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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
e7d274de-3f56-4485-ba4c-7a36650acd63 | journeydb-a-benchmark-for-generative-image | 2307.00716 | null | https://arxiv.org/abs/2307.00716v1 | https://arxiv.org/pdf/2307.00716v1.pdf | JourneyDB: A Benchmark for Generative Image Understanding | While recent advancements in vision-language models have revolutionized multi-modal understanding, it remains unclear whether they possess the capabilities of comprehending the generated images. Compared to real data, synthetic images exhibit a higher degree of diversity in both content and style, for which there are significant difficulties for the models to fully apprehend. To this end, we present a large-scale dataset, JourneyDB, for multi-modal visual understanding in generative images. Our curated dataset covers 4 million diverse and high-quality generated images paired with the text prompts used to produce them. We further design 4 benchmarks to quantify the performance of generated image understanding in terms of both content and style interpretation. These benchmarks include prompt inversion, style retrieval, image captioning and visual question answering. Lastly, we assess the performance of current state-of-the-art multi-modal models when applied to JourneyDB, and provide an in-depth analysis of their strengths and limitations in generated content understanding. We hope the proposed dataset and benchmarks will facilitate the research in the field of generative content understanding. The dataset will be available on https://journeydb.github.io. | ['Hongsheng Li', 'Yu Qiao', 'Jifeng Dai', 'Yi Wang', 'Zipeng Qin', 'Aojun Zhou', 'Renrui Zhang', 'Xiaoshi Wu', 'Haodong Duan', 'Hao Li', 'Yuying Ge', 'Keqiang Sun', 'Junting Pan'] | 2023-07-03 | null | null | null | null | ['visual-question-answering-1', 'image-captioning', 'retrieval', 'question-answering'] | ['computer-vision', 'computer-vision', 'methodology', 'natural-language-processing'] | [ 3.58096600e-01 2.30198905e-01 1.33814469e-01 -6.23517573e-01
-1.07073414e+00 -8.49734366e-01 9.29495931e-01 -1.96211070e-01
1.16667099e-01 4.54072833e-01 5.69548786e-01 -2.55283266e-01
2.16828749e-01 -7.70193040e-01 -1.07754135e+00 -2.84129739e-01
5.41662395e-01 6.07386649e-01 -1.80521369e-01 -2.81863242e-01
2.48796225e-01 1.23853842e-03 -1.79223835e+00 9.51444209e-01
1.00690162e+00 1.00367594e+00 4.67543900e-01 9.43573833e-01
-3.86122435e-01 1.05708575e+00 -6.44053996e-01 -6.34864748e-01
-1.67348012e-01 -4.79979128e-01 -1.02018940e+00 3.13036323e-01
8.75677049e-01 -5.75908542e-01 -2.51919717e-01 8.96471858e-01
4.74295974e-01 -2.15550885e-01 9.99883473e-01 -1.52839804e+00
-1.48953617e+00 5.55746377e-01 -3.02554846e-01 -5.09651341e-02
8.18594694e-01 5.14988244e-01 8.59451413e-01 -1.12185907e+00
8.33641350e-01 1.59089780e+00 1.33712798e-01 6.99010789e-01
-1.10885966e+00 -5.03533185e-01 9.19885784e-02 2.95725256e-01
-1.07166898e+00 -5.27722299e-01 6.76917195e-01 -5.27934134e-01
6.21838808e-01 2.48337612e-01 5.40629983e-01 1.75437593e+00
-8.14254954e-02 1.13646901e+00 1.42566144e+00 -3.19017798e-01
-4.56019156e-02 3.48714083e-01 -4.45088707e-02 5.47498941e-01
-1.84438750e-01 1.22087926e-01 -7.75562406e-01 3.25647622e-01
7.16351926e-01 -3.85338366e-01 -3.15207511e-01 -2.19631776e-01
-1.46467173e+00 9.25434351e-01 3.96998227e-01 -2.32557282e-01
-2.73547083e-01 1.98312744e-01 1.51972249e-01 2.02180952e-01
4.72233385e-01 2.66628534e-01 -3.19047309e-02 -1.84785128e-01
-8.00103545e-01 4.36444908e-01 7.30326831e-01 1.31338501e+00
7.05104709e-01 8.87185559e-02 -4.84138608e-01 7.84441054e-01
3.94125193e-01 1.06490934e+00 1.56763166e-01 -1.09586763e+00
4.96947497e-01 5.87993860e-01 1.04755990e-01 -1.04709172e+00
-1.90579109e-02 -1.65613323e-01 -7.94405520e-01 -1.08984476e-02
3.04668278e-01 9.22380984e-02 -1.01660967e+00 1.66184413e+00
5.60327657e-02 -1.82153080e-02 2.41840973e-01 9.46461737e-01
1.55212307e+00 8.96634936e-01 2.61014163e-01 3.76545757e-01
1.53248131e+00 -1.19592607e+00 -6.21754766e-01 -5.23636818e-01
1.11296788e-01 -1.17022836e+00 1.56677532e+00 1.22757174e-01
-1.24086547e+00 -8.23946595e-01 -5.88182211e-01 -3.31994832e-01
-4.83272016e-01 1.34439528e-01 5.39719284e-01 5.18328011e-01
-1.30136037e+00 -1.24153234e-01 -2.77435809e-01 -4.01501447e-01
5.58679283e-01 -4.91340697e-01 -1.53920904e-01 -3.54391843e-01
-1.17863429e+00 7.24664450e-01 3.80877763e-01 -1.36455014e-01
-1.35591233e+00 -1.02094710e+00 -9.78017986e-01 -3.24671775e-01
1.51537970e-01 -1.15619195e+00 1.56167018e+00 -1.14721215e+00
-1.01244140e+00 1.23592317e+00 -2.91373819e-01 -2.83744216e-01
6.78726435e-01 -2.39428759e-01 -4.40154642e-01 3.55066568e-01
3.94883633e-01 1.40523481e+00 9.60192561e-01 -1.86281788e+00
-4.28145021e-01 -1.72051936e-01 4.18467551e-01 3.41628909e-01
-9.87215936e-02 -1.51897997e-01 -6.63798511e-01 -7.80275762e-01
-2.30410054e-01 -8.31427455e-01 1.72300160e-01 1.58880740e-01
-5.87673306e-01 -3.28122303e-02 8.78309846e-01 -6.54119015e-01
7.69675255e-01 -1.94307959e+00 1.81060806e-01 -3.38831991e-01
2.35698983e-01 -3.00083645e-02 -5.39706469e-01 5.75389266e-01
1.17299862e-01 2.56363571e-01 -4.71577346e-02 -5.29229999e-01
2.80901324e-02 2.37240165e-01 -7.67927170e-01 -2.46977270e-01
3.38451207e-01 1.37293720e+00 -9.50048327e-01 -5.59433281e-01
4.40075308e-01 6.16688728e-01 -4.00952905e-01 4.98272657e-01
-7.33332515e-01 5.15185773e-01 -4.47682112e-01 7.40573466e-01
6.77520573e-01 -6.97998047e-01 -2.96034843e-01 -5.03810942e-01
1.71085551e-01 -1.20062165e-01 -8.15265715e-01 1.81797254e+00
-6.13144279e-01 9.57294226e-01 -3.76794904e-01 -3.67816836e-01
7.32535481e-01 1.02034494e-01 -1.00705504e-01 -1.02519906e+00
-1.27904966e-01 -1.68913186e-01 -5.85271060e-01 -7.01872289e-01
8.64227176e-01 1.24752857e-01 -2.03726590e-01 6.56187952e-01
2.94096246e-02 -5.15964150e-01 2.43869916e-01 6.97573066e-01
3.77679139e-01 1.53819397e-01 -1.26244456e-01 -1.76128335e-02
2.00358361e-01 2.85810530e-01 -2.76372463e-01 9.10164118e-01
4.38140258e-02 9.09405351e-01 3.14990580e-01 -4.13213104e-01
-1.15323412e+00 -1.54689419e+00 1.31801784e-01 1.15798104e+00
3.76548439e-01 -3.41719359e-01 -9.57409978e-01 -3.11320454e-01
-7.96791762e-02 1.02308881e+00 -8.35344195e-01 1.19704455e-02
-3.04251518e-02 -5.20164311e-01 6.88448608e-01 5.19836724e-01
7.67416954e-01 -1.41089761e+00 -5.15059948e-01 -2.00357318e-01
-8.50177288e-01 -1.43360555e+00 -3.91200870e-01 -5.88335931e-01
-5.14175177e-01 -1.03314149e+00 -7.60720074e-01 -7.23708749e-01
7.42793620e-01 5.85643649e-01 1.89843285e+00 -1.77108347e-01
-2.79573709e-01 1.00295687e+00 -5.38554728e-01 -6.29046142e-01
-8.70530605e-01 -1.22277744e-01 -4.63486731e-01 -2.18812446e-03
1.64766908e-01 -1.65448666e-01 -7.89987028e-01 3.21109742e-01
-1.38360393e+00 7.96792030e-01 6.68553710e-01 5.51816642e-01
5.89098513e-01 -5.40967405e-01 5.54987073e-01 -8.01565409e-01
9.54320967e-01 -5.75095892e-01 -9.86343324e-02 5.52829146e-01
-3.19328874e-01 3.30484360e-02 3.17898035e-01 -3.70070368e-01
-1.47572339e+00 -2.44527370e-01 4.10948806e-02 -4.08111632e-01
-4.39037412e-01 3.59709412e-01 -4.43718284e-02 9.49188098e-02
7.63443887e-01 5.62221348e-01 -1.08517177e-01 -3.35857213e-01
1.03134489e+00 6.55381382e-01 7.31491089e-01 -7.07459927e-01
7.04404116e-01 5.36036253e-01 -5.08502424e-01 -8.33525956e-01
-1.17313802e+00 -4.66747992e-02 -1.97223142e-01 -5.92193246e-01
9.96642709e-01 -1.20296264e+00 -4.80514079e-01 7.23820329e-01
-1.43641651e+00 -3.77158731e-01 -1.00971684e-01 -3.15124094e-01
-8.63627732e-01 3.31071496e-01 -3.17880064e-01 -5.43468297e-01
-4.04563665e-01 -1.18551433e+00 1.51616871e+00 3.66179585e-01
-3.68487448e-01 -1.02011883e+00 -1.41697958e-01 1.07642126e+00
6.32151186e-01 2.81236172e-01 9.76735711e-01 -5.17476797e-02
-9.08176899e-01 3.98434520e-01 -7.07920134e-01 1.76112786e-01
-1.62924320e-01 6.55338243e-02 -1.15101612e+00 1.10359853e-02
-3.58154356e-01 -9.25548017e-01 8.71264815e-01 3.47601116e-01
1.29717481e+00 -2.02205792e-01 -1.21524907e-03 5.16166866e-01
1.40883338e+00 -5.34804761e-02 8.41253817e-01 2.30058521e-01
8.57302606e-01 7.50769734e-01 6.72620118e-01 3.05406362e-01
1.01026952e+00 4.74040657e-01 6.09896600e-01 -3.06599975e-01
-6.30181313e-01 -6.42272592e-01 2.41446167e-01 5.71874559e-01
1.93524703e-01 -6.80469990e-01 -1.05127287e+00 7.52773404e-01
-1.72534227e+00 -9.17976439e-01 -2.27417380e-01 1.57062852e+00
9.28400218e-01 -1.97376549e-01 -7.02982917e-02 -3.31451744e-01
5.09548962e-01 3.29514027e-01 -6.94213212e-01 -4.76263314e-01
-5.40153682e-01 -3.19409519e-01 2.68829241e-02 3.90873700e-01
-8.69722486e-01 1.09773636e+00 6.63621759e+00 9.18308794e-01
-8.62082958e-01 -6.91816956e-02 9.92771208e-01 1.26285151e-01
-8.46834838e-01 -9.69494134e-02 -6.79847240e-01 3.63702595e-01
8.12197626e-01 -1.02732457e-01 4.42744017e-01 8.37812722e-01
2.42794648e-01 -2.45303437e-01 -1.10612214e+00 1.19546378e+00
5.01717865e-01 -1.49879777e+00 8.53716433e-01 -3.89944702e-01
9.23930228e-01 -1.16947949e-01 5.35046220e-01 1.63673222e-01
2.17279270e-01 -1.24051392e+00 1.04895329e+00 8.20304096e-01
1.09082103e+00 -3.98861438e-01 2.54330784e-01 1.20560534e-01
-9.44351733e-01 2.28292540e-01 -8.43999162e-02 1.36357009e-01
5.16771913e-01 3.77356261e-01 -7.95552731e-01 3.39508653e-01
6.50175273e-01 6.62709713e-01 -1.03549862e+00 5.37039042e-01
-3.09288353e-01 3.67050827e-01 1.83752850e-01 7.35940561e-02
2.39668384e-01 -2.90545542e-03 2.96503752e-01 1.15354180e+00
2.21194744e-01 -1.31759137e-01 -5.50542958e-02 1.35886157e+00
-4.64685932e-02 -5.14600016e-02 -6.22979820e-01 -3.85149866e-01
4.04024452e-01 1.10998952e+00 -3.80134612e-01 -5.60784340e-01
-3.94255608e-01 1.03944910e+00 1.77251205e-01 5.66468954e-01
-1.03445315e+00 6.33351728e-02 6.52791083e-01 9.56879258e-02
-9.89681408e-02 -9.37393606e-02 -4.00303662e-01 -1.19590259e+00
-7.84113407e-02 -1.26015437e+00 2.80935913e-01 -1.74303734e+00
-1.43776083e+00 8.03441048e-01 2.73904860e-01 -9.41863656e-01
-4.93040890e-01 -6.38255298e-01 -3.04636508e-01 7.52940476e-01
-1.56954622e+00 -1.78471279e+00 -1.02179778e+00 7.01000333e-01
1.10649562e+00 -5.33184558e-02 8.82685006e-01 1.74308699e-02
-1.73818439e-01 3.56041968e-01 -1.98015198e-01 -1.93901937e-02
1.01542568e+00 -1.11542416e+00 7.58593917e-01 5.81895590e-01
2.86666870e-01 3.24046791e-01 9.12655413e-01 -6.03337288e-01
-1.42404354e+00 -1.23969305e+00 5.44961214e-01 -7.71059453e-01
5.34405231e-01 -3.64236504e-01 -6.89175010e-01 7.08572507e-01
6.07887328e-01 -5.01569867e-01 6.24382734e-01 -2.27319241e-01
-5.93276620e-01 2.78632998e-01 -9.08911765e-01 8.86669517e-01
1.14710343e+00 -6.70786798e-01 -5.05584121e-01 3.43555152e-01
8.78841758e-01 -4.50735241e-01 -7.27634907e-01 2.19956145e-01
5.81159651e-01 -1.19546616e+00 1.25486422e+00 -5.71334839e-01
1.20478189e+00 -1.71234995e-01 -2.19118208e-01 -1.41651523e+00
-1.43156257e-02 -3.26877445e-01 -4.70614657e-02 1.30107820e+00
5.66869676e-01 -2.13546544e-01 5.30979753e-01 6.06144905e-01
-5.93483560e-02 -5.42433679e-01 -2.20369637e-01 -3.59785974e-01
4.13601585e-02 -5.22906661e-01 5.98777711e-01 6.57215774e-01
-5.86410642e-01 5.57459593e-01 -4.73414958e-01 -8.50388110e-02
7.73831427e-01 5.44847429e-01 1.02837431e+00 -7.78013706e-01
-1.48249745e-01 -2.81373739e-01 -6.01110868e-02 -1.15785563e+00
1.70133829e-01 -6.61401629e-01 -2.63285190e-01 -1.85999584e+00
5.79981804e-01 -1.41547561e-01 3.26479048e-01 2.66593367e-01
-2.79296190e-01 5.36324620e-01 3.48634034e-01 1.21904485e-01
-8.69772255e-01 5.91369271e-01 1.80503619e+00 -2.85498917e-01
1.66493580e-01 -5.13193488e-01 -1.05431759e+00 5.88229775e-01
9.61006641e-01 6.43087924e-02 -7.98741758e-01 -1.00721157e+00
4.30452734e-01 1.11191325e-01 7.72658110e-01 -6.78861737e-01
1.33173950e-02 -2.02714339e-01 4.95332032e-01 -8.19465637e-01
6.47806704e-01 -3.44244897e-01 3.47244233e-01 -5.82172908e-02
-5.50377846e-01 2.38954931e-01 2.97572076e-01 5.55411518e-01
-4.15590495e-01 1.50137588e-01 5.46405673e-01 -2.40842253e-01
-1.17159128e+00 2.18473613e-01 -2.13042662e-01 5.16965866e-01
9.17904615e-01 -3.14528078e-01 -8.43265593e-01 -1.10401380e+00
-5.67876995e-01 3.10625404e-01 7.11730719e-01 1.00477791e+00
1.03549552e+00 -1.46442056e+00 -9.58370745e-01 5.47579192e-02
7.02545166e-01 -1.49971187e-01 7.90923953e-01 2.91590452e-01
-4.22859311e-01 4.19593096e-01 -4.70662445e-01 -8.00824523e-01
-1.18217623e+00 4.07219827e-01 1.17221974e-01 1.76116601e-01
-2.79838443e-01 7.54010558e-01 7.27378845e-01 -2.97791481e-01
-6.12173751e-02 -1.95087388e-01 -8.30428153e-02 6.44273683e-02
7.24540234e-01 1.87139630e-01 -4.14108604e-01 -8.50021124e-01
4.10598479e-02 6.05911076e-01 1.16660669e-02 -2.38501951e-01
9.69014645e-01 -4.84833181e-01 8.92651209e-04 3.65030438e-01
1.10559702e+00 -4.97364223e-01 -1.23606861e+00 -1.33892655e-01
-5.06822824e-01 -5.28395116e-01 -1.56060979e-01 -1.20424187e+00
-8.62160385e-01 9.41281736e-01 4.80953962e-01 1.06611259e-01
1.21079278e+00 3.26574296e-01 8.24313402e-01 1.48026213e-01
2.11849079e-01 -8.26729894e-01 6.31921649e-01 5.36381423e-01
1.40291202e+00 -1.64539015e+00 -4.87887830e-01 -5.28934836e-01
-1.16393292e+00 8.59996021e-01 7.88013577e-01 2.51008421e-01
2.08872885e-01 1.30807031e-02 5.01333296e-01 -3.08535397e-01
-8.41868401e-01 -1.64497286e-01 5.82799971e-01 9.06480670e-01
3.29376221e-01 2.19145387e-01 3.18980128e-01 3.94865096e-01
-5.36619961e-01 -9.20170918e-02 6.11293018e-01 3.41338843e-01
-2.61694312e-01 -7.92116761e-01 -4.01461750e-01 2.93563187e-01
-8.40253085e-02 -2.87679851e-01 -6.70832515e-01 5.92194259e-01
-1.83258384e-01 1.23401344e+00 -1.07353320e-02 -1.08001433e-01
2.35849142e-01 -7.89245963e-02 6.12082303e-01 -4.27597106e-01
-9.50569883e-02 -4.03556108e-01 3.04318577e-01 -7.06961870e-01
-5.28424263e-01 -3.05667639e-01 -8.41645420e-01 -3.37224215e-01
2.71264464e-01 -3.65108192e-01 8.14217448e-01 6.56911969e-01
5.46838224e-01 5.42207837e-01 1.76944539e-01 -7.81359971e-01
-2.48644248e-01 -9.17685509e-01 -3.26341391e-02 7.96371579e-01
1.66097924e-01 -5.01972139e-01 -2.60714233e-01 5.74461699e-01] | [11.031265258789062, 1.3205134868621826] |
b7ceb855-15ca-4e61-8870-bf9d15898a59 | densely-deformable-efficient-salient-object | 2102.06407 | null | https://arxiv.org/abs/2102.06407v1 | https://arxiv.org/pdf/2102.06407v1.pdf | Densely Deformable Efficient Salient Object Detection Network | Salient Object Detection (SOD) domain using RGB-D data has lately emerged with some current models' adequately precise results. However, they have restrained generalization abilities and intensive computational complexity. In this paper, inspired by the best background/foreground separation abilities of deformable convolutions, we employ them in our Densely Deformable Network (DDNet) to achieve efficient SOD. The salient regions from densely deformable convolutions are further refined using transposed convolutions to optimally generate the saliency maps. Quantitative and qualitative evaluations using the recent SOD dataset against 22 competing techniques show our method's efficiency and effectiveness. We also offer evaluation using our own created cross-dataset, surveillance-SOD (S-SOD), to check the trained models' validity in terms of their applicability in diverse scenarios. The results indicate that the current models have limited generalization potentials, demanding further research in this direction. Our code and new dataset will be publicly available at https://github.com/tanveer-hussain/EfficientSOD | ['Sung Wook Baik', 'Khan Muhammad', 'Amin Ullah', 'Saeed Anwar', 'Tanveer Hussain'] | 2021-02-12 | null | null | null | null | ['rgb-d-salient-object-detection', 'salient-object-detection'] | ['computer-vision', 'computer-vision'] | [ 3.25434297e-01 2.88890123e-01 7.53892213e-02 -3.50087762e-01
-2.93140382e-01 -4.12589341e-01 4.68441039e-01 -4.00227755e-01
-3.84801537e-01 7.53995299e-01 2.40624279e-01 -1.75528765e-01
-3.41630429e-02 -5.39113581e-01 -6.66635752e-01 -7.79690981e-01
-9.12651345e-02 -1.58454463e-01 9.77397919e-01 -5.26050448e-01
1.61747620e-01 7.34994233e-01 -1.74799955e+00 2.24061921e-01
1.06721449e+00 1.03952765e+00 5.73383033e-01 5.75061083e-01
2.15793952e-01 7.12558448e-01 -3.30625266e-01 -3.08352321e-01
5.92504680e-01 -1.58266649e-01 -8.47801566e-01 -2.22312436e-02
4.59565520e-01 -5.55387080e-01 -4.09535676e-01 1.02790725e+00
6.39298558e-01 1.88444689e-01 1.47761479e-01 -1.35838163e+00
-1.07232320e+00 2.21823692e-01 -4.21023369e-01 8.28847885e-01
7.37123936e-02 2.73483962e-01 6.38452351e-01 -9.07430410e-01
7.08456814e-01 1.10844052e+00 7.69554675e-01 8.78940046e-01
-8.48216236e-01 -5.64801276e-01 3.47953320e-01 1.30314261e-01
-1.18500912e+00 -4.69905466e-01 7.90960431e-01 -1.82318807e-01
9.20366764e-01 5.57479680e-01 6.12943292e-01 1.21995842e+00
5.56370504e-02 9.92979944e-01 1.19004881e+00 -1.92770869e-01
2.24305078e-01 3.83742154e-01 -1.57235667e-01 8.70808125e-01
3.80415618e-01 1.93684742e-01 -3.84761691e-01 1.68395966e-01
9.99193668e-01 9.99851674e-02 -3.15461725e-01 -5.72353721e-01
-9.68282938e-01 6.99961960e-01 9.48933184e-01 2.84689575e-01
-3.75004739e-01 -5.88191450e-02 -1.85208395e-03 -1.51850164e-01
7.38039613e-01 3.98930460e-01 -4.51280147e-01 8.19800273e-02
-9.77033675e-01 3.35182160e-01 3.31270099e-01 9.73963261e-01
5.50758123e-01 1.91992521e-01 -3.07286590e-01 7.03159750e-01
2.29905993e-01 4.54668611e-01 3.86849284e-01 -7.25259900e-01
2.73755997e-01 8.22468102e-01 2.78260916e-01 -1.15866983e+00
-5.57220519e-01 -3.60845476e-01 -5.86854100e-01 4.90362793e-01
3.14614803e-01 -3.21114093e-01 -1.19909883e+00 1.47904396e+00
4.55179572e-01 2.67593920e-01 3.34833711e-02 1.41406429e+00
1.15096664e+00 3.97120804e-01 3.80439281e-01 3.25207710e-01
1.09396601e+00 -9.92479086e-01 -4.44412887e-01 -3.84637386e-01
3.31177056e-01 -5.97028077e-01 1.01587451e+00 9.91760492e-02
-1.19546747e+00 -5.67919314e-01 -9.08972561e-01 -1.88219652e-01
-5.69578886e-01 2.67530680e-01 9.48639274e-01 6.96316302e-01
-1.38026047e+00 5.76920271e-01 -1.04298747e+00 -6.68736637e-01
9.08934116e-01 4.53386515e-01 -9.00936201e-02 2.74744362e-01
-1.24381924e+00 1.03057528e+00 4.61933762e-01 4.12375122e-01
-9.90333021e-01 -5.82872808e-01 -6.30700052e-01 -1.66387513e-01
2.36952290e-01 -4.66864109e-01 1.09441090e+00 -1.23830247e+00
-1.27033436e+00 9.08495724e-01 1.06498405e-01 -6.60478950e-01
5.18164158e-01 -3.19530398e-01 -4.21051472e-01 3.01895797e-01
-5.16817570e-02 1.16076636e+00 7.02855289e-01 -1.28847146e+00
-7.43638992e-01 -8.40843618e-02 3.96969944e-01 2.47223809e-01
-4.10748065e-01 3.41213763e-01 -2.14023918e-01 -8.69839549e-01
6.02718517e-02 -1.01441133e+00 -2.85226017e-01 3.68348122e-01
-3.00165951e-01 6.87466562e-03 1.04927397e+00 -7.32100189e-01
1.06897759e+00 -2.04390764e+00 -1.71503954e-04 -3.04509819e-01
2.42094189e-01 8.27938914e-01 -1.06540710e-01 1.36169776e-01
2.00820761e-03 -7.06722289e-02 -4.67302680e-01 -1.50505826e-01
-4.55124676e-02 1.62283450e-01 -3.01535040e-01 4.93783325e-01
8.80501628e-01 1.16774011e+00 -8.91819298e-01 -5.36373913e-01
3.73603374e-01 5.37481725e-01 -4.26788837e-01 6.57460913e-02
-1.04485162e-01 6.56220913e-02 -5.87896526e-01 9.93141055e-01
8.73902619e-01 -2.79421955e-01 -3.86349142e-01 -2.77422279e-01
-3.70834857e-01 -1.20525889e-01 -9.93864119e-01 1.50126660e+00
1.29560724e-01 8.51668954e-01 -1.49435631e-03 -9.16875184e-01
8.06123018e-01 -7.33075812e-02 2.70463347e-01 -8.70214462e-01
1.14886574e-01 1.47514418e-01 -5.86770065e-02 -7.14368880e-01
7.21691608e-01 5.71265258e-02 2.60036349e-01 -7.77440146e-02
-2.08171410e-03 -8.48247483e-02 5.29185347e-02 1.81217581e-01
8.65039587e-01 5.28637409e-01 1.83965974e-02 -5.82504869e-01
3.15507948e-01 3.23115528e-01 4.35651511e-01 6.99280202e-01
-7.76662886e-01 9.91336524e-01 9.73419473e-02 -5.63139439e-01
-8.43916655e-01 -1.18284118e+00 -2.22766533e-01 1.11551118e+00
6.22904778e-01 2.67550915e-01 -7.34909296e-01 -7.60703504e-01
-1.02292918e-01 5.06406724e-01 -8.85505497e-01 -8.28123391e-02
-5.66970348e-01 -8.78455520e-01 5.33547342e-01 7.36071885e-01
8.36856484e-01 -1.53209543e+00 -1.20086217e+00 5.79773299e-02
-3.05153038e-02 -1.02010822e+00 -1.70097098e-01 1.39677808e-01
-8.42359900e-01 -1.03317189e+00 -1.10158992e+00 -8.54602039e-01
7.18239307e-01 6.02800667e-01 7.86149144e-01 7.42678531e-03
-4.76952940e-01 3.09625059e-01 -5.12873352e-01 -7.63586283e-01
-4.93157320e-02 -2.82300748e-02 -7.04714283e-02 -2.09132731e-01
2.16692224e-01 -2.17381358e-01 -1.01957810e+00 5.00250041e-01
-1.17340934e+00 2.64135420e-01 7.71660447e-01 3.94858837e-01
2.85725385e-01 -2.92203486e-01 5.29452443e-01 -4.12822008e-01
4.54451650e-01 -5.51984549e-01 -5.85614145e-01 1.89872012e-01
-3.06277931e-01 -1.62622601e-01 5.89407422e-02 -3.99248779e-01
-1.40785038e+00 5.89212812e-02 -2.45120339e-02 -3.69564593e-01
-3.61560732e-01 7.70124868e-02 2.09611118e-01 -4.15448755e-01
7.76920438e-01 1.72985673e-01 -1.30777836e-01 -4.24794614e-01
6.38075843e-02 5.51463723e-01 5.36515951e-01 -3.11044455e-01
7.88605154e-01 8.22081089e-01 -4.07355785e-01 -5.24687111e-01
-9.82961416e-01 -2.34212890e-01 -5.90681553e-01 -2.95960933e-01
9.80221927e-01 -9.19626892e-01 -2.06886947e-01 6.37305796e-01
-8.97605181e-01 -5.84732413e-01 -2.88883179e-01 2.71254718e-01
-2.61445284e-01 1.83292001e-01 -4.57575172e-01 -7.62353718e-01
-3.95694792e-01 -1.07970619e+00 1.27530706e+00 8.23752820e-01
2.09769197e-02 -1.03008270e+00 -1.32047564e-01 3.37799132e-01
7.61172295e-01 5.07685184e-01 2.04454809e-01 -4.87994969e-01
-6.79953039e-01 4.31347638e-02 -6.72736108e-01 4.50873256e-01
1.98990628e-01 4.77014296e-02 -1.20390737e+00 -2.91644484e-01
-2.25914806e-01 -1.27180398e-01 8.84931505e-01 4.68471974e-01
1.18613780e+00 -2.33375862e-01 -3.80088151e-01 5.13573587e-01
1.37104058e+00 7.74243400e-02 7.49993801e-01 7.38277793e-01
6.22034371e-01 5.45371890e-01 7.69699872e-01 3.80616665e-01
2.79793054e-01 4.61420804e-01 7.95340240e-01 -6.21513426e-01
-4.89931583e-01 4.55173925e-02 4.08911228e-01 1.56189129e-01
-4.58442867e-01 -3.18202674e-01 -9.30278003e-01 8.18319678e-01
-1.77247155e+00 -1.05769002e+00 -8.18798095e-02 1.70199800e+00
6.17585540e-01 1.27167627e-01 2.98594385e-01 -2.28048921e-01
7.53744304e-01 2.20561743e-01 -5.92966735e-01 -4.05556262e-01
-4.41863507e-01 3.23226810e-01 5.66948116e-01 1.66674942e-01
-1.37587833e+00 1.04445446e+00 5.87698364e+00 7.14085281e-01
-1.26377988e+00 5.10150343e-02 6.95538998e-01 -2.41542622e-01
-8.41246620e-02 -1.37210459e-01 -8.71701181e-01 5.56513250e-01
6.67214692e-01 9.18843299e-02 8.09424520e-02 7.97390461e-01
2.98568368e-01 -4.34849560e-01 -5.17337203e-01 4.90230829e-01
-3.83075178e-02 -1.44710243e+00 -6.62356168e-02 -1.36883527e-01
9.09443080e-01 2.70838857e-01 2.95665532e-01 2.95509309e-01
3.41480464e-01 -9.08694029e-01 9.25325096e-01 5.29202819e-01
3.52007568e-01 -3.62500191e-01 8.28175306e-01 7.91546144e-03
-9.60874319e-01 -2.35304192e-01 -5.44023633e-01 -1.34603813e-01
-3.94098870e-02 1.99367791e-01 -8.50926161e-01 4.86202806e-01
1.19329166e+00 7.45279849e-01 -1.06089830e+00 1.22986662e+00
4.77795377e-02 5.78487515e-01 -3.04220438e-01 -1.65361434e-01
6.56612277e-01 5.36382711e-03 6.08996511e-01 1.41208076e+00
2.66815156e-01 2.23064989e-01 -9.96733010e-02 1.04335535e+00
3.32175434e-01 -1.86390996e-01 -4.15273428e-01 1.98692873e-01
1.68257192e-01 1.41242731e+00 -1.11868489e+00 -2.90640652e-01
-4.50659573e-01 9.51389611e-01 2.57866919e-01 3.85879040e-01
-1.26196241e+00 -2.16342449e-01 7.31610000e-01 1.73068732e-01
6.11743987e-01 -2.49062367e-02 -1.60505742e-01 -9.71689522e-01
7.43596032e-02 -5.63037038e-01 3.56408805e-01 -1.19754040e+00
-1.15982890e+00 7.06032515e-01 2.72806883e-01 -1.28899121e+00
3.86274070e-01 -6.88857734e-01 -4.27179486e-01 7.46329844e-01
-1.69708276e+00 -1.32488453e+00 -6.91597402e-01 5.99940717e-01
5.36534965e-01 7.40591437e-02 3.94696951e-01 2.41962373e-01
-6.74302101e-01 3.73050809e-01 -1.48899078e-01 1.40392393e-01
4.47066635e-01 -1.00869155e+00 4.48659420e-01 1.15118074e+00
-2.42805749e-01 3.87681007e-01 7.77365744e-01 -6.39420927e-01
-1.00657642e+00 -1.22597349e+00 4.12201792e-01 -5.37504375e-01
6.83491468e-01 -3.07558924e-01 -9.87882853e-01 5.91947258e-01
3.17442209e-01 3.23459178e-01 1.33414581e-01 -5.49956024e-01
7.67512619e-02 1.34813368e-01 -1.46226215e+00 7.76076198e-01
1.38969529e+00 3.24255764e-03 -6.08948112e-01 2.13882998e-01
7.92791724e-01 -6.77142441e-01 -5.30783415e-01 6.99202120e-01
3.79609406e-01 -1.28129888e+00 1.08884263e+00 -4.57956940e-01
4.72185671e-01 -4.32453275e-01 -6.20082691e-02 -9.20337439e-01
-3.97867769e-01 -4.07399505e-01 -2.33168118e-02 1.14372456e+00
1.65153980e-01 -7.66664624e-01 8.08738291e-01 8.41809273e-01
-4.52062398e-01 -9.17054951e-01 -8.44303846e-01 -7.63702393e-01
-1.90151379e-01 -1.85598552e-01 5.42868376e-01 8.87255490e-01
-4.09308046e-01 -3.82468879e-01 -1.59276426e-01 4.60822076e-01
3.59936625e-01 -7.73069405e-05 3.89888853e-01 -1.04796243e+00
1.38842136e-01 -4.52453643e-01 -5.70770025e-01 -6.43960595e-01
-2.47528985e-01 -6.38532996e-01 8.83297250e-03 -1.51892722e+00
1.64568961e-01 -6.12980902e-01 -4.27510887e-01 8.47408891e-01
-3.92554283e-01 5.86726546e-01 2.31915161e-01 -5.55031970e-02
-8.01569760e-01 5.78213334e-01 1.49165487e+00 1.03481546e-01
-3.70936841e-01 -5.05302586e-02 -7.95799851e-01 7.03221679e-01
1.20192230e+00 -2.77804881e-01 -3.59096020e-01 -3.65574509e-01
-2.54834026e-01 -4.66757953e-01 9.78331745e-01 -1.21317899e+00
-4.90698311e-03 -3.56136531e-01 6.17972493e-01 -3.80616248e-01
3.40519398e-01 -6.64330125e-01 7.63564557e-02 4.55380231e-01
-1.48092419e-01 2.93689724e-02 7.99975991e-01 3.21386456e-01
2.92037260e-02 -5.43380007e-02 8.02034616e-01 -3.90588231e-02
-1.29875135e+00 3.20496053e-01 -1.75442562e-01 4.51561883e-02
1.24251020e+00 -7.45574057e-01 -4.26618010e-01 4.08627503e-02
-7.16017425e-01 2.70640794e-02 5.92832625e-01 7.88539588e-01
7.25192130e-01 -1.09656823e+00 -5.74492693e-01 2.13114053e-01
6.68096542e-03 4.23924811e-02 3.84312689e-01 8.83014023e-01
-6.74379826e-01 4.12654161e-01 -6.52959943e-01 -5.36492586e-01
-1.24113059e+00 5.54112196e-01 4.05763328e-01 1.61939710e-01
-7.33440042e-01 1.09717536e+00 3.71885538e-01 -1.83556363e-01
6.41464069e-02 -6.71716988e-01 -1.38684437e-01 -1.86740339e-01
3.38850260e-01 4.71371800e-01 -3.78169827e-02 -5.82194209e-01
-6.68467104e-01 2.34564364e-01 -5.10866083e-02 2.19901636e-01
1.60287082e+00 -1.65744320e-01 1.60446271e-01 -1.48088142e-01
8.45504820e-01 -2.24290952e-01 -1.84389520e+00 -2.20202953e-02
3.08905076e-02 -6.12343550e-01 1.10819347e-01 -7.85794199e-01
-1.27635562e+00 5.28025091e-01 1.06530976e+00 3.25306833e-01
1.31935132e+00 1.83280066e-01 5.48015594e-01 4.46820678e-03
2.79596746e-01 -9.73446846e-01 3.99443470e-02 1.37967184e-01
1.02553570e+00 -1.47926414e+00 5.11100441e-02 -3.77174973e-01
-8.62516940e-01 8.00356209e-01 8.83572578e-01 -3.61139476e-01
4.75408375e-01 2.84794152e-01 9.05567184e-02 -3.32461089e-01
-3.67660165e-01 -5.84764540e-01 4.29354131e-01 8.96318495e-01
2.57309347e-01 -1.10561438e-01 2.45073773e-02 3.05233538e-01
-2.94591431e-02 1.37214586e-01 5.36202669e-01 1.14350724e+00
-5.11228442e-01 -5.21131694e-01 -3.05844754e-01 2.74843305e-01
-3.55768085e-01 -1.36065900e-01 -4.38373297e-01 1.14647269e+00
2.05845341e-01 8.97236645e-01 -9.16792750e-02 -1.13757469e-01
3.86302263e-01 -2.68814415e-01 4.03631687e-01 -4.78201151e-01
-5.74627697e-01 -3.39601874e-01 -6.41788393e-02 -6.72522128e-01
-9.29263592e-01 -7.71866441e-01 -1.24607098e+00 -1.95888028e-01
-2.93623418e-01 -2.31563166e-01 4.52495366e-01 6.69385314e-01
6.06203318e-01 5.90206861e-01 2.45387897e-01 -1.21535826e+00
-2.93195218e-01 -7.68016756e-01 -3.00397396e-01 3.58015537e-01
4.76758689e-01 -9.09214020e-01 -2.68169492e-01 1.98898941e-01] | [9.776304244995117, -0.4616542160511017] |
db25b031-cc35-4f69-85c9-900d3b767d1f | deformirisnet-an-identity-preserving-model-of | 2207.08980 | null | https://arxiv.org/abs/2207.08980v2 | https://arxiv.org/pdf/2207.08980v2.pdf | DeformIrisNet: An Identity-Preserving Model of Iris Texture Deformation | Nonlinear iris texture deformations due to pupil size variations are one of the main factors responsible for within-class variance of genuine comparison scores in iris recognition. In dominant approaches to iris recognition, the size of a ring-shaped iris region is linearly scaled to a canonical rectangle, used further in encoding and matching. However, the biological complexity of the iris sphincter and dilator muscles causes the movements of iris features to be nonlinear in a function of pupil size, and not solely organized along radial paths. Alternatively to the existing theoretical models based on the biomechanics of iris musculature, in this paper we propose a novel deep autoencoder-based model that can effectively learn complex movements of iris texture features directly from the data. The proposed model takes two inputs, (a) an ISO-compliant near-infrared iris image with initial pupil size, and (b) the binary mask defining the target shape of the iris. The model makes all the necessary nonlinear deformations to the iris texture to match the shape of the iris in an image (a) with the shape provided by the target mask (b). The identity-preservation component of the loss function helps the model in finding deformations that preserve identity and not only the visual realism of the generated samples. We also demonstrate two immediate applications of this model: better compensation for iris texture deformations in iris recognition algorithms, compared to linear models, and the creation of a generative algorithm that can aid human forensic examiners, who may need to compare iris images with a large difference in pupil dilation. We offer the source codes and model weights available along with this paper. | ['Adam Czajka', 'Patrick Tinsley', 'Siamul Karim Khan'] | 2022-07-18 | null | null | null | null | ['pupil-dilation'] | ['computer-vision'] | [ 2.41975278e-01 3.11305914e-02 -1.53314963e-01 -2.89572597e-01
7.86607563e-02 -5.14398456e-01 1.59038514e-01 -3.46465111e-01
-1.45094037e-01 2.74045199e-01 2.97788709e-01 -1.91234633e-01
-5.15764177e-01 -6.62031531e-01 -6.74371183e-01 -9.48499799e-01
5.84992906e-03 3.40921074e-01 -3.64136696e-01 -8.47596750e-02
4.70136076e-01 9.02209580e-01 -1.93030751e+00 2.52256840e-01
8.72673929e-01 8.92925918e-01 -3.33194971e-01 8.25152993e-01
1.81270644e-01 3.63620311e-01 -5.35516858e-01 -2.61669844e-01
7.83219278e-01 -7.74264872e-01 -7.56060421e-01 5.09568036e-01
9.88676846e-01 -5.33423424e-01 -3.74611080e-01 1.10975814e+00
8.10101390e-01 1.90139323e-01 5.54392099e-01 -4.44114506e-01
-8.13780606e-01 2.32337162e-01 -6.41508281e-01 4.95962426e-02
2.79805541e-01 4.49340641e-01 3.92099112e-01 -3.92112762e-01
7.16267943e-01 9.54150438e-01 5.61321914e-01 8.37487161e-01
-1.30359101e+00 -3.18649083e-01 -5.95643878e-01 -7.42679387e-02
-1.30454075e+00 -4.91593510e-01 6.45516455e-01 -5.35909176e-01
5.80954552e-01 6.52870834e-01 9.44106460e-01 5.68782151e-01
1.43669054e-01 2.32722059e-01 1.44044256e+00 -7.04227924e-01
-3.15165251e-01 2.35269561e-01 -2.02756345e-01 6.14865124e-01
-7.62030408e-02 8.43392372e-01 -1.69291362e-01 -5.56866825e-02
1.22264516e+00 -1.90615788e-01 -3.19351196e-01 -2.29739413e-01
-8.83647621e-01 5.37007093e-01 4.53790694e-01 5.06091297e-01
-2.98863202e-01 -2.92151392e-01 -1.09976381e-01 3.46918494e-01
6.85490146e-02 7.78843582e-01 2.41697226e-02 -5.87831438e-02
-9.20368493e-01 9.58818868e-02 5.30109644e-01 3.20577264e-01
4.97473747e-01 1.73713211e-02 -7.80200213e-02 7.30011165e-01
2.04283237e-01 2.75416166e-01 7.65400410e-01 -6.30702496e-01
-1.90140352e-01 9.17865396e-01 -1.82516202e-01 -9.67597365e-01
-2.91506946e-01 -6.33195460e-01 -7.19151020e-01 6.41999245e-01
8.51634383e-01 2.13492420e-02 -9.45187390e-01 1.39133525e+00
6.63741171e-01 2.03698561e-01 -1.01755897e-03 1.15143013e+00
8.02928686e-01 1.26079082e-01 -5.01671314e-01 -2.04403270e-02
1.13079238e+00 -4.38670129e-01 -3.18504632e-01 1.93572760e-01
3.91992003e-01 -1.21615422e+00 8.23725700e-01 1.09244905e-01
-1.25755405e+00 -6.53604090e-01 -8.03471386e-01 -1.34062693e-01
-9.50418785e-02 4.82073247e-01 3.98639560e-01 8.00563514e-01
-1.03239787e+00 7.47783661e-01 -7.30511844e-01 -3.78734440e-01
1.63409874e-01 7.68228292e-01 -4.00120616e-01 3.75842273e-01
-6.65422320e-01 8.55817080e-01 1.78860351e-01 2.83493131e-01
4.18739617e-02 -7.26381123e-01 -7.38857388e-01 -8.56256709e-02
-3.00879508e-01 -7.53212154e-01 6.20279491e-01 -1.45788884e+00
-1.77629590e+00 1.33804297e+00 -1.44581854e-01 -2.70336926e-01
4.55600262e-01 2.50991911e-01 -3.21396142e-01 2.20006388e-02
-5.21551192e-01 5.54096282e-01 1.09946680e+00 -1.06610835e+00
-2.17758611e-01 -5.87546825e-01 -1.55482367e-01 1.56833917e-01
-1.29870325e-02 2.00052679e-01 -1.81700096e-01 -7.50930965e-01
2.01136455e-01 -9.80971217e-01 5.30075468e-02 9.81982052e-02
-3.58298600e-01 7.81160817e-02 6.06179118e-01 -8.79564703e-01
1.23949587e+00 -2.35526943e+00 3.63824740e-02 5.45570016e-01
1.80850685e-01 6.09112203e-01 -2.66443640e-01 4.39652726e-02
-5.02107680e-01 -7.30831772e-02 -5.07039800e-02 9.41928104e-03
-3.24901015e-01 5.76065034e-02 -1.83056816e-01 7.63749361e-01
9.14054140e-02 8.66480350e-01 -4.59377885e-01 -2.21634299e-01
3.91289562e-01 6.15910590e-01 -4.37863350e-01 1.61779121e-01
2.67503649e-01 6.54640317e-01 -1.17896318e-01 7.33944833e-01
8.26904893e-01 3.13123055e-02 -9.32362080e-02 -3.55230540e-01
-6.14789315e-02 3.45923826e-02 -1.18933272e+00 1.26896036e+00
-2.61491314e-02 7.09034741e-01 -4.67515737e-02 -7.10990906e-01
1.07876849e+00 3.22495461e-01 4.91708100e-01 -7.48251140e-01
3.36570799e-01 2.13124454e-01 6.28192544e-01 -5.45020103e-01
2.72704214e-01 -2.96868831e-01 8.58920753e-01 6.38980269e-01
-1.19843744e-01 -3.25915776e-02 5.38227297e-02 -6.21056318e-01
2.92754561e-01 2.67476439e-01 2.10232079e-01 -4.21459638e-02
5.97408712e-01 -2.10838675e-01 3.17475140e-01 4.81472254e-01
-1.03820287e-01 8.70701194e-01 4.05627012e-01 -1.15024388e+00
-1.37256253e+00 -6.62226200e-01 -7.17916131e-01 1.85393453e-01
-2.40013748e-03 2.71278858e-01 -1.05334604e+00 -1.86118871e-01
1.85774371e-01 1.81598619e-01 -8.89255583e-01 -1.28360137e-01
-5.41110992e-01 -7.97026813e-01 6.48075044e-01 9.73716304e-02
2.35204086e-01 -9.07606244e-01 -6.93449020e-01 -1.49619251e-01
2.43133068e-01 -5.07127345e-01 -5.43105066e-01 -5.30560791e-01
-1.08309698e+00 -1.37751484e+00 -5.53284824e-01 -7.20889568e-01
1.28514600e+00 -2.46534109e-01 6.27570331e-01 4.61828023e-01
-9.00852263e-01 1.59025148e-01 3.88052082e-03 -3.98057252e-01
-6.38419449e-01 -4.86710787e-01 2.42839545e-01 5.94763935e-01
5.70304871e-01 -4.04913396e-01 -6.76685333e-01 3.68223637e-01
-8.76348078e-01 9.38207284e-03 5.08535683e-01 1.10107577e+00
6.72797084e-01 1.39120430e-01 -3.11494023e-01 -3.86849791e-01
4.78122413e-01 9.87668633e-02 -7.02998161e-01 1.13763973e-01
-7.88270891e-01 -1.41094094e-02 2.97387630e-01 -8.30451190e-01
-7.33332932e-01 5.48614003e-02 1.17286958e-01 -6.07493401e-01
-3.29153061e-01 1.01451509e-01 3.72169137e-01 -7.56750762e-01
1.12515295e+00 4.76302236e-01 7.45022893e-01 -4.40287560e-01
4.31071781e-02 7.08212078e-01 6.31922424e-01 -4.84427631e-01
8.28592181e-01 5.52342474e-01 3.02610546e-01 -7.63684630e-01
-1.25247866e-01 -3.01574081e-01 -7.59316325e-01 -1.28568351e-01
5.18696904e-01 -3.84855598e-01 -1.06746745e+00 7.16478586e-01
-8.56312037e-01 -1.41641051e-01 -7.60712326e-01 8.04140449e-01
-5.78653216e-01 3.04104120e-01 -5.63066542e-01 -5.72220206e-01
-3.51217359e-01 -1.29609895e+00 7.34117568e-01 6.84382617e-01
3.10738645e-02 -1.08861911e+00 3.55386078e-01 5.24585545e-01
4.32608068e-01 3.78793418e-01 1.06149983e+00 -3.50521058e-01
-5.37720799e-01 -4.30838197e-01 -6.17043488e-02 6.36219442e-01
3.38401943e-01 4.42319393e-01 -1.12970674e+00 -4.58816767e-01
2.26439223e-01 -6.84615076e-02 4.55740780e-01 7.48388946e-01
9.53714788e-01 -5.24544895e-01 -3.84034477e-02 1.26496637e+00
1.38830125e+00 3.73496562e-01 1.04591465e+00 2.65923023e-01
4.41081822e-01 1.00624073e+00 6.94606779e-03 1.42176509e-01
-2.83887953e-01 6.82563186e-01 1.79134145e-01 -7.09955513e-01
-4.52173501e-01 -9.95108187e-02 -6.06093965e-02 3.91449094e-01
-7.06543148e-01 3.58908534e-01 -7.74688721e-01 4.88940954e-01
-1.24071908e+00 -1.07883835e+00 5.17930556e-03 2.68820977e+00
1.05396545e+00 -4.96521771e-01 4.36747409e-02 -3.27767469e-02
7.39126623e-01 -2.50057191e-01 -6.34687364e-01 -6.57747328e-01
-3.06021452e-01 6.36303544e-01 4.27103877e-01 7.21843421e-01
-9.14499938e-01 5.29690683e-01 6.67048168e+00 5.40129781e-01
-1.69944465e+00 -4.96081293e-01 6.45049989e-01 -2.80351698e-01
-1.06487796e-01 8.13686699e-02 -5.10621190e-01 3.48900169e-01
6.63616717e-01 -1.41280219e-01 8.44402015e-01 3.66812289e-01
2.27141932e-01 4.83006909e-02 -9.59330797e-01 1.11757040e+00
8.92244652e-02 -1.28065395e+00 -8.24152306e-03 4.84999895e-01
7.56299198e-01 -3.58608991e-01 5.58426976e-01 -5.50482035e-01
-3.71070236e-01 -1.47054732e+00 2.80082915e-02 8.55425179e-01
1.15212476e+00 -5.72299302e-01 7.58535028e-01 -4.57667150e-02
-6.13853395e-01 -1.36481524e-01 -3.91484529e-01 -1.74648345e-01
-5.49075782e-01 2.95020461e-01 -1.01193869e+00 2.20915556e-01
3.78971100e-01 3.93036038e-01 -4.77990270e-01 1.25312591e+00
5.22336513e-02 4.47420031e-01 -2.34823495e-01 3.44181627e-01
-1.89796433e-01 -6.08551502e-01 9.04724121e-01 6.38475776e-01
1.95151031e-01 9.17379744e-03 -3.96383017e-01 1.38114417e+00
1.08015634e-01 4.29565340e-01 -5.13317347e-01 -1.85979396e-01
7.30434209e-02 9.78963614e-01 -2.46836230e-01 5.71734942e-02
-3.08695138e-01 6.71377599e-01 -2.02280208e-01 4.91137505e-01
-5.59841692e-02 -3.81610543e-01 9.28234279e-01 4.70886409e-01
-2.07564309e-02 3.69636148e-01 -4.20540273e-01 -8.71663988e-01
2.21766338e-01 -1.32665467e+00 -5.11775203e-02 -6.49243414e-01
-9.94327426e-01 5.85042298e-01 -4.31408852e-01 -1.41224027e+00
-4.06286836e-01 -6.50595963e-01 -5.82275093e-01 1.58880031e+00
-1.17968011e+00 -1.43096161e+00 -1.87015221e-01 6.15964234e-01
-1.38103753e-01 -4.80838865e-01 9.42498446e-01 1.06145628e-01
-3.94825399e-01 8.08089316e-01 1.49263203e-01 4.66690719e-01
6.97349072e-01 -1.19734168e+00 3.20712894e-01 9.06747401e-01
2.74086416e-01 1.09123349e+00 5.19219875e-01 -5.06521463e-01
-1.27067792e+00 -4.97730762e-01 1.01832163e+00 -4.38639313e-01
2.37811506e-01 1.73898190e-01 -8.77695799e-01 4.55162346e-01
-3.89128700e-02 6.88163787e-02 8.56437445e-01 -3.85670806e-03
-1.19449213e-01 -2.22536087e-01 -1.24744141e+00 4.31329638e-01
3.93408030e-01 -6.71740830e-01 -6.16988957e-01 2.91246563e-01
1.68743655e-02 -1.14894915e+00 -1.06429815e+00 2.72269607e-01
9.87708926e-01 -1.17528737e+00 8.69938552e-01 -8.48943055e-01
3.39639932e-01 -5.03590047e-01 5.17041326e-01 -1.00300658e+00
-3.55819046e-01 -7.64228523e-01 1.67614475e-01 6.00080550e-01
1.20259248e-01 -6.65167868e-01 9.24330652e-01 8.18938553e-01
2.45854512e-01 -8.97446513e-01 -1.01064146e+00 -3.44238877e-01
6.30276278e-02 1.81155503e-01 7.27139115e-01 1.02500319e+00
-1.93001643e-01 -5.93709528e-01 -1.47326529e-01 2.56248325e-01
5.19012272e-01 2.85992384e-01 9.09465849e-01 -1.20404398e+00
-5.18982232e-01 -6.00444734e-01 -8.91599178e-01 -7.16573536e-01
-2.29254246e-01 -7.90358663e-01 -5.59173226e-01 -7.39798069e-01
4.80503552e-02 -4.58190531e-01 5.05369855e-03 4.95175600e-01
3.87801044e-02 1.98388904e-01 -4.80044670e-02 2.86352843e-01
7.60669768e-01 -5.47599271e-02 1.69142103e+00 -3.01960260e-02
-7.00541198e-01 2.92164505e-01 -5.88845015e-01 5.13473749e-01
5.13180673e-01 -2.48107716e-01 -2.40297854e-01 -2.80673742e-01
2.79088300e-02 1.49120139e-02 4.63077486e-01 -7.57298470e-01
3.40564251e-01 -1.27516285e-01 6.16462886e-01 9.53669250e-02
-8.20569992e-02 -6.81950569e-01 4.35395062e-01 5.19943893e-01
-2.43303820e-01 -1.51375726e-01 2.32646152e-01 -8.58974233e-02
-4.42654252e-01 -1.24533363e-01 1.01911223e+00 1.25095751e-02
-1.28519759e-01 2.43616834e-01 1.79944709e-01 -3.56979698e-01
8.00673366e-01 -8.94467533e-01 -3.11704218e-01 -1.55695841e-01
-7.46492267e-01 -3.61809790e-01 8.72651398e-01 3.62945974e-01
5.89540958e-01 -1.02827179e+00 -8.33982944e-01 1.00919032e+00
-2.75388993e-02 -1.73063502e-01 3.85286838e-01 1.16146004e+00
-8.62374663e-01 3.25991213e-01 -4.97536004e-01 -6.00864649e-01
-1.57968986e+00 5.21447718e-01 1.17636454e+00 1.30534858e-01
-5.13213277e-01 9.07513678e-01 6.86846673e-02 -3.13276559e-01
1.87860727e-01 -2.07044989e-01 -1.91082492e-01 -3.05783331e-01
7.51853704e-01 1.52948543e-01 1.05396174e-01 -8.04635108e-01
1.84071481e-01 1.14745450e+00 8.32708851e-02 3.06244999e-01
1.01233077e+00 8.57946649e-02 -5.47125816e-01 -1.08334690e-01
8.07867587e-01 3.09371144e-01 -1.05929029e+00 -2.41274208e-01
-7.21344292e-01 -9.09350216e-01 2.75266737e-01 -9.97930348e-01
-1.12281334e+00 7.81532407e-01 1.23736966e+00 -1.81255996e-01
1.47316563e+00 -4.45191890e-01 4.36271608e-01 -1.83309868e-01
-5.34473658e-02 -9.04119074e-01 -3.90411526e-01 -1.65227160e-01
8.57590437e-01 -9.56975162e-01 -5.88196851e-02 -2.13341764e-03
-3.24734509e-01 1.42133439e+00 1.96075767e-01 -2.30877995e-02
4.43727672e-01 1.50218219e-01 5.39700031e-01 -9.20112804e-02
-1.74072951e-01 -1.39261946e-01 9.75463867e-01 8.40946734e-01
5.61262071e-01 2.34594680e-02 -4.53452170e-01 -5.75992949e-02
-3.29030663e-01 6.34801313e-02 4.20733958e-01 3.72954220e-01
-4.71724011e-02 -1.56961834e+00 -7.18339562e-01 4.44601864e-01
-4.48068261e-01 -1.71324134e-01 -2.96917439e-01 5.71405828e-01
5.71100831e-01 4.97245044e-01 4.12755609e-01 -3.17614555e-01
2.56485045e-01 3.08893248e-02 6.60402536e-01 -2.65703201e-01
-8.67187321e-01 2.74495333e-01 -4.97680962e-01 -4.91804183e-01
-3.99741948e-01 -7.07256615e-01 -9.35140193e-01 -5.63223004e-01
-1.94798306e-01 -1.46304756e-01 8.36432457e-01 7.22413957e-01
3.91970634e-01 -1.79117620e-02 7.48417914e-01 -4.97982264e-01
-5.87357044e-01 -9.60892618e-01 -8.82760763e-01 7.14832723e-01
7.05802560e-01 -3.19202602e-01 -5.61342120e-01 2.68048286e-01] | [3.7424685955047607, -3.632662773132324] |
dc9c85b1-2004-44a6-8f7c-3a6279cdf2d2 | reducing-conservativeness-oriented-offline | 2103.00098 | null | https://arxiv.org/abs/2103.00098v1 | https://arxiv.org/pdf/2103.00098v1.pdf | Reducing Conservativeness Oriented Offline Reinforcement Learning | In offline reinforcement learning, a policy learns to maximize cumulative rewards with a fixed collection of data. Towards conservative strategy, current methods choose to regularize the behavior policy or learn a lower bound of the value function. However, exorbitant conservation tends to impair the policy's generalization ability and degrade its performance, especially for the mixed datasets. In this paper, we propose the method of reducing conservativeness oriented reinforcement learning. On the one hand, the policy is trained to pay more attention to the minority samples in the static dataset to address the data imbalance problem. On the other hand, we give a tighter lower bound of value function than previous methods to discover potential optimal actions. Consequently, our proposed method is able to tackle the skewed distribution of the provided dataset and derive a value function closer to the expected value function. Experimental results demonstrate that our proposed method outperforms the state-of-the-art methods in D4RL offline reinforcement learning evaluation tasks and our own designed mixed datasets. | ['Xiangyang Ji', 'Shuncheng He', 'Yuhang Jiang', 'Jianzhun Shao', 'Hongchang Zhang'] | 2021-02-27 | null | null | null | null | ['d4rl'] | ['robots'] | [-1.38210759e-01 3.33818823e-01 -7.80390680e-01 -3.22111517e-01
-6.82155907e-01 -3.52828056e-01 1.47570014e-01 3.07996333e-01
-6.49756014e-01 1.25381851e+00 -1.75922409e-01 -1.83235288e-01
-2.75532335e-01 -8.20974767e-01 -7.47587323e-01 -1.01538622e+00
-9.36024636e-02 5.69444835e-01 5.73577061e-02 -2.97242314e-01
4.58348155e-01 3.63326341e-01 -1.53806353e+00 -9.67433453e-02
1.15826142e+00 1.36823332e+00 3.44836950e-01 1.28077269e-01
-5.13613820e-02 9.88754749e-01 -8.15751433e-01 -1.23117901e-02
4.13423866e-01 -3.96474183e-01 -6.18180156e-01 6.56425506e-02
1.07797354e-01 -6.44854307e-01 -1.44393649e-02 1.27299643e+00
6.96469128e-01 1.90046564e-01 4.29489970e-01 -1.24173152e+00
-4.18598115e-01 8.77648532e-01 -9.20242608e-01 1.95132658e-01
-8.00069720e-02 3.19402456e-01 8.15593481e-01 -2.12780014e-01
4.26212251e-01 1.12577927e+00 2.55190581e-01 9.02539968e-01
-9.98660147e-01 -5.81951916e-01 6.58882201e-01 1.60452098e-01
-1.01950312e+00 -1.40983939e-01 8.46700311e-01 4.03685793e-02
5.67415297e-01 8.06596689e-03 7.14353025e-01 1.13300908e+00
-1.67003304e-01 1.14611280e+00 1.35567701e+00 -1.83822036e-01
6.50692523e-01 2.43544653e-01 -2.49615967e-01 5.80946028e-01
2.78758794e-01 2.92492419e-01 -4.60581243e-01 -1.39558092e-01
3.82649869e-01 -2.75120646e-01 -1.12195306e-01 -5.25974929e-01
-6.66060090e-01 6.84068859e-01 2.78823137e-01 2.57752649e-03
-5.20634234e-01 1.88089952e-01 5.38887680e-01 4.42571223e-01
3.36217761e-01 5.18687665e-01 -5.99597812e-01 -5.73796570e-01
-5.45189083e-01 6.98698282e-01 5.83400607e-01 7.20629156e-01
6.52550638e-01 2.06464559e-01 -4.41628933e-01 7.79512942e-01
5.56260310e-02 2.25860536e-01 6.01827323e-01 -8.29909921e-01
8.48681748e-01 6.43405318e-01 4.82486576e-01 -6.00318789e-01
-2.31667817e-01 -7.21458852e-01 -5.42999446e-01 4.14997965e-01
7.92729497e-01 -4.25364971e-01 -6.92821860e-01 1.78707135e+00
5.99348068e-01 -2.32168823e-01 3.17316741e-01 1.11158812e+00
5.66676818e-02 4.36497629e-01 -1.61591414e-02 -6.30060673e-01
7.84633636e-01 -8.18630815e-01 -7.50143945e-01 -1.48790509e-01
5.32210350e-01 -2.49654993e-01 1.19922984e+00 6.35746300e-01
-1.04569995e+00 -5.31379879e-01 -1.11107683e+00 5.85086167e-01
9.39373523e-02 2.69526154e-01 3.71624440e-01 5.35767436e-01
-5.29396474e-01 8.98692012e-01 -8.01635563e-01 1.55215055e-01
7.10936725e-01 3.25792968e-01 3.94675672e-01 2.10446283e-01
-1.16065967e+00 7.42882133e-01 7.62884915e-01 3.75045389e-02
-1.16433024e+00 -7.38704920e-01 -3.60926986e-01 2.65606446e-03
8.08813810e-01 -1.89864244e-02 1.38656509e+00 -1.25985599e+00
-1.84947813e+00 4.58164811e-01 5.60967147e-01 -7.10466087e-01
9.56211269e-01 -2.52625227e-01 5.03499396e-02 1.36141013e-02
-3.09923530e-01 5.70785820e-01 1.00934732e+00 -1.32396066e+00
-8.53780925e-01 -4.12884742e-01 1.88496232e-01 4.57704484e-01
-6.95160985e-01 -6.83360338e-01 7.23555386e-02 -4.95064676e-01
-3.77657861e-01 -7.58113086e-01 -3.30110341e-01 -9.86325219e-02
-1.42809004e-01 -4.39500630e-01 7.01431632e-01 -2.34647512e-01
1.41322684e+00 -2.00503469e+00 1.19418181e-01 2.02732131e-01
-2.13170692e-01 3.58078837e-01 3.05823833e-02 2.75684565e-01
3.35782856e-01 -1.85801923e-01 -1.88916415e-01 9.86310020e-02
3.11045926e-02 4.46299136e-01 -3.92845720e-01 5.57650924e-01
1.62259508e-02 3.43355417e-01 -1.10079622e+00 -4.17908639e-01
-1.68176681e-01 -1.66098222e-01 -7.66798019e-01 4.08166677e-01
-6.25665843e-01 4.29087192e-01 -8.89180541e-01 6.75297022e-01
8.75066996e-01 1.10377803e-01 5.52872956e-01 1.60701156e-01
-1.03305303e-01 -3.48728034e-03 -1.24459505e+00 1.40953720e+00
-3.76348287e-01 -6.22716174e-02 -3.57619114e-02 -1.44509351e+00
1.31918931e+00 -8.55647698e-02 6.21268332e-01 -9.69374239e-01
1.70745358e-01 3.34866792e-01 1.37845695e-01 -6.02081120e-01
4.06317770e-01 -1.64569229e-01 6.58024698e-02 3.58163565e-01
-2.84916967e-01 8.79506320e-02 1.31966427e-01 -3.15064430e-01
8.26888740e-01 5.73201060e-01 3.36112857e-01 -4.13108140e-01
6.35408878e-01 -1.18803091e-01 8.02319884e-01 7.81845808e-01
-5.31608760e-01 -4.33867276e-02 1.05395520e+00 -4.48517680e-01
-9.40602422e-01 -7.29268909e-01 -1.49151161e-01 1.05339503e+00
3.71987194e-01 1.78969979e-01 -8.51624966e-01 -1.20531988e+00
3.77763987e-01 7.75650799e-01 -5.57796180e-01 -3.46847922e-01
-6.97753608e-01 -8.52256060e-01 4.44775671e-01 3.38053524e-01
6.60957515e-01 -1.18182731e+00 -1.04470432e+00 3.00292671e-01
7.01881945e-02 -7.11255550e-01 -2.19058990e-01 3.58789563e-01
-9.86094117e-01 -1.01383805e+00 -7.77473330e-01 -5.55522501e-01
7.47604072e-01 -2.91400343e-01 8.73671472e-01 -2.10471079e-02
3.48540880e-02 1.07159510e-01 -5.04270136e-01 -5.06315053e-01
-3.49699765e-01 2.62985140e-01 8.24976712e-02 4.63467166e-02
1.25402525e-01 -3.66606295e-01 -7.81764984e-01 3.94978791e-01
-7.84614146e-01 -4.15480793e-01 4.80231971e-01 1.06765175e+00
6.99438512e-01 2.85578460e-01 1.23668468e+00 -8.08374047e-01
8.24597180e-01 -5.62367976e-01 -9.93865371e-01 3.72380793e-01
-1.10460985e+00 5.45206368e-01 1.26733541e+00 -7.99571693e-01
-1.21216786e+00 -4.26607840e-02 6.97614700e-02 -4.26262945e-01
2.09171042e-01 1.29145116e-01 -1.15962744e-01 1.95582584e-01
6.89473808e-01 3.24702591e-01 2.47771874e-01 -4.15895015e-01
7.02507887e-03 5.81324458e-01 1.23187371e-01 -1.26954794e+00
4.10713017e-01 3.24490070e-01 9.94300768e-02 -2.39793792e-01
-9.34704900e-01 2.08878163e-02 -1.00838609e-01 -4.95904118e-01
2.07130179e-01 -6.45552754e-01 -1.08191192e+00 3.27345759e-01
-5.57882488e-01 -5.59618771e-01 -5.82579792e-01 3.66594791e-01
-1.00258160e+00 7.97480047e-02 -1.69962555e-01 -1.29775560e+00
-2.82711834e-01 -1.10354149e+00 8.01844060e-01 4.08913553e-01
2.59117812e-01 -5.78907549e-01 2.22213030e-01 1.32825702e-01
1.86830491e-01 3.93247515e-01 9.36400294e-01 -4.85207230e-01
-2.87754089e-01 2.29324028e-01 6.16770498e-02 5.07265151e-01
1.00639679e-01 -2.01434508e-01 -8.04742157e-01 -5.81328869e-01
1.16384268e-01 -1.02297473e+00 6.36173010e-01 2.11873382e-01
1.80523825e+00 -5.58891594e-01 2.38942709e-02 3.72059494e-01
1.42288923e+00 5.34055293e-01 5.35164595e-01 6.88863158e-01
1.48883328e-01 5.68410635e-01 1.43645430e+00 1.20133293e+00
1.55934989e-01 5.09284437e-01 9.15312171e-01 3.75817657e-01
3.05094182e-01 -5.38089156e-01 4.70216036e-01 1.13357641e-01
-2.18016040e-02 -1.48828566e-01 -5.82141578e-01 3.91565382e-01
-1.99911392e+00 -9.53808606e-01 5.86886585e-01 2.42078686e+00
1.26439178e+00 4.21157151e-01 4.74655300e-01 2.78171040e-02
6.19649291e-01 1.91069216e-01 -1.19000876e+00 -4.99027461e-01
1.43412292e-01 -6.09842800e-02 7.03843594e-01 1.55243173e-01
-6.96721315e-01 7.88248479e-01 5.40141392e+00 1.14550924e+00
-1.24685633e+00 -2.30386645e-01 9.48090374e-01 -4.50998366e-01
-3.38879317e-01 -2.60937452e-01 -9.30557609e-01 6.33188963e-01
6.26752675e-01 -1.26834914e-01 8.11138928e-01 1.19148076e+00
2.87843108e-01 -2.27042615e-01 -9.83136535e-01 8.07164848e-01
-3.11756313e-01 -1.03740847e+00 -2.74458140e-01 1.65224388e-01
8.87988389e-01 -3.06918502e-01 2.03735426e-01 6.97384000e-01
1.76068008e-01 -7.95393527e-01 9.42875981e-01 5.04749835e-01
5.59683025e-01 -1.31446493e+00 5.64226210e-01 7.76017249e-01
-5.45515716e-01 -5.90479732e-01 -7.01154709e-01 -4.75195386e-02
-3.63552660e-01 4.48285848e-01 -8.35960448e-01 2.48017281e-01
5.83218932e-01 4.26183343e-01 -3.03719014e-01 7.17000246e-01
-5.37984259e-03 4.61642474e-01 -1.28668502e-01 -5.53764820e-01
5.49561381e-01 -3.42263967e-01 3.42173666e-01 5.04471660e-01
1.79514825e-01 -9.43016186e-02 4.25228089e-01 5.70748568e-01
-1.57833174e-01 3.22651088e-01 -4.44366604e-01 -1.03499450e-01
4.81124461e-01 9.58098590e-01 -5.12360513e-01 -1.44991845e-01
2.69632358e-02 5.21384895e-01 1.01226580e+00 1.81410134e-01
-9.70403075e-01 -1.40018463e-01 4.13024813e-01 9.01970267e-03
4.55661684e-01 1.67874202e-01 -1.65604323e-01 -8.29909384e-01
2.64594227e-01 -1.15773249e+00 6.06206119e-01 6.93528876e-02
-1.24823678e+00 3.12487721e-01 -1.94200110e-02 -1.46409166e+00
-3.88974875e-01 -3.46595228e-01 -3.20227742e-01 3.35189819e-01
-1.72938919e+00 -4.37338948e-01 6.01681285e-02 4.78453279e-01
5.06746054e-01 -4.15451199e-01 3.16169381e-01 1.74547046e-01
-6.06734455e-01 8.04395795e-01 4.09335732e-01 -4.38556910e-01
6.28210545e-01 -1.35722470e+00 -4.72703606e-01 2.75486708e-01
-5.34861982e-01 9.08863991e-02 8.32472861e-01 -5.54402649e-01
-1.59067094e+00 -8.73580873e-01 -5.41140102e-02 2.45028660e-01
6.41040027e-01 -6.96512088e-02 -7.00864255e-01 1.31350681e-01
-7.76848644e-02 1.27296001e-01 2.83118933e-01 -2.20332965e-02
9.08264611e-03 -7.38064229e-01 -1.54456151e+00 5.47596633e-01
8.49570513e-01 1.39265284e-01 -1.64795652e-01 2.08659858e-01
3.51277649e-01 -6.70311630e-01 -9.01568532e-01 4.52830851e-01
5.67846656e-01 -9.41009700e-01 6.35216832e-01 -7.45659471e-01
4.97590005e-01 -8.82084593e-02 -9.24269482e-02 -1.46410584e+00
2.35860988e-01 -6.89674854e-01 -4.60700423e-01 1.20121956e+00
2.22644329e-01 -5.58479369e-01 1.05436909e+00 2.03825444e-01
4.79467809e-02 -1.48247385e+00 -1.08024848e+00 -9.51967120e-01
2.73113161e-01 -7.72420987e-02 7.01219976e-01 7.35269129e-01
-2.97480375e-02 -1.80910558e-01 -6.82519019e-01 -6.27127066e-02
8.51372659e-01 4.43779975e-01 5.90456069e-01 -6.24858320e-01
-5.88194430e-01 -3.31761181e-01 9.85604674e-02 -1.17633653e+00
2.50655949e-01 -5.64812243e-01 1.13919571e-01 -9.14499700e-01
-6.18643723e-02 -1.00659919e+00 -4.85880435e-01 3.68694246e-01
-6.53711036e-02 -3.79346341e-01 1.88707978e-01 -7.57000521e-02
-6.99015379e-01 1.00387907e+00 1.77628696e+00 -2.04376265e-01
-5.06063104e-01 2.41268367e-01 -8.21522951e-01 3.79198581e-01
1.03122640e+00 -6.15605295e-01 -8.21920812e-01 -7.50563070e-02
3.68404180e-01 3.71705055e-01 4.34551090e-02 -7.38763392e-01
-1.24477051e-01 -8.29873860e-01 4.00818855e-01 -5.66044867e-01
-1.38227299e-01 -8.53179157e-01 -4.29837704e-01 7.80015051e-01
-6.25071943e-01 -2.27070063e-01 -2.32687276e-02 6.94681168e-01
-3.67134884e-02 -5.44635892e-01 1.06063616e+00 -1.68814763e-01
-4.37027812e-01 2.12096885e-01 -7.42822960e-02 4.41761374e-01
1.25948298e+00 1.14804864e-01 -4.00095075e-01 -2.02777460e-01
-2.91602880e-01 7.71675467e-01 2.56696105e-01 4.49971646e-01
5.15294433e-01 -1.32722330e+00 -4.66456354e-01 7.58506532e-04
-3.59039451e-03 9.27079916e-02 2.95280725e-01 5.90132654e-01
-2.59154230e-01 1.62586104e-02 -4.18378562e-01 -4.02013630e-01
-6.32459283e-01 7.31526792e-01 5.34970641e-01 -5.33021331e-01
-5.04539847e-01 3.50632370e-01 -2.85617888e-01 -3.78936380e-01
5.58533728e-01 -3.20871979e-01 -2.48839185e-01 3.35040450e-01
4.51406807e-01 6.58618271e-01 -1.48527622e-01 1.65195584e-01
-2.65129179e-01 1.61186144e-01 -1.28492802e-01 1.54061586e-01
1.40992868e+00 4.55971658e-02 3.61251682e-01 3.04425597e-01
9.86494362e-01 -2.02242613e-01 -1.80870473e+00 -1.18792271e-02
-2.47306153e-02 -6.09651625e-01 1.18615450e-02 -9.64022815e-01
-1.15546286e+00 3.12909901e-01 7.93991029e-01 3.45875740e-01
1.26776767e+00 -3.16801667e-01 6.50638819e-01 5.81159353e-01
4.19735938e-01 -1.88204050e+00 3.03611338e-01 2.64727741e-01
8.34200084e-01 -1.24899733e+00 1.35960281e-01 2.76986271e-01
-1.06507301e+00 1.15426540e+00 1.16710615e+00 -2.82898128e-01
2.73345411e-01 3.42846274e-01 -2.61638146e-02 7.80108199e-02
-9.91410136e-01 -1.36232495e-01 -2.20899999e-01 5.38062215e-01
9.02833696e-03 5.50219007e-02 -6.54728770e-01 3.44754487e-01
2.45521287e-03 3.24310130e-03 4.12978947e-01 9.27657306e-01
-5.61518848e-01 -1.21875012e+00 -4.62241739e-01 3.62474769e-01
-3.34893346e-01 3.91312599e-01 6.17863275e-02 9.23627794e-01
1.42117754e-01 6.14808381e-01 2.43534204e-02 6.45181537e-02
3.26145768e-01 -2.02675253e-01 6.90870941e-01 -7.85662010e-02
-5.02467275e-01 4.61219437e-02 3.80469561e-02 -6.06986582e-01
-3.42705399e-01 -7.63085127e-01 -1.49687970e+00 -2.72339545e-02
-5.57603799e-02 1.43089086e-01 4.86588985e-01 8.92053246e-01
3.87184359e-02 5.49920738e-01 1.13721514e+00 -3.64453405e-01
-1.63402367e+00 -7.02136159e-01 -7.17299819e-01 2.86106557e-01
4.23940599e-01 -9.35726941e-01 -1.72999606e-01 -6.11208677e-01] | [4.115314960479736, 2.3282577991485596] |
29f11db7-a91a-4926-a03e-f6ab9475b7eb | a-sentence-interaction-network-for-modeling | null | null | https://aclanthology.org/P16-1053 | https://aclanthology.org/P16-1053.pdf | A Sentence Interaction Network for Modeling Dependence between Sentences | null | ['Minlie Huang', 'Biao Liu'] | 2016-08-01 | null | null | null | acl-2016-8 | ['sentence-pair-modeling'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.479804992675781, 3.632047653198242] |
6b5fb9ef-4941-4876-8bd8-0daf801feb48 | 190909910 | 1909.09910 | null | https://arxiv.org/abs/1909.09910v2 | https://arxiv.org/pdf/1909.09910v2.pdf | Deep learning approach to control of prosthetic hands with electromyography signals | Natural muscles provide mobility in response to nerve impulses. Electromyography (EMG) measures the electrical activity of muscles in response to a nerve's stimulation. In the past few decades, EMG signals have been used extensively in the identification of user intention to potentially control assistive devices such as smart wheelchairs, exoskeletons, and prosthetic devices. In the design of conventional assistive devices, developers optimize multiple subsystems independently. Feature extraction and feature description are essential subsystems of this approach. Therefore, researchers proposed various hand-crafted features to interpret EMG signals. However, the performance of conventional assistive devices is still unsatisfactory. In this paper, we propose a deep learning approach to control prosthetic hands with raw EMG signals. We use a novel deep convolutional neural network to eschew the feature-engineering step. Removing the feature extraction and feature description is an important step toward the paradigm of end-to-end optimization. Fine-tuning and personalization are additional advantages of our approach. The proposed approach is implemented in Python with TensorFlow deep learning library, and it runs in real-time in general-purpose graphics processing units of NVIDIA Jetson TX2 developer kit. Our results demonstrate the ability of our system to predict fingers position from raw EMG signals. We anticipate our EMG-based control system to be a starting point to design more sophisticated prosthetic hands. For example, a pressure measurement unit can be added to transfer the perception of the environment to the user. Furthermore, our system can be modified for other prosthetic devices. | ['Mohsen Jafarzadeh', 'Daniel Curtiss Hussey', 'Yonas Tadesse'] | 2019-09-21 | null | null | null | null | ['electromyography-emg'] | ['medical'] | [ 8.04576948e-02 -1.02115117e-01 -3.81003439e-01 -1.24789648e-01
-9.48762894e-03 -2.90834427e-01 -2.97898483e-02 -9.42959905e-01
-5.67822218e-01 7.12935388e-01 1.75788984e-01 -1.78537920e-01
-1.09877087e-01 -7.17912734e-01 -5.47512770e-01 -5.85078239e-01
4.16186377e-02 1.56934336e-02 1.14835136e-01 -3.40402335e-01
2.18263596e-01 6.97582781e-01 -1.68607557e+00 2.41864562e-01
6.44810736e-01 1.00178361e+00 7.04037070e-01 5.26260614e-01
4.41819914e-02 2.49588311e-01 -5.20806253e-01 2.28332326e-01
3.73272568e-01 -1.25051379e-01 -4.77985322e-01 6.02732152e-02
-2.27900282e-01 -5.06190658e-01 -5.19643664e-01 7.87420034e-01
9.79893029e-01 -1.29645437e-01 2.72395790e-01 -1.10496354e+00
-2.82173008e-01 4.00697291e-01 -1.06341369e-01 -1.31441087e-01
3.65072519e-01 7.24798083e-01 5.71378231e-01 -6.30580962e-01
5.52193344e-01 9.75740075e-01 5.41243732e-01 8.32549572e-01
-9.53273952e-01 -6.21342957e-01 -1.97601035e-01 5.81801653e-01
-9.97255981e-01 -4.64536309e-01 1.00039756e+00 -5.47889948e-01
1.17732012e+00 4.24288243e-01 9.78508949e-01 1.37719274e+00
5.74444413e-01 1.01651239e+00 8.74207139e-01 -2.81647235e-01
1.03110984e-01 -2.17995077e-01 -4.97331731e-02 2.98655808e-01
1.51643291e-01 1.83774188e-01 -4.79966402e-01 9.71732140e-02
1.15337777e+00 1.32561922e-01 -5.94226778e-01 -7.60638863e-02
-1.16289318e+00 3.32875490e-01 3.64191204e-01 3.06613952e-01
-8.61264527e-01 3.68534207e-01 4.59444106e-01 3.50682884e-01
-8.80020484e-02 5.00515878e-01 -5.46138763e-01 -7.71276116e-01
-3.04260492e-01 2.51433194e-01 8.41575682e-01 7.89477348e-01
3.08811441e-02 1.10168159e-01 -1.94138288e-01 8.14344466e-01
1.18348002e-01 2.76887983e-01 6.52182877e-01 -1.03049588e+00
3.77474308e-01 5.93937337e-01 1.44241035e-01 -5.79034746e-01
-7.53107011e-01 -2.14147821e-01 -6.66744292e-01 7.90102065e-01
3.23552668e-01 -5.18373847e-01 -8.23068738e-01 1.30142987e+00
1.12950332e-01 -3.02327543e-01 -2.98490316e-01 1.40215218e+00
3.19670588e-01 9.09116177e-04 -3.30104940e-02 9.80149880e-02
1.29217720e+00 -5.05286276e-01 -8.81884098e-01 -1.53873444e-01
3.56048286e-01 -5.47017813e-01 1.42254639e+00 7.47018099e-01
-8.82345498e-01 -6.64845049e-01 -1.09942544e+00 8.97447467e-02
-4.46884073e-02 5.44202626e-01 9.05648291e-01 5.73211551e-01
-4.65122700e-01 1.34184742e+00 -1.41869640e+00 -2.66579449e-01
2.96142787e-01 1.06019223e+00 -4.07305747e-01 6.51592016e-01
-9.93800819e-01 9.04020071e-01 2.96304464e-01 4.59709853e-01
-9.06351283e-02 -3.00930500e-01 -3.03535640e-01 -3.87373269e-02
2.92116195e-01 -8.36961031e-01 9.39194739e-01 -7.38909245e-01
-2.24739861e+00 5.00906587e-01 1.82327107e-01 2.16645375e-01
4.44565952e-01 -6.51154280e-01 -3.24344426e-01 -2.91139513e-01
-2.60946631e-01 1.94419593e-01 8.52949977e-01 -5.18563747e-01
-4.23742294e-01 -5.48810601e-01 -9.56050307e-02 1.07475452e-01
-6.71461046e-01 -1.12557538e-01 -3.75810385e-01 -6.64088607e-01
-2.03310754e-02 -1.10965419e+00 -3.72416787e-02 3.55853826e-01
-4.20491189e-01 -3.26655716e-01 7.30972648e-01 -6.70207739e-01
1.02300298e+00 -2.08734679e+00 6.41828477e-01 1.78187191e-01
-3.82299758e-02 4.36730415e-01 1.10400550e-01 2.22677693e-01
8.75036642e-02 -2.71546304e-01 2.56876886e-01 2.32524112e-01
-4.81381714e-02 2.81817228e-01 2.23771766e-01 3.06352347e-01
2.17506066e-01 7.40686655e-01 -6.09195173e-01 -9.01168510e-02
4.99804884e-01 4.79496330e-01 -5.96385777e-01 3.30464780e-01
1.66497380e-02 6.71091557e-01 -5.62757909e-01 6.96515143e-01
3.03533137e-01 1.75682381e-01 3.50224465e-01 -6.32392466e-01
-1.42599002e-01 3.54633808e-01 -1.28143907e+00 1.78315783e+00
-4.39084172e-01 5.69157362e-01 3.94091189e-01 -8.88410926e-01
7.60314703e-01 4.57822949e-01 7.19560981e-01 -4.66318846e-01
5.91166139e-01 4.53325272e-01 4.08601463e-01 -1.19816005e+00
1.42285720e-01 5.81517741e-02 3.20868701e-01 3.06845993e-01
3.64972698e-03 1.84015587e-01 -1.19027860e-01 -6.97630048e-01
1.10276043e+00 7.72279859e-01 1.34287313e-01 -3.64586338e-02
4.14920747e-01 -1.76660474e-02 5.88117778e-01 3.03847969e-01
-4.77043018e-02 3.90682191e-01 1.28549218e-01 -3.63949418e-01
-9.92872179e-01 -9.64192629e-01 -6.41838163e-02 1.00054622e+00
-1.35789320e-01 -2.10615143e-01 -6.01472020e-01 -1.13826051e-01
3.95385295e-01 2.17961907e-01 -9.22164917e-02 -3.18130404e-01
-8.97901893e-01 -3.31394225e-01 4.00642663e-01 1.11418748e+00
2.74748176e-01 -1.41469157e+00 -8.23623180e-01 6.29133284e-01
-1.88449887e-03 -9.40600991e-01 -2.45683506e-01 2.95439035e-01
-1.08564937e+00 -9.51702893e-01 -5.81403136e-01 -7.99589992e-01
3.23311776e-01 -1.50133222e-01 2.15492353e-01 -2.66918559e-02
-5.11885941e-01 1.80718169e-01 -3.70705992e-01 -3.53678703e-01
-1.03936844e-01 2.90491253e-01 5.94465733e-01 -3.24954718e-01
4.99377698e-01 -9.70120728e-01 -8.53552222e-01 3.67885649e-01
-3.85502994e-01 1.19621791e-01 7.72446394e-01 8.90639782e-01
4.77208495e-01 -3.52679253e-01 4.35682237e-01 -1.93581536e-01
1.17856145e+00 -5.04228845e-02 -2.64023453e-01 4.10982035e-02
-2.85120130e-01 1.65763289e-01 8.28939140e-01 -8.17807674e-01
-8.52184832e-01 4.19347405e-01 -6.52800262e-01 -2.68815637e-01
-7.97118619e-02 5.06117761e-01 -3.41115683e-01 -1.79428816e-01
7.05621243e-01 1.88289031e-01 4.89099324e-01 -7.97945797e-01
4.44940813e-02 1.50912905e+00 8.04132402e-01 -6.36547267e-01
2.48367608e-01 1.87760219e-01 1.52930524e-02 -9.22355711e-01
1.29055277e-01 -2.70529062e-01 -7.01263309e-01 -4.89256769e-01
5.23807526e-01 -5.49598634e-01 -1.41275275e+00 6.44914091e-01
-1.22132885e+00 -2.49637574e-01 -4.16211896e-02 9.15264487e-01
-8.14188480e-01 3.33533466e-01 -5.61564386e-01 -6.84804916e-01
-7.74523795e-01 -1.26589239e+00 9.93189275e-01 2.53687859e-01
-7.12721765e-01 -4.02652472e-01 -2.49983609e-01 2.05130830e-01
5.64151883e-01 2.13454321e-01 5.47317147e-01 -9.69714150e-02
-1.49922997e-01 -6.00286782e-01 2.41380587e-01 6.11569881e-01
6.32200897e-01 -1.92179754e-01 -8.62196803e-01 -3.52403790e-01
9.11466926e-02 -1.14654668e-01 3.17081839e-01 4.58006829e-01
1.52449608e+00 -1.01070832e-02 -3.90437007e-01 6.54663384e-01
1.04287112e+00 3.93796563e-01 8.61229300e-01 4.20635492e-01
8.24978948e-01 5.25573552e-01 5.08022308e-01 4.67220575e-01
-1.62700132e-01 1.05475163e+00 2.12202251e-01 1.09585956e-01
-1.78782284e-01 1.89440161e-01 5.11694789e-01 7.50926137e-01
-1.07636070e+00 3.10094118e-01 -5.09643734e-01 1.50234342e-01
-1.74556017e+00 -7.43444145e-01 -1.40944153e-01 2.07440352e+00
8.49732220e-01 2.12332442e-01 1.57598332e-01 4.10446703e-01
3.59291166e-01 -5.22300541e-01 -9.46581721e-01 -2.25164592e-01
3.62204880e-01 6.30989432e-01 4.70674366e-01 1.28806680e-01
-8.01589549e-01 7.36904621e-01 5.40469217e+00 4.00736064e-01
-1.66543496e+00 -1.29954293e-01 -5.99716306e-01 -3.70035022e-01
1.86232805e-01 -3.34165514e-01 -5.06597579e-01 6.18689001e-01
4.31021124e-01 -1.28419787e-01 6.56536937e-01 1.12438881e+00
4.53506768e-01 2.60308415e-01 -1.35408580e+00 1.03929615e+00
-4.84415472e-01 -1.22264016e+00 -6.76173568e-01 2.12455213e-01
-2.87938695e-02 6.92130402e-02 -3.74010354e-01 1.82652235e-01
-4.86939162e-01 -8.84097219e-01 3.90756398e-01 7.55956411e-01
8.33574831e-01 -4.78914231e-01 6.93611264e-01 2.77380168e-01
-1.02521968e+00 -1.94598317e-01 -2.31432691e-01 -5.66432416e-01
3.54859143e-01 1.56677917e-01 -6.82686388e-01 1.63644940e-01
6.47750258e-01 5.01953900e-01 8.20727795e-02 9.01583970e-01
-2.31712878e-01 3.04763198e-01 -4.98002738e-01 -5.66555798e-01
-4.10469204e-01 -4.14265245e-02 8.10647070e-01 8.72324228e-01
2.97925353e-01 1.66765191e-02 -1.60185829e-01 9.45089400e-01
1.52683258e-01 -3.67243290e-02 -5.03368258e-01 -2.85713196e-01
2.10186005e-01 1.06264544e+00 -2.06828892e-01 1.26736900e-02
-1.68988153e-01 1.29419363e+00 1.94285661e-02 2.27158606e-01
-4.71951097e-01 -9.51931655e-01 1.14948213e+00 3.00718427e-01
1.62144024e-02 -5.40125668e-01 -6.94285810e-01 -1.25109136e+00
6.86241508e-01 -8.40111732e-01 -3.75395119e-01 -6.35442793e-01
-1.19990635e+00 3.39941889e-01 -3.12113374e-01 -1.55669916e+00
-5.83090603e-01 -1.33814621e+00 -6.60936832e-01 1.07686377e+00
-8.29327464e-01 -6.07928872e-01 -4.87806737e-01 6.99102163e-01
5.41513860e-01 -2.08394423e-01 1.08617449e+00 3.57349038e-01
-4.14345622e-01 4.14519638e-01 4.44636680e-02 2.10972443e-01
5.01111329e-01 -9.58332956e-01 4.12139803e-01 5.64666033e-01
-5.31413615e-01 1.02324009e+00 6.96522593e-01 -6.53937399e-01
-2.00248170e+00 -4.89281148e-01 3.94643962e-01 -4.11928408e-02
5.54884374e-01 -2.01905981e-01 -7.60328054e-01 4.66869026e-01
-1.45424068e-01 -2.01043203e-01 5.88751376e-01 1.28233925e-01
4.65587795e-01 -2.92484492e-01 -9.07175958e-01 9.24591660e-01
1.29944003e+00 -4.65450555e-01 -7.35628784e-01 3.24621618e-01
1.84267834e-01 -5.94410360e-01 -1.20642078e+00 4.20233756e-01
1.47474599e+00 -4.23234999e-01 8.35332751e-01 -7.70120203e-01
2.20453620e-01 -1.38554364e-01 7.48295709e-02 -1.43401206e+00
-4.55751389e-01 -6.62550807e-01 -2.10674062e-01 6.19136274e-01
-2.60116793e-02 -5.58596611e-01 1.07028043e+00 7.33861446e-01
-4.19853270e-01 -8.35898936e-01 -9.94077742e-01 -8.77964437e-01
-2.80304790e-01 -7.05248654e-01 5.29212952e-01 3.55630875e-01
7.33236611e-01 2.14208201e-01 -6.87953472e-01 -1.05582252e-01
4.62916076e-01 -1.22871175e-01 9.70545411e-01 -1.30395031e+00
-5.75673819e-01 -4.92256999e-01 -7.72845507e-01 -1.03482258e+00
-2.30707843e-02 -7.26692617e-01 4.72622523e-05 -1.63760197e+00
-3.61910194e-01 -2.20373139e-01 -2.20586032e-01 6.65030181e-01
-5.36211953e-02 -1.57140344e-01 1.69260457e-01 1.08561829e-01
5.27357996e-01 4.88219082e-01 1.51311207e+00 -1.22969449e-01
-6.88936412e-01 5.15228450e-01 -3.47134382e-01 6.97408557e-01
1.02226317e+00 -2.85670429e-01 -1.25133604e-01 -4.11775678e-01
-2.29125485e-01 -4.67141084e-02 3.02155852e-01 -1.17102408e+00
2.56442457e-01 -1.97819009e-01 4.50675339e-01 -2.29357168e-01
3.75759482e-01 -1.02909827e+00 3.91492546e-01 8.78729641e-01
5.85464984e-02 -4.58925545e-01 4.31312583e-02 6.05434291e-02
1.02147624e-01 3.80105786e-02 2.74802417e-01 1.89651921e-02
-8.90019357e-01 -2.32647527e-02 -4.75934029e-01 -7.49280572e-01
8.76359284e-01 -7.86926389e-01 3.56475525e-02 8.20613727e-02
-1.02354896e+00 8.15148950e-02 1.65777862e-01 8.65501761e-01
7.89692342e-01 -1.42902160e+00 -3.88018698e-01 6.39692962e-01
-1.02209099e-01 -4.00064230e-01 1.27993941e-01 9.64847684e-01
-6.15701020e-01 3.45079780e-01 -1.04243112e+00 -6.53195322e-01
-1.28550196e+00 -5.95801882e-03 3.56606156e-01 2.74762481e-01
-1.01619828e+00 3.89290988e-01 -6.30282581e-01 -2.64381707e-01
5.49410343e-01 -4.29669082e-01 -1.96893409e-01 -4.07318980e-01
3.76545936e-01 5.00198483e-01 1.19821563e-01 -7.19578064e-04
-4.62795615e-01 5.15869498e-01 2.49194354e-01 1.06522039e-01
1.64138949e+00 2.09048808e-01 1.08616641e-02 1.95487753e-01
9.46505666e-01 -3.15059304e-01 -1.21232367e+00 2.88611501e-01
-6.64185584e-02 -4.95411724e-01 1.10147245e-01 -7.65854180e-01
-1.11883652e+00 7.46448934e-01 1.00564444e+00 -3.83996189e-01
1.28605497e+00 -4.21711385e-01 9.98254418e-01 6.10776663e-01
7.74518907e-01 -1.50370169e+00 -2.11906582e-01 7.23134950e-02
1.24139166e+00 -1.00086367e+00 -7.45004136e-03 -3.14777702e-01
-4.82151985e-01 1.58989429e+00 6.62205338e-01 -3.46464813e-01
6.82551980e-01 6.10135972e-01 -1.05930746e-01 1.31440465e-03
-2.92439461e-01 -1.21122740e-01 4.74465758e-01 7.60803938e-01
7.25100100e-01 3.92197013e-01 -9.76579010e-01 1.08197868e+00
-1.16314404e-01 1.03870499e+00 1.61429152e-01 1.24304867e+00
-3.94637436e-01 -1.36608982e+00 -2.37660274e-01 7.08250105e-01
-4.03955549e-01 3.26920539e-01 -1.63403019e-01 8.26328397e-01
1.80370137e-01 7.74709105e-01 -2.49465629e-01 -1.07191896e+00
9.40800607e-01 1.42534122e-01 7.72299170e-01 -4.56636876e-01
-6.70730591e-01 1.24217518e-01 2.04617471e-01 -9.39946890e-01
-2.35820621e-01 -3.60521942e-01 -1.40546620e+00 -1.90093949e-01
-2.07950324e-01 -4.44486707e-01 1.09982145e+00 8.99309695e-01
4.61257339e-01 8.25003982e-01 3.22493672e-01 -1.44720733e+00
-9.05660152e-01 -1.39039660e+00 -6.77804053e-01 2.82712579e-01
-6.59982711e-02 -9.74810898e-01 3.20214257e-02 -8.63489360e-02] | [6.814927101135254, 0.17762640118598938] |
05c9fbbf-cf2f-423a-b216-1eb03e555e69 | towards-view-invariant-vehicle-speed | 2206.00343 | null | https://arxiv.org/abs/2206.00343v2 | https://arxiv.org/pdf/2206.00343v2.pdf | Towards view-invariant vehicle speed detection from driving simulator images | The use of cameras for vehicle speed measurement is much more cost effective compared to other technologies such as inductive loops, radar or laser. However, accurate speed measurement remains a challenge due to the inherent limitations of cameras to provide accurate range estimates. In addition, classical vision-based methods are very sensitive to extrinsic calibration between the camera and the road. In this context, the use of data-driven approaches appears as an interesting alternative. However, data collection requires a complex and costly setup to record videos under real traffic conditions from the camera synchronized with a high-precision speed sensor to generate the ground truth speed values. It has recently been demonstrated that the use of driving simulators (e.g., CARLA) can serve as a robust alternative for generating large synthetic datasets to enable the application of deep learning techniques for vehicle speed estimation for a single camera. In this paper, we study the same problem using multiple cameras in different virtual locations and with different extrinsic parameters. We address the question of whether complex 3D-CNN architectures are capable of implicitly learning view-invariant speeds using a single model, or whether view-specific models are more appropriate. The results are very promising as they show that a single model with data from multiple views reports even better accuracy than camera-specific models, paving the way towards a view-invariant vehicle speed measurement system. | ['Iván García Daza', 'David Fernandez Llorca', 'Antonio Hernández Martínez'] | 2022-06-01 | null | null | null | null | ['vehicle-speed-estimation'] | ['computer-vision'] | [ 1.00023281e-02 -3.03218246e-01 -1.72129542e-01 -5.14763534e-01
-5.05982459e-01 -5.43408215e-01 7.72489846e-01 -1.98196724e-01
-7.49836981e-01 5.78226566e-01 -5.15995800e-01 -3.55968386e-01
1.40353162e-02 -9.18479085e-01 -9.51894343e-01 -6.52883887e-01
2.17063829e-01 6.92684948e-01 2.75698125e-01 -4.14495468e-01
2.56563723e-01 9.35544729e-01 -1.81447005e+00 -4.40823138e-01
4.48967159e-01 7.61828482e-01 2.95343935e-01 8.94198716e-01
-1.13776196e-02 2.89813131e-01 -4.22051370e-01 -3.01481545e-01
4.70582366e-01 -2.84630004e-02 -1.30362391e-01 1.07043637e-02
7.88540184e-01 -6.86037004e-01 -5.06762445e-01 8.38299274e-01
1.58078536e-01 -3.13210883e-03 3.46511245e-01 -1.40672719e+00
3.19523722e-01 2.23982520e-02 -3.79279077e-01 8.33153725e-02
3.47878873e-01 3.85297716e-01 4.01657820e-01 -5.02205431e-01
5.41480720e-01 8.97553086e-01 6.19380593e-01 4.61486191e-01
-1.36433661e+00 -5.29109359e-01 -3.14400584e-01 5.25842607e-01
-1.42233312e+00 -5.70965052e-01 9.17729497e-01 -5.39275587e-01
6.20513141e-01 -9.02450681e-02 7.15548456e-01 1.46450734e+00
1.23322874e-01 9.76433456e-02 1.25211358e+00 -2.93509305e-01
1.78342208e-01 5.32134116e-01 -7.89532736e-02 4.67744231e-01
6.56573474e-01 5.43734372e-01 -2.93377012e-01 3.27395976e-01
6.42504215e-01 -1.49117932e-01 -2.01528087e-01 -9.07469690e-01
-1.33127308e+00 8.01414788e-01 2.86314845e-01 1.60035491e-01
-1.51953772e-01 3.90311867e-01 2.74060756e-01 3.38025928e-01
1.15587026e-01 3.65698934e-01 -3.82275432e-01 -3.72609556e-01
-1.04509616e+00 3.96043897e-01 5.24003267e-01 9.42749023e-01
9.86989319e-01 3.11371207e-01 5.32435000e-01 1.29382610e-01
3.14690590e-01 9.92098868e-01 1.13245972e-01 -1.16839933e+00
5.77706814e-01 3.07944447e-01 2.64842957e-01 -1.02400422e+00
-6.48305178e-01 -2.72518694e-01 -5.22830129e-01 7.77313709e-01
9.12125051e-01 -1.28036842e-01 -6.08203113e-01 1.60278571e+00
2.13902950e-01 3.50099415e-01 2.62387931e-01 1.13999629e+00
3.39000851e-01 3.81777346e-01 -3.65780085e-01 3.69277894e-02
1.05203462e+00 -3.44508022e-01 -4.70130831e-01 -3.59057724e-01
7.08487868e-01 -6.25104368e-01 8.04496527e-01 5.12128592e-01
-6.44159555e-01 -7.00521767e-01 -1.36583579e+00 1.89365566e-01
-6.08907998e-01 1.20674193e-01 3.91384214e-01 9.09889698e-01
-8.36499453e-01 3.60784948e-01 -7.79071510e-01 -5.10434687e-01
-3.87994051e-02 2.13525653e-01 -6.20253146e-01 -1.79361165e-01
-1.20310760e+00 1.32936847e+00 2.95001328e-01 1.90305725e-01
-8.40025187e-01 -5.62099218e-01 -9.34275329e-01 -3.46211106e-01
2.63549685e-01 -5.19494474e-01 9.40690577e-01 -8.74906838e-01
-1.76409018e+00 8.25513601e-01 -7.29324445e-02 -7.59698689e-01
8.70518029e-01 -2.10140437e-01 -4.04000729e-01 2.33223423e-01
-1.80836260e-01 7.21743762e-01 8.46815526e-01 -1.30890703e+00
-4.25597340e-01 -3.09510469e-01 2.40978584e-01 -1.89776018e-01
1.75024167e-01 -4.68751490e-01 -2.40150705e-01 1.38353527e-01
-1.59831986e-01 -1.21501780e+00 -1.54648051e-01 2.23203730e-02
-1.23289153e-01 4.10475522e-01 1.00927079e+00 -3.02169740e-01
4.57992256e-01 -1.81195116e+00 -2.05264911e-02 1.33717924e-01
2.57706344e-02 4.54409510e-01 -5.33110313e-02 3.37393016e-01
1.15744635e-01 -2.72844940e-01 -3.10107437e-03 -4.66189943e-02
-3.67015660e-01 1.72490522e-01 2.01560520e-02 7.68639922e-01
7.72768930e-02 6.53751731e-01 -7.71322727e-01 -2.87832230e-01
1.05433083e+00 7.95189977e-01 -2.00874344e-01 -5.84955625e-02
-3.81875671e-02 7.29732990e-01 -2.92520016e-01 5.78297824e-02
8.21227252e-01 3.33215684e-01 -3.59678715e-02 -3.19209456e-01
-4.50843066e-01 3.34536694e-02 -1.30954814e+00 1.52649283e+00
-8.75203788e-01 1.33425534e+00 1.40114620e-01 -1.05524194e+00
1.07827401e+00 2.34410152e-01 2.61818856e-01 -9.46779847e-01
3.63730341e-01 2.42523730e-01 -6.08986542e-02 -5.76039076e-01
5.64958930e-01 -1.42700613e-01 8.56845230e-02 -5.84852174e-02
-6.17108159e-02 -4.73811507e-01 1.94563314e-01 -2.63990853e-02
8.00696731e-01 2.97060817e-01 -7.00654536e-02 6.07558340e-02
7.04851747e-01 1.76822260e-01 2.37965822e-01 4.91433442e-01
-1.51000768e-01 5.92418551e-01 1.89258739e-01 -5.20927250e-01
-1.39897716e+00 -1.04704261e+00 -1.96246132e-01 5.32860719e-02
4.09218073e-01 1.52279045e-02 -7.09919810e-01 -3.41596633e-01
-7.02520013e-02 7.28871882e-01 -3.87572289e-01 -2.30628490e-01
-6.40656888e-01 -2.36416593e-01 5.03502786e-01 4.05372113e-01
5.65987170e-01 -2.12827042e-01 -1.08583605e+00 9.66572240e-02
-8.21192414e-02 -1.64099681e+00 1.69913098e-01 -1.06409281e-01
-8.94334197e-01 -1.07827246e+00 -6.48816645e-01 -3.74735892e-02
4.31958467e-01 6.59313798e-01 7.86943316e-01 -2.25507334e-01
6.21532090e-02 5.40038943e-01 -2.23103777e-01 -4.33630437e-01
-5.92028499e-01 4.69007269e-02 2.74079949e-01 5.97205199e-02
3.59080136e-01 -4.01681304e-01 -5.08843064e-01 5.21485627e-01
-5.96239030e-01 2.98250839e-02 5.24582267e-01 2.42933422e-01
2.50927836e-01 -3.22901219e-01 4.46009219e-01 -5.64722836e-01
5.09729944e-02 -2.12144509e-01 -1.27840459e+00 -3.80212426e-01
-6.49348736e-01 8.04759264e-02 9.18474197e-01 -2.42224008e-01
-8.25400889e-01 3.51908684e-01 -1.33271039e-01 -6.93655193e-01
-5.44420540e-01 1.48930654e-01 -1.67361557e-01 -2.12740868e-01
8.43666673e-01 5.88170364e-02 3.72348815e-01 -4.53339927e-02
3.55653733e-01 5.00211596e-01 5.04409373e-01 -9.65572894e-02
1.20743835e+00 8.72976601e-01 5.48996866e-01 -1.34010375e+00
-3.87121201e-01 -4.48065877e-01 -7.81687975e-01 -6.94354355e-01
8.85341346e-01 -1.00810242e+00 -5.88620782e-01 4.79698181e-01
-1.12630022e+00 -2.93356836e-01 -1.19964018e-01 9.99035895e-01
-8.22282553e-01 3.55707854e-01 -8.60757083e-02 -6.31392419e-01
2.75244415e-01 -1.35899329e+00 1.02589095e+00 3.48602444e-01
-3.47933434e-02 -1.00013196e+00 7.91704059e-02 5.26079237e-01
5.59175849e-01 3.43815714e-01 1.76793769e-01 -3.05405799e-02
-8.84905219e-01 -6.12661123e-01 -2.97708716e-02 3.48957151e-01
-1.15450695e-01 2.64170110e-01 -1.21939993e+00 -2.26293266e-01
-4.28369828e-02 -6.97764680e-02 6.17584527e-01 4.47927862e-01
6.61813498e-01 2.94680178e-01 -3.25058103e-01 6.47288144e-01
1.69045997e+00 8.10657255e-03 7.30816007e-01 5.09271145e-01
6.43506289e-01 7.25497663e-01 6.55102193e-01 1.87494736e-02
4.66369897e-01 1.15853655e+00 6.24664009e-01 3.70564312e-02
-1.06820716e-02 -1.60007671e-01 3.49677682e-01 3.98941368e-01
-1.37039319e-01 -1.15369685e-01 -8.15980375e-01 3.87192100e-01
-1.34529543e+00 -1.17633033e+00 -6.26696765e-01 2.62844729e+00
2.28173211e-02 4.75669354e-01 1.99462801e-01 3.68121773e-01
4.83997613e-01 5.93450479e-02 -2.62280643e-01 -6.03797853e-01
5.50474785e-02 -2.32996285e-01 1.15574539e+00 5.94027460e-01
-8.64437282e-01 5.64990580e-01 5.83755493e+00 2.05181047e-01
-1.49535000e+00 -9.02413428e-02 1.35124773e-01 -5.03035001e-02
-1.01220839e-01 1.54041424e-01 -9.25020874e-01 5.26209176e-01
1.47038364e+00 3.11204903e-02 3.60022366e-01 6.78352058e-01
6.33263946e-01 -4.48725045e-01 -1.02977765e+00 1.04547346e+00
1.35331333e-01 -1.23153007e+00 -2.09273055e-01 3.17301363e-01
4.68063980e-01 3.11071664e-01 -1.97039321e-01 2.02669669e-02
-2.50502050e-01 -6.60287678e-01 7.14957356e-01 7.62725413e-01
6.97265327e-01 -7.34494805e-01 7.99612403e-01 5.60692668e-01
-1.04636514e+00 2.59806007e-01 -2.93014646e-01 -1.02577478e-01
4.25434381e-01 4.70236659e-01 -8.76699150e-01 5.47324479e-01
4.71302092e-01 5.16367495e-01 -5.36626875e-01 1.01288795e+00
-2.09317566e-03 4.60471064e-01 -5.97211480e-01 2.96018627e-02
2.25919485e-01 -4.06779379e-01 7.00961292e-01 1.07730424e+00
5.12256324e-01 -6.08108878e-01 -3.58080566e-01 7.36924350e-01
3.31722647e-01 -3.70163292e-01 -1.31864023e+00 4.59411293e-01
1.95885539e-01 1.43250537e+00 -5.64009368e-01 -6.90594018e-02
-7.50234663e-01 3.81459415e-01 2.54793279e-02 3.46024781e-01
-1.03768241e+00 -3.04241627e-01 7.25772321e-01 4.43834215e-01
3.05963695e-01 -7.93318570e-01 1.10445574e-01 -9.92763042e-01
1.82672497e-02 -3.78748357e-01 -4.16369170e-01 -8.78889740e-01
-4.65940356e-01 2.72159874e-01 2.84372807e-01 -1.69778860e+00
-4.02415484e-01 -8.52542937e-01 -4.77561116e-01 7.22050309e-01
-1.76349854e+00 -7.99710870e-01 -6.00116670e-01 4.04702753e-01
4.14005816e-01 -5.79836294e-02 3.84687275e-01 3.92351002e-01
-3.12451273e-01 1.95217416e-01 7.15250075e-02 -2.31578574e-02
5.44536293e-01 -1.10226917e+00 2.81945616e-01 8.42951179e-01
1.77137628e-01 1.03212602e-01 1.09396720e+00 -1.53044060e-01
-1.76862979e+00 -9.00304496e-01 5.28064549e-01 -6.77616596e-01
4.49950725e-01 -4.23008651e-01 -7.35895872e-01 5.10858834e-01
1.04768150e-01 8.74539539e-02 9.29413959e-02 -3.08779329e-01
-9.26730707e-02 -6.26540244e-01 -8.86400640e-01 3.86814952e-01
5.22999346e-01 -4.92649168e-01 -3.21416885e-01 -8.42402279e-02
3.47416013e-01 -4.04648095e-01 -5.55326343e-01 1.92126408e-01
7.35456586e-01 -1.18994999e+00 9.26581979e-01 3.84555310e-02
8.73704627e-02 -4.80438560e-01 -1.23294011e-01 -1.34732985e+00
2.23597914e-01 -2.35965535e-01 9.60152075e-02 1.08453381e+00
3.57526243e-01 -8.46022308e-01 9.88740146e-01 3.30060095e-01
9.31533575e-02 -2.01028422e-01 -1.17461526e+00 -1.03603220e+00
-7.19222352e-02 -8.48066092e-01 3.49526852e-01 6.28936887e-01
-4.66339946e-01 4.61962551e-01 -4.10486460e-01 3.71687263e-01
7.63766706e-01 -2.10975587e-01 1.32390177e+00 -1.37758613e+00
2.12580711e-02 -3.27916354e-01 -9.24943686e-01 -9.06943798e-01
2.80609369e-01 -5.18276393e-01 -7.42003247e-02 -1.27508497e+00
-3.67175013e-01 -4.20742303e-01 3.12723160e-01 -5.70874035e-01
4.26532745e-01 5.53446747e-02 9.81910825e-02 -1.10442773e-01
-1.56948492e-01 2.37194195e-01 1.09213006e+00 -4.08619680e-02
8.32656026e-02 2.99695492e-01 -1.36674410e-02 7.68228352e-01
1.03839004e+00 -2.64025897e-01 -4.97753739e-01 -5.04824102e-01
4.65073556e-01 2.02017412e-01 6.80574298e-01 -1.49207294e+00
3.49065989e-01 1.49774374e-02 3.92114192e-01 -5.84961832e-01
6.68231428e-01 -1.20766509e+00 1.99793532e-01 4.30449009e-01
-1.87675301e-02 1.42231241e-01 2.55412638e-01 6.41233087e-01
-2.17073262e-01 -3.39649349e-01 8.28445494e-01 -1.06532700e-01
-9.82040524e-01 3.22510861e-02 -6.60966575e-01 -1.52180761e-01
1.16022944e+00 -7.18247354e-01 -1.61848858e-01 -5.54888487e-01
-1.23921059e-01 3.28123085e-02 7.29422748e-01 5.46973348e-01
5.72818339e-01 -1.12104595e+00 -5.60682118e-01 3.67400378e-01
3.08955789e-01 -5.12086973e-02 1.34027779e-01 8.23203743e-01
-8.31914902e-01 7.67347097e-01 -2.58042395e-01 -1.09365773e+00
-1.01736629e+00 6.31526589e-01 6.86040342e-01 2.69804716e-01
-5.83001673e-01 5.04161566e-02 -2.03866497e-01 -4.75385040e-01
-1.93036512e-01 -3.84643942e-01 -1.20821044e-01 -3.83494571e-02
3.68089676e-01 5.35120368e-01 3.09004664e-01 -1.03642356e+00
-3.55283529e-01 1.08784330e+00 2.87039846e-01 -2.91644156e-01
8.69226754e-01 -3.53439093e-01 7.34244406e-01 5.98577321e-01
1.23327768e+00 -3.71101461e-02 -1.57748616e+00 1.31297350e-01
-1.16459504e-01 -4.56548721e-01 4.30121303e-01 -2.42054179e-01
-1.14584970e+00 1.16237080e+00 8.86870742e-01 3.09876651e-02
7.60872722e-01 -1.99480951e-01 4.74211127e-01 4.71489549e-01
6.74027801e-01 -1.12295830e+00 -3.76098186e-01 2.47896582e-01
5.35206139e-01 -1.47980893e+00 -8.89823064e-02 -1.42167002e-01
-2.95837194e-01 1.28945518e+00 5.76364458e-01 -9.24736187e-02
3.70967418e-01 2.37797663e-01 2.99030274e-01 1.91177130e-02
-5.39317906e-01 -3.72343957e-01 -1.29839346e-01 7.81233132e-01
1.66765139e-01 7.52223507e-02 4.25097458e-02 -4.94567305e-01
-6.66208491e-02 1.39986426e-01 1.00229013e+00 7.24119484e-01
-2.56197065e-01 -1.00853109e+00 -4.92245138e-01 3.10543031e-01
-2.89967563e-02 5.47533214e-01 -3.82991545e-02 1.37103355e+00
4.89559956e-02 1.04100788e+00 1.63852602e-01 -4.12438422e-01
7.01586008e-01 -6.18605502e-02 6.64913356e-01 -1.96400672e-01
-4.43891808e-02 -4.74419892e-01 1.07936673e-01 -6.20417774e-01
-6.51905417e-01 -8.66513193e-01 -9.15754437e-01 -4.68622059e-01
-3.30430120e-01 -1.79347143e-01 1.19989014e+00 9.92954373e-01
1.99278057e-01 1.29398152e-01 7.79248118e-01 -1.16292322e+00
-2.96308547e-01 -6.20125830e-01 -3.58967632e-01 2.82147318e-01
6.91134751e-01 -8.45222890e-01 -4.66225922e-01 -7.04504326e-02] | [7.958740234375, -1.3945425748825073] |
12e83557-e797-4c5b-80d0-003674e68450 | orthogonal-attention-a-cloze-style-approach | 2103.04294 | null | https://arxiv.org/abs/2103.04294v1 | https://arxiv.org/pdf/2103.04294v1.pdf | Orthogonal Attention: A Cloze-Style Approach to Negation Scope Resolution | Negation Scope Resolution is an extensively researched problem, which is used to locate the words affected by a negation cue in a sentence. Recent works have shown that simply finetuning transformer-based architectures yield state-of-the-art results on this task. In this work, we look at Negation Scope Resolution as a Cloze-Style task, with the sentence as the Context and the cue words as the Query. We also introduce a novel Cloze-Style Attention mechanism called Orthogonal Attention, which is inspired by Self Attention. First, we propose a framework for developing Orthogonal Attention variants, and then propose 4 Orthogonal Attention variants: OA-C, OA-CA, OA-EM, and OA-EMB. Using these Orthogonal Attention layers on top of an XLNet backbone, we outperform the finetuned XLNet state-of-the-art for Negation Scope Resolution, achieving the best results to date on all 4 datasets we experiment with: BioScope Abstracts, BioScope Full Papers, SFU Review Corpus and the *sem 2012 Dataset (Sherlock). | ['Vahida Attar', 'Aditya Khandelwal'] | 2021-03-07 | null | null | null | null | ['negation-scope-resolution'] | ['natural-language-processing'] | [ 3.56994003e-01 2.62431409e-02 -2.66514719e-01 -3.29687655e-01
-8.64957988e-01 -2.99027473e-01 4.32897925e-01 1.28603995e-01
-7.38500416e-01 8.87499571e-01 5.55386364e-01 -1.77427813e-01
1.43451616e-01 -5.45909107e-01 -9.38983679e-01 -2.78744072e-01
2.82345951e-01 3.41913998e-01 1.58639699e-01 -8.66289020e-01
5.54932237e-01 1.82137176e-01 -1.31529057e+00 7.02070653e-01
8.19668353e-01 9.17425692e-01 7.40617514e-02 6.34551585e-01
-1.11691132e-01 7.61226356e-01 -8.75532150e-01 -5.94101191e-01
-1.08980075e-01 -6.13706648e-01 -8.13808382e-01 -6.56231761e-01
6.61095321e-01 -6.67254105e-02 -1.22354180e-01 1.24997962e+00
7.94251740e-01 2.07429212e-02 2.55267590e-01 -8.40571642e-01
-1.36645818e+00 1.04245520e+00 -5.14667928e-01 7.70968974e-01
4.45413560e-01 8.10082778e-02 1.48115063e+00 -1.13791931e+00
8.68156374e-01 1.45870924e+00 7.19375134e-01 9.71171379e-01
-1.01736879e+00 -5.98498225e-01 4.57358778e-01 7.09783077e-01
-1.12724721e+00 -3.39225024e-01 5.58949590e-01 -3.92004289e-02
1.75718415e+00 1.61615148e-01 5.44991016e-01 1.41695333e+00
8.30500484e-01 9.02280748e-01 7.45696425e-01 -5.77119887e-01
4.71688621e-02 -2.95012176e-01 4.13859129e-01 4.07656580e-01
2.13379353e-01 -1.92570776e-01 -8.27259004e-01 7.51996487e-02
1.29039228e-01 -2.10591540e-01 -7.29949534e-01 9.85862389e-02
-1.11349010e+00 8.41885746e-01 4.96606082e-01 5.40296614e-01
-5.66619277e-01 1.88237697e-01 8.12574387e-01 4.17033851e-01
3.50142568e-01 6.88285291e-01 -7.68215716e-01 -4.36508656e-02
-8.01163614e-01 5.35364151e-01 6.61868095e-01 9.69343722e-01
3.36507410e-01 -5.59446476e-02 -8.04783285e-01 5.09692907e-01
1.06765248e-01 2.74878591e-01 7.11009622e-01 -4.27502334e-01
5.49071729e-01 5.06850660e-01 2.45846491e-02 -5.61752021e-01
-6.14049733e-01 -6.04911923e-01 -7.35851705e-01 -1.49615958e-01
-1.64322630e-01 -1.49081811e-01 -9.24018025e-01 2.09556150e+00
-1.96876243e-01 -2.06812799e-01 2.66580105e-01 8.75400186e-01
1.15149510e+00 3.73769224e-01 9.86389071e-02 -3.46261024e-01
1.74615729e+00 -1.04592562e+00 -1.46626604e+00 -3.95145357e-01
4.77237731e-01 -4.05061066e-01 1.36461949e+00 5.69016874e-01
-1.35424817e+00 -5.14505148e-01 -1.43337166e+00 -5.90830445e-01
-6.52728975e-01 -4.44388799e-02 4.58737999e-01 3.73662978e-01
-1.24653733e+00 7.75344014e-01 -3.92643690e-01 -5.33185780e-01
3.23689729e-01 3.59839648e-01 -3.28373641e-01 1.10552654e-01
-2.03592134e+00 1.33295178e+00 3.43540102e-01 3.03231031e-01
-7.18092263e-01 -8.96132231e-01 -7.79646337e-01 4.50307280e-01
5.53416431e-01 -8.88324976e-01 1.18280625e+00 -7.10002124e-01
-1.16057885e+00 1.09901416e+00 -2.31864542e-01 -8.77139807e-01
6.79841116e-02 -6.48293972e-01 -5.51044047e-01 -1.50270715e-01
3.77872407e-01 6.56653941e-01 6.09550357e-01 -6.49095535e-01
-4.36623305e-01 -3.65505487e-01 1.74141139e-01 1.40005246e-01
-7.74844587e-02 3.66662592e-01 -3.74018967e-01 -7.49476433e-01
-3.34983081e-01 -5.26546896e-01 2.51170918e-02 -3.75110179e-01
-6.04262173e-01 -6.53274596e-01 6.93838298e-01 -4.00152653e-01
1.64352691e+00 -1.96010113e+00 3.87914449e-01 -4.26439464e-01
1.75707266e-01 4.29905653e-01 -2.38853946e-01 3.78161848e-01
-6.89925253e-01 4.74223971e-01 -2.62295812e-01 -3.57079446e-01
4.87196892e-02 3.33045907e-02 -4.55386341e-01 2.14201868e-01
6.78062916e-01 1.28813279e+00 -9.13697183e-01 -9.42746624e-02
-3.22947860e-01 3.57868522e-01 -7.57446647e-01 1.44273162e-01
-4.06125188e-01 -1.41821876e-01 3.25380601e-02 7.61745155e-01
5.17465591e-01 -1.43527478e-01 2.04370350e-01 -5.44886470e-01
-8.18651617e-02 5.65257967e-01 -6.36763394e-01 1.67949867e+00
-1.06099680e-01 6.23324394e-01 1.12115145e-01 -6.88803732e-01
8.65287185e-01 3.75850886e-01 -2.47513667e-01 -9.81664479e-01
3.51043254e-01 2.00767443e-01 1.55523822e-01 -5.60112059e-01
6.46692753e-01 -2.76500911e-01 -3.26702774e-01 2.29329571e-01
4.31133389e-01 7.36748204e-02 5.35997748e-01 1.53869167e-01
1.31353927e+00 -9.64907091e-03 6.17664516e-01 -5.46970129e-01
8.45335901e-01 -3.56254071e-01 7.36956239e-01 9.51915443e-01
-3.75256628e-01 6.58930302e-01 1.07114375e+00 -4.17181671e-01
-5.97383559e-01 -7.43069291e-01 -1.67852249e-02 1.28959000e+00
-2.77758688e-01 -6.63508356e-01 -9.35085118e-01 -8.35662723e-01
-8.54311585e-02 9.28317785e-01 -1.17359066e+00 -4.27680463e-01
-7.47063696e-01 -7.97356606e-01 7.41706014e-01 8.85270655e-01
4.13940281e-01 -1.84524179e+00 -6.14011109e-01 3.02494556e-01
-4.05814677e-01 -9.34517086e-01 -6.77908838e-01 7.90847421e-01
-4.79250044e-01 -1.00454760e+00 -4.34263676e-01 -5.40079653e-01
3.68366875e-02 -3.90340239e-01 1.48214722e+00 7.49088079e-02
-1.60355359e-01 -1.64766148e-01 -2.15428904e-01 -7.15375185e-01
1.06237493e-01 1.88024402e-01 7.28476867e-02 -1.14937007e-01
6.99436367e-01 -3.34183872e-01 -3.03767085e-01 -3.89805853e-01
-9.89214778e-01 -6.31473005e-01 6.45555437e-01 1.31862366e+00
6.60756409e-01 -6.16967499e-01 8.19330812e-01 -1.20936763e+00
1.13654029e+00 -3.37364048e-01 -3.16117585e-01 1.16372243e-01
-7.04167724e-01 1.93338320e-01 5.92680097e-01 -1.06266409e-01
-5.93436480e-01 -4.72757608e-01 -5.05996823e-01 -5.31575143e-01
2.04421923e-01 7.95316875e-01 -4.38714236e-01 2.77886301e-01
7.03835547e-01 1.43024102e-01 -3.72004449e-01 -3.16168487e-01
2.73485839e-01 3.43702674e-01 7.06150651e-01 -8.78488421e-02
1.56484932e-01 2.88089067e-01 -9.03267786e-02 -5.35254180e-01
-1.30233610e+00 -1.93871036e-01 -6.39651835e-01 2.52408236e-01
1.17417932e+00 -7.19034433e-01 -7.47370362e-01 2.08973646e-01
-1.41100800e+00 -1.50268868e-01 -2.16125935e-01 -4.28169444e-02
-3.87536407e-01 2.14118212e-01 -5.80773652e-01 -3.57415885e-01
-8.73708308e-01 -1.15857077e+00 1.19327092e+00 1.21486925e-01
-5.47181785e-01 -7.39342213e-01 -1.19753936e-02 1.45982265e-01
5.50256610e-01 6.91016242e-02 1.18985665e+00 -9.91625488e-01
-3.19076478e-01 1.06377549e-01 -2.76383311e-01 2.63299078e-01
-1.71524599e-01 -3.41834486e-01 -9.96307373e-01 -1.66127965e-01
1.92343548e-01 -4.50665891e-01 1.46314013e+00 3.87043893e-01
1.21738768e+00 1.22649018e-02 -3.35747451e-01 6.52080774e-01
1.23987329e+00 5.45953400e-02 8.79662812e-01 5.29787421e-01
4.20742273e-01 1.40024260e-01 6.91561520e-01 1.04955330e-01
2.66316622e-01 5.35078526e-01 6.66362345e-01 -2.65756045e-02
-2.04118714e-01 1.10967018e-01 3.31921458e-01 4.35448796e-01
3.01949590e-01 -7.78196812e-01 -7.98637807e-01 7.95736670e-01
-1.81854796e+00 -1.16416097e+00 -1.93493571e-02 1.70294273e+00
1.01900291e+00 6.15939677e-01 -4.55898523e-01 1.93301290e-01
5.77140272e-01 2.56762058e-01 -5.40548801e-01 -9.95195210e-01
-5.51223457e-01 5.07638216e-01 6.45893961e-02 6.19354844e-01
-1.19717097e+00 1.13665795e+00 6.72158861e+00 5.77336252e-01
-1.21658325e+00 2.01744229e-01 2.74946958e-01 -2.16548741e-01
-2.68672854e-01 -3.38230342e-01 -1.19531858e+00 1.08325161e-01
1.15470314e+00 -1.11231126e-01 1.39379367e-01 6.01677060e-01
8.42460990e-02 9.56841111e-02 -1.45861244e+00 6.01738453e-01
6.71308875e-01 -1.42132998e+00 2.63885260e-01 -3.77642632e-01
5.41319430e-01 2.03307584e-01 1.62128378e-02 7.71045506e-01
-2.22399428e-01 -1.16263437e+00 7.10619390e-01 6.78772271e-01
8.31527650e-01 -7.67364562e-01 1.13492286e+00 -8.00988153e-02
-8.12019825e-01 -1.38912454e-01 -3.86877656e-01 -1.05644222e-02
2.81209173e-03 5.08082390e-01 -6.03543818e-01 5.49490988e-01
8.74672055e-01 7.94322252e-01 -5.67670643e-01 6.43360436e-01
-6.16235197e-01 5.15405059e-01 8.02660733e-02 -2.79427528e-01
2.40648955e-01 4.04869080e-01 6.44108415e-01 1.61953521e+00
-6.17924053e-03 1.24739841e-01 -2.43012145e-01 1.20175588e+00
-5.67680955e-01 2.41177790e-02 -3.99025470e-01 -5.12778647e-02
2.75667101e-01 1.08861351e+00 -3.41744959e-01 -3.08136791e-01
-3.57003033e-01 1.01652205e+00 6.33388758e-01 2.10202098e-01
-9.44988012e-01 -8.26313794e-01 7.62093604e-01 -2.20741242e-01
8.82162690e-01 5.25687754e-01 -2.60039419e-01 -1.12861061e+00
-3.52395978e-03 -1.16637456e+00 4.50980723e-01 -1.16349089e+00
-1.44986463e+00 7.70498395e-01 -4.17999834e-01 -7.19697475e-01
-1.41237956e-02 -7.63115585e-01 -7.01554954e-01 1.13158441e+00
-1.82515299e+00 -8.16468716e-01 -6.49742857e-02 4.00222808e-01
5.66485882e-01 1.08437026e-02 1.06366837e+00 2.46585011e-01
-6.16212666e-01 9.17164862e-01 -3.12612832e-01 2.26329863e-01
9.65951264e-01 -1.41760039e+00 4.84493881e-01 7.78548300e-01
-3.34690779e-01 1.08181643e+00 6.50729716e-01 -8.79816175e-01
-1.42196810e+00 -1.13405681e+00 1.62325966e+00 -4.72406894e-01
7.02916086e-01 -6.55732453e-01 -1.09867203e+00 8.97571504e-01
7.31935740e-01 1.10295124e-01 4.04845417e-01 3.74012589e-01
-3.93187135e-01 -9.89544168e-02 -1.05216324e+00 7.97943413e-01
1.11268127e+00 -5.74030161e-01 -1.20023274e+00 2.20679030e-01
1.06776190e+00 -6.46963716e-01 -8.06636512e-01 6.43402517e-01
2.07297087e-01 -1.09808338e+00 7.53524542e-01 -7.53375828e-01
6.81826591e-01 -2.66795337e-01 -2.27485165e-01 -1.17984843e+00
-6.68099463e-01 -6.08605504e-01 -2.40999475e-01 1.11321604e+00
7.43915200e-01 -3.63978267e-01 4.58876580e-01 -1.61714882e-01
-5.73878527e-01 -1.10481739e+00 -1.01773643e+00 -2.98854142e-01
2.38162547e-01 -2.64719903e-01 6.54572666e-01 7.53408253e-01
1.37475714e-01 9.20616090e-01 -6.72724545e-02 -4.22772728e-02
1.54680267e-01 -6.75195549e-03 1.09729484e-01 -1.00109780e+00
2.62245182e-02 -7.02240407e-01 -1.82484210e-01 -8.07095528e-01
4.44776386e-01 -9.77514982e-01 1.34645909e-01 -1.53708982e+00
2.86639631e-01 6.03557289e-01 -7.21368968e-01 7.57796526e-01
-6.86084807e-01 1.87658548e-01 1.69351876e-01 -2.99399734e-01
-7.90641487e-01 7.22711265e-01 9.80469584e-01 -3.19077313e-01
9.73863453e-02 -5.02903402e-01 -1.16076458e+00 4.41684753e-01
4.88586038e-01 -3.77823174e-01 -2.88495179e-02 -4.74342197e-01
4.97632474e-01 -1.28889859e-01 1.97664365e-01 -6.87608302e-01
1.83163032e-01 3.12636048e-01 5.04823506e-01 -1.06482184e+00
2.13885814e-01 -1.99421793e-01 -6.94737911e-01 5.61387479e-01
-5.95213294e-01 5.09482861e-01 5.70576966e-01 3.91394675e-01
-2.54761368e-01 -1.55848533e-01 6.91242993e-01 -3.11525434e-01
-7.20610559e-01 -2.56155152e-02 -5.23098588e-01 5.68218350e-01
5.07927001e-01 2.11905584e-01 -5.47278404e-01 -1.07186235e-01
-8.64919126e-01 5.17890811e-01 8.98203440e-03 6.87104106e-01
7.93566585e-01 -1.25676143e+00 -8.67981732e-01 2.78309852e-01
2.07473129e-01 -3.63361567e-01 1.28089502e-01 9.72468436e-01
-3.57926860e-02 9.89871383e-01 -1.84879005e-01 -4.65070754e-01
-1.13789201e+00 8.65150869e-01 5.19839287e-01 -6.00948930e-01
-3.57936382e-01 1.17647505e+00 1.56152666e-01 -5.76925814e-01
2.39509761e-01 -8.35694909e-01 -5.83590031e-01 2.52538353e-01
7.18569517e-01 -3.47543992e-02 6.75321877e-01 -1.71078593e-01
-7.74543822e-01 4.59483862e-01 -3.88446391e-01 1.77595064e-01
1.19673562e+00 5.30349985e-02 -4.55253929e-01 4.52510118e-01
9.71219957e-01 7.10861832e-02 -4.59356308e-01 -1.17656671e-01
2.58791953e-01 2.04522282e-01 1.37869373e-01 -1.19480121e+00
-7.60373056e-01 8.52857530e-01 3.40137452e-01 3.58950533e-02
1.37541556e+00 -1.80662513e-01 6.12012565e-01 4.79033232e-01
-1.81306422e-01 -8.56180012e-01 1.08923554e-01 1.19086075e+00
1.47817862e+00 -1.10255742e+00 -2.18227908e-01 -5.33006079e-02
-6.09914243e-01 1.02924526e+00 9.10737514e-01 -1.53325453e-01
3.82761180e-01 3.99080157e-01 1.10592172e-01 -4.81753528e-01
-1.32205951e+00 -2.56872922e-01 3.31414998e-01 4.19994414e-01
9.24953878e-01 -3.76530915e-01 -6.04687214e-01 1.25042641e+00
-1.42482504e-01 8.10898934e-03 6.47854388e-01 8.48895431e-01
-3.63768578e-01 -9.96564567e-01 -2.85666406e-01 3.53964984e-01
-9.62303698e-01 -8.28720331e-01 -7.63811111e-01 9.46326733e-01
2.63596684e-01 6.51812017e-01 -8.49924013e-02 -1.85942888e-01
9.30171072e-01 4.42027897e-01 4.41083252e-01 -8.66200626e-01
-1.27132404e+00 -3.86090204e-02 1.42375797e-01 -8.11885417e-01
-1.51238695e-01 -5.54226220e-01 -1.37723362e+00 4.49619368e-02
-4.55356628e-01 1.18674681e-01 1.03682421e-01 9.42099094e-01
5.42766035e-01 1.21357846e+00 5.57764200e-03 -6.09531045e-01
-7.57744253e-01 -1.23240817e+00 -2.44653001e-01 2.97282010e-01
6.29262030e-01 -6.34513795e-01 -1.96176678e-01 -4.82621670e-01] | [8.780658721923828, 8.799766540527344] |
0466f568-917a-45ad-ae4c-74669c615bce | understanding-and-mitigating-copying-in | 2305.20086 | null | https://arxiv.org/abs/2305.20086v1 | https://arxiv.org/pdf/2305.20086v1.pdf | Understanding and Mitigating Copying in Diffusion Models | Images generated by diffusion models like Stable Diffusion are increasingly widespread. Recent works and even lawsuits have shown that these models are prone to replicating their training data, unbeknownst to the user. In this paper, we first analyze this memorization problem in text-to-image diffusion models. While it is widely believed that duplicated images in the training set are responsible for content replication at inference time, we observe that the text conditioning of the model plays a similarly important role. In fact, we see in our experiments that data replication often does not happen for unconditional models, while it is common in the text-conditional case. Motivated by our findings, we then propose several techniques for reducing data replication at both training and inference time by randomizing and augmenting image captions in the training set. | ['Tom Goldstein', 'Jonas Geiping', 'Micah Goldblum', 'Vasu Singla', 'Gowthami Somepalli'] | 2023-05-31 | null | null | null | null | ['image-captioning', 'memorization'] | ['computer-vision', 'natural-language-processing'] | [ 4.48416501e-01 2.65335798e-01 -2.47218564e-01 -9.32376757e-02
-4.06614095e-01 -8.39421749e-01 1.05805695e+00 1.59645960e-01
-5.11080444e-01 9.43757296e-01 3.11221927e-01 -6.11620009e-01
7.86276013e-02 -6.78474903e-01 -1.14189625e+00 -8.22711945e-01
1.79064840e-01 5.47066808e-01 1.78741530e-01 1.13951616e-01
4.27019238e-01 2.35349596e-01 -1.42882752e+00 1.89816937e-01
9.79731441e-01 2.79906511e-01 2.50648439e-01 8.81447434e-01
5.78602441e-02 1.37947166e+00 -9.65367258e-01 -6.50966704e-01
2.30577946e-01 -7.58804262e-01 -7.91666031e-01 4.53492433e-01
6.30787730e-01 -6.19782448e-01 -4.41707909e-01 1.28189576e+00
2.52833933e-01 -2.67920736e-02 9.76274252e-01 -1.44191384e+00
-1.32027805e+00 9.24479187e-01 -8.82684231e-01 2.49347389e-01
6.18909113e-02 2.20231220e-01 6.96279347e-01 -6.11224949e-01
1.02978027e+00 1.04645240e+00 2.71707028e-01 7.06182539e-01
-1.28381538e+00 -5.34980476e-01 9.77489799e-02 4.56538796e-03
-1.19917512e+00 -5.08743405e-01 5.59251010e-01 -5.65277338e-01
5.80399573e-01 2.75078624e-01 4.02530402e-01 1.35332620e+00
3.83783966e-01 9.25467849e-01 1.33930659e+00 -6.87394798e-01
2.64545083e-01 4.55956310e-01 1.13964872e-02 6.20218456e-01
6.05031312e-01 -8.88774842e-02 -4.92462575e-01 -2.80349582e-01
8.80669177e-01 -2.62821108e-01 -3.38633239e-01 -1.14519343e-01
-8.58832717e-01 9.72786546e-01 1.04440503e-01 4.25343156e-01
-5.98681867e-01 4.94260013e-01 2.38905177e-01 4.45061624e-01
6.65747821e-01 1.33784309e-01 -8.60419124e-02 -1.41926959e-01
-1.00201964e+00 2.54918188e-01 6.25500321e-01 9.18678164e-01
7.31251657e-01 -2.40638971e-01 -2.70242859e-02 5.91862440e-01
2.67877012e-01 5.66989481e-01 7.23770916e-01 -8.11297655e-01
3.69080335e-01 1.36876702e-01 2.29192585e-01 -1.00046933e+00
3.00384253e-01 -1.77846014e-01 -8.44075203e-01 1.08093470e-01
5.59501529e-01 -2.48893216e-01 -1.03266716e+00 1.99638844e+00
1.36720344e-01 1.74057916e-01 4.05153744e-02 4.45799857e-01
7.95291066e-02 6.62256837e-01 1.78270921e-01 -3.50408763e-01
8.81714046e-01 -8.62483501e-01 -8.72203052e-01 -9.33194458e-02
6.94871664e-01 -8.20376813e-01 9.36937034e-01 2.77944416e-01
-1.05728841e+00 -2.62985140e-01 -8.81885767e-01 -5.96050844e-02
-1.37146980e-01 -2.31179401e-01 3.22730273e-01 9.59549606e-01
-1.15689898e+00 4.63166833e-01 -5.87384820e-01 -5.38789988e-01
5.06180167e-01 -9.36748087e-02 -1.77471027e-01 -2.88807064e-01
-1.10278738e+00 9.18120682e-01 -2.36479193e-02 -2.37327442e-01
-1.07599330e+00 -4.52037781e-01 -4.27354306e-01 -1.75581813e-01
3.24842006e-01 -8.39024842e-01 1.26154959e+00 -1.45391130e+00
-1.16145599e+00 8.78799140e-01 -4.38452214e-01 -6.13553107e-01
1.03120363e+00 -2.09090739e-01 -8.74884799e-02 7.84896463e-02
1.14060476e-01 6.83046043e-01 1.37682343e+00 -1.73085535e+00
-3.83828521e-01 -3.01048458e-01 1.30184755e-01 4.53332663e-02
-5.03724098e-01 -1.07035093e-01 -3.58829767e-01 -7.47509360e-01
-3.61241907e-01 -1.10536027e+00 -1.96888596e-01 1.81246966e-01
-7.85188198e-01 -1.42394528e-01 7.59761691e-01 -5.76315463e-01
1.15660286e+00 -2.11621046e+00 -3.32243219e-02 3.48371416e-01
3.26044530e-01 5.92170767e-02 -2.31539488e-01 5.73555052e-01
-9.27308425e-02 5.56799173e-01 -2.60448426e-01 -7.02624857e-01
-7.82331377e-02 4.40136015e-01 -6.42036319e-01 5.26680410e-01
-7.65733570e-02 9.02056992e-01 -6.05781496e-01 -4.55615968e-01
-1.71006531e-01 4.70046073e-01 -4.89888191e-01 8.32033623e-03
-3.34362149e-01 3.44947219e-01 -2.62632608e-01 1.27173156e-01
8.32874417e-01 -5.84217548e-01 2.03966618e-01 4.69561160e-01
3.17238346e-02 -1.08504713e-01 -9.57333326e-01 1.19654739e+00
-1.96975306e-01 9.69648659e-01 -2.78130978e-01 -6.63056910e-01
2.36851141e-01 3.60621929e-01 7.97825586e-03 -6.12361312e-01
7.07300901e-02 1.62224062e-02 -1.52847311e-02 -5.84496140e-01
1.03552914e+00 -3.05226326e-01 3.83406132e-01 9.77847397e-01
-3.11186433e-01 -3.99006605e-02 2.12174818e-01 7.54396737e-01
9.63117540e-01 -2.66456664e-01 -1.12402767e-01 -1.13534741e-01
8.79409313e-02 1.25286102e-01 7.64061138e-02 1.34658289e+00
1.11946903e-01 5.22577465e-01 6.22730792e-01 1.11412011e-01
-1.49499512e+00 -9.26445603e-01 -5.63049205e-02 7.45721579e-01
5.80679663e-02 -3.76384377e-01 -1.07578456e+00 -6.57270253e-01
-1.24825485e-01 1.04193544e+00 -8.74577343e-01 -1.38421118e-01
-6.84942901e-02 -5.49885035e-01 5.50602138e-01 3.13423067e-01
3.68640095e-01 -8.27763081e-01 -5.12003720e-01 2.76288111e-02
-1.56421661e-01 -8.53217542e-01 -5.26776016e-01 -2.94247448e-01
-8.93244445e-01 -7.86510646e-01 -1.15889919e+00 -4.01682466e-01
1.14781034e+00 3.30379903e-01 1.02236080e+00 4.77523714e-01
-2.72445023e-01 6.47505760e-01 -3.42444152e-01 -5.51177502e-01
-1.03129208e+00 -3.48401740e-02 -1.13665096e-01 -1.23201525e-02
3.40154432e-02 -2.35595673e-01 -5.69453418e-01 1.77212581e-01
-1.50920618e+00 6.33340776e-02 5.01710176e-01 6.09265208e-01
1.47407904e-01 3.36236626e-01 2.73359060e-01 -1.35915685e+00
9.53000784e-01 -5.88169336e-01 -4.08151865e-01 3.37307602e-01
-8.21387708e-01 1.05165038e-02 1.99457064e-01 -6.95306957e-01
-1.23155236e+00 -1.81014031e-01 7.63666853e-02 -4.38372761e-01
-2.02272236e-01 7.95695484e-01 4.00584638e-01 1.48778230e-01
9.18917418e-01 3.63800138e-01 1.65507004e-01 -2.58232057e-01
6.23852253e-01 5.63751698e-01 6.18600026e-02 -5.56585491e-01
8.54405046e-01 7.53910482e-01 -4.15677220e-01 -1.06341791e+00
-5.10172069e-01 9.38983932e-02 -3.61459792e-01 -3.64323348e-01
8.36194754e-01 -8.27741265e-01 -2.26353601e-01 8.08174193e-01
-1.30278134e+00 -6.04810715e-01 -5.21253526e-01 1.64318979e-01
-3.88747185e-01 5.75655997e-01 -7.55067170e-01 -9.27810550e-01
1.46089047e-01 -1.14917815e+00 6.52131557e-01 1.31523743e-01
-1.68710306e-01 -1.12024820e+00 1.94659993e-01 3.28853816e-01
5.57604671e-01 -2.31216267e-01 1.12361085e+00 -4.26934004e-01
-8.25395644e-01 -2.70154834e-01 -2.11773083e-01 2.91870117e-01
8.78154486e-02 4.13481951e-01 -9.32323158e-01 -2.38207877e-01
1.48156807e-01 -3.27270746e-01 9.83557463e-01 5.14582932e-01
8.49500775e-01 -5.92907667e-01 -3.64185542e-01 -3.16205502e-01
1.44523585e+00 -1.06957264e-01 1.10963988e+00 4.11234125e-02
4.32342350e-01 7.61279464e-01 2.66829729e-01 3.85610789e-01
2.72810340e-01 2.48606563e-01 5.59205301e-02 -1.42299175e-01
-1.09787956e-01 -5.49719810e-01 4.02324736e-01 5.33495188e-01
2.31560674e-02 -7.96203494e-01 -6.89199746e-01 5.78036427e-01
-1.84278131e+00 -1.13512075e+00 -3.77816409e-01 2.14999533e+00
9.95530128e-01 9.13069993e-02 1.64504722e-02 -1.08494788e-01
9.21534240e-01 3.86559479e-02 -4.34533894e-01 -7.92457163e-02
-2.57418871e-01 -2.58758664e-01 7.32176781e-01 7.29902506e-01
-5.94309449e-01 8.17059457e-01 7.27898836e+00 9.25692141e-01
-1.00571692e+00 4.47290331e-01 1.01409853e+00 -9.02694017e-02
-7.00139582e-01 1.49314627e-01 -4.44460571e-01 5.16107202e-01
8.66371870e-01 -4.63873923e-01 2.61155009e-01 6.02969646e-01
7.11196586e-02 -6.57619715e-01 -9.61582601e-01 6.50191367e-01
3.01632464e-01 -1.34641206e+00 3.64370525e-01 3.35972548e-01
1.07799757e+00 -1.35027170e-01 5.57154715e-01 -1.94497302e-01
6.31402433e-01 -8.43528330e-01 1.03416812e+00 4.67490435e-01
5.16863465e-01 -4.67067271e-01 2.71848798e-01 5.67475498e-01
-3.31071079e-01 1.70051400e-02 -4.90964741e-01 1.03266321e-01
2.99432218e-01 8.39926958e-01 -5.79580307e-01 1.44884348e-01
2.98613846e-01 2.92780519e-01 -8.39009881e-01 8.70598376e-01
-4.72979486e-01 8.15647423e-01 -1.48123607e-01 1.29031032e-01
9.03427377e-02 -7.68764094e-02 4.99243766e-01 7.87418127e-01
3.43026131e-01 -1.86709166e-01 -4.62801427e-01 9.37672973e-01
-2.82984823e-01 9.97110549e-03 -1.08705688e+00 -4.13521707e-01
2.17013851e-01 7.35906005e-01 -8.85020256e-01 -6.17033243e-01
-3.56284916e-01 1.44012058e+00 2.25822866e-01 5.89117885e-01
-9.57255721e-01 1.32365286e-01 2.85717696e-01 3.02389681e-01
2.41411924e-01 -3.95773947e-01 -2.04173133e-01 -1.19388020e+00
-6.16115257e-02 -7.03109920e-01 8.03021714e-02 -1.04031467e+00
-1.47112131e+00 4.27878767e-01 1.54545866e-02 -5.76205015e-01
-2.08326846e-01 -1.48202837e-01 -4.82441664e-01 7.66658843e-01
-1.44314110e+00 -8.43036592e-01 1.67328700e-01 6.71679556e-01
4.63059366e-01 2.40092218e-01 4.68570501e-01 2.01290220e-01
-6.00239217e-01 4.66489017e-01 4.61624473e-01 -7.68917752e-03
8.26967478e-01 -9.73499715e-01 4.38230842e-01 1.07998979e+00
4.26051885e-01 1.07236278e+00 1.12672055e+00 -1.03613603e+00
-1.20295405e+00 -8.05166960e-01 1.00862372e+00 -5.07242143e-01
7.16937304e-01 -2.94589221e-01 -9.18747306e-01 1.00488389e+00
7.78599679e-01 -5.39745688e-01 6.12255037e-01 -7.34062344e-02
-5.25425255e-01 4.26712513e-01 -1.05872524e+00 9.50636983e-01
8.05523932e-01 -8.10931921e-01 -3.03546131e-01 6.86605275e-01
5.14641881e-01 -5.44381514e-03 -3.22519183e-01 -3.04057539e-01
2.21603811e-01 -9.06490386e-01 4.62641537e-01 -5.77118874e-01
9.07577515e-01 6.84402362e-02 9.38662887e-02 -1.17751634e+00
-6.33212551e-02 -7.17352629e-01 8.23182911e-02 1.39149165e+00
8.40977192e-01 -6.84731185e-01 9.63026285e-01 1.15799630e+00
5.30116856e-01 8.55109468e-02 -1.00520897e+00 -9.30554330e-01
4.38712835e-01 -3.50547850e-01 6.00253455e-02 1.11152506e+00
-1.35421887e-01 3.36465925e-01 -6.58888757e-01 1.98018625e-02
6.91744685e-01 -2.30239078e-01 8.58904123e-01 -9.44901288e-01
-3.32483947e-01 -2.91562319e-01 2.43110675e-02 -1.34991550e+00
-4.33597015e-03 -5.82497954e-01 1.00838736e-01 -1.46682477e+00
5.27703524e-01 -5.48740327e-01 1.68391392e-01 1.49965659e-01
-2.57723629e-01 3.00666064e-01 1.66084915e-01 5.76183736e-01
-3.69371891e-01 2.66516268e-01 1.39136422e+00 -1.26176849e-01
1.00541838e-01 -2.56104320e-01 -7.75421023e-01 3.71159405e-01
7.08729088e-01 -8.55645835e-01 -7.01006651e-01 -6.33274317e-01
5.64758122e-01 -8.27246606e-02 4.35844004e-01 -6.01358891e-01
4.18572962e-01 -8.76236558e-02 6.39031678e-02 -2.57038057e-01
8.86862725e-02 -7.95416534e-01 3.59268755e-01 5.26519120e-01
-6.42679691e-01 2.35855177e-01 2.64908615e-02 9.66071486e-01
1.05967209e-01 -7.56751895e-01 6.59753442e-01 -2.27434456e-01
-2.72087365e-01 1.44317538e-01 -9.93796766e-01 -2.78858235e-03
1.09797800e+00 -6.45096824e-02 -6.51096463e-01 -8.76544714e-01
-4.91397500e-01 -1.95255682e-01 7.32848167e-01 2.67727315e-01
4.93937314e-01 -9.47465181e-01 -6.37359142e-01 2.25660317e-02
-7.76260421e-02 -3.46622795e-01 4.45820630e-01 8.70927334e-01
-3.79201323e-01 -9.12539754e-03 8.54705945e-02 -4.77934718e-01
-1.35159707e+00 7.80216753e-01 -2.60166042e-02 5.98552227e-02
-3.11586618e-01 9.25924122e-01 2.38368168e-01 2.10896730e-01
6.66136891e-02 -1.11541562e-01 3.08044195e-01 2.04819992e-01
5.36211550e-01 1.18683122e-01 -3.14626306e-01 -5.41773856e-01
1.87168539e-01 2.27754027e-01 -7.23517418e-01 -6.30488455e-01
1.05962288e+00 -3.75895411e-01 -2.45143771e-01 6.08462572e-01
9.86169755e-01 3.46362829e-01 -1.16274118e+00 -2.29009241e-01
-2.48001173e-01 -6.28668308e-01 -1.53913066e-01 -5.70336998e-01
-9.36220407e-01 8.57323766e-01 4.11942124e-01 7.62538016e-01
7.05931425e-01 1.45455301e-01 5.40651262e-01 2.32204527e-01
1.54095918e-01 -1.11708331e+00 2.47222781e-01 2.42761344e-01
8.54823828e-01 -1.10554612e+00 1.30303591e-01 -2.92307764e-01
-9.18763518e-01 6.02160335e-01 3.01760942e-01 3.36677977e-03
7.37914383e-01 2.54251480e-01 1.51592255e-01 -1.91399291e-01
-9.06808853e-01 1.92547202e-01 -3.69468570e-01 6.11551762e-01
2.98032433e-01 -2.91551083e-01 -3.64254534e-01 -1.64680004e-01
1.21479481e-01 6.37570694e-02 9.61451828e-01 1.16333246e+00
-3.48511666e-01 -1.25303352e+00 -3.98316413e-01 3.25608045e-01
-4.90238994e-01 -2.92890072e-01 -6.46517158e-01 6.56584501e-01
-1.37572989e-01 1.09460056e+00 1.91763774e-01 -5.00610396e-02
-1.63950861e-01 1.27601042e-01 5.75201094e-01 -4.42496866e-01
