e-01 2.39348635e-02 1.04958236e+00
7.85407364e-01 5.29710343e-03 1.43044901e+00 -5.49588263e-01
6.48784459e-01 -1.07258976e+00 -9.81259525e-01 3.44093323e-01
-8.78766596e-01 6.13417774e-02 1.06084490e+00 -5.84862709e-01
-1.18567050e+00 -3.82731229e-01 -4.19818796e-03 -4.49025929e-01
-1.97047114e-01 8.11947286e-01 3.24377060e-01 5.49776673e-01
5.77967823e-01 3.13957304e-01 -1.64846972e-01 -5.18822849e-01
1.57918617e-01 7.64469326e-01 4.27452117e-01 -6.92002535e-01
2.84558117e-01 3.56493443e-01 -6.60493672e-02 -2.92919993e-01
-6.61142170e-01 -1.00924440e-01 -4.72762018e-01 -2.06919357e-01
6.71005487e-01 -1.09149480e+00 -6.68564737e-01 4.49701279e-01
-8.48743975e-01 -3.75336528e-01 -2.08846748e-01 8.07173729e-01
-7.43953943e-01 -1.87378973e-01 -5.49446702e-01 -1.01325095e+00
-6.27218306e-01 -5.39316595e-01 1.10574341e+00 2.62378186e-01
-3.15691791e-02 -1.19596124e+00 4.11871552e-01 -4.02758211e-01
6.87360048e-01 4.58567321e-01 1.16200042e+00 -1.11731791e+00
-1.08591542e-02 -5.35099864e-01 2.44670920e-02 4.92119879e-01
-1.71108082e-01 5.93008757e-01 -8.83854449e-01 -3.91790062e-01
-3.39457601e-01 4.99422364e-02 7.35384524e-01 9.59327400e-01
1.77441645e+00 -9.37538892e-02 -3.14599723e-01 7.09394515e-01
1.01779675e+00 3.66481900e-01 7.18050480e-01 2.37195000e-01
-1.35537805e-02 6.45803154e-01 4.32093322e-01 7.65425503e-01
6.50222898e-01 3.43699813e-01 3.38009715e-01 1.52085245e-01
8.26359093e-01 -3.36283147e-01 4.73616213e-01 7.34943569e-01
-2.80915171e-01 -3.31987560e-01 -9.16621625e-01 3.76088202e-01
-1.91527593e+00 -1.28132129e+00 -2.23481134e-01 2.48476100e+00
5.25404811e-01 -1.27779484e-01 2.00614884e-01 -2.17643365e-01
6.97098374e-01 -1.93116635e-01 -7.43074358e-01 -7.35569537e-01
-3.02253842e-01 3.76593709e-01 4.01682436e-01 1.08702317e-01
-1.26888859e+00 3.78985733e-01 6.80573893e+00 8.15368831e-01
-1.34151518e+00 -6.72786608e-02 8.95951688e-01 1.98210970e-01
-5.16407251e-01 -2.20035136e-01 -6.87155783e-01 8.05660665e-01
1.57281983e+00 -2.68068939e-01 -2.83030812e-02 1.18056893e+00
2.17867717e-01 2.79955626e-01 -1.01986229e+00 6.10849977e-01
-3.39009583e-01 -1.26845586e+00 -2.14821562e-01 -1.44593239e-01
8.72800052e-01 2.52070218e-01 4.94512409e-01 9.35635388e-01
4.51035142e-01 -1.17270947e+00 6.65253460e-01 1.40666246e+00
6.75665617e-01 -1.26044965e+00 1.24799514e+00 5.97802103e-01
-8.63319159e-01 -3.31538647e-01 -5.03242850e-01 7.71959350e-02
-1.45955151e-02 8.57865334e-01 -2.51519769e-01 7.74367154e-01
1.10687399e+00 6.06543601e-01 -2.27950379e-01 8.11643779e-01
2.36294165e-01 9.84245241e-01 -3.04394841e-01 1.25534147e-01
-1.53502405e-01 -3.85533452e-01 1.45475626e-01 9.32328463e-01
1.15586615e+00 -5.09030148e-02 4.34090458e-02 5.45529485e-01
1.87666014e-01 2.75928766e-01 -3.75113338e-01 2.62073159e-01
6.72600448e-01 8.56710553e-01 -1.87950924e-01 -3.47987920e-01
-3.09092820e-01 3.95650387e-01 1.08644716e-01 6.08337939e-01
-1.00389373e+00 -2.48423070e-01 3.15369159e-01 -1.45388663e-01
5.09386182e-01 -6.82772696e-02 -4.11412448e-01 -1.14359546e+00
-2.71466017e-01 -6.86946690e-01 5.67194641e-01 -8.07242274e-01
-1.99458063e+00 8.90336752e-01 2.07019106e-01 -1.36661541e+00
-7.32532084e-01 -4.83205646e-01 -9.66162801e-01 1.10936332e+00
-1.26600957e+00 -1.07030654e+00 -1.58206791e-01 6.60311699e-01
7.33986571e-02 -4.19608712e-01 1.10448241e+00 -1.19441003e-01
-4.28397089e-01 6.64564192e-01 1.08132231e+00 -1.84723035e-01
8.45170736e-01 -1.22649348e+00 2.21101180e-01 2.43782833e-01
-3.21842164e-01 4.80536669e-01 7.00009525e-01 -6.81787133e-01
-7.16181397e-01 -1.31947112e+00 7.40687072e-01 -4.59183782e-01
6.07325613e-01 -7.28122741e-02 -1.25538623e+00 8.51581872e-01
1.06327176e-01 -1.24346599e-01 1.17861462e+00 5.54353654e-01
-6.32800877e-01 -5.62255919e-01 -1.10335684e+00 1.86130285e-01
2.61999726e-01 -2.23917395e-01 -4.70715255e-01 1.33258492e-01
5.60792208e-01 -4.53329422e-02 -1.47442555e+00 7.23629236e-01
1.03807211e+00 -1.02242279e+00 6.52210176e-01 -8.15132082e-01
3.77537757e-01 -1.94104054e-04 -2.81124413e-01 -1.53208077e+00
-5.08568406e-01 -1.71946451e-01 -2.34162793e-01 1.13843560e+00
5.53376198e-01 -1.14655197e+00 2.82703698e-01 5.82143843e-01
-2.19567373e-01 -6.59993112e-01 -1.20585573e+00 -9.41962183e-01
7.74268806e-01 -4.59717333e-01 1.38213539e+00 9.00642633e-01
-2.65808642e-01 -2.74900645e-01 -6.18775487e-01 3.74107480e-01
5.69042385e-01 6.35958076e-01 7.13798106e-01 -1.75869107e+00
-1.44964859e-01 -3.73259962e-01 -3.96464169e-02 -6.05518460e-01
5.19117713e-01 -5.45369208e-01 -1.67432830e-01 -8.13494027e-01
-1.37384832e-01 -4.87269282e-01 -7.72190094e-01 2.17478022e-01
7.60625750e-02 -1.61983833e-01 -3.40931952e-01 2.03348711e-01
-1.26303107e-01 1.05584896e+00 5.02389312e-01 1.56273752e-01
-3.90903234e-01 6.89419091e-01 -3.21315736e-01 6.14531994e-01
1.17044568e+00 -4.08257872e-01 -4.05003846e-01 -3.61889414e-02
1.92578673e-01 5.76992989e-01 3.64887297e-01 -6.69627011e-01
-1.55176997e-01 -2.84577847e-01 8.83718073e-01 -1.12776542e+00
-1.11825451e-01 -7.72513628e-01 6.70948088e-01 -2.88385525e-02
-2.17030659e-01 5.82063913e-01 4.05599117e-01 8.19293797e-01
-2.97744572e-01 2.84652740e-01 3.65729839e-01 6.35450780e-01
-2.21736863e-01 4.33546841e-01 -3.23201597e-01 -5.00532568e-01
1.02056420e+00 -8.44770819e-02 -2.11827174e-01 -6.63130641e-01
-9.17999268e-01 3.62113476e-01 -8.24627876e-02 4.05351162e-01
4.35535192e-01 -1.43612635e+00 -1.09267163e+00 4.46667910e-01
5.25581464e-02 -4.44629639e-01 7.19194829e-01 6.77492082e-01
-3.40570174e-02 3.67178380e-01 -1.71054125e-01 -7.14345515e-01
-5.69506586e-01 6.57856703e-01 5.37207544e-01 -1.17654853e-01
-4.28834796e-01 4.17308599e-01 2.06821904e-01 -8.62336397e-01
4.14986201e-02 -3.92106026e-01 -2.41155922e-01 1.68267161e-01
5.28969526e-01 5.08505940e-01 1.28332391e-01 -2.53254056e-01
-2.03609660e-01 3.04625869e-01 6.91459514e-03 1.73244625e-01
1.52332139e+00 -8.60572457e-02 -1.22406311e-01 8.37089121e-01
7.70756662e-01 -6.43881559e-02 -1.52033412e+00 -4.53546643e-02
1.02096580e-01 -3.40843014e-02 -6.12859949e-02 -9.80354726e-01
-7.53927112e-01 8.39468122e-01 6.69666231e-01 4.74470824e-01
1.33891881e+00 -3.36210996e-01 3.47545803e-01 -4.64524366e-02
4.52069521e-01 -8.86876583e-01 -6.10743821e-01 6.73259914e-01
1.15949285e+00 -1.06420553e+00 -3.48437756e-01 3.99242818e-01
-6.09125435e-01 1.57832479e+00 3.02502692e-01 -3.70751947e-01
1.23454463e+00 2.45424584e-01 -1.72824025e-01 4.80939060e-01
-1.14442563e+00 4.23860133e-01 5.99806368e-01 6.70116127e-01
5.77236652e-01 4.59968090e-01 -2.79036134e-01 1.25697231e+00
-4.26784456e-01 -3.99407782e-02 1.15382321e-01 1.57622159e-01
-1.69655293e-01 -8.37552845e-01 -4.65940297e-01 5.35390615e-01
-4.69259501e-01 -3.62441167e-02 3.15389276e-01 8.34726930e-01
-1.36633655e-02 8.21311116e-01 6.89775407e-01 -3.18958610e-01
2.81709999e-01 3.98352146e-01 -1.03231870e-01 1.78784534e-01
-4.14088190e-01 4.45591956e-02 -2.46575490e-01 -4.08139735e-01
-6.08193763e-02 -6.91969514e-01 -8.73810887e-01 -7.36445665e-01
-2.86129862e-01 4.24734503e-01 5.89148819e-01 9.36505318e-01
7.02181637e-01 4.37088758e-01 9.90294695e-01 -7.21627474e-01
-1.09684896e+00 -1.14503419e+00 -8.70984614e-01 -7.39082322e-02
4.06467676e-01 -7.76248932e-01 -4.73254472e-01 -1.37870982e-01] | [7.142037868499756, 3.5163469314575195] |
9affd78d-f5bb-4c50-9656-a186895b0277 | dasha-distributed-nonconvex-optimization-with | 2202.01268 | null | https://arxiv.org/abs/2202.01268v2 | https://arxiv.org/pdf/2202.01268v2.pdf | DASHA: Distributed Nonconvex Optimization with Communication Compression, Optimal Oracle Complexity, and No Client Synchronization | We develop and analyze DASHA: a new family of methods for nonconvex distributed optimization problems. When the local functions at the nodes have a finite-sum or an expectation form, our new methods, DASHA-PAGE and DASHA-SYNC-MVR, improve the theoretical oracle and communication complexity of the previous state-of-the-art method MARINA by Gorbunov et al. (2020). In particular, to achieve an epsilon-stationary point, and considering the random sparsifier RandK as an example, our methods compute the optimal number of gradients $\mathcal{O}\left(\frac{\sqrt{m}}{\varepsilon\sqrt{n}}\right)$ and $\mathcal{O}\left(\frac{\sigma}{\varepsilon^{3/2}n}\right)$ in finite-sum and expectation form cases, respectively, while maintaining the SOTA communication complexity $\mathcal{O}\left(\frac{d}{\varepsilon \sqrt{n}}\right)$. Furthermore, unlike MARINA, the new methods DASHA, DASHA-PAGE and DASHA-MVR send compressed vectors only and never synchronize the nodes, which makes them more practical for federated learning. We extend our results to the case when the functions satisfy the Polyak-Lojasiewicz condition. Finally, our theory is corroborated in practice: we see a significant improvement in experiments with nonconvex classification and training of deep learning models. | ['Peter Richtárik', 'Alexander Tyurin'] | 2022-02-02 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [-2.11669847e-01 1.61486030e-01 -7.47695491e-02 5.57101108e-02
-1.06804454e+00 -8.34290087e-01 -1.46121820e-02 7.33574629e-02
-8.65144372e-01 1.10322011e+00 -4.08173472e-01 -7.35286772e-01
-7.56098330e-01 -8.98969948e-01 -1.05562997e+00 -1.34800816e+00
-7.43665040e-01 2.95307964e-01 -1.09698139e-01 -1.27016991e-01
-1.16381928e-01 2.93491751e-01 -1.13756144e+00 -3.04101318e-01
8.16398919e-01 1.43550706e+00 1.22128926e-01 7.98344433e-01
2.94606268e-01 6.49072826e-01 -4.30019706e-01 -4.91945893e-01
6.69450760e-01 -4.86595094e-01 -6.40484095e-01 -1.95847958e-01
2.98792958e-01 -3.15028816e-01 -4.98178035e-01 1.41673887e+00
7.51419187e-01 3.72439027e-01 4.55222011e-01 -1.01792097e+00
-1.72007978e-01 7.87109673e-01 -8.24433982e-01 1.17888272e-01
-1.01363458e-01 -9.39166993e-02 8.03316295e-01 -8.99100900e-01
3.59586000e-01 6.49643898e-01 1.01308239e+00 1.43336058e-01
-7.22805142e-01 -8.27917635e-01 2.15327755e-01 1.41134560e-02
-1.58512688e+00 -2.22392261e-01 2.41125628e-01 8.25850219e-02
5.40896535e-01 6.93863392e-01 6.10057592e-01 2.08475530e-01
1.79001004e-01 7.37203896e-01 1.01310587e+00 -4.22176749e-01
5.59344888e-01 -2.54862547e-01 -1.93260640e-01 9.13546979e-01
6.16376698e-01 -8.66644457e-02 -3.88337463e-01 -2.94935465e-01
7.07078755e-01 1.01218954e-01 -3.75379443e-01 2.08265945e-01
-9.98571754e-01 1.00221884e+00 1.91151559e-01 3.45798194e-01
-3.72935861e-01 7.87414670e-01 2.16742128e-01 4.04203147e-01
4.80968088e-01 -3.48902382e-02 -4.73146439e-01 -3.16803306e-01
-1.26521194e+00 3.35828513e-01 1.07425499e+00 1.20329440e+00
7.40456760e-01 3.14170271e-01 -8.82482827e-02 4.53877985e-01
2.43018448e-01 1.13141608e+00 1.17198415e-01 -1.37651813e+00
7.74714649e-01 -1.03504643e-01 1.26489010e-02 -9.77128088e-01
-4.53622699e-01 -8.77016306e-01 -1.23335004e+00 1.14127703e-01
6.14207447e-01 -7.62796402e-01 -4.17462587e-01 1.57990658e+00
3.60960215e-01 1.31825447e-01 4.60063033e-02 8.95266950e-01
4.14920628e-01 7.69460976e-01 -6.25948966e-01 -5.45491457e-01
1.16598833e+00 -8.90801132e-01 -3.58643353e-01 8.39203373e-02
8.89777958e-01 -7.69141674e-01 8.23234856e-01 5.63888431e-01
-1.57246470e+00 -2.44730804e-02 -9.76282299e-01 3.52666080e-01
-1.41310086e-02 6.26538098e-02 6.02821887e-01 1.04948318e+00
-1.25526714e+00 8.33209038e-01 -8.94083619e-01 4.31526810e-01
5.79187095e-01 6.45163476e-01 -2.08091617e-01 -1.95416868e-01
-7.58867681e-01 2.07725406e-01 7.58550316e-02 2.81577826e-01
-1.00294793e+00 -6.01382315e-01 -6.10762537e-01 2.17097595e-01
4.65016991e-01 -2.57193446e-01 8.76788676e-01 -5.43281972e-01
-1.48445606e+00 4.63170171e-01 1.78463981e-01 -5.94953477e-01
6.24526203e-01 7.31737465e-02 1.97594956e-01 2.80950040e-01
-1.11760736e-01 1.43299356e-01 4.31443840e-01 -9.10696626e-01
-6.52848601e-01 -6.39443219e-01 1.90001309e-01 1.77982241e-01
-6.64673030e-01 -2.31480494e-01 -3.92463446e-01 -7.51065314e-01
1.14504047e-01 -1.06956255e+00 -5.05665123e-01 8.18403997e-03
-3.42308402e-01 3.08151990e-02 4.67393816e-01 -4.96661633e-01
1.07832599e+00 -2.08188319e+00 7.07545877e-02 7.85075307e-01
4.71996903e-01 2.98141152e-01 1.23270288e-01 4.21432406e-01
2.35953182e-01 2.09952250e-01 -1.65064842e-01 -4.43281054e-01
1.62464395e-01 4.25058514e-01 1.60810482e-02 1.03891754e+00
-8.91994774e-01 3.72302711e-01 -8.09012055e-01 -3.54169101e-01
-2.48376280e-01 3.92832130e-01 -7.86664903e-01 -2.59798020e-01
1.29142925e-01 2.72771955e-01 -6.93775594e-01 4.57418144e-01
1.04346514e+00 -4.91698027e-01 2.87507385e-01 1.08588494e-01
-7.72054791e-02 -1.01165501e-02 -1.77916026e+00 1.48301435e+00
-4.79687929e-01 3.81721824e-01 7.84859955e-01 -1.31452215e+00
4.77173090e-01 3.13303262e-01 9.03896987e-01 -4.90576237e-01
2.81272262e-01 5.41043103e-01 -4.75624204e-01 -7.23098367e-02
2.74130970e-01 1.28178217e-03 7.61408359e-02 7.23803580e-01
-1.21014901e-01 -5.90091087e-02 1.87294692e-01 3.13529931e-02
1.36057889e+00 -3.02708685e-01 -9.55138505e-02 -5.82295120e-01
4.01793718e-01 -3.74309540e-01 5.81125617e-01 1.11006939e+00
-2.41830517e-02 3.54045331e-01 7.15463221e-01 -2.33750358e-01
-9.37756121e-01 -8.48385394e-01 -1.82149231e-01 1.15677631e+00
2.00502411e-01 -4.78380889e-01 -8.55480075e-01 -6.32967889e-01
-1.78180337e-01 5.32792270e-01 -4.25167739e-01 2.25482300e-01
-4.70040619e-01 -8.40982199e-01 8.80811036e-01 4.86920923e-01
5.49667835e-01 -5.28323054e-01 -6.52747869e-01 -7.58536234e-02
2.78575774e-02 -9.88229334e-01 -7.66185164e-01 6.52242541e-01
-7.07642972e-01 -7.44937956e-01 -1.00710106e+00 -5.43403804e-01
9.00562525e-01 1.87495992e-01 7.71136820e-01 2.02580273e-01
-1.94852408e-02 3.34156603e-01 -2.07614228e-01 -3.94583583e-01
1.90269023e-01 1.11446485e-01 1.02853432e-01 -1.49691612e-01
-3.76986265e-01 -7.31966257e-01 -8.57705772e-01 1.15033522e-01
-9.82445061e-01 -3.08256239e-01 2.60340124e-01 8.59340608e-01
1.04216075e+00 1.08513512e-01 2.73875445e-01 -6.51968122e-01
3.35247576e-01 -3.63153994e-01 -1.21414006e+00 -5.36248535e-02
-7.13276684e-01 2.45525911e-02 9.99732673e-01 -2.75403321e-01
-3.89333755e-01 1.22274633e-03 -2.06907690e-01 -7.46962011e-01
5.48195660e-01 5.98400831e-01 2.14104906e-01 -4.14984465e-01
4.58325565e-01 3.56200725e-01 -1.09888278e-01 -2.39730582e-01
3.64254743e-01 4.42924738e-01 3.87005717e-01 -6.78582907e-01
8.34490955e-01 7.05823660e-01 4.27668780e-01 -5.65304160e-01
-7.09401369e-01 -1.38224453e-01 3.34270932e-02 -1.37872040e-01
6.72584057e-01 -8.86355281e-01 -1.46524525e+00 1.64098978e-01
-8.17946553e-01 -4.98373866e-01 -7.31345594e-01 7.18070209e-01
-6.70971036e-01 5.23919344e-01 -5.16373217e-01 -1.05291271e+00
-6.86943412e-01 -1.16880584e+00 7.87207365e-01 2.98312217e-01
3.47002059e-01 -1.19465995e+00 -3.41287762e-01 3.23915213e-01
4.55172032e-01 1.58985108e-01 5.62189937e-01 -7.46832192e-01
-6.92272961e-01 -3.34917456e-01 -1.64177105e-01 6.55481637e-01
-2.82521546e-01 -5.62847912e-01 -4.40885037e-01 -6.89532995e-01
1.60700470e-01 -2.32420266e-01 5.27756333e-01 3.77014875e-01
1.60511172e+00 -9.62018847e-01 -1.24848634e-01 1.03094232e+00
1.59396195e+00 1.69022568e-02 5.67282677e-01 1.35308104e-02
2.34995648e-01 -2.96486527e-01 2.60885268e-01 1.06444621e+00
3.21298718e-01 4.00807649e-01 7.81557560e-01 -3.35652195e-02
7.87228420e-02 2.02761352e-01 4.00230467e-01 9.67773080e-01
-2.81323165e-01 -4.38331753e-01 -5.61504126e-01 6.55504882e-01
-1.70666933e+00 -7.71479845e-01 -1.08565450e-01 2.40288901e+00
1.00869179e+00 -1.78022280e-01 -2.67800659e-01 1.77736819e-01
3.87415260e-01 3.01320016e-01 -3.71060371e-01 -3.56811434e-01
-2.65788287e-01 8.06278765e-01 1.37372935e+00 3.41132045e-01
-7.02461481e-01 5.03548622e-01 4.41756296e+00 1.41773534e+00
-9.43783283e-01 4.43641156e-01 5.36539435e-01 -4.71598268e-01
-4.95456994e-01 3.45140958e-04 -8.06055784e-01 6.19583368e-01
8.68878901e-01 -1.04223356e-01 9.14164007e-01 9.49019372e-01
-6.54647723e-02 -2.52545267e-01 -8.64351809e-01 1.15771353e+00
2.12126777e-01 -1.49232292e+00 -6.63557053e-01 3.93605232e-01
1.13471603e+00 8.59610811e-02 2.99609154e-01 -2.97082122e-02
7.17535794e-01 -1.04305220e+00 6.62650883e-01 1.38519973e-01
8.58657062e-01 -8.17784607e-01 8.32511902e-01 4.22374874e-01
-1.19192159e+00 -2.09177583e-01 -3.73721421e-01 1.53540030e-01
1.20537549e-01 9.88366187e-01 -4.34687138e-01 9.13443089e-01
1.00241947e+00 8.64663124e-02 2.05279797e-01 9.75706220e-01
1.53024402e-03 7.38661826e-01 -9.43951786e-01 -2.54617423e-01
5.12637734e-01 -3.76068652e-01 7.38869071e-01 8.17072809e-01
6.79382026e-01 3.83833528e-01 4.03225869e-01 2.98046947e-01
-4.88097250e-01 1.66487470e-01 -1.48445815e-01 1.96824655e-01
5.74589312e-01 9.89520967e-01 -7.73175359e-01 -1.16258189e-01
-1.94992468e-01 5.48078477e-01 4.06332053e-02 4.10553157e-01
-1.06798589e+00 -6.98481321e-01 4.55620199e-01 2.02491693e-02
8.41579556e-01 -4.58707988e-01 -5.75222492e-01 -9.47105348e-01
3.35040867e-01 -7.96594918e-01 4.65026855e-01 -1.73306108e-01
-9.66734648e-01 3.16377759e-01 3.26329581e-02 -1.17746055e+00
9.89244226e-03 -4.73342448e-01 -2.61132002e-01 7.03702390e-01
-1.27572060e+00 -6.38444424e-01 -2.09501967e-01 9.84950304e-01
-6.31155819e-02 -4.07027513e-01 7.36581504e-01 7.46488988e-01
-2.27765501e-01 9.62598503e-01 8.34540009e-01 1.40191913e-01
2.00582400e-01 -8.27601135e-01 -3.86208385e-01 7.27306247e-01
1.47410454e-02 1.60326257e-01 6.07948780e-01 -3.16745669e-01
-1.83159709e+00 -9.81437087e-01 5.03306150e-01 2.26415545e-01
4.86955583e-01 -1.79432288e-01 -3.32430154e-01 7.63768315e-01
2.30226278e-01 5.35167515e-01 5.48599005e-01 -2.62611091e-01
1.34395182e-01 -5.45484662e-01 -1.21425438e+00 4.17479008e-01
9.24104869e-01 -1.88076884e-01 2.76788354e-01 9.10381734e-01
3.52014601e-01 -8.64335895e-01 -1.12491536e+00 2.89063245e-01
2.53832817e-01 -8.79682302e-01 7.20315278e-01 -1.89270735e-01
6.14380557e-03 -1.12908326e-01 -5.68170846e-01 -8.85642946e-01
2.81908125e-01 -1.27823973e+00 -2.39655226e-01 5.50995231e-01
4.74732667e-01 -9.30457890e-01 9.08542991e-01 2.10334092e-01
-2.95787454e-01 -9.93217766e-01 -1.72555006e+00 -8.30599904e-01
3.93855214e-01 -4.53966528e-01 3.51267517e-01 7.10180879e-01
-1.88686743e-01 -4.52824086e-01 -4.20766890e-01 1.26589239e-01
6.39585018e-01 -1.28039226e-01 6.46237314e-01 -6.57746613e-01
-7.50539124e-01 -2.55964637e-01 -5.53939268e-02 -1.15339041e+00
-1.84639394e-01 -1.06826234e+00 -3.18113603e-02 -1.40355778e+00
-5.64616499e-03 -9.84454334e-01 -2.32292339e-01 7.88855314e-01
4.15327132e-01 1.24887101e-01 3.70188713e-01 1.59153268e-02
-6.77539468e-01 5.52655578e-01 1.22380292e+00 -2.70629954e-02
-7.82653987e-02 2.25033402e-01 -6.12904370e-01 7.41450489e-01
4.72815752e-01 -6.65049434e-01 -4.81393427e-01 -8.06931853e-01
8.11611354e-01 5.01692593e-01 3.52669835e-01 -8.85672092e-01
5.75109899e-01 -1.09010592e-01 3.25877033e-02 -5.39770544e-01
4.54949886e-01 -7.58587420e-01 7.07853138e-02 5.31182289e-01
-5.02917394e-02 4.07253414e-01 -3.48501727e-02 4.65694308e-01
1.94412306e-01 -4.48771477e-01 6.84110761e-01 -2.17941124e-02
2.92649984e-01 5.31897724e-01 -3.22108954e-01 2.63164520e-01
1.25548589e+00 -4.99730706e-02 -4.35166538e-01 -7.71556258e-01
-7.21799314e-01 3.65680784e-01 -4.40932103e-02 -4.47302520e-01
3.02311778e-01 -1.16868937e+00 -7.64812410e-01 8.97112265e-02
-7.72766232e-01 5.71535766e-01 4.35026795e-01 1.43483996e+00
-7.16926754e-01 2.20449299e-01 3.20321023e-01 -5.44291317e-01
-7.60392666e-01 1.20668709e-01 5.00135481e-01 -4.95373189e-01
-4.07709360e-01 1.16658211e+00 -4.17104512e-01 -2.68353373e-01
4.64060992e-01 -3.12237471e-01 5.64632118e-01 -1.13915317e-01
2.40700692e-01 9.05066788e-01 3.13543111e-01 -1.02993950e-01
-2.96757072e-01 2.98285335e-01 2.73907095e-01 -2.42756054e-01
1.57954144e+00 -5.27628660e-02 -3.75073940e-01 -1.00759350e-01
1.37090600e+00 3.84057552e-01 -1.18627572e+00 -5.96710555e-02
-6.13605142e-01 -2.14763388e-01 7.48434514e-02 -4.78273034e-01
-1.64704144e+00 7.86040246e-01 6.95983231e-01 2.30070964e-01
1.13358271e+00 -1.59514427e-01 9.69594002e-01 4.79239255e-01
7.42926538e-01 -1.40985811e+00 -3.98731790e-03 6.84271574e-01
4.71783191e-01 -6.13409996e-01 4.03744578e-01 -1.12800896e-01
-3.72742265e-01 1.19019783e+00 4.46190126e-02 -2.11032793e-01
1.04610550e+00 5.88852465e-01 -3.59599888e-01 -4.11259308e-02
-4.02736843e-01 1.58129364e-01 -1.15868405e-01 1.22424297e-01
1.11222923e-01 9.72705930e-02 -6.21911943e-01 8.89630139e-01
-4.82024074e-01 -1.73204631e-01 5.49757779e-01 9.78528380e-01
-2.60941267e-01 -8.89367700e-01 -3.86237264e-01 3.72592062e-01
-7.72205412e-01 -1.31054983e-01 4.38483298e-01 7.14374781e-01
1.81866288e-01 9.42246258e-01 -8.73865783e-02 -2.23968197e-02
-1.25588253e-01 -2.70554781e-01 3.83107573e-01 -2.07201764e-01
-5.90835154e-01 3.63875479e-01 2.15758570e-02 -3.82001281e-01
-1.35205507e-01 -5.81072450e-01 -1.81164527e+00 -7.12215960e-01
-2.84978092e-01 4.88275915e-01 8.17694783e-01 1.12751877e+00
3.05680871e-01 3.39707613e-01 9.18112576e-01 -5.54925323e-01
-1.20062029e+00 -5.34955025e-01 -7.71694005e-01 -1.74797177e-01
5.07314764e-02 -1.56348646e-01 -3.51525873e-01 -2.82509595e-01] | [6.373004913330078, 4.621700286865234] |
44a6ac7f-d39e-4965-9694-b5d29a61f0f7 | leveraging-mocap-data-for-human-mesh-recovery | 2110.09243 | null | https://arxiv.org/abs/2110.09243v1 | https://arxiv.org/pdf/2110.09243v1.pdf | Leveraging MoCap Data for Human Mesh Recovery | Training state-of-the-art models for human body pose and shape recovery from images or videos requires datasets with corresponding annotations that are really hard and expensive to obtain. Our goal in this paper is to study whether poses from 3D Motion Capture (MoCap) data can be used to improve image-based and video-based human mesh recovery methods. We find that fine-tune image-based models with synthetic renderings from MoCap data can increase their performance, by providing them with a wider variety of poses, textures and backgrounds. In fact, we show that simply fine-tuning the batch normalization layers of the model is enough to achieve large gains. We further study the use of MoCap data for video, and introduce PoseBERT, a transformer module that directly regresses the pose parameters and is trained via masked modeling. It is simple, generic and can be plugged on top of any state-of-the-art image-based model in order to transform it in a video-based model leveraging temporal information. Our experimental results show that the proposed approaches reach state-of-the-art performance on various datasets including 3DPW, MPI-INF-3DHP, MuPoTS-3D, MCB and AIST. Test code and models will be available soon. | ['Grégory Rogez', 'Yannis Kalantidis', 'Romain Brégier', 'Philippe Weinzaepfel', 'Thibault Groueix', 'Fabien Baradel'] | 2021-10-18 | null | null | null | null | ['3d-human-reconstruction', 'human-mesh-recovery'] | ['computer-vision', 'computer-vision'] | [-9.10491198e-02 -1.95647940e-01 8.52208585e-03 -2.15556383e-01
-7.66251743e-01 -1.35094985e-01 3.25426817e-01 -5.92253685e-01
-4.20895785e-01 3.37476283e-01 1.68762848e-01 2.73006439e-01
3.20191324e-01 -6.48436546e-01 -1.07461190e+00 -4.15005356e-01
-9.68028232e-02 7.55044460e-01 7.54266381e-01 -5.44885218e-01
-5.51048107e-02 3.86098891e-01 -1.77538693e+00 5.65744758e-01
2.68486321e-01 9.34963882e-01 3.58845782e-03 8.56655598e-01
1.12649873e-01 6.44680798e-01 -2.30165839e-01 -3.01857501e-01
5.66675782e-01 -3.19335252e-01 -8.45351577e-01 1.14386991e-01
9.98902321e-01 -5.16279638e-01 -6.17175221e-01 4.62231159e-01
8.88149738e-01 8.65067318e-02 5.07354498e-01 -9.79013145e-01
-1.51054174e-01 -1.53437711e-03 -6.50989175e-01 -3.04634608e-02
5.46983361e-01 4.31017220e-01 2.99174935e-01 -9.99827981e-01
9.39291179e-01 1.42685306e+00 9.68636572e-01 8.84161353e-01
-1.26892662e+00 -6.38607144e-01 -9.97319669e-02 1.19314395e-01
-1.37116921e+00 -5.68045676e-01 7.72882760e-01 -6.18046641e-01
9.73503709e-01 3.03192466e-01 1.02674651e+00 1.41254234e+00
1.33036733e-01 8.25552583e-01 1.03074205e+00 -3.65474761e-01
-1.31487563e-01 -3.22097540e-01 -2.72923678e-01 1.03676939e+00
-1.39049411e-01 1.79370642e-01 -8.39736581e-01 -1.17181037e-02
1.18186414e+00 -1.02866329e-01 -3.40635777e-01 -7.87883162e-01
-1.35098302e+00 5.24792969e-01 2.94351369e-01 1.49130104e-02
-3.02390665e-01 5.33659399e-01 3.21836233e-01 2.15263888e-01
6.12774611e-01 -8.68151635e-02 -5.41843891e-01 -2.54978657e-01
-1.24426568e+00 7.22486675e-01 6.62640214e-01 7.74665713e-01
5.23594260e-01 1.08119780e-02 -1.68239251e-01 8.29283953e-01
2.46140599e-01 6.63659453e-01 3.72268856e-01 -1.25511873e+00
4.27696288e-01 3.09167743e-01 6.38052672e-02 -9.76536810e-01
-6.44255161e-01 -1.32413089e-01 -4.79656219e-01 3.77600074e-01
6.02293909e-01 -1.12396507e-02 -1.29341424e+00 1.56214643e+00
7.39831865e-01 4.37986225e-01 -3.71697962e-01 1.45707738e+00
9.72822011e-01 3.79133791e-01 -6.85958341e-02 2.24185526e-01
1.29061222e+00 -1.15993547e+00 -3.52356374e-01 -2.94870555e-01
4.29632366e-01 -8.09017599e-01 1.25653136e+00 4.64427471e-01
-1.40173960e+00 -6.89394295e-01 -9.52060759e-01 -2.24428892e-01
8.06403607e-02 2.50922740e-02 5.98113418e-01 4.07070577e-01
-8.87623727e-01 9.23572183e-01 -1.44852567e+00 -5.59917212e-01
3.42701882e-01 5.09089887e-01 -6.43695593e-01 -1.06076514e-02
-8.40853393e-01 7.47296631e-01 -8.27092081e-02 8.77927318e-02
-1.07342207e+00 -6.40218318e-01 -9.31517303e-01 -5.03007412e-01
4.65358496e-01 -1.16358399e+00 1.12405157e+00 -6.85171425e-01
-1.68727708e+00 1.16446316e+00 5.27433194e-02 -2.93075383e-01
8.38667929e-01 -4.94369626e-01 -8.15537795e-02 3.50537449e-01
-1.41352445e-01 8.18967998e-01 1.02752388e+00 -1.01948977e+00
-2.53282309e-01 -4.88808960e-01 8.31083860e-03 9.20251757e-02
-1.60487458e-01 1.48810610e-01 -1.08498955e+00 -8.05870295e-01
1.18905716e-01 -1.36753654e+00 -1.65205255e-01 3.55792880e-01
-1.78909957e-01 3.17590356e-01 6.71364486e-01 -1.02574706e+00
8.57612610e-01 -1.72691369e+00 6.24695003e-01 -2.37602759e-02
-5.98621517e-02 3.09867352e-01 -1.54030979e-01 2.66976833e-01
1.44765839e-01 -1.45040289e-01 -1.69677585e-01 -5.84446549e-01
-1.08457983e-01 3.96262139e-01 2.29556426e-01 8.03196907e-01
3.38825583e-02 9.28451538e-01 -5.04890442e-01 -6.75058365e-01
5.55839419e-01 8.05006981e-01 -9.52011168e-01 1.93355277e-01
-2.91026503e-01 9.34027076e-01 -2.02265188e-01 6.94239557e-01
6.41848028e-01 -2.39276052e-01 -8.08520708e-03 -3.87702137e-01
1.62114948e-01 7.38365129e-02 -1.38660073e+00 2.39661837e+00
-3.32450569e-01 2.60240138e-01 2.10393474e-01 -8.21713924e-01
3.29044104e-01 3.31751466e-01 6.80526257e-01 -5.76240003e-01
2.10564867e-01 2.49343589e-02 -1.85435787e-01 -7.28057444e-01
2.82105118e-01 -1.18880145e-01 -2.83703376e-02 5.44583350e-02
3.35254967e-01 -2.86543705e-02 3.01931202e-02 -1.88369930e-01
1.16617119e+00 8.14737618e-01 -1.99512765e-01 -5.66769531e-03
4.49576646e-01 5.00862710e-02 5.10363519e-01 3.82523417e-01
1.86473373e-02 1.20547378e+00 -3.93899344e-02 -5.40845454e-01
-1.05414462e+00 -1.00853705e+00 4.77509899e-03 1.17057133e+00
-5.87039590e-02 -5.65149963e-01 -8.64826322e-01 -4.95696217e-01
1.37729617e-02 -8.88944864e-02 -7.19973564e-01 1.24781668e-01
-1.03847647e+00 -6.76693678e-01 4.93121058e-01 6.83134913e-01
4.29983258e-01 -9.57328618e-01 -6.56741083e-01 1.68624580e-01
-3.68163496e-01 -1.18895090e+00 -5.04270732e-01 -2.01814070e-01
-1.07254219e+00 -1.04823399e+00 -1.10333109e+00 -6.16826236e-01
5.15231311e-01 -7.70068020e-02 1.11598110e+00 2.02654064e-01
-4.36441481e-01 4.82799739e-01 -3.69088024e-01 -1.02298908e-01
-2.43742391e-01 1.71931654e-01 1.49240062e-01 -1.34070456e-01
-2.04530165e-01 -8.48569512e-01 -8.87894392e-01 6.13613248e-01
-7.49899864e-01 3.13934445e-01 2.17299372e-01 6.95202589e-01
8.57552290e-01 -5.09714544e-01 -7.97996521e-02 -7.53121078e-01
-1.16340049e-01 -2.37858985e-02 -3.31031322e-01 -1.34297311e-01
-1.51294738e-01 1.23745225e-01 1.95961356e-01 -7.27766156e-01
-7.70831108e-01 5.00709653e-01 -5.30094862e-01 -9.82025743e-01
-8.90200660e-02 2.22805858e-01 -9.28279981e-02 -4.41210389e-01
7.82502413e-01 -7.79099613e-02 1.54739410e-01 -9.64226544e-01
3.67767334e-01 2.62933522e-01 9.74831104e-01 -7.01623201e-01
8.72648776e-01 7.61485994e-01 5.30373529e-02 -7.28338122e-01
-7.53577769e-01 -5.36485910e-01 -7.99157143e-01 -3.98986280e-01
9.80761230e-01 -1.10495615e+00 -5.10997057e-01 7.01403499e-01
-9.91038263e-01 -6.73075378e-01 -1.97821558e-02 4.21778917e-01
-7.75213242e-01 2.58297235e-01 -8.66613209e-01 -4.14754540e-01
-3.90551716e-01 -1.11678338e+00 1.51321840e+00 -4.21042517e-02
-1.94758788e-01 -7.08477616e-01 2.81741768e-01 7.88871765e-01
2.67221361e-01 6.65266633e-01 3.37292194e-01 -1.60235494e-01
-5.23465812e-01 -3.11746210e-01 2.39872292e-01 2.46550590e-01
-2.27824733e-01 -1.78674698e-01 -1.00122023e+00 -4.15927261e-01
-1.48933232e-01 -4.69237536e-01 8.21387768e-01 5.28354526e-01
1.14914024e+00 -2.56739229e-01 -2.94130266e-01 1.05069959e+00
1.11493111e+00 -5.94707310e-01 9.58661079e-01 3.69866312e-01
1.16121638e+00 4.51184154e-01 6.49427295e-01 4.27292377e-01
5.53728998e-01 1.30564296e+00 3.98887008e-01 -2.11225659e-01
-5.58286667e-01 -2.96939164e-01 4.11132276e-01 9.22165871e-01
-6.75511181e-01 2.15709120e-01 -9.56130147e-01 4.90315467e-01
-2.01848245e+00 -9.02485847e-01 -2.64827818e-01 2.27095532e+00
9.20918286e-01 7.61497580e-03 4.82337534e-01 7.48881400e-02
3.10430586e-01 -2.61134631e-03 -1.66357383e-01 5.49420416e-02
2.24715352e-01 5.34132302e-01 4.10076827e-01 3.61974478e-01
-1.24084675e+00 1.05022407e+00 6.18785238e+00 8.35448086e-01
-1.31413364e+00 3.54473174e-01 2.27933198e-01 -5.51173568e-01
2.86191672e-01 -1.21992156e-01 -6.68049693e-01 2.82759488e-01
9.59682286e-01 5.29725969e-01 5.27250230e-01 7.68443942e-01
2.75171250e-01 6.74874112e-02 -1.23228109e+00 1.15764415e+00
1.76303372e-01 -1.31012821e+00 -1.85933322e-01 1.10804617e-01
6.04643524e-01 2.23060116e-01 -2.75329500e-01 2.77811319e-01
-4.50894311e-02 -8.96975517e-01 9.96339023e-01 7.30417311e-01
7.63742626e-01 -5.77958882e-01 5.19892156e-01 3.70862544e-01
-1.16708553e+00 3.95138472e-01 -3.30332100e-01 -3.98532897e-02
3.57616544e-01 2.20797062e-01 -4.28764760e-01 5.44352770e-01
1.08355165e+00 5.18483162e-01 -7.62045801e-01 9.81747925e-01
-4.57199141e-02 5.03562570e-01 -4.98031735e-01 5.46529174e-01
-2.66785353e-01 3.22106570e-01 4.94742721e-01 1.04597855e+00
1.87760487e-01 -3.03262509e-02 4.50835317e-01 5.72293341e-01
-2.64158789e-02 6.23511113e-02 -2.37065032e-01 2.68604815e-01
-1.17448419e-01 1.22109544e+00 -6.81577981e-01 -2.96206802e-01
-3.93675536e-01 1.21208537e+00 2.83466786e-01 4.70861308e-02
-1.12521935e+00 1.69125676e-01 5.67858517e-01 9.48891163e-01
5.45922339e-01 -3.61201704e-01 9.98114273e-02 -1.29928493e+00
5.53413965e-02 -1.04294956e+00 2.61876076e-01 -8.09797585e-01
-1.22735715e+00 3.91458035e-01 2.42289126e-01 -1.28560388e+00
-4.85339880e-01 -7.14703798e-01 -1.72659099e-01 5.30661583e-01
-1.07627845e+00 -1.48663378e+00 -5.41325510e-01 9.45484698e-01
4.61343259e-01 2.21864656e-01 8.04635465e-01 6.50780439e-01
-4.20502633e-01 5.56930542e-01 -4.22530472e-01 3.92402291e-01
8.87625813e-01 -9.38562989e-01 5.99756539e-01 5.89790761e-01
2.95396447e-01 2.85634309e-01 7.94736326e-01 -7.03027427e-01
-1.78731322e+00 -9.99472558e-01 2.94576555e-01 -8.08238029e-01
2.40667358e-01 -5.52017510e-01 -7.30811357e-01 5.89825511e-01
-1.80374116e-01 5.77728868e-01 3.04105520e-01 -9.92416441e-02
-7.92552531e-02 5.95143847e-02 -1.07078040e+00 5.68601847e-01
1.49777031e+00 -3.10596049e-01 -5.34801185e-01 2.52240777e-01
6.04586840e-01 -1.16402185e+00 -1.08064198e+00 7.28747964e-01
8.67283821e-01 -8.98297727e-01 1.45080101e+00 -5.44712186e-01
5.69853008e-01 -3.65584403e-01 -1.87308162e-01 -9.57268119e-01
7.79185491e-03 -5.83579957e-01 -4.53487217e-01 9.04730141e-01
5.46935536e-02 -3.03354878e-02 1.10090280e+00 5.33291340e-01
-1.10431187e-01 -7.01189816e-01 -9.68831956e-01 -7.41147339e-01
5.42161092e-02 -7.08546281e-01 2.95582145e-01 6.48449898e-01
-2.48726532e-01 1.04880989e-01 -7.90829718e-01 1.02065749e-01
5.92708230e-01 -4.27965261e-02 1.40431786e+00 -9.88121748e-01
-6.48121595e-01 2.02935830e-01 -6.22186780e-01 -9.86690164e-01
-3.86991166e-02 -6.77486181e-01 1.07994184e-01 -1.45415163e+00
1.17503628e-01 -2.26158947e-01 1.25647917e-01 6.75442398e-01
6.03664927e-02 8.18846881e-01 3.24764550e-01 2.30826139e-01
-5.58824301e-01 5.30536890e-01 1.50299668e+00 1.48814246e-02
-1.06119074e-01 -8.91345665e-02 -2.32875496e-02 1.03445554e+00
3.03551942e-01 -5.94243109e-01 -1.21096581e-01 -5.25394857e-01
-7.67638907e-02 3.25897574e-01 7.96336234e-01 -1.45884848e+00
2.30386294e-02 5.62354997e-02 5.70532262e-01 -6.01164579e-01
8.09697151e-01 -7.06163704e-01 5.73240995e-01 3.44596028e-01
4.60564159e-02 8.64940695e-03 3.46050739e-01 4.31918621e-01
1.16716459e-01 1.01144880e-01 6.59882307e-01 -3.79475623e-01
-5.74236989e-01 4.86151904e-01 -1.14817768e-02 3.40686619e-01
5.66253185e-01 -1.54529810e-01 1.71095952e-01 -3.21106136e-01
-9.11654949e-01 -4.38845605e-02 8.47569644e-01 5.32034457e-01
5.99643826e-01 -1.48461890e+00 -8.09256315e-01 1.50782317e-01
5.93681075e-02 1.28822327e-01 2.84049511e-01 1.00847995e+00
-8.02955627e-01 4.29372154e-02 -3.64556134e-01 -1.09277761e+00
-1.45925593e+00 3.97271246e-01 4.28778887e-01 -3.18573922e-01
-9.98214602e-01 7.53511488e-01 1.16942994e-01 -5.66999316e-01
2.51085818e-01 -4.18561459e-01 2.96295226e-01 -3.59401286e-01
4.48650330e-01 3.24368924e-01 3.31345052e-01 -8.85345757e-01
-5.41165650e-01 9.67361927e-01 3.31404239e-01 -2.84147024e-01
1.49600959e+00 8.00392181e-02 2.72862464e-01 1.58170834e-01
1.09049165e+00 1.17722424e-02 -1.50745690e+00 6.76242784e-02
-4.92889971e-01 -6.31570876e-01 7.98649639e-02 -7.20992446e-01
-1.46082497e+00 8.59025538e-01 9.30291533e-01 -6.04882538e-01
1.02446473e+00 1.10843569e-01 1.12854719e+00 6.25646533e-03
8.57977331e-01 -1.05228841e+00 2.50448734e-01 2.46353790e-01
1.09454346e+00 -1.10861897e+00 1.83185339e-01 -5.33596039e-01
-4.03261840e-01 8.23045313e-01 6.62188113e-01 -3.07079732e-01
6.35351956e-01 3.72339249e-01 -4.96816114e-02 -4.76022959e-01
-4.96273726e-01 -2.46245354e-01 6.58865094e-01 6.62570775e-01
4.72341537e-01 -1.97050899e-01 -1.78455532e-01 6.71169162e-01
-3.33511829e-01 3.19611251e-01 1.17512368e-01 1.09718144e+00
-6.49540573e-02 -1.37777293e+00 -6.91537857e-01 1.28914982e-01
-6.93758428e-01 7.03089312e-02 -9.20661837e-02 9.41196442e-01
2.83559114e-01 5.27776241e-01 -2.24267066e-01 -6.63281858e-01
6.15740657e-01 5.18493839e-02 1.09492409e+00 -5.78917682e-01
-5.79458117e-01 2.75232255e-01 3.60449642e-01 -1.02744043e+00
-7.10439146e-01 -5.46941817e-01 -1.27290440e+00 -5.25376379e-01
-2.83726215e-01 -5.40017784e-01 5.06210506e-01 8.43722820e-01
3.57051820e-01 5.07312119e-01 -9.65995938e-02 -1.65779305e+00
-1.64788708e-01 -9.77032959e-01 -2.31961459e-01 5.87926507e-01
8.36912915e-02 -1.08032084e+00 -5.30249700e-02 4.10234272e-01] | [7.075483798980713, -0.911954402923584] |
8506d3a9-47fe-4ccf-b2ce-0dd1f444cf11 | logical-activation-functions-logit-space-1 | 2110.11940 | null | https://arxiv.org/abs/2110.11940v2 | https://arxiv.org/pdf/2110.11940v2.pdf | Logical Activation Functions: Logit-space equivalents of Probabilistic Boolean Operators | The choice of activation functions and their motivation is a long-standing issue within the neural network community. Neuronal representations within artificial neural networks are commonly understood as logits, representing the log-odds score of presence of features within the stimulus. We derive logit-space operators equivalent to probabilistic Boolean logic-gates AND, OR, and XNOR for independent probabilities. Such theories are important to formalize more complex dendritic operations in real neurons, and these operations can be used as activation functions within a neural network, introducing probabilistic Boolean-logic as the core operation of the neural network. Since these functions involve taking multiple exponents and logarithms, they are computationally expensive and not well suited to be directly used within neural networks. Consequently, we construct efficient approximations named $\text{AND}_\text{AIL}$ (the AND operator Approximate for Independent Logits), $\text{OR}_\text{AIL}$, and $\text{XNOR}_\text{AIL}$, which utilize only comparison and addition operations, have well-behaved gradients, and can be deployed as activation functions in neural networks. Like MaxOut, $\text{AND}_\text{AIL}$ and $\text{OR}_\text{AIL}$ are generalizations of ReLU to two-dimensions. While our primary aim is to formalize dendritic computations within a logit-space probabilistic-Boolean framework, we deploy these new activation functions, both in isolation and in conjunction to demonstrate their effectiveness on a variety of tasks including image classification, transfer learning, abstract reasoning, and compositional zero-shot learning. | ['Sageev Oore', 'Thomas Trappenberg', "Jason d'Eon", 'Robert Earle', 'Scott C. Lowe'] | 2021-10-22 | logical-activation-functions-logit-space | https://openreview.net/forum?id=Ck_iw4jMC4l | https://openreview.net/pdf?id=Ck_iw4jMC4l | null | ['compositional-zero-shot-learning'] | ['computer-vision'] | [ 4.91023183e-01 -9.36624706e-02 1.96476594e-01 -4.82361794e-01
-2.20950067e-01 -6.40112400e-01 5.95509470e-01 4.49698299e-01
-8.64981472e-01 8.19063902e-01 -4.05837238e-01 -6.44110918e-01
-5.43203533e-01 -1.20370197e+00 -9.74111736e-01 -8.61163855e-01
-4.87898409e-01 1.51300684e-01 3.06254834e-01 -1.90961212e-01
1.88770950e-01 6.86920524e-01 -1.86452460e+00 2.60291338e-01
3.53695452e-01 1.50643826e+00 -1.10319233e-03 5.18184662e-01
-2.05618262e-01 7.80482411e-01 -6.08351111e-01 -3.96000415e-01
1.99709364e-04 -1.64322972e-01 -4.16362703e-01 -7.05888569e-01
2.21320882e-01 -4.53035198e-02 -4.41692740e-01 1.32761967e+00
2.79967874e-01 2.65461326e-01 1.23063624e+00 -1.28481591e+00
-8.54764760e-01 7.93388546e-01 -2.14798406e-01 4.30108011e-01
6.14060685e-02 2.17519417e-01 1.09183931e+00 -8.32071602e-01
2.04830393e-01 1.16373253e+00 6.90761805e-01 4.70506638e-01
-1.59170616e+00 -6.34106398e-01 2.61449635e-01 -4.78761736e-03
-1.71965992e+00 -3.21947277e-01 3.42670858e-01 -4.59795445e-01
1.10993612e+00 2.46430650e-01 6.78437948e-01 8.78405511e-01
4.95461434e-01 7.95146346e-01 1.20528030e+00 -4.58548188e-01
7.23230541e-01 7.05987168e-03 4.08192545e-01 8.14713657e-01
3.30804288e-01 9.91672426e-02 -5.89081526e-01 -1.76300049e-01
1.09613943e+00 2.95329750e-01 -2.57371873e-01 9.64591056e-02
-9.72646475e-01 9.51437175e-01 5.19102335e-01 2.41595551e-01
-3.94450814e-01 9.19153571e-01 1.86076507e-01 1.53926998e-01
-6.31822720e-02 3.54814798e-01 -2.61052072e-01 4.33491990e-02
-8.55939746e-01 4.95324165e-01 7.24932909e-01 8.76908481e-01
9.16723907e-01 1.82494894e-01 -2.04179451e-01 7.32595325e-01
2.90804118e-01 5.49626112e-01 2.24624500e-01 -1.01766503e+00
-8.64510834e-02 4.25238639e-01 -1.96440101e-01 -6.11356974e-01
-2.53518641e-01 -1.76357254e-01 -8.94232988e-01 6.40577257e-01
5.17216206e-01 4.67272568e-03 -1.09714305e+00 2.11727571e+00
-3.94372463e-01 -1.16345704e-01 -2.02949960e-02 4.49821085e-01
4.45437402e-01 6.68843865e-01 3.69902849e-01 -1.90074772e-01
1.53472412e+00 -1.21344544e-01 -4.69091058e-01 -3.10321540e-01
3.38243544e-01 -1.76492587e-01 1.30094624e+00 4.07360256e-01
-1.25736630e+00 -1.70283124e-01 -1.16225719e+00 -7.50043988e-02
-9.34758782e-01 -4.93267089e-01 9.40281451e-01 6.21553719e-01
-1.31145203e+00 6.78546906e-01 -8.39580297e-01 3.30253085e-03
8.13120365e-01 5.88416100e-01 -9.34002176e-02 -5.89997135e-02
-1.32884955e+00 8.11572075e-01 3.59607667e-01 1.59655929e-01
-7.71588743e-01 -3.61358613e-01 -9.58284736e-01 3.96989465e-01
-3.25668952e-03 -5.45807719e-01 9.43316042e-01 -6.84867680e-01
-1.12676346e+00 7.93256044e-01 -2.34146893e-01 -6.92308843e-01
-7.97462277e-03 3.32206517e-01 -9.47751403e-02 -1.10170599e-02
-8.80035087e-02 1.08978283e+00 7.28488386e-01 -8.31887424e-01
-4.03080374e-01 -5.53470016e-01 3.85213733e-01 -7.13443607e-02
-4.28856045e-01 -4.14553471e-02 -1.33194521e-01 -7.08699942e-01
3.54678571e-01 -5.94828129e-01 -1.50066525e-01 4.11965311e-01
-2.02100277e-01 -3.48848104e-01 1.57792166e-01 1.00379445e-01
1.21429658e+00 -2.13074708e+00 -6.14779182e-02 4.73013312e-01
3.31701100e-01 -1.72089174e-01 1.75471216e-01 -8.65397006e-02
-1.11504719e-01 4.09418911e-01 -6.73510551e-01 2.05469325e-01
2.20677942e-01 4.57551330e-01 -3.53329569e-01 3.86385024e-01
5.34665465e-01 8.93427670e-01 -6.84142590e-01 -3.67187828e-01
2.13291757e-02 6.25048757e-01 -5.77514946e-01 -3.74641001e-01
-3.90082151e-01 -4.00837779e-01 -3.50662053e-01 6.46424532e-01
2.78190821e-01 -2.07988411e-01 -2.60625869e-01 -7.73632303e-02
-1.31314605e-01 1.80847049e-01 -1.28937638e+00 1.23857200e+00
-4.58416305e-02 6.43729985e-01 -9.23630297e-02 -1.02902973e+00
8.20817530e-01 1.09462075e-01 1.83115095e-01 -4.39672679e-01
3.89315009e-01 1.59084350e-01 1.65705845e-01 1.16790578e-01
1.29890859e-01 -4.98096704e-01 -2.82507926e-01 5.42137027e-01
3.01104605e-01 -2.40626022e-01 5.24813175e-01 1.43118843e-01
1.36159182e+00 -6.87656775e-02 1.04130872e-01 -5.46866417e-01
3.14647347e-01 -2.82995105e-01 2.76182592e-01 1.10250628e+00
-1.62617311e-01 2.76114881e-01 9.61234570e-01 -2.89741278e-01
-7.14293182e-01 -1.57495463e+00 -5.88677406e-01 1.24344969e+00
-4.29404862e-02 -7.71069974e-02 -7.47809112e-01 -8.28845054e-03
-4.84261699e-02 8.56068909e-01 -8.02198708e-01 -2.56539136e-01
-8.99541453e-02 -9.83897805e-01 1.04873812e+00 8.17970753e-01
1.75416261e-01 -1.26080799e+00 -8.98639441e-01 2.54797965e-01
4.39881206e-01 -6.12859488e-01 -6.51360154e-02 1.20475399e+00
-7.82339096e-01 -6.14186764e-01 -6.24048054e-01 -5.93870342e-01
7.49557376e-01 -3.45790774e-01 8.02685261e-01 -2.00391740e-01
-5.80881536e-01 2.19411314e-01 1.76617697e-01 -7.43446946e-01
1.64787292e-01 -2.47357458e-01 1.87911600e-01 -1.94701806e-01
5.80502570e-01 -8.83517981e-01 -5.20333111e-01 2.52022278e-02
-1.34854245e+00 -3.01954895e-01 4.57968354e-01 1.00114501e+00
9.98543918e-01 1.70128763e-01 4.26038861e-01 -7.35867798e-01
7.47753978e-01 -2.36260697e-01 -7.60625660e-01 2.73035821e-02
-5.01362681e-01 5.19393444e-01 5.64026415e-01 -7.07200587e-01
-4.84287143e-01 -1.96160093e-01 -1.75002798e-01 -5.57694554e-01
-1.79041252e-01 7.16330290e-01 4.23455983e-03 8.52327794e-02
9.67316508e-01 2.28694588e-01 -2.62326241e-01 -7.69539736e-03
3.46849859e-01 1.73465595e-01 5.30446887e-01 -7.37398267e-01
3.09619278e-01 5.30449808e-01 2.53226697e-01 -5.05383790e-01
-4.77091044e-01 1.36065260e-01 -2.89210320e-01 -3.08102164e-02
8.71656001e-01 -5.98942637e-01 -1.22645736e+00 4.20511544e-01
-1.02434790e+00 -3.30212921e-01 -5.68173110e-01 3.10944557e-01
-6.43460512e-01 -1.97540253e-01 -7.63732910e-01 -1.17937851e+00
-1.59184173e-01 -1.27557337e+00 6.98583961e-01 3.71416897e-01
-2.84590900e-01 -9.80529785e-01 -6.25978887e-01 -4.87602472e-01
5.13780534e-01 -8.54547136e-03 1.59967911e+00 -7.15393543e-01
-4.93810296e-01 -3.81321698e-01 -4.41299200e-01 4.32340384e-01
-3.32167566e-01 1.28700808e-01 -1.15586078e+00 1.72472075e-01
9.30864438e-02 -3.82666111e-01 1.08479500e+00 4.28216368e-01
1.63015902e+00 -4.60514396e-01 -2.13132277e-01 3.98013085e-01
1.37796688e+00 2.49149591e-01 7.20078945e-01 -1.40203838e-03
1.34377301e-01 5.11780798e-01 -2.20436439e-01 5.38919568e-01
2.04169318e-01 6.86515495e-02 4.15854394e-01 3.96499395e-01
2.97721893e-01 -1.20149581e-02 3.61774385e-01 4.00462955e-01
-5.43293962e-03 -2.88887322e-01 -9.87817764e-01 2.96889752e-01
-1.54180872e+00 -1.19522870e+00 1.22654356e-01 2.38001180e+00
1.07960463e+00 6.74873650e-01 -4.17274594e-01 3.94712299e-01
6.05022907e-01 1.02924965e-01 -7.47674286e-01 -5.41800082e-01
-4.14701015e-01 1.04691553e+00 6.12297654e-01 3.62095565e-01
-8.20692539e-01 7.76074588e-01 5.93224859e+00 9.74871755e-01
-1.23905003e+00 -4.34783325e-02 8.53673518e-01 -3.44789743e-01
-5.72766006e-01 3.85667272e-02 -8.22674632e-01 3.83279473e-01
1.04755497e+00 1.54299200e-01 8.04644048e-01 7.05269933e-01
-3.26269567e-01 -3.94565135e-01 -1.52324474e+00 1.05401206e+00
-2.18303755e-01 -1.45183146e+00 7.17328191e-02 5.92971444e-02
4.80536342e-01 4.17426080e-02 3.69841039e-01 5.06288469e-01
4.74424809e-01 -1.52431536e+00 1.02462983e+00 5.47575533e-01
9.89546597e-01 -4.83984828e-01 3.43481749e-01 1.79707304e-01
-1.04657412e+00 -9.90737006e-02 -4.59549189e-01 -3.68486971e-01
-6.28274009e-02 6.83711112e-01 -3.06063831e-01 -2.87200660e-01
8.89435649e-01 1.04614578e-01 -1.49103761e-01 6.09135330e-01
-3.59397948e-01 4.37123299e-01 -8.03470433e-01 -5.98239660e-01
2.72016913e-01 5.93448356e-02 2.17091069e-01 1.18320405e+00
2.33859211e-01 2.45806292e-01 -1.60898358e-01 1.43242574e+00
-3.29245806e-01 -2.76554018e-01 -5.70766151e-01 -3.67017955e-01
6.51401043e-01 9.24517572e-01 -1.06925797e+00 -2.99412936e-01
2.23736726e-02 5.96415579e-01 3.93985540e-01 7.09230900e-01
-8.72884333e-01 -1.01069713e+00 7.26410806e-01 1.86130881e-01
2.27523103e-01 -3.11740696e-01 -7.38671958e-01 -7.41231322e-01
-1.12480093e-02 -3.96321714e-01 1.66305959e-01 -7.90996671e-01
-1.30034256e+00 5.32205820e-01 3.00406218e-01 -6.13389075e-01
-1.96331576e-01 -1.35935748e+00 -4.36375231e-01 1.14047921e+00
-1.21777558e+00 -4.71820980e-01 8.82786736e-02 7.38612413e-01
2.94158421e-02 1.64857898e-02 1.07882130e+00 9.70663726e-02
-4.34738785e-01 5.47359824e-01 9.10829082e-02 1.35253787e-01
7.90773332e-02 -1.07002664e+00 8.42385963e-02 4.52065527e-01
1.10308178e-01 1.10902405e+00 5.64781547e-01 -2.82228351e-01
-1.48083365e+00 -6.66804969e-01 7.11154938e-01 -7.14684278e-02
6.51440978e-01 -6.70515716e-01 -9.92190063e-01 6.39074683e-01
-3.10336679e-01 2.85060823e-01 5.87774158e-01 -5.94742782e-02
-5.67740738e-01 -1.80516005e-01 -1.30239391e+00 9.65439081e-01
9.91582453e-01 -7.74966002e-01 -4.92836028e-01 2.73795277e-01
4.93161470e-01 4.43293713e-02 -9.19150531e-01 3.67380530e-01
7.30136812e-01 -8.38660955e-01 9.08818781e-01 -4.81944472e-01
2.38145784e-01 -1.93074971e-01 -5.29083133e-01 -7.26120889e-01
-1.77669629e-01 -4.47461307e-01 -1.06584851e-03 8.47827613e-01
6.95506096e-01 -9.73367929e-01 5.80719292e-01 8.69913340e-01
-3.76233459e-02 -1.02126765e+00 -1.28715527e+00 -5.35527110e-01
3.20943743e-01 -8.64221931e-01 2.97321141e-01 5.70923030e-01
1.20762430e-01 -5.49732149e-02 4.79378074e-01 -1.47603869e-01
5.36307812e-01 -4.58811283e-01 -1.75917625e-01 -1.18926251e+00
-3.58745188e-01 -1.13699400e+00 -6.41245723e-01 -8.81341994e-01
3.73489298e-02 -9.32384729e-01 3.49694192e-01 -1.38537252e+00
-7.34771192e-02 -7.94851303e-01 -8.04834366e-01 9.42250013e-01
2.01473475e-01 2.84591734e-01 -2.08997488e-01 2.25698501e-02
-1.35712102e-01 4.42700833e-01 6.20916903e-01 -2.50608772e-01
1.07685536e-01 -2.65811354e-01 -6.23104155e-01 1.08500087e+00
5.51488638e-01 -6.12848818e-01 -3.50263000e-01 -5.26255608e-01
6.16025031e-01 -1.08016580e-01 6.24334753e-01 -1.10650074e+00
5.70248663e-01 -1.63868219e-01 6.72708631e-01 -2.11793557e-01
8.24607849e-01 -4.97638404e-01 -1.13757744e-01 3.27669829e-01
-5.93554258e-01 2.46313438e-01 2.61961788e-01 3.81298542e-01
-9.52514762e-04 -6.20564699e-01 6.66333258e-01 -3.62695187e-01
-4.50637549e-01 3.21420670e-01 -7.14612901e-01 7.08948150e-02
9.17747200e-01 -4.65047181e-01 -2.10529998e-01 -9.43863392e-02
-7.34711230e-01 -5.78970425e-02 2.34469563e-01 -8.17079768e-02
8.50977302e-01 -1.08610678e+00 -3.18681896e-01 2.92685866e-01
4.24235947e-02 2.76881576e-01 -3.30920927e-02 6.17710054e-01
-4.43469048e-01 3.10234547e-01 -3.41484994e-01 -4.93298441e-01
-4.37479317e-01 3.52704823e-01 3.44936788e-01 8.49528983e-02
1.17558070e-01 1.44441211e+00 1.92600176e-01 -7.72598311e-02
4.94774848e-01 -6.62071586e-01 1.95027202e-01 6.53554872e-02
5.06497145e-01 1.46580741e-01 -7.56889358e-02 -1.13521405e-01
-4.38106805e-01 1.26693487e-01 5.34684472e-02 -7.05299199e-01
1.09124541e+00 3.20328534e-01 -4.39561754e-01 9.55077231e-01
9.83858407e-01 -6.83825374e-01 -1.00586176e+00 -1.30092487e-01
-6.19521067e-02 1.44783244e-01 2.08612420e-02 -6.99630857e-01
-6.20850861e-01 1.34914410e+00 6.50649309e-01 1.75441369e-01
9.31460977e-01 2.24959385e-02 1.73671767e-01 6.81363881e-01
5.20510137e-01 -8.87313783e-01 -1.07817821e-01 8.15724015e-01
5.39689243e-01 -7.48879135e-01 -1.66995287e-01 5.85559607e-02
-1.10501386e-01 1.12181520e+00 4.63398844e-01 -2.78877050e-01
8.37735415e-01 6.38406992e-01 -4.47508484e-01 -2.09876806e-01
-7.15696454e-01 -1.79190010e-01 2.18993165e-02 2.98648238e-01
5.47066212e-01 2.43355911e-02 -2.12977558e-01 7.70956039e-01
-7.20229670e-02 7.37768412e-02 2.41429955e-01 1.26742566e+00
-6.60309196e-01 -6.69334292e-01 -1.72129065e-01 8.64232302e-01
-4.58809197e-01 -6.11548245e-01 1.63887948e-01 3.41874450e-01
2.80676544e-01 6.22396350e-01 3.36782932e-01 -1.10814795e-01
2.20000967e-02 4.80793744e-01 7.16538131e-01 -6.53155446e-01
-5.13222516e-01 -2.71540642e-01 -3.12764585e-01 -7.70979896e-02
-1.23311058e-02 -4.90109712e-01 -1.83634233e+00 -2.29344353e-01
-2.14859918e-01 -8.30431655e-02 8.87968540e-01 9.35637236e-01
7.65342340e-02 6.51568234e-01 5.07313712e-03 -7.16821671e-01
-8.55952740e-01 -7.26788938e-01 -6.98248088e-01 1.50134876e-01
-2.42041573e-02 -8.22993338e-01 -2.37217650e-01 2.09584106e-02] | [8.423966407775879, 3.172776460647583] |
bf134129-bf45-44ba-a7fc-c73971436b5a | avtpnet-convolutional-autoencoder-for-avtp | 2202.00045 | null | https://arxiv.org/abs/2202.00045v2 | https://arxiv.org/pdf/2202.00045v2.pdf | Unsupervised Network Intrusion Detection System for AVTP in Automotive Ethernet Networks | Network Intrusion Detection Systems (NIDSs) are widely regarded as efficient tools for securing in-vehicle networks against diverse cyberattacks. However, since cyberattacks are always evolving, signature-based intrusion detection systems are no longer adopted. An alternative solution can be the deployment of deep learning based intrusion detection system which play an important role in detecting unknown attack patterns in network traffic. Hence, in this paper, we compare the performance of different unsupervised deep and machine learning based anomaly detection algorithms, for real-time detection of anomalies on the Audio Video Transport Protocol (AVTP), an application layer protocol implemented in the recent Automotive Ethernet based in-vehicle network. The numerical results, conducted on the recently published "Automotive Ethernet Intrusion Dataset", show that deep learning models significantly outperfom other state-of-the art traditional anomaly detection models in machine learning under different experimental settings. | ['Jean-Luc Danger', 'Hadi Ghauch', 'Maria Mushtaq', 'Natasha Alkhatib'] | 2022-01-31 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [-2.46454716e-01 -3.18898648e-01 -2.40030661e-02 -3.89015704e-01
6.17236085e-02 -2.97569185e-01 7.88303733e-01 4.68463600e-01
-4.02767181e-01 5.96397400e-01 -6.39600098e-01 -1.01215434e+00
-4.11423832e-01 -7.77390480e-01 -5.82944632e-01 -7.21994817e-01
-4.90612149e-01 6.61274970e-01 5.98217666e-01 -4.75850314e-01
5.18039465e-01 1.08958447e+00 -1.62531447e+00 1.56296059e-01
2.30997071e-01 1.40830672e+00 -9.28077936e-01 7.19002604e-01
-4.44999188e-01 8.02996337e-01 -9.22110379e-01 -4.86686751e-02
2.75190115e-01 1.23892561e-01 -6.25492036e-01 -2.20962659e-01
3.15447569e-01 -2.47571453e-01 -7.50591397e-01 1.14624906e+00
-2.75513083e-01 1.55428380e-01 5.82273245e-01 -2.02198458e+00
2.72066802e-01 1.84988379e-01 -3.23245287e-01 9.77511466e-01
2.12050229e-01 3.09493929e-01 3.89227957e-01 -3.71729821e-01
3.23368907e-01 1.19906771e+00 2.48034567e-01 3.79323214e-01
-9.21755135e-01 -1.14558673e+00 7.42866769e-02 1.18235087e+00
-9.21845198e-01 -2.29877859e-01 8.71822238e-01 -3.33697498e-01
9.35964525e-01 1.62472367e-01 4.75420058e-01 1.37464857e+00
1.05135989e+00 4.19210285e-01 7.20175326e-01 -1.61925450e-01
4.33811575e-01 -1.84871504e-04 6.95203781e-01 3.61301720e-01
4.23634946e-01 5.04195154e-01 3.72986346e-02 -4.23379451e-01
5.26941009e-02 1.22586772e-01 3.97562772e-01 -5.49462531e-03
-3.99096042e-01 1.05756712e+00 -4.01978977e-02 4.09904659e-01
-7.17879057e-01 -1.57333873e-02 1.25079370e+00 1.24304080e+00
3.97321463e-01 3.26832026e-01 -6.72514439e-01 -3.04270148e-01
-3.13017279e-01 1.56651855e-01 9.86630201e-01 3.85864854e-01
4.26134825e-01 9.02994812e-01 6.09489620e-01 2.51786768e-01
5.36731780e-01 1.76078618e-01 3.92101943e-01 -6.00878298e-01
-2.24995643e-01 5.86932898e-01 -3.42412323e-01 -9.80636060e-01
-5.01826227e-01 -3.62899825e-02 -8.62193167e-01 9.66119766e-01
3.36417347e-01 -3.35007757e-01 -9.14777756e-01 1.05367839e+00
4.22268629e-01 7.25614071e-01 1.34964466e-01 3.28413129e-01
4.53950167e-02 6.41491294e-01 3.69132310e-01 -1.39252007e-01
9.73785579e-01 -2.97060847e-01 -1.00045824e+00 2.59773284e-01
5.91539025e-01 -7.93323040e-01 2.26557940e-01 8.42527092e-01
-2.69816905e-01 -5.12115538e-01 -1.47965193e+00 9.50586200e-01
-1.06649458e+00 -9.88373339e-01 6.24489903e-01 1.04934466e+00
-6.01385593e-01 4.46655869e-01 -8.06100905e-01 -2.20345095e-01
3.27444702e-01 6.93049669e-01 -5.35746098e-01 1.57405153e-01
-1.25339651e+00 8.92369449e-01 6.56853855e-01 1.75415259e-02
-1.11341047e+00 -4.90221083e-01 -6.94525301e-01 -2.31085569e-01
6.38246536e-01 2.42590427e-01 1.26534176e+00 -1.04126012e+00
-1.42961347e+00 4.48609382e-01 3.72744113e-01 -1.07728076e+00
2.13748455e-01 -2.38879427e-01 -1.45537317e+00 9.57508162e-02
-5.07557034e-01 -2.22760707e-01 1.06366205e+00 -8.34209561e-01
-7.85521507e-01 -3.18693459e-01 -5.05334027e-02 -1.12044811e+00
-1.13070510e-01 4.63379264e-01 8.17372501e-01 -2.74228007e-01
-2.49370828e-01 -8.46280873e-01 -1.27442241e-01 -3.84298444e-01
-2.34365717e-01 -7.17426658e-01 2.03325176e+00 -3.65851760e-01
9.33409929e-01 -2.19927883e+00 -5.43616831e-01 9.41811740e-01
-2.80614570e-02 9.05196309e-01 -4.56263777e-03 3.41484368e-01
-5.76654255e-01 -1.21417053e-01 1.09432228e-01 2.02160016e-01
-1.13482110e-01 6.37467980e-01 -4.12203640e-01 6.26191318e-01
2.63544261e-01 1.72997996e-01 -4.89768565e-01 -1.44770443e-01
6.83778644e-01 -9.60113853e-02 -1.28771663e-01 1.29281029e-01
-4.58941519e-01 3.96298707e-01 -6.36294961e-01 7.07561731e-01
6.04477704e-01 4.86671805e-01 -6.07370958e-02 1.92130953e-01
-1.66158810e-01 -1.08530797e-01 -1.04416919e+00 6.99243724e-01
-3.10311671e-02 1.13799393e+00 -8.57149959e-02 -1.65475392e+00
1.01570249e+00 6.47910297e-01 7.67518103e-01 -7.80973971e-01
6.37645364e-01 2.30992824e-01 8.35134804e-01 -6.37963355e-01
-1.08384982e-01 1.88652933e-01 6.99766562e-04 5.30384600e-01
2.20772952e-01 6.62901461e-01 2.63997853e-01 1.22849056e-02