-4.78899658e-01 2.36344896e-02 -1.33416085e-02 8.87103472e-03
-5.81418931e-01 -7.71700084e-01 -7.77649581e-01 -7.10262537e-01
-6.30065143e-01 1.05269015e-01 6.37842119e-01 8.88743222e-01
4.47160453e-01 2.47678325e-01 4.37119752e-01 -4.05855715e-01
-5.58132887e-01 -9.13884461e-01 -7.92606115e-01 6.73314691e-01
2.77657211e-01 -2.74773151e-01 -5.34448385e-01 5.73161840e-01] | [11.443683624267578, -0.2229757308959961] |
5691eec6-3be3-4cde-a878-dfc540184494 | machine-unlearning-via-gan | 2111.11869 | null | https://arxiv.org/abs/2111.11869v1 | https://arxiv.org/pdf/2111.11869v1.pdf | Machine unlearning via GAN | Machine learning models, especially deep models, may unintentionally remember information about their training data. Malicious attackers can thus pilfer some property about training data by attacking the model via membership inference attack or model inversion attack. Some regulations, such as the EU's GDPR, have enacted "The Right to Be Forgotten" to protect users' data privacy, enhancing individuals' sovereignty over their data. Therefore, removing training data information from a trained model has become a critical issue. In this paper, we present a GAN-based algorithm to delete data in deep models, which significantly improves deleting speed compared to retraining from scratch, especially in complicated scenarios. We have experimented on five commonly used datasets, and the experimental results show the efficiency of our method. | ['Yiwen Wang', 'Yao Huang', 'Kongyang Chen'] | 2021-11-22 | null | null | null | null | ['membership-inference-attack'] | ['computer-vision'] | [ 1.31950229e-01 5.34641027e-01 -2.84861028e-01 -3.13986510e-01
-4.13741022e-01 -7.96467066e-01 4.60867643e-01 -2.87064970e-01
-5.72859406e-01 1.06738734e+00 -7.09152371e-02 -6.76186502e-01
2.97451288e-01 -1.09670651e+00 -7.59678125e-01 -6.01198375e-01
3.57721061e-01 2.09570661e-01 -3.45209002e-01 2.89942890e-01
2.56115139e-01 5.18952191e-01 -8.58849168e-01 1.55806288e-01
1.01204538e+00 5.04846454e-01 -6.23629570e-01 1.73807845e-01
2.14750573e-01 6.82827234e-01 -9.62643802e-01 -1.33533347e+00
5.54375887e-01 -3.16347897e-01 -8.26813698e-01 -4.51359093e-01
1.62736550e-01 -7.04107821e-01 -4.50478882e-01 1.43562341e+00
2.84719169e-01 -1.76502347e-01 4.02924061e-01 -1.50119960e+00
-7.61486888e-01 5.72462857e-01 -3.79980177e-01 -1.55749276e-01
-1.19406238e-01 2.93820471e-01 4.20029074e-01 -2.96978056e-01
3.37422729e-01 9.59809482e-01 6.65289998e-01 1.19142830e+00
-1.32466459e+00 -1.22355223e+00 -1.03231706e-01 -1.78114697e-01
-1.68650353e+00 -5.11414170e-01 6.77812219e-01 -4.80638266e-01
4.73177731e-01 8.25938165e-01 4.74820316e-01 1.47729719e+00
2.63180554e-01 5.05918086e-01 1.16167700e+00 -1.77647159e-01
2.34164119e-01 4.65588063e-01 1.70348689e-01 4.97080684e-01
1.18203139e+00 2.61499107e-01 -6.81286007e-02 -7.69635260e-01
6.58681452e-01 -9.77338199e-03 -1.90805003e-01 -1.86972499e-01
-3.10804993e-01 1.07974470e+00 1.73618302e-01 -2.15016920e-02
-1.13267265e-01 2.73338556e-01 3.71773839e-01 2.85895050e-01
5.39971769e-01 2.65523046e-01 -5.48828006e-01 1.46700397e-01
-3.17090631e-01 2.46115342e-01 8.17493558e-01 6.66715205e-01
6.65530920e-01 -7.05685019e-02 3.16781491e-01 4.58660692e-01
4.18400884e-01 3.95774543e-01 4.84984726e-01 -6.44780934e-01
5.84405363e-01 6.80083454e-01 2.86588103e-01 -1.17501175e+00
1.46033376e-01 -3.74762297e-01 -1.24312496e+00 2.08663598e-01
3.02724838e-01 -6.51564300e-01 -9.85504150e-01 1.97201908e+00
3.14229906e-01 3.08750778e-01 2.06195489e-01 4.38256592e-01
5.36658525e-01 1.76125541e-01 3.91564846e-01 1.72623277e-01
1.06603575e+00 -2.80507982e-01 -7.86244154e-01 -1.98563337e-02
8.27000260e-01 -1.61581293e-01 8.08954597e-01 5.84460080e-01
-7.05496788e-01 -3.20570737e-01 -8.96819472e-01 -9.37644318e-02
-7.24543452e-01 -1.41724408e-01 8.77077281e-01 1.48599362e+00
-4.50468570e-01 5.74787378e-01 -8.24874997e-01 9.97381881e-02
1.23537791e+00 5.83366394e-01 -6.71579301e-01 1.69647276e-01
-1.74660182e+00 7.19870746e-01 5.61180174e-01 2.04686105e-01
-1.05976617e+00 -9.03515875e-01 -8.36336255e-01 1.32924080e-01
2.23676354e-01 -8.29958439e-01 8.94439578e-01 -7.67981350e-01
-1.01759171e+00 1.13907921e+00 1.67936862e-01 -1.02262950e+00
9.73492622e-01 -3.52687031e-01 -8.64307165e-01 -4.10002261e-01
-3.02894205e-01 5.07334888e-01 9.16760325e-01 -1.27054811e+00
-3.40875000e-01 -5.91291368e-01 2.68439084e-01 -3.21280330e-01
-4.63256061e-01 -1.40644163e-01 8.58922005e-02 -7.86319733e-01
-4.30273056e-01 -9.47560430e-01 -4.12653029e-01 -2.38488823e-01
-8.50481689e-01 1.35199949e-01 1.10153818e+00 -1.03598130e+00
1.61584437e+00 -2.11815071e+00 -4.43080246e-01 4.12016153e-01
2.53312647e-01 9.67889190e-01 3.61411870e-01 2.92896062e-01
-5.28120920e-02 8.97510648e-01 -3.74946892e-01 -4.84115452e-01
-8.17620568e-03 4.14895147e-01 -5.56157410e-01 4.95337248e-01
-2.26559669e-01 1.03226340e+00 -2.28988171e-01 -9.96197090e-02
1.87818818e-02 4.28511232e-01 -5.68248332e-01 2.66639311e-02
-1.53104290e-01 6.66808963e-01 -4.67017800e-01 3.19978595e-01
1.29275680e+00 -5.51449880e-02 2.26783514e-01 2.97153652e-01
4.34617907e-01 2.48661309e-01 -8.33440065e-01 1.25472426e+00
-2.10297897e-01 4.28294450e-01 4.47448306e-02 -6.11994147e-01
7.53244638e-01 2.53595233e-01 -1.11140579e-01 -4.19013053e-01
2.61336476e-01 -2.18707964e-01 -4.43023890e-02 -6.13157928e-01
2.82530695e-01 -1.46126300e-01 -3.34072560e-01 5.79171538e-01
-5.14791429e-01 5.34002006e-01 -7.30157137e-01 8.90828967e-02
8.07735920e-01 -1.95887387e-01 1.85339078e-01 -1.54060468e-01
4.80840683e-01 -2.71300226e-01 8.08511496e-01 7.70197451e-01
2.59512383e-02 1.29510671e-01 6.42142594e-01 -6.30858541e-01
-7.54613519e-01 -8.79626095e-01 -4.00555283e-02 5.19197464e-01
-2.92885989e-01 -3.64255667e-01 -1.13922536e+00 -1.58120573e+00
3.92335922e-01 1.19606698e+00 -9.18675363e-01 -5.07604718e-01
-4.69752997e-01 -6.80937529e-01 1.24437606e+00 3.51842642e-01
9.98564243e-01 -6.77677155e-01 -4.37619656e-01 -1.86851352e-01
-1.66631848e-01 -6.60325229e-01 -3.23865116e-01 -1.74298763e-01
-7.99903393e-01 -1.14529538e+00 -1.36546016e-01 -2.29025111e-01
1.11506283e+00 -9.51316059e-02 6.65346622e-01 5.19612730e-01
-1.30047455e-01 -6.76082969e-02 1.47841558e-01 -8.75895977e-01
-7.57503092e-01 3.35187465e-01 -8.81253332e-02 1.46361932e-01
7.52681553e-01 -4.59451407e-01 -4.17781889e-01 1.64648339e-01
-1.24677420e+00 -3.38041410e-02 1.97048873e-01 4.92037743e-01
5.29259481e-02 3.97997767e-01 8.23286414e-01 -2.06845188e+00
6.40863359e-01 -3.25792938e-01 -8.07317376e-01 3.45863640e-01
-6.83358312e-01 -1.20965488e-01 5.90898037e-01 -4.02274042e-01
-1.21172845e+00 -1.19289771e-01 -3.66191208e-01 -3.01476091e-01
-4.21261758e-01 4.03518975e-01 -9.87192571e-01 -2.48248458e-01
5.11596560e-01 2.89390851e-02 2.37655550e-01 -7.92022526e-01
1.70289963e-01 7.97058225e-01 4.13432658e-01 -3.77586633e-01
1.10282314e+00 5.58365464e-01 -5.13813198e-02 -4.94153857e-01
-5.99222124e-01 2.38596410e-01 -3.55497211e-01 1.01898730e-01
6.40088379e-01 -8.73068929e-01 -1.05005527e+00 9.22755778e-01
-1.08916855e+00 -1.14018545e-01 5.22774048e-02 3.30700845e-01
1.49258599e-01 5.42336881e-01 -5.14850914e-01 -6.77986026e-01
-4.77525681e-01 -6.32605374e-01 1.67087704e-01 2.27391705e-01
-2.97299445e-01 -1.24146485e+00 -9.62016825e-03 7.18190730e-01
5.03991306e-01 6.35043263e-01 9.45950270e-01 -1.05276597e+00
-4.45491850e-01 -5.41245520e-01 1.88939810e-01 7.30256915e-01
1.10281557e-01 -1.47500306e-01 -1.26502371e+00 -4.91453737e-01
3.51002038e-01 -4.71212342e-02 8.06847513e-01 -8.95667523e-02
1.70453322e+00 -1.08215523e+00 -4.52345073e-01 6.58510387e-01
1.27538347e+00 2.46654242e-01 1.35599172e+00 -4.59246822e-02
7.69497991e-01 4.63285327e-01 2.77072608e-01 5.76696992e-01
-2.95466441e-03 2.85475373e-01 6.08235657e-01 -1.34006172e-01
4.66392815e-01 -7.72619784e-01 4.94790450e-02 1.15956001e-01
7.55881816e-02 -2.81974226e-01 -5.39272785e-01 1.51804328e-01
-1.69009471e+00 -9.66591895e-01 -2.43173704e-01 2.50269508e+00
7.57652521e-01 2.25623459e-01 -1.19666226e-01 -1.60736457e-01
7.14227140e-01 -3.01399119e-02 -7.61249006e-01 -4.10574913e-01
2.42475450e-01 3.24997276e-01 9.13394213e-01 5.61692774e-01
-1.16404855e+00 1.07916546e+00 6.21876001e+00 8.25994313e-01
-9.19157028e-01 2.74513304e-01 1.01464951e+00 -1.66157808e-03
-6.68295205e-01 1.20339952e-01 -8.57821286e-01 7.56817102e-01
7.98049092e-01 -4.19626445e-01 3.71851802e-01 9.83786523e-01
-3.66211832e-02 1.71048701e-01 -8.56095135e-01 8.11735213e-01
-8.81260857e-02 -1.38423657e+00 2.06803471e-01 7.66942561e-01
4.96808618e-01 -5.67385912e-01 3.94894212e-01 4.58933502e-01
7.86337435e-01 -1.43402982e+00 2.08507806e-01 6.05394185e-01
7.80415595e-01 -1.29567850e+00 6.40368164e-01 6.48500323e-01
-3.60035151e-01 6.99545443e-02 -5.87581158e-01 -8.45658854e-02
-1.06458016e-01 4.77873713e-01 -8.35103333e-01 6.50532246e-01
4.97254133e-01 2.40208179e-01 -8.36279333e-01 5.38292348e-01
-5.44319034e-01 7.36491144e-01 -1.46201089e-01 1.62837118e-01
-6.36976659e-02 -3.81973535e-01 1.41945228e-01 6.86593711e-01
8.84904638e-02 1.50394812e-01 -5.45454502e-01 1.10135603e+00
-6.08056724e-01 -4.38906223e-01 -1.07010269e+00 -8.10220167e-02
5.74009359e-01 1.00123644e+00 -7.68737569e-02 -1.88454926e-01
-1.90298855e-01 8.29335511e-01 7.76624009e-02 2.38137588e-01
-1.19188917e+00 -2.54487962e-01 9.41665530e-01 2.69789726e-01
3.88213969e-03 6.53312579e-02 -3.93538892e-01 -1.16565132e+00
-6.06510304e-02 -1.05710185e+00 6.24131024e-01 -5.00122905e-01
-1.17972410e+00 2.86495894e-01 -1.90977082e-01 -9.47097957e-01
-1.63692571e-02 -3.04989547e-01 -5.84965110e-01 1.05842412e+00
-9.79655504e-01 -1.26597035e+00 1.93217650e-01 7.30665684e-01
-3.14165443e-01 -3.26707423e-01 1.05624413e+00 3.30990195e-01
-8.03111017e-01 1.15517902e+00 -7.72284493e-02 7.79027343e-01
3.06942970e-01 -7.79094219e-01 7.23746598e-01 1.03671932e+00
6.40269741e-03 1.26781893e+00 4.10286874e-01 -1.06425595e+00
-1.10340631e+00 -1.34003592e+00 9.05830622e-01 -7.71555781e-01
1.35527831e-02 -7.66593039e-01 -1.15241420e+00 1.44737327e+00
3.01333934e-01 -3.26843679e-01 1.31416094e+00 1.17846698e-01
-5.15510201e-01 -4.46796417e-02 -1.84773517e+00 6.21822715e-01
8.18579793e-01 -6.17664337e-01 -4.55942512e-01 5.44840191e-03
7.74768829e-01 -1.86865076e-01 -6.83359563e-01 3.42356086e-01
5.83913684e-01 -8.44541728e-01 8.99961472e-01 -1.41260326e+00
-1.60673842e-01 -1.35388494e-01 4.36282977e-02 -1.03918338e+00
2.78092362e-02 -7.93720126e-01 -1.65789217e-01 1.62471521e+00
4.24180984e-01 -1.23996472e+00 1.28634715e+00 1.51041710e+00
4.74785388e-01 -8.58031362e-02 -1.02516437e+00 -5.20108759e-01
3.20745349e-01 -3.69271874e-01 1.26580024e+00 1.50586772e+00
-4.27903473e-01 6.62362874e-02 -9.59483802e-01 6.24718070e-01
6.74012363e-01 -4.30546701e-01 1.18847096e+00 -1.20855844e+00
-2.00466067e-01 7.09464699e-02 -1.98562816e-01 -6.29352331e-01
3.19965929e-01 -7.98041940e-01 -9.04004931e-01 -8.95850778e-01
6.63775951e-02 -4.48914498e-01 -2.19259053e-01 7.31874108e-01
9.28179454e-03 -3.56179588e-02 2.33251676e-02 9.73986834e-03
-8.99052024e-02 2.86545843e-01 1.03228164e+00 -7.63752386e-02
7.12084472e-02 5.13423681e-01 -1.33280241e+00 8.24533522e-01
1.13287103e+00 -1.00782216e+00 -5.19539475e-01 -3.43674719e-01
3.70229036e-01 -2.20322281e-01 6.31145418e-01 -8.94795179e-01
1.08797461e-01 -1.72179267e-01 6.06873333e-01 -4.55604613e-01
1.39966667e-01 -1.28477621e+00 1.09057128e+00 9.32905853e-01
-4.28103089e-01 -2.57708669e-01 1.60891518e-01 5.85318387e-01
8.04672241e-02 -4.36848283e-01 8.13296258e-01 -1.31068513e-01
-2.42223531e-01 4.84763533e-01 -1.35505706e-01 -4.69679162e-02
1.26879168e+00 1.41375929e-01 -7.15773404e-01 -2.75348693e-01
-9.45398092e-01 9.09325704e-02 6.74844444e-01 2.63893366e-01
5.26337266e-01 -1.25663149e+00 -5.36734521e-01 7.83754766e-01
-1.62071243e-01 -1.74588617e-02 4.53216940e-01 2.86169291e-01
-5.49827695e-01 2.40619496e-01 -1.53629184e-01 1.43739089e-01
-1.52002466e+00 9.57886875e-01 3.42153043e-01 -2.85668910e-01
-3.62011284e-01 5.20515323e-01 1.46323621e-01 -5.42723477e-01
2.03316957e-01 -2.67625488e-02 2.32851915e-02 -1.32370427e-01
6.87892973e-01 4.24791843e-01 -1.02026984e-01 -2.99527913e-01
-2.53317595e-01 -9.55867171e-02 -5.87205231e-01 1.34551287e-01
9.57218885e-01 2.45635360e-02 -2.14047849e-01 -3.46981347e-01
1.31291151e+00 2.92564958e-01 -1.10504496e+00 -4.30867821e-03
-2.35798240e-01 -8.88475060e-01 -3.73795718e-01 -9.05519307e-01
-1.21408689e+00 8.84248734e-01 2.90848583e-01 5.70347071e-01
8.94269288e-01 -3.24289143e-01 7.28122532e-01 2.16347560e-01
4.94055510e-01 -8.30248475e-01 -4.50919002e-01 1.85829494e-02
6.74105287e-01 -9.20047402e-01 2.04608198e-02 -4.72768158e-01
-7.81863451e-01 4.64192480e-01 7.40229666e-01 5.60054295e-02
7.63344347e-01 2.11399749e-01 1.84577942e-01 1.86786652e-01
-5.25035143e-01 4.93469059e-01 -2.74516016e-01 9.25025582e-01
-3.33408475e-01 3.21274787e-01 -2.21176341e-01 1.03762090e+00
-2.10555598e-01 2.23233446e-01 4.66925025e-01 8.42823446e-01
-1.19036343e-02 -1.58187497e+00 -3.47443342e-01 6.33965850e-01
-9.78898644e-01 1.89944787e-03 -6.54642999e-01 1.10700178e+00
3.10049713e-01 7.58558452e-01 -2.57207781e-01 -5.17529488e-01
2.06837162e-01 1.51316896e-01 1.39532599e-03 -4.45084482e-01
-8.59079778e-01 -4.31776166e-01 3.30067366e-01 -2.06532732e-01
6.64874353e-03 -4.71369863e-01 -9.32126224e-01 -1.18401980e+00
-2.94079036e-01 3.06734711e-01 3.27289551e-01 5.92054725e-01
7.25177526e-01 2.86249854e-02 7.36566722e-01 1.59436822e-01
-7.14352071e-01 -9.01265681e-01 -6.09110296e-01 6.57713592e-01
3.08178533e-02 -2.63807923e-01 -4.34295118e-01 -7.78011698e-03] | [5.9437055587768555, 7.143743991851807] |
fa662589-4bea-45df-8e3e-314b13c4037a | action-anticipation-by-predicting-future | 1808.00141 | null | http://arxiv.org/abs/1808.00141v1 | http://arxiv.org/pdf/1808.00141v1.pdf | Action Anticipation By Predicting Future Dynamic Images | Human action-anticipation methods predict what is the future action by
observing only a few portion of an action in progress. This is critical for
applications where computers have to react to human actions as early as
possible such as autonomous driving, human-robotic interaction, assistive
robotics among others. In this paper, we present a method for human action
anticipation by predicting the most plausible future human motion. We represent
human motion using Dynamic Images and make use of tailored loss functions to
encourage a generative model to produce accurate future motion prediction. Our
method outperforms the currently best performing action-anticipation methods by
4% on JHMDB-21, 5.2% on UT-Interaction and 5.1% on UCF 101-24 benchmarks. | ['Basura Fernando', 'Hongdong Li', 'Cristian Rodriguez'] | 2018-08-01 | null | null | null | null | ['action-anticipation'] | ['computer-vision'] | [ 2.80011654e-01 5.82470715e-01 -4.84056860e-01 -3.21526051e-01
-3.93663853e-01 -1.65460035e-01 9.63853717e-01 -2.41994843e-01
-6.07698143e-01 8.62658978e-01 6.30854249e-01 -9.83695760e-02
2.62429953e-01 -3.67239237e-01 -7.23608494e-01 -3.31849396e-01
-3.48890901e-01 5.28585017e-01 5.09522617e-01 -2.38826230e-01
8.02873150e-02 5.48192680e-01 -1.39514768e+00 5.43257058e-01
5.38188457e-01 5.38491786e-01 3.17704678e-01 1.22465444e+00
5.95124900e-01 1.50328410e+00 -1.07850038e-01 -2.27710068e-01
2.34811306e-01 -5.85256577e-01 -1.09011078e+00 2.46964589e-01
-8.44421163e-02 -6.90578043e-01 -6.47653103e-01 5.27735472e-01
3.17880303e-01 6.79943442e-01 7.01876700e-01 -1.48085213e+00
-1.56170025e-01 5.74438930e-01 -4.60302234e-01 1.51364937e-01
6.55370176e-01 7.45521724e-01 6.61870301e-01 -5.96361041e-01
1.21756732e+00 1.52719557e+00 4.08340156e-01 1.16833603e+00
-7.72131860e-01 -1.34224713e-01 2.10073039e-01 6.94235682e-01
-7.55075932e-01 -5.31785309e-01 6.26216769e-01 -4.66012359e-01
1.40656865e+00 7.78970793e-02 7.18612075e-01 1.30206215e+00
7.14296937e-01 1.51487803e+00 4.55157697e-01 -2.09122792e-01
2.32644975e-01 -6.24054611e-01 -3.16578954e-01 5.21803558e-01
-2.91311443e-01 2.82929331e-01 -3.97599161e-01 -2.46782415e-02
3.98989916e-01 -6.20317273e-02 -2.07233410e-02 -1.61609799e-01
-1.68368566e+00 6.49804115e-01 3.79023820e-01 -9.17235911e-02
-1.06350100e+00 8.32229793e-01 4.68237311e-01 -1.31760016e-01
2.34996244e-01 3.76770288e-01 -2.48387232e-01 -8.47850800e-01
-6.36241257e-01 6.41532302e-01 4.90047425e-01 8.18470299e-01
3.03534269e-01 1.17739461e-01 -4.27550495e-01 2.92231023e-01
2.66198903e-01 3.15087885e-01 5.64669788e-01 -1.53070033e+00
4.35079157e-01 2.24401414e-01 7.33892858e-01 -5.68758488e-01
-4.29789871e-01 2.14053363e-01 -6.48624122e-01 3.52184385e-01
2.23837122e-01 -3.53184193e-01 -1.02840972e+00 1.34092712e+00
3.49573731e-01 4.03771758e-01 2.73060977e-01 9.44050610e-01
1.85996741e-01 8.22789729e-01 5.19694388e-01 -5.32518402e-02
9.31042492e-01 -1.33127809e+00 -5.90431869e-01 -8.04814160e-01
9.94255245e-01 -6.11069441e-01 7.34136283e-01 3.62554580e-01
-1.07326818e+00 -7.16170609e-01 -6.94191575e-01 -9.16694254e-02
2.56751120e-01 2.15713024e-01 7.67190039e-01 -7.27684295e-04
-8.29038322e-01 8.81200492e-01 -1.41967964e+00 -5.09436667e-01
3.27064842e-01 2.66605496e-01 -4.15095419e-01 -1.65808469e-01
-9.69971538e-01 1.05172253e+00 6.46903813e-01 1.32951573e-01
-1.07591391e+00 -2.45680362e-01 -8.44425201e-01 -3.08293194e-01
2.07537189e-01 -8.01260173e-01 1.56773829e+00 -8.32533658e-01
-1.44191945e+00 5.00664234e-01 -2.74766713e-01 -1.20947886e+00
1.04685831e+00 -8.16368222e-01 -4.72270846e-01 8.84749442e-02
-1.34351049e-02 1.49241197e+00 5.77662647e-01 -7.42168784e-01
-9.10662055e-01 4.68732119e-02 2.01135539e-02 4.29180652e-01
3.56978059e-01 -1.37131199e-01 -4.55280691e-01 -3.53759021e-01
-2.35330328e-01 -1.71583116e+00 -7.71403968e-01 6.42740652e-02
-4.26618487e-01 -4.04467821e-01 1.09275043e+00 -6.10638976e-01
8.92403305e-01 -1.82508135e+00 2.61446625e-01 -4.33639914e-01
-9.90764201e-02 2.75439829e-01 -2.39037573e-01 5.50253391e-01
2.36441478e-01 -2.43328556e-01 7.02937543e-02 -4.33286577e-01
1.73467413e-01 4.14449632e-01 -4.94879693e-01 4.70731974e-01
3.01844954e-01 1.07860911e+00 -1.11868227e+00 -4.91920203e-01
5.68168700e-01 4.46094185e-01 -6.51952863e-01 2.20291659e-01
-6.02432072e-01 7.51673996e-01 -5.70773780e-01 3.52922440e-01
5.28863166e-03 -1.65157884e-01 1.65824667e-02 2.33699024e-01
4.39081602e-02 1.68486089e-01 -7.50361145e-01 1.66901243e+00
-3.35057914e-01 9.65371490e-01 -5.60488939e-01 -6.53704703e-01
6.29624784e-01 3.21933419e-01 8.02885532e-01 -4.96679902e-01
-5.35593405e-02 -5.21986187e-02 1.78055197e-01 -6.55480504e-01
9.15826082e-01 -1.59779917e-02 -1.29626006e-01 2.94040889e-01
-4.71659184e-01 -5.19805960e-02 3.20998728e-01 1.96841136e-01
1.35968351e+00 5.48537254e-01 5.08372307e-01 2.52329707e-01
2.95688659e-01 4.29528654e-01 7.08717763e-01 6.86883926e-01
-7.61330962e-01 6.23991907e-01 3.50628376e-01 -7.58527994e-01
-1.14712155e+00 -7.78471649e-01 6.14011884e-01 9.15943921e-01
1.27014190e-01 -1.85017705e-01 -4.26778793e-01 -6.03779018e-01
-1.69093356e-01 1.36786342e+00 -5.58796942e-01 -4.11372989e-01
-1.13120091e+00 -2.67280877e-01 2.22136676e-01 1.04596305e+00
5.26708424e-01 -1.56350231e+00 -1.32229280e+00 4.07431930e-01
-4.63051349e-01 -1.32845867e+00 -4.03513700e-01 -3.40761602e-01
-8.80113780e-01 -7.97901571e-01 -8.13664675e-01 -4.35394228e-01
2.64344186e-01 1.16157662e-02 1.12246025e+00 -4.29389849e-02
-2.04604998e-01 4.58592325e-01 -5.03267169e-01 -2.92116493e-01
-7.42946208e-01 -2.62938470e-01 1.61349699e-01 -3.18479389e-01
2.37149939e-01 -2.17527136e-01 -9.58234310e-01 1.44430384e-01
-3.78111720e-01 4.58111048e-01 5.68183959e-01 7.65732288e-01
5.43611050e-01 -1.11184463e-01 2.09626228e-01 -6.19536281e-01
3.57617974e-01 -6.18943095e-01 -1.13495000e-01 1.50679126e-01
-3.75283867e-01 5.70194237e-02 5.77110648e-01 -6.87275589e-01
-1.25206923e+00 8.46418440e-01 -7.60363862e-02 -5.41067421e-01
-2.64841050e-01 5.76102473e-02 1.50749460e-01 6.04319155e-01
7.76142657e-01 2.00671196e-01 -3.10764074e-01 -7.85653889e-02
5.47340631e-01 2.64847994e-01 8.95844936e-01 -3.16105843e-01
2.42819190e-01 5.95920086e-01 5.42134419e-02 -8.43505085e-01
-4.45195943e-01 -3.76156360e-01 -7.19438076e-01 -6.11670792e-01
1.21574700e+00 -8.63433123e-01 -8.60828221e-01 4.90623415e-01
-1.52035749e+00 -9.21049476e-01 -3.17425013e-01 7.60767102e-01
-1.26676321e+00 3.46454114e-01 -6.45715058e-01 -9.60157931e-01
-2.38274112e-01 -1.18616164e+00 1.30803657e+00 1.07006125e-01
-8.98774028e-01 -7.84124017e-01 1.62184313e-01 2.91264683e-01
-8.42164457e-02 5.32060146e-01 1.79693267e-01 -2.21228182e-01
-6.83747053e-01 -2.69468099e-01 3.14201653e-01 2.01135781e-02
-1.03537485e-01 1.29395813e-01 -4.09276277e-01 -1.26978546e-01
-4.88372445e-01 -3.97771388e-01 8.77807617e-01 5.99584699e-01
1.00484037e+00 -3.74696493e-01 -6.86347306e-01 1.31894678e-01
8.21290433e-01 5.48499703e-01 1.16202223e+00 1.47981077e-01
6.49875820e-01 5.94685018e-01 1.40473652e+00 6.76092148e-01
4.12952602e-01 7.53654897e-01 5.07026076e-01 3.63613069e-01
-1.51382938e-01 -5.06572485e-01 6.71873450e-01 1.70076847e-01
-2.44861215e-01 -7.86465585e-01 -1.09143329e+00 7.63061941e-01
-2.33907175e+00 -1.33879077e+00 -2.76251912e-01 1.78369975e+00
3.27999413e-01 3.75261068e-01 2.79126614e-01 -9.79410559e-02
4.64164823e-01 2.54037440e-01 -9.35527384e-01 -4.09573704e-01
2.37296596e-01 -3.51878583e-01 5.26021540e-01 5.07186830e-01
-1.28642261e+00 1.25665355e+00 6.27277899e+00 4.56832260e-01
-9.27096009e-01 -4.85355221e-02 6.43901110e-01 -1.18910968e-01
3.48073803e-02 1.01484738e-01 -8.57063890e-01 4.85376358e-01
1.01910758e+00 -1.92494065e-01 8.52269903e-02 1.01701224e+00
6.96841896e-01 -5.00715435e-01 -1.20655608e+00 8.82788599e-01
-4.12952155e-01 -1.31316507e+00 -2.35616937e-01 -6.47096783e-02
7.41740584e-01 1.43395752e-01 -1.45481065e-01 2.93547869e-01
6.76360071e-01 -1.03244305e+00 8.86145234e-01 9.67818260e-01
2.71369666e-01 -6.97717011e-01 4.02485490e-01 7.51560330e-01
-1.09673905e+00 -1.86089009e-01 -9.77844819e-02 -3.17588955e-01
8.78827453e-01 2.77501434e-01 -1.22716522e+00 7.07380800e-03
4.17171299e-01 8.56632411e-01 -2.46804059e-01 8.50292623e-01
-4.13420320e-01 5.77366114e-01 -1.71875760e-01 -2.27149710e-01
4.18308407e-01 3.06347460e-01 5.94889939e-01 1.07008326e+00
5.50139666e-01 4.26231682e-01 3.23703319e-01 2.70695478e-01
3.42767775e-01 -3.65189910e-01 -7.52304733e-01 -2.82650292e-01
1.92541137e-01 7.45830834e-01 -5.87736666e-01 -5.18786967e-01
2.65040365e-03 1.59254754e+00 9.07952413e-02 1.45672634e-01
-1.07712889e+00 6.59275725e-02 7.19718277e-01 1.47211671e-01
1.60848081e-01 -6.06833160e-01 7.22719654e-02 -8.19263160e-01
-9.13420320e-02 -5.92139959e-01 2.73337185e-01 -1.28213537e+00
-5.21198988e-01 5.79862893e-01 1.95512623e-01 -1.54957736e+00
-1.36164308e+00 -4.16880816e-01 -7.59809852e-01 4.29427505e-01
-7.54869640e-01 -1.16243005e+00 -1.17719509e-01 1.39400169e-01
1.00449181e+00 2.32765879e-02 4.75163788e-01 -1.61830977e-01
-2.38873586e-01 -2.42475271e-02 -2.43249342e-01 -8.83981064e-02
4.92593318e-01 -1.07147574e+00 1.05374253e+00 9.51636076e-01
8.78495947e-02 5.16032018e-02 1.26005507e+00 -1.01725233e+00
-1.19259405e+00 -1.20336854e+00 1.12765002e+00 -5.09684920e-01
4.95862216e-01 1.87822998e-01 -7.07927465e-01 1.06168044e+00
2.07315102e-01 2.37587392e-01 -1.69671327e-02 -4.30142224e-01
3.07123721e-01 2.63772815e-01 -9.61200714e-01 1.19441819e+00
1.36528194e+00 -7.43731186e-02 -4.29770529e-01 7.31248498e-01
7.41326153e-01 -5.72671473e-01 -5.36767542e-01 4.69232261e-01
6.27889931e-01 -9.23085749e-01 1.01914215e+00 -9.01791334e-01
7.59627283e-01 -1.24265514e-01 3.02327964e-02 -1.21280527e+00
-4.30058420e-01 -8.93414676e-01 -4.96834069e-01 5.00796974e-01
1.33123904e-01 -4.83474620e-02 1.24182081e+00 8.48636746e-01
-1.66000009e-01 -7.68555284e-01 -7.17410564e-01 -8.72402608e-01
-2.27545440e-01 -6.66307807e-01 3.56786847e-01 3.49667490e-01
-7.62740672e-02 -3.00220735e-02 -9.78314042e-01 -1.73173994e-01
4.57851499e-01 -1.91797569e-01 1.11751795e+00 -6.02775395e-01
-3.78919601e-01 -2.28927508e-01 -5.88310778e-01 -1.44682705e+00
4.67542589e-01 -3.54829341e-01 4.96286392e-01 -1.68843925e+00
-3.63558903e-03 3.93731669e-02 1.00805506e-01 5.49038887e-01
-1.27971902e-01 -1.09648399e-01 3.31930667e-01 1.73752606e-01
-8.98986816e-01 7.87323654e-01 1.46575677e+00 -6.45746812e-02
-2.99978346e-01 1.67238802e-01 2.84900470e-03 9.47592199e-01
8.78982425e-01 -3.39115113e-01 -6.04452133e-01 -3.19460034e-01
-1.62063986e-01 5.48432350e-01 6.55602932e-01 -1.37941718e+00
4.26920205e-01 -6.86823726e-01 2.94584990e-01 -8.56244445e-01
7.73665249e-01 -4.76508737e-01 3.97093087e-01 1.13350606e+00
-4.60780025e-01 1.58685029e-01 3.62298489e-02 8.99365604e-01
1.59154702e-02 7.81545341e-02 7.22376466e-01 -1.13662735e-01
-1.69976985e+00 2.67701805e-01 -6.51893973e-01 -4.69404571e-02
1.36601031e+00 -1.11028217e-01 -1.61920950e-01 -7.74044633e-01
-9.23543274e-01 4.63203073e-01 3.44078958e-01 7.98417628e-01
7.87579179e-01 -1.37147164e+00 -7.95381129e-01 -4.66465533e-01
-3.89776118e-02 -3.04613143e-01 5.01845121e-01 8.42229009e-01
-7.71474719e-01 5.16576231e-01 -4.68266696e-01 -4.64519262e-01
-1.33187830e+00 4.62773442e-01 1.01006307e-01 -4.15737331e-01
-9.62335110e-01 7.99713016e-01 1.11199565e-01 4.61020246e-02
6.60666497e-03 -9.16467905e-02 -6.34131655e-02 -5.67805469e-01
4.64443088e-01 7.60015607e-01 -5.75960815e-01 -9.41917360e-01
-5.33673227e-01 -5.76347895e-02 7.19151692e-03 -4.26076263e-01
1.08784556e+00 -9.90440324e-02 4.78040695e-01 3.11678767e-01
9.37900364e-01 -5.10141373e-01 -2.00053406e+00 1.47933915e-01
1.73782185e-01 -3.48874420e-01 -3.11250120e-01 -6.89058244e-01
-6.86797798e-01 8.81848812e-01 3.77054125e-01 -3.55926931e-01
9.82620060e-01 -4.58474196e-02 1.15402794e+00 6.74483716e-01
6.07470989e-01 -1.31756556e+00 3.54930669e-01 5.00761330e-01
1.17814362e+00 -1.32347453e+00 1.46629468e-01 -9.64771137e-02
-1.24844027e+00 8.53286147e-01 7.55207002e-01 -1.89903021e-01
4.39355761e-01 5.79760820e-02 -2.29236290e-01 1.32828414e-01
-1.22388768e+00 -1.36293843e-01 4.77295071e-01 5.09145856e-01
3.15827966e-01 3.49140704e-01 -3.83945793e-01 1.81234598e-01
-9.26350728e-02 3.52605045e-01 5.45596302e-01 1.01266026e+00
-5.52131474e-01 -9.43277359e-01 -7.53468722e-02 2.92080790e-01
-2.25739479e-01 3.98311168e-01 -3.64483327e-01 7.95131147e-01
-5.66729009e-02 9.18141186e-01 1.84228897e-01 -5.80997288e-01
1.59684226e-01 4.77032550e-02 5.01414657e-01 -4.12554473e-01
6.84792828e-03 -1.15010552e-01 4.69217628e-01 -9.90108073e-01
-4.52566236e-01 -1.11743438e+00 -1.69865561e+00 -3.71325344e-01
4.24282759e-01 -4.45328027e-01 2.63427228e-01 9.62313712e-01
5.60356617e-01 3.49227309e-01 3.07524920e-01 -1.21842480e+00
-4.26948547e-01 -8.02401066e-01 -6.48034438e-02 5.10918796e-01
6.02731965e-02 -6.32083476e-01 -2.05072090e-02 4.39290375e-01] | [7.827841758728027, 0.40753284096717834] |
42263546-08d4-4518-b8dc-23f2f24a36c2 | tie-a-framework-for-embedding-based | 2104.08419 | null | https://arxiv.org/abs/2104.08419v3 | https://arxiv.org/pdf/2104.08419v3.pdf | TIE: A Framework for Embedding-based Incremental Temporal Knowledge Graph Completion | Reasoning in a temporal knowledge graph (TKG) is a critical task for information retrieval and semantic search. It is particularly challenging when the TKG is updated frequently. The model has to adapt to changes in the TKG for efficient training and inference while preserving its performance on historical knowledge. Recent work approaches TKG completion (TKGC) by augmenting the encoder-decoder framework with a time-aware encoding function. However, naively fine-tuning the model at every time step using these methods does not address the problems of 1) catastrophic forgetting, 2) the model's inability to identify the change of facts (e.g., the change of the political affiliation and end of a marriage), and 3) the lack of training efficiency. To address these challenges, we present the Time-aware Incremental Embedding (TIE) framework, which combines TKG representation learning, experience replay, and temporal regularization. We introduce a set of metrics that characterizes the intransigence of the model and propose a constraint that associates the deleted facts with negative labels. Experimental results on Wikidata12k and YAGO11k datasets demonstrate that the proposed TIE framework reduces training time by about ten times and improves on the proposed metrics compared to vanilla full-batch training. It comes without a significant loss in performance for any traditional measures. Extensive ablation studies reveal performance trade-offs among different evaluation metrics, which is essential for decision-making around real-world TKG applications. | ['Jackie Chi Kit Cheung', 'Mark Coates', 'Chen Ma', 'Yingxue Zhang', 'Yishi Xu', 'Jiapeng Wu'] | 2021-04-17 | null | null | null | null | ['temporal-knowledge-graph-completion'] | ['knowledge-base'] | [-7.36985728e-02 2.53938109e-01 -4.68897581e-01 -2.00115010e-01
-4.83215332e-01 -3.92435044e-01 5.92828870e-01 3.26858610e-01
-4.33100045e-01 9.36365008e-01 4.27587122e-01 -3.07964206e-01
-4.76957142e-01 -7.79217243e-01 -9.47361648e-01 -4.58526105e-01
-3.09587598e-01 4.93927985e-01 2.12115422e-01 -1.24280415e-01
1.12168625e-01 5.68890721e-02 -1.40008831e+00 1.88773423e-01
1.01193309e+00 1.03053117e+00 6.65139556e-02 3.25038016e-01
-1.71371147e-01 1.08032799e+00 -2.39312261e-01 -8.13115835e-01
1.21511877e-01 -2.35299513e-01 -1.02416015e+00 -3.55795085e-01
3.66232872e-01 -4.67520058e-01 -7.06699133e-01 9.48979735e-01
2.16309309e-01 3.27840745e-01 4.24726456e-01 -1.06877601e+00
-1.01934552e+00 8.75180781e-01 -4.14140314e-01 4.57717270e-01
1.30081639e-01 -2.18718380e-01 1.02788842e+00 -8.53378117e-01
7.96069205e-01 1.13016999e+00 8.73242438e-01 3.43542784e-01
-9.30299699e-01 -4.75120813e-01 5.33537865e-01 7.44263113e-01
-1.46617472e+00 -3.61615092e-01 6.75406694e-01 -2.93117672e-01
1.11943138e+00 3.96569557e-02 5.35016418e-01 1.13704777e+00
2.74715275e-01 6.65670872e-01 6.82468653e-01 -4.23257947e-01
3.17370623e-01 -3.42732519e-02 4.06944543e-01 9.99791384e-01
4.53071266e-01 -6.43343627e-02 -9.64907825e-01 -1.38423026e-01
4.25059527e-01 -5.20694517e-02 -2.80799180e-01 -3.58239233e-01
-9.24513400e-01 6.51588440e-01 3.99200678e-01 2.44711429e-01
-3.69233757e-01 5.13827026e-01 3.74058694e-01 6.46991909e-01
4.15672898e-01 1.92774907e-01 -7.71608055e-01 -2.77164280e-01
-7.52735674e-01 1.17451876e-01 7.26633728e-01 1.02463627e+00
6.87240660e-01 -2.89466307e-02 -2.42962718e-01 6.80015624e-01
6.32775109e-03 1.83155507e-01 6.74338579e-01 -7.76668906e-01
6.30141318e-01 7.17173398e-01 1.58008114e-01 -1.15275979e+00
-3.11861008e-01 -5.67821920e-01 -5.51338017e-01 -4.40174788e-01
3.87381315e-01 8.92769545e-02 -1.02343857e+00 2.15332818e+00
3.92176777e-01 6.01887465e-01 1.68942496e-01 5.20625234e-01
5.80760539e-01 4.95710284e-01 7.58873075e-02 -1.90932974e-01
1.20032084e+00 -9.38468277e-01 -9.07614410e-01 -3.78262192e-01
7.81557739e-01 -3.39788496e-01 9.44636643e-01 2.00612321e-02
-8.99399996e-01 -1.51643932e-01 -1.08612549e+00 -2.93203950e-01
-6.37961686e-01 -8.79164487e-02 1.05745411e+00 4.56016809e-01
-9.06527102e-01 7.82212079e-01 -9.98995483e-01 -6.26802087e-01
2.67519951e-01 3.38535815e-01 -2.64899105e-01 -3.45470667e-01
-1.58483744e+00 1.00610101e+00 6.35386467e-01 2.25161090e-01
-6.91112399e-01 -9.39934134e-01 -8.04017007e-01 2.61034310e-01
7.15911329e-01 -7.66294777e-01 1.03472090e+00 -5.85799456e-01
-1.40168643e+00 5.16552448e-01 -1.49294823e-01 -6.25769496e-01
6.29421175e-01 -4.28556412e-01 -7.19338179e-01 -1.44771069e-01
1.07435912e-01 1.44404963e-01 8.49424422e-01 -9.45158899e-01
-5.03785193e-01 -4.68150914e-01 1.12094410e-01 3.15336734e-01
-5.23569345e-01 -6.33663952e-01 -8.73266339e-01 -6.50399029e-01
2.18389511e-01 -8.74868751e-01 1.62405834e-01 -1.93912864e-01
-1.99959725e-01 -1.90442041e-01 6.96217716e-01 -9.50731337e-01
1.51837873e+00 -2.16516280e+00 1.65114522e-01 1.45819828e-01
-7.83330016e-03 2.52707601e-01 -6.52756691e-02 5.42969704e-01
2.06947818e-01 -6.70999736e-02 6.14508949e-02 -3.18301052e-01
-1.24021627e-01 4.32067871e-01 -4.64179397e-01 2.74449140e-01
-1.89202368e-01 1.01090240e+00 -1.15135145e+00 -3.30704004e-01
-5.10722585e-02 3.46021563e-01 -7.50378549e-01 -2.55345643e-01
-2.60088444e-01 8.99377912e-02 -4.55490559e-01 6.69604719e-01
3.51959080e-01 -4.64971632e-01 5.71439266e-01 -3.79592389e-01
2.68199027e-01 3.56456757e-01 -1.09206450e+00 1.97919011e+00
-4.69314843e-01 1.52010545e-01 -5.54327011e-01 -8.06829929e-01
5.35345674e-01 2.45411932e-01 7.51020312e-02 -9.60442603e-01
-9.95546877e-02 1.80981770e-01 -3.18932891e-01 -4.34845954e-01
7.13160515e-01 1.36554912e-01 -1.31378278e-01 4.67233777e-01
3.17939430e-01 5.98590791e-01 1.40816486e-02 5.26564717e-01
1.34464777e+00 2.90542841e-02 1.95002854e-01 -4.94398810e-02
1.65137559e-01 -2.21577063e-01 9.46756542e-01 8.95576477e-01
-1.22957312e-01 5.50183393e-02 5.13146162e-01 -5.65248728e-01
-6.97742462e-01 -1.01979315e+00 2.37354472e-01 1.00654912e+00
2.47344121e-01 -5.97408712e-01 -2.41351992e-01 -8.30391824e-01
4.16094691e-01 1.04033685e+00 -9.65463400e-01 -7.96433985e-01
-4.59930778e-01 -6.41020715e-01 5.58057487e-01 4.91228491e-01
6.04491472e-01 -7.54406035e-01 -3.89610887e-01 3.64738733e-01
-4.12907064e-01 -1.05572617e+00 -4.20084566e-01 1.41947642e-01
-9.34488237e-01 -9.72372591e-01 -2.49229386e-01 -6.79304659e-01
7.01819360e-01 2.14536577e-01 8.18714440e-01 1.36568636e-01
-9.14855860e-03 5.82179546e-01 -4.04062182e-01 -3.06635518e-02
-6.81644604e-02 1.83990985e-01 2.83426046e-01 3.62108722e-02
2.93206066e-01 -7.80203104e-01 -6.58443689e-01 -4.70507480e-02
-8.01124573e-01 -5.31782843e-02 4.85543400e-01 1.03430986e+00
4.03148115e-01 4.47171718e-01 8.30704212e-01 -1.27996397e+00
5.29635727e-01 -6.38481438e-01 -4.21664476e-01 7.36424029e-01
-1.16246748e+00 4.00702089e-01 4.89348948e-01 -4.34084326e-01
-1.40855515e+00 -3.26132208e-01 4.07422751e-01 -5.56528032e-01
7.87995279e-01 9.39417481e-01 1.59860283e-01 8.70213434e-02
4.57177907e-01 2.31137529e-01 -2.95018107e-01 -5.28812349e-01
7.37926543e-01 1.57202870e-01 7.15990365e-01 -6.81774616e-01
7.03699291e-01 5.02262294e-01 -3.80645484e-01 -4.79092240e-01
-1.06354797e+00 -2.78423876e-01 -4.17989075e-01 -3.61677557e-02
2.66513109e-01 -1.05264318e+00 -6.69628501e-01 3.04790080e-01
-8.69282365e-01 -3.07694614e-01 -2.85973907e-01 4.51406002e-01
-3.62860441e-01 4.36270863e-01 -6.80207849e-01 -5.56364655e-01
-3.37791562e-01 -4.35500681e-01 4.79487181e-01 1.54867217e-01
-1.13672033e-01 -1.23490632e+00 5.91566600e-02 4.04437989e-01
4.60517049e-01 -2.05660630e-02 1.37654865e+00 -3.33777398e-01
-7.08107114e-01 -2.25428909e-01 -5.09370267e-02 1.21696688e-01
2.18800172e-01 -4.03927386e-01 -9.36764896e-01 -5.31162202e-01
-2.10211352e-01 -3.68944556e-01 1.16453421e+00 6.42591715e-02
1.04119372e+00 -5.30645669e-01 -4.74510521e-01 6.19862854e-01
1.62447095e+00 3.74434263e-01 5.61188579e-01 3.51404309e-01
6.17642224e-01 1.90829560e-01 5.87180376e-01 5.29345155e-01
7.35459208e-01 6.04288936e-01 1.54012561e-01 5.19705296e-01
-2.44671211e-01 -8.67915034e-01 3.84972125e-01 1.07004106e+00
-2.00163890e-02 -2.01292560e-01 -7.99232662e-01 8.97530437e-01
-2.21265721e+00 -8.66528153e-01 4.45728302e-01 2.30736589e+00
9.90523398e-01 1.99489087e-01 -5.85111916e-01 -8.03546980e-03
3.73151124e-01 2.02905923e-01 -7.96742737e-01 -2.41280705e-01
-2.13251077e-03 6.79240152e-02 6.35078430e-01 7.21733332e-01
-7.80249834e-01 1.34529269e+00 6.13819170e+00 6.45436645e-01
-9.28655624e-01 4.17065918e-01 2.41138220e-01 -1.86876819e-01
-3.95981520e-01 3.37696791e-01 -7.81738520e-01 4.63758081e-01
7.74867237e-01 -5.04495382e-01 7.13028848e-01 6.45890832e-01
-2.14887917e-01 -8.22025910e-02 -1.12528658e+00 9.71826077e-01
2.30005071e-01 -1.32804680e+00 1.89659312e-01 -3.02131206e-01
7.82492995e-01 -1.19622275e-01 1.08601354e-01 7.40684032e-01
4.37078059e-01 -6.24667585e-01 7.47001529e-01 7.01262116e-01
6.71001554e-01 -5.97368658e-01 5.56537390e-01 9.76040587e-02
-1.05731559e+00 -3.01776081e-01 -3.31997275e-01 -4.41345647e-02
2.42995061e-02 6.57855809e-01 -9.11894500e-01 8.40327263e-01
7.75764704e-01 7.56852806e-01 -7.44096756e-01 6.70964420e-01
-3.28262836e-01 6.32933140e-01 -2.88429499e-01 2.50852257e-01
7.91411921e-02 1.90337915e-02 3.57393026e-01 8.70460927e-01
3.35206419e-01 8.43132883e-02 -2.33831197e-01 5.57155013e-01
-3.89741987e-01 -1.15306638e-01 -5.87732315e-01 -2.62991726e-01
7.89940476e-01 7.34124124e-01 -4.87410724e-01 -2.86747992e-01
-4.21187699e-01 1.28229797e+00 8.64248633e-01 6.74695790e-01
-9.05796647e-01 -2.32923403e-01 6.14486516e-01 -9.20539275e-02
5.83100259e-01 -1.13926813e-01 -7.85449296e-02 -1.37742341e+00
3.20887983e-01 -4.47535068e-01 8.19511414e-01 -4.88562018e-01
-1.12055933e+00 2.52852082e-01 -6.23239763e-02 -7.13737369e-01
-1.83739305e-01 -5.64822182e-02 -2.32834190e-01 4.28481430e-01
-1.62645102e+00 -1.10320234e+00 -1.23500183e-01 6.15074337e-01
1.42456263e-01 -6.12051338e-02 7.02465296e-01 6.33787513e-01
-6.53341830e-01 9.34575438e-01 3.10093910e-01 -1.30259767e-01
5.46599925e-01 -1.04059148e+00 3.05497378e-01 1.00194621e+00
9.99151394e-02 7.88305700e-01 5.43757737e-01 -8.55154335e-01
-1.58037508e+00 -1.27764094e+00 1.30286789e+00 -4.46110755e-01
6.21591508e-01 -2.80534267e-01 -9.56719756e-01 1.36274827e+00
-2.05958590e-01 -1.10304564e-01 3.64120364e-01 6.37635529e-01
-6.50406778e-01 -3.83951664e-01 -8.69233727e-01 7.36884594e-01
1.26987433e+00 -6.31881475e-01 -7.62644172e-01 3.81784827e-01
8.03946853e-01 -3.40235412e-01 -8.15003753e-01 3.54584157e-01
5.12366831e-01 -5.69484115e-01 9.65737820e-01 -8.33748817e-01
1.37019530e-01 -2.53739119e-01 -8.01881030e-02 -1.20096469e+00
-5.57152808e-01 -6.79189205e-01 -6.21798396e-01 1.30663085e+00
2.94692278e-01 -6.93863273e-01 7.31004179e-01 7.24688828e-01
2.47853473e-02 -6.26992881e-01 -1.14920235e+00 -9.30510759e-01
-2.26608843e-01 -3.14047545e-01 5.64466119e-01 1.32126844e+00
1.38590485e-01 4.87221807e-01 -8.29999685e-01 3.50124329e-01
4.66537833e-01 1.12404816e-01 4.16574359e-01 -1.02496457e+00
-1.47757754e-01 -4.56787273e-02 -4.12076056e-01 -8.34078610e-01
7.19152838e-02 -1.04935598e+00 -2.45667174e-01 -1.63339162e+00
2.27495536e-01 -5.25795400e-01 -6.44515455e-01 8.27598631e-01
-2.85637766e-01 -3.15925330e-01 -1.44324124e-01 1.87936738e-01
-8.00602734e-01 8.82074475e-01 9.24965560e-01 -9.63042825e-02
-2.35843211e-01 -4.74982530e-01 -7.33761191e-01 5.22423148e-01
5.41191697e-01 -6.30017579e-01 -9.09817815e-01 -6.71803832e-01
8.14431489e-01 3.73792797e-02 1.96446449e-01 -8.91334534e-01
6.53843582e-01 -2.95210164e-02 1.53566882e-01 -3.47459942e-01
4.44073021e-01 -7.63389289e-01 3.86370242e-01 4.79919940e-01
-2.60305166e-01 1.17295869e-02 1.73947245e-01 1.17508328e+00
-1.36936501e-01 1.60993706e-03 4.03874546e-01 2.86635719e-02
-1.11445856e+00 2.83028454e-01 -8.13638344e-02 1.51025400e-01
9.20735121e-01 1.39240865e-02 -4.99637246e-01 -4.50204194e-01
-7.47899055e-01 5.22745252e-01 1.84864894e-01 6.44212842e-01
5.52781463e-01 -1.44922721e+00 -2.94056654e-01 1.05776349e-02
3.50854516e-01 -1.88816875e-01 5.16353369e-01 7.78800428e-01
-3.70121717e-01 4.29142267e-01 1.05481606e-03 1.42620481e-03
-8.45989227e-01 5.97239673e-01 2.43886814e-01 -5.72716236e-01
-8.38865340e-01 8.38311434e-01 -6.66589737e-02 -3.55486751e-01
4.33002472e-01 -4.56816405e-01 -8.36442709e-02 1.06483839e-01
2.99947351e-01 3.39121610e-01 3.35692197e-01 -6.28406405e-02
-3.60626996e-01 1.90043241e-01 -6.45360827e-01 1.19804665e-01
1.29772580e+00 -3.84523779e-01 1.37599170e-01 6.37276411e-01
9.43687022e-01 -2.65632898e-01 -9.92860019e-01 -8.03167701e-01
2.17135176e-01 -4.68693972e-01 1.93028837e-01 -9.93415356e-01
-9.78347421e-01 3.55940521e-01 4.94404674e-01 -2.61451066e-01
9.23367620e-01 -1.16067357e-01 9.35560167e-01 7.67757654e-01
6.25774145e-01 -1.34780908e+00 1.44187942e-01 6.68264925e-01
6.67820752e-01 -1.00199556e+00 1.16850682e-01 -3.03779513e-01
-6.55579627e-01 7.56960988e-01 5.09880781e-01 3.35575640e-01
8.77485693e-01 -3.24947417e-01 -1.20985612e-01 -4.57625180e-01
-1.20831120e+00 1.65974498e-02 7.06370687e-03 1.31058410e-01
-5.65047711e-02 2.30891500e-02 -4.34766680e-01 7.01255977e-01
-2.04696432e-01 3.23946953e-01 3.19408029e-01 1.10844851e+00
-8.34115446e-02 -8.42350781e-01 2.33265907e-01 6.05542362e-01
-2.93336987e-01 -1.64739177e-01 -9.64816585e-02 7.73904324e-01
2.75931973e-02 6.27423108e-01 -1.31674781e-01 -4.74349469e-01
2.88977236e-01 4.96108741e-01 5.19164622e-01 -3.34340930e-01
-2.59797573e-01 -5.12195647e-01 1.84306785e-01 -6.14703178e-01
-1.43097505e-01 -6.07678592e-01 -1.13882482e+00 -5.12966752e-01
-2.54322827e-01 1.31719425e-01 2.60983318e-01 9.31743026e-01
6.60966754e-01 5.37667751e-01 2.19690278e-01 9.73559842e-02
-5.71896195e-01 -7.75355160e-01 -8.14124286e-01 6.00124180e-01
1.25069454e-01 -1.03373861e+00 -3.27229381e-01 8.91542714e-03] | [8.613642692565918, 7.867116451263428] |
c588ca27-817c-4984-8808-bd33847c29ef | joint-layout-analysis-character-detection-and | 2007.06890 | null | https://arxiv.org/abs/2007.06890v1 | https://arxiv.org/pdf/2007.06890v1.pdf | Joint Layout Analysis, Character Detection and Recognition for Historical Document Digitization | In this paper, we propose an end-to-end trainable framework for restoring historical documents content that follows the correct reading order. In this framework, two branches named character branch and layout branch are added behind the feature extraction network. The character branch localizes individual characters in a document image and recognizes them simultaneously. Then we adopt a post-processing method to group them into text lines. The layout branch based on fully convolutional network outputs a binary mask. We then use Hough transform for line detection on the binary mask and combine character results with the layout information to restore document content. These two branches can be trained in parallel and are easy to train. Furthermore, we propose a re-score mechanism to minimize recognition error. Experiment results on the extended Chinese historical document MTHv2 dataset demonstrate the effectiveness of the proposed framework. | ['Weihong Ma', 'Lianwen Jin', 'Hesuo Zhang', 'Sihang Wu', 'Yongpan Wang', 'Jiapeng Wang'] | 2020-07-14 | null | null | null | null | ['line-detection'] | ['computer-vision'] | [ 5.02755821e-01 -3.43357652e-01 -1.00809038e-01 -4.18801457e-01
-5.69153547e-01 -7.59411693e-01 4.41608131e-01 -8.72537047e-02
-3.22106928e-01 2.55066723e-01 1.42945245e-01 -4.59714323e-01
8.01915973e-02 -9.33472514e-01 -8.32853198e-01 -5.04452169e-01
5.31556904e-01 4.88059223e-02 4.32559490e-01 3.81958857e-02
6.34647548e-01 7.10970461e-01 -9.35316086e-01 7.54353583e-01
1.00182617e+00 8.65125179e-01 5.51244318e-01 9.56382096e-01
-4.67590243e-01 9.13234890e-01 -8.00938308e-01 -3.32677931e-01
2.54923433e-01 -4.18787211e-01 -6.40141547e-01 6.29767120e-01
5.64496636e-01 -8.18246424e-01 -6.35644794e-01 1.06278157e+00
4.41510081e-01 2.62160581e-02 5.89257717e-01 -6.04853809e-01
-9.67342913e-01 8.08296978e-01 -7.52126813e-01 3.72667536e-02
1.71826020e-01 -3.19912761e-01 8.10851872e-01 -1.08270359e+00
5.77937603e-01 8.79114270e-01 5.21546483e-01 3.46865386e-01
-5.89500904e-01 -2.93490976e-01 1.85092688e-01 1.82258531e-01
-1.17507064e+00 -2.20910460e-01 7.60308087e-01 -2.80057132e-01
7.36714184e-01 3.32353383e-01 2.34784573e-01 4.96120065e-01
2.87819505e-01 1.18706274e+00 5.98188579e-01 -6.91341937e-01
-1.40840560e-01 -1.24490760e-01 4.73860681e-01 1.02443719e+00
5.48177250e-02 -4.22662288e-01 -1.75268576e-01 3.73579562e-01
8.95840049e-01 1.56756103e-01 -6.48893595e-01 -2.18009055e-01
-1.02484381e+00 5.51917255e-01 5.54600596e-01 2.95775533e-01
-1.11034602e-01 1.96570437e-02 3.83886367e-01 7.67790526e-02
5.39669096e-02 5.82271218e-02 -2.48769596e-01 2.75463492e-01
-1.20030475e+00 -3.81662510e-02 4.92987156e-01 1.03889680e+00
5.44323504e-01 -1.52943075e-01 -4.79679257e-01 1.13891351e+00
2.17866793e-01 4.51554716e-01 5.83253622e-01 -3.96687388e-01
9.63745832e-01 6.14551246e-01 -4.17082720e-02 -1.10047448e+00
-3.01175475e-01 -5.54075778e-01 -8.96117687e-01 7.80790076e-02
2.92266123e-02 -5.32115102e-02 -1.25581920e+00 8.61904681e-01
-9.88073945e-02 -7.86567703e-02 -3.10529210e-02 8.29339325e-01
6.75491214e-01 1.02811599e+00 -6.91662014e-01 8.61861035e-02
1.24651074e+00 -1.59539592e+00 -7.79971719e-01 -1.43358901e-01
6.35709822e-01 -9.81617153e-01 1.10368073e+00 5.45599461e-01
-1.02624333e+00 -7.33298957e-01 -1.53959322e+00 -4.15172726e-01
-1.86324969e-01 9.86522734e-01 2.01559097e-01 5.69722891e-01
-7.81008244e-01 5.64476252e-01 -8.32619011e-01 -1.39635623e-01
3.40396434e-01 1.02775097e-01 -2.86261886e-01 -2.32524708e-01
-5.39891839e-01 5.30648530e-01 6.46494746e-01 5.80898643e-01
-6.04577780e-01 -4.74697277e-02 -7.56039143e-01 4.59026128e-01
1.81703299e-01 -2.13776544e-01 1.27850318e+00 -9.25588846e-01
-1.60130715e+00 5.33993006e-01 -1.31616369e-01 -2.62729228e-01
6.82594717e-01 -5.05586565e-01 -6.20302737e-01 4.07887250e-01
-4.90176817e-03 3.73959273e-01 8.52099121e-01 -1.26034963e+00
-1.07681251e+00 -5.25915883e-02 -4.20495719e-01 1.88625008e-01
-5.97762644e-01 -1.95808802e-02 -1.12563598e+00 -8.47291112e-01
4.85929668e-01 -5.12937784e-01 7.56284446e-02 -3.19661014e-02
-8.35561812e-01 9.47228968e-02 1.20923853e+00 -1.18183160e+00
1.33352637e+00 -2.06480551e+00 -1.13121077e-01 5.24739504e-01
4.70878556e-02 3.24566424e-01 -3.40325922e-01 2.96646327e-01
1.43873453e-01 7.23847523e-02 -3.11713547e-01 -2.13595182e-01
-2.55253136e-01 -2.03922048e-01 -5.75418890e-01 3.23475897e-01
2.10304484e-01 9.03042197e-01 -3.94569635e-01 -4.48240906e-01
1.82890177e-01 1.75804704e-01 -1.54952675e-01 1.40432015e-01
-1.60330966e-01 -1.02922499e-01 -3.14356059e-01 5.88513792e-01
1.19287932e+00 -7.23677278e-02 1.57673687e-01 -3.21042955e-01
-3.27831835e-01 1.13507845e-01 -1.01239455e+00 1.57862055e+00
-1.12669006e-01 8.57749462e-01 -2.16888607e-01 -8.82718503e-01
1.30139840e+00 -1.56195298e-01 -9.53490287e-02 -8.76792669e-01
2.40177318e-01 1.58332616e-01 -1.61770165e-01 -4.45779324e-01
9.88906384e-01 5.85723221e-01 3.10017560e-02 3.29699159e-01
-1.55781016e-01 1.76458031e-01 3.03948224e-01 1.83210939e-01
9.30577636e-01 3.29439968e-01 -1.15730911e-01 3.93268391e-02
8.56395066e-01 4.46710736e-03 4.90598470e-01 7.14785635e-01
2.13656858e-01 1.00781727e+00 4.32681590e-01 -2.98157126e-01
-1.16366231e+00 -9.03552234e-01 1.00874521e-01 8.09270442e-01
2.43907318e-01 -3.32281590e-01 -9.54166710e-01 -8.53354096e-01
-3.17389220e-01 5.61622500e-01 -3.91914278e-01 2.18420532e-02
-1.00786865e+00 -3.77502799e-01 6.80032134e-01 9.01566923e-01
9.31594312e-01 -1.00004053e+00 -4.50349271e-01 1.68882608e-01
-9.94978100e-02 -1.16410089e+00 -7.77211010e-01 1.71756163e-01
-6.63529158e-01 -9.11031961e-01 -1.03557813e+00 -1.51981390e+00
1.05513763e+00 5.65201998e-01 4.08448398e-01 4.40404683e-01
-1.52888879e-01 -1.97858326e-02 -5.92637360e-01 7.53724277e-02
-3.82140368e-01 1.64071873e-01 -6.05653763e-01 1.28871188e-01
-1.53035857e-02 1.15255035e-01 -5.27378500e-01 1.89791739e-01
-9.46460724e-01 3.07367414e-01 8.44733894e-01 8.61457527e-01
6.26568735e-01 3.08878332e-01 -8.82824138e-03 -7.99496412e-01
7.13771820e-01 1.58214539e-01 -7.15014040e-01 7.30971277e-01
-3.49518895e-01 -6.97942972e-02 8.61090720e-01 -2.63894498e-01
-1.22713256e+00 3.08337837e-01 -5.68162836e-02 -2.06905276e-01
2.26812176e-02 7.13593245e-01 -3.61099184e-01 6.67553097e-02
2.83505589e-01 6.70503259e-01 -5.24250805e-01 -6.76082015e-01
3.87032986e-01 1.12203479e+00 1.29995072e+00 -3.25590253e-01
1.01626730e+00 3.11426163e-01 -4.39924777e-01 -8.72159660e-01
-5.42070925e-01 -2.81459987e-01 -9.04661715e-01 -4.98230085e-02
8.60603571e-01 -7.07418323e-01 -1.83217004e-01 8.69463682e-01
-1.29616117e+00 -2.08146304e-01 2.03844860e-01 3.04845810e-01
-1.78483352e-01 9.05318201e-01 -9.90563929e-01 -5.82861245e-01
-7.14495242e-01 -9.20246720e-01 1.01051068e+00 6.90470994e-01
4.27739888e-01 -6.05232179e-01 -2.13989988e-01 1.56327024e-01
-1.80575449e-03 -4.48939390e-02 1.09409297e+00 -5.12676954e-01
-5.62657952e-01 -5.32685399e-01 -5.96585870e-01 5.61817050e-01
1.83473095e-01 4.64099914e-01 -7.00382531e-01 -2.39859939e-01
-3.29030395e-01 6.34303922e-03 1.21582484e+00 1.63506225e-01
1.36311460e+00 -2.96675805e-02 -3.95866930e-01 7.59148121e-01
1.46545482e+00 4.81573999e-01 1.09022260e+00 6.47216976e-01
1.00927436e+00 2.08100513e-01 4.73827809e-01 3.40178996e-01
3.25019896e-01 3.28388542e-01 1.05791949e-01 -3.82754385e-01
-2.44202957e-01 -5.03530741e-01 4.40829337e-01 9.81261671e-01
2.81605899e-01 -7.39269793e-01 -8.12153161e-01 2.34241605e-01
-1.76455998e+00 -8.81772339e-01 -4.82680529e-01 2.01069260e+00
6.37753308e-01 4.29427236e-01 -3.78699064e-01 3.17967206e-01
9.89412427e-01 1.12860203e-01 -2.72837579e-01 -5.34204245e-01
-5.25506556e-01 2.07710508e-02 6.65212929e-01 4.19030249e-01
-1.21118343e+00 1.22991371e+00 5.73990297e+00 8.71154785e-01
-1.24265325e+00 -4.63766396e-01 6.49279594e-01 4.30357277e-01
-2.76012421e-01 1.77007169e-01 -9.70360875e-01 3.42172533e-01
3.01569343e-01 1.55723438e-01 2.18882650e-01 5.98934174e-01
6.83287671e-03 -1.13960050e-01 -8.51413548e-01 7.93594658e-01
3.83580178e-01 -1.44151974e+00 2.05812827e-01 -1.93659589e-01
7.68861890e-01 -2.91464090e-01 -1.12345725e-01 1.16969123e-01
-5.08382432e-02 -8.07403088e-01 1.01464272e+00 5.92438638e-01
6.58243120e-01 -8.91686141e-01 5.39015114e-01 3.78604919e-01
-1.31051314e+00 -2.09775776e-01 -7.15444148e-01 3.06726962e-01
2.50785768e-01 4.85792875e-01 -7.56264567e-01 7.82287598e-01
4.19518441e-01 5.99301219e-01 -8.60903323e-01 1.31905174e+00
-5.08845508e-01 4.58431959e-01 1.12841688e-01 7.63943940e-02
3.75807285e-01 -3.41402113e-01 3.00743654e-02 1.44219947e+00
5.31049907e-01 -7.97694176e-02 6.05222620e-02 6.28816962e-01
-2.71731585e-01 1.97772563e-01 2.26081014e-02 -1.44810408e-01
3.78674388e-01 1.47015846e+00 -1.17609644e+00 -4.38253403e-01
-4.66036588e-01 1.77509809e+00 4.17018265e-01 3.32109928e-01
-8.10319006e-01 -1.10702646e+00 -3.75906914e-01 -4.27828848e-01
8.03651929e-01 -2.69320190e-01 -4.16610539e-01 -1.27433622e+00
3.55526984e-01 -8.70496035e-01 1.08731307e-01 -1.01088822e+00
-7.88604915e-01 6.61818206e-01 -6.38066709e-01 -1.27828944e+00
3.05258155e-01 -8.62361908e-01 -9.01837170e-01 6.74306214e-01
-1.44722080e+00 -1.12950778e+00 -5.25065899e-01 4.37371999e-01
8.08710992e-01 -2.25374460e-01 5.54364681e-01 2.28623673e-01
-1.05085146e+00 7.91146815e-01 6.45067394e-01 6.84989274e-01
5.11842012e-01 -1.14540684e+00 7.57005870e-01 1.38461089e+00
2.68869370e-01 5.69582343e-01 1.12993345e-01 -8.93796802e-01
-1.07926452e+00 -1.00304174e+00 7.68863499e-01 1.71787128e-01
2.01982215e-01 -4.92468923e-01 -1.02834046e+00 6.06010854e-01
3.31090271e-01 -3.70810658e-01 4.26120639e-01 -4.81228769e-01
-4.94273126e-01 2.77813096e-02 -8.30595851e-01 5.43318331e-01
6.52654231e-01 -3.54832172e-01 -6.72515869e-01 3.84044200e-01
4.66748983e-01 -6.51894867e-01 -4.31755751e-01 2.56892204e-01
6.21251941e-01 -7.85323441e-01 5.56238234e-01 -2.44140193e-01
7.36158907e-01 -6.20871305e-01 -9.10310596e-02 -1.07368267e+00
-5.44073582e-01 -3.46606463e-01 1.13017261e-01 1.49985898e+00
5.50572753e-01 -1.36785895e-01 7.53727734e-01 9.08615515e-02
-3.96993726e-01 -5.03202319e-01 -2.46642992e-01 -6.88055813e-01
-4.73738350e-02 -1.88680470e-03 6.58880889e-01 7.50376761e-01
-1.61592588e-01 2.30794594e-01 -4.61342007e-01 3.61395061e-01
2.69251853e-01 3.06294799e-01 7.08233654e-01 -8.98864865e-01
-3.07462931e-01 -6.18207395e-01 -7.82599673e-02 -1.69094539e+00
-6.50002509e-02 -8.56515050e-01 3.35779786e-01 -1.95131183e+00
3.30086917e-01 -1.20902605e-01 -2.14240804e-01 4.70431298e-01
-3.62224102e-01 4.73706797e-02 2.81804085e-01 2.57779002e-01
-3.96246254e-01 4.80238944e-01 1.05949342e+00 -3.99289519e-01
-1.56499296e-01 -9.89792049e-02 -3.76073748e-01 5.43341219e-01
9.87221718e-01 -1.16542339e-01 -1.34308904e-01 -1.00243926e+00
-1.45199507e-01 -7.02835545e-02 7.23007098e-02 -1.07927811e+00
4.46815580e-01 1.99505836e-01 1.13454998e+00 -1.38174963e+00
-1.52627483e-01 -6.53061807e-01 -6.44028783e-01 5.02939939e-01
-3.92182827e-01 2.16947824e-01 1.76367313e-01 4.05196786e-01
-7.86553472e-02 -6.81402087e-01 5.42562902e-01 1.79883957e-01
-8.42858493e-01 1.32196750e-02 -3.87653798e-01 -6.13699675e-01
8.66394937e-01 -4.39412773e-01 -6.91176713e-01 -2.27847934e-01
-1.96968675e-01 3.22364867e-01 3.51087332e-01 5.06447136e-01
9.95346546e-01 -1.28037894e+00 -7.07399368e-01 3.39159489e-01
-4.44516502e-02 -1.96226105e-01 2.68327892e-01 4.15123671e-01
-1.34010673e+00 4.46421444e-01 -2.58223325e-01 -4.39651847e-01
-1.34770036e+00 4.27349240e-01 3.36526573e-01 -1.80708155e-01
-8.29793990e-01 8.97028208e-01 3.50826569e-02 -3.30489188e-01
4.50239033e-01 -3.76114696e-01 -3.65904629e-01 -2.09884182e-01
8.97845030e-01 3.80195081e-01 1.25123948e-01 -4.65891570e-01
-1.92495421e-01 8.69094014e-01 -6.21378422e-01 -1.97068408e-01
1.18997884e+00 -1.02053508e-01 -1.93651304e-01 -3.15969181e-03
1.19640529e+00 3.42988223e-01 -1.19819462e+00 -1.57384470e-01
2.94147104e-01 -3.69987398e-01 5.48301116e-02 -9.06640351e-01
-1.27361846e+00 8.98952723e-01 6.12211585e-01 5.82892187e-02
1.30101621e+00 -6.15855515e-01 9.59198356e-01 5.97834766e-01
-2.29719609e-01 -1.36771894e+00 4.66728173e-02 5.09510875e-01
7.55508125e-01 -6.95538521e-01 1.27448454e-01 -5.03141940e-01
-6.18318856e-01 1.77545297e+00 8.13630521e-01 -2.20234811e-01
-6.54167915e-03 3.52387846e-01 1.92939490e-01 1.70135722e-01
-3.22844267e-01 -6.49002120e-02 3.54772240e-01 4.31808025e-01
3.89993459e-01 -2.21015930e-01 -5.60003757e-01 5.75921357e-01
-2.31636912e-01 -6.16958141e-02 8.44227135e-01 1.15018058e+00
-8.78839910e-01 -1.09624636e+00 -4.90091205e-01 3.50236207e-01
-3.49677831e-01 -1.83550194e-01 -5.92798769e-01 3.99820030e-01
-1.37596011e-01 9.82402980e-01 3.02684039e-01 -5.69736302e-01
4.32061762e-01 -1.60635278e-01 3.95272225e-01 -2.95589924e-01
-4.78608370e-01 5.50565779e-01 -9.99364704e-02 -1.59187227e-01
1.70665860e-01 -2.94566810e-01 -1.59278095e+00 -3.72443199e-02
-6.48490310e-01 -3.49159054e-02 7.17485368e-01 8.31389844e-01
2.03673720e-01 7.33560920e-01 8.49044502e-01 -4.88823622e-01
-5.57250917e-01 -7.96638787e-01 -4.48935419e-01 3.27166580e-02
2.92845756e-01 2.12965459e-01 1.25044718e-01 2.93329000e-01] | [11.928507804870605, 2.229860544204712] |
dcf09d35-9ed3-4487-b696-37bba1282e3f | split-localized-conformal-prediction | 2206.13092 | null | https://arxiv.org/abs/2206.13092v2 | https://arxiv.org/pdf/2206.13092v2.pdf | Split Localized Conformal Prediction | Conformal prediction is a simple and powerful tool that can quantify uncertainty without any distributional assumptions. Many existing methods only address the average coverage guarantee, which is not ideal compared to the stronger conditional coverage guarantee. Existing methods of approximating conditional coverage require additional models or time effort, which makes them not easy to scale. In this paper, we propose a modified non-conformity score by leveraging the local approximation of the conditional distribution using kernel density estimation. The modified score inherits the spirit of split conformal methods, which is simple and efficient and can scale to high dimensional settings. We also proposed a unified framework that brings together our method and several state-of-the-art. We perform extensive empirical evaluations: results measured by both average and conditional coverage confirm the advantage of our method. | ['Qiang Liu', 'Joydeep Ghosh', 'Ziyang Tang', 'Xing Han'] | 2022-06-27 | null | null | null | null | ['prediction-intervals'] | ['miscellaneous'] | [-7.29355663e-02 2.74576366e-01 -5.81077456e-01 -5.09714544e-01
-1.30116868e+00 -5.23953497e-01 6.14315450e-01 2.88491547e-01
-1.58185840e-01 1.06898057e+00 4.84991670e-01 -1.64584666e-01
-3.03578138e-01 -1.05035436e+00 -7.97948241e-01 -6.58809066e-01
-9.51752663e-02 5.20190418e-01 6.31714284e-01 2.01006815e-01
2.07056236e-02 1.43658295e-01 -1.21564448e+00 -1.54020682e-01
1.19019783e+00 1.09661376e+00 -1.44905582e-01 1.18049294e-01
1.50316641e-01 5.58915317e-01 -1.75715625e-01 -5.89314759e-01
-8.97606760e-02 -3.12596709e-01 -4.51197833e-01 -6.60127878e-01
8.28300416e-02 -4.11203265e-01 -1.58704773e-01 1.05378401e+00
3.01614612e-01 -2.68167466e-01 1.17114317e+00 -1.07532334e+00
-6.44491136e-01 7.65643179e-01 -5.95873296e-01 5.85442372e-02
4.18528169e-01 -6.75519943e-01 9.71821964e-01 -7.72773445e-01
1.70087799e-01 9.82782066e-01 1.29760039e+00 3.17220330e-01
-1.15519297e+00 -5.66113114e-01 1.80660747e-02 -3.25287074e-01
-1.61978650e+00 -1.79063752e-01 3.65924627e-01 -4.96039510e-01
3.27889472e-01 2.63354540e-01 3.96923721e-01 8.89667511e-01
4.65178043e-01 6.50151491e-01 1.20515692e+00 -4.38434571e-01
5.50568044e-01 2.52302319e-01 -2.70956978e-02 5.03573835e-01
5.55601299e-01 4.50015217e-01 -2.04340607e-01 -6.92190766e-01
7.05952764e-01 1.39779389e-01 -2.59140640e-01 -6.67361557e-01
-9.51617181e-01 1.04032207e+00 1.56508446e-01 1.12097919e-01
-5.46268886e-03 2.49523580e-01 5.04132845e-02 -1.84843421e-01
8.48374486e-01 -1.33456141e-01 -1.16657913e-01 -1.87412024e-01
-1.20314360e+00 2.08462641e-01 6.73745334e-01 1.11091912e+00
4.14080769e-01 -5.28208494e-01 -6.92197442e-01 4.57808763e-01
3.81972253e-01 7.08379924e-01 -2.75100261e-01 -6.70159757e-01
7.85001069e-02 2.83713073e-01 4.13947821e-01 -7.08917141e-01
-5.40680885e-01 -5.61925828e-01 -9.02848721e-01 -5.31225614e-02
2.34497413e-01 -1.29893824e-01 -4.97944832e-01 1.89533401e+00
4.63746578e-01 2.48655140e-01 -5.94865121e-02 7.28852391e-01
3.08201700e-01 3.14440250e-01 7.52732111e-03 -3.41733545e-01
9.96992767e-01 -7.81310022e-01 -6.78988874e-01 3.77122909e-01
5.79659641e-01 -5.59541821e-01 1.08587289e+00 8.96539167e-02
-1.09879792e+00 3.99658754e-02 -1.21039844e+00 2.64236540e-01
4.70968857e-02 -2.15778619e-01 7.68247008e-01 1.07854402e+00
-9.07691538e-01 5.56259036e-01 -8.38951170e-01 -1.04475923e-01
6.25618577e-01 -1.13566391e-01 -1.43222451e-01 -3.38034213e-01
-1.15566528e+00 7.83159852e-01 1.10878140e-01 -5.62243879e-01
-8.06394517e-01 -1.15537536e+00 -8.06006670e-01 1.83169022e-01
4.12395209e-01 -5.88213503e-01 1.28697383e+00 -1.86951384e-01
-1.34015501e+00 1.88089430e-01 -2.40139179e-02 -3.23833346e-01
7.33337581e-01 -2.38024816e-01 -2.91053236e-01 4.89365533e-02
1.64734393e-01 1.01440430e-01 2.42877796e-01 -1.12297821e+00
-4.99856353e-01 -7.29725584e-02 1.47329167e-01 -5.26044257e-02
-4.92301792e-01 -1.77740663e-01 -5.08339047e-01 -7.27046132e-01
8.53875373e-03 -6.55358851e-01 -2.32597589e-01 6.68191463e-02
-4.17112887e-01 -1.76108122e-01 1.79810390e-01 -3.48355383e-01
1.64118791e+00 -2.03890872e+00 -9.48471054e-02 3.53935063e-01
1.65679271e-03 -4.58487540e-01 2.96850890e-01 6.14467084e-01
3.61669928e-01 8.13767910e-02 -8.14305902e-01 -3.20636153e-01
3.76415193e-01 -5.91838621e-02 -3.72190446e-01 6.93451226e-01
1.16166030e-03 8.46318901e-01 -8.67371559e-01 -5.89629412e-01
-4.53775749e-02 6.77899837e-01 -7.06957996e-01 -7.48987347e-02
-2.12599829e-01 2.81419188e-01 -4.13506866e-01 6.96334600e-01
9.81921256e-01 -3.90835255e-01 1.18439123e-01 9.27576143e-03
9.87580270e-02 7.30568096e-02 -1.04127526e+00 1.54869425e+00
-3.11432749e-01 -1.53096661e-01 -3.11793476e-01 -8.71405065e-01
7.75843620e-01 2.60034442e-01 4.89383519e-01 -4.48385060e-01
-1.90329533e-02 1.42579436e-01 -6.85195744e-01 1.58695832e-01
2.56975561e-01 -6.64385617e-01 -4.58660156e-01 5.86763203e-01
-1.39479622e-01 1.28812909e-01 -3.06676716e-01 1.75072864e-01
1.10201371e+00 1.87462106e-01 4.87373680e-01 -9.87338722e-01
1.73369385e-02 -3.90541792e-01 5.05805910e-01 8.46113265e-01
1.62451124e-05 7.67768681e-01 7.97328711e-01 -1.14791758e-01
-1.04391170e+00 -1.48906767e+00 -7.77732611e-01 6.50141835e-01
3.21548223e-01 -5.00562847e-01 -9.73005354e-01 -9.79271352e-01
1.22830704e-01 8.47399771e-01 -1.06616759e+00 -7.97459185e-02
1.80287868e-01 -9.62765634e-01 5.26212454e-01 8.25615764e-01
1.22829460e-01 -1.02766752e-01 -2.78687954e-01 -1.23842679e-01
-9.11953226e-02 -9.70996857e-01 -5.59592307e-01 -2.86138445e-01
-7.92191744e-01 -9.63455379e-01 -8.65121841e-01 -1.28851369e-01
5.27432561e-01 1.78598017e-02 1.09130085e+00 -1.82267532e-01
1.31349429e-01 5.02947688e-01 -3.70700419e-01 -5.84097803e-01
-6.96661919e-02 -9.77370236e-03 2.54142046e-01 -1.84574589e-01
2.17732430e-01 -5.62927425e-01 -7.36464024e-01 5.16234457e-01
-1.04900873e+00 -9.92968529e-02 6.58625424e-01 6.96601510e-01
9.20923054e-01 -1.04383059e-01 9.39966798e-01 -1.12930548e+00
4.62514699e-01 -7.79136240e-01 -7.48553455e-01 3.81444275e-01
-9.73727882e-01 1.65375203e-01 1.54225513e-01 -4.57304828e-02
-1.16161644e+00 -2.87187666e-01 -1.25885397e-01 -1.11864015e-01
3.20966065e-01 7.13774443e-01 -1.67299971e-01 7.63677582e-02
5.34141600e-01 -1.74199849e-01 -2.32028186e-01 -4.61885303e-01
1.99520141e-01 4.97992933e-01 4.14550543e-01 -7.40510762e-01
7.34031796e-01 7.89336264e-01 2.47896627e-01 -2.16523096e-01
-1.28237188e+00 -1.62485436e-01 -4.48785394e-01 -4.47375746e-03
6.91001773e-01 -8.26102674e-01 -4.88882720e-01 1.48530526e-03
-6.45555556e-01 -2.68251836e-01 -5.23780107e-01 6.07858181e-01
-6.95494652e-01 4.64349091e-01 -3.48603010e-01 -9.68917072e-01
-2.45657608e-01 -7.29148746e-01 1.20296431e+00 3.93980816e-02
1.61404490e-01 -1.17353737e+00 6.19623899e-01 -1.74177200e-01
6.74610674e-01 5.10679126e-01 9.01406646e-01 -2.37983212e-01
-3.00638825e-01 -5.15521288e-01 -3.97570431e-01 8.14620331e-02
-1.07061543e-01 -3.44832897e-01 -8.78287137e-01 -3.80819589e-01
-9.23167467e-02 -2.53924042e-01 8.72432172e-01 6.62620068e-01
1.34382331e+00 -7.93844238e-02 -6.64434850e-01 4.20453101e-01
1.64229298e+00 -4.26080406e-01 1.00063396e+00 -4.08199988e-02
1.61803275e-01 3.88638675e-01 8.83081496e-01 7.03456342e-01
5.94316542e-01 8.57573509e-01 2.32308656e-01 7.56856278e-02
1.17654940e-02 -4.67497617e-01 1.93076476e-01 6.70773268e-01
-9.82889831e-02 1.29189398e-02 -8.59199286e-01 5.76913774e-01
-1.96311629e+00 -8.77949893e-01 4.17037122e-02 2.60526299e+00
8.21794927e-01 1.53091997e-01 2.48383194e-01 -1.83351457e-01
6.32886827e-01 -2.59252578e-01 -2.13048995e-01 2.23198719e-02
-7.84481168e-02 2.20409364e-01 5.60294211e-01 6.77170932e-01
-1.07781959e+00 3.42914492e-01 7.84971046e+00 1.17795086e+00
-4.18241650e-01 7.04302430e-01 6.08721673e-01 -1.98031187e-01
-1.06214201e+00 1.47093266e-01 -7.30711162e-01 6.68584466e-01
9.34419572e-01 -2.59352773e-01 -2.52065714e-02 8.03817093e-01
-4.50918317e-01 -2.75955349e-01 -1.11419213e+00 4.68810946e-01
4.11163829e-02 -1.49220514e+00 -3.20187211e-01 2.14224994e-01
9.58591938e-01 6.03965111e-02 2.54619345e-02 2.86798418e-01
3.95707011e-01 -1.15561283e+00 7.68692613e-01 1.17662179e+00
1.19085348e+00 -9.94939327e-01 1.05209565e+00 3.45875919e-01
-1.12368464e+00 2.56414771e-01 -5.70487082e-01 1.37186572e-01
4.08710063e-01 1.15512860e+00 -1.46259606e-01 8.20371985e-01
8.56148899e-01 3.14542830e-01 -1.43400177e-01 1.14337540e+00
-2.05054563e-02 7.28156626e-01 -5.15559018e-01 -3.03847566e-02
-1.42231315e-01 -3.23286839e-02 4.01498795e-01 1.24919486e+00
8.50239158e-01 1.99056454e-02 8.00141767e-02 8.86195779e-01
3.21124904e-02 9.02931169e-02 -5.44396400e-01 4.05527502e-01
7.50661790e-01 9.81042087e-01 -5.76387763e-01 -3.99773046e-02
-7.81788349e-01 4.77565199e-01 4.74425733e-01 1.10176817e-01
-1.14157462e+00 -1.89833328e-01 2.55151898e-01 3.20048153e-01
4.53798234e-01 1.02302330e-02 -3.78657788e-01 -1.11515093e+00
1.57212272e-01 -2.27271259e-01 6.05151832e-01 -2.73072213e-01
-1.65228522e+00 3.17385197e-01 4.80776489e-01 -1.30371118e+00
1.77055895e-02 -4.43096191e-01 -4.66345906e-01 6.60505176e-01
-1.50587332e+00 -1.14879417e+00 -1.62783727e-01 3.44443023e-01
-2.64966846e-01 9.82127562e-02 1.14566755e+00 2.88165838e-01
-2.59053409e-01 1.05768776e+00 4.30392891e-01 -3.52430314e-01
7.46364415e-01 -1.30251062e+00 -4.06704992e-02 6.34360850e-01
-1.66713282e-01 3.11575234e-01 6.38760865e-01 -7.20357418e-01
-1.10917008e+00 -9.53650594e-01 6.34596109e-01 -7.44704545e-01
6.33664548e-01 -4.07649815e-01 -6.25437856e-01 5.65820694e-01
-7.71119073e-02 1.70936629e-01 1.12320864e+00 5.19195139e-01
-5.51941812e-01 -1.40039787e-01 -1.43417978e+00 3.71876866e-01
1.01812029e+00 -2.27973789e-01 -2.78559774e-01 3.53002995e-01
7.61768699e-01 -1.98827624e-01 -1.25136435e+00 9.74248290e-01
7.26094782e-01 -1.25102651e+00 6.94588840e-01 -1.03542462e-01
4.38977867e-01 -1.91444084e-01 -8.22711945e-01 -9.79627490e-01
-5.11561513e-01 -3.00663352e-01 -4.16647255e-01 1.24554229e+00
6.87879920e-01 -8.38421881e-01 6.94989026e-01 7.00812876e-01
-2.62338936e-01 -1.13686371e+00 -1.21162868e+00 -1.00709879e+00
5.86494267e-01 -6.81287527e-01 9.02068615e-01 8.67680073e-01
4.10310864e-01 -1.41387239e-01 -5.27907908e-01 2.83629328e-01
9.65162158e-01 1.22495897e-01 3.48914087e-01 -1.44381368e+00
-5.97006559e-01 -3.15681785e-01 -4.06539023e-01 -7.73586571e-01
-1.08051918e-01 -8.35355163e-01 -1.36272376e-02 -1.60554624e+00
8.69342208e-01 -8.57951462e-01 -5.21188557e-01 3.23539585e-01
-7.30443299e-02 3.35850239e-01 -4.06840771e-01 1.16503470e-01
-7.01451600e-01 7.98558712e-01 9.19422507e-01 4.87333313e-02
2.91804314e-01 2.41668895e-02 -1.07813740e+00 7.05933511e-01
6.93612576e-01 -5.68628490e-01 -5.32680929e-01 -1.08351700e-01
6.08905077e-01 9.20858309e-02 2.53430367e-01 -1.04920304e+00
6.96243271e-02 -1.93737566e-01 2.90844500e-01 -5.87131798e-01
8.51810202e-02 -7.51780927e-01 2.33317643e-01 2.44958743e-01
-8.38777944e-02 -3.03462654e-01 1.94680274e-01 1.10508859e+00
-1.05357081e-01 -3.26895416e-01 7.58673251e-01 4.81535405e-01
5.31351492e-02 5.14419734e-01 2.01979592e-01 7.06472844e-02
1.42460620e+00 2.54158467e-01 -4.38884020e-01 -6.15730703e-01
-4.03882116e-01 9.08126980e-02 7.87251711e-01 1.91970784e-02
5.87496877e-01 -1.94887543e+00 -8.62316668e-01 -2.53403693e-01
5.16877890e-01 -1.55209109e-01 3.91555309e-01 1.28139639e+00
-2.40338027e-01 5.08836091e-01 3.47371250e-01 -6.64479733e-01
-4.79033411e-01 8.12694192e-01 1.47570670e-01 -2.95672864e-01
-7.30315506e-01 6.13599181e-01 4.64099467e-01 -5.31732380e-01
7.71311149e-02 -1.84063673e-01 1.38368958e-03 -3.44560564e-01
7.11797833e-01 3.64621341e-01 -1.59492493e-01 -2.54000187e-01
-5.06474853e-01 5.66190898e-01 2.33913347e-01 -2.39668384e-01
1.18193519e+00 -8.05548653e-02 -7.33003467e-02 4.79518920e-01
8.78347039e-01 3.84350896e-01 -1.33513010e+00 -2.61391997e-01
-1.31679416e-01 -6.07560873e-01 1.24175102e-01 -8.61796498e-01
-6.54301822e-01 6.31064832e-01 4.95917410e-01 2.75228411e-01
1.12061715e+00 4.41739649e-01 5.68347692e-01 -1.78585619e-01
8.18421185e-01 -8.71507764e-01 -1.72408283e-01 1.03446655e-01
8.44756901e-01 -1.02546370e+00 2.07215786e-01 -6.38150275e-01
-4.77766037e-01 5.46739876e-01 3.26249778e-01 -1.15982763e-01
1.27887762e+00 7.49692798e-01 -3.97904336e-01 -7.84846172e-02
-5.27700722e-01 1.51338890e-01 6.04449987e-01 5.89017272e-01
4.88132745e-01 3.73430043e-01 -5.86639404e-01 1.27021289e+00
3.05265263e-02 -9.28318501e-02 2.35001072e-01 6.33910179e-01
-4.89078850e-01 -9.99692678e-01 -1.40821710e-01 7.19489753e-01
-4.99104530e-01 -3.06317490e-02 1.54784292e-01 6.32668257e-01
-1.84995711e-01 7.27078378e-01 3.39950714e-03 -4.02323306e-01
1.53045997e-01 -1.14693753e-01 6.18170798e-01 -3.05893838e-01
4.77166533e-01 4.35115993e-02 8.00935738e-03 -5.67678928e-01
-5.26569366e-01 -8.43544245e-01 -1.07670581e+00 -4.25274223e-01
-6.26853108e-01 3.21178973e-01 4.60038841e-01 8.87570918e-01
3.11023891e-01 3.19838881e-01 4.58501488e-01 -4.84008849e-01
-7.92521775e-01 -9.48650777e-01 -8.27161968e-01 -6.40242547e-02
1.23499654e-01 -1.21046281e+00 -1.47320330e-01 -5.08933067e-01] | [7.882724285125732, 4.429717540740967] |
fcc350a2-7096-4eb5-8f51-402061b97a53 | a-survey-on-phrase-structure-learning-methods | 1406.5598 | null | http://arxiv.org/abs/1406.5598v1 | http://arxiv.org/pdf/1406.5598v1.pdf | A survey on phrase structure learning methods for text classification | Text classification is a task of automatic classification of text into one of
the predefined categories. The problem of text classification has been widely
studied in different communities like natural language processing, data mining
and information retrieval. Text classification is an important constituent in
many information management tasks like topic identification, spam filtering,
email routing, language identification, genre classification, readability
assessment etc. The performance of text classification improves notably when
phrase patterns are used. The use of phrase patterns helps in capturing
non-local behaviours and thus helps in the improvement of text classification
task. Phrase structure extraction is the first step to continue with the phrase
pattern identification. In this survey, detailed study of phrase structure
learning methods have been carried out. This will enable future work in several
NLP tasks, which uses syntactic information from phrase structure like grammar
checkers, question answering, information extraction, machine translation, text
classification. The paper also provides different levels of classification and
detailed comparison of the phrase structure learning methods. | ['Mary Priya Sebastian', 'Reshma Prasad'] | 2014-06-21 | null | null | null | null | ['genre-classification'] | ['computer-vision'] | [ 5.90330958e-01 -8.35356191e-02 -5.14968157e-01 -3.33269268e-01
-4.23574150e-01 -6.13752842e-01 7.33719647e-01 1.03011513e+00
-3.34013581e-01 8.52659702e-01 4.00506318e-01 -7.86330998e-01
-4.33444470e-01 -9.05436516e-01 1.45538002e-01 -6.34275556e-01
-3.03427950e-02 8.62151444e-01 3.19571465e-01 -2.36792102e-01
1.06123304e+00 1.91829011e-01 -1.82678485e+00 7.32204139e-01
4.38866794e-01 6.55418932e-01 1.66520819e-01 9.82168436e-01
-1.07122171e+00 7.53601849e-01 -6.73962533e-01 -1.25942186e-01
-2.99447536e-01 -3.20732117e-01 -1.42886484e+00 2.19832122e-01
7.34683126e-02 6.14027619e-01 6.90738022e-01 1.05146587e+00
1.64624050e-01 1.82901829e-01 8.19460392e-01 -9.72664118e-01
-1.85164809e-02 6.41103208e-01 -6.45653307e-01 5.52151084e-01
8.72421265e-01 -7.42142737e-01 1.34984565e+00 -6.80355251e-01
5.20629287e-01 1.41650259e+00 6.01374269e-01 1.09629519e-01
-1.02334952e+00 -6.69244170e-01 8.61614570e-02 2.26735741e-01
-9.93655384e-01 5.70861548e-02 5.38896739e-01 -4.09155369e-01
1.38387871e+00 5.31864464e-01 3.04074049e-01 5.97800136e-01
7.26741910e-01 7.77769268e-01 1.31763220e+00 -1.07212591e+00
-7.48971254e-02 4.73965287e-01 1.16939998e+00 6.26063466e-01
1.03367873e-01 -3.59215796e-01 -5.87364018e-01 -5.50871849e-01
3.58201861e-02 -2.39572152e-01 4.67853956e-02 1.57075256e-01
-8.73465836e-01 1.06076205e+00 -3.81749332e-01 9.05186772e-01
-9.67970192e-02 -4.20242310e-01 7.27256536e-01 7.56550908e-01
4.93620127e-01 4.72699046e-01 -9.99727249e-01 -4.27932024e-01
-6.74740016e-01 3.38712633e-01 1.35983431e+00 1.04432237e+00
4.92978454e-01 -6.60391450e-01 -4.27443609e-02 9.32881951e-01
4.58687723e-01 2.80096501e-01 9.07401860e-01 -2.04162046e-01
7.13034093e-01 1.02887154e+00 -6.72036558e-02 -1.08823919e+00
-9.55743849e-01 -1.95176661e-01 -6.86354637e-01 -4.00173306e-01
3.45093280e-01 1.05711259e-02 -7.12498605e-01 8.97768319e-01
2.57494241e-01 -4.95494843e-01 -6.49001300e-02 1.28057376e-01
8.96908641e-01 1.01956022e+00 3.51916373e-01 -5.60272038e-01
1.97283483e+00 -5.31399429e-01 -8.19076061e-01 1.10312931e-01
1.06635392e+00 -1.21565092e+00 8.33932936e-01 5.07086992e-01
-4.26786214e-01 -4.47973192e-01 -4.67078298e-01 1.23401396e-01
-8.62723112e-01 -1.35959998e-01 6.05461180e-01 1.01584756e+00
-6.22455239e-01 2.63332754e-01 -4.27946478e-01 -9.00756121e-01
1.41542807e-01 8.58601213e-01 -3.61319900e-01 2.93193966e-01
-1.10791683e+00 6.41345084e-01 7.14153826e-01 -7.14633584e-01
4.94640827e-01 -2.86209643e-01 -6.78987086e-01 -1.63378324e-02
1.20618306e-01 -4.56808388e-01 1.21109462e+00 -8.79328787e-01
-1.00250101e+00 1.22823834e+00 -5.65852463e-01 -3.73190254e-01
-1.71468839e-01 2.59957090e-02 -4.42436844e-01 -2.23483115e-01
2.75778502e-01 2.88446456e-01 9.60439622e-01 -7.16276050e-01
-1.31119716e+00 -4.97352451e-01 -4.84574795e-01 2.85699517e-01
-2.81308383e-01 5.81929922e-01 1.88674062e-01 -7.28158534e-01
2.14352995e-01 -7.88512468e-01 8.18169191e-02 -9.87896502e-01
-3.89314175e-01 -1.11451948e+00 1.27566206e+00 -5.04366636e-01
1.39228618e+00 -1.65268922e+00 2.85928766e-03 3.79255384e-01
-3.76147334e-03 -1.16078675e-01 3.29154640e-01 7.89694726e-01
-2.79514313e-01 4.88696963e-01 2.25340156e-03 6.40490353e-02
-5.24070673e-02 2.14375168e-01 -4.02873248e-01 1.23830110e-01
4.11967337e-02 5.32569647e-01 -6.83859825e-01 -7.37013996e-01
8.72933567e-02 -1.49135411e-01 -2.43333489e-01 -2.44375333e-01
-2.54585445e-01 3.82937849e-01 -7.72786558e-01 5.25475502e-01
1.34257600e-01 -1.90428630e-01 1.73981801e-01 2.06486732e-01
-3.18475991e-01 8.19413006e-01 -1.17655957e+00 1.10766792e+00
-4.60116506e-01 7.12760448e-01 -3.55405599e-01 -1.33929062e+00
1.10872412e+00 6.28010452e-01 5.56584299e-01 -6.33001328e-01
2.90287584e-01 1.04818836e-01 1.74034074e-01 -7.36843646e-01
6.32686615e-01 -1.87725812e-01 -3.49200457e-01 9.21984851e-01
1.36260949e-02 2.10852429e-01 6.21466279e-01 -2.03376785e-01
8.79341841e-01 -2.12829039e-01 8.67022634e-01 -7.26730108e-01
1.26944304e+00 3.74149054e-01 2.35638674e-02 8.67932975e-01
-5.81644624e-02 5.52296191e-02 5.00120282e-01 -6.00325108e-01
-9.47169602e-01 -2.46587858e-01 -6.41231239e-01 1.69353116e+00
-3.34820300e-01 -6.18660152e-01 -5.40698707e-01 -7.98100591e-01
-3.60096991e-02 5.01414597e-01 -2.85155743e-01 1.98366463e-01
-7.31377184e-01 -1.16723478e+00 2.51081079e-01 2.67714977e-01
6.29469037e-01 -1.37067378e+00 -1.30469292e-01 4.39774901e-01
-1.15219958e-01 -8.02233517e-01 -6.72701895e-02 6.11602366e-01
-1.28479290e+00 -1.06970167e+00 -1.04430698e-01 -1.46042442e+00
4.78259146e-01 9.16796550e-02 1.19131505e+00 4.30405140e-01
-1.44583836e-01 2.34890655e-01 -1.00846899e+00 -8.06749701e-01
-7.64993250e-01 8.44967544e-01 -5.79462461e-02 -4.23114896e-01
1.26128030e+00 -2.74281263e-01 1.58524051e-01 3.54338974e-01
-7.61043608e-01 -1.78377599e-01 4.53003168e-01 1.13682139e+00
2.31375828e-01 8.11556458e-01 4.93156344e-01 -1.46117127e+00
1.03769732e+00 -2.77768761e-01 -3.64664048e-01 2.55059570e-01
-7.03186989e-01 4.05165017e-01 5.57925284e-01 -1.27842247e-01
-1.02115822e+00 -4.12853435e-02 -4.45005983e-01 1.01751494e+00
-7.39573777e-01 6.37487054e-01 -1.61089525e-02 -2.63652354e-02
6.02816939e-01 5.12607992e-01 -2.37542450e-01 -5.71518481e-01
-2.19796881e-01 1.27589345e+00 -2.46484622e-01 -4.59339768e-01
6.14030838e-01 -1.39839366e-01 1.55065864e-01 -1.22820163e+00
-8.82799149e-01 -1.50365603e+00 -1.07315266e+00 1.40616447e-01
7.50125229e-01 -2.13325992e-01 -6.11314833e-01 2.98544317e-01
-1.15481114e+00 3.60774875e-01 2.65021652e-01 1.34607092e-01
-4.21554565e-01 4.25680727e-01 -5.17977118e-01 -8.32731426e-01
-6.68223739e-01 -7.77294695e-01 1.32180500e+00 3.65349092e-02
-8.28343809e-01 -1.57323968e+00 1.02151245e-01 6.16551936e-01
-1.30318597e-01 -4.48132217e-01 1.54954517e+00 -1.28522277e+00
1.10290229e-01 -3.75161946e-01 1.33709326e-01 -7.96335116e-02
4.44996536e-01 -3.41787279e-01 -7.85349190e-01 -6.52559102e-02
2.45666459e-01 5.54255617e-04 6.67927206e-01 2.06581056e-01
9.76915956e-01 -4.94637012e-01 -4.94684815e-01 -2.49080285e-02
1.19432402e+00 5.51916838e-01 3.12731743e-01 8.46426845e-01
5.10011077e-01 1.07068086e+00 8.09640646e-01 1.57354400e-01
-7.76905268e-02 3.15821022e-01 -3.02098840e-01 5.15025735e-01
2.84961641e-01 1.38850570e-01 1.38770610e-01 8.85318935e-01
1.46206319e-01 -3.84468853e-01 -1.28910470e+00 4.44882289e-02
-1.75020576e+00 -1.03037429e+00 -8.20063949e-01 1.74492264e+00
8.67009640e-01 3.97444189e-01 2.64131367e-01 1.07703292e+00
7.19900846e-01 -2.12963820e-01 3.23154360e-01 -9.11166608e-01
2.68893242e-01 4.41050708e-01 4.48082298e-01 5.39696038e-01
-1.66583693e+00 1.00696778e+00 5.87024784e+00 8.63252640e-01
-7.50477612e-01 -1.76781863e-01 4.74502563e-01 8.39641929e-01
2.79118270e-01 -1.53144717e-01 -1.39698637e+00 4.13009495e-01
1.01038527e+00 -8.81904364e-02 -1.37823716e-01 6.16621733e-01
3.27223152e-01 -5.58751941e-01 -9.20283377e-01 8.48734915e-01
1.90635435e-02 -1.14796197e+00 5.07656038e-01 -5.35303578e-02
5.63461125e-01 -4.51136470e-01 -1.76942140e-01 3.48381251e-01
-1.92107126e-01 -8.84029329e-01 9.98785496e-02 -4.18679137e-03
-1.11457847e-01 -1.01237631e+00 1.18062460e+00 7.74526775e-01
-1.05025613e+00 -2.42927909e-01 -2.81287700e-01 -3.71187866e-01
-5.00406250e-02 3.98838252e-01 -1.03197289e+00 4.98949885e-01
7.18728304e-01 7.38707066e-01 -6.66994631e-01 9.82454360e-01
1.59017127e-02 9.30727303e-01 -2.47211263e-01 -7.09636331e-01
3.09999138e-01 -2.87205070e-01 7.69711196e-01 1.57363510e+00
-2.42418855e-01 3.90958115e-02 3.69508028e-01 -1.60190202e-02
1.59388945e-01 9.27543879e-01 -4.57198948e-01 -9.65832826e-03
-4.06336449e-02 1.04075491e+00 -1.36008322e+00 -2.01391950e-01
-5.80604672e-01 3.85729373e-01 -3.41743261e-01 -1.51635498e-01
-4.62708287e-02 -7.16812849e-01 1.78698033e-01 1.67429417e-01
-5.84444180e-02 -2.39390850e-01 -6.13839149e-01 -7.79185414e-01
-7.08513334e-02 -9.68234956e-01 7.52640545e-01 -1.77593842e-01
-1.18004227e+00 4.09190625e-01 2.92505115e-01 -1.00894737e+00
-4.32134330e-01 -1.13554442e+00 -6.42202497e-01 7.57878721e-01
-1.17572320e+00 -9.08597231e-01 1.66152626e-01 3.36499631e-01
1.05608010e+00 -6.09299242e-01 1.09377122e+00 7.93335810e-02
-3.97253454e-01 3.08699787e-01 1.61556482e-01 2.88884848e-01
6.75529540e-01 -1.63529563e+00 1.83941975e-01 4.16593581e-01
4.76948619e-01 5.82883716e-01 9.22554314e-01 -6.86497927e-01
-1.00697982e+00 -7.72184849e-01 1.61686623e+00 -7.38348484e-01
9.50104713e-01 -3.92699063e-01 -1.06737828e+00 2.49514818e-01
2.35042691e-01 -8.73587370e-01 1.18957078e+00 5.72779000e-01
7.11406916e-02 3.77729945e-02 -8.42104554e-01 3.33723724e-01
4.34893847e-01 -3.54156643e-01 -1.15215611e+00 8.27725053e-01
3.81252021e-01 2.25033760e-02 -7.33423948e-01 -1.59591556e-01
3.93555880e-01 -5.53058743e-01 5.64795971e-01 -9.77085888e-01
4.46030274e-02 -8.31897259e-02 2.81357855e-01 -8.45061004e-01
-3.49837631e-01 -6.34311199e-01 2.27595240e-01 1.37552011e+00
8.45784426e-01 -6.33334339e-01 1.10876405e+00 1.55734688e-01
3.70050132e-01 -3.53805900e-01 -7.03372896e-01 -4.83103335e-01
2.78400064e-01 -5.00935197e-01 2.12871864e-01 8.58675778e-01
4.56526726e-01 1.08016121e+00 7.56845698e-02 -2.74022251e-01
5.28821588e-01 3.73812318e-01 5.35394907e-01 -1.84567416e+00
-1.48585096e-01 -6.41446829e-01 -7.23723292e-01 -8.99591029e-01
1.76926762e-01 -8.98877978e-01 -1.07675850e-01 -1.39440978e+00
1.83871657e-01 -4.65730637e-01 2.86368608e-01 3.58353496e-01
-3.46031427e-01 -2.01800808e-01 -2.64915347e-01 4.31831270e-01
-4.63454127e-01 -1.16532609e-01 1.04950893e+00 -2.51901954e-01
-3.60010296e-01 7.21353889e-01 -5.46818078e-01 7.03875899e-01
1.26945198e+00 -1.02959085e+00 -3.81631017e-01 9.48569104e-02
5.26876211e-01 -3.09128881e-01 -3.61928642e-01 -6.24819100e-01
4.06441092e-01 -1.74934283e-01 1.62756190e-01 -8.27790082e-01
-3.20294470e-01 -9.48561192e-01 -4.85050589e-01 5.06006718e-01
-3.91236097e-01 4.06924069e-01 1.52846262e-01 5.66084385e-01
-4.44280714e-01 -1.01686966e+00 5.17755270e-01 -3.00357401e-01
-7.78855622e-01 -1.04390740e-01 -1.08766162e+00 -1.07234024e-01
9.24854696e-01 -3.76288027e-01 2.67652571e-02 4.58287708e-02
-6.13942266e-01 2.66211212e-01 -7.35817775e-02 6.49234474e-01
7.16564134e-02 -7.32340932e-01 -5.15215039e-01 1.41106084e-01
2.56983757e-01 -9.01732370e-02 -5.57584286e-01 7.57742822e-01
-6.66444838e-01 1.04688871e+00 1.01164393e-01 -8.08781326e-01
-1.95082521e+00 3.62117648e-01 -1.37269214e-01 -8.47635090e-01
-5.11339962e-01 5.46361446e-01 -2.38622889e-01 -4.22589988e-01
4.49756026e-01 -3.46855253e-01 -1.20053375e+00 2.19877169e-01
5.73187768e-01 1.23414479e-01 5.00104070e-01 -6.98856533e-01
-1.79131463e-01 7.87132680e-01 -4.80489194e-01 1.09803259e-01
9.84027028e-01 -1.86686322e-01 -6.20039880e-01 6.40407622e-01
1.14003193e+00 -1.83115125e-01 1.12882674e-01 -3.25617120e-02
1.27122951e+00 2.73168012e-02 1.24594383e-01 -6.08031273e-01
-4.42887917e-02 7.62901306e-01 3.79846156e-01 1.01114810e+00
1.06851697e+00 -7.62750059e-02 5.47252119e-01 9.79828179e-01
5.14980018e-01 -1.35730374e+00 -2.23580524e-01 1.14569175e+00
4.53795761e-01 -1.36912835e+00 1.02717668e-01 -6.81568444e-01
-1.81367189e-01 1.52985346e+00 7.60393813e-02 1.34678125e-01
1.29867721e+00 3.34928304e-01 -8.71160626e-02 -2.57255673e-01
-8.06390405e-01 -2.56130993e-01 4.80248600e-01 6.72336221e-01
1.02029943e+00 -2.51841396e-01 -8.88498902e-01 3.21649879e-01
-5.10175705e-01 -4.92155105e-01 2.43539497e-01 1.44058013e+00
-1.07571793e+00 -1.87837172e+00 -7.60209560e-01 1.04113078e+00
-9.67823982e-01 -2.11047411e-01 -8.71406555e-01 7.75344849e-01
-3.48552689e-02 1.18550694e+00 1.49248749e-01 -6.62925392e-02
1.23493552e-01 6.39133513e-01 3.02582115e-01 -1.22473872e+00
-1.01414239e+00 -6.04020990e-02 4.06741261e-01 1.68115079e-01
-8.15723598e-01 -9.12640452e-01 -1.16589844e+00 -2.35352278e-01
-6.06994629e-01 8.49260449e-01 6.95191383e-01 1.33255196e+00
-2.40466014e-01 1.71066061e-01 4.84463692e-01 -1.41821086e-01
-2.38231227e-01 -1.23214364e+00 -5.91445506e-01 3.51760536e-01
4.80088741e-02 -5.13772964e-01 -2.46054813e-01 5.02821207e-01] | [10.498367309570312, 8.375215530395508] |
795a1a0a-6d9d-427a-bc02-b44c5bb47b5c | samscore-a-semantic-structural-similarity | 2305.15367 | null | https://arxiv.org/abs/2305.15367v1 | https://arxiv.org/pdf/2305.15367v1.pdf | SAMScore: A Semantic Structural Similarity Metric for Image Translation Evaluation | Image translation has wide applications, such as style transfer and modality conversion, usually aiming to generate images having both high degrees of realism and faithfulness. These problems remain difficult, especially when it is important to preserve semantic structures. Traditional image-level similarity metrics are of limited use, since the semantics of an image are high-level, and not strongly governed by pixel-wise faithfulness to an original image. Towards filling this gap, we introduce SAMScore, a generic semantic structural similarity metric for evaluating the faithfulness of image translation models. SAMScore is based on the recent high-performance Segment Anything Model (SAM), which can perform semantic similarity comparisons with standout accuracy. We applied SAMScore on 19 image translation tasks, and found that it is able to outperform all other competitive metrics on all of the tasks. We envision that SAMScore will prove to be a valuable tool that will help to drive the vibrant field of image translation, by allowing for more precise evaluations of new and evolving translation models. The code is available at https://github.com/Kent0n-Li/SAMScore. | ['You Zhang', 'Alan C. Bovik', 'Jun Ma', 'Kai Wang', 'Wenxuan Yang', 'Meixu Chen', 'Yunxiang Li'] | 2023-05-24 | null | null | null | null | ['style-transfer', 'semantic-textual-similarity', 'semantic-similarity'] | ['computer-vision', 'natural-language-processing', 'natural-language-processing'] | [ 5.08623481e-01 -7.18948767e-02 -2.98084825e-01 -3.52832139e-01
-8.80907476e-01 -7.09966004e-01 8.85959685e-01 1.29480273e-01
-3.39670151e-01 6.43923938e-01 4.11753446e-01 -1.36898279e-01
2.87882119e-01 -6.74142122e-01 -7.13393807e-01 -3.90566617e-01
3.97827625e-01 4.17408139e-01 2.05368742e-01 -4.54944789e-01
2.11781904e-01 3.19158196e-01 -1.44640362e+00 5.01278639e-01
6.99688911e-01 8.30843091e-01 3.11587363e-01 4.28094059e-01
2.06324495e-02 5.41752577e-01 -3.55896026e-01 -8.06130707e-01
7.69201964e-02 -9.87774432e-01 -1.03668833e+00 -2.08843946e-01
6.77210450e-01 -1.33409686e-02 -1.33777976e-01 1.18039227e+00
5.36034048e-01 1.22924903e-02 6.15419924e-01 -1.22430074e+00
-8.93424213e-01 3.21029812e-01 -2.66091943e-01 2.06812501e-01
5.87080359e-01 1.76120281e-01 1.03072238e+00 -9.80425477e-01
1.14000416e+00 1.11824894e+00 6.61837161e-01 5.51411986e-01
-1.43768418e+00 -3.68748933e-01 -3.59843791e-01 4.33398813e-01
-1.15479743e+00 -4.87760723e-01 6.47147357e-01 -2.92864740e-01
6.38588786e-01 4.83694851e-01 7.03738153e-01 1.36485851e+00
1.97410882e-01 7.53298044e-01 1.56772161e+00 -5.32438576e-01
6.24100678e-02 2.87006140e-01 -5.12836039e-01 6.22353315e-01
-5.92526086e-02 1.68015197e-01 -7.41865098e-01 1.28440201e-01
7.31327295e-01 -2.77624279e-01 -4.18212563e-01 -4.25201267e-01
-1.78512859e+00 8.86827767e-01 7.11966515e-01 6.31858647e-01
-2.74221599e-01 1.34717524e-01 3.17637235e-01 4.16068673e-01
4.04614747e-01 7.26537943e-01 -5.91731966e-02 -4.08847362e-01
-1.05242419e+00 2.85368264e-01 4.62415129e-01 8.38892102e-01
4.81285781e-01 -1.67068720e-01 -3.11222106e-01 7.90840447e-01
-9.89012718e-02 4.44125414e-01 4.86845523e-01 -1.19410002e+00
9.50330049e-02 3.11150640e-01 -1.17294461e-01 -1.06172621e+00
-6.87554330e-02 -4.14349586e-01 -7.88545132e-01 3.45556587e-01
2.64421910e-01 3.68662447e-01 -6.98405206e-01 1.94468033e+00
1.25432387e-01 7.80387968e-03 -1.32052496e-01 1.07910490e+00
9.43986952e-01 5.17618060e-01 1.44736990e-01 4.45120633e-02
1.38120735e+00 -8.84050190e-01 -5.22951007e-01 -1.52710468e-01
3.72310549e-01 -1.26182544e+00 1.58984375e+00 5.39601129e-03
-1.55529654e+00 -5.87092280e-01 -1.03898346e+00 -3.10135096e-01
-3.47931921e-01 -1.81393743e-01 4.46920007e-01 5.14861345e-01
-1.36280620e+00 6.10297501e-01 -5.00963509e-01 -7.51171887e-01
5.49409628e-01 -1.79017372e-02 -6.42481089e-01 -2.10644469e-01
-1.26652253e+00 1.36993110e+00 3.08494925e-01 -1.19982280e-01
-6.62974417e-01 -7.22570121e-01 -7.30404615e-01 -1.96021557e-01
-7.82361776e-02 -1.11511695e+00 1.20659447e+00 -1.17602098e+00
-1.29483557e+00 1.49164677e+00 -7.11015314e-02 -2.24579811e-01
8.44832897e-01 2.25271314e-01 -3.07004124e-01 2.92260081e-01
3.03653270e-01 1.18648982e+00 7.33303130e-01 -1.33431137e+00
-3.72308761e-01 -1.76975623e-01 -8.46936554e-02 3.66334498e-01
-3.89963895e-01 1.58270568e-01 -3.86357039e-01 -8.45041692e-01
-7.36527191e-03 -1.07522511e+00 6.27648532e-02 4.18966204e-01
-9.80379879e-02 2.41544083e-01 5.73103189e-01 -6.96043015e-01
7.07024276e-01 -1.88463318e+00 3.62795889e-01 -5.46213463e-02
8.97748247e-02 3.83259833e-01 -3.04532349e-01 5.35977960e-01
2.15480719e-02 -1.83339475e-03 -5.59789300e-01 -1.77296996e-01
-9.59111825e-02 7.32248947e-02 -2.37991735e-01 2.77966529e-01
2.48977154e-01 1.43727231e+00 -1.02400374e+00 -5.86101890e-01
3.92516941e-01 6.33224010e-01 -3.58332038e-01 1.18981093e-01
-9.47053358e-02 7.43943751e-01 -1.13308914e-01 4.86228317e-01
4.97335583e-01 -2.83253402e-01 -1.03208870e-01 -3.61902744e-01
1.03317566e-01 9.98737011e-03 -6.56847835e-01 2.10453796e+00
-5.77887058e-01 8.75731468e-01 -2.37876639e-01 -6.90361798e-01
7.41146624e-01 1.80908695e-01 3.74351442e-01 -1.09982002e+00
1.08214460e-01 3.87338847e-01 -1.36493564e-01 -3.31588387e-01
6.12380385e-01 -4.55133319e-01 7.41646904e-03 3.77181053e-01
7.14357123e-02 -7.29710460e-01 1.37285024e-01 2.80177653e-01
7.52881587e-01 3.73418808e-01 2.64971584e-01 -2.50585049e-01
5.61845958e-01 2.75855243e-01 8.66952837e-02 4.99270946e-01
-3.35235536e-01 1.20430028e+00 1.98176846e-01 -3.43174368e-01
-1.27710903e+00 -1.48971128e+00 -1.24770150e-01 8.40724170e-01
2.84571469e-01 -2.07254991e-01 -8.54556322e-01 -3.85681957e-01
-3.81443381e-01 7.02562690e-01 -7.20503330e-01 -3.81402493e-01
-2.29885370e-01 -4.71758872e-01 5.69291294e-01 2.97185361e-01
5.70386589e-01 -1.21619296e+00 -5.22439897e-01 1.15135647e-01
-7.74846673e-01 -1.24788344e+00 -5.63179910e-01 -3.37948084e-01
-8.08620811e-01 -7.42350817e-01 -1.16908002e+00 -8.96864235e-01
5.53168237e-01 3.89077544e-01 1.44294715e+00 1.47722527e-01
-3.28893125e-01 3.74769121e-01 -5.24697125e-01 -1.78544372e-01
-7.78247058e-01 1.74197569e-01 -2.66692638e-01 -1.77157536e-01
4.74354513e-02 -6.02618217e-01 -8.13851297e-01 5.89083493e-01
-1.22615635e+00 4.13378954e-01 5.56003571e-01 8.11055183e-01
5.52275956e-01 -4.91648734e-01 3.84456009e-01 -7.46018171e-01
7.87675023e-01 -2.31724918e-01 -8.27865228e-02 2.58671910e-01
-5.89228511e-01 -1.92987218e-01 4.39705342e-01 -2.41248682e-01
-8.88461769e-01 -8.28739256e-02 -3.42736870e-01 -2.85854578e-01
-2.41643377e-02 3.05156320e-01 1.24513634e-01 -3.09290618e-01
9.39930022e-01 4.06428188e-01 2.06308573e-01 -2.03517899e-01
6.28402233e-01 4.55148578e-01 8.46817374e-01 -5.35457313e-01
7.52638400e-01 6.42579734e-01 5.39231338e-02 -5.76108158e-01
-6.31748796e-01 -2.22876996e-01 -6.34592652e-01 -3.08900326e-01
9.13451493e-01 -7.65667439e-01 -2.30061904e-01 2.41285473e-01
-1.11168659e+00 -3.19593728e-01 -4.48523909e-01 3.30571502e-01
-8.90298665e-01 3.48050416e-01 -4.99177575e-01 2.35796850e-02
-2.05531850e-01 -1.21204281e+00 1.15020812e+00 5.23722954e-02
-6.08038843e-01 -1.15801871e+00 1.89480782e-01 6.91308856e-01
7.16440737e-01 3.94751430e-01 8.44791114e-01 -9.21266451e-02
-3.64601851e-01 -8.69918913e-02 -3.29196781e-01 2.69876570e-01
2.53367722e-02 -3.50356027e-02 -8.97445261e-01 -2.51322627e-01
-1.11149609e-01 -5.55057824e-01 8.06525826e-01 2.17391923e-01
8.75456214e-01 -1.24124112e-02 -3.10626794e-02 5.84746122e-01
1.33073151e+00 -2.18462095e-01 9.13140059e-01 4.19181585e-01
6.01034641e-01 6.43522024e-01 5.72972357e-01 -8.31206590e-02
4.69539165e-01 1.07242930e+00 3.28274369e-01 -5.55378973e-01
-6.31283522e-01 -4.96435136e-01 2.90102035e-01 7.53804386e-01
-2.47804284e-01 -9.93360057e-02 -8.36360574e-01 4.78549153e-01
-1.76745820e+00 -1.01250267e+00 -2.84045696e-01 2.11801291e+00
9.64942515e-01 -3.40847373e-02 1.87707737e-01 -8.03832486e-02
6.45174921e-01 1.43996492e-01 -2.78273493e-01 -5.37029028e-01
-4.78117317e-01 3.03684056e-01 2.15382680e-01 4.82963115e-01
-8.38803172e-01 1.10224795e+00 6.57131290e+00 9.48553026e-01
-1.16624880e+00 4.13794488e-01 7.43490636e-01 1.24444872e-01
-5.69232762e-01 5.15559167e-02 -1.11443892e-01 5.37679255e-01
8.02543819e-01 -1.69137254e-01 3.68525833e-01 4.21465963e-01
2.62992114e-01 -1.40690953e-01 -1.01836026e+00 1.03839648e+00
2.98029572e-01 -1.50102437e+00 1.32327452e-01 -1.54281408e-01
1.03349388e+00 7.08013698e-02 3.94079179e-01 -8.60210508e-02
4.34411253e-04 -1.18129420e+00 8.09073746e-01 4.43296671e-01
1.09605479e+00 -4.76354241e-01 5.50972700e-01 -1.19514719e-01
-8.02701712e-01 5.34426928e-01 -3.23251307e-01 1.12875201e-01
3.24457526e-01 4.00455713e-01 -6.07987165e-01 3.50989133e-01
6.87776804e-01 7.78769135e-01 -6.43795073e-01 8.54114234e-01
-2.74575353e-01 1.93354070e-01 4.51781861e-02 1.59594327e-01
2.31861159e-01 -2.36232489e-01 4.23306495e-01 1.14924085e+00
4.20982271e-01 -1.67462200e-01 -1.93547517e-01 1.03754628e+00
-2.31861666e-01 2.92297691e-01 -7.33477354e-01 -4.36326787e-02
2.07349494e-01 1.26987493e+00 -9.88312066e-01 -2.45748341e-01
-4.21915084e-01 1.61220014e+00 8.67109001e-02 8.77779648e-02
-9.24847126e-01 -7.46918544e-02 5.90490639e-01 2.20266834e-01
-1.12669319e-02 -1.51327029e-01 -5.81643343e-01 -1.14982677e+00
3.49865705e-02 -1.04110777e+00 2.25355551e-01 -1.29099309e+00
-1.14928925e+00 8.27062905e-01 -1.12116963e-01 -1.48131454e+00
-1.67102233e-01 -3.97740930e-01 -4.09608275e-01 7.16237068e-01
-1.38928950e+00 -1.53045750e+00 -5.38354695e-01 6.80906653e-01
5.00367224e-01 -8.41771951e-04 9.13187861e-01 3.14723074e-01
3.46940309e-02 5.93919098e-01 -2.83151045e-02 -1.96983442e-01
1.01388657e+00 -9.31062222e-01 5.73308289e-01 7.62071073e-01
5.55337369e-01 3.31314504e-01 1.00066209e+00 -3.46502155e-01
-1.14725125e+00 -9.32561338e-01 1.05513370e+00 -7.79060125e-01
5.38313150e-01 -1.61312789e-01 -8.08758318e-01 4.47348177e-01
4.47402686e-01 -4.84891869e-02 6.27458990e-01 -3.40760946e-01
-4.88532454e-01 1.07371800e-01 -1.17145419e+00 7.83604443e-01
1.23988998e+00 -6.44545257e-01 -5.13010919e-01 3.97262901e-01
4.93204534e-01 -3.95762950e-01 -9.89187419e-01 3.60620588e-01
6.46704793e-01 -1.26344979e+00 1.15075791e+00 -3.71643752e-01
9.77573991e-01 -2.84201801e-01 -2.31012315e-01 -1.35825455e+00
-4.83593494e-01 -4.53939080e-01 4.03552413e-01 1.08490658e+00
5.16310811e-01 -5.82460523e-01 6.36672080e-01 3.54209125e-01
-5.38858660e-02 -6.01719260e-01 -8.83110464e-01 -8.35677743e-01
2.37164885e-01 -2.39794940e-01 4.81424749e-01 1.13937986e+00
-9.60222334e-02 3.78260016e-01 -3.96589249e-01 -4.99956578e-01
4.62412894e-01 1.55583993e-01 5.95881581e-01 -8.06467175e-01
-1.63659766e-01 -8.68799806e-01 -7.30110824e-01 -5.75101078e-01
5.99090010e-02 -1.22598493e+00 -1.20708622e-01 -1.61624944e+00
4.64403450e-01 -1.92933396e-01 -2.42851734e-01 3.78658623e-01
-1.01406336e-01 1.27662396e+00 4.00190532e-01 4.71778095e-01
-4.69605356e-01 6.14181042e-01 1.58725214e+00 -1.75867364e-01
1.88914463e-01 -1.87692136e-01 -8.24817836e-01 5.17323375e-01
8.95350277e-01 -3.51350635e-01 -4.61608827e-01 -3.76927584e-01
2.17560321e-01 -2.06714272e-01 4.98068124e-01 -9.42436874e-01
1.51785677e-02 -2.14932948e-01 3.67652625e-01 -3.75576280e-02
4.40758646e-01 -7.27022827e-01 5.28726697e-01 5.66776335e-01
-5.27708828e-01 2.80710697e-01 1.78135186e-02 8.52276236e-02
-4.95988190e-01 -4.05600406e-02 1.04231346e+00 -2.31081694e-01
-8.94904494e-01 2.63162076e-01 1.25756785e-02 2.25611508e-01
9.80682969e-01 -4.63351488e-01 -2.29671836e-01 -5.62547684e-01
-5.57830811e-01 -2.21386671e-01 1.08569014e+00 8.21336925e-01
7.11717248e-01 -1.61120808e+00 -9.57029760e-01 3.10035776e-02
5.63739777e-01 -5.82184315e-01 1.87169984e-01 1.00450194e+00
-7.71632731e-01 2.08177552e-01 -7.43245780e-01 -7.45029926e-01
-1.32485008e+00 3.44750792e-01 2.25588620e-01 1.02731377e-01
-5.49722910e-01 7.51747549e-01 1.21053346e-01 -3.46415460e-01
-2.95349419e-01 -4.39846627e-02 1.59313649e-01 -9.10821855e-02
4.40716803e-01 2.10643381e-01 1.93535015e-01 -9.76406157e-01
-3.35772634e-01 7.39826083e-01 1.16505422e-01 -5.09616733e-01
1.19250166e+00 -1.78349838e-01 -2.04418540e-01 3.35355550e-01
1.25070369e+00 -2.64580816e-01 -1.07391667e+00 -1.41615912e-01
-1.90424815e-01 -7.82667994e-01 -5.07605635e-02 -1.05856407e+00
-9.11964536e-01 8.18022609e-01 7.87760615e-01 -1.07821360e-01
1.27542472e+00 2.35370234e-01 9.51479197e-01 -1.87827542e-01
5.41885734e-01 -8.83848071e-01 2.77878731e-01 1.79541662e-01
1.18220758e+00 -1.41914701e+00 -1.37169749e-01 -3.03164691e-01
-8.88233721e-01 8.32734227e-01 2.79392481e-01 8.47120062e-02
1.67792201e-01 -6.43625297e-03 2.54405946e-01 -4.35688756e-02
-4.44199353e-01 -3.28013808e-01 5.15308797e-01 8.14700544e-01
6.74497604e-01 2.15393901e-01 -5.63481212e-01 -1.32186979e-01
-4.13765728e-01 1.17689274e-01 3.58651102e-01 6.19662642e-01
-1.97684407e-01 -1.42165613e+00 -2.78368592e-01 2.27278888e-01
-3.40403318e-01 -1.92622796e-01 -6.35009646e-01 5.67775905e-01
1.05492204e-01 7.43294895e-01 -1.55754089e-01 -3.75966340e-01
3.30754161e-01 -1.76061049e-01 9.39439654e-01 -3.35923910e-01
-4.70431834e-01 -2.65377223e-01 6.85352180e-03 -6.47545099e-01
-5.77503145e-01 -7.06486225e-01 -8.85070801e-01 -5.93845963e-01
1.87529877e-01 -1.02829330e-01 8.34757626e-01 5.72265506e-01
3.61993045e-01 1.87323779e-01 5.05453110e-01 -8.32264483e-01
-1.87596038e-01 -7.55509973e-01 -2.63681710e-01 8.79330158e-01
-4.09485400e-02 -4.87296760e-01 7.44727161e-03 2.71134704e-01] | [11.494417190551758, -0.024170823395252228] |
3770f5c3-da93-4801-9a08-7d8ef7606646 | sequential-diagnosis-by-abstraction | 1401.3892 | null | http://arxiv.org/abs/1401.3892v1 | http://arxiv.org/pdf/1401.3892v1.pdf | Sequential Diagnosis by Abstraction | When a system behaves abnormally, sequential diagnosis takes a sequence of
measurements of the system until the faults causing the abnormality are
identified, and the goal is to reduce the diagnostic cost, defined here as the
number of measurements. To propose measurement points, previous work employs a
heuristic based on reducing the entropy over a computed set of diagnoses. This
approach generally has good performance in terms of diagnostic cost, but can
fail to diagnose large systems when the set of diagnoses is too large. Focusing
on a smaller set of probable diagnoses scales the approach but generally leads
to increased average diagnostic costs. In this paper, we propose a new
diagnostic framework employing four new techniques, which scales to much larger
systems with good performance in terms of diagnostic cost. First, we propose a
new heuristic for measurement point selection that can be computed efficiently,
without requiring the set of diagnoses, once the system is modeled as a
Bayesian network and compiled into a logical form known as d-DNNF. Second, we
extend hierarchical diagnosis, a technique based on system abstraction from our
previous work, to handle probabilities so that it can be applied to sequential
diagnosis to allow larger systems to be diagnosed. Third, for the largest
systems where even hierarchical diagnosis fails, we propose a novel method that
converts the system into one that has a smaller abstraction and whose diagnoses
form a superset of those of the original system; the new system can then be
diagnosed and the result mapped back to the original system. Finally, we
propose a novel cost estimation function which can be used to choose an
abstraction of the system that is more likely to provide optimal average cost.
Experiments with ISCAS-85 benchmark circuits indicate that our approach scales
to all circuits in the suite except one that has a flat structure not
susceptible to useful abstraction. | ['Sajjad Ahmed Siddiqi', 'Jinbo Huang'] | 2014-01-16 | null | null | null | null | ['sequential-diagnosis'] | ['medical'] | [ 3.26491147e-01 5.83190024e-01 -1.35598570e-01 -1.60480246e-01
-6.46200001e-01 -4.13912535e-01 2.14710355e-01 2.03098044e-01
4.31982666e-01 5.47771811e-01 -4.26123977e-01 -6.93507195e-01
-5.63904881e-01 -1.09185064e+00 -3.52698833e-01 -5.67863762e-01
-1.73480242e-01 9.48651552e-01 8.44721258e-01 5.10823391e-02
1.61971003e-01 7.13623405e-01 -1.60567689e+00 9.92534235e-02
8.35027695e-01 7.81000793e-01 -1.12469904e-01 6.77523255e-01
2.05303311e-01 7.76600540e-01 -8.05888116e-01 -1.55255683e-02
2.32594684e-01 -8.73893023e-01 -1.18423533e+00 2.09972069e-01
-1.02260970e-01 -4.47443515e-01 3.36344764e-02 1.31134939e+00
7.16272444e-02 -4.83731180e-01 5.79035282e-01 -1.46278906e+00
-1.14265032e-01 1.11369514e+00 -2.60269016e-01 -1.17862128e-01
5.65978825e-01 -9.24616084e-02 8.13169658e-01 -1.52020723e-01
3.18614483e-01 1.44699919e+00 5.03181636e-01 5.07218003e-01
-1.62323070e+00 -5.32946944e-01 1.16590470e-01 1.82204545e-01
-1.67942560e+00 -1.79823160e-01 3.33596796e-01 -2.86554635e-01
1.30212069e+00 5.50429881e-01 7.68496275e-01 3.91628385e-01
4.66137081e-01 1.87701166e-01 1.01223052e+00 -6.42163813e-01
8.64331603e-01 -3.15794609e-02 3.47180665e-01 8.59884679e-01
6.99029565e-01 7.55567774e-02 1.62920892e-01 -6.33955717e-01
3.48525405e-01 1.18309654e-01 -2.13849649e-01 -4.13939029e-01
-8.70267272e-01 5.09252191e-01 -7.46561885e-02 5.39482117e-01
-2.42642954e-01 3.17068517e-01 1.50741860e-01 5.67685187e-01
4.18920536e-03 7.35799313e-01 -4.74484682e-01 -2.40273867e-02
-9.66894925e-01 3.12897205e-01 1.31994617e+00 7.92299390e-01
6.77824199e-01 -4.70030308e-02 1.50397541e-02 2.41154492e-01
2.87605554e-01 6.39590859e-01 1.03707209e-01 -1.01231933e+00
-7.64943138e-02 1.02808857e+00 -1.36192620e-01 -5.11658251e-01
-4.54391867e-01 -1.51064515e-01 -4.71946567e-01 4.88437265e-01
1.66840315e-01 5.80696575e-02 -9.62264776e-01 1.69990838e+00
9.30824801e-02 7.77734742e-02 2.10284844e-01 4.23503548e-01
-1.07088592e-02 6.01642728e-01 -5.77125132e-01 -2.87163466e-01
1.13778305e+00 -3.15439045e-01 -4.13002551e-01 5.03171124e-02
8.38810325e-01 -3.43874156e-01 4.40530717e-01 9.64825511e-01
-1.24013531e+00 -3.54948156e-02 -1.50992560e+00 6.23212636e-01
-1.45977080e-01 -1.45468324e-01 3.77173692e-01 9.15462494e-01
-1.50528884e+00 6.83701634e-01 -1.01431608e+00 -4.81317163e-01
-1.37647465e-01 7.03591108e-01 1.07482977e-01 -2.94167250e-01
-9.89553332e-01 1.20380986e+00 6.09121680e-01 -1.21787667e-01
-1.19248688e+00 -2.87162095e-01 -7.64652431e-01 4.70460474e-01
6.76778138e-01 -7.41201222e-01 1.26016295e+00 -7.23376036e-01
-1.48539364e+00 4.57334444e-02 1.89877488e-02 -4.72490609e-01
1.65509611e-01 5.97550809e-01 -5.42918682e-01 2.02291504e-01
-5.92106208e-02 3.40198278e-01 5.45858860e-01 -1.18457603e+00
-9.15458500e-01 -2.14380622e-01 5.30802608e-01 -4.34026003e-01
-2.80885816e-01 -1.96123123e-01 -1.26071095e-01 1.75942004e-01
5.54606557e-01 -1.20206094e+00 -4.65762287e-01 -4.98031437e-01
-7.87892401e-01 -1.87649369e-01 6.22841597e-01 -2.54597634e-01
1.46447027e+00 -1.85669506e+00 5.17970741e-01 9.66181576e-01
3.86529297e-01 -1.01579040e-01 1.99683636e-01 5.44137299e-01
-3.31277996e-01 3.06874156e-01 -5.04298031e-01 -2.65244376e-02
1.46692812e-01 3.36475223e-01 -2.88209990e-02 4.36686516e-01
3.66392404e-01 2.84846783e-01 -8.66780937e-01 -3.69724661e-01
9.36683118e-02 -1.51964754e-01 -6.73095763e-01 2.42943726e-02
-1.08666718e-01 -1.33980289e-01 -5.07516623e-01 6.36115789e-01
4.85184103e-01 -1.53406322e-01 3.98615569e-01 1.66130349e-01
1.46538481e-01 3.34605366e-01 -1.43159485e+00 9.40837979e-01
-5.07378697e-01 1.25124156e-01 -2.54635662e-01 -1.06123447e+00
9.91330028e-01 5.28191090e-01 4.81398374e-01 -1.29110307e-01
1.80558518e-01 3.95582408e-01 4.35826719e-01 -2.85143286e-01
1.05498269e-01 -6.80222735e-02 -3.80117655e-01 5.63594997e-01
-5.70507273e-02 -6.99854076e-01 3.60971898e-01 2.31671259e-01
2.00718069e+00 -5.46080172e-01 4.36139435e-01 -5.58885634e-01
5.09812236e-01 1.36631271e-02 6.99157894e-01 6.73842192e-01
2.59842962e-01 1.34582013e-01 1.19745135e+00 1.50697050e-03
-9.95645225e-01 -1.18304873e+00 -2.04493508e-01 6.24638423e-02
8.90168101e-02 -5.74497700e-01 -9.40499425e-01 -8.79339218e-01
7.12086074e-03 1.01271844e+00 -4.49555665e-01 -6.61911845e-01
-1.16579451e-01 -4.79415685e-01 5.29068649e-01 4.46540236e-01
6.96987659e-02 -6.78391755e-01 -6.87111974e-01 1.56825095e-01
3.88425648e-01 -6.96633995e-01 2.68105149e-01 4.55976427e-01
-1.01864791e+00 -1.13896716e+00 -8.91758874e-02 -4.21677023e-01
9.96254861e-01 -2.77947992e-01 8.52159560e-01 3.56981546e-01
-1.92886725e-01 1.88436106e-01 -3.46203089e-01 -1.53897852e-01
-9.93230343e-01 -9.84621644e-02 9.91286784e-02 -4.49586600e-01
2.35669002e-01 -5.34459829e-01 1.02769956e-01 2.66032994e-01
-1.11140382e+00 -3.34918976e-01 5.38825989e-01 8.91893804e-01
3.08021188e-01 9.59441364e-01 5.81553757e-01 -1.05808556e+00
5.13076544e-01 -4.41413730e-01 -8.96759748e-01 2.57484555e-01
-1.01968598e+00 4.16863292e-01 8.98841858e-01 -3.15917313e-01
-6.80729032e-01 1.15459733e-01 -4.18648347e-02 -2.81777412e-01
-7.18966648e-02 7.15150118e-01 -4.92900044e-01 -2.58577280e-02
6.06801510e-01 -2.20539212e-01 -2.91253388e-01 -8.65594894e-02
-4.97912942e-03 6.33865893e-01 1.16331592e-01 -4.57486480e-01
6.69343591e-01 1.93522617e-01 5.47448337e-01 -2.66693383e-01
-3.21635872e-01 -3.75226438e-02 -4.05780226e-01 -1.71516072e-02
2.52346903e-01 -2.32580572e-01 -8.29774857e-01 1.30621538e-01
-8.41816425e-01 -1.68450713e-01 -4.19834048e-01 3.59245509e-01
-4.12012309e-01 1.75773084e-01 -5.95413744e-01 -9.17414725e-01
1.17811449e-01 -1.56821573e+00 9.75741029e-01 -1.46553755e-01
-6.64523542e-01 -8.00131619e-01 5.98907471e-03 -4.05291945e-01
2.12389410e-01 1.32625684e-01 1.42886174e+00 -7.63737261e-01
-5.24062574e-01 -6.58054292e-01 8.37428942e-02 4.96548295e-01
2.09167480e-01 3.96113127e-01 -6.58899009e-01 -3.91285509e-01
2.76733130e-01 1.46792740e-01 4.49122488e-01 2.38705069e-01
1.15696216e+00 -1.17320985e-01 -7.92816579e-01 1.04388567e-02
1.43675470e+00 5.89692712e-01 7.00099945e-01 -3.36052217e-02
4.50806797e-01 5.45898080e-01 6.04426801e-01 3.04549277e-01
2.13026851e-01 5.27093530e-01 4.54333752e-01 1.40227571e-01
1.64229646e-01 2.47812122e-01 3.96725684e-01 7.80814171e-01
2.60284871e-01 -1.57214135e-01 -1.15823495e+00 6.79554880e-01
-1.76507998e+00 -7.69650996e-01 3.47599358e-04 2.41810274e+00
6.62272811e-01 6.74994588e-01 -2.83118873e-03 8.25679004e-01
5.94049215e-01 -6.81697905e-01 -4.65483397e-01 -8.19442987e-01
5.96231461e-01 3.80786657e-01 6.54093444e-01 7.61111796e-01
-3.85230988e-01 3.21980804e-01 6.90984869e+00 6.36773765e-01
-1.00561893e+00 1.52955912e-02 3.98005784e-01 6.64794147e-02
-5.07259429e-01 5.51318645e-01 -7.78786421e-01 4.11602050e-01
1.62459898e+00 -4.47014660e-01 2.44864389e-01 8.83086205e-01
-4.36168835e-02 -5.04761696e-01 -1.90147221e+00 4.03572291e-01
-8.59641954e-02 -7.52002001e-01 -2.82041162e-01 2.03064352e-01
6.74138665e-01 -6.02289438e-01 -2.35495850e-01 2.47432873e-01
7.18167961e-01 -8.23999286e-01 7.78928399e-01 4.33241457e-01
6.31148100e-01 -1.08962703e+00 1.13627076e+00 3.86270463e-01
-1.02528417e+00 -4.33422744e-01 -1.23022065e-01 -1.32175431e-01
7.41065964e-02 1.05143070e+00 -1.22077239e+00 5.50552666e-01
4.97989148e-01 3.60075027e-01 -4.80848491e-01 1.04004204e+00
-3.44726771e-01 6.27836287e-01 -4.85203922e-01 -1.37902126e-01
-1.68161049e-01 2.05877364e-01 4.95068789e-01 8.67523253e-01
7.86687016e-01 -5.20527028e-02 4.25574720e-01 1.04183447e+00
4.69562918e-01 -5.51616728e-01 -7.14051366e-01 1.37420326e-01
8.55215549e-01 9.19509172e-01 -1.08533144e+00 -7.31739163e-01
-5.72882220e-02 6.10283375e-01 -1.83276590e-02 6.45317584e-02
-8.88603389e-01 -4.13432419e-01 8.09941664e-02 3.64009961e-02
1.63726449e-01 2.60594934e-01 -2.50325143e-01 -7.50889242e-01
-1.82789954e-04 -9.33273613e-01 1.32977679e-01 -5.04299402e-01
-9.82101679e-01 6.70883298e-01 5.96373856e-01 -1.19280255e+00
-8.52969766e-01 -3.03797871e-01 -3.37969512e-01 8.55856717e-01
-7.66947806e-01 -5.60832679e-01 6.43104687e-02 2.97593623e-01
1.76399108e-02 1.90241128e-01 1.06951344e+00 3.03956240e-01
-5.68462431e-01 4.49414253e-01 -1.94339067e-01 -6.15499973e-01
3.69338900e-01 -1.54135180e+00 2.27644011e-01 1.13322163e+00
-5.41815162e-01 4.60599989e-01 9.17433143e-01 -8.14675987e-01
-1.35219181e+00 -1.02296865e+00 8.95070732e-01 -1.66550711e-01
6.86658442e-01 -1.90985650e-01 -7.10466325e-01 7.37749815e-01
-1.62047550e-01 -3.11956286e-01 2.13028908e-01 1.91618443e-01
-1.09182633e-01 -4.05315876e-01 -1.64017320e+00 6.21418059e-01
6.72750771e-01 -4.24229056e-01 -5.05714238e-01 2.93442458e-01
6.29725277e-01 -2.96633780e-01 -1.15232301e+00 4.43749875e-01
1.55041337e-01 -9.73439634e-01 3.91117632e-01 1.30849838e-01
1.85425937e-01 -6.56350076e-01 -2.19873395e-02 -1.47447658e+00
-4.77651834e-01 -3.49143296e-01 -3.00138474e-01 1.11687899e+00
7.05140352e-01 -1.16370702e+00 4.58438665e-01 4.73543942e-01
-3.53189021e-01 -9.52097774e-01 -9.67184901e-01 -9.82665300e-01
-2.30199546e-01 -2.76082933e-01 1.06173468e+00 6.67922676e-01
5.79004109e-01 1.07752480e-01 2.82555491e-01 5.48441470e-01
4.65099424e-01 1.06422514e-01 2.79476017e-01 -1.47135472e+00
-7.31533825e-01 -4.49619740e-01 -8.35731208e-01 -4.31365520e-01
8.90803114e-02 -7.22416103e-01 4.59955245e-01 -1.44963920e+00
2.40622535e-01 -5.89680135e-01 -2.56504774e-01 7.92121649e-01
2.20326215e-01 -1.84890181e-01 -1.17843084e-01 7.28330854e-03
-3.58560950e-01 -2.06614435e-01 7.46894538e-01 -1.75995409e-01
-1.92055315e-01 1.26779929e-01 -5.85407674e-01 6.72636807e-01
5.69240212e-01 -7.34874666e-01 -5.79382539e-01 2.08223149e-01
2.68429250e-01 5.16132712e-01 1.99449986e-01 -1.40325582e+00
1.84333041e-01 -3.56440768e-02 3.64186764e-02 -4.52915460e-01
1.71035588e-01 -1.27674675e+00 6.71077907e-01 9.84067202e-01
-2.39368230e-01 2.05826625e-01 5.32760248e-02 1.70238718e-01
-1.46064624e-01 -7.93693364e-01 6.68006182e-01 2.58118004e-01
-3.67612660e-01 -1.35307685e-01 -7.63779521e-01 -6.30655646e-01
1.39130032e+00 -1.35033950e-01 -1.65923029e-01 -1.03522703e-01
-8.67385387e-01 3.19332987e-01 8.38145912e-01 -1.31501272e-01
4.86545175e-01 -1.04991341e+00 -3.11877429e-01 8.69610831e-02
5.35724573e-02 7.38938451e-02 -7.88055733e-02 9.27876949e-01
-5.04481494e-01 4.37295467e-01 4.89937216e-02 -8.42381060e-01
-1.15374649e+00 6.89630210e-01 4.17315096e-01 -4.19972211e-01
-5.15415132e-01 4.90048110e-01 -1.54326245e-01 -8.18103254e-02
-4.73116189e-02 -1.07362032e+00 1.19432576e-01 -3.00165415e-01
4.26637858e-01 3.79731536e-01 4.44644153e-01 1.42198876e-02
-4.29230571e-01 3.05671811e-01 -1.72212236e-02 -1.81840897e-01
1.12728381e+00 -1.94886364e-02 -6.66996837e-01 6.17887259e-01
8.20037067e-01 -1.60773769e-01 -7.16726959e-01 2.28065073e-01
1.50614873e-01 -2.29592219e-01 2.23390102e-01 -6.93598032e-01
-1.04773378e+00 4.34093416e-01 3.91627520e-01 1.04684961e+00
1.61714053e+00 9.86670107e-02 3.93247455e-01 5.32028377e-01
8.92830670e-01 -6.94671035e-01 -4.49337691e-01 4.14720684e-01
3.02617759e-01 -5.10036945e-01 1.24806918e-01 -5.21169662e-01
-5.01594432e-02 1.21716011e+00 4.10861731e-01 -2.90824741e-01
6.97526574e-01 1.06048059e+00 -5.57205260e-01 -2.82421410e-01
-9.51916635e-01 3.63823362e-02 -1.11293823e-01 4.22547877e-01
-1.04520500e-01 3.54077578e-01 -3.41933042e-01 4.70273405e-01
-2.21438155e-01 2.02015564e-01 1.00585699e+00 1.07097924e+00
-6.09639943e-01 -1.35344708e+00 -6.26276672e-01 7.79339492e-01
-1.63792565e-01 1.56139418e-01 -3.76340508e-01 6.99431300e-01
2.97343075e-01 1.03414834e+00 1.48713693e-03 -6.00002706e-01
4.54441160e-01 -3.99343148e-02 7.13201582e-01 -1.04642785e+00
-3.26245278e-01 -6.73608705e-02 2.96127975e-01 -7.42837846e-01
-7.09710643e-02 -7.03820765e-01 -1.43501401e+00 -4.30230945e-01
-7.60011494e-01 3.03766042e-01 3.63210410e-01 1.23272491e+00
1.13072999e-01 1.02960670e+00 6.74541175e-01 -5.60999095e-01
-6.19639456e-01 -7.95771480e-01 -9.99047935e-01 -1.31554872e-01
1.88052788e-01 -7.83511758e-01 -4.37205791e-01 -2.47782797e-01] | [5.375762462615967, 2.7212107181549072] |
903b602f-6622-44d5-bb7f-b151a21307f7 | matrix-recovery-using-split-bregman | 1312.6872 | null | http://arxiv.org/abs/1312.6872v1 | http://arxiv.org/pdf/1312.6872v1.pdf | Matrix recovery using Split Bregman | In this paper we address the problem of recovering a matrix, with inherent
low rank structure, from its lower dimensional projections. This problem is
frequently encountered in wide range of areas including pattern recognition,
wireless sensor networks, control systems, recommender systems, image/video
reconstruction etc. Both in theory and practice, the most optimal way to solve
the low rank matrix recovery problem is via nuclear norm minimization. In this
paper, we propose a Split Bregman algorithm for nuclear norm minimization. The
use of Bregman technique improves the convergence speed of our algorithm and
gives a higher success rate. Also, the accuracy of reconstruction is much
better even for cases where small number of linear measurements are available.