1.49924886e+00 -5.51173687e-01 7.36771077e-02 -1.45212039e-01
1.17940962e+00 -9.18646902e-02 4.17884737e-01 5.47142804e-01
-5.40556848e-01 -3.86718571e-01 9.07023668e-01 -1.04779291e+00
-1.03711748e+00 -1.08902264e+00 -1.49582312e-01 8.07463050e-01
-4.91117924e-01 -9.49689001e-03 -6.41631067e-01 -1.10120666e+00
1.34592548e-01 9.04006600e-01 -3.22291821e-01 -6.04760230e-01
-8.37465584e-01 -4.70373869e-01 9.44006443e-01 2.93259710e-01
4.33592409e-01 -1.23582923e+00 -2.05357373e-01 5.26465595e-01
6.33122206e-01 -1.47875082e+00 4.89660889e-01 2.01771319e-01
-6.17214322e-01 -1.65962994e+00 4.75775391e-01 -3.41233730e-01
1.88187771e-02 -1.81261808e-01 5.82603812e-01 1.58273369e-01
-5.30881047e-01 5.00746131e-01 -3.37744504e-01 -9.59671259e-01
-1.10808492e+00 -7.26711899e-02 6.93924665e-01 3.94652665e-01
1.08651352e+00 -7.51480162e-01 -1.22047819e-01 2.88269758e-01
-1.08642435e+00 -1.29931140e+00 5.14851570e-01 2.75585264e-01
8.30956176e-02 5.35583556e-01 9.68949556e-01 -8.16627443e-01
5.17297566e-01 -9.74994957e-01 -1.02433455e+00 -2.53992528e-01
-7.51374662e-01 -2.58148834e-02 8.95008802e-01 -5.05443335e-01
-6.92272961e-01 -5.56458235e-01 -6.35267735e-01 -6.51317716e-01
-1.09937763e+00 2.55796790e-01 -2.58004576e-01 -4.56373453e-01
4.82566833e-01 1.73245911e-02 4.81822014e-01 -4.10359979e-01
-3.43826979e-01 6.73784792e-01 3.92480105e-01 -3.85876805e-01
9.72341299e-01 2.48498172e-01 3.59445304e-01 -1.43883562e+00
-1.32795036e-01 -5.77253163e-01 -3.94149184e-01 -4.42446858e-01
7.07243443e-01 -2.79332310e-01 -8.78480554e-01 9.09555495e-01
-1.19088864e+00 1.35395586e-01 -8.08458775e-02 5.68248749e-01
-2.68666059e-01 4.43086296e-01 -6.19247079e-01 -8.48729849e-01
-7.77456388e-02 -9.60858047e-01 2.93913066e-01 7.56230652e-02
-2.91300863e-01 -8.87872756e-01 3.09148967e-01 6.87201694e-02
6.46397769e-01 4.82819647e-01 1.31954992e+00 -1.87130535e+00
-4.54051018e-01 -8.59947622e-01 4.93445434e-03 8.81130755e-01
-4.81233597e-02 5.45883536e-01 -8.07431936e-01 -1.53301671e-01
2.20351875e-01 6.18593059e-02 2.10761428e-01 6.99738180e-03
1.29384947e+00 -7.29844347e-02 -1.79934412e-01 3.62913400e-01
1.24158609e+00 9.65737700e-01 9.49943125e-01 7.77804434e-01
3.81445289e-01 4.52797532e-01 6.31902277e-01 3.39658678e-01
-4.84468490e-01 3.85074049e-01 8.59122515e-01 3.74524921e-01
5.89124501e-01 5.51680803e-01 4.89246398e-01 4.34720337e-01
-4.94040437e-02 -2.44372889e-01 -9.42650974e-01 -2.03940228e-01
-1.57098675e+00 -1.12744820e+00 -3.73123288e-01 1.91894531e+00
-1.57904580e-01 9.23235178e-01 4.72906470e-01 8.36562753e-01
4.32681680e-01 7.40140751e-02 -4.27252442e-01 -1.10258496e+00
1.46191418e-01 7.36928940e-01 3.62277597e-01 1.67363867e-01
-1.35062242e+00 7.18411624e-01 5.46055412e+00 5.40929615e-01
-1.19171262e+00 8.46368372e-02 3.10489178e-01 2.36710966e-01
6.23721659e-01 -3.29625219e-01 -4.76860136e-01 5.43879449e-01
1.95490742e+00 1.02616996e-01 -1.89539313e-01 1.04866314e+00
3.45540851e-01 1.98232949e-01 -8.81087899e-01 4.45254296e-01
-1.72428384e-01 -1.12428272e+00 1.74210832e-01 2.29027659e-01
-2.77613308e-02 2.24951684e-01 -9.56123844e-02 5.25930941e-01
7.16775507e-02 -8.04032087e-01 5.48013449e-02 4.16033447e-01
-8.41752589e-02 -1.29875553e+00 1.34416187e+00 -9.78160352e-02
-8.37308586e-01 -4.08876538e-01 -1.09375447e-01 1.17974058e-01
1.11344174e-01 9.56037939e-02 -5.41596413e-01 4.52896386e-01
7.08088219e-01 3.03909868e-01 -4.01285052e-01 1.12714982e+00
1.79475725e-01 9.13838267e-01 -3.08197588e-01 2.21256409e-02
7.48521984e-01 -9.95190740e-02 9.23726380e-01 1.00613785e+00
-6.63393661e-02 -1.95799366e-01 2.81871140e-01 2.00065702e-01
4.33284461e-01 -2.68448353e-01 -1.11494279e+00 -2.15421781e-01
2.63539851e-01 9.99520957e-01 -5.77511489e-01 -1.15881599e-01
-5.53392172e-01 1.80268452e-01 -5.01757503e-01 4.24841285e-01
-9.38126147e-01 -6.85972810e-01 1.53349531e+00 3.61159563e-01
2.60001719e-01 -4.50325012e-01 2.56692946e-01 -4.88921165e-01
-3.47582936e-01 -1.24889660e+00 7.51080990e-01 -8.41574967e-02
-1.26709461e+00 5.37415147e-01 2.18968317e-01 -1.17463315e+00
-5.16011000e-01 -1.27757704e+00 -1.12565279e+00 4.61136289e-02
-1.20890820e+00 -7.23873198e-01 1.40868187e-01 1.05076134e+00
8.43406916e-01 -1.16369474e+00 7.35201538e-01 6.06766880e-01
-6.94790363e-01 6.57567501e-01 5.96272238e-02 4.43170726e-01
1.91102788e-01 -8.50966573e-01 2.76193470e-01 9.25204337e-01
5.30494936e-02 1.40049279e-01 9.11190271e-01 -5.21945596e-01
-1.36064339e+00 -8.85524631e-01 2.79647186e-02 -6.50906935e-02
1.30935562e+00 2.39800233e-02 -1.40245831e+00 8.54702532e-01
4.57374394e-01 3.92297506e-02 8.13386142e-01 -5.74689806e-02
-3.98227274e-01 -4.13135260e-01 -1.18301070e+00 4.44680214e-01
1.69427201e-01 -3.78217280e-01 -5.49828589e-01 2.27295563e-01
6.60553157e-01 1.75739348e-01 -7.01809943e-01 5.40355980e-01
3.01789731e-01 -9.54011977e-01 1.01840246e+00 -1.50272357e+00
-3.58515888e-01 -2.09597677e-01 -2.38682270e-01 -8.38815212e-01
-3.38200778e-02 -8.21499765e-01 -5.94718754e-01 8.67185533e-01
1.00056231e-01 -1.32540965e+00 8.27162683e-01 8.62207711e-02
-2.69213587e-01 -4.72689033e-01 -1.57740498e+00 -8.22702289e-01
-1.87437877e-01 -9.07566547e-01 6.21594667e-01 8.77367079e-01
-4.83979851e-01 3.37843411e-02 4.83642109e-02 4.78777081e-01
7.42259502e-01 -6.91750169e-01 7.91982472e-01 -1.71199155e+00
2.30868936e-01 -5.63305914e-01 -1.60463297e+00 1.98256850e-01
7.03076720e-01 -2.21159905e-01 -4.03366357e-01 -2.26350293e-01
-9.17047203e-01 -1.62515104e-01 -9.81007278e-01 1.48356989e-01
6.67395532e-01 -1.82722479e-01 -3.78216505e-01 -3.07084948e-01
-3.60837579e-01 2.11323962e-01 9.76693407e-02 -2.89572001e-01
2.99378008e-01 5.46694279e-01 2.00641543e-01 1.23722243e+00
1.11666536e+00 -6.51687145e-01 -3.49530369e-01 3.95150602e-01
-1.82211280e-01 -1.65891498e-01 4.40126359e-01 -1.43751240e+00
1.58060595e-01 -1.88203752e-01 1.24702409e-01 -6.35532200e-01
-3.40948328e-02 -1.31208730e+00 -2.11308986e-01 7.83803165e-01
4.66551296e-02 7.06210136e-01 5.13244390e-01 8.70304406e-01
-3.76501352e-01 -1.91520870e-01 9.07163024e-01 2.00840473e-01
-1.12970412e+00 6.03051424e-01 -1.37069654e+00 -4.50045049e-01
1.46593273e+00 -2.52080739e-01 -3.05956304e-01 -2.38676414e-01
-7.55567074e-01 -2.27560140e-02 -4.25223187e-02 7.87536263e-01
8.82794857e-01 -1.14998651e+00 -3.44344586e-01 6.91293538e-01
3.41544524e-02 -6.44116580e-01 7.57770687e-02 6.43422306e-01
-7.05916524e-01 5.03747940e-01 -7.25636184e-01 -4.98495430e-01
-1.41147792e+00 7.81497061e-01 3.22454870e-01 -7.55303875e-02
-5.09993672e-01 1.30862221e-01 -6.81625605e-01 -6.60438091e-02
3.86219591e-01 4.82093282e-02 -3.14319909e-01 -2.34662339e-01
6.75963998e-01 7.45499611e-01 3.95562381e-01 -5.15806496e-01
-3.57617915e-01 -7.74147660e-02 -6.29856467e-01 2.80900598e-01
9.77273822e-01 3.63355011e-01 -1.99075550e-01 6.01492822e-01
1.24774086e+00 -6.67565644e-01 -3.98439229e-01 -6.34722114e-02
8.26872408e-01 -2.55467802e-01 -4.26719747e-02 -2.21661597e-01
-1.10741413e+00 7.64670730e-01 1.12474477e+00 8.75000536e-01
8.97826791e-01 -3.45832109e-01 9.94854867e-01 9.50570762e-01
2.86547065e-01 -1.07060814e+00 2.24167883e-01 8.14267695e-01
2.81728536e-01 -1.25668073e+00 -3.00206900e-01 -3.80649529e-02
-6.24490380e-02 1.53522015e+00 9.48507249e-01 -5.10860920e-01
1.09763718e+00 5.39575875e-01 1.16456345e-01 -5.02738833e-01
-7.77993858e-01 2.94719577e-01 -2.18250994e-02 7.37374365e-01
5.05869798e-02 -1.83589503e-01 1.87707990e-01 -1.65225372e-01
1.61780372e-01 -4.56452519e-01 5.12332380e-01 9.89568710e-01
-6.54026151e-01 -1.24250281e+00 -5.18746018e-01 6.48411989e-01
-7.23789752e-01 5.38375080e-01 -2.16992661e-01 1.38390732e+00
-8.14817548e-02 1.03218400e+00 3.73089939e-01 -6.30126238e-01
3.46892416e-01 5.93676269e-01 4.16837335e-02 -4.29208800e-02
-2.68181950e-01 -5.84762931e-01 1.28396228e-01 -8.61235142e-01
-5.96138164e-02 -5.09940386e-01 -1.06207204e+00 -6.85594082e-01
-6.87660724e-02 3.05340230e-01 8.17376971e-01 1.50667202e+00
1.56643406e-01 7.80471385e-01 8.14985871e-01 -4.56081599e-01
-6.30334854e-01 -8.91280770e-01 -5.76735556e-01 2.85655558e-01
5.75077772e-01 -9.85201716e-01 -5.56742787e-01 -6.20232582e-01] | [5.259235858917236, 7.279350280761719] |
3f9fb7aa-1061-4991-aaa8-cbf8c37098a4 | ischemic-stroke-lesion-segmentation-using | 2204.04993 | null | https://arxiv.org/abs/2204.04993v1 | https://arxiv.org/pdf/2204.04993v1.pdf | Ischemic Stroke Lesion Segmentation Using Adversarial Learning | Ischemic stroke occurs through a blockage of clogged blood vessels supplying blood to the brain. Segmentation of the stroke lesion is vital to improve diagnosis, outcome assessment and treatment planning. In this work, we propose a segmentation model with adversarial learning for ischemic lesion segmentation. We adopt U-Net with skip connection and dropout as segmentation baseline network and a fully connected network (FCN) as discriminator network. Discriminator network consists of 5 convolution layers followed by leaky-ReLU and an upsampling layer to rescale the output to the size of the input map. Training a segmentation network along with an adversarial network can detect and correct higher order inconsistencies between the segmentation maps produced by ground-truth and the Segmentor. We exploit three modalities (CT, DPWI, CBF) of acute computed tomography (CT) perfusion data provided in ISLES 2018 (Ischemic Stroke Lesion Segmentation) for ischemic lesion segmentation. Our model has achieved dice accuracy of 42.10% with the cross-validation of training and 39% with the testing data. | ['Hongliang Ren', 'V Jeya Maria Jose', 'N Rajiv Vaidyanathan', 'Mobarakol Islam'] | 2022-04-11 | null | null | null | null | ['ischemic-stroke-lesion-segmentation'] | ['medical'] | [ 3.24366689e-01 3.05767000e-01 -1.43295810e-01 -5.90127468e-01
-7.85382092e-01 -6.92793489e-01 2.03683242e-01 -1.74287006e-01
-7.05057621e-01 9.65194106e-01 3.47056836e-01 -5.65199852e-01
3.83068949e-01 -1.03695560e+00 -7.45206892e-01 -6.36096895e-01
-2.29686216e-01 5.51171780e-01 5.42452812e-01 3.28471214e-01
-3.46620567e-03 8.92813921e-01 -3.79448742e-01 6.94341898e-01
1.10517478e+00 8.75609100e-01 -2.89373815e-01 1.03998125e+00
8.34168214e-03 1.07350373e+00 -5.54534018e-01 -1.44788161e-01
5.87003767e-01 -7.44810283e-01 -1.17355084e+00 -4.46185917e-01
4.09032792e-01 -8.71856332e-01 -9.48304653e-01 9.78709161e-01
8.35307837e-01 -3.00743394e-02 9.09882605e-01 -1.11790693e+00
-2.88826048e-01 7.82496631e-01 -8.39476883e-02 7.90158749e-01
-3.81645203e-01 4.50751901e-01 1.71446249e-01 -3.78196359e-01
9.10847664e-01 9.26899195e-01 7.60268033e-01 6.43189967e-01
-1.06245470e+00 -5.61704397e-01 -3.10918123e-01 2.72599787e-01
-1.04985058e+00 -2.57044029e-03 1.63042471e-01 -6.30332649e-01
7.25027204e-01 2.14987159e-01 7.76863575e-01 1.21293974e+00
5.21929979e-01 6.38992012e-01 1.06039870e+00 8.61517563e-02
2.79882759e-01 -3.28295380e-01 2.82268137e-01 3.89712214e-01
1.79592475e-01 2.68160969e-01 3.85393679e-01 7.33853951e-02
1.14636457e+00 1.80906802e-01 -5.21556020e-01 5.66686243e-02
-9.02983308e-01 7.71786571e-01 1.23332644e+00 2.11583197e-01
-5.84587693e-01 2.63138503e-01 6.60664260e-01 2.83431292e-01
2.98143446e-01 2.05179930e-01 -2.71582931e-01 1.33916602e-01
-1.21987283e+00 -3.78387012e-02 5.32749474e-01 5.47582567e-01
6.77285045e-02 1.55978292e-01 -6.89904988e-01 4.76629078e-01
-1.58857271e-01 5.54136395e-01 8.65518093e-01 -1.07201648e+00
4.00074929e-01 5.34682691e-01 -1.98980078e-01 -4.03614610e-01
-6.94064140e-01 -5.11979163e-01 -1.13041449e+00 8.03721547e-01
9.50845957e-01 -7.84771025e-01 -1.27830708e+00 1.32974505e+00
-8.70689154e-02 3.74032676e-01 -7.56540895e-03 1.26888549e+00
8.79520476e-01 2.01134831e-01 2.34669954e-01 1.73378825e-01
7.94293523e-01 -1.07800579e+00 -2.99291342e-01 -3.79126035e-02
8.12259018e-01 -3.30940813e-01 7.42459714e-01 -3.15629914e-02
-1.36268067e+00 -1.34316370e-01 -8.70766997e-01 -1.69379354e-01
-3.29688013e-01 -1.15168750e-01 1.33154660e-01 6.08366609e-01
-9.97004032e-01 9.46369410e-01 -1.06777155e+00 -1.70453593e-01
1.30033612e+00 2.57319719e-01 -4.17006284e-01 -2.90028863e-02
-1.15935707e+00 1.19769311e+00 3.75329435e-01 6.30838200e-02
-1.05401421e+00 -1.08590555e+00 -3.75851005e-01 -8.10078159e-02
-2.68462658e-01 -8.48211884e-01 8.28579426e-01 -1.23682427e+00
-1.42411292e+00 8.66730154e-01 2.81345665e-01 -1.20968425e+00
1.49771035e+00 -9.88822728e-02 1.00955576e-01 7.79048324e-01
1.96344964e-02 7.56412148e-01 3.33143294e-01 -7.57612407e-01
-3.05486619e-01 -3.17142189e-01 -2.84865677e-01 -2.41614450e-02
4.73021746e-01 -8.42051804e-02 4.77666140e-01 -6.37464106e-01
3.82020980e-01 -6.22211754e-01 -3.64526719e-01 3.03778023e-01
-6.80653870e-01 5.64506590e-01 7.46895790e-01 -1.28803694e+00
7.34473825e-01 -1.75978792e+00 -1.69192374e-01 4.00158495e-01
3.07502031e-01 4.11524206e-01 -1.36524081e-01 -3.37037802e-01
-4.23291177e-01 2.22154826e-01 -8.89970958e-01 1.18491709e-01
-5.04238904e-01 2.59346217e-01 -2.77742416e-01 8.81515384e-01
8.44331607e-02 1.47458172e+00 -8.14209580e-01 -4.12253052e-01
2.91574657e-01 5.84882319e-01 -4.82876539e-01 2.82819450e-01
2.01874718e-01 1.03601742e+00 -4.85831857e-01 2.74551541e-01
8.14986348e-01 2.89055228e-01 -3.86764228e-01 -1.44129358e-02
6.77804346e-04 -1.23127833e-01 -7.24895358e-01 1.53592753e+00
-2.29525324e-02 6.61856890e-01 -6.90422505e-02 -1.20437026e+00
7.03034580e-01 3.47529352e-01 6.86347306e-01 -6.05618596e-01
5.98108888e-01 2.88964897e-01 4.01847124e-01 -7.37596929e-01
-4.04594719e-01 -2.20103726e-01 3.59876156e-01 6.03728056e-01
-9.90075096e-02 -1.64509580e-01 -8.98084790e-02 2.13001460e-01
1.36449647e+00 -1.12164408e-01 -2.48927221e-01 -2.70166397e-01
3.47669840e-01 1.84801280e-01 3.88210714e-01 9.14747715e-01
-9.11965251e-01 1.19379938e+00 1.02171409e+00 -5.39369762e-01
-1.29625678e+00 -1.35144687e+00 -5.15298069e-01 3.35438490e-01
-1.98907658e-01 6.89980209e-01 -1.36374021e+00 -7.29172528e-01
-1.36408776e-01 6.48610532e-01 -9.25472260e-01 -3.45590979e-01
-9.84923542e-01 -9.67225254e-01 1.26744342e+00 9.58389044e-01
1.07997251e+00 -1.20095313e+00 -8.72503340e-01 2.45013475e-01
-1.76392585e-01 -8.93142581e-01 -3.85295212e-01 1.36713192e-01
-1.19350648e+00 -1.64342427e+00 -1.23571384e+00 -7.48830676e-01
8.22220802e-01 -3.96265328e-01 8.49958599e-01 1.58935875e-01
-4.92310554e-01 -2.29548022e-01 -1.67310223e-01 -4.00789417e-02
-5.71049035e-01 8.88925418e-02 -5.46218872e-01 -1.94745079e-01
-3.85761410e-02 -7.22981751e-01 -1.17834353e+00 5.14358170e-02
-9.95982409e-01 -1.37051657e-01 4.88301665e-01 7.20946908e-01
6.08645439e-01 -5.96755862e-01 6.17555141e-01 -1.01500225e+00
3.56068909e-01 -6.04993641e-01 -2.98825949e-01 1.53276017e-02
-2.66109288e-01 -3.55553240e-01 6.08024895e-01 -2.90652901e-01
-7.09847331e-01 -2.67320257e-02 -1.85373813e-01 -2.90321976e-01
-3.34427625e-01 9.34281945e-02 5.34774885e-02 -9.17007104e-02
1.09029138e+00 -9.37387720e-02 8.06934610e-02 -1.78532720e-01
4.82522994e-01 5.20571291e-01 1.12222934e+00 -1.21231720e-01
2.37683579e-01 7.18852639e-01 2.71858424e-01 -2.14866117e-01
-5.85160017e-01 4.64881994e-02 -1.24112988e+00 -2.18347266e-01
1.24306071e+00 -4.04077470e-01 -8.95327702e-02 6.48901105e-01
-1.13935006e+00 -1.03815114e+00 -6.53471768e-01 6.80561006e-01
-4.68235016e-01 7.59855285e-02 -9.45287466e-01 4.19298895e-02
-7.39102900e-01 -1.34853077e+00 2.28012711e-01 2.43513197e-01
1.43363774e-01 -9.74075437e-01 -5.53873368e-02 2.71685533e-02
7.57152915e-01 1.10731840e+00 9.60409045e-01 -7.87277639e-01
-3.20975065e-01 -5.28791010e-01 -6.77376151e-01 6.93991482e-01
-1.57863364e-01 -1.54317826e-01 -9.39335942e-01 -5.61538674e-02
-1.13271065e-01 -1.68366790e-01 1.02526522e+00 1.22798038e+00
1.16734362e+00 -9.17310417e-02 -9.55031067e-02 1.10067427e+00
1.22041368e+00 1.55825406e-01 9.40544009e-01 2.68732190e-01
5.73229909e-01 2.88983047e-01 -4.36470002e-01 -7.97743648e-02
-8.73670797e-04 -7.49199092e-02 4.18650925e-01 -4.15315777e-01
-9.01519835e-01 1.12648405e-01 -1.26891047e-01 7.27857798e-02
-3.66473012e-02 -1.58169508e-01 -1.08872080e+00 3.90789568e-01
-1.48487592e+00 -9.82470691e-01 -7.47575879e-01 2.11711907e+00
6.86868191e-01 2.14048818e-01 1.31754473e-01 -5.73476404e-02
9.31522489e-01 -3.51310194e-01 -9.48453665e-01 -4.76275533e-01
-3.89146954e-01 4.96630073e-01 8.83117616e-01 8.85519564e-01
-1.18881607e+00 9.02339578e-01 6.48248291e+00 2.40190536e-01
-1.45322144e+00 4.51778561e-01 1.17617583e+00 -1.49282236e-02
1.74936906e-01 -2.16154635e-01 1.92843169e-01 5.16664624e-01
9.17302012e-01 1.08041428e-01 2.07948714e-01 4.37596768e-01
4.56486613e-01 -2.24491879e-01 -9.10841525e-01 2.46746704e-01
-3.06186318e-01 -1.23056436e+00 -7.44227841e-02 -4.45143789e-01
1.00095022e+00 6.67477250e-01 -1.33148834e-01 3.23796421e-02
2.06384569e-01 -1.40988851e+00 4.62948561e-01 8.17121983e-01
9.90915537e-01 -5.49088299e-01 1.08127201e+00 1.98739424e-01
-5.11704743e-01 2.46246949e-01 -2.62525707e-01 1.21386185e-01
3.19992423e-01 3.52153063e-01 -7.09249437e-01 1.20316185e-01
4.96814340e-01 5.22414804e-01 -4.75081772e-01 1.47786272e+00
-5.07474959e-01 9.24327374e-01 -3.17579389e-01 7.07697034e-01
4.81889755e-01 -2.86397219e-01 6.64739370e-01 1.30280924e+00
-9.34736878e-02 3.10420424e-01 1.02340624e-01 1.10645461e+00
-2.63185173e-01 3.46551389e-02 -2.72000849e-01 6.79560602e-01
1.40931889e-01 7.73178220e-01 -1.12629795e+00 -7.87280381e-01
-6.39864579e-02 1.07653272e+00 7.88963959e-02 6.04119837e-01
-7.06337392e-01 -4.49216723e-01 1.16479225e-01 1.30266115e-01
-1.41496211e-01 1.23533383e-01 -1.19883025e+00 -1.02733052e+00
-2.80954570e-01 -2.45973498e-01 4.40544486e-01 -6.89051449e-01
-1.06556547e+00 7.84473419e-01 -1.52439952e-01 -7.55703032e-01
5.39506376e-02 -4.37859923e-01 -1.24988484e+00 1.40518105e+00
-1.61417019e+00 -7.93869317e-01 -5.23394942e-01 5.98512232e-01
1.53365970e-01 -1.08273722e-01 6.96105480e-01 3.81859064e-01
-6.93574607e-01 3.68648320e-01 5.59962280e-02 8.31058443e-01
6.61540389e-01 -1.34999895e+00 3.13418299e-01 1.23425066e+00
-6.63810253e-01 2.25878228e-02 4.11620796e-01 -8.85982871e-01
-2.43132859e-01 -1.54329383e+00 5.61234117e-01 -3.12762111e-01
6.18251503e-01 2.88882136e-01 -1.05243766e+00 9.67490792e-01
1.05110563e-01 6.86825871e-01 3.58088970e-01 -1.18345344e+00
-9.91635025e-02 3.41686189e-01 -1.82567036e+00 3.59316111e-01
7.13675678e-01 -3.15499574e-01 -6.70217037e-01 5.04948258e-01
3.89863074e-01 -8.58727932e-01 -9.25994456e-01 2.85653412e-01
6.06278539e-01 -7.31172442e-01 9.95704234e-01 -1.07666099e+00
7.55554914e-01 1.28773838e-01 2.64835656e-01 -1.27446210e+00
-1.67731177e-02 -2.47671336e-01 3.28688562e-01 4.79903162e-01
3.85266751e-01 -7.71029949e-01 9.02218342e-01 9.37284350e-01
-6.09889030e-01 -6.22243464e-01 -1.04307079e+00 -4.35578674e-01
1.25155020e+00 -2.95178384e-01 4.36365128e-01 7.26503670e-01
-3.61257613e-01 -3.13312292e-01 2.45378911e-01 1.17669672e-01
6.33600295e-01 -2.00189739e-01 9.89483222e-02 -9.19486284e-01
3.01382989e-01 -8.27297628e-01 -4.39731240e-01 -4.93111163e-01
1.91958800e-01 -1.57558513e+00 -2.12743297e-01 -1.74345076e+00
7.90536702e-02 -4.14132535e-01 -3.49352121e-01 5.78858495e-01
6.34720847e-02 5.27378738e-01 6.42537419e-03 1.90195113e-01
2.44025156e-01 8.07608739e-02 1.88287890e+00 -2.57606298e-01
-3.38080615e-01 1.71554282e-01 -1.19597405e-01 8.21624160e-01
1.12480390e+00 -6.86528563e-01 -1.07118927e-01 -5.76072276e-01
-6.91075504e-01 2.10552856e-01 8.71980727e-01 -9.88039434e-01
1.86312139e-01 2.43163049e-01 7.78675079e-01 -3.00417900e-01
-4.29797143e-01 -7.21668541e-01 -2.66485691e-01 1.15204453e+00
-6.96333528e-01 -1.47238016e-01 1.07714884e-01 -4.90750838e-03
9.73710865e-02 -4.17743832e-01 1.23913491e+00 -2.40902022e-01
-1.38712496e-01 6.80556655e-01 -6.34065092e-01 5.85863471e-01
9.99480844e-01 -1.42620709e-02 -5.37600875e-01 -6.39450848e-02
-1.29537642e+00 1.45625427e-01 6.71574920e-02 -1.45395488e-01
5.90048313e-01 -1.04459608e+00 -1.05961061e+00 1.61800712e-01
-6.22361660e-01 1.73177928e-01 2.66473830e-01 1.21297729e+00
-1.40266299e+00 1.93469241e-01 -8.39076936e-01 -3.51569414e-01
-6.44267678e-01 -4.06910330e-02 1.20264566e+00 -1.84571102e-01
-1.26271653e+00 8.55074048e-01 -2.58718342e-01 -1.98971540e-01
2.34694466e-01 -7.65063465e-01 -1.41551509e-01 -1.36809468e-01
5.61621487e-01 6.86349690e-01 8.75178650e-02 -6.59610271e-01
-1.71589658e-01 1.00797959e-01 2.25071818e-01 -1.71749100e-01
1.18345964e+00 2.31778756e-01 -1.81569815e-01 -1.29807174e-01
1.41561830e+00 -8.47858787e-01 -1.47124732e+00 6.48697987e-02
-3.02176923e-01 -1.45820305e-01 6.16606355e-01 -1.35880411e+00
-1.67355180e+00 1.13156462e+00 1.13714683e+00 -2.61461064e-02
6.89753234e-01 -3.50534260e-01 1.22088909e+00 -2.57433578e-02
-1.79143041e-01 -6.74956441e-01 -3.43667656e-01 2.78827250e-01
8.79645169e-01 -1.08757949e+00 -3.44348341e-01 -5.50091006e-02
-4.51336116e-01 1.41566765e+00 4.30928677e-01 -8.81743968e-01
8.23690355e-01 4.22066092e-01 4.76138562e-01 4.73366231e-02
1.21537894e-01 -8.05572867e-02 8.99055973e-02 7.30972886e-01
2.01548919e-01 3.05795640e-01 -4.24182028e-01 5.56824744e-01
-1.51254967e-01 2.35243708e-01 6.48448169e-01 6.41307771e-01
-4.44197923e-01 -7.30098665e-01 -5.04687950e-02 7.92745054e-01
-6.55208647e-01 -1.94957301e-01 -2.32041538e-01 6.21290684e-01
2.03073069e-01 6.28037930e-01 1.92680955e-01 3.06622535e-01
4.57357973e-01 2.25815445e-01 4.31815177e-01 -2.42128551e-01
-1.07959461e+00 -2.28640065e-01 -3.92963797e-01 -6.88200176e-01
-1.69171765e-01 -6.58581913e-01 -1.75436568e+00 -1.59354091e-01
2.42360950e-01 -1.67812064e-01 5.30678153e-01 1.08496690e+00
-6.48504402e-03 9.91047502e-01 4.33385640e-01 -5.70093811e-01
-3.74072462e-01 -1.15987897e+00 -5.32311797e-01 4.25409555e-01
4.70919073e-01 -2.04067588e-01 -3.09045255e-01 2.58800209e-01] | [14.24228572845459, -2.0783257484436035] |
1c508d28-dc9b-40b6-934d-97860a2500b2 | viseret-a-simple-yet-effective-approach-to | 2110.05146 | null | https://arxiv.org/abs/2110.05146v2 | https://arxiv.org/pdf/2110.05146v2.pdf | ViSeRet: A simple yet effective approach to moment retrieval via fine-grained video segmentation | Video-text retrieval has many real-world applications such as media analytics, surveillance, and robotics. This paper presents the 1st place solution to the video retrieval track of the ICCV VALUE Challenge 2021. We present a simple yet effective approach to jointly tackle two video-text retrieval tasks (video retrieval and video corpus moment retrieval) by leveraging the model trained only on the video retrieval task. In addition, we create an ensemble model that achieves the new state-of-the-art performance on all four datasets (TVr, How2r, YouCook2r, and VATEXr) presented in the VALUE Challenge. | ['Minjoon Seo', 'Hanseok Oh', 'Aiden Seungjoon Lee'] | 2021-10-11 | null | null | null | null | ['moment-retrieval', 'video-text-retrieval'] | ['computer-vision', 'computer-vision'] | [-1.00337565e-01 -8.95182610e-01 -4.76420164e-01 1.49433557e-02
-1.17881906e+00 -4.94747132e-01 1.03132808e+00 -1.32847711e-01
-6.62491798e-01 1.50766939e-01 3.47811878e-01 8.24973825e-03
-2.43793070e-01 -1.98909909e-01 -5.09673178e-01 -3.56999874e-01
-3.99800390e-02 1.63468987e-01 5.31930089e-01 -3.38348866e-01
2.85146445e-01 -1.41523898e-01 -1.98096943e+00 6.46079659e-01
1.84697866e-01 1.49556124e+00 1.41720995e-01 1.06617796e+00
1.31277651e-01 1.22798967e+00 -3.84396315e-01 -4.54206228e-01
3.40734690e-01 1.78243741e-01 -6.65833235e-01 -3.85747135e-01
8.78827274e-01 -8.46779287e-01 -1.20937049e+00 7.26754904e-01
6.03348374e-01 4.58616614e-01 8.43991220e-01 -1.25871146e+00
-8.65097106e-01 3.00834775e-01 -6.49072111e-01 5.39502382e-01
6.57112658e-01 -1.17639802e-01 1.01129711e+00 -1.03872609e+00
1.18877494e+00 1.18438244e+00 2.19585419e-01 5.29830575e-01
-1.41213879e-01 -4.49058533e-01 1.02909580e-01 8.73707294e-01
-1.77778554e+00 -5.97334325e-01 2.11391494e-01 -4.59815174e-01
1.01764131e+00 3.12641591e-01 6.43311977e-01 1.17915070e+00
1.99955255e-01 1.52395868e+00 2.53793448e-01 -3.80840115e-02
-2.65672654e-02 -3.59782726e-01 2.34326795e-01 3.12632531e-01
-1.24119997e-01 -3.22855145e-01 -8.48211229e-01 -2.95651518e-02
2.16091171e-01 3.20940137e-01 -3.67597520e-01 6.53325170e-02
-1.32873559e+00 5.41459739e-01 -2.41365284e-02 2.46945918e-01
-3.19262981e-01 7.16737092e-01 9.27489340e-01 7.54565120e-01
7.26688385e-01 -6.61847144e-02 -2.05307737e-01 -4.69675183e-01
-1.33057666e+00 4.71097469e-01 5.09527087e-01 1.43119645e+00
2.43662968e-01 -1.66357294e-01 -5.36849678e-01 8.24912190e-01
6.69686675e-01 1.06032777e+00 6.30699575e-01 -8.91882241e-01
5.75237274e-01 -5.75576201e-02 7.13973418e-02 -1.08619106e+00
3.22430819e-01 2.39520386e-01 -3.79172415e-01 -7.42782235e-01
-4.38285649e-01 3.35475415e-01 -1.27717113e+00 9.67979789e-01
8.36885199e-02 6.04688346e-01 2.85650138e-02 1.04080689e+00
1.65070617e+00 1.22249198e+00 6.96688099e-03 -1.25834912e-01
1.23077047e+00 -1.32581520e+00 -9.42192733e-01 7.31161684e-02
6.04198277e-01 -7.95666039e-01 4.17103887e-01 4.40386653e-01
-1.06510139e+00 -3.53548497e-01 -1.08212042e+00 -3.89877945e-01
-6.28100932e-01 1.11273760e-02 3.93960625e-01 3.45540404e-01
-1.41507339e+00 1.40799105e-01 -4.12500560e-01 -5.78351676e-01
2.07034990e-01 -6.21324480e-02 -5.70730686e-01 -6.50315523e-01
-1.33738875e+00 7.46915877e-01 3.64581436e-01 1.09201983e-01
-1.31693435e+00 -5.04325807e-01 -6.14235342e-01 -5.89129180e-02
6.32390738e-01 -4.81101960e-01 1.19698370e+00 -6.21766746e-01
-1.25488627e+00 1.13015139e+00 -2.23858859e-02 -5.11118531e-01
3.95272315e-01 -6.69123232e-01 -6.99106812e-01 6.95533037e-01
-4.87476178e-02 8.05239916e-01 1.16450536e+00 -5.99066377e-01
-5.48271060e-01 -2.30692625e-01 -3.44600640e-02 2.79237926e-01
-5.57722628e-01 5.29432595e-01 -1.54279065e+00 -5.75100780e-01
-2.82461494e-01 -9.93913114e-01 3.05971861e-01 -2.17730310e-02
1.64003864e-01 -5.07870674e-01 1.28579152e+00 -8.21454644e-01
1.41203916e+00 -2.44935489e+00 4.32445079e-01 7.15665706e-03
3.51514876e-01 5.15207887e-01 -6.17958724e-01 6.95690155e-01
2.93418884e-01 3.53187144e-01 6.05756700e-01 -3.70046705e-01
2.78361533e-02 -2.15340361e-01 -5.84860623e-01 5.51360130e-01
-3.15559447e-01 1.10416305e+00 -8.28164935e-01 -6.74966693e-01
3.16383481e-01 5.49653053e-01 -2.66224235e-01 2.16401845e-01
-3.46582830e-01 -1.85744479e-01 -7.19708800e-01 9.67096090e-01
4.85100865e-01 -2.18939930e-01 3.80773470e-02 -1.98329359e-01
1.63918480e-01 -2.39518344e-01 -8.46813738e-01 1.94470155e+00
-8.04694518e-02 1.18289685e+00 -2.61127967e-02 -6.92148387e-01
4.53794211e-01 5.46576381e-01 9.46967900e-01 -1.07310748e+00
2.30628416e-01 1.24395415e-01 -8.30676675e-01 -9.04527128e-01
1.32553637e+00 6.88982546e-01 -1.41788229e-01 2.01061040e-01
4.20954823e-01 -4.72044460e-02 5.18917024e-01 7.10604846e-01
1.39679861e+00 1.29658461e-01 -5.83670102e-02 1.06623203e-01
5.80187380e-01 -2.42147729e-01 -1.50753155e-01 8.73477459e-01
-2.60972261e-01 8.64132822e-01 -1.42091410e-02 -4.47127253e-01
-9.06116784e-01 -9.27524686e-01 2.64076918e-01 1.33283997e+00
5.09363890e-01 -1.14514041e+00 -3.07549655e-01 -4.87973422e-01
-6.31104708e-02 1.44085422e-01 -5.22604644e-01 -2.14696765e-01
-4.22090948e-01 -2.79708743e-01 7.82873631e-01 1.24583468e-01
5.22115409e-01 -8.85053515e-01 -2.68180221e-01 -1.92543089e-01
-7.01755464e-01 -1.64575446e+00 -7.60983050e-01 -7.69528747e-01
-4.79098260e-01 -1.02028298e+00 -1.14575779e+00 -5.79559684e-01
-1.22951038e-01 1.11963284e+00 1.12347400e+00 5.80504835e-01
-4.44976807e-01 1.29305196e+00 -1.16938138e+00 -1.79300800e-01
1.00851148e-01 8.54226276e-02 -9.98090357e-02 -5.11838822e-03
4.27358627e-01 2.25850761e-01 -5.07332742e-01 2.53136992e-01
-1.29273748e+00 -5.57159960e-01 6.34552687e-02 4.35833991e-01
6.13755345e-01 -2.00741187e-01 2.61140436e-01 -2.54036486e-01
5.45920074e-01 -8.34493995e-01 -5.12123942e-01 6.96638823e-01
-3.08613658e-01 -4.08700317e-01 -1.50763974e-01 -5.39027512e-01
-6.17829561e-01 -4.25296187e-01 -4.58719656e-02 -1.27250409e+00
4.82517779e-01 9.15958941e-01 3.56876791e-01 -2.20511053e-02
8.59701261e-02 4.10204649e-01 -3.61828059e-01 -3.56571198e-01
5.02285898e-01 8.83026004e-01 4.19963181e-01 -4.41595316e-01
6.89100921e-01 5.10673821e-01 -2.00334266e-01 -1.20601273e+00
-4.30248737e-01 -1.16705072e+00 -1.59890413e-01 -9.07058001e-01
1.07932496e+00 -1.49141300e+00 -5.58962822e-01 6.57943487e-01
-1.18199730e+00 2.63358373e-02 4.99302298e-02 4.97916996e-01
-4.36948985e-01 6.89858735e-01 -4.68648642e-01 -7.01077759e-01
-7.59922385e-01 -1.19631088e+00 1.68250346e+00 -1.38044640e-01
4.21440452e-01 -5.64517081e-01 8.03491250e-02 5.34331322e-01
5.85480452e-01 -1.04656696e-01 1.52694970e-01 -6.38892174e-01
-9.73417819e-01 -5.83989263e-01 -3.07445914e-01 2.25339472e-01
-5.45341074e-01 5.69438219e-01 -7.42504299e-01 -5.81614852e-01
-4.30615157e-01 -6.88231826e-01 1.46929443e+00 3.30623418e-01
1.16766560e+00 1.77291110e-01 -3.35157812e-01 5.98273098e-01
1.38818669e+00 1.53903425e-01 1.09296024e+00 3.88110995e-01
4.88594413e-01 2.54711479e-01 9.34699237e-01 4.16902244e-01
5.85134327e-01 8.56874049e-01 4.85550344e-01 6.74006820e-01
-1.58191219e-01 -1.54041991e-01 6.48532212e-01 1.17701602e+00
-1.54785126e-01 -9.50947762e-01 -7.73818135e-01 6.48648560e-01
-2.27550244e+00 -1.21199989e+00 -2.87974421e-02 2.02406883e+00
-6.00732267e-02 -3.65873486e-01 -2.00171202e-01 -3.16279858e-01
5.00701427e-01 7.55626678e-01 -1.16967224e-01 1.27374277e-01
-1.91409186e-01 -1.47110611e-01 3.75057369e-01 -8.49572644e-02
-1.31697512e+00 1.08178961e+00 7.36633015e+00 1.32664824e+00
-1.03490925e+00 3.17783624e-01 1.86248794e-01 -4.80150580e-01
-1.60097163e-02 -2.71447599e-01 -7.43310332e-01 5.15916765e-01
1.11668563e+00 -3.76349121e-01 5.76337278e-01 7.52302468e-01
-2.97741562e-01 -8.45964178e-02 -9.68470216e-01 1.52830422e+00
7.64719129e-01 -1.58154082e+00 3.72575134e-01 -1.46924078e-01
6.18772447e-01 6.28863275e-01 3.06242615e-01 7.25823998e-01
-2.73972422e-01 -8.05919826e-01 8.46695185e-01 7.00731814e-01
1.10353792e+00 -3.09902459e-01 8.54230583e-01 -4.06608805e-02
-1.49779713e+00 4.23280485e-02 -3.08885992e-01 7.29198217e-01
3.20400327e-01 2.13767454e-01 -2.23655924e-01 8.79124403e-01
1.15365887e+00 1.31625724e+00 -7.53625989e-01 1.14091718e+00
3.55388343e-01 1.88995063e-01 -2.89336354e-01 -1.35788232e-01
2.15886891e-01 1.91018403e-01 7.65985250e-01 1.26673615e+00
5.97345591e-01 2.49292180e-02 1.93721637e-01 -1.03660926e-01
-6.06967866e-01 1.66304260e-01 -7.70105839e-01 -4.62048620e-01
4.40853357e-01 1.11472988e+00 -4.55530047e-01 -5.16098499e-01
-4.78136122e-01 8.11047256e-01 -8.59048739e-02 4.32579458e-01
-9.84747708e-01 -1.54044032e-01 4.34807509e-01 -2.37599656e-01
6.39394701e-01 -2.32269540e-01 8.91735494e-01 -1.69398177e+00
1.31551966e-01 -8.86447787e-01 5.08513927e-01 -1.31019926e+00
-1.12859941e+00 5.54434419e-01 4.27749693e-01 -1.48756146e+00
-4.18598324e-01 -5.61032712e-01 -1.47397101e-01 1.58154726e-01
-1.58004034e+00 -1.15864062e+00 -4.57651973e-01 8.57455790e-01
8.06925058e-01 -5.50902367e-01 4.43091810e-01 9.65947151e-01
-2.32752904e-01 5.35462260e-01 5.10858119e-01 6.63501918e-02
1.03467023e+00 -4.15234447e-01 3.09876829e-01 7.03068435e-01
1.32866070e-01 3.48030686e-01 3.63754809e-01 -5.35299897e-01
-2.27556992e+00 -9.56553817e-01 5.49888015e-01 -6.46768928e-01
8.80644798e-01 -1.06437832e-01 -3.41620117e-01 6.70197487e-01
3.42752129e-01 8.98573175e-02 2.88682133e-01 -2.23229870e-01
-6.50702178e-01 -5.79248220e-02 -7.55913615e-01 5.69155514e-01
8.31465244e-01 -8.88820052e-01 -4.20504093e-01 4.67489123e-01
1.13151956e+00 -4.51815188e-01 -1.04058111e+00 4.44087476e-01
8.47097278e-01 -3.34308505e-01 1.20158446e+00 -5.20424545e-01
5.17279863e-01 -1.19644567e-01 -7.31675148e-01 -6.75084352e-01
1.63504004e-01 -6.76431417e-01 -6.11969471e-01 1.09576905e+00
-7.43663013e-02 5.16786240e-02 5.58296263e-01 2.92756557e-01
8.84077474e-02 -2.76748806e-01 -1.16585886e+00 -8.61673355e-01
-4.22650695e-01 -7.57053733e-01 2.62366325e-01 6.95692658e-01
-2.60277510e-01 9.56799313e-02 -9.23784554e-01 -4.28830564e-01
4.04781491e-01 -3.57562691e-01 9.32769537e-01 -9.60282147e-01
7.53104761e-02 -2.51017421e-01 -9.16967988e-01 -1.53490484e+00
2.45107293e-01 -7.97373176e-01 -6.54410757e-03 -1.51086593e+00
4.91452515e-01 2.74044305e-01 -4.38091367e-01 -7.14544207e-02
6.73309490e-02 3.69320005e-01 8.92946124e-01 6.14508271e-01
-1.80052066e+00 5.21318674e-01 9.38633204e-01 -4.97086257e-01
2.98958749e-01 -4.82520312e-01 -1.93804830e-01 2.18266845e-01
1.17848828e-01 -3.18263233e-01 -4.37595814e-01 -1.01308632e+00
4.78563458e-01 3.00649256e-01 2.11205319e-01 -7.83188820e-01
5.98840177e-01 6.41898289e-02 -1.27268538e-01 -1.10815954e+00
7.54449487e-01 -9.44829583e-01 1.20270215e-01 -9.16517898e-02
-3.37932199e-01 4.20521349e-01 1.41047448e-01 8.20009291e-01
-5.28059065e-01 -3.60060364e-01 9.61397737e-02 3.84994131e-03
-1.28700829e+00 6.94849432e-01 -7.42856801e-01 1.29694059e-01
9.06164706e-01 1.56562045e-01 -9.67554271e-01 -8.79979432e-01
-3.38637859e-01 7.91913867e-01 2.02765658e-01 1.15979099e+00
1.14728332e+00 -1.31118476e+00 -1.02590466e+00 -3.34179521e-01
5.53953886e-01 -5.09160161e-01 6.10091746e-01 6.42121434e-01
-5.87750137e-01 7.30174363e-01 5.96629754e-02 -8.16235244e-01
-1.52369165e+00 7.20192730e-01 -5.86724281e-02 -4.06725794e-01
-5.32980204e-01 6.21471822e-01 -2.82941312e-01 6.90617487e-02
5.21828830e-01 1.75966740e-01 -4.26363558e-01 2.67493457e-01
1.11024868e+00 5.24748564e-01 1.13102667e-01 -9.15687740e-01
-4.14952606e-01 6.72305286e-01 -4.05930310e-01 -7.17559904e-02
1.32188666e+00 -2.87567288e-01 -1.50647804e-01 2.22350404e-01
1.66876018e+00 -5.26939511e-01 -4.45859164e-01 -2.85266668e-01
-2.21359320e-02 -5.86131334e-01 4.10016000e-01 -4.11327869e-01
-1.23016655e+00 6.44233108e-01 7.95621216e-01 1.51738256e-01
1.24198925e+00 -2.38281470e-02 1.04288018e+00 9.20050442e-01
4.68468130e-01 -1.41104913e+00 2.31760889e-01 7.05721259e-01
8.64849627e-01 -1.30848467e+00 3.97584587e-01 -3.29058245e-02
-5.23731172e-01 1.01994038e+00 2.59627551e-01 -2.30850428e-01
9.26248729e-01 -1.74383208e-01 -3.05902064e-01 -4.10582095e-01
-1.37897289e+00 -2.20403478e-01 6.70622528e-01 4.04085129e-01
1.99659243e-01 -2.14509457e-01 -3.21835935e-01 1.45520851e-01
6.13317907e-01 3.81506175e-01 3.01491857e-01 1.09956110e+00
-3.11796337e-01 -6.54926717e-01 -8.51225629e-02 4.90643680e-01
-7.58203864e-01 -4.13411975e-01 -3.94692928e-01 7.98976302e-01
-7.08790123e-01 9.97260153e-01 -1.87259894e-02 -9.32667136e-01
1.66922614e-01 -9.87811238e-02 3.81825000e-01 -1.60788760e-01
-5.24456084e-01 2.22451910e-01 1.83507115e-01 -9.15146589e-01
-9.26593125e-01 -4.09073919e-01 -5.93749881e-01 -6.19222879e-01
-3.17824960e-01 6.30510449e-02 8.58425319e-01 8.73025000e-01
4.59257871e-01 3.88738275e-01 4.43684489e-01 -8.80267739e-01
-1.54654875e-01 -7.97888756e-01 -4.28531080e-01 3.84256095e-01
3.50474238e-01 -5.37141681e-01 -3.15073848e-01 1.42921016e-01] | [10.382040023803711, 0.836878776550293] |
e8cd254a-cadb-45b0-8760-e09446ce9be6 | ssn-mlrg1-dravidianlangtech-acl2022-troll | null | null | https://aclanthology.org/2022.dravidianlangtech-1.21 | https://aclanthology.org/2022.dravidianlangtech-1.21.pdf | SSN_MLRG1@DravidianLangTech-ACL2022: Troll Meme Classification in Tamil using Transformer Models | The ACL shared task of DravidianLangTech-2022 for Troll Meme classification is a binary classification task that involves identifying Tamil memes as troll or not-troll. Classification of memes is a challenging task since memes express humour and sarcasm in an implicit way. Team SSN_MLRG1 tested and compared results obtained by using three models namely BERT, ALBERT and XLNET. The XLNet model outperformed the other two models in terms of various performance metrics. The proposed XLNet model obtained the 3rd rank in the shared task with a weighted F1-score of 0.558. | ['Angel S', 'Rajalakshmi Sivanaiah', 'Saritha Madhavan', 'Sarika Esackimuthu', 'Shruthi Hariprasad'] | null | null | null | null | dravidianlangtech-acl-2022-5 | ['meme-classification'] | ['natural-language-processing'] | [-5.58252931e-01 -5.07808268e-01 4.60443236e-02 1.19947188e-01
-5.56283593e-01 -4.27035123e-01 1.03092313e+00 4.67117399e-01
-9.11704183e-01 1.23014915e+00 4.27012533e-01 -2.15839565e-01
1.09873101e-01 -5.57968616e-01 -3.08820903e-01 -2.78575361e-01
1.83956549e-01 5.56870639e-01 3.12130541e-01 -6.54194832e-01
1.10419369e+00 -2.10542306e-01 -8.34648073e-01 8.13302219e-01
4.32694286e-01 8.07681739e-01 -3.80676568e-01 1.21357787e+00
2.59591592e-03 1.82985449e+00 -1.11187518e+00 -6.99656963e-01
-7.77620226e-02 -4.38133717e-01 -9.20982122e-01 -7.07956493e-01
9.23140883e-01 2.60549903e-01 -6.53589308e-01 6.17444336e-01
5.79825222e-01 2.21932754e-01 7.60756314e-01 -1.12966514e+00
-2.42450714e-01 6.89293444e-01 -4.59790468e-01 8.84186924e-01
3.52921575e-01 -4.31855582e-02 1.08851230e+00 -1.13659644e+00
6.94774330e-01 1.48115230e+00 9.08041954e-01 4.15803492e-01
-1.22552264e+00 -9.38962400e-01 -5.06798923e-01 2.70589650e-01
-1.36313581e+00 -3.26669216e-01 5.05179465e-01 -8.28361034e-01
9.58668470e-01 3.22039843e-01 1.68275550e-01 1.53173852e+00
1.29777110e+00 7.37687886e-01 1.81625628e+00 6.97144568e-02
-7.41097555e-02 2.13556990e-01 7.46105015e-01 7.72381186e-01
1.62068099e-01 -3.15240234e-01 -1.48684692e+00 -5.30353904e-01
1.81841135e-01 -3.73280346e-01 1.05053917e-01 7.41231799e-01
-1.44365644e+00 1.03063416e+00 3.93387467e-01 4.44551885e-01
1.41923055e-02 4.23544556e-01 9.91330147e-01 8.78876090e-01
1.14175773e+00 6.57405615e-01 -5.43153137e-02 -2.86806345e-01
-9.60757852e-01 4.84973282e-01 9.38248992e-01 3.00638676e-01
4.71825302e-01 2.31230333e-02 -4.59598042e-02 9.64591324e-01
-1.05571384e-02 5.24787128e-01 4.62921619e-01 -2.15116084e-01
6.46362364e-01 5.49255371e-01 6.19373703e-03 -1.58716416e+00
-8.62484574e-01 -7.50846982e-01 -8.79712284e-01 1.00688256e-01
2.99149126e-01 -1.94343656e-01 -5.59275746e-01 1.34987295e+00
-3.62333119e-01 -1.96522474e-02 8.82817581e-02 9.21628535e-01
1.29656470e+00 5.99694133e-01 1.47350565e-01 2.32935607e-01
8.91971588e-01 -9.89771783e-01 -6.90917850e-01 -6.18715107e-01
8.70983720e-01 -1.10283291e+00 9.01496828e-01 4.50588584e-01
-6.30927980e-01 -4.02978927e-01 -1.19561625e+00 4.77252295e-03
-2.71982789e-01 -3.61165367e-02 3.49313378e-01 3.93223673e-01
-6.68618858e-01 5.46069264e-01 -3.45753819e-01 -6.14786506e-01
1.10080168e-01 6.01046905e-02 -5.03921211e-01 3.70641112e-01
-1.20366716e+00 1.05325127e+00 5.53821623e-01 -1.20992675e-01
-9.74994302e-01 -4.33787316e-01 -2.62134492e-01 -6.74860597e-01
1.51190326e-01 -4.06822294e-01 8.42373908e-01 -7.26901650e-01
-1.19857669e+00 1.36388624e+00 3.09544235e-01 -7.04117119e-01
1.08822167e+00 -6.18621349e-01 -5.90441883e-01 -1.04024708e-01
2.33720064e-01 1.54410914e-01 5.10620654e-01 -8.90849590e-01
-5.71007013e-01 5.03144935e-02 -5.28072678e-02 4.12967503e-02
-2.87861943e-01 3.34566504e-01 5.28664947e-01 -6.52604103e-01
1.86687782e-01 -9.66646135e-01 3.09958190e-01 -9.33096051e-01
-6.81062341e-01 -4.38451767e-01 7.73509860e-01 -7.83705711e-01
1.41982865e+00 -1.81019950e+00 7.19582364e-02 -1.01894781e-01
9.55370903e-01 -1.02911182e-01 1.99344322e-01 8.97656500e-01
3.98422062e-01 3.28111947e-01 1.23094559e-01 -3.78789902e-01
-1.46961361e-01 -3.09018761e-01 -1.83467746e-01 8.56200576e-01
-4.01921749e-01 7.38043249e-01 -8.96704733e-01 -6.29712403e-01
-1.04298219e-01 -1.16283014e-01 -1.90787464e-01 2.00546905e-02
-7.77310356e-02 6.71105266e-01 -2.23964706e-01 3.64379138e-01
2.21430868e-01 -4.11704987e-01 1.24503627e-01 2.20984191e-01
-1.65939420e-01 5.97113431e-01 -6.02319539e-01 1.18612218e+00
-3.70510519e-01 1.24978006e+00 -2.28706405e-01 -1.78101346e-01
1.48061299e+00 2.00271398e-01 8.54001418e-02 -8.28716397e-01
6.36671901e-01 6.22656345e-01 4.59907167e-02 -2.95569062e-01
9.25354481e-01 -5.17581344e-01 -7.65307724e-01 8.52098942e-01
-2.00143054e-01 3.30481976e-01 1.77720815e-01 8.65774214e-01
1.51955819e+00 -3.81570816e-01 2.52066165e-01 -7.90807724e-01
7.24413335e-01 1.62187412e-01 5.37123978e-01 1.13636792e+00
-2.85461247e-01 2.83170700e-01 6.50763214e-01 -8.84327114e-01
-1.08949578e+00 -7.41131485e-01 3.11827045e-02 1.49760067e+00
5.09441569e-02 -7.49105394e-01 -1.95971504e-01 -8.08456659e-01
-9.09045637e-02 7.42293358e-01 -7.81936884e-01 -2.04776853e-01
-7.46420085e-01 -6.18402004e-01 1.14908206e+00 7.96746016e-02
9.36122894e-01 -1.10668409e+00 -5.77912629e-01 2.69039035e-01
-7.98078120e-01 -1.05785763e+00 -3.04186851e-01 3.51312533e-02
-6.15639269e-01 -1.22340870e+00 -4.98795360e-01 -4.53275740e-01
-5.09195663e-02 -1.73587620e-01 1.34051096e+00 5.28788269e-01
3.02519858e-01 -1.38384134e-01 -5.59477925e-01 -1.26184642e-01
-7.65551984e-01 6.40516579e-01 2.75894552e-01 9.20047387e-02
4.03154999e-01 -7.45652691e-02 -3.50658834e-01 5.25738418e-01
-2.97572851e-01 4.57655877e-01 1.36662379e-01 8.49538624e-01
-3.20169538e-01 -4.47949320e-01 8.22973907e-01 -1.21545625e+00
8.79377425e-01 -8.02515924e-01 2.94083565e-01 -2.41091713e-01
-2.73203850e-01 -3.76901805e-01 4.62254494e-01 -3.00883204e-01
-6.90589190e-01 -6.90823138e-01 6.26026317e-02 1.44638285e-01
2.88461000e-01 5.44777215e-01 6.38834059e-01 -5.94831146e-02
1.09044755e+00 -2.11803243e-02 -1.56241596e-01 -6.70898199e-01
-4.07231212e-01 9.72000837e-01 5.67429781e-01 -4.49550003e-01
4.86175478e-01 1.37848273e-01 -9.26391629e-04 -9.03004766e-01
-1.22748816e+00 -4.46193814e-01 -4.26289946e-01 -6.56453550e-01
7.84853578e-01 -7.19082594e-01 -1.06811571e+00 9.60568726e-01
-1.38497734e+00 -2.63928324e-01 6.56291783e-01 3.50425601e-01
-3.37505311e-01 1.45843048e-02 -1.25402784e+00 -7.26697087e-01
-5.15175462e-01 -5.83452523e-01 3.68424326e-01 -1.89266473e-01
-7.04780340e-01 -1.12551904e+00 4.14849550e-01 5.95211148e-01
4.65504795e-01 6.13228440e-01 8.72286439e-01 -1.20791090e+00
8.90203472e-03 -1.91749081e-01 -1.90243796e-01 2.12490693e-01
-5.54141402e-01 -6.04770899e-01 -7.14026451e-01 -4.69353616e-01
-1.28945321e-01 -9.11687613e-01 1.23658669e+00 -2.67003208e-01
3.50634813e-01 -4.67595041e-01 1.36402426e-02 -1.79778952e-02
1.35712528e+00 -3.24309617e-01 5.70989370e-01 9.28909183e-01
6.97805464e-01 3.53625774e-01 4.36670780e-01 5.56669950e-01
1.82481453e-01 4.62962419e-01 3.18802416e-01 4.82543677e-01
-1.69889733e-01 -2.98809975e-01 5.49048901e-01 1.14692664e+00
2.03484148e-02 -4.44184840e-01 -1.53728831e+00 2.29041442e-01
-2.09914756e+00 -8.33656371e-01 -1.11096597e+00 1.83099473e+00
4.40230966e-01 6.09498143e-01 3.57905835e-01 3.55767190e-01
7.71620333e-01 4.28148329e-01 1.23814642e-01 -7.87062526e-01
-5.68357170e-01 -4.67669964e-02 5.48315942e-01 6.29753232e-01
-1.07283843e+00 1.23963809e+00 6.15014839e+00 8.14844906e-01
-9.61588800e-01 6.95880651e-01 4.08170044e-01 -1.97671860e-01
3.43789309e-01 -1.53102502e-01 -6.28277600e-01 7.04616308e-01
1.33526242e+00 -1.43239290e-01 -6.73717167e-03 1.55582145e-01
-9.80947986e-02 -4.61671233e-01 -9.67648089e-01 9.19842005e-01
5.44981003e-01 -1.64720249e+00 -1.38802066e-01 6.18004724e-02
9.01171744e-01 3.06045294e-01 -7.76667446e-02 5.50657988e-01
2.35235527e-01 -1.19510877e+00 1.29999363e+00 9.16938543e-01
4.20734584e-01 -7.79810250e-01 1.09342730e+00 5.97070098e-01
-6.05736434e-01 -5.43224588e-02 -3.10807645e-01 -6.86395586e-01
-1.06160596e-01 3.36063176e-01 -6.51485205e-01 2.70807236e-01
6.26865506e-01 9.17909443e-01 -9.02464807e-01 9.93479848e-01
-1.19845986e-01 1.15369749e+00 -8.42132512e-03 -5.97802818e-01
4.86106932e-01 4.54423547e-01 1.09273994e+00 1.65807915e+00
-2.21390963e-01 -3.18329334e-01 3.56250763e-01 4.88996685e-01
-2.51420856e-01 3.70750427e-01 -3.68552834e-01 -1.99597061e-01
2.55484879e-01 1.14316022e+00 -8.01165223e-01 -2.68742204e-01
1.98175699e-01 8.81179094e-01 1.10714607e-01 -1.17492855e-01
-7.69931555e-01 -4.61914122e-01 3.81984678e-03 3.45684946e-01
-5.15265048e-01 -5.38435698e-01 -6.47397876e-01 -7.47420490e-01
-5.86063862e-01 -1.01129699e+00 5.63351333e-01 -4.47874367e-01
-1.60465384e+00 9.90757823e-01 -3.73609722e-01 -9.43832517e-01
2.16407329e-01 -6.13933086e-01 -5.84934235e-01 6.69512987e-01
-6.19724512e-01 -1.36293101e+00 -4.89014834e-01 3.50866556e-01
5.85591793e-01 -7.51551867e-01 4.10066366e-01 3.24811876e-01
-5.31587183e-01 5.49061775e-01 -7.04499558e-02 3.46052080e-01
1.05641747e+00 -9.20309722e-01 -9.69120953e-03 2.28745133e-01
-1.20511979e-01 3.78964037e-01 1.19250679e+00 -1.07319498e+00
-8.66777182e-01 -7.92224348e-01 1.23942101e+00 -8.88173759e-01
1.34441340e+00 -4.99542475e-01 -7.46363103e-01 7.11791754e-01
2.42152736e-01 -5.55198908e-01 6.30899966e-01 4.37030494e-01
-5.89546740e-01 4.73698586e-01 -8.74760807e-01 1.56281412e-01
7.13142037e-01 -5.61082959e-01 -7.93911636e-01 6.66650116e-01
2.54852384e-01 -2.38975286e-01 -7.32133090e-01 -1.10474214e-01
5.01205385e-01 -1.01693165e+00 1.87657565e-01 -9.11673367e-01
1.12461257e+00 4.07159142e-02 -4.44556266e-01 -1.05151439e+00
-1.22010171e-01 -4.84060735e-01 1.51573911e-01 9.75683987e-01
3.68084043e-01 -6.85059667e-01 6.34469748e-01 -3.40356439e-01
1.56920124e-02 -3.97238135e-01 -1.05083692e+00 -8.58118594e-01
5.61872900e-01 -4.78395045e-01 -2.16227964e-01 1.10600007e+00
5.98406084e-02 1.17096794e+00 -1.07124078e+00 -5.97303629e-01
7.33530521e-01 -3.65138531e-01 1.00734544e+00 -1.44727528e+00
-2.82089561e-02 -5.14626920e-01 -8.44193637e-01 -5.12896776e-01
4.03805941e-01 -1.39277589e+00 -2.44390771e-01 -1.31403553e+00
7.28636861e-01 -4.02225554e-01 -4.95865107e-01 2.97585547e-01
1.52079329e-01 1.06723952e+00 4.88271654e-01 7.43590772e-01
-1.07982445e+00 5.66097610e-02 7.68108487e-01 -2.78787136e-01
5.29557429e-02 -2.86640227e-01 -2.75584847e-01 6.81319594e-01
9.54318523e-01 -9.48736966e-01 4.44434345e-01 -1.02926284e-01
1.01334810e+00 2.98653115e-02 7.63649583e-01 -1.26498115e+00
3.09221327e-01 -1.56795532e-01 -4.61976836e-03 -8.85571361e-01
3.20210606e-01 1.42683148e-01 1.46872714e-01 7.07904518e-01
-6.04776680e-01 2.59374529e-01 1.10738181e-01 4.59071636e-01
1.66005548e-03 -1.10027187e-01 9.90819514e-01 -1.20745353e-01
-5.46695948e-01 -2.54526168e-01 -8.17247391e-01 3.41170341e-01
6.77865922e-01 -1.61879778e-01 -9.75068748e-01 -4.06737417e-01
-7.45574594e-01 -3.46717499e-02 2.76586711e-01 6.29195988e-01
4.81036365e-01 -1.11400807e+00 -1.56107318e+00 -4.93659049e-01
1.70655876e-01 -1.23490834e+00 2.37186074e-01 1.31100798e+00
-9.31152105e-01 3.16657990e-01 -5.31216562e-01 -4.56546038e-01
-1.37422991e+00 -1.36493683e-01 4.80901778e-01 -4.88736421e-01
-8.68907094e-01 7.41794348e-01 -1.83723241e-01 -4.25979048e-01
-2.59060293e-01 1.10445249e+00 -1.54946566e-01 5.14129438e-02
6.00499034e-01 1.12698960e+00 1.34256199e-01 -1.11930084e+00
-6.08835220e-01 1.36254996e-01 -2.42622375e-01 -2.85015464e-01
1.19576836e+00 -5.01642609e-03 -4.80117917e-01 1.33906257e+00
1.21857023e+00 1.87086821e-01 -1.97532073e-01 -3.01918358e-01
4.35283482e-01 -5.65259874e-01 2.42722169e-01 -1.17791653e+00
-4.88638550e-01 6.82273388e-01 3.45679939e-01 1.87001899e-01
2.95111015e-02 -1.10756524e-01 9.74472523e-01 4.56664652e-01
5.66949904e-01 -1.34816289e+00 8.65274370e-01 1.34194827e+00
1.16858447e+00 -1.09292328e+00 1.20633990e-01 1.79258510e-01
-8.18609893e-01 1.15146661e+00 8.75863612e-01 -5.34231067e-01
2.82081127e-01 -6.69196695e-02 2.41877615e-01 -5.25699377e-01
-1.04426932e+00 3.84480864e-01 4.99701977e-01 -2.64813185e-01
9.03645277e-01 1.27502009e-01 -1.13298380e+00 3.77095014e-01
-5.59119523e-01 -4.40698117e-01 9.05917108e-01 9.21465874e-01
-7.82964528e-01 -4.28177565e-01 -4.45834965e-01 7.33399987e-01
-7.63275981e-01 2.62174644e-02 -1.00040627e+00 8.01750183e-01
-8.68137777e-02 1.34991086e+00 5.41031994e-02 -1.31256795e+00
-2.68342793e-01 8.72224048e-02 2.82323174e-02 -5.39221883e-01
-1.39030254e+00 -2.74202406e-01 7.53813744e-01 -1.95545226e-01
-3.40324529e-02 -2.82413632e-01 -1.16201210e+00 -1.35670805e+00
-2.01873213e-01 2.55768806e-01 6.88817680e-01 9.02054191e-01
-1.92456558e-01 4.64491308e-01 5.22616684e-01 -4.63627160e-01
-3.57176125e-01 -1.44121623e+00 -5.98559916e-01 4.12161678e-01
1.07813321e-01 -6.98002338e-01 -5.71015954e-01 -5.13283372e-01] | [8.634638786315918, 10.847969055175781] |
94977595-3adf-48f9-9cb3-f36bce03bef9 | natcs-eliciting-natural-customer-support | 2305.03007 | null | https://arxiv.org/abs/2305.03007v1 | https://arxiv.org/pdf/2305.03007v1.pdf | NatCS: Eliciting Natural Customer Support Dialogues | Despite growing interest in applications based on natural customer support conversations, there exist remarkably few publicly available datasets that reflect the expected characteristics of conversations in these settings. Existing task-oriented dialogue datasets, which were collected to benchmark dialogue systems mainly in written human-to-bot settings, are not representative of real customer support conversations and do not provide realistic benchmarks for systems that are applied to natural data. To address this gap, we introduce NatCS, a multi-domain collection of spoken customer service conversations. We describe our process for collecting synthetic conversations between customers and agents based on natural language phenomena observed in real conversations. Compared to previous dialogue datasets, the conversations collected with our approach are more representative of real human-to-human conversations along multiple metrics. Finally, we demonstrate potential uses of NatCS, including dialogue act classification and intent induction from conversations as potential applications, showing that dialogue act annotations in NatCS provide more effective training data for modeling real conversations compared to existing synthetic written datasets. We publicly release NatCS to facilitate research in natural dialog systems | ['Saab Mansour', 'Yi Zhang', 'Arshit Gupta', 'Wesley Rose', 'Emily Moeng', 'James Gung'] | 2023-05-04 | null | null | null | null | ['dialogue-act-classification'] | ['natural-language-processing'] | [ 9.50098932e-02 6.28223062e-01 3.88329215e-02 -8.55658770e-01
-7.18623757e-01 -8.60581160e-01 1.42263341e+00 -1.57012284e-01
-2.37732634e-01 1.04711509e+00 8.55938315e-01 -3.76625508e-01
3.99687320e-01 -4.40943688e-01 8.13444927e-02 -1.45728961e-01
1.47110477e-01 1.50669932e+00 5.73231131e-02 -9.44089055e-01
2.22186506e-01 1.50114283e-01 -1.00941980e+00 9.17919993e-01
6.74234152e-01 3.13078046e-01 -1.20059783e-02 1.11003160e+00
-4.38473046e-01 1.42829132e+00 -1.24015260e+00 -4.65576023e-01
2.05698647e-02 -8.67675364e-01 -1.39062095e+00 5.04307628e-01
3.17918882e-02 -6.65336728e-01 -2.48654753e-01 3.37576091e-01
4.75849956e-01 2.19271019e-01 7.45613515e-01 -1.69914436e+00
-4.80623513e-01 9.01725471e-01 3.71298730e-01 -4.13369797e-02
9.97420847e-01 6.42778277e-01 1.13388264e+00 -4.99834806e-01
7.93074727e-01 1.76767170e+00 6.31630301e-01 1.11596048e+00
-1.40582454e+00 -2.91606754e-01 -1.76977664e-01 -3.54670167e-01
-2.82031953e-01 -8.89179528e-01 4.36372399e-01 -4.11177695e-01
1.23140407e+00 4.46881354e-01 5.71858644e-01 1.98403549e+00
-4.26294595e-01 1.16233718e+00 1.04509139e+00 -3.35171908e-01
-4.66556773e-02 5.69685102e-01 4.07989353e-01 3.15504223e-01
-5.29704571e-01 -2.38810778e-01 -4.87192839e-01 -6.08605921e-01
4.25004959e-01 -4.49075252e-01 -2.61481792e-01 3.74362767e-02
-1.23516095e+00 1.13106573e+00 -1.52526364e-01 2.14670971e-01
-1.66109309e-01 -2.61962056e-01 9.17696416e-01 6.21267021e-01
5.68079233e-01 7.85411417e-01 -5.00111699e-01 -9.35740471e-01
-5.67744039e-02 8.19102347e-01 2.00901389e+00 1.28736877e+00
4.26178336e-01 -6.85156435e-02 -3.04367721e-01 1.33408737e+00
5.53239845e-02 3.22936326e-01 4.63273525e-01 -1.43470144e+00
7.33674407e-01 6.58189058e-01 4.76877630e-01 -5.13184011e-01
-4.57702577e-01 5.48261762e-01 -4.01528150e-01 -3.71278912e-01
9.09814179e-01 -4.89377022e-01 -6.03286773e-02 1.35598493e+00
-1.19076818e-01 -4.05318320e-01 7.59150207e-01 5.05958855e-01
1.11459005e+00 7.70491242e-01 -2.01859325e-01 -2.72122890e-01
1.24549615e+00 -1.33492136e+00 -8.67863417e-01 -2.65566319e-01
9.78837252e-01 -7.44306445e-01 1.50798309e+00 1.98368639e-01
-1.07319200e+00 -3.76594998e-02 -2.79025763e-01 5.28348349e-02
-2.39503756e-01 -3.47457677e-01 7.92248189e-01 5.23753285e-01
-1.05375600e+00 3.03401239e-02 -3.75813484e-01 -6.81530416e-01
-2.67276704e-01 -4.99552637e-02 -1.08168714e-01 1.95237964e-01
-1.15476799e+00 1.07166398e+00 -1.80174664e-01 -1.64073318e-01
-8.36896062e-01 -3.37522537e-01 -1.00396144e+00 -2.13206768e-01
3.41318637e-01 -1.47895530e-01 2.51142740e+00 -7.30152547e-01
-1.80314279e+00 8.44631851e-01 -1.35162100e-01 -7.78803170e-01
7.54334390e-01 1.41173705e-01 -1.77090377e-01 9.75383818e-03
1.93486512e-01 6.02249026e-01 -5.85252903e-02 -1.53349113e+00
-5.77619553e-01 1.40676558e-01 4.26376641e-01 2.02633634e-01
-9.00764316e-02 2.22930253e-01 -5.23350686e-02 -1.61679238e-01
-6.56379640e-01 -9.83617365e-01 -9.95576531e-02 -5.05236089e-01
-7.44794130e-01 -6.22308075e-01 9.21307325e-01 -2.30518267e-01
6.00980461e-01 -1.54732716e+00 -2.50614285e-01 -3.73158187e-01
2.20670313e-01 2.97518402e-01 -2.02096760e-01 1.24330449e+00
4.42197740e-01 2.31318295e-01 -5.98695539e-02 -8.60931039e-01
3.71975958e-01 4.96610582e-01 -4.40479249e-01 -1.06634863e-01
1.81572467e-01 8.31759512e-01 -9.79793668e-01 -4.21337575e-01
3.01766992e-01 -1.26487553e-01 -5.31295657e-01 7.90832222e-01
-7.02891290e-01 7.60127306e-01 -5.07510483e-01 1.34827226e-01
1.70709640e-01 -9.11081582e-02 2.66333133e-01 2.93655097e-01
-3.54574472e-02 9.60465074e-01 -1.47676334e-01 1.16374707e+00
-7.25731611e-01 9.83402133e-01 4.66077447e-01 -6.99308991e-01
1.08559620e+00 6.99286580e-01 3.27464402e-01 -4.39882278e-01
7.51572549e-02 -2.56793182e-02 3.08766901e-01 -6.79633200e-01
7.85205960e-01 -1.54917985e-01 -5.36160588e-01 1.18518269e+00
9.39724967e-03 -7.66502678e-01 5.02696097e-01 5.41594744e-01
1.18380415e+00 -6.90629065e-01 1.11756653e-01 2.99101938e-02
3.84279549e-01 4.89132851e-01 3.47663581e-01 9.01229620e-01
-5.53771853e-01 4.68056232e-01 9.07155156e-01 -3.96132588e-01
-1.15157115e+00 -6.87196851e-01 -2.38132346e-02 1.26572919e+00
-1.79143205e-01 -3.65571797e-01 -8.71157050e-01 -7.52120376e-01
-1.18388176e-01 1.13635564e+00 -3.26757014e-01 3.40059280e-01
-6.52612746e-01 -5.00629544e-01 1.18719411e+00 1.60998717e-01
6.31828427e-01 -1.65091252e+00 -1.53586969e-01 3.60375702e-01
-5.81844747e-01 -1.77543664e+00 -5.94832778e-01 -1.15237191e-01
-2.69344360e-01 -1.38784671e+00 -5.08464992e-01 -7.58117259e-01
1.44272700e-01 2.69763112e-01 1.58264053e+00 4.24605086e-02
-8.79389867e-02 8.15509677e-01 -6.54744506e-01 -5.44047475e-01
-1.63171649e+00 1.25255480e-01 3.50426137e-02 -2.96567947e-01
5.84286749e-01 -1.82445139e-01 -2.23935679e-01 7.16024995e-01
-4.94522393e-01 3.72208774e-01 -1.37867197e-01 1.11004901e+00
-9.55165267e-01 -6.65747225e-01 9.45779264e-01 -1.22691917e+00
1.66145062e+00 -4.94774282e-01 1.45695789e-03 1.18378915e-01
-1.40528888e-01 -3.43760341e-01 7.70835340e-01 -2.74223119e-01
-1.39111710e+00 -3.85521710e-01 -1.93670526e-01 3.99952739e-01
-5.95541239e-01 8.57899373e-04 2.88080752e-01 5.52312553e-01
7.50296354e-01 3.49569529e-01 5.24610460e-01 -3.25972676e-01
2.56800383e-01 1.19587755e+00 3.50050300e-01 -9.41897511e-01
7.46464580e-02 5.41504286e-03 -8.82663369e-01 -1.05151904e+00
-4.91385341e-01 -4.43959087e-01 -4.21177417e-01 -1.71071872e-01
8.03508162e-01 -6.73698425e-01 -1.38006711e+00 5.54007411e-01
-1.57467973e+00 -1.03121102e+00 -1.72325760e-01 2.46094197e-01
-9.43409026e-01 2.38910139e-01 -1.41827452e+00 -1.41022873e+00
-2.98014134e-01 -1.26443529e+00 1.01336968e+00 8.14480036e-02
-9.52814996e-01 -1.39630151e+00 1.90527126e-01 7.67350435e-01
4.74356800e-01 2.21065320e-02 8.67261887e-01 -1.42982602e+00
1.71017144e-02 -1.91293344e-01 3.88132781e-02 3.88627172e-01
4.85064209e-01 1.36947930e-01 -9.85781133e-01 -2.63811126e-02
-8.33503529e-02 -1.25486362e+00 1.88793212e-01 -1.86814964e-01
3.50548297e-01 -7.42814362e-01 -1.05099186e-01 -4.73831624e-01
2.44630978e-01 4.69703734e-01 2.57238537e-01 1.75621539e-01
7.05144480e-02 1.21058226e+00 5.97370028e-01 6.26047254e-01
9.26477134e-01 7.04903543e-01 2.24856175e-02 -1.68259010e-01
1.98404491e-01 -1.28471822e-01 4.40747291e-01 9.21997070e-01
3.35028857e-01 -6.79410219e-01 -1.17093742e+00 5.65708280e-01
-1.91447091e+00 -1.04901910e+00 -2.72489488e-01 1.51532221e+00
1.30224085e+00 2.32580811e-01 7.08434701e-01 -2.66701967e-01
6.83701277e-01 2.25055784e-01 -2.98424274e-01 -8.73656929e-01
-9.35866460e-02 -3.03137839e-01 -1.73892319e-01 8.60179484e-01
-7.18699753e-01 9.20165420e-01 7.04645014e+00 9.99658629e-02
-6.08435988e-01 -4.68443595e-02 7.70238161e-01 6.69198781e-02
-2.02935457e-01 -9.06843096e-02 -7.59968877e-01 3.00441444e-01
1.07908368e+00 -3.57827723e-01 4.76622254e-01 7.93991804e-01
5.73735952e-01 1.98697075e-02 -1.71406865e+00 5.47503710e-01
-6.86754510e-02 -1.33530045e+00 -1.27259359e-01 4.34952602e-02
4.93372023e-01 -2.94694584e-02 -4.20576036e-01 7.85879135e-01
1.30108178e+00 -1.06258249e+00 2.62629032e-01 1.09478086e-01
2.67953724e-01 -3.01785052e-01 7.95331538e-01 9.17584658e-01
-4.92989331e-01 -1.08134293e-03 -1.43838510e-01 -2.76440859e-01
6.03707254e-01 -6.00599125e-02 -1.98364902e+00 -1.10133439e-01
2.57022798e-01 5.14376402e-01 -4.85365018e-02 2.17613712e-01
3.10479701e-01 6.65635347e-01 -4.32055518e-02 -5.79674661e-01
5.20428598e-01 -2.41310999e-01 4.55089033e-01 1.66433656e+00
-3.41758221e-01 1.65203273e-01 5.68727374e-01 8.08936119e-01
-1.12053044e-01 -1.69423506e-01 -1.21841729e+00 -4.74780887e-01
6.88521624e-01 1.13632047e+00 -2.19148427e-01 -4.23788607e-01
-4.49170262e-01 8.36940408e-01 2.26091713e-01 2.60909379e-01
-3.87916923e-01 -1.02581486e-01 8.43213916e-01 -1.21670686e-01
-5.39668620e-01 -2.65667289e-01 -3.59345577e-03 -1.02805245e+00
-2.49593273e-01 -1.65443659e+00 -7.47814262e-03 -6.33469343e-01
-1.77779090e+00 8.71651649e-01 9.99986082e-02 -9.91738677e-01
-1.23827004e+00 -5.45627356e-01 -9.79970217e-01 7.59083509e-01
-7.49233782e-01 -9.90958631e-01 -3.94887924e-01 4.46674407e-01
1.46984351e+00 -6.29643321e-01 1.36883056e+00 -2.24510282e-01
-5.17335653e-01 4.66266006e-01 -6.90620020e-02 4.96579915e-01
8.74821126e-01 -1.36782718e+00 9.46047187e-01 -2.71440208e-01
-2.38132223e-01 6.12949371e-01 9.54025388e-01 -3.20281714e-01
-1.14477050e+00 -6.86265588e-01 6.97620928e-01 -8.40402067e-01
8.17336023e-01 -7.54599869e-01 -8.48420799e-01 9.79919553e-01
8.25852811e-01 -6.87068939e-01 7.68292069e-01 2.75183707e-01
-6.47588968e-02 4.60463226e-01 -1.22111011e+00 8.10900152e-01
8.33510637e-01 -6.58283889e-01 -8.25832486e-01 7.39083529e-01
7.40144193e-01 -3.74368876e-01 -8.34399283e-01 1.71813950e-01
3.78550589e-01 -1.13730514e+00 6.13326192e-01 -9.62310255e-01
6.50337279e-01 5.04776657e-01 -1.54961482e-01 -1.68185604e+00
4.49157029e-01 -1.02561009e+00 3.39610428e-01 1.59733140e+00
6.50771201e-01 -7.18913615e-01 7.31261194e-01 1.12632978e+00
-1.19705431e-01 -4.38055396e-01 -5.06966472e-01 -4.13618207e-01
2.03020468e-01 -2.19437063e-01 5.16546011e-01 1.14502335e+00
7.36071229e-01 9.87671316e-01 -4.14032280e-01 -6.47729993e-01
1.42575815e-01 -1.03677236e-01 1.65313554e+00 -1.11641574e+00
-3.52967918e-01 -4.07070458e-01 1.65668741e-01 -1.62176538e+00
6.53031707e-01 -4.94898170e-01 6.16117895e-01 -1.54021072e+00
1.74398236e-02 -6.63314223e-01 1.09556794e+00 2.63760746e-01
1.22823283e-01 -2.59079754e-01 2.47207239e-01 3.47740620e-01
-4.90789413e-01 7.81963706e-01 1.33631265e+00 -2.19680965e-01
-1.93025813e-01 3.98102820e-01 -5.78670025e-01 7.67427564e-01
7.79831529e-01 -7.07234219e-02 -6.79343879e-01 -1.02041520e-01
-5.25658250e-01 5.93553305e-01 7.21209347e-02 -3.24937463e-01
1.98392615e-01 -4.06333178e-01 -4.81141239e-01 -3.10668200e-01
8.85431647e-01 -3.13336402e-01 -6.90156877e-01 1.37472421e-01
-9.62314188e-01 5.41262999e-02 3.73426601e-02 3.59015614e-01
-2.73196906e-01 -3.57923955e-01 3.94264728e-01 -5.75591028e-01
-2.74686933e-01 -3.20900619e-01 -1.11520720e+00 4.46206301e-01
8.02280843e-01 3.68281342e-02 -9.70067382e-01 -1.55982888e+00
-5.50403953e-01 7.85763741e-01 3.45494479e-01 5.41787028e-01
2.71693528e-01 -8.74851942e-01 -9.92720246e-01 -1.52491301e-01
2.85950869e-01 1.14848595e-02 -3.63939941e-01 3.82064670e-01
-5.92191100e-01 7.25625575e-01 5.60706574e-03 -6.47575676e-01
-1.36883605e+00 -6.23244792e-02 5.42317629e-01 -3.08744282e-01
-6.10674977e-01 5.17449796e-01 2.98915416e-01 -1.37568581e+00
5.31500936e-01 -2.34411642e-01 -3.69634070e-02 -8.21337849e-02
4.45148617e-01 1.18211083e-01 -3.04748803e-01 -5.13036609e-01
-1.71586797e-02 -5.51651776e-01 -3.11938614e-01 -6.19477332e-01
1.16757154e+00 -2.52998710e-01 3.94800492e-02 9.38885093e-01
1.05122423e+00 -7.86766484e-02 -1.11490440e+00 -3.39073390e-01
2.95248866e-01 -2.38900125e-01 -9.83842969e-01 -7.19770730e-01
-2.49492794e-01 6.56222105e-01 -2.49157831e-01 1.23711324e+00
3.14796954e-01 1.38376579e-01 8.97336423e-01 1.14915299e+00
3.78329813e-01 -1.04051173e+00 6.03333712e-01 1.21948695e+00
1.30794585e+00 -1.60311520e+00 -5.29845178e-01 -5.89725077e-01
-1.51738119e+00 1.16533995e+00 9.74867582e-01 2.46597379e-01
1.78285420e-01 5.05659819e-01 6.56268358e-01 -1.55033424e-01
-1.44574785e+00 4.66603972e-02 -5.66544831e-01 7.81121254e-01
8.65662813e-01 -1.07720651e-01 -1.61859691e-02 4.11464185e-01
-6.59327924e-01 -3.29756707e-01 1.13199604e+00 8.59758496e-01
-2.56395608e-01 -1.31552839e+00 -1.65185556e-01 3.36883754e-01
-1.50570884e-01 3.15671810e-03 -1.36265564e+00 1.00660872e+00
-9.17252481e-01 1.68600726e+00 2.94907749e-01 -3.68957043e-01
5.41872263e-01 3.92626762e-01 4.62594032e-02 -1.00867784e+00
-1.06032240e+00 -4.46078748e-01 1.38366556e+00 -4.14138809e-02
-3.75070363e-01 -5.53529203e-01 -1.16814089e+00 -7.68971801e-01
-2.41068587e-01 6.89409316e-01 3.88041228e-01 9.69338477e-01
1.61481388e-02 1.73073307e-01 9.68667984e-01 -1.11211443e+00
-7.73389220e-01 -1.67729115e+00 -2.34652981e-01 8.28335404e-01
3.17455500e-01 -2.22683609e-01 -4.67672795e-01 1.44556845e-02] | [12.833279609680176, 8.039114952087402] |
d2567f19-6969-4656-9582-08cd20818e34 | adapt-and-align-to-improve-zero-shot-sketch | 2305.05144 | null | https://arxiv.org/abs/2305.05144v2 | https://arxiv.org/pdf/2305.05144v2.pdf | Adapt and Align to Improve Zero-Shot Sketch-Based Image Retrieval | Zero-shot sketch-based image retrieval (ZS-SBIR) is challenging due to the cross-domain nature of sketches and photos, as well as the semantic gap between seen and unseen image distributions. Previous methods fine-tune pre-trained models with various side information and learning strategies to learn a compact feature space that is shared between the sketch and photo domains and bridges seen and unseen classes. However, these efforts are inadequate in adapting domains and transferring knowledge from seen to unseen classes. In this paper, we present an effective ``Adapt and Align'' approach to address the key challenges. Specifically, we insert simple and lightweight domain adapters to learn new abstract concepts of the sketch domain and improve cross-domain representation capabilities. Inspired by recent advances in image-text foundation models (e.g., CLIP) on zero-shot scenarios, we explicitly align the learned image embedding with a more semantic text embedding to achieve the desired knowledge transfer from seen to unseen classes. Extensive experiments on three benchmark datasets and two popular backbones demonstrate the superiority of our method in terms of retrieval accuracy and flexibility. | ['Xinbo Gao', 'Heng Yang', 'Nannan Wang', 'Mingrui Zhu', 'Shiyin Dong'] | 2023-05-09 | null | null | null | null | ['sketch-based-image-retrieval'] | ['computer-vision'] | [ 3.31830323e-01 -4.36066180e-01 -4.08892721e-01 -4.19992864e-01
-9.43283975e-01 -9.55060184e-01 8.90815973e-01 -2.37357318e-01
-7.61425495e-02 4.35137033e-01 3.97570044e-01 3.47295731e-01
-2.48414889e-01 -7.19606936e-01 -6.80180371e-01 -4.75639969e-01
4.31913793e-01 4.82437283e-01 2.30059385e-01 -2.51510262e-01
2.28299052e-01 4.28105563e-01 -1.77976561e+00 6.53069675e-01
4.79503214e-01 9.00369108e-01 2.34840959e-01 4.81078625e-01
-4.93948460e-01 4.80673969e-01 -3.75513494e-01 -4.90286946e-01
4.06506449e-01 -3.29570860e-01 -7.08854377e-01 2.20636979e-01
1.00460720e+00 -6.44977570e-01 -7.41496325e-01 7.20821321e-01
5.11743486e-01 1.39831632e-01 9.16037381e-01 -1.47257125e+00
-1.31689906e+00 4.48886752e-02 -3.55345160e-01 -2.04336777e-01
3.83514702e-01 -9.30092414e-04 1.04021180e+00 -1.30474889e+00
1.15091681e+00 1.24723268e+00 4.42303836e-01 9.14549172e-01
-1.40606821e+00 -8.47709775e-01 3.12097132e-01 3.19694340e-01
-1.51230252e+00 -5.58605671e-01 9.98101473e-01 -5.10981917e-01
5.65387607e-01 1.48503020e-01 4.48944509e-01 1.62744951e+00
-3.84901345e-01 1.04748333e+00 7.44346738e-01 -4.10840064e-01
2.41751194e-01 5.06882191e-01 -2.26912618e-01 3.19575608e-01
1.51118755e-01 1.30050397e-02 -9.25599039e-01 -2.78199643e-01
9.22329783e-01 4.72365022e-01 -1.99279904e-01 -1.17980206e+00
-1.13977504e+00 7.76967347e-01 3.12362373e-01 3.03102016e-01
-4.40120362e-02 1.90801039e-01 3.94238114e-01 4.36220706e-01
3.95331264e-01 3.60294759e-01 -2.28260532e-01 7.00612068e-02
-8.61044586e-01 2.24939987e-01 6.62320733e-01 1.43383563e+00
9.56323564e-01 -2.83594429e-01 -3.30579340e-01 1.19817579e+00
-4.18079831e-02 7.83677936e-01 2.84936428e-01 -8.27657461e-01
3.77967298e-01 4.98298287e-01 6.91875219e-02 -9.92037892e-01
4.49664384e-01 -9.63291973e-02 -5.17752767e-01 3.79309170e-02
2.08406776e-01 4.09486800e-01 -9.96436656e-01 1.81846333e+00
1.64751470e-01 1.59229457e-01 4.29099426e-02 8.86692524e-01
8.40421021e-01 5.21093130e-01 2.30398849e-01 4.74759221e-01
1.28308511e+00 -8.94803762e-01 -5.17683744e-01 -4.89252508e-01
2.49888226e-01 -8.26741517e-01 1.35407269e+00 -4.93973158e-02
-1.00379729e+00 -5.75870991e-01 -9.44735110e-01 -4.10381287e-01
-8.29993963e-01 1.97080836e-01 2.81833708e-01 3.21641445e-01
-8.27137589e-01 4.66675937e-01 -3.59388083e-01 -8.29273164e-01
7.21047699e-01 -1.35381306e-02 -6.55338466e-01 -7.19689012e-01
-1.02585006e+00 5.66887677e-01 2.87616879e-01 -5.21471322e-01
-1.03358424e+00 -9.39521432e-01 -8.23921263e-01 1.44870967e-01
4.43857729e-01 -8.05110872e-01 9.41730917e-01 -1.12526500e+00
-1.29998219e+00 9.77689564e-01 -3.52584152e-03 1.88924849e-01
2.99760550e-01 -2.64122635e-01 -1.85051039e-01 5.49789667e-01
6.07634373e-02 9.67302501e-01 1.29226780e+00 -1.43368065e+00
-3.93577635e-01 -4.88447696e-01 1.22179382e-01 3.11158895e-01
-9.47267234e-01 -2.64443159e-01 -7.22593725e-01 -8.58304322e-01
-9.66596827e-02 -7.53651619e-01 3.31430793e-01 7.83676088e-01
2.08721444e-01 -1.45535573e-01 1.16020036e+00 -3.97353947e-01
9.35146987e-01 -2.39336014e+00 3.35541517e-01 1.47894826e-02
1.14275768e-01 2.64512628e-01 -6.58609152e-01 1.04994690e+00
-9.59675163e-02 -1.93319216e-01 4.94972654e-02 -2.80738235e-01
2.88697779e-01 5.29663742e-01 -8.10430050e-01 1.60935909e-01
4.46717113e-01 1.07572365e+00 -1.10058165e+00 -4.55824316e-01
2.56062895e-01 6.33720756e-01 -5.94433904e-01 4.63183552e-01
-2.02498034e-01 1.84658274e-01 -4.09862489e-01 7.95622706e-01
6.63015783e-01 -3.96407664e-01 1.75589338e-01 -3.66097987e-01
4.12881523e-01 1.47138082e-03 -1.04620469e+00 2.30189919e+00
-5.28446972e-01 6.01907670e-01 -6.50461856e-03 -9.63752925e-01
9.10140812e-01 2.14670002e-01 2.74014026e-01 -8.08023334e-01
-2.90896863e-01 9.39505994e-02 -8.28814387e-01 -5.05114734e-01
4.79937166e-01 -2.30121329e-01 -9.78623852e-02 5.52008450e-01
5.91029644e-01 -2.23395869e-01 -1.69619992e-01 5.54677725e-01
9.04238105e-01 3.34787995e-01 1.15413770e-01 -4.94472571e-02
3.03173542e-01 -6.36805315e-03 1.60509616e-01 7.86998808e-01
-6.79078624e-02 9.29444015e-01 1.40389681e-01 -4.60149497e-01
-1.24596822e+00 -1.44698298e+00 -1.25325434e-02 1.57773292e+00
4.69063550e-01 -4.20770377e-01 -3.62729579e-01 -7.74028361e-01
3.48061979e-01 5.36575198e-01 -6.28883660e-01 -4.77938145e-01
-3.22197407e-01 1.51022717e-01 3.84542823e-01 7.60809541e-01
1.76092595e-01 -7.83765256e-01 -1.87982351e-01 -5.59898578e-02
-1.34893805e-01 -1.15220344e+00 -7.21135497e-01 -2.76676416e-01
-5.83853066e-01 -9.36703563e-01 -1.00466394e+00 -1.04991531e+00
7.42505789e-01 8.65569055e-01 1.20099401e+00 -7.63308182e-02
-6.51635110e-01 1.19779682e+00 -5.26674390e-01 -1.27774745e-01
-8.47035497e-02 -3.26497294e-02 -2.57447600e-01 8.87240916e-02
5.50111294e-01 -6.12682998e-01 -8.59599352e-01 3.75454277e-01
-1.29855907e+00 -6.59964755e-02 6.68696463e-01 1.05819952e+00
3.99168253e-01 -4.91179675e-01 5.58489263e-01 -6.61170542e-01
4.64125425e-01 -4.86480027e-01 -2.50984371e-01 7.27167189e-01
-4.37074482e-01 1.21906437e-01 4.12196368e-01 -8.06716204e-01
-1.08224499e+00 4.83929738e-02 4.46534336e-01 -9.54704762e-01
-1.26946718e-01 7.50998408e-02 -2.52275705e-01 -1.32968396e-01
6.00686014e-01 4.31382507e-01 1.00468926e-01 -5.47132611e-01
9.24289107e-01 7.76309252e-01 4.98683989e-01 -9.77787197e-01
1.03059733e+00 9.00515795e-01 -3.81723642e-01 -7.33273327e-01
-7.49629140e-01 -7.99108148e-01 -6.72079086e-01 -6.53448477e-02
6.54103816e-01 -1.20331836e+00 -1.64999455e-01 1.68483585e-01
-1.09782219e+00 -1.14824437e-01 -6.34899676e-01 1.28624842e-01
-6.91983521e-01 4.70196098e-01 -3.24301153e-01 -4.36278343e-01
-3.26118559e-01 -7.81677365e-01 1.67327952e+00 5.84100969e-02
-1.20881878e-01 -1.01581991e+00 1.46484420e-01 1.69721559e-01
5.27956784e-01 -2.60011572e-02 9.68195975e-01 -5.93528986e-01
-8.04765224e-01 -3.03478628e-01 -6.75723791e-01 3.86758208e-01
7.60146379e-02 -2.99378455e-01 -9.62380111e-01 -6.00402534e-01
-3.92322630e-01 -8.52124989e-01 8.57371509e-01 -2.21322894e-01
1.15287602e+00 -2.32267871e-01 -5.00976980e-01 5.04287839e-01
1.64466357e+00 -1.60579503e-01 6.94331527e-01 1.22439951e-01
4.74092871e-01 6.21846855e-01 6.26345038e-01 4.59189862e-01
3.49338681e-01 6.93267822e-01 1.18826285e-01 1.64754596e-03
-4.77624774e-01 -8.17982793e-01 2.64639825e-01 5.59943020e-01
4.35947299e-01 -1.24529392e-01 -7.03076065e-01 9.41428244e-01
-1.87400973e+00 -1.12498069e+00 8.09871495e-01 2.14959025e+00
8.91762435e-01 -4.61187899e-01 -7.90264755e-02 -3.96169662e-01
6.20896518e-01 2.63470978e-01 -6.14320278e-01 1.05399247e-02
-7.71340132e-02 3.04093808e-01 1.90506995e-01 1.50353178e-01
-9.07873094e-01 1.02906358e+00 6.03748751e+00 1.10487890e+00
-1.01511168e+00 9.41865891e-02 -2.98722996e-03 -3.91893417e-01
-4.87263680e-01 8.47365111e-02 -6.60429537e-01 2.23538995e-01
4.42330718e-01 -2.02046216e-01 6.07372999e-01 1.00020981e+00
-6.89547241e-01 4.57898647e-01 -1.54207015e+00 1.36879885e+00
5.60112178e-01 -1.51660478e+00 6.44859672e-01 -8.44425261e-02
8.29619408e-01 -2.08618671e-01 2.62880504e-01 4.42115277e-01
2.18679115e-01 -7.37367928e-01 6.10909104e-01 5.68563879e-01
1.27607977e+00 -3.57003480e-01 1.45006761e-01 5.96774258e-02
-1.11308277e+00 -1.54009983e-01 -6.24407589e-01 2.01649129e-01
-2.00395986e-01 3.40248793e-02 -6.31841600e-01 4.77011114e-01
6.58889055e-01 1.04865360e+00 -4.33319956e-01 5.68176866e-01
-2.20545605e-02 -8.90054926e-02 -7.02734441e-02 2.14711636e-01
1.15311339e-01 -1.58401862e-01 3.64320070e-01 1.10339737e+00
2.86750019e-01 6.61360249e-02 5.39756231e-02 1.12491310e+00
-2.30083838e-01 -9.94695500e-02 -1.01334691e+00 -3.82909566e-01
6.51486456e-01 1.15552938e+00 -2.01585054e-01 -5.00483751e-01
-6.52283311e-01 1.29318225e+00 5.37069499e-01 6.82223797e-01
-5.91504455e-01 -5.43516695e-01 9.28872406e-01 1.92804188e-01
6.34763598e-01 -1.11865856e-01 1.03238277e-01 -1.49957240e+00
1.00271471e-01 -7.65438318e-01 4.46927577e-01 -9.53598678e-01
-1.87409544e+00 2.36281544e-01 1.35991544e-01 -1.43178260e+00
-8.21413100e-02 -6.79547846e-01 -2.50715584e-01 7.17726648e-01
-1.78446841e+00 -1.70889151e+00 -5.04839480e-01 9.34218764e-01
7.17473686e-01 -2.16160044e-01 1.01668167e+00 4.21849728e-01
2.53582783e-02 7.68324256e-01 4.98650074e-01 2.21456379e-01
1.34364569e+00 -9.30135846e-01 2.45950848e-01 2.55092680e-01
2.75898039e-01 7.34456956e-01 3.23371261e-01 -3.74620974e-01
-1.73951805e+00 -1.08917737e+00 5.09964108e-01 -7.74825871e-01
7.67561913e-01 -8.70708287e-01 -9.27523553e-01 7.00059772e-01
1.85751006e-01 2.38438547e-01 8.61820042e-01 3.74961756e-02
-1.20468235e+00 -2.89844900e-01 -9.67718720e-01 6.50268555e-01
1.29930592e+00 -1.14144158e+00 -7.52862334e-01 3.03639024e-01
6.09424889e-01 -5.89546934e-02 -9.48221505e-01 2.18239486e-01
1.02865672e+00 -7.18030870e-01 1.49317551e+00 -9.50022817e-01
6.64744735e-01 6.98821843e-02 -5.23999035e-01 -1.15743041e+00
-3.87006730e-01 -2.46046603e-01 -1.10626267e-02 1.37377799e+00
-9.08708647e-02 -3.47297907e-01 7.82652676e-01 6.40164554e-01
3.25756162e-01 -3.82212311e-01 -7.29149342e-01 -9.48112547e-01
1.80134848e-01 -5.79975322e-02 6.75077617e-01 1.04680264e+00
3.58452424e-02 3.89701247e-01 -4.52032566e-01 -8.43931213e-02
6.65567935e-01 4.86652464e-01 1.08617926e+00 -1.11833322e+00
-1.65060103e-01 -2.39988446e-01 -5.37957728e-01 -1.19969761e+00
1.78402647e-01 -9.46735799e-01 -1.80868372e-01 -1.38462782e+00
6.47003233e-01 -3.35348666e-01 -4.67791378e-01 4.76001173e-01
-8.24079849e-03 3.39867622e-01 3.83036196e-01 4.22039866e-01
-9.18691337e-01 8.20678830e-01 1.17731225e+00 -3.62681627e-01
1.98077276e-01 -6.44054532e-01 -6.77914798e-01 2.76303262e-01
3.60767722e-01 -3.36616725e-01 -7.42031693e-01 -8.70309174e-01
5.37039898e-02 -1.44569054e-01 6.31148815e-01 -6.65493608e-01
2.39723101e-01 -2.04452485e-01 4.32492971e-01 -5.23858249e-01
7.00687349e-01 -1.07762718e+00 -7.55238608e-02 2.05161073e-03
-5.67200124e-01 -4.20675248e-01 2.36972034e-01 1.03198683e+00
-2.63619006e-01 -6.16408624e-02 6.63488448e-01 -1.45108134e-01
-9.66100216e-01 4.40005749e-01 3.21852863e-01 1.53418168e-01
1.05294561e+00 -3.89264375e-01 -7.06615090e-01 -4.91893739e-01
-4.94010150e-01 8.73247907e-02 8.89186859e-01 8.35897148e-01
8.71585131e-01 -1.72839582e+00 -4.13481504e-01 4.49656725e-01
9.27848756e-01 -3.52175474e-01 5.02515495e-01 2.71373630e-01
-6.16267622e-02 4.41603810e-01 -4.56909239e-01 -6.42820537e-01
-1.15601277e+00 8.49781215e-01 -6.44866154e-02 5.58314733e-02
-5.86851954e-01 7.16613650e-01 6.73941433e-01 -6.10378921e-01
3.89193892e-01 1.43485755e-01 3.61615896e-01 2.73034181e-02
7.05435216e-01 1.37070745e-01 -2.92555749e-01 -2.45531365e-01
-2.65071779e-01 8.72177064e-01 -3.19078147e-01 2.02223063e-02
1.38648891e+00 -2.65715659e-01 1.23001933e-02 4.20944393e-01
1.55023408e+00 -3.45199436e-01 -1.41740811e+00 -7.94023573e-01
-1.01443239e-01 -9.56122100e-01 -2.19017968e-01 -8.05916190e-01
-7.11099327e-01 1.13505709e+00 5.92679560e-01 -2.20969632e-01
1.03335571e+00 3.97124439e-01 8.63429189e-01 4.21190649e-01
3.10142159e-01 -1.30755699e+00 7.39342928e-01 2.03178063e-01
1.06273210e+00 -1.32893169e+00 -5.86828068e-02 -2.24045828e-01
-6.31147027e-01 1.12857759e+00 5.95444620e-01 -1.81897208e-01
5.20786881e-01 -2.09018394e-01 -1.49237186e-01 -2.40490213e-01
-8.75532508e-01 -1.79975033e-01 3.82485181e-01 8.19679677e-01
3.60516198e-02 -3.14502537e-01 1.47652879e-01 3.95412743e-01
4.93629754e-01 2.76395679e-01 6.27027899e-02 1.12620592e+00
-2.61395782e-01 -1.20883262e+00 -2.69164234e-01 1.46396905e-01
1.21520169e-01 7.14595392e-02 -6.74855411e-01 8.85758340e-01
-1.49510980e-01 5.23369730e-01 1.33152306e-01 -3.00524563e-01
3.36893886e-01 3.33013147e-01 7.01301098e-01 -7.76230037e-01
8.44380073e-03 -8.77709240e-02 -3.17152858e-01 -6.33030713e-01
-4.06954050e-01 -4.03532445e-01 -6.33332729e-01 -1.40981361e-01
-1.89382315e-01 -1.55843899e-01 6.17860377e-01 6.50441885e-01
8.34146202e-01 6.38654828e-02 5.01286983e-01 -8.78859997e-01
-9.08953369e-01 -5.79732478e-01 -6.04495287e-01 1.01359010e+00
1.83950752e-01 -8.53258431e-01 -2.39867613e-01 2.20499530e-01] | [11.561333656311035, 0.6827563047409058] |
9fa456dd-be6b-47c7-b72a-6a57e0e7af8c | automatic-photo-adjustment-using-deep-neural | 1412.7725 | null | http://arxiv.org/abs/1412.7725v2 | http://arxiv.org/pdf/1412.7725v2.pdf | Automatic Photo Adjustment Using Deep Neural Networks | Photo retouching enables photographers to invoke dramatic visual impressions
by artistically enhancing their photos through stylistic color and tone
adjustments. However, it is also a time-consuming and challenging task that
requires advanced skills beyond the abilities of casual photographers. Using an
automated algorithm is an appealing alternative to manual work but such an
algorithm faces many hurdles. Many photographic styles rely on subtle
adjustments that depend on the image content and even its semantics. Further,
these adjustments are often spatially varying. Because of these
characteristics, existing automatic algorithms are still limited and cover only
a subset of these challenges. Recently, deep machine learning has shown unique
abilities to address hard problems that resisted machine algorithms for long.
This motivated us to explore the use of deep learning in the context of photo
editing. In this paper, we explain how to formulate the automatic photo
adjustment problem in a way suitable for this approach. We also introduce an
image descriptor that accounts for the local semantics of an image. Our
experiments demonstrate that our deep learning formulation applied using these
descriptors successfully capture sophisticated photographic styles. In
particular and unlike previous techniques, it can model local adjustments that
depend on the image semantics. We show on several examples that this yields
results that are qualitatively and quantitatively better than previous work. | ['Hao Zhang', 'Zhicheng Yan', 'Yizhou Yu', 'Sylvain Paris', 'Baoyuan Wang'] | 2014-12-24 | null | null | null | null | ['photo-retouching'] | ['computer-vision'] | [ 4.65152681e-01 -2.21349508e-01 8.40989202e-02 -4.36801523e-01
-1.60274267e-01 -8.01726341e-01 8.05536926e-01 -1.11033417e-01
-3.83226156e-01 5.69671988e-01 5.30140400e-02 -4.50998247e-02
1.68993603e-03 -5.83714008e-01 -6.57975614e-01 -5.33511758e-01
4.58324492e-01 2.53390998e-01 1.44212052e-01 -5.01355946e-01
5.36837816e-01 1.04467356e+00 -1.73727095e+00 1.30339414e-01
4.64383662e-01 6.70551419e-01 1.09140679e-01 6.25983596e-01
-3.96092653e-01 5.63240528e-01 -6.44399166e-01 -5.77318907e-01
5.82826734e-01 -3.20232183e-01 -5.58574200e-01 5.92713773e-01
1.09712577e+00 -3.57619762e-01 -1.06810004e-01 1.14045250e+00
3.87282193e-01 1.63746327e-01 5.18323958e-01 -1.02661633e+00
-1.08231509e+00 1.16943002e-01 -6.52817667e-01 -6.25129268e-02
3.63166392e-01 1.79450899e-01 7.16249585e-01 -6.90099061e-01
7.23570764e-01 1.09272814e+00 9.14311409e-01 5.34317732e-01
-1.46769226e+00 -3.80577236e-01 9.23939869e-02 1.33198544e-01
-1.08088970e+00 -4.02775317e-01 9.55536664e-01 -5.14764309e-01
6.06094956e-01 4.35495377e-01 8.41404736e-01 8.70742142e-01
1.90590590e-01 4.37601864e-01 1.57513475e+00 -6.59626245e-01
1.87302917e-01 3.09809148e-01 -2.62722373e-01 5.68415523e-01
-5.92909902e-02 8.07731785e-03 -2.30622366e-01 2.86530435e-01
1.13757610e+00 2.36108199e-01 -2.82461286e-01 -6.52993858e-01
-1.05671394e+00 7.23828852e-01 5.14498651e-01 4.36591744e-01
-1.85833320e-01 3.24558258e-01 1.08138599e-01 4.54743445e-01
3.40210050e-01 9.13145661e-01 -2.44457573e-01 -6.66270852e-02
-1.12328053e+00 8.62825289e-02 6.20942712e-01 7.20291257e-01
1.06853259e+00 3.54695842e-02 1.38944000e-01 9.14707661e-01
-3.23889256e-01 1.59785792e-01 4.03357387e-01 -1.08473694e+00
-1.61267117e-01 6.86897278e-01 2.51900315e-01 -1.33786035e+00
-3.15330446e-01 -1.74367607e-01 -9.54092383e-01 1.02107537e+00
6.02545142e-01 1.66722566e-01 -1.04012060e+00 1.52211189e+00
2.53636330e-01 -3.16000789e-01 -3.75191748e-01 1.02505398e+00
3.28772247e-01 3.49379778e-01 3.66258919e-02 -1.67050250e-02
1.11148751e+00 -8.68530393e-01 -8.60316277e-01 -1.84850454e-01
1.26186505e-01 -1.28983581e+00 1.49848819e+00 4.99640226e-01
-1.23690462e+00 -6.30014002e-01 -1.13280344e+00 -4.73496675e-01
-5.41368961e-01 3.26214254e-01 6.29520535e-01 6.41943574e-01
-1.56151628e+00 8.99902105e-01 -3.09130281e-01 -7.81504691e-01
2.61103451e-01 3.83789122e-01 -4.72733974e-01 3.15209851e-02
-8.73103619e-01 1.13988507e+00 -1.76625371e-01 1.35433108e-01
-5.61456010e-03 -6.49337649e-01 -7.29517579e-01 -6.64898157e-02
2.89123803e-01 -7.10875988e-01 1.20782864e+00 -1.69912457e+00
-1.95518482e+00 1.23422909e+00 7.30424374e-03 -2.57800549e-01
1.00460458e+00 -6.78986087e-02 -1.99975982e-01 1.37224138e-01
-2.14869201e-01 7.61451900e-01 1.28033257e+00 -1.26797140e+00
-3.62207949e-01 -2.14888677e-01 5.11080801e-01 1.91049024e-01
-4.42265123e-01 -2.12651193e-02 -5.40789664e-01 -8.50522876e-01
-1.42941818e-01 -1.00438356e+00 -2.38860756e-01 6.25725448e-01
-1.71757743e-01 8.30114335e-02 7.38032699e-01 -4.55591738e-01
9.24883485e-01 -2.11477709e+00 1.57999128e-01 8.86386633e-02
2.62618840e-01 5.58948994e-01 -7.97752738e-02 6.23351812e-01
-1.22217290e-01 2.11732686e-01 -2.44094312e-01 -4.05499279e-01
8.54256600e-02 1.59486886e-02 -2.30736285e-01 4.80465651e-01
1.66725814e-01 7.94563174e-01 -6.75734401e-01 -3.04031312e-01
7.10543036e-01 5.33051372e-01 -3.15594435e-01 2.22470332e-02
-4.43490073e-02 1.75392419e-01 4.74950001e-02 4.62213397e-01
7.11892962e-01 -5.16709536e-02 1.16872482e-01 -5.15596747e-01
-3.40314060e-01 -2.71769822e-01 -1.11842752e+00 1.61770809e+00
-6.58484697e-01 1.05368435e+00 -4.86091599e-02 -8.06150794e-01
9.20455098e-01 -2.43692100e-01 3.93760651e-01 -6.22569323e-01
-3.14568728e-02 1.60927624e-01 -2.82249391e-01 -5.15749037e-01
6.78237200e-01 -5.12228012e-01 1.87032521e-01 5.62932730e-01
-4.16128397e-01 -6.49223685e-01 5.06621879e-03 3.49815004e-02
8.17007244e-01 2.12856889e-01 5.52398086e-01 -1.80288389e-01
4.52003419e-01 -1.15016766e-01 1.88545004e-01 6.18461192e-01
-1.17011294e-01 1.06282282e+00 1.55263141e-01 -9.24398482e-01
-1.28983641e+00 -8.30932319e-01 2.64679715e-02 8.87256980e-01
2.26134554e-01 -8.90345201e-02 -8.25183868e-01 -3.79530311e-01
1.15373321e-02 4.48003560e-01 -8.37251246e-01 -6.35854602e-02
-5.13691425e-01 -2.55460620e-01 1.07726872e-01 4.55339193e-01
5.11377811e-01 -1.04528105e+00 -6.20767474e-01 9.24745202e-03
3.18437517e-01 -1.21410251e+00 -6.55969143e-01 -3.61775398e-01
-6.38642669e-01 -9.67196703e-01 -9.78311360e-01 -7.49005735e-01
7.12880015e-01 4.97164637e-01 1.01834393e+00 2.50532240e-01
-5.24257183e-01 7.12325156e-01 -3.06502640e-01 -3.43181670e-01
-5.14255285e-01 -5.62773086e-03 1.57516692e-02 2.80824482e-01
2.62712151e-01 -7.69558966e-01 -7.77342558e-01 2.77033091e-01
-1.32448876e+00 2.46161371e-01 8.31952155e-01 6.03676975e-01
4.29059207e-01 -1.61036894e-01 1.94394231e-01 -1.03038442e+00
5.75402260e-01 2.65393227e-01 -5.55924654e-01 2.81456292e-01
-4.05769944e-01 2.67037377e-03 6.28297925e-01 -5.75021148e-01
-1.08227229e+00 3.79452616e-01 1.86153889e-01 -4.35803324e-01
-1.69291317e-01 1.09860085e-01 3.55272740e-03 -6.32575870e-01
6.06480598e-01 2.56507117e-02 1.95944250e-01 -4.39151376e-01
4.76867348e-01 5.54329395e-01 4.90326047e-01 -1.76538289e-01
1.04513323e+00 7.83882499e-01 8.19312856e-02 -1.07596028e+00
-7.28984535e-01 -1.98898077e-01 -1.00362277e+00 -4.64845181e-01
7.68224478e-01 -4.84039426e-01 -4.46781337e-01 6.26108289e-01
-1.03971541e+00 -4.90920186e-01 -5.21564603e-01 8.06835294e-02
-6.62322760e-01 6.42354608e-01 -3.92144561e-01 -5.31135798e-01
2.46387515e-02 -1.00651348e+00 1.08319855e+00 2.50452518e-01
-4.26551461e-01 -1.24887025e+00 3.63678485e-02 3.43693644e-01
6.48647785e-01 4.44215238e-01 7.36978650e-01 1.61113560e-01
-6.00203216e-01 -4.13970709e-01 -3.88772607e-01 4.75510448e-01
4.77549046e-01 2.96475977e-01 -1.06039262e+00 -7.16542453e-02
-1.09670863e-01 -1.71882689e-01 6.71615362e-01 2.13113219e-01
1.31669271e+00 -2.06533387e-01 6.74934313e-02 7.51134038e-01
1.56631160e+00 -8.97831097e-02 8.95044327e-01 5.91947317e-01
8.61660242e-01 6.61445498e-01 3.66159171e-01 2.20293611e-01
1.06611885e-01 1.02662468e+00 2.53862292e-01 -5.77612519e-01
-4.65519637e-01 -6.72772154e-02 2.61181235e-01 3.97509754e-01
-2.54807830e-01 -4.06147018e-02 -5.91030419e-01 3.98370266e-01
-1.59306204e+00 -9.63027477e-01 -4.47747000e-02 2.15833497e+00
9.59018588e-01 -1.84638381e-01 -3.04059125e-03 1.69891402e-01
7.91754007e-01 2.03401536e-01 -4.20013547e-01 -8.96839499e-01
-3.05387676e-01 3.18881869e-01 5.60106397e-01 4.79060680e-01
-1.10407341e+00 1.07818818e+00 6.16093016e+00 5.74453652e-01
-1.32289290e+00 -2.59878159e-01 5.33922493e-01 6.57206587e-03
-3.24487507e-01 -8.55119675e-02 -3.61569643e-01 4.11036074e-01
2.45127112e-01 1.37440443e-01 7.22573638e-01 5.47009587e-01
3.15519154e-01 -1.90136552e-01 -1.24532342e+00 1.24450409e+00
3.56697857e-01 -1.37206101e+00 2.84956098e-01 -1.92570128e-02
9.86791313e-01 -5.20002663e-01 3.54050696e-01 -2.28379041e-01
1.86981767e-01 -9.97552276e-01 6.41330481e-01 7.46682644e-01
9.84453440e-01 -3.59530389e-01 3.89938414e-01 -1.81488007e-01
-7.41698086e-01 -1.31017491e-02 -3.12047660e-01 -4.35096622e-01
1.22146457e-01 4.89399582e-01 -6.25931561e-01 1.51239410e-02
5.77625990e-01 6.85107172e-01 -8.47389340e-01 1.08244658e+00
-1.47889405e-01 -1.04192741e-01 -1.44640401e-01 -2.52597723e-02
2.65489429e-01 -3.60034883e-01 3.43137443e-01 1.17526770e+00
4.37633783e-01 6.51910454e-02 -1.04452789e-01 7.90974021e-01
-1.03432924e-01 1.09840252e-01 -7.90944338e-01 4.55674604e-02
1.33067653e-01 1.39359808e+00 -7.46721506e-01 -1.71399370e-01
-2.73913324e-01 1.43641949e+00 3.54891300e-01 3.93558055e-01
-5.08052349e-01 -2.63959825e-01 7.83142090e-01 3.42346549e-01
1.89090520e-01 -3.85483205e-01 -3.73611629e-01 -1.11982632e+00
9.73790139e-02 -8.87166142e-01 -8.14005136e-02 -1.15395439e+00
-1.36472726e+00 4.58753854e-01 -2.34653234e-01 -1.25912547e+00
-8.96946341e-02 -7.85861969e-01 -5.72388887e-01 6.68029666e-01
-1.48108244e+00 -1.35259461e+00 -6.44202352e-01 5.69092095e-01
6.61251545e-01 6.39127642e-02 6.21531665e-01 1.96976900e-01
-2.10661203e-01 4.97760534e-01 8.63241851e-02 -6.27319366e-02
1.05248106e+00 -1.47721636e+00 3.64812493e-01 7.75723696e-01
3.32190514e-01 4.79193509e-01 9.41432834e-01 -1.54544294e-01
-1.22250867e+00 -7.85826206e-01 8.73653829e-01 -4.89908993e-01
5.67084551e-01 -2.38353983e-01 -7.60190725e-01 5.99550009e-01
4.26180512e-01 -3.34113657e-01 4.52225149e-01 -1.61454022e-01
-2.20402092e-01 -2.50822544e-01 -9.39483404e-01 9.20950532e-01
9.82315004e-01 -6.27417207e-01 -5.33806920e-01 3.68477106e-01
2.39899486e-01 -3.87006044e-01 -5.06236911e-01 3.32442597e-02
7.28091717e-01 -1.29341459e+00 1.02444005e+00 -3.89064729e-01
3.95577312e-01 -2.99695492e-01 1.21414915e-01 -1.36942220e+00
-3.39720607e-01 -8.59901190e-01 5.69120586e-01 1.28815126e+00
-7.24459160e-03 -5.05406082e-01 4.72870857e-01 9.84633684e-01
7.42387027e-02 -3.46761942e-01 -4.88830626e-01 -5.94216943e-01
-5.60391955e-02 -1.76817134e-01 5.21618247e-01 1.12626743e+00
-3.70663047e-01 4.41178121e-02 -5.10621548e-01 -9.21221226e-02
4.48820025e-01 3.05228263e-01 1.09823632e+00 -1.17839766e+00
-8.54314640e-02 -7.46641397e-01 -7.17083931e-01 -6.74997211e-01
1.36361778e-01 -3.57652307e-01 -9.26210284e-02 -1.32934546e+00
3.92877236e-02 -3.31983060e-01 4.59014028e-02 3.42521250e-01
-1.06446087e-01 8.55785728e-01 4.31027830e-01 2.03645259e-01
-2.60166496e-01 2.52148151e-01 1.40721631e+00 -1.99133456e-01
-2.57736981e-01 -1.81478076e-02 -6.64976776e-01 9.62097406e-01
9.94318128e-01 1.18080199e-01 -3.30944657e-01 -5.83081782e-01
4.05656666e-01 -5.81423879e-01 5.35582423e-01 -9.91712153e-01
1.67705879e-01 -2.88700700e-01 5.22844553e-01 -2.34730598e-02
5.32711983e-01 -8.99122596e-01 1.61837250e-01 1.81776121e-01
-4.82462049e-01 2.18664885e-01 2.20278233e-01 5.90871990e-01
-1.50168881e-01 -1.72044128e-01 1.00826597e+00 -3.00109446e-01
-9.76176739e-01 1.67347848e-01 -4.31553632e-01 -3.10411155e-01
1.01939940e+00 -5.52805126e-01 -3.23266461e-02 -7.76797414e-01
-8.36010218e-01 -3.85851949e-01 1.14659607e+00 3.95987540e-01
5.38075566e-01 -1.27705276e+00 -2.73309946e-01 1.86314270e-01
4.31714170e-02 -3.90644670e-01 3.65248770e-01 7.12531090e-01
-1.05266500e+00 4.59612980e-02 -6.45446420e-01 -4.13366497e-01
-1.41638732e+00 6.70930564e-01 4.13844734e-01 2.12470174e-01
-6.76694214e-01 5.95315874e-01 3.43890488e-01 -1.48518786e-01
9.61799249e-02 -4.53565717e-01 -2.30377257e-01 9.20645222e-02
5.47693253e-01 1.98964313e-01 -2.22307593e-02 -5.74721932e-01
-2.41770782e-02 1.13630426e+00 1.45293865e-03 -1.31494269e-01
1.31846464e+00 -5.39275825e-01 -1.67685255e-01 3.16869020e-01
1.10619092e+00 1.93179563e-01 -1.46688437e+00 -2.00484663e-01
-2.25411370e-01 -9.36375916e-01 3.51395682e-02 -8.43881726e-01
-1.01023149e+00 1.12405884e+00 5.94534278e-01 3.81005168e-01
1.38004410e+00 -3.02386403e-01 6.44975424e-01 2.91419744e-01
1.75630122e-01 -1.23485029e+00 2.18508229e-01 9.64376926e-02
1.17443419e+00 -1.34456003e+00 2.63657212e-01 -3.45835626e-01
-6.13882899e-01 1.46610975e+00 3.77762765e-01 -2.57998705e-01
3.02952468e-01 -3.20788100e-02 5.37693143e-01 -1.25243783e-01
-1.16855972e-01 -2.59013176e-01 3.62331569e-01 6.52025282e-01
3.38321507e-01 -3.94287556e-02 -2.65114605e-01 -2.16335744e-01
-2.58357167e-01 8.14041048e-02 8.39257121e-01 5.58642924e-01
-2.95196950e-01 -1.29264379e+00 -4.00530517e-01 6.26998693e-02
-3.18972498e-01 1.68453213e-02 -9.54254329e-01 1.10428393e+00
1.42091855e-01 6.45563185e-01 1.51215028e-02 -2.79540956e-01
2.42845893e-01 -2.00157732e-01 7.03953743e-01 -4.13195431e-01
-4.89240885e-01 -1.68766707e-01 -2.11703837e-01 -6.70526266e-01
-7.56279349e-01 -6.33904517e-01 -6.32737637e-01 -5.51705599e-01
7.17126429e-02 -4.42275703e-01 8.07320476e-01 8.29576850e-01
2.54699558e-01 2.73320079e-01 5.86500108e-01 -1.30475450e+00
-4.32409883e-01 -6.22196198e-01 -5.34127355e-01 7.99109221e-01
3.67383301e-01 -6.49166644e-01 -3.05721104e-01 4.18122828e-01] | [11.41006851196289, -0.6946847438812256] |
aeef3325-f14b-421b-96e9-66caa212c56d | physically-primed-deep-neural-networks-for | 2209.00462 | null | https://arxiv.org/abs/2209.00462v1 | https://arxiv.org/pdf/2209.00462v1.pdf | Physically-primed deep-neural-networks for generalized undersampled MRI reconstruction | A plethora of deep-neural-networks (DNN) based methods were proposed over the past few years to address the challenging ill-posed inverse problem of MRI reconstruction from undersampled "k-space" (Fourier domain) data. However, instability against variations in the acquisition process and the anatomical distribution, indicates a poor generalization of the relevant physical models by the DNN architectures compared to their classical counterparts. The poor generalization effectively precludes DNN applicability for undersampled MRI reconstruction in the clinical setting. We improve the generalization capacity of DNN methods for undersampled MRI reconstruction by introducing a physically-primed DNN architecture and training approach. Our architecture encodes the undersampling mask in addition to the observed data in the model architecture and employs an appropriate training approach that uses data generated with various undersampling masks to encourage the model to generalize the undersampled MRI reconstruction problem. We demonstrated the added-value of our approach through extensive experimentation with the publicly available Fast-MRI dataset. Our physically-primed approach achieved an enhanced generalization capacity which resulted in significantly improved robustness against variations in the acquisition process and in the anatomical distribution, especially in pathological regions, compared to both vanilla DNN methods and DNN trained with undersampling mask augmentation. Trained models and code to replicate our experiments will become available for research purposes upon acceptance. | ['Moti Freiman', 'Nitzan Avidan'] | 2022-08-31 | null | null | null | null | ['mri-reconstruction'] | ['computer-vision'] | [ 4.77852315e-01 2.30958924e-01 2.23268956e-01 -4.08532679e-01
-4.29478258e-01 -1.39507458e-01 5.21254122e-01 -4.64383453e-01
-6.10114455e-01 7.26313889e-01 2.14610115e-01 -3.62257361e-01
-6.24391377e-01 -3.97085637e-01 -7.74648130e-01 -8.30068588e-01
-4.64666307e-01 4.67582822e-01 1.48647711e-01 -3.28560442e-01
-8.04634914e-02 8.40551257e-01 -1.12094784e+00 2.49612123e-01
5.73859751e-01 9.11726773e-01 4.89557922e-01 4.79149193e-01
2.95801967e-01 4.76864517e-01 -2.97468394e-01 2.46866375e-01
4.29279983e-01 -2.96301723e-01 -8.76937509e-01 -1.32271752e-01
3.32860023e-01 -6.02445602e-01 -7.16509461e-01 9.07125354e-01
8.20256352e-01 2.92574257e-01 5.12804806e-01 -5.81052780e-01
-6.84831977e-01 4.71370459e-01 -2.89910257e-01 8.29270482e-01
-6.59425333e-02 3.61349016e-01 1.42280817e-01 -9.17151749e-01
6.81940198e-01 7.37249911e-01 1.04723120e+00 8.90727043e-01
-1.36175513e+00 -5.04152000e-01 -2.87609041e-01 -8.03659577e-03
-1.09012246e+00 -2.87024587e-01 6.80088341e-01 -3.59245181e-01
8.39692652e-01 2.66206920e-01 5.71361601e-01 1.32186317e+00
4.59496111e-01 3.84962916e-01 1.39798009e+00 -2.09086239e-01
2.79115438e-01 -1.90177172e-01 -1.94734037e-01 1.16876163e-01
1.10887676e-01 5.25501728e-01 -6.78755939e-02 -1.76393494e-01
1.33866143e+00 5.65716773e-02 -5.42643666e-01 -4.85397458e-01
-1.35386682e+00 6.66398346e-01 7.64925599e-01 5.11333406e-01
-6.50978088e-01 4.98570465e-02 4.69220787e-01 8.63105953e-02
4.51908022e-01 8.04618478e-01 -4.87291127e-01 5.20400368e-02
-1.10074234e+00 2.70954758e-01 3.92513394e-01 4.37376797e-01
1.05253220e-01 3.60576093e-01 9.95185077e-02 9.28040504e-01
-9.39504988e-03 3.16481054e-01 9.59758461e-01 -9.43573236e-01
2.20041439e-01 2.81228665e-02 -2.04267174e-01 -7.77512133e-01
-9.93605554e-01 -8.53983521e-01 -1.18248892e+00 2.65921205e-01
3.90955061e-01 -2.56629527e-01 -1.16401541e+00 1.68120241e+00
5.08541226e-01 2.18764171e-01 3.77235524e-02 1.32008994e+00
7.63331950e-01 2.57294834e-01 5.11874855e-02 1.32236294e-02
1.21585059e+00 -4.13857907e-01 -7.68391967e-01 -2.44097352e-01
6.01966977e-01 -4.98100609e-01 9.40408051e-01 4.68085855e-01
-1.15789235e+00 -4.01644260e-01 -1.13524985e+00 2.64739931e-01
-6.74674846e-03 -2.82284915e-01 4.53259081e-01 5.03163278e-01
-1.29825711e+00 1.07696557e+00 -1.10290599e+00 -1.08332053e-01
6.58305049e-01 6.22566581e-01 -5.40380597e-01 -1.92623466e-01
-1.45675576e+00 1.13830137e+00 4.38049704e-01 4.28615332e-01
-8.92902553e-01 -1.20032215e+00 -8.52128446e-01 -2.64316827e-01
1.20183788e-01 -6.04162872e-01 1.14915037e+00 -8.61451566e-01
-1.39299893e+00 6.75387919e-01 5.14073908e-01 -6.69550061e-01
6.68441772e-01 -3.55484225e-02 -5.80973327e-01 4.54506755e-01
-2.23689571e-01 6.11127853e-01 7.75710166e-01 -1.07279205e+00
4.53310579e-01 -4.43865776e-01 -1.22164458e-01 1.09113790e-01
1.54502049e-01 -2.02122733e-01 2.03961268e-01 -9.57961679e-01
4.34191525e-01 -1.07309115e+00 -6.34612679e-01 1.29153669e-01
-1.89101741e-01 6.63983107e-01 7.64093399e-01 -8.66944611e-01
6.92407727e-01 -2.07215071e+00 -6.29862621e-02 8.07530582e-02
3.79472762e-01 5.03163695e-01 -1.16897374e-01 8.51043612e-02
-8.28521609e-01 -2.22070485e-01 -5.91378629e-01 8.81951749e-02
-5.12497604e-01 4.89856780e-01 -7.15081692e-02 7.26140022e-01
2.61946142e-01 8.99551511e-01 -1.02111697e+00 -1.19094752e-01
2.98709303e-01 7.78478384e-01 -6.52451694e-01 1.47480026e-01
1.68848753e-01 1.03434360e+00 -2.12878957e-01 8.31930265e-02
1.04659331e+00 -1.39813244e-01 1.39913186e-01 -3.57460916e-01
1.29664034e-01 7.34278411e-02 -8.78164291e-01 1.94925249e+00
-6.24976099e-01 3.83137494e-01 3.50118637e-01 -1.40339637e+00
6.96382165e-01 5.30541122e-01 6.83274448e-01 -8.07141304e-01
1.13226905e-01 3.42583895e-01 5.55338681e-01 -7.78670728e-01
2.21304163e-01 -1.02179050e+00 2.45557562e-01 4.29489642e-01
2.69292027e-01 -3.55262369e-01 -3.90809566e-01 2.31179199e-03
1.02047956e+00 1.76582024e-01 7.20644696e-03 -8.07582319e-01
4.43225503e-01 -1.62638888e-01 1.24699779e-01 9.47476506e-01
-3.26883465e-01 1.00486672e+00 1.83565795e-01 -7.97218025e-01
-1.18941951e+00 -1.13185501e+00 -8.71297836e-01 4.56980973e-01
-1.62656799e-01 3.61644447e-01 -7.81148434e-01 -5.01704335e-01
-2.01010063e-01 6.16622567e-01 -8.32594216e-01 -4.43034887e-01
-8.85352314e-01 -1.17463958e+00 7.04293609e-01 6.72116041e-01
3.64329994e-01 -1.18834424e+00 -6.57261968e-01 4.97387379e-01
-1.83782101e-01 -1.20470035e+00 -5.11903092e-02 4.54158157e-01
-1.20941794e+00 -8.31398726e-01 -1.04568911e+00 -4.57808137e-01
6.60271823e-01 -2.53186256e-01 8.83363783e-01 1.54775716e-02
-6.43810749e-01 1.99653387e-01 -3.41182537e-02 -1.25259429e-01
-7.48453856e-01 -2.16327980e-01 4.22586441e-01 -2.76442796e-01
-1.61918789e-01 -8.09555173e-01 -1.03430724e+00 1.91296577e-01
-1.44174492e+00 -2.35712882e-02 5.27402937e-01 1.08382428e+00
3.95092487e-01 -1.87521726e-01 7.83425033e-01 -6.95752263e-01
6.42120600e-01 -6.64169908e-01 -2.54345864e-01 -2.93557793e-01
-2.32595190e-01 3.67587447e-01 8.06975961e-01 -7.11450756e-01
-9.60146785e-01 -1.62699223e-01 -6.57493293e-01 -3.70639801e-01
-2.72977293e-01 4.70200509e-01 1.48065895e-01 -3.94491971e-01
9.90777314e-01 2.66748995e-01 6.07253730e-01 -4.51203972e-01
1.13007464e-01 3.17878962e-01 7.79434919e-01 -6.04180992e-01
4.30416495e-01 7.38066971e-01 2.57026762e-01 -7.93607175e-01
-6.18388653e-01 3.98429669e-02 -6.52630508e-01 -2.30703145e-01
7.83949554e-01 -5.64090848e-01 -4.17805284e-01 5.57027876e-01
-8.16514671e-01 -5.49380243e-01 -6.21876478e-01 8.08056533e-01
-6.56955302e-01 5.16177237e-01 -8.07564139e-01 -3.82036418e-01
-3.42059106e-01 -1.46779478e+00 7.92302132e-01 -1.04336888e-01
-2.53921658e-01 -1.21585464e+00 -1.82714034e-02 1.14519000e-01
9.63176668e-01 4.80309546e-01 1.05962265e+00 -6.82903230e-01
-1.27059877e-01 -6.55706078e-02 -1.51936457e-01 6.56637192e-01
5.75630702e-02 -7.98750043e-01 -1.00465345e+00 -2.23034620e-01
6.59276962e-01 -3.14539045e-01 6.33973777e-01 8.72543156e-01
1.58657539e+00 -1.47541538e-01 -4.90551367e-02 8.09130728e-01
1.37324667e+00 1.44467428e-02 7.15567112e-01 2.66086161e-01
4.27974075e-01 4.16111618e-01 -5.42760305e-02 1.37958616e-01
-1.73153430e-01 6.78791285e-01 5.68585813e-01 -3.57433438e-01
-3.54414880e-01 1.24550551e-01 -2.18142316e-01 7.11324155e-01
-9.90636349e-02 2.43271783e-01 -1.10197747e+00 4.86596286e-01
-1.34027278e+00 -6.36796057e-01 -3.70554216e-02 2.07487416e+00
8.08414221e-01 6.30492046e-02 -1.68220580e-01 5.56454211e-02
4.48195219e-01 7.02166557e-02 -7.37777591e-01 -3.12442034e-01
-3.35236229e-02 5.33340693e-01 4.73296881e-01 4.63471681e-01
-9.20788765e-01 2.95639366e-01 7.05072689e+00 7.20799148e-01
-1.37285662e+00 4.37613577e-01 7.35280812e-01 -2.41475001e-01
-7.05698654e-02 -3.53360415e-01 -1.74850121e-01 3.88376832e-01
1.03692758e+00 2.67919153e-01 4.78581727e-01 5.05662739e-01
3.60331357e-01 -4.53819819e-02 -9.49127078e-01 7.07396865e-01
-2.03671634e-01 -1.54557383e+00 -1.93782061e-01 2.41758395e-02
5.98004103e-01 4.44742441e-01 1.87228680e-01 1.82770386e-01
-5.70953786e-02 -1.17529607e+00 4.14289117e-01 5.27446985e-01
1.18367445e+00 -4.21371847e-01 6.76260114e-01 3.98293644e-01
-4.26722527e-01 3.16912606e-02 -2.26448953e-01 -8.10614079e-02
2.28565097e-01 7.00223207e-01 -9.53982651e-01 5.56825101e-01
7.27440119e-01 2.34777078e-01 -1.78979948e-01 7.07407773e-01
2.18776867e-01 4.66367990e-01 -3.59684497e-01 4.77457821e-01
4.71972466e-01 -6.32732883e-02 7.02886760e-01 1.17486274e+00
1.09047920e-01 2.95889169e-01 -1.28243849e-01 1.06420493e+00
2.53465563e-01 -3.07639837e-01 -6.34734154e-01 4.13042426e-01
-2.37562999e-01 1.14379501e+00 -7.55430222e-01 -2.32491672e-01
2.80682463e-02 7.12083399e-01 1.24865294e-01 5.83314121e-01
-7.44705141e-01 -7.89698437e-02 3.60481948e-01 6.90162838e-01
1.90893710e-01 -2.41551384e-01 -2.80731767e-01 -1.00217199e+00
1.16920143e-01 -8.45187485e-01 1.60572141e-01 -9.03551817e-01
-1.31038272e+00 1.10272372e+00 3.92563671e-01 -1.04065967e+00
-3.38079691e-01 -8.17757308e-01 -3.05864990e-01 1.12252223e+00
-1.21536732e+00 -6.46146417e-01 -1.28958493e-01 4.40313280e-01
1.33055896e-01 1.82865098e-01 8.38292122e-01 4.34477925e-01
-5.13842069e-02 1.33708790e-01 1.95113987e-01 4.61602286e-02
2.61959046e-01 -1.05403984e+00 2.88059890e-01 6.79141462e-01
-5.24791539e-01 6.99757397e-01 9.35433626e-01 -5.54112196e-01
-1.03546524e+00 -8.87947738e-01 2.16801375e-01 -1.79427728e-01
8.43667209e-01 -4.12350416e-01 -1.22389710e+00 5.33103466e-01
1.57892853e-02 6.21792674e-01 4.96985316e-01 -4.51859564e-01
1.07642390e-01 4.55134735e-02 -1.57176459e+00 5.16202211e-01
1.03678989e+00 -4.59562898e-01 -7.92777896e-01 4.76746529e-01
5.49928963e-01 -7.63789952e-01 -1.03183675e+00 7.46530235e-01
4.28761125e-01 -8.07672739e-01 1.29375744e+00 -7.35818207e-01
5.77972591e-01 6.28709495e-02 1.37904668e-02 -1.61843741e+00
-3.03509116e-01 -3.99339765e-01 -1.17554598e-01 2.37466469e-01
2.00276077e-01 -7.98325062e-01 6.99777067e-01 7.27720976e-01
-3.45341176e-01 -1.04625928e+00 -1.33306503e+00 -8.02475929e-01
5.22693634e-01 -6.80592537e-01 4.48433876e-01 8.89581680e-01
-1.15526833e-01 -2.76322544e-01 -2.53593564e-01 2.44577885e-01
6.53517008e-01 -4.39860135e-01 -2.61882339e-02 -8.96458864e-01
-3.80367786e-01 -2.62250930e-01 -5.36066234e-01 -8.83418739e-01
1.48664370e-01 -1.07760572e+00 -4.47134562e-02 -1.28526735e+00
-4.26783040e-02 -7.13098645e-01 -3.50551933e-01 1.43700510e-01
1.70824099e-02 5.09390652e-01 -3.27840750e-03 1.39448002e-01
1.44837961e-01 3.47618163e-01 1.71279275e+00 1.56000763e-01
-7.46776685e-02 -5.19288667e-02 -5.29970407e-01 6.87559664e-01
6.11304045e-01 -5.91162264e-01 -4.27764863e-01 -4.86148208e-01
-1.43505841e-01 4.24011827e-01 8.30412209e-01 -1.21073687e+00
-1.12372339e-01 2.44399205e-01 6.98916674e-01 -1.97662070e-01
3.82681936e-01 -9.88677502e-01 3.88730109e-01 7.99956322e-01
-3.79901022e-01 2.12720521e-02 6.19808137e-01 2.67119735e-01
5.26348054e-02 -2.13907301e-01 1.04994941e+00 -4.33992803e-01
-2.10612461e-01 3.56965601e-01 -4.00976419e-01 8.42729509e-02
6.07103705e-01 -3.22386980e-01 2.02684492e-01 -2.90545791e-01
-1.42966855e+00 -3.81488562e-01 2.08555236e-01 1.81632936e-01
7.24002302e-01 -1.23615873e+00 -6.54186308e-01 6.65967584e-01
-3.55780363e-01 8.56394097e-02 9.17709291e-01 1.38718271e+00
-7.43491054e-01 2.80812055e-01 -4.83520627e-01 -8.23831022e-01
-4.87333149e-01 7.43396997e-01 1.05096948e+00 -4.72285777e-01
-1.18069458e+00 5.94625115e-01 3.41033876e-01 -8.76609921e-01
-1.38382256e-01 -6.73756838e-01 5.90921082e-02 -5.78803480e-01
5.08762181e-01 -9.54937190e-03 5.40874779e-01 -4.79497492e-01
-5.27132034e-01 2.01386914e-01 -1.78023607e-01 -1.47089466e-01
1.64574718e+00 9.95701924e-02 2.52247214e-01 1.53398916e-01
1.21909440e+00 -5.47530770e-01 -1.45125914e+00 -3.19686264e-01
-3.21697354e-01 -8.95668119e-02 3.79431456e-01 -1.00687885e+00
-1.32320893e+00 9.42485869e-01 8.60512257e-01 1.22959297e-02
1.11830866e+00 -1.03041962e-01 6.95703208e-01 -2.69042030e-02
2.41786107e-01 -8.12414050e-01 -8.23840722e-02 2.97597051e-01
1.13580585e+00 -1.02139497e+00 5.38684875e-02 -7.24248067e-02
-6.38785422e-01 1.23145735e+00 3.96782100e-01 -4.82765436e-01
9.51943099e-01 6.08478725e-01 2.39619389e-01 -5.66030264e-01
-2.41166726e-01 5.21258593e-01 2.83890337e-01 8.36406767e-01
2.44704112e-01 -5.66729568e-02 -1.18545614e-01 4.84603196e-01
-7.71396533e-02 3.32027853e-01 6.64596558e-01 8.71280551e-01
1.48857430e-01 -6.29259109e-01 -4.81926262e-01 4.38318819e-01
-6.57776058e-01 -1.60917714e-01 3.35747629e-01 1.05084801e+00
8.97984430e-02 2.19961479e-01 -1.68717932e-03 4.59787101e-02
3.13126743e-01 -4.29760106e-02 7.33003497e-01 -4.25207645e-01
-6.53400004e-01 -1.48137445e-02 -2.31679499e-01 -5.91803432e-01
-3.29495579e-01 -5.01985669e-01 -1.30837297e+00 1.48455845e-02
-1.66575760e-01 -9.63124633e-02 7.77574539e-01 1.09884381e+00
3.39371771e-01 7.45877683e-01 1.85466751e-01 -1.25766373e+00
-9.06824887e-01 -9.99652922e-01 -7.50284791e-01 6.12124860e-01
5.91294706e-01 -7.41804838e-01 -3.37761670e-01 -3.59359920e-01] | [13.533040046691895, -2.4218926429748535] |
8e653daf-8a28-41dd-9b14-7d6162766c37 | highly-accurate-dichotomous-image | 2203.03041 | null | https://arxiv.org/abs/2203.03041v4 | https://arxiv.org/pdf/2203.03041v4.pdf | Highly Accurate Dichotomous Image Segmentation | We present a systematic study on a new task called dichotomous image segmentation (DIS) , which aims to segment highly accurate objects from natural images. To this end, we collected the first large-scale DIS dataset, called DIS5K, which contains 5,470 high-resolution (e.g., 2K, 4K or larger) images covering camouflaged, salient, or meticulous objects in various backgrounds. DIS is annotated with extremely fine-grained labels. Besides, we introduce a simple intermediate supervision baseline (IS-Net) using both feature-level and mask-level guidance for DIS model training. IS-Net outperforms various cutting-edge baselines on the proposed DIS5K, making it a general self-learned supervision network that can facilitate future research in DIS. Further, we design a new metric called human correction efforts (HCE) which approximates the number of mouse clicking operations required to correct the false positives and false negatives. HCE is utilized to measure the gap between models and real-world applications and thus can complement existing metrics. Finally, we conduct the largest-scale benchmark, evaluating 16 representative segmentation models, providing a more insightful discussion regarding object complexities, and showing several potential applications (e.g., background removal, art design, 3D reconstruction). Hoping these efforts can open up promising directions for both academic and industries. Project page: https://xuebinqin.github.io/dis/index.html. | ['and Luc Van Gool', 'Ling Shao', 'Deng-Ping Fan', 'Xiaobin Hu', 'Hang Dai', 'Xuebin Qin'] | 2022-03-06 | null | null | null | null | ['dichotomous-image-segmentation'] | ['computer-vision'] | [ 4.38223243e-01 -8.44170973e-02 -3.04335833e-01 -2.41055101e-01
-7.96647012e-01 -5.31515956e-01 2.79603601e-01 -4.12109375e-01
-3.40480477e-01 6.01431429e-01 -1.38121441e-01 -2.71442264e-01
2.38320053e-01 -5.18844783e-01 -1.00897133e+00 -6.59183800e-01
2.47205943e-01 2.80029178e-01 5.83899796e-01 1.74851641e-01
2.71442950e-01 5.90210974e-01 -1.53652191e+00 2.42365167e-01
1.06859612e+00 1.11502814e+00 3.88074279e-01 5.73808253e-01
1.15836672e-01 4.25824136e-01 -7.49266922e-01 -5.76889277e-01
3.97100002e-01 -1.77076548e-01 -8.27855051e-01 2.24439338e-01
7.97937036e-01 -4.05969232e-01 -1.52540073e-01 1.28246033e+00
5.44738173e-01 -1.16187952e-01 5.41103542e-01 -1.28571570e+00
-9.08295572e-01 4.21538264e-01 -9.73879337e-01 4.31132197e-01
5.31132333e-02 5.80110848e-01 7.42324173e-01 -9.82163370e-01
4.98152345e-01 1.20360470e+00 6.18753493e-01 6.71596825e-01
-1.14522588e+00 -9.40664649e-01 3.48430276e-01 7.19624907e-02
-1.22070789e+00 -2.61999071e-01 5.06685376e-01 -4.65417802e-01
4.55467373e-01 5.42513072e-01 4.32693481e-01 1.08906746e+00
-6.19153343e-02 1.26544285e+00 1.16158044e+00 -1.81333348e-01
-2.79645082e-02 5.91402985e-02 3.34762335e-01 5.70838511e-01
4.95483905e-01 -5.48633561e-02 -2.75699586e-01 1.75473973e-01
1.09398985e+00 -7.08044274e-03 -4.10721570e-01 -1.86603591e-01
-1.36127830e+00 5.13121367e-01 6.53936267e-01 9.21483859e-02
-1.24653839e-01 7.80533850e-02 3.80487651e-01 -1.01075135e-01
5.15923917e-01 5.80102801e-01 -6.56626701e-01 6.77026734e-02
-8.02699447e-01 1.82915494e-01 2.99207389e-01 1.22056246e+00
6.51485205e-01 -3.92990597e-02 -6.09514117e-01 8.73867273e-01
-8.83563794e-03 5.90042531e-01 1.81221277e-01 -1.20419383e+00
3.46084684e-01 6.77523792e-01 2.20318735e-01 -8.44416201e-01
-2.04568431e-01 -4.93897468e-01 -1.06696653e+00 2.28816085e-02
2.95882285e-01 -1.26483947e-01 -1.30493200e+00 1.43082321e+00
5.49510896e-01 5.45612156e-01 -3.20442438e-01 1.00041926e+00
1.02011597e+00 4.30038542e-01 -5.01738451e-02 -9.44290161e-02
1.37840796e+00 -1.55179405e+00 -6.19547904e-01 -3.63856286e-01
3.88764113e-01 -7.96604037e-01 1.53453362e+00 5.76404274e-01
-1.13719392e+00 -7.26891518e-01 -7.49034405e-01 -1.28843173e-01
-3.39523852e-01 4.06213969e-01 7.41954923e-01 4.11730766e-01
-8.91717672e-01 4.10784096e-01 -7.41683125e-01 -1.47358283e-01
9.49852467e-01 1.84178472e-01 3.19131054e-02 -2.25570723e-01
-8.48132789e-01 4.72024322e-01 1.65261343e-01 2.22550035e-01
-1.03005779e+00 -7.76349306e-01 -6.81535721e-01 -2.98533410e-01
7.02417612e-01 -5.55510283e-01 1.28085208e+00 -9.23282802e-01
-1.13044035e+00 1.18166029e+00 -8.52855369e-02 -3.29457432e-01
6.00046754e-01 -5.25074601e-01 -3.79422992e-01 5.13723716e-02
4.21958923e-01 8.84453893e-01 9.11680758e-01 -1.53501844e+00
-7.66653657e-01 -2.87316620e-01 1.16355322e-01 1.44328261e-02
-2.60731041e-01 2.03376114e-01 -1.11793172e+00 -1.16799402e+00
-3.62133943e-02 -9.40982580e-01 -3.76270860e-01 1.70448259e-01
-8.85164618e-01 -1.08410157e-01 7.98749566e-01 -7.13889897e-01
1.23346400e+00 -2.15529537e+00 1.94789357e-02 -1.72750697e-01
3.61070365e-01 6.61799550e-01 -2.20678598e-01 -2.56805956e-01
6.82136044e-02 4.01812375e-01 -3.94195020e-01 -3.34123194e-01
-2.29289681e-01 -1.97620355e-02 -1.83367237e-01 1.56444073e-01
4.22942728e-01 1.23884130e+00 -7.01435685e-01 -5.70906818e-01