Our claim is supported by empirical results obtained using our algorithm and
its comparison to other existing methods for matrix recovery. The algorithms
are compared on the basis of NMSE, execution time and success rate for varying
ranks and sampling ratios. | ['Ankita Shukla', 'Anupriya Gogna', 'Angshul Majumdar'] | 2013-12-17 | null | null | null | null | ['video-reconstruction'] | ['computer-vision'] | [ 6.92969024e-01 -2.57799119e-01 -3.37203592e-02 -1.58711020e-02
-6.10729039e-01 -5.22605658e-01 4.06886965e-01 -6.01012073e-02
-5.41598856e-01 1.02366352e+00 4.73083735e-01 -7.42858574e-02
-7.37235188e-01 -6.66801691e-01 -4.30059642e-01 -9.04760957e-01
-2.25205317e-01 3.89355958e-01 -3.52075659e-02 -1.02483958e-01
3.97897720e-01 4.22326654e-01 -1.18708432e+00 -1.51254490e-01
6.51875198e-01 8.42264593e-01 2.22629219e-01 6.45166457e-01
2.92881012e-01 7.26730287e-01 -4.96466272e-02 -6.75718859e-02
7.17206061e-01 -3.04854453e-01 -5.36228955e-01 2.76750535e-01
5.29435389e-02 -6.36938736e-02 -1.92810103e-01 1.36779416e+00
4.25949425e-01 3.68100166e-01 8.10508311e-01 -8.29792440e-01
-7.99648985e-02 4.98209000e-01 -1.08569586e+00 8.76468197e-02
5.38345575e-01 -4.72799212e-01 8.52372348e-01 -1.13669550e+00
4.59110349e-01 1.25866234e+00 6.96529627e-01 2.52005216e-02
-1.36495590e+00 -7.87117839e-01 -2.40865603e-01 1.44716382e-01
-1.67414320e+00 -5.84471107e-01 5.44691384e-01 -3.30597371e-01
3.38732958e-01 5.18401206e-01 2.40628913e-01 5.87982237e-01
-1.03946760e-01 4.40576226e-01 1.46266603e+00 -4.25787002e-01
1.30310923e-01 -1.68955058e-01 1.31937131e-01 4.61121976e-01
7.93400288e-01 1.18161492e-01 -4.01463419e-01 -4.75900084e-01
6.08020365e-01 1.87251717e-01 -4.80074018e-01 -5.37186265e-01
-1.45973408e+00 7.99910367e-01 1.93869770e-02 4.18696612e-01
-6.42810822e-01 1.37764374e-02 9.38894153e-02 4.65963781e-01
1.67098716e-01 1.26466125e-01 -6.67283759e-02 9.45977345e-02
-1.11434948e+00 6.43373430e-02 1.09889376e+00 7.67653763e-01
5.48771262e-01 -1.46603426e-02 2.12416291e-01 8.26603830e-01
4.75081652e-01 7.29990602e-01 3.26579809e-01 -8.93123806e-01
4.75509018e-01 7.85590261e-02 1.91691861e-01 -1.39266336e+00
-3.78121823e-01 -4.10299480e-01 -1.42842364e+00 3.62554193e-02
2.92364299e-01 -1.02030456e-01 -5.12482584e-01 1.44173265e+00
4.86412287e-01 6.34862125e-01 4.13529277e-01 1.05773127e+00
2.74301231e-01 7.29102254e-01 -5.48676014e-01 -8.64337921e-01
1.03398860e+00 -4.86146390e-01 -8.55114877e-01 8.60443637e-02
2.85081774e-01 -1.20201719e+00 3.81467313e-01 8.91663253e-01
-8.54726553e-01 -1.51634842e-01 -1.17178118e+00 3.94565761e-01
2.15190291e-01 2.59844810e-01 5.56813478e-01 7.54936814e-01
-6.97922349e-01 7.41978168e-01 -6.36742771e-01 -5.35679936e-01
-1.43056944e-01 7.16154754e-01 -6.47589326e-01 -3.41176838e-01
-7.11577713e-01 6.47733927e-01 3.73556882e-01 2.51802981e-01
-4.66670126e-01 -5.80659993e-02 -4.18424219e-01 -2.37797320e-01
4.19906288e-01 -3.27248007e-01 7.57593095e-01 -4.32239205e-01
-1.42211580e+00 3.68377924e-01 -1.97848320e-01 -3.83824646e-01
4.02930558e-01 -1.76578522e-01 -2.65105903e-01 -6.54642284e-02
-2.21952200e-02 -1.91902831e-01 9.92158890e-01 -9.00432110e-01
-2.90294945e-01 -5.42602003e-01 -2.17878625e-01 2.65079409e-01
-1.08855806e-01 -3.63184154e-01 -6.04112633e-02 -4.19761837e-01
8.22480798e-01 -1.22336185e+00 -6.41334951e-01 -3.59359413e-01
-3.89399737e-01 2.77346700e-01 6.90127015e-01 -6.20812535e-01
1.14034057e+00 -2.11835623e+00 5.17166197e-01 9.09406900e-01
1.81895316e-01 5.80647476e-02 5.11404127e-02 8.19809794e-01
-7.42311999e-02 -2.75886923e-01 -4.01021749e-01 1.16074957e-01
-3.18794519e-01 1.27266765e-01 -1.09503135e-01 1.03066349e+00
-4.82847601e-01 6.97617829e-02 -6.57485604e-01 -2.60632694e-01
2.49220148e-01 3.96337330e-01 -4.60930705e-01 -6.37333244e-02
4.25798833e-01 3.97342801e-01 -3.96235466e-01 4.17849243e-01
8.53664994e-01 -3.93939048e-01 6.54229224e-01 -6.93595469e-01
-6.10497221e-02 -2.75959820e-01 -2.14983439e+00 1.35357273e+00
-2.95009822e-01 4.75773364e-01 5.85729420e-01 -1.26483870e+00
8.65669250e-01 4.49048907e-01 7.39723623e-01 -4.23221827e-01
1.96588740e-01 3.78380030e-01 1.16088197e-01 -4.42522794e-01
5.25877893e-01 -3.72913778e-01 2.31803685e-01 5.62971771e-01
-3.81789237e-01 2.52586365e-01 5.00078440e-01 1.95958883e-01
1.21435332e+00 -3.42866927e-01 8.20695102e-01 -5.78321934e-01
8.71579766e-01 -3.69378865e-01 4.96426374e-01 4.52353090e-01
1.60566285e-01 3.91298681e-01 1.69356093e-01 -6.09854832e-02
-1.04116905e+00 -7.33273327e-01 -2.27032349e-01 6.04932129e-01
1.42619923e-01 -2.75693357e-01 -3.24545652e-01 -2.92099249e-02
-8.57031122e-02 -9.57757514e-03 -3.78656387e-01 2.13846892e-01
-5.66947758e-01 -1.07672215e+00 1.83712602e-01 1.13292672e-02
5.29493093e-01 -4.37732875e-01 -2.39820272e-01 2.88975328e-01
-4.20052037e-02 -1.21375823e+00 -2.88598359e-01 -6.22177534e-02
-1.27939093e+00 -1.38531268e+00 -6.62929058e-01 -4.93568629e-01
8.28974545e-01 7.43578196e-01 6.25377476e-01 -1.01650683e-02
-1.01849429e-01 3.06622326e-01 -4.26785082e-01 -1.91813391e-02
-4.46643569e-02 -2.66075790e-01 5.73954940e-01 1.74496457e-01
1.94224659e-02 -8.63644302e-01 -5.30649960e-01 2.98210979e-01
-1.07893586e+00 -2.68373728e-01 1.01165211e+00 8.67367387e-01
6.90112174e-01 3.32017273e-01 3.24916720e-01 -1.25370073e+00
9.81354117e-01 -5.09747326e-01 -8.47764909e-01 3.18703428e-02
-7.76580334e-01 2.99663991e-01 4.09757286e-01 -4.67173696e-01
-5.93128145e-01 4.41722542e-01 1.23054720e-01 -8.73965472e-02
3.37595910e-01 7.55368173e-01 4.11891937e-03 -3.72691333e-01
8.19408059e-01 3.10416460e-01 2.25320399e-01 -6.55440032e-01
2.54416555e-01 5.39261460e-01 4.84480977e-01 -4.47388411e-01
1.09033847e+00 5.59925497e-01 7.15415478e-01 -1.29647565e+00
-7.59369552e-01 -8.20170283e-01 -4.31513548e-01 6.50565699e-03
4.12394822e-01 -8.20804119e-01 -7.26790726e-01 -1.02502458e-01
-6.79088652e-01 3.77569944e-01 1.41082898e-01 9.95482087e-01
-3.70384216e-01 7.92737722e-01 -4.01702255e-01 -1.03898144e+00
-4.87055898e-01 -1.04887509e+00 6.34223461e-01 -1.60058647e-01
-6.72953650e-02 -8.79031003e-01 2.95383275e-01 2.73120493e-01
2.94319719e-01 5.01016259e-01 5.46338677e-01 -3.87925178e-01
-5.01042426e-01 -2.63322055e-01 -3.87982786e-01 3.18630725e-01
1.15641199e-01 -5.24169445e-01 -4.32642311e-01 -6.54493570e-01
2.69368887e-01 -1.70599446e-02 5.73092580e-01 3.45091641e-01
6.39868259e-01 -7.23414063e-01 -2.47834414e-01 4.28953677e-01
1.76680660e+00 -5.44864312e-03 5.16559005e-01 -5.49798366e-03
6.11962914e-01 4.62391675e-01 7.26886690e-01 7.16244340e-01
-1.61688432e-01 4.30005938e-01 2.92861044e-01 2.32397780e-01
1.94185749e-01 2.44054005e-01 1.91393018e-01 1.35796571e+00
-1.44135475e-01 -1.60100199e-02 -5.99217713e-01 2.82191455e-01
-2.10578442e+00 -1.04333341e+00 -5.17220140e-01 2.54486012e+00
5.17447829e-01 -2.44708925e-01 6.34372160e-02 6.02126122e-01
6.66294932e-01 2.07334012e-02 -2.32776314e-01 5.65578006e-02
-3.94984595e-02 1.28534436e-01 9.27824855e-01 7.94343710e-01
-7.65683651e-01 4.06785876e-01 6.75120735e+00 7.32205331e-01
-9.46837068e-01 1.25135630e-01 7.98940882e-02 -1.18071415e-01
5.31082563e-02 9.60393548e-02 -5.12056351e-01 2.30677933e-01
7.69081771e-01 -2.16375455e-01 7.01216161e-01 4.46196377e-01
2.79227346e-01 -4.21867341e-01 -9.26318347e-01 1.45087862e+00
1.32118389e-01 -1.05748415e+00 -2.49010280e-01 4.49877828e-01
9.37711537e-01 -1.65950544e-02 -2.13938013e-01 -2.31051251e-01
1.77530512e-01 -1.05540752e+00 4.73275818e-02 5.56244373e-01
5.13632178e-01 -9.21772182e-01 7.49118209e-01 5.85349977e-01
-1.12138760e+00 2.74297148e-02 -5.44762731e-01 -2.60804176e-01
1.53569341e-01 1.12732911e+00 -6.94423616e-01 5.44314981e-01
2.98942268e-01 5.71847975e-01 1.00860044e-01 1.22539425e+00
2.07100198e-01 5.04423618e-01 -7.77618051e-01 -1.33774159e-02
2.48384953e-01 -9.20295894e-01 7.10859537e-01 1.01961315e+00
6.91804290e-01 5.78189433e-01 3.15716505e-01 1.01638794e-01
3.75428400e-03 5.03816783e-01 -8.07143152e-01 6.45481236e-03
3.34323376e-01 1.15362895e+00 -6.06937051e-01 -2.13914603e-01
-2.87448585e-01 6.84017241e-01 -9.37474072e-02 3.69420469e-01
-1.81104124e-01 -2.64936507e-01 3.66531909e-01 2.30291694e-01
1.10777095e-01 -5.42150259e-01 -2.52662823e-02 -1.14400101e+00
-1.75859034e-02 -1.15891683e+00 2.90120691e-01 -1.42669529e-01
-1.02533436e+00 3.68570775e-01 1.49836928e-01 -1.31576693e+00
-2.88373232e-01 -4.91821826e-01 -2.76841342e-01 6.18896604e-01
-1.04781544e+00 -5.02326906e-01 -1.39380172e-01 6.79234684e-01
2.63553619e-01 -2.90071368e-01 6.46708369e-01 6.72751069e-01
-4.84797686e-01 3.72392684e-02 6.24515593e-01 -2.92664975e-01
5.00620842e-01 -9.15855646e-01 -4.85437512e-01 1.10017788e+00
2.31243670e-01 7.32564926e-01 1.18190837e+00 -6.60246491e-01
-1.73540211e+00 -6.87970340e-01 4.72991019e-01 1.72329307e-01
6.53161168e-01 1.98020358e-02 -5.66920102e-01 3.20566744e-01
3.69509049e-02 -7.03279898e-02 9.36346471e-01 2.45139986e-01
-1.12994455e-01 -2.99712181e-01 -1.29335523e+00 3.78110349e-01
7.34554946e-01 -3.70868683e-01 -2.46052787e-01 7.09645867e-01
-3.56797576e-02 -2.07243249e-01 -1.04412937e+00 5.69771707e-01
6.74914300e-01 -8.57576132e-01 1.02862477e+00 -9.49719697e-02
-1.51243135e-01 -7.51994491e-01 -7.84613431e-01 -1.00746882e+00
-4.77922350e-01 -8.04957271e-01 -2.03285277e-01 7.80138910e-01
2.12610126e-01 -3.53445500e-01 9.67365623e-01 1.53723761e-01
4.89051491e-01 -5.57076812e-01 -8.11829388e-01 -7.28397548e-01
-5.76124549e-01 -2.48053923e-01 1.36688143e-01 8.88129056e-01
-1.47412390e-01 6.83525205e-01 -1.07831585e+00 2.58509368e-01
1.28922307e+00 9.37262550e-02 7.59434760e-01 -1.39225507e+00
-6.23351991e-01 1.15012027e-01 -4.49099243e-01 -9.13870752e-01
-2.48336881e-01 -7.74487853e-01 -4.47281636e-02 -1.48386836e+00
2.87475944e-01 -4.11428541e-01 -2.36829713e-01 -4.82371673e-02
2.56103009e-01 5.76689422e-01 1.78585589e-01 4.42532420e-01
-5.57085514e-01 1.86350197e-01 9.12607789e-01 1.21915713e-01
-1.09968930e-01 2.24987283e-01 -5.40022969e-01 7.23204792e-01
7.63975561e-01 -6.07918561e-01 -5.82835734e-01 -2.73616999e-01
4.21681195e-01 3.81628484e-01 -2.61183500e-01 -1.21971714e+00
3.21012259e-01 -1.85339838e-01 1.88043714e-01 -6.22201800e-01
5.27991176e-01 -1.13413680e+00 7.32333660e-01 5.95619440e-01
-6.78607151e-02 2.07973152e-01 -3.95875275e-01 9.11438227e-01
-1.56712681e-01 -6.63739264e-01 9.16356981e-01 -1.84018090e-01
-4.43321645e-01 2.57397354e-01 -3.33410621e-01 -4.12499309e-01
8.36254597e-01 -2.48806626e-01 8.55281055e-02 -6.40620172e-01
-7.48669505e-01 -3.82090127e-03 3.58030021e-01 -6.77806213e-02
6.67840958e-01 -1.43769658e+00 -7.05316186e-01 5.45284860e-02
-1.16548188e-01 -3.34850878e-01 -1.64636523e-02 1.24691129e+00
-5.38559258e-01 3.31142455e-01 -7.57952780e-02 -3.97129476e-01
-1.48772252e+00 3.67527843e-01 -2.66178578e-01 -4.78670210e-01
-4.48632002e-01 2.53282338e-01 -1.62088111e-01 -1.88799113e-01
1.16428457e-01 4.56237383e-02 -1.57573745e-01 -5.28831519e-02
7.11696208e-01 8.65382016e-01 -1.65934548e-01 -9.12507236e-01
-1.90639004e-01 8.90768647e-01 5.97612709e-02 -4.52847987e-01
1.38797092e+00 -2.43762910e-01 -5.41471303e-01 3.43185931e-01
1.22845042e+00 4.68635321e-01 -6.64737523e-01 -4.86365527e-01
2.03617513e-01 -6.12005174e-01 1.99255407e-01 -1.45750031e-01
-9.53023553e-01 4.50462550e-01 7.49761105e-01 1.76041707e-01
1.06157339e+00 -5.58132529e-01 5.55688262e-01 9.45744991e-01
5.77799022e-01 -9.90417957e-01 1.83578767e-02 4.24740463e-01
7.47944236e-01 -1.22558880e+00 8.37268353e-01 -5.18577933e-01
-2.10811660e-01 9.57224250e-01 4.34225164e-02 -6.29054010e-01
7.94496715e-01 1.93339884e-01 -2.55076975e-01 -7.08941137e-03
-5.46051443e-01 -1.65425584e-01 3.28715831e-01 4.84510303e-01
5.84460437e-01 1.29121944e-01 -9.71853197e-01 -9.26271528e-02
-1.67233109e-01 -2.54021615e-01 8.60304713e-01 9.21588242e-01
-7.18920827e-01 -1.42430711e+00 -7.81651616e-01 8.09420407e-01
-7.32322097e-01 2.76872870e-02 -4.88168821e-02 4.26855803e-01
-3.71790886e-01 1.11070931e+00 -5.89820206e-01 -2.82903552e-01
1.32968485e-01 -3.89629185e-01 6.35246933e-01 -4.28666145e-01
-3.68479602e-02 4.49299842e-01 2.00177565e-01 -5.47668099e-01
-8.39257717e-01 -7.91415334e-01 -1.09184778e+00 -3.14549953e-01
-3.88015121e-01 6.17393732e-01 9.15955603e-01 8.06372046e-01
9.17665381e-03 5.67886867e-02 6.82395875e-01 -6.18692577e-01
-7.34623432e-01 -8.94993007e-01 -9.72654283e-01 2.74802774e-01
3.19363505e-01 -5.43277740e-01 -2.14874387e-01 -2.08449364e-01] | [7.0486159324646, 4.577120780944824] |
89d0d111-7b02-4337-854e-1a0f2b591630 | multi-task-temporal-shift-attention-networks | 2006.03790 | null | https://arxiv.org/abs/2006.03790v2 | https://arxiv.org/pdf/2006.03790v2.pdf | Multi-Task Temporal Shift Attention Networks for On-Device Contactless Vitals Measurement | Telehealth and remote health monitoring have become increasingly important during the SARS-CoV-2 pandemic and it is widely expected that this will have a lasting impact on healthcare practices. These tools can help reduce the risk of exposing patients and medical staff to infection, make healthcare services more accessible, and allow providers to see more patients. However, objective measurement of vital signs is challenging without direct contact with a patient. We present a video-based and on-device optical cardiopulmonary vital sign measurement approach. It leverages a novel multi-task temporal shift convolutional attention network (MTTS-CAN) and enables real-time cardiovascular and respiratory measurements on mobile platforms. We evaluate our system on an Advanced RISC Machine (ARM) CPU and achieve state-of-the-art accuracy while running at over 150 frames per second which enables real-time applications. Systematic experimentation on large benchmark datasets reveals that our approach leads to substantial (20%-50%) reductions in error and generalizes well across datasets. | ['Josh Fromm', 'Shwetak Patel', 'Xin Liu', 'Daniel McDuff'] | 2020-06-06 | null | http://proceedings.neurips.cc/paper/2020/hash/e1228be46de6a0234ac22ded31417bc7-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/e1228be46de6a0234ac22ded31417bc7-Paper.pdf | neurips-2020-12 | ['photoplethysmography-ppg-heart-rate'] | ['medical'] | [ 3.60485166e-01 -2.36337006e-01 1.05335504e-01 -3.34143102e-01
-9.30364490e-01 -5.21977901e-01 1.04181536e-01 4.68223244e-01
-6.72233462e-01 5.02096057e-01 2.94804275e-01 -7.57536948e-01
6.13477230e-02 -3.60914558e-01 -5.34827232e-01 -3.69567573e-01
-2.79301494e-01 3.13685328e-01 8.57944563e-02 2.43359923e-01
-2.35079259e-01 5.43943942e-01 -8.18743348e-01 4.27631348e-01
1.17563374e-01 1.14947808e+00 -3.41577321e-01 1.27421808e+00
7.03113317e-01 7.79839635e-01 -6.62376761e-01 1.90710723e-01
4.06647265e-01 -2.79887289e-01 -5.94360113e-01 -4.03060228e-01
3.50368589e-01 -8.78088176e-01 -1.34127378e-01 2.30855212e-01
1.02593541e+00 -2.44723573e-01 1.16256103e-01 -7.33988285e-01
-4.44826633e-02 -1.90491900e-01 -3.14458311e-01 7.98015893e-01
3.49919260e-01 5.49135029e-01 9.87914652e-02 -2.65757531e-01
2.17168614e-01 7.57098019e-01 1.11949587e+00 7.21796572e-01
-1.10568643e+00 -3.86812210e-01 -6.18721485e-01 -1.18671805e-01
-1.08611393e+00 -5.21484911e-01 -1.34076789e-01 -4.11353678e-01
1.28705978e+00 2.98476219e-01 7.96095371e-01 1.10080063e+00
8.29276085e-01 7.02073574e-02 9.74740624e-01 -7.35247089e-03
-4.28919680e-02 -7.68267065e-02 -2.47039739e-02 8.11190724e-01
3.87311667e-01 7.67723098e-02 -3.53373051e-01 -2.35969052e-01
7.47529447e-01 6.77886665e-01 -7.39677668e-01 1.42591879e-01
-1.58074617e+00 4.43155855e-01 1.07440777e-01 2.48343512e-01
-8.25527370e-01 3.52885544e-01 6.65220976e-01 9.15554166e-02
1.80765882e-01 3.57530534e-01 -7.42133439e-01 -6.74151361e-01
-6.33718371e-01 -2.25907922e-01 5.26768327e-01 3.72410864e-01
-5.43472916e-02 -1.04356073e-01 -5.52535236e-01 2.59157680e-02
4.10996983e-03 1.04177701e+00 2.62555748e-01 -1.11949575e+00
1.96967095e-01 2.07484171e-01 3.95789683e-01 -6.59353673e-01
-9.22632039e-01 -3.26172411e-01 -8.67194474e-01 4.39197384e-02
1.93965361e-01 -7.51058877e-01 -7.65087366e-01 1.33779359e+00
4.30703551e-01 5.71670055e-01 -8.66962746e-02 7.61188567e-01
7.29381979e-01 4.87968951e-01 1.80314034e-01 -4.70105141e-01
1.61743772e+00 -4.90613669e-01 -6.10751569e-01 2.14784041e-01
7.45444834e-01 -5.13035953e-01 1.01520395e+00 3.46909940e-01
-9.43081617e-01 -3.77449781e-01 -9.61451113e-01 1.07484609e-01
3.28323878e-02 -3.29861194e-01 3.43363285e-01 1.00813448e+00
-1.03582847e+00 7.47561753e-01 -1.46796525e+00 -6.27239347e-01
5.73391020e-01 5.59430480e-01 -2.04730168e-01 -8.43007118e-02
-7.42897272e-01 7.17263877e-01 -1.73835292e-01 -7.22238421e-03
-9.32975352e-01 -1.45094454e+00 -4.23709214e-01 1.86482836e-02
1.97662920e-01 -1.32006204e+00 1.32771599e+00 -9.32597071e-02
-1.54583859e+00 7.16919541e-01 -5.72446100e-02 -5.70135832e-01
3.20851147e-01 -7.06707537e-01 -4.91014510e-01 5.76425791e-01
-3.68942648e-01 6.08569920e-01 6.63649201e-01 -4.02528524e-01
-4.71524298e-01 -4.72520143e-01 -1.88994825e-01 -6.39384463e-02
-3.49453092e-01 3.44180733e-01 -4.69864234e-02 -1.59058988e-01
-5.98852932e-01 -1.07079268e+00 -4.08304147e-02 1.32669121e-01
-8.78643095e-02 4.56162184e-01 1.06672704e+00 -6.96888626e-01
7.19678879e-01 -1.95397246e+00 -5.37479520e-01 -1.55076951e-01
5.94959140e-01 8.08875561e-01 2.42425278e-01 1.89849928e-01
1.42516896e-01 -1.42851789e-02 -2.57271063e-02 -8.99504647e-02
-5.55077612e-01 2.39048200e-03 -6.13714680e-02 6.33932471e-01
-1.79414451e-01 1.22060394e+00 -6.51097238e-01 -3.05864066e-01
5.81734598e-01 1.25460303e+00 -4.91247445e-01 5.16447365e-01
3.62858534e-01 1.11913824e+00 -1.84865102e-01 5.26028514e-01
8.55489969e-02 -7.19944835e-01 1.09658144e-01 -1.75304428e-01
4.83036153e-02 2.54955322e-01 -3.22829455e-01 1.52167237e+00
-4.74781334e-01 5.15000045e-01 4.46107015e-02 -5.74529529e-01
3.25019598e-01 8.19971323e-01 9.10893381e-01 -6.42977178e-01
5.50948501e-01 -2.08780393e-01 1.29363239e-02 -9.31256473e-01
-9.57552418e-02 -3.77186686e-01 1.93380386e-01 4.65751827e-01
-5.59394598e-01 1.45169258e-01 -4.77338612e-01 -7.52056688e-02
1.59740114e+00 -1.18571721e-01 2.79599667e-01 -3.87857556e-01
1.54169619e-01 -8.62391666e-02 4.21136200e-01 6.37252271e-01
-5.99895775e-01 5.21761179e-01 9.38138738e-02 -9.65790987e-01
-7.40023792e-01 -1.10216999e+00 2.05184426e-02 6.52018368e-01
-4.37904179e-01 -1.44608185e-01 -7.62672782e-01 -4.57898647e-01
-2.47724965e-01 4.42617238e-01 -5.71771145e-01 -1.30857110e-01
-6.78222358e-01 -7.71568179e-01 7.08203614e-01 9.59412396e-01
3.14812213e-01 -8.19931865e-01 -1.90687466e+00 4.63425100e-01
-2.62513608e-01 -1.29963505e+00 -5.18193364e-01 -2.12709531e-01
-9.78431761e-01 -1.20282495e+00 -8.20244908e-01 -1.12256959e-01
2.61546165e-01 3.22822601e-01 8.69367301e-01 -4.36566025e-02
-9.84265387e-01 8.01034868e-01 -6.67486340e-02 -6.92422688e-01
-1.72135159e-01 -2.70489544e-01 1.51390314e-01 -1.92121282e-01
5.30501306e-01 -2.77902305e-01 -1.23307526e+00 -2.02088021e-02
-5.46481550e-01 -2.45517030e-01 1.04430459e-01 5.25683820e-01
5.26608884e-01 -5.65365255e-01 4.27747995e-01 -8.33459198e-01
5.68094015e-01 -2.90798724e-01 -6.05489075e-01 -1.29920952e-02
-6.91466808e-01 -1.47908226e-01 6.22761607e-01 -1.64958104e-01
-6.67353868e-01 -1.20830514e-01 2.16556229e-02 -6.08712733e-01
-2.66118526e-01 1.60852417e-01 6.72041833e-01 3.78203616e-02
6.55334532e-01 -1.94233224e-01 2.67142653e-01 -1.19600028e-01
-2.64989495e-01 6.90311253e-01 7.00599194e-01 -2.24842839e-02
2.19058827e-01 7.86258042e-01 4.59912807e-01 -9.18844342e-01
-6.36416137e-01 -5.75693786e-01 -2.36941740e-01 -2.57064104e-01
1.30753458e+00 -9.39941585e-01 -1.36510122e+00 3.71315986e-01
-1.08712220e+00 -5.81332862e-01 -8.20974410e-02 7.65612483e-01
-1.23382904e-01 2.52244800e-01 -7.53839016e-01 -5.61186969e-01
-9.48197544e-01 -1.10104132e+00 1.06717539e+00 5.80828376e-02
-5.12864649e-01 -9.20157194e-01 1.91270933e-01 5.28623879e-01
1.04169655e+00 7.41576374e-01 6.76117539e-01 -4.31423575e-01
-5.00335753e-01 -3.13018471e-01 -1.87146842e-01 1.60315782e-01
3.39950889e-01 -2.09893525e-01 -1.02869749e+00 -6.52478397e-01
3.08737397e-01 -2.28621185e-01 5.56624234e-01 8.18383813e-01
1.12714124e+00 -1.83858931e-01 -4.83175486e-01 7.88604498e-01
1.17840230e+00 4.17983770e-01 7.58857846e-01 -2.49111846e-01
6.63973927e-01 1.65485606e-01 2.60917932e-01 6.08746469e-01
3.19349617e-01 4.61643338e-01 2.85191000e-01 -3.44338119e-01
1.03943102e-01 4.65697169e-01 2.30121136e-01 6.83229387e-01
-2.65802622e-01 -2.55418539e-01 -1.22354007e+00 3.68652463e-01
-1.37037301e+00 -7.74923801e-01 -2.93244094e-01 2.63526130e+00
3.70018095e-01 -1.46406099e-01 2.17806339e-01 -1.24955818e-01
5.68786323e-01 -2.87629247e-01 -5.62042832e-01 -3.66697192e-01
6.32806659e-01 6.66252315e-01 6.61403239e-01 4.42793399e-01
-1.01270604e+00 3.35599363e-01 6.89616776e+00 -3.34691554e-01
-1.50222731e+00 5.79018354e-01 8.96327257e-01 -6.51287794e-01
4.34799790e-01 -6.98237240e-01 -4.79531080e-01 2.50082850e-01
1.64445043e+00 2.85456330e-01 2.19762236e-01 3.70673865e-01
5.40142894e-01 -5.87231256e-02 -1.06978977e+00 1.17139769e+00
-1.75140947e-02 -1.32006896e+00 -5.97543538e-01 1.03009112e-01
3.74864191e-01 4.64159071e-01 -1.79366115e-02 -7.97659904e-02
-8.95346180e-02 -1.06947613e+00 -1.43672824e-01 5.51104784e-01
1.35291159e+00 -6.66141033e-01 1.00176311e+00 3.20654586e-02
-9.64929104e-01 4.46363948e-02 1.76416546e-01 5.83261549e-02
3.86379182e-01 5.21714449e-01 -1.10140395e+00 4.88006286e-02
8.99110854e-01 1.73683941e-01 -1.42846733e-01 8.66166294e-01
2.73722470e-01 8.15069556e-01 -5.12306809e-01 8.26708227e-02
2.91728284e-02 3.35802227e-01 3.10862422e-01 1.23145151e+00
3.97231489e-01 5.90901673e-01 8.05268735e-02 2.35490009e-01
5.84485680e-02 -1.63274094e-01 -6.51141644e-01 -6.68775514e-02
2.54615843e-02 1.19937766e+00 -8.35549891e-01 -5.06722212e-01
-4.60185111e-01 9.64833260e-01 -3.43937397e-01 6.95569515e-02
-1.19653535e+00 -4.28957850e-01 8.16988349e-01 4.60411042e-01
2.04387993e-01 -2.68999875e-01 -8.51848349e-02 -7.75415242e-01
-5.56574576e-02 -8.24084461e-01 3.42133701e-01 -5.14072657e-01
-5.39074421e-01 6.62773073e-01 -2.95958072e-01 -1.03164482e+00
-2.76257038e-01 -5.76675236e-01 -3.71571928e-01 5.61095476e-01
-1.51373088e+00 -6.32393658e-01 -7.86542594e-01 6.03006780e-01
5.44849224e-02 2.18296528e-01 1.26926029e+00 5.66311896e-01
-4.25022483e-01 5.11895776e-01 -1.46130741e-01 -6.32989109e-02
7.15239704e-01 -7.38047957e-01 4.51882690e-01 6.90326571e-01
-4.41375732e-01 7.14941382e-01 3.29540461e-01 -5.37132323e-01
-1.64731371e+00 -1.38056743e+00 6.45455718e-01 -8.56098831e-01
1.17762037e-01 9.33067966e-03 -6.74793303e-01 6.51154399e-01
2.79573619e-01 3.70109588e-01 9.06943560e-01 -3.52552801e-01
-1.92400500e-01 -2.61002868e-01 -1.34645402e+00 2.93315023e-01
8.31481636e-01 -6.12779498e-01 -3.09556752e-01 5.22404790e-01
8.75267088e-01 -6.79029882e-01 -8.60833526e-01 5.14280856e-01
6.53586686e-01 -1.02059758e+00 8.18625689e-01 -7.74549365e-01
1.98039040e-01 -1.81567982e-01 7.86151066e-02 -9.19829845e-01
-2.12709188e-01 -1.03948188e+00 -7.97877833e-02 3.65883619e-01
1.59233391e-01 -9.96800721e-01 6.11538589e-01 6.64409101e-01
-1.33322537e-01 -8.71861100e-01 -9.58056450e-01 -6.42169952e-01
-4.78105754e-01 -5.35471201e-01 3.38786453e-01 7.21076250e-01
-2.59419978e-02 2.11530089e-01 -4.10795301e-01 4.38007861e-01
3.12011838e-01 -2.44253844e-01 5.76086402e-01 -1.00759232e+00
-4.85904872e-01 1.78375393e-01 -3.96696925e-01 -7.36389458e-01
-6.62440062e-01 -3.38050008e-01 2.89395656e-02 -1.50527883e+00
1.64049134e-01 -5.03957011e-02 -6.25697255e-01 6.29844546e-01
-9.08733234e-02 6.73174322e-01 -9.73071977e-02 -3.55332159e-02
-5.98920703e-01 -4.73189652e-02 8.39035094e-01 1.72275066e-01
-3.08344871e-01 -2.13404641e-01 -3.70595068e-01 4.72848922e-01
9.17276740e-01 -7.60781825e-01 -2.61929721e-01 -5.13234437e-01
3.79350968e-02 4.44721967e-01 6.17877543e-01 -1.57407570e+00
3.74016464e-02 1.08211003e-01 3.86518419e-01 -2.31309623e-01
4.76123333e-01 -8.17633271e-01 2.35075802e-01 1.23233831e+00
3.44896466e-02 4.48221475e-01 4.68501925e-01 4.83298123e-01
4.26417708e-01 6.18310332e-01 9.84900713e-01 -1.13315769e-02
1.28380582e-01 3.41263205e-01 -6.67929888e-01 3.89128417e-01
1.14594817e+00 -2.29536407e-02 -4.94728088e-01 -4.39395845e-01
-3.62200260e-01 5.71456179e-02 1.51043072e-01 3.03783953e-01
5.02174139e-01 -7.55379379e-01 -7.13508844e-01 2.20551118e-01
-1.70916840e-02 -1.62304401e-01 5.05553901e-01 1.39358902e+00
-1.06821060e+00 8.83409500e-01 -4.72339951e-02 -1.00069022e+00
-1.59894204e+00 6.36123836e-01 4.43592310e-01 -6.50877878e-02
-9.19614732e-01 7.56260037e-01 6.42465800e-02 1.42749429e-01
7.61004314e-02 -6.93935156e-01 2.27841273e-01 -4.42183435e-01
1.05134261e+00 6.57160997e-01 3.54977727e-01 -1.90987274e-01
-5.95881462e-01 5.30128658e-01 -8.39407742e-02 4.88633960e-02
1.28216290e+00 -7.54705220e-02 1.51821300e-01 3.41414720e-01
1.13939536e+00 -2.31360018e-01 -1.34542513e+00 2.28701040e-01
-5.20104229e-01 -2.91529953e-01 2.37286031e-01 -9.25761044e-01
-9.73176062e-01 1.16983545e+00 1.27899837e+00 3.10000181e-02
1.10461271e+00 -2.80153155e-01 1.43070722e+00 5.49956679e-01
3.48195910e-01 -5.47419667e-01 2.20457464e-01 1.88143179e-01
2.79346675e-01 -1.13872278e+00 -8.70821923e-02 -1.86430857e-01
-6.21997654e-01 9.33631778e-01 9.73962173e-02 3.41451466e-01
5.78882933e-01 4.61060703e-01 6.52710557e-01 -4.23216432e-01
-7.12328076e-01 4.39752191e-01 -1.60011619e-01 7.05447912e-01
5.06258011e-01 2.03088567e-01 1.43100068e-01 3.31065506e-02
3.44007462e-01 4.58294988e-01 6.16181016e-01 1.18583977e+00
-2.36580402e-01 -7.04462290e-01 -2.23476082e-01 5.42671025e-01
-1.09281206e+00 -3.26120794e-01 2.58197099e-01 4.66024518e-01
-2.58601867e-02 1.10331631e+00 1.05953731e-01 -2.03300700e-01
2.08775908e-01 4.06815559e-01 5.39747953e-01 -6.09667480e-01
-8.26068640e-01 -3.09280157e-02 7.99608454e-02 -1.06427848e+00
-4.02161598e-01 -5.09316027e-01 -1.33779788e+00 -4.64641958e-01
1.60347775e-01 -3.70630115e-01 6.89459264e-01 9.43042815e-01
1.21686387e+00 8.81956875e-01 2.92961419e-01 -4.77845311e-01
-4.22385722e-01 -6.95501447e-01 -1.02028206e-01 1.98710784e-01
7.34598517e-01 -1.39002830e-01 2.50607505e-02 3.15554261e-01] | [13.929140090942383, 3.0280354022979736] |
15d9690c-6c2c-4593-9139-5da1da224629 | corgi-content-rich-graph-neural-networks-with | 2110.04866 | null | https://arxiv.org/abs/2110.04866v1 | https://arxiv.org/pdf/2110.04866v1.pdf | CoRGi: Content-Rich Graph Neural Networks with Attention | Graph representations of a target domain often project it to a set of entities (nodes) and their relations (edges). However, such projections often miss important and rich information. For example, in graph representations used in missing value imputation, items - represented as nodes - may contain rich textual information. However, when processing graphs with graph neural networks (GNN), such information is either ignored or summarized into a single vector representation used to initialize the GNN. Towards addressing this, we present CoRGi, a GNN that considers the rich data within nodes in the context of their neighbors. This is achieved by endowing CoRGi's message passing with a personalized attention mechanism over the content of each node. This way, CoRGi assigns user-item-specific attention scores with respect to the words that appear in an item's content. We evaluate CoRGi on two edge-value prediction tasks and show that CoRGi is better at making edge-value predictions over existing methods, especially on sparse regions of the graph. | ['Miltiadis Allamanis', 'Cheng Zheng', 'Simon Peyton Jones', 'Simon Woodhead', 'Angus Lamb', 'Jooyeon Kim'] | 2021-10-10 | null | null | null | null | ['value-prediction'] | ['computer-code'] | [ 3.04208934e-01 8.07194531e-01 -6.44757271e-01 -3.69669110e-01
-1.28596500e-01 -2.62404412e-01 4.29971009e-01 7.04666078e-01
-5.33175431e-02 6.40445292e-01 1.04768836e+00 -2.18653902e-01
-1.33803278e-01 -1.19991016e+00 -7.97853470e-01 -2.53108442e-01
-1.52799889e-01 5.10768592e-01 -4.55829114e-01 -3.29016417e-01
-1.65129274e-01 1.76418684e-02 -1.07063556e+00 4.14346069e-01
6.30644441e-01 6.73163474e-01 1.26483828e-01 2.63830364e-01
-5.00607014e-01 1.18914545e+00 -4.15282100e-01 -8.92732024e-01
-2.25655437e-02 -3.60250294e-01 -8.21869433e-01 1.35946989e-01
1.95935279e-01 -8.25518072e-02 -6.02273405e-01 9.26555455e-01
1.81822758e-02 3.40064675e-01 8.61496031e-01 -1.47158647e+00
-1.06819415e+00 1.66098940e+00 -6.29601955e-01 -2.99347541e-03
4.33553904e-01 -1.19407669e-01 1.85838401e+00 -8.27395439e-01
8.33718538e-01 1.12550831e+00 6.85441911e-01 2.24895656e-01
-1.47914982e+00 -2.36078933e-01 6.05363429e-01 8.11052173e-02
-1.10538042e+00 -1.39631152e-01 9.25807059e-01 -3.86474669e-01
1.33305895e+00 2.38120079e-01 5.31774819e-01 1.40466046e+00
8.27037692e-02 6.80534899e-01 1.60348099e-02 -2.41025269e-01
-1.96905229e-02 -1.63142439e-02 6.43168628e-01 4.36629146e-01
6.07111752e-01 -4.12375629e-01 -4.41195399e-01 -1.25420094e-01
4.61974651e-01 1.78301930e-01 -1.73974156e-01 -4.53577369e-01
-1.11021829e+00 9.35551763e-01 9.26791430e-01 2.36576945e-01
-8.81197155e-01 3.18293035e-01 2.79412359e-01 2.89171666e-01
6.78284109e-01 4.80972558e-01 -2.94526577e-01 2.31987849e-01
-3.77495080e-01 -1.37684336e-02 9.48922575e-01 1.16018116e+00
7.87912130e-01 1.36926482e-02 -5.81280529e-01 8.28881323e-01
2.68918991e-01 -2.34209746e-02 3.33362669e-01 -2.93285102e-01
8.10482740e-01 9.20820355e-01 -8.71083662e-02 -1.35527468e+00
-5.75146317e-01 -7.11198747e-01 -1.03886902e+00 -2.90422946e-01
1.03025958e-01 -3.82096797e-01 -8.34249258e-01 2.04763532e+00
1.69488475e-01 1.57333672e-01 -6.58087879e-02 6.95765555e-01
1.05074501e+00 6.87943697e-01 5.26414096e-01 1.24936618e-01
1.19227374e+00 -9.51435864e-01 -6.34183884e-01 -6.77269697e-01
8.06934297e-01 -5.40756285e-02 9.15161014e-01 4.98117171e-02
-8.56993258e-01 -3.05927485e-01 -7.84688175e-01 -2.22277164e-01
-3.64218146e-01 -2.26177067e-01 8.68723392e-01 3.79491478e-01
-9.54912364e-01 6.63821578e-01 -5.18137217e-01 -3.49389851e-01
5.73247373e-01 1.55261517e-01 -7.21946537e-01 -2.61952490e-01
-1.32960403e+00 8.42669070e-01 4.36949730e-01 -7.35742822e-02
-3.97062391e-01 -7.39565492e-01 -1.27571559e+00 7.20444381e-01
4.74662095e-01 -1.02626371e+00 7.68837690e-01 -9.46210504e-01
-6.64082825e-01 4.73224282e-01 -1.87424928e-01 -5.69981277e-01
-1.04146324e-01 -7.11179376e-02 -4.04146135e-01 -3.25094759e-01
8.90166238e-02 4.04830992e-01 7.31051266e-01 -1.01785135e+00
-2.09273368e-01 -3.91550332e-01 2.81567127e-01 2.77837902e-01
-5.65928936e-01 -2.10318923e-01 -5.32877982e-01 -7.29347348e-01
8.33213180e-02 -6.26050413e-01 -4.66662526e-01 -3.00366938e-01
-9.22263205e-01 -3.64616722e-01 1.63784519e-01 -7.34593868e-01
1.54526591e+00 -1.91587353e+00 5.58090091e-01 5.19083619e-01
7.63570368e-01 -2.06332996e-01 -5.62099874e-01 8.01964045e-01
-3.82390410e-01 2.78924257e-01 -6.90276399e-02 -4.95467395e-01
1.38734221e-01 2.83569604e-01 -1.37041444e-02 3.41119081e-01
3.07813168e-01 1.35340607e+00 -8.05278242e-01 2.17199940e-02
9.60445777e-03 5.61938524e-01 -7.16820896e-01 8.64122249e-03
-4.33997691e-01 -1.72505334e-01 -3.43964070e-01 3.86615694e-01
4.20036465e-01 -7.74401248e-01 5.89976490e-01 -2.77525097e-01
6.81665123e-01 6.00687563e-01 -1.02184761e+00 1.36510050e+00
-5.18029153e-01 4.45860773e-01 -8.55790675e-02 -1.10295963e+00
8.16456199e-01 2.45197043e-02 4.68818992e-01 -5.14504910e-01
4.41824235e-02 -4.56555784e-01 3.24835181e-02 -2.71701247e-01
6.66689217e-01 6.25815615e-02 -1.04998969e-01 6.57171011e-01
2.66696662e-01 6.41651392e-01 1.39761284e-01 8.73897910e-01
1.40497363e+00 -2.08244696e-01 5.25342822e-01 8.64608884e-02
2.02516526e-01 -2.12539360e-01 3.54993880e-01 7.35608816e-01
3.78627330e-01 5.86076498e-01 1.12297237e+00 -2.69624174e-01
-1.00476646e+00 -6.97466731e-01 3.61225784e-01 1.30047810e+00
-1.56975046e-01 -7.31116235e-01 -3.81326199e-01 -8.38678539e-01
3.65956783e-01 8.69091392e-01 -1.04863048e+00 -3.39552492e-01
-3.15955430e-01 -6.94088995e-01 -2.10730415e-02 7.29906142e-01
-4.51247126e-01 -1.32127488e+00 2.20938146e-01 5.07166624e-01
-2.41997361e-01 -1.01830065e+00 -3.88046116e-01 3.26201797e-01
-5.97935677e-01 -9.33791697e-01 -3.45238239e-01 -7.29371786e-01
8.94818664e-01 2.62029171e-01 1.71634591e+00 4.17462200e-01
1.30007699e-01 2.07837328e-01 -7.01731801e-01 -1.81268185e-01
-1.08832277e-01 5.12559652e-01 -2.72695810e-01 1.53594866e-01
4.34139967e-01 -7.89699972e-01 -2.47401655e-01 -1.56142280e-01
-6.84696198e-01 2.46123314e-01 5.21848977e-01 8.96192670e-01
3.69570315e-01 -8.89559090e-02 7.33509362e-01 -1.77674830e+00
7.83731997e-01 -1.31748450e+00 -1.90051898e-01 1.47770450e-01
-6.87256277e-01 2.48893842e-01 6.01615608e-01 -3.39078516e-01
-6.59187675e-01 -1.51839390e-01 -2.48528004e-01 -2.47389510e-01
1.38852030e-01 1.20246398e+00 -4.19754982e-01 4.33565587e-01
7.81386733e-01 -1.96642891e-01 -1.03176855e-01 -5.11041582e-01
6.80834949e-01 2.72304863e-01 3.44879150e-01 -2.43757650e-01
6.69678330e-01 -1.35305062e-01 -8.06625113e-02 -5.50948203e-01
-9.66345549e-01 -5.53390622e-01 -3.69458228e-01 1.37098908e-01
4.47320461e-01 -1.03098583e+00 -5.91960728e-01 -1.13296090e-02
-8.41938972e-01 -1.78199142e-01 -4.23628777e-01 3.32749009e-01
-4.48515750e-02 1.80942923e-01 -6.97930098e-01 -5.90407133e-01
-3.57441962e-01 -6.78152263e-01 7.45066345e-01 -1.41852126e-01
-4.90282297e-01 -1.26463854e+00 -1.34246975e-01 2.18316242e-01
4.44466203e-01 2.26578847e-01 1.33439612e+00 -1.03237486e+00
-4.81175661e-01 -3.24584305e-01 -3.98477823e-01 -1.01439869e-02
5.00135981e-02 -3.36900949e-01 -6.37924552e-01 -1.37303695e-01
-6.17162824e-01 4.16093878e-03 1.11844862e+00 4.48912621e-01
1.25250506e+00 -8.24498713e-01 -5.03352225e-01 5.21307588e-01
1.51245284e+00 -5.55176616e-01 5.64006269e-01 2.76809596e-02
1.16201437e+00 6.63606107e-01 9.31751952e-02 5.87492585e-01
6.23318613e-01 5.97265065e-01 8.96442652e-01 -1.68627977e-01
1.93261243e-02 -7.03257680e-01 2.72342324e-01 4.47257429e-01
1.51665583e-01 -9.69750166e-01 -5.96416831e-01 6.91511095e-01
-1.83143747e+00 -9.32844520e-01 -5.20846665e-01 1.89127219e+00
5.22643745e-01 4.71077599e-02 1.42668650e-01 -1.91213623e-01
6.44260287e-01 4.68979388e-01 -4.73701209e-01 -2.98913479e-01
-2.25162297e-01 1.43263787e-01 4.48185205e-01 5.69338977e-01
-8.82287741e-01 7.65372872e-01 5.76233387e+00 4.00945246e-01
-4.97593880e-01 -3.34509611e-02 6.01314068e-01 -1.01279810e-01
-1.11164713e+00 4.17089872e-02 -5.73509812e-01 4.46940988e-01
1.04510808e+00 -3.69581848e-01 4.54575330e-01 8.39343667e-01
-2.09851548e-01 2.59687901e-01 -1.26392519e+00 6.85708523e-01
9.14166048e-02 -1.40908861e+00 3.46160084e-01 2.69979179e-01
6.92113757e-01 1.63334131e-01 -5.36024123e-02 6.46360993e-01
7.07617879e-01 -1.30277622e+00 3.56784552e-01 5.21812797e-01
5.46785831e-01 -7.79939294e-01 8.75491321e-01 2.36977711e-01
-1.01418138e+00 -6.45939335e-02 -7.85734177e-01 -2.04112843e-01
2.30366543e-01 8.10945690e-01 -1.10276091e+00 6.58643484e-01
1.70179620e-01 1.01719522e+00 -5.91505706e-01 7.91642070e-01
-3.15654308e-01 7.71745741e-01 -9.07127857e-02 -7.81943500e-02
2.91104078e-01 -3.65280330e-01 3.99845719e-01 1.08620918e+00
3.29183012e-01 4.78980802e-02 1.21642418e-01 9.44142878e-01
-8.30534935e-01 3.60333502e-01 -8.86127055e-01 -3.60828966e-01
4.41502780e-01 1.41791010e+00 -3.75203371e-01 -2.45653987e-01
-8.65872025e-01 7.76318192e-01 8.51166070e-01 4.95825827e-01
-5.69963038e-01 -1.99940652e-01 9.14246917e-01 3.79657924e-01
3.32621574e-01 2.38219365e-01 -4.34063971e-01 -1.12563264e+00
-1.23382933e-01 -5.99221349e-01 6.79151475e-01 -1.03905189e+00
-1.53425694e+00 5.70529878e-01 -2.83772528e-01 -7.69685805e-01
-3.50902110e-01 -4.18482840e-01 -5.80812871e-01 1.05629766e+00
-1.12299645e+00 -1.43511379e+00 -2.35055119e-01 4.75368321e-01
2.17252567e-01 -1.94097713e-01 8.11934888e-01 1.83308557e-01
-4.47777897e-01 6.57065451e-01 -6.74446896e-02 4.89476144e-01
2.06903622e-01 -1.30604768e+00 1.01352656e+00 7.74274588e-01
6.75069034e-01 5.96278846e-01 6.23158991e-01 -9.43228304e-01
-1.33615863e+00 -1.34013355e+00 1.35710251e+00 -6.61031544e-01
8.33771467e-01 -4.26132470e-01 -1.16752350e+00 1.35154033e+00
1.74768195e-01 8.53548348e-02 8.76866460e-01 9.67056811e-01
-5.41819632e-01 2.98745871e-01 -8.73676896e-01 5.16764343e-01
1.39045894e+00 -4.49646443e-01 -4.86817539e-01 4.55269128e-01
9.95925069e-01 -2.02129975e-01 -9.12064254e-01 -2.50573233e-02
2.81429440e-01 -4.93404031e-01 1.12298775e+00 -1.29560566e+00
7.52551079e-01 1.77915141e-01 -1.37595266e-01 -1.71961403e+00
-8.61794293e-01 -2.33496845e-01 -4.97369558e-01 1.19844985e+00
8.28702331e-01 -3.16827834e-01 1.05348420e+00 7.33447731e-01
-7.80398622e-02 -6.38915181e-01 -4.10703093e-01 -1.81548044e-01
-2.23737940e-01 -4.54642951e-01 9.02804971e-01 1.13997281e+00
2.85337120e-01 8.74630868e-01 -7.64909446e-01 1.15897357e-01
4.69771922e-01 1.19314089e-01 5.38618863e-01 -1.62696922e+00
-2.59500206e-01 -2.46063486e-01 -5.02490878e-01 -7.37772524e-01
4.47617620e-01 -1.49399626e+00 -3.37791711e-01 -2.12384868e+00
5.16518652e-01 -3.52082223e-01 -4.12717998e-01 8.18784118e-01
-4.51867431e-01 6.84720948e-02 1.13439873e-01 -2.36295834e-01
-5.27858615e-01 4.43707943e-01 9.53947783e-01 -4.10194278e-01
-1.08536124e-01 9.99529846e-03 -1.35717297e+00 4.37587708e-01
6.09626114e-01 -6.07447207e-01 -5.51870227e-01 -5.66453636e-01
7.83947945e-01 2.20799536e-01 2.17015043e-01 -5.07940114e-01
2.22034141e-01 -1.05716035e-01 4.20680404e-01 -4.74849761e-01
2.41783231e-01 -9.22956884e-01 6.03081882e-01 -2.38188356e-02
-6.72724724e-01 -7.24042132e-02 -2.82184690e-01 8.83445680e-01
-1.33617163e-01 -3.18583399e-01 1.62730068e-01 -1.46813855e-01
-7.29313016e-01 5.00005960e-01 -1.86201304e-01 1.19153209e-01
5.53507209e-01 2.24351119e-02 -3.02757263e-01 -7.59571731e-01
-1.04128146e+00 3.06059211e-01 2.78559297e-01 5.57819188e-01
6.23489618e-01 -1.42243814e+00 -8.02558422e-01 4.20106024e-01
4.23665732e-01 -7.67639205e-02 3.38379353e-01 5.87025166e-01
1.01714700e-01 1.01468310e-01 -8.35184231e-02 -1.53724970e-02
-9.92369354e-01 8.32401752e-01 -5.74560612e-02 -5.14427066e-01
-8.87245893e-01 9.07809794e-01 2.85954446e-01 -4.83696967e-01
1.46742165e-01 -2.63560295e-01 -7.56689489e-01 3.83669049e-01
2.73752183e-01 5.70059381e-02 1.67995602e-01 -7.44907200e-01
-1.12459347e-01 -8.25007036e-02 -2.02607602e-01 3.54987472e-01
1.66479135e+00 -1.38172820e-01 -1.90242544e-01 2.09103778e-01
1.15064716e+00 1.48590848e-01 -8.67218018e-01 -5.99066377e-01
1.39327019e-01 -4.36022133e-01 2.21789721e-02 -5.95989585e-01
-1.24633348e+00 5.46000004e-01 -3.17731798e-01 2.48718411e-01
7.81653523e-01 2.15706870e-01 5.27757525e-01 3.95606577e-01
2.31121615e-01 -5.69875658e-01 -2.30270997e-01 4.96296614e-01
8.70697856e-01 -1.28244638e+00 3.36873531e-02 -5.50243974e-01
-9.47715580e-01 6.44166887e-01 5.43270767e-01 -2.82104239e-02
7.42246985e-01 8.16310290e-03 -4.58948076e-01 -3.05515856e-01
-1.22663081e+00 -3.14976931e-01 5.32540560e-01 7.77651131e-01
5.04933774e-01 3.00424844e-01 -1.13859341e-01 1.05803502e+00
-1.36270121e-01 -3.58389884e-01 7.04078555e-01 3.38167340e-01
-3.31688344e-01 -1.03930652e+00 1.96726069e-01 1.16980481e+00
-4.95344192e-01 -4.85382646e-01 -4.50364351e-01 5.18501520e-01
-3.84405941e-01 8.24991703e-01 1.68215454e-01 -4.71931309e-01
3.00255001e-01 1.22140579e-01 -1.42252639e-01 -1.03775704e+00
-7.91598260e-01 -2.58704871e-01 5.05918920e-01 -5.67888737e-01
7.56438673e-02 -4.07488108e-01 -1.03473592e+00 -4.90426809e-01
-4.12726179e-02 1.29333720e-01 3.54989380e-01 6.25488937e-01
5.03133357e-01 8.95542920e-01 3.11757773e-01 -4.44020510e-01
-2.63947666e-01 -1.00382388e+00 -8.31398726e-01 8.80815446e-01
1.78573266e-01 -4.50610757e-01 -2.12311551e-01 -2.96678215e-01] | [8.03140926361084, 6.833065509796143] |
dbc297ae-85df-440e-80e5-af8edfde303b | parallel-sentence-level-explanation | 2302.10707 | null | https://arxiv.org/abs/2302.10707v1 | https://arxiv.org/pdf/2302.10707v1.pdf | Parallel Sentence-Level Explanation Generation for Real-World Low-Resource Scenarios | In order to reveal the rationale behind model predictions, many works have exploited providing explanations in various forms. Recently, to further guarantee readability, more and more works turn to generate sentence-level human language explanations. However, current works pursuing sentence-level explanations rely heavily on annotated training data, which limits the development of interpretability to only a few tasks. As far as we know, this paper is the first to explore this problem smoothly from weak-supervised learning to unsupervised learning. Besides, we also notice the high latency of autoregressive sentence-level explanation generation, which leads to asynchronous interpretability after prediction. Therefore, we propose a non-autoregressive interpretable model to facilitate parallel explanation generation and simultaneous prediction. Through extensive experiments on Natural Language Inference task and Spouse Prediction task, we find that users are able to train classifiers with comparable performance $10-15\times$ faster with parallel explanation generation using only a few or no annotated training data. | ['Qi Dai', 'Xiaokang Chen', 'Yan Liu'] | 2023-02-21 | null | null | null | null | ['explanation-generation'] | ['natural-language-processing'] | [ 5.43807805e-01 1.03750408e+00 -4.44483370e-01 -7.05047667e-01
-5.09325862e-01 -1.92180470e-01 5.73108137e-01 1.70438170e-01
1.11996412e-01 8.85003269e-01 3.88228416e-01 -7.90892601e-01
-1.71542335e-02 -5.80919564e-01 -5.15937924e-01 1.80465039e-02
2.96320438e-01 5.70019007e-01 -1.67620853e-01 -2.47373387e-01
5.16149223e-01 -7.70372078e-02 -1.62144840e+00 5.89465439e-01
1.41220033e+00 5.26338398e-01 2.88885742e-01 7.40516245e-01
-4.39127237e-01 9.13784921e-01 -4.16111469e-01 -6.73103869e-01
-2.46898755e-01 -7.62644112e-01 -1.20639265e+00 5.88403270e-03
1.42129481e-01 -2.89652556e-01 1.42036542e-01 6.33131683e-01
4.96923514e-02 -1.81075379e-01 6.96744025e-01 -1.35937321e+00
-1.08713686e+00 1.12794745e+00 8.28564167e-04 -4.94819432e-02
7.21061230e-01 1.97918072e-01 1.27377069e+00 -8.75313461e-01
2.86778718e-01 1.04625428e+00 4.64327723e-01 9.04407680e-01
-1.09065282e+00 -4.11514610e-01 5.64258635e-01 3.55800241e-01
-6.93304956e-01 -2.26769283e-01 7.71133244e-01 -1.88252017e-01
1.19808710e+00 5.07992923e-01 6.25311673e-01 1.27317321e+00
2.01744601e-01 9.64186132e-01 1.11552489e+00 -7.44807422e-01
9.92717519e-02 4.15286779e-01 5.87715089e-01 9.17247295e-01
1.54216006e-01 -7.45300949e-02 -5.70530891e-01 6.72176480e-02
4.98843879e-01 7.05849156e-02 -2.25866392e-01 1.69120610e-01
-1.11557734e+00 8.96649837e-01 2.94429451e-01 1.41727626e-01
-1.88993886e-01 -8.73341337e-02 1.43456519e-01 4.88768101e-01
5.63157439e-01 8.81348670e-01 -8.37296009e-01 -3.51544946e-01
-5.35078943e-01 1.23825304e-01 1.06654251e+00 1.12088025e+00
6.91646755e-01 1.29228488e-01 -8.97980630e-02 5.30843079e-01
4.07872915e-01 3.46120507e-01 6.65302455e-01 -9.53130484e-01
5.01051426e-01 8.92946541e-01 1.57290716e-02 -9.06763732e-01
-3.53312731e-01 -4.10983235e-01 -1.03420031e+00 -1.20607890e-01
3.76082689e-01 -1.63962081e-01 -5.45752764e-01 1.48166847e+00
-1.63159966e-01 -2.94055846e-02 1.36423513e-01 9.89924908e-01
7.68865287e-01 5.42078078e-01 8.26331899e-02 -3.87614876e-01
1.32941198e+00 -1.37024629e+00 -7.49156475e-01 -3.66962761e-01
8.76245201e-01 -5.42481840e-01 1.67911756e+00 5.95651388e-01
-1.01779652e+00 -8.58612359e-01 -8.31223071e-01 -3.05274010e-01
-2.92588085e-01 1.39880329e-01 1.13179910e+00 5.49084067e-01
-7.78949916e-01 5.27196109e-01 -7.23361731e-01 -3.24187547e-01
-7.14064017e-03 3.62217516e-01 -4.39334065e-02 8.29552636e-02
-1.28826571e+00 9.16538715e-01 2.20934674e-01 5.92630217e-03
-3.02737474e-01 -5.54957211e-01 -8.82568419e-01 1.95370436e-01
3.11741590e-01 -1.00881755e+00 1.40808475e+00 -1.30817163e+00
-1.67851317e+00 3.35015893e-01 -6.58239424e-01 -6.69253767e-01
4.21597332e-01 -5.60086370e-01 -2.83188850e-01 -3.34209234e-01
5.90184741e-02 6.90464199e-01 4.21980888e-01 -1.10065901e+00
-3.91766399e-01 -9.09568295e-02 4.26502675e-01 1.78659201e-01
-5.20893395e-01 -2.38340855e-01 -8.01562890e-02 -4.18744415e-01
1.99270681e-01 -7.99555123e-01 -4.65034336e-01 -3.33545446e-01
-6.02605641e-01 -5.22430956e-01 4.83831435e-01 -6.64336741e-01
1.38008320e+00 -1.75921619e+00 9.23906341e-02 -9.97302830e-02
3.22547525e-01 7.28829950e-03 -1.41379908e-01 3.70477796e-01
-2.53079206e-01 4.87369537e-01 -1.69817492e-01 -6.40380502e-01
2.47223064e-01 3.89949590e-01 -7.91608512e-01 -2.01802060e-01
3.18178535e-01 1.00711727e+00 -1.07527697e+00 -4.29590136e-01
1.77889336e-02 8.91990140e-02 -8.46225441e-01 4.38809067e-01
-5.47693014e-01 5.21355689e-01 -6.17196381e-01 4.02110189e-01
3.04161280e-01 -6.49517477e-01 1.76662374e-02 2.86276758e-01
-1.90201309e-02 6.82772458e-01 -8.41800869e-01 1.61070108e+00
-6.73110485e-01 6.85291827e-01 -6.86376989e-01 -9.31158066e-01
9.17471468e-01 4.04811621e-01 -1.26787975e-01 -4.43437487e-01
-2.30679080e-01 2.60539442e-01 1.27027452e-01 -6.25384271e-01
6.49369597e-01 -6.87329471e-02 3.65942679e-02 6.69097722e-01
-2.11264923e-01 -3.99711216e-03 1.13595247e-01 2.52418488e-01
9.47664678e-01 2.25012660e-01 5.00853300e-01 2.99681127e-02
5.94371140e-01 5.04768305e-02 2.87293792e-01 1.08794153e+00
2.43200719e-01 6.88248396e-01 5.27806818e-01 -7.61939228e-01
-8.95055592e-01 -8.31280529e-01 1.79510698e-01 1.04670346e+00
1.31788164e-01 -7.93409646e-01 -7.75464833e-01 -9.02389467e-01
-3.87342364e-01 1.33952618e+00 -3.16219330e-01 7.94456825e-02
-3.62058282e-01 -3.38847041e-01 2.92169482e-01 6.39787138e-01
3.57955247e-01 -1.31270278e+00 -4.09567803e-01 1.76375732e-01
-3.55801225e-01 -8.60449374e-01 -2.38679573e-01 1.06750585e-01
-1.19911921e+00 -8.21047306e-01 -1.44975051e-01 -5.25867283e-01
9.67105448e-01 1.56446055e-01 1.31719673e+00 7.66123474e-01
1.10271074e-01 1.31562218e-01 -4.89819854e-01 -6.05853140e-01
-5.18798769e-01 4.05603975e-01 3.64324339e-02 -4.78021145e-01
3.13448250e-01 -6.66755974e-01 -4.58610773e-01 1.74386024e-01
-6.50995195e-01 8.07937324e-01 8.09958160e-01 9.97516811e-01
2.17087522e-01 -3.07335049e-01 5.66756010e-01 -1.26269865e+00
8.27216923e-01 -3.29587072e-01 -2.11342931e-01 2.54028887e-01
-1.17832470e+00 5.49040854e-01 1.16611898e+00 -3.24046135e-01
-1.27314246e+00 -2.57696509e-02 -1.48291245e-01 2.24629819e-01
-6.14669144e-01 6.57804251e-01 2.29357034e-02 6.11637115e-01
7.48520195e-01 3.57514769e-01 -1.62464485e-01 -2.58697003e-01
5.38675785e-01 6.60079420e-01 2.20528379e-01 -4.40070391e-01
9.44732487e-01 -7.19512394e-03 -3.20001602e-01 -5.50683081e-01
-1.17603493e+00 -1.82831753e-02 -5.47369003e-01 1.84156485e-02
8.12073052e-01 -6.25511110e-01 -6.46028221e-01 -3.09633851e-01
-1.65990734e+00 -2.85608888e-01 -6.05464168e-02 4.23170328e-01
-5.94321549e-01 4.92329270e-01 -6.10120952e-01 -1.03693700e+00
-4.59573239e-01 -9.78200972e-01 9.82457340e-01 1.76948309e-01
-9.23553407e-01 -1.10443521e+00 -1.55311331e-01 7.16993272e-01
3.53041500e-01 -2.40152881e-01 1.20035255e+00 -9.12462056e-01
-8.79581153e-01 -1.98264554e-01 -1.78226486e-01 1.72963981e-02
2.29062513e-03 -1.37498140e-01 -8.87268305e-01 2.27499142e-01
-1.23915449e-02 -2.65934795e-01 5.75308859e-01 8.33136514e-02
1.56848907e+00 -6.81167126e-01 -2.49183252e-01 2.36572906e-01
8.99378479e-01 -1.36191517e-01 6.84515893e-01 1.71071842e-01
5.80403864e-01 5.66854775e-01 7.48653114e-01 3.43065828e-01
5.91132641e-01 4.63862687e-01 3.25010121e-01 -3.76814365e-01
1.74420401e-01 -6.07712090e-01 4.40226793e-01 9.43437994e-01
-4.40855414e-01 -4.34337050e-01 -7.01849043e-01 2.73516148e-01
-2.00932908e+00 -1.04874766e+00 -4.93527502e-01 1.90737116e+00
7.82992780e-01 3.65252405e-01 -2.50483781e-01 8.28481764e-02
2.13709399e-01 -3.36786248e-02 -3.24141294e-01 -7.70632446e-01
2.61723757e-01 8.79560411e-02 -4.64443378e-02 7.90792048e-01
-5.81613481e-01 8.67770195e-01 6.30658102e+00 4.97377157e-01
-8.84417474e-01 -9.87946913e-02 8.70153427e-01 3.64177138e-01
-9.16879833e-01 4.08956140e-01 -6.86077654e-01 3.81937832e-01
9.87738788e-01 -1.07066512e-01 3.00989300e-01 1.19784153e+00
5.01449764e-01 1.42126873e-01 -1.34151328e+00 6.32448971e-01
5.46993464e-02 -1.18520367e+00 3.32825512e-01 -1.24564372e-01
6.35389030e-01 -4.66929555e-01 -7.03180060e-02 5.65527022e-01
1.15327977e-01 -1.29608405e+00 4.42968935e-01 5.49216568e-01
3.14483523e-01 -5.26642323e-01 9.88951862e-01 9.45884883e-01
-8.48382056e-01 -6.58806935e-02 -4.26638097e-01 -9.20674980e-01
1.50472388e-01 5.50964952e-01 -1.08944499e+00 5.62514007e-01
1.65908083e-01 5.74645162e-01 -6.10649168e-01 5.24170935e-01
-8.80301714e-01 8.87574196e-01 7.26834014e-02 -4.18137461e-01
-4.40178514e-02 7.25896144e-03 1.37447283e-01 1.03244293e+00
3.88165623e-01 4.23587650e-01 1.49354503e-01 1.03192842e+00
7.17497393e-02 1.36565492e-01 -6.54178917e-01 5.78227900e-02
1.32185102e-01 1.13934183e+00 -5.35512388e-01 -4.74505991e-01
-5.20182967e-01 1.32121956e+00 4.55618054e-01 2.73144037e-01
-9.90508139e-01 -1.20364828e-02 2.54753858e-01 2.44766340e-01
-3.54750961e-01 -2.72177100e-01 -9.50125039e-01 -1.35724258e+00
9.38447099e-03 -9.33958590e-01 2.04300061e-01 -1.09609842e+00
-1.19574678e+00 7.69341886e-01 -1.84494928e-02 -1.02014410e+00
-7.16523409e-01 -5.43881238e-01 -8.13606441e-01 8.78380775e-01
-1.47391653e+00 -9.76239204e-01 -4.06165451e-01 -1.01294219e-01
1.23323762e+00 -1.51515022e-01 1.15756881e+00 -3.44501764e-01
-3.59256595e-01 5.69542468e-01 -4.75682020e-01 -1.16080143e-01
5.56636274e-01 -1.55896664e+00 6.42844379e-01 7.42081642e-01
4.30178434e-01 1.02314043e+00 1.06309080e+00 -5.77537239e-01
-1.27955997e+00 -6.74978256e-01 1.77786970e+00 -7.89500535e-01
4.83076960e-01 -1.92176893e-01 -9.85890687e-01 7.47438610e-01
6.66551650e-01 -5.01082003e-01 8.07159960e-01 5.65463781e-01
-1.45742252e-01 1.46439433e-01 -5.93008816e-01 9.55778897e-01
1.16652012e+00 -4.82309729e-01 -8.90806615e-01 4.79896456e-01
9.88154948e-01 -2.93767124e-01 -5.87058723e-01 2.43607283e-01
3.90767008e-01 -1.16139615e+00 5.72092533e-01 -8.61429632e-01
1.23278606e+00 -2.12515041e-01 4.03082937e-01 -1.19108093e+00
-1.85416862e-01 -6.62496269e-01 -1.56772301e-01 1.22923458e+00
1.03795016e+00 -5.42602062e-01 9.69237328e-01 1.04675233e+00
-3.71114105e-01 -9.28879678e-01 -2.51972556e-01 -4.61281210e-01
-1.96159616e-01 -6.54544353e-01 5.86589396e-01 7.72195339e-01
5.38986862e-01 7.51754940e-01 -5.19633353e-01 1.77351713e-01
3.25630099e-01 4.53715295e-01 9.83452499e-01 -1.10185003e+00
-7.14122355e-01 -2.19943285e-01 8.70507360e-02 -1.61841071e+00
4.27285671e-01 -8.01679611e-01 2.18882576e-01 -1.57468033e+00
3.44531149e-01 -5.12405448e-02 2.18281895e-03 5.96590102e-01
-5.84440529e-01 -1.85996756e-01 1.69829931e-02 1.06720544e-01
-6.69595659e-01 4.86968189e-01 1.15068889e+00 2.02398729e-02
-1.21448942e-01 2.04987004e-01 -1.01247478e+00 8.87476206e-01
8.61859083e-01 -3.46220970e-01 -8.26132298e-01 -6.64022148e-01
2.48649880e-01 2.90789992e-01 2.61040926e-01 -7.80660629e-01
1.35458156e-01 -4.45534259e-01 3.27513188e-01 -4.00108576e-01
4.86470051e-02 -6.13658845e-01 2.76668798e-02 4.63571429e-01
-9.16796386e-01 3.11968982e-01 -1.08296394e-01 4.92769390e-01
-2.92450339e-01 -3.63609701e-01 8.67515728e-02 -1.96638346e-01
-6.16366565e-01 5.48875928e-02 -5.67035317e-01 -2.90012956e-01
5.30845344e-01 -3.05346310e-01 -2.22587898e-01 -8.25204313e-01
-5.19566298e-01 1.94577277e-01 2.71308631e-01 3.46749157e-01
7.70998299e-01 -9.88039672e-01 -5.56848109e-01 1.86087057e-01
5.30542731e-02 -6.84827790e-02 5.96915185e-02 6.25818551e-01
-2.37164482e-01 6.96887195e-01 1.53779313e-01 -3.88460964e-01
-1.23786044e+00 4.48433399e-01 -5.92250526e-02 -5.84992230e-01
-5.39064825e-01 5.51688135e-01 5.29533438e-02 -5.86766422e-01
8.12168494e-02 -5.07382810e-01 -1.48819581e-01 -5.14745474e-01
4.92771775e-01 2.61672437e-02 -2.62789994e-01 2.10920379e-01
-2.94340163e-04 1.57041207e-01 -4.02271412e-02 -4.00541574e-02
1.13550901e+00 -1.84098303e-01 4.44985367e-02 5.70689738e-01
6.38531864e-01 -8.81214663e-02 -9.90071952e-01 1.03875615e-01
3.16247821e-01 -2.44007945e-01 -6.26312792e-01 -8.24401617e-01
-4.80306953e-01 1.11307895e+00 -9.03223734e-03 5.72188258e-01
1.01328981e+00 2.12191343e-02 6.61290765e-01 7.21199036e-01
2.37323627e-01 -7.92997360e-01 1.89915851e-01 6.99350119e-01
1.05177712e+00 -1.45527709e+00 4.84226048e-02 -6.59462452e-01
-7.70372868e-01 1.37784004e+00 1.01228344e+00 1.81516007e-01
8.95126387e-02 3.49785239e-02 1.17595375e-01 1.05424024e-01
-1.25842571e+00 5.18602096e-02 4.04220670e-01 4.97845262e-01
1.10277700e+00 -6.67820976e-04 -4.87057418e-01 9.71046269e-01
-6.17611647e-01 8.68709758e-02 4.99209493e-01 2.78276324e-01
-5.60348213e-01 -1.49900830e+00 -8.35668519e-02 4.71683979e-01
-8.59089792e-02 -5.45120180e-01 -6.09426677e-01 8.17905962e-01
-8.16223994e-02 1.11878574e+00 -1.27523139e-01 -2.86854625e-01
1.73745275e-01 2.43534133e-01 8.55999887e-02 -7.42230654e-01
-5.53800166e-01 -4.46074009e-01 3.26026559e-01 -5.07024527e-01
-2.28579447e-01 -3.22298676e-01 -1.55331421e+00 -3.47907931e-01
-2.75941819e-01 4.65198189e-01 2.65353113e-01 1.45801485e+00
3.03266615e-01 4.90890414e-01 3.53954792e-01 -4.83640432e-01
-6.81059837e-01 -1.06893098e+00 -7.70159438e-02 4.06772852e-01
3.32797435e-03 -1.18439682e-01 -4.06870604e-01 3.72875124e-01] | [9.450061798095703, 6.679744243621826] |
af7db1a1-4195-4ff4-b8e8-ad0255113291 | ieee-big-data-cup-2022-privacy-preserving | 2211.11565 | null | https://arxiv.org/abs/2211.11565v1 | https://arxiv.org/pdf/2211.11565v1.pdf | IEEE Big Data Cup 2022: Privacy Preserving Matching of Encrypted Images with Deep Learning | Smart sensors, devices and systems deployed in smart cities have brought improved physical protections to their citizens. Enhanced crime prevention, and fire and life safety protection are achieved through these technologies that perform motion detection, threat and actors profiling, and real-time alerts. However, an important requirement in these increasingly prevalent deployments is the preservation of privacy and enforcement of protection of personal identifiable information. Thus, strong encryption and anonymization techniques should be applied to the collected data. In this IEEE Big Data Cup 2022 challenge, different masking, encoding and homomorphic encryption techniques were applied to the images to protect the privacy of their contents. Participants are required to develop detection solutions to perform privacy preserving matching of these images. In this paper, we describe our solution which is based on state-of-the-art deep convolutional neural networks and various data augmentation techniques. Our solution achieved 1st place at the IEEE Big Data Cup 2022: Privacy Preserving Matching of Encrypted Images Challenge. | ['Vrizlynn L. L. Thing'] | 2022-11-18 | null | null | null | null | ['motion-detection'] | ['computer-vision'] | [ 2.21100956e-01 2.52807550e-02 6.85119852e-02 -4.85370725e-01
-5.96917391e-01 -6.59421146e-01 7.45406687e-01 2.99836218e-01
-8.04615796e-01 4.76639658e-01 5.26582599e-01 -3.04756016e-01
4.38791662e-02 -1.09402454e+00 -3.11335176e-01 -6.65849328e-01
4.12611067e-02 -1.52400851e-01 7.52881095e-02 -6.39810935e-02
-4.87324819e-02 7.96905339e-01 -1.46567190e+00 7.42976725e-01
3.65033537e-01 1.34552503e+00 -5.41217685e-01 8.43420088e-01
2.47851148e-01 6.94215536e-01 -5.20081401e-01 -7.64046669e-01
8.94781947e-01 4.80670631e-01 -6.37133956e-01 -5.03197551e-01
6.84933543e-01 -8.70177984e-01 -6.78579330e-01 1.13731849e+00
6.81310773e-01 -1.34614199e-01 1.18783243e-01 -1.79261410e+00
-4.26525950e-01 3.71015102e-01 -4.60469365e-01 1.03968129e-01
2.26811290e-01 2.78358668e-01 2.73163289e-01 -2.86307305e-01
3.51308733e-01 8.35680068e-01 9.21824276e-01 7.26479709e-01
-6.31604373e-01 -1.10485673e+00 -4.29217696e-01 5.43011367e-01
-1.46861792e+00 -5.47123313e-01 4.49671626e-01 -3.77305329e-01
9.04728055e-01 6.18182123e-01 5.85703433e-01 1.14419854e+00
-1.18370719e-01 6.13145471e-01 5.82161069e-01 5.48678078e-02
3.70359838e-01 3.11872363e-01 2.40209445e-01 2.48230666e-01
9.10598338e-01 4.45565164e-01 -2.32443765e-01 -3.60050142e-01
-2.03781668e-03 3.50711226e-01 -2.99033877e-02 -1.12288624e-01
-6.67724609e-01 6.82997823e-01 2.81445593e-01 2.10897461e-01
-3.41434568e-01 2.83057451e-01 7.11147845e-01 8.45676288e-02
1.68364987e-01 -4.02569845e-02 -3.42713743e-01 -4.39542532e-02
-6.98870361e-01 3.08112144e-01 4.74406868e-01 9.40544724e-01
6.09197438e-01 -3.93171489e-01 -2.35815883e-01 -9.60189626e-02
3.11396867e-01 4.15452003e-01 2.79383272e-01 -1.00269246e+00
9.16718304e-01 1.02356088e+00 2.54418373e-01 -1.31870675e+00
-4.93211299e-01 1.77473307e-01 -1.29907763e+00 4.39963013e-01
1.52178735e-01 -3.31052542e-01 -6.90009177e-01 1.47276330e+00
5.24478912e-01 6.13217615e-02 2.41540268e-01 3.43467504e-01
6.07769489e-01 4.02784675e-01 3.98475975e-01 4.05707687e-01
1.50944793e+00 -1.65491194e-01 -7.83202946e-01 -2.16827825e-01
8.53451431e-01 -2.45970637e-01 5.40334761e-01 9.53788385e-02
-8.18504214e-01 -3.59018624e-01 -1.10237992e+00 -1.53513178e-01
-1.03708279e+00 -2.01261550e-01 4.86069441e-01 1.64076459e+00
-8.35374415e-01 5.19172132e-01 -8.33174646e-01 -3.92634302e-01
1.14021659e+00 7.51512051e-01 -7.07475722e-01 -1.86066285e-01
-1.32611477e+00 6.69758320e-01 4.15283561e-01 -4.83455174e-02
-5.59179604e-01 -8.88957798e-01 -9.55386341e-01 -8.21962580e-03
-4.09199983e-01 -4.21683043e-01 1.01156020e+00 -3.44954491e-01
-6.13407612e-01 1.12907553e+00 4.69191790e-01 -1.04106867e+00
6.74555779e-01 -3.15034688e-01 -8.17327559e-01 1.47990659e-01
5.31164138e-03 6.96995795e-01 6.60048366e-01 -7.87101686e-01
-1.10726893e+00 -6.10567629e-01 -5.76028638e-02 -3.75119090e-01
-1.07166350e+00 2.71896720e-01 1.24860130e-01 -2.01755419e-01
-5.46049595e-01 -6.37247562e-01 -3.71682882e-01 1.35508731e-01
-1.12375297e-01 2.85985112e-01 1.75521493e+00 -9.92084384e-01
1.11329436e+00 -2.38530612e+00 -9.90170062e-01 3.43175024e-01
4.28888261e-01 8.86203647e-01 -5.05643114e-02 3.37023854e-01
4.40467935e-04 4.09894705e-01 -2.01484844e-01 -6.38299108e-01
2.50977218e-01 -1.61662191e-01 -2.99980283e-01 8.60483050e-01
-2.18682751e-01 1.06381667e+00 -5.33979356e-01 -1.92803249e-01
6.84566379e-01 8.55728984e-01 -2.09804431e-01 2.17048172e-02
2.14375094e-01 2.43007138e-01 -4.54407066e-01 6.07959151e-01
1.29508901e+00 2.25291371e-01 -1.35747626e-01 -2.48995900e-01
-2.71320373e-01 -4.85963225e-02 -1.16408145e+00 1.22616744e+00
-1.12789802e-01 6.77479148e-01 4.00218010e-01 -7.69528568e-01
6.37322128e-01 6.64601982e-01 9.71833527e-01 -9.71794665e-01
5.96712291e-01 -1.42899118e-02 -6.86417818e-01 -6.02677345e-01
6.23953879e-01 6.63089216e-01 -2.57726312e-01 3.33175033e-01
-7.02846646e-01 2.67219633e-01 -2.09974676e-01 1.02070726e-01
1.42059016e+00 -3.02093834e-01 1.05338395e-01 4.85519916e-02
5.20602703e-01 -1.82416275e-01 4.72492516e-01 5.06525695e-01
-6.82195902e-01 5.15073359e-01 -1.65694088e-01 -1.09972131e+00
-1.12043631e+00 -4.78424579e-01 8.05260092e-02 4.29365188e-01
-1.10186398e-01 -5.18905401e-01 -9.66294646e-01 -6.15155876e-01
2.34316057e-03 5.20003855e-01 -6.45874918e-01 -4.66752708e-01
-8.15631390e-01 -7.83535838e-01 1.14770997e+00 5.18734276e-01
1.31721628e+00 -8.01499069e-01 -1.40224755e+00 -4.28377353e-02
-4.24535107e-03 -1.58589542e+00 -2.66032845e-01 -2.31836960e-01
-2.96513796e-01 -1.35709560e+00 -2.45201066e-01 -4.37283903e-01
5.40692866e-01 2.73101181e-01 5.64187706e-01 1.85190775e-02
-5.95227063e-01 3.30979735e-01 -3.64172429e-01 -7.11032987e-01
-2.11403310e-01 -2.16760542e-02 -1.66514859e-01 3.97807866e-01
6.79408789e-01 -5.25956690e-01 -9.55324769e-01 2.40882441e-01
-1.14439774e+00 -3.09336036e-01 2.32391596e-01 3.86744179e-02
5.55917807e-03 6.08141065e-01 -2.06646025e-01 -4.79196161e-01
6.11066699e-01 -3.43468964e-01 -8.64635348e-01 1.57084808e-01
-4.55592543e-01 -2.04201952e-01 6.66336894e-01 -1.99782312e-01
-8.22637141e-01 5.05425990e-01 -1.83298558e-01 -1.34725243e-01
-5.70240617e-01 -3.59734327e-01 -7.31876314e-01 -6.14734411e-01
4.86430794e-01 -1.99006259e-01 -3.89244407e-01 -2.94394314e-01
3.73611450e-01 1.17228186e+00 7.85472453e-01 4.71571609e-02
9.97331679e-01 1.02129984e+00 2.43755594e-01 -8.06938887e-01
-1.42615557e-01 -6.98087871e-01 -5.31208515e-01 1.84082612e-02
1.14951146e+00 -7.78187394e-01 -1.27714658e+00 9.87501860e-01
-1.28021443e+00 9.49588120e-02 -5.14556348e-01 1.27193570e-01
-1.39466312e-03 6.22331321e-01 -9.32531506e-02 -1.02303362e+00
-8.42574775e-01 -8.01753283e-01 1.10075092e+00 2.59773880e-01
-4.20337468e-01 -6.44945323e-01 9.87198576e-02 6.18527174e-01
7.11449504e-01 8.50035608e-01 2.65072078e-01 -6.45093620e-01
-6.57831252e-01 -1.10440826e+00 -3.67863864e-01 3.50643486e-01
2.52694815e-01 -2.83595741e-01 -1.29262602e+00 -4.10184264e-01
2.40751468e-02 -6.51308522e-02 7.37707913e-01 2.73488134e-01
1.30725431e+00 -8.36188555e-01 -5.49967647e-01 1.06091774e+00
1.19594777e+00 3.48125279e-01 1.30005169e+00 6.34520233e-01
7.19395638e-01 7.65896857e-01 8.85988697e-02 6.32672250e-01
4.08543825e-01 3.32616419e-01 8.61956418e-01 -1.68138891e-02
3.01625967e-01 -2.10914671e-01 2.09931940e-01 -2.85156578e-01
7.17497617e-02 -3.12970042e-01 -8.80881608e-01 7.63551354e-01
-1.93274558e+00 -1.42125225e+00 -4.18336987e-01 2.39040112e+00
1.99660137e-01 -1.36439800e-01 1.26631141e-01 4.87704068e-01
5.77447653e-01 2.62775719e-01 -2.98921674e-01 -2.41302192e-01
-3.05189133e-01 2.28969261e-01 1.30236030e+00 3.61710221e-01
-1.64251840e+00 5.68314314e-01 5.96128511e+00 3.85662287e-01
-8.73658240e-01 2.90216565e-01 5.34429491e-01 -4.98235635e-02
-9.25563499e-02 -2.35990047e-01 -9.40540195e-01 6.02809310e-01
1.21406090e+00 1.11986147e-02 6.11910746e-02 8.84985983e-01
2.80373633e-01 2.46333286e-01 -6.82641149e-01 1.11602378e+00
-1.92434028e-01 -1.56568730e+00 -3.32874566e-01 4.51598495e-01
5.94055235e-01 -1.59639433e-01 2.57287800e-01 -2.52823621e-01
5.74475884e-01 -1.15453517e+00 4.12287563e-01 2.29442388e-01
7.41892815e-01 -9.41955268e-01 8.47111523e-01 2.15086743e-01
-1.45405650e+00 -4.63984489e-01 -3.12548995e-01 3.46508585e-02
2.86933929e-01 2.76734054e-01 -6.35341048e-01 3.69530708e-01
1.07469821e+00 4.12544787e-01 -4.71476942e-01 9.13264453e-01
-1.04629382e-01 2.32991382e-01 -5.10719597e-01 -8.98278668e-05
9.25815105e-02 1.62637219e-01 2.28619307e-01 1.21295834e+00
4.47604686e-01 2.34933957e-01 -2.39627421e-01 4.85755503e-01
-1.56613350e-01 -3.49937022e-01 -8.74164045e-01 6.76705986e-02
5.10295749e-01 1.34544611e+00 -1.48847491e-01 -1.74874216e-01
-4.55683947e-01 8.98974895e-01 -2.26078644e-01 -6.52175397e-02
-6.88999116e-01 -3.86934787e-01 1.44056022e+00 4.49598491e-01
1.57964870e-01 -1.70611829e-01 -6.26002789e-01 -7.28043675e-01
-3.28719392e-02 -7.20181882e-01 6.77278042e-01 -3.44461769e-01
-9.14079189e-01 2.85570264e-01 3.71054672e-02 -1.10726047e+00
-3.77479270e-02 -5.84590673e-01 -6.38024390e-01 5.88400424e-01
-1.26024389e+00 -1.58423400e+00 -6.60329580e-01 1.16107726e+00
-1.81188628e-01 -3.89068186e-01 1.00451660e+00 8.92671585e-01
-3.37148428e-01 7.31248140e-01 5.59065007e-02 5.06864727e-01
1.59957007e-01 -6.28615439e-01 1.08396256e+00 1.23616731e+00
-2.02584133e-01 3.09682161e-01 4.18293178e-01 -7.56705940e-01
-1.36871445e+00 -1.55543900e+00 9.63516653e-01 -5.58507025e-01
2.07603291e-01 -6.09860361e-01 -4.78484780e-01 8.50488424e-01
3.55396360e-01 2.16486618e-01 7.43535459e-01 -7.98590720e-01
-3.26754600e-01 -4.31982279e-01 -2.00380826e+00 3.44598949e-01
7.59657085e-01 -6.65678024e-01 -5.14899604e-02 2.83138096e-01
6.71852171e-01 -7.39824772e-02 -4.46991682e-01 3.45597774e-01
6.77989483e-01 -9.47047532e-01 1.28640413e+00 -7.21769452e-01
-2.43441865e-01 -3.59087974e-01 -4.94751692e-01 -4.58469361e-01
-1.69365063e-01 -9.18418527e-01 -3.10458571e-01 1.25518036e+00
-6.90209568e-02 -7.40186632e-01 1.37734652e+00 1.54642427e+00
1.99808791e-01 1.62826836e-01 -1.19281197e+00 -5.18554747e-01
-2.26381630e-01 -8.05507004e-01 1.31155813e+00 9.60255563e-01
-2.69707471e-01 -5.28194308e-01 -7.19863951e-01 7.84431815e-01
1.18629944e+00 -6.30166233e-01 8.43924582e-01 -1.12638724e+00
4.88334924e-01 -3.21388617e-03 -1.04018033e+00 -2.19179913e-01
-1.76325321e-01 -3.16789836e-01 -6.60668612e-01 -1.49154019e+00
-3.05480331e-01 5.71087562e-02 -2.47570708e-01 7.62123466e-01
4.36752409e-01 4.23751682e-01 4.39744666e-02 -3.05151612e-01
-2.45438933e-01 1.80571407e-01 2.48935625e-01 -5.38365424e-01
-2.24613678e-02 2.65346736e-01 -7.64989674e-01 6.10588014e-01
1.31556118e+00 -5.08552194e-01 -3.48607689e-01 -5.90282083e-01
3.37546349e-01 -5.85445702e-01 9.21942949e-01 -1.49080873e+00
6.58741176e-01 -1.97011814e-01 4.29111123e-01 -8.29428673e-01
3.95973861e-01 -1.62953615e+00 6.10245049e-01 7.76706874e-01
2.75108926e-02 2.41541237e-01 3.87176007e-01 5.96243106e-02
5.47609851e-02 2.32124656e-01 8.35403681e-01 -1.20305017e-01
-9.88987207e-01 6.91668987e-01 -2.31037125e-01 -2.59980589e-01
1.50486791e+00 -6.02294981e-01 -6.91663563e-01 -4.10789102e-01
-9.51933712e-02 2.49525875e-01 4.72940922e-01 3.59850049e-01
7.96894550e-01 -1.44283402e+00 -5.89241326e-01 4.70742226e-01
1.06172808e-01 -1.19284265e-01 3.20905954e-01 2.32550889e-01
-6.56651974e-01 3.61732274e-01 -5.36680698e-01 1.21386409e-01
-1.67744732e+00 6.62392616e-01 3.01988810e-01 -1.84592128e-01
-6.18821204e-01 6.27445340e-01 -3.66464704e-01 -2.58780003e-01
5.78742146e-01 -2.07386628e-01 -1.09617606e-01 -4.35032621e-02
1.32216692e+00 8.50418389e-01 2.56614298e-01 -7.05528557e-01
-7.48288691e-01 5.92382252e-01 1.61719158e-01 -7.56061152e-02
1.43829155e+00 -2.21816972e-01 1.64458156e-01 -7.83063710e-01
1.39600456e+00 2.13718805e-02 -1.13229120e+00 3.97437841e-01
8.98447707e-02 -7.30111361e-01 -1.20405722e-02 -5.23697615e-01
-1.33477151e+00 9.84606624e-01 1.08935010e+00 4.54235494e-01
1.29717851e+00 -5.72485685e-01 1.39336705e+00 2.51957864e-01
3.83096248e-01 -1.21099806e+00 -6.68639183e-01 9.08257738e-02
3.95775348e-01 -1.13355196e+00 -5.86759895e-02 -1.78577214e-01
-2.70665318e-01 7.67761409e-01 3.99616033e-01 2.22278550e-01
7.41004109e-01 7.22833574e-01 2.87069827e-02 -1.00892052e-01
-1.21068455e-01 3.98647077e-02 -1.37747824e-01 1.27583551e+00
-2.71522313e-01 5.49611636e-02 8.63566771e-02 5.50453246e-01
-2.84114480e-01 8.68182778e-02 1.36404440e-01 1.16524470e+00
-2.95510918e-01 -1.26596904e+00 -6.28084779e-01 -1.80051532e-02
-5.02970517e-01 5.73249236e-02 -6.77344441e-01 5.80780983e-01
5.56927264e-01 1.08387256e+00 -8.73888731e-02 -7.75679529e-01
4.67253357e-01 -1.95758760e-01 -1.32199243e-01 1.16658099e-01
-9.61795151e-01 -1.02564669e+00 2.22265780e-01 -9.36979890e-01
-3.05924773e-01 -6.16459787e-01 -1.10793507e+00 -8.73436868e-01
4.35823739e-01 -1.98347479e-01 1.02245843e+00 5.56424558e-01
6.94101632e-01 -5.99273890e-02 5.84922075e-01 -5.66130519e-01
-3.75866532e-01 -3.90397906e-01 -3.81804794e-01 8.42730999e-01
6.46808088e-01 9.51611251e-02 -1.75816938e-01 3.89271863e-02] | [12.633734703063965, 0.8388638496398926] |
381dc9da-5398-455c-9e96-b3a2919b3aa1 | a-survey-on-distributed-evolutionary | 2304.05811 | null | https://arxiv.org/abs/2304.05811v1 | https://arxiv.org/pdf/2304.05811v1.pdf | A Survey on Distributed Evolutionary Computation | The rapid development of parallel and distributed computing paradigms has brought about great revolution in computing. Thanks to the intrinsic parallelism of evolutionary computation (EC), it is natural to implement EC on parallel and distributed computing systems. On the one hand, the computing power provided by parallel computing systems can significantly improve the efficiency and scalability of EC. On the other hand, data are collected and processed in a distributed manner, which brings a novel development direction and new challenges to EC. In this paper, we intend to give a systematic review on distributed EC (DEC). First, a new taxonomy for DEC is proposed from top design mechanism to bottom implementation mechanism. Based on this taxonomy, existing studies on DEC are reviewed in terms of purpose, parallel structure of the algorithm, parallel model for implementation, and the implementation environment. Second, we clarify two major purposes of DEC, i.e., improving efficiency through parallel processing for centralized optimization and cooperating distributed individuals/sub-populations with partial information to perform distributed optimization. Third, noting that the latter purpose of DEC is an emerging and attractive trend for EC with the booming of spatially distributed paradigms, this paper gives a systematic definition of the distributed optimization and classifies it into dimension distributed-, data distributed-, and objective distributed-optimization problems. Formal formulations for these problems are provided and various DEC studies on these problems are reviewed. We also discuss challenges and potential research directions, aiming to enlighten the design of DEC and pave the way for future developments. | ['Jun Zhang', 'Kay Chen Tan', 'Tian-Fang Zhao', 'Feng-Feng Wei', 'Wei-neng Chen'] | 2023-04-12 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [-3.61072749e-01 -7.65884995e-01 1.72383651e-01 -1.30015731e-01
-1.33839594e-02 -3.54249835e-01 8.33091885e-02 9.00686011e-02
-3.59391659e-01 8.17290545e-01 -1.15406644e-02 1.96632177e-01
-7.38193214e-01 -1.17066932e+00 -1.94471553e-01 -1.21782863e+00
-2.82713145e-01 5.12895763e-01 -1.37548417e-01 -3.35820973e-01
2.55238980e-01 6.93269312e-01 -2.24333930e+00 1.10766254e-01
1.32188106e+00 1.00360858e+00 4.71677572e-01 5.84781110e-01
-1.43128023e-01 1.94756970e-01 -1.09706080e+00 -2.37139463e-01
2.29460970e-01 -7.55957663e-01 -4.02709931e-01 -6.36743605e-02
-4.06135798e-01 1.25916407e-01 2.34274149e-01 1.14764524e+00
1.13509703e+00 5.15547633e-01 4.31531459e-01 -1.28892350e+00
-7.07054257e-01 4.41070497e-01 -5.71208894e-01 4.61043902e-02
1.91634685e-01 1.37294130e-02 7.65433431e-01 -8.98978174e-01
6.97032273e-01 8.69780362e-01 5.64771593e-01 2.97524065e-01
-7.51234412e-01 -7.38770068e-02 2.56853551e-01 4.26297873e-01
-1.78882301e+00 -1.27518460e-01 6.89287543e-01 -5.05351238e-02
1.12041533e+00 8.03396583e-01 1.19724321e+00 7.43549764e-02
4.49230999e-01 7.12820113e-01 7.61108518e-01 -6.66750252e-01
7.55677760e-01 -4.27329317e-02 -1.04593404e-01 2.30951488e-01
7.02184975e-01 2.65032113e-01 -6.74911737e-01 -3.40894639e-01
8.26102868e-02 -5.04762456e-02 -7.67298713e-02 -4.43960160e-01
-6.99964583e-01 8.16975832e-01 5.39571702e-01 6.66836083e-01
-6.78624809e-01 -3.47516567e-01 6.49966359e-01 2.68043250e-01
4.05536830e-01 5.11988044e-01 -3.04574937e-01 -3.31316322e-01
-7.71196306e-01 5.43761194e-01 6.95962429e-01 5.53807735e-01
5.87712228e-01 2.61831492e-01 1.38202474e-01 1.02280235e+00
1.31335840e-01 5.06012857e-01 7.17537940e-01 -5.34457862e-01
7.15448856e-02 8.44019949e-01 -2.46903419e-01 -1.51450503e+00
-4.87963676e-01 -5.82108200e-01 -1.07844472e+00 2.65783459e-01
-2.98815548e-01 -3.11403930e-01 -8.24821070e-02 1.30705118e+00
7.23048568e-01 -4.53408718e-01 8.87875333e-02 1.18053377e+00
3.84697288e-01 9.42348659e-01 -1.87424108e-01 -5.09462953e-01
1.16819704e+00 -1.11045372e+00 -8.79796624e-01 3.85406345e-01
5.07693708e-01 -5.68281651e-01 4.84900802e-01 4.38990325e-01
-9.62090611e-01 -3.01250249e-01 -8.23673844e-01 1.81575909e-01
-7.86098957e-01 1.74754709e-01 8.61610651e-01 9.57676053e-01
-1.15276289e+00 3.01969916e-01 -9.24710035e-01 -4.32350695e-01
-1.53885232e-02 5.19054472e-01 2.97562748e-01 3.16070579e-02
-8.14823687e-01 6.53484762e-01 4.27516013e-01 3.92174274e-01
-1.67424232e-01 -4.92762446e-01 -3.67482543e-01 1.34103149e-01
1.20514788e-01 -1.03929102e+00 6.00776553e-01 -1.11229396e+00
-1.83828378e+00 3.86796832e-01 -2.29939520e-01 -6.80060834e-02
1.48993671e-01 2.34449863e-01 -6.12618804e-01 -1.58675298e-01
-1.40643731e-01 3.98096032e-02 2.08451420e-01 -9.02538359e-01
-9.40537632e-01 -7.29885697e-01 -5.33674359e-01 5.96588254e-01
-9.80845451e-01 4.55428511e-01 -2.10211799e-01 -6.19070709e-01
-3.09654653e-01 -5.77050149e-01 -5.35578132e-01 -6.20403551e-02
2.77448036e-02 -3.36720139e-01 1.08159590e+00 -4.20281649e-01
1.74677718e+00 -2.08219433e+00 7.40044713e-01 3.74576211e-01
2.68826425e-01 3.87362391e-01 -3.08276080e-02 8.91658843e-01
3.55178654e-01 -7.89982527e-02 -4.34133142e-01 -2.29770616e-01
1.66467950e-01 1.58368751e-01 2.70580985e-02 5.50367713e-01
-3.16657037e-01 7.89093435e-01 -6.57303393e-01 -2.49890715e-01
9.87237096e-02 4.31400627e-01 -7.35248268e-01 2.25408524e-01
-1.40959844e-01 2.51939178e-01 -4.79976296e-01 7.08856761e-01
8.55301678e-01 1.31659610e-02 4.61751848e-01 1.98823601e-01
-7.44057000e-01 -2.99827069e-01 -1.52904177e+00 1.45522082e+00
-3.90131593e-01 3.34730506e-01 6.11924052e-01 -1.23932135e+00
1.14913988e+00 2.21702203e-01 8.67772341e-01 -7.36847043e-01
1.43578008e-01 5.64041972e-01 2.14496225e-01 -3.70897770e-01
8.29625010e-01 3.56863379e-01 -1.22105703e-01 6.59741700e-01
-4.07165974e-01 -2.27333799e-01 5.31462133e-01 -2.85468221e-01
6.03027165e-01 -1.77546129e-01 3.79838437e-01 -5.34262061e-01
6.97186232e-01 2.60363281e-01 9.38645542e-01 3.40207487e-01
-2.81215817e-01 1.69191644e-01 1.28393680e-01 -7.87892103e-01
-7.70363390e-01 -6.01579249e-01 -5.37862666e-02 1.11205673e+00
4.75732684e-01 -7.91381717e-01 -5.58027923e-01 -7.07473084e-02
1.82939991e-02 2.76034504e-01 -2.47611314e-01 -9.28399488e-02
-5.49467564e-01 -1.40166700e+00 2.30152443e-01 2.40848124e-01
7.21531749e-01 -9.62760270e-01 -8.78545880e-01 1.20767571e-01
7.48408446e-03 -4.24165934e-01 -1.41345322e-01 -1.90140799e-01
-9.65129375e-01 -6.57148421e-01 -7.06323326e-01 -9.55875337e-01
5.21465123e-01 2.66868740e-01 1.05524325e+00 1.65671065e-01
-6.94536746e-01 2.49497071e-01 -2.44373232e-01 -6.43053710e-01
9.58996043e-02 -1.11796951e-03 1.94109872e-01 1.21639542e-01
3.51680517e-01 -4.40984130e-01 -5.01459241e-01 3.64535898e-01
-8.31654966e-01 -9.61007774e-02 2.78417677e-01 9.02968824e-01
7.01349854e-01 6.05186403e-01 6.14143789e-01 -1.27287388e-01
1.02623045e+00 -5.68679571e-01 -9.39978063e-01 6.68216109e-01
-6.67116344e-01 -2.47148514e-01 9.70488369e-01 2.40102340e-03
-1.16499865e+00 -1.63782552e-01 -8.73612538e-02 -3.20620909e-02
6.81672022e-02 8.32371652e-01 -1.77910298e-01 -3.22020888e-01
5.50576985e-01 3.74001712e-01 2.26157114e-01 -5.03601551e-01
1.74040094e-01 9.52999175e-01 -1.62769884e-01 -9.81360078e-01
1.59953460e-01 2.66313672e-01 5.11501990e-02 -1.09749687e+00
1.05337009e-01 -2.37300709e-01 -1.92643151e-01 -2.42592037e-01
6.22136176e-01 -3.68712634e-01 -8.74606311e-01 7.41952121e-01
-1.06703973e+00 7.42029771e-02 -5.52443922e-01 3.40748340e-01
-2.66958803e-01 6.83698207e-02 -4.04134154e-01 -1.02662265e+00
-6.95259631e-01 -1.17492378e+00 7.38022506e-01 6.40632570e-01
1.88267201e-01 -1.11517406e+00 5.08585095e-01 -5.62011451e-03
1.08421159e+00 2.20100880e-01 6.10165715e-01 -2.37051979e-01
-3.17927420e-01 -1.24328271e-01 2.09067211e-01 -3.35749760e-02
-9.27817263e-03 4.40071732e-01 -4.01345730e-01 -3.51156563e-01
1.09960824e-01 -8.40648934e-02 8.53309259e-02 4.48339283e-01
1.19607055e+00 1.66618731e-02 -5.16089797e-01 8.59426975e-01
1.68857670e+00 4.25573051e-01 3.81505281e-01 3.02362442e-01
1.50379822e-01 6.81377947e-01 6.42906070e-01 8.27576578e-01
2.97794819e-01 8.85617077e-01 1.13241039e-01 5.08753769e-02
-8.82122666e-03 3.06059182e-01 1.87823981e-01 1.45926762e+00
-5.71517885e-01 -4.15890574e-01 -9.07109439e-01 2.77254015e-01
-2.01059723e+00 -8.01261961e-01 -9.43354741e-02 2.14805794e+00
4.52234805e-01 -1.12056231e+00 3.40529025e-01 -8.96434113e-02
1.11366475e+00 -9.29707512e-02 -4.39230323e-01 -8.08653593e-01
-5.44977129e-01 8.21276382e-02 -4.11780477e-02 9.68384296e-02
-6.30540431e-01 5.54449379e-01 6.46846962e+00 9.32066083e-01
-1.47454476e+00 1.55465230e-01 4.96243119e-01 -3.62008631e-01
-3.17928135e-01 -4.66671467e-01 -7.70939112e-01 8.22191179e-01
9.48485851e-01 -7.78276801e-01 8.18114340e-01 8.58588398e-01
8.32365304e-02 -1.38832569e-01 -3.40443641e-01 1.27147448e+00
1.33321330e-01 -1.40095127e+00 -4.99296412e-02 2.48494536e-01
1.20232093e+00 1.21470183e-01 -7.45693147e-02 -2.39605494e-02
1.59898147e-01 -6.89041674e-01 5.20803928e-01 2.28864029e-01
3.79594356e-01 -1.05490577e+00 8.62567246e-01 4.02577311e-01
-1.43083453e+00 -3.78070116e-01 -7.13700175e-01 -3.85169059e-01
3.04764420e-01 7.56290734e-01 3.23323756e-01 1.31020641e+00
1.08060217e+00 6.47561729e-01 -1.99846104e-01 1.50081336e+00
-2.88510788e-02 -1.36032132e-02 -6.13759220e-01 -7.13065147e-01
-7.89365098e-02 -7.97811151e-01 7.29316831e-01 6.76091671e-01
6.49599671e-01 4.55079833e-03 6.73466325e-02 7.77401567e-01
2.72785246e-01 5.17055273e-01 -4.79300618e-01 -1.85795546e-01
5.77169001e-01 1.21796119e+00 -5.72364032e-01 -8.78079012e-02
-2.38457531e-01 7.02687621e-01 3.93728882e-01 2.45258167e-01
-6.38993859e-01 -8.81402433e-01 7.79330134e-01 -5.50507903e-01
1.75650582e-01 -3.33544433e-01 -7.43861437e-01 -1.21580946e+00
1.24337077e-01 -1.04708469e+00 6.90199316e-01 -2.88340598e-01
-1.25078344e+00 8.69356573e-01 -1.92429945e-01 -9.09861326e-01
-8.56160671e-02 -6.19139731e-01 -7.52972901e-01 9.39135134e-01
-1.13299096e+00 -7.77163327e-01 -4.50516313e-01 4.84576941e-01
2.76386410e-01 -4.64533538e-01 1.02940750e+00 3.83395612e-01
-1.02166319e+00 3.37698072e-01 8.77672672e-01 -5.16924679e-01
3.58546823e-01 -6.48804367e-01 -2.11064234e-01 9.62254882e-01
-1.87478289e-01 6.05203748e-01 3.22422832e-01 -5.64392686e-01
-1.93980384e+00 -6.07605577e-01 1.01605821e+00 2.86147982e-01
3.75021636e-01 -3.64499331e-01 -4.35000181e-01 -6.56287223e-02
4.18713778e-01 -9.22955871e-02 8.99256349e-01 3.24831277e-01
4.79371220e-01 -5.14712989e-01 -1.18518555e+00 6.05755389e-01
8.80885184e-01 1.10369340e-01 -1.24762610e-01 4.01532978e-01
1.35397330e-01 -2.09511459e-01 -9.02670979e-01 6.88619167e-02
4.04562771e-01 -9.93713200e-01 8.41427267e-01 -1.95417136e-01
1.16231412e-01 -4.83934969e-01 -1.42974302e-01 -1.45300794e+00
-4.00856256e-01 -7.14814365e-01 5.43357525e-03 9.80415761e-01
-4.47130352e-02 -1.23015678e+00 7.15397775e-01 7.34221399e-01
-2.89082587e-01 -9.36918676e-01 -1.07585907e+00 -9.79768455e-01
1.73770636e-01 3.99995566e-04 1.10801828e+00 6.80381656e-01
2.57832557e-01 -1.01207703e-01 -1.26206830e-01 -4.19001468e-02
4.58748639e-01 6.71963632e-01 7.15645373e-01 -9.65551317e-01
-3.18738967e-01 -6.73769236e-01 -4.56682175e-01 -8.00656140e-01
-2.58691758e-01 -7.78725386e-01 -2.42244661e-01 -1.41929948e+00
1.46184191e-01 -6.98861182e-01 -5.26176430e-02 1.98239591e-02
-2.49457918e-02 1.39384329e-01 1.65500924e-01 3.91040176e-01
-5.97258091e-01 1.12509990e+00 1.07575572e+00 1.14145346e-01
-4.95873272e-01 -1.42263308e-01 -4.67828631e-01 2.14243650e-01
8.04600894e-01 -3.30734253e-01 -2.99194694e-01 -8.59406352e-01
3.12609643e-01 -2.62964845e-01 -1.52752459e-01 -8.70275617e-01
8.42666805e-01 -4.74738300e-01 1.23130120e-01 -4.12685394e-01
1.52348489e-01 -9.59799111e-01 4.20045048e-01 7.09739864e-01
2.12756544e-01 5.58925331e-01 1.04986168e-01 1.95125565e-01
-7.15080738e-01 -1.75634265e-01 5.83905697e-01 8.37594494e-02
-7.68458784e-01 9.63150933e-02 -6.76063538e-01 -2.33819038e-01
1.63678789e+00 -3.12135428e-01 -2.26914033e-01 1.04685187e-01
-4.34076041e-01 4.81738210e-01 5.33901393e-01 1.72801182e-01
4.78053391e-01 -1.11538708e+00 -6.89143896e-01 5.12171865e-01
-3.79689515e-01 -6.52297363e-02 5.59564650e-01 9.03451025e-01
-9.19033051e-01 4.99126315e-01 -4.83042508e-01 -5.73428631e-01
-1.22337592e+00 6.52405262e-01 4.58934516e-01 -9.40204412e-03
-3.74924541e-01 7.48851836e-01 -1.32559061e-01 -6.80830240e-01
-3.97950970e-03 3.54414672e-01 -3.01984474e-02 -2.06049159e-01
5.92294276e-01 8.56599808e-01 3.96776378e-01 -3.27157289e-01
-7.06383705e-01 7.49007106e-01 4.13481206e-01 1.12086780e-01
1.60870361e+00 -2.11696714e-01 -1.04590988e+00 1.98266916e-02
9.20233369e-01 8.15622434e-02 -3.49641234e-01 2.31627211e-01
-2.88905174e-01 -6.71376288e-01 7.24631995e-02 -9.05420303e-01
-1.40151930e+00 5.36173224e-01 4.46184516e-01 1.83253154e-01
1.72888386e+00 -4.19333965e-01 4.49027598e-01 1.64065987e-01
7.35747635e-01 -1.36281514e+00 -3.26516896e-01 5.62135637e-01
9.82470632e-01 -5.91131449e-01 2.80175984e-01 -4.39772010e-01
-4.09519523e-01 1.22508776e+00 6.31465614e-01 -4.13487181e-02
6.39473855e-01 5.59659600e-01 -1.89246401e-01 -2.48362124e-01
-6.87954247e-01 2.33226910e-01 1.26964450e-01 6.70028925e-01
3.00275743e-01 6.31192103e-02 -9.93018746e-01 6.83605552e-01
-1.82860926e-01 -4.67390567e-02 -4.08175550e-02 1.28602159e+00
-2.33682558e-01 -1.31934166e+00 -7.60950685e-01 1.37015581e-01
-9.36536565e-02 2.63864458e-01 -4.46829379e-01 3.98320526e-01
2.35149369e-01 1.02750385e+00 3.39643538e-01 -2.36462057e-01
1.01746079e-02 -3.93885851e-01 1.95700839e-01 -2.29578257e-01
-1.05126011e+00 -1.64074287e-01 -1.69708803e-01 -4.12966490e-01
-1.75165549e-01 -6.54446423e-01 -9.66159225e-01 -7.29011595e-01
-4.59989041e-01 8.59173238e-01 1.15526390e+00 5.87053955e-01
1.13321590e+00 6.11003697e-01 5.64237893e-01 -8.16865146e-01
-5.17046034e-01 -4.12810177e-01 -7.34116137e-01 -2.63322532e-01
-5.63418627e-01 -4.74215120e-01 -2.95162201e-01 -3.79921377e-01] | [5.765742301940918, 3.561239004135132] |
cd1bf0a6-5249-4410-9160-e083aa66a3c3 | effective-and-stable-role-based-multi-agent | 2304.00755 | null | https://arxiv.org/abs/2304.00755v1 | https://arxiv.org/pdf/2304.00755v1.pdf | Effective and Stable Role-Based Multi-Agent Collaboration by Structural Information Principles | Role-based learning is a promising approach to improving the performance of Multi-Agent Reinforcement Learning (MARL). Nevertheless, without manual assistance, current role-based methods cannot guarantee stably discovering a set of roles to effectively decompose a complex task, as they assume either a predefined role structure or practical experience for selecting hyperparameters. In this article, we propose a mathematical Structural Information principles-based Role Discovery method, namely SIRD, and then present a SIRD optimizing MARL framework, namely SR-MARL, for multi-agent collaboration. The SIRD transforms role discovery into a hierarchical action space clustering. Specifically, the SIRD consists of structuralization, sparsification, and optimization modules, where an optimal encoding tree is generated to perform abstracting to discover roles. The SIRD is agnostic to specific MARL algorithms and flexibly integrated with various value function factorization approaches. Empirical evaluations on the StarCraft II micromanagement benchmark demonstrate that, compared with state-of-the-art MARL algorithms, the SR-MARL framework improves the average test win rate by 0.17%, 6.08%, and 3.24%, and reduces the deviation by 16.67%, 30.80%, and 66.30%, under easy, hard, and super hard scenarios. | ['Angsheng Li', 'Hao Peng', 'Xianghua Zeng'] | 2023-04-03 | null | null | null | null | ['starcraft-ii', 'starcraft'] | ['playing-games', 'playing-games'] | [ 2.81399619e-02 1.73242956e-01 -5.73951781e-01 1.53907999e-01
-6.89148664e-01 -6.61172807e-01 6.46772444e-01 2.42271766e-01
-4.55802649e-01 1.15006161e+00 4.83820364e-02 -2.18517795e-01
-8.70090544e-01 -6.71600997e-01 -5.31045794e-01 -1.13351083e+00
-3.73531342e-01 8.14756691e-01 1.11254588e-01 -4.91777927e-01
2.46249750e-01 1.47107560e-02 -1.54039252e+00 1.99501023e-01
1.06103075e+00 7.92099297e-01 1.11632787e-01 3.87630939e-01
2.74177194e-01 1.19358563e+00 -1.01947582e+00 -5.29587120e-02
2.71108955e-01 -4.77846622e-01 -7.17235446e-01 1.67745560e-01
-5.66188931e-01 -3.11486214e-01 1.04742482e-01 6.48017645e-01
4.78889704e-01 3.51367474e-01 3.78682196e-01 -1.67356372e+00
-2.31666431e-01 1.28904104e+00 -8.81485879e-01 -1.23495057e-01
3.88605595e-01 3.09076086e-02 1.21132100e+00 -5.13850987e-01
6.00316882e-01 1.27409208e+00 1.30553797e-01 5.97042799e-01
-1.09537935e+00 -5.82263470e-01 6.40854061e-01 4.00060952e-01
-1.06956577e+00 -9.67642441e-02 7.94903159e-01 -2.28952318e-01
6.72732472e-01 3.13142478e-01 8.51209462e-01 9.23953652e-01
8.82912129e-02 9.71707225e-01 1.17972159e+00 -4.14073110e-01
6.19004667e-01 -2.81394005e-01 5.65866232e-02 8.20378661e-01
6.58391535e-01 3.89937987e-03 -7.90465415e-01 -4.40843463e-01
8.80312502e-01 -2.61640280e-01 8.15136284e-02 -6.65270686e-01
-1.36013401e+00 7.53335893e-01 -1.53803751e-01 1.87424675e-01
-4.46609944e-01 2.55993962e-01 1.85822487e-01 5.23284197e-01
2.25941584e-01 9.69131589e-01 -4.63976502e-01 -3.80828232e-01
-4.39834774e-01 3.86453629e-01 8.05166602e-01 5.09370387e-01
4.80312020e-01 1.48844421e-01 -1.06753603e-01 6.60672247e-01
4.84484345e-01 1.35792613e-01 3.43938470e-01 -1.51236260e+00
2.90509671e-01 8.48813057e-01 3.58684123e-01 -5.11428714e-01
-4.35250044e-01 -6.68800175e-01 -7.09206998e-01 4.31507140e-01
3.89124781e-01 -2.54511982e-01 -4.51406717e-01 1.61971152e+00
6.37441933e-01 1.36921898e-01 3.08936596e-01 8.73755217e-01
5.15866220e-01 3.97206366e-01 -8.94957557e-02 -6.31798446e-01
1.03093672e+00 -1.17327952e+00 -6.24805987e-01 -3.26780789e-02
4.29652274e-01 -2.77794540e-01 9.98131871e-01 8.32974017e-01
-1.18939507e+00 -2.31846750e-01 -1.29668033e+00 7.97513902e-01
2.29618121e-02 2.79829502e-02 1.05739367e+00 7.28874266e-01
-9.15709853e-01 3.49701226e-01 -8.23076427e-01 1.90736905e-01
1.34116173e-01 5.99910736e-01 -1.85385287e-01 -4.94898520e-02
-1.14655864e+00 5.26385307e-01 5.71973920e-01 -2.64066607e-01
-1.59002221e+00 -8.37524772e-01 -7.30226457e-01 4.86196950e-02
1.01273108e+00 -7.79853344e-01 1.19309998e+00 -7.12098718e-01
-1.89024365e+00 2.50387162e-01 2.83983022e-01 -5.56067884e-01
4.93431479e-01 -5.80235943e-02 -8.02850798e-02 3.68053108e-01
7.00391531e-02 2.88862228e-01 9.53965724e-01 -1.71527052e+00
-8.76898646e-01 -8.14384371e-02 8.73569787e-01 7.00400949e-01
-3.21440011e-01 -5.48131578e-03 -1.09787732e-01 -6.24658704e-01
-1.12923414e-01 -6.91060185e-01 -7.72323728e-01 -6.25160694e-01
3.98572162e-02 -4.75051790e-01 3.58855277e-01 -2.05088720e-01
1.22866452e+00 -1.87169862e+00 6.58638954e-01 3.01692039e-01
5.71719885e-01 -9.00923684e-02 -3.82173777e-01 3.92102867e-01
7.50068948e-02 1.65625066e-01 -1.49483696e-01 -1.93026379e-01
1.89522684e-01 3.05883318e-01 5.88569269e-02 4.60255563e-01
-1.22232020e-01 5.17494977e-01 -1.20193112e+00 -2.69853324e-01
1.65617153e-01 3.30706798e-02 -9.16909933e-01 3.16120952e-01
-3.56649786e-01 4.75342184e-01 -4.49676156e-01 9.02438045e-01
4.00088251e-01 -2.56725401e-01 8.99506390e-01 3.71552199e-01
2.97436099e-02 7.85720255e-03 -1.47232676e+00 1.73977542e+00
-3.31829041e-01 -1.64426610e-01 4.19962704e-01 -1.17236853e+00
7.51577854e-01 2.44945332e-01 1.18000281e+00 -6.75997138e-01
-4.21378389e-02 2.93155760e-02 3.56048226e-01 -8.00215527e-02
2.71896869e-01 1.38978720e-01 -3.68857116e-01 8.92439544e-01
-5.28263859e-02 1.60643473e-01 7.74741054e-01 2.20081285e-01
1.38146424e+00 9.95604098e-02 4.77712601e-01 -1.68014884e-01
7.53284574e-01 -9.22743157e-02 9.13955688e-01 9.60127175e-01
-1.03021096e-02 1.43680945e-02 8.04948747e-01 -3.86353225e-01
-4.89271939e-01 -1.02014339e+00 4.43466425e-01 1.62143159e+00
3.88807833e-01 -6.50332987e-01 -6.65719271e-01 -8.31034005e-01
1.63099781e-01 6.09791219e-01 -5.04423440e-01 -2.16376260e-01
-6.76430643e-01 -1.11227369e+00 3.66164774e-01 1.26747683e-01
4.83745039e-01 -9.39348936e-01 -7.36040413e-01 3.76290113e-01
-5.19089513e-02 -8.36036444e-01 -9.16257575e-02 3.10501218e-01
-6.05283141e-01 -1.40809345e+00 -2.01765820e-01 -4.73464966e-01
7.24414706e-01 2.98381269e-01 1.06598866e+00 2.36784786e-01
1.92308184e-02 5.58141947e-01 -8.06283414e-01 -9.99441966e-02
-4.00825113e-01 1.36532605e-01 3.50571901e-01 8.32574889e-02
-3.21078420e-01 -7.48596668e-01 -4.28167254e-01 5.34203172e-01
-6.93282962e-01 -2.07034513e-01 9.38557863e-01 1.04251575e+00
5.53167105e-01 5.68845809e-01 7.28254378e-01 -6.37302101e-01
8.98690045e-01 -5.52202404e-01 -5.35058379e-01 3.90803903e-01
-9.21112776e-01 1.33107752e-01 6.00596368e-01 -4.82400119e-01
-1.13680089e+00 -6.51919469e-02 2.45010898e-01 -3.20224673e-01
1.39507446e-02 6.28291190e-01 -4.35778767e-01 -2.62117069e-02
3.89921606e-01 3.63114387e-01 2.22275272e-01 -2.58733809e-01
3.05065423e-01 2.34522790e-01 5.44124022e-02 -1.18403566e+00
9.29059625e-01 3.28817159e-01 -1.58437774e-01 -4.66596335e-01
-7.33381033e-01 -2.88402021e-01 -3.19705427e-01 -4.46348161e-01
4.69023526e-01 -1.01665759e+00 -1.16611791e+00 4.10541862e-01
-6.66292250e-01 -6.68397009e-01 -3.15598935e-01 2.72471994e-01
-7.62598693e-01 2.67677248e-01 -4.67301041e-01 -8.65957737e-01
-1.63287967e-01 -1.04052627e+00 6.13703370e-01 1.13831311e-01
-4.16393280e-02 -7.33919203e-01 -4.33631390e-02 1.01467156e+00
1.73155695e-01 1.76648363e-01 1.00590253e+00 -8.06701243e-01
-6.47412539e-01 3.73692989e-01 4.87218827e-01 1.16402730e-01
1.54821843e-01 -2.32671157e-01 -2.87276536e-01 -6.83194816e-01
-8.45576599e-02 -5.42257488e-01 4.01286721e-01 3.06003690e-01
1.15668571e+00 -5.01326799e-01 -1.85614824e-02 2.97678500e-01
8.98073673e-01 6.72333121e-01 3.22592407e-01 7.19712913e-01
3.28470856e-01 4.00058985e-01 1.19990945e+00 1.12739551e+00
5.06936550e-01 5.87526500e-01 8.99025738e-01 1.85101390e-01
2.10822135e-01 -1.36852920e-01 7.63023973e-01 6.73346937e-01
-5.04738808e-01 -4.72388938e-02 -7.05583334e-01 2.21731752e-01
-2.38401222e+00 -9.74749446e-01 4.22912955e-01 2.00800395e+00
1.04055071e+00 2.95874774e-01 4.85124290e-01 4.80532587e-01
4.77910250e-01 2.22349614e-01 -7.40951419e-01 3.36754024e-02
3.42757627e-02 1.02339648e-01 1.71887815e-01 4.22846645e-01
-1.02811122e+00 8.69491041e-01 6.02119589e+00 9.35895681e-01
-4.73033607e-01 3.05610269e-01 4.36467081e-01 -2.21897677e-01
-2.13707134e-01 -2.01864522e-02 -6.76226497e-01 2.21257582e-01
6.19449139e-01 -4.40474451e-01 8.87433410e-01 8.68025124e-01
4.09923345e-01 -1.73948392e-01 -9.24002171e-01 9.55345213e-01
8.37338492e-02 -1.34787881e+00 1.13429837e-01 9.19586271e-02
6.91825807e-01 -6.47488236e-01 -5.33498637e-02 5.80933750e-01
8.98996592e-01 -1.17512012e+00 9.35121298e-01 2.04131305e-02
3.49377364e-01 -1.24304974e+00 7.18178511e-01 5.32712758e-01
-1.17334509e+00 -6.08279049e-01 -2.29700133e-01 -4.64069456e-01
-2.45133743e-01 4.40041035e-01 -7.70521700e-01 8.74086440e-01
5.35314441e-01 7.28186190e-01 -3.97122234e-01 8.19056094e-01
-6.05780721e-01 6.62274003e-01 8.86361524e-02 -1.06820844e-01
2.30654180e-01 -3.28796446e-01 6.76744401e-01 3.59659463e-01
-1.93415090e-01 1.56499103e-01 7.42362499e-01 4.45746988e-01
9.93113518e-02 -1.01279929e-01 -8.83180723e-02 -1.03560023e-01
8.14134061e-01 1.32126582e+00 -6.25979781e-01 7.76775628e-02
1.31224960e-01 5.13632417e-01 2.36228243e-01 3.51803333e-01
-1.18331397e+00 -9.21598524e-02 9.76277053e-01 2.33055912e-02
9.13898870e-02 -2.91436106e-01 -1.26847520e-01 -9.88790631e-01
-2.00398490e-01 -1.63460803e+00 6.89708054e-01 -3.17519695e-01
-1.00725400e+00 2.33117431e-01 1.74007237e-01 -1.24910665e+00
-4.05609936e-01 -4.25953895e-01 -4.00475800e-01 1.50414780e-01
-1.33231533e+00 -1.08143950e+00 -1.36005804e-01 5.49594522e-01
7.27162480e-01 -7.87849247e-01 6.44571006e-01 8.78397897e-02
-8.25670958e-01 6.14476144e-01 -2.26277281e-02 -2.32430100e-01
5.38414538e-01 -1.37515998e+00 -3.72363567e-01 6.59640312e-01
-1.35003090e-01 5.72417915e-01 5.98491549e-01 -3.93509895e-01
-1.87073088e+00 -9.79735613e-01 8.70097578e-02 -2.52917022e-01
8.07521522e-01 -3.65493715e-01 -3.26790988e-01 3.07533920e-01
2.68864840e-01 -2.63787001e-01 8.18009555e-01 2.39664778e-01
-1.30232498e-01 -4.17226583e-01 -1.05903196e+00 6.97981060e-01
1.26556158e+00 5.86433187e-02 -5.08777499e-01 2.08178878e-01
1.02867579e+00 -3.64563733e-01 -1.15436208e+00 2.68403709e-01
3.32949698e-01 -9.17576730e-01 1.22165239e+00 -6.74104214e-01
4.46642578e-01 -5.07326663e-01 -1.51680326e-02 -1.46663058e+00
-6.08804584e-01 -1.04498839e+00 -5.46748579e-01 9.72029507e-01
4.28825974e-01 -5.83762467e-01 9.07323897e-01 7.47492537e-02
-7.38936514e-02 -9.97229099e-01 -1.03834105e+00 -8.95136654e-01
-1.27679288e-01 -1.28441229e-01 6.82272255e-01 1.06680715e+00
9.58414972e-02 4.28434253e-01 -4.70194280e-01 1.81469277e-01
9.87463832e-01 2.56075293e-01 7.58813024e-01 -1.23200059e+00
-9.08081293e-01 -6.09801710e-01 2.19541341e-01 -7.34325767e-01
3.68002594e-01 -6.90271616e-01 -2.65667528e-01 -1.62823260e+00
2.07701817e-01 -6.69931650e-01 -7.35093057e-01 8.27625513e-01
-1.57968417e-01 -4.00604665e-01 3.22082281e-01 1.16780899e-01
-1.23017430e+00 7.35316813e-01 1.32099819e+00 -3.74987274e-01
-4.99919921e-01 1.30156159e-01 -1.03401804e+00 7.19334066e-01
1.06692386e+00 -4.34280276e-01 -9.56790209e-01 1.55151158e-03
3.60245705e-01 3.48227233e-01 7.47712329e-02 -8.06456149e-01
1.73099011e-01 -7.46440589e-01 1.00319766e-01 -3.18557799e-01
2.80475348e-01 -5.90984464e-01 1.24429531e-01 8.33079875e-01
-3.91661912e-01 1.89532395e-02 -1.89461321e-01 6.78493917e-01
-1.16694719e-01 -1.97426647e-01 3.54401022e-01 -3.07588100e-01
-8.94347608e-01 -6.71647489e-02 -5.57039976e-01 8.63055885e-02
1.47587276e+00 -1.69970945e-01 -4.12972629e-01 -3.22627187e-01
-5.68319798e-01 7.74404764e-01 1.53057843e-01 4.14651364e-01
6.82801306e-01 -1.16460574e+00 -7.66131401e-01 -9.31781083e-02
-1.75188795e-01 1.01844266e-01 3.88277881e-02 6.90639794e-01
-3.77306528e-02 1.59684345e-01 -4.40043569e-01 -1.29207298e-01
-1.50032794e+00 3.39336932e-01 2.35615551e-01 -9.50010538e-01
-1.39333159e-01 6.28332198e-01 -8.88444707e-02 -5.31948030e-01
4.00879025e-01 -6.45716395e-03 -4.86817271e-01 4.02074426e-01
3.76758963e-01 6.40352547e-01 -1.51214898e-01 1.05174832e-01
-5.11273384e-01 1.54514387e-01 1.19763212e-02 -1.99338302e-01
1.56471741e+00 1.48011595e-01 -3.61194819e-01 1.14394523e-01
6.17080219e-02 6.86909258e-02 -1.29101741e+00 5.20490818e-02
1.49215490e-01 -1.90244481e-01 -9.00778919e-02 -1.02604544e+00
-1.04614401e+00 1.25744671e-01 2.08551753e-02 2.55961061e-01
1.22187471e+00 2.02638586e-03 1.25922993e-01 6.62983119e-01
9.06807780e-01 -1.12416112e+00 7.28101194e-01 5.75974762e-01
1.00940239e+00 -7.61696756e-01 2.40866989e-01 -3.99449110e-01
-8.88660610e-01 7.76221812e-01 9.64772642e-01 1.29662856e-01
2.76047647e-01 2.78710932e-01 -4.07783449e-01 -7.77081028e-02
-1.31685424e+00 -1.66031405e-01 -1.55242622e-01 7.18484879e-01
-3.87566537e-02 1.75711319e-01 -5.69618702e-01 1.09999764e+00
-3.33267227e-02 -4.22615528e-01 7.56188631e-01 1.09900665e+00
-6.17485940e-01 -1.50388300e+00 -5.68913758e-01 2.98667610e-01
-1.85795262e-01 2.36853689e-01 -3.65915686e-01 8.05595398e-01
3.34793955e-01 1.26263928e+00 -2.26508215e-01 -5.38596451e-01
2.38856226e-01 -1.99006602e-01 5.45711994e-01 -7.10542977e-01
-7.42430925e-01 1.69370517e-01 4.22314972e-01 -6.27383828e-01
-6.72113836e-01 -5.52467406e-01 -1.44330132e+00 -1.56477392e-01
1.69495091e-01 5.38405776e-01 1.21942505e-01 8.00591290e-01
3.34920138e-01 8.86268258e-01 8.37842345e-01 -5.62004983e-01
-8.61289680e-01 -6.73532248e-01 -5.81002176e-01 -1.66196246e-02
-8.87082443e-02 -1.13931406e+00 -3.02061170e-01 -2.89849758e-01] | [3.738581657409668, 1.931173324584961] |
2503dcb6-d5d1-4aa9-9689-21f6493094a8 | transfool-an-adversarial-attack-against | 2302.00944 | null | https://arxiv.org/abs/2302.00944v2 | https://arxiv.org/pdf/2302.00944v2.pdf | TransFool: An Adversarial Attack against Neural Machine Translation Models | Deep neural networks have been shown to be vulnerable to small perturbations of their inputs, known as adversarial attacks. In this paper, we investigate the vulnerability of Neural Machine Translation (NMT) models to adversarial attacks and propose a new attack algorithm called TransFool. To fool NMT models, TransFool builds on a multi-term optimization problem and a gradient projection step. By integrating the embedding representation of a language model, we generate fluent adversarial examples in the source language that maintain a high level of semantic similarity with the clean samples. Experimental results demonstrate that, for different translation tasks and NMT architectures, our white-box attack can severely degrade the translation quality while the semantic similarity between the original and the adversarial sentences stays high. Moreover, we show that TransFool is transferable to unknown target models. Finally, based on automatic and human evaluations, TransFool leads to improvement in terms of success rate, semantic similarity, and fluency compared to the existing attacks both in white-box and black-box settings. Thus, TransFool permits us to better characterize the vulnerability of NMT models and outlines the necessity to design strong defense mechanisms and more robust NMT systems for real-life applications. | ['Pascal Frossard', 'Ljiljana Dolamic', 'Sahar Sadrizadeh'] | 2023-02-02 | null | null | null | null | ['nmt', 'semantic-textual-similarity'] | ['computer-code', 'natural-language-processing'] | [ 3.40913564e-01 6.80600628e-02 1.60308748e-01 -1.06134340e-01
-8.54278564e-01 -1.04357255e+00 9.08717036e-01 -3.76426995e-01
-3.11655521e-01 6.04711950e-01 -1.11763403e-01 -6.31348431e-01
3.78201991e-01 -7.13354826e-01 -1.10379577e+00 -5.21618426e-01
2.00394571e-01 5.28500915e-01 -2.55908996e-01 -5.92576087e-01
-2.33414412e-01 5.37394702e-01 -6.98727489e-01 3.12518924e-01
1.03184700e+00 5.49840212e-01 -3.44193399e-01 4.43129897e-01
9.42284390e-02 5.02777934e-01 -9.66076374e-01 -1.12170529e+00
7.20049739e-01 -5.30047476e-01 -6.85365915e-01 -4.42304105e-01
7.01748192e-01 -1.95791245e-01 -4.65081006e-01 1.53366292e+00
5.62153935e-01 -6.33586943e-02 5.57901025e-01 -1.29952705e+00
-1.18129992e+00 8.64979506e-01 -4.85670865e-02 -7.80097321e-02
1.24835908e-01 5.43308675e-01 6.51938021e-01 -9.28150117e-01
6.32303715e-01 1.62940383e+00 6.47595644e-01 1.02630424e+00
-1.45067513e+00 -9.37312126e-01 4.42170948e-02 -2.46169847e-02
-1.03669775e+00 -5.22628546e-01 6.13588572e-01 -1.72369197e-01
8.66232455e-01 5.07156372e-01 9.28304717e-02 2.02179670e+00
5.82046747e-01 6.77615941e-01 1.24897063e+00 -4.01118129e-01
2.03425676e-01 3.80161911e-01 -1.72428638e-01 5.29862940e-01
3.69831733e-02 4.30550754e-01 -1.90390006e-01 -2.98859060e-01
2.99956888e-01 -3.91354680e-01 -3.07182014e-01 -2.16153502e-01
-1.16464829e+00 8.66697729e-01 4.13249880e-01 4.35959458e-01
-9.08871368e-02 1.29140973e-01 5.37332118e-01 7.73723006e-01
5.62326014e-01 9.14523661e-01 -3.42264086e-01 1.57260165e-01
-6.12057328e-01 6.82832524e-02 7.98573077e-01 8.07636321e-01
4.06677872e-01 4.16419715e-01 -3.04141134e-01 6.14078343e-01
-1.36121780e-01 1.06899607e+00 6.22412264e-01 -5.39899468e-01
7.45381415e-01 2.03077570e-01 -1.10661693e-01 -1.07302177e+00
5.47936792e-03 -6.58708453e-01 -8.71266603e-01 3.31582427e-01
2.79646933e-01 -4.45586085e-01 -9.17942524e-01 2.25107288e+00
1.47031378e-02 5.51838353e-02 4.62199241e-01 7.87502587e-01
2.62925059e-01 7.14287877e-01 -1.05747737e-01 -4.80143465e-02
9.34038520e-01 -1.01358426e+00 -7.75616884e-01 -5.72774291e-01
6.29198730e-01 -9.17598844e-01 1.31833494e+00 1.51558504e-01
-1.08952487e+00 -4.14153665e-01 -1.10719299e+00 2.85228282e-01
-5.23020029e-01 -8.94100815e-02 1.63609013e-01 1.07518101e+00
-9.89556551e-01 8.23281467e-01 -7.84284770e-01 -2.07846373e-01
3.84252399e-01 3.96477938e-01 -5.28304040e-01 -1.00035116e-01
-1.78581583e+00 1.44020391e+00 1.67084649e-01 1.51999921e-01
-1.27156568e+00 -5.75318635e-01 -7.43467867e-01 -4.47174460e-02
8.65765288e-02 -7.51010120e-01 1.13698566e+00 -1.26468205e+00
-1.74693489e+00 7.34915912e-01 1.11595683e-01 -7.50235200e-01
9.43575323e-01 -3.81452382e-01 -5.25834978e-01 -4.43064831e-02
-1.24902084e-01 3.87739927e-01 1.23131704e+00 -1.08703864e+00
8.16876218e-02 -2.09307373e-01 3.12689245e-02 -2.01364234e-02
-8.17519009e-01 4.02326018e-01 -1.94990560e-02 -9.49572802e-01
-4.37188447e-01 -1.33909810e+00 -1.99374884e-01 -2.50261068e-01
-7.24067867e-01 2.66559720e-01 7.51690805e-01 -6.78761780e-01
8.17013979e-01 -2.23869205e+00 5.02768695e-01 1.53727487e-01
2.78915074e-02 6.97129071e-01 -6.29402101e-01 5.10887504e-01
-2.80145854e-01 4.18319017e-01 -3.23434144e-01 -3.15011829e-01
2.27033466e-01 1.81404471e-01 -8.50588918e-01 4.11366999e-01
2.96055585e-01 1.22830629e+00 -7.70374835e-01 2.68478319e-02
-6.70048296e-02 5.16358674e-01 -2.95747489e-01 2.80687481e-01
-2.29302794e-01 4.15653348e-01 -2.73394555e-01 2.73892075e-01
6.37271285e-01 2.83412427e-01 -4.58900928e-02 1.44479796e-01
4.57729101e-01 2.03953430e-01 -3.17541063e-01 1.43334556e+00
-5.69703817e-01 7.48853326e-01 -6.70004115e-02 -7.93679655e-01
9.71367002e-01 4.62679446e-01 -1.43550560e-01 -6.27341688e-01
4.53174353e-01 4.08912539e-01 2.06671536e-01 -1.30129740e-01
2.70045310e-01 -3.06047082e-01 -2.12753832e-01 6.52162611e-01
9.08270776e-02 -1.69593155e-01 -1.32207245e-01 1.20542340e-01
1.05622113e+00 -1.51372626e-01 -3.06929559e-01 -2.20125437e-01
4.83248413e-01 -2.12219521e-01 3.61573726e-01 8.03046167e-01
-2.07308218e-01 3.75844806e-01 4.28874344e-01 -3.60478133e-01
-1.10087574e+00 -1.18221939e+00 1.98370039e-01 8.40440392e-01
-5.58369048e-02 -2.62827963e-01 -1.32112861e+00 -1.02078593e+00
-1.40946418e-01 1.00786972e+00 -7.60687888e-01 -9.52743411e-01
-8.87412429e-01 -7.24537611e-01 1.23217118e+00 2.38568231e-01
4.83488292e-01 -1.03427720e+00 -7.42341671e-03 -4.40559760e-02
-2.44463116e-01 -1.40306365e+00 -6.65502310e-01 4.39175516e-02
-7.08694279e-01 -5.76608121e-01 -5.92118680e-01 -7.37974048e-01
7.01658428e-01 -1.05390646e-01 9.98789430e-01 -1.92187175e-01
1.98115140e-01 -1.20740108e-01 -2.22255111e-01 -3.34663898e-01
-1.27113795e+00 1.58265531e-01 5.11992276e-01 1.27063379e-01
1.40720263e-01 -7.09833801e-01 -6.50982112e-02 4.26889241e-01
-1.06738842e+00 -1.13468230e-01 6.38731599e-01 7.46607900e-01
1.10226251e-01 1.77109279e-02 4.30169582e-01 -8.00449491e-01
1.05267656e+00 -2.53636420e-01 -5.38878918e-01 4.15126175e-01
-6.10456824e-01 1.78647086e-01 1.24447477e+00 -1.05064511e+00
-6.54749274e-01 -3.49075019e-01 -2.65778929e-01 -8.80566537e-01
1.13241650e-01 2.05304757e-01 -4.07209188e-01 -2.61490792e-01
1.09606338e+00 3.85116160e-01 2.22285185e-02 -4.49374706e-01
5.91788113e-01 3.93817365e-01 5.56715429e-01 -7.32424974e-01
1.60693908e+00 1.31676182e-01 -2.43275717e-01 -3.76540333e-01
-5.64653754e-01 6.40072346e-01 -3.85077059e-01 8.63660723e-02
6.44553721e-01 -5.93949139e-01 -2.82407701e-01 7.18280911e-01
-1.33976495e+00 -4.24384087e-01 5.23601519e-03 2.15719476e-01
-4.80407536e-01 3.54162991e-01 -8.68134320e-01 -4.53148752e-01
-7.45359898e-01 -1.43099999e+00 6.83325708e-01 -1.93356276e-01
-2.71363586e-01 -9.94230688e-01 1.11014791e-01 3.32758963e-01
7.31694043e-01 2.75133103e-01 1.27764416e+00 -1.08663726e+00
-3.33573878e-01 -4.57265735e-01 1.54045925e-01 8.66728961e-01
5.60808554e-02 -1.03073291e-01 -9.41650033e-01 -6.23484373e-01
3.83051217e-01 -3.79536599e-01 5.76501787e-01 -1.69257760e-01
6.25898898e-01 -8.19674194e-01 -1.54748820e-02 8.10159802e-01
1.05032134e+00 1.57729499e-02 5.70300341e-01 4.47052926e-01
7.72485077e-01 4.91104424e-01 2.67847419e-01 -4.10752028e-01
-4.30114001e-01 7.24706233e-01 4.91472989e-01 -1.69701546e-01
1.50336465e-02 -3.59354526e-01 1.03470004e+00 8.41594398e-01
4.66645539e-01 -3.58257979e-01 -8.22245598e-01 2.76788414e-01
-1.57385790e+00 -9.05740082e-01 2.26392984e-01 2.03298235e+00
9.59053278e-01 4.87458140e-01 -1.45879194e-01 -1.17903300e-01
8.47492456e-01 8.11009705e-02 -5.15347183e-01 -9.15485263e-01
-3.54867280e-01 3.64094228e-01 4.00846332e-01 5.84552228e-01
-1.01075482e+00 1.43073785e+00 6.74058723e+00 9.77090180e-01
-1.32362580e+00 3.38108450e-01 5.06691754e-01 -1.43820181e-01
-3.59168708e-01 -2.23706871e-01 -3.65179658e-01 4.65106845e-01
1.25668502e+00 -4.92514253e-01 7.85975695e-01 7.49657691e-01
5.79082444e-02 9.70360041e-01 -1.16908944e+00 4.30434704e-01
1.48380980e-01 -1.06977177e+00 3.58241916e-01 9.89036709e-02
7.09310412e-01 2.05620527e-01 5.76138496e-01 5.04101753e-01
6.01869524e-01 -1.23108232e+00 7.80093193e-01 2.61681471e-02
7.73611605e-01 -9.42965388e-01 7.38162696e-01 4.74095017e-01
-3.73867512e-01 2.70340536e-02 -4.14597422e-01 8.92144814e-02
-4.66297232e-02 3.72043163e-01 -8.62778544e-01 5.12621284e-01
2.74930179e-01 2.53125340e-01 -6.38756096e-01 2.26036534e-01
-5.97223878e-01 6.02270305e-01 -9.54471156e-02 2.17377525e-02
4.19086486e-01 -2.18731493e-01 8.54133725e-01 1.16326284e+00
2.16729835e-01 -4.95203555e-01 -8.18811953e-02 1.10726142e+00
-4.29343134e-01 1.40216082e-01 -9.55178261e-01 -3.46073210e-01
4.30775702e-01 9.34443474e-01 -1.50371552e-01 -2.22726703e-01
2.47585215e-02 1.42625868e+00 4.60333228e-01 5.21599114e-01
-1.05114591e+00 -3.41564715e-01 1.17047322e+00 -3.81512552e-01
-5.00567593e-02 -1.08884029e-01 -3.48241508e-01 -1.39039552e+00
2.59365469e-01 -1.68187773e+00 -1.81044847e-01 -5.71499944e-01
-1.34833682e+00 1.37303579e+00 -3.16579401e-01 -1.06997013e+00
-3.58155549e-01 -6.95968747e-01 -6.92518771e-01 1.10317218e+00
-1.07894731e+00 -1.29290593e+00 4.09491271e-01 6.95001423e-01
3.76621544e-01 -5.95495939e-01 1.21311867e+00 1.02663711e-01
-8.68053675e-01 1.27133703e+00 2.70663530e-01 5.20229101e-01
6.73050702e-01 -9.68713224e-01 1.19986987e+00 1.33698380e+00
4.15391028e-01 9.05984044e-01 9.64221418e-01 -5.62694132e-01
-1.34381771e+00 -1.37573290e+00 8.32135916e-01 -8.65144372e-01
8.36994886e-01 -7.64678419e-01 -9.47710991e-01 7.43154466e-01
4.31447774e-01 -1.76971376e-01 5.22873819e-01 -1.59074530e-01
-9.08065856e-01 1.51726276e-01 -1.22084641e+00 1.10539973e+00
8.21383953e-01 -9.51602876e-01 -7.15010822e-01 5.24532795e-01
1.17777944e+00 -3.12400877e-01 -6.77479088e-01 3.94737899e-01
4.11321521e-01 -6.52936816e-01 1.04599833e+00 -1.09650636e+00
4.89617109e-01 3.11378986e-02 -2.22661689e-01 -1.79054010e+00
-1.40829742e-01 -1.06589866e+00 -1.33139780e-03 1.01442516e+00
7.33561695e-01 -8.71799350e-01 4.27727282e-01 5.45404971e-01
-5.01975380e-02 -5.20038426e-01 -1.03108275e+00 -1.24235809e+00
9.66313541e-01 -3.84322107e-01 6.70473576e-01 1.20471358e+00
-1.62038788e-01 4.06498015e-01 -7.37367213e-01 3.55919063e-01
5.74806988e-01 -1.33332312e-01 8.02863061e-01 -7.08858907e-01
-3.87766540e-01 -5.61082423e-01 -2.27136329e-01 -4.79121298e-01
7.49833286e-01 -1.19591880e+00 -1.62527740e-01 -8.26759100e-01
-7.40838721e-02 -3.82112786e-02 -3.17838371e-01 5.90704799e-01
-3.66825163e-01 5.42139292e-01 4.99990642e-01 2.30527535e-01
4.26005460e-02 6.57593250e-01 9.93148208e-01 -4.89038914e-01
1.18068241e-01 4.45664376e-02 -6.70747697e-01 4.85218287e-01
9.70206201e-01 -8.12342227e-01 -3.36293459e-01 -8.49820495e-01
7.54164085e-02 -3.28133315e-01 1.56122535e-01 -8.19671452e-01
-1.98104054e-01 -1.62701547e-01 3.91148441e-02 1.77501783e-01
3.76367539e-01 -8.46402884e-01 1.74138304e-02 7.22621441e-01
-5.95357418e-01 2.58918792e-01 3.74408960e-01 2.66522139e-01
1.48561662e-02 -1.86665043e-01 9.17263746e-01 3.38561460e-02
1.40634283e-01 2.72228003e-01 -3.64836454e-01 5.03306091e-02
8.10301542e-01 1.84411511e-01 -4.80065048e-01 -3.86096209e-01
-6.24276638e-01 2.60159541e-02 6.67412579e-01 8.33534002e-01
5.30529320e-01 -1.48168945e+00 -1.04758954e+00 4.12683010e-01
2.02335101e-02 -7.09993601e-01 -1.11322820e-01 5.72749734e-01
-3.45880300e-01 3.69708896e-01 -3.68230194e-01 -1.87719926e-01
-1.25665486e+00 1.02025485e+00 6.49098158e-01 -3.31825823e-01
-5.03378332e-01 7.22949147e-01 3.37694764e-01 -8.01146448e-01
1.43606365e-01 -2.37846449e-02 2.93856204e-01 -4.33298200e-01
4.21943754e-01 -4.06691879e-02 2.19482660e-01 -6.67155504e-01
-2.50570476e-01 2.07741633e-01 -2.57371843e-01 -2.47497350e-01
9.36438084e-01 2.80102789e-01 -1.39527142e-01 1.28155146e-02
1.37692332e+00 1.96587667e-01 -7.70197690e-01 -3.33262056e-01
-1.28350854e-01 -4.91419077e-01 -2.22730339e-01 -1.01915300e+00
-1.02185464e+00 1.05012012e+00 5.03093243e-01 1.92301691e-01
9.81792748e-01 -5.03945112e-01 9.91342366e-01 6.59012079e-01
2.78267026e-01 -5.91296554e-01 1.85643155e-02 6.44792259e-01
1.05551243e+00 -9.04051781e-01 -6.39406085e-01 -1.95386514e-01
-5.71334064e-01 7.05797732e-01 4.97305810e-01 -1.63457751e-01
1.52638048e-01 2.87741989e-01 5.99279225e-01 2.80143619e-01
-8.06977868e-01 3.66862953e-01 2.00863808e-01 7.56354928e-01
-1.51113449e-02 3.24698500e-02 -4.18755412e-02 3.64462554e-01
-5.78239918e-01 -5.84275305e-01 4.41253543e-01 4.64039236e-01
1.78514421e-02 -1.48536313e+00 -5.50848901e-01 -2.21666217e-01
-7.33948112e-01 -3.55331302e-01 -9.91597414e-01 6.28711522e-01
-2.68735617e-01 9.69774067e-01 -5.10732293e-01 -8.86889040e-01
4.36937869e-01 4.33368802e-01 4.02474523e-01 -4.43124622e-01
-1.04659534e+00 -2.92525858e-01 1.67166099e-01 -5.15720010e-01
2.47308508e-01 -2.83077806e-01 -6.60515606e-01 -4.93843764e-01
-2.12683573e-01 2.35499457e-01 7.29270816e-01 1.04278362e+00
3.58609170e-01 3.80450517e-01 9.68178928e-01 -6.29953027e-01
-1.36060655e+00 -1.09520471e+00 4.43139896e-02 7.10474610e-01
3.20805877e-01 -2.21943274e-01 -5.75466692e-01 -2.73852825e-01] | [6.019287586212158, 8.144308090209961] |
6c3104c4-02c7-4bf9-acc3-5f9e03daa974 | lea-meta-knowledge-driven-self-attentive | null | null | https://aclanthology.org/2022.naacl-main.7 | https://aclanthology.org/2022.naacl-main.7.pdf | LEA: Meta Knowledge-Driven Self-Attentive Document Embedding for Few-Shot Text Classification | Text classification has achieved great success with the prosperity of deep learning and pre-trained language models. However, we often encounter labeled data deficiency problems in real-world text-classification tasks. To overcome such challenging scenarios, interest in few-shot learning has increased, whereas most few-shot text classification studies suffer from a difficulty of utilizing pre-trained language models. In the study, we propose a novel learning method for learning how to attend, called LEA, through which meta-level attention aspects are derived based on our meta-learning strategy. This enables the generation of task-specific document embedding with leveraging pre-trained language models even though a few labeled data instances are given. We evaluate our proposed learning method on five benchmark datasets. The results show that the novel method robustly provides the competitive performance compared to recent few-shot learning methods for all the datasets. | ['Tae Young Jang', 'S. K. Hong'] | null | null | null | null | naacl-2022-7 | ['document-embedding', 'few-shot-text-classification'] | ['methodology', 'natural-language-processing'] | [ 2.77924538e-01 -2.26606414e-01 -4.37607765e-01 -3.84170055e-01
-6.92266762e-01 1.46892503e-01 1.02010584e+00 4.32357699e-01
-7.34407306e-01 5.50110936e-01 4.59853441e-01 5.53356670e-02
-1.41202947e-02 -8.41347694e-01 -1.22544900e-01 -5.20493746e-01
4.44689900e-01 3.44034433e-01 2.30178192e-01 -4.37008977e-01
4.86414313e-01 -3.42672765e-01 -1.59389472e+00 2.37580717e-01
8.06518555e-01 7.85928011e-01 1.95774332e-01 4.51235920e-01
-8.49887788e-01 7.59662509e-01 -3.99681211e-01 -4.47339773e-01
-1.07108109e-01 -5.15192509e-01 -6.76731944e-01 -7.11057186e-02
8.81339088e-02 -2.12671831e-01 -4.65673894e-01 9.82636511e-01
6.60235882e-01 6.53825581e-01 7.64145434e-01 -1.18312764e+00
-8.87098491e-01 6.56025052e-01 -6.07435405e-01 3.86011273e-01
-3.35606076e-02 4.52666506e-02 1.20973301e+00 -1.27084827e+00
5.06851733e-01 1.04010081e+00 4.82522100e-01 8.11851740e-01
-9.85222399e-01 -5.03029108e-01 1.95958883e-01 3.76435518e-01
-1.12134051e+00 -4.26123321e-01 8.71823251e-01 -4.00578618e-01
1.06698442e+00 -4.09041882e-01 3.58442575e-01 1.20174825e+00
4.33171988e-01 8.11259985e-01 6.63895130e-01 -7.58168757e-01
4.49258626e-01 3.51197094e-01 7.09925711e-01 6.82215750e-01
3.57656330e-01 -2.60373831e-01 -5.58689833e-01 -7.03093931e-02
9.16963965e-02 7.04644859e-01 -1.35691166e-02 -3.70857745e-01
-1.01045203e+00 1.03948629e+00 1.74911425e-01 6.91214740e-01
-1.46570981e-01 -4.21570381e-03 8.89456153e-01 1.96679026e-01
8.91772985e-01 2.34076381e-01 -3.19461435e-01 -1.85288891e-01
-9.30769145e-01 -1.55172348e-01 5.33303857e-01 9.13985848e-01
7.65538692e-01 2.30136916e-01 -5.90424776e-01 1.23450923e+00
1.67811990e-01 -2.43616421e-02 1.15496433e+00 -3.39164808e-02
6.81889892e-01 7.16273248e-01 -8.21194351e-02 -9.05921996e-01
-2.71182299e-01 -4.08654988e-01 -7.65551269e-01 -9.75567997e-02
-5.94478771e-02 -2.99903154e-01 -1.10627079e+00 1.45914662e+00
8.24045166e-02 2.86999792e-01 3.20754498e-01 5.02383828e-01
8.36604357e-01 1.06758428e+00 5.11881828e-01 -3.12780857e-01
1.35683239e+00 -1.29000700e+00 -1.06157362e+00 -4.80868578e-01
1.06004441e+00 -4.82256055e-01 1.32145917e+00 -2.68997736e-02
-4.54104424e-01 -6.41940892e-01 -1.15385485e+00 -2.66904384e-01
-7.99583018e-01 -6.16370551e-02 5.25447369e-01 7.07599580e-01
-4.20514822e-01 3.93194139e-01 -4.75164682e-01 -8.34430218e-01
6.64086103e-01 -8.58169608e-03 -1.61545590e-01 -2.95902967e-01
-1.32868183e+00 9.21446323e-01 6.32087111e-01 -3.22752416e-01
-8.62539411e-01 -6.46367550e-01 -1.00179422e+00 4.27108735e-01
5.81366122e-01 -4.53323871e-01 1.13008523e+00 -7.23496616e-01
-1.39683878e+00 7.85907507e-01 -1.87469035e-01 -3.56902599e-01
2.64152557e-01 -2.86073178e-01 -5.09118497e-01 -1.57898143e-01
7.10785985e-02 4.33511168e-01 9.14054155e-01 -8.70995283e-01
-6.34604454e-01 -3.38879347e-01 -5.44569902e-02 1.41819403e-01
-1.30819535e+00 -1.21853411e-01 -2.96764493e-01 -7.37332702e-01
-5.19166410e-01 -3.81817579e-01 -2.57587135e-01 1.17883101e-01
-2.65928041e-02 -5.84801197e-01 1.13850820e+00 -2.34686568e-01
1.43298638e+00 -1.98058569e+00 -1.40107915e-01 -4.87273484e-01
1.12114854e-01 7.49710441e-01 -2.63170421e-01 7.53924787e-01
8.65307748e-02 1.29464222e-02 -2.63403356e-01 -4.79723126e-01
1.00713568e-02 2.19658744e-02 -5.47271550e-01 1.63987920e-01
5.49962372e-02 9.40283597e-01 -1.15035474e+00 -6.77095413e-01
5.13640344e-01 3.64850372e-01 -4.06838655e-01 2.70365298e-01
-3.51411939e-01 -1.91124350e-01 -5.62221587e-01 3.80600661e-01
3.74637306e-01 -3.17700058e-01 -1.86855167e-01 5.92579469e-02
3.63408141e-02 2.48931441e-02 -8.86802733e-01 2.03474069e+00
-7.95776904e-01 5.88749230e-01 -5.13922095e-01 -1.43690205e+00
9.63207483e-01 4.43852365e-01 2.35041797e-01 -5.67419410e-01
5.38082778e-01 -6.27933368e-02 -6.90251067e-02 -7.42409348e-01
4.68759984e-01 -5.66146493e-01 -7.22740293e-02 8.19341242e-01
6.41878545e-01 1.39513656e-01 3.07366759e-01 4.40524876e-01
9.34360147e-01 8.04444123e-03 5.16053438e-01 -1.39489993e-01
5.60781240e-01 5.14740311e-02 3.47614676e-01 6.85248792e-01
-4.95628268e-01 3.28909189e-01 2.06257850e-01 -4.81003672e-01
-1.07112050e+00 -4.31420922e-01 -9.11559016e-02 1.60180390e+00
2.71856170e-02 -6.36336625e-01 -5.13101101e-01 -8.38648081e-01
-1.89053267e-01 1.05504990e+00 -9.54381287e-01 -6.35096669e-01
-5.20677790e-02 -1.01014435e+00 2.58026004e-01 5.94431400e-01
5.20406663e-01 -1.09191728e+00 -4.82265949e-01 3.94313812e-01
2.21338764e-01 -9.43064153e-01 -6.75104618e-01 2.80901015e-01
-9.32501197e-01 -7.83424020e-01 -9.91626620e-01 -1.03716385e+00
6.23527110e-01 8.24660480e-01 6.58165216e-01 1.74600318e-01
-6.05923414e-01 3.73510897e-01 -7.11893141e-01 -5.24356484e-01
-1.54013500e-01 2.49505356e-01 -4.43524309e-02 1.58429354e-01
1.05860150e+00 -3.92698437e-01 -3.03343832e-01 -9.18561742e-02
-9.54233766e-01 1.21351350e-02 3.04493397e-01 1.28481030e+00
9.30211991e-02 1.29681081e-01 1.00251985e+00 -1.15345907e+00
9.22845542e-01 -8.67264509e-01 -2.18900889e-01 3.81593764e-01
-9.40213859e-01 2.39139646e-01 8.87382448e-01 -5.03212214e-01
-1.30793715e+00 -3.88808310e-01 1.12474374e-01 -2.63503850e-01
-1.24803491e-01 6.73700035e-01 9.98816937e-02 3.03341031e-01
7.29772568e-01 3.90656620e-01 -1.72519058e-01 -3.69061410e-01
4.86530870e-01 1.01484132e+00 -1.59122318e-01 -4.96678472e-01
7.24207163e-01 4.86311793e-01 -3.81406814e-01 -9.34069157e-01
-1.24179685e+00 -8.08296382e-01 -6.35916173e-01 -6.15456253e-02
8.34664583e-01 -7.69526720e-01 -3.11621837e-02 3.27298194e-01
-1.02895284e+00 1.02994684e-02 -2.71483660e-01 5.28720975e-01
-3.40216875e-01 3.97670150e-01 -4.57103252e-01 -8.93869281e-01
-6.43509448e-01 -7.64558434e-01 7.73238003e-01 2.76940942e-01
1.27131110e-02 -1.35101676e+00 4.38557923e-01 1.77235156e-01
5.69843054e-01 -2.43918300e-01 1.10759234e+00 -1.06057215e+00
1.56289022e-02 -3.72611970e-01 -2.26057157e-01 1.65820166e-01
4.97859448e-01 -2.34801307e-01 -1.13704300e+00 -3.87861073e-01
-9.38025564e-02 -8.05647075e-01 1.12898815e+00 4.97351866e-04
1.06747544e+00 -9.40992031e-03 -3.68880540e-01 2.40841165e-01
1.60675216e+00 2.51075476e-01 2.81523287e-01 2.49465272e-01
6.69014454e-01 6.86639190e-01 6.52343750e-01 5.58093965e-01
1.38308972e-01 4.39795792e-01 1.65770706e-02 2.59008795e-01
7.77207594e-03 -4.85524982e-01 5.62497415e-02 1.08765137e+00
1.67956010e-01 -2.38768980e-01 -1.06129527e+00 7.56597459e-01
-2.07725191e+00 -1.11712754e+00 3.38410139e-01 1.93033457e+00
8.55276406e-01 3.73632163e-01 -1.39177874e-01 1.38444856e-01
8.16256225e-01 5.01913548e-01 -6.10743642e-01 -3.16222161e-01
2.83172250e-01 2.84468472e-01 -4.27249186e-02 3.10110360e-01
-1.12832201e+00 1.16713703e+00 5.47816324e+00 9.60744619e-01
-1.05903423e+00 3.89631599e-01 4.06323642e-01 -9.81753841e-02
-1.35550484e-01 -6.67525083e-02 -1.03003097e+00 4.61006254e-01
1.09365249e+00 -7.57767141e-01 -1.22193478e-01 1.10220349e+00
7.25986585e-02 2.03545690e-01 -9.42457557e-01 9.88278568e-01
6.38844013e-01 -1.41069353e+00 4.83491361e-01 -2.67406821e-01
8.49162161e-01 -1.46041662e-01 -7.14751184e-02 1.07840979e+00
1.40760511e-01 -7.14426637e-01 1.00827754e-01 3.54302913e-01
6.70264661e-01 -9.13645864e-01 8.69105041e-01 6.40212953e-01
-1.09985006e+00 -3.02071780e-01 -7.83468008e-01 -1.80318877e-01
1.70179829e-02 4.06325519e-01 -6.41515672e-01 3.33687693e-01
2.95412332e-01 1.04406261e+00 -5.00449419e-01 1.01895666e+00
-3.56379189e-02 6.29965603e-01 3.45844388e-01 -5.97840548e-01
4.96558100e-01 -1.22121416e-01 1.63969234e-01 1.25629437e+00
2.34308302e-01 2.82588124e-01 7.81455934e-02 7.29122281e-01
-4.04784739e-01 6.42001569e-01 -9.04622257e-01 -5.86663723e-01
2.11823106e-01 1.33814633e+00 -7.09766746e-01 -5.98511398e-01
-9.13509309e-01 8.95687819e-01 4.39034790e-01 2.62972206e-01
-4.49477881e-01 -9.29687500e-01 2.77748674e-01 -8.28213021e-02
2.41218418e-01 2.72157486e-04 -1.36903822e-01 -1.50775063e+00
-1.97664887e-01 -5.30152380e-01 5.29903531e-01 -6.42689824e-01
-1.45747149e+00 4.70784515e-01 -9.57761854e-02 -1.22695160e+00
-2.03098550e-01 -5.90333462e-01 -1.06458580e+00 6.40247703e-01
-1.54463887e+00 -1.14962208e+00 -2.65037239e-01 5.20428300e-01
1.29056704e+00 -5.49268544e-01 1.02634108e+00 3.76582712e-01
-7.45278955e-01 6.50036275e-01 3.95312965e-01 2.59615093e-01
9.40408111e-01 -9.76139545e-01 3.31152231e-01 7.27629244e-01
2.97977120e-01 7.29567468e-01 5.47591150e-01 -5.69299459e-01
-1.27865887e+00 -1.09518230e+00 1.03057635e+00 -2.52842993e-01
8.62611413e-01 -5.96505582e-01 -1.12373161e+00 6.31277382e-01
4.22881812e-01 2.65392900e-01 9.82790411e-01 2.64237583e-01
-3.55827391e-01 -1.09209917e-01 -9.17568922e-01 5.48303783e-01
7.74285674e-01 -6.62524641e-01 -1.19427752e+00 3.30978423e-01
8.13859582e-01 3.66784513e-01 -3.61270308e-01 2.45605260e-02
3.18554968e-01 -3.84534866e-01 5.61627209e-01 -1.23036695e+00
6.51762903e-01 3.46578695e-02 -3.41435373e-02 -1.48404908e+00
-2.05325529e-01 -2.31711254e-01 -2.81098783e-01 1.33579409e+00
2.17126265e-01 -3.73754948e-01 7.91595995e-01 5.18047273e-01
-6.58584312e-02 -6.01115286e-01 -7.48028278e-01 -7.11175025e-01
3.09833556e-01 -2.19008908e-01 1.01072378e-01 1.38770676e+00
5.01815438e-01 9.15902972e-01 -6.50245726e-01 -5.03122568e-01
6.81707740e-01 5.90818934e-02 6.90770805e-01 -1.32616115e+00
-1.40912265e-01 -3.51866245e-01 -4.38903213e-01 -8.60402822e-01
3.86402220e-01 -1.13755870e+00 -5.35286292e-02 -1.67931926e+00
6.80188954e-01 -5.98309226e-02 -5.95906794e-01 4.37992990e-01
-5.95232725e-01 -1.04678504e-01 6.53931350e-02 -9.11661983e-03
-8.15505803e-01 1.11358809e+00 1.02131379e+00 -5.34686327e-01
-8.74043554e-02 -2.20088094e-01 -6.47010148e-01 6.20959997e-01
7.39092350e-01 -6.85858727e-01 -7.70854652e-01 -3.21641594e-01
1.90513823e-02 -1.52679607e-01 -1.33513883e-01 -9.82757330e-01
5.11779487e-01 -1.24212064e-01 1.66779086e-01 -5.10718584e-01
2.30639040e-01 -6.66223049e-01 -7.26155341e-01 5.09494543e-01
-6.71504736e-01 -4.09191906e-01 2.90647894e-02 8.41152072e-01
-2.83638567e-01 -6.62837744e-01 9.42698359e-01 -1.25862539e-01
-1.08804762e+00 5.31243920e-01 -3.37757975e-01 3.60205650e-01
1.02383268e+00 -2.05797311e-02 -3.37727547e-01 -2.02825636e-01
-5.25134921e-01 3.54393035e-01 2.42391732e-02 8.24286103e-01
6.38484716e-01 -1.43384194e+00 -5.22061527e-01 1.48554400e-01
8.23312819e-01 -4.59782720e-01 4.30819273e-01 3.68752569e-01
-3.24851163e-02 5.72231829e-01 -2.14484349e-01 -1.44777074e-01
-8.34845901e-01 9.30245221e-01 8.33115131e-02 -2.26284891e-01
-8.42871308e-01 5.95423102e-01 -9.02595073e-02 -3.49154145e-01
4.41019893e-01 -6.14112616e-03 -4.85091746e-01 3.56100231e-01
1.00578976e+00 2.85720706e-01 2.80275829e-02 -2.76413769e-01
-5.25664911e-02 5.12092769e-01 -6.07099652e-01 -1.40598919e-02
1.38653517e+00 -1.39127886e-02 4.46290791e-01 1.04243970e+00
1.31379116e+00 -3.60932946e-01 -9.60856736e-01 -7.11725295e-01
1.42956674e-01 -4.66598541e-01 3.06491286e-01 -4.46032077e-01
-7.38885581e-01 1.59995365e+00 5.59755027e-01 -3.61712836e-02
5.40655613e-01 -3.60226363e-01 9.83523488e-01 8.39293122e-01
4.33963120e-01 -1.49197114e+00 5.74829638e-01 6.72587872e-01
2.93222100e-01 -1.69411647e+00 -6.98386207e-02 1.51077285e-01
-6.69149578e-01 1.16032994e+00 7.87311673e-01 -1.35836378e-01
1.04042089e+00 -2.04585984e-01 4.83327061e-02 -7.54269361e-02
-1.10652149e+00 -2.06200659e-01 4.76430468e-02 4.81526107e-01
6.06998980e-01 -4.53499228e-01 -5.82741559e-01 7.94946194e-01
3.93485844e-01 1.18190274e-01 4.15110826e-01 1.29332209e+00
-9.95264888e-01 -1.12198222e+00 1.15283988e-01 6.47720695e-01
-2.98843801e-01 -3.04159135e-01 -3.46054211e-02 6.01528287e-01
-1.34557664e-01 8.19772243e-01 -1.04895071e-03 -2.22965196e-01
1.08617522e-01 6.37867868e-01 1.29638135e-01 -1.31174195e+00
-3.40198636e-01 -2.30533302e-01 -1.81362897e-01 2.90108770e-02
-3.45236450e-01 -1.88940793e-01 -1.15100515e+00 -4.95429561e-02
-5.27308226e-01 4.24575895e-01 4.99764591e-01 1.17381358e+00
2.21224755e-01 5.76193810e-01 6.43719733e-01 -5.89645565e-01
-7.89298952e-01 -1.26536059e+00 -6.69106007e-01 4.32802528e-01
1.80236429e-01 -8.81665349e-01 -5.20994127e-01 4.38302606e-02] | [10.254741668701172, 3.608795166015625] |
946e8361-f282-40c5-9701-a93251b363d6 | occlusion-geodesics-for-online-multi-object | null | null | http://openaccess.thecvf.com/content_cvpr_2014/html/Possegger_Occlusion_Geodesics_for_2014_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2014/papers/Possegger_Occlusion_Geodesics_for_2014_CVPR_paper.pdf | Occlusion Geodesics for Online Multi-Object Tracking | Robust multi-object tracking-by-detection requires the correct assignment of noisy detection results to object trajectories. We address this problem by proposing an online approach based on the observation that object detectors primarily fail if objects are significantly occluded. In contrast to most existing work, we only rely on geometric information to efficiently overcome detection failures. In particular, we exploit the spatio-temporal evolution of occlusion regions, detector reliability, and target motion prediction to robustly handle missed detections. In combination with a conservative association scheme for visible objects, this allows for real-time tracking of multiple objects from a single static camera, even in complex scenarios. Our evaluations on publicly available multi-object tracking benchmark datasets demonstrate favorable performance compared to the state-of-the-art in online and offline multi-object tracking. | ['Peter M. Roth', 'Thomas Mauthner', 'Horst Bischof', 'Horst Possegger'] | 2014-06-01 | null | null | null | cvpr-2014-6 | ['online-multi-object-tracking'] | ['computer-vision'] | [-1.21665619e-01 -5.85132003e-01 -6.19939156e-02 1.35227442e-01
-7.82636166e-01 -8.80300760e-01 3.69260162e-01 3.42861980e-01
-4.97466356e-01 5.43901622e-01 -4.16385621e-01 2.80875154e-03
5.62862912e-03 -3.18555862e-01 -7.20237076e-01 -5.94923973e-01
-2.67399158e-02 5.68131328e-01 1.06500673e+00 3.76868248e-01
-1.32026768e-03 7.64922738e-01 -1.50258148e+00 -4.70036343e-02
4.48792756e-01 9.21612442e-01 2.35949725e-01 8.68526995e-01
2.06934929e-01 4.62361932e-01 -4.61562157e-01 -2.72363216e-01
4.03546780e-01 6.20847270e-02 -5.28111123e-02 4.01862174e-01
8.33196700e-01 -6.18186116e-01 -2.66588420e-01 1.02242970e+00
3.57775807e-01 6.96067885e-02 3.67013395e-01 -1.42097402e+00
-1.87661946e-01 7.61977732e-02 -8.57529938e-01 6.34825528e-01
1.62314609e-01 4.22744364e-01 7.19810188e-01 -9.81352389e-01
5.51307857e-01 1.20574188e+00 9.45129395e-01 4.04660016e-01
-1.32599509e+00 -6.84481561e-01 5.92613280e-01 2.67540794e-02
-1.54671824e+00 -4.64321107e-01 2.70052075e-01 -5.97017229e-01
6.48905456e-01 1.48583874e-01 5.82214653e-01 6.59886122e-01
1.71651512e-01 6.43544257e-01 6.89947605e-01 -1.84155181e-01
-7.79887959e-02 1.39520615e-01 2.95245707e-01 6.83094442e-01
8.96279573e-01 4.76647466e-01 -3.13997477e-01 -5.01471102e-01
4.75908399e-01 2.43228748e-01 -3.62791941e-02 -6.76401734e-01
-1.23185968e+00 5.57014942e-01 2.63070315e-01 1.19961195e-01
-3.48556966e-01 2.67424047e-01 3.05851787e-01 6.61313627e-03
2.97428399e-01 -7.12486431e-02 -2.75602579e-01 1.27523765e-01
-1.07003057e+00 4.34251159e-01 4.22329217e-01 1.18235552e+00
4.82250869e-01 1.53470546e-01 -5.34665108e-01 1.24955602e-01
5.38639665e-01 8.61686468e-01 -2.42400303e-01 -7.96955824e-01
3.09394568e-01 4.78511691e-01 6.48051202e-01 -7.87910283e-01
-3.08503300e-01 -6.71611786e-01 -3.23475778e-01 5.03208041e-01
6.06302798e-01 -1.99165419e-01 -8.08027148e-01 1.50360739e+00
8.77365410e-01 6.06321394e-01 -2.81241596e-01 8.87963057e-01
3.98933798e-01 2.18305618e-01 1.35482073e-01 -4.76469994e-01
1.24166334e+00 -8.59192729e-01 -6.11216187e-01 -2.23308206e-01
2.57498473e-01 -8.73546302e-01 6.82051405e-02 3.05006385e-01
-9.63183045e-01 -6.98097765e-01 -8.21515262e-01 4.99594629e-01
-9.51778963e-02 2.66975760e-01 3.23514283e-01 7.67379880e-01
-7.91340232e-01 3.24787825e-01 -1.03450310e+00 -3.95292789e-01
5.41701078e-01 4.15357471e-01 9.22744814e-03 -5.25761247e-02
-2.95697063e-01 9.59598780e-01 4.97939140e-01 1.83345124e-01
-1.19056463e+00 -8.25511813e-01 -5.31180024e-01 -2.11989805e-01
7.44181097e-01 -6.48237109e-01 1.23063743e+00 -5.19217849e-01
-8.91468942e-01 4.51088190e-01 -3.52598190e-01 -6.69489503e-01
9.29678500e-01 -4.77989882e-01 -3.17010403e-01 5.76947518e-02
9.11834985e-02 3.99278641e-01 7.40138590e-01 -1.43340182e+00
-1.17835724e+00 -3.31819177e-01 -2.19198793e-01 -2.86623091e-02
-1.10751264e-01 4.21753794e-01 -8.38681400e-01 -4.72714216e-01
1.49430474e-02 -1.11068726e+00 -4.78647530e-01 5.41246235e-01
-2.26872623e-01 -1.43068105e-01 1.49192989e+00 -2.94272900e-01
9.96308446e-01 -2.05073380e+00 -4.06118482e-02 6.26323149e-02
3.10034305e-01 3.65144342e-01 5.93491010e-02 5.63529991e-02
4.09839451e-01 -4.51680899e-01 3.41063887e-01 -7.69091547e-01
-2.13621795e-01 4.79144752e-02 -4.37237918e-01 9.74190772e-01
1.54297724e-01 7.45379388e-01 -1.02812755e+00 -7.60310113e-01
6.19733512e-01 4.11718607e-01 -3.33077252e-01 2.48488877e-02
-3.00447524e-01 6.34060919e-01 -5.84355891e-01 9.75844204e-01
8.22374940e-01 -3.07452410e-01 1.91441998e-02 -7.95904472e-02
-4.51641202e-01 -3.51184547e-01 -1.67403829e+00 1.12164426e+00
1.71464849e-02 5.25690913e-01 2.83504903e-01 -4.01723713e-01
4.07465488e-01 3.48766297e-01 8.54718685e-01 -1.57923743e-01
1.42248020e-01 1.13067038e-01 -3.22534926e-02 9.22606587e-02
6.48494124e-01 3.13229442e-01 2.86872268e-01 1.83972299e-01
-2.38419414e-01 4.80397791e-01 4.22093719e-01 2.16461033e-01
1.15344179e+00 2.60355681e-01 2.95069695e-01 1.12907819e-01
1.85466245e-01 2.93756366e-01 8.61569762e-01 1.20338523e+00
-5.89618325e-01 4.34840560e-01 -3.51182431e-01 -2.65336186e-01
-8.67028475e-01 -1.23962617e+00 -2.08602905e-01 1.01333666e+00
5.90606689e-01 -2.97049612e-01 -3.01394522e-01 -6.73521757e-01
3.64858568e-01 3.30520451e-01 -3.53928447e-01 1.67953461e-01
-7.80125737e-01 -7.70718634e-01 3.09319556e-01 6.36427641e-01
-6.12920970e-02 -6.63481772e-01 -9.93550062e-01 6.48164213e-01
1.70445502e-01 -1.50627971e+00 -5.48622847e-01 -7.07687885e-02
-9.10053849e-01 -1.28268278e+00 -4.27188665e-01 -2.96800941e-01
5.92015207e-01 8.76369476e-01 9.81703877e-01 3.78401905e-01
-5.71580112e-01 6.36706531e-01 -1.45394266e-01 -5.26404321e-01
-3.99376899e-01 -4.31532323e-01 2.95920968e-01 -1.85678415e-02
1.67062938e-01 2.36185305e-02 -5.40164113e-01 6.07482791e-01
-4.50910836e-01 -2.67774135e-01 5.79005837e-01 4.48461622e-01
7.81309545e-01 1.17598027e-01 2.34803304e-01 -4.02276129e-01
-2.38622889e-01 -4.00039494e-01 -1.45151138e+00 4.60184723e-01
-3.95167142e-01 -3.35087031e-01 2.38536909e-01 -8.63611758e-01
-9.02098179e-01 6.94994986e-01 4.35308158e-01 -9.27967072e-01
-1.50613412e-01 -2.90301204e-01 1.30697280e-01 -5.60293972e-01
3.43376458e-01 6.07246421e-02 -3.32936078e-01 -4.85834181e-01
2.86332041e-01 1.54450402e-01 6.54334962e-01 -3.09946001e-01
1.21282589e+00 9.62962627e-01 1.43291980e-01 -7.37968802e-01
-7.46005118e-01 -1.09408975e+00 -6.90435410e-01 -6.42571032e-01
7.61997461e-01 -1.19492054e+00 -9.42025840e-01 1.83480278e-01
-1.21317089e+00 1.59161747e-01 -8.58482197e-02 6.34371519e-01
-1.77560285e-01 4.97573227e-01 -4.31778252e-01 -1.41020560e+00
-1.25273317e-01 -9.54964042e-01 1.37748885e+00 3.10753047e-01
1.77022278e-01 -7.87659705e-01 -3.25581208e-02 3.19308862e-02
1.96257398e-01 4.13971007e-01 -1.98719859e-01 -4.73842263e-01
-1.38333499e+00 -4.67775255e-01 -2.34761655e-01 -1.97376654e-01
-3.79058346e-02 2.58444160e-01 -7.17533588e-01 -7.67511070e-01
-3.84299666e-01 1.59006789e-02 1.09919846e+00 6.38936162e-01
5.99059463e-01 1.21401148e-02 -1.11124516e+00 1.47610635e-01
1.63966262e+00 -1.20692909e-01 -3.07042785e-02 2.77487904e-01
7.05878019e-01 1.09496564e-01 1.10924852e+00 8.08947265e-01
2.53015101e-01 1.06306732e+00 6.34531677e-01 3.23399641e-02
-2.75532752e-01 6.69656172e-02 3.40493351e-01 -8.34295675e-02
2.19060536e-02 -3.06325316e-01 -8.10471714e-01 7.23489404e-01
-2.18168855e+00 -1.14271748e+00 -6.63646758e-01 2.47491145e+00
3.42533916e-01 4.70854551e-01 5.48995078e-01 -3.51626843e-01
9.99796331e-01 -9.54839215e-02 -4.63488400e-01 6.69722855e-01
-7.24456906e-02 -4.47914988e-01 1.11821353e+00 3.60580623e-01
-1.42839062e+00 7.35484719e-01 6.41363239e+00 4.56867039e-01
-6.57925189e-01 4.63931352e-01 -1.73425302e-01 -4.13604110e-01
4.71458584e-01 8.69849548e-02 -1.70167971e+00 5.30376196e-01
7.30361640e-01 5.35284542e-02 -6.63103759e-02 8.63914907e-01
3.01928312e-01 -2.95036227e-01 -1.06897509e+00 7.22988009e-01
-1.94976836e-01 -1.24959099e+00 -3.28525901e-01 7.56187886e-02
6.88138366e-01 2.76469111e-01 -1.12075172e-02 1.22780785e-01
5.83272874e-01 -3.45911711e-01 1.09427667e+00 4.25509721e-01
1.60294563e-01 -3.86361569e-01 4.66094017e-01 4.23893631e-01
-1.94522691e+00 -2.64755934e-01 -3.09945643e-01 1.69064552e-01
5.54623008e-01 5.45814872e-01 -8.28716755e-01 6.82058811e-01
7.13441610e-01 5.31409919e-01 -7.15988994e-01 1.90166545e+00
1.98755011e-01 3.60154629e-01 -6.92462921e-01 3.30287546e-01
1.09545872e-01 2.96591222e-01 9.93061006e-01 1.31756341e+00
1.30144298e-01 -6.25673458e-02 9.56972539e-01 5.73406935e-01
3.67133468e-01 -3.19211215e-01 -5.39179325e-01 3.43522668e-01
6.36920452e-01 1.27045965e+00 -9.35034513e-01 -4.03969139e-01
-7.52267778e-01 4.88884211e-01 2.61916161e-01 2.11301699e-01
-1.26207936e+00 3.11312586e-01 6.19967818e-01 2.22042277e-02
9.23574567e-01 -5.48852503e-01 2.09384803e-02 -9.24754620e-01
2.04460621e-01 -4.04237628e-01 6.53365254e-01 -2.60329783e-01
-1.20754623e+00 2.89309651e-01 5.23449592e-02 -1.50756741e+00
-1.04043938e-01 -4.56479490e-01 -4.95895684e-01 5.55089295e-01
-1.60072088e+00 -1.48486924e+00 -2.83006757e-01 3.87767047e-01
5.95901787e-01 1.25369877e-01 3.41301769e-01 5.64978242e-01
-5.08477271e-01 2.63064533e-01 2.57062674e-01 5.01753995e-03
6.98405862e-01 -1.02342582e+00 4.03569788e-01 1.29533494e+00
2.66009599e-01 2.90132016e-01 1.00355029e+00 -1.04972315e+00
-1.49423182e+00 -1.42182362e+00 3.45636189e-01 -9.18547750e-01
6.56520009e-01 -2.84126818e-01 -9.53397810e-01 8.41449261e-01
-8.73894393e-02 6.62827313e-01 2.59629428e-01 2.50331238e-02
-1.55197293e-01 2.63966247e-02 -7.93370783e-01 3.66577923e-01
9.34758127e-01 7.33340755e-02 -3.76873583e-01 5.69461286e-01
4.92932349e-01 -7.33589768e-01 -5.10214627e-01 4.98524278e-01
6.45916998e-01 -5.99692941e-01 1.28368604e+00 -4.83049035e-01
-6.61058307e-01 -1.00017548e+00 8.58737808e-03 -6.65162683e-01
-4.66530621e-01 -5.05883157e-01 -6.35408700e-01 1.10039270e+00
1.08135775e-01 -1.79947138e-01 8.03379536e-01 4.59879279e-01
-6.97022527e-02 -1.69520617e-01 -1.04053700e+00 -1.30772150e+00
-6.00675166e-01 -3.28630209e-01 -1.73085518e-02 4.29366589e-01
-8.54454935e-01 -1.64590538e-01 -6.06301248e-01 9.72384095e-01
1.32280529e+00 5.41227281e-01 9.95789230e-01 -1.55777991e+00
-4.18046147e-01 -2.66261846e-01 -4.78883475e-01 -9.37109768e-01
2.55288742e-02 -3.13079625e-01 3.79034728e-01 -1.23135257e+00
5.38870394e-01 -4.97377455e-01 -3.28124315e-01 2.46911779e-01
-4.57329184e-01 5.24518430e-01 4.76971328e-01 3.92691642e-01
-1.46490252e+00 3.05858970e-01 7.20337689e-01 6.40269369e-02
-1.85359016e-01 3.99308860e-01 -5.19984066e-02 7.05504596e-01
3.90291452e-01 -9.21194673e-01 2.61527866e-01 -3.21497083e-01
-3.40099096e-01 -5.30038252e-02 8.62329185e-01 -1.22760332e+00
6.69327557e-01 -3.39539289e-01 5.68405271e-01 -1.32326257e+00
5.71595669e-01 -1.19708848e+00 3.56926680e-01 7.56388247e-01
2.11759314e-01 1.70385450e-01 5.79284370e-01 1.22205400e+00
1.69947684e-01 -4.84979562e-02 9.22357619e-01 -3.66846845e-02
-9.69125450e-01 3.91058981e-01 -1.95374787e-01 -1.69363707e-01
1.47847390e+00 -2.47855797e-01 -3.02681535e-01 1.12223029e-01
-6.93675876e-01 6.04131758e-01 5.40170372e-01 5.33331335e-01
4.09766853e-01 -1.19053972e+00 -7.99441159e-01 -9.05927494e-02
1.72429591e-01 -1.97415322e-01 1.56912103e-01 1.06670356e+00
1.58910397e-02 4.52826053e-01 8.76303837e-02 -1.08121538e+00
-1.81896496e+00 7.68622816e-01 3.01590741e-01 -1.81902885e-01
-8.05247784e-01 4.74121273e-01 1.38266385e-01 2.54824907e-01
4.74832773e-01 -1.37762904e-01 1.34407997e-01 -1.97771251e-01
7.28041172e-01 5.31234920e-01 -5.39247170e-02 -7.16863453e-01
-7.09306598e-01 5.16705871e-01 -2.62439281e-01 -7.48614175e-03
9.85494375e-01 -3.00992936e-01 4.09122020e-01 3.65661025e-01
4.04737264e-01 1.97806060e-01 -1.78390753e+00 -4.87087309e-01
2.66051620e-01 -8.19030762e-01 2.30018124e-02 -5.02996743e-01
-8.46254766e-01 2.06265330e-01 9.36160803e-01 2.39263251e-02
5.69313407e-01 1.55583575e-01 4.80978578e-01 3.70408088e-01
5.80168426e-01 -9.23124790e-01 5.92823736e-02 3.45999032e-01
3.96093309e-01 -1.58133399e+00 3.07156563e-01 -4.94579971e-01
-2.86477417e-01 9.22500491e-01 8.30537498e-01 7.29738399e-02
4.15761590e-01 5.54038167e-01 -1.30762950e-01 1.33272316e-02
-8.19296181e-01 -6.45256937e-01 3.27699095e-01 4.96174783e-01
-1.04238927e-01 -1.75149173e-01 2.44454637e-01 7.73971379e-02
7.08529174e-01 -7.28162974e-02 2.78492093e-01 1.29317963e+00
-8.29575419e-01 -7.92520225e-01 -9.68302190e-01 2.41032109e-01
-4.96518731e-01 2.38178805e-01 -1.44539550e-01 9.18790042e-01
1.74838200e-01 1.15242076e+00 1.18175112e-01 2.20026910e-01
4.16008621e-01 -3.39393288e-01 5.97810924e-01 -6.80626392e-01
-5.55600047e-01 3.81665498e-01 -1.02461152e-01 -5.83023727e-01
-5.57270050e-01 -1.08473766e+00 -1.06844711e+00 -1.61344156e-01
-8.69954884e-01 -1.70057446e-01 5.40467381e-01 9.27714407e-01
3.53606790e-01 6.10748589e-01 1.83283359e-01 -1.02465498e+00
-7.38983035e-01 -6.69518828e-01 -3.28730315e-01 2.05108210e-01
7.84383059e-01 -1.10416889e+00 -1.44373462e-01 -2.90275048e-02] | [6.442364692687988, -2.0462772846221924] |
7f2a7f10-334b-4898-b6ec-62a11bee0fac | uncertain-label-correction-via-auxiliary | 2204.11053 | null | https://arxiv.org/abs/2204.11053v2 | https://arxiv.org/pdf/2204.11053v2.pdf | Uncertain Label Correction via Auxiliary Action Unit Graphs for Facial Expression Recognition | High-quality annotated images are significant to deep facial expression recognition (FER) methods. However, uncertain labels, mostly existing in large-scale public datasets, often mislead the training process. In this paper, we achieve uncertain label correction of facial expressions using auxiliary action unit (AU) graphs, called ULC-AG. Specifically, a weighted regularization module is introduced to highlight valid samples and suppress category imbalance in every batch. Based on the latent dependency between emotions and AUs, an auxiliary branch using graph convolutional layers is added to extract the semantic information from graph topologies. Finally, a re-labeling strategy corrects the ambiguous annotations by comparing their feature similarities with semantic templates. Experiments show that our ULC-AG achieves 89.31% and 61.57% accuracy on RAF-DB and AffectNet datasets, respectively, outperforming the baseline and state-of-the-art methods. | ['Guoying Zhao', 'Janne Kauttonen', 'Xingming Zhang', 'Yang Liu'] | 2022-04-23 | null | null | null | null | ['facial-expression-recognition'] | ['computer-vision'] | [ 3.25596899e-01 4.20463055e-01 -1.30429506e-01 -1.14189112e+00
-4.13789868e-01 -1.97602719e-01 1.71932116e-01 -2.46010587e-01
-1.90546051e-01 7.62102425e-01 -1.77738965e-02 2.78120160e-01
4.30871546e-01 -4.13336217e-01 -6.71915174e-01 -6.13429487e-01
1.96929768e-01 1.04488775e-01 -3.26837510e-01 -1.85086265e-01
-1.66905344e-01 4.86192822e-01 -1.50034821e+00 5.79879820e-01
7.48047888e-01 1.64844477e+00 -5.06891727e-01 -1.43408760e-01
-4.00050789e-01 1.02469611e+00 -5.26654661e-01 -1.07349718e+00
-7.55998641e-02 -4.16991085e-01 -5.83285332e-01 5.88092566e-01
5.85337937e-01 -3.13955426e-01 -6.72077835e-02 1.42256880e+00
4.73635405e-01 -1.01884775e-01 5.76093256e-01 -1.76871371e+00
-6.28447413e-01 3.29532504e-01 -9.16940153e-01 -3.98524314e-01
9.54220742e-02 -9.51597560e-03 9.09316063e-01 -1.15324521e+00
7.31551468e-01 1.36652195e+00 5.77056050e-01 8.96561027e-01
-1.04228377e+00 -1.29020083e+00 5.34654737e-01 2.85598278e-01
-1.44404769e+00 -7.22073853e-01 9.84897971e-01 -2.07716495e-01
4.30781007e-01 -1.90737918e-02 5.00847399e-01 1.34773231e+00
-3.16665441e-01 8.11052978e-01 1.10286355e+00 -9.35523212e-02
2.52615716e-02 8.59171078e-02 -3.66416574e-02 1.17250395e+00
-5.58468513e-02 -3.97470504e-01 -6.37703300e-01 1.61865912e-02
3.84505332e-01 -2.13987008e-01 -2.49387830e-01 -1.69423416e-01
-4.34282422e-01 7.20491707e-01 5.17158449e-01 3.44466306e-02
-4.24206465e-01 1.21449597e-01 6.70375466e-01 8.14370364e-02
8.76611352e-01 1.72198534e-01 -5.00227034e-01 1.66730344e-01
-5.91547072e-01 -2.19366372e-01 3.33727181e-01 1.03183413e+00
8.92591178e-01 3.57111275e-01 -4.22886074e-01 1.11745262e+00
2.80383170e-01 4.24359202e-01 7.51132742e-02 -9.06191170e-01
1.20561875e-01 9.81024742e-01 -1.85915992e-01 -1.32097876e+00
-4.62156415e-01 -5.45165360e-01 -1.05761385e+00 3.59267704e-02
2.41567194e-01 -1.38118759e-01 -1.11544609e+00 2.09659791e+00
4.05692786e-01 4.17951763e-01 -1.25382572e-01 1.13121796e+00
1.10957646e+00 3.90350163e-01 4.96428847e-01 -3.79146159e-01
1.26688588e+00 -9.76003885e-01 -9.85076547e-01 -1.90552518e-01
6.07293963e-01 -5.60385466e-01 1.06630468e+00 3.73250365e-01
-6.43836975e-01 -3.88449192e-01 -7.64513373e-01 5.91464043e-02
-1.16840720e-01 7.19894409e-01 8.70514929e-01 4.14376587e-01
-7.90968776e-01 3.68224084e-01 -4.50194299e-01 2.96126120e-02
1.04522073e+00 3.06503177e-01 -7.04865694e-01 -2.10898206e-01
-1.14575744e+00 3.97373617e-01 2.04937793e-02 4.70473319e-01
-7.78862834e-01 -5.36711693e-01 -1.06077433e+00 -7.25433379e-02
4.13986981e-01 -7.05588758e-02 1.11950421e+00 -2.04218340e+00
-1.60674977e+00 1.30918241e+00 -3.13685060e-01 9.31300409e-03
4.24683630e-01 4.03472111e-02 -5.52402139e-01 1.59442618e-01
4.00806032e-02 9.64442015e-01 1.09428680e+00 -1.25005960e+00
-3.28787178e-01 -6.19166970e-01 -1.50783494e-01 2.72063501e-02
-4.14170206e-01 1.70293003e-01 -6.28862619e-01 -5.48550546e-01
1.73519403e-01 -8.91941071e-01 -6.26704143e-03 3.67639154e-01
-3.58239055e-01 -5.01366317e-01 8.50322664e-01 -8.73168588e-01
1.02194965e+00 -2.29250789e+00 -1.83173019e-04 3.42577666e-01
2.84319997e-01 3.03795874e-01 -3.58025998e-01 -5.14920533e-01
-3.74033093e-01 3.32788192e-02 -1.87859297e-01 -7.15614974e-01
1.10532492e-01 3.80364984e-01 -4.28502224e-02 4.23797607e-01
6.16180062e-01 9.29815054e-01 -7.86635578e-01 -6.01448715e-01
3.26378015e-03 4.48549658e-01 -3.47669601e-01 1.51554123e-01
-2.81538635e-01 4.36279297e-01 -3.31552476e-01 1.06563723e+00
1.02344441e+00 -3.93081069e-01 2.81175435e-01 -5.85610092e-01
5.51530778e-01 -2.64959067e-01 -9.44141150e-01 1.60679781e+00
-2.70619363e-01 4.62939024e-01 3.90211940e-01 -1.14494371e+00
1.29519689e+00 9.82414484e-02 4.77317542e-01 -8.18053842e-01
5.28131962e-01 1.88295022e-01 -2.72917688e-01 -3.58442515e-01
3.44356716e-01 -7.58135617e-02 2.04654615e-02 2.28508306e-03
4.57802922e-01 3.20266306e-01 -5.96012659e-02 1.51567623e-01
7.02527523e-01 2.24856541e-01 4.46825810e-02 -4.78265025e-02
7.35016942e-01 -4.32604462e-01 1.09975421e+00 4.60044593e-02
-5.82192838e-01 4.59352195e-01 9.50807214e-01 -4.66054618e-01
-6.24480605e-01 -6.19621933e-01 3.57553735e-02 1.19502747e+00
1.38853386e-01 -5.69350958e-01 -9.63302433e-01 -1.14785480e+00
-4.18435410e-02 3.77105713e-01 -9.79079127e-01 -4.33396816e-01
-9.64830816e-02 -6.48560524e-01 5.48378348e-01 4.74302322e-01
7.94785917e-01 -1.02195811e+00 1.51489645e-01 3.40682305e-02
-3.40369403e-01 -1.37013161e+00 -3.97682607e-01 -1.43425047e-01
-3.62542331e-01 -1.19394112e+00 -4.30232316e-01 -5.00644982e-01
1.11066163e+00 -2.74791662e-02 1.09709573e+00 2.20609024e-01
-7.04950020e-02 3.74434814e-02 -4.72702563e-01 -4.80458707e-01
-9.33131650e-02 -1.84114620e-01 -1.23611704e-01 8.21279585e-01
6.02918744e-01 -2.47966006e-01 -5.49096286e-01 4.58074152e-01
-5.96777439e-01 3.07545513e-01 5.37653267e-01 1.02724898e+00
8.47080827e-01 -4.39667046e-01 6.83643758e-01 -9.71212089e-01
3.23804051e-01 -3.55290174e-01 -5.36251903e-01 3.13677609e-01
-5.63827813e-01 -1.15658991e-01 4.78042603e-01 -4.26721245e-01
-1.27894282e+00 4.24434185e-01 -2.48924643e-01 -8.61504138e-01
-1.57678396e-01 2.59265095e-01 -7.15799987e-01 -2.67018855e-01
3.70671093e-01 -2.79006660e-01 9.01051536e-02 -6.19507730e-02
1.71510234e-01 7.20891595e-01 5.41515589e-01 -4.78751957e-01
1.93802893e-01 3.33268136e-01 2.75659394e-02 -4.66095001e-01
-1.39002717e+00 -1.35636538e-01 -4.67580527e-01 -6.77734017e-01
8.49355400e-01 -1.00567520e+00 -6.39414251e-01 7.34646916e-01
-1.24042988e+00 -2.48703346e-01 1.35440882e-02 1.54575169e-01
-2.23631620e-01 -4.07906584e-02 -5.66081345e-01 -8.43660295e-01
-3.79571885e-01 -1.16125035e+00 1.45532179e+00 5.22610664e-01
-6.45528510e-02 -4.43526000e-01 -5.95107794e-01 5.89379013e-01
8.37840736e-02 5.72851717e-01 4.98437852e-01 -5.98762691e-01
-2.24881113e-01 -9.15905535e-02 -7.74714828e-01 7.31776178e-01
2.26517111e-01 1.81925625e-01 -1.32570815e+00 4.30844948e-02
-5.74023843e-01 -8.15296590e-01 7.97351599e-01 8.53430256e-02
1.68995118e+00 -2.62376130e-01 -1.74489275e-01 8.01749766e-01
9.28732812e-01 -1.06717303e-01 6.87749505e-01 -1.17998570e-01
8.93890381e-01 7.70717680e-01 8.65249038e-01 6.26982450e-01
2.29108855e-01 5.21458268e-01 6.79862082e-01 -2.87052721e-01
-1.17647253e-01 -3.17061841e-01 1.93466723e-01 3.35067779e-01
2.93300096e-02 -1.25871614e-01 -4.96416807e-01 1.82663351e-01
-1.91294301e+00 -6.15857542e-01 -2.58170180e-02 1.71593559e+00
8.80557060e-01 -5.47182299e-02 -2.18671158e-01 -1.87347621e-01
9.57436442e-01 3.23148936e-01 -7.78848290e-01 -3.15921307e-01
-5.59573054e-01 3.02941978e-01 3.73624980e-01 2.22489551e-01
-1.16730511e+00 1.33720791e+00 4.88619661e+00 1.01103556e+00
-1.28686976e+00 1.89351320e-01 1.34759891e+00 1.75626963e-01
-9.96272415e-02 -4.65940684e-01 -6.70357287e-01 4.39533234e-01
4.68070984e-01 2.05938637e-01 1.81726143e-01 1.25338089e+00
1.29005849e-01 2.92521507e-01 -6.51536286e-01 1.31268358e+00
3.92695308e-01 -1.00658309e+00 1.46464288e-01 -2.54877537e-01
7.55061746e-01 -3.06450754e-01 -2.36247834e-02 4.15734559e-01
2.89186746e-01 -1.10894954e+00 6.19494796e-01 5.78230798e-01
1.20599782e+00 -9.64084804e-01 1.01020563e+00 -2.20389917e-01
-9.77182686e-01 1.48046151e-01 -4.58566785e-01 1.79160014e-01
-1.88116892e-03 6.04272783e-01 -5.61971605e-01 4.69089776e-01
7.17852116e-01 9.82544124e-01 -6.33328855e-01 3.68382424e-01
-7.32142150e-01 6.80092990e-01 -2.67528594e-01 9.12636891e-02
1.34678409e-01 -2.35299647e-01 1.52290940e-01 1.02105069e+00
1.53585091e-01 2.11625591e-01 -5.18664680e-02 8.02888393e-01
-7.03862071e-01 2.49053910e-01 -2.08321169e-01 -5.22251241e-02
1.37245387e-01 1.78742015e+00 -4.08741415e-01 -2.57466078e-01
-2.51960933e-01 1.30740452e+00 6.96890235e-01 3.97050977e-01
-1.10655940e+00 -8.36637393e-02 9.07901525e-01 -3.04318249e-01
-6.92757890e-02 2.43277505e-01 -7.06243515e-02 -1.18393171e+00
1.02571569e-01 -8.82074058e-01 4.84144211e-01 -1.17662609e+00
-1.46355331e+00 7.65712738e-01 -5.47051847e-01 -9.57909405e-01
1.06650636e-01 -7.15954363e-01 -3.42584074e-01 5.39044797e-01
-1.70849848e+00 -1.54233336e+00 -8.61633122e-01 7.42696285e-01
1.76499024e-01 -1.11503676e-01 8.20724905e-01 4.22130704e-01
-9.00262952e-01 1.15404701e+00 -3.72552186e-01 4.17741388e-01
9.69916880e-01 -8.60520065e-01 -2.24571794e-01 7.08902299e-01
2.69002356e-02 2.86586490e-02 4.03884500e-01 -5.19069731e-01
-8.99729908e-01 -1.57945526e+00 5.56233108e-01 5.43265715e-02
5.31792700e-01 -5.45696080e-01 -1.12914324e+00 5.94897747e-01
-1.37251690e-01 6.82582557e-01 7.50531793e-01 3.15351784e-02
-7.96717823e-01 -4.18122947e-01 -1.19662189e+00 4.36767250e-01
1.35557318e+00 -4.39408273e-01 8.81644487e-02 3.56833220e-01
5.58191061e-01 -4.75126296e-01 -6.19725943e-01 7.64775574e-01
5.56651473e-01 -8.11192989e-01 4.30082351e-01 -8.63582790e-01
4.30995435e-01 -2.15736479e-01 -1.62956104e-01 -1.31671798e+00
-1.25007913e-01 -5.55329382e-01 2.03968048e-01 1.52739835e+00
4.01199132e-01 -4.56254333e-01 1.06625819e+00 7.34214425e-01
-2.07123905e-01 -6.95805371e-01 -9.22060490e-01 -3.84972155e-01
-4.62384492e-01 -3.83122325e-01 7.30467796e-01 1.31801438e+00
-3.19829404e-01 4.28195477e-01 -5.63467264e-01 1.02038682e-01
6.21202230e-01 -2.47661266e-02 7.97923684e-01 -1.21828413e+00
2.00331748e-01 -4.04696822e-01 -5.20144522e-01 -5.82066357e-01
9.85030055e-01 -8.66605818e-01 -7.54765123e-02 -1.15377009e+00
3.56637202e-02 -3.42157483e-01 -4.30734903e-01 1.07848549e+00
-3.15759271e-01 6.71264052e-01 -6.78021461e-02 -1.97267994e-01
-1.03331792e+00 1.08252501e+00 1.30467153e+00 -2.51655191e-01
1.25040680e-01 -3.45223099e-01 -6.62271738e-01 9.38922107e-01
9.04227376e-01 -3.16977739e-01 -2.54976064e-01 -2.73373932e-01
1.44848630e-01 -3.21882457e-01 3.16883862e-01 -5.37650824e-01
-1.62232652e-01 -2.32069433e-01 5.12862802e-01 -2.47351944e-01
3.55025977e-01 -7.68058360e-01 -1.34187536e-02 -8.98271576e-02
-3.35823178e-01 -2.61391371e-01 1.80829406e-01 4.62503761e-01
-4.77774799e-01 1.49886921e-01 1.00462830e+00 1.69209212e-01
-9.25969720e-01 8.05319309e-01 8.44920650e-02 -1.39358282e-01
9.47040617e-01 -7.35565228e-03 -3.54012012e-01 -5.56216121e-01
-8.12165141e-01 3.71311456e-01 2.48484209e-01 4.53743517e-01
6.57866836e-01 -1.64548612e+00 -6.81550860e-01 2.73103476e-01
6.79484367e-01 -1.70885660e-02 4.07569796e-01 8.14760149e-01
-1.09508105e-01 -1.78755507e-01 -3.79396468e-01 -4.84301329e-01
-1.68404400e+00 2.10209534e-01 5.94776034e-01 3.75899263e-02
-5.97301051e-02 1.09341037e+00 2.91311234e-01 -4.85468030e-01
2.55311251e-01 2.19311103e-01 -2.47101977e-01 2.83395886e-01
5.02844572e-01 -1.32220253e-01 9.80934128e-02 -1.10647702e+00
-4.59467232e-01 3.87071908e-01 3.54480483e-02 3.72758329e-01
1.14825714e+00 -1.19254157e-01 -3.61562103e-01 6.20959289e-02
1.30187619e+00 -2.31525719e-01 -1.33148277e+00 -3.05391133e-01
-1.79676563e-01 -4.16880459e-01 3.02984454e-02 -8.70106578e-01
-1.71738219e+00 9.55697060e-01 6.24274969e-01 -5.46269715e-01
1.49012446e+00 -2.03422178e-03 3.98910612e-01 2.20989138e-01
1.80979714e-01 -1.18920696e+00 2.62631893e-01 3.66084725e-01
1.01771259e+00 -1.45819807e+00 -2.61442661e-01 -8.90949011e-01
-9.20056045e-01 1.05744827e+00 1.29095161e+00 1.57108575e-01
4.22545195e-01 4.86808158e-02 2.34920040e-01 -2.40293369e-01
-5.40561259e-01 -2.10027784e-01 2.90374964e-01 5.38389564e-01
3.41343403e-01 3.80730219e-02 -2.23067671e-01 1.09326792e+00
1.56005681e-01 1.60881311e-01 1.20099135e-01 4.54708934e-01
-1.54900938e-01 -8.19906592e-01 8.68273675e-02 3.44913334e-01
-5.86898744e-01 3.25538516e-02 -7.07853258e-01 6.71104252e-01
3.51926535e-01 8.33896697e-01 1.37750581e-01 -5.35703599e-01
3.79865825e-01 2.44657174e-01 3.43494296e-01 -2.83287764e-01
-2.96637118e-01 4.47460935e-02 5.02091110e-01 -9.25289989e-01
-5.70471942e-01 -4.54441965e-01 -1.38519239e+00 -7.64442682e-02
-3.73022497e-01 1.52564541e-01 6.59708500e-01 7.14735389e-01
7.05152810e-01 5.05134225e-01 7.41340637e-01 -4.47047204e-01
-1.43534765e-01 -1.08463502e+00 -6.67344511e-01 8.98855865e-01
-8.35032761e-02 -1.08492589e+00 -4.74527359e-01 1.32241026e-01] | [13.630331039428711, 1.647863745689392] |
ca5adf63-4d7b-4cd5-b7f3-528a4685ff68 | anti-koopmanism | 2106.00106 | null | https://arxiv.org/abs/2106.00106v3 | https://arxiv.org/pdf/2106.00106v3.pdf | The kernel perspective on dynamic mode decomposition | This manuscript revisits theoretical assumptions concerning dynamic mode decomposition (DMD) of Koopman operators, including the existence of lattices of eigenfunctions, common eigenfunctions between Koopman operators, and boundedness and compactness of Koopman operators. Counterexamples that illustrate restrictiveness of the assumptions are provided for each of the assumptions. In particular, this manuscript proves that the native reproducing kernel Hilbert space (RKHS) of the Gaussian RBF kernel function only supports bounded Koopman operators if the dynamics are affine. In addition, a new framework for DMD, that requires only densely defined Koopman operators over RKHSs is introduced, and its effectiveness is demonstrated through numerical examples. | ['Joel A. Rosenfeld', 'Rushikesh Kamalapurkar', 'Michael Jury', 'Moad Abudia', 'Efrain Gonzalez'] | 2021-05-31 | null | null | null | null | ['misconceptions'] | ['miscellaneous'] | [-4.84169662e-01 1.84213266e-01 -2.44140495e-02 3.69659632e-01
-1.44802421e-01 -4.72211361e-01 -1.99963041e-02 -6.52088463e-01
7.60818943e-02 7.64434159e-01 7.48405419e-03 -3.26971024e-01
-6.39138758e-01 -2.53003597e-01 -4.19686794e-01 -1.26230252e+00
-7.92745531e-01 -1.46739720e-03 -2.75563329e-01 -3.20776910e-01
-7.27314651e-02 7.22830951e-01 -1.15305424e+00 -1.62401170e-01
8.77391338e-01 8.04151773e-01 -1.99474603e-01 9.36472476e-01
4.03393030e-01 7.41205633e-01 -1.37673527e-01 3.21452320e-02
5.55639625e-01 -1.68934822e-01 -9.20624316e-01 3.16298246e-01
-1.33105993e-01 -9.23565626e-02 -7.98366189e-01 1.21394956e+00
1.64060161e-01 5.76248348e-01 1.01699328e+00 -1.24723184e+00
-1.11578488e+00 2.48431638e-01 -2.44503111e-01 1.26506403e-01
5.54828532e-03 -1.64248630e-01 9.12876904e-01 -1.30881739e+00
2.82789826e-01 8.59555960e-01 1.23698390e+00 5.34281611e-01
-1.53957093e+00 -1.77686468e-01 -3.33967477e-01 -1.25243813e-01
-1.91893554e+00 -2.33781680e-01 4.52305436e-01 -7.26768017e-01
8.38967264e-01 4.79841858e-01 6.37988687e-01 6.91714644e-01
5.15256822e-01 4.30136085e-01 9.23898757e-01 -4.88477945e-01
1.59291998e-01 4.51827824e-01 6.36409163e-01 9.36439872e-01
2.77690023e-01 2.28370741e-01 6.73783422e-02 -6.77903116e-01
1.33138871e+00 -2.07682010e-02 -6.82733536e-01 -1.02124166e+00
-1.12770760e+00 1.00664115e+00 -1.79182023e-01 5.05767405e-01
-4.05407727e-01 -2.10686207e-01 2.37470105e-01 3.43372613e-01
2.20774502e-01 2.93607473e-01 -8.40909220e-03 4.00677975e-03
-6.12499356e-01 1.29302964e-01 8.31904292e-01 1.21573734e+00
5.84183395e-01 2.94982702e-01 2.08206594e-01 5.59459627e-01
1.93557724e-01 1.01245403e+00 3.68158519e-01 -1.13049984e+00
4.71268035e-02 1.80606425e-01 4.13736254e-01 -1.03543854e+00
-5.11714458e-01 -3.21050406e-01 -1.22053790e+00 -1.22220732e-01
1.88526809e-01 -1.07885450e-01 -3.66377920e-01 1.39975595e+00
3.25073570e-01 3.64626378e-01 4.45996016e-01 9.90987599e-01
2.59722590e-01 8.09016883e-01 -6.82523787e-01 -6.42539322e-01
7.20002413e-01 -4.90785420e-01 -9.61416960e-01 7.59383023e-01
6.88726544e-01 -4.57141429e-01 9.19039905e-01 2.89329261e-01
-1.16686022e+00 -4.24208879e-01 -8.61867964e-01 4.82350558e-01
-2.22573467e-02 3.57966423e-01 6.86488926e-01 4.70634907e-01
-1.12643719e+00 7.17484772e-01 -8.74571979e-01 -1.51205391e-01
-3.69494379e-01 3.99325937e-01 -5.17550409e-01 4.55740094e-01
-1.25119090e+00 7.77903378e-01 3.93185705e-01 5.56315124e-01
-6.36010110e-01 -8.79886985e-01 -7.02413023e-01 -1.39674902e-01
-2.73481190e-01 -2.96661496e-01 9.39590573e-01 -6.51253164e-01
-1.59461772e+00 5.48734963e-01 1.18173294e-01 -2.88500100e-01
2.66190141e-01 -6.98375553e-02 -7.04569459e-01 5.77464581e-01
-3.06295782e-01 -3.92586976e-01 1.20480943e+00 -1.01712191e+00
-2.59685349e-02 -1.88925728e-01 3.44920754e-02 -1.47604510e-01
-5.30049622e-01 -1.53116137e-01 3.69083017e-01 -4.91688281e-01
3.66873622e-01 -1.01999819e+00 -2.01300040e-01 -5.80209494e-01
-3.02874804e-01 -7.18127191e-02 7.59687901e-01 -5.48256159e-01
1.71158433e+00 -2.50619078e+00 6.27586663e-01 5.50737500e-01
2.06353426e-01 -8.82929564e-02 1.22661069e-01 6.78400874e-01
-6.70425534e-01 -6.62469864e-02 -3.47227037e-01 1.97832972e-01
-3.25233792e-03 9.33779553e-02 -9.53677416e-01 1.23228478e+00
-1.65273726e-01 4.35823351e-01 -5.07188439e-01 -1.64297298e-01
2.01380923e-02 4.47698057e-01 -4.27817851e-01 2.17762729e-03
5.87298512e-01 1.73092529e-01 -3.21493894e-01 4.52915192e-01
8.91195476e-01 -2.97502607e-01 -1.78972244e-01 -4.72388655e-01
-4.41821128e-01 -4.03261632e-01 -1.61633134e+00 1.01763475e+00
-1.58869147e-01 5.44324577e-01 5.41488469e-01 -1.40802729e+00
6.49877608e-01 5.96324086e-01 1.00855660e+00 1.85770288e-01
-2.45945647e-01 1.76574528e-01 -2.12444976e-01 -5.66174269e-01
2.55678594e-01 -5.97690761e-01 1.54813472e-02 3.49812895e-01
1.98171943e-01 1.42470121e-01 -2.72167623e-02 2.07058832e-01
9.18642819e-01 -3.30989987e-01 2.76936591e-01 -9.51256812e-01
1.00124609e+00 -2.32605487e-02 3.61573905e-01 4.65923816e-01
-3.75670403e-01 3.74800980e-01 2.77153045e-01 -2.91718364e-01
-1.22951972e+00 -1.59240055e+00 -7.11366057e-01 6.35008633e-01
2.18798026e-01 -2.49972582e-01 -6.72622681e-01 -2.73372442e-01
2.10908398e-01 4.92702663e-01 -7.84438729e-01 -3.79226536e-01
-4.45443213e-01 -7.95784354e-01 9.06849504e-01 4.93922144e-01
4.44761306e-01 -4.21277553e-01 -3.34229082e-01 -1.19988635e-01
1.14314049e-01 -8.59195530e-01 -6.93912446e-01 9.14700031e-02
-1.19503570e+00 -1.22287953e+00 -8.88141930e-01 -8.18876266e-01
6.92261100e-01 3.74825478e-01 4.74001467e-01 -6.25877082e-01
-4.31501925e-01 1.25044084e+00 8.44976678e-03 2.01244578e-01
-3.32343102e-01 -6.14520609e-01 7.60471940e-01 1.93724483e-01
-2.52071340e-02 -7.19767749e-01 -2.28834674e-01 6.48858905e-01
-7.52762973e-01 -3.09121281e-01 3.79463792e-01 1.11913931e+00
5.52989841e-01 5.88707507e-01 2.63602912e-01 -2.69556433e-01
7.17144072e-01 -3.95123482e-01 -7.28067398e-01 1.47456348e-01
-6.21182621e-01 1.55354574e-01 6.87195063e-01 -7.67237067e-01
-9.94252324e-01 -1.16972245e-01 6.06515944e-01 -1.08997595e+00
4.49176371e-01 4.71078724e-01 2.80330718e-01 -3.58424038e-01
1.01574719e+00 7.01436996e-01 5.20209730e-01 -4.24925029e-01
2.84094661e-01 5.49832761e-01 5.22632003e-01 -7.82093108e-01
9.67579484e-01 8.92570257e-01 1.53057262e-01 -1.57031691e+00
-4.14899409e-01 -8.40820074e-01 -7.93824792e-01 -2.41997540e-01
5.27162969e-01 -7.01497257e-01 -1.01096952e+00 4.02818471e-01
-9.06488180e-01 -1.56582162e-01 -6.33481026e-01 8.75095189e-01
-1.20104444e+00 6.43921018e-01 -9.48921323e-01 -1.42069805e+00
2.18326189e-02 -5.85545719e-01 4.27269936e-01 -2.01993302e-01
3.54089332e-03 -1.39605618e+00 4.83918726e-01 1.30927479e-02
3.62312138e-01 4.13371343e-03 8.44422221e-01 -2.72119224e-01
-4.15236279e-02 -3.91194791e-01 1.38588667e-01 8.71147513e-01
-1.91949651e-01 2.28614688e-01 -5.21411657e-01 -5.19149303e-01
6.40940249e-01 1.11712150e-01 4.52720076e-01 6.95048928e-01
7.33587444e-01 -6.62929475e-01 -2.62234360e-01 6.00111246e-01
1.32047284e+00 7.76279997e-03 6.56814814e-01 -3.61571535e-02
4.37993139e-01 5.17167330e-01 3.35001439e-01 6.12367749e-01
-1.98937520e-01 2.63158232e-01 -4.25281748e-03 2.50530839e-01
8.08936596e-01 2.03429878e-01 8.94513965e-01 1.32904875e+00
-6.42971933e-01 6.56002581e-01 -5.27862906e-01 3.44610542e-01
-1.99627328e+00 -1.28738093e+00 -2.24014729e-01 2.26544952e+00
6.00404739e-01 -3.74525964e-01 4.83279288e-01 2.05886066e-01
1.00512898e+00 -1.35902599e-01 -4.66851503e-01 -4.55873609e-01
-3.16037118e-01 1.50517868e-02 7.96102524e-01 6.81869149e-01
-1.16033602e+00 4.15933192e-01 7.95037508e+00 5.73062956e-01
-7.67265975e-01 1.77205011e-01 -2.49318078e-01 2.54248977e-01
-9.01080817e-02 -3.40201445e-02 -6.04858518e-01 3.04818094e-01
9.07232285e-01 -3.76982301e-01 5.98677039e-01 1.25358021e+00
3.79459292e-01 1.32288694e-01 -7.16604829e-01 1.21299660e+00
-2.44393274e-01 -1.10822415e+00 -4.40783292e-01 4.17294204e-01
7.06654608e-01 -3.87840331e-01 2.37574309e-01 4.13866907e-01
1.02086468e-02 -1.06532383e+00 3.26546699e-01 9.75255668e-01
4.84994739e-01 -1.02530134e+00 5.10147333e-01 3.80467951e-01
-1.24780440e+00 -2.28824124e-01 -7.92638481e-01 -1.23108312e-01
-1.38054565e-01 7.79676735e-01 -3.92353714e-01 5.09000003e-01
4.43318009e-01 8.01057458e-01 1.40446350e-01 7.34280765e-01
4.71827418e-01 4.42830235e-01 -2.16268644e-01 2.96015918e-01
3.25776972e-02 -7.27827191e-01 8.21917117e-01 1.31554484e+00
4.17645097e-01 4.94927347e-01 1.57481670e-01 1.14284742e+00
7.46904314e-01 2.54334271e-01 -8.36504161e-01 -5.85901320e-01
1.54085770e-01 1.17983615e+00 -9.34261754e-02 -4.55341004e-02
-3.23903024e-01 1.05477488e+00 -5.17260507e-02 8.90907228e-01
-9.21404898e-01 -5.11099517e-01 1.18355119e+00 2.50492036e-01
2.00404838e-01 -6.19654536e-01 -1.32393792e-01 -1.49987984e+00
-1.09262794e-01 -4.86976057e-01 4.58171964e-01 -4.29982901e-01
-1.23375213e+00 9.94132608e-02 3.85179222e-01 -1.28124774e+00
-1.24385737e-01 -9.27442789e-01 -4.59640831e-01 9.33740139e-01
-4.66775805e-01 -7.26002097e-01 2.83028245e-01 9.75780547e-01
-2.31167123e-01 -3.98555785e-01 9.05454814e-01 1.67186439e-01
-8.02780390e-01 2.94735759e-01 5.89714587e-01 7.45987147e-02
-3.84093784e-02 -1.28385162e+00 -3.91681880e-01 6.09055638e-01
-4.07039046e-01 1.28067732e+00 9.62851584e-01 -3.65215808e-01
-1.76677895e+00 -7.98750937e-01 2.90648639e-01 -2.16272339e-01
1.08894813e+00 3.72472852e-02 -1.05371690e+00 8.19221377e-01
-1.24171160e-01 1.68973491e-01 7.42190182e-01 1.44744068e-01
-1.18732288e-01 1.48841381e-01 -9.50459242e-01 5.54905295e-01
7.76530921e-01 -6.73537314e-01 -3.67019296e-01 5.31238079e-01
4.76173908e-01 -2.53748000e-01 -1.42044401e+00 5.73651195e-01
6.51782811e-01 -9.89245653e-01 1.22891772e+00 -1.13473094e+00
-3.11897725e-01 -5.39994955e-01 -4.29044694e-01 -9.56691563e-01
-7.68109083e-01 -1.08386755e+00 -6.42123938e-01 6.19596720e-01
1.35606185e-01 -7.55344748e-01 3.08874160e-01 4.36670870e-01
-3.21420699e-01 -9.40013528e-01 -1.18568206e+00 -1.46163082e+00
1.68862745e-01 -3.46946239e-01 -1.18943021e-01 1.09471500e+00
8.77734840e-01 7.01778084e-02 -3.88666451e-01 5.03415287e-01
5.40282488e-01 -7.23102540e-02 4.42115843e-01 -1.02211809e+00
-4.94705588e-01 -3.68016779e-01 -7.52069294e-01 -7.79746771e-01
3.99142593e-01 -7.82626629e-01 -9.48462337e-02 -8.99066806e-01
5.69045283e-02 -4.10183191e-01 -3.10488880e-01 -1.83626786e-01
5.50160170e-01 -1.93121120e-01 -2.43781403e-01 5.83614111e-01
-2.34807059e-01 6.04023576e-01 1.16543007e+00 2.38317519e-01
-4.89149570e-01 2.05268770e-01 4.43843901e-02 8.83586526e-01
5.17819405e-01 1.98624626e-01 -7.20625818e-01 5.99923491e-01
3.12245749e-02 8.01710337e-02 4.45238978e-01 -1.00480223e+00
-9.27043781e-02 -4.41596597e-01 -2.14403532e-02 -4.54670012e-01
3.64079118e-01 -7.81321526e-01 3.81163895e-01 4.74837542e-01
-2.67723203e-01 1.97140887e-01 2.25193612e-02 7.90091872e-01
-9.56040770e-02 -3.51518661e-01 1.09940195e+00 1.41481265e-01
-3.88790518e-01 8.34577978e-02 -5.53476632e-01 9.42891985e-02
1.20218623e+00 -3.40943992e-01 1.45173252e-01 -4.23223317e-01
-1.24126399e+00 -2.40747258e-01 2.72703081e-01 -3.50483894e-01
7.40963280e-01 -1.61599267e+00 -2.50408173e-01 5.18886685e-01
-1.77555606e-01 -4.46009368e-01 5.22459924e-01 1.66039431e+00
-4.28552300e-01 7.38189042e-01 7.38643035e-02 -6.04778945e-01
-1.03641975e+00 8.30750644e-01 7.31836319e-01 5.40654510e-02
-9.27354455e-01 6.82505190e-01 4.24608350e-01 -3.60786051e-01
1.14846945e-01 -4.62098062e-01 6.09458573e-02 -1.70839176e-01
5.21716595e-01 9.83599484e-01 -1.73602149e-01 -8.30347121e-01
-2.63820589e-01 4.73969072e-01 3.56957525e-01 -3.66879344e-01
9.18199837e-01 -3.12894195e-01 -4.36591834e-01 7.89056182e-01
1.38934970e+00 2.64044940e-01 -9.94905531e-01 7.14621320e-02
-2.26719975e-01 -6.16313629e-02 -2.36381426e-01 7.82633275e-02
-6.39613807e-01 4.93145317e-01 4.30088162e-01 5.60569406e-01
1.04931664e+00 -1.60179913e-01 5.26816428e-01 5.65092564e-01
2.41543397e-01 -1.24577880e+00 -5.83662502e-02 6.59493446e-01
1.17201304e+00 -5.44069588e-01 -2.93752015e-01 -5.97022593e-01
-6.54000998e-01 1.51053011e+00 3.35041612e-01 -5.89456379e-01
1.20901418e+00 9.27812308e-02 -3.94027770e-01 -3.10428515e-02
-5.03884554e-01 1.45755708e-02 3.74323457e-01 5.58876455e-01
3.83307844e-01 7.46614262e-02 -3.76276433e-01 5.67601919e-01
7.88227841e-02 -4.14360166e-01 6.50854826e-01 7.20374405e-01
-5.67703605e-01 -3.45086366e-01 -7.34044194e-01 -4.15038615e-02
-4.31475416e-02 1.11641884e-02 4.30054516e-02 9.90793228e-01
-3.36440623e-01 4.69467878e-01 -1.68760851e-01 -4.46234792e-01
2.10202366e-01 2.86746204e-01 6.01145446e-01 -3.79429549e-01
1.32709771e-01 1.19537368e-01 -2.62058258e-01 -4.51680988e-01
-2.40257621e-01 -7.13320553e-01 -1.52052987e+00 -5.09083748e-01
-4.10897970e-01 6.03175998e-01 2.53075600e-01 8.55051517e-01
2.58156210e-01 6.64150491e-02 9.03955579e-01 -3.52539003e-01
-1.28873634e+00 -9.15205777e-01 -1.52346754e+00 1.57235652e-01
5.65155268e-01 -8.69548023e-01 -9.25301850e-01 1.63886145e-01] | [7.39553689956665, 4.151216983795166] |
c6e3fbd3-ce8e-47c0-8385-a1e20778a47f | an-enhanced-object-detection-model-for-scene | null | null | https://link.springer.com/chapter/10.1007/978-3-031-20601-6_30 | https://rdcu.be/c0bJi | An Enhanced Object Detection Model for Scene Graph Generation | With computer vision improving, a higher level of understanding is needed to solve more complex problems such as semantic image retrieval, image captioning, and scene understanding. Scene understanding has been a long-studied problem due to its complexity and lack of proper data representation. A scene Graph is one of the most powerful data representations that can better understand the scene context. The task of a Scene Graph is to encode the objects presented in the scene, their attributes, as long as the relationships between these objects. With the scene Graph proving its capabilities in complicated tasks, the automation of scene graph generation became a must. Great research has been made to obtain accurate Scene Graphs using different deep learning architectures. The common module among those different architectures is the object detection module, where objects are firstly located in the input image. In this work, we propose using the most recent object detectors from the YOLOv5 family for the scene graph generation task. The proposed YOLOv5x6 achieved a State-Of-The-Art result of 32.7 mean average precision compared to previous works. Furthermore, the paper reviews the different object detectors used in literature for the scene graph generation. | ['Mohamed F. Tolba', 'Howida A. Shedeed', 'Dina Khattab', 'Mohammad Essam'] | 2022-11-18 | null | null | null | international-conference-on-advanced | ['scene-graph-generation'] | ['computer-vision'] | [ 3.65150541e-01 -8.70614797e-02 1.97679311e-01 -4.76923347e-01
-1.10709749e-01 -3.19645494e-01 7.34942734e-01 5.30374825e-01
-4.03533250e-01 3.62877607e-01 -1.49680927e-01 -7.41735771e-02
-2.74292052e-01 -1.00764549e+00 -7.11639702e-01 -5.35932958e-01
2.08006248e-01 4.43728179e-01 5.50984383e-01 -2.35135272e-01
4.44898963e-01 5.26007652e-01 -1.91232121e+00 3.67674857e-01
6.50931835e-01 1.03617275e+00 8.57673407e-01 6.29781842e-01
-5.09822726e-01 8.56589139e-01 -6.03700459e-01 -2.59963602e-01
4.22002263e-02 -3.22962999e-01 -7.41685450e-01 3.48717302e-01
5.65760911e-01 -7.38558918e-02 -3.53130609e-01 1.25453913e+00
9.90066156e-02 1.98444217e-01 6.33805156e-01 -1.41152596e+00
-6.64064229e-01 4.15992737e-01 -2.99622655e-01 2.05802560e-01
3.19254458e-01 -1.26051828e-01 8.79695237e-01 -8.14756989e-01
5.39732933e-01 1.33365250e+00 3.47613133e-02 3.95234168e-01
-7.99631417e-01 -4.10404921e-01 9.73544270e-02 8.53013873e-01
-1.40100086e+00 -8.23859870e-02 9.13861454e-01 -5.29502988e-01
7.98009276e-01 1.80213273e-01 5.89516819e-01 5.41532040e-01
-1.89284328e-02 7.55482256e-01 9.65942085e-01 -6.43245339e-01
2.96525240e-01 3.70600730e-01 4.30273056e-01 6.33749783e-01
4.58601326e-01 -4.22561646e-01 -2.55658567e-01 3.67357910e-01
6.89599991e-01 1.64514631e-01 -7.26820529e-02 -5.33687890e-01
-9.20779228e-01 6.77199364e-01 9.92519617e-01 4.64253843e-01
-2.93849319e-01 1.76742896e-01 3.06421876e-01 -1.98164016e-01
-8.38257372e-02 6.03906989e-01 -6.48075640e-02 2.28580490e-01
-5.89518011e-01 2.87542462e-01 5.13185143e-01 1.03411937e+00
7.15172410e-01 4.37775031e-02 -1.70399159e-01 5.54818571e-01
4.39413309e-01 4.87220466e-01 2.49559164e-01 -4.96700972e-01
5.16193271e-01 1.23728061e+00 -1.25180542e-01 -1.46446419e+00
-4.17446315e-01 -4.30743068e-01 -8.53007853e-01 6.87824041e-02
3.62074435e-01 4.07012522e-01 -1.19155085e+00 1.36994195e+00
3.18133622e-01 9.05185938e-02 1.44455507e-01 1.04303014e+00
1.29403317e+00 9.57488894e-01 2.58809507e-01 2.89704025e-01
1.65513980e+00 -8.81113708e-01 -5.28804362e-01 -5.07136106e-01
3.82599086e-01 -8.14100444e-01 8.62979293e-01 1.85089603e-01
-4.98381823e-01 -8.91605735e-01 -1.14475894e+00 -2.67746836e-01
-8.46489012e-01 3.27760160e-01 8.47880125e-01 4.96653765e-01
-9.56338108e-01 1.86874211e-01 -3.94380510e-01 -8.11617672e-01
4.74214405e-01 2.45123819e-01 -4.71413881e-01 -3.00852567e-01
-9.42614377e-01 9.18233216e-01 9.74318504e-01 2.43676484e-01
-8.59393477e-01 -1.65307611e-01 -9.97351468e-01 3.63491923e-01
4.65310186e-01 -6.85685337e-01 8.95220697e-01 -7.58387864e-01
-7.95838952e-01 1.11124718e+00 -5.78975258e-03 -6.25066638e-01
1.08019821e-01 -2.79017508e-01 -3.06420118e-01 1.41579628e-01
3.62194851e-02 7.37835526e-01 8.40721667e-01 -1.24496245e+00
-6.06051624e-01 -4.92712706e-01 2.27930859e-01 3.70297283e-01
-6.93332916e-03 -1.53384760e-01 -5.62773585e-01 -2.47575358e-01
2.95346111e-01 -8.42045248e-01 -2.20263049e-01 -9.85347331e-02
-5.17183006e-01 -4.70889539e-01 1.20243025e+00 -3.79472196e-01
9.41909075e-01 -2.04585195e+00 -5.11183701e-02 -2.79515255e-02
2.44656637e-01 6.07610345e-01 -1.06459908e-01 5.25461197e-01
-2.51366884e-01 -6.75906837e-02 -2.09799081e-01 -9.32051539e-02
-2.36304224e-01 8.70129317e-02 -2.64708400e-01 1.27293900e-01
2.99059063e-01 6.95990682e-01 -7.58687794e-01 -5.52765667e-01
9.44980383e-01 3.71094495e-01 -4.48853612e-01 3.09736639e-01
-5.64492285e-01 3.29493523e-01 -7.22075343e-01 3.67952704e-01
6.00187123e-01 -4.16239232e-01 -6.13950565e-03 -3.60228211e-01
1.55521929e-01 -1.96332008e-01 -1.27012777e+00 1.65799928e+00
-2.60801166e-01 9.73536551e-01 -2.93398917e-01 -1.28273368e+00
1.20065057e+00 3.32966587e-03 3.21088344e-01 -8.44158292e-01
4.04429883e-01 1.54510671e-02 1.38261050e-01 -6.61357284e-01
5.17498851e-01 2.82823257e-02 -2.31922232e-02 1.39941536e-02
3.55501696e-02 -5.21506786e-01 5.45416772e-01 3.97749662e-01
7.14549601e-01 -9.38319564e-02 5.47450662e-01 -2.00933471e-01
8.27992857e-01 3.01317990e-01 2.81008636e-03 7.01466560e-01
2.33193599e-02 5.67766309e-01 3.06578308e-01 -6.41862273e-01
-1.06409299e+00 -8.17865968e-01 1.00057116e-02 7.05790520e-01
4.98514622e-01 -3.27593327e-01 -7.61566937e-01 -2.33706892e-01
-1.36490896e-01 7.22137034e-01 -4.55962479e-01 -1.20144255e-01
-4.02232945e-01 -5.45967042e-01 5.41592315e-02 3.04985672e-01
1.00641310e+00 -1.40736830e+00 -9.02152658e-01 1.16453744e-01
-9.37572960e-03 -1.56499720e+00 1.44567549e-01 -1.39179230e-01
-7.97781467e-01 -1.40410221e+00 -4.74284351e-01 -1.01069522e+00
8.29392970e-01 5.32087147e-01 9.77645695e-01 2.03556925e-01
-8.03483427e-01 2.55664974e-01 -5.37474036e-01 -7.24710464e-01
-4.09909815e-01 -7.54776001e-02 -2.08791047e-01 2.67074674e-01
3.83492917e-01 -2.10870996e-01 -4.95072544e-01 1.12711722e-02
-1.17854369e+00 4.00067776e-01 7.13780999e-01 4.53647673e-01
5.18164754e-01 2.28754342e-01 3.58389020e-01 -9.64395940e-01
4.50990617e-01 -2.40816802e-01 -7.50504255e-01 3.93835694e-01
-1.75224334e-01 3.42299879e-01 6.42518044e-01 1.22890048e-01
-1.02092826e+00 2.87950844e-01 -9.47093312e-03 -1.98794752e-01
-4.76300716e-01 2.25563183e-01 -3.65755856e-01 1.03187993e-01
5.97700000e-01 4.11359757e-01 -3.21881801e-01 -4.03564006e-01
6.55910194e-01 5.49711525e-01 4.56899166e-01 -2.74137706e-01
5.50785601e-01 4.64852393e-01 2.40894467e-01 -1.20358729e+00
-1.00306952e+00 -8.23705852e-01 -7.77944148e-01 -3.10842484e-01
1.17992783e+00 -6.03321195e-01 -7.10546553e-01 5.60581923e-01
-1.44299603e+00 4.70592417e-02 -6.34939075e-02 3.01814795e-01
-4.07131493e-01 3.48643422e-01 1.67079512e-02 -7.75378942e-01
-3.47294152e-01 -1.13715422e+00 1.20616519e+00 5.86393893e-01
1.54673174e-01 -7.35640645e-01 -2.90993422e-01 4.01475072e-01
2.82965243e-01 3.11576366e-01 1.15118432e+00 -5.36277652e-01
-1.00852263e+00 -2.18576163e-01 -7.16321588e-01 2.43921220e-01
2.01324075e-01 -1.31724820e-01 -1.03336477e+00 3.79758924e-02
1.18884901e-02 5.15196361e-02 7.72834003e-01 3.57398093e-01
1.30452740e+00 -7.90573657e-02 -3.77258688e-01 4.20710981e-01
1.66967618e+00 5.33483267e-01 7.52754033e-01 3.70075464e-01
8.85025203e-01 6.82850301e-01 5.99401593e-01 2.30028301e-01
2.17510834e-01 7.96853483e-01 6.82340145e-01 -1.44632414e-01
-4.04547632e-01 -2.86079377e-01 -1.04442649e-01 5.17021120e-01
1.41583368e-01 -4.40492898e-01 -9.97210205e-01 4.49577451e-01
-1.80386055e+00 -9.18073118e-01 -4.86564189e-01 1.87249231e+00
1.17021501e-01 1.94392920e-01 -2.57399410e-01 3.48269939e-01
7.98084438e-01 1.77099779e-01 -4.39001262e-01 -3.54739964e-01
-8.56522471e-02 1.79813355e-02 2.84558475e-01 2.72230029e-01
-1.25485039e+00 1.28443325e+00 5.32192135e+00 6.75924778e-01
-1.02784002e+00 -3.34195226e-01 5.04494905e-01 5.28196871e-01
1.93292320e-01 -3.68390568e-02 -7.71374941e-01 1.39842972e-01
3.22577953e-01 -2.90918142e-01 1.86922714e-01 1.18178701e+00
6.37734756e-02 -4.59418476e-01 -1.18796301e+00 1.42244267e+00
3.59006435e-01 -1.19903183e+00 5.54796457e-01 -3.57436091e-01
6.22395098e-01 -3.10382038e-01 -1.40988693e-01 9.66394544e-02
-6.05384223e-02 -1.00567067e+00 5.82418442e-01 5.58514237e-01
4.81577992e-01 -3.80140841e-01 8.60864639e-01 3.76264393e-01
-1.40343738e+00 -2.34391112e-02 -6.08645320e-01 -1.64252356e-01
2.33134136e-01 6.21840060e-01 -1.38578653e+00 5.25162995e-01
5.97169280e-01 7.21608818e-01 -9.80048418e-01 1.36234093e+00
-3.56880635e-01 2.44226992e-01 -1.95014104e-01 -4.64143723e-01
3.14332485e-01 -2.17378527e-01 4.27065283e-01 9.76472080e-01
1.80043861e-01 2.32992455e-01 1.11160859e-01 9.53815758e-01
3.61413509e-02 2.14939132e-01 -9.70263958e-01 -1.30536735e-01
2.19813585e-01 1.31409216e+00 -1.11394262e+00 -5.74003279e-01
-3.36701423e-02 8.19048524e-01 1.66491270e-01 1.99573636e-01
-5.92060745e-01 -4.12127703e-01 3.20994377e-01 1.26053080e-01
1.96461584e-02 -5.66603839e-01 -1.71015561e-01 -7.85153627e-01
-2.11825147e-02 -5.41461766e-01 4.57923442e-01 -1.24629855e+00
-9.60639536e-01 7.11502254e-01 2.49429911e-01 -1.03746605e+00
-7.60378465e-02 -1.03018606e+00 -4.34382498e-01 6.48646533e-01
-1.26564062e+00 -1.04526997e+00 -8.39385986e-01 6.19093239e-01
9.41718519e-01 -4.01434749e-02 6.93852484e-01 2.27530643e-01
-2.46949226e-01 -1.62238449e-01 -2.38211393e-01 2.51521647e-01
1.37848958e-01 -1.06246758e+00 4.50555980e-01 9.58981931e-01
6.92041755e-01 3.82012159e-01 6.39751017e-01 -5.28750300e-01
-1.41301596e+00 -9.13369417e-01 7.73088396e-01 -4.38722014e-01
3.45595241e-01 -5.41792035e-01 -8.89887393e-01 3.98217857e-01
7.93793947e-02 -1.46225244e-01 2.52504423e-02 -3.01863104e-01
-1.16551749e-01 -4.03218538e-01 -8.36111128e-01 6.50204062e-01
8.85086894e-01 -3.66639316e-01 -6.50913239e-01 3.44039977e-01
7.79029131e-01 -2.71776766e-01 -1.38646632e-01 3.00363272e-01
1.75059184e-01 -9.93955672e-01 9.25407171e-01 -5.48048139e-01
1.96905434e-01 -6.32498443e-01 -1.35021463e-01 -9.63723063e-01
-1.13520250e-01 7.48632401e-02 3.42074990e-01 9.85139430e-01
1.37893647e-01 -2.87805498e-01 8.36877704e-01 4.81757820e-01
-1.32820979e-01 -3.33631396e-01 -5.82210541e-01 -4.64801431e-01
-7.22928703e-01 -5.71364939e-01 4.40621048e-01 6.24819934e-01
-4.64619517e-01 7.53152370e-01 -1.09929591e-01 2.06636518e-01
6.50727212e-01 3.31361800e-01 9.47105467e-01 -1.38972545e+00
3.06758508e-02 -3.97737324e-01 -1.04426599e+00 -9.07795668e-01
-2.07755357e-01 -9.46961820e-01 -4.58103828e-02 -2.20211291e+00
2.77026951e-01 -4.07081991e-01 7.91676808e-03 2.32092038e-01
-1.89296216e-01 -1.27905563e-01 6.41460299e-01 -8.99832919e-02
-8.25452745e-01 5.49737871e-01 1.45845020e+00 -3.16224426e-01
-1.02350682e-01 -8.31024498e-02 -3.78667057e-01 7.38521039e-01
8.22395146e-01 -3.92813653e-01 -6.23885393e-01 -4.58324611e-01
1.06172800e-01 -1.67743996e-01 5.18901050e-01 -1.35885715e+00
4.41273898e-01 2.16621850e-02 2.86057204e-01 -8.43855083e-01
5.18731356e-01 -1.08788514e+00 2.32125223e-01 5.87337852e-01
-4.17285226e-02 1.10582940e-01 2.88506597e-01 4.62643683e-01
-4.41826016e-01 -3.49977612e-01 7.29100883e-01 -4.34981048e-01
-1.60048020e+00 2.05318376e-01 1.92883052e-02 -1.33429617e-01
1.33726704e+00 -2.30966300e-01 -3.74817878e-01 -3.68528515e-01
-6.60575807e-01 1.38491035e-01 5.50297014e-02 8.23450267e-01
9.22926605e-01 -9.90122676e-01 -5.77353835e-01 1.43942311e-01
4.45591450e-01 2.41903901e-01 4.44789350e-01 2.77337343e-01
-8.68421853e-01 7.32460499e-01 -3.87193739e-01 -9.23351586e-01
-1.43805015e+00 7.59547532e-01 2.60555744e-01 -2.90057715e-02
-5.96305370e-01 7.39051104e-01 6.57818615e-01 -5.59117049e-02
1.04477577e-01 -5.10363162e-01 -6.16831005e-01 -1.58116594e-01
4.19920266e-01 2.00233653e-01 -2.64484901e-03 -7.57848501e-01
-4.83161122e-01 9.15159166e-01 1.41083315e-01 2.25385413e-01
1.09946048e+00 7.60869682e-02 -2.11340353e-01 2.70150155e-01
9.92989600e-01 -5.28059185e-01 -6.84517801e-01 -1.67494610e-01
2.73660004e-01 -3.96848679e-01 -1.47143021e-01 -5.24419546e-01
-8.53997767e-01 1.29946327e+00 6.88343942e-01 4.77303624e-01
1.16425240e+00 2.36667618e-01 3.33395571e-01 5.47943413e-01
5.92829466e-01 -7.42654502e-01 2.68991679e-01 5.05476475e-01
9.91815805e-01 -1.36208355e+00 1.22963205e-01 -7.12553382e-01
-5.94468355e-01 1.11867964e+00 7.26713657e-01 -2.67551333e-01
4.88880336e-01 -2.28215232e-01 -1.14574283e-01 -5.70104897e-01
-2.22069174e-01 -6.23381615e-01 7.02035606e-01 5.54114997e-01
1.69978201e-01 2.17197299e-01 -1.57671750e-01 3.03698033e-01
-2.36078575e-01 -2.06464455e-01 3.28740209e-01 4.87717271e-01
-7.51856863e-01 -9.33396935e-01 -4.15455669e-01 3.93776953e-01
-1.35518417e-01 -3.68594117e-02 -5.66935539e-01 8.72527957e-01
1.93819299e-01 1.00721824e+00 6.69833124e-02 -1.40796006e-01
4.98858273e-01 -2.35723957e-01 4.53681856e-01 -9.23620224e-01
-8.69725496e-02 -5.30892670e-01 -1.38302684e-01 -4.65251356e-01
-6.01778090e-01 -1.57078683e-01 -1.42745125e+00 2.35729575e-01
-2.40356430e-01 4.58002537e-02 1.13461578e+00 1.02159405e+00
1.96801990e-01 6.44481242e-01 3.36548954e-01 -6.00824237e-01
1.97856620e-01 -9.40572441e-01 -4.14707899e-01 8.40510130e-01
4.14995439e-02 -7.25480378e-01 8.08689892e-02 2.04467744e-01] | [10.251348495483398, 1.5086472034454346] |
b2f81d79-9335-453b-a1be-edf816c3e7ed | see-better-before-looking-closer-weakly | 1901.09891 | null | http://arxiv.org/abs/1901.09891v2 | http://arxiv.org/pdf/1901.09891v2.pdf | See Better Before Looking Closer: Weakly Supervised Data Augmentation Network for Fine-Grained Visual Classification | Data augmentation is usually adopted to increase the amount of training data,
prevent overfitting and improve the performance of deep models. However, in
practice, random data augmentation, such as random image cropping, is
low-efficiency and might introduce many uncontrolled background noises. In this
paper, we propose Weakly Supervised Data Augmentation Network (WS-DAN) to
explore the potential of data augmentation. Specifically, for each training
image, we first generate attention maps to represent the object's
discriminative parts by weakly supervised learning. Next, we augment the image
guided by these attention maps, including attention cropping and attention
dropping. The proposed WS-DAN improves the classification accuracy in two
folds. In the first stage, images can be seen better since more discriminative
parts' features will be extracted. In the second stage, attention regions
provide accurate location of object, which ensures our model to look at the
object closer and further improve the performance. Comprehensive experiments in
common fine-grained visual classification datasets show that our WS-DAN
surpasses the state-of-the-art methods, which demonstrates its effectiveness. | ['Tao Hu', 'Yan Lu', 'Honggang Qi', 'Qingming Huang'] | 2019-01-26 | null | null | null | null | ['image-cropping'] | ['computer-vision'] | [ 1.01947613e-01 -5.42990267e-02 -2.17141241e-01 -3.30123067e-01
-2.35785693e-01 -1.57666981e-01 3.54006737e-01 3.13009024e-02
-4.10553992e-01 5.89840114e-01 3.81018072e-01 -9.41664074e-03
2.90700287e-01 -7.11355329e-01 -7.66676188e-01 -9.56458747e-01
3.79799247e-01 2.41401605e-02 3.92413527e-01 -4.23741043e-02
6.89804852e-02 5.17318726e-01 -1.42401457e+00 1.89415604e-01
9.57685173e-01 1.06809211e+00 5.83046496e-01 1.70363024e-01
-3.57175261e-01 8.53474438e-01 -5.49870253e-01 -2.03784276e-02
8.96872804e-02 -2.07036734e-01 -7.05455601e-01 3.37933153e-01
3.19742858e-01 -4.69849646e-01 -5.06047249e-01 1.12567174e+00
3.17561328e-01 1.86181024e-01 3.51313353e-01 -1.20492315e+00
-1.14067757e+00 3.74629498e-01 -1.01620042e+00 3.60868394e-01
-2.45888412e-01 3.58814418e-01 7.67242074e-01 -1.04684627e+00
2.16229022e-01 1.28855968e+00 3.62776518e-01 7.30177641e-01
-1.14933360e+00 -8.48636508e-01 7.54001796e-01 6.07183091e-02
-1.30193484e+00 -2.80301541e-01 9.28277910e-01 -2.71272570e-01
3.58013719e-01 2.29284585e-01 4.71285790e-01 9.28883374e-01
-3.79997611e-01 1.26053631e+00 8.15160275e-01 -4.28141147e-01
-4.94756959e-02 1.69070363e-01 1.43259317e-01 5.97554803e-01
1.93389192e-01 -1.39391363e-01 -2.48909503e-01 2.10780233e-01
9.70727980e-01 5.07559001e-01 -3.30628902e-01 -3.34650427e-01
-1.29019427e+00 5.16465843e-01 1.12060893e+00 4.10842538e-01
-5.64885914e-01 1.46577612e-01 2.33052477e-01 -2.43559584e-01
5.00870526e-01 3.21985483e-01 -4.40162718e-01 3.15159023e-01
-6.51374996e-01 7.12964460e-02 -1.64181426e-01 8.11065555e-01
7.69081473e-01 1.22305609e-01 -6.06017113e-01 9.92897570e-01
3.70322913e-01 2.24597692e-01 6.22004569e-01 -4.77131099e-01
5.55604637e-01 9.89730000e-01 7.56833330e-02 -8.88687313e-01
-2.61069745e-01 -8.22702706e-01 -1.25731003e+00 1.33094564e-01
1.96037322e-01 -9.07015577e-02 -1.32583165e+00 1.74003065e+00
2.15579256e-01 2.11227208e-01 -1.21410005e-01 1.06821048e+00
9.41633523e-01 7.40511000e-01 4.79047298e-01 -3.39837633e-02
1.36858010e+00 -1.31052721e+00 -8.01523805e-01 -6.68711901e-01
4.49538916e-01 -6.10543489e-01 1.40215898e+00 1.60626203e-01
-7.64944375e-01 -1.07162499e+00 -7.93402195e-01 -1.38503343e-01
-2.19435260e-01 5.05425870e-01 6.82315767e-01 2.33006701e-01
-5.94280064e-01 3.30825597e-01 -6.73025787e-01 -1.16993517e-01
1.10968149e+00 1.91572681e-01 -3.65799159e-01 -3.38050991e-01
-9.59210753e-01 5.18470168e-01 3.97251129e-01 3.15195799e-01
-9.64207590e-01 -5.59246063e-01 -8.20201099e-01 1.90897807e-01
2.56100386e-01 -3.37605953e-01 1.05694330e+00 -1.06048930e+00
-9.64301884e-01 6.30412221e-01 -1.89543396e-01 -1.74002826e-01
3.03556621e-01 -3.79184991e-01 -1.54002070e-01 -1.93876758e-01
2.39768043e-01 1.17066479e+00 8.64504874e-01 -1.54145789e+00
-6.47004128e-01 -3.99721652e-01 7.50171905e-03 1.84506714e-01
-6.49526834e-01 -2.46586770e-01 -7.74356067e-01 -1.00995851e+00
6.01550490e-02 -6.29522979e-01 -5.10520458e-01 3.17556672e-02
-4.97020870e-01 -2.91584522e-01 1.04083490e+00 -5.09057879e-01
1.28557003e+00 -2.65929127e+00 -1.21507794e-01 -5.13261259e-02
2.44611740e-01 6.12624764e-01 -3.46317232e-01 -1.08672038e-01
-2.70624906e-01 3.51404101e-01 -5.98231368e-02 -3.35106373e-01
-3.84021550e-01 3.72421533e-01 -2.77289480e-01 2.18889326e-01
7.34216213e-01 1.17261875e+00 -7.22691715e-01 -3.93644303e-01
2.63452768e-01 2.60744482e-01 -5.40935636e-01 3.41124594e-01
-2.29338244e-01 4.67346340e-01 -6.27568722e-01 5.34697473e-01
7.77123868e-01 -4.95809644e-01 -3.72496784e-01 -3.87412101e-01
-2.70640310e-02 -4.33542170e-02 -8.88826311e-01 1.50298166e+00
-3.13675135e-01 6.53684437e-01 -2.55514324e-01 -9.60683107e-01
9.97041047e-01 -1.98766068e-01 1.14517594e-02 -9.36387062e-01
1.87250689e-01 -2.52052128e-01 1.44054949e-01 -6.47929370e-01
3.21423769e-01 1.25079304e-01 2.87902951e-01 9.79311317e-02
-1.58736989e-01 4.91584569e-01 -6.75629303e-02 1.05812825e-01
5.35946667e-01 1.47661954e-01 3.42360735e-02 -1.60621643e-01
4.71343994e-01 -1.60984010e-01 6.79466128e-01 6.61849022e-01
-1.24726623e-01 6.56731129e-01 3.28366011e-01 -4.68577504e-01
-9.83234048e-01 -5.97580671e-01 -3.76705378e-02 1.30829525e+00
4.49873388e-01 1.42754763e-02 -8.55325878e-01 -9.38442171e-01
-3.48903015e-02 5.63401103e-01 -9.82480526e-01 -4.29918498e-01
-4.84889150e-01 -9.11485076e-01 2.26092249e-01 1.09987104e+00
8.09424818e-01 -1.34418726e+00 -3.03707272e-01 8.65286142e-02
-2.18489498e-01 -9.51649308e-01 -5.42091131e-01 1.17776878e-01
-9.34107721e-01 -8.99474084e-01 -9.73276854e-01 -1.09608257e+00
1.18276858e+00 6.82109058e-01 8.31283510e-01 4.84632760e-01
-1.52692810e-01 -3.51230681e-01 -5.03272891e-01 -4.75812614e-01
7.05061182e-02 6.83786813e-03 -5.36605604e-02 2.66962260e-01
4.69481289e-01 -2.57457525e-01 -8.07346344e-01 3.06698263e-01
-9.80088174e-01 2.11852223e-01 9.97556925e-01 9.49257314e-01
6.65292680e-01 4.39080298e-02 6.22186959e-01 -7.86147654e-01
4.54238892e-01 -3.02091867e-01 -3.33028555e-01 1.20526351e-01
-2.18545586e-01 1.70117840e-01 7.07160473e-01 -7.70161152e-01
-1.00434983e+00 9.74198431e-02 -1.97425708e-01 -5.61178505e-01
-3.35674316e-01 3.53448570e-01 -4.55109417e-01 -6.20729551e-02
5.41611195e-01 3.97690177e-01 -8.75661522e-02 -6.91260338e-01
2.75753647e-01 7.12131679e-01 4.87197727e-01 -3.34506601e-01
7.60304213e-01 3.63421351e-01 -2.87871122e-01 -5.85032940e-01
-1.18497360e+00 -2.97236204e-01 -5.56869507e-01 6.01712195e-03
9.06050742e-01 -9.10596371e-01 -5.41725755e-01 5.46611965e-01
-1.02012384e+00 -3.61229211e-01 -3.41797173e-01 3.77348751e-01
1.07336521e-01 1.61290154e-01 -3.63421917e-01 -7.83434510e-01
-2.92103648e-01 -1.12122726e+00 1.09878528e+00 7.01454043e-01
1.89768434e-01 -6.15215480e-01 -5.29585779e-01 1.66623920e-01
3.56659800e-01 2.10895371e-02 8.98996651e-01 -7.23141849e-01
-5.72303414e-01 -2.35184550e-01 -7.73832262e-01 6.09787762e-01
2.70971030e-01 -9.56497043e-02 -1.15663922e+00 -1.60518825e-01
-2.41616949e-01 -2.94017226e-01 9.12247002e-01 4.03817087e-01
1.86843526e+00 -3.40436131e-01 -5.54867744e-01 5.07298827e-01
1.13578510e+00 2.36519903e-01 7.64771879e-01 4.47369128e-01
9.04605925e-01 4.60214108e-01 8.68767619e-01 1.88732564e-01
1.03969678e-01 4.25163537e-01 6.99950576e-01 -8.14136147e-01
-2.30069175e-01 -3.51841122e-01 -2.70238407e-02 4.08458292e-01
-1.92582048e-02 -7.26124942e-02 -6.98373973e-01 6.90248609e-01
-1.75335085e+00 -5.90100646e-01 -1.37871176e-01 1.98325503e+00
6.50983095e-01 3.67672235e-01 -1.05745941e-02 6.62781298e-02
7.89140880e-01 8.25240538e-02 -6.43402398e-01 2.49706864e-01
-7.80270770e-02 4.65778671e-02 2.40697667e-01 8.88088122e-02
-1.13202929e+00 1.01640761e+00 5.41010761e+00 9.22052443e-01
-1.10351670e+00 -4.50819405e-03 1.17766058e+00 -2.45746821e-02
-1.35473117e-01 -3.13685387e-01 -8.09748352e-01 6.58109009e-01
1.91680238e-01 2.40873367e-01 1.85586169e-01 1.02948558e+00
1.92953914e-01 8.75265226e-02 -8.48818541e-01 8.85167718e-01
-5.29778302e-02 -1.28755641e+00 3.56677532e-01 3.96739803e-02
7.39309549e-01 -1.19923107e-01 9.02663246e-02 4.74731028e-01
1.48713142e-01 -1.02971613e+00 5.29570282e-01 5.46941936e-01
6.17253125e-01 -8.25886667e-01 9.11906838e-01 5.24819493e-01
-1.04133785e+00 -2.61676818e-01 -5.56501448e-01 3.82128023e-02
-8.04285780e-02 4.48784053e-01 -3.80429000e-01 2.42072433e-01
8.66388917e-01 5.93554437e-01 -8.98410916e-01 1.16653597e+00
-3.01997840e-01 5.89215875e-01 -7.73952976e-02 -1.42991450e-02
5.87270856e-01 -2.45132507e-03 1.53878042e-02 1.14545727e+00
3.13722007e-02 3.29913080e-01 3.55082482e-01 7.90343285e-01
-3.96282762e-01 -1.98358744e-02 -3.01114798e-01 4.69356440e-02
3.51237029e-01 1.25963795e+00 -6.65841997e-01 -4.20058578e-01
-4.43775386e-01 9.45575595e-01 4.49101359e-01 5.05014420e-01
-8.60739529e-01 -4.97270763e-01 6.09205365e-01 1.93470657e-01
3.20398957e-01 6.60437271e-02 -3.05095583e-01 -1.03958774e+00
4.85777259e-02 -8.11610162e-01 2.15238154e-01 -1.04565167e+00
-1.19896901e+00 7.47769594e-01 -4.14696634e-01 -1.18860054e+00
4.24080372e-01 -5.53672016e-01 -6.60783947e-01 1.01359892e+00
-1.51297331e+00 -1.29253626e+00 -8.16287637e-01 4.47269708e-01
8.15200686e-01 -3.14299911e-02 4.38093096e-01 4.29294169e-01
-8.39825273e-01 6.43078327e-01 -7.94459432e-02 5.08880615e-01
5.69938958e-01 -1.06989872e+00 4.75737929e-01 1.03402233e+00
1.47116616e-01 7.23196566e-01 3.74098897e-01 -5.92045009e-01
-7.33714163e-01 -1.45215321e+00 3.65975887e-01 -1.22070380e-01
3.84480804e-01 -3.75080317e-01 -1.35518050e+00 5.91786802e-01
5.89218326e-02 4.96684253e-01 4.90849733e-01 5.06855128e-03
-1.90892294e-01 -3.90464097e-01 -9.39773262e-01 6.93766356e-01
9.61251855e-01 -2.56600916e-01 -5.29699981e-01 3.29299748e-01
8.79425466e-01 -2.12020248e-01 -4.25323308e-01 5.45792043e-01
2.20612958e-01 -6.09831572e-01 9.90151584e-01 -1.02328348e+00
5.12839675e-01 -5.09351552e-01 8.98598954e-02 -1.25120044e+00
-7.54468858e-01 -1.51448607e-01 -8.26005414e-02 1.53391349e+00
2.44696602e-01 -3.66309136e-01 8.50280583e-01 3.97048086e-01
-1.17561914e-01 -9.66497302e-01 -4.49258357e-01 -5.21796167e-01
-4.79283407e-02 -2.44059876e-01 7.56398678e-01 9.55313087e-01
-4.38547373e-01 4.33769464e-01 -4.02191013e-01 3.20442140e-01
4.36842054e-01 1.20256282e-01 7.16607809e-01 -1.14729571e+00
8.32110494e-02 -4.52827513e-01 -2.97700256e-01 -1.50066912e+00
-9.75130349e-02 -5.02070844e-01 1.18269913e-01 -1.63141370e+00
4.23193425e-01 -7.55595148e-01 -5.89552164e-01 8.80000472e-01
-7.69566059e-01 5.53201497e-01 1.22094952e-01 1.82407513e-01
-6.82701707e-01 7.73008168e-01 1.66463101e+00 -1.69716403e-01
-1.25858635e-01 -6.48681670e-02 -1.08565640e+00 8.47479522e-01
7.63177812e-01 -2.91730970e-01 -2.96741635e-01 -5.79581201e-01
-2.90338457e-01 -4.01150823e-01 4.68611747e-01 -8.71541500e-01
9.45212469e-02 -1.86527133e-01 1.00599003e+00 -5.71697414e-01
-1.75707939e-03 -8.82810354e-01 -3.10934246e-01 5.07443547e-01
-4.24921423e-01 -1.56008095e-01 4.59536195e-01 5.56306064e-01
-3.00421417e-01 -3.54069144e-01 9.51523185e-01 -1.11914568e-01
-7.90113747e-01 5.89284301e-01 -4.15663831e-02 -1.27689287e-01
1.10902560e+00 -1.75192952e-01 -3.06917548e-01 -6.98521510e-02
-7.06484556e-01 4.48851556e-01 2.41636351e-01 6.39462531e-01
6.30587995e-01 -1.55441368e+00 -6.11861348e-01 4.09709781e-01
3.03010285e-01 4.64052111e-01 4.70901459e-01 6.69065595e-01
-2.30845913e-01 1.84844241e-01 -2.23814473e-01 -6.67171657e-01
-1.21429026e+00 9.89591658e-01 1.35827288e-01 -6.28726929e-02
-5.83968818e-01 1.03100109e+00 8.93471897e-01 -9.14201736e-02
3.79573643e-01 -3.92942637e-01 -4.46603298e-01 -1.50908276e-01
7.76781261e-01 -2.14757547e-02 -1.61570773e-01 -4.75101888e-01
-3.60747695e-01 4.46103126e-01 -5.90126932e-01 5.39363384e-01
1.34584916e+00 -1.91424012e-01 3.93379070e-02 7.72793666e-02
1.02873158e+00 -5.56630567e-02 -1.59020317e+00 -4.16808873e-01
-2.42710084e-01 -7.48571098e-01 2.17207164e-01 -7.69014537e-01
-1.39565396e+00 1.11887336e+00 7.34095216e-01 3.38503659e-01
1.45704734e+00 2.56285276e-02 5.92141151e-01 1.04104616e-01
-1.16306245e-01 -6.53731346e-01 3.22849333e-01 1.90954730e-01
8.98369670e-01 -1.44134295e+00 -1.74220726e-01 -3.51466566e-01
-8.64648163e-01 8.06717396e-01 1.30530894e+00 -9.30527970e-02
3.09921950e-01 1.93241298e-01 1.09939002e-01 3.35402377e-02
-4.53406513e-01 -4.63747829e-01 4.22423035e-01 5.76771677e-01
3.14481795e-01 -2.49833196e-01 1.85963541e-01 7.38156378e-01
2.41402164e-01 -3.87602337e-02 1.07270613e-01 6.35555208e-01
-5.61128259e-01 -7.69763112e-01 -2.74213225e-01 5.91381669e-01
-4.94129628e-01 -2.99036443e-01 -2.84710169e-01 7.83755541e-01
3.87331009e-01 6.86343610e-01 2.77109623e-01 -2.47886956e-01
4.55923587e-01 -2.43358567e-01 1.62127465e-01 -5.65048158e-01
-3.04245770e-01 1.55328229e-01 -3.57993931e-01 -3.34527045e-01
-2.72832870e-01 -3.55283886e-01 -1.19822299e+00 2.43650824e-02
-5.65466046e-01 1.20062888e-01 2.69317031e-01 8.61810446e-01
4.49947655e-01 1.04483509e+00 5.54344475e-01 -7.87564635e-01
-2.62926459e-01 -1.17091703e+00 -2.90475607e-01 5.60270250e-01
4.82449681e-01 -6.93897784e-01 -6.95626661e-02 1.33716643e-01] | [9.539163589477539, 1.8534023761749268] |
3b5d86d5-1acf-476a-9ffc-c2a5f3a2d28b | vsql-variational-shadow-quantum-learning-for | 2012.08288 | null | https://arxiv.org/abs/2012.08288v1 | https://arxiv.org/pdf/2012.08288v1.pdf | VSQL: Variational Shadow Quantum Learning for Classification | Classification of quantum data is essential for quantum machine learning and near-term quantum technologies. In this paper, we propose a new hybrid quantum-classical framework for supervised quantum learning, which we call Variational Shadow Quantum Learning (VSQL). Our method in particular utilizes the classical shadows of quantum data, which fundamentally represent the side information of quantum data with respect to certain physical observables. Specifically, we first use variational shadow quantum circuits to extract classical features in a convolution way and then utilize a fully-connected neural network to complete the classification task. We show that this method could sharply reduce the number of parameters and thus better facilitate quantum circuit training. Simultaneously, less noise will be introduced since fewer quantum gates are employed in such shadow circuits. Moreover, we show that the Barren Plateau issue, a significant gradient vanishing problem in quantum machine learning, could be avoided in VSQL. Finally, we demonstrate the efficiency of VSQL in quantum classification via numerical experiments on the classification of quantum states and the recognition of multi-labeled handwritten digits. In particular, our VSQL approach outperforms existing variational quantum classifiers in the test accuracy in the binary case of handwritten digit recognition and notably requires much fewer parameters. | ['Xin Wang', 'Zhixin Song', 'Guangxi Li'] | 2020-12-15 | null | null | null | null | ['handwritten-digit-recognition'] | ['computer-vision'] | [ 2.54982114e-01 -6.42120689e-02 3.92206898e-03 -3.48423958e-01
-8.73068094e-01 -5.21228552e-01 3.42614859e-01 3.11462004e-02
-5.05569279e-01 8.49148691e-01 -6.62336349e-01 -4.72011298e-01
-1.67958811e-01 -1.40490448e+00 -6.35222256e-01 -1.24052227e+00
5.68098187e-01 7.97807723e-02 -1.08777694e-01 -4.28257138e-01
5.37440538e-01 3.32963765e-01 -1.36962843e+00 2.01150209e-01
9.75840807e-01 1.11611533e+00 -2.30199262e-01 5.39548755e-01
1.78050950e-01 6.38896644e-01 -2.12097362e-01 -5.11077821e-01
9.22074169e-02 -6.92725003e-01 -7.38117516e-01 -1.54959515e-01
2.57824451e-01 -1.56764254e-01 -8.82268727e-01 1.53113806e+00
4.68386084e-01 2.54909337e-01 9.22262251e-01 -8.08333993e-01
-6.82222366e-01 5.83338618e-01 -5.71875274e-03 -2.96330154e-02
-6.16397075e-02 1.46569729e-01 1.36934543e+00 -5.01047850e-01
4.33060467e-01 9.51592803e-01 3.77503723e-01 7.34254003e-01
-1.60605049e+00 -6.50322139e-01 -9.44887280e-01 6.64231062e-01
-1.66515315e+00 -2.39509523e-01 7.12497115e-01 -2.11280495e-01
8.98256540e-01 4.89679612e-02 4.26423669e-01 7.25954592e-01
4.95815188e-01 4.50797111e-01 1.24286377e+00 -8.09255838e-01
4.75160211e-01 1.59079388e-01 5.16304135e-01 1.19106603e+00
3.17344181e-02 5.43137670e-01 -5.99260628e-01 -1.17189139e-01
3.86554390e-01 -1.12553209e-01 -1.69834793e-01 -2.50302225e-01
-1.09994304e+00 1.08953476e+00 6.39208257e-01 9.41980332e-02
-1.47943631e-01 4.07219261e-01 3.04759294e-01 2.36405820e-01
-1.52741652e-02 2.98521340e-01 -1.70541015e-02 1.52955189e-01
-3.55468184e-01 -1.03776373e-01 7.88478136e-01 6.29689634e-01
1.26199794e+00 -2.48236895e-01 -8.12731609e-02 2.71891713e-01
1.10795289e-01 9.16844130e-01 8.19089636e-03 -1.04948997e+00
1.60999715e-01 2.75155634e-01 -6.21028282e-02 -2.89387822e-01
-2.62596339e-01 -1.91955790e-01 -1.02035785e+00 3.55225801e-02
4.34197813e-01 4.71959189e-02 -6.30201101e-01 1.52623856e+00
1.19444467e-01 -5.25149889e-02 6.22571290e-01 7.69183099e-01
5.59700370e-01 7.33833969e-01 -2.70740390e-01 -1.77894145e-01
1.23173523e+00 -4.88420814e-01 -6.38809085e-01 2.93833196e-01
9.34965551e-01 -4.64873344e-01 7.52113581e-01 4.06950563e-01
-5.25696397e-01 -4.38532144e-01 -1.40708816e+00 -3.34907435e-02
-3.44097644e-01 4.16198336e-02 1.16084397e+00 1.11228907e+00
-4.74705368e-01 1.47607565e+00 -8.44627261e-01 1.01650104e-01
4.53012854e-01 6.62404239e-01 -3.42932254e-01 -5.35878725e-02
-1.44670343e+00 7.60314822e-01 6.42155111e-01 2.28353411e-01
-6.14187896e-01 -1.00185961e-01 -6.92751944e-01 -1.71353109e-02
1.36426881e-01 -4.57730144e-01 1.06208074e+00 -3.30522597e-01
-1.92393494e+00 6.58793747e-01 -9.73175019e-02 -2.69985020e-01
-2.02384200e-02 3.77093047e-01 -2.60868669e-01 2.72927105e-01
-1.84667930e-01 2.09838882e-01 6.46051288e-01 -4.76294070e-01
-3.16908717e-01 -5.13524175e-01 3.14214319e-01 -1.36904970e-01
-2.04302594e-01 -7.43938982e-01 1.26590729e-01 2.93952763e-01
4.75892127e-01 -1.16416287e+00 -2.28148609e-01 -4.91385639e-01
-3.83889973e-01 -6.12596393e-01 3.33945006e-01 1.94557756e-01
8.09331715e-01 -2.24785113e+00 2.50534147e-01 4.12899703e-01
2.43318051e-01 6.73499405e-02 2.99291722e-02 3.95075738e-01
1.70000583e-01 -2.11420044e-01 -2.66437382e-01 1.07714921e-01
2.09052861e-01 4.22842145e-01 -3.67716014e-01 7.34062076e-01
2.74899662e-01 1.11331022e+00 -8.54362488e-01 -5.00419259e-01
3.00327748e-01 3.02594602e-01 -8.30377400e-01 -3.28642368e-01
4.97112907e-02 4.78309512e-01 -6.54506028e-01 5.52654445e-01
7.99073815e-01 -5.28144240e-01 2.11127490e-01 -3.68495405e-01
-1.93268090e-01 5.07080734e-01 -1.00119758e+00 1.57820594e+00
-4.16813344e-01 6.23402715e-01 -3.75009596e-01 -1.34073031e+00
6.93912148e-01 5.32835647e-02 1.08741947e-01 -1.08881402e+00
3.58257413e-01 5.40667236e-01 1.72347859e-01 -3.19853216e-01
3.27907205e-01 -5.38709104e-01 -2.34094396e-01 4.34085488e-01
3.65916997e-01 -3.88261974e-01 1.61295667e-01 1.03253625e-01
8.48548591e-01 -5.46390601e-02 2.20122099e-01 -3.02677155e-01
6.88636541e-01 -6.53709471e-02 4.79158670e-01 9.96615708e-01
-4.59256619e-01 1.84483360e-02 6.73509955e-01 -1.18727975e-01
-9.43916678e-01 -1.21574044e+00 -7.36080050e-01 6.52106583e-01
6.64629400e-01 -4.05121118e-01 -8.31790268e-01 -4.10560399e-01
-9.94524062e-02 7.26297617e-01 -4.57779288e-01 -5.43889105e-01
-3.65337878e-01 -1.18749452e+00 6.89425111e-01 1.72885880e-01
6.36346102e-01 -8.68341804e-01 -4.19546962e-01 8.55728313e-02
-3.77828553e-02 -1.17356944e+00 1.42863259e-01 7.76814044e-01
-9.62350965e-01 -1.10577655e+00 -2.14914411e-01 -4.96478736e-01
4.96393681e-01 -2.64862962e-02 4.35513407e-01 -3.69805604e-01
-4.85219777e-01 -1.82746723e-02 -2.51907080e-01 4.17273454e-02
-6.70704901e-01 -7.05522597e-02 2.74531037e-01 1.22687191e-01
6.94757581e-01 -3.72919738e-01 -5.35882294e-01 -9.85760018e-02
-6.42847717e-01 -4.75441590e-02 7.23518252e-01 1.48670888e+00
6.28710449e-01 1.38680637e-01 1.29732385e-01 -7.77564764e-01
1.51941612e-01 1.46109918e-02 -9.70497906e-01 2.85807282e-01
-4.62347060e-01 7.92760611e-01 8.30571890e-01 2.78321579e-02
-7.94766247e-01 7.96700791e-02 -2.45909601e-01 6.94213137e-02
1.04541212e-01 3.76563698e-01 2.53007501e-01 -7.86955595e-01
9.28766251e-01 4.36529487e-01 -1.62826806e-01 -8.33163559e-02
4.06596750e-01 9.53186035e-01 3.65551978e-01 -5.65337598e-01
7.67903090e-01 6.58906043e-01 7.67438829e-01 -1.11746407e+00
-9.95930254e-01 -2.01893017e-01 -8.60474765e-01 1.02741748e-01
1.01502395e+00 -6.34183109e-01 -1.38055873e+00 4.88664240e-01
-1.18163002e+00 -8.87709931e-02 4.67879809e-02 8.31390619e-01
-5.91973901e-01 6.01257682e-01 -1.03385031e+00 -1.14965641e+00
-2.15733871e-01 -1.34963453e+00 8.61330688e-01 3.96519393e-01
6.36522949e-01 -8.61568332e-01 -1.73179567e-01 4.28148240e-01
-4.17961217e-02 -4.08066101e-02 1.18760574e+00 -2.38975808e-01
-9.63554144e-01 -2.89569706e-01 -4.12477583e-01 6.44653440e-01
-2.15556011e-01 -1.88943133e-01 -1.20952451e+00 -2.22342193e-01
-4.21126224e-02 -7.24669337e-01 1.11007643e+00 -5.76881431e-02
1.06440151e+00 3.99854660e-01 -1.02028064e-01 7.61842191e-01
1.45303464e+00 1.91441774e-01 7.19710052e-01 -1.19147375e-01
7.52558112e-01 3.27126622e-01 3.99908066e-01 4.07154143e-01
1.14970252e-01 6.54205680e-01 2.18162194e-01 3.80926073e-01
3.63115996e-01 -1.37753025e-01 3.74819756e-01 1.18495572e+00
-3.36196333e-01 3.04340750e-01 -6.59676254e-01 -2.00186715e-01
-1.49737191e+00 -9.14980829e-01 -4.22888637e-01 2.26768708e+00
7.56031156e-01 6.48604706e-02 -6.00876808e-01 3.56988311e-01
5.94761550e-01 -1.72438607e-01 -4.78451401e-01 -3.96358013e-01
-2.32237622e-01 9.60476875e-01 8.02032828e-01 4.18116629e-01
-1.26057565e+00 1.02334428e+00 5.48066187e+00 9.99426126e-01
-1.18600869e+00 3.21092933e-01 1.07003227e-01 4.59073097e-01
-4.25752625e-02 4.42538083e-01 -7.90667593e-01 7.93191716e-02
1.18173373e+00 -5.29361106e-02 7.56567955e-01 5.53923666e-01
-4.06878740e-01 -1.39547214e-01 -1.25814629e+00 1.39956260e+00
-2.81103134e-01 -1.47412515e+00 -3.09172943e-02 3.21677744e-01
6.68673158e-01 1.60889775e-01 3.54958698e-02 4.12446827e-01
-4.04743403e-01 -8.65559757e-01 5.22127867e-01 4.10986334e-01
9.42886055e-01 -8.58200252e-01 7.80445337e-01 3.85631055e-01
-8.72828186e-01 -1.82499081e-01 -7.46293128e-01 -2.39873245e-01
-2.10446179e-01 3.87143582e-01 -1.67359397e-01 7.16234565e-01
2.11125121e-01 5.09716630e-01 -3.61461908e-01 6.27005219e-01
-1.75571695e-01 3.79478246e-01 -1.99912325e-01 -6.77053988e-01
4.51760411e-01 -7.87422061e-01 1.12366773e-01 7.72371948e-01
2.29440317e-01 5.40522754e-01 -2.24935070e-01 1.15205312e+00
-1.44387335e-01 5.47931902e-02 -5.63323617e-01 -5.30488312e-01
1.20649733e-01 1.08944952e+00 -5.96470773e-01 -3.21575195e-01
-2.69158691e-01 1.22712123e+00 2.61398613e-01 1.95928872e-01
-7.48562098e-01 -9.75629151e-01 2.24569887e-01 -6.59700036e-01
2.28567511e-01 -1.43621951e-01 -1.78318694e-01 -1.70683897e+00
-2.14803562e-01 -2.94565380e-01 -4.22293805e-02 -2.20157683e-01
-1.05357766e+00 2.71433145e-01 -5.30956686e-01 -1.12666273e+00
-1.18699968e-02 -1.23732054e+00 -2.75391996e-01 7.73103952e-01
-1.48551691e+00 -6.27742231e-01 -1.24052092e-01 6.43246472e-01
-4.02929008e-01 -3.93121801e-02 1.25434852e+00 4.12951201e-01
-7.51938760e-01 6.51866317e-01 8.60898554e-01 4.01000530e-01
4.10849065e-01 -1.31278813e+00 6.40958473e-02 4.93583500e-01
4.32476044e-01 6.96225226e-01 3.38835448e-01 -1.18873835e-01
-1.88796246e+00 -6.25866711e-01 6.42906547e-01 -2.88344592e-01
7.63189077e-01 -5.02812028e-01 -5.60608149e-01 1.47526070e-01
-3.31796795e-01 1.69524759e-01 8.41148138e-01 5.57635576e-02
-4.98553157e-01 -2.09426343e-01 -8.39279830e-01 2.76043415e-01
7.08436549e-01 -1.43171442e+00 -4.64209557e-01 7.10713089e-01
3.04971427e-01 -2.70303100e-01 -8.75320196e-01 2.90332109e-01
7.81351864e-01 -1.06730080e+00 6.32120430e-01 -5.55197179e-01
4.27108258e-01 -1.62493676e-01 -5.86817145e-01 -1.00200343e+00
-2.35868283e-02 -5.49615741e-01 1.41971141e-01 4.97959524e-01
3.89232457e-01 -9.35322821e-01 8.30815256e-01 2.26275340e-01
-6.94901794e-02 -4.89583343e-01 -1.30106723e+00 -8.78485858e-01
4.83060658e-01 -5.00133634e-01 6.97956979e-03 7.29765952e-01
2.96650201e-01 5.15817344e-01 -1.71943009e-01 1.03641652e-01
1.28010523e+00 5.01756132e-01 3.39004695e-01 -1.07184052e+00
-4.99642998e-01 -3.23592514e-01 -8.64993691e-01 -9.94487226e-01
3.04128975e-01 -1.34136605e+00 6.01435266e-02 -8.18303883e-01
4.65154082e-01 -4.00936633e-01 -6.96850240e-01 -4.70976420e-02
-6.08570594e-03 4.92636412e-01 -4.07537026e-03 2.89484322e-01
-6.24860406e-01 7.75272429e-01 1.30264139e+00 -4.22131091e-01
2.66808905e-02 4.55809422e-02 -7.09639862e-02 4.02489334e-01
5.04042447e-01 -5.78198493e-01 6.08772375e-02 2.66622175e-02
1.97039559e-01 2.54213452e-01 5.73536336e-01 -1.06398571e+00
2.79705077e-01 6.98562339e-02 1.28393307e-01 -3.92278999e-01
4.17954564e-01 -9.50436592e-02 -4.22205150e-01 9.57751453e-01
-1.09652899e-01 -8.45310509e-01 -2.84798354e-01 6.69772446e-01
-2.11236790e-01 -5.39208412e-01 1.12237036e+00 1.22285858e-01
-7.02056646e-01 1.85520396e-01 -7.76377991e-02 -3.20569038e-01
9.47522163e-01 2.55157948e-01 -4.87076133e-01 1.68476865e-01
-6.75106585e-01 -1.40903130e-01 3.14622521e-01 -2.62224823e-01
5.66323042e-01 -1.33039117e+00 -2.71178693e-01 2.55049139e-01
3.76485765e-01 -5.64461291e-01 5.16609013e-01 9.97389495e-01
-8.82690966e-01 9.55996215e-01 -2.07306951e-01 -7.31258273e-01
-6.36459589e-01 3.55562836e-01 6.14376068e-01 9.09943283e-02
-4.35912758e-01 8.69864762e-01 8.24968517e-02 -5.45760810e-01
-1.34229466e-01 -6.73007593e-02 2.04629838e-01 -8.99545401e-02
3.13972503e-01 2.57529140e-01 8.58745128e-02 -7.77358115e-01
-4.57220912e-01 6.90849483e-01 1.44646093e-01 -1.90030724e-01
9.64065194e-01 2.66044527e-01 -3.83880049e-01 6.36712790e-01
1.55294406e+00 -2.62397110e-01 -9.47081327e-01 -4.01968390e-01
-4.88510728e-03 -3.62778977e-02 3.01837862e-01 -2.43879065e-01
-6.07519269e-01 1.43410563e+00 8.48209143e-01 2.90449649e-01
8.00336063e-01 7.75818108e-03 8.63778293e-01 1.41214204e+00
8.65627289e-01 -1.31901777e+00 -2.56750494e-01 5.66883624e-01
-2.86561041e-03 -1.44747901e+00 -1.48539126e-01 -2.73515373e-01
-1.81229591e-01 1.63169932e+00 1.25241384e-01 -1.01323649e-01
5.31838536e-01 -3.58041137e-01 -1.92028284e-01 -1.72287211e-01
-4.92122710e-01 -4.58384097e-01 2.12190434e-01 9.46351662e-02
2.66418636e-01 5.28745413e-01 -2.87123233e-01 2.87194997e-01
-2.16196060e-01 -1.54671326e-01 6.87317133e-01 8.73076797e-01
-3.84016663e-01 -1.30145478e+00 -1.14355899e-01 3.08151096e-01
-2.88621396e-01 -3.58948708e-01 -4.85386513e-02 4.46363837e-01
1.72852129e-01 9.77148473e-01 -1.89738005e-01 -7.82642365e-01
1.71710774e-01 2.33825460e-01 1.19730043e+00 -5.11994183e-01
5.29569807e-03 -2.33091354e-01 -3.34067911e-01 -5.78810334e-01
-4.31682080e-01 -4.50865120e-01 -1.50979102e+00 -5.71637690e-01
-9.19658124e-01 4.28610355e-01 8.61077189e-01 1.37386727e+00
-3.55624743e-02 4.91158515e-01 7.81802893e-01 -6.43086970e-01
-1.31421065e+00 -7.46407092e-01 -9.54615057e-01 1.31713584e-01
3.30143422e-01 -7.26835608e-01 -5.16964555e-01 -3.11739057e-01] | [5.585540294647217, 4.95892858505249] |
ceba1c09-f039-4979-b413-94c07c8d3d69 | cross-linguistic-syntactic-difference-in | 2212.10879 | null | https://arxiv.org/abs/2212.10879v1 | https://arxiv.org/pdf/2212.10879v1.pdf | Cross-Linguistic Syntactic Difference in Multilingual BERT: How Good is It and How Does It Affect Transfer? | Multilingual BERT (mBERT) has demonstrated considerable cross-lingual syntactic ability, whereby it enables effective zero-shot cross-lingual transfer of syntactic knowledge. The transfer is more successful between some languages, but it is not well understood what leads to this variation and whether it fairly reflects difference between languages. In this work, we investigate the distributions of grammatical relations induced from mBERT in the context of 24 typologically different languages. We demonstrate that the distance between the distributions of different languages is highly consistent with the syntactic difference in terms of linguistic formalisms. Such difference learnt via self-supervision plays a crucial role in the zero-shot transfer performance and can be predicted by variation in morphosyntactic properties between languages. These results suggest that mBERT properly encodes languages in a way consistent with linguistic diversity and provide insights into the mechanism of cross-lingual transfer. | ['Xuanjing Huang', 'Menghan Zhang', 'Jingting Ye', 'Qi Zhang', 'Ruotian Ma', 'Tao Gui', 'Ningyu Xu'] | 2022-12-21 | null | null | null | null | ['zero-shot-cross-lingual-transfer', 'cross-lingual-transfer'] | ['natural-language-processing', 'natural-language-processing'] | [-2.80887842e-01 -1.04757570e-01 -4.15383190e-01 -7.14347482e-01
-6.13489032e-01 -6.17188692e-01 6.18065000e-01 1.18853167e-01
-5.79804659e-01 7.13274181e-01 5.55052817e-01 -4.00577903e-01
-2.19309226e-01 -8.89568269e-01 -1.01320207e+00 -5.65082848e-01
9.89813060e-02 5.56428909e-01 1.20803952e-01 -7.98515856e-01
-1.14484683e-01 1.59861650e-02 -1.14660096e+00 4.82539803e-01
1.32790053e+00 1.93912312e-01 3.87188286e-01 2.15231180e-02
-6.32820368e-01 3.65538418e-01 -4.64876473e-01 -5.59887707e-01
1.04289472e-01 -7.16311276e-01 -7.66556919e-01 -3.50692302e-01
5.97036839e-01 3.06367517e-01 -5.14148250e-02 1.16253567e+00
7.61590153e-02 -1.25755847e-01 9.60735798e-01 -7.56186187e-01
-9.64723945e-01 1.08620572e+00 -3.03871959e-01 3.23351294e-01
1.46390989e-01 6.05081134e-02 1.41002035e+00 -5.46214044e-01
8.85280311e-01 1.64147151e+00 4.86300439e-01 4.33399737e-01
-1.48944795e+00 -6.50343060e-01 -2.00805590e-02 1.74602360e-01
-1.43729305e+00 -3.30668211e-01 6.09340429e-01 -7.51903832e-01
7.41684675e-01 -3.04353803e-01 5.60510218e-01 9.46809947e-01
4.98654455e-01 4.16428983e-01 1.52346325e+00 -9.11357284e-01
-1.50966316e-01 4.49329346e-01 3.01350236e-01 7.56620049e-01
2.71309972e-01 1.57669097e-01 -8.20949078e-01 2.40539461e-01
4.29619998e-01 -5.47161698e-01 -5.03782690e-01 -1.69979826e-01
-1.09936690e+00 1.04602623e+00 6.06661499e-01 9.25845921e-01
2.71341801e-01 -2.15773270e-01 5.92648625e-01 6.88269258e-01
3.99997950e-01 4.61017072e-01 -9.15243864e-01 1.02429569e-01
-2.90442139e-01 -2.59781957e-01 6.73020899e-01 7.23662794e-01
1.10374081e+00 8.92516151e-02 2.56212145e-01 1.14540386e+00
2.20968381e-01 6.81057453e-01 7.75414526e-01 -4.62832212e-01
4.65566874e-01 5.10354698e-01 -5.95801592e-01 -7.73219883e-01
5.74500635e-02 -8.67603347e-02 -4.39763635e-01 6.51774108e-02
7.05241144e-01 -1.22310638e-01 -3.27295154e-01 2.34731698e+00
1.10036552e-01 -7.14900196e-01 3.03758472e-01 5.08730710e-01
7.09153175e-01 5.94694257e-01 4.60611939e-01 -7.97850639e-03
1.49448204e+00 -3.81556094e-01 -3.69870901e-01 -2.35975325e-01
9.42239046e-01 -5.74180305e-01 1.45532525e+00 -2.70516247e-01
-6.84925795e-01 -5.17538369e-01 -8.13264310e-01 -1.33633107e-01
-5.65957129e-01 -2.76226699e-01 7.58155704e-01 6.05281711e-01
-8.38608742e-01 6.19718075e-01 -4.31515276e-01 -6.69501185e-01
8.65935385e-02 8.26872289e-02 -6.82693899e-01 -1.42734442e-02
-1.59538865e+00 1.16969204e+00 6.47179723e-01 -3.68054062e-01
-2.60138452e-01 -7.50330687e-01 -1.15950060e+00 3.66054513e-02
-4.29045856e-02 -4.19792861e-01 9.53357577e-01 -1.16366434e+00
-1.57997537e+00 1.19644284e+00 -2.58129351e-02 -2.28507742e-01
6.13910556e-02 1.16770677e-01 -3.24696034e-01 -1.93425328e-01
4.31734562e-01 6.56563103e-01 3.70992273e-01 -9.92368519e-01
-4.81439590e-01 -5.18790662e-01 -1.84901431e-01 2.94303417e-01
-3.39751542e-01 3.55069578e-01 5.12820594e-02 -6.59932256e-01
-2.08458632e-01 -9.75440919e-01 3.14050645e-01 -4.07275021e-01
8.43729302e-02 -6.35604024e-01 7.98658952e-02 -7.77966499e-01
8.38846087e-01 -2.22253180e+00 1.89573810e-01 4.49171513e-02
-2.95934916e-01 9.30772126e-02 -2.44833171e-01 5.93764305e-01
-1.65114626e-02 2.79578000e-01 -1.48218617e-01 1.64098442e-01
1.31520852e-01 4.35113937e-01 -2.12979376e-01 5.15707433e-01
1.40007854e-01 8.93140376e-01 -8.83672535e-01 -3.75420868e-01
-1.00994572e-01 4.25201625e-01 -4.51077044e-01 -3.44004892e-02
-3.18299025e-01 8.18985999e-01 -3.05905968e-01 2.67248452e-01
2.63519347e-01 2.52895623e-01 5.25609851e-01 -1.00644216e-01
-2.31031388e-01 9.06716585e-01 -3.33785892e-01 1.70448947e+00
-4.92226839e-01 6.85170650e-01 -2.78458834e-01 -9.77906227e-01
9.10221398e-01 1.53772190e-01 -1.72554076e-01 -9.67507839e-01
2.88238317e-01 4.57623273e-01 8.25491011e-01 -2.15147123e-01
2.33214036e-01 -7.29735732e-01 -4.03016269e-01 4.54335570e-01
4.69021738e-01 -1.60581321e-01 3.01105350e-01 -1.45100087e-01
4.45218652e-01 1.35120064e-01 4.61044490e-01 -8.34712982e-01
3.45359653e-01 -3.33455540e-02 8.40901673e-01 4.68694061e-01
-5.42004369e-02 -6.74839094e-02 5.35495222e-01 -7.33949021e-02
-8.08236003e-01 -1.06491470e+00 -7.00464547e-01 1.37084293e+00
7.11339340e-02 -3.13116908e-01 -7.06033468e-01 -6.30209565e-01
1.53376669e-01 9.55611408e-01 -6.92264080e-01 -3.99469405e-01
-5.92305720e-01 -4.95962501e-01 6.68462455e-01 3.05449545e-01
4.43871617e-01 -1.21082115e+00 -6.63722828e-02 -7.47765973e-02
1.24369346e-01 -1.11705160e+00 -4.32655215e-01 2.82911271e-01
-7.56222844e-01 -7.47128427e-01 -3.66113096e-01 -9.54873443e-01
4.54922259e-01 -8.12584609e-02 1.30525422e+00 -6.62181973e-02
5.80717362e-02 1.21862963e-01 -2.32824460e-01 -2.81231701e-01
-9.04863417e-01 4.73172754e-01 1.47283852e-01 -2.50913799e-01
5.79201281e-01 -5.29943764e-01 -1.36794038e-02 2.25116536e-01
-5.80162883e-01 -3.35260212e-01 3.83093119e-01 5.92057407e-01
2.35184386e-01 -1.25889301e-01 5.17074347e-01 -1.15213418e+00
4.97580141e-01 -5.27912021e-01 -4.78977144e-01 4.30974334e-01
-3.86319876e-01 4.67687994e-01 5.54377317e-01 -3.86991985e-02
-1.33968699e+00 -4.53618884e-01 7.57875741e-02 3.22336048e-01
-3.43058169e-01 5.20753503e-01 -3.71714711e-01 4.14787568e-02
5.09106755e-01 4.84460033e-02 -6.60526976e-02 -4.09212023e-01
5.13324261e-01 6.59992099e-01 1.01779550e-01 -1.12812948e+00
6.06657624e-01 1.46499807e-02 -4.61219043e-01 -1.00315070e+00
-9.83774662e-01 -4.70081270e-02 -8.71597707e-01 2.72381932e-01
1.16454005e+00 -1.04354537e+00 -2.69752532e-01 2.95040667e-01
-1.10237002e+00 -6.72754943e-01 -1.82632297e-01 6.82855248e-01
-1.88556731e-01 -2.25126408e-02 -8.51215005e-01 -1.72087580e-01
-4.24825773e-02 -1.02261496e+00 7.79809654e-01 -6.28202111e-02
-4.49376255e-01 -1.53642976e+00 3.73148590e-01 2.14198664e-01
2.42621109e-01 -1.90315634e-01 1.62802708e+00 -6.65139198e-01
-4.89834815e-01 4.09866929e-01 -6.58415258e-02 4.25548315e-01
4.32249248e-01 3.16394754e-02 -7.36082971e-01 -2.21024811e-01
-2.46272936e-01 -4.11953241e-01 7.89523900e-01 1.81627110e-01
2.42415130e-01 1.33858155e-02 -2.06447467e-01 6.90435708e-01
1.49913442e+00 1.23545699e-01 3.63267332e-01 3.56414139e-01
6.49553895e-01 9.45419371e-01 2.32438207e-01 -4.60923404e-01
6.78917468e-01 7.02425480e-01 -4.80872720e-01 2.13439494e-01
-2.17963696e-01 -2.60931104e-01 8.55566502e-01 1.62932491e+00
6.39733821e-02 1.21599339e-01 -1.14801645e+00 5.89283049e-01
-1.41201389e+00 -7.79645324e-01 1.49405366e-02 2.23801064e+00
1.32640660e+00 -6.94832280e-02 -7.02373981e-02 -3.49762082e-01
7.82871723e-01 8.63335282e-02 -1.41518936e-01 -6.90248847e-01
-4.62927341e-01 3.10385734e-01 2.95714200e-01 6.91593468e-01
-4.53880191e-01 1.45843303e+00 6.40964985e+00 6.89309180e-01
-1.26064205e+00 2.45807931e-01 3.42992634e-01 5.13467014e-01
-5.43011785e-01 4.46356237e-02 -1.04111254e+00 5.20480216e-01
9.68913972e-01 -4.77998495e-01 1.76470235e-01 4.65606600e-01
-1.77407920e-01 7.26910755e-02 -1.26187301e+00 3.93522143e-01
3.02064478e-01 -8.01500976e-01 -2.97309496e-02 8.19438547e-02
7.81116724e-01 3.93464267e-01 -4.73595113e-02 6.39975607e-01
6.15749955e-01 -8.38280976e-01 7.71371961e-01 1.70064662e-02
8.73521388e-01 -7.52443671e-01 4.55210447e-01 3.56980264e-01
-1.02892554e+00 3.83679837e-01 -4.92437840e-01 -1.91114888e-01
1.30067347e-02 4.05119121e-01 -8.89639497e-01 2.64010757e-01
5.89418590e-01 5.59551239e-01 -5.97149372e-01 1.91812068e-01
-7.44232178e-01 7.43551195e-01 1.26026422e-02 1.02302097e-02
-4.68160920e-02 -4.35831100e-01 4.77322608e-01 1.27668631e+00
2.01523021e-01 -3.64269704e-01 1.77577510e-01 8.80436897e-01
-8.14127326e-02 6.11073792e-01 -8.10858965e-01 -8.36943015e-02
4.33265477e-01 8.00699532e-01 -5.75521588e-01 -2.41936743e-01
-6.89731002e-01 9.27540064e-01 1.01922321e+00 1.97657391e-01
-5.98001838e-01 -1.36984438e-01 7.43864357e-01 2.29227915e-02
4.19336170e-01 -3.09724718e-01 -1.70104548e-01 -1.42157817e+00
-1.08778074e-01 -8.17631245e-01 2.05962107e-01 -3.05284530e-01
-1.82853866e+00 6.61130607e-01 9.64272860e-03 -6.09752715e-01
-1.85716942e-01 -8.39727044e-01 -4.11156058e-01 1.03781235e+00
-1.40154505e+00 -1.14312506e+00 1.44271180e-01 6.13116741e-01
2.70783842e-01 -4.02981013e-01 1.06879246e+00 2.15144619e-01
-5.02778292e-01 7.79389262e-01 6.89495429e-02 4.16145802e-01
1.05103302e+00 -1.36460865e+00 3.71289819e-01 4.42441791e-01
4.97580677e-01 9.94264662e-01 4.21391904e-01 -6.18906558e-01
-1.06098258e+00 -9.91127491e-01 1.29093671e+00 -5.28042614e-01
9.59290683e-01 -4.47664768e-01 -1.30984604e+00 8.17490935e-01
6.31615877e-01 -8.83060682e-04 1.10426998e+00 6.75369024e-01
-8.55150461e-01 -1.56098604e-01 -8.20901752e-01 6.03317320e-01
1.05581307e+00 -9.29120779e-01 -1.13701594e+00 7.32440501e-02
8.54932010e-01 6.63878843e-02 -9.78888929e-01 1.77540809e-01
2.44708732e-01 -9.36963260e-01 4.51229304e-01 -7.19906747e-01
4.91296411e-01 -1.56061485e-01 -3.30624998e-01 -1.93585706e+00
-3.65984678e-01 -2.02585936e-01 8.53360295e-01 1.64192772e+00
6.02635264e-01 -1.02512181e+00 7.33152628e-02 2.63794065e-01
-4.33082022e-02 -3.13611358e-01 -8.83950114e-01 -1.00601459e+00
9.23491061e-01 -1.74336314e-01 5.19927919e-01 1.52482188e+00
2.07524687e-01 9.51431096e-01 2.42085382e-01 -1.52048960e-01
5.10977983e-01 1.15662806e-01 5.57063580e-01 -1.29713702e+00
-5.07072031e-01 -4.91556466e-01 -5.45275509e-01 -5.77237129e-01
9.84920502e-01 -1.68796468e+00 1.52450740e-01 -9.75979209e-01
2.15466544e-01 -6.59163415e-01 -3.26371133e-01 3.94736171e-01
-1.38377070e-01 -6.04589693e-02 3.24779719e-01 8.04876909e-02
-3.73163044e-01 6.63368046e-01 1.05180311e+00 1.70031637e-01
-1.69864565e-01 -5.83459914e-01 -7.66839504e-01 9.27737296e-01
8.92628491e-01 -5.31324983e-01 -1.80674747e-01 -7.80296862e-01
1.78257823e-01 -2.15260237e-01 -3.09335709e-01 -5.62769413e-01
-1.84483677e-01 -8.30968469e-02 -6.15744181e-02 1.33337513e-01
-2.76895344e-01 -5.97225189e-01 -3.13841611e-01 3.74916375e-01
-3.05345654e-01 9.04051587e-02 1.98632047e-01 2.47269362e-01
-4.76201534e-01 -2.63741791e-01 8.65291417e-01 -3.21272284e-01
-6.87571883e-01 3.85511182e-02 -4.25766796e-01 6.76054001e-01
7.09634483e-01 1.53019756e-01 -3.74312669e-01 4.93529104e-02
-4.04834777e-01 -2.43907675e-01 6.25078857e-01 5.56558430e-01
-3.00920427e-01 -1.55258822e+00 -8.93866241e-01 2.84796476e-01
3.51526171e-01 -5.03406227e-01 -3.20104957e-02 5.72831213e-01
-2.07769081e-01 4.42517638e-01 -3.31197113e-01 -6.46308720e-01
-1.06063116e+00 2.20314071e-01 4.35849041e-01 -1.06388532e-01
-2.73094267e-01 7.69678652e-01 6.03371680e-01 -9.05659676e-01
-2.68499613e-01 -1.30917132e-01 -6.43336326e-02 1.55933782e-01
1.59920812e-01 -8.00560117e-02 -1.05369031e-01 -1.00420439e+00
-2.94793576e-01 9.71645832e-01 -1.94687024e-01 -8.85478407e-02
1.23806214e+00 -2.01706216e-02 -5.21169484e-01 1.01925468e+00
1.22038746e+00 4.99115795e-01 -7.90443897e-01 -4.73937392e-01
3.11000317e-01 -3.47520262e-01 -4.15875763e-01 -6.06385052e-01
-9.93117869e-01 8.65566432e-01 2.13585734e-01 -1.74955353e-02
4.65442628e-01 4.86590058e-01 7.18036592e-01 2.01280028e-01
6.77263916e-01 -9.76757586e-01 -2.69059956e-01 1.14030707e+00
8.23173940e-01 -1.36771488e+00 -4.31201130e-01 -5.86965382e-01
-5.81042171e-01 7.93213069e-01 7.47722208e-01 -2.20946342e-01
5.07862866e-01 8.98287892e-02 2.24049583e-01 -1.08289622e-01
-7.12401807e-01 -3.60931993e-01 4.03569877e-01 6.29329264e-01
1.21518087e+00 3.92707258e-01 -5.87031424e-01 4.37272638e-01
-7.76057541e-01 -4.74447310e-01 1.74503997e-02 4.73195791e-01
-4.23661649e-01 -1.60715699e+00 -1.46682143e-01 -1.99593619e-01
-3.07880998e-01 -4.00738984e-01 -3.64872992e-01 1.04601347e+00
5.51614344e-01 5.56570649e-01 3.36289525e-01 1.63899865e-02
1.17522106e-01 5.05920410e-01 9.73434985e-01 -1.06956089e+00
-6.35363996e-01 -1.64035425e-01 2.11055949e-01 -7.94951469e-02
-2.21051410e-01 -7.63930678e-01 -1.34174287e+00 -1.53674856e-01
-2.29355067e-01 2.46020213e-01 4.26690668e-01 1.14999425e+00
1.65857419e-01 4.15072888e-01 4.25609529e-01 -3.53378087e-01
-2.60213226e-01 -9.65964973e-01 -6.69605553e-01 7.12077618e-01
-5.95837384e-02 -5.67684948e-01 -5.00576913e-01 -4.16458286e-02] | [10.873221397399902, 9.939891815185547] |
8eb6ca6c-bf18-4fa1-9673-cc85588d5ed5 | optimal-transport-graph-neural-networks | 2006.04804 | null | https://arxiv.org/abs/2006.04804v6 | https://arxiv.org/pdf/2006.04804v6.pdf | Optimal Transport Graph Neural Networks | Current graph neural network (GNN) architectures naively average or sum node embeddings into an aggregated graph representation -- potentially losing structural or semantic information. We here introduce OT-GNN, a model that computes graph embeddings using parametric prototypes that highlight key facets of different graph aspects. Towards this goal, we successfully combine optimal transport (OT) with parametric graph models. Graph representations are obtained from Wasserstein distances between the set of GNN node embeddings and ``prototype'' point clouds as free parameters. We theoretically prove that, unlike traditional sum aggregation, our function class on point clouds satisfies a fundamental universal approximation theorem. Empirically, we address an inherent collapse optimization issue by proposing a noise contrastive regularizer to steer the model towards truly exploiting the OT geometry. Finally, we outperform popular methods on several molecular property prediction tasks, while exhibiting smoother graph representations. | ['Octavian-Eugen Ganea', 'Gary Bécigneul', 'Benson Chen', 'Regina Barzilay', 'Tommi Jaakkola'] | 2020-06-08 | null | https://openreview.net/forum?id=o1O5nc48rn | https://openreview.net/pdf?id=o1O5nc48rn | null | ['graph-regression'] | ['graphs'] | [ 1.39291063e-01 4.94601190e-01 -1.80162221e-01 -1.03624128e-01
-3.74211639e-01 -7.16129005e-01 6.91031039e-01 6.15496159e-01
-9.77194235e-02 5.05194485e-01 2.19039679e-01 -5.09720743e-01
-4.81179386e-01 -1.01594996e+00 -1.10076308e+00 -7.92965949e-01
-5.77287853e-01 7.01878071e-01 -8.58884677e-02 -1.09897666e-01
1.53008819e-01 8.57027411e-01 -8.19621742e-01 -7.86512420e-02
8.11825097e-01 8.41489613e-01 3.68944593e-02 7.93253601e-01
-1.38977364e-01 4.20810133e-01 -1.58784956e-01 -5.13242424e-01
3.28121066e-01 -1.14869565e-01 -6.30844295e-01 -1.80268645e-01
9.27789271e-01 2.97548305e-02 -6.50621355e-01 1.06006753e+00
3.07574123e-01 5.43908894e-01 9.25982058e-01 -1.39099872e+00
-1.57739890e+00 5.81905067e-01 -2.44521782e-01 3.80327441e-02
2.19105706e-01 4.60691571e-01 1.61570823e+00 -6.15439951e-01
9.42136288e-01 1.33380878e+00 1.29390478e+00 2.66572237e-01
-2.09098649e+00 -6.48735762e-02 3.48406345e-01 -1.78441271e-01
-1.40645480e+00 -1.21908924e-02 9.17597115e-01 -5.78038573e-01
1.30579662e+00 1.41587734e-01 9.49342370e-01 1.03847682e+00
4.68150705e-01 3.98876309e-01 5.27033150e-01 7.14327246e-02
2.21023425e-01 -3.59454483e-01 2.84579784e-01 1.04114664e+00
6.27176821e-01 -1.35447726e-01 -4.32707876e-01 -4.94042754e-01
9.11531270e-01 1.29434869e-01 -5.44687212e-01 -1.04780996e+00
-1.05398309e+00 9.50826883e-01 1.19030082e+00 1.01947278e-01
-4.96932238e-01 7.72482157e-01 1.88174590e-01 3.12564343e-01
6.35925710e-01 9.20411408e-01 -3.31460059e-01 3.07492882e-01
-5.59966862e-01 2.80243754e-01 7.22745597e-01 1.14520764e+00
1.15418696e+00 1.10834308e-01 6.16755635e-02 3.57032299e-01
2.34150022e-01 2.00260431e-01 -1.64439306e-01 -8.56971562e-01
1.20058969e-01 8.08544397e-01 -2.13902920e-01 -1.28795469e+00
-6.43495381e-01 -5.49789488e-01 -8.63815844e-01 -8.98571536e-02
3.14065725e-01 3.56559038e-01 -9.40182984e-01 1.69446433e+00
2.11523697e-01 2.87802964e-01 -2.54609644e-01 5.40954053e-01
9.21540380e-01 5.12646198e-01 4.02803540e-01 2.71303505e-01
9.30176497e-01 -4.91315156e-01 -3.91767383e-01 1.93827480e-01
1.02576745e+00 -1.86179280e-01 1.19843793e+00 1.58800721e-01
-1.07103455e+00 -2.89350748e-01 -9.88851964e-01 -6.28753066e-01
-7.49955475e-01 -4.29028571e-01 9.41432118e-01 4.00760740e-01
-1.68630564e+00 1.45347214e+00 -8.29406738e-01 -4.72487956e-01
7.77534664e-01 5.64966798e-01 -5.91384411e-01 -5.51693812e-02
-6.68232620e-01 6.51498556e-01 2.66036898e-01 -9.41964984e-02
-7.19673216e-01 -1.20011830e+00 -9.47264493e-01 -4.19654809e-02
2.23126374e-02 -1.26658106e+00 5.80060780e-01 -4.11352009e-01
-1.12919009e+00 5.76701820e-01 3.62208039e-02 -4.82584715e-01
2.42148787e-01 1.76539183e-01 6.47892850e-03 1.91192806e-01
-1.68811053e-01 8.02813590e-01 6.52728796e-01 -1.22146678e+00
3.43072951e-01 -3.34422112e-01 2.14874789e-01 5.06782494e-02
-9.21077803e-02 -7.78774679e-01 -1.16324509e-02 -4.46451306e-01
2.80720115e-01 -8.87765884e-01 -5.02716959e-01 5.43421865e-01
-6.89992845e-01 -3.51622254e-01 7.45320439e-01 -3.11954528e-01
8.75106871e-01 -1.87804198e+00 5.96047878e-01 4.82760817e-01
1.07106304e+00 -1.81895196e-01 -4.28059816e-01 6.13884568e-01
-3.02490026e-01 4.61092979e-01 -3.55135441e-01 -4.21045154e-01
2.54959404e-01 5.38741648e-01 -1.51309922e-01 8.01818609e-01
4.21574742e-01 1.51685512e+00 -1.19782686e+00 -1.63334943e-02
2.50902355e-01 8.29827785e-01 -1.04168451e+00 -1.03719011e-01
-7.21208096e-01 1.47553161e-01 -3.59754950e-01 5.27580500e-01
7.21006870e-01 -7.45269418e-01 1.69097662e-01 -5.38872004e-01
2.74711460e-01 2.73657769e-01 -6.24377549e-01 1.88764131e+00
-1.37417272e-01 6.18736804e-01 -3.26243527e-02 -1.09900188e+00
8.31494391e-01 -2.25343972e-01 6.14816308e-01 -2.45134905e-01
-5.08085676e-02 1.27401939e-02 -4.97599095e-02 -1.69503063e-01
6.35364115e-01 -4.27701846e-02 2.66075611e-01 2.22730920e-01
2.80436099e-01 -2.16404319e-01 -2.28756130e-01 5.44242084e-01
1.36089969e+00 2.76816398e-01 1.55455664e-01 -6.28215015e-01
-2.03356847e-01 9.03550610e-02 -3.15041579e-02 8.49434853e-01
-5.57144098e-02 5.95303833e-01 9.93099153e-01 -6.66395068e-01
-1.39565814e+00 -1.45797908e+00 -1.45143420e-02 1.02032292e+00
2.94952448e-02 -7.60176718e-01 -5.02175152e-01 -6.31168127e-01
5.52852929e-01 6.72259033e-01 -9.16658700e-01 -3.94187570e-01
-4.42985892e-01 -6.98534966e-01 5.74667752e-01 4.85807836e-01
-3.71985853e-01 -5.63901365e-01 2.07429230e-01 3.20367694e-01
2.82856554e-01 -9.17628348e-01 -5.45529008e-01 3.17665070e-01
-9.44730937e-01 -1.08531892e+00 -3.55801076e-01 -5.71274698e-01
7.83157945e-01 3.52689415e-01 1.35000384e+00 2.89898366e-01
-3.91666979e-01 5.06226003e-01 -1.75966755e-01 2.88371760e-02
-2.56329715e-01 2.15047315e-01 1.72615796e-02 -2.39281356e-01
3.01461965e-01 -1.10187066e+00 -8.17522764e-01 -2.00727820e-01
-7.53616869e-01 -1.55402780e-01 4.46940094e-01 5.89651823e-01
7.88464069e-01 -4.92721558e-01 1.16907865e-01 -8.96818340e-01
8.88813913e-01 -6.06551528e-01 -3.77937496e-01 5.29004522e-02
-5.47001600e-01 4.80272532e-01 8.42134595e-01 -3.47983897e-01
1.81904007e-02 -2.87793279e-01 9.02252272e-02 -8.42981339e-01
2.49653757e-01 4.71249551e-01 2.21596006e-02 -6.94003999e-01
9.33154583e-01 8.07169527e-02 3.40189576e-01 -4.33141381e-01
1.06918120e+00 -1.92992464e-01 2.67178595e-01 -8.82229149e-01
7.97159851e-01 6.18152678e-01 6.49967670e-01 -1.01408243e+00
-5.59863865e-01 -3.38710278e-01 -6.84330761e-01 2.52423994e-02
8.34476233e-01 -5.47283292e-01 -1.11309814e+00 -1.37679577e-01
-1.17119813e+00 -4.99816895e-01 -4.37546641e-01 2.82312959e-01
-6.87788367e-01 6.28999114e-01 -6.19853139e-01 -4.98062313e-01
-1.76480934e-01 -1.07655931e+00 1.44050121e+00 -3.84863764e-01
-2.31987089e-01 -1.60528970e+00 1.03040300e-01 -3.51736695e-01
4.14176852e-01 7.06252456e-01 1.22754407e+00 -6.42679274e-01
-9.18437660e-01 -3.89646105e-02 -6.33115709e-01 9.65147093e-02
1.41784651e-02 2.63039559e-01 -7.72452950e-01 -4.53919411e-01
-6.28684461e-01 8.30738097e-02 1.07053530e+00 6.28058314e-01
1.32483554e+00 -5.15840709e-01 -5.11490583e-01 1.26851201e+00
1.52066958e+00 -6.28965735e-01 3.60608935e-01 -1.25868591e-02
1.43126214e+00 3.49004149e-01 -4.11205679e-01 1.21495649e-01
3.82389188e-01 3.42786640e-01 6.67999685e-01 6.53368831e-02
-3.07881832e-01 -7.05421209e-01 1.54233873e-01 8.61573219e-01
-2.70099729e-01 -2.90554047e-01 -8.23868692e-01 3.15999538e-01
-1.83260930e+00 -7.76772141e-01 -4.35400158e-01 1.90692329e+00
2.03570336e-01 4.58140858e-02 3.16645950e-02 -4.51344907e-01
5.56882977e-01 6.11962020e-01 -7.34794259e-01 -3.48877132e-01
-2.56853908e-01 3.75154257e-01 1.01430392e+00 8.08979154e-01
-8.63489628e-01 8.60909998e-01 6.62130260e+00 4.89860833e-01
-8.85670781e-01 -5.42012528e-02 2.88461983e-01 -1.30404219e-01
-1.05167139e+00 1.33866951e-01 -2.89210349e-01 2.11987928e-01
9.25310433e-01 -2.21622556e-01 7.07569659e-01 8.14396799e-01
1.36217456e-02 7.10117400e-01 -1.34773731e+00 8.38988543e-01
-1.45676926e-01 -1.98581874e+00 7.76730776e-01 5.90277493e-01
4.76755857e-01 4.40302193e-01 2.87207663e-01 1.29104167e-01
8.79909456e-01 -1.34317720e+00 4.60486919e-01 7.55849361e-01
8.21685791e-01 -5.11556149e-01 -3.60963792e-02 -1.17978305e-01
-1.44976807e+00 2.48689219e-01 -7.57556975e-01 6.85759112e-02
8.98441523e-02 4.92924660e-01 -9.88130867e-01 8.99380505e-01
3.05414259e-01 1.11259246e+00 -5.13603091e-01 9.09703195e-01
1.02208436e-01 2.73014933e-01 -6.09793842e-01 -1.32037893e-01
7.11145699e-01 -7.17303038e-01 8.24129462e-01 9.66584742e-01
2.79691339e-01 -2.40867093e-01 2.75899261e-01 1.38238215e+00
-3.98843318e-01 7.69959111e-03 -1.23312676e+00 -3.37754667e-01
3.18857282e-01 1.13785744e+00 -6.95260346e-01 -1.84410438e-02
-1.24778450e-01 7.35235512e-01 8.79746437e-01 7.26304948e-01
-6.21800482e-01 -2.89290935e-01 1.08117294e+00 3.40341955e-01
3.34242463e-01 -5.81931889e-01 -2.14947268e-01 -9.31479156e-01
-2.66253091e-02 -2.76426911e-01 1.80718318e-01 -9.93612885e-01
-1.52962887e+00 4.50252056e-01 -3.31846267e-01 -8.05789471e-01
3.60694796e-01 -1.19079340e+00 -5.67400873e-01 7.84279108e-01
-1.39380431e+00 -1.36371696e+00 -1.07276343e-01 2.60951579e-01
-2.57689416e-01 1.90673560e-01 6.67067349e-01 6.99252239e-04
-4.01230872e-01 5.34526944e-01 2.86570132e-01 -6.29070625e-02
1.01263903e-01 -1.64415205e+00 1.09565127e+00 4.18448180e-01
1.83319688e-01 1.14421189e+00 7.34135568e-01 -8.19920778e-01
-1.83653772e+00 -1.56558907e+00 4.89872277e-01 -8.70573461e-01
1.17279983e+00 -6.00991070e-01 -1.10970128e+00 1.01453710e+00
-1.95453510e-01 4.27831233e-01 6.26817703e-01 4.22686160e-01
-8.56551051e-01 2.81722546e-01 -1.09006846e+00 6.97184086e-01
1.69512784e+00 -9.02277470e-01 -9.84583572e-02 7.51397848e-01
1.28064883e+00 -2.03375015e-02 -1.35490263e+00 2.02974364e-01
4.63229209e-01 -6.21558666e-01 1.24262655e+00 -1.14877760e+00
1.66336522e-01 -1.70000464e-01 -3.50532293e-01 -1.56990063e+00
-6.85994983e-01 -1.06634033e+00 -2.46680945e-01 4.06283587e-01
2.99011409e-01 -9.06799734e-01 9.50854778e-01 2.53661245e-01
-4.53287363e-01 -9.87883270e-01 -1.03007257e+00 -8.37820470e-01
4.05292124e-01 -4.38090652e-01 8.04890752e-01 1.10408175e+00
-1.28557980e-01 2.78786570e-01 1.27794400e-01 2.51804560e-01
9.81857896e-01 -3.03564936e-01 6.90633178e-01 -1.44858837e+00
-2.45290115e-01 -9.66813564e-01 -9.69386756e-01 -1.24014974e+00
2.00471178e-01 -1.58266509e+00 -5.23831189e-01 -1.77961206e+00
-7.41453990e-02 -4.19058979e-01 -2.95223117e-01 3.17973047e-01
-8.27616304e-02 1.71605185e-01 -1.84286814e-02 1.89616188e-01
-5.72195768e-01 7.99388230e-01 1.54706407e+00 -1.74023286e-01
5.17854914e-02 -5.21719933e-01 -8.22324634e-01 4.40919369e-01
6.71450496e-01 -2.75955021e-01 -4.90541488e-01 -5.80368102e-01
5.50009310e-01 -3.71323854e-01 8.08847010e-01 -6.48595870e-01
1.98726833e-01 -2.99862609e-03 9.09401700e-02 -3.74009103e-01
4.34790701e-01 -4.52462465e-01 4.04242009e-01 2.63346225e-01
-3.92620832e-01 2.27282688e-01 3.22290063e-02 1.09019792e+00
5.59132934e-01 2.56014884e-01 6.70664847e-01 -2.03177139e-01
-2.26884648e-01 9.55798507e-01 1.85306937e-01 -7.75012523e-02
6.76132739e-01 -4.31600899e-01 -7.29506671e-01 -1.95674986e-01
-8.61525714e-01 1.36931643e-01 9.83010888e-01 2.80246846e-02
5.62265277e-01 -1.43177414e+00 -4.68752295e-01 1.86889872e-01
2.81720281e-01 1.76692918e-01 1.65429056e-01 8.55024517e-01
-8.83936584e-01 2.24048376e-01 1.25511184e-01 -7.60270417e-01
-6.55341506e-01 7.40698457e-01 5.34438014e-01 -4.44498435e-02
-1.02968311e+00 1.06366789e+00 4.09530461e-01 -8.00629199e-01
-9.52117741e-02 -7.78559208e-01 3.76597703e-01 -9.29484740e-02
-1.00558400e-01 2.64497787e-01 7.72945806e-02 -5.72581947e-01
-1.34659171e-01 5.53648949e-01 -9.36366543e-02 4.12374109e-01
1.49099851e+00 2.14841247e-01 -3.41706216e-01 4.01145101e-01
1.66895235e+00 -2.16031849e-01 -1.34108317e+00 -2.02097788e-01
-1.07996874e-01 -2.22725511e-01 -8.97228718e-02 -8.20078701e-02
-7.18060791e-01 8.43745589e-01 1.40115637e-02 6.49228394e-01
1.87896177e-01 3.77781838e-01 6.44100189e-01 6.92839861e-01
2.22811341e-01 -5.96497059e-01 6.18432797e-02 3.37072521e-01
7.99460411e-01 -8.75436723e-01 2.85447627e-01 -3.10865074e-01
-4.61498126e-02 1.05428588e+00 1.57907575e-01 -7.60414898e-01
8.13316405e-01 -2.12258130e-01 -4.24549550e-01 -8.84151399e-01
-7.69629478e-01 -2.70650417e-01 3.99284542e-01 9.28647816e-01
2.64795154e-01 3.42412800e-01 3.75926524e-01 -7.40131140e-02
-4.79148239e-01 -4.87385333e-01 4.20579612e-01 5.12460053e-01
-5.87287843e-01 -7.22368538e-01 2.21578509e-01 8.15965056e-01
-2.34759767e-02 -4.07323033e-01 -5.70240974e-01 9.60943699e-01
-3.24884027e-01 3.58272731e-01 2.41808161e-01 -5.00189245e-01
3.14421088e-01 -1.51679546e-01 5.76514661e-01 -6.21269047e-01
-2.77863204e-01 -2.98563391e-01 -6.16365634e-02 -7.84750581e-01
-6.50698021e-02 -4.37543362e-01 -9.19939160e-01 -7.77626991e-01
-1.72844470e-01 1.35660127e-01 4.90085721e-01 5.28831244e-01
7.62982070e-01 5.75647354e-01 1.67430386e-01 -1.16473091e+00
-5.26285470e-01 -5.84075630e-01 -5.82391560e-01 5.78183591e-01
5.72827637e-01 -7.78471828e-01 -5.03331244e-01 -4.43433374e-01] | [6.845322132110596, 6.121209144592285] |
5058e172-8910-4393-9ff4-460a024bcc3b | a-software-architecture-for-autonomous | 2010.12598 | null | https://arxiv.org/abs/2010.12598v1 | https://arxiv.org/pdf/2010.12598v1.pdf | A Software Architecture for Autonomous Vehicles: Team LRM-B Entry in the First CARLA Autonomous Driving Challenge | The objective of the first CARLA autonomous driving challenge was to deploy autonomous driving systems to lead with complex traffic scenarios where all participants faced the same challenging traffic situations. According to the organizers, this competition emerges as a way to democratize and to accelerate the research and development of autonomous vehicles around the world using the CARLA simulator contributing to the development of the autonomous vehicle area. Therefore, this paper presents the architecture design for the navigation of an autonomous vehicle in a simulated urban environment that attempts to commit the least number of traffic infractions, which used as the baseline the original architecture of the platform for autonomous navigation CaRINA 2. Our agent traveled in simulated scenarios for several hours, demonstrating his capabilities, winning three out of the four tracks of the challenge, and being ranked second in the remaining track. Our architecture was made towards meeting the requirements of CARLA Autonomous Driving Challenge and has components for obstacle detection using 3D point clouds, traffic signs detection and classification which employs Convolutional Neural Networks (CNN) and depth information, risk assessment with collision detection using short-term motion prediction, decision-making with Markov Decision Process (MDP), and control using Model Predictive Control (MPC). | ['Fernando Santos Osório', 'Denis Fernando Wolf', 'Jean Amaro', 'Angelica Tiemi Mizuno Nakamura', 'Tiago Cesar dos Santos', 'Júnior Anderson Rodrigues da Silva', 'Iago Pacheco Gomes', 'Luis Alberto Rosero'] | 2020-10-23 | a-software-architecture-for-autonomous-1 | https://arxiv.org/abs/2010.12598 | https://arxiv.org/abs/2010.12598 | journal-of-systems-architecture-in-submission | ['carla-map-leaderboard'] | ['robots'] | [-5.66522598e-01 4.22178328e-01 2.77068704e-01 -4.18473989e-01
-3.32888961e-01 -3.88852566e-01 1.02037108e+00 -2.38222972e-01
-6.60115659e-01 3.46116424e-01 -3.41056317e-01 -1.01194978e+00
-3.20632339e-01 -1.03553760e+00 -4.79230165e-01 -3.89586568e-01
-2.42129683e-01 1.00218678e+00 6.98060751e-01 -1.11895502e+00
3.04260701e-01 9.84647274e-01 -2.21235585e+00 -8.82397518e-02
7.08921969e-01 1.10129917e+00 1.75402150e-01 7.70694256e-01
3.03050037e-02 6.26687169e-01 -2.62433924e-02 -8.43232498e-03
4.78973150e-01 2.29311958e-01 -5.67607701e-01 -6.32711232e-01
1.70303971e-01 -1.80499375e-01 -5.71837902e-01 2.88551122e-01
2.95709759e-01 2.83090711e-01 7.28503108e-01 -1.76758671e+00
5.82639337e-01 -2.08093703e-01 3.53694201e-01 2.67105937e-01
-4.84196506e-02 8.54838669e-01 3.45465183e-01 -8.13597143e-01
5.97849190e-01 1.31975317e+00 6.27740026e-01 3.96471590e-01
-4.36704338e-01 -5.08063376e-01 -1.04724899e-01 7.10968852e-01
-1.36311817e+00 -4.68905181e-01 2.67431438e-01 -7.58123994e-01
1.21295798e+00 -8.66549611e-02 7.64090657e-01 9.06848073e-01
6.24086201e-01 4.62657481e-01 6.41593993e-01 1.10035412e-01
6.16165280e-01 9.75297838e-02 8.79868586e-03 6.65348589e-01
2.57287145e-01 9.42519307e-01 -7.51647055e-02 3.41729254e-01
-3.76898676e-01 -4.96286988e-01 6.63479567e-01 -5.24034083e-01
-8.65576088e-01 1.01334107e+00 1.56913131e-01 -1.90741047e-01
-6.01563632e-01 2.24791020e-01 5.25713265e-01 2.48680934e-01
-2.83170372e-01 8.09701458e-02 -2.78938621e-01 -5.35987675e-01
-4.62093711e-01 1.00779152e+00 6.59974933e-01 8.39958966e-01
7.33565271e-01 3.22161585e-01 6.38726354e-02 1.78268492e-01
7.82305837e-01 7.82134056e-01 1.15636643e-02 -1.40820289e+00
3.29430372e-01 6.33594573e-01 3.00720215e-01 -9.60467339e-01
-8.58586609e-01 -1.57100841e-01 -2.79272825e-01 1.39784789e+00
3.25155556e-01 -5.53363979e-01 -9.71936107e-01 1.21891463e+00
3.52819473e-01 -1.98083714e-01 3.08953434e-01 8.23208570e-01
4.21712726e-01 7.45306730e-01 1.35920674e-01 7.52323747e-01
1.16311777e+00 -6.71938062e-01 -4.13178623e-01 -2.97440052e-01
1.05013514e+00 -5.02080142e-01 3.56037170e-01 5.17636538e-01
-9.88946319e-01 -9.36170578e-01 -1.57437158e+00 -3.72750871e-02
-9.83636022e-01 -3.78813803e-01 2.56723046e-01 7.06285775e-01
-1.28836787e+00 2.14423820e-01 -5.88538706e-01 -4.61887270e-01
4.80841160e-01 3.36960405e-01 -3.37361068e-01 -1.25287816e-01
-1.50217736e+00 1.70726311e+00 1.70375481e-01 9.50638354e-02
-1.44729269e+00 -6.85251832e-01 -8.59376013e-01 -2.36600086e-01
1.47374704e-01 -4.74642068e-01 1.00012231e+00 -1.31456479e-02
-1.39290619e+00 8.04093719e-01 3.78077710e-03 -1.07014489e+00
9.16297376e-01 -1.86343178e-01 -6.74658537e-01 -3.66899759e-01
3.29995275e-01 1.13673830e+00 1.05940029e-01 -1.03417003e+00
-1.59284008e+00 -2.47515440e-01 4.08361629e-02 2.02550903e-01
6.03824735e-01 -4.27203596e-01 -1.86332107e-01 3.28359127e-01
-2.53074318e-01 -1.17333937e+00 -6.37993455e-01 -2.47708365e-01
1.95193868e-02 -5.38323343e-01 1.37099445e+00 -2.12527171e-01
6.97549045e-01 -2.22429943e+00 -1.81007221e-01 5.30189157e-01
4.19473462e-02 3.74481708e-01 4.80588749e-02 6.65989816e-01
2.91331053e-01 -1.81591973e-01 -6.35929629e-02 -7.04922527e-02
3.66820067e-01 4.93556559e-01 -3.23966086e-01 3.41160029e-01
2.23350048e-01 7.37387240e-01 -6.56989753e-01 -1.77334115e-01
7.67979980e-01 3.30186337e-01 -1.88731372e-01 -1.29908636e-01
-1.66105866e-01 2.49595642e-01 -4.36945885e-01 1.97742701e-01
8.28243911e-01 8.64737213e-01 -4.55551028e-01 3.62346351e-01
-1.09696984e+00 2.30990708e-01 -1.32679939e+00 1.17652369e+00
-4.86598343e-01 1.20120299e+00 3.92190456e-01 -8.82895052e-01
9.75469172e-01 1.81571282e-02 5.93112409e-01 -1.26224625e+00
2.97994405e-01 5.55684686e-01 1.26629829e-01 -7.28311479e-01
6.65401459e-01 1.72109142e-01 -2.78784782e-01 -1.41893759e-01
-4.59253013e-01 -4.08999860e-01 2.71049917e-01 2.56848216e-01
1.20005560e+00 9.73262861e-02 -4.37184751e-01 -4.60024357e-01
8.48415375e-01 8.22051346e-01 3.68095756e-01 6.45996749e-01
-8.82641852e-01 1.30180851e-01 3.73337537e-01 -9.08695817e-01
-1.14950323e+00 -9.78773832e-01 -7.10093230e-02 5.80324352e-01
4.30656403e-01 -6.05608001e-02 -7.25736141e-01 -4.59011525e-01
1.62190005e-01 1.04897881e+00 -5.53581953e-01 -1.81373626e-01
-7.25209355e-01 -1.05771363e-01 6.02814198e-01 1.73145771e-01
5.77314675e-01 -8.74360800e-01 -1.14235425e+00 3.05148125e-01
2.47986689e-01 -1.13542342e+00 4.25074279e-01 2.78008252e-01
-4.54717837e-02 -1.25948858e+00 -2.08822750e-02 -6.69504166e-01
-1.14939079e-01 3.34368050e-01 5.88071406e-01 -7.60422051e-02
-4.35875088e-01 1.84392646e-01 6.09734934e-03 -1.08009708e+00
-7.61750519e-01 1.73823446e-01 1.55054972e-01 -2.73520708e-01
5.10725737e-01 -2.10201114e-01 -7.91638434e-01 7.10437715e-01
-3.26820314e-01 -1.29448925e-03 5.47349632e-01 -7.65013602e-03
1.49090841e-01 1.83427900e-01 4.80516493e-01 -1.08736366e-01
3.77536118e-01 -6.91972077e-01 -1.00974560e+00 -5.55283427e-01
-7.21869469e-01 -3.08815807e-01 3.23285550e-01 4.66656148e-01
-7.03368008e-01 2.64116377e-01 -5.21040201e-01 8.63496512e-02
-6.22530222e-01 1.86251864e-01 -2.43429825e-01 -1.90158233e-01
6.48778498e-01 -1.36435926e-01 5.42406619e-01 1.19529203e-01
4.91552204e-01 6.28735363e-01 5.11767328e-01 -5.39311059e-02
1.04854870e+00 1.00856113e+00 7.61967838e-01 -7.68059552e-01
8.92704725e-02 -4.51637626e-01 -7.54610419e-01 -9.00771499e-01
1.12411737e+00 -9.95489120e-01 -1.14982486e+00 4.31796730e-01
-1.04986739e+00 -5.86950064e-01 -3.06303442e-01 4.48228002e-01
-7.11095512e-01 -1.44036785e-01 2.11264685e-01 -8.56739640e-01
1.46925047e-01 -1.44671428e+00 5.87190390e-01 1.56061828e-01
-1.60673335e-01 -6.69968545e-01 3.49370122e-01 3.91072452e-01
9.95081723e-01 4.30278301e-01 5.99615693e-01 -3.64488840e-01
-8.42804730e-01 -3.94564837e-01 -7.83688873e-02 1.42926380e-01
-4.60863560e-01 9.75038782e-02 -8.11386168e-01 8.98135453e-02
-4.87767458e-01 6.00128099e-02 6.94598079e-01 3.79693151e-01
3.66340101e-01 4.07764882e-01 -6.36243045e-01 2.75295854e-01
1.16061342e+00 7.80435622e-01 8.87953401e-01 9.00671721e-01
2.98189312e-01 1.26603043e+00 8.94739151e-01 -3.52587104e-02
9.59276140e-01 9.82181728e-01 1.19561493e+00 -1.19730458e-02
-8.16292390e-02 7.10566044e-02 4.50775295e-01 2.52934963e-01
1.58221036e-01 -6.25212304e-03 -1.42304051e+00 7.89200902e-01
-1.79576099e+00 -8.26094627e-01 -9.58652437e-01 1.89756215e+00
-3.82325232e-01 6.88455164e-01 2.47831225e-01 3.47267017e-02
2.28782862e-01 -2.09477901e-01 -3.10026318e-01 -1.00377476e+00
5.27617037e-02 -2.18467236e-01 1.07334518e+00 9.11680222e-01
-9.62824285e-01 8.66985857e-01 6.38737583e+00 7.13858306e-01
-8.48894000e-01 9.71743837e-02 2.92961657e-01 -2.23200191e-02
-4.12565209e-02 1.41784117e-01 -1.06185758e+00 5.54691315e-01
1.83581293e+00 -7.84799531e-02 1.00721993e-01 9.66059208e-01
9.50465798e-01 -4.06822592e-01 -5.52912116e-01 3.25575113e-01
-3.78811508e-01 -1.69223845e+00 -1.61061451e-01 3.86464655e-01
4.55213487e-01 8.72444689e-01 1.36999935e-01 8.59129608e-01
6.85597956e-01 -1.08013952e+00 1.06870103e+00 6.40157342e-01
2.21325248e-01 -1.04836345e+00 8.54713261e-01 4.36169952e-01
-1.03270233e+00 -4.75193173e-01 -1.01045975e-02 -2.44459197e-01
5.88455617e-01 -6.61807042e-03 -9.61907506e-01 5.59776068e-01
7.03165770e-01 1.81697011e-01 -3.21150750e-01 1.13854218e+00
2.66094089e-01 2.93375731e-01 -4.17794853e-01 -1.97311595e-01
1.01782358e+00 -5.08274324e-02 9.93707478e-01 9.63065445e-01
2.60516107e-01 -3.43302876e-01 1.74447954e-01 4.21680421e-01
6.18369460e-01 -2.94065684e-01 -1.04102135e+00 8.58065248e-01
2.36466751e-01 1.02746284e+00 -2.49607071e-01 -1.00160502e-01
-1.48818284e-01 6.97682127e-02 -2.50289708e-01 3.11830807e-02
-9.12811577e-01 -6.15257680e-01 1.16160631e+00 7.54776657e-01
2.61503816e-01 -5.09997487e-01 -5.75128615e-01 1.67259306e-01
-1.60195276e-01 -1.45826310e-01 -1.16554953e-01 -7.88890183e-01
-4.35293555e-01 5.49533427e-01 2.64775217e-01 -1.15351129e+00
-3.84244680e-01 -9.96178746e-01 -8.83755326e-01 1.05631852e+00
-1.89934766e+00 -9.19748902e-01 -5.03103733e-01 2.70108879e-01
6.90430284e-01 -6.85013652e-01 4.20950860e-01 5.25867164e-01
-3.37766021e-01 -8.91608223e-02 -1.90399066e-02 -4.45892006e-01
1.66955143e-01 -1.03980303e+00 8.10849428e-01 6.53922915e-01
-8.42363417e-01 -1.92957535e-01 8.28614712e-01 -6.60052419e-01
-1.42565799e+00 -1.28604341e+00 9.53056514e-01 -6.07628644e-01
5.62945843e-01 -3.63405883e-01 -4.15030301e-01 2.90132493e-01
1.94023103e-01 -4.27084893e-01 -5.57031259e-02 -4.08189714e-01
3.18670630e-01 -3.79425108e-01 -9.95929718e-01 6.32808089e-01
5.86405873e-01 -1.01741161e-02 -9.25062671e-02 1.82296842e-01
3.45985264e-01 -3.64220321e-01 -2.03573778e-01 5.54492772e-01
5.74499905e-01 -8.58616769e-01 5.71325660e-01 -3.77247274e-01
8.58870670e-02 -7.09270537e-01 -1.42330945e-01 -1.00270581e+00
-1.59667462e-01 -7.64893353e-01 1.43394753e-01 6.56231642e-01
6.31972849e-01 -7.09000468e-01 9.99540091e-01 5.82902670e-01
-7.88244724e-01 -6.10375941e-01 -1.73945689e+00 -8.19807291e-01
3.90215844e-01 -1.05048001e+00 5.95538378e-01 7.91810527e-02
-5.24112284e-01 4.42392938e-02 8.99578556e-02 3.64187449e-01
5.28647602e-01 -8.86469066e-01 1.17726481e+00 -1.43980777e+00
8.00772190e-01 -8.50519359e-01 -1.01441205e+00 -5.95176458e-01
-3.14267650e-02 -6.14522696e-01 2.69271344e-01 -1.87794256e+00
-7.70052552e-01 -7.10158229e-01 1.64309368e-01 6.28178893e-03
5.53376853e-01 -1.40137881e-01 -1.37070278e-02 3.75568569e-02
-4.28326577e-01 4.05669570e-01 7.07188904e-01 -2.70389408e-01
2.34673917e-02 2.34416202e-01 -4.18299645e-01 4.49104518e-01
9.46817756e-01 -1.07519396e-01 -4.37934697e-01 -1.33542031e-01
5.14637113e-01 -1.43503428e-01 6.34326339e-01 -1.78723824e+00
6.05898023e-01 -1.50348485e-01 -2.11605847e-01 -1.34009826e+00
4.42321062e-01 -1.04978240e+00 1.88001543e-01 1.01184130e+00
1.91682279e-02 1.83701068e-01 6.63498342e-01 3.64029855e-01
-9.64505076e-02 2.43908633e-02 8.14736009e-01 2.08973780e-01
-1.33467734e+00 2.36769587e-01 -1.44544244e+00 -2.35892504e-01
1.84281635e+00 -6.77637517e-01 -2.40513429e-01 -2.68970698e-01
-5.69552720e-01 1.04295623e+00 1.46536574e-01 8.19270611e-01
5.66081285e-01 -1.09515417e+00 -9.90542233e-01 3.99751067e-01
1.29693806e-01 1.43429758e-02 5.95211446e-01 5.93810737e-01
-1.08258343e+00 1.05934989e+00 -6.28519297e-01 -6.01430535e-01
-7.91514277e-01 3.38226974e-01 7.93423831e-01 -6.54247105e-02
-7.47502983e-01 2.85785228e-01 -1.90392420e-01 -5.99899352e-01
9.98339504e-02 1.02449775e-01 -5.75245440e-01 -6.22700006e-02
5.07989287e-01 7.50805199e-01 4.07380790e-01 -1.01319349e+00
-4.41167414e-01 2.79742569e-01 1.33260891e-01 -4.66242760e-01
8.25673223e-01 -4.45232153e-01 6.27606928e-01 3.39214504e-01
9.99760568e-01 -4.58363205e-01 -1.41183531e+00 6.49778128e-01
4.96977125e-04 2.90488023e-02 3.84070873e-01 -8.60339165e-01
-6.82861209e-01 9.99000132e-01 1.12947452e+00 2.59805173e-01
1.84896812e-01 -4.85198557e-01 9.24382150e-01 4.73784506e-01
5.15292585e-01 -1.34426999e+00 -5.79552829e-01 1.08481324e+00
7.28358567e-01 -1.16575003e+00 -4.14164633e-01 -2.72488166e-02
-5.75168550e-01 1.00041032e+00 7.98240542e-01 -1.08847372e-01
8.53737712e-01 2.37559900e-01 3.72760653e-01 -6.03311300e-01
-9.22504723e-01 -4.45526958e-01 -1.82765529e-01 1.14824557e+00
-6.61821902e-01 1.49077639e-01 -1.03915252e-01 3.57910953e-02
-2.89354503e-01 -2.03221411e-01 3.63788456e-01 8.51036370e-01
-9.63039100e-01 -7.33778775e-01 -2.71392256e-01 1.03974007e-01
1.01541772e-01 4.14996058e-01 -7.01575447e-03 1.26229239e+00
5.80300748e-01 1.22815728e+00 2.10151613e-01 -5.09655476e-01
1.02924013e+00 -1.27847910e-01 -4.06849444e-01 4.67038266e-02
-5.31622887e-01 -8.47903192e-01 3.82079452e-01 -8.86420310e-01
3.10777664e-01 -8.23926866e-01 -1.49398458e+00 -4.86004204e-01
4.97983396e-01 1.60221055e-01 1.57084262e+00 1.08731365e+00
6.58313692e-01 5.90268493e-01 8.21022391e-01 -9.91579294e-01
-7.30057731e-02 -5.39618254e-01 -7.01588169e-02 -5.76843657e-02
3.06646675e-01 -8.63982797e-01 -2.01447815e-01 -5.37845492e-01] | [5.652741432189941, 1.0372872352600098] |
443e5dbe-94c9-4950-83a6-8347bcded544 | tbgc-task-level-backbone-oriented-gradient | 2307.03465 | null | https://arxiv.org/abs/2307.03465v1 | https://arxiv.org/pdf/2307.03465v1.pdf | TBGC: Task-level Backbone-Oriented Gradient Clip for Multi-Task Foundation Model Learning | The AllInOne training paradigm squeezes a wide range of tasks into a unified model in a multi-task learning manner. However, optimization in multi-task learning is more challenge than single-task learning, as the gradient norm from different tasks may vary greatly, making the backbone overly biased towards one specific task. To address this issue, we propose the task-level backbone-oriented gradient clip paradigm, compared with the vanilla gradient clip method, it has two points of emphasis:1) gradient clip is performed independently for each task. 2) backbone gradients generated from each task are rescaled to the same norm scale. Based on the experimental results, we argue that the task-level backbone-oriented gradient clip paradigm can relieve the gradient bias problem to some extent. We also propose a novel multi-branch data augmentation strategy where conflict augmentations are placed in different branches. Our approach has been shown to be effective and finally achieve 1st place in the Leaderboard A and 2nd place in the Leaderboard B of the CVPR2023 Foundation Model Challenge. It's worth noting that instead of evaluating all three tasks(detection, segmentation and fine-grained classification) in Leaderboard A, the segmentation task is not evaluated in Leaderboard B, in which our team has a huge advantage. | ['Xue Pan', 'Zelun Zhang'] | 2023-07-07 | null | null | null | null | ['data-augmentation', 'multi-task-learning'] | ['methodology', 'methodology'] | [ 4.58597928e-01 4.41371985e-02 -9.21817869e-02 -3.71667892e-01
-9.92034912e-01 -3.16189975e-01 5.50694048e-01 1.54209554e-01
-6.03296399e-01 6.05523169e-01 2.21687272e-01 -9.69609395e-02
3.71735580e-02 -1.18374936e-01 -7.23676205e-01 -5.80017805e-01
3.97145115e-02 2.57802784e-01 3.26727808e-01 -2.49642059e-01
1.20646469e-01 7.45670497e-02 -1.38894844e+00 5.62313616e-01
8.95595610e-01 9.25225616e-01 3.51076603e-01 6.13174319e-01
-5.67551889e-02 6.00669205e-01 -7.98245072e-01 -3.06569487e-01
4.95079428e-01 -8.88333842e-02 -8.37117374e-01 5.87525219e-02
1.02824962e+00 -7.56755173e-02 2.84595400e-01 8.33808541e-01
5.32036662e-01 1.31552786e-01 4.62592542e-01 -1.41396880e+00
-2.30939776e-01 3.12696576e-01 -8.26220810e-01 2.45377392e-01
-1.77441955e-01 -6.75175190e-02 1.27195585e+00 -9.74167645e-01
5.11675119e-01 1.23656738e+00 8.59113455e-01 4.28543776e-01
-1.35963917e+00 -7.00821221e-01 7.99525678e-01 2.91970395e-03
-9.56387877e-01 -1.25247031e-01 7.22656846e-01 -5.31086206e-01
8.59650075e-01 1.60860404e-01 4.63467747e-01 1.24504852e+00
1.87423497e-01 1.28755975e+00 1.39535308e+00 -2.57122129e-01
-2.05048304e-02 1.36346057e-01 1.55651301e-01 5.31197846e-01
6.30257949e-02 -1.11365743e-01 -5.24842322e-01 -8.01996663e-02
5.25586545e-01 -1.20326743e-01 -1.53036878e-01 -4.76792306e-01
-1.39340353e+00 8.46334577e-01 3.68492633e-01 4.75002378e-01
-2.43315235e-01 1.67928129e-01 6.57383382e-01 3.71085107e-01
8.64181459e-01 6.34504855e-01 -7.69377172e-01 -9.75367501e-02
-1.20357049e+00 4.56081331e-01 5.60166657e-01 6.74645662e-01
7.92145789e-01 4.50502373e-02 -5.89483142e-01 1.02599227e+00
1.67967200e-01 8.34257454e-02 4.97473478e-01 -7.34966099e-01
6.18265152e-01 4.41742212e-01 2.11666729e-02 -9.06843662e-01
-6.76741004e-01 -7.39208162e-01 -7.16707408e-01 5.20204902e-01
5.88560045e-01 -3.92548472e-01 -9.95009422e-01 1.83215237e+00
1.80380657e-01 1.32630855e-01 -3.69138181e-01 8.97365153e-01
6.14056170e-01 3.94871026e-01 2.64501721e-01 1.05015509e-01
1.28184652e+00 -1.53715432e+00 -5.86091757e-01 -7.43100047e-01
6.95947766e-01 -9.00799811e-01 1.34208095e+00 5.58691800e-01
-8.44709039e-01 -7.58271813e-01 -1.22614193e+00 -1.35670602e-01
-3.93208742e-01 2.81448603e-01 7.34107494e-01 4.26932633e-01
-1.08250427e+00 3.32996130e-01 -5.88821232e-01 -6.56312332e-02
4.39160109e-01 1.02315292e-01 -2.60160267e-01 -1.85619537e-02
-1.00546598e+00 1.09390485e+00 3.28076243e-01 -4.72824685e-02
-8.38359892e-01 -9.33754981e-01 -7.04695702e-01 -4.39172499e-02
5.01838207e-01 -5.13160706e-01 1.30113709e+00 -1.01174569e+00
-1.13256049e+00 1.00493109e+00 -1.12055950e-01 -3.77351999e-01
7.86343217e-01 -5.70882440e-01 -7.39403814e-02 -3.08717608e-01
2.28869051e-01 9.01606739e-01 1.12453473e+00 -1.25228107e+00
-8.16889405e-01 -3.26650441e-01 -2.88165938e-02 4.90597606e-01
-5.19474149e-01 -2.49285297e-03 -5.82902014e-01 -8.57042193e-01
-1.47775516e-01 -8.66741240e-01 -4.15517658e-01 -1.24847703e-01
-3.67584586e-01 -2.79899955e-01 1.13528156e+00 -5.05325675e-01
1.11429513e+00 -2.12223768e+00 2.02779099e-01 -1.86803043e-01
4.24784184e-01 4.01670009e-01 -3.43702614e-01 1.11980841e-01
-1.51020497e-01 2.21413746e-01 -3.29780787e-01 -9.33487713e-01
-1.69179127e-01 2.22459510e-02 -1.80342048e-01 2.34561533e-01
3.77661228e-01 8.27756584e-01 -9.04445112e-01 -2.95884341e-01
3.18236649e-02 4.69127864e-01 -4.59320426e-01 -3.77427377e-02
-3.43567699e-01 5.33984065e-01 -4.48404044e-01 4.26826268e-01
6.14476144e-01 -3.69634360e-01 9.64897990e-05 -2.78984785e-01
-2.38697648e-01 2.60091484e-01 -1.30535638e+00 1.91920769e+00
-6.11138225e-01 6.56450927e-01 4.09458309e-01 -1.07420599e+00
7.60686398e-01 3.27384979e-01 4.37010437e-01 -6.39649749e-01
-2.53572881e-01 8.99460390e-02 6.31067529e-02 -8.02212209e-02
6.49530530e-01 -2.01256610e-02 -1.47021145e-01 4.08138037e-01
-3.84777859e-02 -3.59619766e-01 3.44356358e-01 7.33463392e-02
1.01136613e+00 3.79077852e-01 -3.22293420e-03 -3.37440819e-01
3.05184782e-01 4.46928404e-02 7.17947364e-01 8.44260037e-01
-4.97009963e-01 7.72220254e-01 5.50489247e-01 -4.06799018e-01
-9.25094783e-01 -7.55775869e-01 -7.86247998e-02 1.75069654e+00
-2.59009421e-01 -4.57940042e-01 -5.49069226e-01 -1.00374341e+00
1.58976480e-01 2.54668772e-01 -6.96414948e-01 8.85565728e-02
-6.27412200e-01 -1.08301675e+00 4.67424780e-01 5.58716357e-01
6.20238841e-01 -1.01517653e+00 -6.60835803e-01 2.71873862e-01
-1.83061644e-01 -1.30468476e+00 -6.24744713e-01 4.55143929e-01
-7.89066076e-01 -1.04033685e+00 -9.16329563e-01 -7.83759952e-01
4.20367837e-01 3.77981365e-01 1.25634885e+00 -5.36712445e-02
-3.02941442e-01 2.95447350e-01 -3.65457386e-01 -6.44701719e-01
-1.54727697e-01 3.34439546e-01 -2.99676687e-01 1.49463892e-01
4.58158255e-02 -3.99192780e-01 -4.78451848e-01 2.69090652e-01
-6.94905758e-01 1.44520357e-01 7.77494967e-01 8.83195341e-01
6.84942126e-01 -3.49628091e-01 7.60417163e-01 -1.19657564e+00
9.08487439e-01 -1.96740717e-01 -5.01174390e-01 2.10407510e-01
-7.76740789e-01 -1.82248697e-01 4.30799365e-01 -4.57535088e-01
-9.65923369e-01 1.40137762e-01 -1.09989576e-01 -3.91579688e-01
1.42486980e-02 3.37191671e-01 1.19129591e-01 -8.29965547e-02
6.25017524e-01 -3.97951752e-02 -8.92782118e-03 -5.69964528e-01
4.62539256e-01 3.17448467e-01 3.18265945e-01 -5.17346501e-01
6.25228107e-01 4.01212692e-01 -1.32544607e-01 -7.23683357e-01
-1.21763897e+00 -5.47690570e-01 -5.85797787e-01 -1.58424690e-01
1.05809808e+00 -1.09724867e+00 -4.87465799e-01 7.13001847e-01
-1.20823336e+00 -6.51158392e-01 -2.50365525e-01 2.73344338e-01
-2.81260192e-01 2.75760531e-01 -4.88090545e-01 -5.38277268e-01
-3.88747752e-01 -1.29342782e+00 1.26880169e+00 5.03896810e-02
-1.35852486e-01 -1.12882590e+00 -6.70918077e-02 5.47077000e-01
6.28421068e-01 3.57447833e-01 8.05130184e-01 -8.77312183e-01
-2.62830436e-01 -1.13203198e-01 -3.15763474e-01 7.39409149e-01
-2.80250423e-02 -2.17997476e-01 -1.14656067e+00 -5.10631919e-01
2.49699026e-01 -6.56642318e-01 1.10252678e+00 4.16427374e-01
1.15693450e+00 2.43443444e-01 -2.13169202e-01 4.49611574e-01
1.15596414e+00 -2.31759891e-01 3.08433920e-01 4.53587204e-01
8.59555602e-01 6.18773282e-01 6.45744085e-01 1.33203521e-01
4.20566559e-01 8.69433165e-01 5.77900887e-01 -5.70256770e-01
-4.18401271e-01 2.63970755e-02 4.03440714e-01 4.85306144e-01
1.76280469e-01 -1.34887369e-02 -9.89731014e-01 5.45811713e-01
-1.86726606e+00 -5.95199943e-01 -3.42144817e-01 2.16361570e+00
7.45896697e-01 4.38300520e-01 2.91316688e-01 1.28765972e-02
5.32808185e-01 5.19913971e-01 -5.62504709e-01 -3.38998288e-01
-2.61167318e-01 7.95799345e-02 3.42155218e-01 5.07122874e-01
-1.36549938e+00 1.06084561e+00 6.82906055e+00 8.38665247e-01
-1.25902629e+00 5.43047607e-01 7.05757618e-01 -2.95717746e-01
6.53548166e-02 -6.79503977e-02 -1.04591966e+00 3.83979529e-01
4.37909275e-01 2.19871178e-01 1.10070281e-01 8.05156529e-01
9.94261578e-02 -1.65579349e-01 -1.02969301e+00 7.93580890e-01
8.94976929e-02 -1.13969660e+00 -1.53575897e-01 2.89557446e-02
7.23131895e-01 4.36323911e-01 1.41913801e-01 5.54943621e-01
3.49814445e-01 -1.04736209e+00 7.40490377e-01 6.77278042e-02
4.69209880e-01 -5.23598731e-01 4.48420912e-01 3.99911284e-01
-1.15357625e+00 -4.52427082e-02 -1.07407063e-01 -1.17793880e-01
2.21027449e-01 9.03224051e-01 -8.39554608e-01 4.90955353e-01
7.23059654e-01 5.44500291e-01 -7.46060193e-01 1.00385487e+00
-3.91482323e-01 5.64692438e-01 -1.22152857e-01 2.31819347e-01
4.70822126e-01 -1.94726482e-01 6.27045572e-01 1.49318361e+00
-1.90009639e-01 -6.88963115e-01 6.79974854e-01 7.17640638e-01
-1.48981437e-01 1.14658974e-01 -3.31574529e-01 1.96204454e-01
1.64587483e-01 1.70066524e+00 -5.80964386e-01 -2.43316427e-01
-6.56658351e-01 1.05737782e+00 4.03362989e-01 3.51947486e-01
-7.59871304e-01 -2.43263096e-01 5.76665938e-01 -7.79494941e-02
3.69570196e-01 -2.84844637e-01 -6.57059073e-01 -8.88209879e-01
1.18852206e-01 -1.00681591e+00 2.76339382e-01 -5.40748775e-01
-1.22749412e+00 5.80845952e-01 -1.18245542e-01 -8.57853591e-01
-5.44466972e-02 -6.61675096e-01 -6.83653235e-01 1.05283105e+00
-1.75413740e+00 -1.33898079e+00 -2.08548844e-01 4.49942291e-01
7.91272581e-01 -1.02208361e-01 6.55327559e-01 4.85332221e-01
-7.13263929e-01 7.27834821e-01 -4.72972244e-02 -6.11416064e-03
1.01179516e+00 -1.48482609e+00 4.10010338e-01 8.02014291e-01
9.31871384e-02 3.14693630e-01 5.83111048e-01 -5.15427351e-01
-9.96401250e-01 -1.10094941e+00 8.38756263e-01 -4.83695030e-01
6.94950819e-01 -6.50581002e-01 -9.82043564e-01 7.99610198e-01
2.00533479e-01 1.00616053e-01 5.22499740e-01 5.81957638e-01
-4.63843167e-01 -3.59069780e-02 -9.21725988e-01 4.68397379e-01
8.54390562e-01 -4.10535991e-01 -3.96659940e-01 7.20396399e-01
6.82715654e-01 -5.46797872e-01 -7.78298318e-01 5.11644006e-01
2.93598443e-01 -8.78414571e-01 8.68865490e-01 -5.39827943e-01
4.35006738e-01 -1.63014755e-01 2.93765236e-02 -1.53537810e+00
-1.81711242e-01 -5.47905684e-01 -3.85975018e-02 1.13367319e+00
6.91984117e-01 -5.72437108e-01 7.79937506e-01 3.14494878e-01
-4.60726529e-01 -1.03055310e+00 -9.42778170e-01 -8.64327669e-01
3.72858524e-01 -4.56900954e-01 1.18674114e-01 9.73988295e-01
-3.75911951e-01 6.83589518e-01 -6.58678710e-01 -1.93300292e-01
4.62059528e-01 7.54199456e-03 8.35195303e-01 -1.37328947e+00
-4.80108112e-01 -6.35976374e-01 9.89357382e-02 -1.17926693e+00
6.49217218e-02 -1.05753028e+00 1.35224253e-01 -1.50727308e+00
1.62346065e-01 -6.35693431e-01 -3.93534988e-01 7.81825900e-01
-4.99736845e-01 2.05488160e-01 5.99940717e-01 2.59968579e-01
-5.95684469e-01 3.51653516e-01 1.43090403e+00 -5.00282347e-02
-1.50787175e-01 1.34905055e-01 -9.23284352e-01 6.75075293e-01
6.24231160e-01 -4.42662269e-01 -3.95470142e-01 -7.09615231e-01
9.91901085e-02 -4.85464662e-01 3.05435240e-01 -9.58742142e-01
8.35591480e-02 1.20171800e-01 2.35023305e-01 -2.33431861e-01
4.71886188e-01 -4.69327748e-01 -4.07231927e-01 4.15982008e-01
-3.54992330e-01 1.31249681e-01 4.79359746e-01 4.13076073e-01
-2.53597260e-01 -1.54896662e-01 7.89751172e-01 -1.17871486e-01
-8.24455500e-01 2.04577178e-01 -1.61541626e-01 4.18423295e-01
8.46817553e-01 -4.83611748e-02 -2.61288047e-01 -1.11263707e-01
-6.56764746e-01 5.20201266e-01 1.90331504e-01 6.58409357e-01
2.56274641e-01 -1.04186618e+00 -9.58178580e-01 6.87578470e-02
5.80157340e-02 1.86848745e-01 1.51269540e-01 1.05887449e+00
1.96901157e-01 3.02446395e-01 -3.76894586e-02 -6.71298563e-01
-1.17302155e+00 1.69517845e-01 2.71319360e-01 -1.02497888e+00
-5.92486143e-01 1.07552838e+00 2.68635273e-01 -4.86977875e-01
4.22345519e-01 -2.53164560e-01 -2.32197344e-01 4.06795710e-01
3.62167180e-01 2.96688080e-01 3.11381340e-01 -4.08257186e-01
-2.21649632e-01 4.68817472e-01 -6.02911353e-01 -1.11503363e-01
1.36783791e+00 2.30643615e-01 1.45280540e-01 4.68867391e-01
1.07595789e+00 -6.11495413e-02 -1.52976155e+00 -2.63382524e-01
2.77974665e-01 -2.10632473e-01 2.79386252e-01 -1.19007754e+00
-1.07863820e+00 9.51354027e-01 6.09481156e-01 1.21717937e-01
1.09072137e+00 -2.23534286e-01 7.92723119e-01 1.78531155e-01
1.56647816e-01 -1.39166057e+00 4.26824778e-01 7.25531340e-01
1.05084550e+00 -1.40551805e+00 1.75275374e-02 -3.86892200e-01
-9.11859930e-01 7.32513785e-01 8.23363900e-01 -2.48735305e-02
4.22332913e-01 3.35848004e-01 1.20192043e-01 -1.87376872e-01
-7.11068451e-01 -3.23486656e-01 3.77015591e-01 4.51315790e-01
6.87073648e-01 -1.65454224e-01 -3.81171763e-01 3.62603515e-01
1.24901854e-01 -4.78795618e-02 2.70312309e-01 9.81891215e-01
-4.61500973e-01 -1.33569121e+00 -2.86041677e-01 4.75150913e-01
-6.18348539e-01 -2.52305627e-01 -2.71804482e-01 9.06489789e-01
2.59437472e-01 8.89797449e-01 -1.47697136e-01 -3.16902608e-01
5.24221778e-01 3.33548069e-01 3.35838377e-01 -8.55476797e-01
-1.01990235e+00 9.98533294e-02 1.12342693e-01 -5.54533064e-01
-2.35605687e-01 -5.73189020e-01 -1.08045626e+00 1.85230926e-01
-6.00836799e-02 -9.94215757e-02 8.86424363e-01 9.69455004e-01
5.44587970e-01 8.13938320e-01 2.81013757e-01 -9.84562457e-01
-7.10209429e-01 -1.08257914e+00 -3.12033892e-01 6.35643482e-01
3.65745723e-01 -7.64458895e-01 -3.54895383e-01 -1.19053356e-01] | [9.499124526977539, 2.658980131149292] |
2723da6a-8eac-41c7-b509-d6e503ba2dd0 | copner-contrastive-learning-with-prompt | null | null | https://aclanthology.org/2022.coling-1.222 | https://aclanthology.org/2022.coling-1.222.pdf | COPNER: Contrastive Learning with Prompt Guiding for Few-shot Named Entity Recognition | Distance metric learning has become a popular solution for few-shot Named Entity Recognition (NER). The typical setup aims to learn a similarity metric for measuring the semantic similarity between test samples and referents, where each referent represents an entity class. The effect of this setup may, however, be compromised for two reasons. First, there is typically a limited optimization exerted on the representations of entity tokens after initing by pre-trained language models. Second, the referents may be far from representing corresponding entity classes due to the label scarcity in the few-shot setting. To address these challenges, we propose a novel approach named COntrastive learning with Prompt guiding for few-shot NER (COPNER). We introduce a novel prompt composed of class-specific words to COPNER to serve as 1) supervision signals for conducting contrastive learning to optimize token representations; 2) metric referents for distance-metric inference on test samples. Experimental results demonstrate that COPNER outperforms state-of-the-art models with a significant margin in most cases. Moreover, COPNER shows great potential in the zero-shot setting. | ['Chen Li', 'Rui Mao', 'Tieliang Gong', 'Xianli Zhang', 'Yige Wang', 'Kai He', 'YuCheng Huang'] | null | null | null | null | coling-2022-10 | ['few-shot-ner'] | ['natural-language-processing'] | [ 7.71023333e-02 -3.38672772e-02 -1.64187416e-01 -5.14253676e-01
-1.19022810e+00 -2.26878285e-01 6.07935131e-01 3.91027242e-01
-7.87017226e-01 5.04024208e-01 1.44080579e-01 2.61617303e-01
-2.27297708e-01 -8.85165453e-01 -3.89847994e-01 -5.30438364e-01
1.99064195e-01 4.56929415e-01 1.53515458e-01 -1.46227852e-01
1.65711790e-01 3.50605875e-01 -1.58083260e+00 -3.54734273e-03
9.80980635e-01 1.00250077e+00 3.12053472e-01 3.96491915e-01
-5.44052780e-01 6.13583028e-01 -8.76632273e-01 -6.12307549e-01
1.11240707e-01 -5.46601057e-01 -6.57240510e-01 -2.10992411e-01
1.21538706e-01 1.09407213e-03 -1.84318244e-01 1.11460805e+00
7.62111962e-01 7.75885701e-01 8.47619772e-01 -1.25498796e+00
-7.35796571e-01 7.68262148e-01 -2.16372073e-01 2.53049821e-01
4.12393808e-02 -8.45168158e-02 1.38796723e+00 -1.31514215e+00
6.14348829e-01 1.01878774e+00 4.77762371e-01 7.99771786e-01
-1.01307559e+00 -4.76392537e-01 1.53921321e-02 4.13260579e-01
-1.61156249e+00 -5.55978537e-01 6.12155497e-01 -2.29888335e-01
8.08639705e-01 8.84517729e-02 5.24888784e-02 1.08965802e+00
-3.48862648e-01 8.67055237e-01 6.12364650e-01 -6.04959130e-01
6.52142406e-01 2.09113300e-01 3.94926816e-01 5.54203451e-01
3.37496512e-02 -1.41164899e-01 -5.26598513e-01 -1.55341193e-01
1.66316591e-02 7.21358433e-02 -1.88681319e-01 -3.30177516e-01
-9.51881528e-01 9.03739750e-01 4.62722093e-01 6.20664477e-01
-3.49022001e-01 -2.23968491e-01 4.23250169e-01 2.50874192e-01
5.09019911e-01 8.24546993e-01 -2.34292507e-01 -2.10146606e-01
-9.26136494e-01 5.55172050e-03 7.64206827e-01 9.46991265e-01
8.20257187e-01 -6.32527098e-02 -8.56409311e-01 1.18783450e+00
-3.36594433e-02 1.90488711e-01 6.73549354e-01 -3.98811430e-01
7.54304588e-01 3.38358164e-01 9.82592255e-02 -5.67776978e-01
-2.20245197e-01 -1.86325103e-01 -4.69533861e-01 -1.07843392e-01
3.91985685e-01 -1.97230235e-01 -8.12367141e-01 1.71691489e+00
3.41819733e-01 6.02722824e-01 1.55226335e-01 7.81916797e-01
7.79249966e-01 7.21874416e-01 3.34460288e-01 -4.31550182e-02
1.25078654e+00 -8.38331699e-01 -6.36178792e-01 -2.75287986e-01
1.08876014e+00 -5.40404081e-01 1.22220564e+00 -1.15522526e-01
-6.73361123e-01 -4.56181675e-01 -1.06478214e+00 7.78045803e-02
-6.75600290e-01 -4.45648432e-02 2.49436945e-02 6.97292864e-01
-5.14841795e-01 9.55358207e-01 -6.70037270e-01 -4.12970245e-01
5.07379472e-01 1.03478860e-02 -1.00180097e-01 -4.14483070e-01
-1.43533349e+00 1.00819826e+00 5.18324792e-01 -1.48849353e-01
-8.57957780e-01 -1.04653132e+00 -1.08810830e+00 6.77973390e-01
4.64624375e-01 -1.59554049e-01 1.28239691e+00 -2.27351204e-01
-1.26891267e+00 8.26858342e-01 -8.88583064e-02 -3.51410329e-01
3.43886107e-01 -1.54084578e-01 -6.12773716e-01 -1.49615854e-01
3.30363959e-01 4.05166626e-01 4.33697224e-01 -9.85652745e-01
-6.79844797e-01 -1.55656874e-01 -2.15864740e-02 2.40759671e-01
-8.58481705e-01 1.32428780e-01 -3.97628784e-01 -6.05593920e-01
-2.67410696e-01 -4.98859674e-01 -1.18006669e-01 -1.15814455e-01
-5.12546241e-01 -6.03692949e-01 6.40405595e-01 -1.94749907e-01
1.35977852e+00 -2.21466327e+00 -8.38665366e-02 5.61075918e-02
-3.36661823e-02 6.82586670e-01 -5.05711854e-01 3.32603216e-01
-2.76051350e-02 7.12705478e-02 -2.67038226e-01 -4.97207850e-01
3.82473290e-01 5.62842786e-02 -2.15424120e-01 3.14744174e-01
4.52592999e-01 7.43261397e-01 -1.29239738e+00 -4.82510507e-01
2.00389937e-01 4.07466263e-01 -2.40308896e-01 6.20432675e-01
1.98239014e-02 -3.55854221e-02 -4.59258467e-01 5.05326271e-01
4.00272757e-01 -1.64294280e-02 -1.61004677e-01 -1.61427617e-01
-3.54603007e-02 3.00757408e-01 -1.28144157e+00 1.68800080e+00
-6.31488919e-01 4.26710397e-01 -4.28560168e-01 -1.15152466e+00
1.16005671e+00 3.93046767e-01 3.69020700e-01 -7.55682707e-01
1.82324454e-01 1.53008282e-01 -2.08606124e-01 -4.50874358e-01
4.39481884e-01 -4.14411485e-01 -1.62564456e-01 4.53049511e-01
3.69923294e-01 8.95380508e-03 4.85714734e-01 1.67415906e-02
1.25796890e+00 -1.72599062e-01 4.11316246e-01 2.27734982e-03
3.04914445e-01 -3.02881390e-01 9.76165473e-01 8.31225574e-01
-4.13599849e-01 6.67884767e-01 4.02710885e-01 8.49059522e-02
-8.83051574e-01 -1.20427799e+00 -2.11914435e-01 1.29701793e+00
1.36692733e-01 -4.77126390e-01 -8.32028627e-01 -9.10509348e-01
-2.49520048e-01 1.27455318e+00 -5.44765770e-01 -5.37905753e-01
-3.62503469e-01 -5.80497503e-01 6.77690148e-01 6.55911863e-01
2.68021345e-01 -1.04589367e+00 -5.06960273e-01 2.45389104e-01
-8.47937260e-03 -1.09423637e+00 -6.78976834e-01 4.15403724e-01