2.55843312e-01 3.64059091e-01 -3.90195698e-01 1.94085494e-01
-2.41340533e-01 5.90184450e-01 -5.55721521e-01 9.94639874e-01
9.24831033e-01 -6.37342334e-01 -4.87891614e-01 -4.23215866e-01
5.39508760e-02 -7.51179010e-02 -1.14314699e+00 1.48173153e+00
-1.36398658e-01 5.89869618e-01 1.25169620e-01 -7.56663561e-01
6.85812593e-01 -2.35451937e-01 2.25676894e-01 -7.45417953e-01
1.63709417e-01 -1.48940869e-02 -3.89262199e-01 -2.97565252e-01
4.10260767e-01 4.85590816e-01 1.10869287e-02 8.86042863e-02
-2.35348448e-01 -2.59615388e-03 2.08559349e-01 2.06990808e-01
9.82020557e-01 -5.35480827e-02 1.61476687e-01 -1.19574003e-01
4.00130093e-01 5.63087203e-02 8.67477238e-01 9.05951023e-01
-5.95087290e-01 9.14880097e-01 3.37422550e-01 -3.33521813e-01
-8.33886027e-01 -1.04142702e+00 -3.55106324e-01 1.01254928e+00
7.66306579e-01 -2.22073779e-01 -1.11792481e+00 -9.02896583e-01
1.88725200e-02 4.76970792e-01 -6.98340893e-01 -6.45887852e-02
-6.11330926e-01 -8.60020936e-01 5.81224322e-01 6.48394048e-01
9.46611464e-01 -1.24674642e+00 -2.32266888e-01 -1.45945624e-01
-3.30575854e-01 -1.28096235e+00 -8.69530976e-01 -6.51399931e-03
-9.00831461e-01 -1.08495212e+00 -9.76982355e-01 -8.49765539e-01
6.23501897e-01 7.44029105e-01 1.38094020e+00 2.29347333e-01
-4.71674114e-01 2.19531283e-01 -4.04756844e-01 -4.43721443e-01
-1.54621555e-02 1.98761374e-02 -9.35774520e-02 -2.42175639e-01
4.07542348e-01 -1.46127284e-01 -8.62209320e-01 8.81119847e-01
-9.41591263e-01 2.98365176e-01 7.29639411e-01 6.89862669e-01
1.06073880e+00 -2.34724078e-02 4.82463449e-01 -9.39626455e-01
2.40444586e-01 -1.84942290e-01 -7.20530748e-01 4.49420452e-01
-5.30413270e-01 -3.75023335e-01 3.82100314e-01 -4.97546285e-01
-1.05088043e+00 -8.20280835e-02 -1.44770173e-02 -5.06706834e-01
-3.63050133e-01 -7.52893537e-02 -4.37716901e-01 -2.11682469e-01
4.96061742e-01 2.28040561e-01 -3.80050659e-01 -7.08125710e-01
3.32040161e-01 7.15344310e-01 8.26512754e-01 -5.07365286e-01
7.90227473e-01 4.75282609e-01 -4.16738957e-01 -5.80398738e-01
-1.31233144e+00 -5.72445154e-01 -6.50137901e-01 -6.05339222e-02
9.11170006e-01 -1.01980281e+00 -4.27400917e-01 8.25196147e-01
-1.04079485e+00 -6.11079872e-01 -3.08039427e-01 8.53486806e-02
-3.53336662e-01 3.40793997e-01 -8.90162349e-01 -4.68494445e-01
-3.95158917e-01 -1.20621979e+00 1.42125499e+00 5.35425365e-01
2.42898479e-01 -5.82710505e-01 -3.85030597e-01 8.28632116e-01
1.93587095e-01 2.01298714e-01 4.30890054e-01 -3.45765322e-01
-9.40399289e-01 3.87498960e-02 -8.33782136e-01 7.35022485e-01
9.69217718e-02 1.79803267e-01 -9.61180270e-01 -2.39716917e-01
-2.29798362e-01 -2.96162874e-01 1.03424811e+00 6.92032754e-01
1.88309312e+00 -2.05328375e-01 -6.06505752e-01 8.38949680e-01
1.11585546e+00 8.39863196e-02 8.27835083e-01 4.81882274e-01
1.06374907e+00 2.58815199e-01 1.15014493e+00 2.29771987e-01
2.12708369e-01 7.43805408e-01 4.60738242e-01 -5.93769670e-01
-5.22682667e-01 -9.36016515e-02 1.82042196e-01 5.27793348e-01
-1.81432217e-01 -3.06401819e-01 -8.54894519e-01 5.05653977e-01
-1.65844953e+00 -5.56414902e-01 -3.13317269e-01 2.05498266e+00
9.29022074e-01 4.38294709e-01 1.43295780e-01 -8.05578828e-02
1.09271574e+00 2.80972775e-02 -1.04501307e+00 1.37434661e-01
-2.79992431e-01 -1.59341842e-02 6.47334635e-01 2.03667551e-01
-1.51178312e+00 1.28249443e+00 6.03774214e+00 1.39419138e+00
-9.15539086e-01 1.82599530e-01 1.17988348e+00 5.63090816e-02
-3.63162979e-02 -3.52831274e-01 -1.05858195e+00 7.13820636e-01
4.82621729e-01 5.18927164e-02 2.96164751e-01 1.11246705e+00
2.39345580e-01 -1.36857495e-01 -8.53121281e-01 1.10321975e+00
6.25861287e-02 -1.33424842e+00 1.57195821e-01 -8.78923684e-02
1.01287007e+00 1.78136915e-01 1.60844460e-01 2.69700229e-01
2.76687115e-01 -1.07227755e+00 5.57172894e-01 3.51502806e-01
1.09392428e+00 -4.52122808e-01 7.46690035e-01 2.27639735e-01
-1.18850923e+00 9.43959057e-02 -3.01543504e-01 1.96767375e-01
2.72510499e-01 8.01992476e-01 -2.57904440e-01 3.28647941e-01
1.11555493e+00 9.66883361e-01 -9.24172521e-01 1.28903353e+00
-1.62233979e-01 7.53315747e-01 -7.88309574e-02 3.81252527e-01
1.29244015e-01 -1.19535349e-01 4.59996730e-01 1.38341701e+00
2.37434711e-02 1.75723746e-01 1.57810315e-01 8.93917680e-01
-3.06559652e-01 -1.79960728e-01 -2.21887633e-01 1.29018366e-01
5.42164922e-01 1.33342004e+00 -1.16136527e+00 -5.84893942e-01
-3.79439086e-01 1.25479364e+00 3.77083570e-02 4.69768256e-01
-1.22545159e+00 -1.63936049e-01 8.65425467e-01 7.66648278e-02
4.12808418e-01 9.99415293e-02 -5.51974893e-01 -1.26279783e+00
1.54253736e-01 -1.06888485e+00 2.49132693e-01 -9.31374073e-01
-1.41381383e+00 5.41890085e-01 -1.38641924e-01 -1.27381372e+00
6.48769081e-01 -7.37914860e-01 -3.86128485e-01 5.50697684e-01
-1.66935360e+00 -1.08598328e+00 -7.59164929e-01 3.55806142e-01
8.15265656e-01 1.42713144e-01 2.61448711e-01 5.36232173e-01
-1.13600445e+00 7.51369238e-01 5.32743782e-02 3.11291903e-01
8.52482796e-01 -1.19664454e+00 7.90457904e-01 8.28663945e-01
5.63038848e-02 3.91186386e-01 4.14499283e-01 -6.65402770e-01
-8.68010998e-01 -1.56425810e+00 3.27287823e-01 -7.80593574e-01
2.61061549e-01 -3.29139799e-01 -9.96849835e-01 7.25435436e-01
-2.18456641e-01 4.00551438e-01 2.53747076e-01 -2.38975942e-01
-1.82644546e-01 -9.56648067e-02 -1.17280686e+00 5.92501700e-01
1.54590201e+00 -1.25769958e-01 -1.84664533e-01 4.55331534e-01
1.35031283e+00 -6.35476589e-01 -7.83677459e-01 7.38685191e-01
3.34119201e-01 -1.05555379e+00 1.22016096e+00 -4.04411763e-01
5.46819866e-01 -3.50357592e-01 9.98233259e-02 -8.90733302e-01
-4.97153103e-01 -3.95175755e-01 -1.73915759e-01 1.43429828e+00
2.60838538e-01 -3.67974401e-01 1.00551844e+00 5.33895612e-01
-3.67234617e-01 -1.02771032e+00 -5.74052274e-01 -8.90692830e-01
-7.17995465e-02 -4.10280377e-01 6.93069935e-01 8.87258172e-01
-9.42356884e-01 1.96431085e-01 -4.27002758e-01 2.27037415e-01
6.85484648e-01 9.91971269e-02 8.56886327e-01 -1.10076511e+00
-4.96397205e-02 -5.16457796e-01 -2.94248492e-01 -1.61391926e+00
-1.56171843e-01 -5.41413546e-01 6.18552789e-03 -1.38814425e+00
4.42163110e-01 -5.85851431e-01 -3.53365183e-01 4.08521622e-01
-5.97337186e-01 6.74329579e-01 -3.41048539e-02 3.51600528e-01
-1.04546535e+00 5.09748161e-01 1.82382905e+00 -2.55675972e-01
-1.44703388e-01 2.45795652e-01 -9.07918513e-01 1.05763018e+00
7.69916952e-01 -3.88880074e-01 -1.19055279e-01 -6.03692353e-01
-1.61736846e-01 -4.42983240e-01 6.80169225e-01 -8.99668574e-01
-5.21751419e-02 -4.85117286e-01 4.18123543e-01 -8.10563743e-01
4.53945436e-02 -6.70742035e-01 -1.64363638e-01 2.04970837e-01
-2.38080591e-01 -1.37346119e-01 1.46806672e-01 5.97328246e-01
-2.31678277e-01 3.89432460e-02 9.53649580e-01 -1.38746753e-01
-1.05228400e+00 5.89267492e-01 1.83889344e-01 3.95885885e-01
1.25192165e+00 -5.04190385e-01 -4.41173941e-01 5.00905439e-02
-4.99170929e-01 4.58245873e-01 5.64624310e-01 4.38928455e-01
5.27675271e-01 -1.22512245e+00 -5.34226239e-01 7.41221160e-02
2.59060383e-01 5.63070059e-01 3.08702767e-01 7.75477111e-01
-4.79334712e-01 2.93763191e-01 -6.64860457e-02 -7.99331844e-01
-1.38562000e+00 4.92053747e-01 2.40263700e-01 -8.78679007e-02
-6.94271564e-01 1.20453215e+00 6.00702643e-01 -4.30910498e-01
5.20438671e-01 -4.85272378e-01 -1.50914729e-01 -3.02709907e-01
7.04104960e-01 5.53200245e-01 -6.28649741e-02 -4.88204896e-01
-3.57280642e-01 7.45723367e-01 -1.99185833e-01 4.83528495e-01
1.10549128e+00 -2.83417851e-01 -4.55635153e-02 1.97852686e-01
8.40541542e-01 -1.17445454e-01 -1.73246241e+00 -3.17871988e-01
3.01973671e-02 -8.33565533e-01 -8.38428810e-02 -9.30139661e-01
-1.36632907e+00 6.97273016e-01 6.93434715e-01 -2.06053238e-02
1.38274229e+00 2.66030610e-01 9.84595776e-01 2.07895473e-01
2.60061830e-01 -1.07116556e+00 4.24753219e-01 3.19057375e-01
9.30569410e-01 -1.65098083e+00 -8.65740702e-02 -8.12838197e-01
-7.81819463e-01 4.34705853e-01 1.12161469e+00 -3.64262750e-03
4.14410800e-01 2.71617293e-01 9.22532827e-02 -1.34095654e-01
-3.37177038e-01 -3.22551012e-01 4.13976371e-01 7.52961874e-01
2.33659804e-01 1.82604000e-01 -1.90445945e-01 8.54692638e-01
-7.51705766e-02 8.92259851e-02 3.61246616e-01 6.27153039e-01
-4.07435626e-01 -6.95692599e-01 -3.60721678e-01 7.92691171e-01
-4.26486433e-01 -1.52539104e-01 -4.17249620e-01 6.80759490e-01
4.88283753e-01 8.25959980e-01 -5.89264259e-02 -3.22664469e-01
4.66119111e-01 -5.49510777e-01 2.12376460e-01 -7.39147186e-01
-3.44696790e-01 -7.22986646e-03 -1.36340097e-01 -7.94189036e-01
-4.36082840e-01 -4.51157212e-01 -9.83588338e-01 -2.26436436e-01
-6.43042743e-01 -2.21696213e-01 3.09557796e-01 6.69428170e-01
4.99901056e-01 7.12071419e-01 3.96029204e-01 -1.10660553e+00
-2.22660795e-01 -7.52922833e-01 -5.11886895e-01 5.38632929e-01
3.21033925e-01 -7.45561302e-01 -3.91854435e-01 2.11873770e-01] | [9.630363464355469, -0.10245904326438904] |
334087a0-348a-4a48-a4d9-2b65cb50fa8f | bilex-rx-lexical-data-augmentation-for | 2303.15265 | null | https://arxiv.org/abs/2303.15265v1 | https://arxiv.org/pdf/2303.15265v1.pdf | Bilex Rx: Lexical Data Augmentation for Massively Multilingual Machine Translation | Neural machine translation (NMT) has progressed rapidly over the past several years, and modern models are able to achieve relatively high quality using only monolingual text data, an approach dubbed Unsupervised Machine Translation (UNMT). However, these models still struggle in a variety of ways, including aspects of translation that for a human are the easiest - for instance, correctly translating common nouns. This work explores a cheap and abundant resource to combat this problem: bilingual lexica. We test the efficacy of bilingual lexica in a real-world set-up, on 200-language translation models trained on web-crawled text. We present several findings: (1) using lexical data augmentation, we demonstrate sizable performance gains for unsupervised translation; (2) we compare several families of data augmentation, demonstrating that they yield similar improvements, and can be combined for even greater improvements; (3) we demonstrate the importance of carefully curated lexica over larger, noisier ones, especially with larger models; and (4) we compare the efficacy of multilingual lexicon data versus human-translated parallel data. Finally, we open-source GATITOS (available at https://github.com/google-research/url-nlp/tree/main/gatitos), a new multilingual lexicon for 26 low-resource languages, which had the highest performance among lexica in our experiments. | ['Orhan Firat', 'Ishank Saxena', 'Isaac Caswell', 'Alex Jones'] | 2023-03-27 | null | null | null | null | ['nmt', 'unsupervised-machine-translation'] | ['computer-code', 'natural-language-processing'] | [ 5.71023114e-02 -1.86274111e-01 -5.60949385e-01 -1.81438699e-01
-1.49269962e+00 -9.64303076e-01 8.67684424e-01 8.83834288e-02
-4.92375433e-01 1.09255040e+00 5.62337041e-01 -7.57546067e-01
3.87625784e-01 -5.27807355e-01 -9.10218537e-01 -3.31080437e-01
2.72672057e-01 9.03440237e-01 -3.47103357e-01 -6.48731768e-01
-6.76746964e-02 1.88391373e-01 -9.08949554e-01 3.11812252e-01
1.17182469e+00 2.55961388e-01 1.66501865e-01 2.37701952e-01
-1.94214880e-02 2.30993524e-01 -2.71204174e-01 -7.49443829e-01
5.35091996e-01 -7.82142222e-01 -8.13005745e-01 -1.23268232e-01
3.64908308e-01 -3.76452915e-02 -7.35807121e-02 1.05266166e+00
6.08984768e-01 -3.88559431e-01 4.90619987e-01 -7.27922916e-01
-1.01744163e+00 8.33078325e-01 -4.32672203e-01 1.55252844e-01
2.93133467e-01 3.99070501e-01 1.30456221e+00 -1.22727811e+00
9.62998450e-01 1.11643660e+00 6.80351675e-01 3.63943011e-01
-1.35830510e+00 -7.73484051e-01 -2.77859896e-01 -4.29680496e-02
-1.25073814e+00 -7.42613018e-01 2.98449725e-01 -2.55167425e-01
1.37378335e+00 1.25925660e-01 5.61588228e-01 1.42533338e+00
4.47990328e-01 7.04785407e-01 1.62438118e+00 -7.75209546e-01
-2.69716918e-01 1.58490568e-01 -1.16887890e-01 4.78527755e-01
2.35424086e-01 1.54855892e-01 -5.42093396e-01 -1.64348751e-01
5.32687962e-01 -4.16711867e-01 -2.12843522e-01 1.95240647e-01
-1.79119766e+00 9.00177538e-01 5.60532650e-03 6.80228472e-01
-3.43676180e-01 -3.89710873e-01 4.49081272e-01 6.42995477e-01
7.38800704e-01 7.24298537e-01 -7.52937615e-01 -3.34157497e-01
-8.64939213e-01 -3.49169225e-02 8.98128510e-01 1.08159637e+00
7.00445116e-01 8.66560172e-03 1.46594271e-01 1.14398634e+00
-1.85757518e-01 8.62516105e-01 8.03136826e-01 -3.90445948e-01
9.39955771e-01 3.59800965e-01 -4.55570295e-02 -6.50268674e-01
-2.71675527e-01 -5.65517843e-01 -6.91458642e-01 -3.70311320e-01
5.83537638e-01 -2.33626127e-01 -7.94245601e-01 1.92096305e+00
2.74295211e-02 -5.63343406e-01 2.84289420e-01 8.23227286e-01
2.86947340e-01 7.81528473e-01 -1.68854650e-02 -3.43780249e-01
1.31985140e+00 -1.12708247e+00 -6.17517650e-01 -5.53882658e-01
7.62364805e-01 -1.31329095e+00 1.47733700e+00 2.65096277e-01
-1.16615403e+00 -3.82255375e-01 -9.26013470e-01 -2.27102086e-01
-4.70635802e-01 1.69952750e-01 8.44421208e-01 4.41624939e-01
-1.18833041e+00 4.41425651e-01 -8.70069802e-01 -8.53658020e-01
1.11670382e-01 4.34615463e-01 -6.01514339e-01 -1.82591528e-01
-1.39212215e+00 1.24991226e+00 2.27147937e-01 -1.43816784e-01
-7.95851469e-01 -2.97800541e-01 -7.22684085e-01 -2.80751139e-01
2.57112712e-01 -5.73085785e-01 1.17130697e+00 -1.23485374e+00
-1.33416808e+00 1.13323236e+00 -3.33536118e-01 -2.85816431e-01
4.90728348e-01 -2.99848378e-01 -5.24190545e-01 -2.46198162e-01
4.95965928e-01 6.66610658e-01 4.22414839e-01 -8.67424309e-01
-5.44626772e-01 -3.30992818e-01 -2.74354786e-01 3.62431973e-01
-5.44640958e-01 4.62291867e-01 -3.88958693e-01 -8.64224732e-01
4.39109392e-02 -1.11577833e+00 -1.60775766e-01 -6.09106004e-01
-3.76427174e-01 -6.43544644e-02 7.20724985e-02 -1.08948588e+00
1.02894664e+00 -1.82936251e+00 3.40652108e-01 -1.44000247e-01
-5.79599850e-02 1.35462925e-01 -4.58437353e-01 8.06964636e-01
-6.81448169e-03 4.14755106e-01 -3.37476790e-01 -2.31142908e-01
-8.86783078e-02 2.11119309e-01 -2.27332890e-01 3.88358414e-01
2.52575278e-01 1.30084825e+00 -8.88770998e-01 -3.83210808e-01
-2.75014162e-01 2.43174434e-01 -3.54157895e-01 -2.14328915e-01
-1.33310735e-01 6.40692353e-01 -1.66331485e-01 8.30305696e-01
2.41015464e-01 -2.27634341e-01 5.59841454e-01 1.38849869e-01
-3.01724941e-01 8.06818962e-01 -5.21848679e-01 1.82701600e+00
-5.36444485e-01 6.77071095e-01 -1.41724482e-01 -5.38137913e-01
6.66465580e-01 4.66373950e-01 3.19267720e-01 -9.20111775e-01
1.87161848e-01 9.43469763e-01 3.47128451e-01 -2.46223152e-01
5.15811026e-01 -2.32573107e-01 -9.41597670e-02 8.38494182e-01
1.52138948e-01 -1.64384007e-01 4.50961709e-01 3.28401253e-02
9.20016944e-01 2.72519797e-01 5.68758070e-01 -5.59862912e-01
1.19327329e-01 5.58359087e-01 6.91357315e-01 3.52288753e-01
-6.11728430e-02 4.80023533e-01 1.49603352e-01 -3.47019434e-01
-1.38100934e+00 -9.42256093e-01 -1.56696990e-01 1.10862303e+00
-2.77227730e-01 -4.48693097e-01 -7.60501862e-01 -6.54670715e-01
-1.36459365e-01 6.09821916e-01 -2.49142155e-01 1.47912070e-01
-9.29641902e-01 -1.27904487e+00 7.99236774e-01 4.04632688e-01
2.66305208e-01 -1.01277947e+00 1.33427590e-01 3.16345453e-01
-8.40261400e-01 -1.23082447e+00 -7.28132069e-01 3.57524216e-01
-1.08260882e+00 -6.21732712e-01 -7.57884383e-01 -9.73703623e-01
6.49933517e-01 2.32991189e-01 1.42502511e+00 -1.62500516e-02
2.05710217e-01 -9.66676697e-02 -3.76726210e-01 -2.82266229e-01
-8.86453867e-01 5.82991123e-01 4.92192239e-01 -3.63734335e-01
6.64423525e-01 -7.23769426e-01 -1.82426214e-01 3.80833477e-01
-6.04380310e-01 2.57199019e-01 8.43740046e-01 9.67469335e-01
6.57283366e-01 -4.83070672e-01 6.82428539e-01 -7.77950883e-01
8.46407771e-01 -4.87956405e-01 -4.96765673e-01 2.37491429e-01
-8.46916676e-01 1.05165187e-02 7.90611446e-01 -4.74624097e-01
-6.65764749e-01 -3.36288422e-01 -2.49417856e-01 1.72024235e-01
-7.24826008e-02 6.53890789e-01 -5.46699157e-03 1.38632774e-01
9.22628582e-01 3.84739816e-01 -3.17541435e-02 -6.83871090e-01
3.75677884e-01 9.13361549e-01 3.12702149e-01 -8.74277234e-01
8.32248569e-01 1.92568868e-01 -4.41310138e-01 -5.70757270e-01
-5.86149395e-01 -2.90708244e-01 -8.41841698e-01 2.74215072e-01
6.56577051e-01 -1.08067644e+00 -1.78437326e-02 3.32072884e-01
-1.05972815e+00 -5.80063462e-01 1.26398131e-02 7.08965898e-01
-4.78415668e-01 2.15531558e-01 -1.13051081e+00 -2.43382260e-01
-4.74076003e-01 -1.32045925e+00 9.20498431e-01 -1.76473290e-01
-5.01242280e-01 -1.02953494e+00 2.73063898e-01 6.34946585e-01
4.22759145e-01 6.11779355e-02 1.26084805e+00 -9.17847931e-01
-4.98355567e-01 1.06867999e-01 -1.53490439e-01 3.86336505e-01
4.27403718e-01 -2.41765201e-01 -5.65582097e-01 -4.65206385e-01
-1.25109091e-01 -4.90901828e-01 5.06588995e-01 -2.27518678e-02
1.15903169e-01 -3.36196780e-01 -1.86338142e-01 4.98244494e-01
1.30562127e+00 4.64706086e-02 4.06933784e-01 5.37928462e-01
6.04191899e-01 4.96626586e-01 4.18572307e-01 -3.10877681e-01
5.83320558e-01 8.34296882e-01 -1.68916255e-01 -3.42654675e-01
-3.81652355e-01 -3.60568434e-01 6.93916261e-01 1.81246197e+00
-2.99178213e-01 -1.88852206e-01 -1.18421328e+00 8.60399604e-01
-1.72443092e+00 -3.82947803e-01 -1.15700431e-01 2.12603784e+00
1.37281215e+00 1.05718523e-01 1.20626286e-01 -2.68005818e-01
4.61274564e-01 -1.52774215e-01 -3.10112983e-01 -4.51852828e-01
-6.10684514e-01 4.14579481e-01 5.53574085e-01 5.75466692e-01
-7.81242669e-01 1.51979661e+00 6.45912743e+00 8.98835599e-01
-1.30174553e+00 6.35852158e-01 5.41552901e-01 -4.14806418e-02
-3.85673553e-01 7.80164972e-02 -8.08475494e-01 3.58218312e-01
1.15707350e+00 -3.53280216e-01 7.85581887e-01 5.40747583e-01
2.94726700e-01 1.71389580e-01 -1.21275699e+00 4.99044597e-01
1.18043378e-01 -1.09538603e+00 2.12717086e-01 2.93945491e-01
1.06591070e+00 6.70999527e-01 -2.68776845e-02 4.03735071e-01
4.53082472e-01 -9.32222068e-01 6.72174037e-01 -2.01376349e-01
1.08464766e+00 -5.82963288e-01 6.23749316e-01 3.78468692e-01
-8.02837670e-01 2.79212922e-01 -2.94404984e-01 -6.67470768e-02
1.87913507e-01 4.47647363e-01 -8.12700212e-01 7.38362074e-01
4.16268110e-01 7.51045585e-01 -6.43938541e-01 5.20861983e-01
-5.74509621e-01 8.47694159e-01 -2.91276991e-01 2.62911152e-02
4.73929971e-01 -4.38407660e-01 6.17599428e-01 1.40074146e+00
3.22729588e-01 -2.70210743e-01 9.95597690e-02 5.20228982e-01
-4.42896515e-01 7.78196990e-01 -7.17242777e-01 -4.40783918e-01
3.07460815e-01 1.08259976e+00 -6.58937752e-01 -2.80720294e-01
-6.41213477e-01 1.10403180e+00 5.21623969e-01 4.27409798e-01
-7.05211520e-01 6.78245211e-03 6.32656217e-01 7.01686591e-02
-1.95232242e-01 -4.38682288e-01 -4.00978088e-01 -1.55041134e+00
1.93608344e-01 -1.67887342e+00 1.33972749e-01 -6.08414173e-01
-1.34882593e+00 9.49830055e-01 -2.62756765e-01 -1.27923441e+00
-4.05210763e-01 -6.57627344e-01 -1.20757911e-02 1.11976874e+00
-1.39512730e+00 -1.34204888e+00 4.28717911e-01 3.50776225e-01
8.65460575e-01 -3.80095065e-01 1.04101014e+00 6.82963371e-01
-4.30958390e-01 7.47690558e-01 3.45924258e-01 2.48440683e-01
1.15526021e+00 -9.89276946e-01 9.87612545e-01 9.61246490e-01
6.52649999e-01 9.27645683e-01 4.41923231e-01 -8.68144393e-01
-1.45205235e+00 -9.91729200e-01 1.57670057e+00 -7.73716152e-01
9.34945047e-01 -5.79670787e-01 -6.60232127e-01 1.09393847e+00
5.84149897e-01 -6.94642425e-01 6.74027383e-01 3.56390715e-01
-4.27914172e-01 2.31764793e-01 -8.05867374e-01 9.54365671e-01
1.13856757e+00 -6.30020380e-01 -6.30400896e-01 7.24872231e-01
7.54270732e-01 -3.57168943e-01 -9.21606004e-01 4.12888944e-01
6.44044161e-01 -5.72409570e-01 6.72236621e-01 -8.05786490e-01
7.25985944e-01 -5.86881936e-02 -3.59065264e-01 -1.63092053e+00
-2.68122047e-01 -7.37758756e-01 4.02612478e-01 1.09768367e+00
1.12774408e+00 -1.04786479e+00 2.78634697e-01 7.07901046e-02
-1.52932748e-01 -8.74269783e-01 -8.09139669e-01 -9.97871220e-01
6.32144809e-01 -3.18660855e-01 5.92318654e-01 1.39693499e+00
1.93047285e-01 6.34951055e-01 -4.87331688e-01 -1.93775460e-01
3.53240639e-01 1.15711652e-01 6.79802537e-01 -7.64390886e-01
-4.14216787e-01 -5.33277154e-01 8.75494480e-02 -9.69527066e-01
1.31582454e-01 -1.48991573e+00 -5.19801565e-02 -1.47328424e+00
4.30430710e-01 -3.70053858e-01 -6.33921176e-02 7.81991184e-01
-3.17261219e-01 6.32505596e-01 1.73075125e-01 7.05941916e-01
-1.42593011e-01 3.03131789e-01 1.28599215e+00 -1.04754053e-01
-1.56111956e-01 -3.45710307e-01 -8.73742938e-01 4.53232676e-01
9.38731313e-01 -6.60365343e-01 -8.58990592e-04 -9.56769049e-01
3.51861626e-01 -7.61773139e-02 -1.48018286e-01 -5.52791238e-01
-1.40026426e-02 -8.59482661e-02 1.77877739e-01 -1.71667218e-01
6.94636405e-02 -5.09745955e-01 1.43275872e-01 3.21496159e-01
-8.92015547e-02 6.57256901e-01 4.76233393e-01 -4.71610725e-02
-1.60784632e-01 9.95686650e-02 5.48813105e-01 -3.01092178e-01
-3.76483679e-01 1.37361154e-01 -4.34631079e-01 2.37200752e-01
4.48848069e-01 1.41128749e-01 -4.68897909e-01 -2.81260699e-01
-3.42966914e-01 7.68547952e-02 7.43898094e-01 7.26580679e-01
-2.04798323e-03 -1.33588409e+00 -1.10631716e+00 1.90432191e-01
1.81402549e-01 -5.89458525e-01 -4.93420959e-01 1.14339662e+00
-5.28221846e-01 5.57366252e-01 -2.43343666e-01 -5.18410206e-01
-8.49570334e-01 3.47285092e-01 1.10011213e-01 -5.49594045e-01
-5.14615119e-01 3.50572765e-01 -1.26077421e-02 -8.96038115e-01
-2.75044709e-01 -1.60418734e-01 3.66033018e-01 -5.59173413e-02
2.92042702e-01 7.86509663e-02 3.53340447e-01 -9.22529995e-01
-2.04879940e-01 5.13809264e-01 -2.29474053e-01 -4.92587298e-01
1.29395533e+00 -1.00478083e-01 -4.76967096e-01 3.89997303e-01
9.20264900e-01 4.52113777e-01 -5.83722711e-01 -3.42041105e-01
1.61995426e-01 -1.73163980e-01 -4.15460765e-01 -1.23546517e+00
-6.79433465e-01 6.57160521e-01 2.03228220e-01 -2.02624843e-01
9.59643304e-01 -1.45734489e-01 1.03097796e+00 4.26933318e-01
7.56219923e-01 -1.00837016e+00 -4.19977576e-01 7.92015493e-01
7.04207897e-01 -1.41353726e+00 -1.72942147e-01 -2.53015816e-01
-5.88684320e-01 8.71048450e-01 2.06171632e-01 2.79487371e-01
1.44002855e-01 3.47635835e-01 5.97588837e-01 -4.65382449e-02
-6.90766096e-01 -1.62222579e-01 1.11402482e-01 2.83607244e-01
8.22211087e-01 2.83638924e-01 -8.32876384e-01 3.94701958e-01
-5.59166253e-01 -2.38563195e-01 2.60585040e-01 6.98713124e-01
1.01339705e-02 -1.62497354e+00 -3.66036683e-01 3.37861717e-01
-8.07896197e-01 -7.43438661e-01 -6.79386675e-01 1.03078878e+00
-1.20808057e-01 1.11753178e+00 -3.42673302e-01 -3.45576435e-01
2.11476475e-01 4.79946345e-01 5.01030803e-01 -7.73897052e-01
-8.43734503e-01 2.83805758e-01 3.81436765e-01 -3.94521624e-01
-1.75552964e-01 -7.99068451e-01 -8.34732890e-01 -3.66138101e-01
-1.76111326e-01 2.54550099e-01 8.40481102e-01 1.03283215e+00
3.25902969e-01 1.27150401e-01 4.20121253e-01 -6.43950760e-01
-4.68046010e-01 -1.30902719e+00 -8.11356381e-02 3.13199639e-01
-6.36772513e-02 -2.12335095e-01 -3.25906157e-01 2.50176668e-01] | [11.513226509094238, 10.301836013793945] |
6d6be8a0-2da2-4f09-b69c-e00ba27f40b6 | making-the-invisible-visible-action | 1909.09300 | null | https://arxiv.org/abs/1909.09300v1 | https://arxiv.org/pdf/1909.09300v1.pdf | Making the Invisible Visible: Action Recognition Through Walls and Occlusions | Understanding people's actions and interactions typically depends on seeing them. Automating the process of action recognition from visual data has been the topic of much research in the computer vision community. But what if it is too dark, or if the person is occluded or behind a wall? In this paper, we introduce a neural network model that can detect human actions through walls and occlusions, and in poor lighting conditions. Our model takes radio frequency (RF) signals as input, generates 3D human skeletons as an intermediate representation, and recognizes actions and interactions of multiple people over time. By translating the input to an intermediate skeleton-based representation, our model can learn from both vision-based and RF-based datasets, and allow the two tasks to help each other. We show that our model achieves comparable accuracy to vision-based action recognition systems in visible scenarios, yet continues to work accurately when people are not visible, hence addressing scenarios that are beyond the limit of today's vision-based action recognition. | ['Ming-Min Zhao', 'Lijie Fan', 'Tianhong Li', 'Yingcheng Liu', 'Dina Katabi'] | 2019-09-20 | making-the-invisible-visible-action-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Li_Making_the_Invisible_Visible_Action_Recognition_Through_Walls_and_Occlusions_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Li_Making_the_Invisible_Visible_Action_Recognition_Through_Walls_and_Occlusions_ICCV_2019_paper.pdf | iccv-2019-10 | ['rf-based-pose-estimation'] | ['computer-vision'] | [ 6.99339390e-01 -3.02512925e-02 -1.10696837e-01 -2.80698866e-01
-2.20215231e-01 -3.61554027e-01 6.12692475e-01 -4.71499354e-01
-4.74185497e-01 5.11537135e-01 5.39130569e-01 -1.81614801e-01
3.20675403e-01 -6.76688790e-01 -4.15521532e-01 -4.54455733e-01
1.74148217e-01 2.38523290e-01 4.29956228e-01 5.06340265e-02
1.16784628e-02 5.18485665e-01 -1.64822316e+00 7.00662851e-01
1.34894386e-01 6.69702232e-01 -1.10338219e-01 1.30611539e+00
1.07993439e-01 1.10557079e+00 -6.57724917e-01 2.96619475e-01
4.28875774e-01 -4.32101935e-01 -8.06170523e-01 4.37802821e-01
8.11986685e-01 -6.37286544e-01 -7.35791206e-01 3.49780142e-01
2.75557697e-01 2.41776854e-01 5.89610040e-01 -1.19221723e+00
-6.20490789e-01 -3.45846266e-02 -7.10594952e-01 3.91016573e-01
9.16495383e-01 5.57581425e-01 4.99150515e-01 -3.26541394e-01
4.50782329e-01 1.59295416e+00 5.79771876e-01 7.92851567e-01
-1.32984829e+00 -3.41429055e-01 4.11380261e-01 2.43404299e-01
-8.76320601e-01 -7.52586305e-01 3.81531060e-01 -6.10690415e-01
1.23612547e+00 8.06834698e-02 9.34800923e-01 1.52589214e+00
1.34964049e-01 8.56719375e-01 8.86362553e-01 -5.57855606e-01
1.04569621e-01 -3.43098730e-01 1.45263195e-01 7.28219926e-01
2.57282704e-01 2.19109491e-01 -7.81648695e-01 -7.36196041e-02
1.08854640e+00 2.48507753e-01 -4.38938886e-01 -2.54747689e-01
-1.44822431e+00 3.43210876e-01 1.78253010e-01 2.67481506e-01
-4.04617131e-01 8.26965928e-01 -7.50934929e-02 2.12071374e-01
1.81280792e-01 5.12380078e-02 -2.27691084e-01 -3.17999244e-01
-7.75071025e-01 2.56314334e-02 5.27214170e-01 3.60617280e-01
2.69489408e-01 1.14789039e-01 -1.46014825e-01 5.47651887e-01
4.16043162e-01 8.20044279e-01 3.86878997e-01 -1.27509332e+00
3.77318382e-01 5.57324588e-01 5.13050139e-01 -7.81130195e-01
-6.76516652e-01 1.07951783e-01 -3.31722617e-01 6.94206059e-01
9.98723686e-01 -3.50265801e-01 -1.17430091e+00 1.65143061e+00
3.85690957e-01 2.78887153e-01 -6.99327663e-02 9.95040894e-01
6.41873658e-01 1.84924722e-01 1.45715743e-01 5.76038696e-02
1.33595145e+00 -8.62771273e-01 -3.83718848e-01 -7.55659282e-01
5.12171805e-01 -5.40165007e-01 8.47222567e-01 4.91267771e-01
-8.59932065e-01 -8.60395074e-01 -9.74362969e-01 -1.30439460e-01
-2.37368047e-01 2.98543513e-01 7.02066302e-01 7.47460067e-01
-1.11152720e+00 4.35119182e-01 -1.07232094e+00 -9.25673604e-01
4.19364214e-01 4.05816406e-01 -4.42125291e-01 -1.28822267e-01
-9.57734048e-01 9.40578818e-01 -2.34677806e-01 6.19431213e-02
-7.04149008e-01 -1.70657597e-02 -7.62395978e-01 -3.52425009e-01
2.33124644e-01 -9.28882241e-01 1.47035491e+00 -1.24491274e+00
-1.40349233e+00 7.31142759e-01 -2.90127993e-01 -4.94956374e-01
5.72489500e-01 -4.22164142e-01 -5.61604381e-01 4.59373474e-01
-3.71655338e-02 6.88306451e-01 9.36440468e-01 -9.32641864e-01
-6.68692172e-01 -6.62122130e-01 3.85649294e-01 2.72597075e-01
2.82481499e-03 4.82957885e-02 -2.45211855e-01 -2.90271401e-01
1.33587524e-01 -1.00541806e+00 -1.85027659e-01 6.00275338e-01
-1.46830291e-01 -8.26224312e-02 7.99062490e-01 -6.02077067e-01
6.05314076e-01 -2.07913136e+00 -3.04110497e-01 1.09524317e-01
6.78350106e-02 3.72166276e-01 -2.50357449e-01 2.53914267e-01
6.41191080e-02 -2.48902678e-01 2.50543833e-01 -8.29921216e-02
-2.09989771e-01 2.34345153e-01 -6.13657869e-02 6.82833493e-01
-6.52850941e-02 7.48048544e-01 -8.79013598e-01 -3.77033561e-01
5.31552970e-01 8.22868586e-01 -1.59794062e-01 -1.07340299e-01
-5.57130799e-02 5.94606757e-01 -5.67036927e-01 5.94902992e-01
1.76079214e-01 -2.51360148e-01 2.50000328e-01 -4.93113734e-02
4.43549901e-02 -6.55078888e-02 -1.31796765e+00 1.38394523e+00
-4.69820589e-01 9.94024992e-01 1.30263492e-01 -8.76362145e-01
6.79677427e-01 4.38509077e-01 6.49084628e-01 -8.26442480e-01
6.91510213e-04 -3.41910064e-01 -1.22187927e-01 -9.49407279e-01
8.93567577e-02 4.34586890e-02 7.49835595e-02 8.88242304e-01
-3.68026286e-01 3.91682327e-01 1.62526593e-01 -1.32083312e-01
1.74775648e+00 6.43336713e-01 2.27318481e-01 6.55937731e-01
4.12063628e-01 4.12010327e-02 2.30036601e-01 1.08634830e+00
-4.61885273e-01 5.98427534e-01 -6.65155873e-02 -9.36198413e-01
-6.56180024e-01 -1.22454774e+00 2.92201787e-01 1.36332214e+00
1.68537408e-01 -9.32612643e-03 -5.22867084e-01 -7.50857651e-01
8.05763304e-02 4.41965252e-01 -6.54640257e-01 -1.21901602e-01
-6.29006147e-01 -3.21694434e-01 5.49972236e-01 9.80358541e-01
6.90842807e-01 -1.18311107e+00 -1.28749907e+00 1.27281815e-01
-3.88415813e-01 -1.16390133e+00 -2.27920339e-01 -4.14791470e-03
-6.21900082e-01 -1.36463380e+00 -7.37110674e-01 -3.23274106e-01
6.31613910e-01 7.51579404e-01 1.02656937e+00 2.06384405e-01
-7.46029735e-01 1.24006581e+00 -2.30695024e-01 -4.74702507e-01
-4.22274396e-02 -5.94945312e-01 1.06614023e-01 2.45535269e-01
8.89391363e-01 -5.49034595e-01 -7.96966374e-01 5.55689156e-01
-3.72413307e-01 -6.01036809e-02 4.96949971e-01 3.00278932e-01
7.77982548e-02 -1.22103021e-01 -5.89286117e-03 -4.31322247e-01
2.61209279e-01 -5.88123091e-02 -2.09094986e-01 2.53203809e-01
1.31339476e-01 -5.88597693e-02 3.18234831e-01 -5.52538097e-01
-1.19214201e+00 7.15775609e-01 3.27908218e-01 -1.96496069e-01
-8.04496288e-01 -3.39746267e-01 3.00303865e-02 -2.09317897e-02
1.14055490e+00 -1.72493830e-02 -6.00527599e-02 -3.33250403e-01
3.78620058e-01 8.64311397e-01 6.66725934e-01 -2.11184353e-01
5.44049323e-01 1.15638340e+00 2.59317900e-03 -1.14882278e+00
-9.48347867e-01 -6.83995724e-01 -9.63548958e-01 -6.27006888e-01
1.20221341e+00 -8.88556957e-01 -9.32114840e-01 4.40818071e-01
-1.16899526e+00 -6.60603046e-01 -2.73341894e-01 8.26062739e-01
-7.55593717e-01 3.92704993e-01 -2.50273973e-01 -1.19207537e+00
1.76242918e-01 -6.78295255e-01 1.23643482e+00 2.65657008e-01
-6.30308688e-01 -7.22979605e-01 2.75072128e-01 8.63068521e-01
8.07049200e-02 3.32048088e-01 3.35669696e-01 -1.07764371e-01
-5.40186942e-01 -1.54436126e-01 -6.48849532e-02 1.12647943e-01
2.56230503e-01 -2.26333201e-01 -1.15298665e+00 3.01793776e-02
-2.63986081e-01 -4.02004600e-01 1.12241280e+00 6.04212701e-01
8.31832051e-01 1.29801959e-01 -7.60308921e-01 2.60361820e-01
9.50481474e-01 2.40247890e-01 6.89690650e-01 3.54610503e-01
5.51008821e-01 5.71702540e-01 3.10169816e-01 3.52872461e-01
3.11752260e-01 7.99920619e-01 2.83207327e-01 -4.02793616e-01
-5.48798978e-01 4.65871729e-02 6.85318828e-01 -3.67994338e-01
-6.20305955e-01 -1.62354007e-01 -8.64153326e-01 2.83551306e-01
-2.03528881e+00 -1.50997794e+00 -3.71245027e-01 2.11171579e+00
2.26229027e-01 -2.97719948e-02 4.43174720e-01 2.67409086e-01
7.49350250e-01 1.72183365e-01 -6.88122511e-01 -3.88086915e-01
3.45009178e-01 1.51604027e-01 4.55585659e-01 3.27991724e-01
-1.40906727e+00 6.39530003e-01 7.52483988e+00 -1.35686636e-01
-8.23125243e-01 -2.38751411e-01 1.22881189e-01 -3.85014355e-01
3.85624379e-01 -2.10893750e-01 -5.65182865e-01 1.86595932e-01
7.60373294e-01 5.52401185e-01 5.56113243e-01 6.03779078e-01
5.67954659e-01 -5.83416820e-01 -1.32688975e+00 1.10199916e+00
4.37586904e-01 -5.89412689e-01 -4.16820318e-01 2.65542030e-01
2.60474384e-01 -1.24697626e-01 -4.02476713e-02 1.18243448e-01
7.99745619e-01 -1.23466754e+00 4.33320403e-01 8.71608913e-01
4.54925925e-01 -3.21983010e-01 3.01339656e-01 3.39291573e-01
-1.25881159e+00 -3.82020622e-01 -1.31598398e-01 -4.31994915e-01
3.09591651e-01 2.05830932e-01 -7.69975305e-01 -9.70215797e-02
7.25496590e-01 7.29777813e-01 -5.82659543e-01 8.59171927e-01
-4.50072140e-01 4.19895828e-01 -3.99211943e-01 1.16268158e-01
-4.23397757e-02 7.11650029e-02 2.20415458e-01 1.05177855e+00
1.87978268e-01 3.08912575e-01 4.87067014e-01 4.41445708e-01
3.17246825e-01 -4.63399947e-01 -8.85917187e-01 -1.04990555e-02
-3.46122170e-03 9.68937159e-01 -6.98452055e-01 -3.11573118e-01
-7.24464178e-01 1.21449590e+00 7.78114051e-02 6.61818266e-01
-6.24140084e-01 -9.50497314e-02 6.31562352e-01 2.66482115e-01
3.21983814e-01 -5.40320873e-01 8.01401436e-02 -9.90513563e-01
4.92116623e-02 -6.95818841e-01 4.08416003e-01 -1.20894682e+00
-7.68683016e-01 6.06293604e-02 -1.85460597e-01 -1.35141563e+00
-3.42089802e-01 -8.00913095e-01 -4.96438980e-01 4.75231916e-01
-1.04949880e+00 -1.26077735e+00 -6.47504508e-01 7.52931058e-01
5.94609380e-01 -5.78657724e-02 8.68244767e-01 1.76290777e-02
-2.81161934e-01 5.41215427e-02 -1.85930669e-01 4.47604626e-01
6.56061769e-01 -1.01247668e+00 4.70395982e-01 7.53584445e-01
6.65471733e-01 3.53614837e-01 6.03783309e-01 -5.79512417e-01
-1.29487157e+00 -8.41100097e-01 6.54853880e-01 -9.07098353e-01
2.59453207e-01 -1.61498740e-01 -3.21451485e-01 8.30281198e-01
-1.19817361e-01 2.13047594e-01 7.62833595e-01 2.31727108e-01
-4.41900432e-01 4.62716781e-02 -9.85926509e-01 6.79470479e-01
1.44894946e+00 -5.56645751e-01 -8.54860306e-01 5.48102379e-01
-1.30862564e-01 -3.79170701e-02 -2.63650894e-01 -5.22886962e-02
1.21282327e+00 -1.19388688e+00 1.42350352e+00 -7.12632358e-01
-5.53464107e-02 -4.93470222e-01 -1.07591026e-01 -1.02018881e+00
-2.77207524e-01 -1.50146216e-01 -2.53615201e-01 4.13127273e-01
6.44916072e-02 -3.77009362e-01 8.79872739e-01 6.28614962e-01
5.17612875e-01 -1.12782612e-01 -6.65359437e-01 -7.39519715e-01
-6.12763882e-01 -6.13450527e-01 1.71364054e-01 3.00406635e-01
-1.83529295e-02 2.30822638e-01 -5.28838873e-01 1.14987073e-02
5.38701653e-01 9.03835371e-02 1.18897152e+00 -1.35916090e+00
-6.28732622e-01 -1.32134080e-01 -6.82864487e-01 -1.29729772e+00
-6.35976270e-02 -3.82726222e-01 1.95203438e-01 -2.12642694e+00
-3.81074526e-04 1.86371163e-01 -3.03571206e-03 8.51787925e-01
1.86705038e-01 6.26633644e-01 2.20949784e-01 1.17950402e-01
-7.44002998e-01 3.36614735e-02 1.08153605e+00 -2.67398745e-01
-1.35713637e-01 1.41967341e-01 -5.13994098e-01 1.26145267e+00
6.83693588e-01 -1.40159339e-01 -4.17420268e-01 -5.42794347e-01
-1.22318424e-01 -9.92881060e-02 8.70536447e-01 -1.47547245e+00
2.03731030e-01 -4.82881755e-01 1.29860187e+00 -2.63576508e-01
8.00022542e-01 -9.40735936e-01 1.01715766e-01 5.62502861e-01
-4.56414014e-01 -3.71669233e-01 -1.46650106e-01 8.53344560e-01
4.79299694e-01 1.07851222e-01 6.51374698e-01 -6.22283459e-01
-7.05998838e-01 -8.10023174e-02 -9.62474644e-01 -1.24684051e-01
9.29457366e-01 -7.62371600e-01 -4.35326040e-01 -7.08220065e-01
-1.03658760e+00 -8.87061562e-03 3.52119565e-01 5.04032016e-01
5.89176893e-01 -1.27078664e+00 -4.25959170e-01 2.83905715e-01
-3.66949625e-02 -5.99672079e-01 1.67658955e-01 7.45539904e-01
-2.56721914e-01 3.76152456e-01 -2.93790579e-01 -6.83142722e-01
-1.71714807e+00 2.88489908e-01 3.96030486e-01 5.20258546e-02
-8.15500140e-01 5.49100697e-01 8.77971873e-02 2.65487432e-02
4.03109908e-01 -2.44639501e-01 -4.73535776e-01 -2.20013961e-01
9.44055259e-01 6.34660184e-01 -4.73794639e-01 -8.85892212e-01
-4.15120006e-01 9.64148939e-01 2.21633568e-01 -4.57822412e-01
1.03853488e+00 -1.26017287e-01 6.01922870e-01 4.18492049e-01
6.59032345e-01 -2.44192690e-01 -1.56448925e+00 -2.10242376e-01
-2.72942901e-01 -7.22863913e-01 -7.70638436e-02 -8.53944421e-01
-7.96448886e-01 1.03607452e+00 8.38519335e-01 2.37073928e-01
8.61165702e-01 6.06533289e-02 7.41842508e-01 7.74397612e-01
4.65604812e-01 -1.52039897e+00 5.64842939e-01 3.06805819e-01
6.29335582e-01 -1.07380855e+00 2.60837585e-01 -1.88655376e-01
-5.62971413e-01 1.26646292e+00 7.25455463e-01 1.10454280e-02
2.86039501e-01 2.42683470e-01 3.89898360e-01 -2.75032192e-01
-5.68453133e-01 -7.76803672e-01 1.19878739e-01 1.26109159e+00
3.37718606e-01 -2.52940711e-02 2.61702716e-01 -1.55036330e-01
3.82698402e-02 3.57709080e-01 3.40703994e-01 1.30074942e+00
-8.35155606e-01 -9.57037628e-01 -7.47389615e-01 4.21473324e-01
-2.23174781e-01 4.59960341e-01 -8.25693488e-01 6.42964184e-01
4.59912568e-01 1.45083237e+00 1.23797014e-01 -2.83011317e-01
4.28518057e-01 3.11732739e-01 8.80965889e-01 -7.12413311e-01
-2.78017342e-01 -1.04036212e-01 2.84303069e-01 -1.02622509e+00
-9.26903248e-01 -7.74461746e-01 -1.31234097e+00 8.48449916e-02
4.69639115e-02 -6.19588673e-01 5.32357275e-01 1.08501232e+00
1.91566169e-01 4.59901929e-01 4.23177272e-01 -9.88714874e-01
-1.36318833e-01 -7.14007616e-01 -4.30844188e-01 4.09502447e-01
5.41184366e-01 -8.49186957e-01 -4.10010666e-01 4.16136354e-01] | [8.05561351776123, 0.3702184855937958] |
9618451a-e878-4e49-944b-78bb4e1d2f9b | constraint-translation-candidates-a-bridge | 2010.13658 | null | https://arxiv.org/abs/2010.13658v1 | https://arxiv.org/pdf/2010.13658v1.pdf | Constraint Translation Candidates: A Bridge between Neural Query Translation and Cross-lingual Information Retrieval | Query translation (QT) is a key component in cross-lingual information retrieval system (CLIR). With the help of deep learning, neural machine translation (NMT) has shown promising results on various tasks. However, NMT is generally trained with large-scale out-of-domain data rather than in-domain query translation pairs. Besides, the translation model lacks a mechanism at the inference time to guarantee the generated words to match the search index. The two shortages of QT result in readable texts for human but inadequate candidates for the downstream retrieval task. In this paper, we propose a novel approach to alleviate these problems by limiting the open target vocabulary search space of QT to a set of important words mined from search index database. The constraint translation candidates are employed at both of training and inference time, thus guiding the translation model to learn and generate well performing target queries. The proposed methods are exploited and examined in a real-word CLIR system--Aliexpress e-Commerce search engine. Experimental results demonstrate that our approach yields better performance on both translation quality and retrieval accuracy than the strong NMT baseline. | ['Boxing Chen', 'Weihua Luo', 'Haibo Zhang', 'Baosong Yang', 'Liang Yao', 'Tianchi Bi'] | 2020-10-26 | null | null | null | null | ['cross-lingual-information-retrieval'] | ['natural-language-processing'] | [ 2.29111180e-01 -3.52958679e-01 -6.62216723e-01 -2.14742109e-01
-1.68845642e+00 -6.83099627e-01 7.09911466e-01 -1.77261844e-01
-7.03605413e-01 7.55205154e-01 3.02570373e-01 -5.14458299e-01
-7.71632269e-02 -7.41392732e-01 -6.51367605e-01 -5.05888164e-01
6.20719850e-01 1.07087326e+00 1.48827927e-02 -6.31412685e-01
1.77342832e-01 6.76941201e-02 -8.54943454e-01 4.83876169e-01
1.14663553e+00 8.40322316e-01 5.47870100e-01 1.40810341e-01
-4.31334317e-01 8.93141031e-02 -4.62957025e-01 -6.37531579e-01
4.43648636e-01 -4.25354809e-01 -7.76604950e-01 -5.11439383e-01
2.15216860e-01 -3.28392386e-01 -2.92427689e-01 1.09692526e+00
9.99707222e-01 4.94948663e-02 6.83723271e-01 -7.90774405e-01
-1.19286954e+00 7.25332797e-01 -3.96939605e-01 3.55272323e-01
1.03522331e-01 -1.68431595e-01 1.43239605e+00 -1.20099783e+00
6.85436726e-01 1.24091780e+00 1.00920551e-01 5.79139233e-01
-8.87230754e-01 -6.32653058e-01 -2.79679716e-01 1.15290962e-01
-1.56807888e+00 -4.93078262e-01 4.61261451e-01 4.94564287e-02
1.21976244e+00 2.83680320e-01 1.28902361e-01 1.05171168e+00
3.50851864e-01 1.01200712e+00 6.37803197e-01 -7.14218259e-01
-1.96291685e-01 2.57340670e-01 -1.50463805e-01 3.73164803e-01
-5.28005324e-02 1.96649447e-01 -5.30545890e-01 -2.64305353e-01
7.33576357e-01 -1.88205913e-01 -7.10164681e-02 6.98002353e-02
-1.30738521e+00 1.02103770e+00 3.42750877e-01 3.45686287e-01
-4.68151301e-01 -8.19291025e-02 4.55002904e-01 6.67127371e-01
5.47462702e-01 5.20723045e-01 -6.90879643e-01 2.60118067e-01
-9.38606203e-01 2.02168092e-01 5.29833615e-01 1.31909382e+00
5.80107570e-01 -2.54090428e-01 -6.59056187e-01 1.06641352e+00
3.65607530e-01 1.07235861e+00 8.77431989e-01 -3.46365243e-01
8.91807199e-01 4.91336852e-01 1.92076415e-01 -7.62259066e-01
2.45289266e-01 -6.91941142e-01 -7.47076273e-01 -7.79427052e-01
8.49678218e-02 7.14176595e-02 -8.33176076e-01 1.59464324e+00
8.29675794e-02 -4.79107559e-01 3.76913607e-01 1.09404218e+00
6.64151013e-01 9.69733953e-01 -8.62324536e-02 -3.16015512e-01
1.49251056e+00 -1.17770767e+00 -8.88476312e-01 -3.69356990e-01
9.62255299e-01 -1.35426366e+00 1.35466206e+00 -2.64921561e-02
-1.20194590e+00 -5.93078852e-01 -7.85319567e-01 -3.87083709e-01
-4.06870961e-01 4.08567607e-01 3.30811918e-01 1.79105505e-01
-1.01966870e+00 5.83449639e-02 -4.97334421e-01 -3.69729280e-01
-1.26034170e-01 3.16456705e-01 1.20913498e-02 -3.60332400e-01
-1.84693956e+00 1.09225476e+00 1.77323073e-01 2.10591584e-01
-6.75896943e-01 -3.16253096e-01 -4.43922490e-01 5.58274128e-02
2.02622622e-01 -8.46662283e-01 1.56072891e+00 -9.38728154e-01
-1.22150719e+00 8.97726178e-01 -4.97937113e-01 -3.12006265e-01
4.85436469e-01 -4.69211161e-01 -4.34041023e-01 -8.19897726e-02
4.78784323e-01 6.83710754e-01 6.57915711e-01 -7.33732700e-01
-6.44193411e-01 -3.82036954e-01 -3.03220332e-01 6.03145838e-01
-4.42471802e-01 4.94396687e-01 -9.28384066e-01 -6.57577932e-01
-1.72871270e-03 -8.78532767e-01 -1.57864973e-01 -4.99177277e-01
-2.59890914e-01 -8.02559495e-01 4.54160422e-01 -7.11936772e-01
1.61000228e+00 -1.74324667e+00 1.25653937e-01 7.48839304e-02
-3.45317662e-01 4.43964422e-01 -3.83629054e-01 8.66953552e-01
3.07711869e-01 1.16427399e-01 2.47422650e-01 -5.79905733e-02
1.06857397e-01 2.10304886e-01 -7.36152291e-01 -9.91861057e-03
1.08763956e-01 1.53093004e+00 -8.19558382e-01 -8.47798169e-01
-2.24428505e-01 2.55354404e-01 -1.95267186e-01 3.36317450e-01
-4.35130268e-01 2.29536965e-01 -1.10744381e+00 6.32170260e-01
2.21605778e-01 -2.16405809e-01 2.43998524e-02 -1.89394988e-02
4.54912037e-02 8.47812772e-01 -4.24879551e-01 2.12074018e+00
-7.27900088e-01 2.88809747e-01 -3.42188269e-01 -8.62238884e-01
9.85259593e-01 6.69479489e-01 2.75753111e-01 -1.41862643e+00
9.87059325e-02 7.49541044e-01 7.60898972e-03 -6.34420395e-01
6.75835669e-01 5.83502613e-02 -2.29779720e-01 7.65461326e-01
-2.16071889e-01 3.19346227e-02 1.91290841e-01 -1.47355366e-02
6.55075908e-01 3.01910639e-01 -1.42985269e-01 -3.29784095e-01
6.66038752e-01 4.13044930e-01 3.81716758e-01 7.01774955e-01
9.99637917e-02 3.88095498e-01 -2.11019099e-01 -3.74639004e-01
-1.24909127e+00 -7.02955961e-01 -9.11983196e-04 1.43203628e+00
1.97865680e-01 -2.84803398e-02 -6.57229245e-01 -5.06383717e-01
-3.10708374e-01 8.01168025e-01 -1.93994552e-01 -3.23569536e-01
-1.07320309e+00 -6.68665230e-01 6.88213468e-01 2.06675947e-01
4.10226166e-01 -1.31910217e+00 -1.31426930e-01 3.64174753e-01
-8.06911647e-01 -1.07740879e+00 -1.27679598e+00 -6.27689436e-02
-8.96761298e-01 -4.97197360e-01 -8.89966309e-01 -1.41558063e+00
4.80458409e-01 5.22569418e-01 1.37216878e+00 1.47872463e-01
1.30744567e-02 -1.38374776e-01 -4.29773599e-01 -2.62774229e-01
-5.62482059e-01 5.89572966e-01 -3.95687222e-02 -2.65138835e-01
1.08582234e+00 -1.21068694e-01 -8.29831839e-01 5.67170143e-01
-9.68544364e-01 6.56921230e-03 1.04506660e+00 1.05817330e+00
7.59077847e-01 -3.54569882e-01 9.02699769e-01 -3.50210965e-01
1.19820547e+00 -3.71975243e-01 -7.20049918e-01 8.43371034e-01
-1.03490949e+00 3.76239359e-01 4.47641522e-01 -5.62653005e-01
-9.56038892e-01 -3.92927021e-01 -2.48677842e-03 -2.96725899e-01
4.01235163e-01 7.91980982e-01 -1.33937821e-01 4.68771368e-01
6.35488033e-01 7.75266945e-01 -1.84778735e-01 -5.79673409e-01
3.52328956e-01 9.94178295e-01 1.14340588e-01 -6.50998294e-01
7.04458475e-01 -1.81723759e-01 -4.34991509e-01 -1.94503456e-01
-7.94628501e-01 -6.78091109e-01 -5.29832602e-01 2.36117020e-01
7.64900744e-01 -9.92735505e-01 -2.32395768e-01 -7.94247612e-02
-1.30494928e+00 4.76497523e-02 2.01798424e-01 5.26403069e-01
-2.79639155e-01 7.55425766e-02 -7.28152573e-01 -5.77041626e-01
-1.19535339e+00 -1.31179690e+00 1.60863590e+00 -1.55766467e-02
-1.26547739e-02 -8.70575130e-01 1.54345632e-01 4.85263467e-01
5.27766824e-01 -8.17440689e-01 1.18842089e+00 -1.01151490e+00
-7.33404398e-01 -3.41086715e-01 -2.63335794e-01 1.65392101e-01
1.56540751e-01 -4.59519774e-01 -5.67280889e-01 -3.35075349e-01
-9.43653956e-02 -4.99843985e-01 5.67691267e-01 8.64866003e-02
6.30033255e-01 -4.76438344e-01 -4.37007576e-01 2.11566478e-01
1.46728885e+00 2.39555389e-01 4.58279431e-01 4.48297322e-01
3.07514548e-01 6.80939555e-01 1.00216436e+00 -2.85103053e-01
2.34048635e-01 1.10400963e+00 -1.50580928e-01 -1.68587893e-01
-1.00490354e-01 -6.59809530e-01 3.32705289e-01 1.22783816e+00
3.43130082e-01 -5.27512014e-01 -8.31784129e-01 6.42234623e-01
-1.90665686e+00 -6.88094318e-01 1.65830359e-01 2.22289777e+00
1.41558015e+00 -6.90461472e-02 -3.91051352e-01 -5.50952852e-01
6.40224874e-01 -3.41624647e-01 -5.97726047e-01 -2.66006857e-01
-2.00300172e-01 3.37175131e-01 5.13533831e-01 6.28922880e-01
-6.76008940e-01 1.41027915e+00 5.49745131e+00 1.27319527e+00
-1.04342830e+00 2.65684426e-01 4.56549674e-01 8.15022811e-02
-5.05418181e-01 -1.35791808e-01 -1.07662249e+00 1.83723748e-01
9.36215878e-01 -4.00444388e-01 4.69652593e-01 5.90878010e-01
3.80665898e-01 4.05864954e-01 -1.14041007e+00 8.83832455e-01
1.20499739e-02 -1.11653745e+00 4.39584166e-01 1.55988876e-02
6.26463532e-01 3.71744871e-01 2.18619987e-01 6.87277734e-01
2.71234959e-01 -1.04390264e+00 4.96961445e-01 1.62813038e-01
9.69414771e-01 -6.58919215e-01 8.15713286e-01 7.39459157e-01
-9.36675072e-01 2.70302057e-01 -6.45947576e-01 4.55793023e-01
3.27760400e-03 1.62142336e-01 -1.08313143e+00 6.63480520e-01
3.08574110e-01 3.51583332e-01 -1.92147940e-01 5.26690543e-01
-2.16275379e-01 3.90518993e-01 -1.98904246e-01 -3.78095090e-01
5.99503338e-01 -4.70995158e-01 3.39916587e-01 1.06267405e+00
4.89188731e-01 -1.78093538e-01 2.43871689e-01 8.17398489e-01
-3.57884854e-01 6.61976695e-01 -6.71009719e-01 -2.30679974e-01
6.23946786e-01 1.08077407e+00 -1.45185485e-01 -4.34731066e-01
-4.59279537e-01 1.18269515e+00 2.34301433e-01 6.12404048e-01
-5.96603811e-01 -2.29794472e-01 4.82029080e-01 -5.92018403e-02
3.05345245e-02 -5.92405982e-02 3.45670544e-02 -1.26941741e+00
3.93231899e-01 -1.29043949e+00 2.92873859e-01 -7.39292204e-01
-1.40402818e+00 9.15461123e-01 -1.19592257e-01 -1.40822864e+00
-6.49104834e-01 -2.05055952e-01 -9.34272632e-02 1.56317294e+00
-1.88625145e+00 -1.43614769e+00 4.65770066e-01 6.59878850e-01
1.06601346e+00 -3.42460692e-01 1.03721809e+00 6.80118680e-01
-3.88411999e-01 9.08941984e-01 6.14709496e-01 3.48416924e-01
9.98858035e-01 -7.19821990e-01 5.53907037e-01 6.36212051e-01
2.38735557e-01 1.09289563e+00 4.67165738e-01 -7.07421124e-01
-1.66667688e+00 -1.02275515e+00 1.64097476e+00 -5.58985353e-01
6.61058247e-01 -3.82333010e-01 -6.99271321e-01 4.51998383e-01
5.87213874e-01 -4.89939749e-01 5.94780982e-01 -8.27082321e-02
-2.65937150e-01 -1.06417499e-01 -7.12174118e-01 7.94501126e-01
7.62867212e-01 -6.64811969e-01 -8.85983527e-01 6.94838762e-01
1.26421881e+00 -2.46541619e-01 -6.54899061e-01 4.33682621e-01
6.57747328e-01 -1.43764034e-01 1.08509386e+00 -8.51392448e-01
3.91159147e-01 -6.08376041e-02 -2.92446405e-01 -1.14476359e+00
-5.12142964e-02 -5.64840972e-01 1.68323547e-01 9.96468306e-01
9.02358472e-01 -5.43638885e-01 4.41905737e-01 4.45478022e-01
-1.70322150e-01 -7.52362132e-01 -9.87313926e-01 -6.86441362e-01
5.23788273e-01 -2.32340246e-02 6.64966106e-01 6.82935834e-01
-1.01428010e-01 1.02945101e+00 -1.92893878e-01 -9.21466425e-02
3.24587852e-01 4.71122682e-01 5.52984059e-01 -7.71031380e-01
-2.20153928e-01 -4.68964547e-01 5.05680799e-01 -1.67847896e+00
1.87345386e-01 -1.21559131e+00 1.60822064e-01 -1.46992588e+00
3.38039696e-01 -6.80187047e-01 -4.41269457e-01 2.20264480e-01
-3.40696990e-01 7.17535466e-02 -2.81293094e-01 8.15003693e-01
-4.16957527e-01 6.73004031e-01 1.59913206e+00 -2.89805025e-01
-1.18349887e-01 1.17538936e-01 -5.71498454e-01 -2.84894537e-02
6.95604801e-01 -6.19796157e-01 -6.43996000e-01 -1.13811922e+00
4.49524313e-01 3.49868506e-01 -1.52553916e-01 -7.97751620e-02
2.05970913e-01 -2.92507172e-01 1.02282584e-01 -7.59603202e-01
1.79795146e-01 -7.99129426e-01 -2.57379889e-01 4.20286745e-01
-9.03418481e-01 6.05002582e-01 -8.17735940e-02 2.76234329e-01
-4.29640323e-01 -2.38735527e-01 2.59854674e-01 -2.60100156e-01
-2.85284966e-01 4.65337873e-01 -5.00169992e-02 1.76060468e-01
2.75715232e-01 2.13155180e-01 -2.59802401e-01 -3.99585158e-01
-1.74591348e-01 4.97677118e-01 8.00936967e-02 8.66983950e-01
5.12395740e-01 -1.41178906e+00 -1.07962859e+00 2.27869093e-01
4.44607466e-01 -1.00535125e-01 -4.10268158e-01 6.44244850e-01
-1.97159976e-01 1.27342558e+00 2.44012401e-01 -4.88981962e-01
-9.64150310e-01 6.98517084e-01 2.69129544e-01 -7.67798424e-01
-2.31716707e-01 8.65911484e-01 1.69635132e-01 -6.99972332e-01
2.83390701e-01 -5.95436692e-02 -1.17999934e-01 -2.58512825e-01
4.89111006e-01 -2.80333966e-01 3.84039909e-01 -5.95537961e-01
4.87373769e-02 4.41899449e-01 -4.61028129e-01 -4.18931484e-01
9.16193068e-01 -5.35718918e-01 -3.76373380e-01 2.81727733e-03
1.38511193e+00 -1.67777076e-01 -1.96969420e-01 -9.28969562e-01
3.25110674e-01 -3.91862959e-01 5.76952882e-02 -9.74592388e-01
-6.89288616e-01 9.81917977e-01 5.10407388e-01 -1.78533942e-01
1.13550174e+00 -9.46894735e-02 1.52531970e+00 8.37936342e-01
5.64445138e-01 -1.21949673e+00 -2.50477381e-02 5.46661139e-01
8.86955976e-01 -1.43369913e+00 -4.39053178e-01 2.72452720e-02
-4.85765517e-01 9.01087999e-01 5.80432117e-01 2.69876689e-01
2.82713652e-01 -2.92327285e-01 5.25679410e-01 -1.14969157e-01
-9.70081151e-01 -5.01001067e-02 5.97950101e-01 2.08020568e-01
7.14083612e-01 -2.13888288e-01 -8.33839774e-01 2.83094585e-01
-1.65529132e-01 -9.71859694e-02 -2.73458093e-01 7.60806024e-01
-4.79247779e-01 -1.57293332e+00 -1.77062199e-01 3.06082577e-01
-7.96933830e-01 -8.90222728e-01 -4.17593181e-01 4.19812024e-01
-4.44508970e-01 1.05776989e+00 -2.14395672e-01 -7.82659948e-02
2.10742608e-01 3.86002541e-01 2.16535553e-01 -5.57460070e-01
-5.98820150e-01 5.76538324e-01 3.66744138e-02 -3.13481212e-01
-1.68207332e-01 -3.13531220e-01 -9.35343206e-01 -1.61761642e-02
-8.01690757e-01 7.26624310e-01 7.59486854e-01 9.60075319e-01
4.43248332e-01 1.00261554e-01 7.37517774e-01 -1.89001411e-02
-1.14093435e+00 -1.36682916e+00 -1.11869231e-01 2.74422079e-01
4.81502116e-02 -1.57187134e-01 9.30669755e-02 9.40552633e-03] | [11.565766334533691, 10.009098052978516] |
e314bd0c-23e6-4ac2-91df-cbed759b1578 | temporal-roi-align-for-video-object | 2109.03495 | null | https://arxiv.org/abs/2109.03495v2 | https://arxiv.org/pdf/2109.03495v2.pdf | Temporal RoI Align for Video Object Recognition | Video object detection is challenging in the presence of appearance deterioration in certain video frames. Therefore, it is a natural choice to aggregate temporal information from other frames of the same video into the current frame. However, RoI Align, as one of the most core procedures of video detectors, still remains extracting features from a single-frame feature map for proposals, making the extracted RoI features lack temporal information from videos. In this work, considering the features of the same object instance are highly similar among frames in a video, a novel Temporal RoI Align operator is proposed to extract features from other frames feature maps for current frame proposals by utilizing feature similarity. The proposed Temporal RoI Align operator can extract temporal information from the entire video for proposals. We integrate it into single-frame video detectors and other state-of-the-art video detectors, and conduct quantitative experiments to demonstrate that the proposed Temporal RoI Align operator can consistently and significantly boost the performance. Besides, the proposed Temporal RoI Align can also be applied into video instance segmentation. Codes are available at https://github.com/open-mmlab/mmtracking | ['Huamin Feng', 'Nenghai Yu', 'Dahua Lin', 'Feng Zhu', 'Qi Chu', 'Xinjiang Wang', 'Kai Chen', 'Tao Gong'] | 2021-09-08 | null | null | null | null | ['video-instance-segmentation'] | ['computer-vision'] | [ 2.11476237e-02 -5.15159905e-01 -4.13302749e-01 -2.93269694e-01
-4.93814081e-01 -4.73175615e-01 1.53255731e-01 -2.01954674e-02
-4.79865491e-01 3.16867769e-01 -1.23187862e-01 1.50612429e-01
9.99784097e-02 -5.04652560e-01 -6.04694545e-01 -7.30694354e-01
4.08263020e-02 -3.83731037e-01 1.11093223e+00 1.37771562e-01
2.38206893e-01 5.12004614e-01 -1.33729994e+00 3.48304778e-01
5.06081462e-01 1.23798788e+00 4.56791371e-01 4.77608621e-01
-9.21237096e-02 9.09723461e-01 -3.05188030e-01 9.03301910e-02
3.24938148e-01 -4.55689162e-01 -6.95320129e-01 5.08020997e-01
4.38363552e-01 -7.72494972e-01 -7.77206302e-01 1.12360263e+00
9.84596685e-02 3.98648739e-01 2.65026391e-01 -1.34537816e+00
-7.72209018e-02 4.51204598e-01 -8.28557014e-01 9.48485315e-01
5.57111382e-01 2.05331638e-01 7.54601419e-01 -1.02479994e+00
7.91855335e-01 1.16436899e+00 2.38505885e-01 2.28038043e-01
-6.41324997e-01 -6.75430954e-01 6.52294457e-01 5.84785759e-01
-1.49875331e+00 -4.70229149e-01 7.76881337e-01 -4.10246164e-01
4.86817539e-01 1.95552155e-01 8.12357008e-01 5.94273627e-01
7.94982444e-03 1.36360955e+00 5.87442517e-01 -1.22753128e-01
-5.28988726e-02 -9.50159505e-02 2.04668328e-01 8.03816795e-01
-5.36704846e-02 -3.60598974e-02 -6.00367129e-01 1.21506266e-01
7.16523051e-01 4.77444410e-01 -5.14150381e-01 -1.88745573e-01
-1.25053120e+00 4.63864595e-01 4.36739773e-01 4.83303577e-01
-3.88989836e-01 1.04402080e-01 3.40125293e-01 6.37511462e-02
2.10474923e-01 -3.65665287e-01 -3.22185457e-01 -1.36310190e-01
-1.20266736e+00 2.19593659e-01 1.91349000e-01 1.02223933e+00
7.88960755e-01 2.08417494e-02 -3.25385749e-01 4.67144638e-01
3.97229433e-01 3.80535543e-01 4.08804566e-01 -1.03612089e+00
4.62522835e-01 6.29877985e-01 1.19950175e-01 -1.17974222e+00
-2.74536282e-01 -7.41692930e-02 -3.74268472e-01 -2.58915931e-01
4.04726237e-01 -8.78481343e-02 -7.77239919e-01 1.37403738e+00
7.28084445e-01 6.03197634e-01 -2.50682324e-01 1.21293604e+00
1.02819228e+00 8.74916732e-01 2.66494602e-02 -7.06150353e-01
1.34067261e+00 -1.02181113e+00 -6.74586594e-01 -1.19057991e-01
3.58903408e-01 -9.80320275e-01 3.94585133e-01 1.51984662e-01
-1.11140478e+00 -7.98004627e-01 -9.13219750e-01 1.77880570e-01
3.37444544e-02 2.57532448e-01 3.08038950e-01 2.24160239e-01
-7.49877512e-01 3.44616443e-01 -8.68992805e-01 -3.61755848e-01
4.57801074e-01 2.87321657e-01 -3.03252965e-01 -3.39700550e-01
-9.43748951e-01 3.27132165e-01 6.39030993e-01 2.68586367e-01
-8.50960851e-01 -3.49462450e-01 -7.91489601e-01 -2.09649086e-01
8.38139415e-01 -3.33735406e-01 1.19553792e+00 -1.10585976e+00
-9.53443110e-01 3.80843461e-01 -5.61211407e-01 -2.22418621e-01
4.93420750e-01 -2.09080189e-01 -4.55907345e-01 5.75632095e-01
1.90419585e-01 7.89983213e-01 1.10192931e+00 -8.76244187e-01
-1.18924344e+00 -1.88334003e-01 3.69600840e-02 1.23771377e-01
-3.09921950e-01 4.61809427e-01 -1.12726128e+00 -6.92405820e-01
3.54253352e-01 -8.79388034e-01 -1.33051425e-01 8.24152455e-02
-2.18661815e-01 -3.88911992e-01 1.20896173e+00 -6.98985457e-01
1.43580353e+00 -2.25527906e+00 3.52076557e-03 5.44218421e-02
1.50669917e-01 3.00240010e-01 -3.85474749e-02 -1.30813092e-01
9.73366499e-02 -6.62647337e-02 -4.93074916e-02 1.88612461e-01
-5.04145384e-01 -1.39375120e-01 2.43405234e-02 5.51625609e-01
2.98981339e-01 7.23393321e-01 -8.31880689e-01 -9.96928096e-01
4.96487409e-01 1.31559014e-01 -4.74273264e-01 2.15224743e-01
-3.05959256e-03 5.56602120e-01 -7.99441516e-01 8.59192967e-01
7.72800446e-01 -1.96318212e-03 -5.01732826e-02 -6.36326909e-01
-2.19471171e-01 -1.93252131e-01 -1.50239313e+00 1.56658876e+00
2.51673549e-01 7.19369709e-01 -1.08041540e-01 -1.06122923e+00
6.17238939e-01 3.83433998e-01 9.34000373e-01 -6.31147265e-01
2.95954227e-01 1.80677220e-01 1.53040543e-01 -7.78177321e-01
5.49470425e-01 4.06705409e-01 1.66546956e-01 2.65087217e-01
6.58012703e-02 3.57497931e-01 8.17047536e-01 1.32399037e-01
1.11914825e+00 2.03219920e-01 3.80257130e-01 8.63847062e-02
8.51699829e-01 -1.58033222e-01 1.06643176e+00 6.07184947e-01
-6.74292207e-01 6.57303810e-01 2.80135535e-02 -5.53502381e-01
-8.22507203e-01 -9.51150715e-01 -1.24730282e-01 9.32477593e-01
7.08584249e-01 -3.98108572e-01 -7.54831016e-01 -8.96324575e-01
-1.71811968e-01 -6.24605604e-02 -3.00048769e-01 1.06661305e-01
-8.09708118e-01 -5.10174870e-01 1.94999292e-01 6.28582537e-01
6.18146479e-01 -8.98657143e-01 -7.25015759e-01 4.45550680e-01
-4.39533919e-01 -1.43663383e+00 -8.89484286e-01 -3.05536836e-01
-8.85694921e-01 -1.24194932e+00 -8.14494550e-01 -8.58444691e-01
6.74991488e-01 9.08711851e-01 5.80719531e-01 3.55500907e-01
-4.32522982e-01 3.74746799e-01 -6.27461731e-01 2.12945968e-01
-1.44288272e-01 -2.35482961e-01 1.69343892e-02 3.63048762e-01
4.11781251e-01 -1.80639029e-01 -8.73059154e-01 7.88503528e-01
-1.00826824e+00 7.25380257e-02 3.36864322e-01 4.87897545e-01
6.28804147e-01 1.36571541e-01 3.02575916e-01 -3.33840221e-01
-1.35557260e-02 -4.06550467e-01 -5.95670342e-01 1.64867878e-01
1.14825167e-01 -4.15647954e-01 3.54732752e-01 -4.58869517e-01
-8.94040942e-01 2.44101152e-01 8.73204544e-02 -7.01275826e-01
-1.91007510e-01 1.98369741e-01 -1.71948239e-01 -2.94068959e-02
7.95981660e-02 4.04011637e-01 -1.69780895e-01 -2.67618716e-01
2.19363734e-01 5.57382286e-01 4.00007546e-01 -3.23237985e-01
8.20999146e-01 4.99709994e-01 -2.51157224e-01 -7.83101201e-01
-7.29469419e-01 -9.42824364e-01 -7.91632950e-01 -7.05850601e-01
9.40056503e-01 -1.00906372e+00 -5.65709710e-01 4.79277074e-01
-1.16603100e+00 1.67229086e-01 5.48315719e-02 7.02201426e-01
-3.91439348e-01 6.91902697e-01 -6.22166753e-01 -5.90990484e-01
-1.27729699e-01 -1.51210582e+00 9.36781287e-01 7.64326632e-01
-7.39429072e-02 -6.42650187e-01 -4.44966167e-01 3.41629684e-01
-3.26642133e-02 8.07894468e-02 2.08251566e-01 -4.25109535e-01
-1.04262841e+00 -1.58171728e-01 -2.33849391e-01 3.57783526e-01
3.72844487e-01 5.29268682e-01 -7.31476903e-01 -2.67582506e-01
1.55816585e-01 1.47317290e-01 9.99177516e-01 7.23958194e-01
9.86797869e-01 -1.63902909e-01 -5.91682732e-01 4.19065624e-01
1.19884384e+00 6.42517388e-01 3.70887697e-01 3.71351510e-01
7.80151665e-01 4.90491569e-01 1.32900870e+00 4.90871727e-01
1.61654264e-01 8.51513207e-01 1.92216411e-01 1.72451828e-02
5.45765348e-02 1.95064381e-01 7.68924117e-01 7.78783679e-01
-1.77189812e-01 -1.81019038e-01 -6.84677660e-01 4.95180130e-01
-2.02520418e+00 -1.16616762e+00 -1.38228968e-01 2.00101948e+00
4.50923592e-01 9.83099416e-02 4.15277660e-01 9.68820304e-02
1.25998855e+00 1.53306708e-01 -5.66551685e-01 1.71302557e-01
-8.09351057e-02 -3.46330673e-01 3.44314545e-01 8.04691091e-02
-1.45075822e+00 8.19850862e-01 5.10746241e+00 9.05005991e-01
-1.06181645e+00 1.61357269e-01 6.68198466e-01 -2.31622651e-01
2.83779711e-01 1.57764673e-01 -8.83948624e-01 7.67462134e-01
3.29216957e-01 -2.91404873e-01 1.14578165e-01 8.85257542e-01
6.19354248e-01 -4.53066438e-01 -1.13825428e+00 9.74930167e-01
8.19769129e-02 -1.21523452e+00 -2.94729881e-02 -1.70801148e-01
7.16261029e-01 -1.46846995e-01 -2.77784895e-02 -2.24442575e-02
-4.45493221e-01 -4.72041190e-01 9.15605247e-01 3.73250693e-01
2.48424262e-01 -6.27814412e-01 7.30776668e-01 7.52660036e-02
-1.88370156e+00 -3.29558372e-01 -3.14448088e-01 2.45077848e-01
3.78973335e-01 4.20084268e-01 -6.21220469e-01 6.01981580e-01
9.40364838e-01 1.29133534e+00 -6.83268607e-01 1.48206890e+00
8.04856867e-02 4.28924143e-01 -2.42934704e-01 3.64675462e-01
2.40959063e-01 -1.05033092e-01 6.95544660e-01 1.09239566e+00
5.16154051e-01 3.07963192e-01 7.01546788e-01 5.73171973e-01
8.54148045e-02 2.52014607e-01 -4.18664396e-01 5.27815819e-02
5.28787196e-01 1.30026996e+00 -9.86380398e-01 -5.04655302e-01
-7.70299137e-01 8.30638587e-01 -1.95077136e-01 3.78492057e-01
-1.14927948e+00 -7.24884793e-02 5.68795204e-01 1.14856109e-01
6.16674125e-01 -2.17148259e-01 3.64339441e-01 -1.19019628e+00
4.37593311e-01 -7.21104681e-01 5.73791444e-01 -7.22767949e-01
-9.39605117e-01 4.51854229e-01 -4.83229384e-02 -1.95365655e+00
-1.20509811e-01 -3.40901911e-01 -6.88355863e-01 2.71107823e-01
-1.37987769e+00 -8.74625862e-01 -6.55927658e-01 7.57562101e-01
1.04474866e+00 -1.78684853e-02 -5.70645630e-02 4.39393908e-01
-9.65524614e-01 3.91245037e-01 -1.98019650e-02 5.04918694e-01
5.39605260e-01 -6.34116471e-01 3.42154577e-02 1.30997205e+00
8.96557570e-02 5.30967414e-01 4.39339548e-01 -5.91209590e-01
-1.36007822e+00 -1.24096358e+00 8.32984969e-02 6.50159195e-02
5.52761018e-01 1.29823461e-01 -1.03395677e+00 5.19056797e-01
1.40960449e-02 5.26158035e-01 3.64841461e-01 -6.07142448e-01
1.83944613e-01 -1.92642540e-01 -8.69173110e-01 4.41189259e-01
1.03908706e+00 -1.83866844e-01 -4.81116354e-01 1.34900555e-01
7.31881440e-01 -5.55097163e-01 -9.24455523e-01 6.05734169e-01
5.31086862e-01 -9.15068209e-01 9.83444750e-01 -1.97187871e-01
1.93699136e-01 -9.29142952e-01 -1.17766276e-01 -6.52554810e-01
-1.13317423e-01 -6.15317464e-01 -2.23528743e-01 1.52206516e+00
-6.80001602e-02 -2.19439179e-01 6.21862113e-01 5.38343132e-01
5.92958666e-02 -4.92149919e-01 -9.59206462e-01 -7.57183611e-01
-5.54653049e-01 -5.55083692e-01 2.64934063e-01 6.15754545e-01
-1.52830467e-01 -1.58609837e-01 -2.88481921e-01 2.91211188e-01
5.86775303e-01 2.56119400e-01 7.43785858e-01 -9.27902639e-01
-3.15513276e-02 -3.83991927e-01 -8.01203430e-01 -1.22332466e+00
1.86863379e-03 -6.42142296e-01 2.21442699e-01 -1.29156601e+00
5.40927768e-01 -2.61372358e-01 -5.55391729e-01 3.02170217e-01
-4.96667802e-01 3.74154001e-01 6.09888554e-01 3.22234482e-01
-9.38403487e-01 4.56243247e-01 1.24384415e+00 -2.64216781e-01
-2.44625881e-01 4.56286855e-02 -1.01573810e-01 8.87889028e-01
6.67663693e-01 -6.14853799e-01 -2.36992255e-01 -3.41667235e-02
-4.45304245e-01 2.55447596e-01 2.15294123e-01 -1.27654254e+00
3.80967468e-01 -3.99463207e-01 5.71075797e-01 -8.05791557e-01
2.24759430e-01 -8.62769067e-01 9.36822966e-02 4.50062215e-01
9.97296795e-02 1.16422042e-01 2.47312397e-01 6.55161560e-01
-5.03507435e-01 -2.64901370e-01 7.35082507e-01 -7.67510310e-02
-1.32704246e+00 6.20506167e-01 -5.52756786e-01 3.41554172e-02
1.44653451e+00 -6.08180642e-01 -1.96614131e-01 -1.31574813e-02
-6.88821912e-01 3.77639741e-01 4.99815553e-01 6.97502971e-01
9.41298366e-01 -1.25619757e+00 -5.86883724e-01 -8.13205075e-03
2.86395047e-02 -3.91289592e-03 6.52541280e-01 1.18445396e+00
-4.51487988e-01 2.86088645e-01 -2.88140804e-01 -1.11113584e+00
-1.74233305e+00 6.69740677e-01 2.58305967e-01 2.63063401e-01
-6.37549698e-01 6.22082293e-01 5.37045062e-01 4.41675842e-01
6.88623870e-03 -4.58029896e-01 -2.98960954e-01 2.68725246e-01
7.84016728e-01 2.49097124e-01 -2.43952900e-01 -1.00592875e+00
-5.15978336e-01 8.61440957e-01 -3.54934841e-01 2.11611629e-01
1.07909787e+00 -5.17485738e-01 2.53390856e-02 3.73654038e-01
1.14241993e+00 -1.20562248e-01 -1.49044228e+00 -4.58416432e-01
-7.03290328e-02 -8.85777771e-01 -4.27416302e-02 2.18431074e-02
-1.55234885e+00 5.78210950e-01 6.94382131e-01 -6.96817711e-02
1.34256065e+00 2.64081955e-02 9.44726765e-01 3.58094797e-02
4.03514236e-01 -1.05074203e+00 3.07506949e-01 1.36746123e-01
4.35455859e-01 -1.26999962e+00 1.74543455e-01 -7.32977033e-01
-5.52262783e-01 1.32129312e+00 8.13446641e-01 2.45904215e-02
4.85082090e-01 3.35382298e-02 -1.23343267e-01 5.97140417e-02
-5.23522794e-01 -4.20938075e-01 3.96633387e-01 2.44054481e-01
3.19680691e-01 -2.10092857e-01 -3.43444973e-01 2.97153533e-01
5.06567836e-01 9.36833993e-02 7.19947934e-01 9.71236289e-01
-6.09755218e-01 -8.59340370e-01 -4.25930470e-01 3.96610737e-01
-5.31265259e-01 1.66543409e-01 1.01886764e-01 7.37320781e-01
2.04617575e-01 1.15733647e+00 2.27624662e-02 -3.78054887e-01
8.20616633e-02 -1.04788110e-01 3.72455865e-01 -3.25066090e-01
-3.27641487e-01 4.18027878e-01 -2.59095103e-01 -7.77992606e-01
-8.96798134e-01 -1.06252539e+00 -1.47897744e+00 -1.29542142e-01
-4.42015916e-01 1.27143655e-02 2.18316093e-01 1.01462960e+00
1.41688332e-01 5.53847730e-01 7.66463816e-01 -9.61250842e-01
1.48778766e-01 -7.08058655e-01 -4.26323503e-01 4.65359747e-01
3.30692947e-01 -8.39105546e-01 -1.12992898e-01 4.57779348e-01] | [9.143132209777832, -0.29532381892204285] |
283be3e4-9c2b-49a2-a945-c5d7155e5e00 | hybrid-indoor-localization-via-reinforcement | 2210.15132 | null | https://arxiv.org/abs/2210.15132v1 | https://arxiv.org/pdf/2210.15132v1.pdf | Hybrid Indoor Localization via Reinforcement Learning-based Information Fusion | The paper is motivated by the importance of the Smart Cities (SC) concept for future management of global urbanization. Among all Internet of Things (IoT)-based communication technologies, Bluetooth Low Energy (BLE) plays a vital role in city-wide decision making and services. Extreme fluctuations of the Received Signal Strength Indicator (RSSI), however, prevent this technology from being a reliable solution with acceptable accuracy in the dynamic indoor tracking/localization approaches for ever-changing SC environments. The latest version of the BLE v.5.1 introduced a better possibility for tracking users by utilizing the direction finding approaches based on the Angle of Arrival (AoA), which is more reliable. There are still some fundamental issues remaining to be addressed. Existing works mainly focus on implementing stand-alone models overlooking potentials fusion strategies. The paper addresses this gap and proposes a novel Reinforcement Learning (RL)-based information fusion framework (RL-IFF) by coupling AoA with RSSI-based particle filtering and Inertial Measurement Unit (IMU)-based Pedestrian Dead Reckoning (PDR) frameworks. The proposed RL-IFF solution is evaluated through a comprehensive set of experiments illustrating superior performance compared to its counterparts. | ['Arash Mohammadi', 'Mohammad Salimibeni'] | 2022-10-27 | null | null | null | null | ['indoor-localization'] | ['computer-vision'] | [-2.14126810e-01 -4.12540644e-01 -2.00806230e-01 -1.28989816e-01
-6.29471600e-01 -4.60476428e-02 7.97879815e-01 2.28938252e-01
-5.44105232e-01 1.36462653e+00 1.57823935e-01 -5.00018179e-01
-5.46597064e-01 -1.34976423e+00 -3.76813769e-01 -9.85944808e-01
-1.71762947e-02 2.39472508e-01 3.05796593e-01 -4.41974044e-01
1.19709976e-01 4.01632518e-01 -1.70745695e+00 -5.55189133e-01
1.19122720e+00 9.69421387e-01 3.85018528e-01 7.45383263e-01
1.55188311e-02 3.96846294e-01 -6.89288795e-01 2.03924447e-01
-3.06658987e-02 -3.36214639e-02 -1.09123245e-01 -5.80070376e-01
-3.20596457e-01 6.50398359e-02 -1.69838145e-02 7.48810709e-01
9.18134451e-01 2.50937849e-01 2.09657371e-01 -1.38271654e+00
-1.32605031e-01 3.02443206e-01 -2.78748691e-01 3.99684250e-01
8.56503844e-01 -1.30211845e-01 1.87797084e-01 -3.71651620e-01
9.41054001e-02 8.57416809e-01 9.77628350e-01 1.02353893e-01
-6.42028034e-01 -6.52426600e-01 -5.68688773e-02 5.81364334e-01
-1.56365812e+00 -3.60715061e-01 6.01688921e-01 4.28873077e-02
8.52813482e-01 6.45698726e-01 1.28298497e+00 8.89247954e-01
5.15051544e-01 4.96624380e-01 1.20323670e+00 -4.72664237e-01
8.61125350e-01 -1.86815053e-01 -5.97068109e-02 1.11045346e-01
9.89411175e-01 2.55655169e-01 -2.25227386e-01 6.86790934e-03
3.12339038e-01 3.23681384e-02 1.89755615e-02 -1.75879776e-01
-1.17884171e+00 5.16456366e-01 5.78737617e-01 7.48497188e-01
-7.12047160e-01 3.32409590e-01 -2.43691251e-01 -7.09829405e-02
1.98972166e-01 -1.65419295e-01 -4.09136862e-01 -4.71804649e-01
-9.39587951e-01 2.04879623e-02 5.82624078e-01 8.15958977e-01
5.63539088e-01 2.40912825e-01 7.96519816e-02 2.89604574e-01
1.14278948e+00 1.23180044e+00 4.39854562e-01 -8.27063859e-01
2.80710369e-01 3.32460791e-01 6.87029481e-01 -1.02817059e+00
-8.56605768e-01 -8.81517828e-01 -8.81573915e-01 -5.33732660e-02
3.21286291e-01 -6.85894608e-01 -2.90059686e-01 1.43034697e+00
7.27130055e-01 6.83861494e-01 9.28676724e-02 2.67031878e-01
6.27360642e-01 5.34690440e-01 3.23169500e-01 -3.49361837e-01
1.17929828e+00 -3.85664284e-01 -1.02692950e+00 -1.67382076e-01
4.39316928e-01 -6.62580609e-01 2.59887129e-01 3.00361484e-01
-7.21832573e-01 -6.23712838e-01 -1.21680319e+00 6.68519974e-01
-8.71443868e-01 -1.51702598e-01 4.76788878e-01 1.58320904e+00
-1.17901433e+00 1.65667996e-01 -1.07133877e+00 -8.94303739e-01
1.23277694e-01 6.57332420e-01 2.74146199e-01 1.21129140e-01
-1.39152837e+00 1.02984309e+00 3.92779820e-02 3.63839045e-02
7.83441588e-02 -4.31428462e-01 -5.38789809e-01 -2.20878601e-01
-1.22662961e-01 -9.48167741e-01 9.26862955e-01 -1.05566032e-01
-1.81028354e+00 -2.22045302e-01 -3.40532511e-01 -8.78458679e-01
4.99198526e-01 -2.18223751e-01 -1.16898644e+00 -2.47362718e-01
3.06238115e-01 2.01687142e-01 4.40993339e-01 -1.23752129e+00
-1.33254707e+00 -3.46614540e-01 -3.13300759e-01 -3.25890817e-03
-1.44229293e-01 -5.57211816e-01 2.14347094e-01 -2.86308408e-01
2.31888667e-01 -8.30452323e-01 -3.74269545e-01 -5.42244792e-01
-3.55890580e-02 -3.53683949e-01 8.31680954e-01 -3.29782695e-01
1.59217739e+00 -1.50127900e+00 -9.38385904e-01 3.73675644e-01
-2.31146961e-01 3.97720784e-01 3.64298552e-01 6.11609161e-01
5.16015887e-01 -1.96399271e-01 3.11177075e-01 -2.88093239e-01
1.01699084e-01 3.91351879e-01 1.74332038e-01 6.23079538e-01
-7.05841064e-01 8.89643669e-01 -1.22606277e+00 -2.45709598e-01
1.08574331e+00 9.69651997e-01 -9.10871923e-02 -4.88211930e-01
4.88323182e-01 8.41565371e-01 -8.90157402e-01 6.68201685e-01
9.42450225e-01 3.29979919e-02 -9.73894447e-02 -8.72312635e-02
-6.58465147e-01 1.33959308e-01 -1.74470222e+00 1.32036650e+00
-7.65750349e-01 4.01647389e-01 -9.67466831e-02 -8.07966232e-01
8.86175990e-01 3.63472551e-01 9.43641484e-01 -1.07437587e+00
2.73824751e-01 3.82868677e-01 -2.90898710e-01 -2.88365453e-01
4.98365223e-01 3.42714399e-01 3.82011198e-02 9.41517726e-02
-4.65513319e-01 2.45152086e-01 9.20047164e-02 -1.73096433e-01
1.29460061e+00 1.30401686e-01 8.31761241e-01 -3.85336637e-01
1.01126552e+00 -3.35148036e-01 5.71210265e-01 8.16520870e-01
-5.64871073e-01 -1.67095378e-01 -9.40087199e-01 -3.25922996e-01
-3.41011018e-01 -1.09446120e+00 -2.69864827e-01 4.56974894e-01
4.59957600e-01 -3.35579038e-01 -4.27246600e-01 -1.66857004e-01
1.20778933e-01 1.15204573e+00 -2.73395866e-01 9.82914269e-02
-4.24664259e-01 -1.00055313e+00 3.29346627e-01 1.92957029e-01
1.18416500e+00 -2.15850711e-01 -9.59266365e-01 7.42274404e-01
-3.63996446e-01 -9.88793612e-01 4.62974936e-01 -6.64787441e-02
-8.72232914e-01 -8.28882277e-01 -4.31846946e-01 -1.70477495e-01
2.75001764e-01 9.53519464e-01 7.21107543e-01 6.65377676e-02
4.31984693e-01 1.04686654e+00 -5.46159744e-01 -6.87536061e-01
2.36443076e-02 -1.44239878e-02 2.60560274e-01 2.77927071e-02
6.09370291e-01 -7.31478989e-01 -1.18632245e+00 4.88869190e-01
-3.13652642e-02 -4.11685407e-01 4.39519644e-01 1.63748875e-01
3.84766012e-01 5.47884464e-01 9.93767619e-01 -9.80892256e-02
4.81162190e-01 -8.53065491e-01 -5.79849958e-01 9.45681408e-02
-7.66574442e-01 -4.00154799e-01 3.11075538e-01 1.00849204e-01
-1.11555505e+00 9.88999829e-02 -5.98327279e-01 7.83859313e-01
-5.05140960e-01 -2.44512670e-02 -4.54961479e-01 -3.91048342e-01
3.99608314e-01 3.39779437e-01 -6.07376337e-01 -2.62718588e-01
2.23642305e-01 1.04718292e+00 3.68119448e-01 -2.54282117e-01
9.24878597e-01 9.55595672e-01 2.02357203e-01 -1.20296133e+00
-3.71094435e-01 -8.66201937e-01 -3.20966274e-01 -5.65038621e-01
7.21723318e-01 -1.06214595e+00 -1.11617160e+00 2.46085197e-01
-7.66429007e-01 1.11078665e-01 -2.36907184e-01 8.23747635e-01
-3.07402283e-01 3.85494977e-01 1.82520822e-01 -1.37198055e+00
-2.37853974e-01 -9.36766505e-01 8.03587735e-01 6.96520090e-01
-1.44011468e-01 -1.09563780e+00 3.60135972e-01 4.42058921e-01
1.02680266e+00 3.98384571e-01 -9.05952454e-02 -3.39584857e-01
-5.14021516e-01 -4.49272662e-01 2.84265578e-01 -2.72744685e-01
3.93584758e-01 -4.01266932e-01 -7.99667597e-01 -2.15907395e-01
3.30167487e-02 7.87796676e-01 2.75107026e-01 7.42591262e-01
4.23289925e-01 -1.11801557e-01 -7.75546014e-01 2.30013728e-01
1.60673487e+00 5.18248618e-01 7.90295184e-01 1.03672111e+00
2.79035896e-01 -2.05239698e-01 7.04773188e-01 8.29973519e-01
1.10984337e+00 7.11102545e-01 6.63887441e-01 2.83992857e-01
-1.82184651e-01 -2.02818125e-01 2.39303276e-01 6.75453961e-01
-4.14781600e-01 -5.45593917e-01 -6.66315615e-01 3.90780926e-01
-1.89158094e+00 -1.18740642e+00 -7.06963658e-01 2.47100067e+00
2.37139240e-02 7.12364763e-02 -2.55054887e-02 7.72798419e-01
6.65594935e-01 1.76916812e-02 -1.80714995e-01 -4.55681384e-02
4.24468294e-02 2.20795527e-01 1.05296898e+00 7.56495059e-01
-9.99135852e-01 3.13496500e-01 5.33993196e+00 8.02420259e-01
-8.40724230e-01 4.17348117e-01 -1.80304781e-01 6.84085190e-01
-3.17223907e-01 -1.44272491e-01 -8.98427784e-01 1.01654100e+00
1.38358986e+00 7.94951096e-02 2.54594624e-01 7.44553924e-01
1.01210463e+00 -7.88271308e-01 -2.45102271e-01 1.10868704e+00
-5.06099463e-01 -1.21236122e+00 -4.49707359e-01 2.94104517e-01
8.15914690e-01 3.09965283e-01 -9.24731269e-02 9.37163159e-02
4.76584077e-01 -2.75361806e-01 4.33664799e-01 8.22176874e-01
2.46115904e-02 -8.24094355e-01 8.65346611e-01 4.11473244e-01
-1.80956292e+00 -3.68587196e-01 -2.33498245e-01 -3.07408124e-01
5.79543293e-01 1.02167022e+00 -7.32888579e-01 1.10608435e+00
7.64427304e-01 7.17565298e-01 -5.45799613e-01 1.70853138e+00
-3.08730662e-01 6.65480077e-01 -8.33808124e-01 -3.83814633e-01
1.05642095e-01 -1.25761464e-01 7.03842461e-01 1.00493979e+00
1.03285480e+00 -6.68364540e-02 -6.18330650e-02 -3.51723172e-02
5.12475848e-01 -8.93651024e-02 -5.71774781e-01 8.33560288e-01
1.07169819e+00 1.17769015e+00 -8.20231140e-01 -3.46688837e-01
-4.05162275e-01 3.20871711e-01 -6.98520899e-01 2.39344671e-01
-7.77733684e-01 -6.27974719e-02 3.93105179e-01 3.49747390e-01
3.19950730e-01 -6.55348063e-01 -3.80626827e-01 -7.72726595e-01
-5.74229836e-01 -9.42995995e-02 1.83965534e-01 -6.09813929e-01
-8.72102082e-01 -2.29026116e-02 -2.96152895e-03 -1.57517838e+00
-7.16178343e-02 -3.38790491e-02 -6.91861689e-01 2.95867354e-01
-1.76634622e+00 -1.11033964e+00 -5.60413539e-01 7.46092021e-01
4.08890635e-01 3.97036150e-02 7.73819327e-01 8.07982266e-01
-3.21892858e-01 1.50912583e-01 8.01268518e-01 -5.38794518e-01
1.87372833e-01 -1.23318124e+00 1.29630059e-01 9.50626493e-01
-1.86621547e-02 5.39603353e-01 1.04596341e+00 -7.97240078e-01
-1.49180722e+00 -1.01550329e+00 1.15054071e+00 -3.20900589e-01
4.91097689e-01 2.13128507e-01 1.01803439e-02 4.35414940e-01
2.83497959e-01 -2.10662354e-02 7.83666134e-01 -2.37303093e-01
7.03628838e-01 -6.08880758e-01 -1.67628515e+00 4.86550808e-01
8.61527383e-01 2.06798136e-01 -3.70216370e-01 8.28849003e-02
7.39536881e-02 1.11778200e-01 -9.05447364e-01 4.93412703e-01
7.23140001e-01 -1.13964605e+00 1.39027393e+00 5.66899657e-01
-1.16410995e+00 -8.76640439e-01 -5.66670597e-01 -1.20239890e+00
-2.99534202e-01 -7.47387290e-01 -5.77367723e-01 1.28462946e+00
-1.92299992e-01 -1.14848053e+00 6.77188218e-01 2.09453285e-01
2.06272602e-01 -3.46049130e-01 -1.54832172e+00 -8.07046115e-01
-5.25203049e-01 -8.72450888e-01 1.16383529e+00 4.35452282e-01
-2.94692099e-01 6.65197670e-02 -2.64895290e-01 5.39903224e-01
1.05659747e+00 -5.74892282e-01 8.05786610e-01 -1.50718403e+00
1.58858702e-01 -4.98147160e-02 -5.98208606e-01 -1.07675207e+00
-6.62999928e-01 -2.66889721e-01 -1.46949604e-01 -2.19597149e+00
-6.92614138e-01 -7.30156898e-01 -5.11634827e-01 -1.90190747e-01
2.46200308e-01 1.92113400e-01 -5.06999064e-03 -3.34229246e-02
-9.50895071e-01 7.12134600e-01 7.70651758e-01 -7.06650019e-02
-3.87037784e-01 9.65523362e-01 -4.24143076e-01 9.70791638e-01
1.30357075e+00 -4.36442226e-01 -4.65088278e-01 -2.20661998e-01
5.06612182e-01 -7.32239708e-02 3.84134978e-01 -2.02253342e+00
7.20230699e-01 3.51900123e-02 6.66821778e-01 -1.00642562e+00
3.16585004e-01 -1.45146370e+00 4.04459924e-01 1.07189083e+00
6.33678794e-01 1.99664593e-01 -7.65463561e-02 8.71507883e-01
3.74578178e-01 1.22294046e-01 5.55401027e-01 1.76935777e-01
-7.98681736e-01 -1.45242453e-01 -9.14268255e-01 -6.49002969e-01
1.20229375e+00 -7.81867683e-01 -3.70332271e-01 -6.20384037e-01
-5.85042894e-01 4.87034507e-02 1.47635698e-01 1.89798906e-01
3.12197328e-01 -1.56264699e+00 -6.07479140e-02 1.92688722e-02
-2.82283306e-01 -5.20599067e-01 2.17565298e-01 1.00606894e+00
-2.59142786e-01 8.31187844e-01 -1.52186215e-01 -5.63304603e-01
-9.39121187e-01 -5.74837159e-03 1.92984998e-01 -3.60586733e-01
-2.38283917e-01 1.55130208e-01 -9.70580280e-01 -3.94718379e-01
2.43144080e-01 -3.60348850e-01 -5.21346033e-01 -8.11796412e-02
5.52470863e-01 1.30759382e+00 1.97849467e-01 -9.94159818e-01
-8.11846912e-01 9.48102355e-01 7.82335162e-01 -2.07977146e-01
1.24296355e+00 -1.06062305e+00 3.99539620e-01 2.67238319e-01
5.45561075e-01 4.04022813e-01 -8.89015555e-01 1.36375010e-01
3.42590332e-01 -4.03049856e-01 2.29301140e-01 -9.03512955e-01
-5.39396405e-01 3.63714039e-01 1.39354825e+00 4.74026889e-01
8.64579499e-01 -4.78692770e-01 1.14426661e+00 5.77128351e-01
1.24667406e+00 -1.35528398e+00 -4.41905439e-01 1.68552682e-01
9.58986804e-02 -1.08865154e+00 3.00791919e-01 -6.56912662e-03
1.60112455e-01 8.48011792e-01 7.22095743e-02 -2.83083729e-02
1.25163150e+00 2.45023564e-01 -1.28905490e-01 2.27029026e-01
-4.04482260e-02 -8.25433135e-01 -1.64482996e-01 1.24112940e+00
6.25991523e-02 3.00619751e-01 -7.87863970e-01 4.34083790e-01
-2.86076695e-01 1.84758425e-01 2.16686368e-01 1.25920272e+00
-1.17066801e+00 -1.42736864e+00 -9.14401472e-01 2.28415877e-01
-2.57361829e-01 4.98560220e-01 5.33523381e-01 8.84583890e-01
6.41397893e-01 1.80489898e+00 -3.31928611e-01 -4.91377622e-01
2.63526022e-01 -4.80958790e-01 3.49247962e-01 2.14355975e-01
-2.65032470e-01 -1.36048943e-01 7.12901400e-03 -5.67711830e-01
-1.03693008e+00 -8.89361560e-01 -1.31891477e+00 -5.06160200e-01
-2.98103571e-01 3.45784962e-01 1.16199744e+00 1.20839727e+00
4.00539935e-01 6.24925613e-01 6.88020051e-01 -9.08666790e-01
8.15008804e-02 -6.08006239e-01 -3.90769511e-01 -2.82792807e-01
2.42788002e-01 -9.85883236e-01 -1.02159187e-01 -5.81078887e-01] | [6.342438220977783, 1.0813220739364624] |
f69a8b2d-dc44-45c1-83a5-dac14d714ffb | a-negation-detection-assessment-of-gpts | 2306.16638 | null | https://arxiv.org/abs/2306.16638v1 | https://arxiv.org/pdf/2306.16638v1.pdf | A negation detection assessment of GPTs: analysis with the xNot360 dataset | Negation is a fundamental aspect of natural language, playing a critical role in communication and comprehension. Our study assesses the negation detection performance of Generative Pre-trained Transformer (GPT) models, specifically GPT-2, GPT-3, GPT-3.5, and GPT-4. We focus on the identification of negation in natural language using a zero-shot prediction approach applied to our custom xNot360 dataset. Our approach examines sentence pairs labeled to indicate whether the second sentence negates the first. Our findings expose a considerable performance disparity among the GPT models, with GPT-4 surpassing its counterparts and GPT-3.5 displaying a marked performance reduction. The overall proficiency of the GPT models in negation detection remains relatively modest, indicating that this task pushes the boundaries of their natural language understanding capabilities. We not only highlight the constraints of GPT models in handling negation but also emphasize the importance of logical reliability in high-stakes domains such as healthcare, science, and law. | ['Ken Satoh', 'Kostas Stathis', 'Francesca Toni', 'Randy Goebel', 'Ha Thanh Nguyen'] | 2023-06-29 | null | null | null | null | ['negation-detection', 'natural-language-understanding'] | ['natural-language-processing', 'natural-language-processing'] | [ 4.36768591e-01 8.86193216e-01 -3.55475806e-02 -4.68145847e-01
-7.76029527e-01 -6.43586278e-01 6.39752746e-01 7.09176242e-01
-5.10813236e-01 7.21058071e-01 6.15426362e-01 -8.84976327e-01
-9.20518786e-02 -8.77147615e-01 -6.96684420e-01 1.07959904e-01
2.21968308e-01 5.00423789e-01 -5.66429608e-02 -9.48765516e-01
2.03250289e-01 -1.53627381e-01 -1.12663865e+00 5.27273417e-01
1.04092002e+00 6.63915634e-01 -3.81018311e-01 5.35690248e-01
-1.31243259e-01 1.44904292e+00 -7.71832108e-01 -1.10185301e+00
-2.58949399e-01 -4.27870244e-01 -8.24458361e-01 -6.18288934e-01
5.17844558e-01 -2.98809499e-01 -2.56672204e-02 1.00247085e+00
4.73815233e-01 -2.62522370e-01 5.55023968e-01 -1.23068976e+00
-6.52136862e-01 1.23715639e+00 6.81583807e-02 4.38708901e-01
9.37509954e-01 3.63097638e-01 1.58603013e+00 -3.70452821e-01
7.68229365e-01 1.40019727e+00 7.87501156e-01 7.44631827e-01
-1.07126737e+00 -6.91805243e-01 -2.52825897e-02 -2.90744081e-02
-1.07872081e+00 -5.96888304e-01 3.73915106e-01 -5.49677253e-01
1.68470228e+00 1.07419208e-01 7.64385343e-01 1.37580717e+00
7.87558615e-01 6.22849882e-01 1.15403771e+00 -3.80807221e-01
3.46804529e-01 5.80467060e-02 2.00040132e-01 7.89836705e-01
2.47251809e-01 9.44462568e-02 -9.72777367e-01 -3.17869753e-01
1.34321555e-01 -5.82554162e-01 -3.08518946e-01 3.60616118e-01
-1.08658755e+00 8.61678541e-01 1.00783907e-01 3.34381312e-01
-3.87219638e-01 -7.21003562e-02 6.17725253e-01 7.11520314e-01
2.88379371e-01 6.16018355e-01 -5.11730015e-01 -7.24677980e-01
-6.93525672e-01 3.14937085e-01 1.12385702e+00 1.03397787e+00
1.03554018e-01 -2.46681482e-01 -4.30173874e-01 4.17028517e-01
5.79723269e-02 1.79832712e-01 3.51555347e-01 -8.84839773e-01
7.05105364e-01 6.80650115e-01 -1.48794353e-01 -9.40390050e-01
-4.03383255e-01 -4.49236333e-01 -5.71315467e-01 -1.88557252e-01
2.90976197e-01 -2.28428632e-01 -5.35648823e-01 1.98605335e+00
-2.14061871e-01 -4.23811615e-01 5.30861318e-01 2.44774222e-01
1.06375289e+00 2.78263718e-01 4.55951124e-01 -2.45115627e-02
1.40598845e+00 -2.70016879e-01 -6.73284054e-01 -6.57085598e-01
1.18370020e+00 -5.39114177e-01 1.10425735e+00 3.89655590e-01
-1.13479531e+00 -3.39854062e-02 -1.06066561e+00 -2.03984380e-01
-2.19171643e-01 -3.68639052e-01 6.43512249e-01 7.42771268e-01
-1.13807440e+00 5.26780128e-01 -4.15405333e-01 -3.68595034e-01
4.98565555e-01 1.55655339e-01 -4.52763677e-01 -1.49859339e-01
-1.56677711e+00 1.43399811e+00 2.74139464e-01 -1.75864592e-01
-6.61343277e-01 -6.98051631e-01 -1.30232549e+00 4.38288361e-01
2.65286565e-01 -5.66049278e-01 1.53917801e+00 -6.13843501e-01
-1.06789398e+00 9.95819390e-01 -2.89169431e-01 -8.06931496e-01
7.36021638e-01 -2.07784534e-01 -4.32424664e-01 -2.96915174e-02
4.95763302e-01 6.45723939e-01 2.78019547e-01 -6.21607900e-01
-4.96533185e-01 6.21527657e-02 3.30845326e-01 1.39384642e-01
-4.52829525e-03 1.93306223e-01 4.33271050e-01 -4.31438804e-01
-6.31540492e-02 -2.87051052e-01 4.69606221e-02 -1.22601800e-01
-5.69271743e-01 -5.98986089e-01 3.52660000e-01 -4.92582232e-01
1.09079778e+00 -1.84198475e+00 -2.87168443e-01 4.95894216e-02
3.74602556e-01 -4.04430460e-03 -2.89965537e-03 6.58065259e-01
-1.75160676e-01 4.51859534e-01 -1.92963839e-01 -5.09711318e-02
2.18223393e-01 3.45760643e-01 -3.71321172e-01 4.99537326e-02
5.88087440e-01 1.18298697e+00 -1.11677313e+00 -4.96103972e-01
-1.01609387e-01 1.81934595e-01 -7.97949672e-01 -1.98432267e-01
-5.20302176e-01 7.92038292e-02 -1.16798118e-01 8.78399193e-01
1.92579567e-01 -2.61036307e-01 2.93251991e-01 -8.00289735e-02
8.83145332e-02 8.77006590e-01 -4.79270130e-01 1.14752400e+00
-1.28320903e-01 6.55407488e-01 -1.37870267e-01 -6.43075764e-01
7.47590542e-01 5.79963505e-01 -1.39484957e-01 -1.03945756e+00
2.62992650e-01 2.72312909e-01 5.55837631e-01 -6.81056499e-01
4.46616441e-01 -8.52587342e-01 -2.64330208e-01 5.42933881e-01
3.31524432e-01 -4.84785974e-01 3.50573957e-01 5.78943133e-01
1.38374376e+00 -3.15232903e-01 4.93393958e-01 -3.78591806e-01
3.26761186e-01 9.01569650e-02 5.62357903e-01 9.53505158e-01
-2.68619418e-01 2.85771430e-01 1.28830695e+00 1.11775585e-02
-7.62318313e-01 -7.73857296e-01 -3.59204151e-02 8.04793417e-01
-3.34236145e-01 -7.25074172e-01 -4.89028007e-01 -7.29919434e-01
6.08984903e-02 1.47697544e+00 -6.73634231e-01 -5.76425791e-01
-3.87652159e-01 -4.48295861e-01 1.01895988e+00 5.20693779e-01
4.62240666e-01 -1.04566634e+00 -8.34222674e-01 2.54674643e-01
-4.78464603e-01 -1.18066943e+00 -4.92701456e-02 3.80933106e-01
-5.54928541e-01 -1.08751750e+00 -1.51352510e-01 -7.51063287e-01
4.42648262e-01 -6.13996565e-01 1.55623233e+00 1.17555238e-01
1.94830492e-01 6.15147769e-01 -3.05566221e-01 -6.98020995e-01
-9.74140525e-01 -4.47468236e-02 -2.47066796e-01 -7.80525982e-01
7.44409859e-01 -3.25788349e-01 -2.76138820e-02 1.89334154e-02
-6.71667814e-01 -9.99678597e-02 3.57767910e-01 9.84221339e-01
3.18557117e-03 6.41706437e-02 6.12771630e-01 -1.01411831e+00
1.29098201e+00 -4.80752289e-01 -1.32531971e-01 4.02796328e-01
-6.35138094e-01 -8.13005641e-02 3.11480224e-01 -1.35580823e-01
-9.14565265e-01 -7.05721319e-01 -5.12250602e-01 3.14411789e-01
9.72677916e-02 8.59356046e-01 -1.07439920e-01 5.16599119e-02
7.78254449e-01 -1.24012465e-02 5.17541803e-02 9.67743695e-02
4.55028042e-02 3.24625850e-01 5.73518813e-01 -6.11820996e-01
4.97300953e-01 -5.73691679e-03 -2.92247057e-01 -6.94712520e-01
-1.04223955e+00 2.17837796e-01 -5.00439927e-02 -3.20910215e-02
6.85188532e-01 -8.35885823e-01 -9.60619032e-01 1.15006372e-01
-1.09765816e+00 -3.19185764e-01 -5.70201099e-01 3.49522054e-01
-3.40176523e-01 2.49929845e-01 -8.10479701e-01 -7.64862180e-01
-6.11851215e-01 -9.01581466e-01 9.01872754e-01 -9.27293003e-02
-9.15724039e-01 -9.40851986e-01 -1.19302668e-01 4.93955463e-01
2.54697859e-01 -3.66046727e-02 1.60906637e+00 -1.12133276e+00
-1.90022826e-01 -2.34918371e-01 1.84350852e-02 4.74948585e-01
-1.96063686e-02 -1.70754358e-01 -7.31875062e-01 -1.32611051e-01
2.93400019e-01 -7.21521139e-01 5.30756474e-01 6.94901943e-02
4.10303831e-01 -2.61541963e-01 -1.43853566e-02 7.77667761e-02
1.08609784e+00 2.05958098e-01 4.58338678e-01 3.72278869e-01
9.95646939e-02 6.36697888e-01 5.08004963e-01 1.20980740e-01
7.09681749e-01 -7.83313587e-02 2.80718207e-01 2.26089373e-01
2.26636574e-01 -5.86734653e-01 4.11181509e-01 3.99615198e-01
3.98822755e-01 -6.05087221e-01 -1.32155538e+00 5.58367074e-01
-1.40822494e+00 -9.37566161e-01 9.71260741e-02 1.58244145e+00
9.66733813e-01 8.68855059e-01 -4.77555633e-01 4.05197978e-01
2.58799911e-01 3.10460012e-02 -3.14836204e-01 -7.43812919e-01
-5.36059737e-01 3.91064644e-01 -3.39053459e-02 5.07206142e-01
-5.74470460e-01 8.50310028e-01 7.23192549e+00 3.28429818e-01
-7.67314374e-01 -6.58549443e-02 7.74214327e-01 8.69271830e-02
-6.85367942e-01 1.09836996e-01 -5.11892557e-01 3.24083835e-01
9.87342894e-01 -5.13294876e-01 3.67404222e-02 3.97151679e-01
9.93035361e-02 -4.08966780e-01 -1.55296171e+00 4.54998910e-01
2.08836302e-01 -1.22578895e+00 3.13078761e-01 -1.49671331e-01
3.05644274e-01 1.36237191e-02 -4.06143479e-02 6.68339789e-01
2.76579052e-01 -1.43759131e+00 9.86713290e-01 4.20330286e-01
4.53041196e-01 -5.76893270e-01 1.10740697e+00 5.15355766e-01
-3.82962197e-01 -9.39351767e-02 -6.24352917e-02 -4.81187493e-01
3.04730088e-01 5.90178192e-01 -9.84127700e-01 2.54184186e-01
5.83697438e-01 5.29367983e-01 -5.00838697e-01 4.94809449e-01
-6.76783681e-01 8.28040481e-01 -2.77111173e-01 -2.04358444e-01
1.89352334e-01 1.32036343e-01 4.77748752e-01 1.08977902e+00
1.18860453e-01 3.35750848e-01 -2.86422610e-01 9.36905980e-01
-3.44227582e-01 2.37402190e-02 -6.01757526e-01 -5.63742399e-01
4.29301262e-01 6.64010227e-01 -3.81099403e-01 -6.75433993e-01
-4.89636868e-01 5.08466244e-01 2.34460086e-01 -5.12243211e-02
-5.46151280e-01 -2.87130684e-01 5.28269589e-01 5.21782413e-02
1.23927429e-01 1.77915260e-01 -3.69451195e-01 -1.10211420e+00
2.90823936e-01 -1.22728336e+00 4.14027601e-01 -9.88067508e-01
-1.33399332e+00 4.93287891e-01 -5.34724956e-03 -7.22104251e-01
-5.28489828e-01 -6.38313413e-01 -6.69930935e-01 8.19915354e-01
-1.25397265e+00 -7.93426931e-01 -8.33328813e-02 3.37458402e-01
8.78808945e-02 3.09140719e-02 9.68218863e-01 2.74356753e-02
-2.50057667e-01 7.38741040e-01 -6.18685663e-01 3.70944738e-01
3.66373479e-01 -1.08415008e+00 6.46026015e-01 7.22017229e-01
-2.96435922e-01 1.05184805e+00 9.38334405e-01 -9.51735497e-01
-1.02185309e+00 -5.73042154e-01 1.77503181e+00 -5.63212574e-01
7.00720668e-01 -4.06374782e-01 -8.44634056e-01 1.07674217e+00
3.68643373e-01 -5.77571273e-01 8.78242373e-01 2.37117842e-01
-4.38521683e-01 2.34418407e-01 -1.56443369e+00 6.08133078e-01
1.01590145e+00 -9.62112188e-01 -1.31445789e+00 3.90898049e-01
8.14135969e-01 -4.70870465e-01 -9.40901637e-01 4.08940226e-01
4.44889933e-01 -1.08185363e+00 5.05539000e-01 -6.79474235e-01
9.25253749e-01 2.58970499e-01 -1.34082034e-01 -1.02804494e+00
-3.43130454e-02 -5.57810009e-01 2.16062367e-01 1.25840402e+00
7.45330274e-01 -9.47589338e-01 6.89797997e-01 1.07567787e+00
-2.28710808e-02 -6.85608506e-01 -1.16412437e+00 -4.61271137e-01
4.55839545e-01 -5.73174953e-01 4.48348612e-01 1.00800884e+00
5.51760316e-01 5.44447124e-01 1.30241036e-01 -1.85722098e-01
3.34369779e-01 -3.04130733e-01 2.01657712e-01 -1.20968437e+00
-2.25875124e-01 -3.50781232e-01 -5.02333701e-01 -7.09634542e-01
1.78057626e-01 -9.47613001e-01 -1.62893593e-01 -1.58090138e+00
2.15178877e-01 1.00047864e-01 1.29453555e-01 9.14383709e-01
6.32170867e-03 -2.37852886e-01 -3.68604288e-02 -3.60229611e-01
-1.59786522e-01 5.65358520e-01 9.37392652e-01 -3.93492401e-01
1.91653013e-01 -3.27051014e-01 -1.28077805e+00 6.70724750e-01
7.21144319e-01 -4.41942483e-01 -5.50870895e-01 -5.33999383e-01
7.99531877e-01 4.44878861e-02 5.18592715e-01 -7.43022442e-01
3.12644124e-01 1.46985307e-01 2.44369283e-01 -4.38737959e-01
2.23853171e-01 -4.63391274e-01 -3.12086880e-01 7.08339572e-01
-6.05504036e-01 4.82409149e-01 3.87392253e-01 1.31971255e-01
-2.63676047e-01 -3.43386173e-01 3.70466977e-01 -4.39612210e-01
-3.93065244e-01 -5.85868299e-01 -7.93364465e-01 6.39328063e-01
6.40668094e-01 -2.73445487e-01 -6.58057690e-01 -5.80012143e-01
-5.85370898e-01 4.67484295e-01 2.90929317e-01 3.10378224e-01
8.61615837e-01 -6.39378309e-01 -7.67471910e-01 2.67888188e-01
2.04423770e-01 -2.59762593e-02 1.05519462e-02 8.29753816e-01
-2.76260525e-01 6.01922631e-01 -4.23338637e-02 -3.68690282e-01
-1.15661550e+00 5.08268289e-02 5.90716064e-01 -1.58161521e-01
-5.89391768e-01 1.16210234e+00 -1.45273507e-01 -5.49033701e-01
1.89010873e-01 -7.54296184e-01 -1.58101320e-01 -7.00023398e-02
1.82059586e-01 2.93368809e-02 2.70353168e-01 -4.96622652e-01
-5.15277445e-01 1.17203854e-02 -2.59053886e-01 -1.90638304e-01
1.22954166e+00 2.91130751e-01 -3.45072925e-01 3.95577073e-01
7.88776875e-01 4.43693511e-02 -1.77217469e-01 -5.62672317e-03
2.22632706e-01 5.72667234e-02 -1.42548919e-01 -1.17446077e+00
-3.63270432e-01 5.31157136e-01 8.52098241e-02 -5.93273863e-02
8.21148753e-01 1.91203281e-01 5.77630877e-01 6.07684612e-01
5.02749920e-01 -9.45676088e-01 -2.07864419e-01 8.63295436e-01
7.33582318e-01 -1.21457577e+00 -5.56864291e-02 -3.01620841e-01
-5.74027419e-01 7.65119791e-01 7.13001728e-01 3.23503107e-01
2.54140794e-01 2.86651015e-01 5.72478659e-02 -6.59403801e-01
-1.23811948e+00 1.62262514e-01 -2.05790520e-01 3.00856620e-01
8.39520633e-01 -2.98076123e-02 -3.94069463e-01 6.67292535e-01
-1.01126993e+00 7.40447268e-02 7.10793793e-01 1.38419378e+00
-7.74970502e-02 -1.02781785e+00 1.50514580e-02 7.57767618e-01
-4.75427210e-01 -7.10506439e-01 -6.86632156e-01 7.29796410e-01
4.15383019e-02 1.25534797e+00 7.68542057e-03 -3.68192315e-01
6.26068354e-01 2.80327916e-01 3.77165943e-01 -7.92867720e-01
-1.09876668e+00 -5.07485032e-01 7.85436153e-01 -4.56729442e-01
-1.67148441e-01 -6.58131897e-01 -1.05077517e+00 -3.77619773e-01
-1.58427954e-01 3.72184694e-01 1.49906367e-01 1.16683137e+00
1.62407875e-01 4.75705028e-01 -3.13434809e-01 2.47573122e-01
-8.23623717e-01 -9.63178813e-01 -1.97381452e-01 1.95482314e-01
4.04054999e-01 -3.57275069e-01 -2.83551663e-01 -4.65727359e-01] | [10.108912467956543, 8.129762649536133] |
3d552716-f55f-46bb-9a78-9dca645c397e | what-my-motion-tells-me-about-your-pose-self | 2007.14812 | null | https://arxiv.org/abs/2007.14812v2 | https://arxiv.org/pdf/2007.14812v2.pdf | What My Motion tells me about Your Pose: A Self-Supervised Monocular 3D Vehicle Detector | The estimation of the orientation of an observed vehicle relative to an Autonomous Vehicle (AV) from monocular camera data is an important building block in estimating its 6 DoF pose. Current Deep Learning based solutions for placing a 3D bounding box around this observed vehicle are data hungry and do not generalize well. In this paper, we demonstrate the use of monocular visual odometry for the self-supervised fine-tuning of a model for orientation estimation pre-trained on a reference domain. Specifically, while transitioning from a virtual dataset (vKITTI) to nuScenes, we recover up to 70% of the performance of a fully supervised method. We subsequently demonstrate an optimization-based monocular 3D bounding box detector built on top of the self-supervised vehicle orientation estimator without the requirement of expensive labeled data. This allows 3D vehicle detection algorithms to be self-trained from large amounts of monocular camera data from existing commercial vehicle fleets. | ['Cédric Picron', 'Tinne Tuytelaars', 'Punarjay Chakravarty', 'Tom Roussel'] | 2020-07-29 | null | null | null | null | ['monocular-visual-odometry'] | ['robots'] | [-4.24653053e-01 2.39889428e-01 -2.39501044e-01 -6.69930637e-01
-6.31578863e-01 -9.98712361e-01 6.37158334e-01 -3.09321672e-01
-4.54704672e-01 2.13004962e-01 -3.08575720e-01 -5.10193646e-01
6.59006596e-01 -3.88298780e-01 -1.36264837e+00 -3.65404725e-01
7.86575601e-02 9.68761146e-01 4.01080281e-01 -1.68123338e-02
1.73660338e-01 9.75819826e-01 -1.53233087e+00 -3.20347518e-01
2.45947525e-01 8.25741112e-01 4.03683692e-01 8.37736130e-01
3.34781796e-01 3.78904909e-01 -2.43344471e-01 -4.22722735e-02
7.20368385e-01 2.20044300e-01 -3.10589075e-01 5.09513557e-01
1.13987362e+00 -9.99060333e-01 -6.04971349e-01 9.11116362e-01
-6.06012233e-02 -2.52642959e-01 8.42324555e-01 -1.66720545e+00
7.85625204e-02 -1.54382885e-01 -3.28021497e-01 5.66258132e-02
1.20247029e-01 3.56552511e-01 6.94435120e-01 -1.05102634e+00
9.92027104e-01 1.23657739e+00 8.36674511e-01 3.66651744e-01
-1.05653036e+00 -7.54272163e-01 3.86492871e-02 2.34902814e-01
-1.51916885e+00 -7.13539302e-01 7.12555885e-01 -8.53405654e-01
1.22750771e+00 -3.40680152e-01 6.76909685e-01 9.14987862e-01
2.05945536e-01 6.50319338e-01 6.40907466e-01 6.71907095e-03
1.26781836e-01 5.34918249e-01 -6.38329908e-02 9.62676525e-01
6.68357909e-01 6.78823411e-01 -1.53827444e-01 1.03404976e-01
5.77787817e-01 5.04299961e-02 3.00120533e-01 -1.57373476e+00
-7.89087772e-01 9.12223101e-01 5.32052100e-01 -5.47722161e-01
1.16965855e-02 2.73389518e-01 1.96941361e-01 2.67776966e-01
3.53989601e-01 2.81068563e-01 -6.09674215e-01 1.83221698e-01
-7.55909204e-01 3.52295548e-01 4.77663517e-01 1.38811457e+00
1.34197760e+00 3.09280872e-01 4.76382971e-01 1.61857620e-01
8.07748079e-01 1.09332132e+00 -1.26148760e-01 -1.16806066e+00
4.02858496e-01 7.78808177e-01 2.92310506e-01 -6.04882896e-01
-6.77479446e-01 -3.90854448e-01 4.31345515e-02 8.45895469e-01
3.15930039e-01 -4.38479871e-01 -1.05160141e+00 1.34158909e+00
8.06257427e-01 -7.94342980e-02 2.02041671e-01 1.00596559e+00
7.44599581e-01 4.09476370e-01 -3.93640995e-01 4.71320540e-01
1.02324390e+00 -8.06109428e-01 -5.01904190e-02 -1.04358518e+00
9.60705936e-01 -5.54882228e-01 2.42985919e-01 1.06433630e-02
-4.48456675e-01 -5.11859953e-01 -1.52237463e+00 -1.16251983e-01
-3.29865456e-01 3.36453825e-01 2.65926123e-01 6.47117555e-01
-1.02686214e+00 3.91272269e-02 -9.52630103e-01 -4.28687721e-01
3.14667314e-01 4.28632826e-01 -8.17988575e-01 -1.90070555e-01
-6.54735386e-01 1.42540920e+00 2.58834928e-01 -9.30631533e-02
-1.67700446e+00 -5.30883193e-01 -1.40459907e+00 -4.93611366e-01
3.89177412e-01 -4.02831197e-01 1.30807805e+00 -6.03579462e-01
-1.11672735e+00 1.17445481e+00 -2.13161036e-01 -6.93432748e-01
6.61866546e-01 -3.47984463e-01 -4.11087871e-02 1.34707138e-01
3.00212950e-01 1.06641352e+00 1.14904344e+00 -1.57607830e+00
-7.89725602e-01 -7.60382473e-01 -7.32462779e-02 3.62748474e-01
4.33599621e-01 -3.67146581e-01 -4.41000372e-01 2.35042349e-01
3.55082959e-01 -1.32526028e+00 -1.45647898e-02 2.73055166e-01
-2.63081819e-01 1.92213893e-01 1.39392626e+00 -5.35867929e-01
1.50383577e-01 -2.16529727e+00 -2.25445941e-01 2.48493012e-02
4.04638886e-01 1.31345734e-01 -3.93506512e-03 2.46258900e-01
3.54664437e-02 -5.00671089e-01 2.90206313e-01 -5.14899850e-01
-1.41358897e-02 1.66847348e-01 -1.92274764e-01 1.25196731e+00
2.93405861e-01 9.42883670e-01 -6.93489552e-01 -1.88912436e-01
5.72622120e-01 4.04491514e-01 -5.97383320e-01 3.43779176e-01
-1.15094602e-01 2.96831578e-01 -2.14988396e-01 7.46626377e-01
1.02425373e+00 8.97724330e-02 1.20937131e-01 -1.75397992e-01
-2.32448146e-01 2.74177641e-01 -1.07471192e+00 1.20472646e+00
-3.20943922e-01 1.05912650e+00 3.48958015e-01 -7.49316871e-01
1.22485399e+00 -4.35734093e-02 2.65486836e-01 -5.67077100e-01
3.24431807e-01 1.46181181e-01 -2.68552423e-01 -3.45822304e-01
7.06616640e-01 2.91244984e-02 -1.79294929e-01 8.04028139e-02
2.32669875e-01 -5.41048527e-01 5.00424467e-02 1.05427481e-01
9.19023693e-01 3.49000365e-01 3.09277862e-01 -1.11629903e-01
2.74564028e-01 5.83808124e-01 4.21921164e-01 4.12378639e-01
-3.97220612e-01 4.46354955e-01 2.92395264e-01 -6.56184375e-01
-1.75461423e+00 -1.01625276e+00 -2.82532215e-01 6.08071268e-01
5.55354416e-01 1.39047289e-02 -5.52057385e-01 -6.76020384e-01
6.79815292e-01 4.37005132e-01 -4.21235174e-01 -9.54340175e-02
-5.88446558e-01 -1.42568424e-01 3.55053425e-01 7.04553127e-01
2.36353830e-01 -2.40833163e-01 -8.69719505e-01 1.25486543e-02
4.18061376e-01 -1.67250311e+00 -3.10449243e-01 4.29526567e-01
-7.15176642e-01 -1.21343374e+00 -1.89217344e-01 -8.06719959e-01
8.25489759e-01 7.26334095e-01 8.67025077e-01 -3.69772106e-01
-3.25833559e-02 3.20857257e-01 4.28184262e-03 -5.69611073e-01
-7.42736340e-01 5.63108213e-02 3.89876842e-01 -1.58737972e-01
7.11714506e-01 -2.17934102e-02 -5.42009532e-01 5.73414445e-01
1.97684648e-03 2.61320472e-02 6.01411402e-01 4.55515653e-01
3.41963142e-01 -5.30191898e-01 3.33322994e-02 -5.43740332e-01
-4.04040307e-01 -3.90576839e-01 -1.44968760e+00 -5.22287548e-01
-6.29558682e-01 1.81102917e-01 3.70049685e-01 -1.98895693e-01
-6.47506416e-01 9.18861926e-01 8.18447545e-02 -7.75832176e-01
-3.83777708e-01 -1.26096413e-01 -1.00333869e-01 -4.62328166e-01
7.67837703e-01 -6.84585795e-02 4.52662975e-01 -3.09338331e-01
3.65896463e-01 7.27357149e-01 5.65246642e-01 1.31279930e-01
1.37599909e+00 8.62358332e-01 1.36883020e-01 -7.49429703e-01
-5.50563633e-01 -8.31723392e-01 -9.07434165e-01 -2.25439474e-01
8.64304245e-01 -1.66514051e+00 -7.04153478e-01 2.83594459e-01
-1.13097823e+00 -4.41952616e-01 9.21059772e-02 6.11654222e-01
-6.15836680e-01 2.34053850e-01 -1.40305042e-01 -6.89735353e-01
4.07969467e-02 -1.29521251e+00 1.74267435e+00 -1.10197291e-02
1.26004005e-02 -6.80496514e-01 2.70442367e-01 5.55672705e-01
-3.58990319e-02 8.99639502e-02 2.21892998e-01 -4.34573501e-01
-9.77368176e-01 -6.84022665e-01 -3.25399548e-01 2.38858104e-01
-4.32378918e-01 -1.50454655e-01 -1.11196196e+00 -5.83583415e-01
-2.70402640e-01 -3.64100456e-01 7.36927688e-01 2.03051165e-01
7.71571919e-02 1.33925378e-01 -7.10722625e-01 9.30761337e-01
1.30936587e+00 1.04178764e-01 4.50408868e-02 5.13941288e-01
8.62257898e-01 4.42265034e-01 7.99318492e-01 1.72063112e-01
8.12478065e-01 8.17522824e-01 1.09617555e+00 -9.57623944e-02
4.41153646e-02 -4.89500672e-01 4.91161972e-01 -2.61223805e-03
4.74534154e-01 2.75993913e-01 -1.06621456e+00 7.27510035e-01
-1.53511715e+00 -7.29239047e-01 2.55504400e-02 2.06711793e+00
2.07063481e-01 4.23963636e-01 1.45459935e-01 -2.17720166e-01
3.58076900e-01 7.97167048e-02 -7.86981404e-01 -2.43145153e-01
9.52510238e-02 -7.57156432e-01 1.51780510e+00 7.93817103e-01
-1.58292484e+00 1.18516004e+00 6.31184912e+00 5.72924204e-02
-1.19563878e+00 -3.54031213e-02 7.13446066e-02 7.53809810e-02
-6.29075021e-02 2.55198687e-01 -1.69231069e+00 -6.11210428e-02
8.80738437e-01 8.47202316e-02 3.48658532e-01 1.45162487e+00
1.44048885e-01 -2.00769499e-01 -1.39483130e+00 1.00157511e+00
3.25827092e-01 -1.19366491e+00 -4.48020309e-01 4.31897134e-01
7.37018168e-01 8.79893541e-01 -1.22970052e-01 3.64787370e-01
5.28501928e-01 -7.52957642e-01 1.01397884e+00 -1.26781240e-01
8.37481260e-01 -4.81690794e-01 6.58155143e-01 6.83804095e-01
-1.19241297e+00 -2.39463851e-01 -5.08674920e-01 -1.01660736e-01
1.81633607e-01 -3.65978964e-02 -1.77089274e+00 5.69509603e-02
5.39706349e-01 6.87115252e-01 -6.23428404e-01 9.33492362e-01
-1.23533450e-01 2.01942250e-01 -5.51463127e-01 1.94006398e-01
3.70689660e-01 3.66566591e-02 8.16235065e-01 8.56617868e-01
1.44878611e-01 -6.49673522e-01 3.09631854e-01 7.25539863e-01
-7.03068152e-02 -4.69073296e-01 -1.34973013e+00 2.48089731e-01
3.88416529e-01 1.33489895e+00 -2.93010741e-01 -2.09799990e-01
-5.40781319e-01 4.06027883e-01 3.12637061e-01 1.96043476e-01
-8.48870933e-01 1.23845646e-02 1.09731042e+00 3.74227434e-01
7.99398720e-01 -6.98083639e-01 -1.52980402e-01 -1.07407236e+00
-1.32718608e-02 -5.59333622e-01 -3.13089222e-01 -8.92538786e-01
-5.84266245e-01 3.86332899e-01 1.42770112e-01 -1.51023901e+00
-7.75664091e-01 -1.04685438e+00 -1.12868704e-01 6.51542604e-01
-1.68522739e+00 -1.20100296e+00 -5.74098289e-01 1.75538018e-01
5.02901852e-01 -4.46125478e-01 2.99381882e-01 9.65679884e-02
-8.16109851e-02 3.59979182e-01 2.18905643e-01 2.35725820e-01
6.11120403e-01 -1.09190476e+00 9.18284714e-01 9.59411442e-01
2.40069106e-01 1.34059504e-01 9.97083902e-01 -7.13093817e-01
-1.90846145e+00 -1.43871748e+00 6.27656460e-01 -1.14646351e+00
4.93719459e-01 -7.74847209e-01 -4.97469902e-01 1.17983091e+00
-1.79059222e-01 3.20117801e-01 -1.69171691e-01 -1.71784446e-01
-5.68499148e-01 -4.31301028e-01 -1.02085602e+00 4.25845385e-01
9.31464136e-01 -7.45920181e-01 -5.39548814e-01 2.66983181e-01
4.22772557e-01 -9.77245092e-01 -3.54205698e-01 5.17952859e-01
6.63024545e-01 -7.52373517e-01 1.09909737e+00 -4.31408048e-01
-1.85808167e-01 -6.47239029e-01 -1.93675324e-01 -1.16255033e+00
1.24241859e-02 -2.61427522e-01 -1.41119957e-01 4.81958985e-01
3.20905715e-01 -4.88743871e-01 1.24509287e+00 2.72650987e-01
-3.16118479e-01 -1.16294771e-01 -9.52406585e-01 -7.70618021e-01
-1.21685274e-01 -4.44845825e-01 3.69300693e-01 3.38373572e-01
-4.40375179e-01 5.21879792e-01 -2.06431672e-01 8.80887687e-01
9.49309289e-01 1.28873542e-01 1.62350845e+00 -1.22811377e+00
-7.14493776e-03 9.23426524e-02 -1.08567464e+00 -1.52862704e+00
3.47623080e-01 -8.72713625e-01 3.25650364e-01 -1.16648066e+00
9.36081409e-02 -1.58343062e-01 4.03383881e-01 1.53345466e-01
3.65806341e-01 4.13748950e-01 -2.02028570e-03 1.63658515e-01
-5.06583452e-01 4.14544344e-01 8.44483316e-01 -1.45396829e-01
1.26895398e-01 -4.23872657e-02 -2.09845200e-01 8.90643358e-01
5.48685551e-01 -5.58380485e-01 -3.81156296e-01 -6.29017055e-01
5.94676733e-02 3.82146388e-02 7.89398849e-01 -8.18046749e-01
2.73084402e-01 1.66857436e-01 5.23243308e-01 -1.16761410e+00
5.69936037e-01 -1.02682579e+00 -4.51462567e-02 3.85298491e-01
2.06992269e-01 6.66099638e-02 2.89180011e-01 5.51039517e-01
-5.77122569e-02 -1.04505964e-01 7.93707550e-01 2.73734182e-02
-1.35145605e+00 2.52019197e-01 -4.98606712e-01 -1.02882665e-02
1.26429379e+00 -5.48587263e-01 -3.41470152e-01 -2.07145512e-01
-3.84790808e-01 4.34275001e-01 9.87200558e-01 6.24001324e-01
5.84810555e-01 -1.08881760e+00 -4.68746096e-01 6.94757938e-01
5.01659334e-01 9.40033793e-02 -2.15035398e-02 2.94732004e-01
-1.07372391e+00 9.28225040e-01 -3.06672961e-01 -1.13620913e+00
-1.24831295e+00 7.57514179e-01 5.82262099e-01 3.65735918e-01
-7.72941411e-01 5.21122098e-01 5.06994188e-01 -9.25200760e-01
8.89711007e-02 -2.15279236e-01 5.92611581e-02 -2.40184143e-01
2.05767512e-01 1.31019711e-01 2.86146432e-01 -1.12361479e+00
-5.76800108e-01 7.39722908e-01 2.98339035e-02 -1.17058136e-01
1.09418869e+00 -4.38236833e-01 6.40138805e-01 1.83989987e-01
1.41026878e+00 -1.86364099e-01 -2.20257783e+00 -1.38759315e-01
-1.46122977e-01 -4.07019645e-01 2.60773152e-01 -6.08793683e-02
-7.44481623e-01 8.64952147e-01 7.68039703e-01 -3.81503969e-01
1.50258496e-01 2.82597899e-01 4.37031180e-01 8.67194355e-01
5.32524109e-01 -1.00265598e+00 -1.29230231e-01 7.19465733e-01
5.22261560e-01 -1.79768944e+00 6.39381632e-02 -3.13021719e-01
-6.39638066e-01 1.12534642e+00 7.34621525e-01 -4.64577198e-01
4.80521470e-01 5.00262141e-01 5.10013998e-01 -3.03633928e-01
-7.66854048e-01 -2.98026413e-01 2.71579266e-01 8.72499943e-01
-2.23666981e-01 -4.22316529e-02 6.25102818e-01 -3.70817930e-01
-2.57627040e-01 -4.69971418e-01 5.92216909e-01 7.80273199e-01
-7.87430823e-01 -4.44976836e-01 -4.46151763e-01 -2.96086017e-02
2.67214328e-01 3.07923108e-01 -2.65180022e-01 1.09647834e+00
4.68306057e-02 6.88448191e-01 3.67447466e-01 -4.15580243e-01
3.06363553e-01 -4.41954024e-02 3.98785204e-01 -6.26553237e-01
1.27576873e-01 -1.66023728e-02 3.02219242e-01 -5.89191973e-01
6.09908104e-02 -9.78228807e-01 -9.89669025e-01 -9.03817043e-02
-1.68536007e-01 -3.92053217e-01 1.07685792e+00 1.07893562e+00
3.30972075e-01 -1.14068009e-01 8.45659077e-01 -1.39894187e+00
-7.49911487e-01 -5.68988204e-01 -4.08331722e-01 3.78712937e-02
8.61120164e-01 -9.51573551e-01 -4.16906923e-01 3.34078819e-02] | [7.897592544555664, -2.2931013107299805] |
877b9ad0-35d6-4afe-aa86-671446965844 | collection-and-validation-of | 2011.00958 | null | https://arxiv.org/abs/2011.00958v2 | https://arxiv.org/pdf/2011.00958v2.pdf | Collection and Validation of Psychophysiological Data from Professional and Amateur Players: a Multimodal eSports Dataset | Proper training and analytics in eSports require accurately collected and annotated data. Most eSports research focuses exclusively on in-game data analysis, and there is a lack of prior work involving eSports athletes' psychophysiological data. In this paper, we present a dataset collected from professional and amateur teams in 22 matches in League of Legends video game with more than 40 hours of recordings. Recorded data include the players' physiological activity, e.g. movements, pulse, saccades, obtained from various sensors, self-reported aftermatch survey, and in-game data. An important feature of the dataset is simultaneous data collection from five players, which facilitates the analysis of sensor data on a team level. Upon the collection of dataset we carried out its validation. In particular, we demonstrate that stress and concentration levels for professional players are less correlated, meaning more independent playstyle. Also, we show that the absence of team communication does not affect the professional players as much as amateur ones. To investigate other possible use cases of the dataset, we have trained classical machine learning algorithms for skill prediction and player re-identification using 3-minute sessions of sensor data. Best models achieved 0.856 and 0.521 (0.10 for a chance level) accuracy scores on a validation set for skill prediction and player re-id problems, respectively. The dataset is available at https://github.com/smerdov/eSports Sensors Dataset. | ['Andrey Somov', 'Paul Lukowicz', 'Bo Zhou', 'Anton Smerdov'] | 2020-11-02 | null | null | null | null | ['physiological-computing', 'sensor-modeling', 'skills-evaluation', 'skills-assessment', 'real-time-strategy-games'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'playing-games'] | [-2.31220990e-01 -1.29955485e-01 -2.93593794e-01 -1.46098062e-01
-4.86447692e-01 -6.35594249e-01 -2.57721901e-01 6.32292747e-01
-7.34088480e-01 6.98168278e-01 -7.38413706e-02 3.52040052e-01
-2.77960777e-01 -5.66910446e-01 -5.94584107e-01 -3.37335914e-01
-2.03924015e-01 2.93572903e-01 4.42535847e-01 -5.45782030e-01
3.93125236e-01 3.51028115e-01 -1.88003910e+00 3.49835247e-01
2.40599483e-01 9.78910148e-01 -5.39497919e-02 8.23460162e-01
6.55590713e-01 8.66309822e-01 -7.48556495e-01 -6.07770741e-01
4.84095663e-01 -5.01275420e-01 -3.55454445e-01 -1.42966717e-01
1.67210653e-01 2.52252400e-01 -4.71587121e-01 4.21894014e-01
8.76593590e-01 4.93263572e-01 -1.77923054e-01 -1.44692767e+00
1.52017355e-01 6.45699918e-01 -3.84067655e-01 7.25947201e-01
5.22349119e-01 3.99298549e-01 6.40659273e-01 -2.68183202e-01
2.58982778e-01 4.90765601e-01 1.04397416e+00 3.82519156e-01
-1.32556713e+00 -1.07352340e+00 -4.44560647e-01 5.11798918e-01
-1.42528391e+00 -5.03633499e-01 8.19609761e-01 -5.64243555e-01
5.19252837e-01 3.28058660e-01 1.28601885e+00 1.04983938e+00
2.37850234e-01 3.39622915e-01 1.32597136e+00 7.57084135e-03
2.58476287e-01 2.29414105e-01 3.11155260e-01 1.38092071e-01
2.99927741e-01 3.72779429e-01 -1.38532150e+00 2.20506728e-01
7.06232369e-01 -2.12720364e-01 1.13810413e-01 1.95669100e-01
-7.26806641e-01 4.90151942e-01 -1.66050896e-01 1.39029220e-01
-5.19526243e-01 2.00526148e-01 7.04486787e-01 5.84540009e-01
1.90530121e-01 5.58490336e-01 -2.69359589e-01 -1.09332824e+00
-7.89763570e-01 5.16126633e-01 6.72663748e-01 3.14680010e-01
3.39904726e-01 5.13534285e-02 2.70402618e-02 8.38447034e-01
-3.38464111e-01 3.84694159e-01 6.87542796e-01 -1.15052640e+00
2.18106315e-01 5.57814956e-01 -8.54994953e-02 -1.37470829e+00
-8.71856272e-01 -5.01296043e-01 -8.78140777e-02 1.33327931e-01
5.82774341e-01 -1.11763805e-01 -1.58843338e-01 1.61323786e+00