-6.08470619e-01 -9.67630625e-01 -9.31187570e-01 -7.81852007e-01
5.37439048e-01 3.97611082e-01 1.01855600e+00 -2.84668177e-01
-2.46107385e-01 2.48332694e-01 -6.29853249e-01 -3.85343581e-01
-2.10894540e-01 1.92652151e-01 5.30422479e-02 1.20067209e-01
8.76484454e-01 -2.48753101e-01 -3.36866975e-01 3.86735380e-01
-8.51003945e-01 -5.53773344e-01 4.73686516e-01 1.03281534e+00
5.51538527e-01 -2.14932598e-02 6.93711877e-01 -1.02202857e+00
5.65406382e-01 -7.80566514e-01 -3.64923507e-01 5.31397879e-01
-5.64715564e-01 1.90078095e-01 8.86393905e-01 -5.56083322e-01
-1.04380882e+00 -3.33526075e-01 6.53393045e-02 -5.64351857e-01
-1.68783575e-01 5.64012110e-01 -5.07581115e-01 3.11730325e-01
8.64688396e-01 -8.53918269e-02 -4.97287601e-01 -6.30356133e-01
4.02297735e-01 7.89268434e-01 3.12075377e-01 -6.51076615e-01
8.22890699e-01 1.33370623e-01 -2.49815091e-01 -8.80590796e-01
-1.35213840e+00 -8.07468653e-01 -6.28637433e-01 4.92816418e-03
8.36543620e-01 -6.84271336e-01 -7.44539797e-02 3.58496085e-02
-1.05008066e+00 -1.63757250e-01 -6.53983295e-01 7.04311907e-01
-4.23452616e-01 9.93237048e-02 -4.64181572e-01 -6.91943288e-01
-2.99276888e-01 -8.67965341e-01 8.29242468e-01 4.47088480e-01
-2.01572269e-01 -1.04663026e+00 1.94690377e-01 1.02919325e-01
3.01796466e-01 4.68933620e-02 9.79174376e-01 -1.33006823e+00
-8.08498412e-02 -4.29555297e-01 -1.71652809e-01 3.57455909e-01
2.22440630e-01 -3.53920013e-01 -1.26202857e+00 -2.37406746e-01
-5.70537150e-02 -4.53170300e-01 7.10930288e-01 1.62796653e-03
1.07492411e+00 -2.03615744e-02 -2.91725606e-01 4.08945471e-01
1.30555451e+00 9.46184397e-02 3.88559639e-01 2.32900590e-01
6.88260496e-01 6.86740696e-01 1.06530082e+00 5.52687705e-01
7.37183392e-02 8.25136185e-01 3.49901691e-02 1.73593581e-01
-1.84619009e-01 -3.09333414e-01 2.74681538e-01 8.94331574e-01
3.02591681e-01 -4.30149198e-01 -1.08143687e+00 7.18776226e-01
-1.69895077e+00 -1.13139451e+00 3.84412229e-01 2.45639062e+00
7.15069592e-01 2.27858648e-02 -1.54863924e-01 5.07373847e-02
1.19388711e+00 2.48378500e-01 -7.74587274e-01 -1.41956508e-01
4.10987139e-02 3.25839043e-01 1.14257358e-01 4.60853912e-02
-1.12682021e+00 9.06904221e-01 4.84389639e+00 1.04630339e+00
-8.44969511e-01 3.33984405e-01 3.46929431e-01 -8.51527229e-02
-1.51158154e-01 -1.24477677e-01 -1.09303606e+00 5.04744530e-01
1.10929871e+00 -6.39141142e-01 7.09142536e-02 9.61754203e-01
7.67900702e-03 3.30939531e-01 -1.31315374e+00 1.09872806e+00
2.76899666e-01 -1.04051864e+00 -1.24992251e-01 -1.95010021e-01
8.45149517e-01 1.13718174e-01 -2.02536806e-01 8.77627254e-01
1.84182927e-01 -7.78615952e-01 5.47551632e-01 5.19235313e-01
7.73266137e-01 -9.94235039e-01 7.65185952e-01 3.19081008e-01
-1.19033265e+00 -7.11046085e-02 -8.94447863e-01 4.04510021e-01
1.28406897e-01 6.97290599e-01 -9.76623237e-01 5.34379840e-01
2.70121634e-01 5.69969118e-01 -3.67487580e-01 1.34751487e+00
-3.63911867e-01 6.31942570e-01 -2.05443497e-03 -2.53867328e-01
3.89705747e-01 -1.57237072e-02 4.93061423e-01 1.29760039e+00
6.67552769e-01 7.14360029e-02 2.19493583e-01 9.41388965e-01
-4.76458341e-01 3.25654954e-01 -5.79595447e-01 -2.87034899e-01
1.00849462e+00 1.26603353e+00 -6.02546513e-01 -3.42720002e-01
-5.85029662e-01 1.00723743e+00 5.69068670e-01 2.89810449e-01
-8.29793155e-01 -1.08559370e+00 8.19447577e-01 -1.56538054e-01
5.59143007e-01 1.60648808e-01 -5.73442765e-02 -1.27826107e+00
-1.55283036e-02 -3.95255595e-01 6.85155451e-01 -4.10951078e-01
-1.60449684e+00 5.32507062e-01 -2.59364516e-01 -1.50313699e+00
-1.39769971e-01 -5.54987431e-01 -9.86303329e-01 6.93062067e-01
-1.62095690e+00 -5.98382711e-01 -1.45290390e-01 4.43711549e-01
6.76131427e-01 -2.37235889e-01 1.14534235e+00 3.71374011e-01
-7.62806654e-01 9.35890436e-01 3.74720335e-01 4.66850847e-01
9.10569429e-01 -1.39281547e+00 2.54282057e-01 8.91575933e-01
4.78463471e-01 4.16810036e-01 5.41057765e-01 -3.43907177e-01
-1.02941430e+00 -1.35526454e+00 1.13338757e+00 -3.13697726e-01
5.76795638e-01 -4.29541886e-01 -1.08444893e+00 3.05104256e-01
-2.82677472e-01 3.53520870e-01 1.16530490e+00 3.96000028e-01
-4.64083254e-01 -1.62594303e-01 -1.18240237e+00 6.86783850e-01
8.90050590e-01 -6.98971331e-01 -9.17510033e-01 3.99743438e-01
7.12222993e-01 -1.17470115e-01 -9.17959929e-01 5.15776724e-02
1.04112111e-01 -6.36459172e-01 7.17374861e-01 -7.55525470e-01
2.55050868e-01 -2.98511386e-01 -3.42100769e-01 -1.68498695e+00
-2.02695400e-01 -3.00660729e-01 -3.25503945e-01 1.81611741e+00
3.55934232e-01 -1.53998852e-01 5.91569960e-01 8.98371577e-01
-2.28201970e-01 -5.82043469e-01 -9.59052444e-01 -1.19962490e+00
5.75873256e-02 -4.55356389e-01 6.95092976e-01 1.16446245e+00
1.62713453e-01 5.81594110e-01 -1.93143070e-01 9.87755805e-02
6.17402554e-01 -1.29889324e-01 3.36370319e-01 -1.31818807e+00
-1.16216414e-01 -3.25424731e-01 -6.66419029e-01 -7.31251359e-01
3.98687601e-01 -1.16173625e+00 4.92281914e-01 -1.35640764e+00
2.07865834e-01 -5.10803401e-01 -9.06092942e-01 4.73370761e-01
-4.87298965e-01 -3.74009646e-02 3.81320089e-01 -3.36523466e-02
-8.86983156e-01 8.53894174e-01 8.84472668e-01 -2.46085480e-01
-1.16683096e-01 1.34109388e-04 -5.60342014e-01 4.66702759e-01
6.42507851e-01 -8.23799372e-01 -4.52838033e-01 -3.51827472e-01
-1.19597457e-01 -1.22804888e-01 1.04680927e-02 -1.11478591e+00
5.32063961e-01 -5.60741536e-02 1.80251554e-01 -1.46288544e-01
2.23933220e-01 -5.65221012e-01 -5.07976234e-01 1.07459903e-01
-6.47968292e-01 -3.43926668e-01 -2.77739704e-01 5.97294867e-01
-3.87180239e-01 -8.53605986e-01 1.02977693e+00 6.25707954e-02
-9.27865803e-01 5.18593788e-01 9.12074298e-02 6.87278211e-01
9.66860294e-01 -6.32502064e-02 -2.92897284e-01 -7.65158795e-03
-4.68346208e-01 3.21186244e-01 1.53966725e-01 5.29713094e-01
5.64038873e-01 -1.51661468e+00 -6.36090219e-01 -1.18724424e-02
6.45197630e-01 -9.18174088e-02 2.63421416e-01 5.03186822e-01
1.74441397e-01 2.79496491e-01 6.58603832e-02 -2.64782071e-01
-9.28923905e-01 6.33843422e-01 1.68087915e-01 -2.70213693e-01
-5.40065587e-01 1.01044261e+00 1.26943171e-01 -5.07077515e-01
3.69318783e-01 8.36084932e-02 -2.96866477e-01 4.35149759e-01
8.85506451e-01 5.48192143e-01 2.15060115e-01 -4.50838953e-01
-2.70102292e-01 2.62224078e-01 -3.19929659e-01 -5.62868416e-02
1.47503829e+00 2.07952354e-02 4.66097623e-01 7.35540390e-01
1.43968809e+00 -2.53172398e-01 -1.21573234e+00 -5.99553883e-01
5.75657010e-01 -4.90249723e-01 2.18669120e-02 -4.95107174e-01
-7.67240644e-01 9.64359224e-01 5.88223040e-01 -4.37099189e-02
6.90145850e-01 -7.16450363e-02 9.44710851e-01 7.78476238e-01
2.88762420e-01 -1.49913466e+00 1.08401537e-01 6.55158937e-01
3.21533710e-01 -1.26656377e+00 -4.92121190e-01 -2.18303222e-02
-6.92529976e-01 1.07301652e+00 5.59934139e-01 -4.33547609e-02
5.45948684e-01 -3.22792791e-02 4.83665168e-02 -5.65442406e-02
-6.16174579e-01 -4.64607298e-01 4.16782320e-01 6.15749121e-01
5.73426604e-01 -1.55398235e-01 -1.69710562e-01 7.61599183e-01
2.87731253e-02 -2.90578127e-01 2.77296305e-01 8.90046656e-01
-6.43616676e-01 -1.03031206e+00 -9.79304537e-02 4.25913870e-01
-3.32080781e-01 -2.47760385e-01 5.21226367e-03 4.60628241e-01
3.73435132e-02 1.03473401e+00 9.43811610e-03 -2.20652655e-01
6.72392786e-01 4.47997600e-01 1.14222370e-01 -1.11623847e+00
-4.13142502e-01 -4.67401415e-01 7.43209347e-02 -4.70548540e-01
6.02012221e-03 -5.45931995e-01 -1.29036427e+00 1.21017814e-01
-5.18292844e-01 5.49391210e-01 5.35626709e-01 1.23426187e+00
3.16074282e-01 5.32410026e-01 9.85451519e-01 -5.34527540e-01
-9.63770866e-01 -1.01855803e+00 -8.89588058e-01 6.98078632e-01
-5.73777929e-02 -8.35451782e-01 -4.66837376e-01 -2.81410217e-01] | [9.697123527526855, 9.328908920288086] |
7bfee4a4-c5dd-4a5f-8417-ab3f8ffff21b | batch-bayesian-optimization-via-particle | 2209.04722 | null | https://arxiv.org/abs/2209.04722v2 | https://arxiv.org/pdf/2209.04722v2.pdf | Batch Bayesian Optimization via Particle Gradient Flows | Bayesian Optimisation (BO) methods seek to find global optima of objective functions which are only available as a black-box or are expensive to evaluate. Such methods construct a surrogate model for the objective function, quantifying the uncertainty in that surrogate through Bayesian inference. Objective evaluations are sequentially determined by maximising an acquisition function at each step. However, this ancilliary optimisation problem can be highly non-trivial to solve, due to the non-convexity of the acquisition function, particularly in the case of batch Bayesian optimisation, where multiple points are selected in every step. In this work we reformulate batch BO as an optimisation problem over the space of probability measures. We construct a new acquisition function based on multipoint expected improvement which is convex over the space of probability measures. Practical schemes for solving this `inner' optimisation problem arise naturally as gradient flows of this objective function. We demonstrate the efficacy of this new method on different benchmark functions and compare with state-of-the-art batch BO methods. | ['Andrew B. Duncan', 'Konstantinos Zygalakis', 'Simon L. Cotter', 'Enrico Crovini'] | 2022-09-10 | null | null | null | null | ['bayesian-optimisation'] | ['methodology'] | [ 2.72624433e-01 2.45856971e-01 3.42309862e-01 -3.70932579e-01
-1.18787193e+00 -6.15099370e-01 5.63982010e-01 4.51549470e-01
-7.06004739e-01 8.25727940e-01 -1.42161697e-01 -2.73853660e-01
-8.50045741e-01 -6.03421390e-01 -8.06554258e-01 -1.00005841e+00
-2.46231079e-01 7.55606413e-01 -4.37072385e-03 7.63028348e-03
4.50278580e-01 4.45021927e-01 -1.25302887e+00 -1.84411883e-01
9.17109847e-01 1.34379327e+00 4.97830659e-02 8.38302672e-01
1.08372800e-01 9.35956836e-02 -7.07063437e-01 -4.74119574e-01
2.50212461e-01 -3.58450323e-01 -8.16553771e-01 1.17797472e-01
-6.93947449e-02 2.06369072e-01 3.44951123e-01 1.25400126e+00
5.29512346e-01 4.96517897e-01 8.70657086e-01 -1.12813175e+00
4.92597707e-02 1.63690627e-01 -3.60046923e-01 1.30097270e-01
3.16933900e-01 8.21909606e-02 1.24079311e+00 -7.64669180e-01
6.34161606e-02 1.18308198e+00 7.52142787e-01 1.23547569e-01
-1.77490079e+00 9.90659222e-02 -1.28992140e-01 -5.05486690e-02
-1.33775246e+00 -4.01869744e-01 3.32089841e-01 -4.87648338e-01
6.95697665e-01 4.96498019e-01 5.16725659e-01 3.54335725e-01
3.11134279e-01 6.98147297e-01 1.13428354e+00 -4.04262424e-01
6.14618003e-01 7.91147128e-02 -6.74742237e-02 4.72259492e-01
9.25815478e-02 3.12074423e-01 -4.45777923e-01 -2.74983615e-01
3.89720738e-01 -2.35159650e-01 -2.90481925e-01 -5.49079537e-01
-1.06473374e+00 9.90066767e-01 3.60908389e-01 -1.12243630e-01
-6.12400115e-01 2.49405995e-01 1.11352839e-01 1.40084550e-01
4.86294091e-01 6.63929224e-01 -4.15241271e-01 -3.01125348e-01
-1.03826404e+00 4.27006274e-01 1.09479225e+00 4.00797725e-01
6.60573840e-01 -3.36862445e-01 -3.27230841e-01 7.56623685e-01
8.20566356e-01 3.20694119e-01 -1.63156968e-02 -9.29513097e-01
3.95791650e-01 1.64382443e-01 4.01542693e-01 -7.21412480e-01
-2.55426198e-01 -6.69224501e-01 -8.18684280e-01 3.95556182e-01
7.57032871e-01 -3.01227033e-01 -6.18025124e-01 1.37344038e+00
7.03544438e-01 -1.49843261e-01 -8.23193416e-02 8.46439183e-01
1.72160283e-01 8.80286515e-01 -1.93266869e-01 -4.99772936e-01
9.97501969e-01 -5.79938829e-01 -5.45532346e-01 -1.74557278e-03
3.36154401e-01 -7.18947172e-01 6.78988934e-01 8.10338140e-01
-1.28489602e+00 -1.52558193e-01 -1.00180626e+00 4.87361729e-01
-8.52029175e-02 -9.55223888e-02 3.04758132e-01 9.01473343e-01
-9.32052970e-01 9.74274516e-01 -8.11183929e-01 2.67470777e-01
3.37013751e-01 7.69316256e-01 6.50505871e-02 1.39499575e-01
-9.00242746e-01 1.00752175e+00 7.19043732e-01 7.61329889e-01
-9.91647422e-01 -7.50974894e-01 -7.74876058e-01 3.49877141e-02
6.67402744e-01 -4.18787688e-01 1.29158270e+00 -7.54759371e-01
-1.87732422e+00 6.03130460e-01 2.14882363e-02 -3.63600761e-01
7.13741601e-01 -2.33308434e-01 -2.83716302e-02 -1.75289690e-01
-1.74349129e-01 2.71403253e-01 1.10167456e+00 -1.12622368e+00
-5.55873156e-01 -2.85628051e-01 1.78923503e-01 1.23575270e-01
2.55842179e-01 1.40909299e-01 -2.74999380e-01 -5.63094497e-01
5.05600404e-03 -7.07938910e-01 -5.90509772e-01 -1.75545514e-01
-4.33972120e-01 -2.11484551e-01 2.02066988e-01 -7.54775405e-01
1.48190379e+00 -1.89827621e+00 4.64682728e-01 7.21076787e-01
3.09157018e-02 4.74325679e-02 2.31695712e-01 1.80872962e-01
1.21900588e-01 3.44520882e-02 -8.97035182e-01 -3.00156921e-01
3.15737516e-01 2.41141915e-01 1.88879550e-01 9.08168793e-01
3.35212886e-01 7.36144245e-01 -1.09972394e+00 -4.31051314e-01
1.89450949e-01 3.37849975e-01 -6.38818741e-01 3.21115136e-01
-4.07097369e-01 4.52669412e-01 -2.71629035e-01 3.39459062e-01
5.24045587e-01 -1.36276186e-01 2.60306448e-02 1.45878166e-01
-1.18607849e-01 1.07772410e-01 -1.67045546e+00 1.48030734e+00
-4.51396763e-01 4.16146487e-01 3.02071363e-01 -1.45166314e+00
9.30509627e-01 1.64835542e-01 6.37684703e-01 -7.79985189e-02
1.50458813e-01 2.11096659e-01 -1.23218261e-02 -6.29456863e-02
2.67328054e-01 -5.91155231e-01 -1.80530958e-02 2.42481455e-01
-1.61942970e-02 -4.67083365e-01 3.45042586e-01 -3.19974899e-01
1.00966501e+00 1.98657513e-01 2.99235851e-01 -5.28637946e-01
7.82189071e-01 -2.33698606e-01 3.23045641e-01 9.37018394e-01
1.40701383e-01 6.23748958e-01 7.84382284e-01 -1.54475316e-01
-7.82063901e-01 -1.10865641e+00 -5.52207768e-01 6.79502249e-01
-1.83449358e-01 -3.66837710e-01 -6.37135208e-01 -7.57980943e-01
-5.98581694e-02 7.63584554e-01 -4.57895696e-01 5.87673150e-02
-3.85176659e-01 -1.31504357e+00 -3.22984643e-02 1.77403912e-02
1.90492302e-01 -6.63236856e-01 -8.11010957e-01 5.58867395e-01
1.55476883e-01 -6.74003780e-01 -3.67497653e-01 3.49982709e-01
-9.74892318e-01 -8.35046589e-01 -7.53940940e-01 -1.82094708e-01
6.76172674e-01 -5.37861526e-01 1.30396950e+00 -1.44741729e-01
-2.76351035e-01 2.68478215e-01 -3.32803093e-02 -5.84652424e-01
-4.74615753e-01 -2.33906016e-01 -1.97869509e-01 4.28892642e-01
-6.82702437e-02 -1.77250937e-01 -5.28612912e-01 5.38270652e-01
-9.81057525e-01 -4.21613395e-01 4.33874488e-01 1.04263198e+00
7.94952095e-01 4.28858072e-01 3.50206226e-01 -5.24408817e-01
8.26706707e-01 -5.39506733e-01 -1.33168888e+00 4.16589230e-01
-6.16506040e-01 5.83790541e-01 2.35946238e-01 -1.79848611e-01
-1.02373862e+00 1.11090876e-01 -9.65731442e-02 -5.35756862e-03
1.56665340e-01 8.86612594e-01 -2.15253662e-02 -1.29226401e-01
4.78203475e-01 -2.76224077e-01 1.57505602e-01 -5.76908946e-01
2.70192623e-01 5.87197065e-01 5.45555055e-01 -8.91627669e-01
4.63865429e-01 2.01768160e-01 6.53291643e-01 -6.56458557e-01
-1.08027613e+00 -8.51301491e-01 -5.63913703e-01 -4.53212976e-01
6.40726447e-01 -1.45278946e-01 -1.15263021e+00 2.02096507e-01
-1.10770822e+00 -2.38509715e-01 -6.36017203e-01 5.61894417e-01
-7.71377087e-01 2.25195870e-01 9.09896716e-02 -1.32454765e+00
-7.91747272e-02 -1.20708609e+00 1.11535025e+00 2.19098032e-02
-1.18250966e-01 -1.33613431e+00 2.74354696e-01 7.94649720e-02
2.63773650e-01 4.32759494e-01 5.22360742e-01 -3.70062143e-01
-3.15932572e-01 -4.55255926e-01 -1.23962343e-01 5.84825277e-01
-1.97407141e-01 1.17815062e-01 -7.31116474e-01 -3.04080665e-01
4.00011957e-01 -7.49812275e-02 7.48548567e-01 7.07334757e-01
1.22012126e+00 -4.26281899e-01 -1.14770792e-01 5.40953159e-01
1.51308525e+00 -2.64875777e-02 4.84451473e-01 3.34698021e-01
2.08680421e-01 6.81863070e-01 6.75286770e-01 5.90644181e-01
-3.43641937e-02 8.82643521e-01 5.27356505e-01 2.90521890e-01
5.09568691e-01 2.89022803e-01 3.75660688e-01 6.41401350e-01
-2.00690597e-01 -1.39378086e-01 -1.01209188e+00 5.13662577e-01
-1.97115922e+00 -5.94519973e-01 -1.39802009e-01 2.81436253e+00
1.00409591e+00 2.88581878e-01 2.54614681e-01 2.94235736e-01
7.16580808e-01 -2.28444397e-01 -3.10313553e-01 -5.67104995e-01
2.15416029e-01 4.33365375e-01 6.32961452e-01 8.13115597e-01
-1.18279052e+00 1.52728051e-01 6.74573851e+00 9.12589252e-01
-7.35555172e-01 1.26616165e-01 6.02365851e-01 -1.53869420e-01
-2.80311834e-02 1.90713018e-01 -8.48144829e-01 6.52870595e-01
1.21029675e+00 5.00901341e-02 5.79900324e-01 4.35356706e-01
1.09679490e-01 -4.66644526e-01 -1.22864366e+00 8.55555296e-01
-3.38074863e-01 -1.00345266e+00 -7.41877794e-01 3.28234613e-01
9.07886863e-01 -5.62382974e-02 -1.77661344e-01 1.14753760e-01
4.53071624e-01 -1.41619051e+00 6.13468170e-01 7.63120234e-01
4.35636401e-01 -8.92257035e-01 8.31949234e-01 3.20163816e-01
-7.50463247e-01 -8.28844681e-02 -1.83768958e-01 1.76290944e-01
5.22418737e-01 1.07356858e+00 -7.33382523e-01 5.67985237e-01
5.93230307e-01 2.70144641e-01 -1.76251069e-01 1.72172523e+00
-2.62113005e-01 5.29651523e-01 -1.00119221e+00 -2.90518433e-01
5.45846581e-01 -6.29022539e-01 9.38518763e-01 1.16855872e+00
2.90972918e-01 -2.09508374e-01 2.51775142e-02 9.61888254e-01
1.60914257e-01 7.47426692e-03 -1.30604744e-01 4.20981199e-02
-2.31186785e-02 1.00997710e+00 -7.02851534e-01 2.72069639e-03
-2.00979739e-01 4.56714571e-01 2.62771070e-01 2.94770896e-01
-5.78596115e-01 -4.07925546e-01 3.83910418e-01 -2.75260597e-01
4.62605298e-01 -2.73387700e-01 -3.44494671e-01 -7.37816811e-01
1.42326832e-01 -8.08773041e-01 6.06819153e-01 -2.83615023e-01
-1.12253153e+00 2.07408920e-01 4.55353230e-01 -6.60317481e-01
-5.95002413e-01 -7.97795713e-01 -4.29636538e-01 9.87560987e-01
-1.19312930e+00 -3.53195727e-01 1.46056592e-01 2.18960941e-01
2.92272419e-01 2.10894927e-01 5.45876861e-01 -1.77786953e-03
-5.04330814e-01 3.12097192e-01 6.66997671e-01 -5.06886661e-01
7.17629269e-02 -1.70477629e+00 6.90800399e-02 7.51670718e-01
2.53641363e-02 3.69841158e-01 1.11956692e+00 -4.04358089e-01
-1.32685149e+00 -5.76274753e-01 6.40266776e-01 -6.31279707e-01
7.65925765e-01 -1.68539137e-01 -7.15970218e-01 2.09782466e-01
-2.72596985e-01 6.87825009e-02 2.99561977e-01 1.81149378e-01
4.10463184e-01 -1.03177711e-01 -1.22805488e+00 3.39055717e-01
4.43975776e-01 -7.44978040e-02 -4.08896983e-01 4.54138696e-01
7.24635422e-02 -3.86420935e-01 -1.31156337e+00 5.25220752e-01
4.79196817e-01 -7.31744349e-01 1.13470447e+00 -5.59396446e-01
1.40514284e-01 -4.25574303e-01 -8.71780813e-02 -1.45987213e+00
1.96478635e-01 -1.29686153e+00 -4.29167897e-01 9.97601688e-01
3.93571228e-01 -7.01261699e-01 5.71553409e-01 6.56759441e-01
1.16721347e-01 -1.16107059e+00 -1.23904991e+00 -1.01321542e+00
-4.54486199e-02 -7.23954380e-01 5.83481371e-01 2.96830714e-01
-2.21511349e-01 2.17753321e-01 -9.24909040e-02 1.31302148e-01
1.08939862e+00 -1.74825370e-01 5.55426776e-01 -1.29620183e+00
-7.58579850e-01 -7.24255085e-01 -4.48160738e-01 -1.02841711e+00
-1.50415644e-01 -7.04968989e-01 6.71013653e-01 -1.30613732e+00
-1.43907443e-01 -5.49735308e-01 -2.12344036e-01 5.43124080e-02
-3.06157887e-01 -1.19801193e-01 -1.58616230e-02 -1.27789333e-01
-5.39155781e-01 6.26977563e-01 1.05546713e+00 -4.65385662e-03
-2.41295695e-01 4.73699152e-01 -4.13562834e-01 7.69383967e-01
6.57364726e-01 -8.33210528e-01 -5.01941517e-02 -4.40570600e-02
6.91506326e-01 9.43597630e-02 4.06527191e-01 -7.15690494e-01
1.46077648e-01 -2.58432418e-01 8.97396803e-02 -6.15452468e-01
3.94800097e-01 -6.81365073e-01 2.82038778e-01 2.41593465e-01
-2.17345923e-01 -1.20825753e-01 3.26415198e-03 6.26248479e-01
-2.77725458e-01 -9.99218404e-01 8.65283310e-01 -6.76846355e-02
-1.93753794e-01 1.33928834e-02 -4.85719554e-02 2.30076507e-01
8.35323453e-01 -1.24417320e-01 3.01905125e-01 -2.83606887e-01
-8.54488254e-01 4.89119798e-01 1.32293120e-01 -3.19833338e-01
5.02466559e-01 -1.05167043e+00 -8.70050371e-01 -1.19847693e-01
-3.75498921e-01 5.11129975e-01 -2.25826055e-01 1.27153063e+00
-5.39766133e-01 3.51272702e-01 3.87107134e-01 -8.32522631e-01
-9.49531913e-01 3.80335033e-01 4.99757409e-01 -6.90424681e-01
-1.80352658e-01 1.03298438e+00 2.96283010e-02 -4.43195254e-01
2.59828329e-01 -1.89109370e-01 3.41305286e-02 -2.77080480e-02
2.30453297e-01 7.80991733e-01 2.85560071e-01 -4.70632076e-01
-3.18411827e-01 4.59724694e-01 3.92214954e-01 -6.61574662e-01
1.35124028e+00 -1.28999174e-01 -3.69758725e-01 3.36458266e-01
1.30977845e+00 -3.49783897e-01 -1.40717709e+00 -3.91115509e-02
4.06423032e-01 -5.54028392e-01 4.24974650e-01 -7.29993582e-01
-7.49915302e-01 6.69587374e-01 3.91349137e-01 5.05100787e-01
1.05883646e+00 -1.51321366e-01 2.69347548e-01 3.81840646e-01
-1.81276444e-02 -1.24770916e+00 -2.30733886e-01 4.11852717e-01
1.13705742e+00 -1.24220526e+00 2.79514730e-01 -1.60728619e-01
-2.42200106e-01 1.17169082e+00 -1.92689583e-01 -6.33399934e-02
8.92931402e-01 -3.72203402e-02 -3.91219348e-01 -3.55483413e-01
-4.05572623e-01 -2.29365453e-01 8.80066931e-01 2.79115558e-01
1.67510480e-01 9.19632539e-02 -4.46389496e-01 2.39920557e-01
-2.53187597e-01 -3.10490787e-01 1.23081930e-01 1.00125968e+00
-1.91003501e-01 -1.08686101e+00 -7.29506910e-01 5.97049952e-01
-8.56526077e-01 -6.27794191e-02 5.47603145e-02 4.21906233e-01
-2.67557621e-01 1.08158386e+00 -1.28064409e-01 2.81726301e-01
1.90423310e-01 -7.26867095e-03 5.84071219e-01 -5.34282148e-01
-4.32668358e-01 2.28044569e-01 2.32743695e-01 -5.38345397e-01
-3.10981363e-01 -1.06931055e+00 -6.84141219e-01 -2.05075461e-02
-6.82219863e-01 4.26159024e-01 1.23508966e+00 1.26069319e+00
-1.88402206e-01 4.74356085e-01 6.46532834e-01 -1.05609417e+00
-1.24284375e+00 -7.37473130e-01 -4.40273672e-01 4.16418910e-02
4.79775310e-01 -6.49544001e-01 -4.94707167e-01 -2.01542571e-01] | [6.324435710906982, 3.8555662631988525] |
72baa2f9-f1a9-4e03-b419-9d9c62c4ed15 | evaluating-machine-learning-models-with-nero | 2305.19889 | null | https://arxiv.org/abs/2305.19889v1 | https://arxiv.org/pdf/2305.19889v1.pdf | Evaluating Machine Learning Models with NERO: Non-Equivariance Revealed on Orbits | Proper evaluations are crucial for better understanding, troubleshooting, interpreting model behaviors and further improving model performance. While using scalar-based error metrics provides a fast way to overview model performance, they are often too abstract to display certain weak spots and lack information regarding important model properties, such as robustness. This not only hinders machine learning models from being more interpretable and gaining trust, but also can be misleading to both model developers and users. Additionally, conventional evaluation procedures often leave researchers unclear about where and how model fails, which complicates model comparisons and further developments. To address these issues, we propose a novel evaluation workflow, named Non-Equivariance Revealed on Orbits (NERO) Evaluation. The goal of NERO evaluation is to turn focus from traditional scalar-based metrics onto evaluating and visualizing models equivariance, closely capturing model robustness, as well as to allow researchers quickly investigating interesting or unexpected model behaviors. NERO evaluation is consist of a task-agnostic interactive interface and a set of visualizations, called NERO plots, which reveals the equivariance property of the model. Case studies on how NERO evaluation can be applied to multiple research areas, including 2D digit recognition, object detection, particle image velocimetry (PIV), and 3D point cloud classification, demonstrate that NERO evaluation can quickly illustrate different model equivariance, and effectively explain model behaviors through interactive visualizations of the model outputs. In addition, we propose consensus, an alternative to ground truths, to be used in NERO evaluation so that model equivariance can still be evaluated with new, unlabeled datasets. | ['Gordon L Kindlmann', 'Michael Maire', 'William Irvine', 'Takumi Matsuzawa', 'Zhuokai Zhao'] | 2023-05-31 | null | null | null | null | ['3d-point-cloud-classification', 'point-cloud-classification'] | ['computer-vision', 'computer-vision'] | [-3.50851685e-01 -4.05213714e-01 2.46970709e-02 -1.02229476e-01
-3.16691279e-01 -9.46633995e-01 7.72755384e-01 5.94822764e-01
1.11041710e-01 4.99330848e-01 -2.34023049e-01 -8.07290733e-01
-3.60982984e-01 -5.55274665e-01 -3.43808323e-01 -7.70855844e-01
-2.45112196e-01 5.36465228e-01 6.29240423e-02 -1.92337379e-01
4.60516691e-01 1.06668878e+00 -1.72865748e+00 -1.65332630e-01
8.65772843e-01 7.36176312e-01 -2.47918323e-01 5.23325026e-01
2.13034630e-01 3.85952324e-01 -7.12756395e-01 -1.98299214e-01
1.27168857e-02 -4.61150892e-02 -4.00199741e-01 -2.02478647e-01
2.90595710e-01 6.26578405e-02 2.80987799e-01 1.01526570e+00
2.44344354e-01 -9.71222445e-02 8.26233208e-01 -1.59407890e+00
-2.88412899e-01 7.30917901e-02 -3.15939158e-01 3.20052445e-01
2.85990179e-01 7.83771098e-01 7.98208058e-01 -6.23165190e-01
5.27845383e-01 1.33625531e+00 9.72348154e-01 7.04886615e-02
-1.50844717e+00 -7.22701967e-01 2.29421370e-02 -5.98260239e-02
-1.41583049e+00 -1.46871125e-02 5.99187076e-01 -9.34367418e-01
7.89572239e-01 1.00404024e+00 7.27876604e-01 7.95060098e-01
3.10484946e-01 2.93605387e-01 1.32556736e+00 2.18465216e-02
3.51908684e-01 3.20973486e-01 3.76465410e-01 5.79850495e-01
6.58656836e-01 4.77778643e-01 -3.87109965e-01 -3.33843827e-01
6.31562650e-01 -6.38992712e-02 -3.10121715e-01 -4.37217176e-01
-1.18012297e+00 4.51566190e-01 3.66441280e-01 -9.75398049e-02
-8.73806179e-02 -4.25045341e-02 2.36198589e-01 1.38477400e-01
4.75709021e-01 1.01414335e+00 -4.18704897e-01 -3.06846768e-01
-7.49711037e-01 6.16012156e-01 4.30883288e-01 4.06735152e-01
8.66429806e-01 1.68640018e-01 -1.15234166e-01 3.64592820e-01
4.40793574e-01 6.50060415e-01 1.53311104e-01 -8.76002252e-01
2.09327668e-01 1.10779357e+00 4.68052477e-01 -1.17248905e+00
-7.70868778e-01 -7.65928864e-01 -6.89489782e-01 1.05908132e+00
4.46301371e-01 1.55882597e-01 -8.82399976e-01 1.41260958e+00
4.51052666e-01 4.93254885e-02 7.12751299e-02 1.10117102e+00
7.39035249e-01 3.97049099e-01 1.17525294e-01 1.12325676e-01
1.19788742e+00 -4.88652498e-01 -3.62681478e-01 -1.20753825e-01
9.64298606e-01 -6.46086812e-01 1.47274303e+00 1.94753811e-01
-7.98862815e-01 -3.30203682e-01 -1.14938414e+00 2.97159821e-01
-5.72223604e-01 -3.30900624e-02 4.87068862e-01 5.98456323e-01
-6.88438952e-01 1.07088137e+00 -1.26832366e+00 -3.09100986e-01
2.99716175e-01 7.48756155e-02 -2.29011938e-01 2.80234337e-01
-9.07391012e-01 1.12798750e+00 9.79522802e-03 1.00888684e-01
-8.26121092e-01 -1.18351483e+00 -7.41665423e-01 2.47875929e-01
8.99866968e-02 -6.36214018e-01 1.15607429e+00 -4.26512837e-01
-9.54402089e-01 4.42552656e-01 -1.90422162e-02 -2.31413692e-01
6.71456218e-01 8.26749504e-02 -5.41891277e-01 -3.40975374e-01
-4.07417528e-02 3.18827033e-01 5.18394470e-01 -1.48955238e+00
-1.60272717e-01 -2.80294508e-01 -2.53799185e-02 8.40010345e-02
1.58963740e-01 6.72661439e-02 -1.77502528e-01 -5.14440715e-01
2.92222410e-01 -8.53863716e-01 -2.02280864e-01 2.89703518e-01
-4.98947471e-01 1.40037343e-01 1.30920839e+00 -5.01587570e-01
1.48556077e+00 -1.96924090e+00 -3.58261466e-01 3.96351188e-01
4.35439318e-01 3.07207912e-01 1.85026094e-01 5.77192783e-01
-3.45804513e-01 6.27321959e-01 -6.01481259e-01 -2.57289767e-01
5.09954020e-02 4.00287360e-02 -2.98093408e-01 3.10982496e-01
6.25266790e-01 6.59048080e-01 -8.59535694e-01 -1.32135734e-01
5.40678859e-01 4.23681796e-01 -2.50329554e-01 -9.13438946e-03
-1.11874558e-01 7.17554390e-01 -2.16710925e-01 7.39044189e-01
8.82554173e-01 -4.49778795e-01 -9.51710939e-02 -1.55908123e-01
-4.55200106e-01 2.48589337e-01 -1.34232414e+00 6.64058506e-01
-3.30252886e-01 9.65733409e-01 -1.15951061e-01 -6.24145389e-01
1.02241313e+00 -6.85589090e-02 2.18600094e-01 -6.30873263e-01
-3.42982672e-02 2.50051711e-02 -3.25247087e-02 -2.37453163e-01
4.01867449e-01 4.07748073e-02 3.02074432e-01 7.13688016e-01
-5.37576735e-01 -1.80040568e-01 1.47000834e-01 1.23375617e-01
7.14534342e-01 1.44304886e-01 7.98093602e-02 -4.99522358e-01
1.72386572e-01 3.14573705e-01 4.72217560e-01 5.38738549e-01
1.07511185e-01 7.19526768e-01 8.64513159e-01 -5.55060506e-01
-1.02443540e+00 -1.08033514e+00 -4.96517926e-01 1.59917697e-01
3.20908666e-01 -9.18173134e-01 -5.24726033e-01 -5.23656368e-01
2.69852847e-01 8.87166739e-01 -6.83988214e-01 -3.22476089e-01
-2.83357769e-01 -9.80862141e-01 4.61502165e-01 5.64385951e-01
1.60663232e-01 -4.63357359e-01 -9.35275495e-01 -2.58887947e-01
8.11517388e-02 -6.16354287e-01 8.39071274e-02 -4.77594920e-02
-1.21705341e+00 -1.36866558e+00 -3.20719928e-01 7.47775659e-02
7.80204177e-01 3.10057938e-01 1.33870625e+00 4.80636179e-01
-2.51818031e-01 2.20120445e-01 -8.03200752e-02 -6.57223582e-01
-6.10151470e-01 -3.73823732e-01 2.18329936e-01 -2.38116443e-01
-2.74101403e-02 -5.06759763e-01 -5.55632770e-01 9.54353154e-01
-8.74941766e-01 2.38862932e-01 2.28871286e-01 6.73561513e-01
5.33748686e-01 -1.30732562e-02 6.43319115e-02 -6.12463236e-01
7.31170654e-01 -3.85426521e-01 -9.93763030e-01 4.44915950e-01
-1.30091608e+00 2.70318776e-01 6.78892195e-01 -3.34030956e-01
-7.61499047e-01 -3.44736278e-01 2.95598984e-01 -6.81943536e-01
1.21190976e-02 6.81324184e-01 6.31515756e-02 -1.19999141e-01
1.01581609e+00 -1.88172936e-01 2.01624408e-01 -6.70190752e-01
-8.98922309e-02 3.33757162e-01 2.85135537e-01 -4.53848124e-01
9.20328379e-01 4.44094539e-01 2.80887693e-01 -7.78434992e-01
-1.95321277e-01 -2.24768594e-01 -4.11267817e-01 -5.34145355e-01
3.62997204e-01 -5.52899837e-01 -8.74261856e-01 2.97602654e-01
-9.19241905e-01 -4.00259465e-01 -2.14383602e-01 5.16921878e-01
-5.87841570e-02 3.80637527e-01 5.60591333e-02 -9.13435638e-01
6.35755658e-02 -1.53002024e+00 1.15679836e+00 3.92373651e-01
-4.40328807e-01 -1.30053806e+00 1.33444056e-01 1.26258120e-01
5.48652470e-01 4.53545421e-01 1.13386953e+00 -4.19559300e-01
-5.25819182e-01 -3.63176763e-01 -3.10231537e-01 2.55471619e-04
-4.49284017e-02 7.38780439e-01 -1.19281971e+00 -3.18725348e-01
-2.60530770e-01 5.33964634e-02 4.72042441e-01 2.02133790e-01
1.18249261e+00 -2.18827367e-01 -5.30595601e-01 5.43452263e-01
1.04016304e+00 -3.78382020e-02 5.48112214e-01 4.19100434e-01
4.19998556e-01 6.42562866e-01 6.44673288e-01 3.02801728e-01
1.73937067e-01 8.43392074e-01 7.40167201e-01 -5.56443095e-01
9.18796733e-02 -1.91044673e-01 8.53957683e-02 3.12078178e-01
-2.40575150e-01 6.01865314e-02 -1.44995284e+00 1.43096410e-02
-1.74419475e+00 -6.67519391e-01 -5.96736550e-01 2.61124420e+00
2.45624647e-01 1.67170867e-01 3.46636400e-02 1.93560734e-01
5.93350291e-01 3.23164389e-02 -7.42490053e-01 -3.46292287e-01
-7.33777955e-02 -1.92342490e-01 5.04212022e-01 5.69125831e-01
-7.66427219e-01 4.54446495e-01 6.68729973e+00 5.25137007e-01
-1.65041924e+00 -1.05779722e-01 6.51600003e-01 9.75564215e-03
-5.43387890e-01 4.75336343e-01 -5.00372708e-01 2.92816341e-01
9.35158908e-01 -4.02066946e-01 1.57735556e-01 9.27465796e-01
6.24128640e-01 -2.70428777e-01 -1.17299116e+00 1.16029608e+00
-3.57250065e-01 -1.56733203e+00 1.77265368e-02 9.39167514e-02
4.21270072e-01 -5.70901595e-02 6.28764629e-02 1.04983643e-01
2.20676929e-01 -1.15312243e+00 8.32844019e-01 7.97343314e-01
6.85597539e-01 -5.03222287e-01 5.63499868e-01 1.77358359e-01
-1.19502258e+00 1.60132349e-01 -2.05003932e-01 -2.45852828e-01
6.33482561e-02 5.33864439e-01 -9.92898941e-01 7.41648912e-01
9.32585537e-01 6.88607216e-01 -1.04494977e+00 1.33574975e+00
3.37551720e-02 6.44630849e-01 -4.31510866e-01 2.09474936e-02
9.76690929e-03 -2.26787537e-01 8.24160516e-01 1.09519231e+00
2.02951595e-01 -2.09077895e-01 -2.49462202e-01 1.33542347e+00
4.83350933e-01 -1.65362105e-01 -6.78743958e-01 -6.12250976e-02
4.37975973e-01 1.33340204e+00 -1.08077514e+00 -1.70721069e-01
-3.12452149e-02 2.97394872e-01 7.99490325e-03 5.65647721e-01
-8.25713277e-01 -1.38676047e-01 1.01061094e+00 3.59753579e-01
-2.56011963e-01 -5.29022992e-01 -6.01683795e-01 -1.01248491e+00
3.43607292e-02 -9.24191356e-01 1.14353016e-01 -1.07452655e+00
-9.44006801e-01 3.62917006e-01 3.26587945e-01 -1.68628824e+00
-9.83095542e-02 -8.29401016e-01 -1.03538823e+00 9.24715102e-01
-1.13984692e+00 -8.19626629e-01 -6.27635360e-01 2.22800523e-02
6.13190643e-02 1.27384484e-01 7.03146696e-01 1.29520774e-01
-8.87169898e-01 4.23022151e-01 3.37021470e-01 -4.18116987e-01
4.78093982e-01 -1.26201391e+00 7.23745167e-01 7.28905380e-01
1.32062003e-01 7.72454202e-01 1.05819523e+00 -7.22981870e-01
-1.20316792e+00 -9.54219818e-01 3.78500521e-01 -9.12413180e-01
6.76097512e-01 -1.82964087e-01 -1.23208809e+00 3.11351836e-01
-3.38731825e-01 3.41616683e-02 4.60122854e-01 2.51635551e-01
-3.09817940e-01 -9.32832807e-02 -9.76267517e-01 8.32526088e-01
6.24673247e-01 -3.44131947e-01 -1.05475023e-01 3.32291037e-01
3.78639609e-01 -3.36377203e-01 -8.09068382e-01 4.94434595e-01
5.33914626e-01 -1.14891148e+00 8.84192228e-01 -9.83462989e-01
-7.07374215e-02 -7.59217441e-01 1.89572424e-01 -1.33734632e+00
-7.77207762e-02 -4.83649850e-01 1.18664987e-01 1.07770634e+00
7.53027499e-01 -7.50516832e-01 4.24807847e-01 9.79686141e-01
-2.19194159e-01 -8.08512390e-01 -7.22215295e-01 -1.02323568e+00
-1.16699915e-02 -9.63425398e-01 9.97012258e-01 1.11342013e+00
-2.59474903e-01 8.29748213e-02 1.02675065e-01 5.46259344e-01
4.67693835e-01 -4.56868559e-02 9.90945339e-01 -1.64693677e+00
6.97202682e-02 -7.22991645e-01 -5.76037288e-01 -4.76362467e-01
-4.16572422e-01 -7.39283681e-01 -4.38025802e-01 -1.33721185e+00
-8.73259231e-02 -8.48848403e-01 -2.97612101e-02 4.49530423e-01
-9.63589251e-02 2.03000531e-01 2.15239406e-01 6.22193396e-01
-3.06876600e-01 3.92697126e-01 1.08889091e+00 -1.26623854e-01
-2.33276695e-01 -1.17684910e-02 -4.49297041e-01 8.22316289e-01
6.89267218e-01 -5.34520864e-01 -3.10604125e-01 -2.82813907e-01
5.03270268e-01 -1.19305067e-01 1.03531241e+00 -1.27252817e+00
-2.63421759e-02 -2.95712173e-01 3.93581241e-01 -6.04418099e-01
9.95997712e-02 -6.67272091e-01 5.49920738e-01 4.82919008e-01
7.43246898e-02 5.91977835e-01 6.32704973e-01 3.10141325e-01
-1.39136642e-01 -1.85434207e-01 6.10665858e-01 2.19747052e-01
-3.85563463e-01 1.12918891e-01 -3.26960325e-01 -1.85457975e-01
9.52367783e-01 -2.23311633e-01 -6.21425152e-01 -3.31941426e-01
-6.50406420e-01 5.42860389e-01 1.02007306e+00 2.58297056e-01
3.36769879e-01 -1.11360228e+00 -4.78830457e-01 4.51145411e-01
3.55065763e-01 -1.20924246e-02 4.27011907e-01 1.00411081e+00
-7.50312686e-01 3.46282393e-01 -6.67977184e-02 -1.20445585e+00
-1.41926491e+00 3.16742092e-01 5.95152378e-01 -2.58834928e-01
-4.88326013e-01 2.66142726e-01 2.60621458e-01 -5.80624878e-01
-5.10114841e-02 -7.19503343e-01 -1.43712789e-01 -5.70193827e-02
4.40697640e-01 6.57746732e-01 4.70719844e-01 -4.29406315e-01
-4.69967812e-01 7.29215145e-01 1.80667073e-01 -1.40361367e-02
9.96419907e-01 -2.50721853e-02 1.34391367e-01 5.59381962e-01
6.80747390e-01 -9.60723683e-02 -1.56677842e+00 3.86307567e-01
5.79839833e-02 -5.06434977e-01 -3.97126526e-02 -1.02633476e+00
-7.70577908e-01 1.22691214e+00 8.69758427e-01 5.02005637e-01
7.99102604e-01 -5.49216196e-02 3.78871635e-02 4.10003752e-01
-1.05152873e-03 -8.03124011e-01 -8.17942619e-02 3.06600958e-01
1.19415915e+00 -1.25980449e+00 1.72605842e-01 -1.37709409e-01
-6.64002717e-01 1.16271043e+00 7.13152468e-01 3.62160295e-01
6.75696671e-01 2.49061301e-01 3.12947541e-01 -4.53612268e-01
-7.79474020e-01 3.61352116e-02 3.47247809e-01 4.75444168e-01
1.32317439e-01 1.65955722e-02 1.75294578e-02 3.84375095e-01
-4.08571869e-01 -4.00047451e-01 3.95593166e-01 8.34399462e-01
-2.48751864e-01 -1.01212692e+00 -7.51335382e-01 4.81891215e-01
-1.00452472e-02 1.50348559e-01 -5.38875937e-01 1.11722839e+00
-2.82234818e-01 7.58118510e-01 2.52198726e-01 -6.31381929e-01
4.77144659e-01 -6.44425536e-03 -1.43202201e-01 -3.32839161e-01
-4.41853285e-01 -2.93289870e-01 1.01547264e-01 -4.91915584e-01
-2.47851759e-02 -5.36183655e-01 -1.06501865e+00 -5.47024667e-01
-4.96010035e-01 3.55765671e-01 9.03202891e-01 9.04987276e-01
5.82653463e-01 5.02300501e-01 5.51748276e-01 -9.69277740e-01
-4.24899608e-01 -8.88213813e-01 -2.52535760e-01 3.35843951e-01
4.02203530e-01 -1.19152510e+00 -6.98830605e-01 -2.91093141e-01] | [8.005343437194824, 4.600368499755859] |
c403333c-c5b8-41ea-94af-e0a49cc1d1d8 | segmentation-of-photovoltaic-module-cells-in | 1806.06530 | null | https://arxiv.org/abs/1806.06530v4 | https://arxiv.org/pdf/1806.06530v4.pdf | Segmentation of Photovoltaic Module Cells in Uncalibrated Electroluminescence Images | High resolution electroluminescence (EL) images captured in the infrared spectrum allow to visually and non-destructively inspect the quality of photovoltaic (PV) modules. Currently, however, such a visual inspection requires trained experts to discern different kinds of defects, which is time-consuming and expensive. Automated segmentation of cells is therefore a key step in automating the visual inspection workflow. In this work, we propose a robust automated segmentation method for extraction of individual solar cells from EL images of PV modules. This enables controlled studies on large amounts of data to understanding the effects of module degradation over time-a process not yet fully understood. The proposed method infers in several steps a high-level solar module representation from low-level edge features. An important step in the algorithm is to formulate the segmentation problem in terms of lens calibration by exploiting the plumbline constraint. We evaluate our method on a dataset of various solar modules types containing a total of 408 solar cells with various defects. Our method robustly solves this task with a median weighted Jaccard index of 94.47% and an $F_1$ score of 97.62%, both indicating a very high similarity between automatically segmented and ground truth solar cell masks. | ['Christian Riess', 'Florian Gallwitz', 'Ansgar Steland', 'Claudia Buerhop-Lutz', 'Sergiu Deitsch', 'Andreas Maier', 'Evgenii Sovetkin'] | 2018-06-18 | null | null | null | null | ['solar-cell-segmentation'] | ['computer-vision'] | [ 6.65234149e-01 -7.96622112e-02 3.68140340e-01 -8.46865177e-02
-5.09932816e-01 -1.12221003e+00 2.58290619e-01 2.60418683e-01
-7.34769851e-02 8.24783504e-01 -4.89894420e-01 -1.93895936e-01
-1.48153141e-01 -8.64206970e-01 -5.69139242e-01 -1.07458103e+00
4.55706418e-01 2.99868524e-01 2.33764112e-01 3.26205492e-01
5.47346950e-01 7.31923997e-01 -1.87954473e+00 1.97761003e-02
1.61720157e+00 1.07444060e+00 6.02273583e-01 8.17827046e-01
6.58977404e-02 1.78827569e-01 -6.81606114e-01 -2.20043227e-01
2.00438842e-01 -5.12903988e-01 -3.12205493e-01 5.06174803e-01
5.14245749e-01 2.03475714e-01 6.48337126e-01 1.28504336e+00
3.94923776e-01 -5.34082770e-01 1.04428256e+00 -9.19885039e-01
-1.98493734e-01 -1.98870912e-01 -3.67969483e-01 -1.44580156e-01
3.96612972e-01 4.07471180e-01 7.69734383e-01 -5.57603538e-01
5.42935908e-01 4.23057705e-01 7.53205061e-01 -3.89275365e-02
-1.34953439e+00 -3.37960809e-01 -4.50577617e-01 2.68644784e-02
-1.27690125e+00 -3.40557873e-01 7.34876812e-01 -9.22258675e-01
7.67572045e-01 4.02869374e-01 1.04982162e+00 3.24846834e-01
3.40436757e-01 4.52504260e-03 1.68270123e+00 -6.16869271e-01
4.95998293e-01 3.56187016e-01 -9.62945893e-02 7.68106163e-01
8.29994798e-01 -1.12996958e-01 -1.34049311e-01 1.92405339e-02
4.77322936e-01 -1.72546461e-01 -5.56509793e-01 -4.99681026e-01
-6.40983284e-01 4.33785647e-01 2.84344047e-01 4.57474858e-01
-3.36619318e-01 -3.67994338e-01 -1.30004361e-01 1.48244262e-01
2.66910821e-01 5.06428719e-01 -2.20001370e-01 1.60867855e-01
-9.66983020e-01 -4.13091272e-01 7.75715649e-01 7.11180806e-01
9.00666654e-01 -2.94845521e-01 -1.42847657e-01 6.69865251e-01
4.04893130e-01 6.27028048e-01 6.81028068e-02 -8.88425112e-01
-1.39371246e-01 1.23908770e+00 2.33126849e-01 -6.36638820e-01
-1.67081371e-01 -3.16990197e-01 -8.18142831e-01 6.71703100e-01
3.61977726e-01 -5.90553954e-02 -1.00831389e+00 1.05123317e+00
2.83213675e-01 -3.01249683e-01 -8.93417671e-02 6.21692598e-01
5.06531715e-01 3.89662117e-01 -1.83571279e-01 -6.02537632e-01
1.34573364e+00 -3.77276391e-01 -5.10299504e-01 -1.78540871e-02
3.19251984e-01 -7.84901440e-01 8.22458625e-01 4.53982860e-01
-9.92342591e-01 -2.60137439e-01 -1.31815922e+00 1.95527613e-01
-4.65529859e-01 7.19079435e-01 4.74413812e-01 1.11126339e+00
-1.04287589e+00 4.24871892e-01 -5.28464317e-01 -6.19561672e-01
6.48061633e-01 5.14460266e-01 -2.90615231e-01 -6.38361275e-02
-1.20405637e-01 6.94950521e-01 2.62866944e-01 1.14619710e-01
-5.59879363e-01 -5.63725352e-01 -5.51611960e-01 4.81464155e-02
3.14034551e-01 -6.66694701e-01 6.33686662e-01 -8.24772537e-01
-1.47289491e+00 1.25241542e+00 -4.26567644e-01 -2.51603395e-01
4.76429820e-01 3.83081615e-01 -1.94369197e-01 2.15925962e-01
-5.44499308e-02 2.41507098e-01 7.11820066e-01 -1.70850766e+00
-5.73870957e-01 -5.53331017e-01 -2.99653977e-01 -1.13569796e-01
-1.90340415e-01 -2.40337402e-01 -2.51839072e-01 -1.01629548e-01
1.04262777e-01 -7.50761330e-01 1.50905147e-01 1.36705548e-01
-4.02017266e-01 -1.27170816e-01 5.14000237e-01 -8.75165582e-01
9.10761774e-01 -1.96729696e+00 -5.42874262e-02 3.75387549e-01
-1.92725006e-02 2.19834551e-01 2.60748833e-01 2.90370792e-01
1.10617295e-01 -1.11109875e-01 -7.84281731e-01 -1.24452442e-01
-1.29864454e-01 -6.22052625e-02 2.61375546e-01 6.03595436e-01
2.01662779e-01 7.91736543e-01 -5.93341291e-01 -3.78138602e-01
4.07626599e-01 4.24120218e-01 -1.55931726e-01 3.01029921e-01
-3.33895355e-01 3.78596961e-01 -4.42090482e-02 1.14571822e+00
8.77662480e-01 2.65628174e-02 2.28012457e-01 -4.72082883e-01
-6.29307508e-01 -2.67555892e-01 -1.02508867e+00 1.50857544e+00
-1.53274968e-01 7.30737805e-01 1.61046565e-01 -9.12997723e-01
1.00628150e+00 -1.40508940e-03 4.71846014e-01 -7.32381105e-01
2.46211544e-01 3.69291395e-01 -3.17866236e-01 -6.36347950e-01
2.44011685e-01 -5.48904724e-02 2.28602543e-01 3.43184806e-02
2.15421338e-02 -5.46289206e-01 5.20574570e-01 -2.71777719e-01
9.36863244e-01 3.01638711e-02 3.61637294e-01 -5.45400500e-01
5.96603990e-01 1.08397119e-01 5.86725354e-01 4.85914320e-01
1.41146824e-01 5.62592924e-01 4.08564389e-01 1.63765550e-01
-1.02145743e+00 -9.24770892e-01 -3.32830012e-01 4.52325828e-02
1.47058621e-01 7.32666180e-02 -1.09799635e+00 -2.22328618e-01
6.13076314e-02 5.34233272e-01 -2.98413992e-01 1.20417736e-01
1.10864952e-01 -1.05328953e+00 1.75932236e-02 1.15156390e-01
5.69846094e-01 -8.04727256e-01 -9.18445706e-01 -8.19437876e-02
2.05499530e-01 -9.81912494e-01 2.29275286e-01 1.09390154e-01
-9.05998051e-01 -1.62468243e+00 -6.36077583e-01 -7.86802292e-01
1.11537886e+00 6.12479933e-02 1.28486073e+00 6.22377470e-02
-1.07325709e+00 5.05017757e-01 -1.88764229e-01 -3.74174863e-01
-5.06325901e-01 -2.67595023e-01 -3.26576263e-01 -8.32212046e-02
2.51008898e-01 -3.95237654e-01 -6.86745465e-01 3.52712780e-01
-8.74119163e-01 -1.25736758e-01 9.15366352e-01 5.68344951e-01
7.99377799e-01 5.36192000e-01 3.30724984e-01 -7.50930905e-01
4.20167655e-01 -8.79216716e-02 -1.37179029e+00 6.22160494e-01
-6.80150867e-01 -2.77908027e-01 3.70108426e-01 1.53106172e-02
-9.80910778e-01 3.64585936e-01 5.13964286e-03 1.61784187e-01
-5.84553599e-01 1.65430203e-01 -6.39747500e-01 -4.98190671e-01
4.18179154e-01 1.69624850e-01 -1.27943859e-01 -4.04158443e-01
1.36186462e-02 7.69846678e-01 5.26812673e-01 -2.25895911e-01
9.00276840e-01 6.68378770e-01 2.58462340e-01 -1.24177027e+00
-4.45899546e-01 -5.49161792e-01 -6.46373153e-01 -3.78710151e-01
9.87695098e-01 -8.88069928e-01 -9.83806252e-01 7.98673034e-01
-8.81190121e-01 -2.38537401e-01 -3.09576809e-01 1.70230828e-02
-2.72966564e-01 5.24787009e-01 -1.44686624e-01 -1.01427805e+00
-2.66801357e-01 -9.70957994e-01 9.67593133e-01 7.67203152e-01
2.59265929e-01 -9.08253074e-01 2.04024073e-02 6.96138561e-01
1.98893145e-01 5.28898537e-01 1.01020646e+00 1.28374025e-01
-8.84986162e-01 -1.80895731e-01 -3.15192908e-01 5.99906445e-01
3.66307884e-01 3.11447173e-01 -1.23584640e+00 -3.70855033e-01
1.21159464e-01 6.94430806e-03 8.93131196e-01 5.53468764e-01
9.42432523e-01 1.93469450e-01 -3.74052167e-01 4.74423349e-01
2.02152705e+00 4.20989007e-01 8.97417665e-01 1.79273393e-02
4.69347179e-01 6.68837786e-01 6.22828007e-01 2.41496772e-01
9.61627625e-03 4.59207386e-01 7.46936917e-01 -2.09782928e-01
-7.45028928e-02 1.90697938e-01 3.18269223e-01 5.06763399e-01
-1.64364949e-02 -3.59887809e-01 -6.84795201e-01 7.05246389e-01
-1.31751633e+00 -7.34860301e-01 -5.84363699e-01 2.37291384e+00
5.14218807e-01 -8.84355456e-02 -4.22106236e-02 3.28586191e-01
7.95590818e-01 -4.73886549e-01 -5.49581647e-01 -5.67115322e-02
-4.94218141e-01 2.46775120e-01 5.28607905e-01 5.56067526e-01
-7.89840817e-01 4.59289283e-01 5.50082922e+00 4.02826697e-01
-9.63024437e-01 -2.88009942e-01 4.05050039e-01 2.84814805e-01
-4.17304516e-01 1.88479617e-01 -5.89902103e-01 8.25812101e-01
3.63734812e-01 8.74496102e-02 3.65691781e-01 1.70646295e-01
1.77998811e-01 -9.14530158e-01 -8.37604463e-01 1.02166796e+00
3.28816414e-01 -1.13898873e+00 -2.69093335e-01 2.39634037e-01
9.57232893e-01 -1.58139527e-01 -3.44547421e-01 -5.77697754e-01
-1.82402834e-01 -9.41181123e-01 2.61626422e-01 1.04970920e+00
8.91850770e-01 -4.35099036e-01 6.27417624e-01 2.66742051e-01
-1.17484474e+00 -2.71629751e-01 -4.28479582e-01 1.85072213e-01
1.51741505e-02 1.24488699e+00 -8.78497899e-01 7.32257426e-01
7.35457420e-01 5.24109840e-01 -7.52027869e-01 1.57668996e+00
-2.85054147e-01 4.43769664e-01 -3.75173688e-01 1.06726736e-01
-5.13759494e-01 -8.84601057e-01 4.79430795e-01 9.52783763e-01
8.21420789e-01 -2.06315309e-01 -3.46991628e-01 1.32949412e+00
-3.40279602e-02 -1.64584264e-01 -6.40538990e-01 -1.96545258e-01
3.55435073e-01 1.52044141e+00 -1.11513448e+00 3.58015560e-02
-1.79193586e-01 1.02072442e+00 -1.95580989e-01 2.95490116e-01
-4.27192211e-01 -3.24293762e-01 4.53632593e-01 6.02934957e-02
5.35634995e-01 -1.04472429e-01 -4.58780199e-01 -8.81746709e-01
3.07033449e-01 -3.89855236e-01 2.00787023e-01 -1.01926339e+00
-1.19484746e+00 1.78543553e-01 -4.37078297e-01 -1.01505530e+00
2.17837319e-01 -8.21163595e-01 -7.91340709e-01 9.48538303e-01
-1.77807903e+00 -1.02822566e+00 -8.65209877e-01 2.94453740e-01
2.40271360e-01 1.41597748e-01 7.64317572e-01 7.19456840e-03
-6.21286988e-01 -9.78658870e-02 4.25216019e-01 -3.49428058e-01
3.83910447e-01 -1.43862474e+00 -2.29152843e-01 1.02764285e+00
-3.12210117e-02 1.59862891e-01 7.29038656e-01 -6.69947267e-01
-1.39814281e+00 -7.53567934e-01 6.61285043e-01 -3.99098724e-01
2.44214356e-01 -1.61687613e-01 -7.40806162e-01 8.83689150e-02
5.51759005e-01 -3.49224180e-01 7.58319080e-01 -4.99226391e-01
3.37340176e-01 -2.52038002e-01 -1.49476779e+00 1.95092097e-01
7.38735616e-01 -5.19438267e-01 -2.48908713e-01 3.76110226e-01
-1.78244025e-01 -6.50078524e-03 -1.17772341e+00 6.76607788e-01
3.95654112e-01 -1.39653754e+00 8.08513761e-01 3.32283497e-01
3.64175327e-02 -7.85977244e-01 1.57062799e-01 -1.09565890e+00
1.06925100e-01 -4.07773912e-01 2.75461823e-01 1.57208276e+00
2.57661968e-01 -7.13950515e-01 7.41125286e-01 3.37759703e-01
-2.19239846e-01 -4.81350660e-01 -6.82011485e-01 -5.29881835e-01
-4.93846714e-01 -7.99301341e-02 1.36779040e-01 7.15642750e-01
-5.79070210e-01 -7.38309622e-02 3.93771231e-01 5.95671892e-01
9.43280876e-01 5.01099706e-01 5.57135224e-01 -1.85143459e+00
6.71838373e-02 -4.87680852e-01 -4.63945001e-01 -3.10857356e-01
1.07107632e-01 -7.27667332e-01 1.88578833e-02 -1.90747929e+00
4.43470001e-01 -2.98748404e-01 3.70051106e-03 1.54574931e-01
-8.83103162e-02 5.21313548e-01 -2.01288104e-01 -1.38402442e-02
-2.73752451e-01 3.42091918e-01 8.23623300e-01 -3.20410430e-01
-1.42531946e-01 7.65264258e-02 -5.70141315e-01 5.57921886e-01
8.06118309e-01 -2.75311828e-01 -2.22167835e-01 -1.46039799e-01
5.43381453e-01 -4.56319511e-01 6.77540064e-01 -1.35189724e+00
3.14404428e-01 1.06833771e-01 5.18628418e-01 -6.28157496e-01
2.04539761e-01 -1.13632536e+00 4.40899581e-01 5.14849603e-01
5.11079252e-01 -5.08551717e-01 1.49936885e-01 4.23083246e-01
-4.62402962e-02 -6.92894876e-01 7.65892088e-01 -1.29421324e-01
-6.22860849e-01 -1.84608757e-01 -3.06668550e-01 -3.15804511e-01
1.34482956e+00 -7.64029682e-01 -4.78830963e-01 1.85212687e-01
-2.52842039e-01 1.03429109e-01 1.30135691e+00 -1.68690279e-01
4.73170966e-01 -8.85946751e-01 -3.07289183e-01 4.24779058e-01
3.74047130e-01 1.25629023e-01 3.31799388e-01 8.69893610e-01
-7.10825622e-01 1.78539410e-01 -2.98871756e-01 -1.04604459e+00
-1.72408402e+00 6.90872688e-03 5.13526142e-01 7.18554407e-02
-2.22430319e-01 7.32701898e-01 -2.70443916e-01 -5.52098937e-02
-1.21939272e-01 -5.36318481e-01 -4.42677855e-01 2.78274596e-01
-1.59368411e-01 5.37613392e-01 3.44761968e-01 -5.39114892e-01
-1.30621508e-01 1.15271652e+00 5.68942070e-01 4.16168213e-01
1.30977178e+00 -8.00562128e-02 -6.68686509e-01 3.84078503e-01
7.18710721e-01 2.32444391e-01 -1.35082495e+00 4.39688474e-01
1.40944615e-01 -5.46685755e-01 5.97771164e-03 -1.07184863e+00
-8.89600575e-01 6.42099679e-01 9.99844730e-01 4.95902658e-01
1.54592776e+00 -7.21148849e-02 2.15287983e-01 2.26511061e-01
3.05422187e-01 -1.40671122e+00 -3.00441772e-01 -1.85994551e-01
6.51138783e-01 -1.36432171e+00 2.13031799e-01 -7.65379310e-01
-2.55635798e-01 9.52949107e-01 2.68391371e-01 2.20625222e-01
2.14309141e-01 3.49170744e-01 6.02579638e-02 -4.35776621e-01
-3.50469887e-01 -4.05914307e-01 4.16873813e-01 7.02701986e-01
3.11470896e-01 -8.82887989e-02 -4.00022447e-01 1.33029521e-01
1.73606530e-01 -5.38795069e-02 4.14213240e-01 1.00384140e+00
-6.30065382e-01 -1.07017994e+00 -5.00564337e-01 6.42553449e-01
-1.32071644e-01 2.89775938e-01 -6.64573610e-01 3.94616157e-01
4.31562752e-01 1.14023817e+00 -9.11108330e-02 1.24967575e-01
4.03695464e-01 8.60350952e-02 6.94913447e-01 -3.14109445e-01
-3.69615346e-01 2.20209658e-02 -1.37491912e-01 -2.54243821e-01
-8.27239454e-01 -6.43855751e-01 -8.54922295e-01 2.91763544e-01
-5.93343079e-01 4.33368683e-02 1.21183157e+00 7.99393892e-01
2.42978230e-01 5.52555442e-01 7.09136844e-01 -7.35376894e-01
9.19967517e-02 -7.65857637e-01 -7.12559223e-01 4.67830658e-01
3.03383712e-02 -4.98756945e-01 -4.67029184e-01 2.04496458e-01] | [7.2179341316223145, 1.8986417055130005] |
372cfb59-f032-4efb-a41c-13bf4e8b01e5 | the-clrs-algorithmic-reasoning-benchmark | 2205.15659 | null | https://arxiv.org/abs/2205.15659v2 | https://arxiv.org/pdf/2205.15659v2.pdf | The CLRS Algorithmic Reasoning Benchmark | Learning representations of algorithms is an emerging area of machine learning, seeking to bridge concepts from neural networks with classical algorithms. Several important works have investigated whether neural networks can effectively reason like algorithms, typically by learning to execute them. The common trend in the area, however, is to generate targeted kinds of algorithmic data to evaluate specific hypotheses, making results hard to transfer across publications, and increasing the barrier of entry. To consolidate progress and work towards unified evaluation, we propose the CLRS Algorithmic Reasoning Benchmark, covering classical algorithms from the Introduction to Algorithms textbook. Our benchmark spans a variety of algorithmic reasoning procedures, including sorting, searching, dynamic programming, graph algorithms, string algorithms and geometric algorithms. We perform extensive experiments to demonstrate how several popular algorithmic reasoning baselines perform on these tasks, and consequently, highlight links to several open challenges. Our library is readily available at https://github.com/deepmind/clrs. | ['Charles Blundell', 'Raia Hadsell', 'Misha Dashevskiy', 'Andrea Banino', 'Razvan Pascanu', 'David Budden', 'Adrià Puigdomènech Badia', 'Petar Veličković'] | 2022-05-31 | null | null | null | null | ['learning-to-execute'] | ['computer-code'] | [ 3.74725342e-01 3.09825152e-01 -6.68212414e-01 -2.20038116e-01
-5.07585585e-01 -8.16158593e-01 6.30207717e-01 5.85721910e-01
-3.37893993e-01 4.54106510e-01 2.73886502e-01 -8.01951289e-01
-4.88074541e-01 -1.12049270e+00 -1.03804004e+00 -2.77131617e-01
5.81446514e-02 7.91576922e-01 -5.28044440e-03 -2.54829060e-02
5.82649648e-01 7.17819214e-01 -1.62720156e+00 6.41603887e-01
5.35194695e-01 8.64610553e-01 -4.45344776e-01 8.92940402e-01
-4.50797439e-01 1.12456799e+00 -4.09973949e-01 -8.40564787e-01
1.74620494e-01 -7.24670291e-02 -1.20150721e+00 -6.21414840e-01
6.94191396e-01 -1.15707919e-01 -6.02199972e-01 9.86232400e-01
1.75317258e-01 2.13248760e-01 6.93064392e-01 -1.45633543e+00
-7.54900992e-01 1.24341428e+00 -3.26185107e-01 2.71202981e-01
4.73744214e-01 2.10176811e-01 1.60561037e+00 -4.70190257e-01
6.79407537e-01 1.31924665e+00 9.94514048e-01 7.26927757e-01
-1.29122710e+00 -4.58957076e-01 1.21743530e-01 4.67014581e-01
-1.04754794e+00 -3.32699299e-01 7.19199479e-01 -2.21430421e-01
1.11732078e+00 4.46492314e-01 5.37613571e-01 1.02718091e+00
-1.95418954e-01 1.36807477e+00 6.80136204e-01 -4.84351516e-01
2.15209037e-01 -4.50271219e-01 6.38036728e-01 1.06399632e+00
6.10792160e-01 9.82636735e-02 -2.56606102e-01 -4.88520563e-02
5.67081451e-01 8.53262842e-02 -1.86932951e-01 -7.13281512e-01
-1.11167705e+00 8.74107122e-01 6.27335250e-01 2.47497261e-01
1.10799551e-01 5.30376136e-01 6.36621237e-01 3.62271369e-01
-1.24404050e-01 9.59859252e-01 -5.80549538e-01 -3.14510614e-01
-7.07467794e-01 5.92654884e-01 1.17211437e+00 9.81938720e-01
6.82871222e-01 -2.06055731e-01 -1.84612036e-01 6.09042585e-01
-1.18519954e-01 6.54264241e-02 2.06734359e-01 -9.99021113e-01
6.48770213e-01 9.64722812e-01 -3.53235036e-01 -1.10369170e+00
-6.36226952e-01 -2.36656919e-01 -8.22175205e-01 -1.92869008e-01
6.84029937e-01 -4.40920740e-02 -7.87618518e-01 1.60839379e+00
-9.82617661e-02 -2.59369100e-03 7.79989362e-02 6.88489318e-01
1.13751650e+00 5.43861747e-01 -7.35259205e-02 2.40431339e-01
1.01199079e+00 -1.35813153e+00 -1.64608881e-01 -1.67243093e-01
1.14180791e+00 -4.93702143e-01 1.06786561e+00 5.91782272e-01
-1.44347441e+00 -4.33046609e-01 -1.02911615e+00 -2.85726011e-01
-8.07082176e-01 -1.24076359e-01 1.28240001e+00 4.64533865e-01
-1.13633358e+00 8.23210359e-01 -7.84171522e-01 -3.14568758e-01
8.10647666e-01 4.14307714e-01 1.06435664e-01 -2.95163661e-01
-1.03230190e+00 8.50653410e-01 7.42773592e-01 -2.07392901e-01
-3.57742876e-01 -1.08615887e+00 -6.59060657e-01 2.93123096e-01
6.46897614e-01 -9.67387259e-01 1.67296338e+00 -8.40048194e-01
-1.17427707e+00 1.09505987e+00 1.24942683e-01 -9.75123167e-01
6.08904302e-01 -2.31902599e-01 -9.80890691e-02 -2.85513536e-03
-5.14509797e-01 6.41417146e-01 1.93073273e-01 -9.52781022e-01
-5.14732361e-01 -1.90536022e-01 4.62708563e-01 -1.68504361e-02
-1.57765850e-01 -1.60982877e-01 -4.42473412e-01 -5.23690045e-01
3.77373919e-02 -8.09267879e-01 -2.66350538e-01 -1.41087454e-02
-6.18922830e-01 -6.85715377e-01 3.60163838e-01 -2.09651198e-02
1.05737817e+00 -1.81843281e+00 4.08777028e-01 1.53252199e-01
4.26248133e-01 2.08552167e-01 -3.18990052e-02 5.25566697e-01
-1.77939400e-01 3.06971371e-01 -1.43793851e-01 2.03241810e-01
4.82289761e-01 7.08734393e-02 -5.75653136e-01 2.19541818e-01
1.93876162e-01 1.38222885e+00 -7.51562715e-01 -3.88749629e-01
1.18209258e-01 1.98242869e-02 -8.45347047e-01 1.64793339e-02
-9.66649890e-01 -2.78482169e-01 -4.28802222e-01 7.62121499e-01
4.59134400e-01 -5.52365124e-01 2.40205944e-01 -8.61272514e-02
1.25178888e-01 5.69075882e-01 -1.14103520e+00 1.85471272e+00
-3.06946933e-01 6.89925194e-01 -2.33910933e-01 -1.43444788e+00
6.23491406e-01 -2.10444927e-01 2.00066090e-01 -6.65160835e-01
3.31491321e-01 2.55782958e-02 2.22413361e-01 -3.47498655e-01
4.67968196e-01 4.12025601e-01 -3.25305611e-02 6.60566092e-01
-1.30344748e-01 -3.43059093e-01 5.26865005e-01 2.18774214e-01
1.47408974e+00 8.12117532e-02 2.44652882e-01 -1.54320925e-01
5.17922580e-01 4.16735590e-01 1.98086441e-01 1.12328207e+00
1.94235276e-02 2.60901839e-01 6.13076925e-01 -1.00875664e+00
-8.07053745e-01 -1.19921267e+00 1.08963452e-01 1.32541120e+00
-4.62362133e-02 -5.98771453e-01 -6.89455748e-01 -6.13389730e-01
2.10426331e-01 7.54251361e-01 -6.76219106e-01 -2.30950519e-01
-8.09749782e-01 -5.44791102e-01 8.90068948e-01 8.86505425e-01
4.15059716e-01 -1.24378049e+00 -7.10783780e-01 -1.65863857e-01
8.69524926e-02 -6.98343217e-01 1.68093473e-01 4.63790417e-01
-1.13100302e+00 -1.47871637e+00 -1.95706815e-01 -8.43651891e-01
5.99904895e-01 1.69162303e-01 1.78307796e+00 7.72499800e-01
-5.98746061e-01 6.38854265e-01 -1.09259933e-01 -5.73860407e-01
-4.74055111e-01 4.87021446e-01 -3.24497461e-01 -7.69528925e-01
6.39012158e-01 -4.00239736e-01 -4.43878740e-01 -4.70671207e-02
-9.03505325e-01 3.33289862e-01 7.56846189e-01 6.30139887e-01
3.38421196e-01 1.36338249e-01 1.15048639e-01 -1.20242357e+00
5.20096898e-01 -4.12910223e-01 -7.07961261e-01 3.93550903e-01
-6.74407899e-01 3.78111511e-01 9.52569127e-01 -2.13244915e-01
-5.61299086e-01 -3.55047196e-01 4.08482142e-02 -2.61175096e-01
-4.49085087e-01 6.11277103e-01 1.15536541e-01 1.39357850e-01
7.94382989e-01 1.13688640e-01 -1.58746228e-01 -1.81110620e-01
8.03084135e-01 4.40169126e-02 5.50174296e-01 -1.20799220e+00
1.03575861e+00 2.89770603e-01 1.74814671e-01 -3.49742442e-01
-1.01206994e+00 -2.00913474e-01 -4.64248568e-01 1.72568619e-01
4.40395325e-01 -4.07166302e-01 -1.01430082e+00 1.78082734e-01
-1.05277705e+00 -7.31871605e-01 -2.49947801e-01 1.12931401e-01
-8.32292438e-01 -1.39115611e-02 -7.31196761e-01 -4.41062212e-01
-4.02983546e-01 -1.14658439e+00 6.49480104e-01 3.64354104e-01
-4.91212666e-01 -1.32883584e+00 1.22604445e-02 3.50736201e-01
4.13566798e-01 1.28600478e-01 1.49415159e+00 -1.07765877e+00
-9.08565342e-01 -1.25713542e-01 -4.81028855e-01 -8.95261094e-02
-3.30896288e-01 1.61703676e-01 -7.76668966e-01 -1.60201862e-01
-4.65570629e-01 -8.46432805e-01 1.20100927e+00 3.19201142e-01
1.98423946e+00 -2.00262010e-01 -5.98689198e-01 7.69219637e-01
1.39447534e+00 -1.47188604e-02 5.54611444e-01 5.01816332e-01
6.58978581e-01 3.87493253e-01 1.93070576e-01 2.54647508e-02
3.13503474e-01 2.24226132e-01 4.22424912e-01 1.58784181e-01
-5.30607738e-02 -2.96401978e-01 -9.53419730e-02 4.61933345e-01
-1.28891498e-01 -3.50672513e-01 -1.48519421e+00 3.73948038e-01
-1.82434118e+00 -1.04313159e+00 6.03245664e-03 1.91170490e+00
8.67317021e-01 5.06001592e-01 2.21930534e-01 4.47167724e-01
5.88964164e-01 2.26845786e-01 -6.64840937e-01 -7.81587124e-01
1.33498102e-01 5.61795413e-01 4.65497434e-01 5.37963770e-02
-1.07635951e+00 1.01176119e+00 6.45017099e+00 6.65180624e-01
-9.77660835e-01 -5.76499462e-01 7.00958908e-01 -1.57775834e-01
-3.62023115e-01 -1.68211341e-01 -6.75190151e-01 -1.10721029e-01
1.23797297e+00 -2.87001014e-01 8.17445397e-01 1.05940878e+00
-4.77040797e-01 2.36335918e-01 -1.81324947e+00 8.94217730e-01
7.96989202e-02 -1.98825943e+00 2.55766809e-01 -4.25656646e-01
6.14307046e-01 5.32676391e-02 2.92373061e-01 7.22174823e-01
8.06382656e-01 -1.44285369e+00 4.04262424e-01 3.84403914e-01
2.88109601e-01 -7.46659100e-01 4.27279979e-01 5.20716719e-02
-1.03311932e+00 -2.72376895e-01 -2.49349773e-01 -3.46897900e-01
-5.32204390e-01 1.49987444e-01 -8.72702837e-01 4.47308153e-01
7.09634483e-01 7.90390193e-01 -8.42502654e-01 1.17968416e+00
-2.43770152e-01 5.57878137e-01 -1.06915966e-01 -5.48065007e-01
3.56417000e-01 3.90479304e-02 2.02777520e-01 1.27810538e+00
-1.52519122e-01 -4.95668910e-02 1.28037587e-01 1.15132964e+00
-5.67861915e-01 6.84153149e-03 -6.81189120e-01 -4.63507950e-01
4.35733914e-01 1.13616478e+00 -1.01629698e+00 -3.06051463e-01
-5.48990965e-01 4.00551528e-01 7.01306462e-01 1.49959639e-01
-1.02619159e+00 -4.65387613e-01 5.92052758e-01 -9.74191651e-02
1.69431910e-01 -1.25635564e-01 -6.36961579e-01 -8.57642233e-01
-1.01334862e-01 -1.23198807e+00 9.67853725e-01 -6.55834377e-01
-1.19433093e+00 8.49581324e-03 -4.96691838e-02 -5.97566366e-01
-4.85837571e-02 -1.17062795e+00 -6.85276210e-01 2.31854647e-01
-1.40568674e+00 -8.30245972e-01 -4.11591947e-01 3.20945889e-01
6.19008183e-01 -2.34341830e-01 8.23010981e-01 -4.90798466e-02
-7.25366950e-01 7.91494191e-01 3.17547843e-02 4.11169201e-01
3.51147234e-01 -1.42127848e+00 5.87942123e-01 4.42981750e-01
5.57905316e-01 8.41625333e-01 5.99447012e-01 -9.81900096e-02
-2.12688112e+00 -9.69051003e-01 3.05407524e-01 -6.57188594e-01
1.04116130e+00 -1.78979814e-01 -8.91722977e-01 1.03016722e+00
8.37060213e-02 -1.05577350e-01 6.21768296e-01 6.45864129e-01
-8.34761918e-01 1.06619582e-01 -8.84478748e-01 1.08027506e+00
1.37261999e+00 -2.44436324e-01 -8.50126088e-01 4.72720295e-01
7.89942205e-01 -7.12318957e-01 -7.73771584e-01 4.65817034e-01
6.23396039e-01 -1.12517750e+00 1.13066423e+00 -1.10676837e+00
9.83864546e-01 1.50829941e-01 -7.47940466e-02 -1.11275613e+00
-1.85478047e-01 -6.02792501e-01 -5.08826196e-01 9.70155597e-01
5.83031118e-01 -5.20026743e-01 1.20372617e+00 5.61199844e-01
-1.21931806e-01 -1.11489534e+00 -2.33029142e-01 -6.55684173e-01
6.05229616e-01 -5.96849620e-01 6.88074112e-01 8.72279048e-01
3.06263380e-02 2.07803905e-01 4.10248995e-01 -7.22920299e-02
3.76734763e-01 6.16146147e-01 1.00223601e+00 -1.52155101e+00
-3.84843737e-01 -1.43124151e+00 -3.12469274e-01 -9.91748214e-01
4.97753590e-01 -1.44426692e+00 -2.39815563e-01 -1.75292909e+00
1.88110128e-01 -3.54193360e-01 -4.17386144e-01 4.73975450e-01
9.45354998e-02 -8.90880972e-02 -1.54945273e-02 -9.08732787e-02
-9.61462021e-01 3.65904085e-02 1.03130472e+00 -5.03763378e-01
4.31110114e-02 8.62132013e-03 -9.30287719e-01 9.71446872e-01
1.10047805e+00 -1.61451027e-01 -4.99212414e-01 -6.63305163e-01
6.68861449e-01 -2.60425955e-01 3.36884022e-01 -1.27510250e+00
4.55608249e-01 -3.83700758e-01 3.77168030e-01 -5.57313979e-01
-2.98294313e-02 -5.70876598e-01 -3.46686631e-01 7.87907898e-01
-9.61737216e-01 4.35051501e-01 5.24294317e-01 4.59908426e-01
2.09599547e-03 -3.34699869e-01 6.78638577e-01 -3.20779026e-01
-7.68854558e-01 3.91224444e-01 -5.69825023e-02 7.03592420e-01
8.25514853e-01 -1.85131524e-02 -8.68989646e-01 -1.48678273e-01
-3.24464500e-01 3.33320081e-01 5.28864264e-01 4.68350768e-01
4.41843867e-01 -9.41309690e-01 -3.45806569e-01 -5.06699421e-02
1.79842293e-01 2.07077965e-01 -5.75301312e-02 6.53020442e-01
-9.96252656e-01 9.19404924e-01 -1.66792795e-01 -2.48110399e-01
-1.09220672e+00 9.97952759e-01 3.74945998e-01 -3.14534128e-01
-6.63264871e-01 9.55381513e-01 -1.82731062e-01 -6.90254152e-01
8.07997525e-01 -6.47800267e-01 9.96931344e-02 -2.66875565e-01
4.67903942e-01 4.89202797e-01 1.30245360e-02 2.35077992e-01
-2.83054620e-01 3.05618286e-01 -3.15827578e-01 5.17990053e-01
1.51706946e+00 6.55856073e-01 -2.36461967e-01 4.15316224e-01
1.01268423e+00 -2.41729349e-01 -7.21358120e-01 -3.84005904e-01
4.99113709e-01 -1.74393710e-02 -1.85907990e-01 -8.44004929e-01
-9.99793291e-01 9.55843627e-01 1.16825007e-01 5.01167953e-01
9.85021412e-01 -6.78494871e-02 6.00846827e-01 1.14910698e+00
1.92334384e-01 -9.66112792e-01 2.06081957e-01 8.32143664e-01
6.64410233e-01 -1.11472654e+00 1.60202727e-01 -3.17633539e-01
-8.24638754e-02 1.59426141e+00 8.96628201e-01 -2.75055796e-01
4.07195926e-01 5.48046708e-01 -2.95287162e-01 -3.98737669e-01
-1.07432365e+00 -9.61779878e-02 1.97399214e-01 2.96700567e-01
6.46362007e-01 -7.23529980e-02 2.73779519e-02 4.90170240e-01
-6.41750216e-01 2.99206842e-02 4.66148257e-01 8.72616231e-01
-1.84577242e-01 -1.07810605e+00 -1.70610383e-01 7.16681719e-01
-3.29445809e-01 -2.96851844e-01 -7.12420344e-01 1.07284021e+00
-3.21817428e-01 4.85166043e-01 2.69409418e-01 -2.24294245e-01
1.07768387e-01 3.72505575e-01 7.44803667e-01 -4.37686682e-01
-5.06700516e-01 -8.84820819e-01 2.20120400e-01 -5.70649266e-01
-1.96392938e-01 -5.18096030e-01 -1.29187000e+00 -6.87991500e-01
1.36660352e-01 1.30813032e-01 3.38845134e-01 7.56936014e-01
1.85756639e-01 6.55229211e-01 -8.96339044e-02 -6.12781167e-01
-6.25801504e-01 -5.04570842e-01 2.89626569e-02 3.90416503e-01
2.92516612e-02 -6.17489517e-01 -3.76722634e-01 -2.73273349e-01] | [9.158180236816406, 7.157996654510498] |
9858b97b-e6b7-449f-b5bd-ce3eba2e8b5f | invertible-tree-embeddings-using-a | null | null | https://aclanthology.org/2020.coling-main.328 | https://aclanthology.org/2020.coling-main.328.pdf | Invertible Tree Embeddings using a Cryptographic Role Embedding Scheme | We present a novel method for embedding trees in a vector space based on Tensor-Product Representations (TPRs) which allows for inversion: the retrieval of the original tree structure and nodes from the vectorial embedding. Unlike previous attempts, this does not come at the cost of intractable representation size; we utilize a method for non-exact inversion, showing that it works well when there is sufficient randomness in the representation scheme for simple data and providing an upper bound on its error. To handle the huge number of possible tree positions without memoizing position representation vectors, we present a method (Cryptographic Role Embedding) using cryptographic hashing algorithms that allows for the representation of unboundedly many positions. Through experiments on parse tree data, we show a 30,000-dimensional Cryptographic Role Embedding of trees can provide invertibility with error {\textless} 1{\%} that previous methods would require 8.6 {\mbox{$\times$}} 1057 dimensions to represent. | ['Paul Smolensky', 'Coleman Haley'] | 2020-12-01 | null | null | null | coling-2020-8 | ['role-embedding'] | ['graphs'] | [ 4.52651024e-01 2.74926543e-01 -5.67408875e-02 -4.51195948e-02
-1.03141975e+00 -8.67756605e-01 4.33818072e-01 5.73733866e-01
-4.61285442e-01 6.13953114e-01 2.16847315e-01 -9.97360170e-01
-7.74419010e-02 -1.14437306e+00 -6.06626093e-01 -7.09380031e-01
-5.44808805e-01 5.20587444e-01 1.30347878e-01 -4.13316041e-01
4.47925746e-01 3.86265099e-01 -1.53149486e+00 2.04197124e-01
2.66251236e-01 8.41819704e-01 -3.05141270e-01 1.02181053e+00
-3.15765500e-01 6.26880527e-01 -6.78912818e-01 -7.49423623e-01
3.79155666e-01 -5.75163066e-02 -9.14231420e-01 -4.46225017e-01
4.62498605e-01 -7.08458185e-01 -6.55185401e-01 8.22920978e-01
2.04311922e-01 -2.18161657e-01 8.15201759e-01 -1.31143653e+00
-6.28882289e-01 7.09324539e-01 -3.83721888e-01 6.04078360e-02
3.46074253e-01 -3.53185058e-01 1.43915224e+00 -5.94937384e-01
4.45870072e-01 9.74822462e-01 8.25095177e-01 1.38711140e-01
-1.54905105e+00 -4.91739601e-01 -6.15164459e-01 1.41216576e-01
-1.41191089e+00 -3.17818969e-01 6.03319824e-01 -2.32973710e-01
1.00012875e+00 7.03553736e-01 5.52146137e-01 4.81687903e-01
9.00996998e-02 3.89946610e-01 8.15540493e-01 -6.18636072e-01
1.46171242e-01 -8.91634896e-02 4.68686104e-01 9.47789729e-01
8.90815973e-01 3.01012844e-02 -3.08491528e-01 -8.44469607e-01
7.55749464e-01 -7.10725486e-02 -7.30547532e-02 -5.37156820e-01
-1.37544000e+00 1.31561399e+00 3.11659396e-01 3.12966704e-01
1.28103837e-01 7.87379384e-01 5.98884523e-01 5.13035536e-01
7.83583000e-02 2.02662751e-01 -2.28436887e-01 -3.93591255e-01
-6.73145473e-01 4.50845212e-01 9.17406201e-01 7.41439342e-01
8.16085458e-01 -8.72514863e-03 2.52894610e-01 3.54688942e-01
1.76336005e-01 6.18140936e-01 1.60392433e-01 -1.07639074e+00
6.73278809e-01 2.91980118e-01 -3.03818863e-02 -1.08662057e+00
-2.82958634e-02 2.24738166e-01 -6.02954745e-01 -5.36965653e-02
5.59110403e-01 2.61697292e-01 -5.80572546e-01 1.59609866e+00
4.16277140e-01 -2.61815786e-01 1.17161915e-01 2.20140845e-01
4.77103174e-01 9.62457955e-01 -2.16347277e-01 -9.26148593e-02
1.51757181e+00 -2.73961186e-01 -5.18138826e-01 1.42759249e-01
1.15733981e+00 -6.91311836e-01 8.54012609e-01 2.89870590e-01
-1.13611770e+00 -8.62631574e-02 -1.40367484e+00 -3.40415120e-01
-3.63562256e-01 -3.62961292e-01 1.16335702e+00 1.15008855e+00
-1.20218587e+00 5.93381226e-01 -7.18693256e-01 2.17175961e-01
-1.17880553e-01 6.37179673e-01 -7.21931100e-01 7.24701658e-02
-1.13859439e+00 6.67063653e-01 3.73022586e-01 2.71902303e-03
-3.88174742e-01 -4.11796063e-01 -1.04372990e+00 2.81085551e-01
-5.11533618e-02 -2.17935354e-01 9.55481231e-01 1.30204130e-02
-9.23487186e-01 5.66018760e-01 -2.23689258e-01 -7.22252488e-01
-8.70608091e-02 2.10170984e-01 6.42831773e-02 4.78449374e-01
-4.10056114e-02 3.58922780e-01 8.54090691e-01 -9.69002604e-01
-1.58316717e-02 -6.54088974e-01 1.98578224e-01 -1.33634433e-01
-6.02862656e-01 -1.84477180e-01 1.23128004e-01 -7.01174676e-01
6.89683080e-01 -1.10785580e+00 -1.17532670e-01 2.04771414e-01
-2.06399918e-01 -3.00581213e-02 5.86335599e-01 -4.89697784e-01
1.22444355e+00 -2.29388666e+00 1.37495354e-01 4.51177180e-01
5.70303380e-01 1.99937731e-01 2.08458137e-02 8.32981586e-01
-3.07827771e-01 6.09701753e-01 -4.48484182e-01 -2.09013194e-01
4.01518017e-01 5.10224640e-01 -6.96312487e-01 6.47986770e-01
-1.88064292e-01 6.94823682e-01 -5.87897778e-01 -6.08791947e-01
2.23301910e-02 6.65411353e-01 -9.07283843e-01 -2.17132643e-01
5.70823587e-02 -5.75786412e-01 -3.32682610e-01 3.54802340e-01
8.14329803e-01 -1.84474900e-01 5.74062705e-01 -2.08648831e-01
2.14544743e-01 6.46931350e-01 -1.46775746e+00 1.30894911e+00
-2.21455276e-01 5.27368844e-01 2.20125485e-02 -7.53957093e-01
7.82529712e-01 3.36277574e-01 5.96839130e-01 -2.15593055e-01
2.93085575e-02 3.02790314e-01 -2.16607913e-01 2.19470263e-01
7.82021582e-01 -2.79596597e-01 -5.55847347e-01 1.02333164e+00
-3.28324944e-01 -4.16547298e-01 -8.16148799e-03 4.97264773e-01
1.49376762e+00 -2.81936675e-01 -4.39688601e-02 1.56725422e-02
3.77106339e-01 -1.41719609e-01 2.13073730e-01 5.68304718e-01
-1.17509380e-01 4.40126896e-01 8.37744951e-01 -6.49889886e-01
-1.30939090e+00 -1.11052430e+00 -3.07691127e-01 9.98135030e-01
-2.83751488e-01 -1.19168258e+00 -5.79997361e-01 -2.83712864e-01
3.06472421e-01 6.27823889e-01 -7.02263951e-01 -1.30543277e-01
-6.82350159e-01 -7.18726575e-01 1.08067048e+00 5.89395106e-01
1.72797814e-01 -5.90988159e-01 -6.63043380e-01 1.46917015e-01
-1.02609403e-01 -8.90860200e-01 -2.91918486e-01 6.93335652e-01
-1.07248390e+00 -7.59890318e-01 -3.46954107e-01 -5.79872608e-01
6.19970441e-01 3.43045026e-01 7.80176818e-01 2.67800212e-01
-3.04894596e-01 1.40804708e-01 -1.42946437e-01 2.73506761e-01
-4.62361932e-01 -3.54562402e-02 1.86885502e-02 -5.61352134e-01
3.94343406e-01 -8.14642072e-01 -3.71351004e-01 -7.43076652e-02
-1.16755080e+00 -3.48628968e-01 3.30297381e-01 1.08651161e+00
2.76218146e-01 1.32481128e-01 -2.25584477e-01 -8.62327576e-01
4.21767354e-01 -1.57131165e-01 -8.70517612e-01 3.40132490e-02
-5.40429652e-01 8.44903171e-01 6.36595964e-01 -1.46086857e-01
-1.44945800e-01 -4.87840138e-02 -1.51681617e-01 -3.27594876e-01
5.69415450e-01 2.65529037e-01 2.71160901e-01 -3.31023097e-01
6.61134064e-01 5.00109494e-01 2.94732422e-01 -4.19384181e-01
7.53568411e-01 5.46299338e-01 3.20694715e-01 -9.16935384e-01
1.14235556e+00 4.87653881e-01 4.27795172e-01 -6.76326454e-01
-1.84387267e-01 -4.27793153e-02 -5.48171818e-01 6.93763137e-01
4.89030391e-01 -6.65674508e-01 -1.14471447e+00 -9.27242488e-02
-1.25185251e+00 1.63112849e-01 -3.61385226e-01 1.98463947e-01
-6.79422915e-01 8.99589598e-01 -7.03882515e-01 -8.95028412e-01
-4.07112271e-01 -1.13001990e+00 1.12065113e+00 -6.66798234e-01
-2.37152606e-01 -5.92883885e-01 1.92754269e-01 5.08953154e-01
3.42796415e-01 1.70455977e-01 1.57344913e+00 -4.86705273e-01
-8.52204978e-01 -5.45463204e-01 -2.66224533e-01 1.82615191e-01
-1.74608022e-01 6.77676797e-02 -7.22480774e-01 -3.44802022e-01
-1.39955372e-01 -2.54214406e-01 7.64688075e-01 -2.39913210e-01
1.11924732e+00 -6.91046655e-01 -1.34064585e-01 5.26570797e-01
1.34426200e+00 -7.87526593e-02 9.49422956e-01 2.52029061e-01
6.19314730e-01 2.89829791e-01 1.34816915e-01 5.12141228e-01
4.54123974e-01 5.53876638e-01 4.14884061e-01 3.12141836e-01
1.05831780e-01 -4.70315784e-01 3.36703777e-01 8.45832825e-01
1.43921539e-01 1.62822455e-01 -7.44512856e-01 6.28879607e-01
-1.30816317e+00 -9.77824032e-01 -1.46876559e-01 2.59227896e+00
9.38315034e-01 2.55983502e-01 1.50034547e-01 9.13084030e-01
4.66114312e-01 3.53738725e-01 -6.31557107e-02 -1.00685394e+00
1.79669276e-01 9.13791835e-01 9.79345322e-01 7.21265674e-01
-8.26382518e-01 7.69688904e-01 7.36605549e+00 5.93397141e-01
-6.93227112e-01 9.74692032e-02 8.63804668e-02 1.45320103e-01
-9.84640539e-01 4.82548505e-01 -5.91313720e-01 2.39807531e-01
1.23270607e+00 -1.81860521e-01 4.50797766e-01 7.47537315e-01
-7.38910377e-01 1.50255159e-01 -9.53349590e-01 1.09998345e+00
-1.47849739e-01 -1.35609055e+00 1.21394053e-01 5.10063589e-01
-7.22911432e-02 -4.28178281e-01 3.31517965e-01 2.55419016e-01
2.58097917e-01 -1.16313112e+00 4.59889472e-01 -4.65050429e-01
1.07111764e+00 -8.23294461e-01 3.00675482e-01 -5.52839600e-02
-1.50759947e+00 -2.35668227e-01 -6.91756904e-01 -3.90981771e-02
-1.77348122e-01 3.93678159e-01 -1.08071053e+00 3.21343929e-01
3.06380004e-01 1.18602235e-02 -4.11059290e-01 4.03667957e-01
7.24579096e-02 4.48222548e-01 -9.38455880e-01 -3.40025395e-01
2.06695080e-01 -2.12979570e-01 3.50363016e-01 8.88668954e-01
3.59814614e-01 1.69778675e-01 -2.69350231e-01 4.31343168e-01
3.68160307e-02 4.89634871e-02 -1.02678084e+00 -4.78205711e-01
7.74137735e-01 7.73905098e-01 -5.95175564e-01 -2.38953337e-01
8.94394889e-02 8.93703341e-01 2.65182465e-01 -1.36221170e-01
-6.94798529e-01 -8.19551110e-01 6.91089034e-01 2.82329738e-01
8.49850893e-01 -8.25459778e-01 -2.84200549e-01 -1.08300793e+00
1.60242781e-01 -8.31332862e-01 4.13461089e-01 -5.05129457e-01
-6.63058817e-01 4.57017213e-01 1.70369714e-01 -8.99485290e-01
-6.42102420e-01 -6.80452883e-01 -1.49938622e-02 7.06374526e-01
-9.96082485e-01 -9.21051323e-01 4.71210122e-01 5.35803378e-01
-3.13153744e-01 1.13398597e-01 1.36074448e+00 3.32856536e-01
3.57375182e-02 1.00243890e+00 2.06436202e-01 9.01821107e-02
1.57251880e-01 -1.25134122e+00 7.17631459e-01 4.35429275e-01
3.87872100e-01 8.91186655e-01 7.71165311e-01 -2.98857510e-01
-2.17268133e+00 -4.48469669e-01 1.25397766e+00 -6.11151040e-01
7.89019525e-01 -7.56879032e-01 -7.76687801e-01 8.69018853e-01
-2.90432006e-01 -1.36683397e-02 8.39804769e-01 2.03157794e-02
-1.10131991e+00 -2.17133284e-01 -1.32524097e+00 5.91222048e-01
8.90838385e-01 -1.01315773e+00 -5.85703433e-01 4.30279702e-01
9.24283028e-01 -1.73528939e-01 -1.18603611e+00 9.27758142e-02
7.23747551e-01 -7.05308676e-01 1.36866820e+00 -4.32210296e-01
1.70206025e-01 -4.21844751e-01 -8.20185661e-01 -3.83807957e-01
-3.13450217e-01 -8.00469935e-01 -3.01768303e-01 6.95463836e-01
2.92265445e-01 -9.43648577e-01 8.90812576e-01 5.79864442e-01
5.44983149e-01 -5.61530471e-01 -1.32755876e+00 -5.32501161e-01
3.26519758e-01 -3.44339550e-01 9.04781222e-01 7.47223675e-01
7.36804381e-02 3.03617030e-01 -1.97329640e-01 3.13424831e-03
8.82522821e-01 8.84267092e-02 8.35593343e-01 -1.09017849e+00
-5.01305461e-01 -3.64855938e-02 -9.83026445e-01 -9.74668562e-01
6.07806593e-02 -1.03114009e+00 -5.64217448e-01 -1.09906924e+00
6.31918840e-04 -8.58755708e-01 -1.08960487e-01 7.59406507e-01
2.35311240e-01 3.23152125e-01 2.07672894e-01 2.55828887e-01
-8.90263617e-02 6.86820388e-01 6.76821530e-01 -1.73300922e-01
3.80995542e-01 -4.63326812e-01 -9.83101666e-01 2.06237808e-01
3.72330785e-01 -6.66033089e-01 -4.81661648e-01 -4.23759818e-01
5.72626352e-01 5.02810478e-01 3.04922789e-01 -8.07369053e-01
-7.72424787e-02 2.64490038e-01 6.06428534e-02 -9.54066038e-01
8.65781546e-01 -7.39097297e-01 4.93946910e-01 7.78207183e-01
-2.27000386e-01 5.66362143e-01 1.31036520e-01 6.52668357e-01
1.42589420e-01 -5.45955837e-01 3.76640856e-01 7.50465989e-02
-8.48675370e-02 -1.72114577e-02 -3.91789317e-01 -1.63043573e-01
7.98636019e-01 -4.05837089e-01 -3.82816076e-01 -4.20562774e-01
-4.60448205e-01 -2.65433609e-01 6.96919620e-01 -1.09117731e-01
6.61944389e-01 -1.40626359e+00 -5.05848467e-01 3.33373338e-01
-1.01288036e-01 -3.02949220e-01 -1.03507005e-01 1.60830662e-01
-1.01466000e+00 4.86092448e-01 -3.24044824e-02 -3.63294452e-01
-1.34829772e+00 6.62170708e-01 -4.45781872e-02 -2.98979878e-01
-5.74855328e-01 7.94758201e-01 -2.54909009e-01 -3.82870764e-01
7.89012536e-02 -1.97349131e-01 3.15856129e-01 1.58522204e-02
5.29360175e-01 2.22903177e-01 2.54733741e-01 -6.08300567e-01
-4.13736194e-01 6.06349289e-01 -2.85168737e-01 -5.53708613e-01
1.15646136e+00 3.23248208e-01 -6.20599031e-01 2.29482919e-01
1.69563222e+00 -1.78062264e-02 -4.73416865e-01 -8.15051943e-02
-1.69780388e-01 -6.87207758e-01 -3.34270857e-02 6.68053776e-02
-6.36073172e-01 1.10211563e+00 6.47565663e-01 4.67040867e-01
6.26155794e-01 -2.65778959e-01 1.20736718e+00 9.25027549e-01
7.41295874e-01 -5.45167327e-01 -2.97489073e-02 5.09314001e-01
3.54577005e-01 -5.98718047e-01 3.55845511e-01 -5.72593510e-01
-2.29197085e-01 1.16345394e+00 -2.96912521e-01 -1.72935545e-01
8.08194399e-01 1.75835982e-01 -3.82367760e-01 -2.37088516e-01
-7.34880209e-01 7.98659548e-02 -3.46226901e-01 5.06799877e-01
3.58685732e-01 3.19127351e-01 -5.06149590e-01 7.77184069e-02
-7.41470695e-01 -3.84456992e-01 7.03620374e-01 1.40506804e+00
-2.77773529e-01 -1.55019617e+00 -5.35693944e-01 2.40322098e-01
-5.32912612e-01 -4.60687488e-01 -1.22738697e-01 6.24331295e-01
-4.28353757e-01 6.71006739e-01 3.11845634e-02 -5.03418922e-01
-9.24498439e-02 3.18254441e-01 7.47010589e-01 -3.70692790e-01
-5.01509249e-01 -3.08149636e-01 2.55557686e-01 -3.32312137e-01
2.29827613e-01 -4.22008872e-01 -1.32396972e+00 -9.98279989e-01
-1.94180802e-01 4.92536366e-01 9.48564470e-01 2.32871562e-01
3.83909822e-01 -8.75913724e-02 8.51687968e-01 -6.05775058e-01
-1.12013018e+00 -3.16595763e-01 -5.99968433e-01 2.04639852e-01
3.20277929e-01 -5.90311050e-01 -6.20752633e-01 -2.36745566e-01] | [8.190569877624512, 4.099946975708008] |
aba6bfff-6ab2-4f66-967e-a3eabd22e694 | distilling-multi-level-x-vector-knowledge-for | 2303.01125 | null | https://arxiv.org/abs/2303.01125v1 | https://arxiv.org/pdf/2303.01125v1.pdf | Distilling Multi-Level X-vector Knowledge for Small-footprint Speaker Verification | Deep speaker models yield low error rates in speaker verification. Nonetheless, the high performance tends to be exchanged for model size and computation time, making these models challenging to run under limited conditions. We focus on small-footprint deep speaker embedding extraction, leveraging knowledge distillation. While prior work on this topic has addressed speaker embedding extraction at the utterance level, we propose to combine embeddings from various levels of the x-vector model (teacher network) to train small-footprint student networks. Results indicate the usefulness of frame-level information, with the student models being 85%-91% smaller than their teacher, depending on the size of the teacher embeddings. Concatenation of teacher embeddings results in student networks that reach comparable performance along with the teacher while utilizing a 75% relative size reduction from the teacher. The findings and analogies are furthered to other x-vector variants. | ['Tomi Kinnunen', 'Md Sahidullah', 'Xuechen Liu'] | 2023-03-02 | null | null | null | null | ['speaker-verification'] | ['speech'] | [ 1.56250671e-01 5.99007845e-01 -1.21472843e-01 -5.99281490e-01
-1.12333941e+00 -5.43419898e-01 5.46208978e-01 1.16486862e-01
-6.10219061e-01 2.11896464e-01 5.55010140e-01 -6.99387908e-01
2.14060917e-01 -2.05292806e-01 -3.66359115e-01 -6.30382240e-01
9.76302177e-02 4.07929085e-02 -2.46050969e-01 2.22656950e-02
-3.98682095e-02 3.39084178e-01 -1.52788198e+00 -5.81309460e-02
3.86662006e-01 7.87101567e-01 -2.49185860e-01 9.60987866e-01
2.53896359e-02 5.35354197e-01 -8.64897311e-01 -4.03341204e-01
-3.91686447e-02 -2.19708025e-01 -6.68705165e-01 -6.90960437e-02
9.74434257e-01 -7.37581611e-01 -6.32463515e-01 7.54441857e-01
9.40281510e-01 3.51685822e-01 5.16146481e-01 -1.21230471e+00
-8.06974590e-01 9.79137063e-01 -1.93734929e-01 2.84781963e-01
2.25139603e-01 9.31629464e-02 1.22859120e+00 -9.24377620e-01
2.80066699e-01 1.26710367e+00 7.33054757e-01 7.77494431e-01
-1.33981144e+00 -9.11295533e-01 2.96832681e-01 2.14717925e-01
-1.41514635e+00 -1.17254841e+00 6.71152174e-01 -2.47966722e-01
1.49411404e+00 1.97847232e-01 2.90276527e-01 1.15417790e+00
-2.84890413e-01 7.66458333e-01 8.81100476e-01 -5.52481353e-01
5.14340624e-02 6.93489611e-01 5.45326591e-01 6.19361281e-01
4.69806418e-02 1.90530315e-01 -9.39738691e-01 -1.47687942e-01
2.92137325e-01 -3.33941251e-01 -3.90444398e-01 -1.15635907e-02
-8.82788718e-01 9.96621192e-01 2.28255406e-01 2.42200628e-01
-1.82479575e-01 1.61127314e-01 5.40473878e-01 4.73761857e-01
4.38958347e-01 3.05104226e-01 -6.72318697e-01 -5.35994291e-01
-1.28052223e+00 1.93900123e-01 8.17622781e-01 9.71155167e-01
6.95632458e-01 6.50729001e-01 -8.10247958e-02 8.87782156e-01
5.41241109e-01 2.54885107e-01 7.52598166e-01 -6.69585168e-01
4.00859058e-01 3.20016950e-01 -3.58406037e-01 -6.08653367e-01
-1.83806881e-01 -5.00557840e-01 -1.80527344e-01 7.41373524e-02
3.02022070e-01 -3.44179630e-01 -9.94984686e-01 1.96352160e+00
5.32976210e-01 5.15212238e-01 2.85254598e-01 6.03862524e-01
1.13412178e+00 6.84747934e-01 1.30809918e-01 1.05911210e-01
1.75138164e+00 -9.68817651e-01 -6.79362357e-01 -2.20810235e-01
8.49072039e-01 -6.92226231e-01 9.41851199e-01 2.16951326e-01
-9.09045279e-01 -6.56124711e-01 -1.14281046e+00 -3.47816080e-01
-3.96058053e-01 2.07162425e-01 3.56451720e-01 1.25319529e+00
-1.47165632e+00 3.00004214e-01 -8.01566064e-01 -1.81682363e-01
4.04012889e-01 6.22669756e-01 -3.55567902e-01 9.22956690e-02
-1.03511524e+00 1.17185318e+00 2.20204175e-01 2.86158454e-02
-1.00964880e+00 -1.18102765e+00 -1.17929637e+00 3.76366585e-01
-2.35957369e-01 -2.27103576e-01 1.50381505e+00 -5.20732760e-01
-1.87697983e+00 4.95953381e-01 -3.93090069e-01 -5.33945322e-01
1.47525042e-01 -2.29852766e-01 -4.84203339e-01 4.93922345e-02
-3.56873006e-01 1.09433663e+00 8.60193670e-01 -8.12442064e-01
-5.22330463e-01 -2.51806229e-01 7.69546777e-02 3.13112536e-03
-9.95915174e-01 2.72773564e-01 5.35329897e-03 -4.53278363e-01
-5.13791814e-02 -8.61122489e-01 1.77514493e-01 -1.64850816e-01
-2.19647929e-01 -6.78105891e-01 1.33697271e+00 -9.48388875e-01
1.21580303e+00 -2.39726734e+00 -2.40879208e-01 -1.33928731e-01
2.93507010e-01 4.14089411e-01 -4.21329230e-01 3.82258415e-01
-2.75997192e-01 6.36860356e-02 1.02946013e-01 -6.83347464e-01
3.66842955e-01 4.82378006e-02 -5.13672650e-01 5.11644959e-01
4.79560107e-01 7.49269664e-01 -5.07512450e-01 -4.07448858e-01
1.82631105e-01 9.85650301e-01 -7.99325526e-01 2.41512835e-01
2.88214564e-01 -1.01996943e-01 6.39390722e-02 5.08884370e-01
4.93719757e-01 2.22887561e-01 8.25809464e-02 -1.03940755e-01
1.22234106e-01 1.14370334e+00 -1.11674166e+00 1.60623515e+00
-5.57334185e-01 1.16258204e+00 4.31408435e-01 -9.16399002e-01
8.10035884e-01 9.02042925e-01 1.04771681e-01 -1.98448047e-01
1.63265020e-01 -2.72576455e-02 3.44481558e-01 -1.92027912e-01
6.83200836e-01 -4.94706750e-01 7.73006678e-02 7.12884724e-01
6.10153615e-01 -1.36761665e-01 -2.24573314e-01 2.58818358e-01
1.02336836e+00 -1.69974342e-01 -8.59847367e-02 -1.89922765e-01
3.30598950e-01 -2.90983707e-01 3.06736857e-01 4.30908471e-01
-5.20014465e-01 2.75109351e-01 2.56613076e-01 3.31005576e-04
-7.46159375e-01 -1.02350497e+00 -3.45114559e-01 1.50595272e+00
-3.85739237e-01 -6.29988074e-01 -9.18748677e-01 -6.36724055e-01
1.87572733e-01 9.47031438e-01 -4.68269944e-01 -3.34807396e-01
-6.12109065e-01 -3.04476857e-01 1.06837857e+00 8.18049669e-01
-8.97616073e-02 -7.72502720e-01 -3.46004546e-01 2.71183759e-01
2.64843464e-01 -1.25359762e+00 -5.09925902e-01 3.25364381e-01
-8.89400661e-01 -5.47602415e-01 -4.93466020e-01 -7.15698004e-01
5.24574935e-01 4.69764583e-02 8.78362834e-01 -1.69425651e-01
-1.17658332e-01 5.58493674e-01 -7.17897862e-02 -6.64986134e-01
-5.81501067e-01 2.80836731e-01 5.93821347e-01 -3.67704749e-01
8.71310771e-01 -4.93727237e-01 -2.79480815e-01 -7.22845644e-02
-6.96371138e-01 -3.01656276e-01 4.04791474e-01 9.63850558e-01
-1.26012310e-01 -1.28122047e-01 8.95698369e-01 -4.62519735e-01
5.93666434e-01 -1.01064749e-01 -3.95372838e-01 -1.10020489e-01
-8.18675458e-01 9.74227712e-02 2.91716278e-01 -7.49264479e-01
-9.97782946e-01 -1.78984731e-01 -2.88255781e-01 -4.41898346e-01
-3.28930110e-01 2.06274286e-01 -2.95315012e-02 3.19944508e-02
4.58435982e-01 1.05178811e-01 2.48868376e-01 -4.62438315e-01
6.17406487e-01 1.01529706e+00 2.03153834e-01 -4.84570503e-01
7.66471863e-01 -1.24561377e-01 -8.11018229e-01 -1.25655901e+00
-5.61553776e-01 -2.28455082e-01 -5.09216547e-01 1.71857014e-01
6.78073883e-01 -1.19811058e+00 -7.97193944e-01 -7.70937651e-02
-1.08238637e+00 -9.48011130e-02 -4.23586279e-01 9.44469631e-01
-1.37040898e-01 1.43373504e-01 -8.74543965e-01 -8.97000670e-01
-3.05367500e-01 -1.45868063e+00 1.04072106e+00 2.68352985e-01
-6.00001991e-01 -1.03375173e+00 6.68497309e-02 5.73953450e-01
6.63554013e-01 -6.07240438e-01 9.14516211e-01 -1.20708203e+00
-4.02021915e-01 -3.37964654e-01 -1.12531945e-01 6.88367546e-01
-8.37066695e-02 -6.76039755e-02 -1.69715774e+00 -4.73363787e-01
6.87352894e-03 -4.79445726e-01 8.13595533e-01 1.69008628e-01
7.50833452e-01 -2.89294600e-01 -8.00017789e-02 3.17404836e-01
1.02545428e+00 -1.99299484e-01 2.46446654e-02 4.45591584e-02
5.68910360e-01 5.61579764e-01 -1.85407162e-01 9.64974687e-02
4.16142583e-01 6.67188585e-01 -1.34447530e-01 1.57099649e-01
-5.13363779e-01 -3.14817876e-01 9.59088266e-01 1.37079442e+00
4.42261606e-01 1.34183630e-01 -8.34798813e-01 8.33723009e-01
-1.11812818e+00 -8.56756449e-01 3.89354348e-01 1.87736440e+00
1.09196842e+00 -3.91709134e-02 1.65010184e-01 2.62945056e-01
4.06129718e-01 2.84690827e-01 -3.80629331e-01 -8.37308824e-01
2.70844281e-01 4.18807209e-01 1.68705881e-01 7.94864416e-01
-8.50105643e-01 8.73703897e-01 7.38990068e+00 5.97912431e-01
-1.33800435e+00 3.43251050e-01 2.23515645e-01 -4.87099469e-01
-3.04700881e-01 -1.43764719e-01 -1.43600500e+00 1.53760925e-01
1.82208455e+00 -5.84275387e-02 1.10249348e-01 9.41804349e-01
1.16739599e-02 3.34365338e-01 -1.46446919e+00 8.90148520e-01
3.14593911e-01 -1.07119167e+00 -1.71755090e-01 2.96156526e-01
5.73655248e-01 2.28405178e-01 1.79493144e-01 8.66213620e-01
3.67355853e-01 -1.13834584e+00 5.48047125e-01 -2.05105394e-01
6.41437113e-01 -7.60111213e-01 6.52022719e-01 2.00006396e-01
-9.84175920e-01 -1.40286550e-01 -2.77132690e-01 -7.00834095e-02
7.79657066e-02 2.16399193e-01 -1.56562603e+00 9.71330702e-02
4.82905567e-01 1.99940458e-01 -5.45209289e-01 3.88181090e-01
-1.86778247e-01 1.26404607e+00 -5.53093612e-01 -1.00949816e-02
2.76240885e-01 9.78927165e-02 3.15407157e-01 1.65010464e+00
1.77474543e-01 -1.15512416e-01 -9.90526453e-02 7.00869620e-01
-3.79576176e-01 -6.89276755e-02 -5.48142970e-01 -4.22181219e-01
8.20475042e-01 1.17709970e+00 -7.37112612e-02 -7.12191641e-01
-6.16247475e-01 6.99726045e-01 4.30671006e-01 4.54915226e-01
-7.98826456e-01 -5.74245095e-01 1.29454243e+00 -1.48322240e-01
5.65014899e-01 -3.69577676e-01 -2.87188798e-01 -9.74476397e-01
-2.30444013e-03 -9.28894639e-01 2.33361125e-02 -2.85407841e-01
-1.00439787e+00 6.56076849e-01 -2.05259044e-02 -6.78525209e-01
-6.09380007e-01 -6.80853665e-01 -6.26212895e-01 1.02551305e+00
-1.57418358e+00 -1.13870442e+00 1.84578121e-01 2.44772851e-01
6.04821682e-01 -2.64280528e-01 1.28906071e+00 3.19572330e-01
-7.63391495e-01 1.32352114e+00 -2.29293387e-02 4.84260917e-01
6.09937429e-01 -1.15590739e+00 2.74487674e-01 6.48231745e-01
5.49359739e-01 9.57878292e-01 5.79747736e-01 -1.42708682e-02
-1.57644379e+00 -7.22162247e-01 1.22561646e+00 -8.16111326e-01
6.39539599e-01 -6.26586080e-01 -9.63900745e-01 8.86686862e-01
5.17807066e-01 -8.15472938e-03 1.13136208e+00 6.82759225e-01
-8.39135230e-01 -4.26752679e-02 -1.00410724e+00 3.98440808e-01
5.21968067e-01 -1.32458138e+00 -9.12064314e-01 -5.42715192e-02
9.90624368e-01 -2.02076569e-01 -1.13594139e+00 1.00934125e-01
7.57600009e-01 -5.99326074e-01 9.15491641e-01 -7.20851958e-01
1.13762550e-01 -1.25452489e-01 -2.10640207e-01 -1.25832355e+00
-4.07008752e-02 -3.34028095e-01 -5.12404382e-01 1.64167523e+00
7.12847829e-01 -6.32896781e-01 1.00444722e+00 8.93866360e-01
-2.04457939e-01 -6.88124061e-01 -1.27504265e+00 -8.22895765e-01
2.12580577e-01 -6.28188193e-01 5.25394678e-01 1.16116536e+00
1.72298089e-01 6.63756907e-01 1.10001415e-01 4.17023927e-01
3.48875254e-01 -3.15352559e-01 7.70605981e-01 -9.90340590e-01
-2.14336619e-01 -5.99235594e-01 -6.89896286e-01 -1.10371208e+00
5.97948313e-01 -1.00484753e+00 9.19761881e-02 -1.23646545e+00
-8.68420452e-02 -2.73578614e-01 -4.30945665e-01 5.96305549e-01
-3.42222273e-01 2.68352032e-01 1.82612181e-01 -2.59772211e-01
-3.96896480e-03 6.23095334e-01 4.67283607e-01 -3.52848142e-01
-2.26967245e-01 -3.10264528e-01 -8.24810803e-01 5.11800230e-01
6.92694485e-01 -4.25366491e-01 -4.56010163e-01 -6.77142739e-01
-6.58437014e-01 -2.66026586e-01 5.67359924e-02 -8.66622925e-01
3.71952623e-01 5.12179375e-01 4.34555233e-01 -4.51293766e-01
7.94460297e-01 -6.48168206e-01 -3.83365035e-01 1.92996085e-01
-6.14869535e-01 1.06163882e-01 6.67975068e-01 3.65662396e-01
-3.15012336e-01 -3.38985026e-01 7.21861422e-01 4.78590518e-01
-4.22444344e-01 -3.36349159e-02 -5.16559660e-01 -1.99017867e-01
6.85676754e-01 -3.03215325e-01 -1.08134702e-01 -3.65860552e-01
-6.80941582e-01 -1.71647981e-01 -1.28493244e-02 6.01128638e-01
5.13118029e-01 -1.18724835e+00 -7.90937364e-01 5.05405962e-01
5.10774367e-02 -2.39173800e-01 2.23878860e-01 7.45508015e-01
5.13731129e-02 7.14581907e-01 2.96427459e-01 -6.42914951e-01
-1.63353801e+00 2.60878175e-01 1.72662362e-01 8.47788304e-02
-5.66664934e-01 1.27917278e+00 4.77087870e-02 -8.94161284e-01
6.95222497e-01 -4.74707782e-01 1.26939833e-01 2.96329945e-01
8.42824936e-01 3.44411790e-01 3.01516771e-01 -6.31208420e-01
-5.31315744e-01 1.25870168e-01 -4.74378586e-01 -4.08541083e-01
1.36999071e+00 8.23455974e-02 3.22090119e-01 2.96983838e-01
1.57868385e+00 2.25287437e-01 -1.13250268e+00 -3.78274411e-01
2.03787815e-02 -9.28413793e-02 5.81898689e-01 -5.60441554e-01
-6.85551941e-01 1.23260868e+00 8.33971739e-01 -4.93975878e-02
6.32189214e-01 1.29580170e-01 8.52343619e-01 3.60231280e-01
-2.01580182e-01 -1.05317104e+00 -2.87667543e-01 3.96018386e-01
4.62225795e-01 -1.15101278e+00 1.01017520e-01 7.08414018e-02
-4.47741061e-01 9.17390585e-01 6.32160604e-01 8.03412870e-02
9.03145373e-01 4.17577446e-01 3.90353709e-01 -9.43670943e-02
-1.00527620e+00 1.18701391e-01 1.90055996e-01 5.61875284e-01
8.87981832e-01 1.59479529e-01 3.27442884e-01 4.96508211e-01
-6.84665143e-01 -4.59383309e-01 3.18838239e-01 8.70908201e-01
-4.42336738e-01 -1.11469471e+00 -3.52882802e-01 2.78161019e-01
-4.01627421e-01 -3.59985560e-01 -2.22200111e-01 5.93932211e-01
-1.56243905e-01 1.15527821e+00 2.94864982e-01 -5.05556703e-01
3.03345859e-01 7.93084204e-01 3.39168727e-01 -7.78024077e-01
-8.50750208e-01 -1.28808007e-01 2.72801429e-01 -3.22939515e-01
-1.59071892e-01 -6.56897008e-01 -1.07755613e+00 -3.84389192e-01
-8.26519072e-01 3.14210624e-01 1.12197876e+00 8.32546771e-01
5.81821680e-01 5.88418186e-01 3.57315570e-01 -8.36368918e-01
-1.16099596e+00 -1.39324772e+00 -3.99178952e-01 -1.61359265e-01
7.44350612e-01 -4.66112971e-01 -5.00685275e-01 1.19171351e-01] | [14.371315956115723, 6.158485412597656] |
aa1ee96f-422d-4437-bc48-35d76e85488d | cfear-radarodometry-conservative-filtering-1 | null | null | https://arxiv.org/abs/2105.01457 | https://arxiv.org/pdf/2105.01457.pdf | CFEAR Radarodometry - Conservative Filtering for Efficient and Accurate Radar Odometry | This paper presents the accurate, highly efficient, and learning-free method CFEAR Radarodometry for large-scale radar odometry estimation. By using a filtering technique that keeps the k strongest returns per azimuth and by additionally filtering the radar data in Cartesian space, we are able to compute a sparse set of oriented surface points for efficient and accurate scan matching. Registration is carried out by minimizing a point-to-line metric and robustness to outliers is achieved using a Huber loss. We were able to additionally reduce drift by jointly registering the latest scan to a history of keyframes and found that our odometry method generalizes to different sensor models and datasets without changing a single parameter. We evaluate our method in three widely different environments and demonstrate an improvement over spatially cross-validated state-of-the-art with an overall translation error of 1.76% in a public urban radar odometry benchmark, running at 55Hz merely on a single laptop CPU thread. | ['Henrik Andreasson', 'Achim J. Lilienthal', 'Anas Alhashimi', 'Martin Magnusson', 'Daniel Adolfsson'] | 2021-09-16 | null | null | null | ieee-rsj-international-conference-on-3 | ['radar-odometry'] | ['robots'] | [ 1.42777503e-01 -2.05901340e-01 1.49909720e-01 -5.88601649e-01
-1.18913734e+00 -5.20310938e-01 7.18564689e-01 1.03944935e-01
-6.87674999e-01 6.74719572e-01 -5.27415015e-02 -1.24488346e-01
-4.38665420e-01 -1.06519485e+00 -7.21077204e-01 -3.95722598e-01
-6.66279137e-01 1.07929230e+00 5.16483963e-01 -4.97898698e-01
3.61482739e-01 9.17577922e-01 -1.48974204e+00 -7.26495445e-01
8.46296966e-01 9.71164644e-01 -3.60169590e-01 8.59377027e-01
5.24304211e-01 2.31306821e-01 -4.58541155e-01 -2.05053449e-01
8.02266359e-01 3.19705367e-01 -2.74884760e-01 -2.85710841e-01
1.27519572e+00 -2.22368598e-01 -3.55455816e-01 1.00591016e+00
8.35551023e-01 8.01017100e-04 4.45557684e-01 -9.13182795e-01
1.68559343e-01 -1.81990683e-01 -4.27955627e-01 3.36224446e-03
3.63839477e-01 -1.55461118e-01 5.57975292e-01 -9.04864728e-01
8.48091304e-01 8.43049526e-01 1.32526875e+00 -3.30741018e-01
-1.55637419e+00 -8.29751670e-01 -3.22436303e-01 -1.30588889e-01
-1.90442443e+00 -6.63087964e-01 2.06822380e-01 -4.67037588e-01
1.15099704e+00 1.82784140e-01 6.40009284e-01 3.76005054e-01
6.51582360e-01 -1.50573462e-01 8.86127532e-01 -2.30866030e-01
1.94892749e-01 -6.09615982e-01 2.66204566e-01 7.94800878e-01
9.23781037e-01 6.22556388e-01 -5.66950679e-01 -3.48124325e-01
4.89360064e-01 8.62390473e-02 -2.95835175e-02 -8.07847142e-01
-1.30792320e+00 8.47847283e-01 5.25106370e-01 -2.31020942e-01
-6.22069240e-01 3.79155695e-01 4.37422749e-03 5.28859437e-01
5.52375734e-01 4.15668070e-01 -3.06590170e-01 -1.11505754e-01
-1.29906869e+00 6.92576826e-01 8.99823546e-01 9.22523797e-01
1.19649005e+00 2.18547359e-01 5.67359507e-01 1.60119355e-01
1.65483862e-01 1.64653492e+00 -2.85079647e-02 -8.10540318e-01
5.11518478e-01 2.31652185e-01 3.78837794e-01 -1.14568651e+00
-1.00550199e+00 -5.39066732e-01 -7.32196569e-01 4.60867673e-01
2.83462763e-01 -4.87663269e-01 -1.06405175e+00 1.35222733e+00
5.63449025e-01 3.45839739e-01 2.05462977e-01 9.51288164e-01
2.15185568e-01 1.09780557e-01 -4.36505079e-01 -1.39600066e-02
1.36744630e+00 -1.86022878e-01 -4.50235486e-01 -6.12715781e-01
6.50338531e-01 -8.97098005e-01 1.65310919e-01 5.34647942e-01
-6.07783079e-01 -3.60304534e-01 -1.73299313e+00 2.42093712e-01
-1.73338011e-01 -2.95497924e-01 7.03997314e-01 7.78755426e-01
-9.57437217e-01 5.68718970e-01 -1.02623355e+00 -3.42857242e-01
1.13099612e-01 3.67836744e-01 -3.80297482e-01 -1.51025668e-01
-1.16050673e+00 1.35416269e+00 9.90333222e-03 -3.41343619e-02
-6.00533247e-01 -9.07147646e-01 -1.00170040e+00 -7.01916218e-01
2.40641665e-02 -7.22989321e-01 1.01613891e+00 -2.31258780e-01
-1.29209232e+00 6.26820266e-01 -3.99128824e-01 -1.06274629e+00
4.18041438e-01 -6.26783311e-01 -8.72933686e-01 -8.15297440e-02
1.95296466e-01 5.27551651e-01 8.68152201e-01 -9.42057490e-01
-7.00302660e-01 -6.40592933e-01 -6.21320605e-01 1.47366703e-01
5.73290527e-01 -4.38912988e-01 -1.30570278e-01 -4.80390966e-01
1.03012478e+00 -1.35746264e+00 -3.70640159e-01 -1.96861461e-01
1.70132741e-01 6.15601659e-01 4.10061359e-01 -6.35696948e-01
8.54350090e-01 -1.77414346e+00 -1.97596818e-01 7.76863277e-01
7.66183808e-02 -1.99364603e-01 1.96264923e-01 2.91715473e-01
3.00594151e-01 -6.09539568e-01 -3.50756288e-01 -1.49517670e-01
-9.62899625e-02 1.82791054e-01 -5.44955492e-01 1.20696175e+00
-2.07810581e-01 6.37077928e-01 -7.65128016e-01 -4.76506278e-02
1.88019890e-02 5.86571574e-01 -3.82151127e-01 -3.32213104e-01
1.76434219e-01 1.81525290e-01 -1.50024369e-01 7.13137746e-01
1.13129878e+00 4.02874112e-01 -5.73626161e-02 -9.81055051e-02
-3.49576205e-01 4.76574153e-01 -1.67356586e+00 1.87696338e+00
-4.00136977e-01 7.60688186e-01 2.87978407e-02 -4.28278089e-01
1.53086388e+00 -1.44246921e-01 5.44796228e-01 -9.95832622e-01
-1.79251492e-01 4.78623837e-01 -2.05840215e-01 1.34061620e-01
8.56845319e-01 -1.32087022e-01 -4.14884686e-01 3.18977326e-01
-3.71539146e-01 -7.37639308e-01 -1.68892279e-01 -1.78841233e-01
1.28307283e+00 6.00108355e-02 4.91165161e-01 -4.91675943e-01
1.31319433e-01 5.33021092e-01 4.26555604e-01 6.71142638e-01
8.71363655e-02 4.05130655e-01 -3.65624398e-01 -7.89866328e-01
-8.03277791e-01 -1.34362912e+00 -4.81251627e-01 5.67518711e-01
3.14605534e-01 -2.12897092e-01 -2.20873691e-02 -6.67376295e-02
5.41843116e-01 5.15113592e-01 -2.17583522e-01 9.62341111e-03
-8.13142717e-01 -9.07654583e-01 9.34658110e-01 2.47614726e-01
3.53633493e-01 -2.01354221e-01 -1.06801009e+00 4.87183094e-01
1.08274758e-01 -9.79074001e-01 -9.34715047e-02 3.51954371e-01
-1.30720425e+00 -9.14410472e-01 -1.39491037e-01 -4.14150655e-01
3.94361615e-01 4.78921622e-01 1.24664938e+00 -3.92589867e-01
-2.09650084e-01 -4.33056988e-02 -5.66084757e-02 -4.91095781e-01
3.88711065e-01 1.12452246e-01 4.73505735e-01 -3.79870892e-01
4.16613162e-01 -6.97771966e-01 -3.22388053e-01 2.66137600e-01
-2.91260928e-01 -3.69581848e-01 4.72877353e-01 5.84083855e-01
6.32106960e-01 -3.59724313e-01 8.93263053e-03 -5.70600569e-01
1.51670858e-01 -2.72714376e-01 -1.31722903e+00 -4.22993958e-01
-9.23714757e-01 5.11735439e-01 -6.77811801e-02 2.71815378e-02
-5.84224939e-01 4.88396585e-01 -5.57670649e-03 2.71928497e-02
2.96743922e-02 4.35123444e-01 2.95370400e-01 -7.50257075e-01
1.00725543e+00 1.19821742e-01 2.14888245e-01 -2.23525450e-01
5.54995179e-01 3.66481751e-01 7.98783600e-01 -3.08263659e-01
1.38166714e+00 9.75656450e-01 5.94601452e-01 -1.00589132e+00
-5.98810792e-01 -7.13139296e-01 -7.54353344e-01 7.53126517e-02
4.88372654e-01 -1.32679963e+00 -6.15886867e-01 2.72938401e-01
-9.00906324e-01 -1.22065417e-01 -1.84805110e-01 8.57790411e-01
-5.38344443e-01 2.15702310e-01 8.22422430e-02 -6.56231761e-01
-5.56044161e-01 -7.99741387e-01 1.30037189e+00 -1.56843811e-01
-3.82485688e-01 -7.50313103e-01 1.04549408e+00 -1.04635917e-01
4.35425580e-01 5.35885453e-01 -5.53336255e-02 -2.19929591e-01
-9.20560718e-01 -5.19305348e-01 -2.24684943e-02 -4.78342712e-01
-2.66819507e-01 -4.26205456e-01 -8.04576814e-01 -5.85746288e-01
-2.89170235e-01 6.50271773e-02 9.90652740e-01 3.50899756e-01
-2.00002164e-01 -3.75056602e-02 -4.98762757e-01 1.19956625e+00
1.69844377e+00 -1.97759315e-01 8.04233134e-01 7.51748562e-01
4.19366241e-01 3.19061607e-01 1.00107753e+00 3.43064338e-01
5.71703672e-01 1.04785645e+00 5.66648006e-01 9.70547572e-02
2.72201598e-01 9.95329767e-03 4.22406286e-01 3.70724112e-01
-4.18798655e-01 3.78242135e-01 -1.01076555e+00 6.75316274e-01
-1.76822710e+00 -1.07207227e+00 -3.88068289e-01 2.59721708e+00
3.67802203e-01 3.83995026e-01 -2.08289951e-01 3.73502560e-02
2.56696790e-01 4.22226608e-01 -1.67108968e-01 -3.73822749e-02
-7.51711382e-03 6.66951060e-01 1.78073156e+00 1.12049854e+00
-1.34317672e+00 9.03394103e-01 6.56303167e+00 9.36244279e-02
-1.19116974e+00 3.05792084e-03 -7.29400814e-01 -9.43905953e-03
-1.71844102e-02 2.69411355e-01 -1.16942358e+00 -1.75951105e-02
1.28047490e+00 1.53084695e-01 3.13648343e-01 8.35523427e-01
-1.67129904e-01 -2.45186046e-01 -5.19434810e-01 8.47860038e-01
1.07146636e-01 -1.48371983e+00 -5.89047253e-01 2.10890114e-01
6.45089507e-01 9.46705282e-01 -4.00780261e-01 4.06769887e-02
6.12134993e-01 -8.54326725e-01 7.96900272e-01 8.13108146e-01
7.95094728e-01 -9.00489032e-01 7.62837470e-01 2.42937028e-01
-1.64037931e+00 4.49677616e-01 -3.83876562e-01 -3.99065524e-01
3.82061869e-01 8.61318886e-01 -1.08439696e+00 9.87648487e-01
6.28863215e-01 6.18316710e-01 -4.49111432e-01 1.14763319e+00
-1.70724794e-01 1.27102500e-02 -8.73601794e-01 2.63316125e-01
1.37281418e-01 -2.37685695e-01 7.52323151e-01 1.13001573e+00
8.07175517e-01 -8.68674964e-02 3.18368018e-01 7.71535980e-03
5.00966370e-01 -3.29552859e-01 -8.61292541e-01 7.49897063e-01
8.01724255e-01 1.12284839e+00 -1.85950801e-01 -1.35291412e-01
-5.55480015e-04 5.29134989e-01 -1.38827428e-01 -4.10731174e-02
-6.66892529e-01 -8.39575648e-01 1.25688612e+00 3.84592593e-01
4.75285500e-01 -9.23238039e-01 -3.59290361e-01 -9.36951280e-01
1.99071821e-02 -6.04407072e-01 1.90090865e-01 -3.86595398e-01
-7.20689297e-01 5.75256169e-01 2.63637770e-02 -1.64878285e+00
-5.94331026e-01 -3.45226526e-01 -7.47252852e-02 9.69487488e-01
-1.74287128e+00 -9.74916518e-01 -6.72844172e-01 4.13417429e-01
-2.12175071e-01 -7.63378665e-02 8.45070839e-01 3.64940852e-01
3.92903477e-01 1.69930771e-01 1.49720937e-01 -9.99601632e-02
1.05288911e+00 -1.01619816e+00 1.09679580e+00 1.08284891e+00
3.51533741e-01 5.54460704e-01 1.16074049e+00 -9.84177172e-01
-1.64665341e+00 -1.11376810e+00 1.36629748e+00 -5.74412465e-01
7.54747808e-01 -2.94225931e-01 -6.01337433e-01 8.90103221e-01
-2.45440796e-01 1.71946540e-01 -1.34286210e-01 1.66167811e-01
-4.07639265e-01 -5.57635427e-01 -1.20388889e+00 2.38726377e-01
1.15429282e+00 -2.74516016e-01 -8.51482868e-01 2.38046795e-01
4.54894871e-01 -9.96161103e-01 -9.57231820e-01 6.52434111e-01
7.90591240e-01 -8.45462382e-01 1.30798423e+00 -2.43801245e-04
-7.57243633e-01 -8.32281053e-01 -5.06276667e-01 -1.12396443e+00
-3.37095052e-01 -7.80212224e-01 -4.08881530e-02 5.43709636e-01
1.61403507e-01 -1.10875094e+00 8.32510650e-01 -1.67481169e-01
-2.06783533e-01 -1.13040283e-01 -1.23880470e+00 -1.00298142e+00
-4.28141683e-01 -7.06180632e-01 6.39055610e-01 5.08533895e-01
-4.47219968e-01 3.11531663e-01 -4.68420178e-01 9.36794043e-01
1.24283826e+00 2.45427057e-01 1.28951299e+00 -1.83138323e+00
1.15658857e-01 1.14592828e-01 -9.92574811e-01 -1.25940120e+00
-3.18651311e-02 -7.69838750e-01 3.11840057e-01 -1.07579839e+00
-6.89337909e-01 -7.01087356e-01 3.46641153e-01 6.84420466e-02
3.53139997e-01 6.31727040e-01 -1.33346692e-01 3.49148542e-01
-3.18853855e-01 1.95386171e-01 5.43718994e-01 -3.40945534e-02
-1.78647682e-01 1.40680164e-01 -1.23967163e-01 7.55934715e-01
4.90298837e-01 -8.95993292e-01 6.18259422e-02 -4.74485248e-01
4.68470275e-01 -4.54417914e-02 5.73079705e-01 -1.59568882e+00
5.70942402e-01 -3.91054116e-02 4.45291877e-01 -1.16981053e+00
5.93221366e-01 -8.31737101e-01 6.83241487e-01 8.45929027e-01
5.42191029e-01 4.70355213e-01 2.21999109e-01 5.78728676e-01
-1.38082132e-01 1.62344307e-01 7.98143566e-01 4.34828460e-01
-9.83261585e-01 2.26321012e-01 -9.24426019e-02 1.26202956e-01
8.12650502e-01 -3.98617536e-01 -4.76118803e-01 -3.25741887e-01
-3.64490926e-01 1.90273181e-01 8.35033476e-01 8.75768159e-03
4.88556266e-01 -1.37097418e+00 -7.87884176e-01 6.30297124e-01
4.31552410e-01 -6.57816827e-02 -7.17791021e-02 9.56345856e-01
-9.51668382e-01 6.22546673e-01 -2.51390487e-01 -9.27679181e-01
-9.97172475e-01 -2.08166867e-01 4.56979245e-01 -2.13255897e-01
-7.91039467e-01 5.84641337e-01 -8.92378569e-01 -7.29451418e-01
-6.96969032e-02 -4.54541296e-01 3.13155651e-01 2.92435914e-01
4.57860351e-01 4.69651073e-01 7.19533205e-01 -8.19992602e-01
-1.06593490e+00 1.26966798e+00 3.12116832e-01 -5.46453953e-01
1.37826705e+00 -1.00667765e-02 2.75565743e-01 7.97024071e-02
8.13711882e-01 5.08599102e-01 -1.31627977e+00 -2.63121784e-01
4.16029871e-01 -4.92523462e-01 2.78564155e-01 -2.85643399e-01
-6.10379875e-01 3.23246002e-01 9.62104797e-01 -1.84439167e-01
7.08256602e-01 -4.34233993e-01 8.28908801e-01 9.21225548e-01
8.88251305e-01 -9.27508831e-01 -4.93323505e-01 1.15898144e+00
5.74810624e-01 -1.12643337e+00 7.72907853e-01 -1.38048101e-02
-5.31907201e-01 1.01929486e+00 -2.33288378e-01 -8.15660536e-01
8.43133271e-01 6.95670009e-01 3.70535880e-01 -3.41780752e-01
-3.28744888e-01 -4.77074951e-01 2.60279059e-01 9.31192815e-01
1.67060420e-01 7.91384950e-02 3.20251584e-02 -1.59597546e-01
-6.74963832e-01 -1.51852176e-01 2.21366942e-01 1.20522475e+00
-1.01137626e+00 -9.23757732e-01 -8.08503747e-01 2.50452250e-01
1.76717803e-01 -6.18563071e-02 8.35270956e-02 1.15461636e+00
-6.08524606e-02 4.65082318e-01 4.54320490e-01 -4.21955347e-01
6.54556513e-01 3.11224703e-02 6.89436793e-01 -2.75139242e-01
-3.01217794e-01 -1.49231195e-01 4.24499810e-01 -9.83719945e-01
-4.22054023e-01 -9.73930836e-01 -1.24561632e+00 -5.17809987e-01
-5.22364601e-02 -8.39860812e-02 9.00150120e-01 7.67746150e-01
5.35763502e-01 1.23402022e-01 3.95873368e-01 -1.07483065e+00
-8.17445695e-01 -7.17570066e-01 -5.22135317e-01 -2.50862926e-01
7.28502929e-01 -8.91199589e-01 -1.87807605e-01 -2.27570742e-01] | [7.379060745239258, -2.139268398284912] |
a331b2fc-1742-4ba7-b134-74d70a405319 | shapelet-based-sparse-representation-for | 1708.05974 | null | http://arxiv.org/abs/1708.05974v1 | http://arxiv.org/pdf/1708.05974v1.pdf | Shapelet-based Sparse Representation for Landcover Classification of Hyperspectral Images | This paper presents a sparse representation-based classification approach
with a novel dictionary construction procedure. By using the constructed
dictionary sophisticated prior knowledge about the spatial nature of the image
can be integrated. The approach is based on the assumption that each image
patch can be factorized into characteristic spatial patterns, also called
shapelets, and patch-specific spectral information. A set of shapelets is
learned in an unsupervised way and spectral information are embodied by
training samples. A combination of shapelets and spectral information are
represented in an undercomplete spatial-spectral dictionary for each individual
patch, where the elements of the dictionary are linearly combined to a sparse
representation of the patch. The patch-based classification is obtained by
means of the representation error. Experiments are conducted on three
well-known hyperspectral image datasets. They illustrate that our proposed
approach shows superior results in comparison to sparse representation-based
classifiers that use only limited spatial information and behaves competitively
with or better than state-of-the-art classifiers utilizing spatial information
and kernelized sparse representation-based classifiers. | ['Björn Waske', 'Ribana Roscher'] | 2017-08-20 | null | null | null | null | ['classification-of-hyperspectral-images', 'sparse-representation-based-classification'] | ['computer-vision', 'computer-vision'] | [ 5.53587377e-01 -2.57386178e-01 -2.89421260e-01 -2.03155741e-01
-4.47339505e-01 -3.84792387e-01 4.76980507e-01 1.20903134e-01
-4.81877178e-02 7.19945312e-01 1.09991487e-02 1.41749710e-01
-4.97925073e-01 -9.06756759e-01 -3.78467530e-01 -1.16986573e+00
9.50583667e-02 1.68199554e-01 3.89440432e-02 -8.32004398e-02
2.47753292e-01 8.27321887e-01 -1.96875584e+00 4.13355321e-01
8.29570651e-01 1.18581593e+00 4.86038893e-01 1.89794838e-01
-1.76964507e-01 7.26396859e-01 -3.08482766e-01 4.95414555e-01
2.66750544e-01 -4.75063920e-01 -3.68263751e-01 6.41852081e-01
3.77063245e-01 3.17038178e-01 -3.91201645e-01 1.14223993e+00
3.55761275e-02 2.79565513e-01 8.75491798e-01 -6.43415749e-01
-7.40533590e-01 1.75247937e-01 -5.84481955e-01 1.00070149e-01
3.30983490e-01 -4.61128026e-01 6.34993613e-01 -1.21983802e+00
5.46301842e-01 7.73075521e-01 8.43388796e-01 -2.05501646e-01
-1.31816673e+00 -2.72155762e-01 -3.65649909e-02 3.12846482e-01
-1.71523583e+00 -4.03352737e-01 1.15927172e+00 -6.53170943e-01
9.20850039e-01 3.24195385e-01 6.56352699e-01 3.94250363e-01
9.61459801e-02 3.81110698e-01 1.33231485e+00 -8.47747684e-01
4.47976410e-01 2.25829393e-01 2.07572013e-01 7.34466553e-01
3.32732081e-01 3.05795491e-01 -4.84348118e-01 -3.66525143e-01
6.70730293e-01 3.61349910e-01 -5.12394309e-01 -6.41034365e-01
-8.35531235e-01 9.82129753e-01 5.50266266e-01 8.69628251e-01
-9.34800208e-01 -4.40974653e-01 2.82048862e-02 1.02706179e-01
5.56343913e-01 1.06312044e-01 -1.91884078e-02 5.18176079e-01
-1.35230494e+00 -1.29008904e-01 7.54033923e-01 6.52962446e-01
1.19306326e+00 5.07317603e-01 1.99874952e-01 1.12076354e+00
2.02742428e-01 9.09488738e-01 4.78346407e-01 -5.55788755e-01
-1.28839791e-01 6.87919319e-01 1.16796214e-02 -1.33464038e+00
-2.65760362e-01 -7.91599512e-01 -1.01802731e+00 1.93976909e-01
7.94960409e-02 2.97726572e-01 -8.29653978e-01 1.17235267e+00
2.06947878e-01 5.87237895e-01 2.53078163e-01 7.47342110e-01
8.93954873e-01 8.48926008e-01 -1.66882619e-01 -5.11249959e-01
1.23785222e+00 -6.95162773e-01 -5.71834505e-01 3.82869579e-02
1.38941213e-01 -8.44402015e-01 3.11212838e-01 5.74661136e-01
-6.41405225e-01 -6.68486953e-01 -1.05214500e+00 5.43489575e-01
-4.04413640e-01 5.64870179e-01 6.02584004e-01 5.41358232e-01
-7.73291230e-01 2.68946856e-01 -5.20035982e-01 -4.30709809e-01
2.51931846e-01 2.86935102e-02 -7.30577350e-01 -3.72740418e-01
-5.65076768e-01 7.85121024e-01 4.95135516e-01 -9.32461843e-02
-6.35632277e-01 -6.94687307e-01 -9.92499053e-01 2.74801478e-02
-4.91610207e-02 -3.18842947e-01 5.03044248e-01 -1.08437836e+00
-1.14474130e+00 7.50144064e-01 -4.56427723e-01 -2.67296404e-01
-5.08383036e-01 4.98879790e-01 -6.03559434e-01 7.39615738e-01
1.58849031e-01 -6.41806517e-03 1.35466683e+00 -1.46412671e+00
-6.55448318e-01 -4.17360157e-01 -3.90660167e-01 5.45763709e-02
-3.48262101e-01 -2.32105955e-01 1.75857574e-01 -8.16965342e-01
6.54989660e-01 -6.86948836e-01 -2.06315339e-01 -1.88384712e-01
4.16970812e-02 1.49360850e-01 1.13490939e+00 -6.81998909e-01
1.08864927e+00 -2.45394778e+00 4.30100381e-01 6.24626040e-01
4.02630540e-03 2.48682424e-01 -2.82499254e-01 6.81086183e-01
-3.72303069e-01 -5.24609149e-01 -5.71172833e-01 7.05468729e-02
-4.72824723e-01 3.23811978e-01 -3.71041030e-01 8.93013835e-01
-1.22488309e-02 2.69381016e-01 -9.01011825e-01 -1.27753839e-01
5.55889606e-01 7.81803250e-01 -2.76738107e-01 5.93701601e-02
4.69405092e-02 4.54815179e-01 -4.44289953e-01 7.89201736e-01
8.73561859e-01 -1.51325494e-01 1.16981827e-01 -4.97960150e-01
-3.63136500e-01 -3.84439141e-01 -1.28103340e+00 1.70945120e+00
-4.96291280e-01 2.95749933e-01 2.74255216e-01 -1.73203599e+00
1.31387544e+00 4.36319768e-01 7.88159728e-01 -2.86482036e-01
-1.04921468e-01 3.55584294e-01 -4.04093385e-01 -3.29209387e-01
2.57218599e-01 -3.89117450e-01 4.51702565e-01 1.39196247e-01
4.24529433e-01 -4.92866710e-02 7.70778731e-02 -3.07923973e-01
8.12916517e-01 -1.35641232e-01 7.01658249e-01 -5.39466739e-01
9.07950759e-01 2.67543972e-01 2.46159106e-01 4.51026082e-01
3.12811852e-01 2.72335529e-01 -1.51477069e-01 -6.22333169e-01
-1.04684007e+00 -6.95075452e-01 -5.75030386e-01 6.67181969e-01
-9.86684188e-02 6.41805492e-03 -4.10155952e-01 -2.32925728e-01
1.69769540e-01 5.21059155e-01 -6.81404412e-01 5.61350770e-02
-4.41623740e-02 -7.71085620e-01 1.16049089e-01 1.49395660e-01
4.96125221e-01 -8.56194854e-01 -5.28536439e-01 3.45413953e-01
1.58101961e-01 -8.06088865e-01 1.57728538e-01 3.01878661e-01
-1.06198001e+00 -1.08030343e+00 -7.35821724e-01 -9.28019226e-01
8.89141321e-01 9.23123896e-01 7.19030380e-01 -2.02458709e-01
-4.58736539e-01 5.79559624e-01 -7.64925659e-01 -2.33419016e-01
-1.52171850e-01 -2.09986135e-01 -5.91348074e-02 7.37685263e-01
4.91270661e-01 -9.98679876e-01 -2.87065536e-01 8.36105794e-02
-8.44823837e-01 -3.76967162e-01 7.20290244e-01 1.11754668e+00
9.25360024e-01 6.95509315e-01 5.87583423e-01 -7.56358385e-01
1.91332743e-01 -7.88736463e-01 -5.67290068e-01 1.83973014e-01
-3.44744235e-01 -9.74151939e-02 7.09133387e-01 -3.48041505e-01
-1.12140131e+00 5.45947134e-01 2.35609934e-01 -6.18589997e-01
-4.32046384e-01 9.12053347e-01 8.00055563e-02 -9.18925524e-01
9.98834014e-01 9.99397874e-01 2.17924103e-01 -6.28931761e-01
2.79893249e-01 7.19080865e-01 4.56757098e-01 -6.13548160e-01
8.50012779e-01 7.35577643e-01 2.78708905e-01 -1.55984640e+00
-7.61799932e-01 -1.11749256e+00 -8.51361036e-01 -1.21629938e-01
5.33808947e-01 -1.03411758e+00 5.20196594e-02 3.30386013e-01
-7.84404874e-01 3.04205418e-01 -6.45761907e-01 6.22702539e-01
-3.85701627e-01 5.50862372e-01 -8.52557719e-02 -9.72770989e-01
2.76501365e-02 -7.19020247e-01 9.57956195e-01 9.52613354e-02
2.84455538e-01 -1.16078043e+00 1.35623917e-01 1.97204918e-01
4.99968588e-01 2.66678125e-01 8.06270540e-01 -6.51485562e-01
-3.01208675e-01 -5.15245497e-01 -1.41873896e-01 5.19498825e-01
4.30769622e-01 -3.82756978e-01 -9.74939227e-01 -4.24897999e-01
4.31941777e-01 -5.22749834e-02 9.83461261e-01 5.82828641e-01
1.01208603e+00 -1.84314221e-01 -3.41390401e-01 7.46134877e-01
1.95427346e+00 4.44557458e-01 5.50971627e-01 1.22549966e-01
5.95002830e-01 2.86846697e-01 4.48955685e-01 6.70614183e-01
-9.14044231e-02 4.92943823e-01 2.04373583e-01 -2.64602125e-01
-4.32445258e-02 -6.33649528e-03 -9.80727151e-02 7.34379232e-01
-3.60553294e-01 2.61383682e-01 -5.91005087e-01 5.04962385e-01
-1.76146638e+00 -1.51212716e+00 -1.28898472e-01 2.17285061e+00
3.50218922e-01 -5.53169012e-01 -3.30694802e-02 6.95935249e-01
6.66967452e-01 4.19098794e-01 -2.80067712e-01 2.37468407e-02
-5.20583332e-01 6.29600942e-01 5.68847060e-01 5.42803645e-01
-1.15862513e+00 6.59180760e-01 6.75328493e+00 8.63338888e-01
-1.22721088e+00 9.44623202e-02 -5.43050747e-03 5.10019422e-01
-3.87853622e-01 7.28003383e-02 -3.34010273e-01 1.06267065e-01
5.91577291e-01 -1.06578201e-01 6.86100543e-01 9.18478310e-01
8.28827322e-02 -1.95490912e-01 -5.84959567e-01 1.38964546e+00
5.48273623e-01 -1.41782737e+00 2.03967392e-01 9.01446864e-02
9.31439638e-01 -1.98310554e-01 5.28413132e-02 -1.00985929e-01
-1.66116983e-01 -9.48374212e-01 5.77891707e-01 9.09626901e-01
7.33884096e-01 -6.49779975e-01 7.38286674e-01 4.09864575e-01
-1.51113844e+00 -2.67460167e-01 -7.96280026e-01 -1.71278879e-01
-3.25641274e-01 9.44119453e-01 -4.37951028e-01 9.42929327e-01
3.22527111e-01 1.18653095e+00 -2.92444646e-01 1.21300626e+00
-1.05958767e-01 7.57134795e-01 -2.47364312e-01 2.75903761e-01
4.45063591e-01 -7.78979301e-01 6.50344133e-01 1.24284887e+00
7.49332786e-01 4.65330273e-01 5.05236208e-01 7.67857432e-01
4.58039582e-01 4.94652778e-01 -9.66999710e-01 -9.48001146e-02
5.79387248e-01 1.39768720e+00 -6.11758769e-01 -4.41567391e-01
-6.32081687e-01 7.75050819e-01 -1.27999321e-01 5.15168548e-01
-1.44626051e-01 -1.00861043e-01 2.32638568e-01 3.18575092e-02
8.22466195e-01 -2.44554266e-01 -1.75151020e-01 -1.20255101e+00
-1.53999880e-01 -8.90277445e-01 2.81216741e-01 -7.92920709e-01
-1.40620494e+00 6.71690285e-01 5.20563908e-02 -1.51228154e+00
8.96559134e-02 -7.10551023e-01 -3.56147587e-01 1.25995934e+00
-1.65886486e+00 -1.44388521e+00 -5.84825397e-01 9.18311000e-01
3.48693728e-01 -9.25688982e-01 1.36570358e+00 -4.69647571e-02
-1.19760998e-01 -1.10345200e-01 5.92329919e-01 -2.68504292e-01
1.63400903e-01 -9.08095002e-01 -7.54943132e-01 7.35907793e-01
3.39170992e-01 3.50375384e-01 4.80572850e-01 -4.38651711e-01
-1.44183469e+00 -1.08418727e+00 7.16157377e-01 2.16186196e-01
6.09247625e-01 2.52872527e-01 -9.35665786e-01 3.00009727e-01
6.27129525e-02 3.95581454e-01 1.33593738e+00 1.40887201e-02
-5.80091894e-01 -2.98025727e-01 -1.31999135e+00 -2.98976332e-01
5.90028644e-01 -7.39367843e-01 -7.15837598e-01 5.50189912e-01
-1.03557386e-01 4.92976345e-02 -1.01600063e+00 2.52701789e-01
3.32042336e-01 -8.48229766e-01 1.03707075e+00 -3.15405905e-01
4.88530658e-02 -6.36047959e-01 -7.19348371e-01 -1.42918122e+00
-9.71117973e-01 4.04272899e-02 1.51033655e-01 6.45840406e-01
1.20468251e-01 -7.08731771e-01 6.68851972e-01 -3.26508552e-01
-1.63721591e-01 -4.15289223e-01 -9.15968537e-01 -9.17796433e-01
-2.22512797e-01 1.13736438e-02 4.04842347e-01 1.29454863e+00
1.32906348e-01 2.08313465e-01 -3.32279861e-01 5.58584750e-01
1.02557063e+00 5.88250697e-01 1.48530379e-01 -1.62367690e+00
-2.06014946e-01 -1.79004923e-01 -8.25718582e-01 -4.16574478e-01
4.84330446e-01 -1.11245644e+00 -2.87020355e-01 -1.46923399e+00
2.83295363e-01 -5.40052772e-01 -3.35700840e-01 6.76591039e-01
3.08510453e-01 3.71386737e-01 -2.48937700e-02 5.75968504e-01
-2.00990029e-02 3.56759012e-01 9.34913397e-01 -3.77877086e-01
-1.76375628e-01 -2.00711742e-01 -5.41325867e-01 5.60652435e-01
5.79735339e-01 -2.82207489e-01 -6.17702901e-01 3.62029485e-02
-2.35714361e-01 3.08627427e-01 4.03785646e-01 -1.36634684e+00
2.56876320e-01 -2.88646013e-01 3.75530094e-01 -4.94098932e-01
4.32593495e-01 -1.16236544e+00 7.59149671e-01 3.23677748e-01
4.39492464e-02 -6.46376848e-01 9.22887698e-02 6.47455633e-01
-7.00182974e-01 -5.14621198e-01 1.00804186e+00 -2.72027820e-01
-8.65694642e-01 2.09393367e-01 -4.59149003e-01 -6.59566283e-01
1.05386972e+00 -6.02336049e-01 1.46397933e-01 -1.67557240e-01
-1.04527414e+00 -6.37551844e-01 4.77198750e-01 -1.45951539e-01
8.22468698e-01 -1.56273782e+00 -8.70468140e-01 6.49626493e-01
3.69758457e-01 -3.63426328e-01 3.97241533e-01 6.32071733e-01
-4.88308579e-01 5.50060332e-01 -4.70150322e-01 -7.18516469e-01
-1.07569146e+00 8.05885315e-01 3.41796577e-01 2.74794511e-02
-7.18379736e-01 5.27635276e-01 3.91150787e-02 -5.08343056e-02
-1.06913529e-01 -2.80227885e-02 -5.12472212e-01 2.59041190e-01
6.37988806e-01 1.60809532e-01 1.92063734e-01 -1.22597909e+00
-3.83793622e-01 9.96867537e-01 5.35580158e-01 2.31714353e-01
1.68715203e+00 3.81957978e-01 -5.77737153e-01 3.35293591e-01
1.05320239e+00 1.79480478e-01 -8.09651613e-01 -6.41847014e-01
-2.67346770e-01 -7.58529782e-01 4.86011833e-01 -6.51127577e-01
-1.18443847e+00 5.05299509e-01 6.44982278e-01 3.02243173e-01
1.47873509e+00 -2.52544284e-01 1.88763887e-01 3.62076908e-01
6.94342196e-01 -6.95941389e-01 -2.20331699e-01 4.01035249e-01
1.02669239e+00 -1.03321028e+00 1.62231550e-01 -7.21194625e-01
-2.42971182e-01 1.31338394e+00 -2.51468956e-01 -5.31005502e-01
1.05672848e+00 1.28018796e-01 -2.48173684e-01 -2.72834778e-01
-2.83395469e-01 -4.30634290e-01 5.86882710e-01 9.05798972e-01
2.81340927e-01 2.27739260e-01 -2.79086143e-01 4.11120474e-01
5.71709648e-02 4.50847112e-03 9.24361050e-02 9.54898238e-01
-9.85761940e-01 -1.05689657e+00 -8.77199531e-01 4.31393266e-01
-2.61415727e-02 -6.85818940e-02 -2.08050251e-01 3.85426074e-01
3.49178016e-01 8.77617121e-01 -1.61001310e-01 -3.04474950e-01
2.32847720e-01 1.16937727e-01 5.92620790e-01 -8.55631411e-01
-5.43543026e-02 2.18273938e-01 -1.08528227e-01 -3.29219699e-01
-8.73857498e-01 -7.51183748e-01 -8.79756212e-01 -1.61970332e-02
-1.99280396e-01 4.69968498e-01 6.91528857e-01 8.07765543e-01
7.24947453e-02 2.30328307e-01 8.27967286e-01 -1.23411310e+00
-3.21197778e-01 -8.26012671e-01 -1.24906921e+00 1.55918777e-01
3.51228297e-01 -8.73841047e-01 -6.20220900e-01 4.51238900e-01] | [12.354206085205078, 0.30551379919052124] |
6d6fe063-a1e3-4304-9fa8-75da39436fd3 | single-image-super-resolution-based-on | 2210.03743 | null | https://arxiv.org/abs/2210.03743v1 | https://arxiv.org/pdf/2210.03743v1.pdf | Single Image Super-Resolution Based on Capsule Neural Networks | Single image super-resolution (SISR) is the process of obtaining one high-resolution version of a low-resolution image by increasing the number of pixels per unit area. This method has been actively investigated by the research community, due to the wide variety of real-world problems where it can be applied, from aerial and satellite imaging to compressed image and video enhancement. Despite the improvements achieved by deep learning in the field, the vast majority of the used networks are based on traditional convolutions, with the solutions focusing on going deeper and/or wider, and innovations coming from jointly employing successful concepts from other fields. In this work, we decided to step up from the traditional convolutions and adopt the concept of capsules. Since their overwhelming results both in image classification and segmentation problems, we question how suitable they are for SISR. We also verify that different solutions share most of their configurations, and argue that this trend leads to fewer explorations of network varieties. During our experiments, we check various strategies to improve results, ranging from new and different loss functions to changes in the capsule layers. Our network achieved good results with fewer convolutional-based layers, showing that capsules might be a concept worth applying in the image super-resolution problem. | ['Helio Pedrini', 'George Corrêa de Araújo'] | 2022-10-06 | null | null | null | null | ['video-enhancement'] | ['computer-vision'] | [ 4.42534983e-01 1.85384482e-01 1.56396732e-01 -1.40490249e-01
-2.17485741e-01 -3.34542811e-01 7.38319635e-01 -2.80287415e-01
-5.95460236e-01 9.08430994e-01 1.53110862e-01 1.74341910e-02
-5.02889752e-01 -9.28038776e-01 -5.75299621e-01 -7.73118734e-01
-3.31724674e-01 1.14991784e-01 5.97442448e-01 -6.00049734e-01
2.15674952e-01 7.58281052e-01 -1.70104790e+00 4.40968841e-01
6.62890077e-01 7.68925428e-01 5.05763292e-01 3.69603336e-01
-2.48408675e-01 6.39145792e-01 -4.75200474e-01 -3.08575481e-01
4.20632839e-01 -2.68064499e-01 -8.65688324e-01 1.19668908e-01
1.69216201e-01 3.00384965e-02 -4.56913747e-02 1.14764488e+00
3.88473749e-01 -1.08376443e-01 2.42265686e-01 -5.00615537e-01
-5.51147461e-01 6.85604095e-01 -5.87717175e-01 4.14821655e-01
2.93014646e-01 -1.70588911e-01 4.92227465e-01 -6.26528323e-01
7.88680971e-01 1.24182463e+00 8.83988857e-01 4.38608944e-01
-1.21907854e+00 -3.50718498e-01 7.12828487e-02 1.13112874e-01
-1.40671921e+00 -1.41959801e-01 5.55145860e-01 -2.71553904e-01
5.81075251e-01 3.95894170e-01 6.22107565e-01 9.57951844e-01
8.38420726e-03 3.14112395e-01 1.63255084e+00 -3.62178504e-01
7.31950626e-02 5.09030282e-01 -4.25316542e-01 2.10287467e-01
4.84507799e-01 8.92414749e-02 -2.56040603e-01 2.93292314e-01
1.20710301e+00 -3.47266123e-02 -5.52806616e-01 -2.08129391e-01
-1.06342661e+00 8.16673040e-01 5.62486053e-01 1.10114229e+00
-3.76567960e-01 -2.29335472e-01 1.19166546e-01 3.21304172e-01
5.82554698e-01 8.09571147e-01 -3.55512977e-01 1.91215828e-01
-1.25270927e+00 2.65733659e-01 5.64148545e-01 3.84333909e-01
6.61010683e-01 7.46037513e-02 1.33589089e-01 7.50871062e-01
-1.47536024e-01 -1.23506829e-01 6.30450249e-01 -1.09323180e+00
-2.89476663e-02 6.76374733e-01 6.42390549e-02 -9.34275091e-01
-5.28417647e-01 -7.84288704e-01 -9.88028347e-01 7.93637216e-01
5.61508715e-01 -9.03156959e-03 -8.49519193e-01 1.35559082e+00
7.30770379e-02 -5.01545370e-02 7.54673360e-03 1.12385356e+00
5.64322114e-01 3.22351396e-01 -1.93712741e-01 -1.89660743e-01
1.40419817e+00 -7.06997395e-01 -5.94618857e-01 -2.51763850e-01
1.92794457e-01 -7.63266265e-01 6.71410024e-01 7.48509884e-01
-1.17349112e+00 -7.05685675e-01 -1.00266540e+00 1.62744418e-01
-6.57004237e-01 -9.78750512e-02 7.28350163e-01 5.73848426e-01
-1.37637389e+00 1.27884114e+00 -5.56965172e-01 -4.99188602e-01
4.97689843e-01 3.40693146e-01 -6.11276805e-01 3.19466814e-02
-1.12921584e+00 1.03225064e+00 4.77353632e-01 1.87352061e-01
-3.66716176e-01 -5.67552745e-01 -3.90819639e-01 2.48727426e-01
4.20672804e-01 -5.36871552e-01 6.97592854e-01 -1.24272704e+00
-1.36046827e+00 8.52246225e-01 2.52665609e-01 -8.27413857e-01
7.68604994e-01 6.64161295e-02 -1.87465057e-01 1.83184281e-01
-3.79556447e-01 6.23096526e-01 9.44161713e-01 -1.45068109e+00
-6.40358627e-01 -3.40144038e-01 2.75754631e-01 -3.28295454e-02
-3.41219723e-01 1.02202654e-01 7.72625161e-03 -6.58107162e-01
2.66403824e-01 -7.08933234e-01 -5.32130957e-01 -5.57069220e-02
-4.43230607e-02 6.43752292e-02 6.76984549e-01 -7.66411722e-01
9.03606117e-01 -1.95427263e+00 3.79873872e-01 -7.98303857e-02
2.64095724e-01 4.94051665e-01 -1.81405637e-02 2.09337413e-01
-4.10141140e-01 4.41408068e-01 -4.12816495e-01 -3.36865097e-01
-4.53916699e-01 1.32513121e-01 -1.23446375e-01 4.17437196e-01
2.82001644e-01 6.85549796e-01 -5.43344617e-01 -5.19910336e-01
3.37266713e-01 1.13347924e+00 -1.76874891e-01 -1.33081734e-01
-2.66216416e-02 7.68757701e-01 -3.00653607e-01 3.39830995e-01
9.58818197e-01 -2.40350440e-01 5.31785004e-03 -2.51073271e-01
-5.09507120e-01 -7.74533376e-02 -1.37326837e+00 1.56162286e+00
-4.29710716e-01 5.88836610e-01 3.87908787e-01 -1.26265430e+00
9.58905101e-01 2.73615092e-01 6.28049433e-01 -8.54991436e-01
1.89560056e-02 3.04068506e-01 -2.41653025e-02 -4.88063067e-01
6.04243636e-01 -3.16411495e-01 4.74183857e-01 -7.34644085e-02
-4.91833948e-02 -8.96703079e-02 2.79554218e-01 -2.23976806e-01
8.38460386e-01 3.72347295e-01 2.59411603e-01 -3.19620758e-01
6.71466768e-01 -6.03610352e-02 1.98039934e-01 5.67597032e-01
1.62940890e-01 9.58212435e-01 3.82912308e-01 -5.90288758e-01
-1.22952580e+00 -7.04301238e-01 -4.01429325e-01 6.43498063e-01
6.62337318e-02 -2.46967468e-02 -8.48512530e-01 -2.10421607e-01
-3.96385908e-01 9.62260738e-02 -8.09719980e-01 2.41105512e-01
-6.60606265e-01 -1.04293513e+00 2.67306358e-01 3.15730155e-01
7.97113776e-01 -1.24516869e+00 -1.02157640e+00 1.90060079e-01
-4.74675931e-03 -1.17888844e+00 4.23815250e-01 3.32327217e-01
-1.23215616e+00 -8.96428049e-01 -1.17998636e+00 -5.17596066e-01
2.95428723e-01 3.19755435e-01 1.09420729e+00 1.03704996e-01
-4.24175590e-01 8.26560482e-02 -6.51921391e-01 -1.59013525e-01
-5.06127656e-01 2.15809032e-01 -3.15085799e-01 -4.35689986e-02
-9.29230452e-02 -7.78233707e-01 -5.98764062e-01 8.10980424e-02
-1.19740081e+00 -2.50295959e-02 9.08583939e-01 6.04668796e-01
4.54826087e-01 3.51452440e-01 3.88412952e-01 -9.40921128e-01
5.56681335e-01 -1.85235083e-01 -6.50797009e-01 1.08673617e-01
-6.76628649e-01 1.59356177e-01 6.84143245e-01 -4.11736876e-01
-1.05126584e+00 3.17926928e-02 -3.40407401e-01 -3.30926150e-01
-4.79901373e-01 3.46685767e-01 1.53336957e-01 -4.01260257e-01
7.18042076e-01 2.73909897e-01 1.23524785e-01 -6.66963339e-01
2.02428624e-01 3.35003376e-01 3.61151993e-01 -6.22904599e-02
8.34377527e-01 8.50986302e-01 1.31248608e-01 -8.90108466e-01
-6.28229260e-01 -3.10414910e-01 -6.07127666e-01 -1.10052668e-01
1.19084108e+00 -6.64217532e-01 -5.77262521e-01 2.42568642e-01
-9.65167701e-01 -6.56150430e-02 -4.34733808e-01 4.72944111e-01
-2.70109892e-01 4.28163022e-01 -4.50074285e-01 -8.87027621e-01
-8.36825818e-02 -1.27393365e+00 8.74608099e-01 6.25407636e-01
1.45251900e-01 -1.02632284e+00 -1.36116788e-01 2.25302637e-01
1.09107161e+00 4.33607966e-01 4.71369922e-01 -2.07450569e-01
-7.81733990e-01 6.25888910e-03 -3.85394752e-01 5.83908737e-01
6.51250929e-02 -3.08043033e-01 -1.18958735e+00 -4.22515452e-01
3.24853301e-01 -1.19199902e-02 1.15172553e+00 3.77311826e-01
1.26443338e+00 -5.27115120e-03 -2.39782825e-01 8.47578168e-01
1.92732596e+00 3.55393626e-02 1.19677985e+00 7.53804266e-01
2.45353401e-01 8.41299951e-01 2.94483483e-01 2.59155840e-01
-1.61921740e-01 8.69128346e-01 9.03079510e-01 -5.81636965e-01
-3.35844040e-01 3.45554143e-01 1.56281143e-01 1.91350877e-01
-8.20292413e-01 1.39638543e-01 -4.30208921e-01 4.70373780e-01
-1.50147200e+00 -1.16612566e+00 -2.11682156e-01 2.28217530e+00
5.88853478e-01 2.42687643e-01 1.33841857e-01 1.49778679e-01
6.03630543e-01 2.73759782e-01 -2.50270247e-01 -3.41114223e-01
-5.19936562e-01 4.81065154e-01 6.33042157e-01 3.40094030e-01
-1.09565771e+00 6.53223038e-01 5.99379587e+00 8.17346573e-01
-1.46652865e+00 1.01597525e-01 6.64184451e-01 1.21559836e-01
-1.26247212e-01 -2.08602399e-01 -6.10614479e-01 3.88136685e-01
7.05773652e-01 1.04402088e-01 7.01599121e-01 6.34022892e-01
2.03546032e-01 -2.82386959e-01 -5.74384451e-01 8.52036715e-01
1.42546380e-02 -1.38439715e+00 -1.17185349e-02 1.92325696e-01
7.03048587e-01 1.03604101e-01 4.96740825e-02 4.30448055e-02
-1.27775997e-01 -1.40388179e+00 4.87864941e-01 7.31481016e-01
5.25724471e-01 -5.98332942e-01 9.49924767e-01 1.07303694e-01
-1.17394769e+00 -2.25318238e-01 -5.28915942e-01 -1.27643257e-01
-8.78925435e-03 5.54743767e-01 -4.81945246e-01 8.94896328e-01
1.07872701e+00 4.17061985e-01 -8.32501054e-01 1.18680727e+00
7.35784769e-02 2.80613005e-01 -2.37947941e-01 2.32302874e-01
1.46927536e-01 -3.94443065e-01 5.46778500e-01 1.25625706e+00
4.59871650e-01 -1.92804366e-01 -1.78104043e-01 1.20155382e+00
2.84099370e-01 7.81403575e-03 -6.31642759e-01 1.94684014e-01
5.63007705e-02 1.59151781e+00 -1.06731510e+00 -8.31385106e-02
-5.06372094e-01 8.79327416e-01 7.33603686e-02 3.22480142e-01
-7.85205603e-01 -1.69995159e-01 5.28220177e-01 4.58258390e-01
6.42434776e-01 -1.19617566e-01 -4.03607041e-01 -8.94576550e-01
2.89520621e-02 -7.73480177e-01 5.83264194e-02 -8.36718082e-01
-9.88126218e-01 1.09727645e+00 7.49149993e-02 -1.32145035e+00
3.44456248e-02 -7.22962677e-01 -4.28628445e-01 9.72908139e-01
-1.98139167e+00 -8.97786677e-01 -4.31210965e-01 4.14384842e-01
5.76376557e-01 -1.27979051e-02 6.59939051e-01 4.82964426e-01
-8.97367522e-02 -1.18239624e-02 -4.42267023e-02 -1.21119246e-01
5.03229797e-01 -1.17711055e+00 1.68812256e-02 8.72751355e-01
-1.40981982e-03 5.18619061e-01 9.25100505e-01 -1.52844176e-01
-8.55832994e-01 -7.44176269e-01 5.82305253e-01 -9.44443494e-02
4.10670519e-01 -1.32628173e-01 -1.23822892e+00 2.07109541e-01
5.01955211e-01 -7.62670860e-03 4.72863279e-02 -1.12798445e-01
-3.76976444e-03 -1.02622725e-01 -1.26355588e+00 4.44590271e-01
7.59977520e-01 -4.90006991e-02 -4.86781448e-01 1.00695556e-02
7.43459821e-01 -1.83400646e-01 -1.08943510e+00 5.69086134e-01
3.53788823e-01 -1.48266768e+00 1.21519303e+00 -1.50016695e-01
7.01200664e-01 -3.85163099e-01 1.53702885e-01 -1.34730601e+00
-3.69733870e-01 -1.03423126e-01 2.83150971e-01 1.02647507e+00
3.68450850e-01 -6.07070506e-01 8.03176403e-01 1.97467819e-01
-5.81259020e-02 -8.28971148e-01 -1.02543032e+00 -7.48220444e-01
1.67252377e-01 -7.33437240e-02 4.61066931e-01 1.06802750e+00
-4.21728611e-01 9.29974169e-02 -3.80007207e-01 1.36412978e-01
5.75024605e-01 1.04042597e-01 4.15267885e-01 -1.47007513e+00
-3.80720019e-01 -6.94355071e-01 -3.01396638e-01 -5.37963927e-01
-3.66232395e-01 -7.28834867e-01 -2.81512082e-01 -1.85933447e+00
8.32119510e-02 -3.02947670e-01 -1.57079417e-02 2.93574572e-01
2.33958773e-02 5.61080158e-01 3.50757509e-01 1.02423072e-01
-3.20253938e-01 6.07512668e-02 1.46791184e+00 4.72297370e-02
-2.18168512e-01 -4.27967049e-02 -7.06248403e-01 8.43442976e-01
8.94916713e-01 -1.93887368e-01 -3.21234167e-02 -3.62126231e-01
4.43149984e-01 -7.08016530e-02 5.66898823e-01 -1.27925932e+00
-1.69598889e-02 1.73754901e-01 5.11276484e-01 -2.07185328e-01
4.42531168e-01 -1.02602208e+00 3.20093930e-01 5.54707468e-01
-8.55095983e-02 -1.39614791e-01 1.43151671e-01 1.73844039e-01
-3.80785584e-01 -5.12002468e-01 1.10453236e+00 -8.08224499e-01
-8.71700883e-01 -1.91949215e-02 -2.25925982e-01 -3.40856493e-01
7.84170270e-01 -6.75134420e-01 -2.15885267e-01 -2.49802679e-01
-1.10507607e+00 -2.14717805e-01 5.66836953e-01 4.32737321e-01
5.02294123e-01 -7.07254291e-01 -8.20210397e-01 -1.26164872e-02
-2.35638797e-01 8.16672593e-02 3.11338335e-01 1.02752757e+00
-5.62607944e-01 4.10412878e-01 -4.90420938e-01 -5.08265555e-01
-1.24170887e+00 6.74362779e-01 6.92103505e-01 -4.89737213e-01
-8.96266699e-01 5.50550222e-01 2.93746963e-02 -8.00070912e-02
1.29159704e-01 -3.88409495e-01 -7.45908737e-01 9.78802294e-02
6.20009840e-01 3.06327194e-01 1.50276974e-01 -4.22749460e-01
-1.01894356e-01 7.11445928e-01 1.35867655e-01 1.38739020e-01
1.79114306e+00 -2.43540660e-01 -3.36793840e-01 1.84999213e-01
8.31769705e-01 -1.18348496e-02 -1.36728406e+00 6.16205260e-02
1.31643638e-01 -4.35756624e-01 9.20314416e-02 -8.56920004e-01
-1.35080326e+00 8.07215095e-01 9.40753460e-01 6.72283351e-01
1.34432554e+00 9.42280516e-02 3.05097282e-01 1.02231957e-01
6.76854968e-01 -8.67674351e-01 -1.20338701e-01 2.97982961e-01
9.38255548e-01 -1.22427523e+00 2.42284402e-01 -4.21813607e-01
-4.08940017e-01 1.34595215e+00 3.63289535e-01 -3.67460966e-01
3.58743817e-01 6.27180755e-01 -1.17605902e-01 -1.20300509e-01
-3.44572246e-01 -5.41428149e-01 -8.72101933e-02 7.28037715e-01
6.24193490e-01 -2.72926837e-01 -7.39046395e-01 3.65681976e-01
-9.22044218e-02 3.43433291e-01 8.44649613e-01 4.69358891e-01
-5.69452226e-01 -1.23533475e+00 -6.66964173e-01 3.05221677e-01
-6.83108091e-01 3.79123688e-02 -2.24411160e-01 1.09257245e+00
5.29563189e-01 6.91407979e-01 -1.48579001e-01 -1.56429335e-01
2.78669745e-01 -2.30782539e-01 6.39778316e-01 -3.39452773e-01
-7.49311924e-01 3.28951292e-02 -1.63733065e-01 -5.60447097e-01
-9.71996248e-01 -6.54133499e-01 -9.75994170e-01 -3.52176204e-02
-2.70613283e-01 1.48481220e-01 8.56997192e-01 9.30632055e-01
-1.75868557e-03 9.77600753e-01 4.09546524e-01 -1.21509373e+00
-3.08928758e-01 -1.07078183e+00 -6.84478939e-01 2.92985171e-01
2.53208637e-01 -5.42567372e-01 -4.25665647e-01 8.64797682e-02] | [10.641520500183105, -1.97036612033844] |
429f8b4d-857f-4983-b16a-d672ec17ac1f | class-weighted-convolutional-features-for | 1707.02581 | null | http://arxiv.org/abs/1707.02581v1 | http://arxiv.org/pdf/1707.02581v1.pdf | Class-Weighted Convolutional Features for Visual Instance Search | Image retrieval in realistic scenarios targets large dynamic datasets of
unlabeled images. In these cases, training or fine-tuning a model every time
new images are added to the database is neither efficient nor scalable.
Convolutional neural networks trained for image classification over large
datasets have been proven effective feature extractors for image retrieval. The
most successful approaches are based on encoding the activations of
convolutional layers, as they convey the image spatial information. In this
paper, we go beyond this spatial information and propose a local-aware encoding
of convolutional features based on semantic information predicted in the target
image. To this end, we obtain the most discriminative regions of an image using
Class Activation Maps (CAMs). CAMs are based on the knowledge contained in the
network and therefore, our approach, has the additional advantage of not
requiring external information. In addition, we use CAMs to generate object
proposals during an unsupervised re-ranking stage after a first fast search.