3.56656425e-02 4.32919890e-01 -1.34627640e-01 1.20508635e+00
8.56500924e-01 -2.66869459e-02 -9.86817759e-03 -3.43589455e-01
1.30725956e+00 -2.93673575e-01 -6.69346094e-01 -4.53551829e-01
2.91147500e-01 -4.77348000e-01 8.45498860e-01 9.91295218e-01
-1.36368370e+00 -8.83965492e-01 -8.29290330e-01 3.83328110e-01
-2.99728066e-02 -9.40176174e-02 7.97319174e-01 8.95545900e-01
-3.85671973e-01 1.08857059e+00 -9.60094273e-01 -3.58962774e-01
1.56360376e-03 5.56940794e-01 -5.34754515e-01 6.35229647e-01
-1.28501928e+00 7.16128469e-01 3.66160542e-01 5.70625067e-02
-8.54069352e-01 -8.47866774e-01 -6.53792858e-01 -3.26425761e-01
6.03007019e-01 1.52704790e-01 1.02209234e+00 -7.18278825e-01
-1.53877068e+00 1.23302317e+00 4.15909320e-01 -5.16645730e-01
2.21537173e-01 -1.88308582e-01 -6.08731866e-01 6.96008727e-02
2.01427862e-01 -6.28469512e-02 2.00954199e-01 -6.62958801e-01
-5.49683630e-01 -6.70287669e-01 -1.26607463e-01 2.67679930e-01
-1.24143541e-01 1.00411296e-01 -3.60868156e-01 -3.30612570e-01
1.64710417e-01 -1.06488216e+00 1.49690509e-01 -5.96583247e-01
-1.82196066e-01 4.03704047e-02 -9.19174030e-02 -6.11013591e-01
1.46014690e+00 -2.38086820e+00 -1.02332227e-01 3.89556140e-01
1.87187970e-01 1.42760605e-01 2.28629217e-01 6.96284533e-01
-3.84213537e-01 -1.89906985e-01 3.86465937e-01 5.14759868e-02
-1.57701984e-01 3.21394831e-01 1.54071108e-01 6.50287807e-01
-5.77689171e-01 7.06617892e-01 -5.87387681e-01 -3.81043226e-01
2.52045155e-01 -1.87815651e-01 -4.53544080e-01 7.51412958e-02
4.99196023e-01 5.67737103e-01 -5.56059852e-02 5.56921422e-01
1.72033101e-01 6.41524613e-01 -2.31174659e-02 -6.98666722e-02
-4.84308273e-01 9.91730243e-02 -1.27006018e+00 1.80803299e+00
-4.10013162e-02 7.30528116e-01 2.80998591e-02 -1.25765300e+00
1.13316703e+00 3.53528559e-01 1.05395484e+00 -1.18572164e+00
5.27296245e-01 1.28702521e-01 4.10831660e-01 -9.41080809e-01
5.69584310e-01 -3.57466698e-01 -4.87846404e-01 3.99757445e-01
6.24196194e-02 1.17289275e-01 7.85170972e-01 -2.45055812e-03
1.10432601e+00 -9.10864919e-02 2.44039781e-02 -9.41301584e-02
1.41427387e-02 2.12055072e-01 8.08041811e-01 6.91190183e-01
-4.01097298e-01 3.73095334e-01 6.63396060e-01 -1.35463908e-01
-6.35565042e-01 -8.88579011e-01 2.96936166e-02 1.26782227e+00
2.55107194e-01 -6.64070070e-01 -5.63419938e-01 3.68723363e-01
2.00659648e-01 3.75212371e-01 -7.70415783e-01 -6.48011386e-01
-4.73473072e-01 -4.47591096e-01 9.00241852e-01 7.45797992e-01
1.62111014e-01 -1.08929682e+00 -8.96866024e-01 3.47181916e-01
-3.40299726e-01 -1.06099355e+00 5.25495075e-02 3.26287866e-01
-7.63223231e-01 -1.30771613e+00 1.41361907e-01 -1.50965795e-01
-2.80485183e-01 3.95289343e-03 1.01825321e+00 3.60091850e-02
-5.82761228e-01 5.99351764e-01 -4.45848912e-01 -8.13734412e-01
-1.56420488e-02 -5.57921939e-02 6.09561861e-01 -1.27948830e-02
6.78043187e-01 -1.04387617e+00 -4.32380497e-01 4.95779753e-01
-2.90210009e-01 -2.97703058e-01 2.48522773e-01 3.36219072e-01
6.42871141e-01 -5.49060591e-02 3.99178058e-01 -4.59039927e-01
7.87304282e-01 -4.89610016e-01 -2.78142899e-01 -3.84526521e-01
-4.76421684e-01 -7.55665600e-01 2.63197452e-01 -5.90693533e-01
-3.87039930e-01 -1.01465382e-01 8.41484219e-02 -4.17445391e-01
-3.48824292e-01 4.15023088e-01 2.62035340e-01 1.21231988e-01
1.18208313e+00 -7.35943317e-02 3.44460517e-01 -3.79904270e-01
-3.95531982e-01 6.44618630e-01 1.08417618e+00 -6.77957773e-01
5.10086119e-01 2.12767303e-01 4.67022620e-02 -7.73543417e-01
-7.53566921e-01 -6.87004149e-01 -6.76878393e-01 -9.99167979e-01
7.31693625e-01 -9.62956667e-01 -1.62255013e+00 3.54325205e-01
-2.61659056e-01 -3.29966277e-01 -5.95763862e-01 1.16270089e+00
-6.96169496e-01 -1.33218095e-01 -4.64108974e-01 -1.05212307e+00
-3.33418399e-02 -4.60710377e-01 5.25017738e-01 5.67709267e-01
-6.09051704e-01 -5.67091405e-01 5.28889835e-01 1.01214314e+00
-4.70631793e-02 3.69856507e-01 -3.91630456e-02 -8.99706960e-01
2.57465065e-01 -6.63996100e-01 6.83650374e-01 2.79494345e-01
-8.38632509e-02 -3.41982752e-01 -1.01855028e+00 -1.12730451e-01
2.83236802e-02 -5.73880792e-01 2.77739614e-01 4.00360197e-01
1.03310931e+00 5.27463257e-01 8.26891214e-02 3.33156258e-01
9.67668772e-01 2.40584955e-01 6.96187794e-01 3.95706892e-01
3.93218070e-01 7.68188000e-01 9.28129017e-01 7.30190456e-01
7.03947768e-02 7.54305184e-01 2.96511322e-01 1.19179480e-01
1.21052459e-01 -2.22544834e-01 3.17759663e-01 4.93134916e-01
-8.63484323e-01 7.85074905e-02 -7.24952698e-01 4.81112331e-01
-1.79849374e+00 -1.34816623e+00 -6.55253470e-01 2.42896557e+00
6.43962979e-01 3.73952270e-01 8.26265097e-01 7.09916174e-01
6.49295211e-01 -3.56274366e-01 -5.29681444e-01 -7.10657418e-01
1.04892209e-01 5.27509332e-01 7.23222196e-01 -2.67119706e-02
-6.52792633e-01 5.13755143e-01 6.18503857e+00 6.51021421e-01
-9.01339769e-01 4.66213562e-02 1.51682058e-02 -1.03022444e+00
3.70822608e-01 -2.21097916e-01 -4.68830436e-01 6.68460429e-01
1.30211294e+00 -3.28010380e-01 3.58107507e-01 5.67558587e-01
6.50088072e-01 -6.45090520e-01 -9.16872621e-01 1.28536594e+00
1.19704291e-01 -8.05164814e-01 -1.14651811e+00 1.86420873e-01
9.99191403e-02 -2.06901371e-01 -1.96810976e-01 5.98781884e-01
-2.09629580e-01 -9.36624944e-01 7.60147274e-01 7.66109765e-01
4.38434571e-01 -8.46082628e-01 7.09056318e-01 3.90974909e-01
-9.13702846e-01 -3.44499707e-01 -5.14853746e-03 -7.98607469e-01
4.45984863e-03 3.71488273e-01 -2.12503165e-01 3.56501609e-01
9.58858907e-01 6.44234538e-01 -6.18862331e-01 9.91766512e-01
9.27742198e-03 9.97431874e-01 -5.53256094e-01 6.69703186e-02
-2.64657766e-01 -7.40456954e-02 4.98237342e-01 6.22306406e-01
4.21404913e-02 4.51256335e-01 2.59965032e-01 4.98743176e-01
4.04453427e-01 2.38789797e-01 -3.55326116e-01 -4.43364158e-02
4.12323415e-01 1.30568647e+00 -6.98202431e-01 6.11774847e-02
-1.99898273e-01 4.56563503e-01 2.96633877e-02 -1.61013052e-01
-9.92930591e-01 -1.78428054e-01 8.19830358e-01 6.18277550e-01
-3.75714391e-01 -5.38982786e-02 -4.91907418e-01 -7.39883363e-01
4.61222455e-02 -8.54210794e-01 5.51944435e-01 -7.66907752e-01
-1.17176402e+00 2.08681040e-02 9.73658189e-02 -1.33594680e+00
-2.17210829e-01 -3.60964477e-01 -6.74772382e-01 5.37735820e-01
-4.66817081e-01 -4.69906449e-01 -5.58265269e-01 7.58620083e-01
2.62026489e-01 -1.53046206e-01 5.71504831e-01 6.35008931e-01
-8.72822702e-01 5.11717618e-01 -4.96908754e-01 1.18692882e-01
9.29151475e-01 -9.29046512e-01 -2.37652764e-01 5.41985333e-01
3.49944010e-02 2.47238055e-01 9.41890121e-01 -7.46219397e-01
-1.40239346e+00 -4.34159547e-01 3.55344474e-01 -5.46046734e-01
7.56159663e-01 -2.70281434e-01 -5.14198303e-01 6.27561450e-01
-1.77051723e-01 -1.67148978e-01 1.37211800e+00 4.46186334e-01
4.65021968e-01 -4.27982748e-01 -7.20972657e-01 2.75754154e-01
1.02059460e+00 -4.43525702e-01 -5.36585152e-01 1.41631156e-01
-2.06256226e-01 -7.18088329e-01 -1.29143262e+00 2.06868365e-01
8.93468678e-01 -1.30207908e+00 7.58186400e-01 -6.95312977e-01
3.28880787e-01 6.27433509e-02 1.21721245e-01 -1.12104440e+00
-8.62847641e-02 -4.66936827e-01 3.67278159e-01 1.09940875e+00
1.33646369e-01 -2.48962954e-01 8.77245307e-01 9.65855658e-01
-3.54444087e-01 -5.25024593e-01 -7.09774911e-01 -7.43986249e-01
-2.44225129e-01 -1.15796530e+00 -1.07331805e-01 8.27423692e-01
6.76154435e-01 1.89051583e-01 -6.61990225e-01 -1.53349862e-01
4.73289669e-01 7.88178742e-02 1.11525726e+00 -1.53267860e+00
-6.43699944e-01 -9.35209319e-02 -1.04246879e+00 -2.41691470e-01
-1.25233039e-01 -7.56109238e-01 -1.56680450e-01 -1.07930636e+00
3.52159142e-02 -1.41959384e-01 -3.64289194e-01 5.76887906e-01
2.17987016e-01 8.59582365e-01 1.29240260e-01 -4.27703885e-03
-4.48955715e-01 -5.40644340e-02 9.96206641e-01 5.73607802e-01
-4.94096607e-01 3.59300375e-01 -7.14581251e-01 5.23077846e-01
1.18861377e+00 -6.51636839e-01 -4.42902505e-01 7.12111266e-03
4.23987567e-01 3.87266755e-01 6.23988986e-01 -1.54752958e+00
3.09454948e-01 -2.94144005e-01 2.96403885e-01 -3.50243181e-01
7.77581632e-01 -5.03390014e-01 5.90287805e-01 4.18620646e-01
-2.91496783e-01 -1.88234180e-01 4.77514744e-01 2.26436704e-01
-2.34679416e-01 -2.23205015e-01 3.94203275e-01 -2.35292315e-01
-6.82544768e-01 1.06001310e-01 -6.49813592e-01 1.75527215e-01
1.26706612e+00 -7.26014316e-01 -3.63336392e-02 -3.22204262e-01
-1.30515432e+00 2.84789026e-01 1.73690915e-01 4.29635942e-01
1.93835869e-01 -1.07937145e+00 -7.27020264e-01 1.92360327e-01
1.35654449e-01 -7.13209093e-01 7.03846216e-01 1.48178911e+00
-2.80561656e-01 5.06911501e-02 -9.51000452e-01 -6.05214119e-01
-1.44160640e+00 1.93295538e-01 3.71054262e-01 2.81353176e-01
-5.25264740e-01 7.18764365e-01 -5.93723834e-01 6.29597530e-02
-6.18912354e-02 1.07600562e-01 -3.23465317e-01 4.55559671e-01
4.45707023e-01 7.96544015e-01 3.44166130e-01 -5.78995585e-01
-4.20530856e-01 4.46909398e-01 3.62082869e-01 -2.34671190e-01
1.18946099e+00 -6.47521541e-02 1.79424077e-01 1.07666254e+00
4.91899341e-01 2.35833958e-01 -8.63551378e-01 1.86197013e-01
-4.38085318e-01 -6.03115559e-01 -1.16195075e-01 -4.90526170e-01
-1.16036999e+00 7.59587467e-01 9.04636502e-01 5.40360332e-01
1.06827915e+00 -4.28187437e-02 5.48866212e-01 1.43761382e-01
3.46207559e-01 -1.68972099e+00 1.36446059e-01 1.59540713e-01
5.35782039e-01 -9.39679265e-01 7.23937526e-02 -2.73315549e-01
-1.01769435e+00 7.61631250e-01 8.80832970e-01 -3.07606548e-01
3.66961122e-01 5.06355941e-01 2.10928813e-01 -5.35683393e-01
-8.08363020e-01 -5.05909562e-01 -5.91874458e-02 6.33157551e-01
4.32111740e-01 6.83558434e-02 -7.58041322e-01 1.36420798e+00
-9.42654550e-01 2.45262623e-01 8.18259656e-01 8.85116398e-01
-3.92494142e-01 -7.74590135e-01 -5.30379951e-01 5.50420403e-01
-6.35258257e-01 4.02753413e-01 -4.55988765e-01 1.02895308e+00
4.40158159e-01 1.25754964e+00 3.56554389e-01 -9.16487813e-01
8.91104698e-01 1.45601839e-01 5.04564464e-01 -6.84448779e-01
-1.08380270e+00 -5.98935746e-02 3.53552163e-01 -9.62057292e-01
-5.41056991e-01 -8.50080729e-01 -1.31703317e+00 -8.08701634e-01
-1.00272074e-01 4.46248472e-01 6.78190410e-01 6.93401039e-01
3.33927087e-02 7.04725564e-01 4.98344988e-01 -7.47038960e-01
-1.15140434e-03 -1.14291537e+00 -1.34487116e+00 6.82084799e-01
-4.24917936e-01 -9.25358713e-01 -2.73825973e-01 2.06792504e-01] | [6.85168981552124, 0.37554827332496643] |
1ef2ef44-3f27-4ac1-bf56-a403affdcd71 | implicit-dual-domain-convolutional-network | 1810.08042 | null | https://arxiv.org/abs/1810.08042v3 | https://arxiv.org/pdf/1810.08042v3.pdf | Implicit Dual-domain Convolutional Network for Robust Color Image Compression Artifact Reduction | Several dual-domain convolutional neural network-based methods show outstanding performance in reducing image compression artifacts. However, they suffer from handling color images because the compression processes for gray-scale and color images are completely different. Moreover, these methods train a specific model for each compression quality and require multiple models to achieve different compression qualities. To address these problems, we proposed an implicit dual-domain convolutional network (IDCN) with the pixel position labeling map and the quantization tables as inputs. Specifically, we proposed an extractor-corrector framework-based dual-domain correction unit (DCU) as the basic component to formulate the IDCN. A dense block was introduced to improve the performance of extractor in DRU. The implicit dual-domain translation allows the IDCN to handle color images with the discrete cosine transform (DCT)-domain priors. A flexible version of IDCN (IDCN-f) was developed to handle a wide range of compression qualities. Experiments for both objective and subjective evaluations on benchmark datasets show that IDCN is superior to the state-of-the-art methods and IDCN-f exhibits excellent abilities to handle a wide range of compression qualities with little performance sacrifice and demonstrates great potential for practical applications. | ['Xuesong Liu', 'Yaowu Chen', 'Xiang Tian', 'Fan Zhou', 'Bolun Zheng'] | 2018-10-18 | null | null | null | null | ['color-image-compression-artifact-reduction', 'jpeg-artifact-correction', 'image-compression-artifact-reduction'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 4.20552135e-01 -7.63793886e-01 -2.45475009e-01 -2.90682822e-01
-6.86411679e-01 -1.49371609e-01 3.43956441e-01 -2.91108966e-01
-4.75695878e-01 2.50707358e-01 1.17211670e-01 -1.67562187e-01
1.23436959e-03 -1.00564635e+00 -7.03569174e-01 -6.24308407e-01
1.35219365e-01 -6.55252486e-02 1.99487060e-01 -2.73381352e-01
2.93320477e-01 3.91852528e-01 -1.28975761e+00 5.63800514e-01
1.12963283e+00 1.35646176e+00 6.22313321e-01 6.43182099e-01
-2.13479474e-01 7.88183868e-01 -4.86031264e-01 -4.81821388e-01
5.90375185e-01 -5.49186766e-01 -4.24303293e-01 1.25618860e-01
3.01507354e-01 -7.80263484e-01 -7.86591589e-01 1.40567517e+00
6.42490029e-01 -7.50111765e-04 5.11752307e-01 -9.68420327e-01
-1.26523530e+00 3.80980015e-01 -9.72146630e-01 4.68606614e-02
-8.27062055e-02 1.21249169e-01 7.88665056e-01 -8.73337209e-01
3.00356328e-01 1.28321099e+00 5.91393471e-01 4.32985485e-01
-1.07978868e+00 -8.14607084e-01 -2.18559489e-01 6.43337488e-01
-1.46635365e+00 -1.34340242e-01 7.43113697e-01 1.87686868e-02
8.30067396e-01 1.13630623e-01 4.45169121e-01 6.83030605e-01
2.32193977e-01 7.72502899e-01 9.96611714e-01 -2.87910551e-01
1.80943742e-01 -2.51486957e-01 -2.63963014e-01 4.48569536e-01
2.34304026e-01 7.33613446e-02 -4.81804669e-01 3.12142044e-01
1.09143734e+00 1.93246499e-01 -4.78959948e-01 -2.53100634e-01
-1.10007143e+00 7.65554488e-01 6.67926490e-01 2.04027578e-01
-1.53794482e-01 2.72383809e-01 5.97796500e-01 2.26080313e-01
4.79845144e-02 -1.18536351e-03 -3.15033704e-01 -8.17829370e-02
-1.00797141e+00 1.65387407e-01 5.01969695e-01 1.18241298e+00
5.01031816e-01 2.73965418e-01 -3.27056259e-01 1.17192590e+00
7.92590827e-02 4.73959684e-01 7.21923232e-01 -1.16023445e+00
8.52397501e-01 4.07507092e-01 -1.49424553e-01 -1.26981986e+00
-3.24212424e-02 -5.90626657e-01 -1.57759631e+00 2.71017790e-01
-9.49978009e-02 2.95119643e-01 -9.85586047e-01 1.48971951e+00
-2.41378441e-01 1.73679546e-01 1.93646044e-01 1.16101050e+00
6.19552851e-01 9.54074562e-01 -1.10410668e-01 -1.29244879e-01
1.17609942e+00 -1.05162263e+00 -9.66014326e-01 -1.87365159e-01
2.66742915e-01 -8.74106705e-01 1.10811484e+00 6.19771719e-01
-1.39583600e+00 -1.04039180e+00 -1.62165904e+00 -4.58031982e-01
-1.48789644e-01 4.13259536e-01 2.99667865e-01 4.78677839e-01
-9.17753637e-01 8.77336740e-01 -7.62770474e-01 2.96082795e-01
4.51145947e-01 2.74222672e-01 -1.58005401e-01 -5.18100917e-01
-1.33323061e+00 6.75447106e-01 5.22087276e-01 1.05708294e-01
-6.52925193e-01 -4.15605247e-01 -8.13204587e-01 4.82322723e-01
1.17478468e-01 -3.92140031e-01 1.13149559e+00 -9.37618732e-01
-1.36686254e+00 4.53665614e-01 1.12296194e-01 -4.90166724e-01
4.51964408e-01 -1.69574618e-01 -7.55429447e-01 4.59433287e-01
-4.26284894e-02 5.90551674e-01 1.04466999e+00 -1.15935433e+00
-7.34439552e-01 -2.00781584e-01 -2.49847844e-01 1.29860878e-01
-5.55062294e-01 -2.18580097e-01 -1.10979223e+00 -1.18773532e+00
3.05811703e-01 -5.68473458e-01 7.17473403e-02 4.49650407e-01
-1.23774167e-02 2.02862218e-01 1.11747444e+00 -9.64307189e-01
1.56539309e+00 -2.36958814e+00 1.23025969e-01 -7.94582516e-02
2.50242889e-01 6.91491783e-01 -4.99236882e-01 1.11476295e-01
-6.34068996e-02 6.00904897e-02 -6.01931870e-01 -2.84696639e-01
3.81442606e-02 4.39740688e-01 -1.87636867e-01 4.46205467e-01
3.84647310e-01 5.63160479e-01 -6.65297627e-01 -4.63094562e-01
2.70106584e-01 8.55684102e-01 -8.71677756e-01 3.17404211e-01
-1.18056284e-02 7.02649653e-02 -1.68153852e-01 5.86714685e-01
1.44256699e+00 -3.32472771e-01 2.20097393e-01 -6.32939816e-01
7.92785883e-02 3.98602709e-02 -1.36039066e+00 1.85169554e+00
-4.26851839e-01 5.64038992e-01 3.02530587e-01 -1.02308309e+00
9.07971859e-01 9.64912772e-02 4.22618270e-01 -1.19451821e+00
2.52800018e-01 3.52639616e-01 5.39709963e-02 -3.72152120e-01
6.07449710e-01 2.89344490e-02 2.08510131e-01 2.87871718e-01
2.19041079e-01 -1.08280040e-01 3.09711665e-01 1.73250005e-01
8.37268829e-01 1.26488000e-01 3.27577889e-01 5.51097058e-02
7.21479535e-01 -4.44075346e-01 1.02283072e+00 4.49935347e-01
-2.79997468e-01 1.06897461e+00 2.72336364e-01 -4.47050303e-01
-1.30902028e+00 -9.80011344e-01 -2.33066782e-01 8.21676612e-01
5.41967690e-01 -5.17356992e-01 -7.75056779e-01 -2.21665561e-01
-2.39971548e-01 4.80039567e-01 -2.32786641e-01 -2.87412345e-01
-7.80387163e-01 -7.15052426e-01 4.25096482e-01 7.05605984e-01
1.36140871e+00 -6.52017236e-01 -4.83261377e-01 2.47519344e-01
-5.02768517e-01 -1.39589584e+00 -7.93498337e-01 2.10694224e-01
-8.82489622e-01 -9.04344022e-01 -8.58841121e-01 -9.45814013e-01
3.89038563e-01 5.04269063e-01 9.29087937e-01 3.52186970e-02
-8.98255482e-02 -1.86069146e-01 -5.68014622e-01 -5.14323451e-03
-3.99953097e-01 -2.37123653e-01 -1.49067119e-01 -3.64577994e-02
4.48452383e-01 -6.85273588e-01 -8.47502828e-01 2.90512711e-01
-1.39528322e+00 2.51333386e-01 8.25345218e-01 1.06038511e+00
7.67439008e-01 3.03387851e-01 1.68801382e-01 -6.27716005e-01
6.12061083e-01 -1.49324432e-01 -6.57622516e-01 2.17392847e-01
-9.03770506e-01 1.91002026e-01 1.09213662e+00 -3.10133696e-01
-9.50930536e-01 -1.22365259e-01 -2.53365070e-01 -6.46970689e-01
1.18292421e-01 3.55742306e-01 -3.55923027e-01 -3.43161114e-02
4.58647072e-01 5.45726717e-01 1.99748594e-02 -6.71980798e-01
3.17301631e-01 7.78613329e-01 1.02795494e+00 -4.66309279e-01
9.00315106e-01 2.66741455e-01 2.96661071e-03 -3.05175006e-01
-4.30136621e-01 -1.31307945e-01 -5.39662838e-01 1.63896590e-01
8.92425299e-01 -1.22029233e+00 -4.48561311e-01 6.79942250e-01
-1.26877141e+00 -1.32945657e-01 6.71664700e-02 4.21458870e-01
-5.44267535e-01 6.63029552e-01 -1.04839540e+00 -1.56668112e-01
-5.46030283e-01 -1.40445578e+00 8.11400235e-01 3.06581110e-01
5.59608400e-01 -6.02420807e-01 -3.97456884e-01 3.03333923e-02
7.10001111e-01 2.83244792e-02 1.08844137e+00 -2.89173350e-02
-6.98372066e-01 -6.11995868e-02 -7.21695364e-01 8.43161702e-01
2.34718665e-01 -1.87217861e-01 -8.09661508e-01 -4.85423803e-01
2.13055775e-01 -2.93412089e-01 9.82672155e-01 2.84044176e-01
1.80765569e+00 -4.81011719e-01 1.32296383e-01 1.30446589e+00
1.88708222e+00 3.64481181e-01 1.03427947e+00 4.81648862e-01
5.65859258e-01 -1.41233742e-01 4.48027223e-01 6.16683781e-01
1.14886500e-01 7.54206300e-01 4.82427746e-01 -2.38579825e-01
-5.12344897e-01 -2.47786805e-01 2.27520972e-01 1.05994976e+00
-5.19228168e-02 -2.41623104e-01 -4.58365232e-01 2.26880908e-01
-1.50379002e+00 -8.43294799e-01 2.20546395e-01 1.89392591e+00
1.07846367e+00 1.91735357e-01 -2.79558182e-01 4.14858788e-01
7.43397892e-01 2.46894807e-01 -5.80915570e-01 -4.15603012e-01
-2.86852866e-01 2.95461744e-01 7.59503782e-01 1.64512783e-01
-9.95343983e-01 4.12745148e-01 5.91152096e+00 1.19235587e+00
-1.16262186e+00 4.43161316e-02 6.72461450e-01 1.01718061e-01
-5.88161424e-02 -2.03440189e-01 -5.76180816e-01 5.91424406e-01
7.46577501e-01 7.83153027e-02 7.45706201e-01 7.87408113e-01
9.77167338e-02 1.13324158e-01 -7.49212563e-01 1.37774217e+00
8.01981091e-02 -1.40013671e+00 3.33740592e-01 -1.20078094e-01
7.94415593e-01 -6.00584112e-02 3.40183973e-01 1.92902625e-01
4.21963818e-02 -8.22438061e-01 7.60835946e-01 1.95025697e-01
1.49534297e+00 -7.88849533e-01 6.74737275e-01 7.33513460e-02
-1.38628852e+00 -2.90523708e-01 -8.39461446e-01 2.40893319e-01
8.31236616e-02 4.83744502e-01 -6.24097362e-02 5.86770713e-01
8.52249742e-01 9.91559267e-01 -5.52452505e-01 9.22455430e-01
6.12737499e-02 4.17905092e-01 4.08498757e-02 5.73180795e-01
3.41841847e-01 -4.01391208e-01 1.03381768e-01 1.33803177e+00
6.28539383e-01 2.63909400e-01 -2.05125377e-01 8.46166790e-01
-3.22135031e-01 -2.00774953e-01 -8.09473097e-02 7.40234926e-02
3.64116490e-01 1.12623978e+00 -4.30760205e-01 -3.32475036e-01
-6.97337866e-01 1.30306232e+00 3.13827358e-02 4.12682086e-01
-9.82298791e-01 -6.43812060e-01 6.54532552e-01 -2.13700605e-05
6.91108763e-01 -2.86050856e-01 -4.84573722e-01 -1.20742238e+00
-4.08616699e-02 -1.18022764e+00 2.01563358e-01 -8.19356263e-01
-1.07407367e+00 6.39173925e-01 -2.63882369e-01 -1.94929624e+00
8.30807090e-02 -8.32031190e-01 -4.08626318e-01 9.47103560e-01
-2.07771540e+00 -8.53798032e-01 -6.94061160e-01 7.37572372e-01
6.31555736e-01 -3.58788133e-01 6.12915039e-01 7.71379828e-01
-5.07902205e-01 7.96647668e-01 5.72675586e-01 2.87754297e-01
8.84018779e-01 -1.04790473e+00 3.59632820e-01 1.19296443e+00
-2.54194736e-01 4.97846246e-01 4.12177920e-01 -4.13307041e-01
-1.56864953e+00 -1.35401154e+00 3.18936706e-01 5.75852692e-01
1.88496709e-01 -2.76805237e-02 -1.08160007e+00 3.01997900e-01
3.88818949e-01 3.19947869e-01 4.59355861e-01 -7.30199873e-01
-5.36849856e-01 -5.41574478e-01 -1.14511919e+00 3.95125479e-01
8.84368598e-01 -5.42611897e-01 -1.22516386e-01 3.78397666e-02
8.78782272e-01 -6.17496550e-01 -8.15971255e-01 3.92120451e-01
3.21150005e-01 -1.18748558e+00 1.01523650e+00 7.61471838e-02
1.20109081e+00 -6.16938412e-01 -5.93532622e-01 -1.07333803e+00
-7.02136278e-01 -3.04435343e-01 -4.35218006e-01 1.03525484e+00
-4.46008332e-02 -2.59253591e-01 4.11090195e-01 3.78583819e-01
-2.19621763e-01 -7.01756716e-01 -7.62608647e-01 -6.86198950e-01
-7.79847130e-02 -3.57850820e-01 8.13718140e-01 5.92738807e-01
-3.75421196e-01 3.30740362e-02 -6.85030043e-01 8.39559138e-02
7.06772387e-01 1.17049068e-01 4.98084813e-01 -6.32653654e-01
-5.13120532e-01 -4.08818543e-01 -3.91697288e-01 -1.52110934e+00
-3.73232871e-01 -7.39994466e-01 -9.42082480e-02 -1.50781739e+00
2.21120492e-01 -4.07520205e-01 -5.39431632e-01 2.16889352e-01
-2.23955184e-01 3.35886031e-01 4.50301826e-01 4.33494627e-01
-3.27180237e-01 8.10206413e-01 1.62339294e+00 -4.53840613e-01
1.19881101e-01 -2.81973869e-01 -7.60935843e-01 4.03975338e-01
6.84419215e-01 -2.41592839e-01 -4.08211887e-01 -1.03955364e+00
-1.37276769e-01 1.65904060e-01 1.85205236e-01 -1.42588985e+00
3.67469639e-01 -7.31686503e-02 7.88752198e-01 -7.89036632e-01
2.52742648e-01 -9.84338820e-01 6.23814613e-02 6.29259527e-01
-3.52110893e-01 1.14597797e-01 1.37582868e-01 5.74829578e-01
-6.51092052e-01 -1.95032835e-01 1.25545132e+00 -1.46530122e-01
-9.75474954e-01 4.59965706e-01 8.40966851e-02 -2.44338691e-01
6.98121369e-01 -3.07964206e-01 -2.92508543e-01 -3.93008739e-01
-1.73939928e-01 -7.56024420e-02 4.69337076e-01 2.32243851e-01
1.05986881e+00 -1.64505291e+00 -7.46650517e-01 5.25542378e-01
-1.51810870e-01 1.09202757e-01 4.58982289e-01 3.74223262e-01
-1.02300036e+00 4.03779566e-01 -5.63877702e-01 -4.26502496e-01
-8.80918205e-01 7.64296710e-01 3.50036711e-01 -2.87829578e-01
-6.99069858e-01 6.17289841e-01 1.53470233e-01 -7.43864244e-03
3.84682298e-01 -3.30368906e-01 -8.31812322e-02 -3.84585589e-01
8.48314345e-01 4.33561474e-01 6.63742572e-02 -6.27883434e-01
5.86217381e-02 6.89493179e-01 -1.43054411e-01 1.71105877e-01
1.45174563e+00 -1.82648197e-01 -1.03114136e-01 -2.85728425e-01
1.60139143e+00 -4.20855105e-01 -1.58239937e+00 -4.91383523e-01
-4.78759110e-01 -8.77524793e-01 2.64013886e-01 -7.31354356e-01
-1.66150248e+00 1.02597404e+00 9.97604430e-01 -3.03237408e-01
1.98533082e+00 -8.21409941e-01 1.26470077e+00 3.10303383e-02
1.90574214e-01 -1.32132471e+00 2.15220347e-01 3.63609821e-01
8.95778477e-01 -1.20998001e+00 2.83540815e-01 -3.17339420e-01
-5.57859361e-01 1.48531246e+00 7.51418233e-01 -1.99276000e-01
5.15621305e-01 3.71716529e-01 -1.56439230e-01 3.07858765e-01
-4.92992908e-01 2.84431487e-01 3.23098928e-01 6.95214152e-01
3.17415476e-01 -1.33960217e-01 -5.61374545e-01 6.25220954e-01
2.01620311e-01 3.62147450e-01 5.51210344e-01 8.93131018e-01
-3.58493894e-01 -1.15558171e+00 -4.28370714e-01 3.63043785e-01
-5.00144184e-01 -2.24561915e-01 3.22070152e-01 4.85450029e-01
4.18571025e-01 7.92773128e-01 7.22950697e-02 -7.48015404e-01
2.07562953e-01 -4.80212599e-01 2.33598426e-01 -7.55698532e-02
-3.36727351e-01 3.06143999e-01 -6.37717485e-01 -7.57751346e-01
-4.98177230e-01 -1.21621341e-01 -1.11785972e+00 -5.80256581e-01
-8.72816667e-02 -2.88312525e-01 4.81326312e-01 5.09945333e-01
3.72594684e-01 6.82920396e-01 8.28028321e-01 -7.75101602e-01
-7.00432718e-01 -8.39893341e-01 -8.30032885e-01 5.83414853e-01
4.23038185e-01 -3.20584059e-01 -6.22689947e-02 3.64183545e-01] | [11.320850372314453, -1.6525719165802002] |
25ec00e0-7e4b-405f-baba-35fe35a897b0 | can-machine-learning-discover-the-determining | 2212.03092 | null | https://arxiv.org/abs/2212.03092v3 | https://arxiv.org/pdf/2212.03092v3.pdf | Can Machine Learning discover the determining factors in participation in insurance schemes? A comparative analysis | Identifying factors that affect participation is key to a successful insurance scheme. This study's challenges involve using many factors that could affect insurance participation to make a better forecast.Huge numbers of factors affect participation, making evaluation difficult. These interrelated factors can mask the influence on adhesion predictions, making them misleading.This study evaluated how 66 common characteristics affect insurance participation choices. We relied on individual farm data from FADN from 2016 to 2019 with type 1 (Fieldcrops) farming with 10,926 observations.We use three Machine Learning (ML) approaches (LASSO, Boosting, Random Forest) compare them to the GLM model used in insurance modelling. ML methodologies can use a large set of information efficiently by performing the variable selection. A highly accurate parsimonious model helps us understand the factors affecting insurance participation and design better products.ML predicts fairly well despite the complexity of insurance participation problem. Our results suggest Boosting performs better than the other two ML tools using a smaller set of regressors. The proposed ML tools identify which variables explain participation choice. This information includes the number of cases in which single variables are selected and their relative importance in affecting participation.Focusing on the subset of information that best explains insurance participation could reduce the cost of designing insurance schemes. | ['Simone Severini', 'Luigi Biagini'] | 2022-12-06 | null | null | null | null | ['variable-selection'] | ['methodology'] | [ 3.33566874e-01 -9.83567256e-03 -1.17513800e+00 -4.02661294e-01
-2.35030919e-01 -2.79288411e-01 -3.52208912e-02 7.27644563e-02
-7.08015338e-02 8.83922160e-01 5.26077032e-01 -1.06442904e+00
-2.62724280e-01 -1.11808205e+00 -9.53429103e-01 -4.52075303e-01
-1.32801384e-02 1.07616596e-01 -4.15959507e-01 -5.79835892e-01
1.15569755e-01 1.67484224e-01 -1.63429224e+00 4.49851632e-01
1.14031267e+00 4.87150490e-01 6.06573641e-01 2.59813190e-01
-5.83568253e-02 9.10531461e-01 -2.30003774e-01 -3.23725164e-01
4.87110525e-01 -4.73340243e-01 -3.82886142e-01 -2.98201829e-01
-8.89608711e-02 -4.79675323e-01 3.78397614e-01 4.19072181e-01
1.85472444e-01 -5.21559119e-01 9.11287904e-01 -1.25858986e+00
-6.40772939e-01 1.08262742e+00 -8.32301497e-01 -2.02939928e-01
4.96714205e-01 -1.63857546e-02 9.55858231e-01 -5.03716409e-01
6.35115623e-01 1.59884310e+00 1.17401135e+00 -7.19383284e-02
-1.53480923e+00 -8.39450657e-01 7.10830837e-02 1.18191428e-01
-8.97196829e-01 -2.05654442e-01 5.01311719e-01 -9.65878189e-01
6.67945683e-01 6.00150287e-01 5.44661522e-01 5.96209049e-01
1.97234526e-01 6.40272021e-01 1.36767030e+00 -6.50632799e-01
-1.80835724e-01 5.25951982e-01 2.69674361e-01 4.74457383e-01
7.11954474e-01 5.03524244e-01 -5.17129362e-01 -7.13874221e-01
6.42429113e-01 4.30503905e-01 6.97262585e-02 -3.45310792e-02
-1.10120964e+00 1.43049395e+00 2.48273954e-01 -1.93099603e-02
-8.05093467e-01 7.67805278e-02 -2.97508091e-02 6.68409586e-01
6.77468777e-02 2.35957131e-01 -1.21439314e+00 5.06048501e-01
-6.07281566e-01 5.24386048e-01 7.55051792e-01 6.74785197e-01
1.15259802e+00 -1.44400224e-01 -3.09578866e-01 6.76218867e-01
4.39665794e-01 8.13098133e-01 7.66994506e-02 -8.97665262e-01
6.55044794e-01 1.08292592e+00 3.55322838e-01 -1.23851120e+00
-4.52890694e-01 -1.92650482e-01 -7.06966341e-01 4.12569076e-01
5.81490695e-01 -6.86128795e-01 -7.71722972e-01 1.58247674e+00
1.19076535e-01 -6.37725830e-01 -2.52067804e-01 6.97944641e-01
3.46521229e-01 3.42068553e-01 4.47847873e-01 -4.34348077e-01
1.41182852e+00 -5.01531422e-01 -8.70067298e-01 -1.76578104e-01
7.81019032e-01 -9.36317027e-01 9.13210571e-01 2.19171807e-01
-5.65921307e-01 -6.57936871e-01 -6.16289973e-01 5.38727641e-01
-3.75751942e-01 4.77766484e-01 1.41577101e+00 9.56542194e-01
-4.85104620e-01 5.59048951e-01 -5.04173934e-01 -2.50524670e-01
3.55247855e-01 5.99739730e-01 -1.10378042e-01 2.15576470e-01
-1.26165056e+00 9.75549757e-01 -2.05254015e-02 -9.46719125e-02
-3.23497802e-01 -7.97653615e-01 -5.54812133e-01 -1.79236233e-01
4.30358589e-01 -6.55630529e-01 7.26204932e-01 -1.38716733e+00
-8.84323716e-01 4.55692947e-01 -4.90777373e-01 -4.54726636e-01
5.88692904e-01 -2.56409168e-01 -1.52114347e-01 -6.13566458e-01
3.08536768e-01 3.92330199e-01 4.11853880e-01 -1.07854629e+00
-6.76366866e-01 -5.68261981e-01 -1.65492743e-01 -1.08979143e-01
2.50514355e-02 3.18159938e-01 7.09427059e-01 -8.12645495e-01
4.22749549e-01 -9.83266532e-01 -6.05936825e-01 -4.00717854e-01
-9.42277089e-02 1.51465103e-01 2.64306903e-01 -1.07323301e+00
1.41804898e+00 -1.47656560e+00 -2.22711042e-01 1.72441944e-01
-3.93386960e-01 -9.45186317e-02 2.44216740e-01 8.52953732e-01
2.16676388e-02 5.45773327e-01 -1.78594425e-01 4.43186194e-01
-1.06309228e-01 1.84941456e-01 -2.26949155e-01 2.36568347e-01
3.91345382e-01 7.66475618e-01 -9.87940803e-02 -4.59238827e-01
2.92233258e-01 2.52177175e-02 -7.54763603e-01 1.70734569e-01
-5.74864037e-02 2.25616425e-01 -7.16936707e-01 1.16462958e+00
1.05607295e+00 1.06311336e-01 4.97854620e-01 6.34978386e-03
-3.44201416e-01 4.52205539e-02 -1.22006583e+00 8.33254337e-01
-3.08378130e-01 2.52232969e-01 1.15820169e-01 -1.30591345e+00
1.18851233e+00 2.57239223e-01 5.04985988e-01 -1.61862850e-01
-2.27655217e-01 2.00253814e-01 7.43409023e-02 -1.01261294e+00
-1.67858511e-01 -6.98375106e-02 -7.60716423e-02 1.90159589e-01
-4.80091363e-01 4.63105917e-01 -1.35464370e-01 -3.91321987e-01
8.66926432e-01 2.46337608e-01 9.67367709e-01 -5.40646195e-01
3.59434664e-01 5.00780582e-01 1.25034797e+00 8.71892393e-01
-5.68997264e-02 1.25947714e-01 8.45538259e-01 -7.17331350e-01
-5.79812944e-01 -4.42413718e-01 -4.11659300e-01 1.51527786e+00
-3.45525593e-01 3.92701805e-01 -5.24652489e-02 -4.43235517e-01
1.04308069e+00 4.48771447e-01 -8.91493678e-01 3.88686836e-01
-5.17325461e-01 -1.42551792e+00 2.82546043e-01 2.87743151e-01
3.19277018e-01 -1.07442904e+00 -4.87985432e-01 3.17860842e-01
-2.16244191e-01 -1.75136015e-01 1.44066691e-01 4.98949707e-01
-1.16751897e+00 -1.14248538e+00 -5.05078375e-01 -7.23065794e-01
5.18537283e-01 3.76391113e-01 1.25450742e+00 3.14474970e-01
-6.81851506e-02 -6.96674466e-01 -4.09150273e-01 -9.36310232e-01
-5.63028753e-01 9.94041711e-02 -1.17844410e-01 -2.35728949e-01
8.34129989e-01 -5.38588762e-01 -2.97390223e-01 4.16703284e-01
-4.98616099e-01 1.46048889e-01 9.55138147e-01 8.90562296e-01
2.60688007e-01 -3.76271494e-02 1.25139129e+00 -1.44979179e+00
3.19209337e-01 -8.21498215e-01 -5.22631764e-01 5.43619335e-01
-9.08994913e-01 6.01225868e-02 8.37762430e-02 -5.09480298e-01
-1.02407932e+00 4.31610823e-01 1.97562590e-01 5.01210511e-01
-1.95649132e-01 8.96875679e-01 -1.68099508e-01 4.66090329e-02
6.69167578e-01 -5.62578797e-01 1.52685031e-01 -8.32793951e-01
-5.90123758e-02 7.02046931e-01 -2.93076575e-01 -1.14481315e-01
6.83594465e-01 2.11460650e-01 -1.82644203e-02 -4.37578648e-01
-3.80459458e-01 -3.33165020e-01 -7.01470315e-01 4.25102189e-02
6.07380748e-01 -1.10887504e+00 -1.00845695e+00 1.95162490e-01
-7.19135463e-01 -1.90325096e-01 3.93760830e-01 8.63726437e-01
-2.95912176e-01 -2.00970784e-01 -3.25916201e-01 -1.36165071e+00
-1.09455794e-01 -9.63825881e-01 4.69329357e-01 7.24287480e-02
-4.58511174e-01 -6.95129752e-01 -3.33959050e-03 8.98529589e-01
3.86524290e-01 7.68538594e-01 1.19728339e+00 -1.86428443e-01
-4.87968445e-01 2.22000740e-02 7.38395005e-02 -9.02402326e-02
3.07589442e-01 2.31737480e-01 -6.81253374e-01 -1.38979331e-02
-1.36553356e-02 -1.68388665e-01 1.06817722e+00 1.20607030e+00
7.21957326e-01 -4.55287904e-01 -4.49751735e-01 4.90002126e-01
1.61263919e+00 2.93269366e-01 3.64415646e-01 3.20075125e-01
4.83346343e-01 1.24529743e+00 1.10021269e+00 4.06894296e-01
4.87008154e-01 6.20041132e-01 3.50565881e-01 -5.80986202e-01
4.02515203e-01 -4.63538826e-01 4.35790151e-01 2.12322444e-01
-2.74823606e-01 1.21685743e-01 -1.04106498e+00 4.19189751e-01
-2.14950204e+00 -9.85325873e-01 -8.35238278e-01 2.01071644e+00
8.65153372e-01 -1.46933824e-01 3.76919478e-01 3.50520343e-01
4.46159273e-01 -1.60172984e-01 -4.19993550e-01 -4.90288675e-01
-3.27617377e-01 -1.54477432e-01 1.34828150e+00 6.34604394e-01
-1.00644207e+00 5.94462574e-01 7.18064308e+00 3.64757359e-01
-5.89859426e-01 -1.60053164e-01 8.67757380e-01 2.48338446e-01
-7.34352171e-01 3.63047332e-01 -8.53202879e-01 2.37697303e-01
5.67219675e-01 6.43847734e-02 3.47981095e-01 6.85700953e-01
9.55922425e-01 -5.35223067e-01 -6.08183563e-01 7.20807398e-03
-3.32585424e-01 -1.10600877e+00 -1.61799088e-01 2.09484845e-01
1.04940760e+00 -3.06618601e-01 -1.61885843e-01 2.43816346e-01
6.72378063e-01 -1.30784035e+00 4.59016025e-01 6.65646911e-01
4.29856658e-01 -5.09836674e-01 1.05717027e+00 5.97216249e-01
-1.03145957e+00 -5.25469780e-01 -5.26000440e-01 -9.55149949e-01
5.79375625e-02 7.42224276e-01 -9.81825948e-01 4.60132599e-01
7.46064126e-01 4.14250821e-01 -4.78474885e-01 3.00671548e-01
7.68094733e-02 1.03765500e+00 -1.09536134e-01 -2.06419248e-02
-1.89580247e-01 -4.12866980e-01 -3.38584334e-02 7.99520075e-01
5.13201118e-01 1.40591577e-01 2.05719352e-01 8.65480185e-01
3.77110332e-01 4.57712620e-01 -7.87718236e-01 9.46138501e-02
5.91348946e-01 6.60274565e-01 -4.24722224e-01 5.79613894e-02
-7.31582761e-01 4.45417985e-02 -1.10015944e-01 3.72654259e-01
-5.79128742e-01 1.58399463e-01 8.02404106e-01 3.51677299e-01
2.43932560e-01 1.78685620e-01 -1.02309990e+00 -7.68400431e-01
-1.31558821e-01 -1.36542463e+00 4.09116983e-01 -2.49136955e-01
-1.00773644e+00 -2.79579133e-01 -7.08287358e-02 -8.12008083e-01
-5.13664149e-02 -3.41341615e-01 -4.34061408e-01 1.51228750e+00
-1.44929349e+00 -1.30274928e+00 -1.25943944e-01 1.88767850e-01
3.98174644e-01 -4.32878405e-01 8.64841223e-01 1.43021554e-01
-5.43532252e-01 3.25316727e-01 1.66755825e-01 -3.34009618e-01
5.16374528e-01 -9.62591231e-01 1.79889426e-01 4.76618201e-01
-2.58449703e-01 6.09667778e-01 6.67387545e-01 -1.34170842e+00
-1.27783632e+00 -7.52749920e-01 1.46707690e+00 -4.44286019e-01
1.98688909e-01 -1.65280282e-01 -7.34426856e-01 6.12677872e-01
-3.46685141e-01 -8.14401805e-01 8.57417047e-01 6.80085361e-01
1.76025718e-01 -2.65500873e-01 -1.23360288e+00 2.14370847e-01
7.68571973e-01 -1.36653990e-01 -3.10927629e-01 3.42425108e-01
7.43527412e-01 2.78708488e-01 -9.54826772e-01 6.68790877e-01
9.97767329e-01 -9.79653239e-01 1.06109571e+00 -8.42570126e-01
4.90303457e-01 2.00396758e-02 -2.93218017e-01 -6.00656450e-01
-6.51970208e-01 -5.80324411e-01 3.06608438e-01 1.30047131e+00
9.27829862e-01 -6.62155628e-01 8.85754704e-01 6.08964801e-01
6.92297637e-01 -8.89108717e-01 -2.49781206e-01 -4.19157356e-01
9.34572592e-02 -8.00858587e-02 1.13575351e+00 1.11513913e+00
-2.56285667e-01 -6.16179146e-02 -7.70149767e-01 1.66444391e-01
3.47914815e-01 4.39620376e-01 9.88321900e-01 -1.41202259e+00
-2.26674810e-01 -1.17914744e-01 -4.87457663e-02 -2.21154273e-01
-3.02585550e-02 -5.58211446e-01 -4.93192464e-01 -1.28422642e+00
4.73728925e-01 -9.20694649e-01 -5.17659821e-02 7.14619517e-01
-5.77403128e-01 -3.60310227e-01 3.36307324e-02 3.45688254e-01
6.52113557e-01 -9.47570428e-02 8.79938781e-01 3.96252275e-02
-7.47316718e-01 7.42534459e-01 -1.07170093e+00 7.45060742e-01
1.02215576e+00 -7.10221052e-01 -1.34398550e-01 -3.09659630e-01
4.51144248e-01 4.86154586e-01 2.91752070e-01 -3.12489778e-01
-4.84110028e-01 -1.18346834e+00 5.17569363e-01 -7.77742743e-01
-2.85226196e-01 -9.00213063e-01 9.83203053e-01 9.26515937e-01
-6.82626784e-01 1.11057341e-01 -1.00109927e-01 2.86752582e-01
-1.49468295e-02 -3.43586236e-01 2.34915152e-01 -9.14476216e-02
-1.25754088e-01 -1.12427659e-01 -5.05374014e-01 -4.96794999e-01
8.23897660e-01 -2.87343085e-01 3.59985866e-02 -3.10857505e-01
-4.98803914e-01 2.65749156e-01 5.11988878e-01 3.85839075e-01
-1.12614088e-01 -1.45222890e+00 -1.19900393e+00 3.45592529e-01
-2.89285451e-01 -6.85644746e-01 -7.27414787e-02 7.51830339e-01
-7.74113894e-01 7.09922969e-01 -5.09693503e-01 -1.57847255e-01
-1.44363761e+00 4.17643279e-01 8.38248432e-02 -3.89201224e-01
2.32744999e-02 6.99026406e-01 1.33956790e-01 -4.63682652e-01
4.61566783e-02 -4.64257002e-01 -7.98041523e-01 4.07753825e-01
4.30075265e-02 8.13656330e-01 -3.21124494e-01 -4.05986756e-01
-4.51205283e-01 5.32376647e-01 3.50200325e-01 1.94243580e-01
1.46732581e+00 -2.39677131e-01 -9.76893380e-02 5.34825146e-01
8.63825202e-01 1.20950624e-01 -1.09878325e+00 3.68873868e-03
3.91040713e-01 -6.40703440e-01 -4.06497270e-02 -9.63288665e-01
-1.01621175e+00 4.13837224e-01 7.12485254e-01 3.84554744e-01
1.24390268e+00 -4.68431026e-01 3.04196209e-01 1.84487253e-01
3.21181029e-01 -9.59454179e-01 -8.41563165e-01 -8.92070010e-02
9.22072768e-01 -2.01964450e+00 3.40192348e-01 -6.32651031e-01
-8.09624374e-01 7.15733945e-01 2.09864691e-01 1.02672502e-01
7.46428013e-01 8.92318785e-02 3.28858018e-01 2.34946124e-02
-7.97381818e-01 -1.62883073e-01 6.14907481e-02 7.44183540e-01
9.49702621e-01 8.60759139e-01 -1.00469911e+00 8.96285892e-01
-3.44808161e-01 2.18121976e-01 2.75753736e-01 6.91793621e-01
-5.32357275e-01 -1.40624011e+00 -7.54204214e-01 1.19672322e+00
-7.86741614e-01 3.41561846e-02 -4.29945946e-01 9.59226489e-01
7.41807938e-01 1.37117970e+00 -5.84007382e-01 -2.59369552e-01
3.57974887e-01 5.78253604e-02 -2.27847815e-01 -3.40733200e-01
-1.04104757e+00 -5.89252822e-02 4.34167057e-01 -3.66378486e-01
-7.18595147e-01 -1.10693634e+00 -6.31321430e-01 -7.73772359e-01
-8.23190451e-01 -1.16306499e-01 6.16470218e-01 7.00881720e-01
9.27331448e-02 2.20450118e-01 9.72378075e-01 -3.36987287e-01
-8.17631483e-01 -1.03738737e+00 -6.05194688e-01 2.62404799e-01
3.33825231e-01 -7.29260683e-01 -4.28166181e-01 3.75393838e-01] | [7.818584442138672, 5.032001972198486] |
9e42e065-e235-498c-9b62-4c1837af519a | visual-exploration-and-knowledge-discovery | 2009.13059 | null | https://arxiv.org/abs/2009.13059v1 | https://arxiv.org/pdf/2009.13059v1.pdf | Visual Exploration and Knowledge Discovery from Biomedical Dark Data | Data visualization techniques proffer efficient means to organize and present data in graphically appealing formats, which not only speeds up the process of decision making and pattern recognition but also enables decision-makers to fully understand data insights and make informed decisions. Over time, with the rise in technological and computational resources, there has been an exponential increase in the world's scientific knowledge. However, most of it lacks structure and cannot be easily categorized and imported into regular databases. This type of data is often termed as Dark Data. Data visualization techniques provide a promising solution to explore such data by allowing quick comprehension of information, the discovery of emerging trends, identification of relationships and patterns, etc. In this empirical research study, we use the rich corpus of PubMed comprising of more than 30 million citations from biomedical literature to visually explore and understand the underlying key-insights using various information visualization techniques. We employ a natural language processing based pipeline to discover knowledge out of the biomedical dark data. The pipeline comprises of different lexical analysis techniques like Topic Modeling to extract inherent topics and major focus areas, Network Graphs to study the relationships between various entities like scientific documents and journals, researchers, and, keywords and terms, etc. With this analytical research, we aim to proffer a potential solution to overcome the problem of analyzing overwhelming amounts of information and diminish the limitation of human cognition and perception in handling and examining such large volumes of data. | ['Shashwat Aggarwal', 'Ramesh Singh'] | 2020-09-28 | null | null | null | null | ['lexical-analysis'] | ['natural-language-processing'] | [ 1.43152068e-03 3.02229505e-02 -2.31304973e-01 7.55034313e-02
1.30708925e-02 -6.54021978e-01 5.50275743e-01 1.32910466e+00
-3.38229328e-01 5.95694244e-01 4.67156857e-01 -8.42363775e-01
-4.22503889e-01 -8.54803741e-01 2.18922906e-02 -4.11052346e-01
-2.05332890e-01 1.70946985e-01 1.69499665e-01 -7.18232617e-02
8.10435057e-01 7.12711334e-01 -1.56619227e+00 3.20298404e-01
1.17869234e+00 7.05142975e-01 5.10986209e-01 1.09083630e-01
-9.75839913e-01 3.90253544e-01 -7.50221491e-01 -1.79335639e-01
-2.77978927e-01 -2.14636356e-01 -6.05506837e-01 -4.89809483e-01
-2.51762360e-01 4.35360938e-01 2.14959860e-01 1.26875436e+00
3.88426721e-01 -1.21412158e-01 3.06403607e-01 -9.56128836e-01
-6.10698819e-01 3.46409619e-01 -9.80229437e-01 6.11595988e-01
6.98677063e-01 1.48480595e-03 5.26871383e-01 -8.30062509e-01
1.27361083e+00 1.47410989e+00 2.63918996e-01 -8.01893994e-02
-1.02592957e+00 -5.27057469e-01 -8.21721554e-03 1.74136266e-01
-1.21042299e+00 1.02623738e-01 6.98433816e-01 -8.40189219e-01
7.65179217e-01 4.55288351e-01 9.52394545e-01 7.75651574e-01
3.06894720e-01 3.04198027e-01 1.21284795e+00 -6.43793046e-01
1.13910839e-01 5.02560019e-01 6.33567572e-01 7.81950057e-01
5.99376738e-01 -4.30531412e-01 -6.93250537e-01 -2.64097154e-01
3.51704806e-01 4.21197712e-01 -4.34985042e-01 2.94196652e-03
-1.45884836e+00 6.07629418e-01 4.51620847e-01 6.14454269e-01
-5.54958642e-01 -6.40984237e-01 5.60536563e-01 9.82867554e-02
4.42524523e-01 6.51018023e-01 -4.87402901e-02 -2.24053949e-01
-7.99993455e-01 4.14295346e-02 5.76826096e-01 3.88746053e-01
6.04115129e-01 -4.19810146e-01 6.41145036e-02 4.77016211e-01
4.11128014e-01 1.49071589e-01 8.28376472e-01 -3.10005367e-01
4.86330301e-01 1.60044992e+00 -7.79120997e-02 -1.50984180e+00
-7.41593003e-01 -3.27895522e-01 -9.13203299e-01 4.51719463e-01
2.22470820e-01 2.44740203e-01 -8.85328770e-01 1.25478005e+00
6.50592089e-01 -5.20479620e-01 -3.24886322e-01 6.88838959e-01
1.12635076e+00 5.02871931e-01 5.15500247e-01 -2.13852286e-01
2.07531261e+00 1.99240074e-02 -1.15602422e+00 2.84396172e-01
7.64306664e-01 -7.17409909e-01 1.24949384e+00 4.88555074e-01
-6.20889306e-01 -3.46979558e-01 -1.20582473e+00 -3.35728228e-01
-1.29912817e+00 -4.31124009e-02 5.89389145e-01 3.29234421e-01
-8.04110944e-01 5.28508365e-01 -7.70777881e-01 -7.36091435e-01
9.04460728e-01 -1.16830952e-01 -5.55383742e-01 -1.18317138e-02
-1.09311318e+00 9.87476826e-01 5.05392671e-01 1.93802211e-02
-8.58550239e-03 -1.13724411e+00 -5.48898578e-01 -2.18247622e-02
3.59369159e-01 -6.48715019e-01 3.57355922e-01 -1.10268161e-01
-5.22804499e-01 8.95845234e-01 -3.65219295e-01 -1.77131668e-01
1.83937013e-01 -2.11262122e-01 -6.01745427e-01 -5.91733456e-02
1.94527417e-01 1.67711347e-01 1.52420297e-01 -9.18032169e-01
-5.26281536e-01 -8.47062767e-01 -2.35743508e-01 -1.61631718e-01
-4.54426140e-01 2.92947471e-01 -2.24263474e-01 -6.69191182e-01
1.60397038e-01 -2.95502394e-01 -3.24333996e-01 1.86257333e-01
-6.93879902e-01 -2.66033828e-01 1.27264118e+00 -8.08600068e-01
1.65920138e+00 -2.29008627e+00 4.46837768e-02 4.74427789e-01
9.20060694e-01 1.95690244e-01 6.45685017e-01 6.92039669e-01
-8.73143077e-02 6.49982035e-01 -1.43587098e-01 1.96606070e-01
-3.11367571e-01 -1.57015234e-01 -1.50426373e-01 2.40468621e-01
-6.48298264e-02 7.53337681e-01 -9.46468234e-01 -7.06622243e-01
1.08733565e-01 6.47744417e-01 -1.28231421e-01 -3.01897433e-02
-7.36345872e-02 4.59281027e-01 -6.87206209e-01 6.95500851e-01
4.13559884e-01 -6.38297677e-01 1.98990792e-01 -2.76378632e-01
-7.64752984e-01 4.27027881e-01 -1.01913095e+00 1.54377532e+00
-1.63915247e-01 1.13035035e+00 -1.92691460e-01 -9.85270619e-01
9.74814057e-01 1.66591778e-01 2.43569493e-01 -8.98969412e-01
2.22628880e-02 -1.85474847e-02 4.34789099e-02 -7.86228001e-01
3.08583915e-01 1.60461321e-01 2.58972496e-01 5.04472852e-01
-4.93299484e-01 6.39279425e-01 3.56224477e-01 5.53756475e-01
8.57387304e-01 -2.50849396e-01 5.83020747e-01 -4.94796932e-01
3.06392521e-01 3.10597539e-01 2.56400228e-01 1.40483245e-01
3.08609307e-01 3.93418223e-02 9.23219323e-01 -6.34966671e-01
-5.46460390e-01 -8.45774591e-01 -2.84815460e-01 5.28374374e-01
-1.01300903e-01 -6.73549652e-01 -3.08614761e-01 -1.63369462e-01
1.14444993e-01 6.02093160e-01 -8.96762013e-01 2.50431836e-01
-2.34009013e-01 -7.53125072e-01 1.00765869e-01 4.37642410e-02
3.00764501e-01 -1.15749264e+00 -1.17888284e+00 -3.86239029e-02
1.54148648e-02 -5.73645234e-01 2.27813944e-01 -3.37904692e-02
-1.05391765e+00 -1.34761846e+00 -5.75209916e-01 -3.30827892e-01
8.18323612e-01 1.36162683e-01 9.75295424e-01 6.63589984e-02
-1.16236734e+00 -2.99710542e-01 -1.58698678e-01 -9.85558391e-01
-3.19641948e-01 -1.09842777e-01 -3.04926485e-01 -3.57684076e-01
4.69418973e-01 -4.67171878e-01 -8.90530050e-01 -1.06692448e-01
-9.75650012e-01 2.70278484e-01 4.77263957e-01 2.34691799e-01
5.99453926e-01 1.65539205e-01 4.40892249e-01 -1.15126455e+00
1.06144094e+00 -9.61235225e-01 -4.52873677e-01 3.66295010e-01
-1.05102074e+00 1.92527741e-01 2.18350708e-01 -5.95380999e-02
-8.98191988e-01 -5.70319951e-01 4.90089893e-01 -6.85061067e-02
-1.19770296e-01 1.20871222e+00 -1.94263667e-01 3.59348357e-01
5.31253636e-01 -1.82543427e-01 1.48696512e-01 -9.15716887e-01
3.97104174e-01 7.08114505e-01 2.91711122e-01 -5.64940274e-02
3.55510205e-01 4.35288817e-01 2.05891654e-01 -8.82603109e-01
-2.79949993e-01 -6.06590211e-01 -6.54107690e-01 -3.44530016e-01
8.45313191e-01 -4.24142271e-01 -7.99068868e-01 -2.22140536e-01
-1.07871771e+00 3.94089997e-01 -5.82912676e-02 2.68982947e-01
3.92190427e-01 3.96622032e-01 -8.74272212e-02 -5.66451967e-01
-4.36712354e-01 -9.24294949e-01 5.43403208e-01 4.85725641e-01
-7.13761091e-01 -1.09071684e+00 1.58782691e-01 6.07896410e-02
4.66454685e-01 7.78624296e-01 1.64970827e+00 -5.98403037e-01
-4.49853569e-01 -1.97585344e-01 -5.03942907e-01 -6.29135489e-01
4.49733257e-01 2.75067717e-01 -6.04367137e-01 1.67951465e-01
-3.74139786e-01 2.41093457e-01 4.13668841e-01 1.41801447e-01
1.30534208e+00 -3.71408015e-01 -9.87270892e-01 1.61149770e-01
1.20300055e+00 6.49227619e-01 4.92448270e-01 5.56811273e-01
7.54720569e-01 1.11478257e+00 4.20814335e-01 3.48034561e-01
2.90937036e-01 5.82218528e-01 2.51632452e-01 -3.12383741e-01
4.87119965e-02 -1.66314051e-01 -5.59780419e-01 6.59310937e-01
-2.60111421e-01 4.45048474e-02 -1.33882201e+00 4.42592293e-01
-1.64793456e+00 -7.30051816e-01 -4.18975025e-01 2.16333556e+00
7.07321882e-01 1.51666626e-01 -1.92067008e-02 1.95506632e-01
5.62467873e-01 -2.45465130e-01 -4.18728232e-01 -3.54498208e-01
2.15899628e-02 -3.90790440e-02 -9.20180604e-03 4.70012948e-02
-5.92702985e-01 3.73586625e-01 5.22296047e+00 5.24394751e-01
-1.37879014e+00 -1.26550838e-01 6.26364231e-01 -1.23769782e-01
-5.40424526e-01 -1.14586763e-01 -4.57356930e-01 5.02416015e-01
8.90221536e-01 -5.65340042e-01 -1.36356130e-01 5.94815135e-01
6.91976607e-01 -4.81690824e-01 -8.26672435e-01 1.06905973e+00
-4.78932202e-01 -1.78547680e+00 2.40860745e-01 3.14249545e-01
3.70069444e-02 -3.84586632e-01 -2.44278759e-02 -2.71272659e-01
1.66011229e-01 -1.08551633e+00 3.57973009e-01 7.78712153e-01
7.68993318e-01 -4.56725508e-01 4.78318781e-01 9.55348983e-02
-9.47361529e-01 5.34031801e-02 -1.43287390e-01 2.57365089e-02
1.50717394e-02 9.37478542e-01 -8.08653593e-01 7.64368117e-01
1.04030240e+00 7.61263371e-01 -7.39297390e-01 1.40292597e+00
1.57590911e-01 3.50567967e-01 -1.16576217e-01 -1.99561745e-01
-1.69812605e-01 -1.60370499e-01 4.50263798e-01 1.30662978e+00
4.02110159e-01 7.69337863e-02 -2.19651148e-01 9.23850358e-01
7.44650736e-02 5.47384202e-01 -6.45921230e-01 -6.39172733e-01
5.69163203e-01 1.31018877e+00 -1.43259799e+00 -3.05494726e-01
-3.42558950e-01 2.32225046e-01 2.76593596e-01 2.83171713e-01
-9.99367163e-02 -7.04733849e-01 4.31063652e-01 5.79658747e-01
-3.42339247e-01 -2.74649620e-01 -5.52854359e-01 -6.68997526e-01
-6.16068542e-02 -8.58980298e-01 6.38637304e-01 -8.05524051e-01
-8.58736634e-01 5.87383449e-01 2.53728509e-01 -1.00120401e+00
1.11634098e-01 -4.94618624e-01 -6.63802624e-01 1.28376174e+00
-9.75194812e-01 -5.78221083e-01 -3.64047855e-01 2.75719732e-01
3.29793125e-01 -2.10451148e-02 8.52314174e-01 2.41893321e-01
-7.54397035e-01 -1.94926769e-01 1.86447978e-01 -1.04261279e-01
5.04265845e-01 -1.19198442e+00 1.99504152e-01 4.39860791e-01
1.55902356e-01 1.14193499e+00 7.63674080e-01 -9.23407555e-01
-1.24341702e+00 -3.06403220e-01 8.56993556e-01 -3.08201641e-01
8.76738548e-01 -3.92585129e-01 -1.30040586e+00 6.79766983e-02
2.35681430e-01 -4.00286824e-01 1.18086815e+00 3.66427004e-01
-2.46192411e-01 6.39267564e-02 -8.89948487e-01 7.35783160e-01
6.52158916e-01 -2.57281899e-01 -7.25869536e-01 1.25730261e-01
5.12901485e-01 -7.49103054e-02 -8.31852615e-01 7.35799745e-02
5.93877017e-01 -6.17425501e-01 8.27790260e-01 -7.41050720e-01
2.84201831e-01 -2.28951737e-01 4.23737109e-01 -1.03453159e+00
-3.70198935e-02 -5.03349841e-01 -2.99226288e-02 1.22596192e+00
7.00058937e-01 -6.36440635e-01 5.17896414e-01 7.23005831e-01
1.97685920e-02 -9.88260150e-01 -7.39421546e-01 7.61951208e-02
-3.50073487e-01 -7.98967257e-02 4.60294485e-01 1.20095706e+00
5.92553496e-01 1.69474870e-01 3.57361048e-01 -2.20105782e-01
3.88175368e-01 4.16972548e-01 3.53566140e-01 -1.83280611e+00
4.27530527e-01 -7.21536160e-01 -5.29369295e-01 -8.42987448e-02
-6.15160227e-01 -8.23498487e-01 -8.15516889e-01 -2.17329288e+00
2.38871917e-01 -4.99041706e-01 -4.09098715e-01 4.77588832e-01
-1.09044023e-01 -1.65149555e-01 -2.49300785e-02 5.58430851e-01
-2.56610483e-01 -5.18189929e-02 1.19978571e+00 -6.17685262e-03
-3.90766472e-01 -3.94901067e-01 -1.02033401e+00 6.70506477e-01
5.22583663e-01 -4.69688118e-01 -5.63496113e-01 2.05179509e-02
8.07949901e-01 -1.07993171e-01 3.32438290e-01 -6.49037540e-01
2.99281299e-01 -1.27174154e-01 6.44793868e-01 -8.44163179e-01
-2.57780850e-01 -6.02568746e-01 3.78988117e-01 6.20539188e-01
-3.10344934e-01 4.13064152e-01 5.95375836e-01 6.16513848e-01
-3.38964671e-01 4.79165427e-02 1.40876964e-01 -1.86275795e-01
-6.64158762e-01 -1.50633961e-01 -4.08967227e-01 -1.22629553e-01
9.22116518e-01 -3.69941324e-01 -7.55389333e-01 -7.21237883e-02
-7.88737714e-01 2.98592448e-01 2.84807116e-01 6.26513422e-01
5.16180158e-01 -9.13515031e-01 -3.10268521e-01 6.53666854e-02
3.36044729e-02 1.26160547e-01 3.81820530e-01 8.54615331e-01
-7.38742709e-01 5.46664000e-01 -4.46462393e-01 -3.52840453e-01
-1.53432477e+00 7.38645315e-01 -4.25027192e-01 -4.01660174e-01
-9.92948055e-01 4.62855905e-01 3.95687640e-01 2.59257168e-01
2.60599285e-01 -3.11880141e-01 -9.79116559e-01 6.27914965e-01
8.12494576e-01 5.55202484e-01 7.16859996e-02 -3.59376311e-01
-6.04416192e-01 4.10872817e-01 -2.50287205e-01 1.51301429e-01
1.34075749e+00 -1.92639083e-01 -3.49819869e-01 7.93235660e-01
9.52011466e-01 2.41314515e-01 -3.51813585e-01 6.76564872e-02
6.16309702e-01 -4.53476131e-01 -1.42033696e-01 -1.10875010e+00
-6.68623507e-01 1.21090078e+00 6.07422709e-01 7.34991372e-01
9.15730894e-01 1.61704570e-01 2.51386344e-01 8.48481432e-02
-1.96174048e-02 -6.06026173e-01 -9.78897810e-02 -1.61066473e-01
9.42276478e-01 -1.17523241e+00 4.36150879e-01 -5.42619705e-01
-3.01692307e-01 1.37707055e+00 1.57692254e-01 6.28154278e-01
8.44099939e-01 3.27580333e-01 3.51340234e-01 -1.03309369e+00
-4.90026504e-01 9.15536731e-02 5.03145158e-01 4.53544110e-01
7.38339901e-01 -1.24499358e-01 -5.67606926e-01 6.20405614e-01
-1.95537552e-01 -2.15573292e-02 8.29184800e-02 9.42062080e-01
-5.50754666e-01 -9.92602825e-01 -4.32694256e-01 5.99602163e-01
-8.61609757e-01 -1.08504333e-01 -4.50495690e-01 9.27398145e-01
-2.45577451e-02 6.87961817e-01 1.56536996e-01 5.08040525e-02
3.20365936e-01 1.27973989e-01 -1.71220735e-01 -4.22830820e-01
-2.92203337e-01 -4.98454012e-02 -4.71005812e-02 -4.22517508e-01
-4.60070968e-01 -4.02087480e-01 -1.44866920e+00 -1.31029546e-01
-4.37909551e-03 4.25852001e-01 1.17237246e+00 8.18829238e-01
8.03089738e-01 8.85132790e-01 -8.66341963e-02 -2.95662731e-01
3.29505116e-01 -8.28859270e-01 -1.49434239e-01 3.16909224e-01
6.48081005e-02 -7.63205230e-01 3.13962884e-02 1.26208678e-01] | [9.560006141662598, 7.948827743530273] |
3dc116b8-cc1f-4b66-a2b6-206b5c381519 | monoise-modeling-noise-using-a-modular | 1710.03476 | null | http://arxiv.org/abs/1710.03476v1 | http://arxiv.org/pdf/1710.03476v1.pdf | MoNoise: Modeling Noise Using a Modular Normalization System | We propose MoNoise: a normalization model focused on generalizability and
efficiency, it aims at being easily reusable and adaptable. Normalization is
the task of translating texts from a non- canonical domain to a more canonical
domain, in our case: from social media data to standard language. Our proposed
model is based on a modular candidate generation in which each module is
responsible for a different type of normalization action. The most important
generation modules are a spelling correction system and a word embeddings
module. Depending on the definition of the normalization task, a static lookup
list can be crucial for performance. We train a random forest classifier to
rank the candidates, which generalizes well to all different types of
normaliza- tion actions. Most features for the ranking originate from the
generation modules; besides these features, N-gram features prove to be an
important source of information. We show that MoNoise beats the
state-of-the-art on different normalization benchmarks for English and Dutch,
which all define the task of normalization slightly different. | ['Rob van der Goot', 'Gertjan van Noord'] | 2017-10-10 | null | null | null | null | ['lexical-normalization'] | ['natural-language-processing'] | [ 3.16476852e-01 1.93293408e-01 -3.39940488e-01 -3.14578980e-01
-6.46463215e-01 -8.10657024e-01 1.04672647e+00 7.83172905e-01
-8.32937360e-01 5.65879405e-01 5.49129248e-01 -1.60470784e-01
-1.42544627e-01 -9.58325744e-01 -4.83471930e-01 -5.43165922e-01
3.02683353e-01 8.55854213e-01 4.33167845e-01 -8.63849103e-01
1.37413949e-01 5.29771805e-01 -1.44263875e+00 3.28152210e-01
8.18618894e-01 4.71450627e-01 -7.14310631e-02 6.10432744e-01
-4.23058748e-01 2.69113362e-01 -6.29746795e-01 -8.31814229e-01
2.65368015e-01 -5.16075492e-01 -9.44184244e-01 -3.87602389e-01
3.29222739e-01 2.48378530e-01 1.42504007e-01 1.11096692e+00
5.93920171e-01 1.12896830e-01 9.68028069e-01 -1.01073623e+00
-3.54141206e-01 1.05933106e+00 9.36003868e-03 6.42911643e-02
4.36585516e-01 -3.28074634e-01 1.24769211e+00 -7.31196702e-01
8.59564602e-01 1.10655940e+00 6.17174506e-01 7.26948619e-01
-1.29552066e+00 -2.59745449e-01 1.26102149e-01 1.53195828e-01
-9.46749806e-01 -5.14724195e-01 1.84636489e-01 -5.76222539e-01
8.15253615e-01 5.37285805e-01 1.92332387e-01 1.26807356e+00
1.61617175e-01 3.86869818e-01 9.18571353e-01 -7.77724087e-01
1.41069725e-01 1.88831568e-01 3.10881823e-01 1.33467466e-01
5.38722634e-01 -2.18873367e-01 -3.19747180e-01 7.24420603e-03
1.99589550e-01 -2.70022929e-01 -2.40464449e-01 -3.85123730e-01
-1.43338156e+00 6.92434430e-01 1.75607204e-01 6.65724397e-01
-1.62191972e-01 -2.85085291e-01 6.85991526e-01 5.43648303e-01
3.90872389e-01 8.17978740e-01 -6.45076990e-01 -2.12844238e-01
-7.53222466e-01 3.27653497e-01 1.07000721e+00 7.92630076e-01
8.97134960e-01 -4.57357019e-01 -6.27308965e-01 7.92480767e-01
-8.05379748e-02 3.32584441e-01 7.80568659e-01 -2.53640890e-01
6.49696290e-01 8.30509365e-01 2.15675142e-02 -5.88204622e-01
-6.38470411e-01 -6.55315638e-01 -8.68824482e-01 -5.26052080e-02
6.12016559e-01 -1.33161396e-01 -6.52976632e-01 1.93146551e+00
2.95624495e-01 -2.89056331e-01 1.26472920e-01 5.49052954e-01
7.31768787e-01 4.28445011e-01 7.47636035e-02 -6.44771308e-02
1.27994752e+00 -9.19890821e-01 -5.28638124e-01 -1.70036688e-01
9.62875485e-01 -1.09383690e+00 1.14496100e+00 2.28013128e-01
-9.38171506e-01 -3.70772064e-01 -9.65240180e-01 -1.73512742e-01
-8.08677793e-01 1.27945468e-01 1.93593144e-01 8.60387564e-01
-8.89228880e-01 9.01841402e-01 -6.07699335e-01 -8.06322873e-01
-2.28525072e-01 3.09608668e-01 -7.53613889e-01 1.95774704e-01
-1.29030132e+00 1.19666910e+00 6.45828545e-01 -3.59548897e-01
-1.67189613e-01 -5.10968328e-01 -7.97595620e-01 6.41981810e-02
1.71984300e-01 -7.08175123e-01 1.22295952e+00 -1.06850159e+00
-1.55238402e+00 1.07221782e+00 -4.79674991e-03 -4.44377422e-01
7.72588313e-01 -1.80233896e-01 -4.91556883e-01 -4.05706674e-01
2.76485234e-01 5.77954911e-02 7.08273888e-01 -7.67448485e-01
-7.54960597e-01 -1.89837083e-01 8.22442174e-02 3.28224339e-02
-6.77895010e-01 2.64521271e-01 -4.99828905e-01 -8.57471049e-01
-1.84926674e-01 -8.64115000e-01 -1.40603259e-01 -6.04112506e-01
-4.11957234e-01 -1.58179611e-01 1.82243571e-01 -6.68972850e-01
1.71490717e+00 -1.84493756e+00 5.09141922e-01 3.44158798e-01
9.84599218e-02 3.20675075e-01 -4.21633452e-01 7.03857541e-01
-4.53909785e-01 2.68230617e-01 -2.20442981e-01 -4.44231451e-01
3.04228514e-01 1.26513526e-01 -7.47081116e-02 3.75014961e-01
3.09690535e-01 7.94930995e-01 -9.52737927e-01 -8.87286738e-02
-1.97774265e-03 2.29186252e-01 -6.29794538e-01 5.42672612e-02
-8.65142941e-02 2.29004070e-01 -1.27879798e-01 -3.49757751e-03
6.74087644e-01 2.17465088e-01 1.35221660e-01 -1.83554098e-01
-2.62443572e-01 6.17486179e-01 -1.36467290e+00 1.66497982e+00
-5.74368179e-01 1.57019988e-01 -2.07485840e-01 -9.73367989e-01
8.49570513e-01 9.40556750e-02 2.44232029e-01 -5.85871339e-01
2.72506088e-01 4.01139528e-01 1.36399150e-01 -1.66763395e-01
7.21702635e-01 2.04983447e-02 -4.21032071e-01 5.33024251e-01
4.67345595e-01 1.77970529e-01 9.73356545e-01 1.18211798e-01
1.20976079e+00 9.08082351e-02 7.64033437e-01 -5.30649602e-01
8.36100221e-01 -1.76541656e-01 6.07738376e-01 7.36328542e-01
2.21115083e-01 6.30393922e-01 7.39597082e-01 -2.92303979e-01
-1.13570142e+00 -1.00180566e+00 -4.04217318e-02 1.57357168e+00
-2.10670248e-01 -8.49307477e-01 -1.03277361e+00 -8.03273201e-01
-2.17055306e-01 7.89106727e-01 -7.93655634e-01 -4.50709134e-01
-5.96365929e-01 -8.80079448e-01 4.66392636e-01 2.06867069e-01
-2.01578975e-01 -8.60537171e-01 -2.73577929e-01 2.04915285e-01
-1.58568889e-01 -1.00670981e+00 -5.06271362e-01 5.14370203e-01
-5.72624505e-01 -1.08750558e+00 -4.89393026e-01 -6.24587059e-01
5.41233182e-01 -2.28710577e-01 1.47270310e+00 -1.31580189e-01
1.56019598e-01 1.59163460e-01 -7.75937140e-01 -6.09878123e-01
-1.14968801e+00 1.01767778e+00 1.34125814e-01 1.87971964e-01
3.75248909e-01 -6.17634058e-01 -1.06834337e-01 2.84064591e-01
-1.23869550e+00 -5.36173172e-02 6.22542739e-01 6.92316532e-01
3.30315918e-01 -4.38142717e-01 2.73297995e-01 -1.26725709e+00
8.55476141e-01 -3.45504045e-01 -4.12116796e-01 3.71823072e-01
-4.73566294e-01 5.37509441e-01 1.04587400e+00 -3.12070012e-01
-6.91627443e-01 5.80728762e-02 -3.97375673e-01 4.76676792e-01
-3.09013247e-01 4.08969283e-01 -5.89661896e-01 1.53469920e-01
1.11060131e+00 -7.55933672e-02 -2.80856162e-01 -8.30746233e-01
6.74482465e-01 5.50723255e-01 5.63997805e-01 -6.38663828e-01
1.24147236e+00 1.82509068e-02 1.60980299e-02 -6.24744654e-01
-7.18823493e-01 -6.15803182e-01 -1.09602582e+00 1.37594536e-01
6.93219244e-01 -5.09709597e-01 -1.40581995e-01 1.74022660e-01
-1.39370131e+00 2.08284743e-02 -6.17342412e-01 3.06531012e-01
-3.76569420e-01 2.96834320e-01 -4.36024785e-01 -9.43733081e-02
-3.51724565e-01 -1.00488234e+00 9.10084546e-01 3.76390442e-02
-3.86792958e-01 -9.30250585e-01 5.97685218e-01 -1.73257723e-01
5.08142531e-01 1.64309293e-01 1.12621224e+00 -1.35157466e+00
6.60854653e-02 -4.37421709e-01 2.06207968e-02 6.30557895e-01
1.84572160e-01 5.47066927e-02 -9.53594565e-01 -1.48241624e-01
-2.78069407e-01 1.14482522e-01 9.99379694e-01 -2.40455151e-01
5.92079043e-01 -3.69502455e-01 -9.40501317e-02 6.25971973e-01
1.22817397e+00 -3.44909638e-01 7.62269258e-01 6.59309626e-01
5.33594251e-01 4.31911051e-01 3.59179944e-01 2.41662398e-01
2.99632728e-01 8.85712683e-01 4.24751908e-01 -1.78817675e-01
-1.10578410e-01 -1.86458886e-01 6.80110157e-01 1.16052163e+00
-1.77820191e-01 -3.49862546e-01 -9.05680239e-01 4.91304278e-01
-1.79928184e+00 -8.63007009e-01 -2.58898795e-01 2.63910866e+00
9.51484323e-01 1.90166682e-01 2.76516587e-01 4.15496081e-01
6.72664702e-01 5.88780381e-02 1.96044847e-01 -8.47753525e-01
-3.46704125e-01 6.40119255e-01 5.58957577e-01 4.97420311e-01
-1.30924582e+00 9.29379940e-01 5.63783026e+00 9.60513532e-01
-1.17648780e+00 3.00255537e-01 4.27857190e-02 1.62904665e-01
-3.54224592e-01 -5.27025526e-03 -9.98283982e-01 2.62540996e-01
9.45618629e-01 -4.68700856e-01 2.49722078e-01 6.85519159e-01
-8.48206058e-02 3.03280741e-01 -1.38216722e+00 5.57000816e-01
1.43087298e-01 -9.91865456e-01 4.02084202e-01 -1.61487654e-01
6.44806862e-01 2.26897344e-01 -4.56134975e-01 3.22857946e-01
2.62263507e-01 -8.01471472e-01 9.20758069e-01 5.87980747e-01
6.52136266e-01 -7.99836457e-01 1.02546167e+00 4.11008298e-01
-1.06640017e+00 1.37303934e-01 -3.90565455e-01 -2.75635906e-02
1.21511154e-01 7.23018408e-01 -6.02786958e-01 9.09645975e-01
2.65687048e-01 5.72228312e-01 -8.92253101e-01 1.03251040e+00
-4.69360054e-01 3.83344024e-01 -2.52763212e-01 8.51836726e-02
-1.14485156e-02 -1.27464235e-01 7.44864583e-01 1.76400638e+00
4.31627065e-01 -6.05927467e-01 2.04017889e-02 4.62983072e-01
-1.78480595e-01 7.48716056e-01 -3.90598327e-01 9.84176695e-02
1.55524656e-01 1.53377378e+00 -6.97477520e-01 -2.26201043e-01
-3.69016707e-01 1.10712039e+00 4.42443788e-01 -8.62999354e-03
-8.20564032e-01 -6.39295280e-01 8.06746244e-01 3.22402805e-01
3.66755962e-01 2.12611500e-02 -5.52179553e-02 -1.54825878e+00
1.59883410e-01 -1.00218606e+00 3.47564518e-01 -1.44632950e-01
-1.42718709e+00 9.08105195e-01 -6.67036101e-02 -1.31728721e+00
-3.96591514e-01 -8.48126888e-01 -5.36750257e-01 9.88479316e-01
-1.58788586e+00 -1.01122081e+00 -2.10239559e-01 4.26938891e-01
7.35632181e-02 -2.24515244e-01 1.35456038e+00 4.72319275e-01
-7.50955284e-01 9.15900052e-01 3.01541120e-01 2.03556985e-01
1.25133729e+00 -1.41795671e+00 8.38077903e-01 1.20865643e+00
3.38861853e-01 7.39249229e-01 7.49050260e-01 -3.70068312e-01
-9.34741855e-01 -1.38033795e+00 1.63210559e+00 -5.36347091e-01
7.95194328e-01 -7.25071490e-01 -7.39608228e-01 4.46266800e-01
1.35123447e-01 -2.74929643e-01 6.84499800e-01 2.43402198e-01
-6.26151741e-01 -2.44408876e-01 -7.18368948e-01 6.37630880e-01
1.07210708e+00 -3.36139053e-01 -7.68359959e-01 1.89628243e-01
4.67689157e-01 -3.92708272e-01 -7.50916660e-01 1.79506660e-01
3.52347285e-01 -9.50248718e-01 7.18880296e-01 -8.55362236e-01
4.94161636e-01 -3.36103708e-01 -3.39545049e-02 -1.55758405e+00
-5.07294536e-01 -8.47267747e-01 2.73655385e-01 1.62525892e+00
7.81394601e-01 -6.73720896e-01 3.56642485e-01 3.06025118e-01
-6.19139634e-02 -1.99341416e-01 -8.80123258e-01 -9.89754975e-01
2.79273272e-01 -5.33922672e-01 8.81176710e-01 9.48357522e-01
4.64479327e-02 7.70388722e-01 -1.34455040e-01 -1.39735088e-01
1.99194089e-01 -1.32336229e-01 1.02261305e+00 -1.66663337e+00
-1.91769555e-01 -7.96013117e-01 -5.44562280e-01 -7.15295911e-01
1.37529060e-01 -1.45316529e+00 -1.95703357e-01 -1.44421768e+00
2.91722268e-02 -1.02922454e-01 -3.05653393e-01 5.24855316e-01
-1.73288286e-01 2.32943267e-01 3.50519270e-01 -7.43446276e-02
-6.85111463e-01 4.25175786e-01 7.21340299e-01 -1.23652995e-01
-1.64658055e-01 1.34240970e-01 -8.72786641e-01 6.57995820e-01
7.47606814e-01 -7.18801200e-01 4.80449460e-02 -5.78438938e-01
6.64978623e-01 -6.63195789e-01 -1.09076262e-01 -9.40031528e-01
2.58401990e-01 -8.57534707e-02 -7.89245218e-02 5.21493405e-02
-1.87179714e-01 -6.50429368e-01 -4.65923063e-02 2.66384512e-01
-3.06950688e-01 3.30131680e-01 8.59871954e-02 1.45516634e-01
-1.68694139e-01 -5.15292048e-01 6.88438475e-01 4.94742393e-02
-4.56681848e-01 1.82314456e-01 -2.58735299e-01 4.15201008e-01
6.23950124e-01 -9.70423128e-03 -1.97812453e-01 -2.05397326e-02
-7.52694368e-01 -2.26937190e-01 7.11144745e-01 6.78396642e-01
-2.29885161e-01 -1.38262701e+00 -9.87007976e-01 2.81515598e-01
5.61633289e-01 -2.94078887e-01 -3.03625077e-01 7.06656575e-01
-4.59130555e-01 2.29019612e-01 -1.85144410e-01 -2.83851743e-01
-1.19837666e+00 3.31453353e-01 2.42496461e-01 -7.84422338e-01
-2.26122260e-01 4.49012041e-01 -6.77850768e-02 -7.55572200e-01
8.80536214e-02 -5.94921947e-01 -5.11630952e-01 5.00294149e-01
5.85663855e-01 4.25818056e-01 8.26787055e-01 -9.48260128e-01
-3.86367291e-01 6.21021926e-01 -4.39125858e-02 -8.06339364e-03
1.44089210e+00 -1.99164674e-02 -4.96047914e-01 3.25831205e-01
1.02105904e+00 3.33294541e-01 -3.35374445e-01 -3.56571019e-01
4.09016728e-01 1.89491287e-02 -3.79658222e-01 -7.79493392e-01
-6.83117092e-01 7.51052976e-01 3.07862043e-01 1.35043338e-01
1.15905380e+00 -1.94886997e-01 3.97940308e-01 4.97794509e-01
2.06815481e-01 -1.18946815e+00 -5.07399857e-01 1.11505473e+00
8.87701035e-01 -9.06757891e-01 -1.56994015e-01 -2.90074319e-01
-3.28676730e-01 1.39941704e+00 2.18895212e-01 -1.47123728e-02
5.52877128e-01 3.14104468e-01 1.01140179e-01 3.15813959e-01
-4.24826980e-01 -4.92372990e-01 5.73062956e-01 6.35190785e-01
7.12022781e-01 1.08751528e-01 -8.84199142e-01 1.13683963e+00
-5.59120297e-01 -2.16408938e-01 4.81198639e-01 3.85294974e-01
-2.52081484e-01 -2.07255769e+00 -3.58214587e-01 3.41130525e-01
-4.22469050e-01 -1.66245252e-01 -5.43921769e-01 8.11897993e-01
4.30938423e-01 6.52453661e-01 -1.53297976e-01 -4.14945811e-01
8.00608158e-01 2.78552681e-01 2.90243387e-01 -9.86414492e-01
-1.13424003e+00 -2.06707597e-01 1.03524812e-01 -6.25438333e-01
-4.08524334e-01 -7.06204474e-01 -8.03043246e-01 -3.29881549e-01
-2.35273123e-01 2.77091175e-01 5.91833532e-01 1.09540415e+00
1.55708611e-01 4.76826876e-01 4.19422150e-01 -7.39685774e-01
-6.78227842e-01 -9.80646789e-01 -3.88391465e-01 8.70897770e-01
1.30213529e-01 -3.99906576e-01 -2.32655242e-01 6.91761088e-04] | [10.260762214660645, 9.85561752319336] |
dc1dc959-6bec-4f97-ba5b-e2b090ff12bf | coherent-online-video-style-transfer | 1703.09211 | null | http://arxiv.org/abs/1703.09211v2 | http://arxiv.org/pdf/1703.09211v2.pdf | Coherent Online Video Style Transfer | Training a feed-forward network for fast neural style transfer of images is
proven to be successful. However, the naive extension to process video frame by
frame is prone to producing flickering results. We propose the first end-to-end
network for online video style transfer, which generates temporally coherent
stylized video sequences in near real-time. Two key ideas include an efficient
network by incorporating short-term coherence, and propagating short-term
coherence to long-term, which ensures the consistency over larger period of
time. Our network can incorporate different image stylization networks. We show
that the proposed method clearly outperforms the per-frame baseline both
qualitatively and quantitatively. Moreover, it can achieve visually comparable
coherence to optimization-based video style transfer, but is three orders of
magnitudes faster in runtime. | ['Dongdong Chen', 'Jing Liao', 'Nenghai Yu', 'Lu Yuan', 'Gang Hua'] | 2017-03-27 | coherent-online-video-style-transfer-1 | http://openaccess.thecvf.com/content_iccv_2017/html/Chen_Coherent_Online_Video_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Chen_Coherent_Online_Video_ICCV_2017_paper.pdf | iccv-2017-10 | ['video-style-transfer', 'image-stylization'] | ['computer-vision', 'computer-vision'] | [ 1.98700443e-01 -1.06331825e-01 5.54477647e-02 -3.14932406e-01
-7.57707953e-01 -6.46706939e-01 7.33609140e-01 -6.75984740e-01
-3.62102509e-01 9.71427321e-01 4.95712578e-01 -2.16081947e-01
4.98510987e-01 -6.27970815e-01 -1.14165139e+00 -4.90055799e-01
5.20205013e-02 4.24257368e-02 2.98484623e-01 -9.27066654e-02
8.34365860e-02 3.84760052e-01 -1.05265355e+00 4.93894011e-01
4.67542082e-01 8.68997395e-01 2.23907620e-01 9.91733372e-01
-1.38921022e-01 1.11279488e+00 -6.00452304e-01 -2.37827107e-01
3.25126797e-01 -8.63611579e-01 -9.57560837e-01 2.44479135e-01
8.36068928e-01 -9.16252732e-01 -7.14760780e-01 8.44233155e-01
5.68265438e-01 3.10834497e-01 5.47540128e-01 -1.07980371e+00
-9.78655636e-01 3.95267367e-01 -5.55247426e-01 1.30115867e-01
3.76502931e-01 4.45782185e-01 6.84024394e-01 -8.59065652e-01
8.72062325e-01 1.51543343e+00 4.96428877e-01 8.31125855e-01
-1.20679629e+00 -7.66389549e-01 1.06071793e-01 7.59975761e-02
-9.83175099e-01 -6.01806819e-01 7.63076484e-01 -2.77787536e-01
8.02769303e-01 1.55111700e-02 7.22912431e-01 1.35218644e+00
2.59792835e-01 8.90000165e-01 1.09293580e+00 -2.70914644e-01
-9.39458702e-03 -2.58168668e-01 -3.33425522e-01 5.29150009e-01
-2.58298784e-01 4.28727001e-01 -7.66234756e-01 2.46012539e-01
1.31758916e+00 -1.13087192e-01 -4.19459641e-01 -3.64717320e-02
-1.22628355e+00 6.71061039e-01 4.43696350e-01 9.02854651e-02
-2.82684743e-01 7.22205281e-01 6.44538701e-01 6.33785367e-01
6.39940560e-01 3.80133301e-01 -1.54036552e-01 -4.04318333e-01
-1.29459035e+00 2.77871430e-01 5.15757024e-01 1.16903520e+00
6.57169223e-01 5.51179409e-01 -4.54853624e-01 6.05687261e-01
3.75557654e-02 4.68080133e-01 2.83402115e-01 -1.70120311e+00
3.42131376e-01 -3.57299149e-01 3.14680964e-01 -7.88736165e-01
7.17628747e-02 -2.03894541e-01 -1.00572371e+00 4.56404209e-01
5.14206767e-01 -5.06843805e-01 -7.95021415e-01 1.91270411e+00
-1.04968548e-01 4.90544617e-01 -2.28404239e-01 1.06680095e+00
7.28419602e-01 9.72850442e-01 4.04241588e-03 -2.58540809e-01
9.82084155e-01 -1.22734427e+00 -8.95716310e-01 -5.63055985e-02
1.69640511e-01 -8.85702908e-01 1.24248219e+00 9.94478166e-02
-1.51551282e+00 -8.02182198e-01 -9.39429462e-01 -2.16050580e-01
2.37422541e-01 -1.75324053e-01 4.03098583e-01 1.63762122e-01
-1.56920743e+00 8.94793987e-01 -6.48108065e-01 -4.23908144e-01
5.69205880e-01 1.84857816e-01 -3.21855158e-01 1.39319226e-01
-1.10753930e+00 5.91119111e-01 3.28611284e-01 -1.20462179e-01
-9.17403102e-01 -9.28374410e-01 -8.81633759e-01 8.05544630e-02
2.17063218e-01 -1.04950845e+00 1.38244939e+00 -1.68587947e+00
-2.18310213e+00 7.37695217e-01 -5.18717945e-01 -5.43305993e-01
8.09736848e-01 -5.98597109e-01 -1.93418771e-01 3.26969564e-01
-7.56746754e-02 1.27243841e+00 1.02346385e+00 -1.24992120e+00
-5.13857841e-01 3.30182076e-01 1.56997174e-01 3.70175451e-01
-4.31018054e-01 2.25434154e-01 -6.40656888e-01 -1.07511234e+00
-5.31790376e-01 -9.06062484e-01 5.55081479e-02 4.48499918e-01
-1.03470460e-01 8.76062363e-02 1.07397389e+00 -8.47089946e-01
9.83776450e-01 -2.03992391e+00 1.81603506e-01 -3.92758876e-01
2.81007767e-01 3.54882002e-01 -5.05421579e-01 3.41390431e-01
-1.77045330e-01 4.81163934e-02 1.55871082e-02 -4.27296549e-01
-2.78261960e-01 1.05196282e-01 -6.49547458e-01 1.96102828e-01
3.45447242e-01 1.32375753e+00 -1.03520656e+00 -3.91663194e-01
2.91766435e-01 4.19790447e-01 -7.15883970e-01 4.58698273e-01
-2.87598193e-01 6.24471068e-01 -2.33758930e-02 2.18529463e-01
6.13660276e-01 -3.50926459e-01 -9.74059105e-02 -3.98770332e-01
9.70855169e-03 1.94346756e-01 -7.23010421e-01 1.84309292e+00
-5.59065342e-01 1.35376143e+00 -5.37579879e-02 -5.85311234e-01
5.16223967e-01 4.85510051e-01 3.70093703e-01 -5.21040738e-01
1.63081497e-01 -2.09292509e-02 -4.51513052e-01 -2.82012165e-01
7.06342101e-01 -2.48764426e-01 2.87873864e-01 5.32245159e-01
3.52362156e-01 -1.51233405e-01 1.78515151e-01 2.77058423e-01
8.81658196e-01 6.11400187e-01 5.02019748e-02 -2.71609753e-01
3.02953660e-01 -4.34485137e-01 4.53765094e-01 5.58636069e-01
-3.03701699e-01 9.77800727e-01 4.57046330e-01 -5.60780883e-01
-1.36166549e+00 -1.18758333e+00 5.11439323e-01 1.01788902e+00
1.66693091e-01 -2.89822102e-01 -8.39327991e-01 -6.89025283e-01
-3.29549551e-01 6.41387343e-01 -4.91142273e-01 -3.03283776e-03
-8.19504857e-01 2.40275194e-03 5.95493972e-01 7.25392640e-01
7.83778489e-01 -1.31705892e+00 -3.39713067e-01 2.98748553e-01
-3.37675631e-01 -1.22624171e+00 -1.19980705e+00 -3.16451788e-01
-8.88117671e-01 -5.81364810e-01 -1.32455993e+00 -8.71953785e-01
4.15491581e-01 5.00784099e-01 1.46544385e+00 3.57591026e-02
2.61542499e-01 1.86022699e-01 -6.96162805e-02 -2.03243308e-02
-5.53971946e-01 -1.28264025e-01 -7.88612217e-02 -1.05637468e-01
-5.64064905e-02 -7.49909580e-01 -7.50323832e-01 2.54789740e-01
-1.08435917e+00 3.96187067e-01 2.49411851e-01 1.12090039e+00
2.83540606e-01 -4.40510094e-01 4.93186951e-01 -6.83629751e-01
6.55698299e-01 1.06724836e-02 -4.48112488e-01 1.71738118e-02
-1.72064587e-01 1.24236263e-01 8.03180516e-01 -4.34068859e-01
-1.27329767e+00 -1.19133666e-01 -1.53453648e-01 -9.10692513e-01
-1.67091683e-01 1.07992865e-01 1.85166463e-01 -1.16337195e-01
7.27296174e-01 1.41410589e-01 9.91513282e-02 -2.00384229e-01
7.53620803e-01 2.07727194e-01 8.32904041e-01 -5.53491116e-01
9.62747693e-01 5.54126024e-01 -1.54474646e-01 -6.47627890e-01
-8.05062294e-01 -9.13892388e-02 -4.93721187e-01 -4.71022040e-01
9.69551086e-01 -1.08630621e+00 -4.65201467e-01 6.84139609e-01
-1.52218115e+00 -8.79226506e-01 -3.91684055e-01 2.98786044e-01
-8.52040887e-01 5.58112860e-01 -1.11207700e+00 -2.12517738e-01
-5.21638215e-01 -1.01135314e+00 1.04572034e+00 2.91186035e-01
-3.53769124e-01 -1.05698752e+00 1.13138985e-02 -8.09932202e-02
5.96226811e-01 1.58513948e-01 4.26469475e-01 1.39839560e-01
-7.25717485e-01 2.90546209e-01 -5.93164980e-01 4.70324814e-01
3.20364207e-01 1.41344190e-01 -9.72594261e-01 -4.93003279e-01
-2.22289041e-01 -4.88783062e-01 9.55449402e-01 6.25124276e-01
1.22693264e+00 -3.75778139e-01 8.63175020e-02 9.15691972e-01
1.38509858e+00 1.48772150e-01 9.00567234e-01 7.20542222e-02
9.42888081e-01 3.67309690e-01 3.96779329e-01 5.71173914e-02
7.93186799e-02 7.02168465e-01 3.25049832e-02 -4.46674168e-01
-6.78331196e-01 -1.48733914e-01 7.13259816e-01 7.59157658e-01
-3.32028806e-01 -4.69080210e-01 -3.48958492e-01 4.94101048e-01
-1.79469645e+00 -1.31813359e+00 2.87538707e-01 1.94584250e+00
9.29213643e-01 2.44385526e-01 4.22492683e-01 -3.68909776e-01
6.82746530e-01 3.78482103e-01 -4.69656050e-01 -4.71195757e-01
-2.28569835e-01 1.38983175e-01 5.53181827e-01 6.11891925e-01
-9.96290982e-01 1.32650912e+00 7.19286346e+00 1.01879239e+00
-1.44013667e+00 1.39255092e-01 9.29271340e-01 -3.69333982e-01
-3.72857332e-01 -2.00789332e-01 -2.88300902e-01 5.59063077e-01
8.47647488e-01 -2.31887966e-01 7.36944795e-01 5.33768952e-01
5.20029724e-01 9.62862745e-02 -1.23851538e+00 1.07101226e+00
7.13006705e-02 -1.84567606e+00 4.02245820e-01 -1.89555407e-01
1.16212749e+00 -8.53559561e-03 6.86529502e-02 5.99886291e-02
4.22771454e-01 -1.00573075e+00 9.76079345e-01 5.27491689e-01
1.26988983e+00 -8.18129539e-01 4.48268801e-01 -1.26664728e-01
-1.21291578e+00 2.72793591e-01 -2.10438281e-01 -1.22964859e-01
6.42446637e-01 3.90745699e-01 -2.04868987e-01 4.38894928e-01
6.73315823e-01 9.88452911e-01 -3.38377237e-01 8.01909804e-01
-1.51480630e-01 7.78500557e-01 8.65150522e-03 2.18183726e-01
4.68051821e-01 -3.53060633e-01 5.18673778e-01 1.42968106e+00
4.88219440e-01 -4.80746105e-02 -9.61113423e-02 9.73088741e-01
-3.91696095e-01 -3.55439365e-01 -8.83148789e-01 -3.75603437e-02
2.81240284e-01 1.04734111e+00 -6.61859810e-01 -6.95707560e-01
-5.04188657e-01 1.59198284e+00 2.72315174e-01 7.46337056e-01
-1.04691291e+00 -4.05129433e-01 6.32267356e-01 -4.48822528e-02
3.03520411e-01 -3.76970619e-01 -3.65215868e-01 -1.17651594e+00
3.91063467e-02 -9.79369760e-01 -1.93371661e-02 -1.10578954e+00
-1.31523669e+00 7.22249329e-01 -1.49920061e-01 -1.24850500e+00
-5.66999376e-01 -5.40851295e-01 -8.23678017e-01 8.13513875e-01
-1.37124765e+00 -1.16880608e+00 -4.89647388e-01 5.88860393e-01
9.33198154e-01 -1.38592690e-01 5.35708487e-01 3.45065445e-01
-1.03690073e-01 7.51189530e-01 1.19232729e-01 1.74667895e-01
1.24287534e+00 -1.20978475e+00 7.24526525e-01 1.00402522e+00
3.27814817e-02 3.09900522e-01 8.25684905e-01 -5.00045478e-01
-1.09382749e+00 -1.30987573e+00 6.95029259e-01 -2.98022807e-01
5.32777488e-01 -3.20744187e-01 -8.80854309e-01 7.31735587e-01
9.48890090e-01 -7.11871386e-02 1.10998347e-01 -3.05518895e-01
-2.62889147e-01 -1.22534417e-01 -7.78687418e-01 1.06748235e+00
1.27100968e+00 -7.43970215e-01 -3.07675391e-01 2.81289101e-01
1.07155871e+00 -4.50327724e-01 -6.10780120e-01 9.74006802e-02
5.44632018e-01 -1.08555520e+00 9.34598327e-01 -5.22327900e-01
9.63450432e-01 -2.59163648e-01 1.69571117e-01 -1.44719708e+00
-4.96345609e-01 -1.22113156e+00 -1.00682437e-01 1.31330538e+00
1.89158365e-01 -3.44409466e-01 7.47944117e-01 3.52162331e-01
-1.07616425e-01 -4.47857201e-01 -6.39419019e-01 -9.59330380e-01
3.04167986e-01 -1.16652369e-01 2.47512668e-01 7.28160679e-01
-3.94491017e-01 4.10896391e-01 -9.79925334e-01 -3.20762306e-01
5.14094651e-01 9.29909013e-03 8.49101841e-01 -5.89083672e-01
-5.19523025e-01 -5.73621929e-01 -1.38871998e-01 -1.54256463e+00
2.05187947e-01 -4.29217011e-01 1.17804788e-01 -1.24701905e+00
1.98707476e-01 5.06441072e-02 -1.05679594e-01 1.20161012e-01
-3.18947554e-01 6.39215171e-01 4.17762846e-01 9.04338360e-02
-6.05087161e-01 6.75967395e-01 1.62809062e+00 -4.31207418e-02
-1.03118561e-01 -3.58165830e-01 -3.87092739e-01 6.87443018e-01
6.92395389e-01 -1.70377448e-01 -4.76030976e-01 -7.08581269e-01
-5.00938520e-02 1.70377851e-01 3.44129533e-01 -9.00199115e-01
-1.10328391e-01 -2.56684124e-01 4.43289936e-01 -5.09046793e-01
3.26248407e-01 -3.68785203e-01 2.15224788e-01 2.40531653e-01
-5.61281502e-01 1.91407129e-01 3.85238677e-01 5.93775928e-01
-3.72138977e-01 1.00724205e-01 1.08307958e+00 -1.00611582e-01
-6.68364406e-01 5.04226685e-01 -4.85668749e-01 2.26481229e-01
7.31656134e-01 -1.97005630e-01 -1.28074914e-01 -1.04209542e+00
-3.64228547e-01 -6.68566814e-03 6.96171463e-01 4.71957445e-01
4.39317703e-01 -1.72222674e+00 -7.82915473e-01 -5.77557571e-02
-3.99880469e-01 -2.43631080e-01 2.19161913e-01 5.07924139e-01
-8.57748866e-01 2.63975590e-01 -4.75091785e-01 -6.66118920e-01
-1.23057950e+00 5.73943734e-01 3.43005955e-01 -1.89325124e-01
-6.90833926e-01 9.30552781e-01 6.39484406e-01 3.92416008e-02
2.34569222e-01 -1.53356045e-01 1.75410077e-01 -2.72290766e-01
6.24176502e-01 2.48951226e-01 -3.90657127e-01 -4.64948177e-01
5.13215363e-02 7.69028008e-01 3.37434188e-02 -5.17160535e-01
1.19075370e+00 -3.15838903e-01 1.14167091e-02 3.83013368e-01
1.32763219e+00 -1.86651558e-01 -2.06259704e+00 -2.66549647e-01
-4.37380254e-01 -5.21338344e-01 -5.93558475e-02 -5.17572105e-01
-1.32494617e+00 8.85279834e-01 4.67669785e-01 -1.48007199e-01
1.21118331e+00 -3.86285305e-01 1.07388687e+00 3.64491791e-01
2.14001805e-01 -1.06312883e+00 5.15070915e-01 6.20974660e-01
1.08742034e+00 -1.19989872e+00 -1.77039772e-01 -2.97886550e-01
-5.77477396e-01 1.28364778e+00 5.80933750e-01 -4.88119513e-01
3.77239376e-01 3.03884715e-01 2.48692900e-01 2.64470279e-01
-7.79552162e-01 2.24521682e-01 4.17594910e-01 6.08449996e-01
6.09329402e-01 -2.34461635e-01 -6.56504408e-02 -1.26654655e-01
-5.11419773e-02 3.30153555e-01 4.78598773e-01 6.16258979e-01
-1.75210074e-01 -1.03606248e+00 -2.39463896e-01 1.94970697e-01
-5.64543843e-01 -2.29758665e-01 -2.13674977e-01 6.68037415e-01
-2.71949351e-01 8.28313172e-01 3.22134793e-01 -2.06580117e-01
-2.38613468e-02 -2.30428856e-02 7.44716585e-01 -1.82233110e-01
-4.94561493e-01 3.98495615e-01 2.13217020e-01 -8.68231416e-01
-6.32232070e-01 -4.10935819e-01 -9.43071961e-01 -7.98239768e-01
-2.79896636e-03 -3.27569276e-01 1.08834743e-01 9.75459933e-01
4.70741421e-01 8.23112130e-01 7.27737367e-01 -1.45685422e+00
-1.79330721e-01 -9.42594230e-01 -2.70288408e-01 6.68320417e-01
4.20786381e-01 -3.15083176e-01 -1.89962775e-01 6.66151166e-01] | [10.997218132019043, -0.8071110844612122] |
963b909e-3655-4b4a-82ed-8fb1b4fcd319 | quantum-contextual-bandits-and-recommender | 2301.13524 | null | https://arxiv.org/abs/2301.13524v1 | https://arxiv.org/pdf/2301.13524v1.pdf | Quantum contextual bandits and recommender systems for quantum data | We study a recommender system for quantum data using the linear contextual bandit framework. In each round, a learner receives an observable (the context) and has to recommend from a finite set of unknown quantum states (the actions) which one to measure. The learner has the goal of maximizing the reward in each round, that is the outcome of the measurement on the unknown state. Using this model we formulate the low energy quantum state recommendation problem where the context is a Hamiltonian and the goal is to recommend the state with the lowest energy. For this task, we study two families of contexts: the Ising model and a generalized cluster model. We observe that if we interpret the actions as different phases of the models then the recommendation is done by classifying the correct phase of the given Hamiltonian and the strategy can be interpreted as an online quantum phase classifier. | ['Marco Tomamichel', 'Josep Lumbreras', 'Shrigyan Brahmachari'] | 2023-01-31 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [ 2.92077780e-01 2.83990979e-01 -4.53476816e-01 -2.33841464e-01
-5.26196718e-01 -4.15786743e-01 6.47031426e-01 1.99666858e-01
-4.63962793e-01 5.05472183e-01 2.72554696e-01 -4.17672634e-01
-5.39193392e-01 -1.22734523e+00 -7.74201274e-01 -1.29000962e+00
3.84291977e-01 7.57491708e-01 -1.44700661e-01 -2.56660372e-01
7.51826465e-01 1.39698377e-02 -1.29939699e+00 2.49238878e-01
5.55501044e-01 8.61428022e-01 1.84173062e-01 8.04861963e-01
1.00082532e-01 7.02336729e-01 2.75299437e-02 -1.51253626e-01
3.25612068e-01 -8.47894311e-01 -1.14579952e+00 -2.91826189e-01
2.57870425e-02 2.41727218e-01 -6.68401897e-01 1.47976661e+00
1.98985562e-01 6.93583429e-01 7.77374566e-01 -6.24204218e-01
-5.03858924e-01 9.42741692e-01 2.99558669e-01 2.27822870e-01
3.33133727e-01 -2.27095708e-01 1.78223550e+00 -3.60836610e-02
6.40812039e-01 8.00558090e-01 -8.12626332e-02 7.05497861e-01
-1.63131607e+00 -2.42287487e-01 -1.59964412e-01 7.17808545e-01
-1.05363774e+00 -3.06145042e-01 5.18011868e-01 -3.43386889e-01
5.02332270e-01 1.77347258e-01 6.48630023e-01 7.89275229e-01
3.04806888e-01 4.21130806e-01 1.37315166e+00 -7.25470960e-01
1.09141374e+00 1.58021659e-01 6.73430920e-01 7.09544539e-01
-6.59964383e-02 5.80578864e-01 -8.12199295e-01 -9.22551081e-02
-1.00417241e-01 1.46177346e-02 -2.62698084e-01 -4.40289497e-01
-9.29209828e-01 1.00353038e+00 5.99155664e-01 2.55301446e-01
-5.19958556e-01 2.43829906e-01 -2.79762357e-01 4.77415919e-01
2.18876228e-01 4.72582459e-01 -9.76068974e-02 -6.61028549e-02
-4.67406064e-01 7.34703168e-02 1.06987727e+00 1.92844748e-01
1.21847868e+00 -8.57506156e-01 -6.74477220e-02 7.61083812e-02
4.71553028e-01 6.91873193e-01 1.49561062e-01 -9.96555805e-01
1.37684092e-01 6.35943189e-02 4.77837950e-01 -3.71055841e-01
-2.57177442e-01 -1.39609456e-01 -3.32761347e-01 -7.48369470e-02
5.39423883e-01 -3.48051675e-02 -6.58680737e-01 1.78080964e+00
2.43434563e-01 5.80791354e-01 2.23029196e-01 8.54901850e-01
3.88271689e-01 8.00524533e-01 -4.74541783e-02 -4.46648777e-01
9.85245287e-01 -7.47826397e-01 -3.02447945e-01 -1.42628089e-01
9.21859443e-01 -3.53338927e-01 5.17602086e-01 5.69123864e-01
-7.22071826e-01 -2.10813090e-01 -1.12672889e+00 3.96662503e-01
-1.77659497e-01 -2.99628615e-01 4.78141516e-01 7.35261202e-01
-9.18754458e-01 1.37618041e+00 -7.32803226e-01 -2.99704611e-01
-2.40945980e-01 5.39977014e-01 -9.73799452e-02 -1.49632141e-01
-1.09476316e+00 8.95071387e-01 3.69280815e-01 -6.00120388e-02
-8.26199174e-01 1.21047266e-01 -1.22157201e-01 2.06874624e-01
4.81018305e-01 -6.35245621e-01 1.21496749e+00 -1.06209254e+00
-1.67665696e+00 6.29823685e-01 -1.85003594e-01 -5.15947342e-01
-2.00060412e-01 2.83730358e-01 -9.04091746e-02 1.38300166e-01
-4.47537825e-02 -6.76212534e-02 8.70879233e-01 -4.69784200e-01
-7.28422284e-01 -5.22689521e-01 5.93847394e-01 3.04944307e-01
2.31493786e-01 -4.50570941e-01 4.65434678e-02 3.62797827e-01
6.23134732e-01 -1.81654859e+00 -9.87545922e-02 -9.67253029e-01
-2.64857292e-01 -3.97282541e-01 1.00813806e-01 -1.02509126e-01
9.80836749e-01 -1.94630468e+00 7.05130100e-01 6.05019093e-01
-6.36504591e-02 -4.01828408e-01 -1.47749797e-01 4.90448773e-01
7.27947876e-02 4.93577905e-02 1.22372471e-01 -1.14930212e-01
2.94969469e-01 3.37499142e-01 -3.16078246e-01 6.86834693e-01
-5.02091587e-01 4.77055013e-01 -7.92221785e-01 1.78026839e-03
-3.14851291e-02 -1.40943602e-01 -9.86042142e-01 1.80608317e-01
-5.26410043e-01 5.62389314e-01 -8.00215244e-01 -3.59493911e-01
2.41464034e-01 -4.24113065e-01 5.12697935e-01 1.37808938e-02
-1.01559490e-01 1.01246393e+00 -1.54568613e+00 1.40001047e+00
-2.86714941e-01 3.11263442e-01 -4.02224511e-01 -1.27944720e+00
3.42308909e-01 2.07825556e-01 4.31026608e-01 -7.17640579e-01
1.69643149e-01 6.00967743e-02 3.17344010e-01 -4.70609903e-01
4.73817587e-01 -4.42167044e-01 -2.10775539e-01 9.28741634e-01
2.60284573e-01 2.43778765e-01 8.69370103e-02 4.08212095e-01
1.22166872e+00 3.46576869e-02 1.32802606e-01 -3.72769952e-01
4.34533298e-01 -4.35175240e-01 3.39233369e-01 1.22700548e+00
-1.87485572e-02 -3.66016962e-02 6.24670744e-01 -1.89843103e-01
-7.77946770e-01 -8.03795516e-01 -2.72103138e-02 1.14904583e+00
4.34856683e-01 -4.21746731e-01 -5.72445512e-01 -5.56945860e-01
-2.05874190e-01 7.24231184e-01 -9.42549050e-01 -4.83513951e-01
-1.05375066e-01 -4.49132591e-01 -4.49025631e-01 -3.16423386e-01
2.01296464e-01 -9.86168921e-01 -3.12706351e-01 2.13143099e-02
-2.41098359e-01 -6.08210266e-01 -5.07111907e-01 6.43311024e-01
-7.79333651e-01 -9.33312178e-01 1.98454887e-01 -3.82377177e-01
1.81289196e-01 5.78573234e-02 7.78826177e-01 -1.64959326e-01
3.80714267e-01 3.75546545e-01 -5.14845610e-01 1.52736321e-01
-6.79638565e-01 2.61888448e-02 3.54403585e-01 5.18379450e-01
3.33089560e-01 -5.82381822e-02 -7.27909505e-01 -1.57690123e-01
-5.98531902e-01 8.61606970e-02 2.33112872e-01 8.80743027e-01
7.39345133e-01 1.30385503e-01 6.41811937e-02 -1.25172484e+00
2.42381886e-01 -5.30909300e-01 -6.07626319e-01 4.66640532e-01
-9.10550356e-01 8.46111000e-01 6.37273967e-01 -1.46814615e-01
-9.11747098e-01 -8.05030912e-02 2.65644997e-01 2.93839455e-01
-2.67955214e-01 7.16829360e-01 -6.33798772e-03 -1.77075356e-01
7.24480331e-01 2.17970416e-01 -4.21887398e-01 -2.99792290e-01
5.35048425e-01 6.43029571e-01 2.20607854e-02 -7.35258579e-01
4.74684089e-01 1.52834967e-01 5.31154573e-01 -7.54624307e-01
-1.34029901e+00 -4.49569970e-01 -6.54866397e-01 -1.03961162e-01
9.25668359e-01 -3.90221626e-01 -1.02070212e+00 -7.52173811e-02
-5.30886829e-01 -4.12550271e-01 -4.41712916e-01 8.42491627e-01
-7.65619636e-01 2.66599059e-01 -5.67766964e-01 -1.22592485e+00
-3.96712348e-02 -1.16206157e+00 4.81248289e-01 4.96217251e-01
3.14795852e-01 -9.06888545e-01 3.98995340e-01 4.85680073e-01
8.62755850e-02 -5.07955849e-01 1.04838586e+00 -6.74275696e-01
-9.63189244e-01 -1.64674968e-01 3.17731440e-01 2.06719227e-02
-3.22902769e-01 -2.61450976e-01 -8.77061486e-01 -1.57525182e-01
5.93885519e-02 -2.64380366e-01 8.78903091e-01 4.39256608e-01
6.61296427e-01 -2.97607958e-01 -1.74390912e-01 2.67600924e-01
1.19113064e+00 4.59384292e-01 4.66940165e-01 1.29686609e-01
3.37487996e-01 2.23661393e-01 5.14713466e-01 3.42406891e-02
2.25300401e-01 8.93133521e-01 1.50502682e-01 8.88636887e-01
6.11169279e-01 -3.71082515e-01 4.95556295e-01 7.43798494e-01
-1.99300423e-01 -8.20720345e-02 -5.12398779e-01 -1.25331074e-01
-1.95442510e+00 -1.32541895e+00 2.07047716e-01 2.63863993e+00
6.57746434e-01 4.01306339e-02 1.47780314e-01 -1.20264187e-01
4.79725301e-01 3.47316340e-02 -5.30503690e-01 -3.75149161e-01
3.48260581e-01 5.25824428e-01 4.05684650e-01 7.83086717e-01
-9.88017917e-01 6.79881573e-01 4.95986605e+00 5.81656575e-01
-1.18377817e+00 1.65516198e-01 2.62262583e-01 3.79240811e-02
-2.52003044e-01 7.87415624e-01 -8.31975758e-01 6.44113839e-01
1.55629957e+00 -1.21134885e-01 1.24894226e+00 2.53451139e-01
3.65566798e-02 -5.18415570e-01 -1.36388135e+00 6.46212220e-01
-2.82934159e-01 -1.36560023e+00 -3.88884336e-01 6.13850594e-01
8.26882958e-01 4.34368402e-01 4.04262617e-02 3.87979388e-01
3.13591748e-01 -5.90620160e-01 7.93099999e-01 9.47938442e-01
2.82312423e-01 -6.41213179e-01 4.90691990e-01 8.92162561e-01
-5.76047361e-01 -2.35635370e-01 -3.21760654e-01 -2.93908656e-01
-3.67560297e-01 3.18664946e-02 -5.15096486e-01 4.59646910e-01
1.88863829e-01 7.03954577e-01 7.28924572e-03 7.60293245e-01
-2.61403292e-01 1.00907445e+00 -3.51482630e-01 -5.71578503e-01
1.35389462e-01 -1.34678674e+00 3.94531786e-01 3.97097409e-01
3.67655396e-01 6.13076389e-01 2.49300808e-01 6.34433091e-01
-1.92920208e-01 1.48579270e-01 -4.63102460e-01 -2.93601662e-01
2.20296159e-01 1.01120579e+00 -6.89730346e-01 -1.19521096e-01
-3.65139782e-01 8.86667430e-01 4.59071338e-01 2.22661808e-01
-2.51892149e-01 2.20640495e-01 3.32286835e-01 -3.05289596e-01
3.04690778e-01 2.09432423e-01 6.16184808e-02 -1.44180405e+00
-6.52337134e-01 -3.34150821e-01 6.05205417e-01 -4.55973625e-01
-1.04196668e+00 -2.09345724e-02 -2.98418432e-01 -8.57875705e-01
-3.83140855e-02 -4.24987793e-01 -6.37750983e-01 5.85743070e-01
-1.18799067e+00 -3.67094457e-01 4.37136531e-01 4.45676565e-01
-2.95589358e-01 4.24431674e-02 1.16896224e+00 -9.95131880e-02
-5.82612991e-01 5.80037571e-02 7.70886779e-01 -3.50602686e-01
4.01906103e-01 -1.59172702e+00 -3.19525003e-01 5.74615121e-01
8.35378468e-01 7.05541551e-01 1.01764369e+00 -2.90982425e-01
-1.74627221e+00 -6.36488676e-01 8.48631263e-01 -4.24931705e-01
7.62002945e-01 -2.15901248e-02 -4.04634088e-01 4.90804881e-01
-9.64518413e-02 2.79240329e-02 7.82096207e-01 5.93233645e-01
-1.20950758e-01 -1.60145193e-01 -9.56036568e-01 2.13559896e-01
5.65260530e-01 -9.31183934e-01 -3.36121410e-01 5.88978469e-01
3.60654593e-01 -1.29759669e-01 -6.78954601e-01 -4.15516496e-01
4.76006716e-01 -8.74484956e-01 3.80582839e-01 -1.03516984e+00
3.72373432e-01 -2.21405745e-01 -4.53422546e-01 -1.48058414e+00
-6.02848828e-01 -5.20589650e-01 -1.81583717e-01 3.71774822e-01
4.89619523e-01 -5.28278589e-01 1.01690733e+00 6.96855485e-01
3.34307551e-01 -5.02345145e-01 -1.21786630e+00 -1.11701548e-01
2.61449777e-02 -3.92666757e-01 2.83017695e-01 4.90383953e-01
4.84606743e-01 1.12904227e+00 -4.37282801e-01 3.32457781e-01
7.52884150e-01 8.37738872e-01 3.06874126e-01 -1.20868647e+00
-8.91224742e-01 -1.09411798e-01 -2.47289225e-01 -1.20468056e+00
2.53578693e-01 -1.36049640e+00 3.00216109e-01 -1.07950759e+00
5.32512069e-01 -3.80093932e-01 -7.78886795e-01 -3.05659920e-02
-1.02951387e-02 -1.24418765e-01 2.52733171e-01 3.14588994e-01
-1.12911355e+00 1.83042929e-01 8.60705853e-01 -9.62780565e-02
-1.64385185e-01 7.28986382e-01 -6.00338161e-01 2.51247138e-01
4.64411288e-01 -7.77356207e-01 -6.28347397e-02 1.69675350e-01
9.76439178e-01 7.57763505e-01 4.21980508e-02 -6.26278341e-01
5.09940207e-01 -4.03978109e-01 -2.59124696e-01 -2.09604189e-01
2.66331524e-01 -6.61110580e-01 2.25325435e-01 6.47576034e-01
-9.54514384e-01 -7.84575880e-01 -8.18540573e-01 1.05666184e+00
4.26026553e-01 -7.67511547e-01 8.19751680e-01 -2.34008883e-03
-3.54954064e-01 4.45345759e-01 -2.83120066e-01 -1.75305396e-01
6.76284671e-01 5.82569063e-01 -2.04863667e-01 -5.85314751e-01
-1.28664660e+00 -1.22471295e-01 4.18772787e-01 -8.64888802e-02
1.30743235e-01 -1.06408489e+00 -2.76295751e-01 1.59826260e-02
1.05665825e-01 -7.31063187e-01 3.36778462e-01 8.54146600e-01
-3.17425914e-02 6.40164554e-01 3.35494757e-01 -3.20050389e-01
-9.27487373e-01 5.55342853e-01 6.65033877e-01 -2.43537456e-01
-1.66361019e-01 4.82205063e-01 1.50189484e-02 -8.61558691e-02
-1.41395971e-01 -1.31582737e-01 -3.96315485e-01 6.61255792e-02
2.81452686e-01 3.50531042e-01 1.87216014e-01 -8.29227209e-01
-4.85982336e-02 2.01191172e-01 2.77606882e-02 -3.37706476e-01
1.29681301e+00 -1.07074738e-01 -1.51154041e-01 6.80788815e-01
1.05485082e+00 4.07824628e-02 -9.69685614e-01 -6.67562306e-01
5.29129244e-02 -2.17088297e-01 5.14461875e-01 -5.58719158e-01
-5.92842996e-01 5.81458807e-01 9.48007941e-01 8.63502979e-01
4.46969420e-01 3.32650274e-01 4.09312069e-01 1.03283823e+00
7.10806012e-01 -1.25577605e+00 -1.54271379e-01 8.19032550e-01
2.39623692e-02 -1.13501823e+00 -2.88883954e-01 4.78061169e-01
-4.07011688e-01 1.11438811e+00 -2.19970301e-01 -2.64811188e-01
9.89738822e-01 -4.96904790e-01 -4.82469171e-01 -3.83566588e-01
-8.66766512e-01 -4.73702699e-01 5.30471444e-01 -5.82888313e-02
1.82658792e-01 7.02011526e-01 -3.31094354e-01 3.67346853e-01
-3.36629629e-01 -9.78599414e-02 7.00869322e-01 3.93453717e-01
-8.31148028e-01 -1.30842078e+00 -1.02845654e-01 8.21961701e-01
-4.22783613e-01 8.07598978e-02 -1.82890847e-01 -3.14681679e-01
2.78756730e-02 1.17130148e+00 -2.41174221e-01 -5.87772608e-01
1.76712409e-01 2.99394965e-01 7.08137751e-01 -1.07007802e+00
-1.96076781e-01 -6.71690851e-02 -9.40570794e-03 -4.22193736e-01
-4.85893428e-01 -8.50744247e-01 -1.09814417e+00 -4.43450212e-01
-7.58334517e-01 8.19831312e-01 7.74462998e-01 1.49060297e+00
6.09811880e-02 5.97993582e-02 9.56386805e-01 -6.77869976e-01
-8.62750947e-01 -8.12220514e-01 -9.93455887e-01 5.88672757e-01
3.80516052e-01 -5.40813327e-01 -5.40079474e-01 -2.37187013e-01] | [5.630544185638428, 4.9417338371276855] |
0952acb7-df08-4a36-9ce5-36f9ec9a9a80 | semantic-parsing-for-text-to-3d-scene | null | null | https://aclanthology.org/W14-2404 | https://aclanthology.org/W14-2404.pdf | Semantic Parsing for Text to 3D Scene Generation | null | ['Angel Chang', 'Christopher Manning', 'Manolis Savva'] | 2014-06-01 | null | null | null | ws-2014-6 | ['scene-generation', 'text-to-3d'] | ['computer-vision', 'computer-vision'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.259825229644775, 3.746004104614258] |
ea2ac90b-3533-492f-9ec5-b9260a30ad92 | new-students-on-sesame-street-what-order | 2109.08449 | null | https://arxiv.org/abs/2109.08449v2 | https://arxiv.org/pdf/2109.08449v2.pdf | General Cross-Architecture Distillation of Pretrained Language Models into Matrix Embeddings | Large pretrained language models (PreLMs) are revolutionizing natural language processing across all benchmarks. However, their sheer size is prohibitive for small laboratories or for deployment on mobile devices. Approaches like pruning and distillation reduce the model size but typically retain the same model architecture. In contrast, we explore distilling PreLMs into a different, more efficient architecture, Continual Multiplication of Words (CMOW), which embeds each word as a matrix and uses matrix multiplication to encode sequences. We extend the CMOW architecture and its CMOW/CBOW-Hybrid variant with a bidirectional component for more expressive power, per-token representations for a general (task-agnostic) distillation during pretraining, and a two-sequence encoding scheme that facilitates downstream tasks on sentence pairs, such as sentence similarity and natural language inference. Our matrix-based bidirectional CMOW/CBOW-Hybrid model is competitive to DistilBERT on question similarity and recognizing textual entailment, but uses only half of the number of parameters and is three times faster in terms of inference speed. We match or exceed the scores of ELMo for all tasks of the GLUE benchmark except for the sentiment analysis task SST-2 and the linguistic acceptability task CoLA. However, compared to previous cross-architecture distillation approaches, we demonstrate a doubling of the scores on detecting linguistic acceptability. This shows that matrix-based embeddings can be used to distill large PreLM into competitive models and motivates further research in this direction. | ['Ansgar Scherp', 'Angelina Sonderecker', 'Henrik Ferdinand Nölscher', 'Christoph Meyer', 'Isabelle Cuber', 'Lukas Galke'] | 2021-09-17 | null | null | null | null | ['linguistic-acceptability', 'question-similarity'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.62670180e-01 2.57304639e-01 -6.09633327e-02 -6.83041334e-01
-9.77143645e-01 -7.17292249e-01 5.75005889e-01 6.74484074e-01
-1.07331204e+00 2.11205408e-01 3.82546961e-01 -9.69314039e-01
3.29551727e-01 -7.67178297e-01 -7.21405745e-01 -1.81882292e-01
1.37475237e-01 5.76295435e-01 -8.56085345e-02 -6.45561218e-01
1.52842268e-01 3.30576524e-02 -1.19102025e+00 7.08945692e-01
7.41613209e-01 1.03656423e+00 1.13949813e-01 9.16945159e-01
-4.47527885e-01 7.77651548e-01 -2.86441296e-01 -8.82234097e-01
1.40868664e-01 -2.09538162e-01 -1.07519495e+00 -6.29616559e-01
7.97980845e-01 -2.58755505e-01 -6.34685680e-02 5.82922757e-01
6.50699675e-01 3.62237781e-01 4.46797848e-01 -9.30988252e-01
-9.67455029e-01 1.31887043e+00 -4.07484472e-01 3.91003698e-01
5.54892898e-01 1.80341467e-01 1.80081749e+00 -1.27687871e+00
2.97237515e-01 1.57686830e+00 1.05513012e+00 5.09076297e-01
-1.44445479e+00 -5.29597044e-01 4.19580340e-01 3.05602372e-01
-1.13844156e+00 -4.42139387e-01 2.70152688e-01 -2.71103773e-02
1.98971772e+00 5.49108982e-01 3.29392493e-01 1.10078263e+00
2.92239726e-01 8.93438220e-01 8.93659949e-01 -5.66501737e-01
9.25482363e-02 2.00696096e-01 5.47473729e-01 7.08930790e-01
2.27017105e-01 -2.43895233e-01 -3.02997768e-01 -1.30854174e-01
-1.01626836e-01 -1.41203746e-01 1.24146037e-01 6.66444153e-02
-1.03421354e+00 1.09639168e+00 4.34269696e-01 4.53811705e-01
1.27291143e-01 2.51621217e-01 8.47078919e-01 7.78092802e-01
4.89843637e-01 1.11930823e+00 -6.77372158e-01 -2.53841549e-01
-9.73384678e-01 2.77505040e-01 9.93730366e-01 7.34076500e-01
7.29380727e-01 2.68731359e-02 -2.33091637e-01 1.00910044e+00
5.77643663e-02 3.09349447e-01 9.30676460e-01 -7.36009061e-01
7.39819944e-01 6.91249311e-01 -6.13483787e-01 -8.63597035e-01
-5.16994715e-01 -2.04262987e-01 -5.33424914e-01 -1.73873082e-01
1.69568479e-01 -3.18205833e-01 -6.55702472e-01 1.82365966e+00
1.32639855e-01 -6.27476647e-02 1.10293463e-01 4.69877303e-01
9.14119303e-01 8.05438161e-01 8.66634920e-02 3.87223303e-01
1.59101391e+00 -1.16850531e+00 -2.58384347e-01 -7.87847221e-01
1.47554815e+00 -7.75362909e-01 1.71635783e+00 3.99120390e-01
-1.39511716e+00 -6.56160593e-01 -1.13650227e+00 -8.46430361e-01
-6.39661551e-01 -1.77282050e-01 6.72553003e-01 6.65651798e-01
-1.15984094e+00 6.42051816e-01 -6.48089409e-01 -3.46794039e-01
9.47132558e-02 4.38779145e-01 -3.44076246e-01 -3.06580335e-01
-1.52967346e+00 1.37212610e+00 4.26746517e-01 3.57006788e-02
-2.72652149e-01 -1.16677785e+00 -1.42585826e+00 1.52141511e-01
6.17998764e-02 -8.51518691e-01 1.39147377e+00 -6.14701986e-01
-1.29742610e+00 1.01365912e+00 -1.74724460e-01 -7.77725995e-01
2.76869722e-02 -4.22596157e-01 -3.53874117e-01 -3.47405761e-01
-5.64104691e-02 8.18668902e-01 4.74200100e-01 -5.71336806e-01
-3.81537080e-01 -2.57130638e-02 4.34961438e-01 2.49985710e-01
-6.60748780e-01 1.08377978e-01 -7.82807693e-02 -6.44718230e-01
-3.53046864e-01 -8.81334662e-01 -2.21262440e-01 -2.10629225e-01
-3.32318425e-01 -4.69604045e-01 3.83006066e-01 -5.47448635e-01
1.53582978e+00 -2.05600858e+00 2.00441450e-01 6.72966540e-02
2.23689318e-01 3.84981960e-01 -6.82730138e-01 6.69428587e-01
-2.79332638e-01 4.28772688e-01 -4.59101945e-01 -9.01173472e-01
3.54540527e-01 5.37096739e-01 -3.34969670e-01 2.17902852e-05
4.40536350e-01 1.28248811e+00 -7.89860189e-01 -3.50832045e-01
7.19670430e-02 8.67515728e-02 -1.13644814e+00 1.40978977e-01
-2.40048811e-01 -5.45561671e-01 2.84296066e-01 1.79643512e-01
6.13548994e-01 -7.69863054e-02 1.93285808e-01 -1.90831542e-01
1.78901702e-01 1.02665401e+00 -1.05188966e+00 1.93897402e+00
-1.28819132e+00 6.34874165e-01 4.08825874e-02 -9.62541223e-01
6.51411235e-01 5.78828789e-02 -1.05125025e-01 -7.76208282e-01
8.49972740e-02 3.39390635e-01 2.96415895e-01 -4.96000707e-01
9.24562335e-01 -3.61959159e-01 -4.12236899e-01 7.56207764e-01
1.43908709e-01 -4.60094213e-01 4.19320792e-01 5.35242617e-01
1.16423821e+00 -3.07731301e-01 1.38166785e-01 -3.30030203e-01
5.58697820e-01 -3.17604780e-01 1.30886838e-01 6.92972958e-01
3.07037860e-01 3.89601171e-01 5.82363963e-01 -3.69261742e-01
-1.01841486e+00 -1.02969253e+00 -1.85138896e-01 1.70964277e+00
-3.44152272e-01 -9.44438815e-01 -6.69584632e-01 -6.90377235e-01
2.93602139e-01 1.12145996e+00 -5.71202278e-01 -5.59801340e-01
-8.84734929e-01 -8.01643252e-01 1.06343162e+00 8.05728316e-01
-3.89872193e-02 -8.86630714e-01 -2.90238142e-01 1.81205094e-01
-2.11738110e-01 -1.18174684e+00 -5.86816907e-01 5.23405254e-01
-7.08065271e-01 -5.32674313e-01 -3.57362591e-02 -9.31038797e-01
4.03386325e-01 -1.34138256e-01 1.64657724e+00 2.11053193e-01
-3.31590593e-01 4.97026145e-02 -3.39535564e-01 -3.33367884e-01
-4.92715001e-01 3.72784227e-01 6.98620752e-02 -4.79508311e-01
6.56866252e-01 -5.16000032e-01 -3.80743802e-01 -1.47174641e-01
-9.68201995e-01 -1.76221013e-01 6.15082562e-01 1.01390266e+00
1.91404715e-01 -4.93227601e-01 4.53142911e-01 -1.24284732e+00
1.04574335e+00 -6.97778702e-01 -9.28670019e-02 9.37053263e-02
-7.61918664e-01 4.74726528e-01 8.59412313e-01 -4.76224750e-01
-8.14042568e-01 -4.54516351e-01 -6.95015550e-01 3.47784869e-02
1.28526434e-01 7.53228664e-01 9.44191739e-02 2.43213996e-01
8.53229165e-01 -2.09487617e-01 -2.38636117e-02 -4.88290370e-01
1.02000058e+00 5.76588333e-01 3.97697121e-01 -7.56787419e-01
6.22787714e-01 -1.20848194e-02 -4.57079381e-01 -6.57864809e-01
-9.72258747e-01 -3.82841557e-01 -2.99108148e-01 7.25904882e-01
7.75705814e-01 -8.41253579e-01 -5.73993325e-01 9.34467465e-02
-1.39467728e+00 -3.70987236e-01 -6.12622261e-01 3.30353826e-01
-8.31839368e-02 4.65977609e-01 -1.00641811e+00 -3.54878247e-01
-6.39526546e-01 -9.98635232e-01 1.05496204e+00 -3.70616913e-01
-9.35976326e-01 -1.34715462e+00 2.47964621e-01 3.84090424e-01
6.83245420e-01 -2.79602140e-01 1.40836656e+00 -9.59152579e-01
2.02423409e-01 -1.79388836e-01 -1.92479953e-01 7.37282991e-01
-2.40759358e-01 -2.68626660e-01 -1.00504208e+00 -2.78532982e-01
6.83572516e-02 -8.08624268e-01 1.12655365e+00 5.53797223e-02
1.15306628e+00 -5.22183478e-01 5.43511920e-02 6.48086727e-01
1.19783223e+00 -3.00529420e-01 3.96659166e-01 2.65195310e-01
6.72107816e-01 4.39425021e-01 1.72624663e-01 1.95347890e-01
6.17146850e-01 4.69342142e-01 1.61598325e-01 -2.81467527e-01
-1.00520283e-01 -2.15022698e-01 7.35927880e-01 1.23754644e+00
5.10928392e-01 -2.04881191e-01 -1.01932490e+00 5.82283199e-01
-1.60117877e+00 -6.86246753e-01 -5.49310036e-02 1.80055988e+00
1.23386145e+00 3.42136323e-01 -1.40915141e-01 2.19331205e-01
1.16836920e-01 2.17492670e-01 -4.01521176e-01 -1.59147227e+00
-1.09767750e-01 8.69406760e-01 2.52052516e-01 9.81435418e-01
-7.71427751e-01 1.01561582e+00 6.12172461e+00 9.46083963e-01
-9.08886194e-01 2.26372316e-01 7.11007476e-01 -4.41700816e-01
-8.06713641e-01 -1.10817522e-01 -8.46333861e-01 2.61699229e-01
1.38872516e+00 -1.96645558e-02 3.68392557e-01 8.35623443e-01
-2.84108251e-01 -4.99030165e-02 -1.72216535e+00 8.67091417e-01
3.34096193e-01 -1.34055376e+00 2.14275777e-01 -4.32947934e-01
4.92329985e-01 4.06356543e-01 1.79375127e-01 9.17114556e-01
4.24767673e-01 -1.31126261e+00 5.05291164e-01 -2.32622251e-02
7.50520647e-01 -8.39672625e-01 9.24423039e-01 2.06265211e-01
-1.12104642e+00 -1.84441611e-01 -4.96070981e-01 -4.69778687e-01
1.49462804e-01 4.88938034e-01 -8.24066937e-01 2.76797891e-01
5.28754890e-01 4.95497018e-01 -8.61028969e-01 2.92420328e-01
-1.72201440e-01 6.11030698e-01 -3.08972955e-01 -2.52800167e-01
6.32012188e-01 2.78471205e-02 1.71236768e-01 1.82395494e+00
2.63379328e-02 -2.15580299e-01 3.57047021e-02 7.57762194e-01
-3.98349911e-01 1.77699879e-01 -5.49449801e-01 -4.52108234e-02
5.34227192e-01 1.41574824e+00 -8.77429545e-02 -5.28620899e-01
-5.02697647e-01 9.79613245e-01 7.55862057e-01 -3.04815080e-02
-8.30894470e-01 -7.86919236e-01 9.02249813e-01 8.08301941e-02
2.47672066e-01 -2.98515469e-01 -6.07976317e-01 -1.20873237e+00
1.36936083e-01 -1.16332102e+00 5.21011174e-01 -4.81836528e-01
-1.53214228e+00 9.19122100e-01 -1.12426661e-01 -5.42756140e-01
-6.48789942e-01 -9.27117229e-01 -7.57942140e-01 9.98264790e-01
-1.56749964e+00 -1.07829726e+00 1.40248816e-02 4.74427670e-01
5.89102328e-01 4.25455235e-02 1.14686358e+00 4.21344608e-01
-6.49443865e-01 1.15371668e+00 -8.64754617e-02 2.01581597e-01
8.01133811e-01 -1.51657963e+00 1.10289216e+00 7.35295951e-01
5.00554204e-01 9.85591114e-01 4.86803710e-01 -6.96347877e-02
-1.45389140e+00 -1.03370643e+00 1.50136507e+00 -7.83168316e-01
1.13181794e+00 -8.19257319e-01 -9.73042071e-01 7.66631961e-01
5.52505910e-01 -6.38093278e-02 9.09072399e-01 7.48549342e-01
-8.75009000e-01 -1.18302420e-01 -7.77915478e-01 7.68573165e-01
8.93963337e-01 -9.54604149e-01 -9.93253112e-01 3.12821031e-01
1.15992308e+00 -3.36170793e-01 -8.66771281e-01 1.37121111e-01
5.22775173e-01 -7.00341225e-01 9.24208641e-01 -1.01783574e+00
7.99894691e-01 3.26669216e-01 -2.52690256e-01 -1.44556999e+00
-3.28599930e-01 -4.82264996e-01 -8.29831138e-02 1.26807237e+00
9.38916624e-01 -6.71816289e-01 5.68524659e-01 7.35244989e-01
-3.65403533e-01 -1.26645863e+00 -6.82136834e-01 -8.50540817e-01
7.02431798e-01 -8.77834618e-01 5.45553446e-01 8.95409584e-01
3.51441443e-01 9.37813997e-01 6.29258677e-02 -4.07339454e-01
-2.19967272e-02 1.94995925e-02 3.88775826e-01 -7.24788666e-01
-7.02716768e-01 -6.68857813e-01 -1.78732306e-01 -1.19917297e+00
4.81964618e-01 -1.53273809e+00 -9.91224945e-02 -1.42059290e+00
1.54242031e-02 -5.29401600e-01 -2.48827368e-01 5.33801496e-01
-3.27794999e-01 3.06879163e-01 2.24984542e-01 -5.15672505e-01
-3.63525957e-01 3.13846111e-01 4.86600071e-01 -2.41163403e-01
2.20448468e-02 -5.18566132e-01 -1.06690609e+00 6.58287585e-01
8.06284904e-01 -4.07490075e-01 -5.12362182e-01 -1.17045951e+00
7.20387280e-01 -5.62707603e-01 4.06730659e-02 -7.05159187e-01
2.06656173e-01 2.78737992e-01 -9.95535962e-03 -1.43666804e-01
5.07853448e-01 -4.58097965e-01 -6.28042221e-01 4.14649278e-01
-8.16217363e-01 7.47834861e-01 5.52726686e-01 1.78065285e-01
-2.53587067e-01 -4.02805567e-01 6.15249395e-01 -1.70031443e-01
-4.42130059e-01 4.68611866e-02 -4.48486924e-01 6.65920138e-01
5.21290660e-01 -1.36103377e-01 -2.47588009e-01 -1.11067802e-01
-4.36131001e-01 3.15542489e-01 9.59332809e-02 5.75164258e-01
6.35403931e-01 -1.18269598e+00 -8.08386028e-01 3.60039294e-01
2.35204175e-01 -8.39756988e-03 4.89577986e-02 8.46880198e-01
-4.41152900e-01 5.71379900e-01 1.15523666e-01 -2.85448760e-01
-1.31276643e+00 5.50030589e-01 2.03356788e-01 -6.49999917e-01
-2.82740295e-01 1.38530385e+00 1.08227566e-01 -9.35237885e-01
1.49617434e-01 -9.16548610e-01 1.37079492e-01 2.02373788e-01
4.54561979e-01 2.75918424e-01 5.35637259e-01 -1.96731299e-01
-4.68067795e-01 3.81562471e-01 -4.13500577e-01 6.07251050e-03
1.31466937e+00 3.33686955e-02 -4.31826591e-01 4.68084306e-01
1.66294909e+00 6.98131397e-02 -4.10396695e-01 -3.45620573e-01
2.93017089e-01 2.33798604e-02 -6.34578168e-02 -6.43729091e-01
-6.69382989e-01 1.31680572e+00 4.87882979e-02 2.41451580e-02
7.92643726e-01 -3.87964323e-02 1.35558987e+00 9.22204792e-01
-2.54663259e-01 -1.11476636e+00 1.65851533e-01 9.75623548e-01
8.62660110e-01 -1.23586726e+00 -2.02345133e-01 1.24666151e-02
-7.22495437e-01 9.86985028e-01 7.76700974e-01 -2.39031225e-01
5.98807216e-01 6.94628179e-01 -8.75931010e-02 5.05095869e-02
-1.25210452e+00 9.64296833e-02 3.41223508e-01 1.66111499e-01
9.57313001e-01 1.07939234e-02 -2.67474979e-01 7.61707366e-01
-6.69516146e-01 -6.16734743e-01 2.85217673e-01 7.42088616e-01
-3.55585575e-01 -1.47320211e+00 1.47471130e-01 5.90814948e-01
-4.05462265e-01 -1.03755987e+00 -3.55483055e-01 6.45288289e-01
2.00300783e-01 9.08001125e-01 3.32799077e-01 -5.65953612e-01
4.21932280e-01 2.47843876e-01 3.54800016e-01 -1.05457592e+00
-1.17556167e+00 -7.28809714e-01 5.90028584e-01 -7.71993756e-01
2.30927408e-01 -3.37460726e-01 -1.28759944e+00 -7.10272014e-01
-3.16878110e-01 2.10934430e-01 5.79805195e-01 9.34095562e-01
4.85471934e-01 4.02767718e-01 3.57280403e-01 -5.65453649e-01
-9.72717106e-01 -9.96110559e-01 -1.07720472e-01 4.91966337e-01
3.39601517e-01 -7.94769600e-02 -2.85224438e-01 -3.87302130e-01] | [10.790111541748047, 8.714434623718262] |
17ab0007-3d1a-4075-ae99-e26935d72079 | learning-latent-representations-in-neural | 1802.03063 | null | http://arxiv.org/abs/1802.03063v1 | http://arxiv.org/pdf/1802.03063v1.pdf | Learning Latent Representations in Neural Networks for Clustering through Pseudo Supervision and Graph-based Activity Regularization | In this paper, we propose a novel unsupervised clustering approach exploiting
the hidden information that is indirectly introduced through a pseudo
classification objective. Specifically, we randomly assign a pseudo
parent-class label to each observation which is then modified by applying the
domain specific transformation associated with the assigned label. Generated
pseudo observation-label pairs are subsequently used to train a neural network
with Auto-clustering Output Layer (ACOL) that introduces multiple softmax nodes
for each pseudo parent-class. Due to the unsupervised objective based on
Graph-based Activity Regularization (GAR) terms, softmax duplicates of each
parent-class are specialized as the hidden information captured through the
help of domain specific transformations is propagated during training.
Ultimately we obtain a k-means friendly latent representation. Furthermore, we
demonstrate how the chosen transformation type impacts performance and helps
propagate the latent information that is useful in revealing unknown clusters.