Our experiments on two public available datasets for instance retrieval,
Oxford5k and Paris6k, demonstrate the competitiveness of our approach
outperforming the current state-of-the-art when using off-the-shelf models
trained on ImageNet. The source code and model used in this paper are publicly
available at http://imatge-upc.github.io/retrieval-2017-cam/. | ['Xavier Giro-i-Nieto', 'Jose M. Alvarez', 'Albert Jimenez'] | 2017-07-09 | null | null | null | null | ['instance-search'] | ['computer-vision'] | [ 5.67013063e-02 -3.42488348e-01 -2.84250170e-01 -5.03828287e-01
-8.70647371e-01 -7.25691259e-01 7.50345826e-01 3.87233734e-01
-9.36119616e-01 4.56794232e-01 -1.40895113e-01 3.33647691e-02
-3.26092511e-01 -8.43566597e-01 -9.91242707e-01 -5.62238574e-01
9.52566341e-02 6.36261284e-01 3.51142406e-01 -2.32224599e-01
4.43709791e-01 7.23143160e-01 -1.86786354e+00 5.52267194e-01
5.95917881e-01 1.36421800e+00 5.92622280e-01 5.19356847e-01
-2.72953212e-02 3.75366867e-01 -4.07412618e-01 -2.23294616e-01
4.24650937e-01 3.53008918e-02 -8.46003592e-01 -2.60330379e-01
5.28816044e-01 -4.04983908e-01 -5.51359713e-01 8.09074759e-01
4.67648476e-01 1.04754314e-01 6.23071373e-01 -8.64699841e-01
-6.14366114e-01 2.21784130e-01 -2.07807183e-01 2.99249679e-01
3.23419981e-02 7.26182535e-02 1.06889009e+00 -1.20997345e+00
9.63495612e-01 9.21597838e-01 1.39201358e-01 3.63767058e-01
-9.28218484e-01 -5.24352312e-01 1.45884559e-01 5.65959096e-01
-1.60254586e+00 -3.66169751e-01 6.79829299e-01 -2.77865022e-01
9.06876266e-01 3.10600489e-01 5.95044613e-01 8.20744216e-01
2.35722610e-03 1.08829010e+00 7.90474534e-01 -4.35273230e-01
1.20014235e-01 1.76217586e-01 1.31767511e-01 6.67449176e-01
7.65143111e-02 1.72049686e-01 -4.89125848e-01 -5.36454991e-02
5.03090084e-01 3.41061950e-01 -2.77443886e-01 -5.99434078e-01
-1.26177287e+00 7.47656524e-01 1.17053628e+00 4.39509660e-01
-5.65120399e-01 3.26880395e-01 3.68510485e-01 4.03855145e-01
3.84835333e-01 4.92810041e-01 -5.40939808e-01 2.89676249e-01
-1.11667418e+00 3.30772191e-01 3.77855897e-01 6.89091921e-01
1.02298594e+00 -7.15827823e-01 -1.76165298e-01 8.61239910e-01
2.91464776e-01 3.86479199e-01 7.76287138e-01 -5.91399491e-01
3.70144755e-01 8.17996323e-01 9.65798497e-02 -1.02585590e+00
-3.00069898e-01 -5.86583376e-01 -6.60560489e-01 -1.23988949e-01
1.84910640e-01 3.85724515e-01 -1.32890546e+00 1.34689236e+00
2.59655744e-01 8.72441232e-02 4.73954901e-02 1.00606084e+00
8.35491836e-01 6.82707846e-01 -1.24662727e-01 3.64695966e-01
1.19325733e+00 -1.11936641e+00 -2.83814996e-01 -1.92966416e-01
6.19714975e-01 -8.99821818e-01 8.54317248e-01 2.75033087e-01
-8.60539615e-01 -7.75201976e-01 -9.66590047e-01 -1.73238050e-02
-9.37988520e-01 5.86629987e-01 4.90628242e-01 1.16809651e-01
-1.33393180e+00 5.38588107e-01 -5.88672221e-01 -4.62193191e-01
5.56502342e-01 6.03317559e-01 -4.76987779e-01 -3.80516946e-01
-1.26111686e+00 6.95746005e-01 7.05980778e-01 3.81858051e-01
-1.00057352e+00 -4.42754745e-01 -5.84691823e-01 3.05933803e-01
4.05407488e-01 -3.17682713e-01 1.16389167e+00 -1.21960545e+00
-1.15996420e+00 9.18034136e-01 7.40223899e-02 -6.43064737e-01
5.18863022e-01 -2.41111889e-01 -1.44447103e-01 4.77646321e-01
2.77435221e-02 1.19579208e+00 9.02749360e-01 -1.19077289e+00
-5.05429685e-01 -2.63986796e-01 2.65148222e-01 3.59188989e-02
-5.78070343e-01 -7.30885342e-02 -1.10139322e+00 -5.10701835e-01
-6.33502081e-02 -1.14645016e+00 -2.73888499e-01 4.44693081e-02
-3.48953664e-01 -3.21766376e-01 7.71671176e-01 -4.07132387e-01
1.02243710e+00 -1.96046531e+00 1.83020145e-01 5.23705006e-01
-9.34278741e-02 6.91788018e-01 -5.37775517e-01 5.16046166e-01
9.52575356e-02 -1.71773992e-02 -3.02916616e-02 -1.49145216e-01
-1.02809638e-01 2.46997625e-02 -2.02632442e-01 3.35244715e-01
3.79459918e-01 1.32984293e+00 -7.25271165e-01 -4.14420098e-01
4.60272402e-01 4.95516390e-01 -6.05166912e-01 6.40120283e-02
-2.89314121e-01 3.33875597e-01 -5.63430309e-01 5.25789678e-01
5.09857595e-01 -4.99385744e-01 5.77101000e-02 -2.79833496e-01
3.11110429e-02 3.80740352e-02 -9.72383380e-01 2.04019332e+00
-5.47892213e-01 6.08215809e-01 -3.88310641e-01 -1.22787213e+00
7.73999333e-01 1.34848468e-02 3.72070819e-01 -1.23631454e+00
-3.90689932e-02 5.63484609e-01 -2.13497579e-01 -2.48361781e-01
6.34968281e-01 7.52166629e-01 -1.44917697e-01 3.10874820e-01
3.02121103e-01 3.00835222e-01 5.01176655e-01 4.01942968e-01
9.38384533e-01 4.40763123e-02 1.40120700e-01 -2.53109038e-01
7.87644744e-01 8.22496414e-02 8.08515176e-02 1.00830686e+00
1.87503770e-01 7.81404614e-01 1.30847707e-01 -7.15429962e-01
-1.00129938e+00 -6.45052850e-01 -2.00043514e-01 9.41619396e-01
3.40459377e-01 -4.41126049e-01 -6.55010641e-01 -7.53983438e-01
9.44409519e-02 1.10591911e-01 -8.55466485e-01 -1.78304285e-01
-6.50516748e-01 -4.21118826e-01 2.82628477e-01 3.82898957e-01
5.93067706e-01 -1.33177567e+00 -7.46887207e-01 2.21197948e-01
-6.84457943e-02 -9.73527968e-01 -3.07094038e-01 1.38523921e-01
-6.64992392e-01 -1.14470243e+00 -1.14727724e+00 -8.78097594e-01
9.74245727e-01 3.22851270e-01 1.08620775e+00 5.68282247e-01
-7.21222937e-01 5.14942348e-01 -5.07206142e-01 -2.97507256e-01
-2.70750616e-02 5.96740842e-01 -3.66319239e-01 1.58042908e-01
3.16523850e-01 -9.45453346e-02 -1.18483853e+00 4.29601520e-01
-1.21386182e+00 -1.05155535e-01 8.89250278e-01 9.52099502e-01
9.22657788e-01 -2.40662768e-01 3.99011850e-01 -8.27840030e-01
3.83842885e-01 -4.54001844e-01 -8.82870972e-01 5.92789054e-01
-5.81140220e-01 2.14543357e-01 4.24953371e-01 -2.55436122e-01
-7.63482749e-01 3.36004615e-01 4.36611660e-03 -5.53717256e-01
-1.84656426e-01 6.23396218e-01 2.76803374e-01 -1.97890952e-01
5.29395759e-01 4.30469871e-01 -3.22090864e-01 -6.65116429e-01
3.61534208e-01 6.36262596e-01 3.24122727e-01 -3.05749476e-01
6.85904264e-01 4.84667093e-01 -2.36684501e-01 -4.51993436e-01
-7.16988444e-01 -7.84077823e-01 -8.08859110e-01 -2.61665732e-01
5.63776612e-01 -9.22346950e-01 -5.69229901e-01 4.58468616e-01
-9.95926499e-01 -3.08384389e-01 -1.58345535e-01 3.15915108e-01
-2.71458268e-01 1.07121475e-01 -3.45928937e-01 -2.80037135e-01
-5.15463889e-01 -1.32984388e+00 1.41390181e+00 3.38090211e-01
2.67811358e-01 -5.99918902e-01 -7.01403096e-02 2.84245431e-01
7.32493758e-01 -2.03809127e-01 6.37427926e-01 -8.60702693e-01
-1.01832354e+00 -6.19288564e-01 -4.59356129e-01 2.78020561e-01
-2.11534962e-01 -4.17341739e-02 -9.35174346e-01 -4.77751404e-01
-7.77016759e-01 -6.15938663e-01 1.32440794e+00 2.88871080e-01
1.78703988e+00 -2.46849507e-01 -5.14102280e-01 4.48178738e-01
1.66683650e+00 -9.34842601e-02 6.32695138e-01 5.75448990e-01
3.84882897e-01 5.28046489e-01 9.11968708e-01 2.84415632e-01
1.45811900e-01 9.70537901e-01 6.69533968e-01 -1.92988560e-01
-2.01465145e-01 -9.67899188e-02 -4.66675386e-02 4.97056782e-01
1.12136334e-01 -5.45409679e-01 -9.77082849e-01 7.76333988e-01
-1.93214953e+00 -6.22590721e-01 3.51111948e-01 2.18328238e+00
5.85807025e-01 2.90532671e-02 -3.45447809e-01 -1.97815925e-01
3.90112191e-01 7.08460659e-02 -4.96367455e-01 3.20593943e-03
-1.51223503e-02 3.94454092e-01 6.80995226e-01 3.05752963e-01
-1.25629807e+00 1.09536839e+00 4.85077286e+00 1.04518270e+00
-1.29524493e+00 4.20752428e-02 7.95239687e-01 -1.98646128e-01
-7.52753541e-02 -1.52060494e-01 -8.71257544e-01 3.84157389e-01
7.83060849e-01 9.76311788e-02 2.51036912e-01 7.96398461e-01
-1.94024608e-01 -2.01755121e-01 -1.00336897e+00 9.04464781e-01
1.42765105e-01 -1.60331213e+00 3.50480020e-01 6.43956140e-02
7.60488689e-01 6.00498438e-01 2.62200773e-01 2.96338618e-01
-6.54357076e-02 -9.41311657e-01 6.41757965e-01 7.10763633e-01
7.16416776e-01 -6.34473383e-01 9.42419708e-01 1.05879731e-01
-1.06995094e+00 -9.60954502e-02 -6.17008388e-01 5.73558569e-01
-2.52040863e-01 4.31672424e-01 -7.66624689e-01 4.96139914e-01
9.38370824e-01 7.64467239e-01 -1.09324956e+00 1.35728717e+00
-1.89098030e-01 2.95876741e-01 -4.45399791e-01 6.11858442e-02
6.39754117e-01 1.28187105e-01 1.50959894e-01 1.19993234e+00
2.51684606e-01 -2.72597581e-01 1.74545348e-01 5.73983192e-01
-4.43000555e-01 3.65556031e-01 -5.43549359e-01 -1.21825989e-02
2.76723623e-01 1.45303369e+00 -8.67506862e-01 -4.81269032e-01
-1.78845957e-01 1.08731341e+00 4.50253427e-01 4.71594274e-01
-5.84249377e-01 -3.64418894e-01 3.76210719e-01 -4.29639295e-02
6.57350719e-01 6.58482313e-02 4.11623508e-01 -1.11283112e+00
1.94043174e-01 -8.04354072e-01 4.45054293e-01 -6.77282214e-01
-9.74339068e-01 7.93211579e-01 -3.65847908e-02 -1.15765762e+00
-3.89414907e-01 -7.60090828e-01 -1.41840398e-01 6.70629382e-01
-1.99670255e+00 -1.13307369e+00 -5.61047554e-01 6.60242081e-01
6.81320250e-01 -1.86449766e-01 8.38220954e-01 5.59656918e-01
-2.24061787e-01 5.79361916e-01 4.76158619e-01 2.48454109e-01
7.53086388e-01 -9.16191101e-01 3.15470368e-01 4.59666044e-01
6.11036122e-01 5.56508362e-01 1.46106884e-01 -3.04560721e-01
-1.29168284e+00 -9.93850231e-01 7.22088039e-01 -1.82270184e-01
4.26722497e-01 -4.57782507e-01 -9.18513834e-01 3.75278234e-01
1.83799490e-01 4.59485888e-01 3.53576928e-01 -1.75865963e-01
-3.69585156e-01 -3.45578760e-01 -8.17735434e-01 2.78976262e-01
7.69069433e-01 -4.41955835e-01 -9.33029279e-02 5.42633533e-01
4.96384054e-01 -3.64038438e-01 -7.18495071e-01 5.79618990e-01
7.05223918e-01 -7.74346232e-01 1.17528760e+00 -5.35141945e-01
3.04801911e-01 -2.67151773e-01 -1.43952280e-01 -1.01401973e+00
-1.28618538e-01 2.49570422e-02 1.98906288e-01 7.89314747e-01
5.38003504e-01 -5.53378046e-01 7.40140438e-01 1.49722844e-01
1.44516751e-01 -8.80289137e-01 -7.28263855e-01 -5.11368394e-01
-1.80740595e-01 -1.45443439e-01 5.31685472e-01 5.57034016e-01
-5.85277200e-01 -8.15714300e-02 -1.65418237e-01 2.10837331e-02
3.69299173e-01 3.96684259e-01 7.44006395e-01 -1.31859863e+00
-2.03580216e-01 -3.57242346e-01 -5.77364862e-01 -9.64508414e-01
2.26230189e-01 -1.04867363e+00 2.34651249e-02 -1.54870248e+00
3.85036081e-01 -7.38317728e-01 -9.09594417e-01 6.98105156e-01
-4.67799753e-02 7.06294775e-01 4.06451821e-01 7.05197394e-01
-1.11691964e+00 4.39809054e-01 9.11013901e-01 -4.13736820e-01
3.39030661e-02 -1.26130164e-01 -2.18758523e-01 3.59520823e-01
1.00376940e+00 -5.37348688e-01 -3.81407171e-01 -6.25901520e-01
2.53432631e-01 -3.83415312e-01 6.29053712e-01 -1.06127322e+00
3.94175589e-01 3.22655320e-01 5.67776084e-01 -7.23989248e-01
4.34336603e-01 -9.21442211e-01 -2.08692229e-03 6.39094889e-01
-6.29825234e-01 2.22338978e-02 2.75498241e-01 6.52458429e-01
-6.53703690e-01 -4.18237507e-01 4.26875800e-01 -2.33999297e-01
-1.19169760e+00 4.81233984e-01 -1.25528723e-01 -3.72718394e-01
8.99449229e-01 1.65274873e-01 -2.88986504e-01 -3.11871231e-01
-5.64188778e-01 2.89306521e-01 4.65314776e-01 7.29485452e-01
7.07342207e-01 -1.10056555e+00 -4.71702367e-01 1.63202360e-01
6.66892946e-01 -1.47076070e-01 3.12721282e-01 5.57083189e-01
-6.98494196e-01 8.82786691e-01 -1.39007926e-01 -7.31588185e-01
-1.29639614e+00 6.14464700e-01 3.50012273e-01 -5.12118280e-01
-3.48279804e-01 7.84686744e-01 2.61100292e-01 -4.00037259e-01
2.76451379e-01 -6.51342198e-02 -2.78439790e-01 1.17325097e-01
5.43716133e-01 -2.42136672e-01 4.76526558e-01 -6.12938881e-01
-4.03893620e-01 6.12020910e-01 -6.06226444e-01 1.30439296e-01
1.45316327e+00 4.91951499e-03 -8.74677822e-02 1.61970407e-02
1.57828963e+00 -4.79131222e-01 -8.96540403e-01 -5.87620556e-01
2.08188400e-01 -5.92567444e-01 1.26451284e-01 -8.91830862e-01
-1.44144797e+00 8.40185225e-01 9.83816385e-01 -1.90026298e-01
1.23973942e+00 2.00544164e-01 5.66961169e-01 8.44932437e-01
3.85794044e-01 -1.10735798e+00 2.51284212e-01 3.23369294e-01
9.25408721e-01 -1.45047235e+00 -5.85981086e-02 -9.43821073e-02
-3.19893569e-01 1.10814953e+00 5.45845032e-01 -3.31266403e-01
7.42613852e-01 -3.15687686e-01 9.95948985e-02 -4.22548115e-01
-7.47829199e-01 -3.67149889e-01 6.53814793e-01 1.14222728e-01
3.22729886e-01 -1.11666657e-01 -3.58077496e-01 -9.71513316e-02
2.04785362e-01 1.24324419e-01 -1.34167103e-02 9.80750799e-01
-2.95425743e-01 -1.31465280e+00 -6.96464926e-02 5.80936134e-01
-4.75346893e-01 -2.75400639e-01 -5.52703857e-01 7.14612663e-01
-2.41576564e-02 5.04410744e-01 -1.06223859e-02 -1.70052588e-01
2.20469773e-01 -1.22971170e-01 3.01860154e-01 -4.67306733e-01
-6.12433434e-01 -6.75347745e-02 -2.88699806e-01 -8.81005824e-01
-5.67601323e-01 -3.14712822e-01 -9.24998403e-01 1.97720498e-01
-4.57047522e-01 2.63905555e-01 9.78765368e-01 6.19398355e-01
7.16627002e-01 3.94074798e-01 5.42144477e-01 -1.14829421e+00
-3.81322891e-01 -7.93680847e-01 -9.56416950e-02 4.90928888e-01
1.53814435e-01 -5.64337611e-01 -3.35268048e-03 -2.17742026e-01] | [10.657520294189453, 0.6769074201583862] |
4abb7bcb-2679-4a34-b8db-f6d6d8fed9b4 | are-chatgpt-and-gpt-4-general-purpose-solvers | 2305.05862 | null | https://arxiv.org/abs/2305.05862v1 | https://arxiv.org/pdf/2305.05862v1.pdf | Are ChatGPT and GPT-4 General-Purpose Solvers for Financial Text Analytics? An Examination on Several Typical Tasks | The most recent large language models such as ChatGPT and GPT-4 have garnered significant attention, as they are capable of generating high-quality responses to human input. Despite the extensive testing of ChatGPT and GPT-4 on generic text corpora, showcasing their impressive capabilities, a study focusing on financial corpora has not been conducted. In this study, we aim to bridge this gap by examining the potential of ChatGPT and GPT-4 as a solver for typical financial text analytic problems in the zero-shot or few-shot setting. Specifically, we assess their capabilities on four representative tasks over five distinct financial textual datasets. The preliminary study shows that ChatGPT and GPT-4 struggle on tasks such as financial named entity recognition (NER) and sentiment analysis, where domain-specific knowledge is required, while they excel in numerical reasoning tasks. We report both the strengths and limitations of the current versions of ChatGPT and GPT-4, comparing them to the state-of-the-art finetuned models as well as pretrained domain-specific generative models. Our experiments provide qualitative studies, through which we hope to help understand the capability of the existing models and facilitate further improvements. | ['Sameena Shah', 'Xiaomo Liu', 'Zhiqiang Ma', 'Xiaodan Zhu', 'Xianzhi Li'] | 2023-05-10 | null | null | null | null | ['named-entity-recognition-ner'] | ['natural-language-processing'] | [-2.05486521e-01 2.85524011e-01 8.95443931e-02 -4.90423441e-01
-1.06072354e+00 -7.53218174e-01 9.12647605e-01 -4.27229553e-02
-3.08458060e-01 7.97953963e-01 2.99589783e-01 -6.38665497e-01
4.06599371e-03 -9.15221632e-01 -6.22575462e-01 -2.44560540e-01
-3.34135965e-02 1.03782010e+00 -6.39974102e-02 -4.54040051e-01
3.70045424e-01 1.95581064e-01 -1.10587764e+00 6.61310852e-01
7.96516061e-01 8.07798862e-01 -1.22758105e-01 6.42497540e-01
-3.89159560e-01 1.44237673e+00 -7.73378551e-01 -7.53929853e-01
1.02499254e-01 -4.42230463e-01 -1.07999194e+00 -1.39242753e-01
3.93033624e-02 -2.00908154e-01 -4.87974845e-02 6.05595529e-01
6.72441065e-01 3.46755385e-01 5.66263676e-01 -1.21427953e+00
-1.20229578e+00 1.10275209e+00 -1.38269246e-01 3.33475560e-01
6.01824641e-01 4.72900987e-01 1.24076021e+00 -9.71546113e-01
7.55342305e-01 1.39822757e+00 8.43578696e-01 6.92985475e-01
-1.13312006e+00 -5.90492010e-01 2.45780870e-01 -7.65725225e-02
-1.06161308e+00 -2.84748167e-01 4.16601747e-01 -3.12653124e-01
1.93319428e+00 -5.84027031e-03 4.54195470e-01 1.49387193e+00
3.16428483e-01 1.01939678e+00 1.18927491e+00 -3.47310215e-01
3.34700137e-01 3.58902603e-01 -2.08020005e-02 4.01321322e-01
-6.56141490e-02 -2.01594740e-01 -7.78677344e-01 -1.50760219e-01
6.05211377e-01 -4.30327684e-01 5.29745482e-02 2.07934514e-01
-9.70787823e-01 1.20609307e+00 1.28484249e-01 5.85252821e-01
-4.69214708e-01 7.00865537e-02 2.88541704e-01 3.73694748e-01
9.34749544e-01 9.71187770e-01 -6.44386888e-01 -6.26840293e-01
-1.01456451e+00 7.13717163e-01 1.51040554e+00 1.07191598e+00
3.74487340e-01 2.27810979e-01 -1.74714938e-01 6.08238399e-01
5.29159978e-02 1.66498587e-01 7.54114866e-01 -6.79227650e-01
8.78525496e-01 6.09484673e-01 6.33137599e-02 -8.33453774e-01
-3.77252698e-01 -2.30700821e-01 -2.03009531e-01 -8.78705606e-02
5.46656668e-01 -5.24062932e-01 -8.41127872e-01 1.58517814e+00
-2.01311231e-01 1.27261952e-01 4.43053186e-01 6.47539258e-01
8.92791688e-01 1.02825153e+00 3.23374838e-01 9.24727991e-02
1.27675533e+00 -9.73649442e-01 -3.88776571e-01 -8.31175327e-01
8.46697330e-01 -8.84863734e-01 1.19039989e+00 4.22699034e-01
-1.51690412e+00 -3.68043751e-01 -5.35105646e-01 9.18344874e-03
-5.08696079e-01 -2.23120481e-01 1.06868684e+00 7.96396911e-01
-1.13925231e+00 5.72103679e-01 -8.78787577e-01 -5.38778961e-01
1.46090522e-01 8.80194008e-02 -6.76537156e-02 -1.15205728e-01
-1.37870300e+00 1.27846074e+00 2.91351497e-01 -1.01928920e-01
-6.32963240e-01 -9.10161138e-01 -7.84261644e-01 2.11951882e-01
4.12574947e-01 -8.54667723e-01 1.79256475e+00 -6.68577194e-01
-1.69074357e+00 8.68917167e-01 1.37247503e-01 -7.86412001e-01
5.05238950e-01 -2.02016190e-01 -1.82365268e-01 -1.95416398e-02
4.54683341e-02 6.39563441e-01 2.20880240e-01 -7.95736611e-01
-4.85488385e-01 7.73562342e-02 9.74969715e-02 3.01866353e-01
-1.70490682e-01 4.19871658e-01 -1.09957054e-01 -7.20307231e-01
-3.84589881e-01 -7.74395645e-01 -2.97474742e-01 -7.96750724e-01
-2.08697319e-01 -4.61322576e-01 5.25855899e-01 -6.14831805e-01
1.03246045e+00 -1.63694108e+00 -2.82669868e-02 -1.53736368e-01
-2.17894778e-01 1.95008665e-01 -2.44126484e-01 9.13979530e-01
-5.20370807e-03 5.17030120e-01 -2.00167626e-01 -5.44528782e-01
2.93684751e-01 3.97024661e-01 -4.49881971e-01 -1.58467352e-01
7.73675919e-01 1.43314743e+00 -8.13640833e-01 -3.89187098e-01
-3.97257283e-02 3.00465524e-01 -6.97136283e-01 2.01500133e-01
-6.18577302e-01 1.48897365e-01 -6.67232633e-01 6.93574607e-01
4.81171720e-02 -4.26857173e-01 5.90544380e-02 3.47987920e-01
-9.92569327e-02 6.95726871e-01 -9.12951767e-01 1.47404253e+00
-4.93269145e-01 3.54300410e-01 -2.06307843e-01 -1.01511097e+00
9.29243207e-01 3.76850367e-01 1.94854081e-01 -8.59291315e-01
2.69301444e-01 2.76666939e-01 -1.05617650e-01 -5.31102836e-01
6.98027790e-01 -6.13263667e-01 -4.75964576e-01 8.50080132e-01
2.95933634e-01 -4.52629238e-01 5.75213432e-01 4.42531198e-01
1.12130928e+00 2.56311923e-01 2.49026015e-01 -2.21458197e-01
1.11994095e-01 2.85454333e-01 2.61135519e-01 1.00312531e+00
2.27130696e-01 6.79494798e-01 6.00962222e-01 -2.53458560e-01
-1.01463377e+00 -5.35285711e-01 1.98372215e-01 1.26243937e+00
-5.71649969e-01 -4.23070371e-01 -9.01646733e-01 -3.91767621e-01
-5.56657799e-02 1.34041357e+00 -5.38490057e-01 1.76075287e-02
-5.87982953e-01 -1.31388032e+00 7.27395952e-01 7.93239951e-01
4.23089385e-01 -1.42347014e+00 -7.33651161e-01 6.06492758e-01
-3.41756850e-01 -1.28609395e+00 -1.50438756e-01 3.03366393e-01
-8.44975889e-01 -8.33668172e-01 -6.77895010e-01 -5.73910594e-01
2.15400696e-01 -3.59312922e-01 1.62020254e+00 -3.61267954e-01
-1.16072685e-01 5.91964304e-01 -5.51779270e-01 -4.96043533e-01
-7.83734083e-01 4.16620493e-01 -4.14753020e-01 -4.33502227e-01
6.09826982e-01 -4.34625566e-01 -1.29186079e-01 9.06488150e-02
-8.87003839e-01 9.47975069e-02 5.93082607e-01 8.31598341e-01
-5.67139722e-02 -2.60802712e-02 6.07708097e-01 -1.23185241e+00
1.31050515e+00 -7.14006007e-01 -2.50107199e-01 2.34762520e-01
-7.49081552e-01 2.49813702e-02 6.08253777e-01 -4.66259927e-01
-1.35189283e+00 -5.62384486e-01 -1.78705707e-01 -1.70144185e-01
6.71775341e-02 1.06510353e+00 1.85482696e-01 1.08603187e-01
8.75379324e-01 1.83635369e-01 -3.57026070e-01 -3.46551955e-01
2.56076723e-01 3.30469280e-01 4.25267935e-01 -1.05377483e+00
6.87462687e-01 -2.19093084e-01 -3.47067654e-01 -3.56621951e-01
-1.01195025e+00 -9.34510529e-02 -2.18612418e-01 1.05048448e-01
6.63057685e-01 -8.88323188e-01 -4.45867836e-01 5.74626923e-01
-1.01539135e+00 -7.10303247e-01 -4.31268901e-01 2.31620878e-01
-5.82839787e-01 1.02614127e-01 -1.22086918e+00 -9.57841694e-01
-4.68310714e-01 -9.61126864e-01 9.42061007e-01 2.67219633e-01
-5.86161077e-01 -1.50795031e+00 1.10684745e-01 4.60654825e-01
6.94148898e-01 3.97207081e-01 9.24776673e-01 -1.27743733e+00
-4.69623506e-01 -3.15050840e-01 2.79481560e-02 3.16751987e-01
-4.20345813e-01 4.35125865e-02 -8.29820633e-01 -1.28905550e-01
2.18687430e-01 -7.43710339e-01 6.21677577e-01 1.96011394e-01
6.06566727e-01 -4.88067687e-01 4.59569730e-02 3.46337110e-01
1.16991949e+00 2.50553668e-01 5.26441097e-01 6.60453677e-01
3.57102156e-01 5.34946978e-01 5.11098981e-01 4.52454269e-01
5.93253374e-01 2.74747372e-01 1.28396973e-01 2.10298419e-01
2.50436097e-01 -3.99292946e-01 4.81468916e-01 6.93897486e-01
-1.94704548e-01 -6.03846550e-01 -1.25379682e+00 4.62741733e-01
-1.93382812e+00 -8.91477168e-01 6.11627214e-02 1.63097048e+00
1.11591971e+00 3.68140101e-01 -5.52133210e-02 -2.37733424e-01
2.00650275e-01 9.81377065e-02 -4.14552629e-01 -7.53691971e-01
-1.99621871e-01 7.77946651e-01 1.26062870e-01 2.09568888e-01
-7.98484743e-01 1.19175994e+00 7.28950644e+00 6.89848423e-01
-8.45482171e-01 -3.05208843e-02 9.72282350e-01 -3.20191294e-01
-2.48354971e-01 9.69556198e-02 -9.62077737e-01 2.61581838e-01
1.53329313e+00 -3.65544975e-01 4.84282315e-01 1.02914345e+00
-3.69931944e-02 -2.39642169e-02 -1.33720052e+00 5.89440107e-01
1.16396897e-01 -1.27123463e+00 6.47163838e-02 -1.05263427e-01
8.13839376e-01 2.69415230e-02 4.79162745e-02 1.05217481e+00
7.25242674e-01 -1.32567513e+00 8.96094918e-01 3.58995110e-01
3.07227284e-01 -7.05511153e-01 8.24696600e-01 4.76858854e-01
-7.62017488e-01 -3.68995890e-02 -3.63146722e-01 -3.97632092e-01
3.65416199e-01 2.76462853e-01 -1.09418726e+00 6.47227883e-01
4.88418221e-01 6.93385363e-01 -5.29757559e-01 5.34094453e-01
-4.00536150e-01 8.98658693e-01 -2.08698690e-01 -2.11152241e-01
7.53203034e-01 -6.27793744e-02 2.94119477e-01 1.51778388e+00
5.11974573e-01 5.28298438e-01 2.85639971e-01 1.28409481e+00
-1.50219491e-03 2.11336687e-01 -2.31210783e-01 -6.24475241e-01
2.38303900e-01 1.21518075e+00 -7.39072084e-01 -5.62438250e-01
-4.91896749e-01 6.97208524e-01 3.46958309e-01 3.28780651e-01
-7.36483753e-01 -1.84343129e-01 3.82885993e-01 1.02453589e-01
3.24494749e-01 -1.05501831e-01 -3.76748860e-01 -1.33606339e+00
-1.90553099e-01 -1.29144061e+00 5.49443841e-01 -1.10747218e+00
-1.61469615e+00 6.65820837e-01 2.01082170e-01 -5.58835983e-01
-8.85141194e-01 -7.79331625e-01 -6.94087803e-01 1.09451675e+00
-1.30260968e+00 -1.18209636e+00 3.64863649e-02 5.72567165e-01
6.19260848e-01 -2.97374636e-01 9.72521305e-01 -2.65408438e-02
-5.36106944e-01 4.39998090e-01 -3.04812014e-01 2.35037163e-01
6.10797048e-01 -1.27756226e+00 1.01703954e+00 8.48561466e-01
8.15658197e-02 8.72739613e-01 9.11457360e-01 -6.71638787e-01
-1.26826108e+00 -8.34688008e-01 1.23347116e+00 -8.29023302e-01
1.00404561e+00 -3.85066777e-01 -1.04492092e+00 9.53955770e-01
3.42245400e-01 -4.07338858e-01 5.42017460e-01 2.11791441e-01
-3.52450848e-01 4.09778982e-01 -1.15613806e+00 4.32378203e-01
8.64832580e-01 -4.37900275e-01 -1.08639932e+00 3.89871001e-01
6.26397133e-01 -7.07218826e-01 -8.79562676e-01 9.14970040e-02
2.17980593e-01 -1.15713966e+00 8.27385306e-01 -8.66632223e-01
9.56786990e-01 5.37499964e-01 1.07376844e-01 -1.45893598e+00
-2.10866615e-01 -7.71799624e-01 6.27859160e-02 1.45444500e+00
7.07585454e-01 -8.60831201e-01 7.55101323e-01 1.14220095e+00
-2.22801223e-01 -7.56062567e-01 -6.46898150e-01 -6.91480160e-01
6.01695895e-01 -4.96721059e-01 5.36154270e-01 1.06428170e+00
2.04494819e-01 4.31539357e-01 -2.04004750e-01 -3.43808651e-01
8.67923424e-02 9.03255716e-02 6.70926273e-01 -1.00397182e+00
-7.36568153e-01 -5.79390705e-01 2.80022901e-02 -6.18288815e-01
2.68089533e-01 -7.79429138e-01 -1.38597712e-01 -1.73017597e+00
2.27497876e-01 -1.25689238e-01 4.54585850e-02 8.52274716e-01
-2.11414114e-01 1.94187183e-02 2.87001729e-01 9.10709128e-02
-2.84907401e-01 2.33326122e-01 1.02947652e+00 5.47007509e-02
-1.94642574e-01 -1.08163677e-01 -1.14129865e+00 5.00011504e-01
8.21601152e-01 -4.26659226e-01 -4.62144822e-01 -3.23162884e-01
5.60605407e-01 2.39067346e-01 2.02671334e-01 -7.86997616e-01
2.81683743e-01 -3.03381860e-01 2.47952446e-01 -2.19388410e-01
3.52749497e-01 -1.03689946e-01 8.96812901e-02 1.64338619e-01
-3.74804050e-01 2.79901445e-01 4.40971494e-01 1.59091502e-01
-3.92310590e-01 -4.25989449e-01 4.21361417e-01 -7.91908979e-01
-6.64316237e-01 8.03523418e-03 -6.10927761e-01 6.00113630e-01
8.36438060e-01 -1.88283890e-01 -4.94839191e-01 -6.46311045e-01
-4.48672175e-01 2.72910267e-01 1.81777403e-01 5.09834051e-01
3.77789557e-01 -8.14499557e-01 -9.48821962e-01 -1.79465208e-02
-6.61683157e-02 -1.21537494e-02 1.90021172e-02 6.47130013e-01
-3.26962709e-01 9.11547899e-01 -1.74750715e-01 -1.45089388e-01
-8.51303637e-01 4.88383919e-01 3.19987088e-01 -8.92151237e-01
-5.19074559e-01 1.09746158e+00 1.33776963e-01 -4.99175668e-01
-5.72213605e-02 -7.44453192e-01 1.50632877e-02 7.28816316e-02
3.46645683e-01 9.00997818e-02 1.03000268e-01 -1.29558086e-01
-1.52909607e-01 1.65390894e-01 -2.36741230e-01 -3.80150408e-01
1.66052210e+00 2.55996108e-01 7.63781443e-02 4.88307744e-01
7.45146990e-01 -3.53163600e-01 -1.02175820e+00 -1.11500256e-01
2.34352797e-01 -1.76410109e-01 -2.58934885e-01 -1.07031870e+00
-7.52851367e-01 7.15969741e-01 -4.11531955e-01 3.81841898e-01
9.17245924e-01 -3.86412181e-02 8.49096835e-01 3.84640455e-01
5.25883794e-01 -9.06685948e-01 2.47326940e-01 9.04586494e-01
7.05313861e-01 -9.24609840e-01 -2.22912580e-01 -5.20294867e-02
-1.17274535e+00 1.09874785e+00 5.28447747e-01 -1.89684018e-01
5.65120429e-02 5.85868776e-01 -1.70160234e-02 -3.00639063e-01
-1.35484886e+00 8.90215039e-02 9.50809494e-02 2.47470960e-01
8.35963607e-01 -2.15003878e-01 -7.37523064e-02 8.82024527e-01
-7.59498000e-01 1.71133250e-01 5.89408159e-01 1.24611056e+00
-2.71429181e-01 -1.26375675e+00 -2.61365116e-01 2.31074467e-01
-7.02166915e-01 -4.63625759e-01 -6.71493053e-01 1.10160577e+00
-4.01720911e-01 9.23604310e-01 -6.51856512e-02 -1.08054854e-01
5.76255739e-01 7.61945486e-01 4.96976733e-01 -9.64574039e-01
-1.29269421e+00 -6.42473772e-02 4.99487787e-01 -4.19848442e-01
-3.13105941e-01 -8.47580969e-01 -1.01339698e+00 -4.66763675e-01
1.58638135e-02 3.08300793e-01 4.04229909e-01 1.03459847e+00
2.01640382e-01 4.85504329e-01 4.01478335e-02 -7.64650881e-01
-9.58335102e-01 -1.26268458e+00 -4.32772338e-01 2.92094111e-01
-3.43735158e-01 -2.89338201e-01 -3.76727760e-01 1.25771090e-02] | [10.978813171386719, 8.471558570861816] |
7bebc731-7854-4a1d-9b9e-299f72096ee4 | diversity-promoting-gan-a-cross-entropy-based | null | null | https://aclanthology.org/D18-1428 | https://aclanthology.org/D18-1428.pdf | Diversity-Promoting GAN: A Cross-Entropy Based Generative Adversarial Network for Diversified Text Generation | Existing text generation methods tend to produce repeated and {''}boring{''} expressions. To tackle this problem, we propose a new text generation model, called Diversity-Promoting Generative Adversarial Network (DP-GAN). The proposed model assigns low reward for repeatedly generated text and high reward for {''}novel{''} and fluent text, encouraging the generator to produce diverse and informative text. Moreover, we propose a novel language-model based discriminator, which can better distinguish novel text from repeated text without the saturation problem compared with existing classifier-based discriminators. The experimental results on review generation and dialogue generation tasks demonstrate that our model can generate substantially more diverse and informative text than existing baselines. | ['Xu sun', 'Jingjing Xu', 'Junyang Lin', 'Xuancheng Ren'] | 2018-10-01 | null | null | null | emnlp-2018-10 | ['review-generation'] | ['natural-language-processing'] | [ 4.09660876e-01 4.88843173e-01 -1.69881545e-02 -2.32345730e-01
-1.11070585e+00 -5.63376248e-01 1.10092688e+00 -3.69434655e-01
-9.26806256e-02 1.56807160e+00 4.39972192e-01 -2.75033921e-01
4.46376950e-01 -9.35662448e-01 -2.73067296e-01 -5.48962057e-01
5.81958115e-01 6.42294228e-01 -2.53998280e-01 -5.95634699e-01
1.24610595e-01 5.23662940e-02 -1.12420738e+00 2.87621766e-01
1.45206165e+00 4.62056667e-01 1.65001810e-01 9.44286585e-01
-2.93704242e-01 9.33521509e-01 -1.47495806e+00 -7.53361166e-01
9.27177593e-02 -1.21594071e+00 -7.42083073e-01 -1.64473698e-01
4.11156379e-02 -6.08692050e-01 -3.49371791e-01 8.47656429e-01
8.98572981e-01 5.02368033e-01 1.06219018e+00 -1.24449611e+00
-1.29838991e+00 1.13614666e+00 -3.05791706e-01 9.23351385e-03
5.21704435e-01 4.21003163e-01 1.01186299e+00 -7.54412949e-01
6.31984591e-01 1.42951155e+00 3.26245308e-01 1.31821823e+00
-1.08861804e+00 -8.23764026e-01 2.31084749e-01 -5.81985652e-01
-1.01048827e+00 -3.23055714e-01 8.47847164e-01 -1.19806223e-01
6.82706296e-01 4.54516441e-01 4.92850751e-01 1.83511043e+00
1.66480616e-01 1.26061559e+00 9.74359155e-01 -3.58223379e-01
1.89706132e-01 2.46817976e-01 -4.25856262e-01 3.99492770e-01
1.30984873e-01 9.85428393e-02 -2.25663528e-01 -1.57929122e-01
6.61479652e-01 -1.82736531e-01 -2.78680563e-01 4.98895526e-01
-1.11147928e+00 1.12344825e+00 7.78172836e-02 1.55092746e-01
-4.73199576e-01 2.22719729e-01 3.98141235e-01 1.62999064e-01
8.31059039e-01 7.25004435e-01 1.15232626e-02 -3.50581318e-01
-8.36505473e-01 6.75446212e-01 8.14745188e-01 1.35573542e+00
3.16606283e-01 7.19323635e-01 -1.21848404e+00 1.05044389e+00
1.20005250e-01 6.61972582e-01 8.84231150e-01 -5.59855878e-01
6.84195638e-01 4.78642792e-01 1.89124733e-01 -4.78133649e-01
2.61375476e-02 -3.00746769e-01 -1.23535585e+00 -1.92264736e-01
-6.19820282e-02 -8.97412121e-01 -1.13913178e+00 1.78595209e+00
5.33812903e-02 -2.31293693e-01 4.58231807e-01 4.39200759e-01
1.06962812e+00 9.90259588e-01 1.29010230e-01 -3.26653004e-01
9.24980164e-01 -1.06295311e+00 -8.57037961e-01 -9.95695591e-02
3.60689759e-01 -8.35964143e-01 1.27419317e+00 -1.51447147e-01
-1.30673337e+00 -5.67666650e-01 -6.78033113e-01 1.05613254e-01
-2.01671571e-01 2.28776798e-01 4.89682168e-01 8.87950540e-01
-1.02393472e+00 3.37939084e-01 -8.50327015e-02 5.31499535e-02
5.15357852e-01 4.20839675e-02 1.67060450e-01 1.23264387e-01
-1.58306575e+00 6.62028790e-01 4.98015344e-01 -1.80634495e-03
-9.27726626e-01 -5.90845942e-01 -8.19425642e-01 -9.27649662e-02
2.38934588e-02 -1.09291267e+00 1.57609797e+00 -1.10900521e+00
-2.05204058e+00 4.41878825e-01 -1.68784589e-01 -4.06557530e-01
8.78736138e-01 -8.11465159e-02 -4.56003934e-01 9.70464293e-03
1.95120305e-01 9.60744858e-01 1.02211308e+00 -1.46847844e+00
-4.27863717e-01 2.89843887e-01 -6.97299168e-02 3.68091881e-01
-1.45763442e-01 -2.54959255e-01 8.73684660e-02 -1.35735166e+00
-6.95532024e-01 -7.68256247e-01 -3.62096697e-01 -5.70967138e-01
-1.02784860e+00 -6.67784929e-01 5.47691464e-01 -5.38356423e-01
1.20622373e+00 -1.56150091e+00 -1.04910480e-02 -1.38097152e-01
2.03590974e-01 5.90122461e-01 -2.79247433e-01 7.03559458e-01
1.75009027e-01 5.66199481e-01 -1.94698378e-01 -4.84072834e-01
2.45631963e-01 1.17732160e-01 -5.99568605e-01 -3.82954299e-01
5.76156914e-01 1.25896037e+00 -1.25570130e+00 -4.30236399e-01
-3.31914090e-02 3.25397015e-01 -3.94810408e-01 5.06193399e-01
-7.04699099e-01 4.44066465e-01 -7.61454880e-01 4.38885391e-01
4.09476668e-01 -1.17803864e-01 -1.75544739e-01 5.60584247e-01
3.02579045e-01 3.05510253e-01 -5.41661441e-01 1.26635861e+00
-4.82817888e-01 4.87123162e-01 -5.83394945e-01 -5.22071183e-01
1.48756170e+00 4.02666211e-01 -1.14242986e-01 -5.40830791e-01
2.90997893e-01 7.62836188e-02 -2.08924524e-02 -3.04695487e-01
1.11362588e+00 -4.10845548e-01 -2.98857927e-01 7.87572801e-01
-1.63130149e-01 -5.76927900e-01 2.33122528e-01 4.59282935e-01
1.02760279e+00 1.93124592e-01 -1.43928573e-01 6.21871576e-02
4.74999905e-01 -1.93277925e-01 3.82778943e-01 1.03228867e+00
-1.04941174e-01 6.85554802e-01 6.27847314e-01 1.29786983e-01
-1.12677181e+00 -9.06937838e-01 2.79704720e-01 1.07711899e+00
-8.76412168e-02 -1.22440644e-01 -8.14752162e-01 -1.07111514e+00
-2.09072754e-01 1.24610829e+00 -8.70168865e-01 -4.34379488e-01
-5.55683076e-01 -7.65509248e-01 8.92851055e-01 6.32346570e-01
7.10585535e-01 -1.53047562e+00 3.81848477e-02 3.51949602e-01
-5.43121099e-01 -6.07624948e-01 -8.85731995e-01 -2.20321909e-01
-5.20037115e-01 -5.31891108e-01 -1.27031183e+00 -7.91947246e-01
7.22511232e-01 -1.95429504e-01 1.36455274e+00 -1.96957037e-01
-1.74422339e-02 1.82441577e-01 -7.62116313e-01 -6.85585141e-01
-1.17475379e+00 3.67140979e-01 -4.35621500e-01 -8.00931081e-02
1.76049396e-02 -1.50893509e-01 -6.16929650e-01 1.54197402e-02
-1.08637822e+00 2.47039363e-01 6.87371790e-01 1.27182901e+00
3.93440187e-01 -1.98788881e-01 1.46236241e+00 -1.27036119e+00
1.60779548e+00 -6.41252160e-01 8.75531044e-03 2.88335204e-01
-7.71383107e-01 1.94502726e-01 1.36508036e+00 -7.49583840e-01
-1.47556317e+00 -5.00437200e-01 -3.14596713e-01 -1.50401384e-01
-1.06701151e-01 1.47607163e-01 -1.43497676e-01 6.09145463e-01
7.62591124e-01 6.85346901e-01 -1.60812631e-01 1.13311131e-03
4.90869254e-01 7.95583606e-01 4.44794774e-01 -5.94001889e-01
8.63665342e-01 -2.68472545e-02 -3.53192925e-01 -3.84149224e-01
-6.73153818e-01 1.76806405e-01 -7.87080079e-02 -8.88763443e-02
6.70023799e-01 -8.68097305e-01 -4.21280324e-01 5.35486221e-01
-1.32513571e+00 -4.96782839e-01 -7.38600492e-01 7.37165436e-02
-5.28803229e-01 2.69507140e-01 -5.24838209e-01 -1.20073342e+00
-1.07186306e+00 -8.29175115e-01 1.23723340e+00 6.25508666e-01
-4.45918739e-01 -1.06122339e+00 2.09943384e-01 2.63488829e-01
4.47394609e-01 5.37993610e-01 7.57713020e-01 -9.77639675e-01
-3.11902910e-01 -4.33648944e-01 3.58988382e-02 5.62142134e-01
1.77721024e-01 1.05482772e-01 -6.86168313e-01 -9.10382494e-02
-4.87881064e-01 -5.33791363e-01 8.17286551e-01 -8.43686163e-02
8.95260632e-01 -7.91888297e-01 -9.33219492e-02 2.39176840e-01
7.32441306e-01 4.61059868e-01 8.13925505e-01 -1.68449193e-01
6.37833178e-01 4.07848746e-01 6.10115230e-01 7.34792411e-01
4.52112675e-01 2.79384047e-01 3.39956842e-02 -1.37128681e-01
-9.54173505e-02 -8.05580676e-01 5.62531650e-01 6.81351185e-01
1.29587218e-01 -1.03045976e+00 -3.42335880e-01 6.05449498e-01
-1.64654136e+00 -1.25857234e+00 3.11759636e-02 1.63737810e+00
1.36704874e+00 2.83485409e-02 2.33975530e-01 -5.87018318e-02
9.74208653e-01 2.66753942e-01 -6.82495952e-01 -7.84831166e-01
-1.95059672e-01 5.68790734e-01 -1.25716358e-01 4.17950273e-01
-7.10296094e-01 1.19297338e+00 6.60943174e+00 1.06828713e+00
-8.88931811e-01 -2.25083873e-01 8.75925899e-01 -2.51993299e-01
-9.89248872e-01 -2.93917328e-01 -9.54233408e-01 9.10801470e-01
7.25589931e-01 -7.30197847e-01 1.61532551e-01 7.78952301e-01
2.66985595e-01 1.95647717e-01 -8.01122725e-01 6.56218290e-01
3.28345954e-01 -1.21378779e+00 7.11433828e-01 -1.80389494e-01
1.31038654e+00 -6.73055649e-01 1.77701190e-01 7.80773699e-01
1.15668917e+00 -1.31211078e+00 5.43485463e-01 6.51910126e-01
9.49802637e-01 -1.02873409e+00 6.61019981e-01 3.69149894e-01
-6.93209767e-01 2.06650540e-01 -2.63558298e-01 2.25452781e-01
2.03769550e-01 5.53116024e-01 -1.12344098e+00 6.52203500e-01
-1.04885086e-01 3.63145769e-01 -5.62603056e-01 4.16051984e-01
-6.45870388e-01 6.59051836e-01 1.51242852e-01 -8.08048725e-01
3.84749919e-01 -1.04536861e-01 5.83427310e-01 1.25110745e+00
3.01506251e-01 8.47404748e-02 1.42078578e-01 1.15469587e+00
-5.90778530e-01 1.98051304e-01 -7.03632653e-01 -3.06598723e-01
5.43768287e-01 1.14248466e+00 -3.37965339e-01 -5.33242702e-01
2.76389122e-01 1.32786131e+00 2.25587770e-01 5.46651602e-01
-8.18861485e-01 -9.29297268e-01 2.34594986e-01 -3.84262860e-01
1.93271890e-01 2.75215745e-01 -3.24190319e-01 -1.03180826e+00
3.52483913e-02 -1.03827071e+00 9.51415598e-02 -8.57799172e-01
-1.52871192e+00 9.92627740e-01 -2.02762634e-01 -1.02085626e+00
-9.37060952e-01 -5.07505499e-02 -9.42755818e-01 1.29117191e+00
-1.30045700e+00 -1.31282735e+00 -2.17877448e-01 4.64830130e-01
8.53124738e-01 -6.11960113e-01 8.62421691e-01 -1.77817345e-01
-5.72581410e-01 1.15126753e+00 2.81592757e-01 2.94179052e-01
6.70092762e-01 -1.50623918e+00 8.45781744e-01 7.02629089e-01
-2.36352339e-01 5.13557434e-01 6.04808986e-01 -8.59377027e-01
-8.95029426e-01 -1.49609983e+00 1.00347698e+00 -4.49018955e-01
2.00307578e-01 -4.05071706e-01 -6.77825391e-01 4.85018730e-01
4.74608362e-01 -7.09244788e-01 7.89084852e-01 -3.25744331e-01
-1.35621443e-01 3.47807586e-01 -1.26358008e+00 1.01410854e+00
7.88158834e-01 -2.25376531e-01 -3.64999473e-01 5.19541502e-01
9.78387535e-01 -4.10570204e-01 -6.29842341e-01 2.88266033e-01
3.51378471e-01 -6.98468745e-01 6.75501704e-01 -6.51328504e-01
1.04655707e+00 2.05934495e-01 4.57011610e-01 -1.78871381e+00
-2.40559950e-01 -1.27768576e+00 -7.62580112e-02 1.62823510e+00
5.96515238e-01 -7.11047590e-01 7.05723882e-01 4.08793688e-01
-3.74349654e-01 -7.99578726e-01 -5.96984744e-01 -7.90955663e-01
6.93362832e-01 1.40931904e-01 8.91479075e-01 8.89275014e-01
-1.31083161e-01 7.42841423e-01 -7.40566432e-01 -5.94771206e-01
1.21783260e-02 -3.09616607e-02 9.22776699e-01 -7.39298403e-01
-3.48200470e-01 -6.67370319e-01 1.80059731e-01 -1.24024308e+00
4.41748768e-01 -9.94977236e-01 4.69282269e-01 -1.62695980e+00
9.23778936e-02 -5.38097084e-01 3.07025433e-01 2.81472325e-01
-9.52534080e-01 7.82921910e-02 1.16710991e-01 -8.12668204e-02
-4.03068542e-01 1.11274993e+00 1.90755153e+00 -1.72132909e-01
-2.68084466e-01 1.93312451e-01 -1.20930851e+00 1.43285275e-01
9.65615928e-01 -2.61078626e-01 -6.40285015e-01 -1.37801662e-01
7.61674196e-02 2.65104502e-01 1.67696178e-01 -4.77882802e-01
-2.04840481e-01 -3.17895949e-01 5.69836557e-01 -5.69237649e-01
1.29962698e-01 3.18442695e-02 -1.09527498e-01 2.03807756e-01
-9.29609776e-01 3.37242521e-02 1.54870257e-01 5.37059546e-01
-1.80199333e-02 -4.65184927e-01 7.02686846e-01 -2.32842639e-01
4.21685353e-03 3.38239729e-01 -5.63011765e-01 6.44297063e-01
9.64285254e-01 9.29026529e-02 -5.88585556e-01 -8.41581941e-01
-3.13135266e-01 4.75104064e-01 1.24206647e-01 6.80307209e-01
7.25070894e-01 -1.49477661e+00 -1.31720722e+00 -6.40290156e-02
-1.32551198e-04 1.97491780e-01 2.18411759e-01 -8.17636847e-02
-3.06058854e-01 2.96779037e-01 1.58312097e-01 -7.58457184e-03
-8.73880208e-01 2.22397462e-01 2.75242269e-01 -7.46590257e-01
-4.04637963e-01 1.10003531e+00 9.24202129e-02 -4.63068098e-01
-5.39070331e-02 5.51293455e-02 -1.56682327e-01 -7.73239955e-02
3.96030277e-01 3.65464419e-01 -4.37976122e-01 -2.31807753e-01
2.41450757e-01 -2.04186484e-01 -5.11720359e-01 -3.50346386e-01
6.97753012e-01 7.11278766e-02 1.86760157e-01 4.13398206e-01
8.43367338e-01 -4.80148010e-02 -1.12808597e+00 -3.34035419e-02
-4.65755522e-01 -3.05311173e-01 -4.06394839e-01 -1.26250219e+00
-6.64972067e-01 7.74732113e-01 9.63702574e-02 4.78735059e-01
9.16775644e-01 -2.78974861e-01 1.11944687e+00 2.18425363e-01
-1.83504403e-01 -1.09877694e+00 4.70416367e-01 6.75368428e-01
1.16381514e+00 -1.07578850e+00 -2.55299807e-01 -1.38580576e-01
-1.17324972e+00 7.07919836e-01 9.19002473e-01 -2.87246071e-02
-7.01251701e-02 9.02582332e-02 -7.75442133e-03 2.93904424e-01
-1.03654277e+00 -1.46275714e-01 2.21297279e-01 9.97549415e-01
5.80763042e-01 1.10407054e-01 -4.54466134e-01 7.32068896e-01
-6.76581025e-01 -1.36130750e-01 6.22112215e-01 6.92797601e-01
-1.93458095e-01 -1.38730490e+00 -8.73209238e-02 6.98875844e-01
-4.88612682e-01 -4.82191682e-01 -9.13105488e-01 5.86575091e-01
-1.29475683e-01 1.08021438e+00 1.47869112e-02 -3.08909237e-01
1.83629006e-01 3.36995631e-01 2.87611991e-01 -8.39546621e-01
-1.04037535e+00 -7.92957959e-04 3.32124084e-01 3.73537183e-01
1.11078480e-02 -5.26999533e-01 -1.20395350e+00 -3.95527244e-01
-4.60245043e-01 3.61613870e-01 2.58832216e-01 8.28923225e-01
4.08912927e-01 9.21813846e-01 1.13992417e+00 -4.61216390e-01
-8.62335742e-01 -1.30954814e+00 -5.41022897e-01 5.64265013e-01
3.73059399e-02 -1.50775343e-01 -3.36986095e-01 6.40537366e-02] | [11.905096054077148, 9.146808624267578] |
6ecbf7c5-153a-486a-92ac-65a2d4fefa95 | m2r2-missing-modality-robust-emotion | 2205.02524 | null | https://arxiv.org/abs/2205.02524v1 | https://arxiv.org/pdf/2205.02524v1.pdf | M2R2: Missing-Modality Robust emotion Recognition framework with iterative data augmentation | This paper deals with the utterance-level modalities missing problem with uncertain patterns on emotion recognition in conversation (ERC) task. Present models generally predict the speaker's emotions by its current utterance and context, which is degraded by modality missing considerably. Our work proposes a framework Missing-Modality Robust emotion Recognition (M2R2), which trains emotion recognition model with iterative data augmentation by learned common representation. Firstly, a network called Party Attentive Network (PANet) is designed to classify emotions, which tracks all the speakers' states and context. Attention mechanism between speaker with other participants and dialogue topic is used to decentralize dependence on multi-time and multi-party utterances instead of the possible incomplete one. Moreover, the Common Representation Learning (CRL) problem is defined for modality-missing problem. Data imputation methods improved by the adversarial strategy are used here to construct extra features to augment data. Extensive experiments and case studies validate the effectiveness of our methods over baselines for modality-missing emotion recognition on two different datasets. | ['Ning Wang'] | 2022-05-05 | null | null | null | null | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [ 4.94510323e-01 4.18894440e-01 -6.78606555e-02 -9.58015859e-01
-1.02544987e+00 -2.30890602e-01 8.61792803e-01 -5.64630270e-01
-2.15262324e-01 8.26431692e-01 8.03184450e-01 1.86187163e-01
4.43487704e-01 -1.55105904e-01 -5.06948233e-01 -8.31694722e-01
1.51533693e-01 2.46844843e-01 -8.95547032e-01 -3.97140443e-01
-6.60989210e-02 8.01415518e-02 -1.68961799e+00 1.01055598e+00
5.73782682e-01 1.22691524e+00 -2.72646785e-01 7.97607005e-01
-4.93112713e-01 1.44525552e+00 -5.79876661e-01 -4.23304409e-01
-6.87007606e-02 -4.71395046e-01 -1.03164744e+00 1.75910875e-01
-3.38150039e-02 -2.83167779e-01 -2.95654774e-01 6.00126326e-01
7.38133490e-01 4.29883540e-01 7.17175782e-01 -1.92692184e+00
-4.78989571e-01 5.91282308e-01 -3.57286096e-01 -4.13685799e-01
7.64344752e-01 -1.47749484e-01 3.97818506e-01 -1.04281950e+00
4.71264839e-01 1.49286890e+00 6.19383156e-01 1.38133597e+00
-8.03612173e-01 -5.91638803e-01 4.62307721e-01 4.16485488e-01
-1.09810781e+00 -1.05439198e+00 1.35524344e+00 8.90969187e-02
8.92502844e-01 6.44023478e-01 -1.28587067e-01 1.88378763e+00
-2.61080176e-01 1.00940371e+00 1.46132135e+00 -4.98668849e-01
7.12659657e-02 7.04792321e-01 3.07192862e-01 2.42471118e-02
-1.09196937e+00 4.56879940e-03 -8.32158387e-01 -2.96283215e-01
5.87292984e-02 1.31535634e-01 -3.42806935e-01 1.95848435e-01
-1.11675274e+00 5.45116544e-01 -1.58371702e-01 1.65926129e-01
-6.26531005e-01 -2.43265659e-01 8.07119131e-01 9.53152239e-01
5.03223538e-01 -1.11231528e-01 -8.81012857e-01 -4.38914001e-01
-4.09112185e-01 -2.02918082e-01 9.77090955e-01 1.04202652e+00
6.25367105e-01 3.82635117e-01 -2.21566781e-01 1.16567373e+00
3.07170540e-01 3.83895665e-01 6.78358614e-01 -1.02731931e+00
6.39601171e-01 5.77105343e-01 2.25066260e-01 -7.37349570e-01
-4.79404956e-01 -1.74951516e-02 -1.42720473e+00 -7.55767375e-02
8.27908590e-02 -7.01980293e-01 -6.58973038e-01 2.11676574e+00
5.14883399e-01 5.06609321e-01 9.55821574e-01 9.19722676e-01
1.27019262e+00 9.34681594e-01 3.29101413e-01 -6.17717147e-01
1.39993715e+00 -1.06708646e+00 -1.32413983e+00 -6.71766475e-02
4.43335325e-01 -5.19650340e-01 7.63068259e-01 3.46103340e-01
-1.04575813e+00 -5.63765287e-01 -4.89037007e-01 -7.14707598e-02
-5.08125722e-01 -7.46486485e-02 6.29192650e-01 7.73469508e-01
-8.64245176e-01 8.23483337e-03 -3.63456279e-01 -2.38435157e-02
-1.15746222e-01 4.89843398e-01 -8.85120511e-01 6.44920990e-02
-1.49132037e+00 9.71129358e-01 7.20150843e-02 7.96410918e-01
-9.55960512e-01 -3.70413989e-01 -1.06850791e+00 -1.41218156e-01
2.61689246e-01 -3.52092534e-01 1.12113118e+00 -1.71792531e+00
-2.06124830e+00 7.21592665e-01 -6.04261637e-01 -1.60165533e-01
2.70509154e-01 -3.60750817e-02 -1.02276063e+00 4.62713987e-02
-5.49702108e-01 6.60514235e-01 1.12636590e+00 -1.54131079e+00
-3.35146010e-01 -5.45963645e-01 -2.78773040e-01 5.36464810e-01
-1.27185076e-01 2.92472482e-01 -1.94465309e-01 -3.42642337e-01
2.05046251e-01 -7.83425272e-01 -1.23557106e-01 -7.17428505e-01
-4.33479309e-01 -3.55682731e-01 9.65690672e-01 -9.70015168e-01
7.33102202e-01 -2.20282197e+00 3.38059396e-01 6.75223693e-02
-3.59435380e-01 -2.65182137e-01 -5.90558767e-01 5.13603926e-01
-5.81022620e-01 -8.53712410e-02 -1.38615549e-01 -1.06614184e+00
2.31459767e-01 3.59266549e-01 -6.04015887e-01 3.25938880e-01
2.87786990e-01 7.93296099e-01 -8.70914310e-02 -4.17507112e-01
1.81913361e-01 7.73251414e-01 -1.45072490e-01 7.10478485e-01
-4.08325754e-02 9.57007945e-01 -1.52153149e-01 8.52682769e-01
9.88407254e-01 3.01367491e-01 1.83813751e-01 -3.49795371e-01
1.31491020e-01 -5.67078684e-03 -1.42392993e+00 1.95138109e+00
-8.41523230e-01 4.47705910e-02 6.71315849e-01 -1.10446858e+00
1.03708661e+00 1.12491810e+00 3.64347667e-01 -5.77839911e-01
2.42874697e-01 -3.50723118e-01 -3.95932376e-01 -8.76298547e-01
6.34090543e-01 -4.00051713e-01 -6.03035688e-01 4.99185443e-01
3.16808105e-01 2.52275288e-01 -8.73771071e-01 1.23969480e-01
6.36057973e-01 1.73285931e-01 -6.17095493e-02 4.25624043e-01
9.10622180e-01 -4.56890464e-01 9.52279210e-01 7.40299284e-01
-5.09790957e-01 4.01115060e-01 3.55263859e-01 -1.93249494e-01
-5.02909243e-01 -4.69763726e-01 -9.95384157e-03 1.67595208e+00
-5.96494749e-02 2.26210117e-01 -3.97386491e-01 -7.46905625e-01
-5.72782815e-01 7.92307615e-01 -9.14011002e-01 -6.19334169e-02
-3.20662498e-01 -5.26885509e-01 5.78235984e-01 4.42218423e-01
5.37492454e-01 -1.48244214e+00 9.24950093e-03 2.87182093e-01
-8.90306771e-01 -1.09865022e+00 -1.85110390e-01 2.15304375e-01
-5.19609392e-01 -7.54758179e-01 -5.02276719e-01 -6.74102664e-01
4.38159198e-01 -8.58401507e-02 9.14885879e-01 -3.41885507e-01
2.40536317e-01 9.92937207e-01 -6.31264269e-01 -4.15030032e-01
-4.82541412e-01 -2.74148971e-01 6.72228783e-02 7.34950364e-01
5.02607703e-01 -5.77758729e-01 -3.29878956e-01 2.33259186e-01
-7.42220223e-01 6.67809173e-02 3.91357154e-01 1.06801963e+00
2.24757537e-01 -2.62360454e-01 1.31987906e+00 -8.67219627e-01
5.19796729e-01 -8.34972322e-01 2.77281642e-01 5.50476611e-01
-2.19174530e-02 -4.25473191e-02 4.76874292e-01 -7.09367633e-01
-1.93803167e+00 3.14264834e-01 -3.02785695e-01 -4.23613578e-01
-8.44512403e-01 2.58024395e-01 -8.30351889e-01 3.10365170e-01
9.89695489e-02 4.95119691e-01 1.26734361e-01 -3.71564895e-01
6.20349824e-01 1.33124077e+00 8.68050814e-01 -7.66974092e-01
1.29956827e-01 2.62901574e-01 -4.36725467e-01 -6.31270170e-01
-5.84038138e-01 -5.12608349e-01 -4.40691650e-01 -4.00061667e-01
7.93425918e-01 -1.41899478e+00 -1.08329928e+00 5.34828544e-01
-1.32211363e+00 1.60886231e-03 -6.63982183e-02 5.78169107e-01
-7.07032442e-01 2.08521158e-01 -8.10026586e-01 -1.65498614e+00
-5.27135253e-01 -8.48450422e-01 1.10512364e+00 5.70103303e-02
-4.04305421e-02 -9.13232803e-01 -8.44664127e-02 7.92889476e-01
4.07314152e-01 3.39712024e-01 6.37134850e-01 -9.84864473e-01
2.32037678e-02 -1.27970204e-01 1.62369758e-01 5.06332517e-01
-1.01049624e-01 -5.17719328e-01 -1.66345596e+00 -2.71782596e-02
4.81928051e-01 -8.54916811e-01 7.12319911e-01 -2.99408124e-03
1.13232899e+00 -5.74320436e-01 7.56536573e-02 1.18060872e-01
1.08117163e+00 1.69858083e-01 8.58134031e-01 -3.30564052e-01
5.03556907e-01 1.17858958e+00 5.62756479e-01 5.25384605e-01
7.94149697e-01 4.19301808e-01 4.62468594e-01 -4.61433940e-02
2.47256532e-01 -1.91390864e-03 6.41963243e-01 1.09438205e+00
-1.38787925e-02 -3.55169177e-01 -1.84225976e-01 4.73423332e-01
-1.95713675e+00 -1.33098805e+00 -1.06371781e-02 1.76573205e+00
9.42254007e-01 -5.72575927e-01 3.18294689e-02 1.39999643e-01
8.71181846e-01 7.39745349e-02 -5.83907068e-01 -8.19080353e-01
-4.48394269e-01 -1.53760374e-01 -3.87648940e-01 5.57807207e-01
-9.71781731e-01 6.39692903e-01 5.64055204e+00 4.87943590e-01
-8.54451716e-01 4.23947811e-01 7.16625452e-01 6.24959655e-02
-3.87967795e-01 -2.09767371e-01 -5.11426687e-01 2.56790161e-01
1.22868752e+00 3.35990489e-01 6.69753313e-01 7.20495105e-01
1.49257585e-01 2.31736913e-01 -1.18809867e+00 1.39455473e+00
5.66093028e-01 -7.17262268e-01 -3.00298631e-01 -3.28279287e-01
5.63525379e-01 -2.86927432e-01 4.81182411e-02 1.08853161e+00
1.50015857e-02 -1.01053464e+00 2.14074790e-01 9.73532736e-01
7.43160129e-01 -9.76492941e-01 9.80082810e-01 5.72454870e-01
-8.15924704e-01 -1.51757583e-01 -3.27560119e-02 -5.82388453e-02
2.20324442e-01 4.69284840e-02 -5.32539487e-01 7.48518407e-01
4.07618582e-01 2.75638014e-01 -6.02348819e-02 4.17254344e-02
1.65845841e-01 6.02520823e-01 -2.39747152e-01 3.19005340e-01
-1.26421586e-01 -2.77855201e-03 3.99458289e-01 1.23916745e+00
4.39617448e-02 3.59245509e-01 8.99316892e-02 3.32898617e-01
-3.77828658e-01 2.75057524e-01 -7.23440051e-01 3.50925326e-01
5.46861947e-01 1.48434758e+00 2.74501979e-01 -3.79418164e-01
-5.27126014e-01 1.31932831e+00 1.76796168e-01 5.50098777e-01
-7.54827797e-01 -1.48237199e-01 3.99634004e-01 -8.31995726e-01
-1.19091399e-01 4.10248756e-01 8.19087923e-02 -1.35581386e+00
-2.83720382e-02 -1.18727291e+00 5.84837139e-01 -9.35154617e-01
-1.70146263e+00 7.89879739e-01 -3.33264619e-01 -1.12786543e+00
-5.78454673e-01 -1.41625002e-01 -7.39132524e-01 9.39311564e-01
-1.67874825e+00 -1.72074592e+00 -1.30671591e-01 1.25604463e+00
7.33733416e-01 -4.17938501e-01 1.41287410e+00 1.69208899e-01
-6.86864972e-01 8.75782430e-01 -1.29939452e-01 6.12869710e-02
9.49311197e-01 -9.68811691e-01 -6.38464570e-01 2.73351282e-01
-2.75214612e-01 3.30804080e-01 6.20283484e-01 -3.48115712e-01
-1.72256899e+00 -1.08688593e+00 9.30042863e-01 -4.90704894e-01
2.05751240e-01 -4.45459187e-01 -1.06330097e+00 9.87431586e-01
7.25234389e-01 -1.19219616e-01 1.28202581e+00 5.07971108e-01
-5.82608223e-01 -2.68648956e-02 -1.59506500e+00 3.40715110e-01
4.53919917e-01 -8.54557931e-01 -8.24915826e-01 7.53298774e-02
8.93087626e-01 -2.78507173e-01 -1.14436519e+00 5.24638534e-01
5.24440706e-01 -6.22795105e-01 6.90456390e-01 -1.01309526e+00
2.68502116e-01 3.07473410e-02 -7.23242223e-01 -1.14436114e+00
3.79080564e-01 -8.15670967e-01 -3.85525107e-01 1.83833373e+00
3.87900531e-01 -5.43834150e-01 6.03156149e-01 1.29253745e+00
-1.50579035e-01 -1.30174577e-01 -1.22586775e+00 -1.11071363e-01
-1.91702228e-02 -7.99039185e-01 9.57667887e-01 1.49504972e+00
3.65743935e-01 7.16718078e-01 -1.21523929e+00 3.88595432e-01
4.92522776e-01 1.08962789e-01 8.15493941e-01 -8.04131150e-01
-1.56565905e-01 2.92338312e-01 -4.59365174e-02 -7.28469193e-01
9.30072784e-01 -6.12348616e-01 9.49748084e-02 -9.44460750e-01
1.53796986e-01 -1.99460402e-01 -5.57395816e-01 6.43812478e-01
-3.87636498e-02 -1.08383745e-01 1.23893969e-01 -9.41935703e-02
-8.63685012e-01 1.09593284e+00 9.14110124e-01 -2.76926696e-01
-4.28993076e-01 2.21495718e-01 -7.84278989e-01 6.14066243e-01
8.13669920e-01 -2.97144115e-01 -3.37832659e-01 -3.56831960e-02
-9.71199945e-02 1.03815103e+00 5.17304122e-01 -5.49474299e-01
2.81575143e-01 -1.21542506e-01 3.47516209e-01 -9.26013350e-01
1.07917297e+00 -1.24538529e+00 1.72465652e-01 -3.24401528e-01
-8.55345726e-01 -1.73204675e-01 2.12476954e-01 6.89314306e-01
-3.76448631e-01 -1.61032215e-01 3.36775184e-01 -5.35883456e-02
-7.38201857e-01 3.59339342e-02 -3.70312095e-01 -2.59261012e-01
7.89315522e-01 7.61090368e-02 -3.59419614e-01 -9.80862498e-01
-1.53292596e+00 4.45965111e-01 -3.03220928e-01 6.13143325e-01
7.68457532e-01 -1.57922685e+00 -1.00755358e+00 1.29159525e-01
2.27341652e-01 -3.57253373e-01 1.22127342e+00 7.01572180e-01
7.25763917e-01 2.40930039e-02 -6.90689161e-02 -2.80364513e-01
-1.43363309e+00 7.06954360e-01 3.99304897e-01 -1.03506558e-01
-1.63055286e-01 6.18935943e-01 -6.35712594e-02 -1.21473837e+00
5.88552654e-01 1.82321608e-01 -4.91957933e-01 1.78868353e-01
7.39567697e-01 2.44449124e-01 -1.06859669e-01 -1.09836745e+00
-3.21233451e-01 -3.17728966e-02 -2.22543567e-01 -4.54897195e-01
1.37302089e+00 -7.56748319e-01 -1.78078577e-01 8.14495981e-01
1.27194071e+00 -2.86055267e-01 -1.16202402e+00 -4.37511235e-01
-4.57228303e-01 2.00893804e-02 -5.51388338e-02 -1.31386626e+00
-8.23348761e-01 8.77799869e-01 9.84033465e-01 3.62705952e-03
1.35382962e+00 -1.25960141e-01 5.63223958e-01 4.37964380e-01
6.81963637e-02 -1.37197471e+00 -2.65582800e-02 4.27313626e-01
1.32404447e+00 -1.75659680e+00 -6.44273341e-01 -7.34301731e-02
-1.47964752e+00 8.21338296e-01 8.42722356e-01 4.70370591e-01
7.04531431e-01 1.74273744e-01 5.43216765e-01 1.34764224e-01
-1.28460109e+00 -2.85436008e-02 -4.70241643e-02 9.06930685e-01
2.25828797e-01 8.60547572e-02 2.06655130e-01 1.29914165e+00
1.28464833e-01 -7.63212144e-02 4.64344561e-01 7.98497438e-01
7.49099180e-02 -1.02613175e+00 -6.15588248e-01 5.64726330e-02
-4.57112104e-01 2.71960106e-02 -3.19373190e-01 4.90086347e-01
1.31950676e-01 1.66797078e+00 -1.92504749e-02 -4.43497777e-01
3.27465862e-01 8.09435308e-01 1.02825359e-01 -1.33367285e-01
-8.87716234e-01 9.25834179e-02 4.32112753e-01 -4.65924263e-01
-1.01515520e+00 -6.66595459e-01 -9.74229515e-01 1.41671877e-02
-3.10374409e-01 3.21216166e-01 9.63436067e-01 1.02610064e+00
5.91170251e-01 4.86419797e-01 1.28474510e+00 -6.96193576e-01
-5.70259750e-01 -1.35418820e+00 -6.36835814e-01 7.87884057e-01
5.59417367e-01 -2.08668068e-01 -5.67904353e-01 2.81967819e-01] | [13.196829795837402, 5.568807601928711] |
6192e5ac-1309-431f-968f-f57287a7fdae | inferring-preferences-from-demonstrations-in | 2304.14115 | null | https://arxiv.org/abs/2304.14115v1 | https://arxiv.org/pdf/2304.14115v1.pdf | Inferring Preferences from Demonstrations in Multi-objective Reinforcement Learning: A Dynamic Weight-based Approach | Many decision-making problems feature multiple objectives. In such problems, it is not always possible to know the preferences of a decision-maker for different objectives. However, it is often possible to observe the behavior of decision-makers. In multi-objective decision-making, preference inference is the process of inferring the preferences of a decision-maker for different objectives. This research proposes a Dynamic Weight-based Preference Inference (DWPI) algorithm that can infer the preferences of agents acting in multi-objective decision-making problems, based on observed behavior trajectories in the environment. The proposed method is evaluated on three multi-objective Markov decision processes: Deep Sea Treasure, Traffic, and Item Gathering. The performance of the proposed DWPI approach is compared to two existing preference inference methods from the literature, and empirical results demonstrate significant improvements compared to the baseline algorithms, in terms of both time requirements and accuracy of the inferred preferences. The Dynamic Weight-based Preference Inference algorithm also maintains its performance when inferring preferences for sub-optimal behavior demonstrations. In addition to its impressive performance, the Dynamic Weight-based Preference Inference algorithm does not require any interactions during training with the agent whose preferences are inferred, all that is required is a trajectory of observed behavior. | ['Karl Mason', 'Patrick Mannion', 'Junlin Lu'] | 2023-04-27 | null | null | null | null | ['multi-objective-reinforcement-learning'] | ['methodology'] | [ 1.59319565e-02 -2.55943775e-01 -3.48217845e-01 -7.03456879e-01
-5.31878352e-01 -5.95701516e-01 4.35500056e-01 3.78591180e-01
-8.04623425e-01 9.35885727e-01 1.64759293e-01 -4.41136569e-01
-7.60988474e-01 -5.70210993e-01 -1.99793264e-01 -9.96419013e-01
-4.01359051e-01 1.21172380e+00 1.87312603e-01 -4.69334647e-02
6.59094751e-01 1.58150181e-01 -1.59869945e+00 1.25497594e-01
9.52868044e-01 9.33149517e-01 3.53939086e-01 8.94802332e-01
3.62660378e-01 5.74743211e-01 -1.73067048e-01 -1.92885920e-01
2.24992260e-01 7.51412362e-02 -8.25276136e-01 -2.25651860e-02
-1.83451638e-01 -7.24412799e-01 4.38018948e-01 9.23286915e-01
4.07424361e-01 7.78258801e-01 8.43287110e-01 -1.52405298e+00
-3.15833390e-01 5.83978772e-01 -3.64095300e-01 1.38925612e-01
5.42151868e-01 2.26910874e-01 1.08809149e+00 -3.24513078e-01
2.46863529e-01 1.63824666e+00 -6.36208989e-03 4.03291821e-01
-1.14683568e+00 -4.94430661e-01 8.50091994e-01 6.43343449e-01
-9.47883904e-01 -1.39107779e-01 6.94087267e-01 -3.51459980e-01
7.28543222e-01 2.81495333e-01 5.94458282e-01 1.19459498e+00
3.11366022e-01 1.01081264e+00 1.45437312e+00 -6.10890053e-02
9.47684526e-01 4.12531644e-02 3.11783403e-01 3.94256949e-01
2.44356185e-01 4.63832051e-01 -5.81168413e-01 -4.92085755e-01
3.17206383e-01 1.62216276e-01 2.30967067e-02 -4.16317970e-01
-1.10260725e+00 7.50697851e-01 -2.58480161e-01 -1.89933032e-01
-9.13448334e-01 1.47968195e-02 -6.53077215e-02 4.99806464e-01
3.26668888e-01 3.03059578e-01 -4.06378001e-01 -4.19091612e-01
-7.95613110e-01 7.03138351e-01 1.10343003e+00 6.59182608e-01
6.12289369e-01 -8.21282789e-02 -4.11161542e-01 3.73473018e-01
8.96719038e-01 5.39108574e-01 2.35232502e-01 -1.39729106e+00
4.97407049e-01 5.05289435e-01 1.17612875e+00 -9.31250036e-01
-2.19710574e-01 -5.33469208e-02 -9.52585340e-02 9.85128760e-01
4.08566326e-01 -5.60118258e-01 -6.79524720e-01 1.57562840e+00
5.95285833e-01 -9.15703401e-02 3.12002599e-01 1.07348442e+00
-1.18931839e-02 7.66759217e-01 2.10100442e-01 -3.92753839e-01
8.63224149e-01 -7.22434163e-01 -6.21087432e-01 -3.63706142e-01
-4.06994894e-02 -4.85258430e-01 9.26275790e-01 5.93207777e-01
-9.65479732e-01 -1.27212822e-01 -9.23270702e-01 5.81343889e-01
-4.56737764e-02 -3.07804465e-01 7.95769036e-01 5.26225507e-01
-7.00756788e-01 6.36130035e-01 -8.61434698e-01 -2.94232965e-01
1.33335646e-02 6.25894547e-01 1.93716556e-01 -7.78337717e-02
-7.91110098e-01 1.06514883e+00 3.08065414e-01 -5.44811860e-02
-1.51717436e+00 -4.83541995e-01 -6.76265180e-01 2.02778786e-01
5.18578351e-01 -6.71756506e-01 1.68441486e+00 -1.02729690e+00
-1.97653437e+00 6.01316094e-02 -3.25025558e-01 -1.35933340e-01
7.20237136e-01 -1.79349914e-01 -6.10670328e-01 -1.95573688e-01
-7.81340757e-04 3.71333688e-01 4.80450481e-01 -1.50323296e+00
-1.35669935e+00 -3.36296976e-01 6.72051966e-01 7.88260221e-01
3.37030180e-02 4.16902974e-02 2.25802928e-01 -6.60535842e-02
-3.82989109e-01 -9.78676617e-01 -6.40428007e-01 -2.56398052e-01
3.86728533e-03 -3.99899006e-01 6.14872456e-01 -3.34648401e-01
1.03598499e+00 -1.82248080e+00 7.69558176e-02 4.02038157e-01
-3.44644189e-01 -1.24376379e-01 -1.85318455e-01 5.36774635e-01
6.25804603e-01 -2.09402680e-01 -2.90997535e-01 -3.34272712e-01
6.22045040e-01 5.92299461e-01 -1.10979445e-01 4.32083607e-01
-2.44090632e-01 2.76928931e-01 -1.30038071e+00 -4.31914061e-01
2.69908637e-01 -1.17849879e-01 -6.71469569e-01 4.44029331e-01
-5.16039073e-01 4.80516195e-01 -9.33899224e-01 7.07414508e-01
4.89123374e-01 -5.35602830e-02 7.28978693e-01 3.26649725e-01
-5.28300285e-01 3.42689723e-01 -1.56172132e+00 1.47899759e+00
-4.70338464e-01 1.78137809e-01 2.31007919e-01 -8.90285492e-01
4.46028680e-01 5.05478978e-01 4.48789835e-01 -4.78489518e-01
1.65581092e-01 7.81030878e-02 3.18979234e-01 -6.94351435e-01
4.81958866e-01 -2.23746702e-01 -2.81412750e-01 9.87488270e-01
-3.75885129e-01 1.04204290e-01 4.12104726e-01 -2.82707304e-01
6.88849866e-01 2.83287257e-01 4.43329215e-01 -6.40388727e-01
4.45689470e-01 2.49380991e-01 1.10312068e+00 7.47185946e-01
-1.67522192e-01 -2.13738993e-01 1.06728144e-01 -5.22476256e-01
-3.72719944e-01 -1.00933874e+00 4.04733866e-01 1.34213638e+00
5.09226263e-01 2.81767249e-01 -3.22296560e-01 -5.42655110e-01
8.18418935e-02 1.44442737e+00 -5.77932835e-01 2.09306180e-01
-1.72881246e-01 -7.77078867e-01 -1.26603901e-01 3.92556727e-01
4.79668438e-01 -7.31869459e-01 -1.21587038e+00 3.82612467e-01
-4.20524240e-01 -5.51792860e-01 -5.05520344e-01 -1.56862959e-01
-8.92948389e-01 -9.84845698e-01 -3.72554421e-01 -3.03983003e-01
8.66652548e-01 2.01513976e-01 9.18066204e-01 -2.15776041e-01
4.72107440e-01 6.04856491e-01 -1.95996389e-01 -7.42839515e-01
-1.36696562e-01 -4.95895147e-01 5.58105648e-01 3.24440837e-01
5.59241474e-01 -7.49188781e-01 -7.37745166e-01 6.77345216e-01
-6.43770874e-01 -1.34103179e-01 5.80793083e-01 4.63520229e-01
4.21505600e-01 4.16558444e-01 9.26539823e-02 -5.76237798e-01
9.89749253e-01 -6.10630989e-01 -8.56584072e-01 6.70946181e-01
-8.11366558e-01 4.18315411e-01 3.32383543e-01 -7.08848357e-01
-1.66058362e+00 -2.62388319e-01 2.92941600e-01 3.64110246e-02
-4.08248067e-01 6.70538127e-01 -1.91302806e-01 2.76719123e-01
7.23007247e-02 -1.92689434e-01 -2.32470989e-01 -4.06274021e-01
-1.27691865e-01 5.42826951e-01 5.37511311e-04 -1.15765429e+00
5.19778430e-01 3.41444641e-01 -1.27032638e-01 -3.35757911e-01
-6.18281543e-01 -3.34367186e-01 -2.26029053e-01 -5.56554973e-01
7.56287158e-01 -5.43699563e-01 -1.24463367e+00 2.27268696e-01
-8.00032139e-01 -5.93271136e-01 1.49536133e-01 9.18515086e-01
-6.34474397e-01 3.03922057e-01 -1.25037404e-02 -1.38300276e+00
1.21491171e-01 -1.38380229e+00 3.01163614e-01 4.33032244e-01
-6.48923337e-01 -1.40853238e+00 2.37514570e-01 4.93227243e-01
3.45087796e-01 1.73381254e-01 9.95923519e-01 -6.73132539e-01
-4.11830366e-01 2.40513105e-02 4.70976442e-01 -2.86261261e-01
6.52887002e-02 -1.77447610e-02 -6.37243569e-01 -5.17155826e-01
-1.18357748e-01 -3.02296042e-01 1.28462836e-01 5.75894773e-01
8.26014161e-01 -6.66221142e-01 -2.00866446e-01 -7.34457895e-02
1.28724372e+00 8.34305227e-01 8.13468173e-02 4.74703074e-01
-9.92854163e-02 8.56558204e-01 1.27557898e+00 8.63870502e-01
9.01315868e-01 5.71510017e-01 7.55158961e-01 2.99305767e-01
8.26804996e-01 -8.94550532e-02 7.67021835e-01 3.11334521e-01
-8.24524462e-01 -4.62786317e-01 -6.57202661e-01 6.35072351e-01
-2.42346692e+00 -1.06709230e+00 2.17223838e-01 2.46438432e+00
3.94632429e-01 -4.30515558e-02 2.25157246e-01 -6.63248375e-02
5.04662097e-01 -1.51292071e-01 -1.05203211e+00 -7.98909128e-01
4.41361278e-01 -3.94561529e-01 5.00454605e-01 9.96340692e-01
-7.87392735e-01 4.46448565e-01 6.67416859e+00 2.17627063e-01
-7.43671238e-01 2.13669650e-02 4.30813134e-01 -4.45095450e-01
-6.39618576e-01 1.47737369e-01 -6.81994319e-01 3.98671806e-01
7.40071237e-01 -4.85183865e-01 8.36143553e-01 6.48152530e-01
7.10150301e-01 -6.77868485e-01 -1.52880275e+00 7.01839209e-01
-2.22674221e-01 -7.80070364e-01 -8.75228550e-03 1.50647447e-01
6.17409229e-01 -3.68610978e-01 1.47249743e-01 2.42335260e-01
1.56342649e+00 -8.11211109e-01 1.08793771e+00 5.35121322e-01
2.56441254e-02 -9.41877961e-01 9.73393440e-01 7.58313417e-01
-7.84444809e-01 -6.50607169e-01 -2.28322744e-01 -7.27941036e-01
4.70851272e-01 3.51212412e-01 -7.74950743e-01 6.07147574e-01
8.08405101e-01 4.23188299e-01 2.40466967e-01 8.55198503e-01
-4.35049415e-01 5.76575279e-01 -2.09700927e-01 -3.79173517e-01
6.02559626e-01 -6.95908904e-01 6.07935786e-01 8.19142520e-01
5.81168294e-01 4.33714300e-01 5.65342128e-01 7.75840223e-01
5.09455204e-01 -4.46689337e-01 -2.46049061e-01 -2.44972408e-02
5.87505281e-01 8.95627797e-01 -3.81767333e-01 -2.71142483e-01
-1.68072939e-01 6.87222779e-01 -8.85275003e-05 6.98389530e-01
-8.08179200e-01 2.02774435e-01 1.27351785e+00 -3.70800316e-01
9.28273723e-02 -2.75497347e-01 -1.01400726e-01 -6.52583241e-01
-2.18773827e-01 -9.43130791e-01 7.77381122e-01 -4.63873327e-01
-1.06267595e+00 1.70143284e-02 5.03512084e-01 -1.29279256e+00
-4.43483800e-01 -4.03465152e-01 -8.05431962e-01 7.83868015e-01
-1.56405354e+00 -7.77072608e-01 1.65136039e-01 5.25804520e-01
8.29434335e-01 -6.66212514e-02 8.94727468e-01 -5.71801364e-02
-5.44548273e-01 2.86357459e-02 6.56832159e-02 -4.78522748e-01
4.31630254e-01 -1.19490337e+00 -2.80718505e-01 7.00677812e-01
-4.46020424e-01 5.83644927e-01 1.21836913e+00 -5.67342579e-01
-1.59162390e+00 -6.32446885e-01 7.94510841e-01 -2.14276016e-01
4.72332925e-01 2.38445088e-01 -4.22871917e-01 8.06456387e-01
5.24719298e-01 -7.62384832e-01 1.17584598e+00 2.54336119e-01
3.23317319e-01 -1.92947030e-01 -1.43484128e+00 8.75083447e-01
7.19645798e-01 6.06017523e-02 -9.92518008e-01 -7.22280294e-02
1.09660327e-01 -3.92968431e-02 -6.83005571e-01 9.23615247e-02
8.79764557e-01 -8.05496275e-01 1.05171585e+00 -7.64223039e-01
1.81285396e-01 -5.64498782e-01 -4.19816196e-01 -1.82256579e+00
-7.53056884e-01 -5.47444701e-01 4.63203527e-03 9.77062643e-01
7.20709860e-01 -6.90353930e-01 4.15584743e-01 1.31122935e+00
6.73298240e-02 -3.21894825e-01 -1.00927889e+00 -7.05281198e-01
-2.36372963e-01 -2.22226128e-01 9.16585505e-01 8.44408274e-01
1.13170609e-01 4.79282103e-02 -6.43302798e-01 8.18762004e-01
1.09341192e+00 5.56234837e-01 5.33394337e-01 -1.32577693e+00
-5.60852170e-01 -3.85524362e-01 3.72617662e-01 -1.00033474e+00
1.37032658e-01 -4.15369868e-01 3.24260652e-01 -1.92956686e+00
6.71317577e-02 -5.07135510e-01 -7.19970047e-01 2.78448105e-01
-1.26128674e-01 -7.63695776e-01 8.78518969e-02 -9.32555571e-02
-8.87750030e-01 5.11124551e-01 1.23349154e+00 -3.44160944e-01
-3.48725706e-01 4.99057949e-01 -6.15393162e-01 9.34147298e-01
9.48188603e-01 -6.40713155e-01 -8.01430643e-01 -7.29158342e-01
4.44619983e-01 4.43253905e-01 7.98131898e-02 -7.60193348e-01
4.52311814e-01 -1.42478287e+00 5.23684919e-02 -6.92194819e-01
5.19144297e-01 -1.20581961e+00 5.48148811e-01 6.06464148e-01
-4.80264544e-01 3.78506273e-01 -5.63888699e-02 7.61308551e-01
-2.40269173e-02 -3.64884019e-01 4.37611789e-01 -3.68031919e-01
-1.15445685e+00 2.69735098e-01 -1.18182921e+00 -3.41173649e-01
1.28079176e+00 -2.71331400e-01 1.11223347e-01 -4.97746319e-01
-5.20761371e-01 1.06691706e+00 3.21957827e-01 3.45882297e-01
7.17381299e-01 -1.10065591e+00 -7.76023686e-01 -3.81256521e-01
-1.05430745e-01 -6.84337988e-02 1.48496643e-01 7.40785778e-01
-1.03052981e-01 2.32803106e-01 -5.34497380e-01 -2.03881606e-01
-1.40653276e+00 3.39925498e-01 1.35436997e-01 -4.54189211e-01
1.99043546e-02 5.71021736e-01 -3.55791720e-03 -5.71584940e-01
3.55568707e-01 -3.58677596e-01 -4.96760666e-01 2.16933191e-01
6.47888899e-01 9.39533949e-01 -4.60370541e-01 -5.92908412e-02
-3.72665882e-01 3.98098916e-01 1.07837655e-01 -6.25511348e-01
1.51639366e+00 -4.05335128e-01 6.56669065e-02 8.61548841e-01
1.06965438e-01 -2.06434473e-01 -1.76192665e+00 -9.36173201e-02
1.30624235e-01 -8.36332262e-01 2.22470954e-01 -1.11620200e+00
-5.21594524e-01 7.36938834e-01 4.46398735e-01 -1.30250797e-01
9.72071052e-01 -5.24781883e-01 3.21286559e-01 5.26864886e-01
9.30508673e-01 -1.65115798e+00 -1.07663661e-01 4.33785498e-01
6.53997242e-01 -1.30149770e+00 -1.53934099e-02 7.36312717e-02
-1.09804225e+00 8.30068946e-01 3.99158776e-01 1.59924626e-01
5.62569857e-01 -5.30237965e-02 3.01309049e-01 -1.41054653e-02
-1.23292732e+00 3.34063061e-02 1.12086147e-01 7.12044954e-01
-4.43533696e-02 6.89286828e-01 -1.95169479e-01 4.12755460e-01
1.51772857e-01 1.63084611e-01 4.65223849e-01 1.12238264e+00
-5.80290377e-01 -1.25673890e+00 -4.82046068e-01 1.84708536e-01
-1.59879848e-01 2.35305697e-01 -2.31260821e-01 3.84477079e-01
-9.05540958e-02 1.67512941e+00 -2.92717423e-02 -2.27560639e-01
2.81636566e-01 4.35775565e-03 2.27497146e-01 -3.77520323e-01
-4.76092130e-01 -3.04484785e-01 5.49782515e-01 -6.67600870e-01
-9.40925896e-01 -1.10167193e+00 -1.12403822e+00 -6.13889396e-01
1.99878559e-01 4.87339944e-01 4.01162475e-01 1.14935529e+00
7.89547637e-02 4.09293532e-01 6.83356345e-01 -9.08407092e-01
-8.75372648e-01 -7.55533457e-01 -5.78976631e-01 2.86107033e-01
3.76011074e-01 -1.06033850e+00 -2.77779877e-01 -2.91955292e-01] | [4.2480149269104, 2.4015700817108154] |
3852319c-6287-481b-9845-27e89e3b30dc | find-a-reasonable-ending-for-stories-does | 1812.05411 | null | http://arxiv.org/abs/1812.05411v1 | http://arxiv.org/pdf/1812.05411v1.pdf | Find a Reasonable Ending for Stories: Does Logic Relation Help the Story Cloze Test? | Natural language understanding is a challenging problem that covers a wide
range of tasks. While previous methods generally train each task separately, we
consider combining the cross-task features to enhance the task performance. In
this paper, we incorporate the logic information with the help of the Natural
Language Inference (NLI) task to the Story Cloze Test (SCT). Previous work on
SCT considered various semantic information, such as sentiment and topic, but
lack the logic information between sentences which is an essential element of
stories. Thus we propose to extract the logic information during the course of
the story to improve the understanding of the whole story. The logic
information is modeled with the help of the NLI task. Experimental results
prove the strength of the logic information. | ['Mingyue Shang', 'Hongzhi Yin', 'Zhenxin Fu', 'Rui Yan', 'Dongyan Zhao', 'Bo Tang'] | 2018-12-13 | null | null | null | null | ['cloze-test'] | ['natural-language-processing'] | [ 1.66861504e-01 3.73029783e-02 -5.27247488e-01 -7.88843095e-01
-4.97194290e-01 -5.50982356e-01 8.79024148e-01 1.82723567e-01
-7.85160437e-02 7.21915364e-01 7.27202058e-01 -1.70611873e-01
-1.14408910e-01 -9.86240625e-01 -8.69316578e-01 -1.47214830e-01
4.29456502e-01 2.74564028e-01 3.30466509e-01 -5.15652180e-01
4.24163073e-01 -3.34536374e-01 -1.34877515e+00 9.95451689e-01
1.17603767e+00 1.04263628e+00 1.86957106e-01 3.59368533e-01
-5.48803210e-01 1.47803104e+00 -4.31553066e-01 -4.89788681e-01
-1.00714475e-01 -6.57837689e-01 -1.07270277e+00 6.18557446e-02
7.36426711e-02 -3.65480125e-01 -2.22422313e-02 9.21167850e-01
1.15367798e-02 -4.70855534e-02 5.20379424e-01 -1.56869209e+00
-4.56632048e-01 1.18611825e+00 -3.34529608e-01 6.99997880e-03
8.51246238e-01 -1.33225217e-01 1.50814211e+00 -8.28183174e-01
7.58860171e-01 1.45149362e+00 6.13416433e-01 1.74233049e-01
-4.69587833e-01 -6.11595809e-01 3.95415008e-01 7.83702195e-01
-9.55361903e-01 -1.66279063e-01 1.17863071e+00 -6.27890408e-01
7.75600612e-01 -3.36498879e-02 5.09808064e-01 1.05375540e+00
-3.73806991e-02 1.35611415e+00 1.32680178e+00 -5.14125943e-01
8.67703035e-02 2.55811632e-01 7.16389775e-01 6.27817154e-01
2.61260476e-02 -3.21581870e-01 -7.82112300e-01 1.89837798e-01
2.60727137e-01 -4.37600464e-02 -2.61614054e-01 2.01781318e-01
-1.15128744e+00 1.20318234e+00 2.56063849e-01 4.20261770e-01
-6.32059341e-03 -6.65520653e-02 5.51082611e-01 3.49442452e-01
4.35583353e-01 5.23846328e-01 -4.09656644e-01 -3.21788430e-01
-7.05705225e-01 4.85262036e-01 1.07300329e+00 1.05953979e+00
6.31125629e-01 -5.09259880e-01 -4.07326490e-01 7.29456186e-01
2.95309365e-01 1.24245979e-01 1.57139435e-01 -8.55507851e-01
6.88356936e-01 1.11574304e+00 -8.45993906e-02 -1.09581828e+00
-3.28406543e-01 -1.46353945e-01 -5.17509222e-01 -2.83447504e-01
3.64151895e-01 -1.34593815e-01 -4.55745786e-01 1.80818629e+00
3.18434417e-01 1.19720407e-01 1.28139094e-01 7.11525202e-01
1.25663769e+00 6.66247308e-01 -2.17421949e-01 -1.55393004e-01
1.59358668e+00 -1.09827733e+00 -1.14670956e+00 -4.70139354e-01
7.90129423e-01 -5.50425351e-01 1.29235125e+00 4.77829903e-01
-7.25776315e-01 -2.66754955e-01 -1.22522342e+00 -2.59053409e-01
-6.33470356e-01 8.90133902e-02 8.52842331e-01 1.17543533e-01
-4.29958910e-01 2.66291857e-01 -2.60844439e-01 -2.71125436e-01
5.23438692e-01 -1.47626013e-01 -8.40709731e-02 -6.56634510e-01
-1.86315751e+00 8.86829734e-01 6.43166780e-01 -1.69554070e-01
-6.20021999e-01 -5.70613742e-01 -1.05613923e+00 5.61015792e-02
9.92235124e-01 -6.27274454e-01 1.25244570e+00 -9.91119206e-01
-1.20033538e+00 7.01620460e-01 -4.00659621e-01 -2.32027769e-01
6.03078961e-01 -5.17885447e-01 -6.29494041e-02 7.39543363e-02
4.01889026e-01 3.71585578e-01 4.15740073e-01 -1.08792841e+00
-6.42403305e-01 -1.55177817e-01 7.52130091e-01 2.51502842e-01
-2.46903375e-01 2.50106096e-01 -4.60817337e-01 -6.86115921e-01
-8.32726657e-02 -6.51785672e-01 3.32282662e-01 -4.36688006e-01
-6.66632831e-01 -7.85848141e-01 7.52090871e-01 -7.00040579e-01
1.45930803e+00 -2.05401683e+00 4.13412526e-02 -2.74977326e-01
2.29126606e-02 -3.07832420e-01 7.80079663e-02 4.55833375e-01
1.24478318e-01 2.46257246e-01 -1.82166770e-01 -9.40165892e-02
2.65817463e-01 3.03330243e-01 -4.79011089e-01 -6.90132231e-02
3.43893826e-01 1.08616769e+00 -8.39142144e-01 -5.62061548e-01
-3.20642591e-01 -1.21147759e-01 -6.22423530e-01 1.92270547e-01
-7.53911078e-01 3.64812940e-01 -7.48791695e-01 4.23530787e-01
3.99066806e-01 -3.78332466e-01 -2.02631265e-01 -1.11668564e-01
2.53502488e-01 8.44423890e-01 -9.24991608e-01 1.65823698e+00
-3.99931908e-01 6.06862485e-01 -2.81235933e-01 -9.73321378e-01
7.16096222e-01 3.68179619e-01 1.27521545e-01 -3.63640189e-01
3.01816016e-01 -1.17934451e-01 7.89534450e-02 -9.47912157e-01
1.53909802e-01 -4.50073838e-01 -5.52745521e-01 5.65695643e-01
-1.82256743e-01 -4.65916812e-01 4.17225003e-01 5.12100101e-01
9.78687227e-01 -1.03054699e-02 4.82930213e-01 -1.85429692e-01
5.02917290e-01 4.04107213e-01 6.33837342e-01 7.92299867e-01
1.12244219e-01 3.70049626e-01 1.05684376e+00 -1.93553418e-01
-7.23276913e-01 -5.57606399e-01 7.03663053e-03 1.03648269e+00
3.22142631e-01 -6.49624944e-01 -6.76469028e-01 -9.10125554e-01
-1.70415372e-01 1.24837208e+00 -7.35536873e-01 -1.66183680e-01
-4.60938573e-01 -5.00217378e-01 3.46607655e-01 6.77570999e-01
8.80781353e-01 -9.62951362e-01 -3.02689999e-01 -2.32306775e-02
-7.10094273e-01 -1.59673059e+00 -1.72503561e-01 -3.48730348e-02
-4.21330690e-01 -1.05673707e+00 2.96081871e-01 -4.88610387e-01
1.45896375e-01 7.61358440e-02 1.17911458e+00 -3.91847380e-02
2.00302348e-01 1.89597130e-01 -8.34482491e-01 -7.48007357e-01
-4.11756337e-01 -1.01606004e-01 -2.81836241e-01 -9.83318090e-02
6.18326366e-01 -3.07422310e-01 1.17628098e-01 2.31639981e-01
-9.57482636e-01 5.44631839e-01 2.25087419e-01 8.37853611e-01
2.65624702e-01 8.24997127e-01 6.69122934e-01 -1.18508446e+00
7.83317327e-01 -5.37123859e-01 -2.96114304e-04 3.26728046e-01
-2.61122733e-01 5.40876687e-02 5.97036779e-01 -3.90462220e-01
-1.31064272e+00 -4.46386665e-01 6.61231054e-04 8.43698233e-02
4.12816368e-02 1.03199720e+00 -8.02324533e-01 6.32421315e-01
8.34901407e-02 -1.52401831e-02 -4.68645871e-01 -6.16072975e-02
2.76790947e-01 4.26906198e-01 2.32271582e-01 -6.68070436e-01
5.38635075e-01 4.29245412e-01 -3.26471597e-01 -3.62348497e-01
-1.87081647e+00 -3.94680709e-01 -4.51512516e-01 -3.16091627e-01
8.57596815e-01 -9.20814574e-01 -4.54426557e-01 2.56163239e-01
-1.43989992e+00 -2.37593874e-01 -1.32290930e-01 5.73738873e-01
-5.27727306e-01 6.19216003e-02 -5.32867134e-01 -6.84828699e-01
3.41448747e-02 -1.06259716e+00 9.46171224e-01 6.10294081e-02
-6.19119108e-01 -1.02024734e+00 -1.77689716e-01 8.09640110e-01
-1.08685903e-01 3.93938899e-01 1.23729634e+00 -9.16647792e-01
-6.49600685e-01 -3.91141891e-01 -2.65227884e-01 2.74299473e-01
-2.99019963e-02 -1.27095729e-01 -9.68253851e-01 4.16182309e-01
5.88473082e-01 -7.68872082e-01 1.06714439e+00 3.00158411e-01
9.08467472e-01 -4.72086310e-01 -2.13504821e-01 1.29116520e-01
1.34673285e+00 9.88247469e-02 6.76100552e-01 4.66027707e-01
7.70289481e-01 8.45614970e-01 9.71260905e-01 4.30103034e-01
9.18230891e-01 4.43857908e-01 1.03846654e-01 1.87028110e-01
8.40036646e-02 -4.82006043e-01 4.78043407e-01 7.43051291e-01
1.59781665e-01 -3.58542442e-01 -1.00356996e+00 4.45797533e-01
-2.00087500e+00 -9.63299453e-01 -2.66537130e-01 1.30600655e+00
1.10216367e+00 4.96872514e-01 -3.01710546e-01 3.60025972e-01
4.88808483e-01 3.42641443e-01 -3.40544105e-01 -3.42750043e-01
-4.00239408e-01 -3.86600524e-01 -1.09563008e-01 6.01065457e-01
-9.68097091e-01 1.12598205e+00 5.91297197e+00 1.07286894e+00
-6.29174054e-01 1.05252869e-01 5.50349355e-01 8.99318457e-02
-6.58657730e-01 2.77348101e-01 -8.30610454e-01 1.76190302e-01
2.95958251e-01 -1.00605071e-01 2.33056277e-01 8.26355994e-01
3.11377287e-01 -5.86230874e-01 -1.42391002e+00 5.39229155e-01
4.69481021e-01 -9.88369584e-01 1.67561978e-01 -5.08765399e-01
7.47908950e-01 -4.35623914e-01 -3.51966351e-01 5.58992982e-01
3.82429749e-01 -1.13194954e+00 9.08076525e-01 5.18786728e-01
4.36886609e-01 -6.34072483e-01 1.08083308e+00 7.52850413e-01
-9.97588694e-01 -9.95163899e-03 7.23044947e-02 -6.70733333e-01
2.91554123e-01 8.30902576e-01 -7.52060950e-01 5.39660752e-01
6.37388945e-01 1.22314215e+00 -6.82589829e-01 4.02217835e-01
-9.32928860e-01 7.90251315e-01 -1.16693571e-01 -3.56687576e-01
4.15931016e-01 3.04203155e-03 5.22897363e-01 1.07041419e+00
1.33236069e-02 3.32402825e-01 3.94465327e-01 1.08801067e+00
-1.58213124e-01 2.92424392e-02 -7.47316360e-01 -3.56178939e-01
1.81787148e-01 7.79603302e-01 -5.49925029e-01 -6.51111245e-01
-6.06723964e-01 6.42109275e-01 2.39242598e-01 1.68851078e-01
-9.13477838e-01 -2.07434490e-01 1.84726536e-01 1.94701124e-02
2.65973628e-01 5.41501269e-02 -9.28051293e-01 -1.33059597e+00
3.32088590e-01 -7.29260802e-01 4.18214321e-01 -1.03637528e+00
-1.44580925e+00 2.24440649e-01 3.44122827e-01 -1.01439905e+00
-9.84741822e-02 -4.34465230e-01 -1.07593441e+00 4.16508645e-01
-1.69762087e+00 -1.32517993e+00 -4.25290644e-01 3.05825204e-01
9.71868038e-01 8.66416544e-02 2.49878630e-01 -2.04105496e-01
-4.22403902e-01 3.08080196e-01 -4.51671571e-01 3.23621392e-01
6.55851126e-01 -1.10193574e+00 -1.64362267e-01 7.78423369e-01
-1.63795471e-01 5.95136106e-01 8.26984644e-01 -8.84406984e-01
-1.09992516e+00 -9.22289610e-01 1.25297070e+00 -5.39156318e-01
9.00215983e-01 -4.07092661e-01 -1.02506065e+00 7.61730850e-01
4.16448474e-01 -6.22176111e-01 8.21960926e-01 3.56375515e-01
-7.04316676e-01 1.44266576e-01 -7.85204113e-01 5.40740967e-01
9.07514334e-01 -6.67999208e-01 -1.20199966e+00 4.71740782e-01
1.15724921e+00 -1.44562677e-01 -5.90443313e-01 6.37746036e-01
3.52729887e-01 -7.52412021e-01 5.78161478e-01 -6.86942220e-01
1.33946061e+00 -3.29989672e-01 -2.35042363e-01 -1.13086843e+00
-6.38436712e-03 -6.46965802e-02 1.41064718e-01 1.50943375e+00
6.31334603e-01 -1.31050497e-01 3.58518839e-01 6.99593842e-01
7.26112723e-02 -7.95054853e-01 -5.00071764e-01 -4.32913631e-01
4.60671373e-02 -9.84324694e-01 5.04414916e-01 1.16154230e+00
6.60465419e-01 9.59509134e-01 -2.89225578e-01 -4.24652025e-02
1.13171183e-01 5.26927292e-01 6.64998353e-01 -1.20227075e+00
-2.86687076e-01 -4.38446790e-01 -3.00593097e-02 -8.94312561e-01
5.55421650e-01 -1.02994382e+00 1.00456960e-01 -1.91661036e+00
7.75878191e-01 1.38265893e-01 1.12218060e-01 4.60804909e-01
-6.15356922e-01 -4.40700918e-01 8.81353691e-02 -1.78229272e-01
-9.15618300e-01 6.52273953e-01 1.43273139e+00 -3.61287057e-01
-2.82676518e-02 -2.01075211e-01 -9.78500962e-01 1.15082932e+00
6.49791121e-01 -4.64669526e-01 -7.07246304e-01 -4.81306076e-01
6.00613713e-01 1.20592386e-01 3.31861854e-01 -6.61592305e-01
2.97899812e-01 -5.12940347e-01 7.14528337e-02 -7.49936759e-01
1.95523918e-01 -7.06916630e-01 -3.06953460e-01 1.27068814e-02
-6.12838507e-01 -2.55142033e-01 1.45779029e-01 4.86699402e-01
-7.32328176e-01 -4.49100614e-01 4.22603041e-01 -2.46014446e-01
-6.45046890e-01 -8.25545415e-02 -1.52854457e-01 3.45410973e-01
9.35689747e-01 -1.02297612e-01 -3.41394633e-01 -7.03293562e-01
-4.05585080e-01 5.63765585e-01 3.70757014e-01 4.58868831e-01
6.69089854e-01 -1.26471043e+00 -8.82777333e-01 -7.71348253e-02
4.06981617e-01 3.16635430e-01 1.31025985e-01 8.68850291e-01
-1.03058562e-01 2.13094011e-01 1.49724543e-01 -4.49700981e-01
-1.07977796e+00 5.28689861e-01 -9.10391435e-02 -6.09999776e-01
-2.56666511e-01 8.56604397e-01 5.69740117e-01 -2.27801338e-01
1.35700643e-01 -4.42232639e-01 -7.86923528e-01 2.89146453e-01
7.39131749e-01 -1.11164190e-02 -3.70381683e-01 -3.27257544e-01
-1.70787632e-01 5.49349368e-01 -4.82879840e-02 -8.16661492e-02
1.45317900e+00 -1.91318467e-01 -5.54915786e-01 9.11978602e-01
1.10646021e+00 -8.44233185e-02 -8.70555937e-01 -6.98047996e-01
4.81489003e-01 -3.50901276e-01 3.97191569e-02 -9.41685736e-01
-5.03478110e-01 8.78533781e-01 -4.39956725e-01 1.44458890e-01
1.10815871e+00 4.25542206e-01 8.51379991e-01 3.55613530e-01
2.81292200e-01 -1.20021510e+00 2.22060353e-01 9.92572486e-01
1.08422303e+00 -1.45738435e+00 1.04650877e-01 -9.23160076e-01
-1.16156507e+00 9.82393980e-01 6.38730109e-01 1.14966057e-01
5.07132411e-01 5.75842857e-01 -2.47332692e-01 -3.85277480e-01
-9.78534400e-01 -1.68049321e-01 3.03859740e-01 4.15426381e-02
4.35838521e-01 -1.84479926e-04 -3.84816796e-01 1.37064755e+00
-3.02100301e-01 1.96819842e-01 5.32109678e-01 9.61714566e-01
-5.79016447e-01 -6.84403718e-01 -2.14643568e-01 4.94805753e-01
-2.15456292e-01 -1.40423611e-01 -8.80680203e-01 6.23305559e-01
3.14418793e-01 1.19647241e+00 -3.13354880e-01 -4.39353883e-01
2.67397940e-01 2.51433492e-01 1.82921633e-01 -6.32378221e-01
-4.74131286e-01 -1.08286828e-01 5.19152403e-01 -3.36779416e-01
-7.27151453e-01 -5.64701796e-01 -1.34706748e+00 -3.51338685e-01
-3.43955129e-01 6.18498735e-02 2.84212887e-01 1.74783671e+00
-2.49612212e-01 6.43369853e-01 4.55157042e-01 -6.10908754e-02
-4.28013457e-03 -1.18957484e+00 -5.00733316e-01 6.42937839e-01
1.07960559e-01 -8.17743361e-01 -3.54252279e-01 2.10498855e-01] | [11.085335731506348, 8.857656478881836] |
b9359e12-44ca-4f4f-a31f-0e960e0ca274 | mutual-adaptive-reasoning-for-monocular-3d | 2207.07900 | null | https://arxiv.org/abs/2207.07900v1 | https://arxiv.org/pdf/2207.07900v1.pdf | Mutual Adaptive Reasoning for Monocular 3D Multi-Person Pose Estimation | Inter-person occlusion and depth ambiguity make estimating the 3D poses of monocular multiple persons as camera-centric coordinates a challenging problem. Typical top-down frameworks suffer from high computational redundancy with an additional detection stage. By contrast, the bottom-up methods enjoy low computational costs as they are less affected by the number of humans. However, most existing bottom-up methods treat camera-centric 3D human pose estimation as two unrelated subtasks: 2.5D pose estimation and camera-centric depth estimation. In this paper, we propose a unified model that leverages the mutual benefits of both these subtasks. Within the framework, a robust structured 2.5D pose estimation is designed to recognize inter-person occlusion based on depth relationships. Additionally, we develop an end-to-end geometry-aware depth reasoning method that exploits the mutual benefits of both 2.5D pose and camera-centric root depths. This method first uses 2.5D pose and geometry information to infer camera-centric root depths in a forward pass, and then exploits the root depths to further improve representation learning of 2.5D pose estimation in a backward pass. Further, we designed an adaptive fusion scheme that leverages both visual perception and body geometry to alleviate inherent depth ambiguity issues. Extensive experiments demonstrate the superiority of our proposed model over a wide range of bottom-up methods. Our accuracy is even competitive with top-down counterparts. Notably, our model runs much faster than existing bottom-up and top-down methods. | ['Jingyi Yu', 'Lan Xu', 'Fei Gao', 'Ye Shi', 'Jingya Wang', 'Juze Zhang'] | 2022-07-16 | null | null | null | null | ['3d-multi-person-pose-estimation', 'multi-person-pose-estimation'] | ['computer-vision', 'computer-vision'] | [-2.18421683e-01 1.08983237e-02 -9.59517807e-02 -4.10595715e-01
-6.00277305e-01 -4.83932495e-01 3.89224976e-01 -1.78183451e-01
-1.48958117e-01 2.90576935e-01 3.87285858e-01 2.60213882e-01
2.12277561e-01 -6.52280331e-01 -5.89323997e-01 -2.53894657e-01
2.93206483e-01 5.65102041e-01 2.67507792e-01 -1.33143499e-01
7.35456720e-02 4.37664092e-01 -1.57201028e+00 -1.70979239e-02
7.29165971e-01 1.02157044e+00 -9.31950137e-02 6.45847619e-01
3.94942403e-01 4.18013394e-01 -4.08487588e-01 -4.27685052e-01
6.66557431e-01 -3.24947350e-02 -2.54171729e-01 5.21743894e-01
9.44097579e-01 -6.74124658e-01 -5.63436687e-01 7.41139293e-01
7.74082601e-01 1.94705017e-02 1.48294240e-01 -1.29657435e+00
-1.73621446e-01 -2.98859894e-01 -8.89648259e-01 -1.61512628e-01
1.14298940e+00 2.43041500e-01 6.86687827e-01 -1.25180960e+00
4.35667187e-01 1.64792740e+00 8.72835398e-01 3.81336570e-01
-8.43263507e-01 -4.78563428e-01 7.94037044e-01 -2.02605948e-01
-1.51891363e+00 -2.31698394e-01 9.32995379e-01 -5.38269162e-01
8.54045868e-01 1.20527357e-01 1.00660992e+00 9.39177096e-01
-9.76651981e-02 9.07961428e-01 9.50815141e-01 -2.86025852e-01
-5.13247587e-02 5.29180355e-02 -5.55348545e-02 9.67620075e-01
7.18586862e-01 5.45082577e-02 -8.05056691e-01 1.67727191e-02
1.07996416e+00 4.25673544e-01 -1.77318946e-01 -9.11738336e-01
-1.08730078e+00 5.64285696e-01 6.32021368e-01 -3.74212146e-01
-2.38500386e-01 2.15848219e-02 8.96447673e-02 -1.81703806e-01
4.53074217e-01 3.96608673e-02 -4.64712977e-01 8.24787617e-02
-7.01970816e-01 6.46481335e-01 5.69383085e-01 1.41899157e+00
6.86250627e-01 -4.36085999e-01 -1.34493470e-01 4.88854378e-01
5.66498578e-01 5.56249499e-01 -3.28334682e-02 -9.52115476e-01
8.79558563e-01 1.05357766e+00 3.07755530e-01 -1.12925947e+00
-7.04779863e-01 -6.24479711e-01 -5.83021402e-01 1.08897083e-01
5.70000112e-01 -1.06445543e-01 -7.75932431e-01 1.78021181e+00
9.04768050e-01 -2.96431065e-01 -3.26428771e-01 1.36018479e+00
6.98900998e-01 8.21229517e-02 -1.27713874e-01 2.50548065e-01
1.64653385e+00 -1.23437572e+00 -3.80750656e-01 -7.97766328e-01
4.55834717e-01 -4.82925326e-01 7.26725638e-01 3.91893983e-01
-1.16623199e+00 -7.96616614e-01 -1.06411529e+00 -5.38164496e-01
-9.76335928e-02 4.72696126e-01 7.65501082e-01 7.52239943e-01
-7.17755437e-01 1.19475000e-01 -8.00838113e-01 -4.89181727e-01
1.54576257e-01 3.89988512e-01 -3.91291708e-01 -3.59601349e-01
-8.65173638e-01 7.05742598e-01 9.79109854e-02 2.59464622e-01
-5.20935237e-01 -6.49354100e-01 -1.21678030e+00 -3.43751281e-01
7.62577295e-01 -1.54187560e+00 1.23462343e+00 -1.89961568e-01
-1.36763859e+00 9.12832320e-01 -4.50843871e-01 -9.60136484e-03
1.04970920e+00 -8.41045141e-01 2.21714556e-01 3.32485825e-01
2.57937223e-01 6.85190201e-01 6.33819401e-01 -1.36032867e+00
-7.20372677e-01 -1.02016306e+00 4.15069371e-01 8.59789431e-01
-1.40278965e-01 -5.87503910e-01 -1.17318070e+00 -4.86927539e-01
7.28308618e-01 -9.84356999e-01 -2.96123683e-01 5.69836795e-01
-5.65491021e-01 -5.60820922e-02 5.16500771e-01 -6.08100355e-01
9.17080581e-01 -1.77412975e+00 4.15497899e-01 1.26924992e-01
4.77402508e-01 -1.41903222e-01 2.90040046e-01 3.14485610e-01
3.75383049e-01 -3.54199708e-01 1.75572991e-01 -9.60730970e-01
1.86063856e-01 -1.05519816e-01 1.81714311e-01 6.85230553e-01
-2.56900024e-03 7.47793674e-01 -8.55062783e-01 -5.82798779e-01
6.96407199e-01 7.58199096e-01 -7.78990984e-01 2.26950854e-01
1.44543409e-01 4.14764404e-01 -5.25479198e-01 1.01090801e+00
8.36465776e-01 -2.17645556e-01 1.64964214e-01 -4.65490818e-01
-5.38820699e-02 1.63017720e-01 -1.37265563e+00 2.19011378e+00
-3.12680244e-01 9.05498043e-02 6.32724091e-02 -3.84002924e-01
6.57237113e-01 1.87369809e-01 4.29968745e-01 -4.36585665e-01
1.05683938e-01 2.28928085e-02 -5.82805812e-01 -2.15019271e-01
4.45769370e-01 -1.85903274e-02 -3.02044183e-01 -1.07582323e-01
-2.19301879e-01 -9.52982306e-02 -1.06221020e-01 2.31618866e-01
7.26504326e-01 6.99101090e-01 5.74965537e-01 2.83594042e-01
5.56310534e-01 -1.64879799e-01 8.76124322e-01 5.54609835e-01
-4.46511745e-01 8.75683963e-01 2.10886061e-01 -5.10135353e-01
-8.57213497e-01 -1.11017942e+00 1.89429641e-01 7.50795960e-01
6.31125629e-01 -7.84516573e-01 -6.59609199e-01 -5.25208414e-01
3.40431303e-01 -6.54375851e-02 -5.95675886e-01 7.61350840e-02
-4.98965055e-01 -4.45750028e-01 2.56896079e-01 8.73400509e-01
7.73124695e-01 -2.68403459e-02 -8.75745237e-01 -8.94848704e-02
-4.20627713e-01 -1.64448464e+00 -6.59472287e-01 -4.49454188e-01
-9.66994345e-01 -1.16625512e+00 -9.20336127e-01 -5.14738202e-01
7.23634064e-01 8.08319807e-01 9.18405354e-01 5.51700406e-02
-3.44531834e-01 6.31929338e-01 -1.36760280e-01 -2.14821801e-01
5.62423825e-01 -7.13320151e-02 3.67558837e-01 -1.37358829e-01
4.08262789e-01 -4.69457507e-01 -1.12420237e+00 4.69901294e-01
-6.83504567e-02 3.87052059e-01 6.93640053e-01 3.81697685e-01
5.99349260e-01 -1.30525842e-01 -4.08000872e-02 -5.74277818e-01
-8.25136155e-02 -2.84532476e-02 -5.27182519e-01 -1.07603036e-02
-2.65191883e-01 -3.42391938e-01 1.96659297e-01 -1.61902055e-01
-1.20021081e+00 5.14625609e-01 2.54699051e-01 -5.43900907e-01
-1.45845011e-01 5.60418777e-02 -6.71171367e-01 -7.23005086e-02
5.50815701e-01 -1.24168769e-01 -1.63546771e-01 -5.07784486e-01
4.06553298e-01 3.89574796e-01 7.48345673e-01 -6.49809122e-01
1.10792625e+00 8.11060190e-01 1.30623162e-01 -6.22080564e-01
-1.20083642e+00 -8.20117533e-01 -1.01551163e+00 -3.88946921e-01
9.20220733e-01 -1.67193842e+00 -1.08409965e+00 5.84286928e-01
-1.13791561e+00 2.70332962e-01 6.83285594e-02 5.38627565e-01
-4.52915609e-01 6.60072863e-01 -4.89842027e-01 -1.11169577e+00
-1.41568139e-01 -9.88152564e-01 1.76022947e+00 2.20735461e-01
-3.06592375e-01 -7.02651918e-01 -1.88804671e-01 9.42675412e-01
-3.50487888e-01 6.61017716e-01 8.46334398e-02 9.77622122e-02
-8.26840699e-01 -5.25270641e-01 -2.86700010e-01 -2.56207228e-01
-3.19743305e-02 -6.02602363e-01 -1.04544938e+00 -4.39565539e-01
-1.51426688e-01 -2.07807750e-01 6.41865253e-01 3.35867405e-01
7.53466308e-01 -1.04614273e-02 -4.73440796e-01 8.30083609e-01
1.21788859e+00 -4.03304815e-01 3.11839640e-01 4.86946493e-01
1.20965612e+00 8.76483560e-01 8.11148942e-01 7.70981669e-01
1.01997066e+00 1.10010135e+00 4.25013930e-01 -1.56221658e-01
-2.83534646e-01 -6.77167952e-01 3.31512719e-01 4.12771404e-01
-2.64033645e-01 1.01112552e-01 -7.59784937e-01 2.63009250e-01
-1.98371029e+00 -7.77746499e-01 -1.29718184e-01 2.24802876e+00
3.75206769e-01 2.37590834e-01 5.32566786e-01 1.27947658e-01
6.56280994e-01 3.21433209e-02 -8.27142954e-01 4.06928509e-01
3.54720391e-02 -4.02266681e-01 3.83831799e-01 4.68904614e-01
-1.21862435e+00 8.40033174e-01 5.17200613e+00 1.40090033e-01
-3.71941298e-01 -1.95455015e-01 1.25293672e-01 -4.23931479e-01
7.49937892e-02 -1.06595121e-01 -1.14888239e+00 1.60064802e-01
1.40233174e-01 3.63610923e-01 -9.18949097e-02 1.09688699e+00
1.00791529e-01 -2.61052340e-01 -1.35909879e+00 1.48918545e+00
3.76511604e-01 -7.65153706e-01 -1.21158056e-01 3.23393732e-01
5.71862519e-01 -5.26820004e-01 -1.54353350e-01 2.42949277e-01
-5.15383370e-02 -6.43715620e-01 1.01621866e+00 4.39970940e-01
7.16175973e-01 -7.84493446e-01 6.22572839e-01 4.96064514e-01
-1.72010851e+00 -2.03586772e-01 -1.88317806e-01 -5.64561605e-01
3.62070382e-01 6.31907701e-01 -4.06240433e-01 7.78860927e-01
7.35661685e-01 8.10149014e-01 -5.79104900e-01 8.70336890e-01
-3.97386283e-01 -4.00109112e-01 -3.58761728e-01 4.54204351e-01
-1.27592966e-01 4.46478128e-02 5.25597930e-01 9.31361616e-01
1.40151799e-01 2.74212062e-01 7.11168349e-01 5.59928894e-01
1.18455246e-01 -3.12339365e-01 -4.56773221e-01 5.37755728e-01
5.92029274e-01 1.08614814e+00 -4.10941571e-01 -2.79662311e-01
-5.18345296e-01 1.27515805e+00 3.25822473e-01 2.31532633e-01
-9.19120073e-01 -1.18880510e-01 8.79843831e-01 3.56371731e-01
2.34211385e-01 -3.37408543e-01 -4.71976161e-01 -1.64199793e+00
5.82875967e-01 -6.56903744e-01 4.02405143e-01 -7.92506278e-01
-1.20261669e+00 3.10446918e-01 1.36524126e-01 -1.51566815e+00
-1.96442157e-01 -7.53176570e-01 -3.69887277e-02 8.19253147e-01
-1.44408882e+00 -1.51936781e+00 -8.92233968e-01 7.12016761e-01
7.02666640e-01 2.48034835e-01 5.04451871e-01 2.38437876e-01
-5.98383188e-01 7.40076840e-01 -8.45566213e-01 2.45506436e-01
7.13391662e-01 -1.25149119e+00 4.41779315e-01 1.03602886e+00
-2.80227512e-01 8.26165795e-01 4.78508204e-01 -7.14371562e-01
-1.65068412e+00 -8.99478257e-01 9.31375027e-01 -8.28525245e-01
-5.39511293e-02 -6.80305004e-01 -2.40792423e-01 7.20111370e-01
-6.05828881e-01 2.37599630e-02 5.02063394e-01 3.62216592e-01
-6.91417873e-01 -1.39104024e-01 -1.17697108e+00 6.49110973e-01
1.62738597e+00 -4.94537801e-01 -6.61555111e-01 2.64831841e-01
6.12796485e-01 -9.20104802e-01 -7.96352088e-01 3.67807239e-01
9.11333859e-01 -1.20392752e+00 1.55961502e+00 -2.57617459e-02
1.14620775e-01 -6.51928961e-01 -3.09810817e-01 -6.88925326e-01
-3.44082952e-01 -6.16165936e-01 -5.60050845e-01 8.41848850e-01
-1.56511277e-01 -4.42191333e-01 1.05432892e+00 9.08904552e-01
-9.88039672e-02 -7.46396124e-01 -7.08710968e-01 -7.16009676e-01
-4.13919449e-01 -4.00169581e-01 3.34945798e-01 4.42285061e-01
-2.87650973e-02 4.57097501e-01 -6.08501852e-01 5.72100163e-01
1.11601090e+00 2.57347107e-01 1.40883672e+00 -1.43030417e+00
-3.66372138e-01 -6.29348308e-02 -6.95990324e-01 -1.81622410e+00
-3.86113375e-01 -3.30772340e-01 -4.94611624e-04 -1.57948053e+00
3.62246603e-01 -9.44796875e-02 2.02191189e-01 3.31107616e-01
-4.18942958e-01 3.59128505e-01 4.56878871e-01 2.34608263e-01
-6.73142970e-01 4.92569864e-01 1.21322608e+00 2.81589836e-01
-1.59195259e-01 -2.74492763e-02 -8.89954805e-01 1.09115756e+00
4.50787932e-01 -1.04696937e-01 -5.44704616e-01 -5.50832272e-01
2.47484758e-01 1.79161146e-01 8.25150251e-01 -1.35630143e+00
3.04790080e-01 2.64363941e-02 1.01090050e+00 -1.04736030e+00
9.63325739e-01 -6.66224897e-01 -9.90623012e-02 4.66038764e-01
1.65479362e-01 5.40809557e-02 -9.90052596e-02 8.28225434e-01
1.16841115e-01 5.77963948e-01 5.48993289e-01 -3.95724416e-01
-5.55658638e-01 6.29045129e-01 2.18841702e-01 -7.92988464e-02
1.07549334e+00 -8.14150691e-01 1.63660385e-02 -4.55798745e-01
-5.79572260e-01 4.42743212e-01 6.87131226e-01 5.49124539e-01
7.92989373e-01 -1.30193579e+00 -4.31806564e-01 1.73884705e-01
2.27881819e-01 5.11538088e-01 4.26688164e-01 9.14452732e-01
-4.10251111e-01 6.12862349e-01 6.99367076e-02 -1.00097907e+00
-1.39319611e+00 5.06696343e-01 2.99964577e-01 -2.61701226e-01
-7.46427476e-01 1.00520909e+00 6.44326806e-01 -5.34204364e-01
4.04814452e-01 -1.90577567e-01 2.18966529e-01 -1.00153089e-01
6.36765540e-01 5.49580693e-01 -1.39425606e-01 -8.16163957e-01
-6.33052826e-01 1.29657030e+00 -2.92461123e-02 -1.83814958e-01
1.04400206e+00 -6.55676663e-01 3.79358917e-01 2.02882096e-01
9.94389296e-01 5.81919625e-02 -1.76156831e+00 -2.73022354e-01
-4.27065402e-01 -8.98341894e-01 -1.95780829e-01 -6.78094506e-01
-8.85484636e-01 8.63577783e-01 3.29674006e-01 -6.00484073e-01
1.13066339e+00 -5.98648302e-02 8.29900146e-01 2.59034485e-01
6.94939315e-01 -1.14604211e+00 3.23456645e-01 2.48081669e-01
7.14257896e-01 -1.36569703e+00 4.89237964e-01 -1.11501932e+00
-4.37440991e-01 9.53362644e-01 1.11410117e+00 -1.12365196e-02
1.93623945e-01 -3.90892588e-02 -1.51743978e-01 -3.58925939e-01
-4.19442207e-01 -3.35838407e-01 5.65642595e-01 7.01833785e-01
2.53547877e-01 -1.49538869e-03 8.35884735e-02 5.91861546e-01
-3.03465366e-01 -4.12910730e-02 7.85081238e-02 1.07832026e+00
-3.64556879e-01 -9.24228191e-01 -7.42982030e-01 -2.20446378e-01
-1.94245167e-02 3.14790547e-01 -4.99742329e-01 1.03587937e+00
4.05278265e-01 1.08658135e+00 -1.61287487e-01 -4.75198478e-01
6.67080998e-01 -1.53619498e-01 8.67952585e-01 -6.49097145e-01
-2.96907634e-01 2.55539626e-01 4.90542240e-02 -9.84578073e-01
-4.47339445e-01 -7.91733265e-01 -1.09345508e+00 -3.79245341e-01
-1.47627950e-01 -4.65108514e-01 4.62675184e-01 8.22751880e-01
4.66538459e-01 2.16370091e-01 3.83524686e-01 -1.27733099e+00
-3.76716822e-01 -6.52395308e-01 -2.70052940e-01 6.31933212e-01
4.62711513e-01 -1.06026971e+00 -1.77071065e-01 -2.14870751e-01] | [7.024882793426514, -1.0023162364959717] |
42bcc423-f316-4336-9963-00e5efe26428 | inference-and-sampling-of-point-processes | 2306.00762 | null | https://arxiv.org/abs/2306.00762v1 | https://arxiv.org/pdf/2306.00762v1.pdf | Inference and Sampling of Point Processes from Diffusion Excursions | Point processes often have a natural interpretation with respect to a continuous process. We propose a point process construction that describes arrival time observations in terms of the state of a latent diffusion process. In this framework, we relate the return times of a diffusion in a continuous path space to new arrivals of the point process. This leads to a continuous sample path that is used to describe the underlying mechanism generating the arrival distribution. These models arise in many disciplines, such as financial settings where actions in a market are determined by a hidden continuous price or in neuroscience where a latent stimulus generates spike trains. Based on the developments in It\^o's excursion theory, we propose methods for inferring and sampling from the point process derived from the latent diffusion process. We illustrate the approach with numerical examples using both simulated and real data. The proposed methods and framework provide a basis for interpreting point processes through the lens of diffusions. | ['Vahid Tarokh', 'Anderson Schneider', 'Mohamed Abdelghani', 'Yuting Ng', 'Yu Chen', 'Ali Hasan'] | 2023-06-01 | null | null | null | null | ['point-processes'] | ['methodology'] | [ 3.01335514e-01 -1.12561911e-01 3.19516025e-02 1.69457924e-02
-2.80350298e-02 -5.80359161e-01 1.05738401e+00 3.01082015e-01
-4.33299929e-01 7.16671824e-01 1.66837156e-01 -1.81436762e-01
-4.27315086e-01 -9.16028917e-01 -6.92580342e-01 -1.01054013e+00
-8.10812339e-02 5.76950908e-01 -8.81717131e-02 -4.13281880e-02
4.50901866e-01 3.23390692e-01 -1.02238214e+00 -1.82689920e-01
5.10476589e-01 8.15686822e-01 1.19850866e-01 5.86338043e-01
-5.80472350e-01 2.24442691e-01 -6.35531306e-01 -2.72680581e-01
4.35100555e-01 -6.04304671e-01 -2.55010158e-01 2.51937151e-01
-6.00766361e-01 8.94352514e-03 -1.18053436e-01 1.15274620e+00
-3.68303694e-02 -9.47456807e-02 1.20341623e+00 -1.25699449e+00
-9.62331593e-01 5.59558690e-01 -6.00291848e-01 5.80251753e-01
1.14143111e-01 9.80009437e-02 6.95162237e-01 -6.65384114e-01
4.74174738e-01 1.36728132e+00 5.28467178e-01 3.17862213e-01
-1.69531417e+00 -3.15633386e-01 -4.55890745e-02 -4.43542659e-01
-1.04609299e+00 -7.29305819e-02 7.25535274e-01 -8.02415788e-01
1.99136510e-01 -7.44743422e-02 9.45016623e-01 1.48292291e+00
9.74401534e-01 8.31801474e-01 1.21457267e+00 -2.86394030e-01
8.21800590e-01 -2.97842681e-01 2.36947849e-01 5.60422577e-02
5.26808023e-01 2.04147384e-01 -4.09733742e-01 -6.05211496e-01
1.36084914e+00 5.58660567e-01 8.24427465e-04 -1.52143627e-01
-1.07306123e+00 1.02479100e+00 -4.52568270e-02 2.94789582e-01
-8.69060516e-01 3.84436488e-01 -1.29061177e-01 4.86760348e-01
7.09210694e-01 -3.01871486e-02 -1.00757778e-02 -1.31333977e-01
-8.05978954e-01 4.57614988e-01 1.10873055e+00 9.46351409e-01
3.23117495e-01 -1.51425779e-01 -2.25591436e-01 1.02110244e-01
6.52320087e-01 7.49030411e-01 1.82511911e-01 -7.39704549e-01
2.18909219e-01 1.45373523e-01 4.76369619e-01 -6.63504422e-01
5.54243736e-02 -3.92836809e-01 -7.70502865e-01 3.13846588e-01
7.97129095e-01 -1.44905567e-01 -6.59723103e-01 1.66248882e+00
-4.25807573e-02 4.27101791e-01 -1.83661729e-02 3.96694601e-01
-3.88975054e-01 1.01272106e+00 -2.31558480e-03 -7.03603804e-01
1.08791864e+00 -1.02948219e-01 -9.54152465e-01 5.30108333e-01
-2.35262699e-02 -4.48740333e-01 6.65786922e-01 5.05967617e-01
-1.54401302e+00 -1.58726797e-01 -4.88011062e-01 5.22702396e-01
-3.55482288e-02 -5.78776956e-01 2.89591908e-01 2.82017589e-01
-1.22198963e+00 6.56441867e-01 -1.08030784e+00 -5.24771571e-01
4.05372947e-01 -7.33884797e-03 4.48198140e-01 3.33475798e-01
-8.78787994e-01 3.63740921e-01 -1.31832808e-01 2.04741389e-01
-1.05505002e+00 -5.79554856e-01 -2.16978546e-02 6.48914501e-02
-1.25346258e-02 -1.05765128e+00 1.23687744e+00 -7.76410639e-01
-1.45244598e+00 4.14706022e-01 -2.03001171e-01 -8.26143205e-01
9.75150585e-01 3.33049446e-02 -3.13686073e-01 1.43313348e-01
3.71429026e-01 8.71777087e-02 1.40495431e+00 -8.95202994e-01
-5.71226180e-01 -3.86494964e-01 -4.94712174e-01 -1.15452006e-01
2.70175457e-01 -2.81345934e-01 2.43081152e-01 -7.70861804e-01
2.65154898e-01 -9.34289873e-01 -3.56849581e-01 8.77472907e-02
-2.74971247e-01 -2.51041800e-01 5.23775280e-01 -1.24874189e-02
7.75421977e-01 -2.19281077e+00 2.16899559e-01 3.23921502e-01
4.82098967e-01 -4.63975281e-01 1.26616865e-01 1.00907218e+00
3.51283327e-02 3.49934071e-01 -5.26858807e-01 -2.76161700e-01
2.58688126e-02 1.70625269e-01 -8.30477774e-01 7.66715169e-01
1.06495254e-01 7.17256129e-01 -9.73470271e-01 1.87338740e-01
-2.90650934e-01 4.71025646e-01 -1.03819706e-01 -1.63729310e-01
-2.84513026e-01 7.56964147e-01 -6.73418045e-01 1.66197971e-01
3.64175707e-01 -1.90258861e-01 -4.52556521e-01 7.12321579e-01
-4.59966004e-01 -2.46914431e-01 -1.11641240e+00 1.31783533e+00
-1.19428769e-01 5.08443892e-01 -9.45187584e-02 -7.01704800e-01
1.20931304e+00 4.90515471e-01 5.13696849e-01 -1.56353801e-01
1.59336731e-01 3.54756176e-01 2.29769230e-01 -1.29436031e-01
2.46904209e-01 -6.18265510e-01 1.14437424e-01 7.80632377e-01
-1.49960771e-01 1.94897372e-02 1.14697754e-01 -1.72870606e-02
1.30855048e+00 -1.57630503e-01 1.15570225e-01 -6.10125899e-01
1.97168931e-01 -2.87171751e-01 3.94572437e-01 8.56452942e-01
-6.81495741e-02 1.40656695e-01 8.98134291e-01 -3.86769265e-01
-1.40177822e+00 -1.65556169e+00 -3.73520076e-01 1.26037493e-01
-1.65792182e-01 2.06937790e-01 -6.25139713e-01 2.63066262e-01
1.50957599e-01 6.77468598e-01 -8.75650465e-01 -6.64166957e-02
-2.06449136e-01 -7.85223782e-01 2.14739755e-01 2.23135322e-01
1.85741469e-01 -1.11661530e+00 -6.98810399e-01 6.66917443e-01
2.66650379e-01 -6.26186967e-01 -4.09377098e-01 -4.70205583e-02
-1.35588372e+00 -4.51836795e-01 -1.16027212e+00 -3.69595081e-01
6.82539999e-01 -1.97291479e-01 6.09837651e-01 -4.90698695e-01
-2.78024137e-01 7.49577820e-01 1.45459510e-02 -8.67985606e-01
-6.74797356e-01 -3.94546688e-01 2.68538326e-01 5.88464379e-01
2.93319613e-01 -7.37712443e-01 -8.44443500e-01 1.91382188e-02
-1.36817110e+00 -3.10528636e-01 3.60116720e-01 4.34899837e-01
7.07622588e-01 3.72591585e-01 7.32913077e-01 -4.49869663e-01
1.27403307e+00 -8.24264705e-01 -1.00365543e+00 -2.40879521e-01
-5.21763742e-01 3.25607210e-01 2.86936581e-01 -6.53240621e-01
-1.03687489e+00 -3.03116143e-01 5.44686139e-01 -4.73476112e-01
-1.85848281e-01 3.64017516e-01 2.50261307e-01 4.86231595e-01
2.42131323e-01 4.49273139e-01 2.08070755e-01 -4.22083020e-01
4.80339617e-01 1.14029862e-01 2.52749175e-01 -6.52908981e-01
7.41019726e-01 1.10231423e+00 5.37516892e-01 -9.31508958e-01
-2.86580771e-01 -2.94709504e-01 -6.61074698e-01 -4.88969199e-02
1.00587475e+00 -5.14076889e-01 -9.12580431e-01 6.84099257e-01
-1.39914823e+00 -2.55291343e-01 -9.82491672e-01 6.49040222e-01
-1.13912964e+00 -2.45920658e-01 -1.02010441e+00 -1.43953919e+00
2.27605715e-01 -7.90913045e-01 9.39555109e-01 7.05850050e-02
-2.06293479e-01 -1.67972267e+00 4.59376723e-01 -4.75384325e-01
3.55182409e-01 2.21282169e-01 8.87815237e-01 -6.95445001e-01
-8.58434677e-01 -2.74020940e-01 3.47485155e-01 6.84106210e-03
2.20787585e-01 1.44066721e-01 -4.90339786e-01 1.19210236e-01
9.14311469e-01 7.58123159e-01 5.83556652e-01 1.05994034e+00
4.99873877e-01 -2.37713560e-01 -3.61352414e-01 3.46653223e-01
1.48958981e+00 6.12490058e-01 4.60520774e-01 1.62488952e-01
1.43367544e-01 9.10719156e-01 6.86093122e-02 6.39942229e-01
-1.98841378e-01 3.67559493e-02 1.27700314e-01 2.35488191e-01
6.57374203e-01 -5.41774273e-01 2.63316065e-01 8.91827643e-01
-2.01890111e-01 -2.21643865e-01 -9.46503997e-01 5.89149296e-01
-1.81493413e+00 -1.37858963e+00 -6.07842505e-01 2.33064222e+00
5.45287073e-01 3.21572006e-01 3.16114396e-01 -8.05092975e-02
9.82524395e-01 -2.62341201e-01 -6.55810237e-01 -2.83633113e-01
-2.40076274e-01 1.22858919e-01 6.78519189e-01 6.31906152e-01
-4.28495586e-01 4.67481643e-01 7.12805986e+00 3.77329588e-01
-8.87907267e-01 1.98834509e-01 5.41752338e-01 1.10308945e-01
-5.29245853e-01 1.60865724e-01 -9.73602116e-01 6.80813611e-01
1.04828596e+00 -7.82166541e-01 1.26072317e-01 2.23316193e-01
7.78052866e-01 -4.69623171e-02 -1.24307728e+00 9.98907506e-01
-6.07977390e-01 -1.12371838e+00 1.44687653e-01 8.86297822e-01
6.86740041e-01 -3.11325043e-01 5.78278303e-01 -3.39846104e-01
5.08824527e-01 -7.38264978e-01 7.93606162e-01 1.22685254e+00
1.13585420e-01 -5.93500733e-01 2.12722883e-01 7.27690339e-01
-8.72890711e-01 2.57785339e-02 -3.96828413e-01 -4.60600644e-01
7.62433410e-01 9.36724305e-01 -7.99073994e-01 -2.21330710e-02
1.33785337e-01 7.15233326e-01 1.82746470e-01 1.11393094e+00
-2.61159781e-02 9.20408487e-01 -4.02174562e-01 -5.09174280e-02
3.03512096e-01 -1.02963221e+00 9.85611856e-01 7.55105197e-01
8.22257638e-01 1.46755233e-01 -2.92634189e-01 1.60575771e+00
2.65924066e-01 -7.87609741e-02 -9.94869590e-01 -3.66146594e-01
2.11560667e-01 8.76355231e-01 -1.34192050e+00 -2.09968567e-01
-1.75557330e-01 8.84549618e-01 -5.04066825e-01 8.26862574e-01
-7.41909385e-01 5.35853207e-02 6.02590203e-01 4.28109884e-01
1.74606115e-01 -6.82895720e-01 -2.12732986e-01 -9.88081217e-01
-1.43688321e-01 6.80930540e-02 -1.17016084e-01 -7.15546489e-01
-1.85052156e+00 4.87218946e-01 3.52693468e-01 -1.20394552e+00
-2.80228049e-01 -3.34840149e-01 -8.91149938e-01 1.30321014e+00
-1.12070823e+00 -2.81500667e-01 4.25388664e-01 7.74977148e-01
6.16568089e-01 -8.90431106e-02 3.16042662e-01 -2.75860399e-01
-8.73093382e-02 -4.92211550e-01 4.38923299e-01 -2.80741721e-01
5.60610462e-03 -1.41812396e+00 1.05680537e+00 5.42536378e-01
1.42450526e-01 9.08643067e-01 9.53570962e-01 -1.12587464e+00
-1.26336324e+00 -5.71892142e-01 5.23764074e-01 -7.45127141e-01
1.33288562e+00 -4.23346639e-01 -1.01303291e+00 6.07756555e-01
1.30838916e-01 -2.39400074e-01 7.04599857e-01 -4.52342749e-01
4.91926312e-01 2.10473046e-01 -9.37734544e-01 7.60108531e-01
9.63536739e-01 -3.17317307e-01 -6.64382041e-01 2.15609297e-01
6.04041815e-01 3.41399491e-01 -6.46493018e-01 -3.80578429e-01
2.75424421e-01 -5.57607472e-01 5.97429395e-01 -4.28106934e-01
2.12626100e-01 -4.94304858e-02 4.89771254e-02 -1.43736959e+00
-3.88006419e-01 -1.14095592e+00 7.17844740e-02 1.17919743e+00
2.16695294e-01 -1.08397269e+00 3.68613780e-01 4.28674877e-01
4.41785634e-01 -4.56566185e-01 -9.19390738e-01 -1.05578947e+00
2.27287710e-01 -3.62451524e-01 4.95859146e-01 4.64865923e-01
-2.87653923e-01 2.08481327e-01 7.98718184e-02 2.43152995e-02
1.30505192e+00 -3.68425548e-01 2.05332011e-01 -1.62157476e+00
-3.22919935e-01 -4.73860443e-01 -4.05069768e-01 -1.08040226e+00
-9.17015523e-02 -6.87522769e-01 9.58311111e-02 -1.44223988e+00
-1.70038119e-01 -4.49391276e-01 -1.43131018e-01 -6.22264802e-01
3.37443531e-01 -3.87472183e-01 1.99818611e-01 9.50137079e-01
1.56319723e-01 7.34510005e-01 1.32175267e+00 3.84734035e-01
-4.23447818e-01 5.48856258e-01 -3.59234601e-01 9.13141489e-01
6.75122440e-01 -6.91579998e-01 -7.07885563e-01 -1.49367690e-01
5.90772629e-01 5.72301507e-01 4.78649884e-01 -6.84987009e-01
3.44864666e-01 -2.25504443e-01 6.24134578e-02 -5.22823691e-01
2.54230469e-01 -8.86729538e-01 4.11801368e-01 5.75876892e-01
-4.62443918e-01 5.69320321e-01 -2.61837751e-01 1.35360718e+00
-4.92140166e-02 -2.60126859e-01 3.38429153e-01 -1.95874855e-01
-8.65634717e-03 5.17856658e-01 -8.29333901e-01 9.49764531e-03
1.17881870e+00 -1.90911144e-01 -1.14919662e-01 -7.10022449e-01
-1.25593579e+00 -8.67514387e-02 3.21010143e-01 1.07273839e-01
6.59848452e-01 -1.24263394e+00 -4.42324519e-01 2.14358285e-01
-4.61339086e-01 -2.13463098e-01 -1.34374186e-01 1.10627747e+00
-3.66567284e-01 2.65649647e-01 -3.94302696e-01 -8.67337584e-01
-4.44570839e-01 6.97901428e-01 1.21257037e-01 1.62260249e-01
-7.71299183e-01 3.30816597e-01 6.53404355e-01 3.48869711e-01
-2.28986859e-01 -8.23075652e-01 5.90089783e-02 2.04693586e-01
5.50010860e-01 5.14211893e-01 -6.53718352e-01 -4.53898191e-01
1.08431295e-01 4.24293697e-01 1.99030146e-01 -1.15846813e+00
1.32006514e+00 -4.69818890e-01 -2.50142395e-01 1.71085227e+00
8.41118634e-01 -1.18962578e-01 -1.64889097e+00 -1.99107453e-01
3.36177379e-01 -2.18915373e-01 -4.38349813e-01 -1.08560614e-01
-5.32340527e-01 9.72638905e-01 4.77436900e-01 1.00853491e+00
5.66814899e-01 1.38718858e-01 5.96273661e-01 -1.33960769e-01
5.10018706e-01 -7.71177351e-01 9.64283012e-03 1.91632271e-01
8.11443269e-01 -4.59984601e-01 -6.01856887e-01 -1.94915548e-01
-3.05533290e-01 1.16247535e+00 -4.45529938e-01 -7.54985809e-01
1.20115948e+00 3.46670270e-01 -2.74888247e-01 -2.35618591e-01
-9.72132742e-01 1.40363667e-02 -1.53831840e-01 5.20932317e-01
2.51603007e-01 1.97854564e-01 -5.51063061e-01 3.57071310e-01
-2.93552205e-02 3.80611628e-01 1.04649937e+00 9.95722055e-01
-5.14995515e-01 -1.13271356e+00 -5.81537247e-01 2.41273925e-01
-5.98868072e-01 -1.27163172e-01 1.07427046e-01 5.06859899e-01
-2.52994627e-01 5.74563324e-01 4.22451347e-01 4.92820710e-01
1.73204884e-01 2.91005999e-01 4.40094501e-01 -6.56775415e-01
8.07863399e-02 6.55256987e-01 -7.41624892e-01 -5.46481870e-02
-3.55842471e-01 -1.21019506e+00 -1.09764767e+00 -3.04540247e-01
5.80674149e-02 2.94840097e-01 8.31500828e-01 9.77706075e-01
-1.24061136e-02 4.90332186e-01 3.83695155e-01 -5.37928522e-01
-7.66335547e-01 -7.34117925e-01 -1.26507115e+00 1.95497602e-01
5.48935592e-01 -4.12727863e-01 -5.97201705e-01 3.84774506e-01] | [6.7687859535217285, 3.805321455001831] |
a8b614fd-2cf1-4a7c-945c-f4a60423a889 | twise-at-semeval-2017-task-4-five-point | null | null | https://aclanthology.org/S17-2127 | https://aclanthology.org/S17-2127.pdf | TwiSe at SemEval-2017 Task 4: Five-point Twitter Sentiment Classification and Quantification | The paper describes the participation of the team {``}TwiSE{''} in the SemEval-2017 challenge. Specifically, I participated at Task 4 entitled {``}Sentiment Analysis in Twitter{''} for which I implemented systems for five-point tweet classification (Subtask C) and five-point tweet quantification (Subtask E) for English tweets. In the feature extraction steps the systems rely on the vector space model, morpho-syntactic analysis of the tweets and several sentiment lexicons. The classification step of Subtask C uses a Logistic Regression trained with the one-versus-rest approach. Another instance of Logistic Regression combined with the classify-and-count approach is trained for the quantification task of Subtask E. In the official leaderboard the system is ranked \textit{5/15} in Subtask C and \textit{2/12} in Subtask E. | ['Georgios Balikas'] | 2017-08-01 | null | null | null | semeval-2017-8 | ['twitter-sentiment-analysis'] | ['natural-language-processing'] | [ 7.35760629e-02 1.88981369e-02 1.70317236e-02 -7.46592045e-01
-1.09626877e+00 -7.20056832e-01 8.98522973e-01 8.33905458e-01
-8.48281145e-01 5.27011096e-01 1.16145089e-01 -3.80967468e-01
-1.26276925e-01 -6.04028761e-01 -1.95091173e-01 -3.83669376e-01
3.01454127e-01 5.90470016e-01 6.03686161e-02 -8.38572562e-01
5.08690953e-01 -5.85915223e-02 -1.57921708e+00 8.05800855e-01
1.52459636e-01 1.60033894e+00 -3.41747075e-01 1.05059803e+00
-4.89321768e-01 1.17899132e+00 -7.12213635e-01 -7.85613775e-01
1.19681992e-01 -2.25203812e-01 -1.02858627e+00 -3.73857975e-01
3.39914978e-01 2.99579173e-01 3.96384835e-01 8.54964972e-01
2.14531511e-01 1.49594888e-01 9.71126795e-01 -1.50747347e+00
-9.95147079e-02 8.59552085e-01 -5.10046899e-01 1.56756848e-01
4.21901405e-01 -2.53561497e-01 1.40256548e+00 -1.17651236e+00
3.95274013e-01 7.07090378e-01 1.05034626e+00 2.24330664e-01
-8.98371100e-01 -9.36051667e-01 8.62228218e-03 1.63135268e-02
-1.13283694e+00 -3.09717745e-01 3.03292572e-01 -1.10774624e+00
1.10985744e+00 7.25063384e-01 4.72013295e-01 9.70095873e-01
-7.43341669e-02 8.27001929e-01 1.64189303e+00 -2.51703501e-01
5.05673289e-02 8.12654257e-01 7.50636876e-01 3.66620690e-01
4.90254723e-02 -2.81516701e-01 -6.93984091e-01 -1.40540957e-01
-9.47996676e-02 -1.60599142e-01 5.16438365e-01 5.63396692e-01
-1.13003385e+00 1.26863980e+00 9.70030352e-02 5.10860443e-01
-2.82864004e-01 1.74467295e-01 9.69059825e-01 5.15158355e-01
1.08651793e+00 6.46673381e-01 -8.77537608e-01 -4.32986110e-01
-1.54781127e+00 5.23654699e-01 1.03831005e+00 7.52613962e-01
8.86585653e-01 -2.25541160e-01 -3.09114814e-01 8.75287414e-01
1.81362405e-01 3.22234511e-01 5.41530669e-01 -5.65666735e-01
7.66074300e-01 8.05913627e-01 1.68366164e-01 -8.92824173e-01
-9.84253585e-01 -2.33386740e-01 -5.62644422e-01 9.74139646e-02
5.07262051e-01 -5.10079563e-01 -4.40975696e-01 1.11671579e+00
2.37929940e-01 -2.68189013e-01 -2.50517219e-01 3.68764281e-01
1.36452556e+00 5.00377655e-01 2.70031005e-01 -1.97310522e-01
1.60878158e+00 -7.15602994e-01 -4.60715890e-01 -5.30803353e-02
1.03353894e+00 -9.92268205e-01 9.38014269e-01 6.02360904e-01
-1.06700730e+00 -4.02404785e-01 -8.81286919e-01 -9.67757925e-02
-1.07886350e+00 8.73382613e-02 3.19630861e-01 6.70307159e-01
-9.30783749e-01 4.56135690e-01 -4.84399050e-01 -1.90596849e-01
2.38613993e-01 4.49428618e-01 -3.89135003e-01 6.76841557e-01
-1.32306254e+00 9.02348876e-01 2.49133602e-01 2.10861713e-02
-1.83763638e-01 -7.30699301e-01 -9.38055098e-01 -3.53655040e-01
-4.04522121e-02 -1.79269046e-01 1.46858060e+00 -9.46761191e-01
-1.37473619e+00 1.62283552e+00 -2.37015504e-02 -2.85464138e-01
7.76500225e-01 -1.08393207e-01 -2.51571625e-01 -4.71159756e-01
5.46328604e-01 1.28355935e-01 7.31423914e-01 -7.80350745e-01
-1.16530430e+00 -3.14008981e-01 1.32833913e-01 -7.31164515e-02
-2.43360668e-01 7.84726739e-01 4.22635600e-02 -4.51709062e-01
-1.60769656e-01 -9.40223753e-01 -1.22572526e-01 -8.87675047e-01
-6.79862440e-01 -5.76124012e-01 3.69866967e-01 -6.09191835e-01
1.62384093e+00 -1.95266771e+00 2.20211241e-02 2.90039629e-01
4.75972116e-01 -3.17051142e-01 3.90719742e-01 6.63173914e-01
-2.66727030e-01 2.98038870e-01 -3.00909504e-02 -9.57899392e-01
4.83142942e-01 -2.47149959e-01 -2.03266427e-01 6.00772381e-01
-7.26381615e-02 8.16306651e-01 -8.68653357e-01 -5.15234172e-01
9.06811729e-02 5.99850118e-02 -3.68222266e-01 -9.35788266e-03
-2.56069116e-02 1.53135076e-01 -5.08294880e-01 3.34788233e-01
4.53270197e-01 -7.66011700e-02 -2.17174992e-01 -1.10532448e-01
-9.23044026e-01 3.57600987e-01 -9.01444197e-01 1.03681910e+00
-5.97480118e-01 8.25310826e-01 -1.21118296e-02 -8.04597676e-01
8.79837573e-01 2.07057729e-01 6.75356805e-01 -5.43134868e-01
7.20694482e-01 3.37707520e-01 -4.82259035e-01 -4.82426673e-01
9.07662928e-01 -3.17805618e-01 -9.87167895e-01 5.27989328e-01
5.65417521e-02 -7.35718787e-01 4.47780430e-01 1.49597734e-01
9.39493537e-01 2.81746179e-01 3.51897627e-01 -4.94649768e-01
7.99307048e-01 1.68087900e-01 -4.23474945e-02 7.00753570e-01
-2.01917112e-01 5.82116783e-01 8.07603121e-01 -6.79122210e-01
-1.09663880e+00 -2.18623340e-01 -1.42553747e-01 2.14452648e+00
-4.88429755e-01 -7.84522593e-01 -7.08705544e-01 -5.38603604e-01
-1.78253651e-01 8.83345544e-01 -1.12336552e+00 3.98680776e-01
-4.53916222e-01 -1.04609454e+00 6.07576966e-01 1.56399339e-01
1.26498774e-01 -9.22636271e-01 -7.55451322e-01 6.30857646e-02
-5.47302663e-01 -1.03568864e+00 -1.42624170e-01 7.79579222e-01
-9.54777896e-02 -8.44038725e-01 -2.84348935e-01 -6.72666192e-01
2.12146506e-01 -3.55868280e-01 1.06132543e+00 -1.92594498e-01
-2.57640015e-02 -1.09027915e-01 -7.99324393e-01 -1.28157365e+00
-1.79191902e-01 5.22861481e-01 -1.06361821e-01 9.46769714e-02
6.89863384e-01 -6.70515075e-02 -2.49733448e-01 2.39056990e-01
-7.60747135e-01 -5.53200543e-02 4.65138303e-03 5.80708504e-01
4.33472723e-01 -3.41477513e-01 2.60625243e-01 -1.41470385e+00
9.54941809e-01 -6.20041788e-01 -4.26567554e-01 -3.44402164e-01
-3.36423993e-01 -4.61958647e-01 7.22855985e-01 3.89584862e-02
-4.36026931e-01 -1.64326116e-01 -6.78948820e-01 3.05866450e-01
1.15739509e-01 5.92240036e-01 3.50992382e-01 3.52736145e-01
1.01597154e+00 -2.54214797e-02 -5.38771093e-01 -1.86386004e-01
1.19500086e-01 1.00254118e+00 2.18440667e-01 -2.81134784e-01
7.59765923e-01 1.82974964e-01 -8.82791653e-02 -7.27284372e-01
-1.52034354e+00 -9.02518570e-01 -7.09882319e-01 -4.33674961e-01
1.16304600e+00 -1.09303808e+00 -1.02519989e+00 5.54098725e-01
-9.65371072e-01 -4.69067931e-01 -5.33430755e-01 2.64445722e-01
-6.18050694e-01 -3.71754438e-01 -5.39578199e-01 -9.87230897e-01
-4.91539478e-01 -9.67409313e-01 1.51048779e+00 -7.13109002e-02
-9.38905954e-01 -9.68261480e-01 2.52965152e-01 5.65748692e-01
5.38927257e-01 4.70496863e-01 5.84296525e-01 -1.40456057e+00
4.49039042e-01 -6.87412620e-01 -2.53825158e-01 2.92987317e-01
-3.36890787e-01 1.48322061e-01 -1.28710437e+00 8.17318782e-02
-6.81084255e-03 -6.75394535e-01 8.57112408e-01 2.29626596e-01
1.09419084e+00 -3.16379994e-01 3.92294396e-03 2.11316198e-01
1.16879308e+00 -4.61234540e-01 3.65017086e-01 8.51028502e-01
4.17507499e-01 9.24490869e-01 6.62311733e-01 6.90570474e-01
8.56324375e-01 5.85179031e-01 2.37495705e-01 9.60275829e-02
3.25548530e-01 3.24139483e-02 5.24616718e-01 1.09856558e+00
-2.00483575e-01 -1.86493378e-02 -9.05442894e-01 5.36163986e-01
-1.76813030e+00 -9.33131039e-01 -8.78484130e-01 1.75789368e+00
8.14992070e-01 4.22640860e-01 4.60782915e-01 5.28869152e-01
2.81358391e-01 3.55152071e-01 3.40097398e-01 -1.11125863e+00
-8.86888728e-02 2.19375923e-01 6.25401080e-01 5.97662151e-01
-1.43351007e+00 9.63998914e-01 5.33594990e+00 9.99707103e-01
-1.03764093e+00 4.41195101e-01 5.66300094e-01 -4.36250791e-02
-5.96161420e-03 -4.27592695e-02 -1.27957904e+00 5.62158942e-01
1.42627716e+00 2.62985319e-01 -2.57249307e-02 7.67106473e-01
1.64875805e-01 -3.99386585e-01 -9.37194824e-01 8.61880183e-01
2.02754840e-01 -1.18483198e+00 -4.52938855e-01 -1.44257650e-01
6.46811903e-01 4.02244717e-01 1.07559845e-01 8.87543917e-01
2.72105426e-01 -1.00998306e+00 1.25885820e+00 6.85537100e-01
1.10267067e+00 -6.04874551e-01 1.12679291e+00 5.66273570e-01
-1.17645490e+00 -6.99511245e-02 2.80251559e-02 -2.94625103e-01
3.80504839e-02 6.74342453e-01 -6.75592124e-01 3.31937790e-01
9.57260787e-01 6.05164826e-01 -5.11934102e-01 4.93817300e-01
-2.36767158e-01 7.64724672e-01 -2.82187909e-01 -6.74254894e-01
4.40455884e-01 -2.05988333e-01 5.08839428e-01 2.04521966e+00
-4.43416312e-02 -1.43421471e-01 4.92688529e-02 2.24871129e-01
-2.89946318e-01 5.73457241e-01 -9.17374194e-02 9.53967720e-02
3.50503698e-02 1.65874445e+00 -8.99684668e-01 -4.36910331e-01
-3.14426988e-01 6.30026937e-01 3.57657790e-01 -1.55648783e-01
-8.32173288e-01 -7.39100575e-01 1.10849269e-01 3.58136415e-01
9.70896780e-02 -1.85374394e-02 -6.63496792e-01 -8.09490442e-01
-2.11674377e-01 -5.96449018e-01 3.05096745e-01 -6.15559697e-01
-1.49882817e+00 8.48959088e-01 1.97398156e-01 -8.75262916e-01
-2.59763181e-01 -8.36962581e-01 -5.62373638e-01 7.22395658e-01
-1.37273312e+00 -1.30011749e+00 -2.85973787e-01 6.37114763e-01
5.98077834e-01 -1.13095485e-01 8.68470252e-01 4.28442836e-01
-3.83343369e-01 6.12769902e-01 -7.01952130e-02 1.58893839e-01
6.89073026e-01 -1.67134356e+00 2.89335400e-01 9.29746311e-03
-1.19374491e-01 3.62759858e-01 9.99046743e-01 -3.28828037e-01
-7.25540817e-01 -1.33414173e+00 1.92801106e+00 -9.70910490e-01
1.39944041e+00 -6.29704177e-01 -2.43911818e-01 7.34742820e-01
6.96457103e-02 -5.00700712e-01 1.19435346e+00 1.70693576e-01
-1.74033314e-01 9.08168703e-02 -9.97184873e-01 1.42348140e-01
3.11344892e-01 -7.41176724e-01 -5.72647095e-01 7.30229974e-01
4.50296611e-01 -5.19248307e-01 -1.02115035e+00 1.65655017e-02
7.93907285e-01 -9.52867448e-01 4.64124352e-01 -6.80455089e-01
8.34115863e-01 -9.80252996e-02 -3.38340521e-01 -1.03072059e+00
6.85829390e-03 -6.46623075e-01 4.58321989e-01 1.27580011e+00
1.00329578e+00 -4.69020993e-01 5.47592402e-01 3.53586644e-01
-4.07871343e-02 -6.11688375e-01 -9.24308717e-01 -2.51675576e-01
4.94859904e-01 -1.14369678e+00 3.00353527e-01 9.53502178e-01
2.56239206e-01 6.41513765e-01 -3.46024752e-01 -6.29799962e-01
8.26091841e-02 -1.87089965e-01 8.65233839e-01 -1.41483843e+00
-6.55732397e-03 -7.34933078e-01 -2.43913934e-01 -4.75730926e-01
8.68599564e-02 -1.19519389e+00 -8.04452524e-02 -1.49071229e+00
1.93027958e-01 -2.67937452e-01 -9.79116186e-02 4.27838296e-01
5.74256256e-02 7.39627361e-01 3.45065206e-01 1.49119765e-01
-8.60944748e-01 -6.67906255e-02 4.99699861e-01 1.60663412e-03
-6.45450652e-02 3.50627273e-01 -9.18102026e-01 9.31417882e-01
6.75509334e-01 -8.09706151e-01 4.11501050e-01 -1.17406666e-01
1.38743889e+00 -4.27245885e-01 2.21522987e-01 -5.37776351e-01
3.59441459e-01 7.07472637e-02 2.02475563e-01 -7.20141292e-01
3.87218535e-01 -6.23779237e-01 -1.44507781e-01 -1.54317860e-02
-3.86073887e-01 2.91922331e-01 1.38873234e-01 4.17161696e-02
-2.95031488e-01 -3.80737782e-01 7.43439376e-01 -2.15403691e-01
-1.21296808e-01 7.34973326e-02 -6.88748419e-01 3.82203907e-01
9.64757442e-01 -2.20497057e-01 -1.52368844e-01 -2.12483928e-01
-9.32680547e-01 2.82764852e-01 -1.96240991e-01 2.02674031e-01
1.22024613e-02 -9.45362508e-01 -1.17314410e+00 -1.08975433e-01
4.21232820e-01 -1.81593969e-01 -7.47341139e-04 1.41604149e+00
-6.31972253e-01 3.71411979e-01 1.67752728e-01 -3.30513149e-01
-1.12394917e+00 -6.59260452e-02 2.54925191e-01 -8.63285005e-01
2.57643729e-01 1.37720776e+00 -3.13825697e-01 -7.53143132e-01
-1.38757810e-01 -3.81852597e-01 -8.98583770e-01 1.06112802e+00
5.95084965e-01 6.14906013e-01 3.36070627e-01 -1.17531168e+00
-3.38088900e-01 7.58646548e-01 -8.17045756e-03 -3.02260548e-01
1.60959458e+00 -6.45106137e-02 -4.32957917e-01 1.04720140e+00
1.60056162e+00 3.35298508e-01 -4.25104976e-01 1.34261683e-01
3.34799349e-01 1.36898071e-01 -6.28957823e-02 -6.76699638e-01
-3.78222406e-01 8.26024234e-01 8.33096579e-02 7.57680178e-01
5.92074633e-01 -4.79577333e-02 7.01403379e-01 1.54281974e-01
2.10334212e-01 -1.62117672e+00 -2.85447568e-01 1.18100321e+00
9.79698598e-01 -1.49119389e+00 2.69628018e-01 3.15204100e-03
-9.53828990e-01 1.18426180e+00 2.78594228e-03 -2.18711928e-01
1.00584221e+00 2.37849131e-01 3.40411961e-02 -5.62381089e-01
-5.10403693e-01 -2.27149367e-01 4.93893683e-01 6.20069243e-02
7.04802811e-01 3.80086809e-01 -7.03059018e-01 1.08051372e+00
-1.13819265e+00 -2.15018481e-01 4.06075925e-01 7.18312383e-01
-5.14501333e-01 -7.59561658e-01 -1.48165584e-01 7.14550853e-01
-9.51363146e-01 -3.10337901e-01 -8.17136586e-01 9.28355038e-01
5.17809391e-01 1.36720705e+00 -9.29642245e-02 -6.96652770e-01
6.37642443e-01 2.32962117e-01 -3.47372182e-02 -6.92920089e-01
-1.76963985e+00 1.13964258e-02 5.54631710e-01 -5.73386788e-01
-4.95503783e-01 -9.76908922e-01 -1.13271463e+00 -3.97021711e-01
-3.27954024e-01 3.99680018e-01 1.16579700e+00 1.11039555e+00
-4.86328155e-01 4.43911642e-01 7.02364624e-01 -8.40229630e-01
-3.98604751e-01 -1.15072787e+00 -7.51082361e-01 3.99030477e-01
4.31472391e-01 -3.42979312e-01 -6.15116835e-01 3.84439565e-02] | [11.166139602661133, 6.943508148193359] |
4df8b434-55a6-4001-9362-e85775a49d95 | nonparametric-probabilistic-regression-with | 2210.16247 | null | https://arxiv.org/abs/2210.16247v1 | https://arxiv.org/pdf/2210.16247v1.pdf | Nonparametric Probabilistic Regression with Coarse Learners | Probabilistic Regression refers to predicting a full probability density function for the target conditional on the features. We present a nonparametric approach to this problem which combines base classifiers (typically gradient boosted forests) trained on different coarsenings of the target value. By combining such classifiers and averaging the resulting densities, we are able to compute precise conditional densities with minimal assumptions on the shape or form of the density. We combine this approach with a structured cross-entropy loss function which serves to regularize and smooth the resulting densities. Prediction intervals computed from these densities are shown to have high fidelity in practice. Furthermore, examining the properties of these densities on particular observations can provide valuable insight. We demonstrate this approach on a variety of datasets and show competitive performance, particularly on larger datasets. | ['Brian Lucena'] | 2022-10-28 | null | null | null | null | ['prediction-intervals'] | ['miscellaneous'] | [ 1.37773842e-01 1.98022529e-01 -4.99654770e-01 -5.14476657e-01
-1.19158840e+00 -4.10676152e-01 9.15830493e-01 2.49961048e-01
-2.51248926e-01 1.17025137e+00 1.19521491e-01 -2.94598013e-01
-7.24223554e-02 -8.65477324e-01 -8.17150891e-01 -8.35283160e-01
-3.47282380e-01 6.46400154e-01 3.11164349e-01 2.31660187e-01
2.45844439e-01 4.42569792e-01 -1.27568841e+00 3.64716761e-02
8.55800331e-01 1.20202672e+00 -4.62323911e-02 7.96921134e-01
-1.49081033e-02 6.39347136e-01 -4.44840431e-01 -4.73031908e-01
-7.47557580e-02 -1.42522410e-01 -7.20772564e-01 -4.35037799e-02
2.57613748e-01 -2.28303432e-01 1.16662323e-01 9.42293882e-01
-1.77819222e-01 2.22564146e-01 1.55485880e+00 -9.92896855e-01
-4.05045509e-01 2.29355961e-01 -6.15632713e-01 2.97674179e-01
2.90739447e-01 -2.85667062e-01 1.08268547e+00 -7.86706448e-01
1.81104809e-01 1.11321509e+00 1.13947594e+00 1.11058526e-01
-1.78729451e+00 -3.99398208e-01 2.45162798e-03 -7.67405480e-02
-1.63123286e+00 -4.28085119e-01 2.94383109e-01 -5.20952046e-01
7.70128310e-01 -5.26674800e-02 3.46726626e-01 8.62917542e-01
4.41301793e-01 5.23100853e-01 1.23899639e+00 -6.01260364e-01
3.33776832e-01 4.54173595e-01 8.91852155e-02 6.34825945e-01
2.27745831e-01 2.34562516e-01 -3.66334051e-01 -4.56694543e-01
5.76357841e-01 8.57415888e-03 -2.36707494e-01 -5.02599359e-01
-6.38277352e-01 1.09924150e+00 4.28121865e-01 -9.92396325e-02
-4.65475976e-01 1.03207171e-01 1.00074276e-01 -2.19045416e-01
8.21303666e-01 -1.19216383e-01 -3.11296672e-01 -1.07779957e-01
-1.14199185e+00 3.70978057e-01 1.08336043e+00 7.54573405e-01
1.00452924e+00 -3.58643919e-01 -2.07079992e-01 7.66024053e-01
1.87286571e-01 3.12232882e-01 -1.57270916e-02 -7.93349385e-01
1.92508891e-01 1.76894944e-02 5.02332866e-01 -4.30498838e-01
-2.97924653e-02 -3.68459411e-02 -6.92873180e-01 2.59348273e-01
7.93943107e-01 -2.53387451e-01 -1.30443943e+00 1.71478212e+00
2.53167540e-01 2.08264932e-01 -1.82888210e-01 3.56572688e-01
-1.48879319e-01 8.14720154e-01 4.25116390e-01 -1.57949045e-01
9.22090352e-01 -3.38892192e-01 -2.66408890e-01 -3.98088759e-03
3.25923830e-01 -3.65584910e-01 8.87762845e-01 4.21323687e-01
-8.16843808e-01 -1.84082225e-01 -9.17050183e-01 1.86760902e-01
-2.63128489e-01 -1.18775293e-01 5.98748565e-01 8.19673359e-01
-1.03541124e+00 1.00820720e+00 -1.13107502e+00 -1.66895717e-01
5.47639489e-01 4.16317433e-01 -1.76571190e-01 -2.11911753e-01
-9.00670886e-01 1.18097210e+00 2.92166859e-01 -1.17042765e-01
-6.27474964e-01 -8.07139456e-01 -7.81720936e-01 1.83430225e-01
-7.66569376e-03 -4.35101956e-01 1.40507472e+00 -6.94140851e-01
-1.24811256e+00 4.87027764e-01 -4.37595159e-01 -7.56943226e-01
3.94667566e-01 -2.03263476e-01 -1.20722823e-01 -2.40184758e-02
1.07890002e-01 4.31828797e-01 8.21269870e-01 -1.07274890e+00
-6.82741523e-01 -1.79023221e-01 -2.78816789e-01 -1.66422967e-02
2.40035970e-02 -2.56416231e-01 -1.03406146e-01 -4.23347980e-01
-1.62429914e-01 -7.22283542e-01 -3.02873313e-01 -1.06879383e-01
-4.61910188e-01 -2.52938896e-01 4.13876891e-01 -6.73827708e-01
9.54436779e-01 -1.89915752e+00 -1.31366551e-01 5.61911821e-01
-8.09402317e-02 -3.08991730e-01 4.51222062e-01 3.63390595e-01
1.67207360e-01 4.40274216e-02 -6.02425516e-01 -2.10893512e-01
-1.02669643e-02 1.76019713e-01 -4.80456948e-01 6.47933483e-01
5.49985349e-01 4.79063928e-01 -5.54706573e-01 -4.81434882e-01
1.41530916e-01 6.89280987e-01 -5.09719372e-01 1.91199824e-01
-2.03502819e-01 1.70635998e-01 -2.76331663e-01 3.15529734e-01
4.75712389e-01 -2.61299282e-01 1.75724313e-01 2.28639305e-01
1.70827404e-01 5.41973591e-01 -8.96895349e-01 9.66782928e-01
-6.65562451e-01 3.07895213e-01 9.46597755e-02 -1.20896566e+00
1.02577674e+00 5.89022823e-02 2.54684538e-01 4.61027324e-02
-7.10668862e-02 7.94700608e-02 -3.59977156e-01 1.96237370e-01
2.24451572e-01 -9.29453611e-01 -1.49929494e-01 2.25292057e-01
2.92235523e-01 -1.66079432e-01 -3.69297154e-02 -2.97624320e-02
9.16714132e-01 2.70678937e-01 5.96374571e-01 -4.35841501e-01
1.06966794e-01 -3.79497297e-02 2.05589876e-01 7.98752308e-01
-6.38945475e-02 5.64951241e-01 7.60003328e-01 -1.58971727e-01
-1.09065759e+00 -1.63749194e+00 -6.08000517e-01 9.33505416e-01
-1.29449174e-01 -1.85617432e-01 -4.80606079e-01 -7.88641393e-01
2.92214215e-01 9.91524637e-01 -8.96845341e-01 -5.39778844e-02
-5.67937009e-02 -9.69656885e-01 1.72770008e-01 8.56735647e-01
2.34512724e-02 -7.27454007e-01 -1.41501665e-01 9.70432237e-02
1.48271263e-01 -8.08288097e-01 -2.37911850e-01 7.75335908e-01
-9.64319944e-01 -8.82487059e-01 -7.56810486e-01 -5.17075181e-01
3.84289771e-01 -3.40333700e-01 1.22439301e+00 -3.28260154e-01
1.44621162e-02 2.09609836e-01 -3.62417102e-02 -3.09495449e-01
-4.54993188e-01 3.98166552e-02 -1.00500003e-01 -1.96096674e-01
3.45463872e-01 -7.68256187e-01 -1.21957228e-01 4.54831794e-02
-4.82698739e-01 -3.18058729e-01 5.88048756e-01 8.80966187e-01
5.81292450e-01 9.59175602e-02 5.20650327e-01 -1.10448658e+00
4.82215881e-01 -8.80639613e-01 -7.09421098e-01 2.62589693e-01
-6.30534828e-01 4.53840256e-01 6.12380385e-01 -3.49969685e-01
-1.13451910e+00 1.72305629e-01 -1.91263594e-02 -4.14543897e-01
-1.97780713e-01 4.78061438e-01 2.36346796e-01 2.22389624e-01
7.39819884e-01 8.36174712e-02 4.13010363e-03 -5.53322256e-01
4.03767377e-01 5.78229010e-01 8.20587814e-01 -7.83211827e-01
5.99721432e-01 3.28789800e-01 2.43236437e-01 -6.19546592e-01
-1.13518941e+00 -3.76155794e-01 -4.85919505e-01 9.06355307e-02
6.13702595e-01 -7.46126056e-01 -4.89279658e-01 -1.69802681e-02
-6.79165840e-01 -4.43591058e-01 -3.60259324e-01 6.04530990e-01
-7.95761287e-01 3.05526722e-02 -5.03814220e-01 -1.36969090e+00
-6.80392757e-02 -6.10405266e-01 1.03955352e+00 1.58228517e-01
-2.84333855e-01 -1.30074167e+00 3.41154397e-01 -2.74709046e-01
2.53716141e-01 1.77618802e-01 1.14510477e+00 -6.98337674e-01
-2.21211314e-01 -4.92690235e-01 -4.11781669e-01 4.63104248e-01
3.59459408e-02 2.28718907e-01 -9.12996113e-01 4.98552173e-02
-2.57006288e-01 -2.87488133e-01 1.32321227e+00 8.22431087e-01
1.19338834e+00 -2.98282027e-01 -5.09918213e-01 3.69025081e-01
1.39813340e+00 -2.58037865e-01 5.23026884e-01 -1.42426081e-02
3.84207577e-01 6.11582994e-01 5.23166060e-01 4.44355965e-01
3.39718074e-01 5.88858724e-01 2.76149567e-02 2.95567721e-01
1.85848191e-01 -5.37664890e-01 4.04230416e-01 1.64507717e-01
-5.61414398e-02 1.31402791e-01 -8.37947547e-01 6.84070349e-01
-1.53600419e+00 -1.04444981e+00 1.68646276e-01 2.46937346e+00
1.00794089e+00 4.31912720e-01 5.43670237e-01 -2.21873038e-02
1.01219141e+00 -1.15510516e-01 -5.29351294e-01 -5.37092268e-01
3.55088741e-01 7.30210006e-01 7.52147555e-01 7.60814488e-01
-1.24074888e+00 5.40584624e-01 8.12886429e+00 9.46269810e-01
-8.26585054e-01 -7.63333738e-02 8.11193943e-01 1.16486877e-01
-3.10510576e-01 2.02183291e-01 -9.11033809e-01 5.56369364e-01
1.29676318e+00 -2.83271313e-01 4.31104094e-01 1.13141954e+00
-2.21624017e-01 -6.75434768e-01 -1.05508864e+00 3.85039181e-01
-5.11811137e-01 -1.23422527e+00 -2.60995477e-01 1.93905786e-01
5.77328920e-01 1.23437885e-02 6.76779598e-02 5.60640037e-01
8.96576703e-01 -1.27068245e+00 6.56309068e-01 7.73461580e-01
9.71866190e-01 -1.06372094e+00 6.69303179e-01 4.22679245e-01
-8.93356144e-01 1.00676663e-01 -4.30584162e-01 -2.12036356e-01
-6.11900818e-03 8.95455718e-01 -1.23981714e+00 2.31837779e-01
6.73035324e-01 5.26839077e-01 -3.33929628e-01 1.11994958e+00
-1.66881606e-01 9.51346040e-01 -8.60234618e-01 -1.10971376e-01
-3.70633304e-02 -2.55175292e-01 1.23470232e-01 1.43886340e+00
4.21665549e-01 -2.87061036e-01 4.98708040e-02 9.23953712e-01
7.46213272e-02 -9.20257419e-02 -5.99531591e-01 9.43383276e-02
4.26900864e-01 1.14857447e+00 -5.91066658e-01 -3.39614302e-01
-5.07836819e-01 4.60459501e-01 8.20940495e-01 3.60891968e-01
-7.14982569e-01 -1.70129955e-01 5.72421849e-01 2.84940749e-01
8.27734649e-01 -9.82676446e-02 -4.48456824e-01 -9.28001583e-01
-1.02600314e-01 -1.63568854e-01 5.59898198e-01 -5.47799170e-01
-1.69949687e+00 2.74433047e-01 5.10254502e-01 -7.20378578e-01
-7.01188743e-01 -5.93643963e-01 -8.11802506e-01 1.21054029e+00
-1.29201591e+00 -7.35124946e-01 1.36038437e-01 3.74154538e-01
-2.74020713e-02 2.51755029e-01 9.10453081e-01 -2.69927770e-01
-3.36691529e-01 3.44211727e-01 2.68280059e-01 -1.03816278e-01
5.01994610e-01 -1.68409431e+00 2.06253678e-01 4.98808622e-01
1.99844316e-01 4.91904497e-01 8.19863915e-01 -5.95904768e-01
-6.63870633e-01 -1.03849423e+00 7.34837353e-01 -4.88808274e-01
7.53026187e-01 -2.25856826e-01 -1.08347428e+00 7.83765972e-01
-1.97015375e-01 1.50571316e-01 6.89326823e-01 7.80117452e-01
-4.65352565e-01 8.73065740e-02 -1.36029518e+00 1.02113329e-01
4.48531866e-01 -4.47729558e-01 -4.39614862e-01 1.95494130e-01
1.05221748e-01 -1.22684568e-01 -9.06678140e-01 4.02791500e-01
7.83781052e-01 -1.13264191e+00 7.47712612e-01 -6.13755286e-01
4.71058398e-01 -5.74654341e-02 -4.07373875e-01 -1.33872080e+00
-2.39774674e-01 -3.86344582e-01 -1.54738590e-01 1.16562605e+00
6.17210686e-01 -6.23399615e-01 9.31331515e-01 7.99438298e-01
1.93987444e-01 -1.03849494e+00 -9.88777339e-01 -8.27487767e-01
3.89844447e-01 -5.36293685e-01 2.55075604e-01 3.17846060e-01
8.43882933e-02 2.03694075e-01 -2.41102591e-01 8.49357545e-02
7.40663886e-01 8.48696753e-02 5.24748445e-01 -1.27658403e+00
-3.21979910e-01 -2.89025366e-01 -5.18260062e-01 -9.21271503e-01
2.60303080e-01 -7.74627090e-01 3.40175420e-01 -1.18716395e+00
4.07202482e-01 -6.24904573e-01 -3.06433469e-01 3.75442684e-01
-4.25714791e-01 2.70133823e-01 -2.35195890e-01 1.73770323e-01
-2.96380907e-01 6.04923069e-01 6.53733075e-01 3.35745901e-01
-1.28837004e-01 4.36542600e-01 -7.65639067e-01 9.05729413e-01
7.52542078e-01 -5.44144392e-01 -1.47444919e-01 4.17290568e-01
-3.89385313e-01 2.67220616e-01 5.41477323e-01 -9.55297649e-01
-1.02886423e-01 -3.35949510e-01 8.27895701e-01 -5.33912778e-01
4.59445208e-01 -4.74125803e-01 2.27977559e-02 3.95988151e-02
-4.79199857e-01 -3.21109205e-01 6.22026846e-02 9.50075030e-01
-1.16092153e-01 -3.41471672e-01 1.11343563e+00 2.53213584e-01
-2.43896052e-01 1.81082785e-01 -2.00642690e-01 1.53441474e-01
9.17846501e-01 -2.15106700e-02 1.21781826e-01 -7.97651052e-01
-8.51587176e-01 -9.36070457e-02 4.90924180e-01 -3.57406437e-01
1.59825638e-01 -1.26147807e+00 -5.77329338e-01 5.97390644e-02
-1.38093486e-01 -4.21213694e-02 -2.58952349e-01 8.70012581e-01
-1.57237574e-01 2.73786485e-01 3.83793116e-02 -5.58478415e-01
-6.70181274e-01 5.16311944e-01 3.58335406e-01 -4.93771791e-01
-3.93293232e-01 8.44890594e-01 2.31623396e-01 -1.58747897e-01
6.51097149e-02 -3.14322025e-01 1.49610072e-01 -8.45516324e-02
4.14628506e-01 2.33504400e-01 -1.31906435e-01 -4.09034491e-01
-4.72758204e-01 2.53555954e-01 -1.16510339e-01 -4.89470124e-01
1.24014878e+00 2.82083433e-02 3.23578238e-01 6.44659042e-01
1.20121539e+00 1.84768382e-02 -1.85271335e+00 -1.53992683e-01
1.59320563e-01 -3.91051352e-01 7.00514093e-02 -7.18793452e-01
-6.03614032e-01 7.32963920e-01 9.05222148e-02 5.07790089e-01
1.03027451e+00 1.51929632e-01 3.33089679e-01 6.63729161e-02
2.99888194e-01 -6.46985292e-01 -5.25197268e-01 2.52537787e-01
5.36560118e-01 -1.12521827e+00 3.21306288e-01 -4.65862066e-01
-6.43098176e-01 1.16695583e+00 3.51407528e-02 -6.06528461e-01
9.55323696e-01 4.75261778e-01 -3.78081828e-01 2.38450199e-01
-8.82821918e-01 -3.23782533e-01 4.76254612e-01 8.95750165e-01
4.72069174e-01 2.53002107e-01 5.14960401e-02 4.75960523e-01
-3.90361637e-01 -3.38854231e-02 2.17865825e-01 6.40121222e-01
-5.94811261e-01 -9.28583860e-01 -3.21938068e-01 1.01414180e+00
-6.88335657e-01 -1.38742253e-01 -1.01192214e-01 8.57934833e-01
-3.36357772e-01 7.42774785e-01 2.58068115e-01 -1.85024858e-01
2.87623908e-02 3.86333108e-01 7.47469366e-01 -6.88175857e-01
1.09241538e-01 1.03715204e-01 2.67894864e-01 -3.04664522e-01
-2.85122216e-01 -9.76187229e-01 -9.53837454e-01 -4.65631396e-01
-4.74492282e-01 2.52391607e-01 6.26297772e-01 1.03091133e+00
-4.84907143e-02 -6.87657818e-02 6.37205064e-01 -1.09559822e+00
-1.09580755e+00 -1.06185818e+00 -1.01537383e+00 2.18147978e-01
4.03928280e-01 -9.87113357e-01 -5.27095973e-01 1.08056456e-01] | [7.342067718505859, 4.1162028312683105] |