Our results show state-of-the-art performance for unsupervised clustering tasks
on MNIST, SVHN and USPS datasets, with the highest accuracies reported to date
in the literature. | ['Ozsel Kilinc', 'Ismail Uysal'] | 2018-02-08 | learning-latent-representations-in-neural-1 | https://openreview.net/forum?id=HkMvEOlAb | https://openreview.net/pdf?id=HkMvEOlAb | iclr-2018-1 | ['unsupervised-image-classification'] | ['computer-vision'] | [ 6.21617079e-01 4.48621571e-01 -4.76675600e-01 -6.68099344e-01
-6.35904312e-01 -3.56228858e-01 7.44309783e-01 3.27367276e-01
-3.49389017e-01 5.09383678e-01 3.74039710e-02 2.27105856e-01
-3.03577870e-01 -6.00280881e-01 -7.76126802e-01 -1.17330635e+00
-1.36682972e-01 6.40168548e-01 -4.27228361e-02 5.73666096e-01
1.58902511e-01 2.60359794e-01 -1.57046413e+00 5.05434215e-01
7.51289070e-01 8.52415681e-01 1.69299930e-01 3.04897606e-01
-3.68512943e-02 9.90812182e-01 -5.27993739e-01 -1.05797537e-01
2.21307669e-02 -5.23431778e-01 -9.66244817e-01 5.56953251e-01
1.19975835e-01 3.79189432e-01 -4.96571213e-02 8.90993237e-01
6.91723228e-02 5.22096038e-01 1.16143775e+00 -1.20733428e+00
-5.91308653e-01 1.16686380e+00 -5.79391599e-01 -2.16187000e-01
-2.15683058e-01 -1.52206467e-02 1.09799445e+00 -5.89039385e-01
9.12513673e-01 9.16565835e-01 4.06905979e-01 5.07754266e-01
-1.68438089e+00 -5.41578650e-01 3.79013866e-01 1.46379113e-01
-1.50956833e+00 -3.58528882e-01 1.12357676e+00 -5.97572744e-01
7.80185699e-01 -1.00941554e-01 2.36520737e-01 1.14168155e+00
-2.07826331e-01 7.46864438e-01 9.70175564e-01 -4.30814683e-01
5.68354964e-01 4.61997271e-01 3.49170625e-01 6.61923230e-01
-1.33170262e-01 -1.27973706e-01 -5.56238234e-01 1.50170192e-01
2.34830722e-01 -1.87584311e-02 7.65550230e-03 -6.59268320e-01
-9.90970433e-01 9.33592796e-01 6.83167815e-01 4.00140852e-01
-5.44005990e-01 1.19891666e-01 4.08006683e-02 9.29527432e-02
5.24281681e-01 5.08536518e-01 -3.38665456e-01 2.37449482e-01
-1.29657984e+00 -4.10660356e-01 6.07966423e-01 8.18998933e-01
1.15116537e+00 2.18006410e-02 -3.30613852e-01 8.11915874e-01
6.05178356e-01 -9.30398703e-02 5.51505327e-01 -1.05375683e+00
2.88181245e-01 1.06806600e+00 -4.32405770e-01 -8.00690949e-01
-6.58513129e-01 -9.78118598e-01 -8.04542601e-01 1.90889403e-01
2.01520979e-01 1.67604107e-02 -1.00215304e+00 1.95686853e+00
4.38503802e-01 4.00940090e-01 3.44447643e-01 5.68486452e-01
6.31475508e-01 5.17883539e-01 2.85496235e-01 -3.61946136e-01
1.05894899e+00 -1.24800706e+00 -5.88347435e-01 -2.59193808e-01
7.83859193e-01 -4.14328903e-01 8.90725613e-01 3.21267664e-01
-6.82716250e-01 -5.55331349e-01 -1.03668118e+00 1.23666137e-01
-6.76134706e-01 4.02117908e-01 5.08697748e-01 4.92275298e-01
-9.76992726e-01 5.59511364e-01 -1.14207506e+00 -4.05008584e-01
4.99542356e-01 6.31654263e-01 -2.27242991e-01 2.27476478e-01
-7.13075340e-01 4.35890019e-01 7.97140956e-01 -1.01777315e-01
-1.06731117e+00 -5.42406857e-01 -8.49776447e-01 1.46080136e-01
5.19337714e-01 -4.63577151e-01 7.62656748e-01 -1.01483810e+00
-1.54206216e+00 8.85563433e-01 -1.13373958e-01 -6.31021798e-01
6.52777031e-02 4.47227150e-01 -3.63752395e-01 3.07347447e-01
2.65024930e-01 1.03617752e+00 1.08463120e+00 -1.77248275e+00
-5.65411925e-01 -4.03596997e-01 -3.34572464e-01 8.41291249e-02
-6.97675228e-01 -3.55596751e-01 -6.12364054e-01 -7.01416910e-01
4.13720906e-01 -1.14052892e+00 -1.15871854e-01 -6.23253822e-01
-9.52991605e-01 -2.18227223e-01 8.36438239e-01 -4.91103560e-01
1.36792600e+00 -2.16955805e+00 4.74664658e-01 5.94779015e-01
3.39961708e-01 -2.34605551e-01 2.36622058e-02 2.94470102e-01
-1.68216005e-01 -1.07057355e-02 -7.29784369e-01 -8.87507021e-01
1.08497053e-01 2.21158966e-01 6.10277057e-03 3.75593126e-01
2.21485123e-01 7.14031756e-01 -8.24833512e-01 -4.24923569e-01
2.73657799e-01 5.44392645e-01 -4.20624226e-01 1.64998740e-01
-3.93365592e-01 6.84322000e-01 -2.04397842e-01 3.89728844e-01
4.36342031e-01 -7.21892297e-01 4.96486574e-01 -2.18472704e-01
1.33711055e-01 2.04475075e-01 -1.02511108e+00 1.82398844e+00
-2.74349213e-01 5.37743390e-01 -1.54477850e-01 -1.31864750e+00
8.23208869e-01 1.93712130e-01 6.65356338e-01 -2.84906849e-02
1.28090367e-01 -2.22002104e-01 -1.67838931e-01 -2.58885503e-01
2.06165195e-01 2.36432254e-01 8.72945413e-02 6.47345364e-01
5.16298294e-01 3.19152385e-01 4.17610794e-01 4.28554803e-01
1.01066887e+00 2.12767109e-01 -3.39913252e-03 -1.98013991e-01
6.64038658e-01 -4.17928509e-02 2.72555441e-01 8.10360491e-01
2.66047031e-01 5.23689389e-01 5.54768622e-01 2.99398117e-02
-7.47272134e-01 -1.04966962e+00 -5.13617471e-02 1.26187801e+00
-2.23806411e-01 -3.91111195e-01 -9.36212063e-01 -9.70946133e-01
-2.85902947e-01 9.39812839e-01 -8.91472578e-01 -5.04501700e-01
-1.84948385e-01 -8.03072512e-01 4.53437299e-01 4.11873430e-01
3.20103854e-01 -1.27225637e+00 -2.90537894e-01 2.76810229e-01
-5.65885566e-02 -1.15708911e+00 -1.40316531e-01 7.69804537e-01
-1.04492331e+00 -9.43674386e-01 -3.23951393e-01 -9.34117734e-01
1.11539805e+00 -1.90491959e-01 7.77122021e-01 -3.08657527e-01
-7.92764872e-03 1.91546738e-01 -3.09242368e-01 2.16418400e-01
-4.91989851e-01 6.96528554e-01 -5.93567602e-02 6.30117714e-01
6.14371419e-01 -7.17735350e-01 -2.03015909e-01 -2.33705584e-02
-9.68269289e-01 -1.05809281e-02 5.87026179e-01 6.46898329e-01
6.45130634e-01 4.72881287e-01 4.00879562e-01 -1.26377547e+00
3.10372591e-01 -6.78360283e-01 -6.80185914e-01 2.17023626e-01
-9.33583140e-01 5.56158125e-01 7.11660206e-01 -5.93764126e-01
-1.20341468e+00 5.77913284e-01 5.03866494e-01 -5.81312478e-01
-4.76723254e-01 6.79074228e-01 -2.07996815e-01 3.05968016e-01
6.81483090e-01 2.75757492e-01 -3.23098779e-01 -5.49659848e-01
5.22059917e-01 5.73686957e-01 7.07397699e-01 -3.85596544e-01
7.74903357e-01 7.38402843e-01 1.00269832e-01 -5.58998168e-01
-9.41148639e-01 -5.25002539e-01 -1.03496206e+00 -1.42710879e-01
1.40566456e+00 -7.74722636e-01 -5.40042818e-01 4.07132268e-01
-7.84406304e-01 -3.68510604e-01 -3.23089212e-01 6.52084053e-01
-3.80411744e-01 1.88875854e-01 -5.15596628e-01 -8.05495918e-01
1.83239002e-02 -1.25857568e+00 9.18270767e-01 4.76117432e-02
-1.07804164e-01 -1.13803101e+00 -6.20421879e-02 4.97945279e-01
-1.00477859e-01 2.40145341e-01 1.08669817e+00 -8.83648217e-01
-7.05574632e-01 -8.69384184e-02 -7.27285743e-02 3.07550788e-01
3.08126152e-01 -1.47881778e-03 -1.20910585e+00 -2.92053461e-01
-2.12001920e-01 -2.47028112e-01 1.22211182e+00 5.21296799e-01
1.11391783e+00 -4.09279406e-01 -6.77380085e-01 7.31103718e-01
1.19152081e+00 7.41654858e-02 2.01038897e-01 2.29012400e-01
9.03493106e-01 7.65842676e-01 1.85546651e-01 3.39741409e-01
4.43941176e-01 5.98906457e-01 2.77712464e-01 -1.20810226e-01
-8.04192722e-02 -1.34737343e-01 3.47104162e-01 8.10338795e-01
1.80041775e-01 -2.61236697e-01 -7.68624783e-01 4.73685384e-01
-1.98330593e+00 -7.93447793e-01 -5.03069535e-02 2.20105791e+00
9.32674289e-01 2.07275108e-01 -7.16641396e-02 1.84333190e-01
8.21255505e-01 8.95782486e-02 -5.18587291e-01 1.04406446e-01
-1.74516156e-01 6.27218187e-02 5.07809401e-01 6.24836504e-01
-1.37331998e+00 1.21029401e+00 5.03807831e+00 7.35396385e-01
-9.53813612e-01 1.44757509e-01 6.06581688e-01 8.52239132e-02
5.06365746e-02 1.90114245e-01 -6.56576157e-01 4.27613318e-01
9.06662881e-01 3.17866892e-01 5.89116871e-01 6.85669839e-01
2.70248801e-01 -3.85818109e-02 -1.30083811e+00 6.16414428e-01
8.80418494e-02 -1.09989905e+00 5.05284406e-02 2.56348073e-01
1.14024603e+00 -5.20625450e-02 3.28803658e-01 2.31880412e-01
5.48986793e-01 -7.34097064e-01 3.56962711e-01 4.90172684e-01
4.98936325e-01 -7.59178698e-01 4.21532243e-01 2.71774590e-01
-1.01897335e+00 -5.35549782e-02 -2.36711964e-01 2.35899672e-01
-5.99089600e-02 6.02126122e-01 -1.21539509e+00 5.12099743e-01
4.82864320e-01 8.61148655e-01 -7.55909979e-01 6.38312042e-01
-6.76855385e-01 1.03600860e+00 -3.51506531e-01 4.14215654e-01
3.93589973e-01 -3.71254981e-01 4.62495118e-01 1.15505397e+00
-2.56461110e-02 -2.74242163e-01 2.59761155e-01 1.21510124e+00
-4.08579528e-01 -1.01767480e-01 -2.92032659e-01 -2.13298440e-01
4.12324131e-01 1.43890703e+00 -1.21860754e+00 -5.16470253e-01
-1.27118498e-01 1.11886191e+00 5.52923799e-01 6.82660937e-01
-6.87460542e-01 -1.03934571e-01 1.76327929e-01 -1.51542872e-01
4.57410455e-01 4.28162254e-02 -3.02081138e-01 -9.71464932e-01
-1.70354411e-01 -3.90862852e-01 7.41316140e-01 -6.51274204e-01
-1.22978032e+00 6.45808876e-01 -3.75916213e-02 -1.10171688e+00
-4.76951748e-01 -2.23155797e-01 -4.36705887e-01 4.14053023e-01
-1.32012212e+00 -1.35410988e+00 -2.77190834e-01 4.27284986e-01
3.73968631e-01 -3.18813890e-01 8.07412684e-01 2.66535819e-01
-8.51668894e-01 6.55772626e-01 3.78774464e-01 1.68859869e-01
7.63182282e-01 -1.41609144e+00 -1.29304484e-01 8.60749900e-01
3.74771357e-01 6.84848964e-01 4.50863659e-01 -7.73879230e-01
-6.29580557e-01 -1.43241847e+00 8.43182027e-01 -3.47632825e-01
4.71440434e-01 -5.26102006e-01 -8.62898886e-01 8.95445645e-01
4.65480648e-02 -2.53298134e-01 8.44651937e-01 5.87540045e-02
-5.01769125e-01 7.18856752e-02 -1.01375723e+00 3.86215657e-01
8.03169429e-01 -5.50626218e-01 -4.58796203e-01 3.29374909e-01
9.07990634e-01 1.78130671e-01 -8.96950185e-01 2.87436634e-01
1.33535653e-01 -6.54689252e-01 8.27326059e-01 -5.97712934e-01
4.48136955e-01 -5.39449155e-01 2.26970501e-02 -1.26836836e+00
-4.91324484e-01 -3.72155875e-01 -2.55590618e-01 1.42171681e+00
9.29206014e-01 -2.31756955e-01 1.18635607e+00 4.20571774e-01
-2.48862356e-01 -4.11404788e-01 -5.81668437e-01 -6.66399240e-01
-2.91985035e-01 -2.95972228e-01 3.97406459e-01 1.46360719e+00
-6.18379284e-03 6.13773882e-01 -3.26101512e-01 3.62366676e-01
8.83520544e-01 -1.68765094e-02 5.18813372e-01 -1.33117473e+00
-4.76681709e-01 -3.85609090e-01 -1.83805659e-01 -6.59905314e-01
5.08242667e-01 -1.42829025e+00 -3.97529863e-02 -1.32128930e+00
1.75081313e-01 -3.30516964e-01 -6.45536959e-01 8.04694176e-01
-9.71418619e-02 4.39468890e-01 1.06971607e-01 3.99736524e-01
-8.79348993e-01 4.79816318e-01 6.08189523e-01 -2.94421107e-01
-7.61478961e-01 8.26831982e-02 -4.58764285e-01 6.49305880e-01
6.86276674e-01 -9.26657081e-01 -4.31912512e-01 -2.34503403e-01
-1.26798958e-01 -3.95738959e-01 2.03115866e-01 -1.14528096e+00
5.39024830e-01 1.68487787e-01 3.51126939e-01 -6.05276167e-01
5.40230691e-01 -1.07020092e+00 1.78890243e-01 3.17777067e-01
-7.90253460e-01 -5.70377827e-01 -3.27678293e-01 6.47159457e-01
-1.54450446e-01 -2.51847476e-01 7.48384476e-01 4.01940495e-02
-4.34310406e-01 9.51906294e-02 -5.76256275e-01 -2.96782911e-01
1.01557362e+00 -2.88939685e-01 7.65254721e-03 -3.53105515e-01
-1.24368882e+00 2.49965146e-01 3.29860747e-01 3.47208381e-01
1.74369052e-01 -1.14979291e+00 -4.82439965e-01 2.53249675e-01
3.13513666e-01 1.57535318e-02 1.73301160e-01 8.97716224e-01
1.13977261e-01 3.07364017e-01 1.36595264e-01 -8.59076738e-01
-1.07459617e+00 8.39823186e-01 3.91828045e-02 -4.44053024e-01
-4.06813413e-01 6.81838214e-01 1.94180965e-01 -7.57437110e-01
4.97200161e-01 -2.26760671e-01 -4.14934486e-01 3.49496692e-01
-4.82869521e-02 4.96065915e-01 4.80300300e-02 -7.42798150e-01
-2.68932283e-01 3.40980083e-01 -5.10981418e-02 -1.93143070e-01
1.46596026e+00 -1.95287481e-01 -1.77364379e-01 4.88565862e-01
1.44880307e+00 -2.63038754e-01 -1.30788362e+00 -3.95492762e-01
4.39723074e-01 2.14836955e-01 8.97919685e-02 -8.67310643e-01
-1.26686954e+00 7.12076783e-01 6.35539055e-01 3.11731547e-02
1.08354092e+00 4.12791103e-01 2.05769747e-01 5.81425846e-01
-2.42669746e-01 -1.31588149e+00 1.35603577e-01 2.05058709e-01
2.29908690e-01 -1.09375632e+00 -2.20554635e-01 -4.77086604e-01
-6.14221811e-01 8.26368570e-01 4.71435219e-01 1.03842556e-01
6.09258115e-01 8.38318020e-02 1.30918011e-01 -5.05323410e-01
-5.58625460e-01 -2.00156435e-01 3.57335061e-01 6.38510525e-01
2.19678104e-01 7.88614526e-02 2.73312777e-02 5.38235784e-01
-2.09439740e-01 -3.37202430e-01 2.84457803e-01 6.80112302e-01
-3.57863307e-01 -1.09485269e+00 -2.77820289e-01 5.11686921e-01
-3.19191813e-01 -1.37985975e-01 -6.33820713e-01 4.81716514e-01
1.91404149e-01 1.09392738e+00 1.05005868e-01 -3.82851213e-01
-3.03238064e-01 4.05826002e-01 1.39340878e-01 -8.57797325e-01
-6.93503141e-01 3.05717915e-01 -2.91460335e-01 -2.40226090e-01
-7.07019746e-01 -5.88317454e-01 -1.55407488e+00 3.05254102e-01
-4.41492707e-01 3.83070171e-01 6.79422081e-01 1.04612958e+00
3.83128464e-01 8.34666193e-01 6.09627545e-01 -6.52245700e-01
-1.84858650e-01 -1.08395612e+00 -5.49434483e-01 6.38448298e-01
-3.59555297e-02 -6.48189902e-01 -5.61384082e-01 6.06130779e-01] | [9.321417808532715, 3.1222541332244873] |
ef8d62d3-5c77-4ddf-a451-cc883766758a | reference-based-image-super-resolution-with | 2207.11938 | null | https://arxiv.org/abs/2207.11938v2 | https://arxiv.org/pdf/2207.11938v2.pdf | Reference-based Image Super-Resolution with Deformable Attention Transformer | Reference-based image super-resolution (RefSR) aims to exploit auxiliary reference (Ref) images to super-resolve low-resolution (LR) images. Recently, RefSR has been attracting great attention as it provides an alternative way to surpass single image SR. However, addressing the RefSR problem has two critical challenges: (i) It is difficult to match the correspondence between LR and Ref images when they are significantly different; (ii) How to transfer the relevant texture from Ref images to compensate the details for LR images is very challenging. To address these issues of RefSR, this paper proposes a deformable attention Transformer, namely DATSR, with multiple scales, each of which consists of a texture feature encoder (TFE) module, a reference-based deformable attention (RDA) module and a residual feature aggregation (RFA) module. Specifically, TFE first extracts image transformation (e.g., brightness) insensitive features for LR and Ref images, RDA then can exploit multiple relevant textures to compensate more information for LR features, and RFA lastly aggregates LR features and relevant textures to get a more visually pleasant result. Extensive experiments demonstrate that our DATSR achieves state-of-the-art performance on benchmark datasets quantitatively and qualitatively. | ['Luc van Gool', 'Wenguan Wang', 'Yulun Zhang', 'Yawei Li', 'Kai Zhang', 'Jingyun Liang', 'JieZhang Cao'] | 2022-07-25 | null | null | null | null | ['reference-based-super-resolution'] | ['computer-vision'] | [ 7.04618335e-01 -2.66427368e-01 3.50741297e-03 -3.75205129e-01
-1.21299100e+00 -2.27556288e-01 4.02790606e-01 -6.33452475e-01
-8.74541700e-04 7.56042778e-01 2.94172704e-01 2.28818744e-01
-1.39055684e-01 -7.29154587e-01 -7.79054046e-01 -9.83719110e-01
5.67517698e-01 -3.85072008e-02 5.86373568e-01 -5.75173974e-01
3.79458129e-01 7.27614224e-01 -1.73697710e+00 5.81521332e-01
9.95453000e-01 1.20682704e+00 6.75495386e-01 2.94614911e-01
-4.28548083e-02 7.82427907e-01 -2.77003706e-01 -3.65334377e-02
2.54958928e-01 -3.82430583e-01 -8.30439806e-01 -1.19665965e-01
7.61343002e-01 -4.83040273e-01 -5.00848591e-01 1.24906874e+00
5.12046993e-01 3.48195255e-01 4.29574788e-01 -6.90576792e-01
-1.45182228e+00 2.38605469e-01 -9.92986441e-01 5.65442204e-01
3.29103887e-01 7.26678967e-02 6.01819813e-01 -1.05926168e+00
5.52443564e-01 1.61490607e+00 1.62524655e-01 6.76946580e-01
-1.31017649e+00 -7.05243111e-01 6.71747625e-02 1.52282566e-01
-1.39180815e+00 -5.90634942e-01 8.51559877e-01 -5.70488442e-03
6.16035879e-01 4.65418041e-01 1.11566015e-01 1.04779708e+00
4.37012017e-01 4.82793212e-01 1.60841596e+00 -6.93631917e-02
-1.72878698e-01 4.97828349e-02 -9.50758532e-03 3.13662946e-01
-2.95981467e-02 1.98112920e-01 -3.93733442e-01 4.62671101e-01
1.36843026e+00 1.82384342e-01 -6.74777985e-01 8.85184929e-02
-1.09891891e+00 3.79788458e-01 6.65651143e-01 4.67238098e-01
-2.79794782e-01 -2.15871647e-01 -4.79114577e-02 2.15792492e-01
5.60723722e-01 3.59754384e-01 -2.93470263e-01 3.67679536e-01
-4.77223575e-01 2.95515805e-02 -1.16586521e-01 7.28898525e-01
9.71515656e-01 1.20151304e-01 -5.33686936e-01 1.29534864e+00
-9.55402106e-02 6.70113266e-01 6.11848950e-01 -9.82922673e-01
4.16872621e-01 5.01753986e-01 2.74845779e-01 -1.15754843e+00
1.25193179e-01 -2.43266925e-01 -1.20577979e+00 1.56256914e-01
1.36136562e-01 5.88072062e-01 -9.44526255e-01 1.53059173e+00
2.05498248e-01 1.04742572e-01 1.05585881e-01 1.38048804e+00
1.26269126e+00 7.64462590e-01 -2.07947448e-01 -2.39036679e-01
1.33003759e+00 -8.82091701e-01 -7.01381922e-01 -3.02453518e-01
-2.85346895e-01 -8.66919935e-01 1.33105087e+00 1.20601758e-01
-1.34163451e+00 -1.03483462e+00 -8.84968638e-01 -7.33552277e-01
-1.75441742e-01 1.74535304e-01 1.93399981e-01 -1.55264512e-04
-1.17303741e+00 5.68656564e-01 -2.97456563e-01 -8.98155347e-02
4.30870652e-01 3.12745988e-01 -4.86616820e-01 -4.11839455e-01
-1.20058990e+00 7.58894086e-01 1.33741826e-01 3.19673359e-01
-5.93897343e-01 -6.41478419e-01 -8.63287449e-01 -2.88188457e-03
3.88564378e-01 -4.37517673e-01 6.25649333e-01 -9.00588989e-01
-1.59645486e+00 9.39633667e-01 -4.16327566e-01 5.26080206e-02
2.33205989e-01 -9.86250043e-02 -7.09395409e-01 1.54018611e-01
2.20144004e-01 5.00617027e-01 1.14884806e+00 -1.42780423e+00
-6.89438105e-01 -4.59704101e-01 -5.26690260e-02 4.09646273e-01
-8.90389271e-03 1.00849539e-01 -4.23130661e-01 -7.98037589e-01
1.82881355e-01 -4.02795404e-01 7.17394724e-02 -1.59605771e-01
-2.35755175e-01 -1.62797555e-01 1.12491965e+00 -8.26968789e-01
1.03741407e+00 -2.24130774e+00 3.43632281e-01 -2.99256116e-01
3.34813058e-01 4.18496430e-01 -3.93900305e-01 -2.92245567e-01
-2.27998435e-01 9.51121747e-02 -1.20670065e-01 5.75169548e-02
-4.90583569e-01 1.26845554e-01 -7.15121210e-01 2.60346234e-01
5.75902581e-01 1.14452910e+00 -7.65001655e-01 -4.46920931e-01
5.77911615e-01 8.82391155e-01 -1.58547163e-01 3.12189102e-01
3.04816272e-02 8.59888196e-01 -6.25520170e-01 7.42854714e-01
1.12236524e+00 -2.80388117e-01 -2.70230144e-01 -8.68044376e-01
-4.07998741e-01 -4.60294485e-02 -1.00997233e+00 1.27400041e+00
-4.60928142e-01 4.24059510e-01 2.75739208e-02 -7.11057425e-01
1.37202001e+00 -2.33807206e-01 2.70216823e-01 -1.35178614e+00
6.95948601e-02 2.46308625e-01 -4.29748595e-01 -3.42982709e-01
7.19074547e-01 -1.02058575e-01 6.23937026e-02 1.19605809e-01
-2.89698333e-01 4.54329140e-02 -2.97914654e-01 -1.63931549e-01
6.31477237e-01 2.54965067e-01 5.65874651e-02 -2.75998950e-01
1.01445687e+00 -4.64918375e-01 6.86454415e-01 4.23054814e-01
-7.39845559e-02 1.01793468e+00 -2.77854782e-03 -6.66436374e-01
-1.11443758e+00 -1.17795384e+00 -3.13480973e-01 8.25634241e-01
8.59805703e-01 2.77654350e-01 -4.76790607e-01 -4.72240150e-01
-2.04513282e-01 4.43459362e-01 -8.29177082e-01 -1.61058038e-01
-8.41213107e-01 -5.25643229e-01 -1.32617101e-01 5.29334903e-01
1.04652262e+00 -1.18003047e+00 -2.51266122e-01 2.48603038e-02
-5.73733091e-01 -1.07456958e+00 -9.51692939e-01 -2.60360658e-01
-6.12780809e-01 -8.23734403e-01 -8.37122500e-01 -7.99559832e-01
6.76004410e-01 9.79367256e-01 9.42308784e-01 -5.03100343e-02
-3.78910601e-01 -8.20925683e-02 -3.58487874e-01 2.30877995e-01
-2.15799570e-01 -2.56403416e-01 -1.98464096e-01 4.61240590e-01
1.47225544e-01 -3.75609308e-01 -7.88231730e-01 6.59816504e-01
-1.05469275e+00 2.57625878e-01 8.85993481e-01 1.02701211e+00
1.26774311e+00 2.04618260e-01 4.11120921e-01 -6.83387995e-01
2.66081840e-01 -1.25878274e-01 -4.81277585e-01 4.74514723e-01
-3.41173589e-01 1.25718176e-01 8.03224564e-01 -4.27145183e-01
-1.41335595e+00 -1.73340037e-01 1.97899938e-02 -7.50918627e-01
-9.60955471e-02 -2.23785177e-01 -3.97621959e-01 -4.03402209e-01
3.24997336e-01 7.28228867e-01 -5.66367395e-02 -6.22558475e-01
2.94231296e-01 8.75289917e-01 8.58422279e-01 -5.06753564e-01
8.84159803e-01 5.85305691e-01 -3.09659597e-02 -7.26976037e-01
-1.22517824e+00 -1.99225605e-01 -6.62914753e-01 -6.98540062e-02
1.13287640e+00 -9.53161955e-01 -5.74980736e-01 4.86184686e-01
-8.63837004e-01 -2.04648271e-01 -3.61434102e-01 1.58394679e-01
-5.50737381e-01 2.77862281e-01 -5.26634455e-01 -5.44653058e-01
-4.87746626e-01 -1.39928758e+00 1.50923729e+00 7.98729658e-01
4.14420962e-01 -5.22666931e-01 -3.41420829e-01 5.70353568e-01
8.36492538e-01 2.86392510e-01 7.23197937e-01 1.66067794e-01
-9.73564863e-01 2.71881908e-01 -1.07885313e+00 4.80671108e-01
5.56959569e-01 -2.94555336e-01 -9.24749494e-01 -4.60371017e-01
1.11313500e-01 -1.99236244e-01 9.71618176e-01 3.49068224e-01
1.53862810e+00 -3.11322808e-01 -1.09956510e-01 8.79983187e-01
1.53420651e+00 2.44463921e-01 1.15958154e+00 3.65063816e-01
9.56131041e-01 3.86085212e-01 9.48174834e-01 -1.49137855e-01
4.59283888e-01 1.17853129e+00 1.50532290e-01 -4.65037376e-01
-8.05425406e-01 -1.70285642e-01 4.70143497e-01 5.09527445e-01
-3.38632107e-01 1.56217575e-01 -1.86932370e-01 2.96060532e-01
-1.52375233e+00 -1.05200434e+00 -7.16095269e-02 2.22019672e+00
9.55387294e-01 -3.12175900e-01 -2.17987031e-01 -1.47773445e-01
9.17078495e-01 3.10598910e-01 -8.94977450e-01 -1.77817807e-01
-5.48348427e-01 3.07496428e-01 2.76275665e-01 4.25571144e-01
-9.67943549e-01 1.02685654e+00 5.27947664e+00 1.23733974e+00
-1.27452886e+00 1.84555128e-01 9.77325976e-01 2.69725382e-01
-4.04063821e-01 -1.87012941e-01 -8.66355896e-01 7.42231309e-01
3.81410390e-01 9.26947296e-02 8.67030680e-01 5.14128208e-01
6.59957230e-02 5.12435101e-03 -6.35196447e-01 1.12680387e+00
1.81770191e-01 -1.10455084e+00 5.10040939e-01 -5.41791655e-02
7.54964113e-01 -2.20971882e-01 5.11352897e-01 1.63955301e-01
8.02789405e-02 -1.24086046e+00 5.31080484e-01 9.40769315e-01
1.52741826e+00 -8.34658027e-01 6.14113808e-01 -2.16452852e-01
-1.55494356e+00 -1.91217095e-01 -8.07450414e-01 4.75192428e-01
-9.98956561e-02 4.94658589e-01 9.28698182e-02 8.88538718e-01
1.19958830e+00 1.06405663e+00 -7.31943607e-01 3.83781910e-01
-1.90871097e-02 -1.46012008e-01 1.56746343e-01 5.70221663e-01
-1.64654016e-01 -3.79911393e-01 6.29006147e-01 7.51519501e-01
3.07856441e-01 4.79543567e-01 -8.14366639e-02 1.17616522e+00
6.96520880e-02 -1.36040732e-01 -2.71039158e-01 2.96701789e-01
4.48334068e-01 1.48336506e+00 -5.24832785e-01 -1.20636448e-01
-3.52370411e-01 1.31974733e+00 3.31045777e-01 4.56639409e-01
-7.25271046e-01 -4.57362354e-01 5.97594976e-01 3.09062362e-01
3.66599530e-01 1.55121550e-01 -2.13475712e-02 -1.47408855e+00
5.12914397e-02 -9.20948148e-01 2.43874207e-01 -1.44998479e+00
-1.38045466e+00 8.61489296e-01 -3.79180938e-01 -1.39623809e+00
2.64884591e-01 -1.92905098e-01 -3.12658250e-01 1.37496209e+00
-1.98265767e+00 -1.28077281e+00 -7.95061707e-01 7.60409296e-01
7.31627107e-01 5.31348959e-02 3.63365352e-01 2.86611915e-01
-7.28758216e-01 5.93347907e-01 1.17319271e-01 -6.33711964e-02
8.04453492e-01 -1.01643264e+00 3.24152738e-01 7.86370158e-01
-4.17240292e-01 4.97464359e-01 4.11613643e-01 -5.87342381e-01
-1.57345712e+00 -1.43086171e+00 4.44906473e-01 -5.82711339e-01
1.60219923e-01 -4.10922766e-02 -1.36830306e+00 4.07151282e-01
-2.86511064e-01 4.51800168e-01 -1.04340859e-01 -6.02275014e-01
-4.24371481e-01 -5.83729446e-01 -1.30475998e+00 5.32187939e-01
1.11700487e+00 -5.95317006e-01 -5.31134605e-01 -7.48631954e-02
9.03298259e-01 -5.72938263e-01 -1.06022692e+00 6.52603984e-01
4.51137573e-01 -1.19540548e+00 1.36849046e+00 4.18612100e-02
7.32018232e-01 -7.38035977e-01 -4.35911119e-01 -1.03954959e+00
-6.28163874e-01 -3.24065030e-01 7.59585649e-02 1.47218609e+00
-2.26496622e-01 -8.28548789e-01 7.43400827e-02 3.97809088e-01
-6.32097870e-02 -8.39842200e-01 -7.99720883e-01 -6.55434370e-01
-9.76863056e-02 3.88670146e-01 8.71464431e-01 9.34173942e-01
-8.76110792e-01 2.35599518e-01 -5.28587937e-01 2.66816169e-01
8.30672622e-01 5.27916670e-01 4.62557971e-01 -1.16489720e+00
-5.49513735e-02 -4.42571461e-01 -1.38621211e-01 -1.04533267e+00
-5.99377714e-02 -6.86697841e-01 9.77340192e-02 -1.51376486e+00
5.57183027e-01 -5.36944747e-01 -4.86496806e-01 3.30083996e-01
-4.12819475e-01 7.02220201e-01 -2.32371278e-02 4.18088734e-01
-6.75655901e-01 6.53050601e-01 1.91632104e+00 -4.50928807e-02
-1.93259180e-01 -3.81001353e-01 -1.11311221e+00 3.34917188e-01
4.48816746e-01 -3.34043577e-02 -1.42493546e-01 -3.45131934e-01
-2.19880670e-01 2.09484637e-01 6.50403619e-01 -7.12708116e-01
-1.05306648e-01 -3.65451425e-01 8.80560219e-01 -7.11614013e-01
2.72658378e-01 -4.85490322e-01 2.00215787e-01 -7.44438637e-03
-2.94115394e-01 -2.28045017e-01 8.31760317e-02 5.98931611e-01
-3.55861962e-01 4.39670116e-01 1.30542457e+00 -6.93198219e-02
-8.49223554e-01 6.06627464e-01 2.54295737e-01 5.30680753e-02
7.92527318e-01 -3.94318700e-01 -7.40433335e-01 7.41071627e-03
-2.32018724e-01 6.79000989e-02 7.39876688e-01 7.90156722e-01
1.03861094e+00 -1.46202826e+00 -8.28893542e-01 4.89236176e-01
7.49893710e-02 1.97211757e-01 9.76222396e-01 7.37061262e-01
-1.09970525e-01 3.08282882e-01 -4.60164815e-01 -5.20306230e-01
-1.26413929e+00 7.00896740e-01 3.64935040e-01 -1.70035362e-01
-1.00123322e+00 7.51206040e-01 7.69839466e-01 -1.26830012e-01
-1.40543133e-01 -1.94093972e-01 -5.62204361e-01 -2.58023798e-01
9.96091068e-01 3.07269871e-01 -1.44691020e-01 -1.06960750e+00
-2.56412089e-01 1.42486107e+00 -4.80711341e-01 3.40372920e-01
1.47307539e+00 -7.34399796e-01 -3.12461376e-01 1.18728302e-01
1.14386249e+00 -9.55365449e-02 -1.55967641e+00 -5.78788042e-01
-5.80808401e-01 -1.01895201e+00 3.74440312e-01 -7.65952408e-01
-1.32250965e+00 8.20013762e-01 7.43649602e-01 -3.51953991e-02
1.69689083e+00 9.93105024e-02 1.01747727e+00 -2.23573595e-01
5.66366434e-01 -7.77952194e-01 2.80040681e-01 2.29307801e-01
1.25344169e+00 -1.28664410e+00 -1.00284874e-01 -4.85901296e-01
-6.77406073e-01 1.14078891e+00 6.82449579e-01 -2.25166589e-01
2.77185172e-01 1.22270714e-02 -1.60231084e-01 6.92779869e-02
-5.47872424e-01 -2.70752758e-01 6.01000488e-01 6.32795870e-01
1.49369538e-01 -1.64080352e-01 -6.66663498e-02 6.94469273e-01
9.32915807e-02 1.70839224e-02 4.11737084e-01 2.39804894e-01
-4.60131347e-01 -9.02295411e-01 -3.27971697e-01 4.87919569e-01
-4.79412287e-01 -2.12506622e-01 -4.00360338e-02 4.74813014e-01
1.51177384e-02 7.97995508e-01 1.00783922e-01 -5.89946568e-01
3.13056916e-01 -6.22888148e-01 5.88562727e-01 -3.77753258e-01
-1.29903913e-01 9.89196151e-02 -5.44844508e-01 -9.53123927e-01
-4.69654649e-01 -4.02619481e-01 -1.06956100e+00 -1.47819161e-01
-2.03121543e-01 -2.53661722e-01 2.32620537e-01 7.70408094e-01
5.08667529e-01 7.41347790e-01 9.87179995e-01 -1.04830074e+00
-2.74617583e-01 -6.89495683e-01 -7.55330145e-01 5.53882301e-01
5.84029078e-01 -8.35613430e-01 -4.33389544e-01 -7.56513625e-02] | [10.99864673614502, -2.059796094894409] |
a18c714b-2d1b-40f2-b7ac-4a51706fa2aa | current-trends-in-deep-learning-for-earth | 2207.07189 | null | https://arxiv.org/abs/2207.07189v2 | https://arxiv.org/pdf/2207.07189v2.pdf | Current Trends in Deep Learning for Earth Observation: An Open-source Benchmark Arena for Image Classification | We present AiTLAS: Benchmark Arena -- an open-source benchmark suite for evaluating state-of-the-art deep learning approaches for image classification in Earth Observation (EO). To this end, we present a comprehensive comparative analysis of more than 500 models derived from ten different state-of-the-art architectures and compare them to a variety of multi-class and multi-label classification tasks from 22 datasets with different sizes and properties. In addition to models trained entirely on these datasets, we benchmark models trained in the context of transfer learning, leveraging pre-trained model variants, as it is typically performed in practice. All presented approaches are general and can be easily extended to many other remote sensing image classification tasks not considered in this study. To ensure reproducibility and facilitate better usability and further developments, all of the experimental resources including the trained models, model configurations, and processing details of the datasets (with their corresponding splits used for training and evaluating the models) are publicly available on the repository: https://github.com/biasvariancelabs/aitlas-arena | ['Nikola Simidjievski', 'Dragi Kocev', 'Ivan Kitanovski', 'Ivica Dimitrovski'] | 2022-07-14 | null | null | null | null | ['satellite-image-classification', 'remote-sensing-image-classification'] | ['computer-vision', 'miscellaneous'] | [ 1.95506424e-01 -4.53625679e-01 1.13077879e-01 -6.92642570e-01
-7.31730282e-01 -6.37699127e-01 6.45256221e-01 3.10251713e-01
-5.53065181e-01 5.79914808e-01 -2.32757881e-01 -3.98744851e-01
-3.21648121e-01 -7.32265294e-01 -5.99195659e-01 -1.03764451e+00
-6.56118631e-01 5.68810999e-01 -1.49984971e-01 -1.90579742e-01
1.58596374e-02 5.96680760e-01 -1.74678004e+00 5.00931382e-01
6.64214075e-01 1.28469729e+00 -7.08697811e-02 8.29011798e-01
4.78063554e-01 6.94033861e-01 -3.72410655e-01 -1.36680484e-01
4.53464687e-01 -1.53844669e-01 -8.67369652e-01 -3.82427424e-01
9.45829928e-01 -4.25484627e-01 -2.50454485e-01 9.58160222e-01
9.16988075e-01 1.07272640e-01 1.00372529e+00 -1.07933712e+00
-5.30515075e-01 2.94385761e-01 -3.63306701e-01 7.42890656e-01
-4.27844673e-01 1.48548141e-01 1.11613297e+00 -7.66033232e-01
3.07660908e-01 1.03772628e+00 8.20712268e-01 3.39065492e-01
-1.29120719e+00 -7.44302392e-01 5.97006939e-02 2.19810337e-01
-1.41014314e+00 -5.57522714e-01 3.24816555e-01 -8.35162222e-01
1.06912804e+00 4.69042778e-01 5.62753022e-01 1.04368949e+00
2.09850535e-01 6.68839276e-01 1.46657038e+00 -3.95655304e-01
7.08364621e-02 8.63698870e-02 4.38035399e-01 5.69016278e-01
3.19572054e-02 2.82384902e-01 7.46709034e-02 -3.33046466e-01
2.77349949e-01 -1.27141997e-01 -3.03905785e-01 -3.35053205e-01
-1.21816778e+00 9.87167835e-01 8.62411499e-01 3.16331595e-01
-4.80285496e-01 9.07085538e-02 5.06996393e-01 3.87862623e-01
1.11756444e+00 2.70086288e-01 -9.42134440e-01 4.66302156e-01
-1.01289141e+00 5.26088774e-01 5.95395207e-01 6.74483776e-01
9.62694049e-01 1.70212850e-01 -1.25428215e-01 8.98024321e-01
1.98347390e-01 7.33762205e-01 5.06525576e-01 -6.52688086e-01
2.02701181e-01 3.74479294e-01 1.11589963e-02 -7.23482549e-01
-7.18072295e-01 -7.24362612e-01 -9.22305584e-01 2.42955908e-01
6.49126172e-02 -2.30402932e-01 -1.01720273e+00 1.38302851e+00
2.00657770e-01 1.70437679e-01 4.17756550e-02 7.28382766e-01
1.52851737e+00 6.43661559e-01 1.88350186e-01 5.27565777e-01
1.40823269e+00 -1.08669615e+00 -9.67268348e-02 -5.34533560e-01
8.29657435e-01 -5.23387969e-01 9.49391484e-01 5.61302081e-02
-3.30925137e-01 -4.17592108e-01 -9.49668169e-01 1.85307879e-02
-8.18193793e-01 6.23685420e-01 5.90328217e-01 5.18445075e-01
-1.04811978e+00 5.92415571e-01 -8.89300346e-01 -4.49075341e-01
6.64866924e-01 6.14335611e-02 -4.06895190e-01 1.36487126e-01
-1.13992524e+00 9.95358527e-01 4.20990974e-01 3.38785529e-01
-1.46947217e+00 -8.21903408e-01 -7.14578927e-01 8.05982724e-02
-9.18727815e-02 -5.17808855e-01 1.56151116e+00 -1.01651096e+00
-1.17921495e+00 1.48496616e+00 1.22615196e-01 -3.67242843e-01
5.09409070e-01 -3.03324699e-01 -3.17984551e-01 -8.98054764e-02
1.16487563e-01 8.16154838e-01 6.82219565e-01 -1.12230361e+00
-6.18183613e-01 -3.47386718e-01 1.95328593e-02 -2.82548182e-02
-1.06462345e-01 2.37013429e-01 1.34424746e-01 -5.60344517e-01
-2.84209847e-01 -1.13315666e+00 -2.10258991e-01 4.68392819e-02
-2.10125506e-01 -8.31627026e-02 3.82932901e-01 -7.11355448e-01
9.10195827e-01 -2.22171998e+00 1.72948688e-01 -4.94260527e-03
1.21111780e-01 4.30735290e-01 -3.98904830e-01 4.60557222e-01
-4.80145633e-01 1.85436949e-01 -6.24966681e-01 -4.87710208e-01
-3.64653952e-02 -7.17564225e-02 -3.05198967e-01 7.67383993e-01
1.30034149e-01 7.89865136e-01 -8.75337362e-01 -9.45026577e-02
4.61900175e-01 3.37525249e-01 -8.64304155e-02 1.99458376e-01
-2.50679672e-01 7.36064017e-01 -3.42564046e-01 7.29720592e-01
8.01194072e-01 -3.59650820e-01 -1.27335951e-01 -5.99972382e-02
8.62611681e-02 2.42558837e-01 -9.05252337e-01 1.37327826e+00
-6.26021564e-01 8.51231456e-01 -2.20049247e-02 -1.13623440e+00
8.96746993e-01 5.96803874e-02 3.09025705e-01 -5.56223512e-01
1.85301438e-01 3.36665273e-01 3.94967496e-02 -4.12503600e-01
2.93284655e-01 -7.43413195e-02 1.49395615e-01 3.41514617e-01
2.80358791e-01 -1.01484612e-01 3.76850158e-01 -2.48412535e-01
5.86230159e-01 7.72528797e-02 5.97370386e-01 -6.45563185e-01
5.19524813e-01 1.49825051e-01 2.03433499e-01 8.02642882e-01
-2.81179994e-01 5.29452801e-01 8.20576996e-02 -9.97854590e-01
-9.68343973e-01 -7.83937097e-01 -9.46776092e-01 1.43298423e+00
-4.61128235e-01 -1.53291330e-01 -5.00453591e-01 -5.31731606e-01
1.26219019e-01 5.96567869e-01 -1.12584305e+00 3.96987498e-01
4.13343832e-02 -1.46157277e+00 8.98027778e-01 2.95718044e-01
5.64049423e-01 -1.07356799e+00 -7.28722632e-01 -5.62926829e-02
-2.82523930e-01 -8.82121980e-01 4.65951264e-01 3.08714449e-01
-9.27343786e-01 -1.11872482e+00 -7.44065523e-01 -4.35933948e-01
3.46746534e-01 1.35383055e-01 1.52500129e+00 2.79929578e-01
-3.31419379e-01 1.44302383e-01 -4.85342175e-01 -8.03188503e-01
-4.51872349e-01 6.04487538e-01 -3.70802194e-01 -2.06486136e-01
5.72556317e-01 -3.28393638e-01 -7.52723336e-01 1.30400270e-01
-1.02026772e+00 -1.53886378e-01 4.30189669e-01 8.92910123e-01
3.64755392e-01 -2.58477956e-01 1.32501125e-01 -7.71728098e-01
1.37489200e-01 -7.85129249e-01 -8.04095089e-01 3.01039338e-01
-5.02724707e-01 -1.86874986e-01 3.44753832e-01 1.45273000e-01
-8.02634597e-01 -1.74885705e-01 -3.67591530e-01 -1.15022771e-01
-7.55313158e-01 8.33669066e-01 4.03605700e-01 -2.64267862e-01
1.15086353e+00 1.19426146e-01 -2.34383166e-01 -6.85401022e-01
2.19805658e-01 1.11997581e+00 -2.79584271e-03 -3.22846681e-01
3.90148282e-01 6.15742028e-01 -1.78518851e-04 -7.89391279e-01
-1.26107478e+00 -6.46927536e-01 -6.97451651e-01 -1.45438701e-01
7.39295602e-01 -1.47469997e+00 -1.24669215e-02 9.07453060e-01
-7.04239249e-01 -8.92249286e-01 2.26240993e-01 4.40180093e-01
-2.71404773e-01 -5.17860278e-02 -5.18857121e-01 -6.10150754e-01
-6.70583844e-01 -1.16056108e+00 1.46434009e+00 -1.97111964e-02
2.71890432e-01 -1.21978152e+00 5.41697443e-01 1.67195976e-01
7.85366595e-01 3.88264567e-01 6.45817339e-01 -8.16286862e-01
-7.19412416e-02 -1.36092260e-01 -2.53583580e-01 5.22795081e-01
-5.40977493e-02 2.59606421e-01 -1.50884056e+00 -7.40880251e-01
-2.79849261e-01 -7.77941763e-01 1.49484444e+00 4.57824260e-01
1.15585017e+00 6.79850904e-03 -2.66177863e-01 9.05194819e-01
1.54084921e+00 -2.79734761e-01 7.37839937e-01 1.02943480e+00
4.01391983e-01 6.92997694e-01 5.04715919e-01 4.82280672e-01
3.34814370e-01 5.55567563e-01 6.76254094e-01 -4.36246246e-01
2.90345550e-01 4.34823036e-01 -2.90217251e-02 2.76469976e-01
-2.63476640e-01 -3.62461120e-01 -1.39315069e+00 5.07316887e-01
-1.65113842e+00 -9.60143387e-01 -3.24170530e-01 2.01039600e+00
5.47760606e-01 -6.20954752e-01 -1.13632016e-01 3.04134982e-03
4.24672246e-01 7.09230423e-01 -3.10129791e-01 -4.33933781e-03
-2.14518085e-01 1.48448989e-01 7.16611564e-01 4.34825271e-01
-1.90691662e+00 8.52653682e-01 6.53219795e+00 8.02119732e-01
-1.59496617e+00 2.91601479e-01 9.42115545e-01 -9.08899456e-02
1.44718498e-01 -1.95486158e-01 -8.23614478e-01 1.27768666e-01
1.13474369e+00 4.07723993e-01 3.56192768e-01 9.57068861e-01
2.77639963e-02 -5.47035597e-02 -1.00505424e+00 6.74983442e-01
-4.33123000e-02 -1.05909753e+00 -1.26356632e-01 -1.41951442e-01
8.84667754e-01 1.22461510e+00 -1.00605683e-02 3.88744801e-01
2.38313243e-01 -1.00421727e+00 7.32187867e-01 2.08001778e-01
8.83727968e-01 -1.58253878e-01 9.69069541e-01 1.31014585e-01
-9.19286191e-01 -1.58244416e-01 -7.22549975e-01 -1.25419378e-01
-2.86329001e-01 6.45550668e-01 -2.40414158e-01 1.05827475e+00
1.42085958e+00 1.02515519e+00 -9.70424235e-01 1.08418095e+00
-3.34895462e-01 8.45331550e-01 -4.40132350e-01 2.80325770e-01
4.41193253e-01 -1.07625440e-01 2.93905497e-01 1.50633490e+00
3.21565658e-01 -9.14160237e-02 -1.87815782e-02 6.10918403e-01
8.31527039e-02 2.58971065e-01 -5.25162935e-01 1.96636841e-02
3.07710826e-01 1.63025582e+00 -5.86121976e-01 -4.38028365e-01
-4.83756036e-01 5.10440290e-01 5.06093025e-01 3.34692836e-01
-8.80029023e-01 -2.16503859e-01 6.40796065e-01 -1.45854816e-01
3.17575842e-01 3.39915464e-03 -7.79151395e-02 -1.44364893e+00
-2.27832839e-01 -1.10213804e+00 6.40932083e-01 -8.24713349e-01
-1.50460970e+00 1.00846982e+00 5.41921735e-01 -1.33655167e+00
7.90894702e-02 -8.89609873e-01 -5.40432096e-01 8.71646404e-01
-2.17271256e+00 -1.35447526e+00 -8.80697966e-01 1.72435462e-01
3.24812651e-01 -3.14631045e-01 1.14585519e+00 5.22829771e-01
-6.27367318e-01 2.49149024e-01 7.32225776e-01 2.59945452e-01
9.05906498e-01 -1.04593253e+00 6.27186358e-01 7.10535169e-01
1.74520135e-01 2.93912619e-01 4.68112767e-01 -9.80110094e-02
-4.71122503e-01 -1.35931098e+00 5.55550218e-01 -4.00565714e-01
7.94882715e-01 -2.74806857e-01 -9.25142229e-01 1.23579574e+00
3.15157771e-01 3.30808759e-01 9.44595933e-01 2.23267823e-02
-3.10933173e-01 -1.84493870e-01 -8.17796230e-01 5.73209710e-02
6.89072013e-01 -2.94535279e-01 -3.69076580e-01 5.90695500e-01
-3.85786034e-02 -5.73243737e-01 -1.08553743e+00 7.71192491e-01
4.75397825e-01 -1.15311611e+00 8.67903769e-01 -7.82005012e-01
7.06503689e-01 -3.38349551e-01 -3.08876902e-01 -1.57843542e+00
-5.21454871e-01 9.86777022e-02 5.05151153e-01 7.48088479e-01
6.01963878e-01 -8.81530225e-01 1.33993909e-01 -1.58223048e-01
-2.62581646e-01 -6.91056192e-01 -9.06793177e-01 -6.90453708e-01
5.92280805e-01 -1.74916118e-01 7.22576439e-01 1.31160879e+00
-5.84098279e-01 2.20861375e-01 -2.99273610e-01 4.15213764e-01
5.96139431e-01 3.31150770e-01 8.45292687e-01 -1.42064023e+00
-2.86134213e-01 -4.96161729e-01 -2.74170786e-01 -6.84694588e-01
3.11329514e-01 -1.12218463e+00 1.43244624e-01 -1.43820202e+00
2.30661690e-01 -5.69287002e-01 -3.85483474e-01 8.73385906e-01
-3.13128471e-01 5.40365934e-01 5.92106953e-02 5.21579206e-01
-2.69700766e-01 7.14505196e-01 6.48578346e-01 -3.24029088e-01
2.02904940e-01 7.05803707e-02 -2.46957526e-01 4.80084836e-01
1.08677244e+00 -8.58883977e-01 -9.91889536e-02 -8.25872064e-01
5.87812178e-02 -5.35115361e-01 5.61910868e-01 -1.14400029e+00
-3.87708664e-01 1.33223608e-01 3.24557960e-01 -3.57930243e-01
-4.65446990e-03 -6.05636537e-01 3.73523176e-01 4.90373492e-01
-3.60219508e-01 1.88369621e-02 5.20067930e-01 2.10991502e-01
-3.18595797e-01 -3.60672325e-01 1.09755027e+00 -3.23626816e-01
-8.06246340e-01 4.54488426e-01 -4.07857746e-01 -1.02396481e-01
1.02390397e+00 3.24538410e-01 -7.36024261e-01 -1.41315088e-01
-8.28767598e-01 2.02940837e-01 3.22021961e-01 3.13530743e-01
3.41240168e-02 -9.19374108e-01 -1.33073938e+00 1.94681793e-01
6.85364723e-01 -3.21074575e-02 3.73468757e-01 1.01393116e+00
-8.96078944e-01 5.41506588e-01 -5.18623590e-01 -8.19896162e-01
-1.15247953e+00 2.15338841e-01 1.06380105e+00 -4.24849927e-01
-4.26008552e-01 5.62178910e-01 2.61412472e-01 -1.07623661e+00
-4.82876115e-02 -3.50056052e-01 -3.36928695e-01 1.06914066e-01
5.50861537e-01 3.93775076e-01 5.22417307e-01 -6.69539154e-01
-4.72601086e-01 4.16220725e-01 2.11786240e-01 2.12953508e-01
1.69575393e+00 -5.75916730e-02 -3.02289039e-01 7.20630467e-01
9.41660047e-01 -4.75800961e-01 -1.03734040e+00 -3.13179642e-01
-3.12587827e-01 -3.23883951e-01 4.78237063e-01 -9.20584619e-01
-1.21606576e+00 1.18960011e+00 1.09159911e+00 1.09559648e-01
1.00371110e+00 -1.66143492e-01 -1.86482631e-02 5.90143859e-01
2.32387334e-01 -6.52343452e-01 -4.04573321e-01 6.13828599e-01
1.11139667e+00 -1.72170520e+00 2.59918064e-01 -3.90926488e-02
-4.68167633e-01 1.12302554e+00 4.30940986e-01 -6.37262687e-02
1.08517921e+00 2.32158918e-02 5.48675835e-01 -4.16416228e-01
-6.01542115e-01 -4.15475637e-01 3.11590701e-01 5.15907705e-01
7.94294417e-01 2.02427357e-01 1.40610896e-02 1.35116383e-01
1.61565449e-02 -3.00243516e-02 3.34185302e-01 8.53151917e-01
-3.73507589e-01 -8.08951735e-01 -2.48108134e-01 7.13446915e-01
-7.27564394e-01 -4.61468637e-01 -2.88944036e-01 8.34244370e-01
-1.11005723e-01 7.52216160e-01 1.82501808e-01 -1.85110942e-01
1.58474013e-01 -2.63879579e-02 1.16525747e-01 -5.30714512e-01
-7.68481553e-01 -2.83771574e-01 2.97842681e-01 -3.29115510e-01
-9.10072565e-01 -7.29978085e-01 -6.20390534e-01 -3.35607141e-01
-4.04619128e-01 -3.94940898e-02 7.33406305e-01 8.36763501e-01
4.11178112e-01 3.19332570e-01 3.70908320e-01 -1.37999940e+00
-8.10863495e-01 -1.72402191e+00 -6.32530034e-01 2.78274685e-01
4.24129307e-01 -6.46338999e-01 -7.48916566e-01 -1.17962576e-01] | [9.534760475158691, -1.4226665496826172] |
2515f787-658c-4015-a18b-5e57203516b5 | generating-scenes-with-latent-object-models | null | null | https://openreview.net/forum?id=WTXMNULQ3Uu | https://openreview.net/pdf?id=WTXMNULQ3Uu | Generating Scenes with Latent Object Models | We introduce a structured latent variable model that learns the underlying data-generating process for a dataset of scenes. Our goals are to obtain a compositional scene representation and to perform scene generation by modeling statistical relationships between scenes as well as between objects within a scene. To make inference tractable, we take inspiration from visual topic models and introduce an interpretable hierarchy of scene-level and object-level latent variables (i.e., slots). Since generating scenes requires modeling dependencies between objects, we cannot make a bag-of-words assumption to simplify inference. Moreover, assuming that slots are generated with an autoregressive prior requires decomposing scenes sequentially during inference which has known limitations. Our approach is to assume that the assignment of objects to slots during generation is a deterministic function of the scene latent variable. This removes the need for sequential scene decomposition and enables us to propose an inference algorithm that uses orderless scene decomposition to indirectly estimate an ordered slot posterior. Qualitative and quantitative analysis establishes that our approach successfully learns a smoothly traversable scene-level latent space. The hierarchy of scene and slot variables improves the ability of slot-based models to generate samples displaying complex object relations. We also demonstrate that the learned hierarchy of representations can be used for a scene-retrieval application with object-centric re-ranking. | ['Anand Rangarajan', 'Sanjay Ranka', 'Pan He', 'Patrick Emami'] | 2021-09-29 | null | null | null | null | ['scene-generation'] | ['computer-vision'] | [ 5.30800104e-01 3.00202638e-01 -1.01448961e-01 -4.81212735e-01
-6.22165143e-01 -6.98451221e-01 1.09231198e+00 4.78005297e-02
1.16641343e-01 3.30387414e-01 4.53307509e-01 -2.51947194e-01
-2.43032560e-01 -1.09858906e+00 -9.84588206e-01 -6.61095679e-01
1.08815446e-01 8.68587077e-01 1.05312802e-01 1.37375474e-01
1.14362314e-01 6.53559923e-01 -1.97011793e+00 6.89229310e-01
4.52464610e-01 3.78708005e-01 6.37249112e-01 7.75340736e-01
-5.04873276e-01 9.09583330e-01 -6.13570452e-01 -1.38893262e-01
2.62140423e-01 -4.64345336e-01 -8.06649804e-01 8.20919514e-01
6.27579093e-01 -3.65329921e-01 -1.33720577e-01 6.51584089e-01
-1.79641962e-01 3.05676162e-01 1.08323431e+00 -1.23960710e+00
-4.98095363e-01 4.69503194e-01 -3.30538124e-01 -2.66736478e-01
2.49188960e-01 -1.42714262e-01 1.32513690e+00 -7.83075094e-01
9.23110127e-01 1.51128221e+00 1.15104273e-01 4.10183549e-01
-1.80855334e+00 -8.50028545e-02 6.11451387e-01 9.45346281e-02
-1.17792928e+00 -4.07573909e-01 1.03346860e+00 -8.29050124e-01
7.26227760e-01 4.64434981e-01 8.14276457e-01 1.01105273e+00
3.04109566e-02 8.76948535e-01 9.26665604e-01 -6.80823803e-01
5.01741707e-01 3.76431793e-01 9.13884565e-02 5.55309117e-01
1.83748335e-01 -4.25586164e-01 -6.64293289e-01 -1.06740646e-01
1.12948453e+00 1.78387389e-01 2.40484759e-01 -9.93791461e-01
-1.13382339e+00 9.74537313e-01 2.70653814e-01 -3.00584920e-02
-4.43698108e-01 3.64845484e-01 6.42709658e-02 -1.35623842e-01
6.17435694e-01 4.74444866e-01 -2.33657077e-01 4.01591957e-01
-1.06322217e+00 4.66177911e-01 6.29124522e-01 1.25825143e+00
1.00114119e+00 7.14057218e-03 -2.43064761e-01 7.82042801e-01
4.06152248e-01 2.44916588e-01 -6.98313070e-03 -1.15915787e+00
1.92934826e-01 5.54254472e-01 1.78129286e-01 -1.10474455e+00
-6.06174654e-05 -2.09687188e-01 -4.62824464e-01 6.52262717e-02
1.80866525e-01 4.41599220e-01 -1.21110475e+00 1.71373272e+00
3.44582707e-01 -1.36780664e-02 -6.79161996e-02 5.26752889e-01
5.46693206e-01 8.44991505e-01 4.16115582e-01 -2.79498696e-01
1.47620893e+00 -7.19328284e-01 -6.09753788e-01 -1.72981352e-01
3.83964211e-01 -5.10919631e-01 1.39546955e+00 2.23031357e-01
-1.18446398e+00 -5.81497848e-01 -7.06209540e-01 -4.61369514e-01
-3.93416077e-01 4.17021662e-02 1.11633956e+00 2.09620968e-01
-1.04626977e+00 8.50739181e-02 -1.01341438e+00 -4.27808851e-01
4.66969490e-01 4.89545539e-02 -2.91666359e-01 -3.69164348e-02
-6.14898920e-01 5.71582377e-01 2.96083421e-01 -2.42176875e-01
-1.51392019e+00 -7.24339187e-01 -1.04302454e+00 1.48555413e-01
3.64647239e-01 -1.20227575e+00 1.13967216e+00 -8.42264235e-01
-1.20648348e+00 9.25686717e-01 -7.10001886e-01 -2.74851054e-01
1.74893767e-01 -1.45500630e-01 2.49517232e-01 4.34664220e-01
4.53891039e-01 1.07783449e+00 9.64800954e-01 -1.94120908e+00
-5.62923074e-01 -2.40516424e-01 5.90515375e-01 5.71407497e-01
-9.16215032e-02 -2.49553159e-01 -6.53733432e-01 -4.65705425e-01
5.71077526e-01 -7.55788922e-01 -3.80213350e-01 1.20344177e-01
-5.74771702e-01 -1.38403356e-01 6.33796334e-01 -3.24651510e-01
7.88377166e-01 -2.20242500e+00 4.07998919e-01 7.60167688e-02
3.69160593e-01 -8.19919646e-01 -2.23743469e-01 5.65687776e-01
-2.76432216e-01 1.11812457e-01 -2.03285307e-01 -8.16519678e-01
8.99936110e-02 5.33038616e-01 -8.32415164e-01 2.91125536e-01
1.70356169e-01 8.34282935e-01 -9.09619331e-01 -6.75134718e-01
6.21961832e-01 4.65891421e-01 -9.19218659e-01 2.86371976e-01
-7.87316084e-01 2.57343829e-01 -3.40028614e-01 2.70668805e-01
4.34618413e-01 -4.34287369e-01 1.49302080e-01 -2.99020350e-01
-1.28103286e-01 5.60365081e-01 -1.16976273e+00 1.77905297e+00
-6.17436230e-01 7.53959000e-01 -2.20842421e-01 -1.00861728e+00
6.26055062e-01 1.56707451e-01 5.28391361e-01 -2.29650185e-01
-3.37701708e-01 -4.32533801e-01 -5.70223749e-01 -3.74383390e-01
6.62252009e-01 -5.32388270e-01 -2.47183934e-01 3.72008085e-01
1.95751145e-01 -6.81840360e-01 3.47945780e-01 8.39499116e-01
8.30165148e-01 3.39866519e-01 2.54007615e-02 -2.89605618e-01
-2.75031012e-02 3.54104847e-01 2.56379217e-01 9.46018279e-01
5.86770236e-01 5.61729729e-01 7.83323109e-01 -3.78279954e-01
-1.25790918e+00 -1.65538263e+00 -1.68823242e-01 1.23007810e+00
1.57049596e-01 -5.49783170e-01 -5.09684682e-01 -2.99037397e-01
-2.27219298e-01 1.20231247e+00 -7.77482390e-01 1.69746920e-01
-1.86872840e-01 -4.78283465e-01 -1.41819879e-01 3.38388354e-01
2.33009011e-02 -9.44136322e-01 -7.62565136e-01 4.53628749e-02
-3.52748573e-01 -1.00504577e+00 9.08233151e-02 2.62942076e-01
-8.90207827e-01 -7.99149752e-01 -3.09182972e-01 -6.76870346e-01
1.22700870e+00 4.27030504e-01 1.37084496e+00 -3.36135656e-01
-5.13406992e-01 9.84289527e-01 -2.53854305e-01 -3.37375879e-01
-2.47467741e-01 -2.25629687e-01 -1.95763111e-01 1.43722706e-02
4.91955504e-02 -7.38141119e-01 -4.43024069e-01 1.61329973e-02
-1.17012942e+00 7.99416900e-01 3.18584144e-01 6.06449723e-01
8.04884136e-01 5.08961916e-01 -2.83771425e-01 -1.05343258e+00
2.07103819e-01 -4.67019349e-01 -7.38713384e-01 3.36749017e-01
1.24112792e-01 2.37054572e-01 2.40921199e-01 -4.13506091e-01
-1.37375653e+00 3.57516289e-01 6.34950697e-01 -4.65692848e-01
-4.13681626e-01 3.85533094e-01 -3.66973668e-01 7.86351740e-01
5.96969426e-01 3.03682238e-01 -4.04772252e-01 -3.86563241e-01
9.02471960e-01 -1.74657293e-02 3.51447850e-01 -9.10852492e-01
9.07430470e-01 9.77062404e-01 1.92417353e-01 -1.02567613e+00
-1.05293417e+00 -4.60644096e-01 -8.21410775e-01 -1.63940012e-01
1.12070847e+00 -1.23674929e+00 -4.73659277e-01 -7.85558950e-04
-1.36177385e+00 -5.34023225e-01 -7.49247134e-01 2.63095260e-01
-8.55472684e-01 1.00070976e-01 -4.16791111e-01 -9.44491923e-01
5.19314945e-01 -9.29096520e-01 1.52681088e+00 -2.90516436e-01
-5.12779415e-01 -1.15135360e+00 -1.12926051e-01 3.20897251e-01
-6.39654137e-03 1.83627501e-01 1.21681869e+00 7.61542330e-03
-1.31340742e+00 1.56796411e-01 -9.49736685e-02 -1.93600073e-01
3.42009455e-01 2.53006399e-01 -9.92701113e-01 4.94760368e-03
1.01109497e-01 -1.83806762e-01 8.81009221e-01 6.64195657e-01
1.38039362e+00 -5.10377288e-01 -3.75493973e-01 4.42592978e-01
1.40070450e+00 -2.63639186e-02 5.87907076e-01 4.48158421e-02
9.37212646e-01 1.09020746e+00 2.39480346e-01 4.55118865e-01
6.24123037e-01 7.15769231e-01 1.77595973e-01 -2.07761586e-01
-1.35754019e-01 -6.56909406e-01 8.85466337e-02 4.73233342e-01
2.79021502e-01 -2.75274456e-01 -8.99623215e-01 7.42997050e-01
-1.77353203e+00 -1.12315011e+00 -2.69652516e-01 1.98383760e+00
7.85064936e-01 4.47608307e-02 -1.33923396e-01 -1.49881780e-01
3.93024802e-01 1.16922379e-01 -3.71946514e-01 -9.42772180e-02
-1.28407753e-03 3.56003828e-02 2.34144762e-01 9.24018383e-01
-1.03081262e+00 1.23422623e+00 6.95448112e+00 7.49896348e-01
-6.31974638e-01 -5.31870052e-02 7.26041496e-01 -2.05984503e-01
-1.15049672e+00 4.61320400e-01 -8.48880649e-01 -6.72621932e-03
4.74994898e-01 -1.09667920e-01 4.78656858e-01 1.01614571e+00
2.72686332e-01 -3.36547971e-01 -1.43555915e+00 8.36366534e-01
1.11766435e-01 -1.43009269e+00 7.42506623e-01 2.08902895e-01
7.57094026e-01 -5.74774921e-01 1.30135909e-01 1.60027191e-01
7.80793786e-01 -9.60791349e-01 1.04250097e+00 5.85232615e-01
6.51085794e-01 -3.19483042e-01 -1.45317897e-01 2.85355747e-01
-1.15913415e+00 1.64090827e-01 -4.68733549e-01 -3.14793706e-01
2.62027562e-01 7.88274050e-01 -9.58022535e-01 2.06803486e-01
3.50913823e-01 5.39396286e-01 -5.14503598e-01 6.24543726e-01
-4.15663362e-01 5.13595998e-01 -3.69163036e-01 3.53588730e-01
-6.64484575e-02 -1.82141632e-01 4.89627242e-01 1.06980848e+00
-6.45152107e-03 4.59733158e-02 2.43398547e-01 1.44639421e+00
3.37314188e-01 -4.41341028e-02 -8.14476490e-01 4.15249020e-02
3.05048019e-01 1.27128291e+00 -1.19341624e+00 -6.70368612e-01
-7.18635842e-02 9.58063245e-01 2.22490191e-01 7.07408249e-01
-6.33539379e-01 3.39537203e-01 5.02689242e-01 3.57592344e-01
2.75230706e-01 -6.67032659e-01 -6.11408710e-01 -1.25755775e+00
-1.57641158e-01 -5.26894927e-01 8.19325298e-02 -1.16504002e+00
-1.15511048e+00 2.50406414e-01 7.93916464e-01 -1.09033346e+00
-4.20572609e-01 -3.93686324e-01 -2.83244669e-01 9.16706920e-01
-1.06594610e+00 -1.45279396e+00 -2.70103991e-01 6.64812386e-01
9.14638817e-01 2.61749864e-01 8.50476325e-01 -2.41698653e-01
-1.53324857e-01 -8.73914137e-02 -3.08083922e-01 -1.62663028e-01
1.76656753e-01 -1.35838270e+00 2.86596209e-01 7.99420178e-01
6.01706326e-01 1.12148666e+00 9.75482404e-01 -7.92904079e-01
-1.20402992e+00 -8.95532489e-01 8.26778710e-01 -9.14470971e-01
4.83079344e-01 -9.72767293e-01 -5.79497099e-01 8.49185050e-01
-1.69354111e-01 -3.12467664e-01 7.50973284e-01 4.43340719e-01
-4.44227755e-01 8.60074826e-04 -6.21521473e-01 9.47948217e-01
1.09120488e+00 -9.14666057e-01 -5.50826192e-01 5.12387216e-01
8.55530143e-01 -2.31517851e-01 -2.78994411e-01 2.26450965e-01
3.98012787e-01 -8.55238438e-01 1.10228598e+00 -6.17085993e-01
7.61883974e-01 -4.45492387e-01 -3.91845047e-01 -8.93952847e-01
-5.66116393e-01 -1.54663017e-02 3.11847422e-02 1.20335507e+00
3.25148940e-01 -8.85372013e-02 9.30199146e-01 1.00392032e+00
1.36459747e-03 -3.13349128e-01 -5.68195343e-01 -4.48233664e-01
-1.91078350e-01 -7.81369448e-01 2.78271198e-01 9.23946500e-01
-3.25574487e-01 5.66355884e-01 -1.91819653e-01 3.34986508e-01
7.57940888e-01 3.28526586e-01 1.06451786e+00 -1.08987427e+00
-6.68805122e-01 -1.58110812e-01 -3.06492656e-01 -1.39747047e+00
1.98327988e-01 -6.00762606e-01 2.52533883e-01 -2.03102708e+00
6.08662963e-01 -5.95733404e-01 9.60889086e-03 4.34105456e-01
7.80352503e-02 -3.52868736e-02 2.47375757e-01 4.05510068e-01
-6.16524339e-01 6.18466139e-01 1.17734885e+00 -1.64922744e-01
-3.06882322e-01 -2.24129543e-01 -7.06535280e-01 8.01411510e-01
4.22460765e-01 -4.34469074e-01 -1.05546951e+00 -4.99142855e-01
3.66486311e-01 2.73506194e-02 7.29007781e-01 -5.64419091e-01
8.15461483e-03 -4.41357017e-01 5.81845582e-01 -7.93596506e-01
7.64633417e-01 -8.23534548e-01 5.75119197e-01 -6.18621036e-02
-6.92265034e-01 -4.05138344e-01 1.30078301e-01 5.10757208e-01
-2.79334243e-02 -2.24426743e-02 3.03710639e-01 -4.58126336e-01
-5.98505557e-01 1.09111480e-01 -6.09612346e-01 -4.03760165e-01
8.82933378e-01 -4.14290577e-01 4.37203534e-02 -5.33335865e-01
-9.93774056e-01 -2.28430610e-02 5.52814782e-01 3.79730284e-01
5.60052931e-01 -1.33815145e+00 -3.75764489e-01 2.41515696e-01
2.72272974e-01 4.05173659e-01 3.70662123e-01 1.07685618e-01
-3.05509478e-01 3.73260915e-01 -2.14576982e-02 -9.89975810e-01
-1.16495383e+00 5.17274916e-01 -5.76795340e-02 -1.50620669e-01
-4.79732454e-01 9.83409047e-01 1.19330776e+00 -2.81316489e-01
1.44125104e-01 -3.83492619e-01 4.06501591e-02 2.19987258e-01
1.23549752e-01 -6.29036129e-02 -5.40987790e-01 -5.78648508e-01
-1.09681608e-02 3.96714598e-01 -6.48540333e-02 -6.20620549e-01
1.32872891e+00 -2.70633459e-01 -2.42934197e-01 1.10176671e+00
9.22115862e-01 -3.40540931e-02 -1.47509873e+00 -4.72164160e-04
-2.63317794e-01 -6.33837044e-01 -1.11654885e-01 -3.44023615e-01
-3.56644183e-01 1.02349949e+00 5.81981950e-02 2.60567188e-01
8.95260155e-01 6.02804661e-01 -5.72655685e-02 2.49436632e-01
4.62948620e-01 -7.59350121e-01 3.87125224e-01 2.68811345e-01
1.03514135e+00 -8.10215235e-01 2.35378727e-01 -8.19119275e-01
-4.75663573e-01 8.99232268e-01 4.70356703e-01 4.18078527e-02
5.32779455e-01 7.89383948e-02 -2.03405127e-01 -4.89834726e-01
-9.27278936e-01 -3.20213497e-01 4.80338514e-01 4.77343470e-01
3.59408200e-01 2.84960568e-01 1.73498392e-01 -8.40636268e-02
-3.94709080e-01 -3.07824641e-01 4.41068739e-01 6.86715186e-01
-4.40565377e-01 -1.12275660e+00 -2.78716207e-01 3.84798855e-01
2.34393552e-02 -2.67225116e-01 -2.27948204e-01 6.26857579e-01
2.58339107e-01 7.39427388e-01 4.68992859e-01 1.19169943e-01
3.16798910e-02 -2.28613969e-02 6.72724068e-01 -1.26436281e+00
4.04836208e-01 4.01522309e-01 -5.04548363e-02 -4.70720083e-01
-5.94626427e-01 -7.60224462e-01 -1.06461966e+00 9.55903232e-02
-1.04060441e-01 9.76813808e-02 7.42773592e-01 8.89347732e-01
1.93425223e-01 6.40882611e-01 4.33557898e-01 -1.00451326e+00
2.04374015e-01 -5.71068287e-01 -5.99583209e-01 7.35365570e-01
2.58111805e-01 -7.82572925e-01 -3.57736140e-01 8.89054298e-01] | [10.285720825195312, 0.21697717905044556] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.