107119e9-0138-4f5d-9731-8c571ae8b30b | curvature-based-feature-selection-with | 2101.03581 | null | https://arxiv.org/abs/2101.03581v3 | https://arxiv.org/pdf/2101.03581v3.pdf | Curvature-based Feature Selection with Application in Classifying Electronic Health Records | Disruptive technologies provides unparalleled opportunities to contribute to the identifications of many aspects in pervasive healthcare, from the adoption of the Internet of Things through to Machine Learning (ML) techniques. As a powerful tool, ML has been widely applied in patient-centric healthcare solutions. To further improve the quality of patient care, Electronic Health Records (EHRs) are commonly adopted in healthcare facilities for analysis. It is a crucial task to apply AI and ML to analyse those EHRs for prediction and diagnostics due to their highly unstructured, unbalanced, incomplete, and high-dimensional nature. Dimensionality reduction is a common data preprocessing technique to cope with high-dimensional EHR data, which aims to reduce the number of features of EHR representation while improving the performance of the subsequent data analysis, e.g. classification. In this work, an efficient filter-based feature selection method, namely Curvature-based Feature Selection (CFS), is presented. The proposed CFS applied the concept of Menger Curvature to rank the weights of all features in the given data set. The performance of the proposed CFS has been evaluated in four well-known EHR data sets, including Cervical Cancer Risk Factors (CCRFDS), Breast Cancer Coimbra (BCCDS), Breast Tissue (BTDS), and Diabetic Retinopathy Debrecen (DRDDS). The experimental results show that the proposed CFS achieved state-of-the-art performance on the above data sets against conventional PCA and other most recent approaches. The source code of the proposed approach is publicly available at https://github.com/zhemingzuo/CFS. | ['Noura Al Moubayed', 'Han Xu', 'Jie Li', 'Zheming Zuo'] | 2021-01-10 | null | null | null | null | ['breast-cancer-detection', 'cervical-cancer-biopsy-identification', 'breast-cancer-detection', 'diabetic-retinopathy-detection', 'breast-tissue-identification'] | ['knowledge-base', 'medical', 'medical', 'medical', 'medical'] | [ 7.64529780e-02 -2.34123856e-01 8.57741460e-02 -9.23656300e-02
-5.90007901e-01 -1.05235822e-01 3.99857730e-01 6.33508980e-01
-1.87472209e-01 5.36684573e-01 3.54054242e-01 -3.68710250e-01
-6.96583450e-01 -6.53799713e-01 8.82306024e-02 -9.26558912e-01
-6.37056306e-02 3.75249416e-01 -2.81851739e-01 5.70673831e-02
3.04093093e-01 6.31650269e-01 -1.47153282e+00 2.37844497e-01
8.59822512e-01 9.38587666e-01 3.13469172e-01 7.12474525e-01
-5.72294258e-02 5.93105912e-01 2.64498834e-02 -1.14154071e-01
2.00009629e-01 -1.95695266e-01 -5.36591947e-01 -7.42350221e-02
-4.98406857e-01 1.10743694e-01 -5.93461795e-03 8.61534595e-01
6.70713305e-01 -2.04938009e-01 6.77771270e-01 -8.69693637e-01
-2.01426581e-01 -3.76589820e-02 -7.07045376e-01 1.45508587e-01
4.54053938e-01 -1.26788110e-01 4.99933600e-01 -9.91360247e-01
4.42276746e-01 1.02701128e+00 6.88785493e-01 1.62114590e-01
-6.94788516e-01 -4.20981616e-01 -4.72561389e-01 3.71465266e-01
-1.28069615e+00 -3.60623568e-01 6.76437199e-01 -6.68434918e-01
6.57143474e-01 6.51524365e-01 7.46113479e-01 4.96859699e-01
5.18553555e-01 4.80136037e-01 1.18786418e+00 -5.69224536e-01
2.82597691e-01 2.29789704e-01 4.34855014e-01 6.27522349e-01
3.52574199e-01 -2.90811777e-01 -1.10101990e-01 -6.09757125e-01
2.38875791e-01 7.31878400e-01 -4.15751524e-02 -4.16912921e-02
-9.65214074e-01 5.99841774e-01 1.15413293e-01 5.18156171e-01
-8.87943804e-01 -5.56011438e-01 4.19321477e-01 1.57198742e-01
1.42208070e-01 5.08423597e-02 -6.81702495e-01 -3.45528632e-01
-4.09133166e-01 5.99258356e-02 6.63441241e-01 5.26593864e-01
3.86613697e-01 -5.61690271e-01 -8.19179639e-02 6.20952904e-01
4.50880617e-01 3.19823623e-01 7.56304979e-01 -3.94128472e-01
3.82636845e-01 1.19747889e+00 -2.76475046e-02 -1.30831277e+00
-6.81249201e-01 -3.48210603e-01 -1.30162382e+00 -4.36203331e-01
4.15854789e-02 -9.17398259e-02 -7.91701794e-01 9.28245008e-01
7.73441613e-01 2.28243425e-01 2.25718588e-01 5.03691733e-01
6.93027616e-01 3.54341149e-01 1.39998361e-01 -5.44575930e-01
1.56158078e+00 -2.26340175e-01 -6.81189239e-01 4.81552720e-01
6.89298570e-01 -8.91953468e-01 7.41324186e-01 6.53075755e-01
-5.20102918e-01 -3.08947235e-01 -4.33664382e-01 2.32077152e-01
-3.33197534e-01 2.09654778e-01 6.84565485e-01 7.17355609e-01
-4.74532902e-01 2.97055095e-01 -9.56176579e-01 -7.28951871e-01
7.22688377e-01 4.58450079e-01 -4.35249269e-01 -4.12465900e-01
-7.69752204e-01 5.25782228e-01 8.20872560e-02 3.07529140e-02
-2.87111759e-01 -5.95179915e-01 -3.84358764e-01 -9.29606929e-02
1.14586949e-01 -7.23947048e-01 6.37103200e-01 -4.52601075e-01
-1.07962990e+00 6.49786592e-01 -1.66767985e-01 -2.52697676e-01
4.83016312e-01 -3.06408912e-01 -5.05414248e-01 1.81299560e-02
-1.26538366e-01 -1.22519903e-01 5.09840250e-01 -7.70548820e-01
-7.27530122e-01 -8.83398831e-01 -7.14033008e-01 1.09567083e-01
-5.18629313e-01 1.22872874e-01 -1.12138629e-01 -3.59542906e-01
2.04447821e-01 -1.03783977e+00 -4.01493549e-01 -5.24873674e-01
-5.34036338e-01 -2.00150788e-01 9.46673989e-01 -8.46124411e-01
1.54562354e+00 -2.22718096e+00 1.27589777e-01 4.85689431e-01
3.13948065e-01 4.59314942e-01 4.14973557e-01 7.42077172e-01
-3.40488832e-03 -2.33986904e-03 -1.29397675e-01 1.28414646e-01
-5.16294539e-01 1.05939001e-01 4.01912212e-01 6.65621817e-01
7.47333243e-02 4.24725831e-01 -5.58117568e-01 -5.23448825e-01
4.28490490e-01 7.72913754e-01 -3.77216756e-01 2.14747086e-01
3.48010600e-01 7.49781847e-01 -7.96741486e-01 9.39980090e-01
6.70657218e-01 -2.56113559e-01 2.32046530e-01 -3.74180615e-01
-2.32093632e-01 -3.22529405e-01 -1.30039811e+00 1.25038576e+00
-1.42157733e-01 -1.19764894e-01 -2.41646364e-01 -9.44046736e-01
9.71016884e-01 5.63111067e-01 1.08679497e+00 -3.59944075e-01
3.19363147e-01 2.84095585e-01 2.36799479e-01 -1.04239380e+00
-8.41271356e-02 1.88672721e-01 7.78273046e-02 -1.20993450e-01
-5.06494999e-01 5.28770089e-01 3.20028141e-02 5.38725741e-02
1.38983893e+00 -2.73846328e-01 9.83254790e-01 -2.88189024e-01
9.54533279e-01 -3.01312562e-02 7.18228340e-01 1.91139534e-01
-2.01543942e-01 2.98421353e-01 2.29618534e-01 -5.36155999e-01
-8.53427768e-01 -6.51011050e-01 -4.22356009e-01 2.66314656e-01
-2.13517353e-01 -3.22299898e-01 -3.54922771e-01 -4.72019553e-01
1.55529439e-01 2.71631837e-01 -4.18029696e-01 5.71819060e-02
-1.62211716e-01 -1.08926833e+00 1.46004707e-01 7.37564580e-04
4.58269626e-01 -8.60057533e-01 -9.89610910e-01 3.21296662e-01
6.27410263e-02 -6.31533980e-01 9.71696004e-02 1.26795158e-01
-1.00556016e+00 -1.51531899e+00 -6.10997617e-01 -4.42628056e-01
6.96223676e-01 1.00982651e-01 5.89810252e-01 1.99424207e-01
-7.81935692e-01 1.85432166e-01 -5.47575057e-01 -6.57781303e-01
-2.34710440e-01 -2.68918965e-02 1.37766540e-01 1.54818147e-01
8.32137048e-01 -4.42432821e-01 -8.34168613e-01 5.76149188e-02
-8.80435109e-01 -9.68342740e-03 9.32023048e-01 8.81010652e-01
7.29231417e-01 3.54937434e-01 8.74826670e-01 -1.23848808e+00
7.20982969e-01 -8.53192389e-01 -3.21555525e-01 8.92435610e-02
-7.42512941e-01 -2.45941073e-01 6.48191571e-01 -1.21567540e-01
-8.42258453e-01 3.47489774e-01 -2.59742469e-01 -9.33821052e-02
-5.43910861e-01 7.66528904e-01 -1.75966084e-01 1.66251555e-01
3.50289941e-01 1.16116688e-01 1.04199396e-02 -8.06325138e-01
-1.75152123e-01 1.32456934e+00 7.24232420e-02 -4.60224338e-02
4.24248099e-01 3.37417334e-01 4.08230633e-01 -9.15580630e-01
-2.97284573e-01 -1.10215461e+00 -4.96268123e-01 4.90327813e-02
8.61103773e-01 -7.63941526e-01 -8.41027796e-01 4.32317674e-01
-6.57665730e-01 5.61707735e-01 8.16409215e-02 6.34162426e-01
-9.70360339e-02 3.14244270e-01 -4.16091859e-01 -1.16995859e+00
-7.07285047e-01 -1.03884470e+00 7.66374826e-01 4.18510705e-01
-3.55155081e-01 -6.30777121e-01 4.68520150e-02 5.51994264e-01
4.02489185e-01 7.41602480e-01 1.25284660e+00 -8.56831133e-01
-9.49685574e-02 -5.21875679e-01 -9.16792005e-02 1.73614725e-01
6.20516777e-01 -9.49220825e-03 -7.79509723e-01 -2.50002414e-01
3.58429611e-01 2.08265707e-01 2.70405591e-01 3.64042878e-01
1.06502604e+00 -4.27057505e-01 -4.69829291e-01 3.73789489e-01
1.67275858e+00 6.47051454e-01 7.10358977e-01 4.57421750e-01
7.13517904e-01 4.37707961e-01 7.49339163e-01 1.02355433e+00
4.15059358e-01 3.12062770e-01 3.77069205e-01 -1.76762030e-01
2.65023410e-01 3.02830070e-01 -7.23831728e-02 1.19714522e+00
-1.20795764e-01 2.03223020e-01 -1.20373011e+00 5.10943770e-01
-1.85061753e+00 -6.44172072e-01 -5.90842009e-01 2.17686415e+00
5.49738765e-01 -2.85074174e-01 6.39285371e-02 7.57728517e-01
4.36480492e-01 -6.95952892e-01 -4.23242807e-01 -4.18953180e-01
3.15426677e-01 1.45759493e-01 3.22553426e-01 1.66550521e-02
-1.14103591e+00 7.59541914e-02 3.97034287e+00 4.03146148e-01
-9.53436673e-01 5.56207746e-02 8.06018710e-01 2.03006878e-01
7.49735460e-02 -3.79773468e-01 -4.85909998e-01 6.64972425e-01
9.27669823e-01 2.02045888e-02 2.96937108e-01 8.67654741e-01
6.88100755e-01 -3.55542660e-01 -7.15170205e-01 1.13586926e+00
-4.09494549e-01 -1.04992306e+00 -2.64736772e-01 2.87101686e-01
7.18039572e-01 -3.86410058e-02 -2.53721941e-02 -1.53721035e-01
-2.61832058e-01 -9.64085817e-01 -3.65467548e-01 8.03334594e-01
6.07857049e-01 -1.03635979e+00 1.19063294e+00 4.21946555e-01
-1.06350207e+00 -5.13314545e-01 -6.12716600e-02 2.53209807e-02
-2.21318588e-01 1.04166055e+00 -1.01276541e+00 9.86414075e-01
9.06945884e-01 6.49627149e-01 -5.30416369e-01 1.25825191e+00
5.34941971e-01 4.99289066e-01 -1.39550626e-01 5.02039567e-02
-1.89962089e-01 -2.27169320e-01 3.33372593e-01 1.08061743e+00
6.75181746e-01 3.28744560e-01 5.24020195e-02 1.81217969e-03
2.45018378e-01 6.14667833e-01 -6.44657731e-01 -3.06462739e-02
3.58289123e-01 1.37421691e+00 -6.32929981e-01 -2.55633086e-01
-4.73304659e-01 4.17350292e-01 -2.72981286e-01 -8.16111416e-02
-3.44455332e-01 -4.06169891e-01 4.84960288e-01 3.12684804e-01
6.28011525e-02 1.16937287e-01 -3.18291992e-01 -8.58405113e-01
-2.47990161e-01 -1.16403627e+00 5.54672599e-01 -2.98948176e-02
-1.31050062e+00 4.45665210e-01 -1.73022032e-01 -1.30449021e+00
-2.23246180e-02 -3.18385005e-01 -2.84967750e-01 8.32997561e-01
-1.40481806e+00 -8.10404718e-01 -5.70921361e-01 8.05896223e-01
3.15419883e-01 -3.12694460e-01 1.15471041e+00 7.88877547e-01
-6.57821715e-01 1.22918174e-01 5.09091556e-01 -1.46396130e-01
3.83517146e-01 -1.00350332e+00 -4.03523594e-01 7.09248781e-01
-3.93279046e-01 7.40678847e-01 6.09555006e-01 -7.06248224e-01
-1.92246604e+00 -1.01798177e+00 8.21243584e-01 4.24379855e-02
1.36156753e-01 1.55479163e-01 -7.77781904e-01 2.19642028e-01
-1.08703256e-01 6.09422438e-02 1.13977385e+00 -1.23404808e-01
3.56365919e-01 -3.88864994e-01 -1.58357012e+00 2.67240882e-01
3.92024994e-01 -2.38491464e-02 -2.58276910e-01 3.50975037e-01
-1.18794113e-01 -1.77789688e-01 -1.08156252e+00 5.72017968e-01
7.51332700e-01 -1.06487250e+00 7.52765536e-01 -4.78822857e-01
1.32998765e-01 -3.70967835e-01 -2.17346892e-01 -8.98970723e-01
-3.38680267e-01 -4.78662819e-01 -3.10226023e-01 1.16937232e+00
-4.53543253e-02 -6.47668302e-01 7.31327593e-01 6.22714341e-01
1.76627159e-01 -1.24427056e+00 -8.70233655e-01 -9.28719938e-02
-6.70285285e-01 -3.20882574e-02 5.13677955e-01 9.12359476e-01
-7.80846328e-02 6.60054013e-02 -2.10268795e-01 1.21485710e-01
5.15353858e-01 -2.28180438e-01 7.70592868e-01 -1.68606532e+00
-2.89023191e-01 -2.59416997e-02 -7.89113283e-01 -5.05699515e-02
-8.92582238e-01 -6.89647794e-01 -5.04319251e-01 -1.70369661e+00
3.15574914e-01 -7.02324450e-01 -6.26119494e-01 2.35514760e-01
-2.87700325e-01 -1.33960783e-01 3.63357291e-02 4.08554405e-01
-1.92386672e-01 6.05596639e-02 9.67618406e-01 2.05535099e-01
-4.71990436e-01 3.77608746e-01 -8.20424497e-01 6.65547907e-01
9.42994475e-01 -6.00404441e-01 -3.75176787e-01 1.18707038e-01
7.19083846e-02 1.19863719e-01 -1.22791035e-02 -1.04852057e+00
4.27863561e-02 -1.84287578e-01 6.21444345e-01 -5.60730457e-01
1.30957728e-02 -1.30811226e+00 8.00704598e-01 9.12583768e-01
2.93440465e-02 2.45538056e-01 -1.41720220e-01 5.84106326e-01
-1.32142514e-01 3.30608748e-02 6.04349017e-01 -1.74979493e-01
-3.32845509e-01 1.53141990e-01 -1.44790351e-01 -5.36483228e-01
1.45371056e+00 -1.34606317e-01 -1.07961029e-01 1.56800285e-01
-5.42832851e-01 -6.01781942e-02 1.61774486e-01 1.04081407e-01
7.20353127e-01 -1.10769558e+00 -6.40291870e-01 2.88950890e-01
4.90801856e-02 1.32323980e-01 4.28153008e-01 1.48888803e+00
-5.89554548e-01 4.10609603e-01 -1.91103429e-01 -6.34018660e-01
-1.68406212e+00 4.89189476e-01 -2.37655193e-01 -4.77313846e-01
-7.30874956e-01 3.03390265e-01 -2.96054453e-01 -1.08674034e-01
1.54497162e-01 -2.99295276e-01 -6.13191366e-01 4.69262972e-02
5.03773689e-01 7.29814768e-01 4.18227881e-01 -5.73533237e-01
-6.44363940e-01 6.29714012e-01 -1.12655081e-01 5.34937859e-01
1.57446778e+00 -1.67508632e-01 -2.35071227e-01 3.27654153e-01
1.15186548e+00 1.30542845e-01 -4.93560433e-01 1.51710108e-01
4.02121007e-01 -5.47722042e-01 -5.29340170e-02 -7.63017774e-01
-7.88132429e-01 5.68585277e-01 1.08380318e+00 3.00094545e-01
1.53845656e+00 -2.39457101e-01 7.17813790e-01 3.02511334e-01
1.34711832e-01 -7.98097014e-01 -3.19528699e-01 1.60018653e-02
3.87982488e-01 -1.20375347e+00 2.31057942e-01 -3.32828104e-01
-7.05976903e-01 1.13913608e+00 2.19840296e-02 4.32573892e-02
1.02397811e+00 2.32094273e-01 5.88807613e-02 -2.01017961e-01
-6.00376785e-01 1.17889963e-01 1.62181094e-01 4.11923468e-01
5.70525110e-01 2.67747670e-01 -7.19176471e-01 6.68927610e-01
1.37340382e-01 4.84040231e-01 2.88980097e-01 1.18019962e+00
-2.65662730e-01 -1.12406886e+00 -5.29452980e-01 1.07152271e+00
-9.43715513e-01 1.14193186e-01 -3.51107940e-02 5.08080244e-01
3.96789551e-01 1.05672550e+00 -4.92497385e-01 -4.48164493e-01
2.70958006e-01 8.42774510e-02 2.57106237e-02 -3.79170954e-01
-7.64706671e-01 2.33537421e-01 -1.03095785e-01 -3.39584291e-01
-5.07161140e-01 -8.01929593e-01 -1.29619408e+00 3.90526615e-02
-2.11509332e-01 1.85442746e-01 9.67215002e-01 8.80339265e-01
7.10605383e-01 5.51736951e-01 7.95718133e-01 -2.50063509e-01
-4.30887878e-01 -8.49077523e-01 -4.41680819e-01 5.07623613e-01
2.99477428e-01 -5.19259214e-01 5.26851788e-02 1.59624815e-02] | [8.439895629882812, 4.840684413909912] |
c74bc84f-ec38-4933-979d-a830497f6cd1 | a-hybrid-system-of-sound-event-detection | 2210.09529 | null | https://arxiv.org/abs/2210.09529v1 | https://arxiv.org/pdf/2210.09529v1.pdf | A Hybrid System of Sound Event Detection Transformer and Frame-wise Model for DCASE 2022 Task 4 | In this paper, we describe in detail our system for DCASE 2022 Task4. The system combines two considerably different models: an end-to-end Sound Event Detection Transformer (SEDT) and a frame-wise model, Metric Learning and Focal Loss CNN (MLFL-CNN). The former is an event-wise model which learns event-level representations and predicts sound event categories and boundaries directly, while the latter is based on the widely adopted frame-classification scheme, under which each frame is classified into event categories and event boundaries are obtained by post-processing such as thresholding and smoothing. For SEDT, self-supervised pre-training using unlabeled data is applied, and semi-supervised learning is adopted by using an online teacher, which is updated from the student model using the Exponential Moving Average (EMA) strategy and generates reliable pseudo labels for weakly-labeled and unlabeled data. For the frame-wise model, the ICT-TOSHIBA system of DCASE 2021 Task 4 is used. Experimental results show that the hybrid system considerably outperforms either individual model and achieves psds1 of 0.420 and psds2 of 0.783 on the validation set without external data. The code is available at https://github.com/965694547/Hybrid-system-of-frame-wise-model-and-SEDT. | ['Kazushige Ouchi', 'Long Yan', 'Rui Tao', 'Yueliang Qian', 'Hong Liu', 'Xiangdong Wang', 'Zhirong Ye', 'Zhifang Guo', 'Yiming Li'] | 2022-10-18 | null | null | null | null | ['sound-event-detection'] | ['audio'] | [-9.43470374e-03 1.19102411e-01 1.98272243e-01 -4.51374203e-01
-1.48252368e+00 -2.83979416e-01 4.02236879e-01 1.13999687e-01
-6.20268583e-01 6.00199699e-01 8.08701888e-02 -1.40685067e-01
1.87343106e-01 -6.10088825e-01 -7.04269171e-01 -8.40908170e-01
-1.55858219e-01 2.52295792e-01 7.36046135e-01 3.40198934e-01
-3.77758034e-02 2.30460197e-01 -1.70907187e+00 5.80684364e-01
2.84276664e-01 1.48655391e+00 2.66984493e-01 1.17703092e+00
1.21859908e-01 1.20760977e+00 -5.69826186e-01 -1.47603914e-01
1.22568227e-01 -5.84060431e-01 -6.86606884e-01 -2.78279752e-01
1.48669332e-01 -2.52690107e-01 -5.31994343e-01 6.27827883e-01
9.45272863e-01 3.02209735e-01 4.16425288e-01 -1.26047432e+00
9.86622125e-02 3.12782735e-01 -2.51130074e-01 5.57965934e-01
1.90753073e-01 1.44895211e-01 7.39225447e-01 -1.19039702e+00
3.78377527e-01 9.83116627e-01 7.46285856e-01 4.68991280e-01
-7.63810575e-01 -8.32406640e-01 -1.63809195e-01 6.91674829e-01
-1.37679613e+00 -6.34361327e-01 5.29810071e-01 -4.14861143e-01
9.64453936e-01 7.61996955e-02 4.64066803e-01 1.03597617e+00
7.73756430e-02 8.63815069e-01 8.93001020e-01 -6.09903991e-01
5.29472709e-01 -1.26735017e-01 -3.40657532e-02 4.64171290e-01
-4.89885926e-01 3.65227312e-01 -8.17322671e-01 -5.48870489e-02
4.02329743e-01 -2.21398503e-01 -3.05447280e-01 2.96009421e-01
-1.12533128e+00 4.60162699e-01 1.89419448e-01 3.09613317e-01
-4.69163984e-01 1.24656178e-01 7.39529252e-01 2.70398527e-01
8.54394257e-01 -1.99955702e-01 -5.38983643e-01 -4.84384418e-01
-1.53925085e+00 1.87485740e-01 5.96412361e-01 8.27932775e-01
4.78467345e-01 1.52976960e-01 -4.62607861e-01 9.48716402e-01
2.05100238e-01 4.33982342e-01 6.36352241e-01 -1.18136895e+00
2.28651226e-01 -2.29053628e-02 -5.04508475e-03 -5.26984572e-01
-4.59893584e-01 -2.72517413e-01 -5.55326104e-01 1.22105509e-01
4.99370158e-01 -4.10860062e-01 -9.30023193e-01 1.82483315e+00
3.74506205e-01 8.48080873e-01 -1.18768424e-01 8.01259577e-01
9.71531987e-01 8.80860031e-01 3.27027738e-01 -4.63774234e-01
1.29771495e+00 -9.23452377e-01 -8.00257564e-01 2.04704151e-01
5.09684324e-01 -7.84608722e-01 7.05811620e-01 5.43259740e-01
-1.44566727e+00 -7.97487795e-01 -8.45545173e-01 -5.42354733e-02
-1.87635645e-01 2.43763626e-01 -2.17344061e-01 2.27754802e-01
-1.12253726e+00 5.13576746e-01 -1.03760612e+00 -2.86430180e-01
4.11613613e-01 1.03936009e-01 -1.32557705e-01 1.62648141e-01
-1.43501508e+00 5.94890177e-01 3.73362005e-01 -6.37361258e-02
-1.40960371e+00 -7.97978938e-01 -6.81760669e-01 6.45453483e-02
2.53870636e-01 -3.33715290e-01 1.87742710e+00 -9.44491565e-01
-1.75482130e+00 8.20716202e-01 -2.52018154e-01 -7.63187528e-01
5.69612026e-01 -1.07833125e-01 -4.34667826e-01 7.56087840e-01
9.32205543e-02 7.11500108e-01 1.00353944e+00 -9.97952938e-01
-1.23029172e+00 5.83459958e-02 -9.66262817e-02 1.70982644e-01
-5.23932017e-02 4.23829436e-01 -5.95633984e-01 -6.82140470e-01
-1.45302206e-01 -8.06901217e-01 1.63491622e-01 8.14633444e-02
-1.46221682e-01 -4.67845738e-01 8.30396950e-01 -9.97505903e-01
1.49278724e+00 -2.42155385e+00 -3.06901366e-01 -1.26634672e-01
7.67841041e-02 3.48534882e-01 -3.60249318e-02 3.08282197e-01
-3.24379981e-01 -2.57789403e-01 -2.44402140e-01 -6.10770822e-01
-2.03298479e-01 -1.60140038e-01 -1.80890292e-01 3.22566897e-01
2.04766661e-01 3.85189742e-01 -9.98308122e-01 -6.37448370e-01
3.64833146e-01 4.88198906e-01 -2.68063724e-01 5.79482079e-01
-6.43249899e-02 2.99932241e-01 -1.12732321e-01 5.53205788e-01
4.87781376e-01 5.39106056e-02 -1.64902493e-01 -2.54630119e-01
-3.23636562e-01 4.57031786e-01 -1.22026944e+00 1.61608768e+00
-5.46015441e-01 6.55335784e-01 -5.48677752e-03 -1.07346785e+00
6.60731077e-01 9.45357978e-01 5.70994794e-01 -6.23447895e-01
2.20511511e-01 3.41823667e-01 -3.73539925e-01 -6.48213446e-01
2.74106711e-01 -1.11213036e-01 1.05530350e-02 4.26553011e-01
4.49913800e-01 -6.00772835e-02 3.65384549e-01 1.69468120e-01
1.34466147e+00 4.15520102e-01 6.00007512e-02 1.19969249e-02
5.47237515e-01 -1.13469429e-01 7.99082100e-01 6.39151812e-01
-4.85428661e-01 1.06744444e+00 3.57398987e-01 -2.22850263e-01
-7.56365597e-01 -1.08509648e+00 -2.41596445e-01 1.08590961e+00
-7.33625367e-02 -3.09609771e-01 -1.00963593e+00 -7.25083470e-01
-4.13098335e-01 9.91918266e-01 -3.74613613e-01 -2.55876094e-01
-5.76873362e-01 -4.59551454e-01 7.19147384e-01 5.59808373e-01
6.11634731e-01 -1.39385056e+00 -6.97740376e-01 5.37554443e-01
-6.44269884e-01 -1.07412076e+00 -4.06467915e-01 4.02042985e-01
-5.75347483e-01 -1.03791022e+00 -7.15532839e-01 -7.51643300e-01
2.77766824e-01 3.19447033e-02 9.69897985e-01 -1.64619744e-01
-2.29847208e-02 5.56321681e-01 -6.14461720e-01 -5.14648199e-01
-3.99986833e-01 -1.97473854e-01 -4.01566625e-02 2.58889765e-01
4.08669144e-01 -6.27685249e-01 -7.66578972e-01 2.98298597e-01
-7.67288625e-01 6.39033616e-02 2.18380034e-01 7.22207606e-01
7.15302825e-01 -1.11332640e-01 8.72728050e-01 -4.33324695e-01
1.00264005e-01 -6.33350968e-01 -4.48013157e-01 -3.65813583e-04
-3.03530514e-01 -6.19796753e-01 6.50880396e-01 -4.25143242e-01
-1.22498941e+00 -3.19093512e-03 -3.99267197e-01 -7.53080964e-01
-3.50265056e-01 1.03308655e-01 -8.51108506e-02 4.20453548e-01
5.20159125e-01 3.27122837e-01 -2.85007060e-01 -4.14904445e-01
3.46887484e-02 1.10294855e+00 9.08829510e-01 -4.50289935e-01
5.23855627e-01 4.09017771e-01 -4.73371208e-01 -6.82301819e-01
-1.04459441e+00 -4.28998262e-01 -4.47136372e-01 -6.86536849e-01
1.11399901e+00 -1.24109387e+00 -3.04153293e-01 8.47776055e-01
-1.09684300e+00 -6.76799893e-01 -6.60064340e-01 8.01444530e-01
-8.24076116e-01 4.02438454e-02 -9.28505778e-01 -7.39408851e-01
-4.48365122e-01 -7.99639940e-01 1.03884006e+00 4.00115877e-01
-1.61591545e-02 -6.23619258e-01 2.63528138e-01 2.88417131e-01
2.18348697e-01 2.25060731e-01 3.79430950e-01 -8.76433790e-01
-2.63117999e-01 -2.08778217e-01 -3.95057946e-02 8.01766753e-01
-1.99687034e-01 -3.51555869e-02 -1.45950532e+00 -1.11206919e-01
1.46821663e-01 -5.01915693e-01 7.68556118e-01 6.41288698e-01
1.43359411e+00 1.96892843e-01 -2.44085002e-03 4.79027182e-01
1.05587947e+00 6.05673492e-01 7.06379354e-01 2.84534872e-01
3.35591525e-01 4.43287462e-01 8.45258117e-01 7.08190978e-01
7.49650300e-01 4.80309457e-01 2.15420827e-01 -9.62511674e-02
-4.31501716e-01 -1.52589321e-01 5.71091056e-01 8.29536259e-01
-7.07728788e-02 -4.31941986e-01 -7.40436614e-01 6.79465234e-01
-1.87942719e+00 -1.22182667e+00 -1.06121242e-01 2.18335080e+00
9.49744880e-01 2.14260831e-01 2.38740370e-01 4.76813555e-01
1.06654024e+00 3.90744023e-02 -3.27231020e-01 -1.57443047e-01
6.83913231e-02 3.43245655e-01 4.65140902e-02 3.35341424e-01
-1.19653273e+00 7.53643334e-01 4.96989107e+00 1.15712416e+00
-1.09739602e+00 6.22350335e-01 7.35664546e-01 -2.89797425e-01
4.94382828e-01 -1.38908461e-01 -6.80038512e-01 8.42656136e-01
1.58812881e+00 -3.14990357e-02 1.87894821e-01 5.21784663e-01
7.26106822e-01 -2.98795491e-01 -9.91078675e-01 9.03916121e-01
-3.10885273e-02 -1.10515654e+00 -4.21596169e-01 -3.69280010e-01
4.56094891e-01 1.18135653e-01 -3.71125519e-01 5.42168379e-01
-5.70791662e-02 -2.51146168e-01 1.39038420e+00 7.20535517e-01
9.59643185e-01 -7.05976665e-01 7.04745471e-01 3.22801203e-01
-1.42692661e+00 -1.19419277e-01 -1.25327408e-01 1.15754470e-01
4.67848420e-01 8.68720293e-01 -4.48838264e-01 5.25841594e-01
1.17613018e+00 6.65461898e-01 -3.01021099e-01 1.23077130e+00
-3.73000085e-01 1.35514438e+00 -4.58506703e-01 3.88338089e-01
-2.39939690e-02 2.38846883e-01 5.94760478e-01 1.49518859e+00
4.96318072e-01 2.06825092e-01 3.69245857e-02 4.50171977e-01
8.96282401e-03 -1.04046427e-01 -6.57081902e-02 4.63159651e-01
7.12179422e-01 1.44707310e+00 -7.24636078e-01 -7.17661083e-01
-4.41961884e-01 7.23777413e-01 4.65408005e-02 2.30708569e-01
-1.23109448e+00 -5.97499013e-01 1.35732710e-01 5.41892983e-02
5.01401067e-01 1.63652435e-01 9.24857408e-02 -9.94513094e-01
-1.58599317e-01 -6.78731084e-01 6.32866323e-01 -1.04473054e+00
-1.12063146e+00 7.67404616e-01 1.97717980e-01 -1.51986694e+00
-2.55741000e-01 -1.43013015e-01 -9.13958609e-01 4.62519497e-01
-1.57624900e+00 -1.00099385e+00 -2.90548950e-01 7.90081143e-01
9.07102287e-01 3.90934385e-03 4.69426304e-01 6.51571929e-01
-6.20079398e-01 5.96478283e-01 -4.00634930e-02 -4.44995845e-03
8.42630684e-01 -1.28286505e+00 2.10398119e-02 9.36796486e-01
-5.31073771e-02 -2.85200775e-01 4.66911256e-01 -4.90292609e-01
-6.49493277e-01 -1.41760898e+00 1.32652402e+00 -1.61136135e-01
5.85743070e-01 -3.26534450e-01 -9.21956360e-01 4.45973814e-01
7.19263703e-02 3.94251078e-01 6.17935836e-01 -5.15326738e-01
8.88932031e-03 -2.28547543e-01 -1.04438090e+00 8.03616419e-02
8.09371829e-01 -6.46400630e-01 -5.96962273e-01 4.53598946e-01
6.11081541e-01 -5.30520618e-01 -9.32223678e-01 3.98681700e-01
3.14136356e-01 -1.08274865e+00 6.30262256e-01 1.24142189e-02
3.61308873e-01 -5.22647738e-01 -2.26279289e-01 -9.80605781e-01
-1.33454442e-01 -5.78614056e-01 -3.20574224e-01 1.58113217e+00
2.38580257e-01 -3.39908242e-01 3.46818805e-01 1.30144566e-01
-5.41836083e-01 -6.49737597e-01 -1.33155382e+00 -7.29299545e-01
-1.26995012e-01 -9.14765477e-01 3.07618797e-01 6.04312539e-01
-3.10215622e-01 8.02688748e-02 -1.66647956e-01 2.61738718e-01
5.25839388e-01 -2.27836505e-01 3.91269982e-01 -8.83073330e-01
-1.49165034e-01 -3.87798324e-02 -2.13557303e-01 -8.04073691e-01
1.87796250e-01 -6.88899219e-01 3.92634630e-01 -1.13363874e+00
-2.64475401e-02 -1.06285498e-01 -6.30140424e-01 5.69530785e-01
1.19597353e-02 4.76392657e-01 1.53110817e-01 1.21053465e-01
-9.68605101e-01 6.10634565e-01 6.94647610e-01 2.73232102e-01
-1.97362617e-01 3.30976993e-01 -6.82850704e-02 8.82588983e-01
8.48528802e-01 -8.09667289e-01 -1.71270549e-01 -1.06270783e-01
-2.90146142e-01 3.75960827e-01 5.39447665e-01 -1.32546496e+00
3.78473818e-01 3.34041148e-01 1.25790894e-01 -7.17513621e-01
2.17521876e-01 -6.52690232e-01 1.11125879e-01 3.90606105e-01
-4.24897552e-01 -2.50562727e-01 9.35966447e-02 5.15470803e-01
-3.62106770e-01 -2.63931721e-01 1.09700108e+00 6.94352984e-02
-7.01369941e-01 2.60200053e-01 -6.83365703e-01 2.84234166e-01
1.02274680e+00 -9.41186026e-02 -8.01696032e-02 -4.13972348e-01
-1.05545735e+00 1.21364146e-01 -6.30687177e-02 2.45388359e-01
4.83269930e-01 -1.38782740e+00 -8.36037576e-01 -7.35724997e-03
-5.06471423e-03 -1.43951587e-02 4.77990776e-01 1.03098309e+00
-3.63946676e-01 -7.63845444e-02 1.46343365e-01 -6.59809530e-01
-1.15038681e+00 1.58821404e-01 4.29943442e-01 -2.73211658e-01
-6.23444498e-01 8.62880528e-01 8.44389126e-02 -1.92528263e-01
4.64695483e-01 -3.32464397e-01 -1.68557569e-01 2.22763062e-01
6.71508491e-01 7.08352149e-01 2.35685527e-01 -6.59879208e-01
-2.71892369e-01 1.97208777e-01 1.65927157e-01 -2.47600049e-01
1.36625981e+00 -2.58688629e-01 2.45766357e-01 6.06291592e-01
1.09833884e+00 -4.46819484e-01 -1.60589314e+00 -2.91598916e-01
-1.60289362e-01 -4.61302400e-02 3.75366300e-01 -7.62264490e-01
-1.09480584e+00 1.00932395e+00 9.29723859e-01 1.65020853e-01
1.57965827e+00 2.10574307e-02 1.06287420e+00 -3.48122492e-02
1.48222089e-01 -1.25151753e+00 2.95707155e-02 5.82175553e-01
7.58861303e-01 -9.93777752e-01 -4.02802110e-01 -1.09644784e-02
-4.84334201e-01 1.08870900e+00 5.64220011e-01 -1.23072520e-01
8.10773373e-01 4.05550599e-01 1.28980160e-01 2.15511397e-02
-1.06704772e+00 -2.65876293e-01 7.36661553e-02 5.35420418e-01
1.15513846e-01 -1.34559840e-01 -2.07044095e-01 9.27198708e-01
-3.23708653e-02 3.93337220e-01 3.82197022e-01 1.12495053e+00
-3.68762940e-01 -7.09233880e-01 -3.63629818e-01 3.08465540e-01
-6.91109955e-01 -6.90328032e-02 2.82490402e-01 6.23783827e-01
5.61014414e-01 1.20252335e+00 2.47467533e-01 -4.11008298e-01
4.07286167e-01 2.82735586e-01 2.03753993e-01 -5.06125331e-01
-8.23189080e-01 4.57505137e-01 1.04691252e-01 -6.58013105e-01
-5.12624621e-01 -8.44624937e-01 -1.46021450e+00 2.65087411e-02
-4.02853847e-01 1.94712102e-01 5.07064044e-01 8.37277710e-01
3.37214500e-01 6.58446848e-01 9.38696027e-01 -1.16763687e+00
-2.50534683e-01 -1.05227709e+00 -5.30930221e-01 1.52407631e-01
3.18627059e-01 -5.76812983e-01 -7.77188361e-01 5.64972997e-01] | [15.184829711914062, 5.065791606903076] |
651207aa-82ad-4858-bbc7-48f2a620aeb4 | cs-tgn-community-search-via-temporal-graph | 2303.08964 | null | https://arxiv.org/abs/2303.08964v1 | https://arxiv.org/pdf/2303.08964v1.pdf | CS-TGN: Community Search via Temporal Graph Neural Networks | Searching for local communities is an important research challenge that allows for personalized community discovery and supports advanced data analysis in various complex networks, such as the World Wide Web, social networks, and brain networks. The evolution of these networks over time has motivated several recent studies to identify local communities in temporal networks. Given any query nodes, Community Search aims to find a densely connected subgraph containing query nodes. However, existing community search approaches in temporal networks have two main limitations: (1) they adopt pre-defined subgraph patterns to model communities, which cannot find communities that do not conform to these patterns in real-world networks, and (2) they only use the aggregation of disjoint structural information to measure quality, missing the dynamic of connections and temporal properties. In this paper, we propose a query-driven Temporal Graph Convolutional Network (CS-TGN) that can capture flexible community structures by learning from the ground-truth communities in a data-driven manner. CS-TGN first combines the local query-dependent structure and the global graph embedding in each snapshot of the network and then uses a GRU cell with contextual attention to learn the dynamics of interactions and update node embeddings over time. We demonstrate how this model can be used for interactive community search in an online setting, allowing users to evaluate the found communities and provide feedback. Experiments on real-world temporal graphs with ground-truth communities validate the superior quality of the solutions obtained and the efficiency of our model in both temporal and interactive static settings. | ['Milad Rezaei Hajidehi', 'Ali Behrouz', 'Farnoosh Hashemi'] | 2023-03-15 | null | null | null | null | ['community-search'] | ['graphs'] | [-2.83935726e-01 -9.64869708e-02 -2.28421465e-01 9.41012353e-02
2.90272593e-01 -9.44569767e-01 4.07893986e-01 6.15904570e-01
-2.44520873e-01 2.77393937e-01 1.19653605e-01 -1.30390793e-01
-5.59731781e-01 -1.12703514e+00 -3.43670398e-01 -4.83125001e-01
-8.46253395e-01 6.56872213e-01 7.71996021e-01 -3.14041823e-01
-4.75633852e-02 3.59146744e-01 -1.00571930e+00 -1.44325182e-01
8.40534747e-01 5.64911485e-01 1.39394432e-01 4.83292967e-01
2.15902701e-02 4.57567185e-01 -8.85131285e-02 -2.44765803e-01
1.05207644e-01 -1.01918839e-01 -8.59885037e-01 7.00374097e-02
5.24504557e-02 7.83591047e-02 -8.60888422e-01 8.84496570e-01
3.92875254e-01 2.84414832e-02 2.20318392e-01 -1.51228702e+00
-7.36649811e-01 8.05340946e-01 -5.35674572e-01 6.61387980e-01
3.66372138e-01 2.92942613e-01 1.49220085e+00 -5.24467111e-01
1.14565206e+00 1.23186445e+00 6.84885085e-01 3.79270375e-01
-1.57441330e+00 -6.21027172e-01 5.47111988e-01 2.71373540e-01
-1.52557206e+00 4.95329909e-02 8.54323149e-01 -7.39651978e-01
7.71237254e-01 5.56666255e-02 1.37999475e+00 1.12249899e+00
-5.45633882e-02 3.45352441e-01 6.62663043e-01 1.41376004e-01
2.25478873e-01 -3.09546620e-01 1.54719591e-01 8.89303684e-01
1.89006492e-01 8.84384476e-03 -7.76327491e-01 -3.56813401e-01
9.32110310e-01 2.88979381e-01 -2.91956246e-01 -8.33534420e-01
-1.36847091e+00 9.99300957e-01 1.16033959e+00 7.95754313e-01
-1.88709661e-01 3.79170716e-01 3.45281184e-01 6.27249777e-01
5.27881205e-01 1.69617981e-01 -2.43109077e-01 2.69356370e-01
-8.43342125e-01 4.77186479e-02 9.04501021e-01 8.31683517e-01
8.36938262e-01 -4.54012871e-01 -3.29028256e-02 5.13401210e-01
3.13792944e-01 1.84310958e-01 2.09107369e-01 -6.40899241e-01
2.31421232e-01 9.74115610e-01 -3.01671326e-01 -1.59895110e+00
-3.95246804e-01 -6.18392408e-01 -1.21176505e+00 -3.88085008e-01
3.19826663e-01 1.45068392e-01 -5.80795348e-01 1.85353374e+00
4.75984603e-01 5.29780567e-01 -4.90013301e-01 8.68375659e-01
6.32029533e-01 2.30448127e-01 -4.05662119e-01 -1.32733852e-01
1.00285447e+00 -7.53891885e-01 -4.71824825e-01 -4.28049266e-02
4.45626974e-01 -1.23460321e-02 6.30559981e-01 -2.39299551e-01
-5.96805751e-01 -1.06754585e-03 -8.08820248e-01 2.67347038e-01
-3.68218631e-01 -6.93091452e-01 6.41934276e-01 2.98814207e-01
-1.62085497e+00 8.00839543e-01 -1.02150071e+00 -8.52143049e-01
4.84687835e-01 4.74149674e-01 -2.86400169e-01 -1.20778799e-01
-1.13224459e+00 1.83487430e-01 3.03289980e-01 1.93086162e-01
-1.01718986e+00 -5.61859727e-01 -6.83559060e-01 3.21942449e-01
6.87509537e-01 -6.48097217e-01 6.34571195e-01 -8.10060024e-01
-7.79734612e-01 6.22443616e-01 -1.16046838e-01 -4.27707434e-01
4.76192981e-01 4.98670489e-01 -3.70753974e-01 3.71894330e-01
1.59694940e-01 4.15063292e-01 3.05182099e-01 -7.98940659e-01
-1.25178024e-01 -2.37396851e-01 2.80553162e-01 -6.44591749e-02
-6.96031272e-01 -1.85581654e-01 -7.43215084e-01 -4.41040844e-01
4.00819480e-01 -1.01822436e+00 -5.07681847e-01 4.06081349e-01
-3.36391121e-01 -3.39819968e-01 9.30414319e-01 -2.39133447e-01
1.54057312e+00 -1.92414498e+00 4.76253331e-01 8.55548143e-01
1.06246614e+00 -3.07141095e-01 -4.55876887e-01 7.88143277e-01
1.86148182e-01 5.92198610e-01 -9.26135406e-02 -1.24884963e-01
-3.26864868e-01 2.21327007e-01 1.49299264e-01 4.52430844e-01
8.07874501e-02 1.08001411e+00 -1.45335484e+00 -5.97086072e-01
-2.76098430e-01 4.34584081e-01 -7.44276285e-01 -2.13660270e-01
-2.21720994e-01 5.20027578e-01 -5.84477782e-01 4.48412746e-01
1.70738265e-01 -1.07388580e+00 8.20978403e-01 -4.32154834e-02
9.69446376e-02 5.17291464e-02 -1.25315273e+00 1.40263367e+00
1.31461024e-01 8.21158409e-01 3.03886324e-01 -9.76857185e-01
6.62499011e-01 3.20791990e-01 9.58621204e-01 -5.80506921e-01
-1.85554326e-01 9.06934664e-02 1.81074873e-01 -2.52973169e-01
1.59517601e-01 4.28602546e-01 2.75058091e-01 8.08618903e-01
-1.28089625e-03 3.43879431e-01 4.76460338e-01 8.62092495e-01
1.68240464e+00 -6.49015307e-01 -6.11918271e-02 -3.65026563e-01
1.60784408e-01 -2.98507512e-01 6.70103431e-01 6.20212555e-01
-3.30634773e-01 3.26191068e-01 7.99633801e-01 -5.00771284e-01
-9.73816752e-01 -1.08334708e+00 3.18438917e-01 8.89354706e-01
3.77196997e-01 -6.04686916e-01 -2.09343016e-01 -5.16063631e-01
1.75877348e-01 -4.92780596e-01 -7.45936036e-01 -3.56209725e-02
-4.46805298e-01 -4.89903420e-01 5.27606234e-02 2.32997805e-01
1.33020103e-01 -1.00404429e+00 -2.34926671e-01 4.21197593e-01
-5.08308351e-01 -1.03608060e+00 -1.09378684e+00 -2.32803851e-01
-1.11131811e+00 -1.55798018e+00 -3.53908777e-01 -8.88152957e-01
7.16420889e-01 5.12963414e-01 1.12936151e+00 8.68377984e-01
-2.55379975e-01 6.92536950e-01 -4.09326434e-01 3.87854636e-01
-6.47401735e-02 3.27324957e-01 5.96360415e-02 4.13679868e-01
-9.24160331e-03 -1.27733707e+00 -7.78155386e-01 3.60373139e-01
-8.24190974e-01 -1.26212314e-01 1.97654188e-01 7.77897775e-01
3.08387786e-01 2.80109227e-01 4.92948055e-01 -6.66834116e-01
8.69256556e-01 -7.51361668e-01 -7.19932556e-01 2.22387403e-01
-9.97753620e-01 5.07407114e-02 9.28748250e-02 -7.70249665e-01
-1.58288851e-01 -2.39898637e-01 7.11413741e-01 -5.93082547e-01
7.08570778e-01 1.13389909e+00 3.39493811e-01 -2.40003988e-01
6.75351083e-01 1.75368071e-01 1.42819121e-01 -2.88156182e-01
2.22577661e-01 8.61988366e-02 1.84316128e-01 -4.69072163e-01
1.01767123e+00 7.28951335e-01 -3.37735824e-02 -7.07847178e-01
-3.28195184e-01 -8.68087292e-01 -8.63861382e-01 -5.08479118e-01
5.08811712e-01 -8.08248639e-01 -9.39471424e-01 2.88714856e-01
-9.60592628e-01 -4.69103456e-01 2.80650537e-02 1.25669509e-01
-1.52369291e-02 6.01230741e-01 -8.79257381e-01 -7.60616243e-01
-3.71824294e-01 -6.52869880e-01 6.32009625e-01 9.96780843e-02
4.00610119e-02 -1.39672327e+00 2.88688570e-01 -3.46389264e-02
5.20316720e-01 5.42650700e-01 8.19607913e-01 -4.23641622e-01
-1.23779190e+00 -1.07570760e-01 -2.62756228e-01 -3.53262991e-01
2.32428640e-01 2.51988649e-01 -3.81951213e-01 -7.68539310e-01
-9.23766911e-01 -7.88756311e-02 8.58102143e-01 2.29598671e-01
7.18033791e-01 -5.42110860e-01 -8.39691341e-01 3.97682816e-01
1.44998276e+00 -3.54223132e-01 2.67470032e-01 9.51855481e-02
7.31599152e-01 4.91986632e-01 -3.00858188e-02 3.94062161e-01
5.88800669e-01 4.73891199e-01 7.15900064e-01 7.01394975e-02
2.61429101e-01 -3.05159122e-01 1.67153418e-01 1.14760494e+00
-1.63258299e-01 -2.82288551e-01 -1.13183057e+00 1.08933115e+00
-2.23082280e+00 -1.32289624e+00 -6.79175928e-02 2.08722496e+00
9.47488308e-01 2.69706786e-01 4.39770281e-01 -2.14192718e-01
8.99766386e-01 2.80929983e-01 -6.77624881e-01 3.40943873e-01
-3.06117088e-01 -5.48572727e-02 1.83555901e-01 4.32253361e-01
-8.19714606e-01 7.70700693e-01 5.80647707e+00 1.64279714e-01
-1.06485391e+00 1.38782546e-01 3.64470363e-01 -1.44928470e-01
-5.94294846e-01 2.74375856e-01 -2.73846239e-02 3.29998165e-01
6.75712645e-01 -7.28945851e-01 1.00169671e+00 3.61994624e-01
3.23195606e-01 3.86711717e-01 -1.15082645e+00 7.95805037e-01
-4.24164027e-01 -1.67766488e+00 -1.61770448e-01 4.48967725e-01
8.47627699e-01 4.91502345e-01 -1.86596364e-01 8.50607641e-03
6.55295610e-01 -9.58613038e-01 6.24341905e-01 5.33926308e-01
7.13887393e-01 -3.45422238e-01 3.25177908e-01 4.29099202e-01
-1.84972608e+00 -2.62022942e-01 -4.85279299e-02 1.10005952e-01
-5.80186620e-02 6.77028418e-01 -9.24788773e-01 5.61841130e-01
1.03227615e+00 1.18737602e+00 -7.11395383e-01 1.40011489e+00
2.11825222e-01 5.98761439e-01 -6.82432055e-01 -2.62964368e-01
2.16951191e-01 -4.25867558e-01 8.63314509e-01 9.26633894e-01
1.79078519e-01 -2.43653625e-01 3.88593853e-01 1.19988632e+00
-3.04296941e-01 -2.81608161e-02 -6.70165181e-01 -6.65191114e-01
9.04661119e-01 1.19005978e+00 -1.07311332e+00 5.46893701e-02
-2.77811766e-01 8.39215755e-01 7.12075114e-01 6.86741233e-01
-4.10421580e-01 -1.05254129e-02 6.56620562e-01 6.42556310e-01
3.91631633e-01 -6.11339986e-01 3.42429399e-01 -1.31671834e+00
1.26568019e-01 -7.16079175e-01 6.60098076e-01 -5.29865861e-01
-1.52504647e+00 5.76897740e-01 -1.39071658e-01 -9.51434612e-01
1.28577545e-01 -5.94699383e-02 -8.42335999e-01 5.43749988e-01
-1.13655913e+00 -1.05726874e+00 -4.80459213e-01 8.14733326e-01
-2.01616548e-02 8.69499743e-02 4.56705719e-01 3.23671281e-01
-5.14590681e-01 3.47861826e-01 1.17954902e-01 5.74653566e-01
3.02157462e-01 -1.22613573e+00 4.74580884e-01 8.57418597e-01
3.24430913e-01 7.65309930e-01 4.66362625e-01 -8.75996768e-01
-1.28981268e+00 -1.09220243e+00 9.58070099e-01 -3.33264619e-01
1.23393285e+00 -6.72417998e-01 -9.94181335e-01 5.75910509e-01
4.55008447e-02 4.07969892e-01 2.82213330e-01 4.20814604e-01
-6.72242105e-01 -2.03668371e-01 -8.19237530e-01 6.99418008e-01
1.66427994e+00 -7.86716402e-01 2.20884413e-01 3.88831645e-01
1.00412345e+00 4.99474406e-02 -9.82246935e-01 1.25555977e-01
4.82188255e-01 -7.74543583e-01 8.59545052e-01 -5.72547376e-01
1.51340272e-02 -3.13484281e-01 2.63647467e-01 -1.24929965e+00
-6.66543126e-01 -8.83503675e-01 -2.89103985e-01 9.36694384e-01
4.35641289e-01 -7.64113188e-01 8.83769035e-01 1.38839245e-01
5.55916548e-01 -7.03273416e-01 -1.11402798e+00 -7.25603402e-01
-2.18151540e-01 2.72685755e-02 6.68449223e-01 1.37161911e+00
3.41313928e-02 4.95306820e-01 -3.51285338e-02 2.45345891e-01
7.82433510e-01 1.94559187e-01 4.21263725e-01 -1.86304939e+00
-2.47721538e-01 -7.37694979e-01 -5.35125375e-01 -8.53348255e-01
-1.65246874e-01 -1.07455266e+00 -4.39180166e-01 -1.72503209e+00
4.53316063e-01 -8.40950429e-01 -2.53932089e-01 4.54322994e-01
1.64891884e-03 -1.84059098e-01 2.85870023e-02 3.45334917e-01
-8.90438914e-01 5.16178727e-01 1.32146537e+00 -5.60018361e-01
-3.58785659e-01 -1.65361270e-01 -4.55098838e-01 3.40503566e-02
3.89656186e-01 -5.06455243e-01 -6.41313255e-01 -2.76145756e-01
9.27510858e-01 5.20184971e-02 6.65238380e-01 -7.34646618e-01
6.84231877e-01 -2.40665242e-01 -1.95620537e-01 -3.75845730e-01
3.70530933e-02 -9.21685576e-01 4.54094738e-01 6.08794332e-01
-1.57929465e-01 2.34653890e-01 -2.69628674e-01 1.46124876e+00
-7.31424317e-02 3.65088075e-01 4.24201131e-01 -7.17133582e-02
-4.30362552e-01 1.08567035e+00 -1.77736193e-01 1.21200666e-01
7.85773218e-01 -2.84070581e-01 -3.35778385e-01 -6.41784489e-01
-9.99353707e-01 9.33384895e-01 6.73497021e-01 4.97706383e-01
5.49362659e-01 -1.38636959e+00 -6.98190749e-01 -9.59040001e-02
3.41681629e-01 -6.52896427e-03 8.07760730e-02 1.08107054e+00
-3.23190540e-01 2.40378175e-02 1.35728046e-01 -9.30773914e-01
-1.26962066e+00 6.93087518e-01 6.87251449e-01 -5.78307986e-01
-7.93627381e-01 6.48061872e-01 -6.03188351e-02 -4.17856872e-01
2.73268372e-01 -2.21218273e-01 -3.11922729e-01 1.05199315e-01
2.21094564e-02 2.96566244e-02 -2.92133749e-01 -4.44476992e-01
-4.64013666e-01 3.25973064e-01 1.69416904e-01 -1.23686930e-02
1.57806253e+00 -2.38029048e-01 -4.95345443e-01 4.56450254e-01
1.07557440e+00 -2.07697719e-01 -8.90991330e-01 -9.19001222e-01
3.62699181e-01 -3.31803858e-01 -1.18149534e-01 -2.94473886e-01
-1.49144328e+00 5.14508426e-01 4.13730294e-01 6.57260537e-01
9.13433254e-01 2.72354543e-01 5.28642714e-01 4.88820881e-01
6.01736128e-01 -8.45608175e-01 7.06452727e-01 3.97032678e-01
8.28927398e-01 -9.63852704e-01 -1.49829134e-01 -5.08526802e-01
-1.51815549e-01 1.06913054e+00 5.12296617e-01 6.62539378e-02
1.19827092e+00 -1.76873654e-01 -5.40894032e-01 -7.20619082e-01
-1.21202278e+00 -3.23916316e-01 2.09518358e-01 5.87581396e-01
8.05430785e-02 4.16830666e-02 3.41784582e-02 2.21878722e-01
2.10230231e-01 -2.60370404e-01 6.09555185e-01 7.61621177e-01
-2.87741661e-01 -1.08145905e+00 1.50531963e-01 6.77715123e-01
8.01191181e-02 -1.58592835e-01 -7.63141155e-01 5.39202452e-01
-2.35551655e-01 1.01932871e+00 4.61609699e-02 -5.54798722e-01
7.99353644e-02 -2.26920113e-01 2.71518916e-01 -7.31536984e-01
-5.55657744e-01 3.74773927e-02 -3.43773700e-02 -6.44542456e-01
-4.51830387e-01 -7.05683231e-01 -1.01424074e+00 -7.38501072e-01
-4.68833178e-01 2.70619154e-01 2.11237688e-02 5.82787633e-01
7.09829628e-01 3.95697147e-01 7.50192940e-01 -4.62870717e-01
-1.36536211e-01 -8.30067515e-01 -5.88708758e-01 5.23230433e-01
5.97443759e-01 -5.87480724e-01 -5.38406670e-01 -5.40030301e-02] | [7.179445266723633, 5.980305194854736] |
460df0c7-1382-4e11-b8e5-956c13b65983 | fame-for-sale-efficient-detection-of-fake | 1509.04098 | null | http://arxiv.org/abs/1509.04098v2 | http://arxiv.org/pdf/1509.04098v2.pdf | Fame for sale: efficient detection of fake Twitter followers | $\textit{Fake followers}$ are those Twitter accounts specifically created to
inflate the number of followers of a target account. Fake followers are
dangerous for the social platform and beyond, since they may alter concepts
like popularity and influence in the Twittersphere - hence impacting on
economy, politics, and society. In this paper, we contribute along different
dimensions. First, we review some of the most relevant existing features and
rules (proposed by Academia and Media) for anomalous Twitter accounts
detection. Second, we create a baseline dataset of verified human and fake
follower accounts. Such baseline dataset is publicly available to the
scientific community. Then, we exploit the baseline dataset to train a set of
machine-learning classifiers built over the reviewed rules and features. Our
results show that most of the rules proposed by Media provide unsatisfactory
performance in revealing fake followers, while features proposed in the past by
Academia for spam detection provide good results. Building on the most
promising features, we revise the classifiers both in terms of reduction of
overfitting and cost for gathering the data needed to compute the features. The
final result is a novel $\textit{Class A}$ classifier, general enough to thwart
overfitting, lightweight thanks to the usage of the less costly features, and
still able to correctly classify more than 95% of the accounts of the original
training set. We ultimately perform an information fusion-based sensitivity
analysis, to assess the global sensitivity of each of the features employed by
the classifier. The findings reported in this paper, other than being supported
by a thorough experimental methodology and interesting on their own, also pave
the way for further investigation on the novel issue of fake Twitter followers. | ['Marinella Petrocchi', 'Angelo Spognardi', 'Stefano Cresci', 'Roberto Di Pietro', 'Maurizio Tesconi'] | 2015-09-14 | null | null | null | null | ['spam-detection'] | ['natural-language-processing'] | [-2.29081213e-02 2.59353608e-01 -1.17580101e-01 -2.53748715e-01
-4.62089330e-01 -6.74536526e-01 1.17569411e+00 6.83735967e-01
-4.73441482e-01 9.09135163e-01 -7.98048526e-02 -3.90169859e-01
-7.94797465e-02 -1.07325876e+00 -5.49435675e-01 -7.13989258e-01
-3.61305848e-02 3.64830941e-01 4.37772214e-01 -5.92642248e-01
6.37809992e-01 6.08609796e-01 -1.76668000e+00 2.86826640e-01
7.61646569e-01 8.93502533e-01 -5.39119661e-01 2.10894629e-01
1.14741705e-01 6.76206648e-01 -7.16048181e-01 -7.50491560e-01
-2.14493945e-02 -3.02552402e-01 -6.96691155e-01 -9.00777578e-02
2.30995178e-01 -1.58575699e-01 -2.45981500e-01 8.86268020e-01
9.10287052e-02 -2.01890305e-01 6.61727428e-01 -1.28081250e+00
-2.33731240e-01 6.61807716e-01 -2.02243075e-01 2.66983151e-01
4.22899902e-01 8.18048045e-02 8.63713384e-01 -7.45018840e-01
7.16173768e-01 1.08293462e+00 9.05169189e-01 -1.87409092e-02
-8.67892981e-01 -8.95300806e-01 -2.09878143e-02 1.04511261e-01
-1.26506364e+00 -4.03738171e-01 4.94886518e-01 -5.80361724e-01
4.78882611e-01 6.55486524e-01 7.09846795e-01 1.15344608e+00
6.60282671e-02 3.85954320e-01 1.40378332e+00 -3.98410290e-01
-2.60207299e-02 8.50372732e-01 3.63338768e-01 6.76422715e-01
9.26558733e-01 1.75593928e-01 -3.23430717e-01 -7.28649497e-01
1.28030954e-02 -1.72289878e-01 -1.27300441e-01 1.91584900e-01
-7.92311072e-01 1.20415282e+00 4.69295919e-01 7.06607461e-01
-1.24600641e-01 -1.42712936e-01 3.97920012e-01 3.22796613e-01
7.83866882e-01 6.38968587e-01 -1.66529864e-01 -2.28809025e-02
-9.35239375e-01 3.86088997e-01 9.26160753e-01 3.58557105e-01
9.15237308e-01 -2.65827268e-01 5.36150113e-03 3.76794845e-01
1.84645623e-01 5.57975769e-01 5.75094819e-01 -3.44615936e-01
2.69306988e-01 9.47285116e-01 4.46517378e-01 -1.67221200e+00
-5.52989542e-01 -5.23736537e-01 -4.78520811e-01 -1.48712620e-01
6.51404679e-01 -5.29888682e-02 -2.57782459e-01 1.17964661e+00
2.75227696e-01 -3.27410363e-02 -4.11704332e-01 5.37422895e-01
6.90706432e-01 3.59227777e-01 -9.75387692e-02 -2.69781053e-01
1.30859458e+00 -5.08776784e-01 -6.12672269e-01 9.46342200e-02
9.21200037e-01 -8.00148129e-01 7.88037062e-01 3.21475744e-01
-7.10112572e-01 -1.83243647e-01 -9.62911248e-01 4.13086474e-01
-1.01297534e+00 -4.82217520e-02 6.14459932e-01 1.12484562e+00
-6.39536619e-01 9.58277166e-01 -3.33733231e-01 -2.82596380e-01
4.14495498e-01 3.99978161e-01 -1.91199839e-01 3.43847245e-01
-1.48354411e+00 1.05849171e+00 1.17445856e-01 -4.24324907e-02
-2.73185104e-01 -3.34971458e-01 -3.59458476e-01 -9.91411582e-02
1.60091624e-01 -1.65052742e-01 8.35760415e-01 -9.45131898e-01
-1.05178320e+00 1.10708380e+00 1.27789071e-02 -5.83509326e-01
9.99568224e-01 2.11646572e-01 -6.03159666e-01 9.30801108e-02
2.09298372e-01 -8.68913010e-02 8.75816047e-01 -1.27463055e+00
-6.55621588e-01 -4.13600117e-01 -1.93205133e-01 -6.10064983e-01
-5.94287813e-01 2.98444685e-02 1.84841733e-02 -5.81598163e-01
3.61039974e-02 -1.07130075e+00 6.97105676e-02 -6.78525150e-01
-6.73876047e-01 -2.31818631e-01 9.22793686e-01 -4.08048630e-01
1.46891844e+00 -1.80113602e+00 -5.14440000e-01 6.50082886e-01
3.34531575e-01 5.83213866e-01 4.54691380e-01 6.94450796e-01
1.83283947e-02 4.98612285e-01 -1.03954762e-01 -3.30775082e-01
-1.33974209e-01 -1.44446343e-01 -4.12723720e-01 9.23970461e-01
1.01393305e-01 6.97694123e-01 -8.75175118e-01 -2.36570314e-01
1.82160318e-01 3.69681984e-01 -2.68737674e-01 -2.57758826e-01
2.79715508e-01 2.26692855e-01 -6.90653205e-01 5.33301055e-01
6.81253970e-01 -1.92312494e-01 5.92639968e-02 1.07866399e-01
-2.65972286e-01 4.65480924e-01 -9.95616615e-01 3.44855368e-01
-2.19773948e-01 5.61307013e-01 -2.77699471e-01 -8.93562138e-01
1.11339617e+00 5.38553372e-02 4.26259488e-01 -4.22880262e-01
5.78878045e-01 8.03548872e-01 -2.07619220e-01 -3.12296659e-01
6.38521016e-01 -6.05398640e-02 -2.06529796e-01 6.59697294e-01
-3.29070896e-01 -1.07287630e-01 8.58026743e-02 1.66601807e-01
9.84852314e-01 -3.91529322e-01 3.72734576e-01 -3.41974258e-01
9.69032407e-01 1.24893352e-01 -7.72360414e-02 9.08502221e-01
-3.76464933e-01 2.59020030e-01 6.60600603e-01 -6.18022621e-01
-9.14902449e-01 -5.49567401e-01 -3.27528030e-01 9.20039058e-01
5.49472198e-02 -2.54586428e-01 -7.09870458e-01 -8.25881124e-01
2.78617084e-01 6.07803762e-01 -7.09437072e-01 -2.59398848e-01
-5.16316056e-01 -1.26615584e+00 8.04939091e-01 -3.00216943e-01
4.64849144e-01 -7.01329887e-01 -3.31483454e-01 2.19397023e-01
-3.18924546e-01 -1.08664906e+00 2.01667696e-01 -1.29757479e-01
-7.23749816e-01 -1.51822495e+00 -2.50057071e-01 -1.40681431e-01
4.47184294e-01 3.43315095e-01 7.94747472e-01 6.68010056e-01
3.79476286e-02 2.81504411e-02 -5.13043106e-01 -5.87921858e-01
-9.34012115e-01 2.54547387e-01 1.36266336e-01 3.39821219e-01
4.90081787e-01 -4.55919564e-01 -2.57082969e-01 5.37896812e-01
-7.80133128e-01 -6.34231091e-01 3.22367668e-01 5.91197729e-01
-1.66191608e-01 6.24102424e-04 8.43295097e-01 -1.46452498e+00
6.35133445e-01 -7.99778402e-01 -6.21536374e-01 -2.59614766e-01
-9.35268998e-01 -3.07613939e-01 6.13451004e-01 -2.43972510e-01
-5.93321383e-01 -1.76768899e-01 -6.67135864e-02 1.94470659e-01
2.12154053e-02 2.72210985e-01 2.58732945e-01 -3.62393469e-01
1.01148808e+00 1.63612098e-01 1.66672766e-01 -4.44864959e-01
2.26198826e-02 1.01929367e+00 6.63851351e-02 -1.63709614e-02
1.06754041e+00 8.76981974e-01 7.02450275e-02 -8.59527349e-01
-7.66080737e-01 -6.45378590e-01 -3.68197709e-01 -2.46439874e-01
1.94562674e-01 -6.64576769e-01 -9.57183599e-01 6.49444640e-01
-9.49779868e-01 3.17430705e-01 -7.59415329e-03 2.64644831e-01
-2.17539757e-01 5.13564050e-01 -4.55241919e-01 -9.90204990e-01
-1.34576425e-01 -9.58369851e-01 7.10625529e-01 -8.74580070e-02
-3.76518935e-01 -9.78274882e-01 -7.90642649e-02 5.63786685e-01
5.79851985e-01 6.24660611e-01 5.98200858e-01 -1.13998628e+00
-2.66297728e-01 -9.31751728e-01 -2.57476509e-01 2.03975663e-01
2.70062797e-02 1.65091962e-01 -1.27404284e+00 -2.64037728e-01
1.49264380e-01 -9.58434194e-02 1.06471896e+00 -7.19275326e-02
8.17432344e-01 -5.50437450e-01 -6.34754777e-01 -7.96456113e-02
1.18733287e+00 -3.34590375e-01 5.17256021e-01 7.36456096e-01
2.92558581e-01 7.58731246e-01 5.25100768e-01 6.22219265e-01
2.47936189e-01 6.79010570e-01 6.26435220e-01 5.56713110e-03
2.22508699e-01 -2.24435881e-01 3.66173208e-01 4.64750707e-01
-2.50293463e-01 -1.41343907e-01 -7.57261276e-01 4.64573532e-01
-1.60115731e+00 -9.52149391e-01 -8.80880237e-01 2.28723574e+00
3.92353892e-01 3.09635013e-01 6.07935727e-01 5.82311451e-01
8.76755059e-01 8.99401680e-02 4.87138480e-02 -5.35146356e-01
-2.48790294e-01 1.56942606e-02 8.63400817e-01 6.77804828e-01
-1.19867516e+00 7.93187320e-01 6.16852999e+00 7.14767575e-01
-1.16072989e+00 1.16812430e-01 6.47036016e-01 2.16797933e-01
-2.45513648e-01 1.04952734e-02 -9.18335199e-01 5.83397865e-01
1.20054150e+00 -1.71235234e-01 2.30181471e-01 6.43927574e-01
7.30136514e-01 -3.56083810e-01 -5.12936294e-01 5.88019371e-01
3.76954563e-02 -1.42697120e+00 -4.15472798e-02 4.00683552e-01
6.21435702e-01 3.11045907e-02 2.46867654e-03 2.17893139e-01
-3.41017805e-02 -8.89959693e-01 7.09083498e-01 4.14261252e-01
2.19943836e-01 -6.43642128e-01 1.12579072e+00 5.48033118e-01
-4.49629128e-01 -2.71171749e-01 -7.36218989e-02 -4.70298886e-01
-8.34085718e-02 1.08946121e+00 -1.04734600e+00 5.60661972e-01
3.67880493e-01 5.96599162e-01 -9.44646776e-01 7.15660036e-01
-3.36365849e-02 8.36995959e-01 -4.81058657e-01 -4.50778186e-01
3.75463068e-01 -6.94866851e-03 5.86193919e-01 1.21822453e+00
1.64424181e-01 -2.67835081e-01 -1.69221207e-01 7.90816426e-01
-7.64571875e-02 3.52172196e-01 -6.93592370e-01 -8.69200304e-02
4.79760051e-01 1.21051133e+00 -7.18675852e-01 -4.81182128e-01
-1.97141901e-01 4.84866291e-01 1.16442971e-01 -1.10975966e-01
-8.45489740e-01 -2.23379180e-01 3.85252595e-01 8.28160644e-01
-2.15077475e-02 1.46757632e-01 -3.98301572e-01 -1.08836699e+00
-2.01382320e-02 -7.03481734e-01 2.45585978e-01 -6.40820488e-02
-1.33277321e+00 4.79077905e-01 -1.02487892e-01 -1.23900831e+00
-1.55010045e-01 -6.47514641e-01 -4.87237751e-01 6.50829196e-01
-1.56506836e+00 -1.04729378e+00 -2.47450322e-01 3.93361151e-01
-1.34435460e-01 -9.12693366e-02 7.50858843e-01 3.04465145e-01
-4.35698628e-01 5.45898676e-01 2.34487936e-01 7.64464736e-02
5.74570119e-01 -7.45418966e-01 1.91590831e-01 4.22058702e-01
-1.65300027e-01 6.16923869e-01 9.59735692e-01 -6.46060705e-01
-8.96878958e-01 -9.51970696e-01 1.45988274e+00 -8.16157341e-01
9.98605251e-01 -2.59755880e-01 -6.48003221e-01 4.16133225e-01
-3.93994272e-01 -1.01736553e-01 5.53757131e-01 7.45097026e-02
-3.34606946e-01 1.69015124e-01 -1.42516172e+00 1.82999790e-01
5.02494335e-01 -3.02514523e-01 -4.09748614e-01 6.22492850e-01
1.25988185e-01 1.34897698e-02 -6.61356390e-01 4.05420959e-02
5.87293565e-01 -1.55823088e+00 7.56564498e-01 -6.12289727e-01
3.29595923e-01 5.33458367e-02 8.80529955e-02 -9.52040017e-01
-6.59248829e-02 -5.89909315e-01 2.00948581e-01 1.19220293e+00
5.50581694e-01 -1.13555729e+00 8.60747874e-01 2.56564319e-01
2.36984342e-01 -6.44121647e-01 -1.05201292e+00 -7.88270235e-01
3.04979235e-01 -4.42918777e-01 5.06236255e-01 1.16908622e+00
1.32123336e-01 4.37301323e-02 -3.76317531e-01 -1.53183639e-01
4.05229479e-01 -6.04386814e-02 9.77721632e-01 -1.53552973e+00
3.55791152e-02 -5.63196301e-01 -4.55947906e-01 -2.69607186e-01
-5.50055094e-02 -8.03230643e-01 -5.01108825e-01 -7.90167928e-01
9.38194767e-02 -7.18126893e-01 9.16888863e-02 2.92793006e-01
-1.33995414e-02 7.05392599e-01 2.45008722e-01 6.58860326e-01
-1.69588014e-01 5.48671708e-02 1.01179338e+00 1.74167901e-02
-1.76378325e-01 5.11380613e-01 -8.01411450e-01 9.77531612e-01
7.65435100e-01 -6.96419656e-01 2.91504741e-01 4.12471026e-01
4.72270489e-01 -5.08956790e-01 6.26296759e-01 -7.90403366e-01
-2.32370779e-01 7.49590918e-02 1.57792896e-01 -2.99429119e-01
3.17808539e-01 -6.96030736e-01 -1.05691567e-01 7.94304550e-01
1.14893310e-01 -2.57021606e-01 -4.81203198e-02 6.10592782e-01
-1.64693713e-01 -3.82796019e-01 9.17172432e-01 -1.54099837e-01
2.18032058e-02 -1.42872646e-01 -6.84214473e-01 -1.19132541e-01
9.65256155e-01 -2.66191870e-01 -6.42466068e-01 -5.95933557e-01
-7.11177528e-01 -3.02752018e-01 5.13064563e-01 3.33232939e-01
-8.03099498e-02 -9.29928482e-01 -7.80895948e-01 4.84823287e-02
1.72161415e-01 -8.72070312e-01 -2.26310223e-01 1.12065804e+00
-4.56959307e-01 6.15883946e-01 2.90986113e-02 -3.76165092e-01
-1.08389783e+00 3.73581976e-01 2.90397435e-01 -3.98479372e-01
-1.47946894e-01 4.39373583e-01 -5.35113990e-01 -3.80475372e-01
-2.34046698e-01 -2.14426771e-01 -3.06461275e-01 6.12822115e-01
5.31311035e-01 7.84848273e-01 4.99049872e-01 -1.08413935e+00
-4.46948916e-01 1.18910149e-01 -4.03279625e-02 1.88487381e-01
1.25534368e+00 -1.73464984e-01 -2.94889092e-01 3.90885264e-01
1.27815759e+00 5.80267549e-01 -3.24584126e-01 7.38109052e-02
2.83505410e-01 -6.10483766e-01 -1.03300981e-01 -6.31131887e-01
-6.39422774e-01 4.71793592e-01 3.53495091e-01 9.96577263e-01
6.05199158e-01 -8.50942582e-02 5.93703568e-01 2.28999972e-01
3.84097636e-01 -1.01512003e+00 -1.33780450e-01 4.09152776e-01
5.71200788e-01 -1.35771692e+00 2.52889961e-01 -7.77986467e-01
-2.07725510e-01 1.19014418e+00 -3.12595405e-02 -2.90682167e-01
7.62195230e-01 -3.26892853e-01 2.90711857e-02 -4.17420447e-01
-1.63707599e-01 -3.77463475e-02 -4.13676277e-02 3.01257581e-01
3.62886161e-01 2.14241609e-01 -1.09048140e+00 5.76168120e-01
-4.84849274e-01 -6.07436411e-02 1.00920868e+00 6.38499439e-01
-7.14307845e-01 -1.15956175e+00 -7.35036671e-01 7.18290985e-01
-5.76687634e-01 2.02094331e-01 -8.70970547e-01 1.13711381e+00
3.68200898e-01 1.30081868e+00 -1.91647679e-01 -4.45527941e-01
2.94608742e-01 8.12431127e-02 -3.50257456e-02 -2.12210894e-01
-1.14497900e+00 -3.34305584e-01 4.54358995e-01 -3.78409654e-01
-3.82143319e-01 -8.58823538e-01 -8.02385092e-01 -9.25952196e-01
-8.18830788e-01 4.24992174e-01 7.43251681e-01 1.13849926e+00
1.37525275e-01 -4.22720946e-02 1.05702055e+00 -7.72374749e-01
-7.43382454e-01 -1.13294506e+00 -6.01430714e-01 6.61885858e-01
2.61549354e-01 -7.60036707e-01 -9.31784451e-01 -3.64064097e-01] | [8.026815414428711, 10.113916397094727] |
eb8cf580-476a-4981-9fa2-727fc9ad137f | don-t-stop-pretraining-make-prompt-based-fine | 2305.01711 | null | https://arxiv.org/abs/2305.01711v2 | https://arxiv.org/pdf/2305.01711v2.pdf | Don't Stop Pretraining? Make Prompt-based Fine-tuning Powerful Learner | Language models (LMs) trained on vast quantities of unlabelled data have greatly advanced the field of natural language processing (NLP). In this study, we re-visit the widely accepted notion in NLP that continued pre-training LMs on task-related texts improves the performance of fine-tuning (FT) in downstream tasks. Through experiments on eight single-sentence tasks and eight sentence-pair tasks in both semi-supervised and fully-supervised settings, we find that conventional continued pre-training does not consistently provide benefits and can even be detrimental for sentence-pair tasks or when prompt-based FT is used. To tackle these issues, we propose Prompt-based Continued Pre-training (PCP), which combines the idea of instruction tuning with conventional continued pre-training. Our approach aims to improve the performance of prompt-based FT by presenting both task-related texts and prompt templates to LMs through unsupervised pre-training objectives before fine-tuning for the target task. Our empirical evaluations on 21 benchmarks demonstrate that the PCP consistently improves the performance of state-of-the-art prompt-based FT approaches (up to 20.1% absolute) in both semi-supervised and fully-supervised settings, even with only hundreds of unlabelled examples. Additionally, prompt-based FT with the PCP outperforms state-of-the-art semi-supervised approaches with greater simplicity, eliminating the need for an iterative process and extra data augmentation. Our further analysis explores the performance lower bound of the PCP and reveals that the advantages of PCP persist across different sizes of models and datasets. | ['Aldo Lipani', 'Zhengxiang Shi'] | 2023-05-02 | null | null | null | null | ['unsupervised-pre-training'] | ['methodology'] | [ 5.85191309e-01 1.67288885e-01 -1.75317928e-01 -5.72354317e-01
-1.24870825e+00 -8.14602315e-01 9.51137424e-01 3.66080731e-01
-8.26357305e-01 7.28669465e-01 1.49018705e-01 -5.59987843e-01
-1.30597249e-01 -2.16644481e-01 -6.62072778e-01 -3.65741432e-01
1.52116016e-01 6.43729210e-01 3.96541446e-01 -4.03681666e-01
1.08742617e-01 2.87617221e-02 -1.31758249e+00 5.16006529e-01
9.97387826e-01 7.22051561e-01 4.76956427e-01 6.41080439e-01
-3.87549430e-01 5.75474977e-01 -5.78917325e-01 -3.80949497e-01
2.25810543e-01 -2.17257962e-01 -1.00324249e+00 -5.32951355e-02
5.65221429e-01 1.31459251e-01 -6.09517610e-03 6.76800370e-01
6.05433822e-01 3.45755249e-01 5.94310701e-01 -9.25219238e-01
-2.50282556e-01 9.17978764e-01 -3.96390766e-01 4.34253961e-01
1.26276359e-01 2.56339103e-01 1.03792405e+00 -1.03403509e+00
3.77383441e-01 1.33483410e+00 7.45812535e-01 4.71824676e-01
-1.46475768e+00 -4.88345563e-01 2.21704289e-01 -1.78480148e-01
-1.16823852e+00 -6.98664248e-01 3.92096162e-01 -4.93800998e-01
1.30679238e+00 5.62821329e-02 -1.85277432e-01 1.06587219e+00
7.01248599e-03 9.03030336e-01 1.21162415e+00 -8.41576993e-01
-1.00157468e-03 2.60885358e-01 3.64938289e-01 3.46298128e-01
-9.48290434e-03 1.23386271e-01 -6.61870837e-01 -7.19768256e-02
3.21691751e-01 -3.08201224e-01 -1.59034535e-01 5.22032119e-02
-1.46554613e+00 6.83882356e-01 1.15989642e-02 6.71536863e-01
-2.02560425e-01 -1.31724566e-01 6.24611318e-01 5.80754519e-01
7.75363863e-01 7.85491943e-01 -1.09062028e+00 -3.53173733e-01
-1.34516525e+00 6.77732602e-02 8.84953082e-01 8.95315051e-01
7.08931029e-01 -1.55084491e-01 -6.68260276e-01 9.14950132e-01
2.11448204e-02 3.40218872e-01 7.76539683e-01 -5.99986792e-01
1.16201115e+00 6.27235770e-01 5.46620740e-03 -3.96680802e-01
-4.71416295e-01 -6.22538745e-01 -7.31366158e-01 -4.22437400e-01
6.79612100e-01 -3.81723017e-01 -8.26806128e-01 1.83295524e+00
8.92203525e-02 3.56044322e-02 1.61872685e-01 3.65368128e-01
6.92394853e-01 5.35732448e-01 3.53471577e-01 -3.97454828e-01
1.13939273e+00 -1.25552285e+00 -6.66026115e-01 -5.89165866e-01
1.28733349e+00 -8.09902251e-01 1.76842260e+00 3.69461209e-01
-7.93446183e-01 -8.48100066e-01 -7.60513425e-01 -4.16977815e-02
-3.83942962e-01 4.07025784e-01 3.04592013e-01 5.87649167e-01
-9.02274609e-01 7.29912221e-01 -6.73610568e-01 -3.75826776e-01
8.47706273e-02 4.23441201e-01 -3.37393522e-01 -4.48561274e-02
-1.38294804e+00 8.52772057e-01 5.94303548e-01 -4.53831516e-02
-6.78559542e-01 -1.03358626e+00 -7.44416595e-01 2.19387352e-01
6.93945050e-01 -5.34877479e-01 1.69190490e+00 -9.32747662e-01
-1.53146255e+00 7.04838336e-01 -3.90683204e-01 -5.22035420e-01
7.09923029e-01 -6.35099411e-01 -2.20585778e-01 -9.54075605e-02
2.13642642e-01 6.75437033e-01 7.12240875e-01 -9.84639525e-01
-6.61751568e-01 -3.03957146e-02 1.03867397e-01 3.40483487e-01
-5.81741989e-01 7.71920830e-02 -2.18293965e-01 -5.85269749e-01
-1.89750984e-01 -8.12949955e-01 -2.76429534e-01 -7.16781318e-01
-4.87618804e-01 -5.38257957e-01 6.18475616e-01 -4.16727096e-01
1.64214373e+00 -2.13122010e+00 -1.70684934e-01 -1.61925003e-01
3.09322812e-02 6.14582717e-01 -5.31501174e-01 6.71813667e-01
-6.71259984e-02 3.26777607e-01 -3.08586508e-01 -7.72186160e-01
4.41502593e-02 1.80783898e-01 -3.24842453e-01 -8.17858651e-02
3.60180765e-01 1.03264141e+00 -1.08327878e+00 -4.98876095e-01
-4.79247421e-03 1.05145752e-01 -1.86375901e-01 1.75274059e-01
-5.03293693e-01 5.98944902e-01 -2.18924373e-01 1.31855786e-01
2.15585753e-01 -4.31667089e-01 1.50220931e-01 1.58952013e-01
-1.18273068e-02 8.46672833e-01 -1.00810170e+00 1.68905222e+00
-6.47009075e-01 6.26833200e-01 -2.09511936e-01 -9.13223982e-01
7.79845238e-01 5.17108619e-01 9.54668969e-02 -7.54872084e-01
-7.19202682e-02 2.56031275e-01 3.08763981e-02 -5.55125237e-01
5.47555506e-01 -1.94647297e-01 -4.34896685e-02 6.48430586e-01
2.05825418e-01 -1.74636226e-02 6.39680564e-01 2.88975775e-01
1.13941050e+00 3.03523000e-02 2.40658581e-01 -3.22811902e-01
8.17982376e-01 4.40037958e-02 2.89171278e-01 1.04294109e+00
-5.85672036e-02 3.09145778e-01 3.14200133e-01 3.73360217e-02
-6.88012660e-01 -7.56527126e-01 -7.47518195e-03 1.82157147e+00
-4.73013937e-01 -6.55937970e-01 -7.21345842e-01 -1.13833928e+00
-4.40578051e-02 9.07678068e-01 -4.09978002e-01 -1.87401585e-02
-5.43120861e-01 -6.46063447e-01 7.03417838e-01 5.62614799e-01
3.91547590e-01 -1.14094961e+00 -2.08508506e-01 4.12780851e-01
-4.45622772e-01 -1.34199059e+00 -5.16930878e-01 5.95969975e-01
-9.93238986e-01 -7.87047148e-01 -6.28507316e-01 -7.53859162e-01
7.62705922e-01 3.42751145e-01 1.28468287e+00 -1.06929332e-01
2.91702718e-01 3.30380470e-01 -4.63267565e-01 -4.21412140e-01
-7.50562549e-01 5.91968834e-01 1.81657568e-01 -7.20741525e-02
4.89674121e-01 -3.33553493e-01 -1.27967447e-01 3.27175647e-01
-9.02884603e-01 1.14185899e-01 8.95241797e-01 9.52199459e-01
4.64600384e-01 4.23496887e-02 9.11083817e-01 -1.31994593e+00
9.26138580e-01 -1.76314265e-01 -3.14925283e-01 5.34266174e-01
-7.86854088e-01 3.33424121e-01 7.07295775e-01 -7.91924238e-01
-1.13953626e+00 -1.35994954e-02 -4.93860990e-03 -1.66386396e-01
-2.90924013e-01 8.43961596e-01 3.02633569e-02 1.12694584e-01
1.01750195e+00 1.82828188e-01 -3.91445905e-01 -7.24869192e-01
4.43293035e-01 7.80326426e-01 4.05929804e-01 -6.81972086e-01
8.15764129e-01 1.03634596e-02 -2.28963524e-01 -6.82206035e-01
-1.47777605e+00 -8.38518500e-01 -1.06371439e+00 1.31703332e-01
3.59762907e-01 -8.09608698e-01 -2.81089723e-01 1.39920443e-01
-1.06097698e+00 -8.72503340e-01 -3.38935822e-01 3.72168899e-01
-4.54915106e-01 3.95638198e-01 -5.35836756e-01 -7.11904645e-01
-4.84296769e-01 -9.17121112e-01 1.10062039e+00 -1.15151271e-01
-4.42846268e-01 -1.28282833e+00 1.19236067e-01 5.08460402e-01
4.24399674e-01 -3.10580045e-01 7.95131624e-01 -1.18656170e+00
3.42297330e-02 -1.21664844e-01 -7.23534450e-02 5.90733469e-01
1.62390515e-01 -3.22728664e-01 -1.13782144e+00 -3.17318797e-01
-2.44541205e-02 -5.62563419e-01 8.56436849e-01 1.77871376e-01
7.12988079e-01 -2.95202583e-01 -2.70664603e-01 1.70532674e-01
1.23386133e+00 -2.63345242e-01 5.92411309e-02 4.85452741e-01
5.29192924e-01 8.34109724e-01 1.07628667e+00 2.02250674e-01
2.71345824e-01 4.77111369e-01 -1.75107747e-01 -7.69472569e-02
4.48275357e-03 -4.61995661e-01 5.93123257e-01 7.99364388e-01
2.50037581e-01 -3.39685947e-01 -1.18785310e+00 5.20099878e-01
-1.81548893e+00 -5.94761431e-01 -2.97965437e-01 2.27478504e+00
1.38879704e+00 5.18818915e-01 6.35886118e-02 3.74608219e-01
5.30590951e-01 -4.23871949e-02 -2.18609348e-01 -4.42818254e-01
-3.75016928e-02 2.08774671e-01 3.66419792e-01 5.39743662e-01
-1.19740295e+00 1.29525268e+00 6.15446711e+00 1.14876270e+00
-1.10763073e+00 3.66388589e-01 5.25554895e-01 -1.06133819e-01
-8.52293298e-02 1.28504699e-02 -1.34028423e+00 2.54412413e-01
1.38229132e+00 -3.43294889e-01 1.41029730e-01 6.74646974e-01
5.13618946e-01 -8.43659714e-02 -1.54109907e+00 6.29100800e-01
-1.40532732e-01 -1.03904939e+00 -1.35965243e-01 -2.23577753e-01
8.73269856e-01 4.04225498e-01 -1.62579030e-01 7.90843189e-01
1.32333711e-01 -8.22154522e-01 6.82193637e-01 7.45632723e-02
8.34399104e-01 -3.09550196e-01 8.02972615e-01 1.09866536e+00
-9.80525374e-01 -1.06403865e-01 -1.92489833e-01 -2.67085075e-01
1.95245400e-01 7.06367016e-01 -1.22002947e+00 6.17250741e-01
3.93404752e-01 5.11155367e-01 -8.63731503e-01 6.89666450e-01
-5.30150235e-01 1.14801550e+00 -3.39785218e-01 -2.14603846e-03
4.41442490e-01 2.42102444e-01 3.96538168e-01 1.75220263e+00
-1.87433496e-01 -1.35905609e-01 2.82879084e-01 5.59841812e-01
-4.72865701e-02 3.15260202e-01 -3.45241427e-01 -2.97042191e-01
5.12718856e-01 1.22577500e+00 -5.38338900e-01 -6.04078412e-01
-3.48188430e-01 6.27007365e-01 4.47284490e-01 4.52256054e-01
-3.58831257e-01 -2.96808869e-01 -4.79444079e-02 2.68004924e-01
2.57385701e-01 -2.98572481e-01 -3.29385400e-01 -8.86143148e-01
3.56822349e-02 -9.32195127e-01 3.46552879e-01 -4.05333370e-01
-1.41518557e+00 6.35538936e-01 8.38546902e-02 -1.08915389e+00
-3.96078825e-01 -4.04591084e-01 -5.01096904e-01 7.92921245e-01
-1.76577735e+00 -1.00885940e+00 3.87259871e-02 4.69701052e-01
8.43031824e-01 4.56785522e-02 7.63165414e-01 2.08645895e-01
-5.17155230e-01 7.40700722e-01 2.97157079e-01 5.37637733e-02
1.22378516e+00 -1.37626243e+00 4.54989851e-01 8.72122109e-01
3.11613262e-01 7.44150221e-01 6.42291605e-01 -6.40869737e-01
-9.37472880e-01 -1.19633973e+00 1.62626207e+00 -6.31964564e-01
8.17528009e-01 -5.87548137e-01 -8.97485673e-01 7.02954412e-01
6.63939863e-02 -1.99486777e-01 6.75830722e-01 4.58067209e-01
-2.04840764e-01 -5.60828187e-02 -6.81628525e-01 3.92432898e-01
8.67195606e-01 -6.22575521e-01 -9.66500640e-01 6.75866544e-01
1.04819489e+00 -3.72366041e-01 -6.98658288e-01 4.40847605e-01
5.29150628e-02 -6.17040873e-01 6.36449099e-01 -5.94172776e-01
3.29794139e-01 7.18693584e-02 9.69907567e-02 -1.28353560e+00
-1.96562588e-01 -9.04938638e-01 3.48749124e-02 1.46526718e+00
8.21968138e-01 -5.78236163e-01 6.10104561e-01 5.01340926e-01
-2.75224715e-01 -7.79309690e-01 -6.26996338e-01 -1.13831699e+00
2.03416511e-01 -5.94741166e-01 1.06526315e-01 9.34478939e-01
2.50945240e-02 8.86322200e-01 1.27257910e-02 -1.16826959e-01
2.88015425e-01 -1.81308866e-01 8.61038327e-01 -1.30098903e+00
-4.67567176e-01 -2.98158973e-01 3.13565999e-01 -1.30766642e+00
3.53788465e-01 -1.03015292e+00 2.71454662e-01 -1.42953253e+00
9.40406173e-02 -5.01900554e-01 -2.95019627e-01 9.48709309e-01
-5.43600082e-01 -8.89958441e-02 2.62740523e-01 2.42162272e-01
-7.87384510e-01 4.62063760e-01 1.21881604e+00 -6.87399209e-02
-5.90863168e-01 1.47054270e-01 -6.91977561e-01 5.52355528e-01
6.81358755e-01 -8.01098824e-01 -5.69755733e-01 -5.80550432e-01
2.65459001e-01 -1.75367340e-01 5.27760014e-02 -7.58921444e-01
3.91206712e-01 1.41227553e-02 -8.56039748e-02 -5.03970265e-01
1.40110040e-02 -5.74091017e-01 -4.59407538e-01 3.51111472e-01
-8.00199568e-01 -9.20756459e-02 4.16142315e-01 5.71895659e-01
-3.03406954e-01 -4.46798295e-01 4.81695384e-01 -3.62542644e-02
-5.03173470e-01 4.71264776e-03 -2.36148179e-01 5.21297812e-01
6.49880111e-01 -7.70046860e-02 -2.97013253e-01 -1.63479969e-01
-6.87592447e-01 5.11419535e-01 -1.14760242e-01 4.71850276e-01
1.32571220e-01 -7.96343446e-01 -8.83100450e-01 1.24095619e-01
1.85827598e-01 1.46537811e-01 1.36382664e-02 1.20307016e+00
1.14944987e-01 9.61802959e-01 4.46544528e-01 -8.12786102e-01
-1.28235149e+00 3.69096279e-01 1.11830290e-02 -9.80932176e-01
-3.42582643e-01 9.67598617e-01 2.83339322e-01 -8.01695645e-01
4.24660027e-01 -4.76031661e-01 -1.81633055e-01 1.93133745e-02
3.54373306e-01 2.20002308e-01 4.35149461e-01 -2.82306671e-01
-1.10409334e-01 2.16553599e-01 -4.13413733e-01 -3.34755242e-01
1.19659650e+00 -2.28308469e-01 1.93815380e-01 7.95904577e-01
9.81227458e-01 -4.62688878e-02 -1.13993359e+00 -8.22385311e-01
5.48806489e-01 -1.29409745e-01 8.13344270e-02 -1.27982545e+00
-4.74052638e-01 1.07498145e+00 9.42410156e-02 9.46514979e-02
9.72145915e-01 -7.80398026e-02 6.07189715e-01 7.92668939e-01
2.55170554e-01 -1.09809744e+00 1.93229452e-01 9.00530815e-01
7.99569607e-01 -1.32661343e+00 -2.29732409e-01 -4.72585887e-01
-8.06995749e-01 9.31549370e-01 6.22124672e-01 2.94727802e-01
5.17605245e-01 2.07640260e-01 1.98763445e-01 1.36058897e-01
-1.14814782e+00 -1.94701478e-01 5.07099569e-01 2.84485668e-01
7.75417626e-01 -1.17419235e-01 -3.41169834e-01 9.03398454e-01
-3.11398983e-01 1.05543345e-01 6.58460557e-02 1.02941287e+00
-5.09236217e-01 -1.37580073e+00 -1.94574550e-01 5.84457159e-01
-4.94057357e-01 -5.69193661e-01 -4.77439702e-01 7.50973105e-01
-3.65578495e-02 1.24445355e+00 -2.63680428e-01 -2.10532054e-01
3.68431509e-01 4.55802083e-01 2.73702234e-01 -1.15176368e+00
-1.06597984e+00 2.05407411e-01 3.11446577e-01 -3.95468950e-01
-4.38940018e-01 -6.12327576e-01 -1.18585384e+00 4.34368737e-02
-6.97236896e-01 3.36983234e-01 5.30282319e-01 1.32776153e+00
3.59270990e-01 5.12059808e-01 4.63452160e-01 -7.19387233e-01
-1.18793190e+00 -1.48008788e+00 -2.10468739e-01 2.78848141e-01
3.17916691e-01 -4.58390057e-01 -6.66587591e-01 2.70896375e-01] | [10.729679107666016, 8.490121841430664] |
ca7aacbe-22d9-41ea-a6de-ae8d482d8355 | a-probabilistic-deep-learning-approach-to | 2103.16664 | null | https://arxiv.org/abs/2103.16664v1 | https://arxiv.org/pdf/2103.16664v1.pdf | A probabilistic deep learning approach to automate the interpretation of multi-phase diffraction spectra | Autonomous synthesis and characterization of inorganic materials requires the automatic and accurate analysis of X-ray diffraction spectra. For this task, we designed a probabilistic deep learning algorithm to identify complex multi-phase mixtures. At the core of this algorithm lies an ensemble convolutional neural network trained on simulated diffraction spectra, which are systematically augmented with physics-informed perturbations to account for artifacts that can arise during experimental sample preparation and synthesis. Larger perturbations associated with off-stoichiometry are also captured by supplementing the training set with hypothetical solid solutions. Spectra containing mixtures of materials are analyzed with a newly developed branching algorithm that utilizes the probabilistic nature of the neural network to explore suspected mixtures and identify the set of phases that maximize confidence in the prediction. Our model is benchmarked on simulated and experimentally measured diffraction spectra, showing exceptional performance with accuracies exceeding those given by previously reported methods based on profile matching and deep learning. We envision that the algorithm presented here may be integrated in experimental workflows to facilitate the high-throughput and autonomous discovery of inorganic materials. | ['Gerbrand Ceder', 'Qingsong Tu', 'Yan Zeng', 'Christopher J. Bartel', 'Nathan J. Szymanski'] | 2021-03-30 | null | null | null | null | ['probabilistic-deep-learning'] | ['computer-vision'] | [ 5.69822669e-01 -5.36602885e-02 2.89123148e-01 -4.37114835e-01
-8.14084411e-01 -3.78361732e-01 6.53126657e-01 5.13226867e-01
-4.35677767e-01 8.69329810e-01 -1.21476434e-01 -3.99871379e-01
-1.74187720e-01 -6.85507357e-01 -8.75505745e-01 -9.94848371e-01
1.70276657e-01 1.40490711e+00 1.01739913e-01 4.61188629e-02
1.96491525e-01 9.31820631e-01 -1.64568245e+00 5.72721779e-01
7.44462848e-01 1.01418471e+00 5.71244776e-01 4.30777341e-01
-3.41177315e-01 4.06947434e-01 -2.44699344e-01 -6.82652742e-02
1.31051525e-01 -7.14437515e-02 -4.82909828e-01 1.17404938e-01
2.87590865e-02 -1.59448072e-01 -1.63584590e-01 8.94579947e-01
5.46922147e-01 -5.99207766e-02 8.51669550e-01 -6.95486307e-01
-7.16431290e-02 8.53437126e-01 -1.79223955e-01 5.80546223e-02
3.48198563e-01 7.58024991e-01 9.41119790e-01 -8.84247720e-01
7.12393522e-01 8.92823637e-01 6.39672637e-01 1.97842926e-01
-1.57319999e+00 -5.65278888e-01 -1.43220201e-01 2.06811532e-01
-8.38553905e-01 -5.87210655e-01 8.40937972e-01 -6.84211314e-01
1.15864646e+00 -4.04553115e-02 5.08879900e-01 1.14071548e+00
1.17399208e-01 5.13758421e-01 9.84682798e-01 -3.09550166e-01
6.48697197e-01 -1.51582807e-01 -1.86416775e-01 5.31728387e-01
2.76534796e-01 3.16011101e-01 -3.29654038e-01 -2.20247075e-01
1.15468912e-01 -1.42240658e-01 -9.94586274e-02 -6.36796772e-01
-1.07171679e+00 4.90884781e-01 2.17565671e-01 9.52357426e-02
-9.04506564e-01 -1.81637466e-01 3.11646432e-01 -1.89008400e-01
2.53686339e-01 9.58191991e-01 -7.22715318e-01 -7.01087108e-03
-1.22566187e+00 4.55150276e-01 8.13739061e-01 6.11673832e-01
9.19159830e-01 -1.30362555e-01 -1.50242392e-02 5.71308613e-01
5.37363052e-01 2.03559682e-01 2.82649666e-01 -9.06588078e-01
1.67535804e-02 6.80406570e-01 3.78116488e-01 1.93786900e-02
-7.44846761e-01 -3.42804253e-01 -5.79175472e-01 3.82233530e-01
6.01056278e-01 6.25588819e-02 -1.18129909e+00 1.37884641e+00
4.61258024e-01 -2.05457821e-01 4.75317873e-02 6.34705663e-01
6.99932694e-01 3.33162755e-01 1.71665400e-01 -3.91005516e-01
1.02539051e+00 -5.42400956e-01 -4.07589316e-01 8.71021375e-02
3.67099971e-01 -8.35797489e-01 5.52141249e-01 6.71214640e-01
-1.10086858e+00 -2.97344476e-01 -1.34497488e+00 2.22287148e-01
-3.94970894e-01 4.92669791e-02 6.75644279e-01 6.05206966e-01
-4.73448277e-01 1.39331102e+00 -1.01678753e+00 -6.87418282e-02
5.96397281e-01 8.72631371e-01 -2.23073140e-01 2.73513366e-02
-8.44387233e-01 6.52677357e-01 6.36142373e-01 1.33636773e-01
-8.45287442e-01 -1.15508854e+00 -5.90381920e-01 -7.49979988e-02
1.75985143e-01 -5.92161775e-01 1.30314624e+00 -6.45561159e-01
-1.68671596e+00 7.36801803e-01 -1.48497537e-01 -4.93836313e-01
7.02608466e-01 1.36464546e-02 -2.81219333e-01 7.74344802e-02
5.77524491e-02 4.49669003e-01 4.87568349e-01 -1.25519478e+00
-3.14174920e-01 -2.41872251e-01 -4.80540454e-01 -3.12877387e-01
4.13220584e-01 -1.07834160e-01 -1.93284843e-02 -1.00109771e-01
3.65301490e-01 -6.78450406e-01 -5.51842630e-01 -2.02205494e-01
-5.41203380e-01 4.40566428e-02 6.60680175e-01 -4.54809844e-01
3.77197027e-01 -1.74435806e+00 2.27066234e-01 5.02763093e-01
4.68784511e-01 2.29777217e-01 9.19704363e-02 5.94651997e-01
-3.72678638e-01 -3.68628621e-01 -4.35055703e-01 -1.37051538e-01
1.90074906e-01 -1.56660601e-01 -2.94960607e-02 4.80841637e-01
4.06766385e-01 5.86143613e-01 -8.77344072e-01 3.15783881e-02
5.70535839e-01 9.11539569e-02 -5.37524045e-01 4.93857086e-01
-1.17168450e+00 7.53370106e-01 -1.97193876e-01 6.42804027e-01
9.27239180e-01 -5.74248075e-01 4.58072543e-01 -5.02524614e-01
-4.84249592e-01 5.94780147e-01 -1.11560214e+00 1.44980836e+00
-7.40709007e-02 8.06832239e-02 1.93121910e-01 -8.64862382e-01
8.27402055e-01 1.38628274e-01 9.95666325e-01 -5.46509326e-01
2.37377524e-01 5.26859224e-01 4.20393080e-01 -5.82406640e-01
3.03787977e-01 -4.24803108e-01 3.83531809e-01 7.75937378e-01
3.66146863e-01 -2.49805465e-01 3.14433515e-01 3.42409201e-02
1.07213032e+00 5.09705782e-01 1.07712008e-01 -4.01271790e-01
6.10483646e-01 -4.85377237e-02 2.63071448e-01 7.63330221e-01
-4.19419259e-02 5.14930785e-01 3.33295465e-01 -6.52889848e-01
-1.72459209e+00 -1.11283684e+00 -2.52376080e-01 8.60172510e-01
-7.82662407e-02 -3.96681398e-01 -5.05617023e-01 -2.25350633e-01
1.56154156e-01 6.13879383e-01 -3.67847383e-01 -2.17068717e-01
-2.94575870e-01 -1.23311472e+00 1.03058122e-01 3.61477822e-01
-2.54349783e-02 -1.24259067e+00 -2.41389498e-01 8.76278758e-01
3.02870601e-01 -1.11243832e+00 3.78734112e-01 9.84356284e-01
-2.95922220e-01 -1.38858831e+00 -2.61787325e-01 -2.19683304e-01
2.97944695e-01 -3.93910915e-01 9.53106225e-01 -4.06758338e-01
-7.68197894e-01 1.57952443e-01 8.06126520e-02 -4.34027374e-01
-1.15023458e+00 6.09801970e-02 3.74104649e-01 -8.44922587e-02
5.54180026e-01 -9.52312350e-01 -7.77508974e-01 -8.25723186e-02
-7.82149613e-01 1.22333609e-01 6.98940396e-01 7.56401479e-01
6.65397286e-01 1.08519033e-01 3.76923054e-01 -8.73255849e-01
3.75634640e-01 -6.13880396e-01 -9.08053398e-01 1.70555457e-01
-6.66387081e-01 6.66523457e-01 8.77689004e-01 -3.25286090e-01
-1.10692310e+00 3.74745637e-01 -5.79223514e-01 -3.46418321e-01
-6.54204369e-01 3.59574497e-01 -5.17636657e-01 4.45780270e-02
6.03655219e-01 3.10557723e-01 6.38912572e-03 -5.30349195e-01
1.46261826e-01 5.69596887e-01 4.62892652e-01 -1.03438258e+00
3.81976545e-01 3.77655596e-01 4.19057012e-01 -9.65142488e-01
-6.86268330e-01 -3.19435447e-01 -6.67395532e-01 -2.47932449e-01
7.17786193e-01 -6.06807113e-01 -1.05604231e+00 6.48086905e-01
-8.79545927e-01 -3.58321011e-01 -8.03743750e-02 4.86056149e-01
-8.24618101e-01 5.47882736e-01 -5.86060345e-01 -8.07622731e-01
-4.22625124e-01 -1.69966137e+00 1.15787983e+00 5.14848493e-02
-4.26713288e-01 -6.18704140e-01 9.60921645e-02 3.68446618e-01
2.43608713e-01 1.11713596e-01 1.23590887e+00 -1.13794541e+00
-8.42505395e-01 3.11075635e-02 -5.59528405e-03 -7.06741437e-02
3.68022412e-01 4.79805410e-01 -1.07688153e+00 -6.48968592e-02
-6.32692054e-02 -4.34571266e-01 9.61658180e-01 4.89733815e-01
1.26814318e+00 1.81693688e-01 -4.84269023e-01 6.97236657e-01
1.19522929e+00 6.24601729e-02 4.60812658e-01 6.28798664e-01
5.90095043e-01 5.00477433e-01 2.03950569e-01 7.16491103e-01
-1.84122846e-01 3.41394812e-01 6.10052049e-01 3.48852366e-01
1.83230549e-01 -1.52693242e-02 6.50411993e-02 1.98203161e-01
5.70819639e-02 -1.66977912e-01 -9.76832271e-01 -3.99024747e-02
-1.61606777e+00 -9.68949437e-01 -1.57513648e-01 2.10330176e+00
8.96503925e-01 5.98532140e-01 1.61959380e-01 2.30582595e-01
6.69491470e-01 -1.35068908e-01 -9.99409139e-01 2.74226014e-02
-1.90301791e-01 7.01943696e-01 4.02133346e-01 4.32424098e-01
-1.17911184e+00 7.48143375e-01 6.88396454e+00 3.61234546e-01
-1.34179616e+00 -4.33223784e-01 5.42590499e-01 -1.37661263e-01
-3.65592629e-01 -1.80747613e-01 -7.20938563e-01 5.46733081e-01
1.05866468e+00 6.78855181e-02 4.44170147e-01 5.71884990e-01
1.31805331e-01 4.96275537e-03 -1.41025591e+00 6.37546837e-01
-6.24192894e-01 -1.85253370e+00 -1.71220452e-01 6.45682588e-02
4.39029932e-01 4.11957562e-01 5.43168448e-02 -8.08361247e-02
6.36761069e-01 -9.16247427e-01 5.98855555e-01 6.15357399e-01
6.01056635e-01 -5.52055717e-01 3.42377454e-01 1.05528675e-01
-6.05139911e-01 -4.29632924e-02 -1.52117983e-01 9.23463479e-02
1.11874171e-01 8.77800882e-01 -1.41066706e+00 5.19832790e-01
3.91058683e-01 5.54664731e-01 -1.28961235e-01 9.40038741e-01
1.26900569e-01 3.74510527e-01 -7.87630856e-01 -4.56929058e-02
2.44777620e-01 -4.88523096e-01 4.64005470e-01 9.69612777e-01
2.87871361e-01 -2.69673347e-01 6.26162514e-02 1.38418829e+00
-1.20518751e-01 -2.23421931e-01 -2.61615664e-01 -3.65753710e-01
4.33210552e-01 1.18035436e+00 -7.99835503e-01 -1.68777734e-01
-2.23456964e-01 4.95911092e-01 3.34372491e-01 1.72387913e-01
-4.53730941e-01 3.87380600e-01 7.15272665e-01 3.28936279e-01
7.37688065e-01 -3.35631929e-02 -2.02006772e-01 -8.00219297e-01
-1.64245799e-01 -7.13797629e-01 9.61179659e-02 -6.91975474e-01
-1.40535402e+00 2.00289071e-01 -3.22193652e-01 -5.86310267e-01
-2.21420661e-01 -1.16170156e+00 -5.33817112e-01 8.53487194e-01
-1.21496248e+00 -7.44260192e-01 -2.99031362e-02 8.32993463e-02
2.01890841e-01 -2.81660885e-01 9.12035108e-01 1.33711651e-01
-5.95229447e-01 1.65107951e-01 6.31901324e-01 -5.57977021e-01
6.11001730e-01 -1.48265457e+00 2.74086952e-01 4.59526241e-01
-2.93415546e-01 3.32797885e-01 1.39766037e+00 -8.21554840e-01
-1.32845163e+00 -1.19089139e+00 2.08653212e-01 -1.65048987e-01
9.31167781e-01 -3.43216211e-01 -1.01658130e+00 4.35091317e-01
-1.36962458e-01 -2.03746006e-01 1.04557037e+00 -7.87039101e-02
-3.11554462e-01 3.98612581e-02 -1.25354230e+00 5.14006913e-01
7.05946207e-01 -3.04839224e-01 -3.50520104e-01 7.59889126e-01
4.92535979e-01 -4.56849039e-01 -1.04359567e+00 5.06504655e-01
5.58957994e-01 -1.14397192e+00 1.01313984e+00 -8.03420365e-01
5.65943658e-01 -1.16602778e-01 -2.69399971e-01 -9.25157964e-01
-3.69402200e-01 -6.14929080e-01 -2.10378334e-01 7.19489515e-01
5.11317074e-01 -3.41748744e-01 1.12145853e+00 8.44891191e-01
-2.41112426e-01 -5.83070278e-01 -8.58597755e-01 -4.15010512e-01
-1.15689682e-02 -3.35791111e-01 8.37356925e-01 4.10071492e-01
-2.61114448e-01 2.09604532e-01 5.68142459e-02 3.92983943e-01
8.99570584e-01 2.96141833e-01 4.57914770e-01 -1.57420671e+00
-4.94906604e-01 -5.57080507e-01 -2.97596037e-01 -5.98034203e-01
2.99548119e-01 -7.60010540e-01 1.10382922e-01 -1.07249534e+00
5.76337986e-02 -5.20541608e-01 -4.53686208e-01 -2.14066287e-03
-3.95521335e-02 -8.83209556e-02 -3.00006688e-01 2.27200627e-01
-6.35832667e-01 7.25398123e-01 7.88758576e-01 -3.37878048e-01
-1.15428962e-01 1.69045851e-01 -4.74552512e-01 5.18867910e-01
7.11642265e-01 -3.61416221e-01 1.18078768e-01 3.45003486e-01
3.66177380e-01 -4.15662490e-02 1.43795699e-01 -1.30927396e+00
5.69537915e-02 2.11853590e-02 6.13978326e-01 -9.73746896e-01
4.67724979e-01 -8.13198924e-01 6.05382144e-01 6.23926520e-01
-3.05297464e-01 -5.83766401e-01 1.46185651e-01 5.05989492e-01
2.48553663e-01 -2.84755111e-01 9.12049592e-01 -3.28886747e-01
-4.03265864e-01 6.02970302e-01 -6.43587530e-01 -5.77523470e-01
8.96662056e-01 3.24123278e-02 -3.10474411e-02 -1.70051008e-02
-1.20414448e+00 -7.36227855e-02 6.85307264e-01 -1.55784190e-01
3.26668978e-01 -9.16320801e-01 -4.64773178e-01 4.47467297e-01
1.59890398e-01 3.14008206e-01 2.44391903e-01 4.87872630e-01
-8.59442353e-01 2.65973449e-01 -2.39876434e-01 -7.90643275e-01
-8.99622142e-01 8.24806094e-01 7.63826549e-01 -4.21912551e-01
-3.74515653e-01 5.69011986e-01 -4.78422046e-02 -6.81999564e-01
-3.21044810e-02 -1.98873132e-01 9.94282961e-02 -2.44962201e-01
2.58540452e-01 7.69521669e-02 4.99406368e-01 -2.60082632e-01
-3.39450389e-01 -1.82761222e-01 -4.47179794e-01 1.64651975e-01
1.99336219e+00 2.82418877e-01 -2.94592828e-01 3.02768379e-01
8.80954921e-01 -1.25536859e-01 -1.70506203e+00 -3.78144056e-01
2.08043322e-01 2.25915954e-01 -2.23334059e-01 -7.41322875e-01
-6.50280893e-01 5.64670563e-01 5.29176772e-01 -8.70358385e-03
7.52190351e-01 1.49302080e-01 3.82078797e-01 6.25640869e-01
2.82418393e-02 -1.01198292e+00 -1.03357352e-01 4.14054781e-01
5.47203958e-01 -1.34665298e+00 2.56215334e-01 -1.18577704e-01
1.61756977e-01 1.54857481e+00 4.78716850e-01 -1.63873695e-02
6.07094288e-01 6.04013503e-01 -1.69225037e-01 -4.84391391e-01
-8.26500297e-01 -2.88819782e-02 -1.39817730e-01 7.15824842e-01
3.68114859e-01 -1.73514914e-02 2.54060060e-01 6.34877145e-01
-1.83436617e-01 -9.80362073e-02 2.83158541e-01 9.12745535e-01
-5.70883095e-01 -1.42461014e+00 -3.25447261e-01 6.72021568e-01
-3.60052109e-01 -1.63422436e-01 -3.38770777e-01 3.25084358e-01
1.73962221e-01 5.32330930e-01 7.09471554e-02 -1.29628673e-01
1.30451545e-02 4.40788418e-01 8.93164873e-01 -3.84633482e-01
-2.85323679e-01 9.37799662e-02 1.66538328e-01 -4.68932122e-01
-4.48692650e-01 -8.33758354e-01 -1.03952539e+00 -1.02234773e-01
-1.99789435e-01 1.12808719e-01 7.50174701e-01 1.24902105e+00
3.49691153e-01 7.75700152e-01 4.14816618e-01 -1.25210178e+00
-4.61706251e-01 -9.50783968e-01 -6.21429682e-01 4.60527539e-01
2.74337113e-01 -8.00177038e-01 -2.16015205e-01 -1.83268487e-01] | [5.297346591949463, 5.209017276763916] |
490a69e7-f6b4-49c8-808b-91d72e77f620 | persuasion-for-good-towards-a-personalized | 1906.06725 | null | https://arxiv.org/abs/1906.06725v2 | https://arxiv.org/pdf/1906.06725v2.pdf | Persuasion for Good: Towards a Personalized Persuasive Dialogue System for Social Good | Developing intelligent persuasive conversational agents to change people's opinions and actions for social good is the frontier in advancing the ethical development of automated dialogue systems. To do so, the first step is to understand the intricate organization of strategic disclosures and appeals employed in human persuasion conversations. We designed an online persuasion task where one participant was asked to persuade the other to donate to a specific charity. We collected a large dataset with 1,017 dialogues and annotated emerging persuasion strategies from a subset. Based on the annotation, we built a baseline classifier with context information and sentence-level features to predict the 10 persuasion strategies used in the corpus. Furthermore, to develop an understanding of personalized persuasion processes, we analyzed the relationships between individuals' demographic and psychological backgrounds including personality, morality, value systems, and their willingness for donation. Then, we analyzed which types of persuasion strategies led to a greater amount of donation depending on the individuals' personal backgrounds. This work lays the ground for developing a personalized persuasive dialogue system. | ['Zhou Yu', 'Sijia Yang', 'Yoojung Oh', 'Xuewei Wang', 'Jingwen Zhang', 'Weiyan Shi', 'Richard Kim'] | 2019-06-16 | persuasion-for-good-towards-a-personalized-1 | https://aclanthology.org/P19-1566 | https://aclanthology.org/P19-1566.pdf | acl-2019-7 | ['persuasion-strategies'] | ['computer-vision'] | [ 1.89239860e-01 7.93107212e-01 -2.75618404e-01 -9.58058655e-01
-4.24413741e-01 -4.78544027e-01 9.76163447e-01 4.03316259e-01
-6.33148134e-01 1.14100468e+00 1.11471415e+00 -4.53538358e-01
8.68312493e-02 -6.62499487e-01 2.16375247e-01 -4.35192376e-01
5.20387530e-01 3.38192195e-01 -2.46762395e-01 -7.41683185e-01
8.18779051e-01 2.21169628e-02 -1.02749872e+00 3.45389128e-01
1.08007991e+00 5.60184717e-01 -2.84313023e-01 7.32986689e-01
-3.76133233e-01 1.13269424e+00 -6.56804979e-01 -1.05322468e+00
-9.82984230e-02 -9.04662967e-01 -1.10725689e+00 -1.22857161e-01
1.48141921e-01 -4.12743926e-01 1.79816857e-01 6.07986152e-01
5.35799742e-01 9.45226550e-02 8.02751601e-01 -9.24343526e-01
-8.20533633e-01 1.24158919e+00 6.36922494e-02 2.35776141e-01
6.83988810e-01 3.48835737e-01 1.12463069e+00 -1.99166834e-01
6.33030415e-01 1.53359592e+00 5.30559480e-01 1.17322254e+00
-1.23113012e+00 -2.18213931e-01 1.52694322e-02 2.44873434e-01
-2.11569041e-01 -8.26019585e-01 9.76367414e-01 -6.07871473e-01
3.67082238e-01 2.03622431e-01 9.72280622e-01 1.41807270e+00
-2.99161300e-02 6.04818106e-01 1.49654925e+00 -4.97128516e-01
2.70200819e-01 7.03185916e-01 9.33523536e-01 4.78628099e-01
5.57251163e-02 -3.31947595e-01 -5.59815526e-01 -7.19829798e-01
4.41517271e-02 -6.66263998e-01 -1.09260522e-01 2.65668452e-01
-1.08995330e+00 1.38567829e+00 1.08962327e-01 3.15201432e-01
-5.58392406e-01 -3.75570476e-01 6.35797739e-01 3.58153015e-01
4.38748360e-01 9.43480909e-01 -2.91745573e-01 -6.12630725e-01
-5.64396568e-02 1.86881736e-01 1.42036712e+00 -9.21745747e-02
5.80236435e-01 -2.49706969e-01 -3.68822515e-01 1.04621601e+00
1.95472032e-01 1.99712723e-01 2.97590226e-01 -1.03299868e+00
2.03687564e-01 7.29200721e-01 3.96861643e-01 -1.17437291e+00
-3.91820550e-01 2.01386943e-01 -3.53011817e-01 -1.35790154e-01
6.96630538e-01 -8.13193202e-01 1.69357836e-01 1.64697039e+00
4.67739135e-01 -7.86402643e-01 3.69960576e-01 8.64938676e-01
6.76754713e-01 3.65240246e-01 3.69306505e-01 -2.99315900e-01
1.55416632e+00 -5.50927341e-01 -6.65035784e-01 -1.74442127e-01
7.07564294e-01 -6.34664953e-01 1.19981050e+00 -6.84913574e-03
-9.35589850e-01 -2.12613791e-01 -7.28781402e-01 -5.43560907e-02
-2.98054256e-02 -2.36777559e-01 1.11127341e+00 1.07093167e+00
-8.73143315e-01 4.67410773e-01 -9.50792246e-03 -5.96505046e-01
4.14483398e-01 -8.92859399e-02 4.80594710e-02 5.16250372e-01
-1.58845270e+00 1.11587155e+00 -1.65099487e-01 -2.15306073e-01
-7.80156553e-02 -5.02993166e-01 -5.90301275e-01 -9.47470516e-02
2.00036019e-01 -9.58137929e-01 1.21447551e+00 -1.38080144e+00
-1.89305127e+00 1.04508460e+00 1.97681293e-01 -5.59148490e-01
6.88549340e-01 5.41204065e-02 -2.64021665e-01 1.12843215e-01
4.03154790e-01 6.76444292e-01 6.70403302e-01 -8.03540170e-01
-7.12314606e-01 -5.00937641e-01 2.55809814e-01 5.08969188e-01
-4.04590279e-01 2.83341527e-01 7.48209119e-01 -2.05870762e-01
-6.99787676e-01 -5.55793226e-01 -6.58267662e-02 -3.03433359e-01
-3.98675889e-01 -8.45337629e-01 1.55729964e-01 -5.40806830e-01
6.42450690e-01 -1.96337259e+00 -5.91995306e-02 -1.68977510e-02
3.90187860e-01 7.96971768e-02 7.32546300e-02 5.23340940e-01
4.99003261e-01 3.11468244e-01 4.94534552e-01 -3.52220722e-02
2.94360965e-01 -1.11906871e-01 -1.86164156e-01 1.35211527e-01
-1.06790736e-01 6.70970917e-01 -9.27984953e-01 -5.60958445e-01
-6.22505434e-02 1.13840044e-01 -5.37802100e-01 2.05494002e-01
-1.01577774e-01 4.71700728e-01 -8.93460870e-01 1.12305202e-01
1.02425955e-01 -3.07247322e-02 1.97467417e-01 2.31757939e-01
-2.62591243e-01 9.69961226e-01 -3.78611118e-01 8.31133902e-01
-2.05004156e-01 8.41470063e-01 4.47864741e-01 -7.26761997e-01
1.22060990e+00 -4.68472429e-02 2.35329881e-01 -4.52248961e-01
6.02076352e-01 4.60353047e-02 4.22361910e-01 -4.14746851e-01
6.57952785e-01 -3.54675353e-01 -4.05920595e-01 1.00845993e+00
-5.86387575e-01 1.22913837e-01 1.34362012e-01 4.88216668e-01
8.31332743e-01 -5.15201271e-01 4.50651497e-01 -6.40613556e-01
7.69239485e-01 3.44264925e-01 6.14864349e-01 8.83058846e-01
-6.84846938e-01 -3.63642097e-01 8.24061334e-01 -8.16872716e-01
-7.15458751e-01 -5.21313608e-01 1.09148555e-01 1.27379012e+00
8.58629271e-02 -6.20027864e-03 -9.13158536e-01 -7.37414658e-01
-7.88916126e-02 1.36564636e+00 -5.95168293e-01 -1.36878461e-01
-3.53127271e-01 -6.31783307e-01 6.29279613e-01 2.15427130e-02
9.51226771e-01 -1.03079498e+00 -5.28830469e-01 2.78445512e-01
-5.72824657e-01 -7.06529856e-01 -5.61557651e-01 -4.78251874e-01
-3.33308220e-01 -1.12878990e+00 -1.66110009e-01 -4.47006434e-01
2.76823789e-01 1.22915395e-01 1.04211819e+00 1.32002383e-01
2.40430789e-04 5.92768788e-01 -3.53807956e-01 -6.07902527e-01
-1.10708117e+00 1.49466321e-01 3.10708195e-01 2.91843675e-02
6.42273009e-01 -2.23084033e-01 -5.89744449e-01 5.02319574e-01
-2.21291006e-01 2.04501286e-01 1.36570603e-01 6.67813838e-01
-5.30284345e-01 -5.12090921e-01 1.04911876e+00 -9.35760617e-01
1.64455640e+00 -3.53533626e-01 -1.29219340e-02 2.65157282e-01
-7.85282016e-01 -1.59851268e-01 7.16729343e-01 -4.74633336e-01
-1.44194961e+00 -4.41875666e-01 -2.99857706e-01 1.05421305e+00
-2.25741178e-01 1.23823307e-01 -8.18198323e-02 1.89026475e-01
9.76248085e-01 2.81254891e-02 5.05042791e-01 -1.66235670e-01
4.52583879e-01 9.74044383e-01 1.82754621e-01 -8.33358467e-01
1.98416889e-01 1.00902110e-01 -5.86973548e-01 -9.42431390e-01
-8.58620822e-01 6.05590306e-02 -3.49559844e-01 -5.41228473e-01
1.01543486e+00 -4.23729509e-01 -1.42408514e+00 4.91959959e-01
-1.06624424e+00 -4.61431116e-01 -7.35409260e-02 3.93609405e-01
-4.09554988e-01 6.41275465e-01 -6.11693919e-01 -1.07219422e+00
-7.28330910e-01 -7.20156610e-01 3.82465333e-01 4.86727744e-01
-9.92674112e-01 -1.19796360e+00 1.80315971e-01 1.17489803e+00
5.00493646e-01 -2.53145039e-01 1.14834309e+00 -1.09408319e+00
2.53227264e-01 7.09418356e-02 7.31801540e-02 2.41176441e-01
2.84298182e-01 -1.74074471e-01 -4.00260806e-01 3.21748912e-01
2.89147913e-01 -6.57116592e-01 2.45149121e-01 3.16202879e-01
2.93110341e-01 -9.04581189e-01 -9.17961299e-02 -2.29184523e-01
4.24153000e-01 1.38962924e-01 5.99262357e-01 3.45111698e-01
1.29489064e-01 1.24034750e+00 5.87807298e-01 6.11598969e-01
9.99021649e-01 4.55386370e-01 -2.91090399e-01 2.72107631e-01
2.81021148e-01 -4.40676004e-01 6.82503939e-01 2.56887019e-01
2.57575121e-02 -8.20948184e-02 -7.91262805e-01 3.04912180e-01
-1.49135435e+00 -1.11545050e+00 -2.59873271e-01 1.71663678e+00
1.09284687e+00 -1.03010228e-02 4.58429486e-01 -9.97249484e-02
8.74402881e-01 1.01761095e-01 -4.06668991e-01 -9.41214025e-01
-1.83293089e-01 -5.05608261e-01 4.94089276e-02 8.84836853e-01
-5.66510618e-01 1.03102136e+00 6.21688175e+00 1.68534011e-01
-7.69373178e-01 -3.37932348e-01 1.18363070e+00 3.16149026e-01
-5.58575571e-01 -5.00683440e-03 -9.61311817e-01 5.12542248e-01
9.23251450e-01 -4.69004661e-01 4.35825616e-01 6.86779857e-01
5.09865522e-01 -3.70432138e-01 -1.25708437e+00 4.44666684e-01
1.98037937e-01 -1.27007782e+00 -2.63831139e-01 8.99296850e-02
3.43539566e-01 -5.69603860e-01 -2.71229774e-01 3.16388339e-01
9.64187682e-01 -6.18224084e-01 2.32761666e-01 5.80958903e-01
6.33525550e-02 -2.12770253e-01 7.35666752e-01 5.88618815e-01
2.30142996e-02 -5.71167916e-02 -9.66040120e-02 -6.99811876e-01
5.02831936e-01 2.02191457e-01 -1.49519718e+00 -1.49339616e-01
2.65540808e-01 3.62322927e-01 -2.14006931e-01 -1.93453655e-01
-2.75881171e-01 7.24957705e-01 -1.00516886e-01 -8.95955682e-01
7.24986894e-03 -4.73136783e-01 5.61379075e-01 8.19943726e-01
-3.41080278e-01 5.61050534e-01 -4.67679985e-02 7.26604760e-01
-3.88532020e-02 5.14926970e-01 -3.96654636e-01 -3.19620132e-01
5.26562095e-01 1.33167517e+00 -3.88893485e-01 -4.31107700e-01
-1.27725944e-01 7.33931243e-01 2.33362764e-01 1.69466063e-02
-3.96787047e-01 -1.35928877e-02 8.36958289e-01 1.42188430e-01
-4.17363614e-01 2.03514680e-01 -7.40486562e-01 -1.09151685e+00
-4.69631732e-01 -1.04216504e+00 5.05051374e-01 -5.51059425e-01
-1.76951396e+00 2.29842782e-01 -3.25145900e-01 -4.01442982e-02
-4.10011649e-01 -2.77159572e-01 -9.49648738e-01 9.17040706e-01
-1.06954205e+00 -8.14834297e-01 -2.91661859e-01 3.27764660e-01
5.95508456e-01 -3.89935523e-01 7.94225812e-01 -3.13778192e-01
-3.96383107e-01 4.69690263e-01 -4.92173553e-01 2.96418428e-01
9.88661587e-01 -9.44528937e-01 1.15673333e-01 2.24568337e-01
-5.92744052e-01 6.40084803e-01 8.34564626e-01 -7.01589525e-01
-1.16754711e+00 -2.43307218e-01 1.27615297e+00 -5.22552431e-01
9.78627026e-01 -1.72216982e-01 -5.70081592e-01 2.52333641e-01
5.00309587e-01 -1.07818687e+00 1.25887978e+00 5.62725365e-01
-4.11240477e-03 5.76158725e-02 -1.33410180e+00 9.37753379e-01
1.01946557e+00 -3.79132956e-01 -1.06841493e+00 2.24507719e-01
2.62404114e-01 2.59959936e-01 -7.93048382e-01 -4.07383353e-01
4.82306510e-01 -1.11266494e+00 6.74810946e-01 -8.88300002e-01
6.87645376e-01 4.85787898e-01 1.45520829e-02 -1.46568251e+00
-3.67869973e-01 -8.54630411e-01 6.83283508e-01 1.40392303e+00
3.48813176e-01 -1.04333067e+00 6.91106737e-01 1.54386139e+00
9.40836594e-02 -5.60699522e-01 -6.84733808e-01 1.78400010e-01
2.93434173e-01 1.15878761e-01 5.37874579e-01 9.68491971e-01
6.27672136e-01 1.21739912e+00 -4.88451183e-01 -5.68600714e-01
5.08516252e-01 1.39922008e-01 1.15859389e+00 -1.39563775e+00
-8.64952207e-02 -6.67822182e-01 -8.63484964e-02 -1.04698944e+00
5.19356608e-01 -6.14645004e-01 -2.32774451e-01 -1.51771402e+00
2.02748120e-01 -5.29959083e-01 5.17187119e-01 4.69649494e-01
-4.84045535e-01 -5.22316575e-01 5.68712987e-02 6.52926415e-02
-3.66406769e-01 7.20066488e-01 1.03650355e+00 -1.18121495e-02
-5.99562466e-01 1.62867576e-01 -1.74923003e+00 7.04420030e-01
1.09683287e+00 5.50036356e-02 -2.49706760e-01 8.87468364e-03
1.91689104e-01 3.37819993e-01 2.50453830e-01 -2.93739121e-02
1.49821669e-01 -5.52705824e-01 6.93306327e-02 -1.48184076e-01
2.47917295e-01 -2.38804072e-01 -3.64993274e-01 4.60232049e-01
-1.08472228e+00 -2.92143792e-01 -4.01277959e-01 3.87857497e-01
3.36868137e-01 -2.11190969e-01 7.73937345e-01 -4.66513671e-02
-2.95674533e-01 -3.40498388e-01 -9.06659842e-01 1.50258064e-01
9.83067393e-01 -2.28658691e-01 -6.57053888e-01 -9.74411428e-01
-5.26215672e-01 4.73561972e-01 5.82924724e-01 4.03646439e-01
3.45872641e-01 -8.58035207e-01 -1.12885916e+00 -2.77353495e-01
-3.01952153e-01 -8.94001305e-01 2.43493110e-01 7.22232819e-01
-1.33508548e-01 1.69891283e-01 -4.44806099e-01 -4.34144661e-02
-1.50676668e+00 4.79546227e-02 3.85633826e-01 7.49765337e-02
-4.54880834e-01 6.88822865e-01 -7.49335140e-02 -2.97840744e-01
-1.08436957e-01 4.90021527e-01 -6.71070814e-01 3.55761737e-01
6.51790738e-01 6.89019859e-01 -6.03218079e-01 -5.47810316e-01
-1.75115719e-01 -2.25339219e-01 -3.64853114e-01 -3.42836112e-01
1.22321713e+00 -3.72306287e-01 -3.39845121e-01 3.59017760e-01
7.73682773e-01 3.18345875e-01 -9.19433117e-01 -5.13173416e-02
-5.05799837e-02 -7.33841598e-01 -4.20360774e-01 -9.59737539e-01
-3.25934857e-01 4.22893405e-01 -1.46325931e-01 6.64678216e-01
4.85542029e-01 -5.46503859e-03 5.64576685e-01 4.40051675e-01
2.42225334e-01 -1.50689304e+00 1.29332289e-01 4.19917494e-01
8.01811516e-01 -1.38929951e+00 7.86609482e-03 -4.70325857e-01
-1.36535478e+00 1.02978039e+00 6.54669702e-01 3.14571351e-01
2.28208363e-01 -2.29866385e-01 4.07639265e-01 -2.01010212e-01
-9.02925968e-01 3.38404536e-01 -1.18495770e-01 5.81152380e-01
5.11595249e-01 3.49641770e-01 -1.15147913e+00 8.07431042e-01
-7.20927894e-01 -2.88206458e-01 8.48864615e-01 2.43346691e-01
-6.27878189e-01 -1.25944579e+00 -1.79413736e-01 6.40800416e-01
-3.55054289e-01 5.82753718e-02 -1.47056949e+00 2.56490201e-01
-3.50251734e-01 1.43432593e+00 -2.33482957e-01 -2.83312410e-01
6.10763505e-02 2.85621405e-01 -1.53394444e-02 -4.19579536e-01
-1.00549126e+00 -4.26425368e-01 1.11633360e+00 2.83002723e-02
-4.82624978e-01 -1.12714052e+00 -9.98642087e-01 -8.39162886e-01
6.68247715e-02 6.66614830e-01 4.68316048e-01 9.50175941e-01
4.11056548e-01 -2.21381322e-01 8.68144572e-01 -1.75643459e-01
-9.26452518e-01 -1.11083877e+00 -1.50619850e-01 5.31878114e-01
-7.03989863e-02 -2.01008350e-01 -4.79496419e-01 -2.44908139e-01] | [12.795148849487305, 7.9718546867370605] |
aaffcddc-a2ce-4023-8f66-9949f98f5b98 | why-do-cnns-excel-at-feature-extraction-a | 2307.00919 | null | https://arxiv.org/abs/2307.00919v1 | https://arxiv.org/pdf/2307.00919v1.pdf | Why do CNNs excel at feature extraction? A mathematical explanation | Over the past decade deep learning has revolutionized the field of computer vision, with convolutional neural network models proving to be very effective for image classification benchmarks. However, a fundamental theoretical questions remain answered: why can they solve discrete image classification tasks that involve feature extraction? We address this question in this paper by introducing a novel mathematical model for image classification, based on feature extraction, that can be used to generate images resembling real-world datasets. We show that convolutional neural network classifiers can solve these image classification tasks with zero error. In our proof, we construct piecewise linear functions that detect the presence of features, and show that they can be realized by a convolutional network. | ['Tongliang Liu', 'Arush Tagade', 'Vinoth Nandakumar'] | 2023-07-03 | null | null | null | null | ['classification-1'] | ['methodology'] | [ 5.08777440e-01 2.55138546e-01 -3.15405913e-02 -2.85630673e-01
-3.19493592e-01 -5.42477369e-01 4.17450905e-01 4.26165722e-02
-4.62669373e-01 4.91012275e-01 -4.75815445e-01 -4.78874892e-01
-2.93487102e-01 -1.02819097e+00 -1.00919712e+00 -7.42940485e-01
-2.30325013e-01 -3.28166597e-02 3.34308930e-02 -4.01137620e-01
2.18344003e-01 8.13745320e-01 -1.69124782e+00 4.64659572e-01
4.01226878e-01 1.31468225e+00 -9.48609561e-02 1.10931444e+00
1.13970742e-01 9.30154741e-01 -7.21536160e-01 -1.55308113e-01
5.54854691e-01 -1.51831463e-01 -9.40470994e-01 1.61588371e-01
5.92831850e-01 -3.13235074e-01 -5.48081934e-01 1.01290095e+00
1.33915067e-01 -2.92268932e-01 6.89615071e-01 -1.59468830e+00
-9.54578817e-01 1.81515858e-01 -9.85674486e-02 1.08638182e-01
-2.99252868e-02 3.69701348e-02 9.16967154e-01 -6.17129445e-01
2.59060413e-01 1.04192984e+00 1.01691103e+00 5.75192571e-01
-1.48588026e+00 -3.31906021e-01 -3.22658688e-01 2.92671025e-01
-9.90872085e-01 1.61875915e-02 8.26714516e-01 -5.25605142e-01
7.36202061e-01 4.10274297e-01 5.70997238e-01 7.57328570e-01
2.01525763e-01 8.07033777e-01 1.01991570e+00 -5.62860608e-01
-8.71223118e-03 -2.01407652e-02 3.66405338e-01 1.06247616e+00
2.59471714e-01 2.18765110e-01 2.02760562e-01 6.61177263e-02
1.01375294e+00 1.69564977e-01 -4.61153120e-01 -2.52883732e-01
-1.09542871e+00 1.29427683e+00 8.34816575e-01 3.84032398e-01
-2.74116009e-01 3.71490180e-01 3.19536865e-01 7.42102265e-01
2.06406817e-01 6.46109641e-01 -5.79093456e-01 4.21881557e-01
-4.48090464e-01 2.92152345e-01 7.42706835e-01 5.95183849e-01
9.20025170e-01 3.79213169e-02 8.41091871e-02 5.55732489e-01
-2.65899956e-01 2.91769922e-01 4.37989414e-01 -1.17915714e+00
-5.69385849e-02 5.58901548e-01 4.34208801e-03 -1.00564551e+00
-5.89962006e-01 -3.68133098e-01 -1.09680200e+00 5.72906137e-01
6.81574702e-01 -1.57165363e-01 -7.72956669e-01 1.48697841e+00
-1.45055830e-01 1.79481223e-01 1.90241337e-01 6.48570955e-01
7.03068972e-01 7.07422972e-01 -4.03703660e-01 2.07738131e-01
9.27238703e-01 -6.86260939e-01 -3.45111310e-01 2.00620517e-01
6.10185087e-01 -4.70823824e-01 8.38882864e-01 4.92910415e-01
-8.31323743e-01 -8.18853021e-01 -1.30526555e+00 -2.33746663e-01
-5.90025544e-01 1.79078579e-01 8.08129430e-01 6.70110345e-01
-1.22449934e+00 8.73598695e-01 -5.63123882e-01 -5.05788289e-02
6.29605711e-01 6.82573318e-01 -5.58430552e-01 3.85319591e-02
-9.46393371e-01 6.31873608e-01 2.94005156e-01 3.58987659e-01
-7.85254180e-01 -5.69990158e-01 -9.38460410e-01 1.78851441e-01
3.38285677e-02 -5.99179566e-01 1.37904644e+00 -1.48539162e+00
-1.18794537e+00 9.29280043e-01 6.14346303e-02 -8.78932118e-01
5.10063410e-01 1.31555468e-01 -7.77907521e-02 2.17063963e-01
-1.65204003e-01 6.96009517e-01 9.87086952e-01 -1.14607131e+00
-7.37973094e-01 -1.25414014e-01 4.62818563e-01 -6.72219753e-01
-2.50135511e-01 -2.30868667e-01 1.95282891e-01 -4.25816894e-01
-4.71956469e-02 -9.26104665e-01 -4.78933305e-01 5.05381823e-01
-2.87355691e-01 -2.45000497e-01 8.79031837e-01 -2.31569484e-01
2.55578250e-01 -2.19988871e+00 -6.50014877e-02 6.90052882e-02
5.00411093e-01 5.58862925e-01 -4.25446689e-01 3.75197083e-02
-3.11060160e-01 3.65204364e-01 -3.11610848e-01 7.55875334e-02
-2.59341270e-01 4.08038497e-01 -5.14366865e-01 6.14296019e-01
6.53458774e-01 1.21072805e+00 -7.24666953e-01 -1.60149872e-01
2.94502318e-01 5.65570772e-01 -4.56037194e-01 2.45968893e-01
-2.08474889e-01 1.72263011e-01 -3.23711991e-01 1.71060085e-01
6.53738916e-01 -3.24430913e-01 1.64687801e-02 -1.77736834e-01
2.70106643e-02 -1.54802471e-01 -7.92276502e-01 9.24252510e-01
-5.43850064e-01 1.27343798e+00 8.51121694e-02 -1.83389199e+00
8.89754474e-01 1.84575900e-01 5.15107691e-01 -6.55558944e-01
2.52337039e-01 6.45451099e-02 5.42210340e-02 -5.86573482e-01
1.23199008e-01 -3.85426771e-04 4.22322117e-02 1.95978414e-02
7.93168619e-02 -1.85256615e-01 -2.71094944e-02 -3.09586078e-01
1.24865222e+00 -4.01424348e-01 3.28534655e-02 -2.32701287e-01
7.41134048e-01 2.48692095e-01 2.86360443e-01 9.37013090e-01
-1.33427069e-01 7.53344119e-01 7.47784078e-01 -1.08865154e+00
-1.22472978e+00 -1.14594352e+00 -1.12893656e-01 7.34773457e-01
8.87087211e-02 7.08215963e-03 -9.29738939e-01 -5.13334453e-01
-1.67859569e-02 -1.25970766e-02 -1.02895367e+00 -3.02706182e-01
-8.59145939e-01 -6.02558851e-01 6.15355194e-01 5.72667181e-01
7.28786588e-01 -1.22432089e+00 -7.36727476e-01 2.21965596e-01
1.29488155e-01 -1.31541741e+00 8.89456198e-02 4.45862979e-01
-9.00314629e-01 -1.45797718e+00 -9.69318449e-01 -1.37474787e+00
8.25574458e-01 3.57570469e-01 1.21556568e+00 5.53261042e-01
-7.50115573e-01 5.05221605e-01 -2.64968693e-01 -3.65990877e-01
-6.45045042e-01 9.01453495e-02 -3.80078048e-01 1.60171822e-01
5.43644913e-02 -4.91101861e-01 -6.08631909e-01 2.11534902e-01
-1.24542177e+00 1.14399180e-01 6.16028965e-01 1.05553043e+00
3.77581537e-01 1.11355104e-01 6.43947423e-01 -7.77193069e-01
5.49912572e-01 -2.85735965e-01 -7.75479078e-01 1.82381213e-01
-1.90505445e-01 3.54341418e-01 1.14419508e+00 -3.35062027e-01
-2.44105294e-01 4.37210709e-01 -4.36533630e-01 -3.07845652e-01
-2.86521465e-01 4.36780065e-01 3.07508439e-01 -4.95964736e-01
7.27721214e-01 3.72696608e-01 2.91715831e-01 -1.71272561e-01
2.37851590e-01 5.32643616e-01 7.29092717e-01 -3.56284916e-01
6.76624238e-01 6.17526412e-01 5.29569387e-01 -1.12097895e+00
-1.05685055e+00 -1.27057031e-01 -6.73669994e-01 -1.14170030e-01
8.28360975e-01 -5.36100328e-01 -1.02819622e+00 4.99797076e-01
-1.37942958e+00 -4.83515114e-01 -1.94405362e-01 1.12228140e-01
-7.83917427e-01 1.04483530e-01 -6.66864932e-01 -4.88741070e-01
-1.37819305e-01 -1.17643750e+00 8.17320645e-01 7.37797767e-02
2.16362640e-01 -1.04213965e+00 -1.48242205e-01 -1.45814255e-01
4.31192130e-01 5.63622892e-01 1.11269164e+00 -3.62290114e-01
-5.83375394e-01 -3.28833908e-01 -5.05918622e-01 8.41760874e-01
1.02706313e-01 6.61912560e-02 -1.01298618e+00 -1.55853555e-01
2.66803764e-02 -3.77324462e-01 1.05684173e+00 4.27481651e-01
1.78718722e+00 -4.45498496e-01 -1.18248798e-01 8.57009530e-01
1.58457112e+00 -4.34592478e-02 7.51209915e-01 2.50309050e-01
4.19367313e-01 4.79193449e-01 -6.32656142e-02 -5.97083680e-02
-6.49996251e-02 4.78346765e-01 5.91653347e-01 -2.76260763e-01
-1.34351358e-01 2.20637143e-01 2.01129057e-02 3.66159588e-01
-9.94077548e-02 8.52221474e-02 -7.15280116e-01 5.56720376e-01
-1.89205086e+00 -7.98130035e-01 -3.01817209e-01 1.87888777e+00
4.70828444e-01 3.18655968e-01 1.02227069e-01 6.27542794e-01
7.28719413e-01 -2.57924497e-01 -2.91550606e-01 -4.40236926e-01
-2.72649139e-01 6.76044047e-01 4.84510571e-01 3.56844842e-01
-1.25940025e+00 5.70576310e-01 7.37769222e+00 3.10599178e-01
-1.57486057e+00 -2.15523556e-01 8.20387900e-01 5.76698124e-01
4.18046750e-02 -4.37246442e-01 -3.09645981e-01 2.11360529e-01
7.17312336e-01 -6.10782132e-02 2.39093632e-01 9.98226345e-01
-2.06406862e-01 2.80873150e-01 -1.38789630e+00 1.12845051e+00
-1.31066933e-01 -1.74716568e+00 1.56564370e-01 1.96416695e-02
7.22279429e-01 -3.60805951e-02 3.91517103e-01 2.34861076e-01
2.23494753e-01 -1.38818192e+00 4.16435421e-01 5.08487403e-01
5.85871100e-01 -7.11561203e-01 7.11086988e-01 3.60612452e-01
-1.02507043e+00 -3.49731743e-01 -4.88247097e-01 -3.70030284e-01
-5.04500985e-01 5.63193858e-01 -7.75201797e-01 2.38665879e-01
4.38426584e-01 6.02841437e-01 -6.72105968e-01 1.24632645e+00
8.91901366e-03 5.68350315e-01 -2.47863740e-01 -1.24881148e-01
5.26788235e-01 3.97791527e-02 3.24768648e-02 1.07785535e+00
3.17121208e-01 -1.99587598e-01 1.05802976e-01 9.84067559e-01
-2.68041342e-01 -3.08962137e-01 -7.88717031e-01 1.20792970e-01
-2.02841535e-01 1.29072225e+00 -9.17615831e-01 -2.52652019e-01
-3.69118005e-01 1.04775369e+00 3.71131510e-01 2.63708085e-01
-6.67850316e-01 -6.61193132e-01 6.53635323e-01 1.05699793e-01
4.43602085e-01 -4.12678123e-01 -3.09988678e-01 -1.06246471e+00
9.21492931e-03 -6.29405022e-01 1.29490709e-02 -4.84680504e-01
-1.08608294e+00 7.48718858e-01 -3.96935046e-01 -9.57360566e-01
-3.34990084e-01 -1.34765959e+00 -7.18051195e-01 4.14733827e-01
-1.58166826e+00 -9.78235960e-01 -4.82257903e-01 6.75386667e-01
3.72576803e-01 8.46216083e-03 8.52971315e-01 2.03271344e-01
-1.90340623e-01 5.10022759e-01 2.51063704e-02 7.94514716e-01
-6.55750409e-02 -1.37840390e+00 5.78338146e-01 4.61118370e-01
3.56215894e-01 2.22435445e-01 6.15113914e-01 1.92630425e-01
-1.50595784e+00 -1.16719186e+00 6.02490366e-01 -1.92114756e-01
4.92892891e-01 -3.35444599e-01 -8.58259261e-01 5.84814727e-01
-1.37100995e-01 6.90519631e-01 1.43596813e-01 -2.81416923e-01
-5.77331662e-01 -8.65719616e-02 -1.13540387e+00 3.57974887e-01
7.29267120e-01 -5.86564362e-01 -4.24171895e-01 3.39976937e-01
6.45110607e-01 -5.35498261e-02 -5.74076533e-01 5.06562650e-01
6.15511656e-01 -9.70362067e-01 1.16939485e+00 -8.68709028e-01
6.32929027e-01 7.33479112e-03 3.03871036e-02 -1.23659408e+00
-1.56054914e-01 -5.49755275e-01 1.25935316e-01 4.56420839e-01
3.35942388e-01 -6.88672483e-01 8.65396857e-01 1.23758726e-01
1.29732549e-01 -8.93181741e-01 -7.88907230e-01 -1.02682424e+00
4.94853169e-01 -4.12860870e-01 2.80510485e-01 6.45115077e-01
-3.65632504e-01 1.81312159e-01 -1.17078468e-01 4.70456146e-02
4.21693325e-01 1.80037260e-01 6.92575157e-01 -1.49722815e+00
-1.53604448e-01 -7.62611151e-01 -1.00952363e+00 -1.04311979e+00
3.86138588e-01 -8.14421833e-01 1.04715563e-01 -1.40264869e+00
-4.84559722e-02 -5.27719200e-01 -1.92289934e-01 5.22738993e-01
1.79756865e-01 6.33343220e-01 2.15103656e-01 -1.21230908e-01
-2.40773305e-01 1.63841560e-01 1.06968594e+00 -4.86865312e-01
2.47440845e-01 1.83800340e-01 -5.02704024e-01 8.38814795e-01
8.87596548e-01 -3.53401482e-01 6.33413941e-02 -3.70979190e-01
4.40592855e-01 -8.38168897e-03 8.65855277e-01 -1.38699794e+00
1.28173139e-02 5.98938204e-02 6.15573347e-01 -3.77088934e-02
1.98585048e-01 -9.67815101e-01 -2.62279809e-01 8.33047807e-01
-6.93749607e-01 -8.40450078e-02 1.65015087e-01 1.95599005e-01
-2.66155034e-01 -5.47683060e-01 9.29933012e-01 -6.71282634e-02
-5.48033416e-01 3.91479254e-01 -5.80334783e-01 -2.66862810e-01
9.80402946e-01 9.25552621e-02 -1.32921785e-01 -5.63327193e-01
-8.30888212e-01 -1.32043526e-01 7.77039304e-02 2.90850013e-01
7.30878174e-01 -1.24579942e+00 -7.77759135e-01 2.38555372e-01
-4.86596040e-02 -2.74788514e-02 -2.35647470e-01 4.06908303e-01
-9.13513839e-01 4.22411442e-01 -3.82122785e-01 -7.42994845e-01
-1.12936783e+00 8.69558096e-01 7.42163837e-01 7.41346553e-02
-6.78907752e-01 5.33726275e-01 3.29098612e-01 -4.01299894e-01
2.34188199e-01 -7.71960497e-01 3.10188672e-03 -3.63159388e-01
6.42133176e-01 4.20691557e-02 7.54334629e-02 -3.75862688e-01
3.22075225e-02 6.96107924e-01 9.21736807e-02 2.24319741e-01
1.55916834e+00 3.42971176e-01 -3.42731774e-02 2.56440341e-01
1.84607995e+00 -5.48859417e-01 -1.25063694e+00 -8.21780413e-02
1.07640810e-01 -1.24879919e-01 -4.27570604e-02 -4.11037892e-01
-1.26075327e+00 1.06488121e+00 5.94041944e-01 9.81040299e-01
1.13782036e+00 -2.24350719e-03 8.28755140e-01 1.02978384e+00
1.24289155e-01 -6.79241836e-01 2.45199978e-01 7.58611321e-01
1.02266467e+00 -1.39297843e+00 -4.64302808e-01 -4.25124168e-01
9.61553752e-02 1.78768241e+00 2.73151070e-01 -8.84941638e-01
9.80989695e-01 3.49646658e-01 -1.02603763e-01 -7.38059431e-02
-5.79191267e-01 -3.79051208e-01 2.12160230e-01 6.40442908e-01
2.14196995e-01 -1.22584123e-02 1.32879585e-01 2.45678037e-01
-2.48403430e-01 2.76903629e-01 6.48749471e-01 8.42306674e-01
-5.68577170e-01 -9.04485941e-01 -2.40354732e-01 4.94386137e-01
-6.11912847e-01 4.07325402e-02 -4.17871296e-01 6.72115386e-01
2.65068561e-01 7.13149011e-01 3.77301633e-01 -4.45626497e-01
3.77316028e-02 -2.13041648e-01 8.45165849e-01 -2.80202031e-01
-4.44639951e-01 -5.63619614e-01 -3.35645616e-01 -2.96702743e-01
-4.03839916e-01 -1.17763273e-01 -1.09330988e+00 -2.77828395e-01
-1.86771706e-01 7.62224346e-02 8.02504301e-01 1.07416153e+00
6.55616894e-02 5.81928074e-01 1.05829728e+00 -8.55821609e-01
-5.82783937e-01 -6.39822602e-01 -2.94073313e-01 3.74041438e-01
1.02784097e+00 -3.45834851e-01 -4.89197850e-01 2.12312549e-01] | [9.207416534423828, 2.274465560913086] |
90a9cd81-db90-4417-9c86-23991c5c557c | robust-self-supervised-audio-visual-speech | 2201.01763 | null | https://arxiv.org/abs/2201.01763v3 | https://arxiv.org/pdf/2201.01763v3.pdf | Robust Self-Supervised Audio-Visual Speech Recognition | Audio-based automatic speech recognition (ASR) degrades significantly in noisy environments and is particularly vulnerable to interfering speech, as the model cannot determine which speaker to transcribe. Audio-visual speech recognition (AVSR) systems improve robustness by complementing the audio stream with the visual information that is invariant to noise and helps the model focus on the desired speaker. However, previous AVSR work focused solely on the supervised learning setup; hence the progress was hindered by the amount of labeled data available. In this work, we present a self-supervised AVSR framework built upon Audio-Visual HuBERT (AV-HuBERT), a state-of-the-art audio-visual speech representation learning model. On the largest available AVSR benchmark dataset LRS3, our approach outperforms prior state-of-the-art by ~50% (28.0% vs. 14.1%) using less than 10% of labeled data (433hr vs. 30hr) in the presence of babble noise, while reducing the WER of an audio-based model by over 75% (25.8% vs. 5.8%) on average. | ['Abdelrahman Mohamed', 'Wei-Ning Hsu', 'Bowen Shi'] | 2022-01-05 | null | null | null | null | ['lipreading', 'audio-visual-speech-recognition'] | ['computer-vision', 'speech'] | [ 3.51891577e-01 2.38442689e-01 8.86225607e-03 -1.82269543e-01
-1.47015572e+00 -6.21208012e-01 7.01332569e-01 1.08874485e-01
-2.95292586e-01 3.03086311e-01 3.86501610e-01 -5.43509781e-01
4.33854222e-01 -1.34752512e-01 -5.94772220e-01 -6.53795242e-01
3.54041129e-01 2.29914859e-01 3.66526753e-01 -2.61206061e-01
8.59378427e-02 5.82217515e-01 -1.77109969e+00 7.72045434e-01
1.05016418e-01 8.34600210e-01 3.07627857e-01 1.19180048e+00
-1.53491661e-01 1.00715113e+00 -1.15023482e+00 2.55381167e-01
-8.08125734e-02 -3.48741174e-01 -6.74210131e-01 1.14671439e-02
5.08237004e-01 8.47041458e-02 -7.74467230e-01 6.30308926e-01
7.65164793e-01 2.53852904e-01 4.71046388e-01 -1.02339077e+00
-6.13996387e-01 4.19901878e-01 -4.33376551e-01 4.58872885e-01
5.49782097e-01 3.10960084e-01 9.06365037e-01 -1.15052104e+00
3.71353745e-01 1.43483686e+00 2.52255499e-01 8.46422613e-01
-1.39432049e+00 -7.13661611e-01 1.33536831e-01 3.20849955e-01
-1.61293709e+00 -1.42048883e+00 6.27209783e-01 -2.77827203e-01
1.40419555e+00 6.90153837e-01 3.58643711e-01 1.45080018e+00
-3.15496922e-01 9.49326396e-01 1.12384474e+00 -6.93568766e-01
2.92058915e-01 2.19289064e-01 -1.43931676e-02 1.83088809e-01
-5.85334837e-01 4.54324543e-01 -1.06438839e+00 1.19795606e-01
2.25760803e-01 -6.41831040e-01 -4.03190523e-01 7.07379505e-02
-9.81242657e-01 4.95734781e-01 2.84461498e-01 2.96408057e-01
4.46208306e-02 1.45541290e-02 5.36819279e-01 5.82052350e-01
5.39364576e-01 2.17155702e-02 -2.08251953e-01 -3.50188881e-01
-1.33943534e+00 -1.58599049e-01 4.14967239e-01 7.92980433e-01
3.47365260e-01 8.22631598e-01 -2.28685230e-01 1.27727365e+00
8.11578214e-01 7.80182719e-01 4.27211821e-01 -7.16797650e-01
5.97558200e-01 -1.82123692e-03 -2.98367053e-01 -3.65470588e-01
-3.00211296e-03 -3.88477594e-01 -5.18384337e-01 4.80477244e-01
2.47697636e-01 3.61043662e-01 -1.42285872e+00 1.26902449e+00
3.12258415e-02 1.92060191e-02 3.22373658e-01 8.53543162e-01
1.39135468e+00 1.15557384e+00 -2.90220827e-02 -2.58220464e-01
1.07715333e+00 -1.02032888e+00 -8.55282247e-01 -4.95254964e-01
4.44565922e-01 -1.08367610e+00 1.26943254e+00 5.92514753e-01
-1.02652299e+00 -5.72986901e-01 -1.13075972e+00 5.43013886e-02
-1.27999991e-01 1.61683619e-01 -2.33968690e-01 1.00687313e+00
-1.48423934e+00 -1.06140114e-01 -7.83485949e-01 -3.23190212e-01
2.84562349e-01 8.77513960e-02 -5.15029669e-01 -1.53907746e-01
-9.47908163e-01 7.61981726e-01 -1.66159291e-02 -9.06999707e-02
-1.54049385e+00 -6.65806293e-01 -9.98379052e-01 -6.86906353e-02
4.82238859e-01 1.08969130e-01 1.59334791e+00 -8.80900383e-01
-1.93420899e+00 8.22595596e-01 -5.43985248e-01 -5.99012375e-01
5.06852865e-01 -1.63714848e-02 -8.44373703e-01 3.99350852e-01
-4.30655569e-01 7.16214836e-01 1.17138624e+00 -1.48670340e+00
-2.27029666e-01 -1.20931640e-01 -3.20853531e-01 9.52429399e-02
2.36909408e-02 4.70517933e-01 -4.04541343e-01 -7.76338875e-01
2.70539317e-02 -7.17813313e-01 3.38389814e-01 -1.74914420e-01
-3.37890238e-01 -8.72687250e-02 1.36208761e+00 -8.06050658e-01
1.02270937e+00 -2.53273344e+00 -2.34294921e-01 -6.01155534e-02
2.50991434e-02 7.85546243e-01 -3.31107855e-01 6.43365324e-01
-3.69268537e-01 1.05726577e-01 -8.66868421e-02 -7.31482208e-01
-1.94617033e-01 2.44108792e-02 -5.74061275e-01 6.01285756e-01
7.34784901e-02 3.88534337e-01 -6.91580117e-01 -4.28838909e-01
6.38427854e-01 8.48512292e-01 -1.46356374e-01 3.17695469e-01
2.29963154e-01 3.68234783e-01 2.97082692e-01 8.03443789e-01
4.52512830e-01 3.12068850e-01 -2.26178229e-01 2.59209592e-02
-1.27093673e-01 7.14430273e-01 -1.11292934e+00 1.34867799e+00
-6.75869048e-01 1.41336775e+00 3.27370077e-01 -8.29082727e-01
1.31269455e+00 7.93090045e-01 1.48015181e-02 -8.91872704e-01
-1.85167447e-01 1.20963089e-01 -1.02071866e-01 -1.96978241e-01
3.08969289e-01 -5.91250062e-02 3.52223247e-01 2.24520579e-01
2.35776290e-01 -1.01651125e-01 -2.25449324e-01 3.65107834e-01
1.18845236e+00 -3.56865883e-01 1.74279064e-01 1.88226894e-01
7.83755839e-01 -2.89239466e-01 1.55738622e-01 7.79834569e-01
-4.68573689e-01 9.86971498e-01 1.13663226e-01 1.82641065e-03
-9.34114456e-01 -1.34811306e+00 -1.38244599e-01 1.23870671e+00
-2.79694378e-01 -5.89783251e-01 -7.21392989e-01 -4.11179692e-01
-3.17868441e-01 1.09852397e+00 -1.88113227e-01 -2.57432997e-01
-3.90561312e-01 -9.71006081e-02 1.06185281e+00 5.92326701e-01
6.82318434e-02 -1.12111020e+00 -7.62083977e-02 -8.03336874e-03
-6.50616735e-02 -1.21638584e+00 -3.55506688e-01 1.56190723e-01
-4.00681466e-01 -5.91464162e-01 -7.33473003e-01 -5.71356058e-01
2.61042356e-01 5.51041484e-01 8.28641593e-01 -1.92667961e-01
-3.03341419e-01 5.72027266e-01 -5.15965104e-01 -3.86689216e-01
-1.20975125e+00 -2.31626898e-01 6.08912259e-02 6.25165105e-02
1.78926796e-01 -2.58585155e-01 -2.49339327e-01 4.81926560e-01
-5.41257799e-01 -3.91753018e-02 4.22690034e-01 8.40175748e-01
4.63049144e-01 -3.41363460e-01 8.52650881e-01 -2.67156303e-01
2.05673933e-01 -8.42166543e-02 -5.64971328e-01 1.01531610e-01
-4.06524032e-01 -2.32598871e-01 4.41770136e-01 -5.69017768e-01
-1.18783915e+00 1.19293205e-01 -3.58823568e-01 -7.17610717e-01
-5.70406139e-01 1.64713740e-01 -1.40809044e-01 -6.26442209e-02
9.04113472e-01 4.53724504e-01 -2.31551472e-02 -5.88541389e-01
4.46741581e-01 1.50742018e+00 6.55896366e-01 1.91567577e-02
6.18197024e-01 2.01969266e-01 -3.52953136e-01 -1.54354799e+00
-2.90376455e-01 -8.58224034e-01 -3.31769794e-01 -4.31293517e-01
4.87806439e-01 -1.26175511e+00 -4.32244688e-01 3.36558670e-01
-1.05747235e+00 -3.81905019e-01 -2.27516279e-01 3.88210148e-01
-5.73618472e-01 5.15337467e-01 -5.25871634e-01 -1.53886056e+00
-3.61450166e-01 -1.30757535e+00 1.30175054e+00 -3.48846138e-01
-2.74847209e-01 -4.04696405e-01 -2.51230374e-02 7.88424730e-01
3.14702868e-01 -4.68857199e-01 3.27770352e-01 -7.68929482e-01
-3.55340391e-01 -1.91501990e-01 -1.91278085e-01 5.42842209e-01
-3.05762477e-02 1.38100132e-01 -1.90223885e+00 -5.31758428e-01
-2.16246232e-01 -4.52121884e-01 1.11322463e+00 1.55765712e-01
9.48321640e-01 -1.29605353e-01 -9.85055715e-02 1.96020156e-01
8.80714059e-01 4.15279239e-01 7.16504097e-01 2.15316951e-01
6.13522887e-01 5.12828112e-01 5.15350282e-01 1.85050741e-01
-3.78326774e-02 1.01989353e+00 4.65814739e-01 -5.24973385e-02
-8.50939572e-01 -3.10619533e-01 9.17826116e-01 8.64960194e-01
3.49437296e-01 -5.46739936e-01 -1.10731721e+00 7.03988135e-01
-1.37405217e+00 -8.18125486e-01 1.17829219e-02 2.32268548e+00
9.02699888e-01 1.30119622e-01 2.48928055e-01 8.10993135e-01
7.30678499e-01 5.30405581e-01 -1.58183947e-01 -4.99441952e-01
-2.01188847e-01 1.82294473e-01 2.88366914e-01 7.70264089e-01
-1.00969768e+00 1.03477252e+00 6.41290092e+00 1.08216488e+00
-1.30780411e+00 1.45372018e-01 4.23171669e-01 -3.83454293e-01
2.17444878e-02 -3.54566753e-01 -6.94672585e-01 1.75893679e-01
1.64402974e+00 6.78802058e-02 5.85105121e-01 7.18832195e-01
5.69013655e-01 6.11013547e-02 -9.94409740e-01 1.39421999e+00
2.99277842e-01 -1.02913892e+00 -1.05549194e-01 -1.21279523e-01
1.28012016e-01 3.44222903e-01 2.78859377e-01 3.08011472e-01
1.21547505e-01 -1.31940842e+00 1.08554602e+00 9.73249376e-02
1.10807061e+00 -7.77206242e-01 4.66751724e-01 2.82201886e-01
-1.36322153e+00 2.44717300e-02 -1.35138750e-01 3.46449882e-01
1.51631877e-01 1.72904804e-01 -1.33235240e+00 1.95530534e-01
9.42033291e-01 4.73641008e-01 -5.04565537e-01 9.16880906e-01
-3.44140142e-01 1.25574338e+00 -2.34000415e-01 2.27717131e-01
1.43359706e-01 6.31518424e-01 8.99477780e-01 1.61403608e+00
1.32583797e-01 -2.47865215e-01 2.78805681e-02 4.54354078e-01
1.23782698e-02 1.56578124e-01 -6.49328768e-01 -1.95843980e-01
6.23352170e-01 7.92212367e-01 -1.98607236e-01 -1.59495771e-01
-4.00617301e-01 7.28904665e-01 9.11600292e-02 5.12805402e-01
-4.13862079e-01 -3.44616562e-01 5.09174526e-01 1.62006572e-01
2.74355292e-01 -2.18332231e-01 -9.10815075e-02 -6.78451598e-01
-5.95485605e-03 -1.16635752e+00 2.64411718e-01 -1.13020587e+00
-8.54124129e-01 7.88082182e-01 -2.86774009e-01 -1.18246543e+00
-3.55875552e-01 -5.91194034e-01 -3.47338229e-01 9.97681558e-01
-1.56960058e+00 -9.94359732e-01 -7.05468729e-02 4.85976070e-01
1.09912336e+00 -6.98233902e-01 8.81120384e-01 2.26415396e-01
-3.56683165e-01 9.85646725e-01 1.19165152e-01 1.06734239e-01
8.96599174e-01 -1.11224830e+00 6.60771847e-01 8.28863919e-01
5.64292610e-01 2.19397381e-01 8.15394402e-01 -2.68013686e-01
-1.43468595e+00 -1.04876995e+00 9.65009570e-01 -4.09490585e-01
6.43872619e-01 -7.02318668e-01 -1.23545682e+00 5.17861664e-01
2.38349169e-01 3.81687582e-01 6.43092453e-01 -1.35962814e-01
-8.11577618e-01 -2.82820433e-01 -1.05449164e+00 4.76533741e-01
7.63433218e-01 -1.18206263e+00 -6.94888473e-01 7.22705796e-02
1.06189883e+00 -4.09874260e-01 -4.61391181e-01 1.13109037e-01
2.95787662e-01 -6.78597450e-01 1.16632700e+00 -3.80622298e-01
-1.71498075e-01 -5.80556154e-01 -4.61994469e-01 -1.36944318e+00
-2.45627705e-02 -8.21738720e-01 -2.09505334e-01 1.45183265e+00
5.21801353e-01 -4.37757641e-01 4.54273105e-01 -1.51866928e-01
-5.16297519e-01 -1.13966525e-01 -1.49198234e+00 -1.02203560e+00
-3.54259402e-01 -1.02020788e+00 7.14124143e-02 6.30555868e-01
4.12748829e-02 3.25984687e-01 -4.24520850e-01 3.67699862e-01
3.38710368e-01 -6.47382796e-01 7.20860362e-01 -8.30884218e-01
-2.89025366e-01 -2.42167816e-01 -6.23548508e-01 -9.20788169e-01
2.00570211e-01 -7.91563988e-01 2.94676512e-01 -1.40348911e+00
-3.08386058e-01 -3.27950530e-02 -3.58699024e-01 7.22513616e-01
3.04744899e-01 2.86155939e-01 4.06726271e-01 1.34273857e-01
-5.11970699e-01 6.28456771e-01 6.81201458e-01 -5.45371890e-01
-3.70955914e-01 -1.24065531e-02 -3.16047937e-01 3.69216651e-01
6.71362460e-01 -4.69289929e-01 -4.91550267e-01 -3.14552009e-01
-3.28633547e-01 1.86756015e-01 4.18851674e-01 -9.48205709e-01
3.81840289e-01 2.16529518e-01 1.38230771e-01 -7.59684801e-01
7.90385604e-01 -6.16030455e-01 -7.99962282e-02 8.48905519e-02
-5.75847507e-01 -3.04482758e-01 6.19441628e-01 7.27343142e-01
-4.71491575e-01 -4.58040424e-02 9.91238594e-01 2.27783918e-01
-6.40174508e-01 -3.07060510e-01 -9.33559060e-01 -5.63206570e-03
6.63177907e-01 -1.10077940e-01 -5.48259497e-01 -6.49256468e-01
-8.34090352e-01 -2.73402750e-01 2.90796254e-02 6.46979630e-01
9.66191828e-01 -9.84460592e-01 -8.75738502e-01 3.15406889e-01
5.26175380e-01 -2.26296440e-01 2.82866180e-01 5.27583480e-01
-2.80649602e-01 6.61911011e-01 3.67681116e-01 -9.60760236e-01
-1.98250163e+00 4.42605942e-01 4.07572359e-01 3.83889496e-01
-7.26011097e-01 8.17497492e-01 1.08978458e-01 -1.46820635e-01
7.71992922e-01 -4.02010250e-04 -2.82070547e-01 -3.02661564e-02
9.21638548e-01 5.48210442e-01 5.78505099e-01 -1.12499797e+00
-5.61164081e-01 1.66672692e-01 -2.21519083e-01 -6.58662558e-01
1.03180599e+00 -8.30883235e-02 4.99220431e-01 7.94315100e-01
1.14804375e+00 1.00743055e-01 -1.12301815e+00 -3.69108796e-01
-2.19790190e-01 -4.44555640e-01 6.63675010e-01 -1.06489408e+00
-8.91927838e-01 1.31034136e+00 8.73285115e-01 2.80578882e-01
9.30455923e-01 2.81167984e-01 3.09946895e-01 3.98379415e-01
1.10105865e-01 -9.71446514e-01 1.85977101e-01 5.21386206e-01
1.28903913e+00 -1.05258167e+00 -2.64804423e-01 -4.44442153e-01
-8.12819898e-01 1.07090914e+00 3.02731633e-01 4.15287703e-01
3.92827123e-01 4.31780279e-01 6.65300906e-01 1.56890467e-01
-9.42174792e-01 -2.00373903e-01 5.17074168e-01 1.01506090e+00
4.68335152e-01 -1.12913421e-03 5.97501993e-01 2.67020375e-01
-2.67721266e-01 -4.07300383e-01 3.38710874e-01 8.48107398e-01
-6.55032754e-01 -8.39063406e-01 -7.59651721e-01 -1.59497932e-01
-3.73867452e-01 -2.63281465e-01 -6.83951914e-01 4.72480625e-01
-6.56123102e-01 1.69192326e+00 1.61899850e-02 -3.67338419e-01
5.29889643e-01 1.44707620e-01 7.45271146e-02 -7.65979886e-01
-4.52119619e-01 5.83158910e-01 3.35169047e-01 -4.17456061e-01
-2.22545445e-01 -7.08976388e-01 -1.15451670e+00 -9.07185581e-03
-3.26057434e-01 -7.88109004e-02 8.56036365e-01 8.03421557e-01
2.33859763e-01 7.92778850e-01 6.66621625e-01 -7.32524157e-01
-4.82311875e-01 -1.03704166e+00 -4.07120407e-01 -1.80386558e-01
8.32030833e-01 -4.77779001e-01 -8.56618345e-01 7.00796396e-02] | [14.371739387512207, 5.172881126403809] |
c52397e0-fc74-4a3a-b504-124a4ed37a39 | a-model-to-search-for-synthesizable-molecules | 1906.05221 | null | https://arxiv.org/abs/1906.05221v2 | https://arxiv.org/pdf/1906.05221v2.pdf | A Model to Search for Synthesizable Molecules | Deep generative models are able to suggest new organic molecules by generating strings, trees, and graphs representing their structure. While such models allow one to generate molecules with desirable properties, they give no guarantees that the molecules can actually be synthesized in practice. We propose a new molecule generation model, mirroring a more realistic real-world process, where (a) reactants are selected, and (b) combined to form more complex molecules. More specifically, our generative model proposes a bag of initial reactants (selected from a pool of commercially-available molecules) and uses a reaction model to predict how they react together to generate new molecules. We first show that the model can generate diverse, valid and unique molecules due to the useful inductive biases of modeling reactions. Furthermore, our model allows chemists to interrogate not only the properties of the generated molecules but also the feasibility of the synthesis routes. We conclude by using our model to solve retrosynthesis problems, predicting a set of reactants that can produce a target product. | ['José Miguel Hernández-Lobato', 'Marwin H. S. Segler', 'Matt J. Kusner', 'Brooks Paige', 'John Bradshaw'] | 2019-06-12 | a-model-to-search-for-synthesizable-molecules-1 | http://papers.nips.cc/paper/9007-a-model-to-search-for-synthesizable-molecules | http://papers.nips.cc/paper/9007-a-model-to-search-for-synthesizable-molecules.pdf | neurips-2019-12 | ['retrosynthesis'] | ['medical'] | [ 6.84163868e-01 4.47218359e-01 -2.78573781e-01 -2.47284435e-02
-3.29695374e-01 -1.17151415e+00 1.00305223e+00 4.56335634e-01
1.82676703e-01 1.04225457e+00 2.92488426e-01 -4.85300481e-01
3.91942799e-01 -1.37028658e+00 -9.31189299e-01 -7.33642578e-01
5.95844164e-02 7.08476841e-01 -8.90782028e-02 -3.28029960e-01
2.34836057e-01 8.29096019e-01 -1.30996168e+00 2.76936591e-01
9.39697087e-01 3.49055201e-01 3.07895333e-01 6.73509061e-01
-1.20067805e-01 6.48381114e-01 -4.86784279e-01 -4.72990155e-01
2.74052948e-01 -8.52451265e-01 -7.63834357e-01 -2.07044512e-01
-1.18555576e-01 -4.05459031e-02 -1.55773953e-01 7.77951837e-01
2.35991791e-01 3.72283220e-01 1.19855142e+00 -7.85055518e-01
-5.76206982e-01 1.05051529e+00 2.16070324e-01 -4.89939213e-01
4.33306813e-01 4.41307217e-01 1.13448799e+00 -7.51932859e-01
1.10263407e+00 1.14547157e+00 1.06642574e-01 8.95266294e-01
-1.50971889e+00 -7.15366900e-01 2.05404207e-01 -3.13788772e-01
-8.60778332e-01 -5.78385770e-01 5.97444177e-01 -5.15968502e-01
9.46800232e-01 3.93395156e-01 1.09249604e+00 1.35182500e+00
2.87366956e-01 3.41790855e-01 5.54309249e-01 -2.55693436e-01
6.39745593e-01 2.39748865e-01 -3.89521718e-01 6.86620831e-01
2.33409762e-01 3.72524500e-01 -5.51819623e-01 -2.67330348e-01
3.33423734e-01 1.00526303e-01 -1.18189141e-01 -3.55476230e-01
-1.21689594e+00 1.06743693e+00 6.76256537e-01 3.40784848e-01
-5.21604776e-01 1.66785926e-01 -9.34758857e-02 -2.89827228e-01
1.49628103e-01 1.49679101e+00 -2.67526239e-01 4.74059343e-01
-7.38535583e-01 4.78356481e-01 9.36592042e-01 9.90300894e-01
1.01311564e+00 1.15175471e-01 -6.76540136e-02 1.46304712e-01
1.30846784e-01 7.53749683e-02 8.07398036e-02 -5.90101063e-01
1.80576563e-01 3.74337018e-01 3.08991492e-01 -5.26710868e-01
-3.38242710e-01 -4.40048367e-01 -5.05663395e-01 -2.39741400e-01
1.44355446e-01 -3.14531028e-01 -1.13483763e+00 1.81688201e+00
1.87469900e-01 -1.49624839e-01 2.74933457e-01 5.36659062e-01
7.23068714e-01 1.38906229e+00 4.34414685e-01 -4.05603439e-01
6.67336881e-01 -8.80881429e-01 -1.93272546e-01 -5.86009547e-02
8.26057673e-01 -6.18172467e-01 4.62830722e-01 4.92198497e-01
-1.25083005e+00 -3.85265946e-01 -1.10162890e+00 9.48423520e-02
-6.61411524e-01 -2.18075559e-01 1.28294730e+00 7.69575775e-01
-7.25965440e-01 1.19384694e+00 -4.34356242e-01 -8.40323865e-02
3.19432050e-01 4.86856014e-01 -1.68271735e-01 3.12940055e-03
-1.09955788e+00 7.80080438e-01 8.59544575e-01 1.43792272e-01
-1.52798998e+00 -7.45665610e-01 -7.77169704e-01 7.54677807e-04
4.01820838e-01 -1.22913253e+00 1.08179760e+00 -8.28894794e-01
-1.67584276e+00 3.24668348e-01 -3.34010214e-01 -3.47580850e-01
2.33982801e-01 3.96943420e-01 -3.15762013e-01 -1.89321280e-01
-1.79496050e-01 1.06372142e+00 4.51414675e-01 -1.25516331e+00
-4.10176605e-01 -4.49866913e-02 1.69521630e-01 1.05428651e-01
2.58553416e-01 -3.66003960e-01 2.84251004e-01 -2.84474850e-01
5.70371887e-03 -1.09090149e+00 -8.83427501e-01 -4.17375863e-01
-1.02328038e+00 1.00178299e-02 -8.91144350e-02 -1.23292832e-02
8.58071864e-01 -1.49470127e+00 5.64700007e-01 7.10466921e-01
2.53210634e-01 2.69334577e-03 -2.89712727e-01 9.65055585e-01
-3.91399980e-01 7.52818704e-01 -1.83690012e-01 2.26805165e-01
-6.65884390e-02 -1.76324338e-01 -4.75195557e-01 3.04890480e-02
3.89183581e-01 9.79103386e-01 -1.28504312e+00 2.09791407e-01
1.90546185e-01 2.55817801e-01 -7.80263245e-01 1.55235589e-01
-9.81976509e-01 6.32963657e-01 -6.09162092e-01 5.06181359e-01
2.63999403e-01 -1.61565468e-01 6.29288912e-01 -2.59967834e-01
-2.56476462e-01 6.71112537e-01 -7.56538868e-01 1.43375623e+00
-4.13092285e-01 1.14554726e-01 -7.52644837e-01 -5.52356243e-01
1.08787429e+00 1.75005406e-01 1.90301895e-01 -1.91512972e-01
-3.53513435e-02 3.19526911e-01 2.71234512e-01 -3.93538848e-02
6.04937613e-01 -7.77384877e-01 5.18902466e-02 4.19071466e-01
1.08195096e-01 -7.06499755e-01 5.98863840e-01 3.09850782e-01
7.48234630e-01 2.37724081e-01 3.82069260e-01 -1.42255435e-02
2.61988252e-01 6.58987239e-02 3.40298146e-01 7.77243495e-01
5.75283647e-01 3.88884217e-01 4.60178226e-01 -5.08032501e-01
-1.43241048e+00 -1.01105976e+00 3.59598845e-01 7.90529966e-01
-4.79575545e-02 -6.76137388e-01 -4.33205485e-01 -6.13081098e-01
-1.98949069e-01 9.36698258e-01 -6.93830431e-01 -3.47919226e-01
-3.62067342e-01 -8.88982117e-01 1.81723181e-02 2.09744871e-01
-2.50809759e-01 -9.62429106e-01 -8.03022552e-03 4.64196563e-01
1.91011429e-01 -4.60066617e-01 -4.10348147e-01 5.08038521e-01
-6.64390862e-01 -9.63471889e-01 -4.77653354e-01 -6.33759677e-01
8.56278837e-01 -1.20530032e-01 9.66478407e-01 -2.67643064e-01
-3.06384265e-01 -2.85140872e-01 4.25066911e-02 -6.30144060e-01
-1.16737056e+00 2.56248057e-01 -1.25900075e-01 -2.23615199e-01
-3.07080299e-01 -5.77205479e-01 -6.03753269e-01 -2.67604947e-01
-7.92704105e-01 3.99141282e-01 4.95281756e-01 6.55202210e-01
6.10046983e-01 9.65549722e-02 5.77035606e-01 -1.24237251e+00
4.36090618e-01 -7.33153224e-01 -5.69501758e-01 4.38694447e-01
-6.24637425e-01 6.51154518e-01 9.79684114e-01 -5.71079016e-01
-1.08034229e+00 4.30104434e-01 -1.27211764e-01 3.55362564e-01
-5.92005253e-02 4.91157919e-01 -4.05015647e-01 1.19243436e-01
7.96247602e-01 3.19746584e-01 -4.47362572e-01 -1.58934101e-01
9.79593515e-01 -3.39270122e-02 1.52380258e-01 -8.89436781e-01
7.62213171e-01 1.48565933e-01 5.03460705e-01 -6.08513236e-01
-5.23821056e-01 1.15578935e-01 -3.64204854e-01 6.99995235e-02
6.61748230e-01 -7.35135257e-01 -8.90736103e-01 -1.94017977e-01
-1.24624825e+00 -3.08430105e-01 -3.94312084e-01 2.80461520e-01
-4.30988461e-01 1.07063763e-01 -4.43105459e-01 -8.77196789e-01
-3.04818958e-01 -1.19494295e+00 6.79550648e-01 2.36111641e-01
-5.51101029e-01 -9.05019283e-01 3.04620005e-02 -9.14874151e-02
6.13695495e-02 4.54900920e-01 1.43073606e+00 -8.17767143e-01
-1.06700253e+00 7.40391687e-02 3.99369717e-01 -3.11426640e-01
4.21524167e-01 4.00170505e-01 -6.92248702e-01 -2.09532604e-02
-6.26198351e-01 -1.35503158e-01 9.28672552e-01 1.61479935e-01
1.01607728e+00 -6.13778055e-01 -7.56003261e-01 4.89708126e-01
1.06619132e+00 8.45044136e-01 6.41018152e-01 2.14972496e-02
7.00978756e-01 7.80762374e-01 4.04087082e-02 3.35223109e-01
-4.45824163e-03 3.54962438e-01 3.31596196e-01 -1.28392130e-02
2.48803779e-01 -1.02642608e+00 3.33129704e-01 3.47598851e-01
-2.55424619e-01 -8.11393201e-01 -5.75734377e-01 2.49023855e-01
-1.40706706e+00 -1.06276691e+00 -1.08533889e-01 2.21174836e+00
1.04887998e+00 1.27375290e-01 4.95987415e-01 -6.71501756e-02
5.53485334e-01 -3.79894488e-02 -7.89426267e-01 -5.42377651e-01
-1.26666784e-01 9.13533390e-01 5.66875696e-01 6.17338955e-01
-6.82640910e-01 1.15885663e+00 7.79300213e+00 5.74800849e-01
-1.33413982e+00 -5.44315934e-01 8.89322639e-01 -8.09622407e-02
-1.15070891e+00 6.23740077e-01 -8.17527294e-01 3.00442725e-01
9.36024487e-01 -6.01934373e-01 6.69685066e-01 4.50634152e-01
4.45015617e-02 2.72080809e-01 -1.59836757e+00 4.39928949e-01
-3.10697287e-01 -2.10185695e+00 8.70342016e-01 -6.98045548e-03
9.07067239e-01 -7.09496737e-01 1.61394309e-02 1.53951282e-02
6.77168310e-01 -1.38597167e+00 8.96814466e-01 5.01971424e-01
4.48686630e-01 -8.57772529e-01 3.31296064e-02 1.83383435e-01
-7.71046817e-01 -4.19214293e-02 -1.83320358e-01 -1.11617357e-03
3.94077636e-02 5.08653760e-01 -1.54722691e+00 6.53479457e-01
-4.40141797e-01 6.08969688e-01 -2.20655397e-01 5.91099858e-01
-4.67135936e-01 4.45541352e-01 -2.90401489e-01 -6.35961711e-01
3.68078947e-01 -5.83185434e-01 2.22441182e-01 1.08232462e+00
6.86293602e-01 8.61987695e-02 -1.93669926e-02 1.54280782e+00
-3.18890810e-01 1.52779892e-01 -8.59547615e-01 -7.57190049e-01
5.41040003e-01 1.04134703e+00 -6.82909966e-01 -4.24472272e-01
1.98157370e-01 6.32128716e-01 3.04014850e-02 6.42581701e-01
-5.49690306e-01 -1.54312953e-01 4.13823456e-01 4.30654176e-02
1.96083888e-01 -1.50997669e-01 4.85640801e-02 -8.91345620e-01
-6.04508638e-01 -9.47931647e-01 5.22279553e-02 -8.05966794e-01
-9.59549725e-01 3.74447733e-01 -3.42407554e-01 -4.85176623e-01
-2.19005972e-01 -4.40346062e-01 -6.64011776e-01 1.00036204e+00
-1.09315872e+00 -8.97829115e-01 2.84523576e-01 -1.69954091e-01
4.39003706e-01 -8.93664658e-02 9.55571175e-01 -3.33574265e-01
-5.77387333e-01 8.29392895e-02 -1.38932914e-01 -5.98914981e-01
5.47866166e-01 -1.15906918e+00 7.07026184e-01 6.29318535e-01
2.57949740e-01 1.04905784e+00 9.45594966e-01 -7.86022246e-01
-1.46236110e+00 -1.17586350e+00 7.55445302e-01 -3.94901305e-01
3.92275035e-01 -6.50913596e-01 -3.77471894e-01 5.55852234e-01
-1.63551182e-01 -5.82080305e-01 8.29401135e-01 1.11680627e-01
-3.64984185e-01 9.45602283e-02 -1.01335752e+00 1.09080696e+00
1.18842685e+00 -2.43605345e-01 -7.27831153e-03 6.52796149e-01
8.76666188e-01 -2.49116912e-01 -8.75962496e-01 1.20556735e-01
6.08532846e-01 -7.18756199e-01 1.06616867e+00 -1.13796401e+00
8.62042904e-01 -2.92917490e-01 5.53519232e-03 -1.53537107e+00
-4.73645568e-01 -1.13340950e+00 -5.35284495e-03 7.84433544e-01
1.21546757e+00 -6.02287650e-01 7.12305129e-01 8.42945695e-01
-5.04505672e-02 -6.28351390e-01 -1.22389145e-01 -6.42330825e-01
2.68492728e-01 5.33474274e-02 1.21284449e+00 7.30185390e-01
2.90167719e-01 8.59117746e-01 -3.26133132e-01 -1.58148095e-01
1.67302012e-01 3.61951113e-01 5.99148393e-01 -1.09262431e+00
-4.11340445e-01 -5.27252734e-01 1.40225992e-01 -1.14539170e+00
3.51074412e-02 -1.31757951e+00 -2.80176941e-02 -1.55503571e+00
3.25759977e-01 -8.27328563e-01 -4.63994686e-03 2.40978658e-01
-8.69609416e-02 -3.15868825e-01 3.34068775e-01 -3.27163711e-02
-1.47246420e-01 5.40180147e-01 1.31653047e+00 -3.49134535e-01
-5.07505834e-01 2.92772055e-03 -1.23976886e+00 2.12466851e-01
7.76163816e-01 -4.47367132e-01 -4.52031761e-01 2.29916781e-01
8.46435249e-01 2.60960102e-01 -8.41854885e-02 -5.89967370e-01
-1.38841763e-01 -6.10340178e-01 6.09610915e-01 -2.23041579e-01
3.30288947e-01 -2.27353171e-01 9.87933576e-01 6.17038488e-01
-7.82322884e-01 -4.88659054e-01 -1.04430700e-02 5.26942790e-01
3.58378619e-01 -5.04175723e-01 4.23754424e-01 -5.95146775e-01
-1.13373451e-01 6.96796060e-01 -6.41422927e-01 -5.61032414e-01
8.53007913e-01 -6.35972843e-02 -3.18680048e-01 -3.82969737e-01
-8.08612883e-01 -8.74536708e-02 6.53466702e-01 2.20820308e-01
3.36761773e-01 -9.67820585e-01 -3.81227523e-01 -5.51817268e-02
1.65440172e-01 4.00262028e-02 -1.25139952e-01 -5.78494556e-02
-5.65610409e-01 5.53029835e-01 1.03620715e-01 -2.24693920e-02
-8.14886630e-01 1.02548873e+00 3.50786179e-01 -1.15245387e-01
1.47676067e-02 8.38011622e-01 6.42180026e-01 -4.09662992e-01
-2.27919579e-01 -5.13504684e-01 1.77704304e-01 -5.62831685e-02
2.00666919e-01 9.21001006e-03 -9.52113271e-02 -4.62027732e-03
-1.51232362e-01 1.81626454e-01 -1.97896659e-01 7.24511892e-02
1.31812060e+00 5.05379021e-01 -7.26617081e-03 2.68387228e-01
7.94368684e-01 1.57399639e-01 -9.38577116e-01 2.48624802e-01
-2.64248073e-01 5.92587776e-02 -3.60817492e-01 -8.94829750e-01
-4.42584246e-01 7.52458453e-01 -9.99847874e-02 -2.64736474e-03
6.02459490e-01 -1.00236572e-01 7.33945131e-01 7.07545400e-01
2.87322253e-01 -7.29695559e-01 1.51949570e-01 2.07101673e-01
7.25424469e-01 -7.08696127e-01 1.67604446e-01 -6.01334214e-01
-3.69525075e-01 1.39167976e+00 1.80315286e-01 2.82231987e-01
3.23980935e-02 -8.07679966e-02 -5.53271651e-01 -1.33722842e-01
-9.70307648e-01 -1.22924022e-01 2.72602569e-02 7.08402097e-01
6.90131962e-01 3.07580918e-01 -1.90442517e-01 2.40576863e-01
-4.35009331e-01 -2.42701218e-01 6.22463286e-01 6.35937691e-01
-4.12957609e-01 -1.61738551e+00 -2.01349054e-02 1.77057952e-01
-4.82575670e-02 -3.94390881e-01 -9.61966574e-01 3.14892501e-01
4.24882740e-01 9.53522623e-01 -3.77969712e-01 -2.77212739e-01
1.08304232e-01 1.72748700e-01 9.57577407e-01 -9.79675293e-01
-6.33792043e-01 -1.84432462e-01 5.17413318e-01 -1.19865283e-01
-1.56935930e-01 -4.21648204e-01 -1.09293151e+00 -3.36432815e-01
-4.37588930e-01 6.14408731e-01 7.51899958e-01 6.72432542e-01
4.31503832e-01 4.90932733e-01 9.78146374e-01 -8.46235693e-01
-5.44879995e-02 -4.33126032e-01 -3.63625556e-01 1.74362183e-01
6.70538023e-02 -3.31753790e-01 -1.18345879e-01 1.71246991e-01] | [4.541945934295654, 6.071860313415527] |
8d211068-5f5a-4d5d-bbd3-3350297db420 | henry-core-domain-adaptation-and-stacking-for | null | null | https://aclanthology.org/S13-1013 | https://aclanthology.org/S13-1013.pdf | HENRY-CORE: Domain Adaptation and Stacking for Text Similarity | null | ['Michael Heilman', 'Nitin Madnani'] | 2013-06-01 | null | null | null | semeval-2013-6 | ['video-description'] | ['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] | [-7.20218563079834, 3.8159279823303223] |
e3e8e002-e4fe-4dd5-a0c2-b8a881c923ee | exploring-intra-and-inter-video-relation-for | 2203.15251 | null | https://arxiv.org/abs/2203.15251v2 | https://arxiv.org/pdf/2203.15251v2.pdf | Exploring Intra- and Inter-Video Relation for Surgical Semantic Scene Segmentation | Automatic surgical scene segmentation is fundamental for facilitating cognitive intelligence in the modern operating theatre. Previous works rely on conventional aggregation modules (e.g., dilated convolution, convolutional LSTM), which only make use of the local context. In this paper, we propose a novel framework STswinCL that explores the complementary intra- and inter-video relations to boost segmentation performance, by progressively capturing the global context. We firstly develop a hierarchy Transformer to capture intra-video relation that includes richer spatial and temporal cues from neighbor pixels and previous frames. A joint space-time window shift scheme is proposed to efficiently aggregate these two cues into each pixel embedding. Then, we explore inter-video relation via pixel-to-pixel contrastive learning, which well structures the global embedding space. A multi-source contrast training objective is developed to group the pixel embeddings across videos with the ground-truth guidance, which is crucial for learning the global property of the whole data. We extensively validate our approach on two public surgical video benchmarks, including EndoVis18 Challenge and CaDIS dataset. Experimental results demonstrate the promising performance of our method, which consistently exceeds previous state-of-the-art approaches. Code is available at https://github.com/YuemingJin/STswinCL. | ['Danail Stoyanov', 'Pheng-Ann Heng', 'Zixu Zhao', 'Cheng Chen', 'Yang Yu', 'Yueming Jin'] | 2022-03-29 | null | null | null | null | ['scene-segmentation'] | ['computer-vision'] | [ 2.82361895e-01 2.39793789e-02 -5.32987416e-01 -2.77893662e-01
-7.62527883e-01 -3.25875640e-01 3.74289572e-01 2.32112497e-01
-7.01411009e-01 4.53542292e-01 5.85160077e-01 -2.39961892e-01
-2.70292640e-01 -5.56049228e-01 -6.52732551e-01 -9.48731065e-01
7.15259388e-02 -5.44740081e-01 1.90505534e-01 -1.42861754e-01
2.04665154e-01 1.19888589e-01 -1.09691751e+00 5.76917648e-01
7.60498464e-01 1.11292076e+00 4.81195956e-01 4.68632042e-01
-1.49630025e-01 8.28265429e-01 -6.42489567e-02 2.48456206e-02
2.41288617e-01 -4.92188424e-01 -8.28879654e-01 7.83075765e-02
2.57189274e-01 -2.78701305e-01 -5.85279465e-01 1.16823256e+00
4.78131771e-01 1.41192034e-01 5.40008023e-02 -7.60269463e-01
-6.28170907e-01 5.85391879e-01 -5.90611577e-01 4.69552398e-01
1.96949616e-02 3.94337744e-01 9.02067125e-01 -7.79890180e-01
7.73794115e-01 8.83954287e-01 4.97388005e-01 3.84701252e-01
-9.44275677e-01 -6.40587568e-01 5.58577120e-01 1.67304173e-01
-1.12324524e+00 -4.07567993e-02 9.46988523e-01 -4.22910333e-01
5.60800850e-01 2.22536266e-01 9.26477134e-01 9.57471013e-01
6.95190907e-01 9.52874541e-01 1.11940515e+00 -2.34559879e-01
-1.69090971e-01 -2.17609331e-01 1.35241568e-01 1.00600910e+00
-1.22822642e-01 2.99343348e-01 -5.74883401e-01 2.99011797e-01
1.04505754e+00 5.53369522e-01 -7.17457533e-01 -5.40334642e-01
-1.74480402e+00 7.48363495e-01 9.90322053e-01 5.77222645e-01
-4.33562547e-01 2.59365559e-01 5.61155438e-01 -1.47318598e-02
2.11043850e-01 3.88062686e-01 -2.96401560e-01 -5.70791923e-02
-6.69735372e-01 -3.06249678e-01 3.66744220e-01 8.72183979e-01
8.01900804e-01 -3.05055261e-01 -4.48065758e-01 5.90000093e-01
1.53675765e-01 -2.46914163e-01 8.42838347e-01 -8.51693988e-01
3.47792298e-01 6.34719193e-01 -3.06020558e-01 -8.32311571e-01
-5.21871388e-01 -6.82079732e-01 -8.45766366e-01 9.38275158e-02
1.02346428e-01 -2.47846395e-02 -1.01937497e+00 1.54284346e+00
4.47592974e-01 7.34395146e-01 1.04481429e-01 8.83046806e-01
1.06132138e+00 3.67070138e-01 1.20411143e-01 -1.95851609e-01
1.40974438e+00 -1.51582003e+00 -7.74545372e-01 -1.71666697e-01
9.13876235e-01 -5.99387944e-01 1.10373151e+00 1.30771384e-01
-1.01402402e+00 -6.32422686e-01 -1.01394272e+00 -3.05535316e-01
-2.54883528e-01 1.34670809e-01 7.07955897e-01 1.27319559e-01
-1.00868428e+00 4.16614652e-01 -1.14706016e+00 -8.50965530e-02
6.20369375e-01 3.96763831e-01 -4.67387289e-01 -2.87265480e-01
-1.06662452e+00 5.60551047e-01 6.22254610e-01 3.95879537e-01
-8.20959032e-01 -9.61274624e-01 -1.17710340e+00 -1.17171451e-01
4.68612790e-01 -6.42464638e-01 1.22238660e+00 -9.64832067e-01
-1.50233757e+00 8.30599368e-01 -2.63503715e-02 -2.33783618e-01
4.66234028e-01 -1.84749499e-01 -1.06959157e-01 3.31190258e-01
-1.20469809e-01 8.07371080e-01 4.61468190e-01 -1.30149233e+00
-7.84446478e-01 -2.73502350e-01 3.41429979e-01 5.19467056e-01
-5.37290931e-01 -3.39253902e-01 -9.22899485e-01 -9.06099677e-01
2.18213588e-01 -9.09988344e-01 -5.44295132e-01 1.82436034e-01
-2.91464090e-01 1.39538040e-02 4.73492533e-01 -7.84643710e-01
1.45391488e+00 -2.36480737e+00 2.87106484e-01 5.13786077e-02
4.54167724e-01 7.72086307e-02 -4.44275998e-02 2.01866820e-01
-2.52577543e-01 -2.00469658e-01 -3.59971166e-01 -3.91199410e-01
-3.88997853e-01 2.86818177e-01 9.22727287e-02 5.53078115e-01
1.64340511e-01 1.13519681e+00 -1.20391357e+00 -6.83639228e-01
3.78593117e-01 5.46365201e-01 -7.96537101e-01 1.02979608e-01
2.29420438e-01 7.28130579e-01 -5.00462949e-01 6.58084691e-01
4.13279325e-01 -2.56665289e-01 5.48023805e-02 -4.62369502e-01
-1.55697122e-01 1.06855616e-01 -6.25024438e-01 2.62733626e+00
-5.84506571e-01 4.46040422e-01 -1.06042419e-02 -1.14894223e+00
6.07060492e-01 6.82316646e-02 6.81190252e-01 -8.10319662e-01
4.15144295e-01 3.41290236e-01 3.29443105e-02 -6.26728654e-01
2.94794768e-01 -4.60730940e-02 -8.74156728e-02 -9.87693965e-02
1.12869456e-01 1.94674432e-01 1.04576670e-01 1.67392418e-01
8.82679343e-01 2.72464097e-01 3.23767871e-01 -3.92087221e-01
5.49542904e-01 -7.48741329e-02 7.69480765e-01 4.46670175e-01
-4.12866414e-01 6.88569725e-01 5.79094768e-01 -4.18498158e-01
-5.21739602e-01 -1.00873733e+00 -5.94405495e-02 7.42896497e-01
7.52032280e-01 -3.00411046e-01 -6.46464884e-01 -8.67045939e-01
-2.20043570e-01 2.45459348e-01 -1.21381080e+00 -4.49336916e-01
-7.51786172e-01 -6.45969570e-01 -5.58373779e-02 8.26764882e-01
3.41606885e-01 -1.12687409e+00 -6.20692074e-01 3.50287497e-01
-2.45289356e-01 -1.16917408e+00 -8.58067393e-01 2.18206763e-01
-9.71488059e-01 -1.01670992e+00 -7.52389967e-01 -1.20931113e+00
7.42398858e-01 4.89655823e-01 8.48716795e-01 2.82626271e-01
-6.66246653e-01 3.59194219e-01 -4.00136799e-01 -8.74802917e-02
1.18293442e-01 -1.67265590e-02 -4.63620991e-01 2.15025693e-02
1.39335409e-01 -4.47197616e-01 -1.14962137e+00 1.45747095e-01
-1.06595516e+00 5.72412252e-01 1.00607562e+00 1.18767893e+00
9.49577510e-01 -4.26877797e-01 8.33651647e-02 -9.10816669e-01
2.52902150e-01 -5.06390095e-01 -1.44567326e-01 2.55402684e-01
-2.49493182e-01 -7.44652599e-02 3.48866850e-01 -3.73308182e-01
-8.75224590e-01 3.40728611e-02 3.32282786e-03 -6.00819409e-01
-4.22611274e-03 8.53213906e-01 5.70277160e-04 -1.01792656e-01
1.42102867e-01 4.25163776e-01 9.76759195e-02 -1.28583938e-01
2.36849591e-01 1.86127663e-01 7.45565236e-01 -4.73139107e-01
3.44794661e-01 5.94014943e-01 -2.51248896e-01 -5.19712150e-01
-9.17999983e-01 -7.64184773e-01 -6.33288503e-01 -1.62767112e-01
1.20884144e+00 -1.06661928e+00 -5.02556503e-01 4.15907174e-01
-8.84628057e-01 -5.58630824e-01 -3.52186918e-01 8.13198149e-01
-5.94677329e-01 4.06297177e-01 -7.83208489e-01 -5.83809018e-02
-1.98277503e-01 -1.60167217e+00 1.03900588e+00 4.26013589e-01
9.95652974e-02 -1.10805357e+00 1.02292702e-01 3.31157565e-01
1.11231297e-01 5.62422276e-01 6.32730365e-01 -2.27773204e-01
-7.66946137e-01 -1.01540536e-01 -4.20207977e-01 4.12533045e-01
3.31875741e-01 -2.06016839e-01 -6.83034837e-01 -4.67475742e-01
-7.01790825e-02 -3.83919314e-03 1.25821948e+00 6.52253151e-01
1.73824155e+00 7.74285989e-03 -4.54031587e-01 1.21516037e+00
1.55743027e+00 1.77632064e-01 6.66452706e-01 4.81721282e-01
1.00002921e+00 5.04977643e-01 7.26236522e-01 1.99198440e-01
6.32025301e-01 4.47642237e-01 5.25078297e-01 -5.35123765e-01
-3.37826222e-01 -1.13871574e-01 2.10053593e-01 1.09026206e+00
-1.24231286e-01 5.12897111e-02 -8.42685282e-01 6.67071998e-01
-2.00238252e+00 -6.69009030e-01 2.45039836e-01 1.85602558e+00
9.64033008e-01 8.42041597e-02 -3.42771620e-01 -1.39243782e-01
4.75706846e-01 3.33833367e-01 -4.64755416e-01 -4.51247916e-02
1.84120193e-01 2.38613784e-01 6.15330517e-01 5.39171815e-01
-1.35467601e+00 7.66735673e-01 4.50976562e+00 8.62204015e-01
-1.30256069e+00 2.67460227e-01 7.82716334e-01 -2.84198582e-01
-3.33822489e-01 -2.51487941e-01 -3.73617500e-01 5.20588577e-01
3.57567996e-01 2.71836463e-02 5.80234081e-02 5.70895553e-01
1.71923563e-01 -4.80874889e-02 -1.02567053e+00 1.08004630e+00
2.23539650e-01 -1.69110835e+00 -1.49126887e-01 2.02111319e-01
9.29376841e-01 -8.27816352e-02 2.41878405e-01 2.52134293e-01
-1.03211459e-02 -9.76079345e-01 4.97531354e-01 6.46070361e-01
7.39990473e-01 -6.84954286e-01 8.48789275e-01 -1.03541836e-01
-1.37784636e+00 -2.60258049e-01 5.53157441e-02 3.68118167e-01
-4.63468991e-02 2.93835878e-01 -4.51118320e-01 8.77261043e-01
7.75587380e-01 1.25332940e+00 -4.16028202e-01 1.16036272e+00
-1.88263625e-01 2.00646564e-01 7.34611303e-02 1.87243283e-01
7.47763693e-01 -7.01659620e-02 2.14627162e-01 1.34958601e+00
2.46879488e-01 2.70911872e-01 4.38523829e-01 4.08755690e-01
-1.82990097e-02 9.53918919e-02 -2.36262873e-01 1.47383794e-01
5.40630743e-02 1.42288625e+00 -9.05438662e-01 -2.74369270e-01
-6.48682117e-01 1.05666280e+00 2.91348934e-01 4.40228224e-01
-8.84035408e-01 -3.43997657e-01 7.66100168e-01 -1.25785023e-01
3.06666046e-01 -3.28934819e-01 -3.70736659e-01 -1.12292397e+00
1.26471013e-01 -6.62922680e-01 6.18711412e-01 -4.68075216e-01
-8.79626513e-01 6.67259037e-01 -1.93440229e-01 -1.61363482e+00
2.39469230e-01 -7.02288747e-01 -7.79886246e-01 6.49482250e-01
-1.91831815e+00 -1.37180376e+00 -8.35046947e-01 6.98524475e-01
6.86999738e-01 1.19024314e-01 6.47459269e-01 2.94417769e-01
-7.60231256e-01 5.81538975e-01 2.78820340e-02 2.42419705e-01
7.71437705e-01 -1.13805747e+00 -1.40327990e-01 9.35944915e-01
-1.20213538e-01 6.66173577e-01 1.78673610e-01 -3.62845689e-01
-1.35043931e+00 -1.01286304e+00 2.91343063e-01 -2.58173943e-01
7.91303277e-01 -2.06338003e-01 -8.12871158e-01 6.74092293e-01
2.83022761e-01 4.34590548e-01 9.49944496e-01 -1.09095283e-01
-2.15065479e-02 -9.26007479e-02 -6.92133069e-01 6.44507468e-01
1.16624403e+00 -5.05865991e-01 -4.15565401e-01 1.17785394e-01
1.09794080e+00 -7.29431212e-01 -1.00771844e+00 6.13117337e-01
5.51842749e-01 -1.04352367e+00 1.09467208e+00 -4.26058471e-01
1.02010489e+00 -3.56783301e-01 -6.00921772e-02 -1.18463087e+00
-3.11751038e-01 -4.44342762e-01 2.51088254e-02 8.28835905e-01
2.37783924e-01 -5.99111617e-01 6.95246100e-01 3.55398953e-01
-8.32399666e-01 -1.40871716e+00 -8.88432205e-01 -3.83123010e-01
4.16904464e-02 -3.72148633e-01 3.27766925e-01 1.08122301e+00
1.29772052e-01 -2.34569743e-01 -1.15662232e-01 1.88377708e-01
4.02767181e-01 2.37306803e-01 4.47028160e-01 -6.15481675e-01
-3.08161646e-01 -7.40964711e-01 -5.83113790e-01 -1.07356298e+00
-8.72449875e-02 -1.11782348e+00 1.26732454e-01 -1.79553533e+00
3.35213482e-01 -3.15876633e-01 -1.08738637e+00 5.46380639e-01
-5.89170337e-01 3.92519861e-01 1.00521013e-01 1.09556235e-01
-6.53058887e-01 7.43835926e-01 1.77125311e+00 -1.14484958e-01
-2.49811947e-01 -4.67842430e-01 -8.49920511e-01 6.35680318e-01
7.34477997e-01 -1.88017100e-01 -4.25867587e-01 -7.35452175e-01
-2.70578563e-01 -1.29542202e-01 4.98319298e-01 -9.86108303e-01
4.07394111e-01 -1.44346908e-01 3.82279247e-01 -4.17343557e-01
2.11612687e-01 -7.54608393e-01 -2.12337047e-01 7.27620065e-01
-3.73574227e-01 -1.03043228e-01 2.74330765e-01 6.24006510e-01
-6.53942108e-01 1.16177842e-01 7.15363741e-01 -4.96364459e-02
-1.01180625e+00 7.57257044e-01 2.12565362e-01 -3.79871652e-02
1.20854783e+00 -2.43835062e-01 -1.66163325e-01 1.48204640e-01
-7.97071040e-01 4.68782455e-01 3.51657927e-01 6.10284150e-01
8.58000398e-01 -1.44985569e+00 -5.09368300e-01 2.59010524e-01
3.48892093e-01 3.19286555e-01 8.10819566e-01 1.57797694e+00
-7.53054440e-01 2.37917677e-01 -2.55769163e-01 -8.34774792e-01
-1.10213363e+00 5.57459295e-01 3.26476187e-01 -4.20594394e-01
-9.58481312e-01 1.23981333e+00 9.21072364e-01 -1.84985921e-01
2.69977808e-01 -8.12804461e-01 -2.22802579e-01 3.59509662e-02
5.77572167e-01 -2.78014749e-01 -1.27736136e-01 -4.25566047e-01
-3.06535095e-01 7.22588122e-01 -2.60453254e-01 2.50476718e-01
1.27542865e+00 -1.73903510e-01 -8.64351541e-02 3.28379244e-01
1.56074774e+00 -2.25479484e-01 -1.60475421e+00 -5.82244873e-01
-3.21198493e-01 -6.25523746e-01 3.94216180e-01 -4.45038885e-01
-1.40525734e+00 9.56788242e-01 7.57393718e-01 -2.49939084e-01
1.39095008e+00 -4.08309735e-02 1.07497251e+00 -9.89196599e-02
1.71483248e-01 -9.24348652e-01 2.91113675e-01 2.19752252e-01
8.30052733e-01 -1.46304703e+00 -1.06197566e-01 -4.45900291e-01
-7.64855623e-01 1.11685729e+00 6.83319509e-01 -1.79327697e-01
8.09226990e-01 1.73553765e-01 1.81308329e-01 -2.49906346e-01
-4.47049260e-01 -2.02255920e-01 4.53354061e-01 1.40834630e-01
4.43208694e-01 5.84148094e-02 -3.89643639e-01 7.55313873e-01
4.06637043e-02 1.02441214e-01 3.02325100e-01 1.13733566e+00
-1.71284556e-01 -9.60948348e-01 1.71446368e-01 3.18861902e-01
-3.94195586e-01 -2.62146920e-01 2.35305145e-01 7.62449622e-01
4.55286890e-01 4.27372128e-01 8.63057747e-02 -3.34631771e-01
2.79964298e-01 -3.90726030e-01 3.15517217e-01 -6.26655579e-01
-5.80738425e-01 3.95086288e-01 -2.46254876e-01 -9.41116869e-01
-5.80069005e-01 -6.59193575e-01 -1.40127146e+00 1.92953229e-01
-2.69960403e-03 -8.47980604e-02 4.42145675e-01 6.72644973e-01
1.25230923e-01 1.23240089e+00 5.43790460e-01 -1.00907087e+00
3.89742944e-03 -6.01247847e-01 -3.22846621e-01 2.37491086e-01
6.30660892e-01 -6.22188330e-01 -7.68353567e-02 9.37345177e-02] | [14.25064468383789, -3.142805814743042] |
03475fa5-8737-4551-b5de-277eb9d2eac5 | probing-the-information-encoded-in-x-vectors | 1909.06351 | null | https://arxiv.org/abs/1909.06351v2 | https://arxiv.org/pdf/1909.06351v2.pdf | Probing the Information Encoded in X-vectors | Deep neural network based speaker embeddings, such as x-vectors, have been shown to perform well in text-independent speaker recognition/verification tasks. In this paper, we use simple classifiers to investigate the contents encoded by x-vector embeddings. We probe these embeddings for information related to the speaker, channel, transcription (sentence, words, phones), and meta information about the utterance (duration and augmentation type), and compare these with the information encoded by i-vectors across a varying number of dimensions. We also study the effect of data augmentation during extractor training on the information captured by x-vectors. Experiments on the RedDots data set show that x-vectors capture spoken content and channel-related information, while performing well on speaker verification tasks. | ['Sanjeev Khudanpur', 'Daniel Povey', 'David Snyder', 'Desh Raj'] | 2019-09-13 | null | null | null | null | ['text-independent-speaker-recognition'] | ['speech'] | [ 1.00990281e-01 -9.13521349e-02 -1.85181364e-01 -6.80205286e-01
-7.97273993e-01 -6.03475213e-01 8.83433044e-01 3.08081329e-01
-4.31762606e-01 2.32967123e-01 9.55267966e-01 -5.84276259e-01
2.86301404e-01 -1.88368767e-01 -4.03552532e-01 -6.92342401e-01
-3.93157601e-01 4.12240215e-02 -5.17343879e-01 -4.12267596e-02
1.18286878e-01 4.27427620e-01 -1.54028296e+00 1.25221521e-01
7.89245144e-02 1.05182922e+00 -3.53358537e-01 8.54444802e-01
-2.97460824e-01 4.14212197e-01 -1.07602036e+00 -1.95444912e-01
-2.08903223e-01 -2.48308703e-01 -6.23167336e-01 8.45615268e-02
4.25319135e-01 -4.60341811e-01 -8.33783746e-01 5.47141552e-01
7.44686842e-01 5.68922758e-02 7.46117413e-01 -1.29855084e+00
-7.33370185e-01 1.07963264e+00 -2.25176916e-01 5.32851040e-01
4.07955825e-01 7.29768276e-02 1.05109596e+00 -1.03661609e+00
4.98024672e-01 1.37850761e+00 7.00635314e-01 7.46281385e-01
-1.29034126e+00 -6.86580539e-01 8.89960751e-02 2.42113560e-01
-1.26899981e+00 -1.13972175e+00 9.02449548e-01 -4.77705926e-01
1.20236981e+00 3.68467987e-01 6.48230761e-02 1.65462971e+00
4.58935127e-02 8.95417333e-01 8.43395650e-01 -6.01326048e-01
1.97282493e-01 5.77483118e-01 7.37504363e-01 3.13266665e-01
-3.94572914e-01 3.85954559e-01 -8.64446342e-01 -2.61648655e-01
1.78401824e-02 -1.36660099e-01 -4.74204123e-01 1.90260336e-01
-1.16526461e+00 9.75398362e-01 2.70489715e-02 5.68072259e-01
-2.79041648e-01 5.28322570e-02 9.36655521e-01 5.23482263e-01
4.69612718e-01 1.53589815e-01 -8.74949098e-01 -4.59368914e-01
-7.38453448e-01 -1.26596987e-02 7.99784601e-01 7.43984580e-01
5.11037767e-01 5.89942098e-01 -5.39043844e-01 1.04372668e+00
4.04536307e-01 4.65078622e-01 1.05465162e+00 -3.69529247e-01
5.51638901e-01 1.09959327e-01 -2.32387394e-01 -4.55935389e-01
-5.16867161e-01 -3.42363089e-01 -6.31119668e-01 -2.19304711e-01
2.64781594e-01 -4.47655827e-01 -1.09265673e+00 1.92682970e+00
3.08790077e-02 -4.44844477e-02 2.16444686e-01 6.47508204e-01
1.19150674e+00 9.20818090e-01 -4.04606014e-02 -5.82799613e-02
1.61317086e+00 -6.61078513e-01 -1.11765540e+00 -1.91140309e-01
7.17267692e-01 -5.21900594e-01 9.57625329e-01 -9.61083993e-02
-7.78479815e-01 -7.48958230e-01 -1.10362804e+00 -3.45228724e-02
-5.73458076e-01 1.18574542e-04 1.69841260e-01 1.12850165e+00
-1.09969652e+00 3.39282483e-01 -4.22985435e-01 -2.12524757e-01
2.16090679e-01 3.16971183e-01 -5.59877574e-01 2.69566271e-02
-1.43712938e+00 7.63772190e-01 2.99548227e-02 -2.08550856e-01
-9.57908034e-01 -7.85192370e-01 -1.29994285e+00 4.16035205e-01
-2.66477019e-01 8.22126567e-02 1.35485184e+00 -5.99369168e-01
-1.59638822e+00 5.47955871e-01 -5.14333725e-01 -4.12526131e-01
-1.58506483e-01 2.50746876e-01 -8.35299432e-01 -2.64309421e-02
-5.02917349e-01 6.73690379e-01 9.86523688e-01 -9.37237799e-01
-2.63543427e-01 -7.02722251e-01 -3.36627543e-01 -1.81555659e-01
-8.16566586e-01 2.31440216e-01 -1.23648077e-01 -6.27435505e-01
-1.09425411e-01 -7.14420676e-01 3.48931253e-01 -2.04826787e-01
-7.23273754e-01 -4.26888943e-01 1.06335080e+00 -1.04681766e+00
1.23613250e+00 -2.76754117e+00 -2.49347929e-02 2.02024858e-02
-1.16947182e-02 3.00532460e-01 -4.71938640e-01 7.05419660e-01
-2.72419214e-01 3.04252684e-01 -1.44457042e-01 -9.15057898e-01
4.45631206e-01 3.77325624e-01 -3.98020297e-01 5.86966991e-01
3.40356261e-01 8.94067168e-01 -2.55854636e-01 -1.77123860e-01
6.56782761e-02 1.09997392e+00 -2.73073673e-01 1.59914240e-01
1.56663969e-01 1.73267365e-01 1.02906421e-01 4.89060342e-01
6.02391362e-01 4.49525714e-01 -1.07314125e-01 -5.82863316e-02
1.06492906e-03 7.58030474e-01 -6.65249169e-01 1.57238376e+00
-7.18110740e-01 1.43538725e+00 4.82053339e-01 -7.51887262e-01
8.81127715e-01 8.83479118e-01 8.86363834e-02 -6.48245871e-01
3.98211032e-01 -2.81320035e-01 2.91180760e-01 -4.88821834e-01
4.94483113e-01 -4.64255989e-01 -1.12563074e-01 7.59479880e-01
5.16887128e-01 3.13232332e-01 -2.52474248e-01 1.71535566e-01
1.04182410e+00 -7.40014076e-01 -1.24648653e-01 2.22045779e-02
3.27689469e-01 -6.80178761e-01 2.22515650e-02 5.27632594e-01
-3.76898080e-01 2.76931375e-01 5.10930538e-01 6.74633980e-02
-9.37893569e-01 -8.33281577e-01 -6.19555175e-01 1.47035122e+00
-4.69053596e-01 -4.47552413e-01 -5.63950360e-01 -5.68115473e-01
2.98141867e-01 1.00753927e+00 -1.00371265e+00 -3.80992740e-01
-1.97408572e-01 -3.84666443e-01 9.38796222e-01 8.66612017e-01
-2.65350997e-01 -7.99395621e-01 -5.78213111e-02 3.04202050e-01
1.49465948e-01 -1.20233452e+00 -8.46343696e-01 6.37413025e-01
-7.00440884e-01 -4.22126710e-01 -5.82797766e-01 -7.22855330e-01
1.78303972e-01 -1.01068683e-01 8.08489740e-01 -3.37423205e-01
-2.91595817e-01 5.95576704e-01 -4.78077084e-01 -5.15065849e-01
-6.13013625e-01 -3.82030085e-02 3.77761006e-01 2.32962817e-01
4.44658339e-01 -3.39657009e-01 -1.63771883e-01 1.97713673e-01
-8.90402436e-01 -8.72634470e-01 2.97130823e-01 9.38854635e-01
-1.18275605e-01 -1.57434613e-01 6.83912575e-01 -3.82413745e-01
8.49452019e-01 -3.62436086e-01 2.48414725e-02 -1.66050002e-01
-4.06166375e-01 2.53152877e-01 3.60403240e-01 -7.08771348e-01
-7.18119442e-01 -2.87063986e-01 -6.00081980e-01 -3.62148523e-01
-4.31638062e-01 5.94133258e-01 -3.53336871e-01 3.42834473e-01
5.55956483e-01 4.97034043e-01 3.82007986e-01 -7.44959831e-01
5.77233672e-01 1.38069570e+00 3.34503710e-01 -1.03837937e-01
4.42851722e-01 4.82009351e-02 -7.78530836e-01 -1.37475562e+00
-3.02082241e-01 -5.93388081e-01 -4.58872855e-01 2.33783916e-01
7.11339474e-01 -8.12692046e-01 -6.43008828e-01 3.48553002e-01
-1.22733867e+00 3.17011587e-02 -4.37247932e-01 6.69795275e-01
-1.45140573e-01 2.75666744e-01 -8.44088554e-01 -1.14514911e+00
-3.69324207e-01 -1.18130159e+00 1.05218911e+00 -2.06721902e-01
-4.18684185e-01 -1.14053321e+00 1.97065368e-01 1.05097041e-01
6.75272644e-01 -3.37293506e-01 1.02864444e+00 -1.28612447e+00
4.50908989e-02 -5.99754810e-01 3.73934774e-04 6.33595228e-01
1.40637636e-01 -4.80225161e-02 -1.71640456e+00 -3.83281350e-01
1.66041553e-01 -7.70312250e-02 1.02709103e+00 4.34561402e-01
1.13157523e+00 -7.12112308e-01 -1.46392047e-01 4.78482544e-01
8.89728665e-01 1.32763028e-01 3.73172402e-01 -2.54020035e-01
4.43838149e-01 6.03709519e-01 -3.07924390e-01 5.90470850e-01
1.95653826e-01 7.46913850e-01 1.09975614e-01 1.57238200e-01
-9.46177915e-02 -2.37304062e-01 7.19467878e-01 1.04168737e+00
6.66620493e-01 -5.12530625e-01 -7.04711795e-01 6.57144248e-01
-9.21479166e-01 -9.42191899e-01 1.92640483e-01 1.83913887e+00
7.88672507e-01 4.38865088e-02 3.09092283e-01 7.72664547e-01
5.01194537e-01 3.64412427e-01 -5.14895856e-01 -9.03446734e-01
-1.09924689e-01 3.65870595e-01 1.78603306e-01 7.03106940e-01
-9.49116290e-01 3.75069261e-01 7.25005388e+00 4.86231208e-01
-1.54908061e+00 3.76680255e-01 2.95227289e-01 -1.75223216e-01
-4.87183601e-01 -5.32524645e-01 -9.03032005e-01 5.34988225e-01
1.69825840e+00 -8.97706747e-02 2.40785971e-01 5.98877132e-01
1.23638771e-01 4.24526453e-01 -1.64705741e+00 1.04712451e+00
4.20737028e-01 -1.13749313e+00 -9.41506326e-02 4.22586232e-01
2.09445849e-01 1.68872356e-01 4.85045701e-01 6.28378868e-01
-2.63274312e-01 -1.14781010e+00 6.92378521e-01 3.51191573e-02
8.55878353e-01 -5.15834808e-01 8.46047342e-01 7.50374869e-02
-7.18239248e-01 -3.28131139e-01 3.26086720e-03 5.25418557e-02
2.17858940e-01 3.00538838e-01 -1.17539048e+00 -4.90836985e-03
3.23331326e-01 5.14788806e-01 -4.72496033e-01 5.77826381e-01
5.95174506e-02 1.05785978e+00 -1.55738279e-01 -6.34912401e-02
1.68335915e-01 3.95290226e-01 5.03572762e-01 1.58283520e+00
7.23875314e-02 -1.92821041e-01 -4.06071603e-01 6.54208481e-01
-3.22337955e-01 5.32611124e-02 -6.07669771e-01 -5.24593413e-01
5.69765747e-01 9.48423088e-01 5.93498647e-02 -4.12346184e-01
-3.76091093e-01 7.91646183e-01 -6.65387511e-02 5.54154932e-01
-6.51143670e-01 -5.68589330e-01 1.29244614e+00 4.16636467e-04
5.92201769e-01 -3.63056570e-01 -1.97437257e-01 -7.71002352e-01
1.23396851e-02 -7.15338767e-01 2.36962214e-02 -4.35705751e-01
-1.21551454e+00 7.51636386e-01 -1.72762513e-01 -8.85481179e-01
-5.33531070e-01 -6.76803112e-01 -7.80802071e-01 1.05098367e+00
-1.48826551e+00 -7.80489028e-01 2.32066408e-01 3.96680117e-01
5.83650768e-01 -4.12107408e-01 1.09020007e+00 2.77398556e-01
-7.53554285e-01 1.22268629e+00 4.35935766e-01 6.01403654e-01
4.99776959e-01 -1.12719226e+00 6.47179246e-01 2.57692337e-01
5.75656414e-01 7.05614448e-01 6.29581511e-01 1.37801677e-01
-1.69173181e+00 -8.75847876e-01 9.45212901e-01 -5.98973632e-01
4.74801749e-01 -6.75531983e-01 -1.05916619e+00 7.35970438e-01
5.68845749e-01 -8.43629166e-02 1.32033336e+00 5.62545240e-01
-8.61126363e-01 -1.08810522e-01 -1.06564486e+00 -4.93337847e-02
4.71731126e-01 -1.27003753e+00 -6.96346700e-01 3.23013142e-02
9.92363155e-01 2.49408139e-03 -9.22858715e-01 2.95348354e-02
5.22909820e-01 -6.52994335e-01 9.47579622e-01 -9.15001392e-01
1.61295488e-01 3.24056417e-01 -5.61912060e-01 -1.60687411e+00
-1.66786626e-01 -3.81132662e-01 -2.90769458e-01 1.53968513e+00
7.89513350e-01 -6.84256494e-01 3.82324249e-01 5.48613191e-01
-2.37550259e-01 -4.54791665e-01 -1.30672979e+00 -8.44942093e-01
2.21501738e-01 -6.29513800e-01 8.10734212e-01 9.14922535e-01
4.90374237e-01 6.87894940e-01 -1.51144877e-01 8.94613415e-02
3.39805812e-01 -2.02950165e-01 5.48567712e-01 -1.05218065e+00
-8.87852684e-02 -6.16132915e-01 -7.06773341e-01 -9.65466976e-01
5.73310316e-01 -9.92921531e-01 2.71889828e-02 -1.22427011e+00
-8.59409571e-02 -1.17473021e-01 -2.05735415e-01 4.59640443e-01
2.18012743e-02 4.77740914e-02 1.32358581e-01 -1.95789039e-01
9.67611820e-02 8.51476371e-01 6.09400690e-01 -5.73532283e-01
-2.14299932e-01 -1.14560202e-01 -5.89518070e-01 6.13467246e-02
6.74014628e-01 -1.66776359e-01 -5.34413457e-02 -5.18148601e-01
-5.92019439e-01 3.09868634e-01 -7.14769959e-02 -6.05315924e-01
1.52042219e-02 6.44421160e-01 3.93882573e-01 -5.70772529e-01
8.99354100e-01 -6.57850385e-01 -4.87757504e-01 2.02923805e-01
-8.49542975e-01 -1.14696741e-01 6.79906309e-01 4.50730324e-01
-4.88718003e-01 -1.36606842e-01 4.92211431e-01 3.92539233e-01
-2.50353813e-01 2.69384116e-01 -7.84393013e-01 -4.80918773e-02
4.85582024e-01 -1.71080157e-01 -7.99111575e-02 -7.09693968e-01
-9.63432610e-01 -2.49252960e-01 5.31900823e-02 7.49282122e-01
6.22950554e-01 -1.37652421e+00 -8.37677598e-01 7.66080797e-01
3.87962759e-01 -8.35632324e-01 1.74544737e-01 5.03943801e-01
3.17774832e-01 7.06656516e-01 1.31347060e-01 -7.00479567e-01
-1.44581008e+00 7.31296957e-01 1.15657620e-01 4.04231280e-01
-2.68959850e-01 8.96453857e-01 -9.26074460e-02 -4.91931230e-01
8.45637739e-01 -7.51962483e-01 -1.21929280e-01 5.66627324e-01
8.99827659e-01 5.03242165e-02 2.62071878e-01 -8.97778511e-01
-5.45331001e-01 1.98194206e-01 -2.61516541e-01 -5.74016690e-01
1.32332754e+00 -1.41324908e-01 2.46563524e-01 7.38959789e-01
2.00980163e+00 4.33259457e-02 -8.00622404e-01 -4.76118296e-01
-2.80101635e-02 -1.73015222e-01 5.17886698e-01 -7.14686930e-01
-9.68248248e-01 1.47280562e+00 7.97304749e-01 4.05097306e-01
5.67490280e-01 1.88021272e-01 7.28954434e-01 -5.43459654e-02
-3.05143923e-01 -7.54437923e-01 -2.12518960e-01 4.75218087e-01
9.08837378e-01 -1.22367597e+00 -4.84274536e-01 2.54172683e-01
-6.32736325e-01 1.02425361e+00 1.59460574e-01 6.32702291e-01
1.18728721e+00 4.49497074e-01 4.15283740e-01 -4.23941873e-02
-9.09950674e-01 -1.58368628e-02 1.91078991e-01 6.60775006e-01
6.45585954e-01 3.22319567e-01 4.53975767e-01 5.17707586e-01
-3.66513699e-01 -6.37858868e-01 4.85150427e-01 7.48596668e-01
-2.95412570e-01 -1.00738132e+00 -6.52295172e-01 4.79304731e-01
-3.96295339e-01 -2.16194704e-01 -5.99717200e-01 3.74831051e-01
-3.42048287e-01 1.24213815e+00 5.05889595e-01 -6.80232584e-01
4.13585752e-01 8.02829325e-01 2.50381976e-01 -4.82190728e-01
-7.01245129e-01 -3.47296000e-02 1.56860396e-01 -1.76723406e-01
3.97008620e-02 -8.24161649e-01 -9.81954575e-01 -2.14425445e-01
-5.79145670e-01 2.50829667e-01 1.45766318e+00 7.96105623e-01
4.40306723e-01 6.16616666e-01 8.96068633e-01 -8.75726104e-01
-9.38712180e-01 -1.54027092e+00 -7.81241179e-01 3.33648831e-01
1.12806737e+00 -2.56361037e-01 -8.50070894e-01 -1.59651771e-01] | [14.349690437316895, 6.145889759063721] |
d6779e7d-0216-4253-b39a-c9f592d965e1 | positive-negative-and-neutral-modeling | 2205.06058 | null | https://arxiv.org/abs/2205.06058v1 | https://arxiv.org/pdf/2205.06058v1.pdf | Positive, Negative and Neutral: Modeling Implicit Feedback in Session-based News Recommendation | News recommendation for anonymous readers is a useful but challenging task for many news portals, where interactions between readers and articles are limited within a temporary login session. Previous works tend to formulate session-based recommendation as a next item prediction task, while they neglect the implicit feedback from user behaviors, which indicates what users really like or dislike. Hence, we propose a comprehensive framework to model user behaviors through positive feedback (i.e., the articles they spend more time on) and negative feedback (i.e., the articles they choose to skip without clicking in). Moreover, the framework implicitly models the user using their session start time, and the article using its initial publishing time, in what we call "neutral feedback". Empirical evaluation on three real-world news datasets shows the framework's promising performance of more accurate, diverse and even unexpectedness recommendations than other state-of-the-art session-based recommendation approaches. | ['Kenny Q. Zhu', 'Shansan Gong'] | 2022-05-12 | null | null | null | null | ['session-based-recommendations'] | ['miscellaneous'] | [-4.48781341e-01 -7.23271370e-02 -1.01785123e+00 -6.51326537e-01
1.16195427e-02 -5.19299209e-01 7.20775902e-01 2.59641647e-01
-4.92746443e-01 6.10350788e-01 5.63698232e-01 -5.12084067e-01
-7.36227483e-02 -7.32960224e-01 -5.60134470e-01 -3.09982568e-01
2.77068973e-01 2.21591637e-01 5.11128008e-01 -5.22148132e-01
7.40902603e-01 -2.69534856e-01 -1.47784615e+00 2.56949216e-01
9.89909410e-01 7.05180705e-01 1.93742797e-01 3.00656289e-01
-2.83669442e-01 7.32945740e-01 -3.24985951e-01 -6.86069667e-01
5.32736182e-02 -2.92553753e-01 -5.24863958e-01 -2.50629097e-01
9.10231564e-03 -7.63494790e-01 -4.30268139e-01 8.95518601e-01
1.57830164e-01 6.90265715e-01 4.35792416e-01 -9.61808026e-01
-1.32650614e+00 1.09383488e+00 -4.85946745e-01 3.86540800e-01
8.18858445e-01 -3.84446234e-01 1.34518445e+00 -6.37919545e-01
5.94376564e-01 7.24279881e-01 3.04060817e-01 5.36383450e-01
-1.08333790e+00 -4.11654323e-01 9.26608503e-01 2.90440172e-01
-8.23225498e-01 -3.32897574e-01 6.58160210e-01 -4.92789358e-01
4.92832452e-01 9.08193827e-01 5.94942987e-01 1.19457245e+00
-1.69847868e-02 1.10423648e+00 6.68463469e-01 -7.03944564e-02
2.88731933e-01 6.64382517e-01 7.46222794e-01 -1.85137585e-01
1.25223339e-01 -1.50070086e-01 -5.71955800e-01 -3.38656753e-01
4.56576139e-01 7.56954014e-01 -3.34177703e-01 -1.83305070e-01
-1.13899469e+00 7.67083168e-01 2.54504234e-01 1.83449626e-01
-4.07465458e-01 -7.40064800e-01 5.56084365e-02 6.22868299e-01
7.64640629e-01 3.51672381e-01 -5.78914344e-01 -2.03768745e-01
-7.95454323e-01 3.81485730e-01 1.25311601e+00 8.96943808e-01
4.98496920e-01 -5.19781053e-01 -4.97184426e-01 9.79384720e-01
4.13623571e-01 2.15013579e-01 4.34534818e-01 -1.99052855e-01
-1.00498339e-02 5.48745394e-01 7.25339711e-01 -9.06017959e-01
-2.03462660e-01 -8.78743410e-01 -4.43041414e-01 -3.44631672e-01
4.19099569e-01 -2.95805186e-01 -4.50585037e-01 1.47423971e+00
9.78926644e-02 2.07780957e-01 -3.12687069e-01 9.67386961e-01
1.01678431e+00 7.48213530e-01 7.29453862e-02 -7.43717372e-01
1.03935635e+00 -1.13720524e+00 -7.77597904e-01 -4.81588691e-02
4.19762701e-01 -9.13469732e-01 1.14570248e+00 5.55051804e-01
-9.70011234e-01 -3.13155055e-01 -5.80032289e-01 2.47276619e-01
-2.08592013e-01 3.03293824e-01 7.87225068e-01 4.09844875e-01
-7.59259999e-01 6.39431417e-01 -5.09552062e-01 -5.74164748e-01
-1.97611645e-01 1.69112682e-01 1.37596652e-01 2.57477134e-01
-1.21726215e+00 4.11095470e-01 -1.63529277e-01 -1.65424511e-01
-1.06130354e-01 -7.57567227e-01 -3.75758857e-02 2.16392651e-01
5.32919109e-01 -5.32967389e-01 1.40243399e+00 -9.26861763e-01
-1.74942398e+00 3.62917334e-01 -3.60253572e-01 -2.66800374e-01
6.31427288e-01 -5.44339955e-01 -8.44028056e-01 -6.62150681e-01
-2.64391992e-02 -9.76056084e-02 6.68346584e-01 -1.18023717e+00
-9.73088562e-01 -2.00159863e-01 6.47724450e-01 4.44208413e-01
-7.00776994e-01 -4.33950834e-02 -7.26619422e-01 -6.06812835e-01
1.41475320e-01 -9.48693752e-01 -2.13529125e-01 -4.22418684e-01
-4.98066038e-01 -5.49668491e-01 5.64309835e-01 -5.26352346e-01
1.73552084e+00 -2.08999801e+00 -8.14302042e-02 3.84248883e-01
1.60276487e-01 2.33747587e-01 1.79355428e-01 7.16716290e-01
5.11892021e-01 3.08014452e-01 8.32210541e-01 2.81272475e-02
8.65823627e-02 -2.28124186e-01 -5.26317477e-01 1.04460299e-01
-9.26044643e-01 5.24501085e-01 -1.01221311e+00 1.55243069e-01
-2.71520987e-02 1.43696502e-01 -7.14605987e-01 3.98060173e-01
-2.82643557e-01 5.81066370e-01 -9.12887394e-01 2.31244743e-01
6.78782880e-01 -6.08978748e-01 1.64316610e-01 2.59765144e-02
-4.37508017e-01 6.52709126e-01 -9.18804646e-01 1.19496882e+00
-3.53106856e-01 6.68996647e-02 -1.56912878e-01 -5.53853929e-01
7.78117478e-01 1.54069766e-01 3.27752203e-01 -9.37840283e-01
9.52845067e-03 -1.57121003e-01 5.12843616e-02 -4.40681458e-01
7.22236872e-01 4.54261929e-01 1.60662875e-01 1.05979908e+00
-3.88759881e-01 1.02196026e+00 3.98605585e-01 6.95895255e-01
8.35660040e-01 3.37937623e-02 3.20462316e-01 -1.85872212e-01
6.29431605e-01 -4.76568341e-01 5.72363019e-01 1.31429648e+00
3.81920375e-02 1.24098159e-01 4.02733773e-01 -2.54368693e-01
-6.02634072e-01 -8.06310713e-01 1.81912586e-01 2.00524497e+00
5.44999778e-01 -6.29766643e-01 -3.78562689e-01 -8.76116097e-01
1.68787278e-02 9.63355720e-01 -8.50914240e-01 -1.16229020e-02
-2.07754299e-01 -4.23709154e-01 -6.18487775e-01 1.99224725e-01
3.91946621e-02 -8.18922341e-01 4.90600348e-01 3.71442229e-01
-3.76506418e-01 -3.89740467e-01 -1.05501068e+00 -3.92696172e-01
-6.83315098e-01 -7.29399979e-01 -8.65338027e-01 -4.39996749e-01
7.35877991e-01 7.60397434e-01 1.05879581e+00 3.80688518e-01
7.19806910e-01 2.62257814e-01 -8.73432934e-01 -1.95902646e-01
1.80767342e-01 2.14316487e-01 2.45721653e-01 3.07929426e-01
5.73197365e-01 -5.47220647e-01 -1.01459432e+00 9.75225449e-01
-4.57544178e-01 1.36702225e-01 2.75221556e-01 7.55650043e-01
1.88733459e-01 -3.66420776e-01 9.78708327e-01 -1.72692549e+00
9.36716855e-01 -1.01167560e+00 -1.74054548e-01 4.66268808e-01
-9.43467855e-01 -3.62116963e-01 9.17790115e-01 -7.50887394e-01
-1.41876936e+00 -5.32416403e-01 -3.09518963e-01 5.08311749e-01
2.70026773e-02 7.29088783e-01 -1.05834454e-02 1.66945651e-01
6.12134457e-01 3.43246102e-01 -3.48125130e-01 -6.93616569e-01
3.10253471e-01 8.04927528e-01 5.02695888e-03 -1.88237861e-01
7.12315619e-01 2.29412138e-01 -9.78265822e-01 -4.51360375e-01
-1.29716253e+00 -1.00341189e+00 -1.82496235e-01 -4.04577971e-01
1.60173133e-01 -7.05057204e-01 -1.11505938e+00 2.21597135e-01
-6.17064059e-01 7.31777772e-02 2.82214787e-02 8.77348959e-01
3.64638004e-03 2.24869117e-01 -7.31349170e-01 -8.88542116e-01
-2.73144633e-01 -6.24474943e-01 1.30575716e-01 6.49203241e-01
-3.44623238e-01 -1.11285818e+00 4.84556556e-02 4.25108463e-01
7.70852089e-01 -6.31253123e-01 5.46760976e-01 -1.19937861e+00
-5.41832030e-01 -4.70660210e-01 -1.59939930e-01 -3.67168374e-02
1.52270004e-01 -5.14564700e-02 -4.13933516e-01 -3.26243043e-01
-2.46085525e-01 3.06130767e-01 5.53577900e-01 3.82831603e-01
1.04835570e+00 -6.63168967e-01 -5.43246090e-01 3.08185071e-01
7.94997692e-01 4.22146082e-01 5.32687962e-01 3.36043835e-01
3.94096524e-01 3.08306426e-01 9.93450820e-01 8.59214604e-01
5.41022360e-01 7.18071282e-01 2.25970700e-01 4.01930250e-02
3.42169374e-01 -6.26110077e-01 3.60077977e-01 8.95275891e-01
-3.73908281e-01 -7.97847092e-01 -5.62109351e-02 2.03572556e-01
-2.35919952e+00 -1.16212308e+00 -2.64007479e-01 2.60684824e+00
5.26480794e-01 3.52682501e-01 3.87331814e-01 -2.10369155e-01
9.12306130e-01 1.55656978e-01 -8.26870620e-01 -2.29848266e-01
2.98399597e-01 -4.98406649e-01 4.25539285e-01 2.96215683e-01
-7.84581363e-01 7.09144771e-01 5.70747614e+00 6.14469111e-01
-1.04615867e+00 -4.46788482e-02 6.34794891e-01 -2.18301401e-01
-7.34213829e-01 1.35914432e-02 -1.00849497e+00 1.01441979e+00
8.34402978e-01 -7.10516632e-01 6.09992266e-01 9.01876271e-01
5.31381607e-01 -2.71659363e-02 -1.16587710e+00 6.53745592e-01
1.17206104e-01 -1.34447157e+00 -2.89329290e-02 5.35847060e-02
8.52979839e-01 -1.56120732e-01 1.37337670e-01 5.83100975e-01
4.79059458e-01 -2.37944081e-01 6.21385872e-01 8.79774094e-01
1.08927406e-01 -4.97416407e-01 7.03825474e-01 8.27864885e-01
-7.46297061e-01 -1.17832445e-01 -4.38419104e-01 -3.05752635e-01
3.28998566e-01 6.08057678e-01 -5.03415108e-01 5.04021645e-01
6.32272959e-01 1.23186052e+00 -4.03911829e-01 1.39610708e+00
-2.32599676e-01 1.11786449e+00 -5.30580543e-02 -5.80446482e-01
-1.63444057e-01 -5.54166853e-01 4.88658100e-01 7.85955966e-01
6.03967607e-01 3.30524653e-01 2.69049942e-01 2.81657308e-01
-2.60374814e-01 7.41158307e-01 -8.94568563e-02 1.78555414e-01
4.52272147e-01 1.35924327e+00 -4.91805434e-01 -1.75887540e-01
-7.93560982e-01 7.84667969e-01 2.72399932e-01 5.17931402e-01
-5.85256994e-01 -1.30172476e-01 4.34428453e-01 6.51319265e-01
2.55464286e-01 2.83849508e-01 -1.89867213e-01 -1.48895514e+00
1.49069175e-01 -3.91631603e-01 2.86979347e-01 -5.71489215e-01
-1.68830824e+00 2.62982935e-01 -5.62012672e-01 -1.68317032e+00
-4.96155815e-03 6.15699477e-02 -1.18856776e+00 7.15388954e-01
-1.17073071e+00 -7.24695027e-01 -6.59043267e-02 4.14469987e-01
3.74174684e-01 -2.02835843e-01 5.06172419e-01 5.14142275e-01
-4.43253428e-01 7.95464337e-01 6.79193795e-01 -2.09281325e-01
1.09211814e+00 -1.04724455e+00 1.69504583e-01 4.78076100e-01
6.78631896e-03 1.31306374e+00 9.55271184e-01 -6.32073164e-01
-1.39383376e+00 -7.52093256e-01 1.03711641e+00 -3.01164091e-01
6.38725579e-01 -2.25224897e-01 -9.27900732e-01 8.11132133e-01
1.05951093e-01 -3.29202801e-01 1.23519623e+00 1.00728917e+00
-7.41898268e-02 -1.62628651e-01 -8.52730691e-01 7.62103856e-01
1.05501366e+00 -8.34250227e-02 -5.55392444e-01 5.44175684e-01
4.33210462e-01 -2.16474518e-01 -6.62249327e-01 -5.34370095e-02
9.67479706e-01 -1.01356637e+00 8.38495433e-01 -7.46914148e-01
2.10918143e-01 -2.01442286e-01 4.39496905e-01 -1.43490946e+00
-8.25728118e-01 -5.77861011e-01 -3.19649220e-01 1.38926637e+00
7.31161773e-01 -6.65254056e-01 9.66331184e-01 8.52340519e-01
8.80047679e-02 -6.39935493e-01 -1.61962435e-01 -4.33826447e-01
-4.48054940e-01 1.02112256e-01 5.41428506e-01 1.07714415e+00
5.34987509e-01 4.55901057e-01 -1.05488241e+00 -1.19696021e-01
2.73777157e-01 3.33768457e-01 8.36260796e-01 -1.71978700e+00
-3.96573931e-01 -1.89820275e-01 1.97173968e-01 -1.76927090e+00
-1.21386372e-01 -6.36955261e-01 -1.69213012e-01 -1.62558746e+00
4.82439399e-01 -4.65893418e-01 -8.78086150e-01 8.82402658e-02
-3.37385327e-01 -1.18090242e-01 -1.78892732e-01 5.86798489e-01
-1.23448277e+00 3.27189893e-01 1.34990084e+00 1.81942120e-01
-6.41205490e-01 9.39578116e-01 -1.34065151e+00 6.11420691e-01
5.06188810e-01 -3.81524175e-01 -6.42614603e-01 -1.44782677e-01
8.41643274e-01 2.95682222e-01 -1.19287133e-01 -3.90018374e-01
5.89257956e-01 -5.85859358e-01 1.43185794e-01 -6.24241173e-01
-5.62853999e-02 -6.05296493e-01 1.34661257e-01 -2.81895380e-02
-9.53869045e-01 -3.26451987e-01 -6.62530959e-01 1.10503113e+00
8.19655973e-03 -2.14974299e-01 1.70412287e-01 -1.16250990e-02
-5.78517795e-01 7.31581211e-01 -4.02830839e-01 -2.16662928e-01
1.02820504e+00 -3.53976078e-02 -5.73713005e-01 -9.59952176e-01
-1.19043946e+00 3.37600708e-01 5.11096179e-01 8.25136483e-01
3.48829478e-01 -1.13946605e+00 -6.26883805e-01 -3.05042248e-02
3.57594967e-01 -8.85780513e-01 5.97599030e-01 7.75131464e-01
2.93659836e-01 2.93648034e-01 -1.33490516e-02 -1.41101465e-01
-1.31960869e+00 4.67067957e-01 -4.12635654e-01 -1.71232119e-01
-3.06398988e-01 8.41518402e-01 3.28989893e-01 -3.24281305e-01
4.09733295e-01 2.63850540e-01 -1.05109584e+00 1.48131475e-01
9.22196150e-01 3.68712902e-01 -3.01004690e-03 -3.13655913e-01
-1.21012412e-01 -1.21847004e-01 -9.94604409e-01 1.71550438e-01
1.40693760e+00 -6.92556202e-01 2.20652130e-02 6.67120993e-01
6.20680988e-01 4.93367434e-01 -9.05999720e-01 -7.69506156e-01
-1.30799174e-01 -8.81877363e-01 9.31790546e-02 -9.87716436e-01
-9.59678829e-01 2.45708779e-01 2.10683942e-01 8.59669507e-01
5.59773266e-01 -1.73494697e-01 8.80988300e-01 5.44097722e-01
3.80646944e-01 -1.32650173e+00 -2.42042646e-01 4.53161150e-01
6.03311598e-01 -1.32247317e+00 1.95672706e-01 -4.32140023e-01
-8.06387007e-01 8.41825962e-01 7.92871654e-01 3.40010151e-02
1.23659170e+00 -5.00934422e-01 1.86081547e-02 3.21031481e-01
-1.10487688e+00 1.86460689e-01 4.69140738e-01 1.92361236e-01
9.38354015e-01 2.20082060e-01 -1.04894745e+00 1.16671896e+00
-1.42659470e-01 1.13231331e-01 8.04945946e-01 7.70237982e-01
-5.62969208e-01 -1.29759824e+00 2.72294015e-01 1.22879755e+00
-7.04298258e-01 -6.30282685e-02 -2.12243512e-01 9.12477747e-02
-3.05042416e-01 1.22858918e+00 -1.15280353e-01 -5.98448634e-01
3.30188245e-01 -2.78260559e-01 -3.90468448e-01 -8.12945902e-01
-9.16650116e-01 3.19838710e-02 1.19054057e-01 -3.12402040e-01
-2.51317173e-01 -7.74235249e-01 -7.77882040e-01 -7.16791451e-01
-6.70728445e-01 7.75361776e-01 5.40666461e-01 9.37362254e-01
4.63413090e-01 2.65622020e-01 1.05963457e+00 -4.44062561e-01
-4.98563379e-01 -9.35728550e-01 -8.65990460e-01 6.15274191e-01
2.37211332e-01 -5.31610906e-01 -4.60742474e-01 -1.20795943e-01] | [10.110187530517578, 5.678277969360352] |
e1b9cd0c-0957-49c2-9d61-ab9dda951802 | tv-regularized-ct-reconstruction-and-metal | 1810.03275 | null | http://arxiv.org/abs/1810.03275v1 | http://arxiv.org/pdf/1810.03275v1.pdf | TV-regularized CT Reconstruction and Metal Artifact Reduction Using Inequality Constraints with Preconditioning | Total variation(TV) regularization is applied to X-Ray computed
tomography(CT) in an effort to reduce metal artifacts. Tikhonov regularization
with $L^2$ data fidelity term and total variation regularization is augmented
in this novel model by inequality constraints on sinogram data affected by
metal to model errors caused by metal. The formulated problem is discretized
and solved using the Chambolle-Pock algorithm. Faster convergence is achieved
using preconditioning in a Douglas-Rachford spitting method as well as Advanced
Direction Method of Multipliers(ADMM). The methods are applied to real and
synthetic data demonstrating feasibility of the model to reduce metal
artifacts. Technical details of CT data used and its processing are given in
the appendix. | ['Clemens Schiffer'] | 2018-10-08 | null | null | null | null | ['metal-artifact-reduction'] | ['medical'] | [ 2.54434109e-01 1.30625710e-01 2.03856006e-01 -2.09670946e-01
-8.48050594e-01 6.80739209e-02 1.02017745e-01 8.94671157e-02
-5.11224449e-01 1.15616751e+00 2.73899823e-01 -1.94484755e-01
-4.82322693e-01 -2.74134040e-01 -3.36035728e-01 -8.64316463e-01
-8.86052847e-02 3.74001563e-01 1.16589718e-01 6.42798096e-02
2.79401869e-01 5.86512804e-01 -4.27901953e-01 2.10078254e-01
7.39110649e-01 7.83287227e-01 2.04793870e-01 6.74267709e-01
3.26661080e-01 1.02748525e+00 -7.74694979e-02 3.41372818e-01
4.81206238e-01 -4.48121697e-01 -9.32945430e-01 5.91011822e-01
4.17679884e-02 -2.48493478e-01 -2.08503708e-01 6.03308558e-01
5.74951708e-01 2.92629302e-01 9.02036846e-01 -4.92136598e-01
-2.49145046e-01 -7.75761604e-02 -1.11890137e+00 3.33739728e-01
1.79983273e-01 1.27263024e-01 2.91287422e-01 -1.46925545e+00
7.65727997e-01 8.20826709e-01 1.01305926e+00 2.90714353e-01
-1.42980099e+00 -1.84747100e-01 -5.47571957e-01 -2.05503777e-01
-1.32778299e+00 7.02600274e-03 6.43453598e-01 -7.54454732e-01
9.75638390e-01 6.19274855e-01 6.54390812e-01 2.65422344e-01
7.69091904e-01 3.22952539e-01 1.27551436e+00 -5.79633474e-01
2.30453804e-01 2.09673401e-02 1.75197095e-01 7.87266791e-01
3.40645015e-01 1.49828240e-01 -2.15303436e-01 -4.98199165e-01
1.15857041e+00 1.86524764e-02 -3.27834010e-01 -3.54806215e-01
-9.60782886e-01 1.02709937e+00 3.51764083e-01 2.86089212e-01
-6.92255676e-01 4.13345307e-01 5.85623443e-01 -1.72369145e-02
7.08510101e-01 3.33325922e-01 -2.09631562e-01 4.56444234e-01
-1.03785062e+00 1.29926890e-01 1.43783256e-01 7.74883211e-01
1.77658886e-01 4.24273491e-01 -1.62547737e-01 9.18264806e-01
6.26711309e-01 5.38470507e-01 3.25435936e-01 -9.11554933e-01
9.75946113e-02 2.72101164e-01 -8.02543312e-02 -6.24428213e-01
-5.56460202e-01 -3.12159032e-01 -9.71198976e-01 5.97217202e-01
2.81134427e-01 -1.51204005e-01 -1.22922599e+00 8.89578700e-01
5.70800483e-01 1.04186192e-01 -4.29063380e-01 1.03738713e+00
5.40349662e-01 4.43305701e-01 -7.99776092e-02 -7.36550987e-01
1.22946477e+00 -6.30899549e-01 -9.58235979e-01 1.31077886e-01
6.54652476e-01 -1.10564089e+00 4.83646214e-01 5.66180170e-01
-1.60244167e+00 -3.38714093e-01 -7.18576491e-01 -5.49419820e-02
3.90747488e-01 -7.60750100e-02 3.94866168e-01 6.57796144e-01
-7.43189394e-01 8.86626780e-01 -1.15414155e+00 4.60805483e-02
6.07153535e-01 6.16294861e-01 -2.10300878e-01 -8.22063386e-02
-6.09284103e-01 1.14865363e+00 -2.25007236e-01 3.31839770e-01
-6.23411477e-01 -1.04352808e+00 -8.09893727e-01 -7.15423405e-01
4.68559340e-02 -8.10872555e-01 6.86716557e-01 -6.55791819e-01
-1.52216411e+00 6.38835192e-01 -9.36135724e-02 -2.23266706e-01
1.06958127e+00 1.64987578e-03 2.82318220e-02 3.52528393e-01
6.99647665e-02 -1.54629812e-01 7.87965477e-01 -1.29862940e+00
9.50559229e-03 -3.00838917e-01 -9.19331610e-01 2.06042081e-02
4.87859011e-01 -1.17343634e-01 -1.86073974e-01 -9.65796769e-01
8.94382119e-01 -9.93560970e-01 -9.54364538e-01 2.46502250e-01
-4.03666437e-01 4.26260412e-01 5.31885624e-01 -1.04992557e+00
1.25451636e+00 -1.77730131e+00 2.09602322e-02 8.18222582e-01
2.24691153e-01 -6.16190173e-02 2.09364444e-01 4.67178613e-01
-3.12047899e-01 -3.25766653e-01 -1.03852892e+00 -2.27808058e-01
-7.15687692e-01 3.99744123e-01 3.37923706e-01 1.18970835e+00
-2.39465341e-01 4.72188503e-01 -7.08038568e-01 -5.74316680e-01
4.70520526e-01 8.34709942e-01 -5.54650545e-01 -3.39623213e-01
3.82134050e-01 1.16172314e+00 -5.51211357e-01 4.63680595e-01
1.05510271e+00 -1.44044653e-01 -1.35947257e-01 -1.43052161e-01
-4.29364324e-01 -8.54077637e-02 -1.33657038e+00 1.64709067e+00
-5.70523560e-01 2.47342587e-01 5.49608409e-01 -8.35651040e-01
6.58618569e-01 7.31565118e-01 1.05526912e+00 -3.95319343e-01
2.01588556e-01 6.23325348e-01 -9.07564610e-02 -6.72213256e-01
1.97709292e-01 -8.56707752e-01 5.32628715e-01 1.62661538e-01
-5.08069396e-01 -7.27780521e-01 -6.24465235e-02 2.32103929e-01
9.85980570e-01 -9.53245983e-02 3.51199180e-01 -1.00363386e+00
6.16175830e-01 5.31395137e-01 5.18471599e-01 3.41574818e-01
-6.46470189e-02 7.81161964e-01 3.96791622e-02 -5.27309954e-01
-1.24480486e+00 -8.62002790e-01 -1.06122899e+00 -1.01704672e-01
-2.11054593e-01 9.56158116e-02 -4.32181507e-01 -1.35075882e-01
-1.86577439e-01 5.87184727e-01 -8.40304971e-01 3.37591708e-01
-8.99826467e-01 -1.06522894e+00 -1.00139275e-01 5.01978934e-01
1.73967019e-01 -6.14074707e-01 -7.67019212e-01 6.28815114e-01
1.29448697e-01 -6.11460209e-01 -2.97867239e-01 2.38859162e-01
-1.56073427e+00 -1.12795758e+00 -8.94124866e-01 -6.35934532e-01
1.15389216e+00 -2.29661062e-01 8.25964212e-01 2.87541747e-01
-8.36044133e-01 4.12821889e-01 -7.97008127e-02 -2.87967082e-03
-4.87957031e-01 -9.02188420e-01 -2.99132243e-02 -4.39344883e-01
-3.20917308e-01 -3.25693369e-01 -9.38713968e-01 2.50595540e-01
-9.09770787e-01 -1.70102566e-01 1.68252647e-01 1.34515929e+00
9.36358273e-01 -1.15565754e-01 1.47489175e-01 -1.16013265e+00
3.88750672e-01 -2.83610106e-01 -5.85797071e-01 -2.73249894e-01
-7.45860636e-01 -2.29387015e-01 4.71320540e-01 -1.38009816e-01
-1.00520480e+00 2.12958068e-01 -1.62646636e-01 -3.74222606e-01
4.72515583e-01 4.18113440e-01 7.06436217e-01 -7.06160367e-01
8.17272604e-01 -5.81292138e-02 1.88146293e-01 -4.32025015e-01
-1.79070562e-01 2.05736130e-01 1.54411882e-01 -4.08280045e-01
4.20414001e-01 1.07863569e+00 7.95382321e-01 -1.14709198e+00
-3.57465386e-01 -5.22910535e-01 -8.56691599e-01 -2.52306938e-01
1.02192044e+00 -4.07790601e-01 -3.47295970e-01 2.79578399e-02
-8.68094146e-01 -1.98231846e-01 -6.78081155e-01 1.14555109e+00
-7.51562655e-01 5.30323923e-01 -9.79676127e-01 -7.25868225e-01
-4.48659539e-01 -1.51659739e+00 4.44670081e-01 -3.25833470e-01
-1.24147467e-01 -1.30688190e+00 3.81691217e-01 3.41164857e-01
5.42882442e-01 6.85983777e-01 7.53838480e-01 1.51582479e-01
-2.93422073e-01 -2.97228485e-01 4.17012796e-02 5.28632998e-01
7.90104419e-02 6.71820994e-03 -4.75491107e-01 -4.67689008e-01
9.51368630e-01 1.52750865e-01 6.25752389e-01 1.19976556e+00
9.49375212e-01 -2.24864736e-01 -3.78796875e-01 6.85889006e-01
2.11344385e+00 1.29403949e-01 5.62769532e-01 6.29128516e-02
5.82580626e-01 3.83876175e-01 4.37127709e-01 7.00354278e-01
-2.36664712e-01 3.40870291e-01 2.06533030e-01 -2.58760512e-01
-1.41868666e-01 4.45762217e-01 -3.64580005e-01 8.97330582e-01
-3.61156940e-01 4.01505888e-01 -1.18897033e+00 6.77675545e-01
-1.48888469e+00 -6.18830562e-01 -1.35756910e+00 2.00633335e+00
5.04220903e-01 -1.36177063e-01 -2.60372460e-01 4.14149493e-01
4.15174603e-01 -3.31608564e-01 -1.15593277e-01 -5.36386847e-01
3.35120738e-01 5.52287042e-01 9.31978583e-01 1.11455452e+00
-7.34536707e-01 1.65817603e-01 6.96813869e+00 8.32900763e-01
-8.65361154e-01 2.96663374e-01 5.35364866e-01 2.61538811e-02
-2.59177953e-01 9.29074511e-02 -8.29151496e-02 4.11863983e-01
4.93061334e-01 1.15058437e-01 1.05289571e-01 3.53041470e-01
7.60370553e-01 -7.36178219e-01 -4.78159189e-01 7.71532893e-01
-8.92747939e-02 -1.33878541e+00 -5.30921280e-01 5.94875850e-02
9.34411168e-01 5.45297703e-03 5.33549748e-02 -3.39326769e-01
-4.17441763e-02 -8.98630977e-01 2.30821237e-01 4.61706787e-01
7.83565938e-01 -5.89163840e-01 4.58803564e-01 -1.92039479e-02
-1.04599690e+00 2.50395328e-01 -4.15104389e-01 -4.29219231e-02
5.89102328e-01 9.80874479e-01 -9.62294519e-01 6.05855227e-01
2.19640642e-01 3.99212956e-01 5.47750071e-02 9.55570102e-01
1.06109917e-01 4.93239224e-01 -5.32462716e-01 4.95992362e-01
3.19968104e-01 -7.93624341e-01 7.03796804e-01 8.61054301e-01
1.86337471e-01 9.08343613e-01 -1.65634323e-02 5.88156521e-01
5.81942439e-01 3.14315140e-01 -3.16997200e-01 8.17252934e-01
4.01492156e-02 9.50412393e-01 -9.10686314e-01 -3.81788254e-01
-5.22823989e-01 7.89299250e-01 -4.36683297e-01 3.17953020e-01
-4.88370657e-01 1.85226664e-01 3.91050130e-02 6.50403619e-01
-6.97825849e-02 -2.59050101e-01 -1.04995298e+00 -7.49534070e-01
-1.76867977e-01 -2.19970927e-01 6.06806576e-01 -5.37985504e-01
-1.35863638e+00 5.82910657e-01 4.89703976e-02 -1.32376528e+00
2.50066221e-01 -5.49331129e-01 -6.19248867e-01 1.13174784e+00
-1.07437098e+00 -8.99684846e-01 5.95322438e-02 7.23005891e-01
4.89534378e-01 2.76038051e-01 3.92258972e-01 4.39916730e-01
-1.97224632e-01 -9.20356065e-03 5.12136638e-01 -1.63929313e-01
3.69967252e-01 -9.99777555e-01 -3.38979751e-01 6.39531434e-01
-7.44632781e-01 5.49050391e-01 1.05543411e+00 -1.05382895e+00
-1.24625015e+00 -9.20655310e-01 6.19260132e-01 -1.67395428e-01
6.27456725e-01 2.67954588e-01 -9.59113538e-01 9.03337598e-01
1.50218561e-01 6.19808555e-01 7.23141074e-01 -5.69727123e-01
5.64919949e-01 2.96879202e-01 -1.69393659e+00 5.61071821e-02
1.96094409e-01 -4.56967540e-02 -4.04986382e-01 6.65985227e-01
-5.81533881e-03 -7.25898921e-01 -1.09989715e+00 4.51070279e-01
9.51332599e-02 -6.18909478e-01 1.00563467e+00 -2.27372438e-01
4.96851772e-01 -1.61082506e-01 -1.34870142e-01 -1.06359518e+00
-3.05956781e-01 -6.36568725e-01 6.05741382e-01 3.44409823e-01
3.24279249e-01 -4.35263604e-01 7.54952252e-01 8.67469966e-01
-6.05575025e-01 -9.61744368e-01 -1.43875575e+00 -6.47458911e-01
4.61060643e-01 -4.59050477e-01 -5.09357214e-01 1.15539384e+00
3.07525665e-01 -7.20573813e-02 -3.98763001e-01 6.59851730e-02
1.12031734e+00 -4.46091831e-01 1.83499515e-01 -8.51146340e-01
-2.18971714e-01 1.41388342e-01 -1.77411228e-01 -5.11434376e-01
-5.83908677e-01 -9.02592421e-01 4.82886396e-02 -1.65247059e+00
5.80852367e-02 -7.76984394e-01 8.99443477e-02 -8.74507427e-02
-9.54719037e-02 4.21475559e-01 -8.37478265e-02 3.61304581e-01
3.97196114e-01 3.76836777e-01 1.71831155e+00 2.53060102e-01
-3.21544766e-01 -7.35305175e-02 2.59809673e-01 9.43256974e-01
4.01888400e-01 -7.77455151e-01 -1.93259969e-01 -4.42014664e-01
-1.69808362e-02 8.73482108e-01 2.66078621e-01 -8.99966061e-01
2.68255807e-02 -1.00377686e-01 4.51305300e-01 -6.55830383e-01
4.46168780e-01 -1.04123867e+00 5.19464672e-01 1.03780448e+00
-1.11150704e-01 1.25330746e-01 1.66386545e-01 3.31495345e-01
-6.56860024e-02 -5.83589911e-01 1.56452096e+00 -4.07364875e-01
9.74787101e-02 2.87996173e-01 -5.43987811e-01 -1.92979887e-01
1.15284932e+00 -3.55117410e-01 4.01004523e-01 5.39491251e-02
-1.31402266e+00 -5.98227419e-02 3.68622899e-01 -8.71267021e-01
8.62362623e-01 -1.31254256e+00 -9.67594028e-01 3.79277617e-01
-5.35229981e-01 1.77802891e-01 6.83643818e-01 1.67699671e+00
-1.31235480e+00 1.83342695e-01 -3.97312865e-02 -7.86577046e-01
-1.30593979e+00 3.96474689e-01 4.97817367e-01 -4.50184703e-01
-9.67464864e-01 1.08530486e+00 -1.40011394e-02 6.68277033e-03
-3.64137441e-01 -3.20305794e-01 3.54821324e-01 -2.44170472e-01
2.32387364e-01 9.56308365e-01 3.08988839e-01 -6.95843041e-01
-5.35983384e-01 8.37851703e-01 8.19983482e-02 -3.94855887e-01
1.55959511e+00 -3.00732911e-01 -3.24387610e-01 1.05838627e-01
1.30710816e+00 1.58774182e-01 -1.22357404e+00 -3.42001580e-02
-1.73520878e-01 -7.07357645e-01 4.34429914e-01 -5.23648918e-01
-1.08289480e+00 6.79981947e-01 6.14440203e-01 -2.62107581e-01
9.15472686e-01 -4.05884057e-01 5.87458611e-01 -3.76745045e-01
7.32950047e-02 -1.25745392e+00 -6.41126409e-02 2.24706456e-01
9.78146136e-01 -1.13244545e+00 1.01611209e+00 -5.90168774e-01
-7.43140280e-01 1.18963027e+00 -5.24744652e-02 -7.79263794e-01
1.10683894e+00 4.26183879e-01 1.12523176e-01 -3.79425257e-01
-3.55967730e-01 2.82239527e-01 1.24705471e-01 2.46452734e-01
7.74562120e-01 -3.27405855e-02 -1.18799174e+00 -1.34922028e-01
4.00614738e-01 1.59792483e-01 7.44251132e-01 1.45661163e+00
-3.02052766e-01 -9.04553294e-01 -9.53187883e-01 5.83867908e-01
-6.97796464e-01 -6.01780899e-02 5.30308247e-01 9.26898718e-01
-9.93570834e-02 8.26463401e-01 -2.16490984e-01 4.31458443e-01
3.04194212e-01 -4.19874221e-01 6.46250606e-01 -5.50992191e-01
-9.17338252e-01 9.17481065e-01 -9.94096547e-02 -4.71112967e-01
-3.68592948e-01 -7.63437152e-01 -1.60591745e+00 -8.74833390e-02
-5.31882882e-01 2.86680669e-01 8.16242218e-01 8.10341001e-01
-3.23528618e-01 7.34546065e-01 6.69044316e-01 -6.91240251e-01
-2.58756101e-01 -9.54579830e-01 -8.94660711e-01 3.07918936e-01
3.23474735e-01 -5.07427514e-01 -5.42433977e-01 1.52013868e-01] | [13.118269920349121, -2.6102638244628906] |
6e75d3cc-ecff-4fae-9518-cc2063149c84 | distilling-token-pruned-pose-transformer-for | 2304.05548 | null | https://arxiv.org/abs/2304.05548v1 | https://arxiv.org/pdf/2304.05548v1.pdf | Distilling Token-Pruned Pose Transformer for 2D Human Pose Estimation | Human pose estimation has seen widespread use of transformer models in recent years. Pose transformers benefit from the self-attention map, which captures the correlation between human joint tokens and the image. However, training such models is computationally expensive. The recent token-Pruned Pose Transformer (PPT) solves this problem by pruning the background tokens of the image, which are usually less informative. However, although it improves efficiency, PPT inevitably leads to worse performance than TokenPose due to the pruning of tokens. To overcome this problem, we present a novel method called Distilling Pruned-Token Transformer for human pose estimation (DPPT). Our method leverages the output of a pre-trained TokenPose to supervise the learning process of PPT. We also establish connections between the internal structure of pose transformers and PPT, such as attention maps and joint features. Our experimental results on the MPII datasets show that our DPPT can significantly improve PCK compared to previous PPT models while still reducing computational complexity. | ['Feixiang Ren'] | 2023-04-12 | null | null | null | null | ['2d-human-pose-estimation'] | ['computer-vision'] | [ 1.49652809e-01 3.18428576e-01 3.90249230e-02 -1.77363873e-01
-7.50187457e-01 -2.12360248e-01 3.88032466e-01 -1.64619699e-01
-4.43870276e-01 4.70712006e-01 4.89301175e-01 3.17057580e-01
2.75669396e-01 -7.30731606e-01 -1.06792498e+00 -4.83095407e-01
-2.58133262e-02 6.28923416e-01 4.90516245e-01 9.53752548e-03
-1.38706684e-01 -2.42226645e-02 -1.44552684e+00 2.95533806e-01
6.30007923e-01 1.04651582e+00 3.19827467e-01 4.58700210e-01
9.47661474e-02 8.01228404e-01 -6.60846174e-01 -7.21380949e-01
2.95625180e-01 -1.51038915e-01 -9.47966218e-01 -8.92505646e-02
6.80386007e-01 -5.74132979e-01 -5.42128921e-01 6.35236323e-01
6.51664257e-01 5.69652058e-02 4.41985697e-01 -1.27828836e+00
-1.78948507e-01 6.98988199e-01 -7.13779926e-01 1.85869426e-01
1.97337970e-01 2.22019419e-01 1.24119782e+00 -9.48365510e-01
5.21375954e-01 1.50056374e+00 1.05483949e+00 3.04108351e-01
-1.00221860e+00 -6.72118068e-01 4.12294179e-01 2.61539817e-01
-1.24833477e+00 -5.93463965e-02 6.26873434e-01 -6.49542958e-02
1.07980764e+00 1.19336441e-01 1.21065569e+00 1.16903281e+00
3.33356529e-01 1.54852569e+00 7.70965040e-01 -2.57939190e-01
-3.17827880e-01 -1.73412114e-01 -1.50806487e-01 1.03227544e+00
3.52610387e-02 -6.08910806e-02 -9.36695814e-01 9.73113179e-02
9.07843530e-01 -5.22646606e-02 -1.29275799e-01 -6.46220982e-01
-1.22675300e+00 4.98908460e-01 7.78198600e-01 -1.19745798e-01
-2.78722227e-01 5.66744149e-01 3.73586297e-01 -1.80884019e-01
5.02790034e-01 4.62263763e-01 -4.30590689e-01 -2.95827180e-01
-7.42697418e-01 4.95699167e-01 4.48904693e-01 8.15212607e-01
6.49606526e-01 -2.18071252e-01 -5.70616007e-01 7.72899508e-01
3.12595963e-01 4.85878348e-01 2.06952468e-01 -9.63781238e-01
7.08496392e-01 5.60263753e-01 1.86617170e-02 -8.53494167e-01
-3.13435704e-01 -6.63142681e-01 -3.32660079e-01 -4.41083275e-02
4.65501875e-01 -1.34961726e-02 -1.00859427e+00 1.74326265e+00
4.89140868e-01 8.94739702e-02 -5.48105597e-01 8.72354627e-01
6.86789870e-01 5.33161879e-01 2.99428374e-01 5.45767367e-01
1.38533807e+00 -1.24018955e+00 -4.46711212e-01 -4.42968130e-01
4.81330425e-01 -4.94061232e-01 9.85273421e-01 2.78766543e-01
-1.06346512e+00 -5.22485077e-01 -8.95354927e-01 -3.85971636e-01
-4.22166251e-02 3.40471953e-01 6.61638856e-01 5.50579607e-01
-9.22638535e-01 7.60522723e-01 -1.10693312e+00 -4.24281985e-01
7.06115782e-01 5.22401690e-01 -2.68950313e-01 -3.94151136e-02
-1.12364030e+00 1.04270232e+00 1.58810258e-01 2.93588519e-01
-8.54725301e-01 -7.53206253e-01 -1.06273007e+00 3.12778234e-01
5.39782822e-01 -8.23923290e-01 1.47991085e+00 -6.67782128e-01
-1.54790556e+00 6.08169615e-01 -2.03053772e-01 -6.00993574e-01
7.93344021e-01 -1.02871251e+00 4.05445665e-01 2.65937299e-01
2.15166748e-01 1.15723336e+00 8.66159737e-01 -9.63098109e-01
-6.97122872e-01 -2.10479394e-01 9.81310531e-02 5.75609744e-01
-3.40292633e-01 -1.57317832e-01 -1.01583910e+00 -5.05783319e-01
-1.17245011e-01 -9.91558671e-01 -3.15576941e-01 8.83390605e-02
-6.35692716e-01 -3.70010793e-01 7.66816080e-01 -7.27527082e-01
9.34986234e-01 -2.01554704e+00 1.73415691e-01 1.30740508e-01
4.85256106e-01 1.94820136e-01 -2.14226112e-01 2.61408836e-01
1.53279290e-01 -1.82732031e-01 1.61393702e-01 -8.08418155e-01
1.00126758e-01 4.20919836e-01 -2.62383491e-01 3.68531048e-01
3.42425704e-01 1.24505281e+00 -8.32750797e-01 -8.43696892e-01
3.86971295e-01 5.29558241e-01 -9.49625015e-01 1.00750178e-01
-3.10633481e-01 3.42527777e-01 -4.23666477e-01 5.13911784e-01
4.49385107e-01 -3.07854593e-01 1.18656028e-02 -3.54662627e-01
4.26471978e-02 6.86695218e-01 -7.47439265e-01 1.62341058e+00
-2.11044356e-01 6.15038872e-01 -1.86016753e-01 -8.65265071e-01
4.43948418e-01 1.86568350e-01 6.95649326e-01 -6.53903961e-01
3.26639920e-01 -2.41004348e-01 -1.30246803e-01 -2.35360786e-01
7.47373939e-01 1.19287446e-01 -1.35179624e-01 2.38471776e-01
1.51972935e-01 1.13335967e-01 1.38481095e-01 3.59097809e-01
1.05354619e+00 7.20408976e-01 -7.71624828e-03 -2.54639122e-03
-7.88122267e-02 -9.66852382e-02 7.11748302e-01 7.61247277e-01
-3.29495162e-01 7.63282120e-01 5.69114506e-01 -4.35837835e-01
-9.19435203e-01 -1.37280798e+00 1.62556112e-01 1.27525473e+00
1.63535044e-01 -8.21758151e-01 -9.02424157e-01 -8.25898886e-01
3.95907983e-02 1.71147451e-01 -7.31545329e-01 -2.51026869e-01
-8.11227381e-01 -4.82623875e-01 6.15657389e-01 1.02666068e+00
8.02529395e-01 -1.05424511e+00 -1.02063131e+00 1.56418204e-01
-7.10887730e-01 -1.11371827e+00 -6.20129049e-01 1.06321193e-01
-7.05371082e-01 -1.01829469e+00 -9.48353589e-01 -6.22700572e-01
6.40585780e-01 1.69420004e-01 1.22335637e+00 -1.50726989e-01
-4.88326728e-01 5.08999348e-01 -2.62793869e-01 -5.12035370e-01
3.02464247e-01 3.59043479e-01 -3.66092138e-02 -2.49273092e-01
2.18995705e-01 -3.64274383e-01 -7.01856732e-01 3.87264848e-01
-3.84805560e-01 2.89806902e-01 9.58496928e-01 9.55750585e-01
5.02048731e-01 -7.31577948e-02 -4.17774431e-02 -6.73299193e-01
1.46342203e-01 7.81703088e-03 -3.96961838e-01 1.07184865e-01
-7.80900419e-02 3.86324883e-01 2.64553636e-01 -5.09742260e-01
-1.07606900e+00 3.05158287e-01 -1.75381988e-01 -6.05047703e-01
3.29171181e-01 3.39154124e-01 -1.57335013e-01 -2.93219239e-02
4.00754780e-01 1.01401612e-01 -1.53229862e-01 -5.23839235e-01
1.67560905e-01 8.99348259e-02 7.60101259e-01 -8.08209062e-01
8.81606340e-01 4.26707119e-01 -7.78122991e-02 -6.83888495e-01
-1.16248477e+00 -3.83852869e-01 -4.60340083e-01 -2.29411319e-01
8.10139418e-01 -1.03740168e+00 -8.49855125e-01 3.74087602e-01
-1.06088543e+00 -4.54061806e-01 -4.09118384e-01 4.54794347e-01
-5.48646927e-01 3.74318302e-01 -8.01271856e-01 -6.79733217e-01
-3.44359696e-01 -1.02590489e+00 1.37899876e+00 -2.79349517e-02
-4.70341146e-01 -6.11035824e-01 -3.84336039e-02 4.79310662e-01
2.00178102e-01 2.03067422e-01 6.76901817e-01 -2.25818992e-01
-8.46194923e-01 -2.72654384e-01 -1.76423773e-01 5.19395955e-02
-1.64371088e-01 -2.90770054e-01 -1.02994418e+00 -2.63107270e-01
-4.30394262e-01 -5.76565087e-01 1.06387532e+00 5.52141488e-01
1.20209742e+00 -2.67885476e-01 -7.03850269e-01 5.46352267e-01
8.71432066e-01 -4.67629433e-01 7.26001322e-01 3.40734214e-01
9.49451625e-01 5.98109663e-01 6.93067670e-01 3.49486083e-01
6.25136495e-01 9.27585483e-01 4.24064666e-01 -3.39506328e-01
-3.35992873e-01 -7.33540416e-01 6.11690342e-01 3.95080477e-01
-3.97453964e-01 -7.07377791e-02 -7.68334806e-01 5.79411328e-01
-1.81871200e+00 -9.99694109e-01 2.17893362e-01 2.05361032e+00
7.94202924e-01 3.25346202e-01 3.19773763e-01 2.13033184e-01
3.48034471e-01 1.07144572e-01 -3.91991317e-01 2.71129422e-02
2.25594297e-01 4.14329380e-01 5.99977195e-01 3.28631997e-01
-1.33484817e+00 1.19935191e+00 6.45283747e+00 7.39415169e-01
-8.85803401e-01 5.12621030e-02 3.55168760e-01 -3.74684334e-01
1.49700388e-01 -2.25458995e-01 -9.61997032e-01 2.45077848e-01
4.56811607e-01 5.61440438e-02 4.32383735e-03 1.03660166e+00
-1.60732508e-01 -2.90607780e-01 -1.25948286e+00 9.53014374e-01
8.87508169e-02 -9.03964996e-01 1.21157587e-01 2.10366175e-01
3.69978100e-01 1.19057402e-01 3.12384665e-01 6.48375571e-01
4.51285511e-01 -9.99784529e-01 1.12705469e+00 8.65730569e-02
6.27203941e-01 -9.00510311e-01 7.08362162e-01 2.24281669e-01
-1.53723574e+00 -8.00638273e-02 -3.73935193e-01 -7.49799535e-02
8.34922120e-03 4.25758481e-01 -1.03263211e+00 4.03158516e-01
1.05129695e+00 8.07918549e-01 -6.68035209e-01 1.17248702e+00
-5.10352373e-01 4.86482471e-01 -5.96980572e-01 3.74452397e-02
1.79859146e-01 3.36423099e-01 5.33842623e-01 1.06420052e+00
1.01867750e-01 -1.14198111e-01 4.41208571e-01 8.20590913e-01
-2.26811599e-03 -2.77939886e-01 -3.73982489e-01 9.32422057e-02
3.01234782e-01 1.08794248e+00 -6.44629538e-01 -3.53576422e-01
-2.56919146e-01 1.17348742e+00 5.60587645e-01 1.70649961e-01
-1.05114841e+00 -1.85362011e-01 6.33663833e-01 4.50447261e-01
7.33853579e-01 -1.73799172e-01 -1.44321859e-01 -1.00606370e+00
1.89068735e-01 -6.92788303e-01 3.50460231e-01 -6.32262647e-01
-8.53289604e-01 4.36529189e-01 2.57517666e-01 -1.09918416e+00
-1.77543581e-01 -6.12637818e-01 -3.75367284e-01 5.96150935e-01
-1.28574204e+00 -1.50894272e+00 -2.95688480e-01 5.00455439e-01
5.01855433e-01 5.85577190e-01 5.85788786e-01 1.41937003e-01
-6.51165545e-01 9.51889575e-01 -3.83907378e-01 3.74320626e-01
6.06502235e-01 -1.36589432e+00 7.21461594e-01 7.32435703e-01
1.26689255e-01 7.26419806e-01 7.21667111e-01 -6.84210598e-01
-1.08567870e+00 -1.03108466e+00 9.21269238e-01 -7.18228281e-01
2.15578243e-01 -6.38252676e-01 -4.30569857e-01 1.03870630e+00
8.32243711e-02 -9.71581042e-02 3.40339541e-01 5.26247323e-01
-4.30888295e-01 -4.35838364e-02 -7.31592715e-01 7.88373172e-01
1.26641738e+00 -3.11289638e-01 -6.64188266e-01 1.70384929e-01
4.65482354e-01 -6.81296647e-01 -5.48495054e-01 2.78787613e-01
8.40574324e-01 -8.75310779e-01 1.31533599e+00 -1.40687242e-01
5.08776367e-01 -8.83714333e-02 2.82798588e-01 -1.29075003e+00
-4.85468864e-01 -5.00348985e-01 -1.97313488e-01 9.35174823e-01
2.29178503e-01 -3.87250394e-01 1.16454601e+00 6.23398900e-01
-2.02918112e-01 -9.47103381e-01 -9.68823254e-01 -8.99683595e-01
-8.83246213e-02 -3.02112490e-01 3.46600562e-01 4.14257556e-01
1.96494266e-01 4.64178920e-01 -6.13516629e-01 -2.61140987e-02
7.31175661e-01 -1.56098515e-01 1.03979719e+00 -1.19047272e+00
-4.88002747e-01 -1.86023042e-01 -4.71677631e-01 -1.53999853e+00
-1.08120441e-02 -6.29635394e-01 4.91576374e-01 -1.58266449e+00
4.30389494e-01 -3.59139144e-01 4.39287536e-02 8.56546462e-01
-4.53355491e-01 3.27773899e-01 3.83064449e-01 1.12931028e-01
-8.84175062e-01 8.50004137e-01 1.41508734e+00 -1.59891516e-01
6.11923151e-02 -9.07967761e-02 -3.83510381e-01 8.79486263e-01
5.95702171e-01 -5.21399021e-01 -4.21126634e-01 -5.81179380e-01
-2.07784912e-03 -3.11265111e-01 5.47236919e-01 -1.42483485e+00
1.65412173e-01 2.74533689e-01 7.87934601e-01 -1.03236878e+00
6.92843020e-01 -5.91468751e-01 -2.18155101e-01 7.25007892e-01
-2.24348098e-01 7.48969167e-02 1.60507619e-01 4.95813727e-01
-9.85886063e-03 3.02401125e-01 6.62365258e-01 -2.52709955e-01
-5.37702918e-01 4.41751212e-01 -2.80060261e-01 9.96493250e-02
6.47531927e-01 -4.51978356e-01 -4.92522977e-02 -3.63745719e-01
-4.20337170e-01 4.67814296e-01 2.39904776e-01 3.18014294e-01
6.11222386e-01 -1.27220190e+00 -4.50683326e-01 7.18565136e-02
3.42595950e-02 3.43052357e-01 1.59439221e-01 9.62081969e-01
-3.78793150e-01 5.49119949e-01 -2.35902339e-01 -7.13202834e-01
-1.45675051e+00 3.80331695e-01 2.98192561e-01 -7.44953096e-01
-1.06389081e+00 1.21565735e+00 5.32195270e-01 -2.71351337e-01
5.10647058e-01 -5.57856321e-01 1.30305933e-02 -2.69048274e-01
3.64055932e-01 2.04964399e-01 -1.79829165e-01 -5.51308513e-01
-4.16767538e-01 5.37978351e-01 -4.33893174e-01 -2.95900494e-01
1.21178138e+00 6.66031763e-02 2.44922504e-01 4.95302603e-02
1.03623378e+00 -1.26197878e-02 -1.76435137e+00 -4.09729987e-01
-2.27654457e-01 -5.69627941e-01 -2.03410745e-01 -6.36149406e-01
-1.18139744e+00 9.37251747e-01 3.46602410e-01 -4.91632462e-01
8.39918971e-01 1.22029539e-02 1.20910442e+00 4.56252426e-01
5.49408615e-01 -1.23839295e+00 5.69263339e-01 5.90955436e-01
7.22349763e-01 -9.66292620e-01 1.10792227e-01 -6.79106653e-01
-6.36518896e-01 6.03461742e-01 1.02433717e+00 -1.86737314e-01
2.39783779e-01 2.65891373e-01 -2.02492014e-01 -2.66546428e-01
-6.20258868e-01 -2.58384556e-01 3.25233161e-01 7.00571179e-01
3.75460327e-01 -1.73118904e-01 -1.21574961e-02 5.72848976e-01
-5.28637171e-01 -2.70865951e-02 -5.89915030e-02 1.11614394e+00
-4.70945477e-01 -1.01207244e+00 -4.84736770e-01 3.77851129e-01
-4.50363994e-01 -1.92083254e-01 -4.30081278e-01 8.22726011e-01
2.62456447e-01 5.92474461e-01 1.01309657e-01 -5.33502638e-01
3.31270993e-01 -3.48940939e-02 8.91757369e-01 -5.73567390e-01
-7.79770494e-01 1.13476403e-01 1.90067843e-01 -9.60636795e-01
-2.23999947e-01 -6.62645340e-01 -1.12310040e+00 -2.65432924e-01
-3.36329997e-01 4.59981821e-02 1.89082205e-01 1.07682407e+00
2.60427564e-01 6.73323691e-01 7.70420656e-02 -1.17708242e+00
-5.61182141e-01 -1.04072559e+00 -3.43876988e-01 2.82821447e-01
1.45326361e-01 -1.10604012e+00 1.91571981e-01 -6.20555021e-02] | [7.125339508056641, -0.7544150352478027] |
2520f5c5-bed4-4b51-9ff9-46453ca1a2a8 | look-ma-no-hands-agent-environment | 2305.16301 | null | https://arxiv.org/abs/2305.16301v1 | https://arxiv.org/pdf/2305.16301v1.pdf | Look Ma, No Hands! Agent-Environment Factorization of Egocentric Videos | The analysis and use of egocentric videos for robotic tasks is made challenging by occlusion due to the hand and the visual mismatch between the human hand and a robot end-effector. In this sense, the human hand presents a nuisance. However, often hands also provide a valuable signal, e.g. the hand pose may suggest what kind of object is being held. In this work, we propose to extract a factored representation of the scene that separates the agent (human hand) and the environment. This alleviates both occlusion and mismatch while preserving the signal, thereby easing the design of models for downstream robotics tasks. At the heart of this factorization is our proposed Video Inpainting via Diffusion Model (VIDM) that leverages both a prior on real-world images (through a large-scale pre-trained diffusion model) and the appearance of the object in earlier frames of the video (through attention). Our experiments demonstrate the effectiveness of VIDM at improving inpainting quality on egocentric videos and the power of our factored representation for numerous tasks: object detection, 3D reconstruction of manipulated objects, and learning of reward functions, policies, and affordances from videos. | ['Saurabh Gupta', 'Aditya Prakash', 'Matthew Chang'] | 2023-05-25 | null | null | null | null | ['video-inpainting', '3d-reconstruction'] | ['computer-vision', 'computer-vision'] | [-1.63842775e-02 8.33287835e-02 -6.80442452e-02 1.23769499e-01
-1.85632244e-01 -5.78782141e-01 5.20016432e-01 -4.56478029e-01
-3.77617598e-01 3.56677324e-01 6.01896048e-01 1.76779568e-01
-4.23268639e-02 -1.68197304e-01 -9.42417085e-01 -6.81259692e-01
-9.55237076e-03 2.29649678e-01 3.15332673e-02 -2.51558647e-02
1.94053054e-01 5.86144447e-01 -1.32501280e+00 2.32955083e-01
2.11189106e-01 8.90696585e-01 7.33530581e-01 7.96448708e-01
7.14628398e-01 1.18540525e+00 -4.05825943e-01 -1.34755671e-02
5.74242353e-01 -2.19663486e-01 -3.93896103e-01 4.54078883e-01
4.19793397e-01 -9.70769763e-01 -1.08530414e+00 8.89460266e-01
2.46358171e-01 4.45238173e-01 7.49576211e-01 -1.24954009e+00
-3.54497910e-01 3.32597077e-01 -7.72015095e-01 1.34895056e-01
2.84712911e-01 6.19319677e-01 9.13145959e-01 -6.88373983e-01
1.12972105e+00 1.52508092e+00 1.57948583e-01 4.75107521e-01
-1.20237219e+00 -4.26821798e-01 4.63106751e-01 1.45581186e-01
-8.93208265e-01 -6.65394962e-01 7.93012023e-01 -8.53674710e-01
7.42685854e-01 -1.59241274e-01 8.29535067e-01 1.48435855e+00
4.32458252e-01 1.06756377e+00 5.01442134e-01 -2.01074764e-01
1.65903449e-01 -2.99332798e-01 -3.71197104e-01 6.03594124e-01
1.76889505e-02 1.19452663e-01 -6.50923431e-01 -1.47393152e-01
1.34783053e+00 3.78565669e-01 -5.35154521e-01 -1.22712481e+00
-1.23391283e+00 4.47483301e-01 4.10279959e-01 -3.54283564e-02
-8.92327487e-01 4.80504721e-01 1.60150155e-01 6.33522198e-02
1.08123094e-01 4.60832328e-01 -3.58810395e-01 -1.36397704e-01
-3.01611185e-01 4.48396891e-01 5.00768125e-01 1.06466043e+00
3.59748870e-01 6.49712831e-02 -3.23569506e-01 3.57045054e-01
2.91629076e-01 3.58634114e-01 1.69325829e-01 -1.53842223e+00
6.17763579e-01 2.90496141e-01 4.94721264e-01 -9.33594584e-01
-2.80445814e-01 -2.74259329e-01 -5.67348711e-02 5.04418373e-01
6.86102748e-01 -8.47281292e-02 -7.14960635e-01 2.06124902e+00
4.80712026e-01 -7.79588148e-02 -1.35334998e-01 1.34958518e+00
3.10782883e-02 4.61290151e-01 1.48510396e-01 1.68756023e-01
1.32389963e+00 -9.58725393e-01 -6.01820827e-01 -7.05958247e-01
2.16725230e-01 -5.57544291e-01 8.18702757e-01 3.85540307e-01
-1.10303545e+00 -3.13161284e-01 -8.02255452e-01 -3.15257788e-01
2.31683210e-01 1.08611725e-01 5.21350265e-01 -9.89362225e-02
-5.44281900e-01 6.51128829e-01 -1.09742820e+00 -3.31504762e-01
4.10388410e-01 2.20295459e-01 -6.69227779e-01 -3.56240571e-01
-4.36119229e-01 1.06612051e+00 -9.26342979e-02 9.63150039e-02
-1.31077218e+00 -5.23516655e-01 -9.32602227e-01 1.96993575e-01
7.01064527e-01 -7.97864079e-01 1.22678149e+00 -1.20166087e+00
-1.16640401e+00 4.11090791e-01 9.49354917e-02 -2.97702640e-01
7.82382190e-01 -6.30879104e-01 3.90850276e-01 5.32841444e-01
7.50203580e-02 8.14420998e-01 1.45874119e+00 -1.23706293e+00
-4.05774534e-01 -8.04013431e-01 2.90373713e-01 4.09008950e-01
-2.59473082e-02 -3.30377966e-01 -5.31144857e-01 -9.68252778e-01
-5.57927787e-02 -1.12839031e+00 -3.66797931e-02 5.85702240e-01
8.89071822e-03 4.09534909e-02 9.41938043e-01 -1.11666787e+00
4.01191741e-01 -2.36585474e+00 7.70890713e-01 2.12492812e-02
2.77529389e-01 -1.25254780e-01 -4.09135342e-01 3.37455332e-01
-1.16005711e-01 -6.99107468e-01 2.21241355e-01 -3.61264348e-01
-1.15890488e-01 2.21913978e-01 -3.81579459e-01 8.06217253e-01
2.64737070e-01 1.01865244e+00 -1.13006651e+00 -1.03913754e-01
2.70288169e-01 8.14221561e-01 -8.28240931e-01 4.35302913e-01
-4.67761755e-01 8.31913590e-01 -5.06439924e-01 4.10233617e-01
2.43771866e-01 4.35909852e-02 2.04137787e-01 -3.97804946e-01
1.48935691e-01 2.59500444e-01 -9.41698015e-01 2.12874103e+00
-3.48578185e-01 5.49237251e-01 6.13584816e-01 -5.66314876e-01
2.56743848e-01 2.50917554e-01 7.32765496e-01 -5.50181568e-01
4.00257587e-01 -2.20191687e-01 2.31224492e-01 -8.14762831e-01
4.31498826e-01 2.72467285e-01 3.44515204e-01 5.69997430e-01
2.14903474e-01 -1.46524712e-01 1.00079877e-02 3.41785014e-01
1.22008252e+00 8.77151132e-01 2.33853832e-02 6.85568005e-02
-2.40265667e-01 -1.20746903e-01 4.35338646e-01 5.35748363e-01
-3.67843390e-01 5.17394900e-01 6.71403289e-01 -2.00150073e-01
-1.30990863e+00 -7.63440669e-01 5.25069177e-01 1.10466790e+00
7.28099495e-02 -8.18981752e-02 -7.08575189e-01 -4.71482009e-01
2.36490324e-01 6.16211116e-01 -7.44257689e-01 -4.76086795e-01
-7.11829185e-01 7.04452023e-02 -5.00586592e-02 6.72693968e-01
-9.34313238e-02 -9.70095277e-01 -1.08530986e+00 8.81752521e-02
-3.46896827e-01 -1.33669901e+00 -8.71980846e-01 2.29894761e-02
-8.91588688e-01 -1.15403211e+00 -8.12395871e-01 -5.78233480e-01
7.47216642e-01 7.16835856e-01 6.55083835e-01 -5.61198533e-01
-3.76771867e-01 8.65269363e-01 -2.35155880e-01 -9.29623544e-02
-2.05093578e-01 -5.41003346e-01 8.54688436e-02 2.18250573e-01
-1.36193842e-01 -7.71393955e-01 -7.11778402e-01 3.93383838e-02
-6.90970242e-01 1.19392179e-01 7.12702274e-01 7.90342569e-01
4.36304063e-02 -2.60542899e-01 -7.63366222e-02 -2.91253179e-01
3.15196037e-01 -3.68578881e-01 -5.58036625e-01 -4.76041883e-02
7.98434988e-02 2.32020944e-01 3.31621349e-01 -8.65872979e-01
-9.27953541e-01 4.84582186e-01 5.03761590e-01 -1.20988417e+00
2.83655435e-01 -9.28152278e-02 -2.21005574e-01 1.31841868e-01
5.05901396e-01 -1.79291725e-01 4.82791603e-01 -3.93177927e-01
6.34632766e-01 2.73311228e-01 7.36608028e-01 -6.45896614e-01
5.71194649e-01 7.08539307e-01 -3.96895036e-03 -7.07176924e-01
-4.22594666e-01 -3.40114325e-01 -5.70197821e-01 -3.44660252e-01
7.64817119e-01 -1.06553698e+00 -8.87606919e-01 2.63410121e-01
-1.31802547e+00 -4.48177814e-01 -3.80653024e-01 7.17058778e-01
-9.19942737e-01 3.70947987e-01 -6.14383757e-01 -8.59693766e-01
5.99759109e-02 -1.27881360e+00 1.18570256e+00 -3.07322949e-01
-3.20461720e-01 -3.49778235e-01 -3.99808824e-01 4.18190092e-01
1.91528350e-02 1.95249483e-01 8.13557923e-01 9.53132752e-04
-8.26516628e-01 -2.01066568e-01 -5.21317497e-02 2.12772325e-01
7.27183968e-02 -3.53015989e-01 -1.03981090e+00 -3.59008789e-01
3.02001297e-01 -3.36203843e-01 7.76396930e-01 5.57160556e-01
7.87298262e-01 -4.11027014e-01 -1.80433676e-01 4.74559456e-01
7.86272824e-01 1.30574524e-01 3.79304618e-01 1.38029277e-01
8.84233773e-01 1.02460122e+00 7.00869203e-01 4.71434176e-01
2.41023704e-01 7.53491879e-01 8.43577921e-01 4.02021617e-01
-2.72692531e-01 -4.38179255e-01 6.00784004e-01 1.39066190e-01
-1.61507964e-01 -7.17416108e-02 -5.15054047e-01 4.15304124e-01
-2.08985543e+00 -8.64694715e-01 4.69123572e-01 2.05378461e+00
4.07280445e-01 -7.67550692e-02 1.53388903e-01 -8.49025100e-02
4.87115294e-01 1.79267526e-01 -1.13482392e+00 2.39382535e-01
9.10718814e-02 -4.26693887e-01 5.18378377e-01 5.05482137e-01
-9.81412709e-01 7.66840637e-01 5.19477415e+00 2.02349842e-01
-1.10642040e+00 4.39603105e-02 1.18182823e-01 -4.63598847e-01
1.97882101e-01 -3.12728360e-02 -2.43431881e-01 1.55220062e-01
2.41011754e-01 2.37398028e-01 8.47634315e-01 8.55817139e-01
3.67201507e-01 -1.56976655e-01 -1.54039693e+00 1.02197134e+00
1.76171035e-01 -7.04654515e-01 -8.36104900e-02 2.53654212e-01
4.44511682e-01 8.60927477e-02 2.51488656e-01 -2.74192616e-02
1.56371415e-01 -4.60924387e-01 1.35889637e+00 4.84556675e-01
3.28630865e-01 -2.93531984e-01 1.82107747e-01 4.22614843e-01
-7.73907542e-01 -6.14164591e-01 -1.13837816e-01 -2.20895275e-01
7.69156637e-03 1.23569928e-01 -6.40300393e-01 9.22575668e-02
3.82836223e-01 8.80479097e-01 -1.04439758e-01 5.71560085e-01
-2.75134176e-01 -5.61244339e-02 -2.33673811e-01 4.11785811e-01
1.93876415e-01 -1.79850861e-01 9.78070557e-01 5.93773901e-01
-1.60308927e-02 3.43019009e-01 1.41587064e-01 1.00703049e+00
8.74157622e-02 -3.51822913e-01 -7.53660500e-01 -2.43308514e-01
1.49449334e-01 1.02703893e+00 -4.33045119e-01 4.37457301e-02
-3.58495206e-01 1.08726394e+00 3.56810391e-01 7.53394604e-01
-5.79922080e-01 1.46697745e-01 9.83374834e-01 1.70831501e-01
6.65221334e-01 -6.42922759e-01 1.08954191e-01 -1.39658511e+00
3.85603637e-01 -1.06282258e+00 1.00655057e-01 -1.10328758e+00
-7.89323986e-01 1.83970794e-01 -2.72006124e-01 -1.23997116e+00
-4.60799366e-01 -8.55208397e-01 -9.13427398e-02 6.80151820e-01
-1.16346943e+00 -1.12809730e+00 -3.12861025e-01 5.53176999e-01
6.86269701e-01 6.68941289e-02 3.07831436e-01 3.76527458e-02
-2.57073849e-01 -1.29885942e-01 -1.09118141e-01 5.39137535e-02
8.45346332e-01 -7.39504576e-01 3.18997294e-01 7.95115948e-01
1.32182054e-02 6.52974844e-01 8.38538826e-01 -6.40440404e-01
-2.30666471e+00 -7.23346889e-01 1.29655644e-01 -7.48062968e-01
6.30706429e-01 -4.71077979e-01 -4.54298973e-01 8.76611412e-01
-1.85273319e-01 -3.72662172e-02 -7.61373788e-02 -3.25705379e-01
-5.59721768e-01 2.90802214e-02 -9.14331913e-01 8.85102451e-01
1.02471530e+00 -6.60756946e-01 -5.10455668e-01 2.38435477e-01
5.22677362e-01 -4.95848358e-01 -4.27290797e-01 -3.53953540e-02
1.10138607e+00 -6.42127454e-01 1.06466949e+00 -7.78216183e-01
5.99260390e-01 -2.18084291e-01 -3.11437160e-01 -1.21181154e+00
-3.98598045e-01 -7.22536325e-01 -6.34106398e-01 6.76147878e-01
-3.48636866e-01 -1.55649245e-01 6.43523037e-01 9.62359846e-01
1.50184751e-01 -4.42649603e-01 -7.11569965e-01 -5.89947820e-01
-3.66596669e-01 -3.48519295e-01 5.74745983e-02 6.10764563e-01
4.36433172e-03 4.01403934e-01 -5.54110944e-01 1.74582958e-01
6.01083875e-01 -2.10511759e-02 1.10989141e+00 -8.70899856e-01
-5.26973784e-01 -3.33641976e-01 -4.29058641e-01 -1.34197521e+00
3.48059297e-01 -5.98127067e-01 2.72014678e-01 -1.13115370e+00
2.16851324e-01 -1.20221145e-01 -2.08996478e-02 3.29987258e-01
-9.59626306e-03 -2.23088205e-01 5.80270171e-01 4.35816705e-01
-2.25821942e-01 6.66176498e-01 1.60903668e+00 -2.10816652e-01
-1.93419412e-01 -1.99013293e-01 -4.78492826e-01 7.56157041e-01
2.51300097e-01 -4.65774328e-01 -3.86334211e-01 -8.91871810e-01
3.40698548e-02 3.21889073e-01 7.45000899e-01 -6.23871684e-01
2.02902183e-01 -1.60460174e-01 4.97459233e-01 -1.22702517e-01
8.83774638e-01 -1.17040014e+00 1.32738904e-03 5.58369935e-01
-3.70433658e-01 6.01554699e-02 -2.37190374e-03 1.00901353e+00
2.75928140e-01 -4.95140329e-02 5.37561476e-01 -2.87005574e-01
-6.39205217e-01 2.62404978e-01 -9.37942564e-02 -8.83423463e-02
7.96721756e-01 -4.12118202e-03 -1.13328598e-01 -7.40243375e-01
-6.31500483e-01 2.95038074e-01 6.97938740e-01 6.63113713e-01
4.74777788e-01 -8.17615390e-01 -3.44295263e-01 3.27195883e-01
-2.09576730e-02 -1.08337952e-02 1.32759795e-01 9.61065352e-01
-1.89643219e-01 1.80558041e-01 -5.12549937e-01 -5.60381293e-01
-9.89843369e-01 7.84780025e-01 2.78411925e-01 2.78417498e-01
-9.37683225e-01 6.44966662e-01 8.71633649e-01 -1.44615574e-02
6.27821445e-01 -3.69985908e-01 7.78988423e-03 -3.34738381e-02
3.95131052e-01 5.21100938e-01 -3.31800848e-01 -6.75596416e-01
1.97616573e-02 4.47917312e-01 -1.00987576e-01 -3.30949366e-01
1.52695489e+00 -1.09414935e-01 5.89800552e-02 3.57917309e-01
9.21312630e-01 -9.27258059e-02 -2.30274844e+00 -3.12396824e-01
-4.19924289e-01 -7.98698008e-01 1.64858639e-01 -4.89886254e-01
-1.06194901e+00 7.57215559e-01 2.29648635e-01 -4.88104373e-01
7.56322026e-01 1.31038446e-02 6.78803444e-01 6.34875894e-01
4.95286554e-01 -9.91707385e-01 5.68440080e-01 3.74578536e-01
1.27257848e+00 -1.01503754e+00 1.00173466e-01 -2.79245794e-01
-8.31918836e-01 1.01591718e+00 5.29730380e-01 -3.45408171e-01
3.80365819e-01 1.18849292e-01 -8.23929310e-02 -1.03567973e-01
-7.22236514e-01 5.80781698e-02 1.08612441e-01 6.28829002e-01
-2.32812524e-01 -1.07243136e-01 3.88662815e-01 4.84779537e-01
9.51118171e-02 -1.56707898e-01 4.12563205e-01 1.29661334e+00
-3.31394553e-01 -6.15820706e-01 -3.75588536e-01 1.64050758e-01
-1.82243630e-01 2.55681902e-01 -4.53540236e-01 5.77375829e-01
-1.12065248e-01 6.97911382e-01 3.90821137e-03 -3.76948155e-02
4.50655371e-01 -7.82820061e-02 1.06385684e+00 -4.94902372e-01
-2.21896708e-01 3.99247289e-01 -2.33265921e-01 -9.82373118e-01
-3.71048093e-01 -8.96237135e-01 -8.43047559e-01 9.25861225e-02
-2.12567613e-01 -3.21910679e-01 7.49227464e-01 8.51693869e-01
4.83190775e-01 4.06356096e-01 3.77064258e-01 -1.60254383e+00
-8.04436207e-01 -9.43830788e-01 -7.24666834e-01 4.90471095e-01
7.65379846e-01 -1.08698070e+00 -2.27333069e-01 1.69070393e-01] | [4.721270561218262, 0.7644671201705933] |
6e4ca0d3-6c93-4f46-8588-ec2ae86ed0de | that-slepen-al-the-nyght-with-open-ye-cross-2 | 2209.02967 | null | https://arxiv.org/abs/2209.02967v1 | https://arxiv.org/pdf/2209.02967v1.pdf | That Slepen Al the Nyght with Open Ye! Cross-era Sequence Segmentation with Switch-memory | The evolution of language follows the rule of gradual change. Grammar, vocabulary, and lexical semantic shifts take place over time, resulting in a diachronic linguistic gap. As such, a considerable amount of texts are written in languages of different eras, which creates obstacles for natural language processing tasks, such as word segmentation and machine translation. Although the Chinese language has a long history, previous Chinese natural language processing research has primarily focused on tasks within a specific era. Therefore, we propose a cross-era learning framework for Chinese word segmentation (CWS), CROSSWISE, which uses the Switch-memory (SM) module to incorporate era-specific linguistic knowledge. Experiments on four corpora from different eras show that the performance of each corpus significantly improves. Further analyses also demonstrate that the SM can effectively integrate the knowledge of the eras into the neural network. | ['Jun Wang', 'Qi Su', 'Xuemei Tang'] | 2022-09-07 | that-slepen-al-the-nyght-with-open-ye-cross-1 | https://aclanthology.org/2022.acl-long.540 | https://aclanthology.org/2022.acl-long.540.pdf | acl-2022-5 | ['chinese-word-segmentation'] | ['natural-language-processing'] | [ 3.15812752e-02 -6.08489275e-01 -3.92758489e-01 -4.70700711e-01
-3.47553164e-01 -6.67457998e-01 2.34008551e-01 7.02676401e-02
-6.67480886e-01 4.58913714e-01 1.96763322e-01 -6.81498468e-01
5.71814120e-01 -8.92015815e-01 -4.68663216e-01 -4.29022014e-01
3.49855900e-01 3.45735580e-01 4.31575388e-01 -4.45224077e-01
4.20022190e-01 4.82180119e-02 -6.50693595e-01 2.55211473e-01
1.27939737e+00 4.02794272e-01 7.54302561e-01 1.52439296e-01
-7.92375326e-01 2.89294153e-01 -7.01104999e-01 -3.37955892e-01
2.00320423e-01 -6.16687298e-01 -8.78792465e-01 -7.83124939e-02
-2.62565553e-01 1.74362913e-01 -2.57057548e-01 1.19956195e+00
1.56296596e-01 1.36872575e-01 1.83477402e-01 -3.96992713e-01
-1.07931936e+00 1.31105280e+00 -5.72367311e-01 2.80778319e-01
-4.82586585e-02 3.11293732e-02 9.84671175e-01 -7.89255679e-01
6.61062062e-01 1.22950351e+00 5.04842520e-01 6.27564371e-01
-7.59097755e-01 -8.20312679e-01 7.90710032e-01 -1.67009234e-01
-1.27048635e+00 1.10841235e-02 8.21167588e-01 -3.44806105e-01
1.00169623e+00 -8.71825218e-02 7.25432754e-01 8.55467856e-01
4.37494874e-01 1.29131639e+00 1.00610733e+00 -7.18549967e-01
-3.42713073e-02 -9.51777957e-03 4.91904378e-01 4.39773560e-01
2.85774440e-01 -3.83083016e-01 -4.50598478e-01 6.23280108e-01
5.43353677e-01 -3.79214510e-02 -6.23232685e-02 3.36024374e-01
-1.07943726e+00 9.23751771e-01 8.61565992e-02 1.18774664e+00
6.46697134e-02 -2.94028401e-01 4.40619975e-01 3.78222078e-01
4.61039543e-01 6.46828592e-01 -9.17556703e-01 -1.59495175e-01
-8.20822477e-01 -1.84775099e-01 5.90640664e-01 1.05535293e+00
7.88073063e-01 1.79346532e-01 1.32082462e-01 8.20115983e-01
4.12884057e-01 6.16531610e-01 1.04734921e+00 -3.00283134e-01
6.95701957e-01 8.86635482e-01 -3.41534823e-01 -9.95865583e-01
-3.37246090e-01 -2.73350477e-01 -5.66670239e-01 -6.68259740e-01
3.36745769e-01 -3.62837315e-01 -1.12221837e+00 1.96253812e+00
-1.69595443e-02 -1.28169268e-01 1.39327124e-01 4.25984949e-01
4.54678893e-01 1.01423621e+00 4.33416724e-01 -3.73187631e-01
1.20481110e+00 -1.02690220e+00 -8.57693553e-01 -8.78679633e-01
7.33554542e-01 -8.09140921e-01 1.37844563e+00 2.25835055e-01
-1.10274971e+00 -6.36064291e-01 -1.11371922e+00 -1.29079551e-01
-7.79584408e-01 -9.48569924e-02 8.02104592e-01 8.30848038e-01
-7.00699806e-01 2.85552502e-01 -1.00878549e+00 -4.68952417e-01
1.68963432e-01 1.00412600e-01 1.73590288e-01 -6.15057070e-03
-1.64588904e+00 8.36417854e-01 9.58848119e-01 3.48077267e-01
-2.49871641e-01 -3.93162370e-01 -8.66532147e-01 -1.53827369e-01
4.19035465e-01 -2.14957312e-01 1.30444193e+00 -1.35504711e+00
-1.59190798e+00 8.08397174e-01 -4.54887420e-01 -2.24501237e-01
1.23192616e-01 -4.83890504e-01 -7.60078728e-01 -3.41610491e-01
1.51034102e-01 5.08979797e-01 4.75457609e-01 -1.04940569e+00
-9.44784641e-01 -4.84871805e-01 -2.41259530e-01 2.88838595e-01
-3.49983692e-01 4.74886239e-01 -8.74313772e-01 -8.11912239e-01
5.89053817e-02 -7.91863561e-01 -3.85161549e-01 -8.66511166e-01
-9.53551102e-03 -3.52472991e-01 5.63357174e-01 -1.08455205e+00
2.11281848e+00 -2.22958970e+00 2.22710848e-01 -8.40632319e-02
-2.19595715e-01 2.97328770e-01 -1.16679341e-01 1.87484294e-01
2.82092839e-01 4.86146420e-01 -5.54675102e-01 -5.00750467e-02
-3.43806535e-01 5.14928460e-01 -3.71409655e-01 -7.55234435e-02
2.04311505e-01 1.12615418e+00 -9.99086916e-01 -4.40759033e-01
-2.51672655e-01 -6.12987690e-02 -2.46090293e-01 -4.18831455e-03
-4.87814784e-01 4.94373083e-01 -6.47748590e-01 7.61622787e-01
4.25551742e-01 -3.65895331e-02 4.64077622e-01 3.15065712e-01
-6.19024396e-01 4.27203655e-01 -5.12640536e-01 2.00596690e+00
-6.12520278e-01 4.79352862e-01 -2.58534193e-01 -8.85042071e-01
7.13918865e-01 1.31482288e-01 6.54195100e-02 -1.07536292e+00
5.34947932e-01 4.17473584e-01 3.97824228e-01 -4.89906013e-01
7.49631763e-01 -9.95277017e-02 -6.05483472e-01 3.12458336e-01
-2.06526786e-01 -3.32411230e-02 4.82943505e-01 6.60926998e-02
5.53324819e-01 2.09811348e-02 3.33675593e-01 -3.55117500e-01
4.32993799e-01 4.41234082e-01 9.62194741e-01 4.26547974e-01
-2.71098882e-01 3.62663120e-01 1.62325904e-01 -1.56519294e-01
-7.50703216e-01 -9.40663218e-01 -1.11786969e-01 1.32407522e+00
4.97931033e-01 -1.94098860e-01 -9.81785774e-01 -5.88827789e-01
-4.83198792e-01 1.05636418e+00 -3.33122581e-01 -1.23083733e-01
-1.40424037e+00 -9.60221112e-01 4.79849786e-01 6.99502707e-01
7.36000597e-01 -1.37819958e+00 -4.59065616e-01 4.83368009e-01
-1.51425511e-01 -1.19872355e+00 -7.58038640e-01 -6.38576150e-02
-9.96869028e-01 -6.79424286e-01 -5.44689953e-01 -1.35373318e+00
4.80417192e-01 3.34201515e-01 9.44664121e-01 -1.79035347e-02
1.89325605e-02 4.15993817e-02 -4.38855082e-01 -4.52236891e-01
-5.95258474e-01 7.23986328e-01 -8.69895667e-02 -2.64844000e-01
8.94171894e-01 -1.42030716e-01 -2.83537924e-01 6.54602572e-02
-9.69054937e-01 2.35556722e-01 6.55249536e-01 5.54654181e-01
3.18752408e-01 4.43571992e-02 6.54309571e-01 -1.10000718e+00
6.09948754e-01 -6.30783379e-01 -8.14047813e-01 5.07047892e-01
-6.71262622e-01 -7.22625181e-02 6.73583686e-01 -5.31716824e-01
-1.53718376e+00 -2.87982911e-01 -2.66482234e-02 5.01049161e-01
-6.46384656e-02 1.03691733e+00 -5.62511623e-01 5.08384168e-01
1.93868205e-01 3.25746119e-01 -3.52112710e-01 -5.76818109e-01
3.28012675e-01 8.43738019e-01 4.50777382e-01 -5.83223581e-01
6.44720972e-01 7.83177242e-02 -8.86565804e-01 -1.01093531e+00
-1.00017846e+00 -3.28814775e-01 -9.44792986e-01 -1.74631737e-02
1.25449812e+00 -9.72584307e-01 1.90706030e-02 8.93434584e-01
-1.34652114e+00 -5.64328015e-01 1.72549307e-01 4.83370125e-01
2.60016501e-01 4.07691419e-01 -8.60299528e-01 -4.10477400e-01
-2.63116091e-01 -1.00919354e+00 5.82582176e-01 8.22010875e-01
-2.41072863e-01 -1.26387668e+00 1.21496566e-01 2.12000683e-01
4.44697022e-01 -3.24419409e-01 1.23982775e+00 -6.59662187e-01
-6.40970767e-01 1.09081976e-01 3.04027963e-02 2.96840429e-01
3.58389467e-01 -1.07026711e-01 -4.19513166e-01 -1.41441122e-01
9.11226198e-02 1.98777884e-01 8.47805679e-01 2.14312077e-01
9.26852167e-01 9.45321247e-02 -4.10386771e-01 4.18309420e-01
1.34009361e+00 1.00861835e+00 5.84404230e-01 4.79657203e-01
7.21427798e-01 5.22479117e-01 6.39243066e-01 -2.02561796e-01
6.89405084e-01 5.28056920e-03 -3.44626367e-01 -5.41312434e-02
3.79061624e-02 -1.84429035e-01 6.09353364e-01 1.95100319e+00
2.89091736e-01 -3.17543387e-01 -1.34706068e+00 8.01167309e-01
-1.76644266e+00 -3.05019110e-01 7.32857957e-02 1.91340828e+00
1.37730360e+00 2.61955738e-01 -4.50612158e-01 -4.45384562e-01
8.36409390e-01 3.08013439e-01 -6.33577824e-01 -5.44757903e-01
-4.00191367e-01 2.85280049e-01 4.25584376e-01 3.98939103e-01
-1.01245403e+00 1.98346162e+00 6.58793592e+00 7.25624502e-01
-1.57443225e+00 1.77788317e-01 5.99583685e-01 4.52857077e-01
-5.09211183e-01 1.47442833e-01 -1.08101344e+00 5.98665595e-01
8.75176489e-01 -4.06752914e-01 4.30685878e-01 6.06507242e-01
8.90747458e-02 -1.14252508e-01 -5.75603724e-01 7.45036244e-01
1.58700332e-01 -1.09305847e+00 3.18948328e-02 -3.14457625e-01
1.04435468e+00 4.03567106e-01 -9.75537822e-02 5.56362510e-01
4.21016097e-01 -7.28018939e-01 5.67666352e-01 1.36615694e-01
6.31791890e-01 -6.80812001e-01 4.44246948e-01 3.37231576e-01
-1.28338575e+00 1.18801139e-01 -2.80166328e-01 -1.10003829e-01
3.68752152e-01 2.50911295e-01 -3.19377512e-01 4.47731048e-01
5.81119657e-01 6.17508054e-01 -7.28364050e-01 5.94937444e-01
-5.43668628e-01 1.08524227e+00 7.24996850e-02 -2.65240461e-01
4.42800194e-01 -6.14840269e-01 3.52051228e-01 1.49623132e+00
1.99043185e-01 1.50969364e-02 2.81041831e-01 5.35187006e-01
-2.23154187e-01 4.50009823e-01 -2.91194797e-01 -7.13575065e-01
5.55422723e-01 8.16904485e-01 -1.24338055e+00 -3.85548145e-01
-8.30246747e-01 1.05336416e+00 2.44471177e-01 5.18787384e-01
-7.72220969e-01 -4.82731074e-01 5.62943995e-01 -2.54879564e-01
3.28814030e-01 -8.37314308e-01 -7.42608905e-01 -1.30082536e+00
-1.58861428e-01 -8.62555981e-01 3.25368941e-01 -8.90394449e-02
-1.16943383e+00 6.04674876e-01 -1.30376413e-01 -6.15294039e-01
1.47180602e-01 -7.14884341e-01 -6.60165429e-01 7.59866357e-01
-1.55434108e+00 -9.31938410e-01 2.88969874e-01 1.12862669e-01
1.13076532e+00 -3.14975947e-01 5.55633783e-01 4.31809545e-01
-1.03914440e+00 5.56951046e-01 2.02635705e-01 5.84370434e-01
6.03482306e-01 -1.00492775e+00 8.22018325e-01 1.35579169e+00
1.49085850e-01 1.05274749e+00 2.36833125e-01 -1.04132783e+00
-1.33985472e+00 -1.06409562e+00 1.33145177e+00 -1.59459963e-01
9.62562501e-01 -5.97468197e-01 -1.28343308e+00 9.55273390e-01
5.57618260e-01 -7.54485726e-01 7.85931528e-01 1.97981462e-01
-1.42414272e-01 -7.10366433e-03 -3.52130026e-01 1.05075812e+00
1.04958510e+00 -4.60039079e-01 -1.01690114e+00 -2.74547972e-02
1.22929096e+00 -3.54227722e-01 -3.86753559e-01 2.48997509e-01
4.24632698e-01 -2.45800033e-01 3.74147803e-01 -7.13730335e-01
1.41220272e-01 -3.46051037e-01 -4.97504286e-02 -1.15222764e+00
-2.58706033e-01 -7.15095758e-01 3.45956475e-01 1.37548900e+00
7.27877378e-01 -7.15815604e-01 2.70578921e-01 4.73294467e-01
-3.77350062e-01 -5.52650154e-01 -7.00157046e-01 -7.37119317e-01
6.95042849e-01 -6.27574921e-01 6.96574986e-01 1.18311381e+00
-1.59453519e-03 7.58692741e-01 5.87210059e-03 4.90556024e-02
-2.05470219e-01 4.26164299e-01 2.93646872e-01 -1.05493426e+00
-1.20991170e-01 -7.52306640e-01 1.72451615e-01 -1.30307353e+00
3.04363877e-01 -9.59302187e-01 3.90010953e-01 -1.41341817e+00
6.96330667e-02 -3.84469450e-01 -3.99007380e-01 3.70161474e-01
-5.29127002e-01 -2.71111995e-01 2.18661055e-01 2.47742310e-02
-5.53876936e-01 5.08709073e-01 1.22575569e+00 6.34298846e-02
-6.37082994e-01 -2.50502706e-01 -9.65476513e-01 9.49093580e-01
9.72250462e-01 -3.35364431e-01 -8.59393328e-02 -1.17032647e+00
4.57309246e-01 -4.18782324e-01 -9.33397949e-01 -7.08667397e-01
3.61057907e-01 -5.16461194e-01 1.51507244e-01 -4.93998796e-01
-3.42479676e-01 -4.56941456e-01 -2.79222161e-01 4.96931225e-01
3.60417292e-02 4.15017188e-01 4.43840116e-01 1.20135568e-01
-3.27645302e-01 -2.02189013e-01 6.86846256e-01 -3.86796117e-01
-1.09556246e+00 2.46917948e-01 -7.32971132e-01 5.15408397e-01
8.60305548e-01 -3.00099462e-01 -6.43426105e-02 3.29640418e-01
-2.83881038e-01 5.46366274e-01 1.55095190e-01 8.52803290e-01
2.55136698e-01 -8.46479833e-01 -4.93160844e-01 1.89923942e-01
-2.06215963e-01 3.01787019e-01 1.30184487e-01 5.34763873e-01
-5.00509739e-01 4.57694024e-01 2.62238085e-02 -3.08965415e-01
-9.20302927e-01 4.63276416e-01 1.63312808e-01 -3.95074129e-01
-4.06401575e-01 7.84712315e-01 4.03137475e-01 -5.64991355e-01
8.62175822e-02 -5.05849302e-01 -4.21951324e-01 3.47999364e-01
4.61127877e-01 9.49830487e-02 -2.95816898e-01 -6.41983628e-01
-2.36421824e-01 7.21173882e-01 -4.34905827e-01 -5.43482788e-02
1.18577325e+00 -4.34262276e-01 -3.81565511e-01 1.01798928e+00
8.54258657e-01 3.15106958e-01 -8.53539288e-01 -4.72533107e-01
6.40770316e-01 -1.95848674e-01 -2.55118340e-01 -8.15119386e-01
-9.56256866e-01 9.17375267e-01 2.48393312e-01 -2.13882163e-01
1.08043396e+00 1.55732119e-02 1.46112943e+00 3.38739842e-01
3.73660803e-01 -1.78054583e+00 -7.32843801e-02 1.27805257e+00
4.52628791e-01 -1.13551056e+00 -3.67118120e-01 -2.05101579e-01
-7.18253076e-01 9.77572381e-01 8.12187433e-01 5.64534403e-02
7.36231565e-01 1.86885238e-01 4.20729011e-01 2.30151758e-01
-3.13027561e-01 -2.33593956e-01 4.19268683e-02 1.21218905e-01
6.21641397e-01 1.40646294e-01 -9.82186198e-01 1.06588745e+00
-2.73165137e-01 -3.52654606e-01 2.30511039e-01 1.14988756e+00
-6.34924471e-01 -1.56558919e+00 -8.92720968e-02 2.62918230e-02
-4.99498963e-01 -4.21625048e-01 -4.51060921e-01 9.26284790e-01
3.90377164e-01 6.86345816e-01 3.00924093e-01 -1.95900708e-01
4.16084751e-02 4.14800316e-01 2.36681968e-01 -8.25761557e-01
-7.17903376e-01 3.61107618e-01 -2.72545606e-01 -1.19617902e-01
-2.92341739e-01 -9.04037058e-01 -1.96252692e+00 -4.54271585e-02
-3.99695903e-01 9.98039171e-02 5.79435408e-01 1.27609503e+00
1.71153322e-01 5.22492886e-01 4.32684630e-01 -4.59394092e-03
1.17367152e-02 -8.81073236e-01 -3.98220181e-01 3.24578658e-02
-1.81700915e-01 -1.50745437e-01 9.62600857e-03 9.17874090e-03] | [9.951775550842285, 10.155074119567871] |
413d8185-b95a-4b31-922b-e1eaf8aebbe4 | continuous-pseudo-label-rectified-domain | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Gong_Continuous_Pseudo-Label_Rectified_Domain_Adaptive_Semantic_Segmentation_With_Implicit_Neural_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Gong_Continuous_Pseudo-Label_Rectified_Domain_Adaptive_Semantic_Segmentation_With_Implicit_Neural_CVPR_2023_paper.pdf | Continuous Pseudo-Label Rectified Domain Adaptive Semantic Segmentation With Implicit Neural Representations | Unsupervised domain adaptation (UDA) for semantic segmentation aims at improving the model performance on the unlabeled target domain by leveraging a labeled source domain. Existing approaches have achieved impressive progress by utilizing pseudo-labels on the unlabeled target-domain images. Yet the low-quality pseudo-labels, arising from the domain discrepancy, inevitably hinder the adaptation. This calls for effective and accurate approaches to estimating the reliability of the pseudo-labels, in order to rectify them. In this paper, we propose to estimate the rectification values of the predicted pseudo-labels with implicit neural representations. We view the rectification value as a signal defined over the continuous spatial domain. Taking an image coordinate and the nearby deep features as inputs, the rectification value at a given coordinate is predicted as an output. This allows us to achieve high-resolution and detailed rectification values estimation, important for accurate pseudo-label generation at mask boundaries in particular. The rectified pseudo-labels are then leveraged in our rectification-aware mixture model (RMM) to be learned end-to-end and help the adaptation. We demonstrate the effectiveness of our approach on different UDA benchmarks, including synthetic-to-real and day-to-night. Our approach achieves superior results compared to state-of-the-art. The implementation is available at https://github.com/ETHRuiGong/IR2F. | ['Luc van Gool', 'Dengxin Dai', 'Martin Danelljan', 'Qin Wang', 'Rui Gong'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['unsupervised-domain-adaptation', 'pseudo-label'] | ['methodology', 'miscellaneous'] | [ 3.84759098e-01 2.25756764e-01 -2.24258214e-01 -8.63226473e-01
-1.25725043e+00 -6.00954533e-01 4.43771154e-01 -3.83535236e-01
-2.28071839e-01 6.79607987e-01 -2.01376155e-02 4.72940244e-02
1.68285534e-01 -4.97572631e-01 -8.64074469e-01 -8.64246190e-01
5.94741344e-01 6.06601179e-01 9.37613100e-02 7.22236335e-02
7.19402432e-02 1.52928948e-01 -1.17672086e+00 1.48638383e-01
1.16701341e+00 1.07165039e+00 3.91840428e-01 4.46511626e-01
-1.51435703e-01 3.47378641e-01 -5.13049424e-01 -2.08049849e-01
3.85751009e-01 -4.96285588e-01 -7.26891637e-01 4.87894952e-01
3.74299377e-01 -2.87176251e-01 -1.58869371e-01 1.19312906e+00
2.93475866e-01 2.36207992e-01 9.46785867e-01 -1.00108635e+00
-8.40598166e-01 2.95703888e-01 -9.38371062e-01 -1.19092003e-01
-2.41136670e-01 1.43907398e-01 8.16164315e-01 -9.30475414e-01
6.98884964e-01 9.74379599e-01 5.63485086e-01 7.68521070e-01
-1.32650852e+00 -7.86115766e-01 4.20573413e-01 1.05332546e-01
-1.48191679e+00 -4.36991274e-01 8.30008030e-01 -3.99979919e-01
3.93332541e-01 -1.11673385e-01 2.12571099e-01 1.04112554e+00
-2.37715021e-01 7.55506456e-01 1.05761313e+00 -3.97807419e-01
3.32971990e-01 1.54648304e-01 -5.01201376e-02 4.38618422e-01
-1.12339057e-01 -5.62472455e-02 -2.21347868e-01 2.24720627e-01
9.37737644e-01 2.48237960e-02 -3.00717771e-01 -3.99877369e-01
-1.09258008e+00 6.40370786e-01 6.72910988e-01 2.95510609e-02
-4.55434024e-01 -9.08131748e-02 -7.65754469e-03 -1.43279493e-01
8.19319427e-01 2.64339447e-01 -7.59932756e-01 1.68623969e-01
-9.62537587e-01 -8.26686807e-03 2.11275801e-01 1.05854177e+00
9.25628960e-01 -1.50969684e-01 -2.62241989e-01 1.40621424e+00
5.32230139e-01 4.37801600e-01 4.66696173e-01 -1.33067143e+00
4.55982715e-01 4.90315408e-01 2.12865084e-01 -4.10767019e-01
-3.22067410e-01 -5.25316000e-01 -7.33248591e-01 2.12215081e-01
7.38586664e-01 -1.50282249e-01 -1.57510340e+00 1.79505885e+00
5.87450922e-01 5.62002599e-01 1.91444233e-01 1.26885223e+00
5.55583239e-01 7.78097570e-01 2.23332420e-01 1.09058455e-01
1.25234711e+00 -1.20547509e+00 -5.89079440e-01 -5.32046199e-01
5.57310641e-01 -6.72825217e-01 1.13154602e+00 2.33176097e-01
-7.09302425e-01 -6.99091792e-01 -1.02487624e+00 -1.46339774e-01
-1.82224467e-01 5.60796320e-01 1.27850205e-01 3.04847538e-01
-8.51612449e-01 5.74966371e-01 -7.86019862e-01 -8.30065981e-02
7.25108981e-01 2.06574842e-01 -2.84872353e-01 -3.51660192e-01
-1.06103396e+00 5.48777163e-01 2.91846722e-01 8.21480006e-02
-8.28312159e-01 -7.35756993e-01 -9.24444437e-01 -2.30108410e-01
3.34862709e-01 -4.84764218e-01 1.51554573e+00 -1.16174722e+00
-1.52331555e+00 1.19068766e+00 -1.52996823e-01 -4.40106183e-01
4.49519604e-01 -5.01745343e-02 -2.80659854e-01 1.23081498e-01
4.63071167e-01 1.18902409e+00 9.54331338e-01 -1.50076497e+00
-7.19534576e-01 -4.40401137e-01 -2.66528398e-01 4.72348511e-01
1.46721810e-01 -4.65753704e-01 -6.47876978e-01 -7.74384141e-01
4.95815039e-01 -1.11458766e+00 -2.93192804e-01 3.51866148e-02
-6.04298055e-01 -7.03160614e-02 6.49678230e-01 -7.90581226e-01
6.01209342e-01 -2.34829378e+00 -5.94858825e-02 -5.14703840e-02
7.95857385e-02 3.28152448e-01 -2.98552722e-01 -2.59516865e-01
-1.60831705e-01 -1.44608155e-01 -8.79313529e-01 -5.71724951e-01
-9.78677496e-02 1.80497602e-01 -3.86618733e-01 6.49007976e-01
4.86935169e-01 9.28182065e-01 -9.04622495e-01 -3.82889569e-01
3.46450955e-01 6.63751781e-01 -2.99361944e-01 3.96070689e-01
-4.89251107e-01 1.02949739e+00 -5.23114264e-01 5.07726133e-01
9.99209106e-01 -2.25545019e-01 -5.37531227e-02 -1.34614438e-01
1.68779865e-01 1.86972708e-01 -1.03659213e+00 1.96513069e+00
-5.47091424e-01 5.39480507e-01 6.83623627e-02 -9.85802650e-01
1.13001025e+00 1.41937435e-01 4.22322929e-01 -7.85652518e-01
1.68359444e-01 3.53632063e-01 -3.17731589e-01 -2.81831384e-01
3.40601116e-01 -1.01314597e-01 -3.33860256e-02 2.11531535e-01
1.40349969e-01 -2.99651474e-01 -6.28919452e-02 5.10400832e-02
6.26759470e-01 5.95865369e-01 7.71453679e-02 -4.41765450e-02
4.48059946e-01 1.68000370e-01 8.13845456e-01 2.65546054e-01
-2.39723384e-01 1.31965852e+00 1.94775864e-01 -1.29721984e-01
-1.07207620e+00 -1.18286109e+00 -3.01966965e-01 8.86548102e-01
3.40451270e-01 1.95465758e-01 -9.98883188e-01 -9.16186631e-01
-1.71541587e-01 9.27030444e-01 -6.00092053e-01 -1.58634856e-01
-3.89484853e-01 -6.57726407e-01 2.75533438e-01 5.64962983e-01
6.94978237e-01 -9.86191154e-01 -1.78854853e-01 2.16302633e-01
-4.48912024e-01 -1.30018604e+00 -7.98630893e-01 2.76766658e-01
-8.30542922e-01 -7.30963647e-01 -1.18159318e+00 -8.12684357e-01
9.37541246e-01 3.41931671e-01 8.58457446e-01 -4.03897703e-01
-9.68386382e-02 1.00879095e-01 -2.13825911e-01 -1.82868868e-01
-4.37306523e-01 -1.56751592e-02 -2.31943265e-01 2.46446595e-01
2.77131826e-01 -4.91700709e-01 -7.73265719e-01 5.78761280e-01
-7.94254422e-01 1.85357302e-01 6.27549827e-01 8.47418010e-01
1.20185137e+00 -1.82883814e-01 8.05258632e-01 -1.23246777e+00
1.29388437e-01 -6.47357166e-01 -7.24595845e-01 1.16006695e-01
-5.13106227e-01 1.18198715e-01 5.57056010e-01 -5.26416004e-01
-1.46451151e+00 4.76752728e-01 -3.44603062e-01 -7.25304902e-01
-6.28027380e-01 4.25636545e-02 -3.48233014e-01 3.29337269e-01
7.70644724e-01 4.19814885e-02 -1.70936480e-01 -5.36416113e-01
7.92753756e-01 8.27226758e-01 8.72409344e-01 -5.54791629e-01
6.37786865e-01 6.15218878e-01 -3.65658790e-01 -5.74858367e-01
-1.23445261e+00 -7.28159010e-01 -7.40345359e-01 -7.25908652e-02
9.01496172e-01 -1.22420406e+00 1.88843012e-02 6.32992864e-01
-1.05724072e+00 -6.77791238e-01 -4.70927685e-01 3.98167551e-01
-7.10824549e-01 1.08140729e-01 -4.55592304e-01 -6.58153653e-01
-2.08185673e-01 -1.25539422e+00 1.34511042e+00 5.76367199e-01
-4.96080965e-02 -8.85043621e-01 1.82806910e-03 7.18365371e-01
1.25416264e-01 9.57619920e-02 5.61525583e-01 -5.99334419e-01
-5.16741276e-01 -9.08141490e-03 -5.59341252e-01 4.64127690e-01
2.11077228e-01 -1.63283005e-01 -1.38229144e+00 1.26897112e-01
3.61990444e-02 -2.47926667e-01 8.92739117e-01 7.91815758e-01
1.16524899e+00 2.93055009e-02 -3.30804884e-01 7.62823045e-01
1.16056287e+00 3.86213809e-02 5.68273962e-01 1.86570972e-01
8.66811395e-01 6.20895028e-01 1.14056492e+00 1.96254596e-01
5.37066638e-01 7.25550592e-01 4.44180608e-01 -2.89066702e-01
-5.34611523e-01 -3.23278785e-01 5.38302250e-02 4.27641243e-01
3.34916264e-01 -2.25454450e-01 -8.24398994e-01 7.91721880e-01
-1.81303251e+00 -4.38850313e-01 -2.44170666e-01 2.20911551e+00
1.03101552e+00 5.93943633e-02 -2.58286316e-02 -2.09792718e-01
1.03493488e+00 7.61465169e-03 -1.07637286e+00 -4.29628342e-02
-2.59293169e-02 -8.55732616e-03 6.70120835e-01 4.66996670e-01
-1.24383116e+00 1.10438526e+00 5.09614944e+00 9.20430243e-01
-1.19439793e+00 3.05841684e-01 1.17219520e+00 1.34082064e-01
-1.62489191e-01 -1.53849795e-01 -7.64582336e-01 3.93615365e-01
8.38785827e-01 1.75236925e-01 4.02993351e-01 1.07838356e+00
1.11688383e-01 -1.57638982e-01 -8.20546031e-01 8.72778475e-01
-1.43461064e-01 -9.88621712e-01 -3.72645348e-01 2.24067979e-02
1.13542056e+00 2.11938992e-01 2.66536802e-01 3.30575436e-01
3.64015579e-01 -7.57770598e-01 7.54567325e-01 2.21602961e-01
1.25508368e+00 -5.85008323e-01 6.15007460e-01 2.98601568e-01
-9.40017223e-01 1.14672527e-01 -4.37472343e-01 3.32414001e-01
2.38101438e-01 7.79094636e-01 -1.10399127e+00 3.60211253e-01
5.46837807e-01 8.54599059e-01 -3.18789214e-01 8.04354489e-01
-5.80226004e-01 6.85533702e-01 -2.86218345e-01 6.08455122e-01
2.30883881e-01 -5.08327127e-01 2.54797548e-01 1.02322376e+00
3.01840752e-01 1.18918829e-01 -7.63747618e-02 1.11512232e+00
-3.71292710e-01 -4.70508225e-02 -3.19408417e-01 9.23094898e-02
4.94561762e-01 1.39814007e+00 -9.21369255e-01 -3.34092140e-01
-2.46442795e-01 1.23453903e+00 2.99688905e-01 7.37385035e-01
-1.00120485e+00 -1.72404483e-01 7.91944146e-01 9.62451100e-02
3.34058076e-01 -2.71636751e-02 -6.28325164e-01 -1.03481579e+00
1.00799963e-01 -6.12515986e-01 3.51203203e-01 -9.01220858e-01
-1.39117217e+00 6.16750896e-01 -3.75687405e-02 -1.31998861e+00
-2.53573060e-01 -3.32462430e-01 -5.10802388e-01 1.10131466e+00
-1.71814609e+00 -1.17587650e+00 -4.37885880e-01 2.98860997e-01
8.84552717e-01 1.05716623e-01 6.00379705e-01 3.46085399e-01
-5.54780185e-01 4.89063919e-01 1.16665140e-01 1.38046309e-01
1.02867877e+00 -1.22469938e+00 6.20316148e-01 8.06801856e-01
2.69643009e-01 1.46866711e-02 5.79794824e-01 -5.73137760e-01
-4.85879719e-01 -1.47738194e+00 4.04720843e-01 -4.67472941e-01
3.23063254e-01 -2.48697177e-01 -1.17390132e+00 6.55582368e-01
-1.07557133e-01 3.84999484e-01 2.81134784e-01 -2.66923696e-01
-3.50459993e-01 -7.08883181e-02 -1.36527658e+00 5.01098394e-01
9.99903440e-01 -4.07617420e-01 -3.30129445e-01 2.90152103e-01
8.79795671e-01 -6.34788752e-01 -5.49826086e-01 3.88341010e-01
1.93392172e-01 -7.89538562e-01 9.74862576e-01 -1.62275627e-01
4.50289100e-01 -5.50190806e-01 -6.14288030e-03 -1.53045654e+00
-1.59711555e-01 -1.47783160e-01 2.63504148e-01 1.40945613e+00
6.14504158e-01 -6.86320424e-01 9.47613418e-01 7.60386765e-01
-2.13477969e-01 -7.09828377e-01 -9.34879541e-01 -5.50864637e-01
1.16339624e-01 -4.20539826e-01 5.61175942e-01 1.02656448e+00
-5.75016558e-01 3.18171799e-01 -1.20622084e-01 4.40709531e-01
7.12012947e-01 1.30125314e-01 5.29638767e-01 -9.85254586e-01
-1.92399636e-01 -2.18227997e-01 -1.24180079e-01 -1.34315777e+00
3.30426723e-01 -8.96063745e-01 4.85368788e-01 -1.61753011e+00
-1.05306268e-01 -8.77005517e-01 -3.01959485e-01 4.88346756e-01
-2.66142040e-01 6.65317953e-01 -6.23658150e-02 3.38839024e-01
-5.25872052e-01 6.68501854e-01 1.40708220e+00 -2.48878211e-01
-4.26728666e-01 8.40301067e-02 -6.99818015e-01 8.04143786e-01
1.14274597e+00 -7.75582433e-01 -4.69695032e-01 -4.68793660e-01
-4.63922590e-01 -1.46797046e-01 2.21688643e-01 -9.26800430e-01
-5.10689802e-02 -1.33763090e-01 4.05234456e-01 -4.23295856e-01
5.30768752e-01 -7.67779827e-01 2.13875547e-02 -1.18322007e-01
-4.57417339e-01 -5.60715795e-01 3.66761759e-02 5.75587869e-01
-1.84736133e-01 -2.74353892e-01 1.11551142e+00 9.70997736e-02
-7.69320607e-01 3.20788205e-01 8.27455223e-02 2.84474909e-01
1.07750785e+00 -1.12847626e-01 -2.28893369e-01 -3.74198735e-01
-7.46123970e-01 2.44339198e-01 7.04562724e-01 4.00285661e-01
4.75652754e-01 -1.13821769e+00 -6.57430589e-01 2.71511525e-01
2.60331422e-01 6.87342465e-01 5.72682559e-01 7.47810423e-01
-3.26033115e-01 8.46960321e-02 -1.01627015e-01 -7.78461814e-01
-9.09957647e-01 4.46546078e-01 5.04664004e-01 -1.29384607e-01
-6.26888156e-01 9.76800025e-01 6.81906760e-01 -5.92094600e-01
1.02718033e-01 -3.35846961e-01 -5.75942583e-02 -1.84290707e-01
4.47113246e-01 -1.22670466e-02 -8.78309831e-02 -8.11308622e-01
-1.60173297e-01 8.85258555e-01 -2.61482783e-02 -2.54500985e-01
1.20563221e+00 -3.80320758e-01 3.69689137e-01 4.20378536e-01
1.23644221e+00 -8.16827267e-02 -2.12402058e+00 -5.16523838e-01
-1.41079471e-01 -3.53264630e-01 3.82596925e-02 -1.01240993e+00
-1.34867203e+00 1.01185679e+00 6.72687471e-01 -2.91198403e-01
1.22548389e+00 2.65850812e-01 8.26251268e-01 -7.03786090e-02
1.35060772e-01 -1.13044548e+00 -6.75530806e-02 1.78313836e-01
7.32239902e-01 -1.56833982e+00 -2.35180900e-01 -7.68358171e-01
-1.19734073e+00 8.07900488e-01 6.41254306e-01 -1.14625908e-01
4.60151851e-01 -2.36377753e-02 6.75574005e-01 1.29842624e-01
-3.32129747e-01 -3.20677340e-01 3.39274615e-01 8.97666156e-01
1.63856834e-01 2.09371969e-01 1.66840658e-01 6.91580534e-01
1.85096756e-01 -2.06462145e-02 3.31154913e-01 4.63520378e-01
-2.55447865e-01 -1.07573962e+00 -4.85905707e-01 3.37760180e-01
-2.49023542e-01 8.86447653e-02 -1.64583251e-01 3.89070034e-01
2.46629372e-01 9.46774125e-01 1.34200677e-01 -1.34486601e-01
3.61340702e-01 1.13352716e-01 2.41683587e-01 -7.23839581e-01
9.02831648e-03 1.97564632e-01 -7.59150907e-02 -4.06841427e-01
-2.33927339e-01 -5.82075655e-01 -1.72934496e+00 2.78593451e-01
-1.33851007e-01 -3.31689529e-02 9.27638590e-01 7.76190579e-01
5.40137529e-01 5.66281438e-01 6.40925527e-01 -1.05092192e+00
-3.98927599e-01 -1.02011883e+00 -6.26501739e-01 6.38145447e-01
3.19074124e-01 -7.58433163e-01 -2.82459348e-01 3.68949801e-01] | [9.688780784606934, 1.2715874910354614] |
0cc80727-2706-423f-849c-4cab8dba3769 | a-rational-distributed-process-level-account | 1801.10186 | null | http://arxiv.org/abs/1801.10186v1 | http://arxiv.org/pdf/1801.10186v1.pdf | A Rational Distributed Process-level Account of Independence Judgment | It is inconceivable how chaotic the world would look to humans, faced with
innumerable decisions a day to be made under uncertainty, had they been lacking
the capacity to distinguish the relevant from the irrelevant---a capacity which
computationally amounts to handling probabilistic independence relations. The
highly parallel and distributed computational machinery of the brain suggests
that a satisfying process-level account of human independence judgment should
also mimic these features. In this work, we present the first rational,
distributed, message-passing, process-level account of independence judgment,
called $\mathcal{D}^\ast$. Interestingly, $\mathcal{D}^\ast$ shows a curious,
but normatively-justified tendency for quick detection of dependencies,
whenever they hold. Furthermore, $\mathcal{D}^\ast$ outperforms all the
previously proposed algorithms in the AI literature in terms of worst-case
running time, and a salient aspect of it is supported by recent work in
neuroscience investigating possible implementations of Bayes nets at the neural
level. $\mathcal{D}^\ast$ nicely exemplifies how the pursuit of cognitive
plausibility can lead to the discovery of state-of-the-art algorithms with
appealing properties, and its simplicity makes $\mathcal{D}^\ast$ potentially a
good candidate for pedagogical purposes. | ['Ioannis N. Psaromiligkos', 'Ardavan S. Nobandegani'] | 2018-01-30 | null | null | null | null | ['detection-of-dependencies'] | ['methodology'] | [ 7.56449476e-02 4.48580474e-01 4.30663377e-01 -4.78250086e-01
-2.31371745e-01 -5.40435493e-01 7.75097191e-01 4.23662007e-01
-5.87197721e-01 6.83593154e-01 -1.87879652e-01 -8.21517766e-01
-1.02333474e+00 -8.20744634e-01 -5.45526266e-01 -6.67574883e-01
-3.11342925e-01 6.63491428e-01 1.72886357e-01 -2.43060097e-01
5.70997775e-01 3.79166305e-01 -1.86524343e+00 -8.95474181e-02
7.46995807e-01 1.11752248e+00 1.13273859e-01 4.52015370e-01
2.03081876e-01 7.41507351e-01 -2.83992171e-01 -1.05391979e+00
1.92714661e-01 -4.94035900e-01 -6.54022753e-01 -5.28744459e-01
1.08276114e-01 -1.77198976e-01 -1.89233750e-01 1.38690054e+00
1.48425892e-01 5.85020445e-02 9.51970220e-01 -1.16422117e+00
-7.06430793e-01 6.35636747e-01 -2.32992068e-01 6.64101839e-01
3.41194987e-01 2.19707042e-01 1.21585524e+00 -5.10024369e-01
2.27133989e-01 1.21261513e+00 3.99279505e-01 1.50273338e-01
-1.17521846e+00 -6.16338730e-01 3.55063498e-01 2.86202192e-01
-1.38424051e+00 -3.22565049e-01 3.47578645e-01 -5.90337873e-01
1.04613113e+00 4.29406255e-01 8.12705457e-01 7.41381526e-01
6.09189689e-01 6.62714422e-01 1.53229630e+00 -5.70097089e-01
6.68737710e-01 -1.81461573e-01 4.45291966e-01 5.56842506e-01
9.13544178e-01 3.26370507e-01 -7.60526001e-01 -6.65419251e-02
9.33166087e-01 -2.34317943e-01 1.33526446e-02 8.35907012e-02
-1.09677362e+00 4.25020128e-01 -3.80702838e-02 3.99455428e-01
-3.37626129e-01 2.34742865e-01 1.61170661e-01 3.60662043e-01
8.01821891e-03 3.79804790e-01 -2.59703100e-01 -3.34663212e-01
-8.39546323e-01 3.02581519e-01 8.72735322e-01 7.15330899e-01
3.39711428e-01 -2.42783818e-02 2.30705440e-01 2.67805427e-01
6.79906607e-01 3.72187287e-01 3.13079447e-01 -1.17017031e+00
3.04472074e-02 2.43758604e-01 1.21366501e-01 -1.12095428e+00
-4.83067155e-01 -3.39925945e-01 -9.27717566e-01 5.48193097e-01
7.90551782e-01 5.00198938e-02 -4.95121062e-01 1.85008621e+00
3.07821333e-02 -2.70474255e-01 -4.15137643e-03 7.57471263e-01
7.36888945e-02 3.12628090e-01 2.72128403e-01 -2.76324749e-01
1.54250002e+00 -1.23044014e-01 -5.54872751e-01 -5.60811520e-01
9.86152887e-02 -5.71187198e-01 7.55677044e-01 9.31037545e-01
-1.38867378e+00 -3.64581019e-01 -1.27067649e+00 1.57370090e-01
-1.56183094e-01 -5.22186995e-01 1.20915222e+00 1.05945790e+00
-1.10153496e+00 6.90532506e-01 -7.52103925e-01 -1.38495788e-01
3.76625031e-01 3.46549779e-01 -4.66100723e-02 1.41975759e-02
-1.12883782e+00 1.40183413e+00 4.80474293e-01 3.92246246e-01
-9.09586608e-01 -5.05587757e-02 -6.34442389e-01 3.04741025e-01
5.72484434e-01 -6.28116131e-01 1.15849996e+00 -8.30644667e-01
-1.29958153e+00 8.80014598e-01 -1.68027014e-01 -3.72400969e-01
3.99130911e-01 -4.65247780e-02 -4.71517742e-01 1.12733737e-01
-8.46744031e-02 2.36266419e-01 7.18665659e-01 -9.00758922e-01
-4.57619637e-01 -6.71387196e-01 1.59381956e-01 1.62602380e-01
8.12331587e-02 1.03563152e-01 9.35523808e-02 -5.07281363e-01
5.48315585e-01 -6.64875388e-01 -7.63081387e-02 -1.39173552e-01
-1.56620890e-01 -4.41986263e-01 -3.96205038e-01 -2.34316707e-01
1.16489577e+00 -1.99918377e+00 -1.32316366e-01 5.42210281e-01
4.97871786e-01 -8.62868801e-02 2.91127414e-01 3.36226374e-01
-2.69413114e-01 2.91538268e-01 -2.20050529e-01 1.17182508e-01
6.47432029e-01 1.19159862e-01 -2.69214898e-01 6.11855984e-01
-1.69251729e-02 6.90675378e-01 -7.29858756e-01 -3.79168868e-01
8.87962282e-02 1.60715893e-01 -2.46999413e-01 -1.98739216e-01
-1.37427181e-01 -2.83066947e-02 -5.02611160e-01 4.49370891e-01
4.37127888e-01 -2.85880804e-01 3.96638036e-01 5.11274695e-01
-2.59183317e-01 5.04619241e-01 -1.46722054e+00 1.22483051e+00
2.28661060e-01 6.10701740e-01 1.24103062e-01 -1.02915418e+00
7.46944368e-01 3.72426569e-01 -6.38596341e-02 -8.01171839e-01
4.94779110e-01 2.58684993e-01 7.13700473e-01 -2.08816469e-01
1.69496924e-01 -4.96338069e-01 -4.14980203e-02 9.51088488e-01
3.85750383e-02 -5.12211800e-01 3.22574615e-01 2.63933569e-01
1.14541304e+00 7.64127299e-02 4.19559300e-01 -7.65352666e-01
4.19753850e-01 -3.73319745e-01 3.94976854e-01 1.20588350e+00
-4.90252286e-01 1.30066678e-01 7.97817707e-01 -4.90673572e-01
-7.55623877e-01 -1.32743442e+00 -3.22658628e-01 1.06934834e+00
2.53318995e-01 -2.79247463e-01 -7.93913901e-01 7.07308436e-03
-2.23421112e-01 1.16943324e+00 -4.14829880e-01 -2.05246851e-01
-4.44322526e-02 -9.80909705e-01 5.50639629e-01 4.32122618e-01
3.34815413e-01 -1.14451694e+00 -9.54420686e-01 1.71379671e-01
2.03700811e-01 -6.62227273e-01 4.20606196e-01 6.73916042e-01
-7.20222890e-01 -7.77413070e-01 -1.82346106e-01 -2.22080156e-01
4.91487920e-01 -4.98182625e-02 1.10309875e+00 3.61372024e-01
-2.39878923e-01 3.39076042e-01 3.91343795e-02 -9.89119351e-01
-1.90823644e-01 -6.80193484e-01 1.98866516e-01 -3.40630412e-01
8.37882638e-01 -9.29341078e-01 -5.68576753e-01 5.36685390e-03
-9.88210678e-01 -6.51878864e-02 6.58934236e-01 3.73231858e-01
4.41654444e-01 4.16440368e-01 5.15171230e-01 -5.87086856e-01
6.88757956e-01 -2.85679519e-01 -6.71629727e-01 1.62980855e-01
-7.59457707e-01 2.78977394e-01 4.43460971e-01 -1.05119750e-01
-1.13135934e+00 -4.47904110e-01 -2.79901121e-02 2.69070268e-01
-3.90854478e-01 5.04274786e-01 -2.18696147e-01 2.82271832e-01
7.11085141e-01 3.48893613e-01 -1.80343941e-01 -9.43367705e-02
3.92421126e-01 3.09229583e-01 5.19942105e-01 -8.51876199e-01
5.23022294e-01 3.79196078e-01 1.14775389e-01 -6.45832837e-01
-8.18758965e-01 6.71568885e-02 -4.86302018e-01 -2.26588443e-01
6.79960549e-01 -7.70292997e-01 -1.40053856e+00 2.33474985e-01
-1.06831002e+00 -1.15575738e-01 -3.02733600e-01 5.87111235e-01
-9.30211365e-01 2.05954045e-01 -5.59250832e-01 -1.33530724e+00
4.66669686e-02 -9.18111801e-01 2.85980642e-01 3.58663321e-01
-7.09080517e-01 -8.48316729e-01 -1.57495499e-01 3.07012707e-01
3.55934858e-01 -2.17543945e-01 1.16059685e+00 -1.11919749e+00
-6.82813704e-01 -4.57620084e-01 -1.22962900e-01 1.60884336e-01
-5.36739707e-01 -1.62922051e-02 -1.02619994e+00 2.13928014e-01
6.29614592e-01 -2.00132027e-01 7.33247757e-01 2.52985001e-01
1.02723694e+00 -2.21943185e-01 -1.28738686e-01 -8.69106352e-02
1.24684751e+00 3.24147791e-01 7.81982958e-01 2.44803697e-01
-7.52372295e-02 6.89285576e-01 4.41280715e-02 5.25494754e-01
5.72956562e-01 -2.81210542e-02 4.67309594e-01 6.39127970e-01
2.78726727e-01 5.19602187e-03 8.29770342e-02 7.82217145e-01
-3.72935981e-01 -2.63927519e-01 -9.28875685e-01 3.79538983e-01
-1.75344217e+00 -1.40851760e+00 -1.27206504e-01 2.42988920e+00
9.40421045e-01 8.63085091e-01 -2.33015224e-01 3.93553913e-01
6.65361583e-01 -7.12349787e-02 -2.67090231e-01 -6.39898300e-01
-4.97834459e-02 5.54411471e-01 9.55662802e-02 4.71418858e-01
-7.20356464e-01 6.37549043e-01 6.51300478e+00 9.13842022e-01
-4.26125318e-01 -1.19686006e-02 6.93554640e-01 -7.29513839e-02
-4.98612791e-01 2.29640827e-01 -6.02823555e-01 5.59063613e-01
1.28268099e+00 -3.94519269e-01 3.39652807e-01 6.69930935e-01
-4.96030673e-02 -9.61742878e-01 -1.50749922e+00 8.01790178e-01
3.35775018e-02 -9.96828258e-01 -3.45396549e-01 3.71653512e-02
2.79165983e-01 -3.05699110e-01 9.23289359e-02 5.42708747e-02
9.76783097e-01 -1.33739078e+00 1.15297472e+00 5.38600862e-01
2.19869167e-01 -6.09292388e-01 4.91452754e-01 7.49513507e-01
-6.74920440e-01 -1.15378939e-01 -4.78449166e-01 -8.30783665e-01
1.35442868e-01 1.00986254e+00 -2.78182596e-01 3.72679323e-01
7.23803878e-01 7.40617290e-02 -4.41646725e-01 9.14711654e-01
-6.93717360e-01 4.46468353e-01 -5.62559962e-01 -5.01573205e-01
9.01011843e-03 -2.70256430e-01 4.38767523e-01 1.15846789e+00
2.30241463e-01 5.24781823e-01 -3.95654351e-01 1.05149460e+00
3.50412875e-01 -2.87209511e-01 -5.39597750e-01 -6.05211332e-02
5.45145452e-01 1.03254509e+00 -1.18375838e+00 -3.29303145e-01
-9.64573845e-02 6.67606831e-01 2.81371593e-01 8.40675682e-02
-6.17945731e-01 -2.42140174e-01 4.43548590e-01 -1.94124579e-01
1.16621919e-01 -4.95346904e-01 -8.97708893e-01 -1.03644133e+00
5.75494878e-02 -7.51009524e-01 3.28216434e-01 -7.20127404e-01
-1.39275289e+00 5.28643727e-01 8.18349048e-02 -5.96085727e-01
-2.68799216e-01 -7.84815848e-01 -3.38257253e-01 9.58177626e-01
-1.08794498e+00 -4.34495986e-01 1.66720375e-01 4.59583521e-01
9.75064933e-02 5.45617901e-02 9.81009781e-01 -3.24488997e-01
-3.54680508e-01 2.95567393e-01 1.07005738e-01 -8.87409970e-02
3.32504928e-01 -1.22713399e+00 2.48648435e-01 8.10345650e-01
3.51455510e-01 9.49909151e-01 1.02897620e+00 -6.10290170e-01
-1.34205735e+00 -1.75948188e-01 1.12616503e+00 -7.93337643e-01
8.15625727e-01 -2.76821405e-01 -7.67499268e-01 8.22788239e-01
2.79658854e-01 -3.88909340e-01 8.44268918e-01 2.30862781e-01
-5.42999685e-01 -4.73499000e-02 -1.12870586e+00 9.16507006e-01
9.97195244e-01 -4.05179173e-01 -1.39979315e+00 2.76355296e-01
1.64081722e-01 1.68489039e-01 -7.68565476e-01 2.65238527e-02
8.35035205e-01 -1.44894326e+00 8.35558891e-01 -4.00309652e-01
3.35056782e-01 -5.19506484e-02 -2.25275576e-01 -8.22065949e-01
-5.38140178e-01 -7.48845398e-01 1.09221414e-01 8.92311215e-01
4.08761948e-01 -9.23939824e-01 4.22207832e-01 1.45725322e+00
-7.67399296e-02 -5.29816449e-01 -1.24669743e+00 -7.38991916e-01
4.42313939e-01 -9.39462185e-01 1.21516958e-01 6.38970554e-01
4.89196658e-01 1.31743118e-01 2.25861609e-01 -1.23462260e-01
7.92726934e-01 4.97050434e-02 6.57426864e-02 -1.62077475e+00
-4.32490647e-01 -8.35013688e-01 -4.35075849e-01 -8.70257199e-01
-4.77987826e-02 -7.48262405e-01 1.69384047e-01 -1.41142857e+00
2.08034873e-01 -5.82155704e-01 -4.94950384e-01 3.75077248e-01
-9.18660387e-02 -5.11931889e-02 2.09896028e-01 2.44206384e-01
-7.72762358e-01 1.35091931e-01 9.19360638e-01 3.68210077e-01
3.00760359e-01 -2.31254339e-01 -1.23799324e+00 1.29856491e+00
6.62189424e-01 -4.54506487e-01 -3.57064337e-01 -4.19556648e-01
1.08268094e+00 1.04136422e-01 4.43479300e-01 -9.73681450e-01
6.14028871e-01 -4.17443812e-01 7.19621658e-01 -1.97365776e-01
3.92349839e-01 -6.75121903e-01 1.97172388e-01 4.76835072e-01
-4.62351978e-01 1.81913331e-01 1.05448835e-01 5.11259437e-01
2.38983795e-01 -5.85792124e-01 5.32830715e-01 -5.52644372e-01
-5.29896617e-01 -2.44669512e-01 -9.20617163e-01 8.64460021e-02
1.22171402e+00 -2.62520432e-01 -4.20238167e-01 -3.03760469e-01
-6.89800560e-01 -3.50884534e-02 1.57823786e-01 -7.57361948e-02
4.49382901e-01 -5.91231763e-01 -5.36081851e-01 1.59136236e-01
-2.86801994e-01 -9.21318829e-02 3.31178308e-01 9.23607469e-01
-4.40964431e-01 6.38632178e-01 -2.60196954e-01 -5.84769994e-02
-4.67333704e-01 5.88101506e-01 1.42091885e-01 -3.43344659e-02
-2.86170870e-01 1.25976539e+00 8.43003839e-02 2.06238046e-01
1.41526699e-01 -2.14894578e-01 9.60916057e-02 1.16415106e-01
7.23938942e-01 4.88340527e-01 4.00238857e-02 -1.92098469e-01
-5.48258007e-01 3.71183455e-02 3.10997069e-02 -6.33340120e-01
1.17344153e+00 -2.21696019e-01 -5.93281031e-01 6.71730995e-01
2.44622633e-01 -2.15430573e-01 -1.06755686e+00 1.08555332e-01
2.71099687e-01 -2.33209029e-01 -2.13233814e-01 -1.22027695e+00
-3.02329302e-01 1.13026536e+00 1.33392364e-01 5.85151553e-01
7.75902689e-01 -1.26722278e-02 -2.19460335e-02 4.45815861e-01
7.35810518e-01 -1.14482462e+00 -1.52965218e-01 6.42240524e-01
5.74500144e-01 -9.19288278e-01 3.24679077e-01 -1.64185792e-01
-5.40308714e-01 1.03400958e+00 3.64033014e-01 -2.61567622e-01
6.95441961e-01 3.29071224e-01 -4.10359383e-01 -3.06421399e-01
-1.05587077e+00 -8.95528942e-02 1.60178710e-02 4.68355149e-01
5.13244450e-01 1.86976403e-01 -6.27595782e-01 1.01615238e+00
-4.58568305e-01 -3.41997296e-02 5.87111413e-01 1.08421779e+00
-7.80705869e-01 -8.03888857e-01 -4.72669780e-01 4.21798438e-01
-5.99908888e-01 -3.86086464e-01 -3.05857062e-01 7.16377795e-01
2.66684204e-01 1.21244884e+00 1.12052105e-01 9.99255627e-02
-1.10346042e-01 4.27470446e-01 7.83513308e-01 -3.87943596e-01
-5.55695534e-01 -1.92112115e-03 -1.16357289e-01 -4.97745335e-01
-1.20673358e-01 -6.71153307e-01 -1.37456298e+00 -5.18913686e-01
1.11332692e-01 1.79521888e-01 6.05354786e-01 1.25891709e+00
1.27860427e-01 2.41517633e-01 -2.02106312e-01 -6.01069272e-01
-8.48613977e-01 -6.89628720e-01 -7.74628997e-01 9.81495231e-02
-1.65709063e-01 -6.79040134e-01 -5.38988531e-01 -8.36803168e-02] | [8.74931526184082, 6.474265098571777] |
2af986f5-55b7-4b9e-a5da-5295c7653e12 | promix-combating-label-noise-via-maximizing | 2207.10276 | null | https://arxiv.org/abs/2207.10276v2 | https://arxiv.org/pdf/2207.10276v2.pdf | ProMix: Combating Label Noise via Maximizing Clean Sample Utility | The ability to train deep neural networks under label noise is appealing, as imperfectly annotated data are relatively cheaper to obtain. State-of-the-art approaches are based on semi-supervised learning(SSL), which selects small loss examples as clean and then applies SSL techniques for boosted performance. However, the selection step mostly provides a medium-sized and decent-enough clean subset, which overlooks a rich set of clean samples. In this work, we propose a novel noisy label learning framework ProMix that attempts to maximize the utility of clean samples for boosted performance. Key to our method, we propose a matched high-confidence selection technique that selects those examples having high confidence and matched prediction with its given labels. Combining with the small-loss selection, our method is able to achieve a precision of 99.27 and a recall of 98.22 in detecting clean samples on the CIFAR-10N dataset. Based on such a large set of clean data, ProMix improves the best baseline method by +2.67% on CIFAR-10N and +1.61% on CIFAR-100N datasets. The code and data are available at https://github.com/Justherozen/ProMix | ['Junbo Zhao', 'Lei Feng', 'Yiwen Dong', 'Ruixuan Xiao', 'Haobo Wang'] | 2022-07-21 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [-1.30496621e-01 6.67437315e-02 -2.86191821e-01 -6.85989022e-01
-1.54033947e+00 -3.77235860e-01 4.46494311e-01 2.10970774e-01
-5.38812399e-01 9.18873966e-01 -1.12723470e-01 -7.02944249e-02
3.72169018e-02 -6.75506234e-01 -8.86405349e-01 -7.31510043e-01
9.21469480e-02 3.65394592e-01 -2.04410050e-02 2.08266884e-01
-2.33599097e-01 8.17171857e-02 -1.48258007e+00 4.40021753e-01
9.74713504e-01 1.26539814e+00 1.57346725e-02 3.52461696e-01
1.02773242e-01 9.12607968e-01 -5.07427037e-01 -5.82760692e-01
4.13729846e-01 -2.78469086e-01 -5.39089799e-01 -4.86999393e-01
7.54063487e-01 -3.01077396e-01 -1.22468866e-01 1.17856693e+00
6.61757827e-01 9.91998687e-02 5.92290640e-01 -1.09023988e+00
-4.95643944e-01 8.72408271e-01 -4.01016831e-01 -9.84977260e-02
-1.99909031e-01 -5.43660447e-02 1.28160286e+00 -1.18455815e+00
3.23547870e-01 9.91413951e-01 8.22736561e-01 6.27636015e-01
-1.11810136e+00 -1.08684349e+00 9.97864679e-02 1.09393455e-01
-1.46777964e+00 -6.94624960e-01 5.91689169e-01 -1.87451392e-01
6.12303376e-01 2.06867442e-01 2.46305004e-01 1.27908909e+00
-1.58121273e-01 1.09186399e+00 1.19518352e+00 -5.63916743e-01
4.58187491e-01 3.04139018e-01 6.58620834e-01 5.28896332e-01
5.00458062e-01 3.07281494e-01 -5.62861145e-01 -1.27499536e-01
5.94041422e-02 1.85872272e-01 -2.65198588e-01 -1.07031778e-01
-8.24821591e-01 8.76490951e-01 6.36303604e-01 4.64141332e-02
-3.11814427e-01 -2.32223477e-02 4.19466794e-01 4.57494967e-02
6.47122025e-01 5.46929240e-01 -4.11895186e-01 1.01142630e-01
-1.13628662e+00 3.30920875e-01 7.83777297e-01 9.12483275e-01
7.12144554e-01 -1.41658820e-02 -4.92513031e-01 1.01581061e+00
9.25767571e-02 7.85373807e-01 2.44792357e-01 -9.61614311e-01
4.97915387e-01 2.26613164e-01 1.76310956e-01 -6.18318915e-01
-1.96869135e-01 -1.02765906e+00 -1.01281047e+00 3.05757344e-01
2.85356581e-01 -1.81502700e-01 -1.18176198e+00 1.81466162e+00
1.98613375e-01 1.69237345e-01 -3.06312479e-02 8.56452465e-01
7.18984783e-01 6.21719778e-01 9.83561426e-02 -1.04697466e-01
1.17869568e+00 -1.17265320e+00 -5.67508698e-01 -3.79332185e-01
6.75141633e-01 -5.45920849e-01 1.11590934e+00 7.05307186e-01
-8.18184674e-01 -5.13319433e-01 -1.04437363e+00 1.82910472e-01
-1.78281769e-01 3.93859416e-01 4.16287661e-01 5.90071440e-01
-7.53247738e-01 7.41602898e-01 -8.04542065e-01 1.20900124e-01
9.54013944e-01 1.14142448e-01 -3.56020182e-01 -6.35377884e-01
-1.15966177e+00 5.63528895e-01 4.50836003e-01 -2.33488251e-02
-1.21900666e+00 -8.25986624e-01 -7.16111422e-01 2.79147476e-01
5.22921860e-01 -2.69862175e-01 1.49289954e+00 -7.63081908e-01
-1.03215718e+00 7.04005122e-01 -1.17600240e-01 -9.13386047e-01
7.21996486e-01 -6.57571614e-01 -3.01995963e-01 1.00933962e-01
1.82222143e-01 6.99038267e-01 6.66631281e-01 -1.52055860e+00
-8.37255239e-01 -1.27242848e-01 -3.25515717e-01 -9.74356383e-02
-3.98101538e-01 -2.00474486e-01 -2.94069231e-01 -6.05898380e-01
-6.27173185e-02 -8.75358701e-01 -2.24744067e-01 -2.34847188e-01
-6.27108335e-01 -2.45494530e-01 3.25749010e-01 -5.12150407e-01
1.22856617e+00 -2.20527387e+00 -6.16124511e-01 1.90468401e-01
4.99523938e-01 9.10936713e-01 -2.63159186e-01 2.14670613e-01
-9.08408966e-03 2.92590648e-01 -2.72811174e-01 -8.39865565e-01
1.43510789e-01 -7.83560872e-02 -3.72175127e-01 3.95752788e-01
2.30289847e-01 7.55621374e-01 -1.09318197e+00 -2.43156865e-01
1.29023725e-02 5.80516279e-01 -2.61004508e-01 3.52446973e-01
-4.76826988e-02 -2.98002474e-02 -2.91324317e-01 6.44612849e-01
8.07716012e-01 -2.60093898e-01 -8.12865421e-02 -1.50069237e-01
2.52990574e-01 3.71661156e-01 -1.19406450e+00 1.26194835e+00
-4.23812747e-01 4.31834638e-01 -8.50483701e-02 -7.38858700e-01
1.06349087e+00 2.31369168e-01 1.30551457e-01 -6.57004952e-01
1.37180254e-01 4.75627631e-01 -2.82163769e-01 -2.40342945e-01
1.92386001e-01 -1.13689065e-01 4.94074784e-02 3.62982690e-01
1.29249185e-01 1.91771150e-01 3.61191005e-01 3.00267547e-01
1.25343668e+00 -9.27726179e-02 2.45690614e-01 -6.59804866e-02
1.32515356e-01 -2.78617620e-01 9.14321125e-01 1.21080208e+00
-4.97699738e-01 8.20521533e-01 1.85150966e-01 -4.51742232e-01
-9.32172537e-01 -1.00944793e+00 -2.49744102e-01 9.67276871e-01
-1.05319187e-01 -4.24055010e-01 -8.34825277e-01 -9.80489790e-01
-9.07410756e-02 9.74983931e-01 -6.29353106e-01 -3.68832350e-01
-2.85213470e-01 -8.28187346e-01 5.16412377e-01 3.91467452e-01
5.18368483e-01 -1.09136569e+00 -1.94560841e-01 5.21162897e-02
-1.95185840e-01 -9.59778309e-01 -2.34266520e-01 5.97879410e-01
-4.91313487e-01 -9.97090995e-01 -4.43954319e-01 -8.37550879e-01
6.34166777e-01 3.20833504e-01 1.26757705e+00 6.14363924e-02
-1.65723544e-03 -3.89953077e-01 -6.99708760e-01 -5.98301411e-01
-4.49538261e-01 2.96081573e-01 1.51137486e-01 4.16775718e-02
5.60231805e-01 -4.22257125e-01 -4.66233790e-01 6.45542517e-02
-7.25266397e-01 -6.55542174e-03 6.03884518e-01 1.07505798e+00
8.89651179e-01 4.73204926e-02 8.99547100e-01 -1.31005013e+00
3.11493367e-01 -6.09364212e-01 -4.87225085e-01 2.40093902e-01
-9.55769360e-01 1.38094155e-02 9.88991082e-01 -3.34441870e-01
-1.04309595e+00 1.90975294e-01 -3.31921339e-01 -5.85485399e-01
-4.77032810e-01 2.79719174e-01 -6.78888187e-02 1.98658422e-01
9.98481035e-01 -1.24884374e-01 -4.31151390e-01 -8.61546338e-01
2.30906561e-01 8.81429136e-01 5.65814435e-01 -4.18300748e-01
6.60706758e-01 3.43629718e-01 -3.89211208e-01 -2.69587904e-01
-1.64592814e+00 -5.76457918e-01 -3.21618468e-01 1.07466415e-01
3.90472591e-01 -1.11054134e+00 -2.99279511e-01 5.80261827e-01
-7.09230065e-01 -4.99674946e-01 -3.59617323e-01 5.49902439e-01
-1.36691645e-01 1.02508411e-01 -7.25033522e-01 -1.03176105e+00
-7.42558241e-01 -1.01990080e+00 1.05491614e+00 1.55111670e-01
9.65239387e-03 -4.77164954e-01 4.77364063e-02 3.33274007e-01
3.86482298e-01 1.46262392e-01 4.03066605e-01 -1.19549990e+00
-4.42418873e-01 -6.05990648e-01 -2.75644332e-01 8.35944057e-01
-3.40842716e-02 -2.61787623e-01 -1.47623026e+00 -5.12615383e-01
-2.37553138e-02 -9.28299010e-01 1.32268202e+00 2.98512220e-01
1.33940125e+00 -2.29527429e-01 -2.79953539e-01 6.71418667e-01
1.55331886e+00 3.14769847e-03 4.06591803e-01 2.12380528e-01
7.30563402e-01 3.56124997e-01 8.84061158e-01 3.56216550e-01
1.30536690e-01 3.52008671e-01 4.53962743e-01 -1.52337372e-01
-2.96924174e-01 -3.63611639e-01 1.17815919e-01 7.99750090e-01
3.30776393e-01 -3.49061579e-01 -9.36487675e-01 5.93200684e-01
-1.82388544e+00 -7.56618619e-01 -3.29531580e-02 2.35327268e+00
1.21076047e+00 4.06437665e-01 -1.73362255e-01 3.09005439e-01
8.02929342e-01 -6.84556812e-02 -5.18204153e-01 1.90150112e-01
-1.19916117e-02 3.39225978e-01 5.16499400e-01 5.96418321e-01
-1.46311975e+00 8.16437840e-01 5.27559042e+00 1.46228290e+00
-8.48024964e-01 2.60732859e-01 1.10671890e+00 -3.56194645e-01
-2.69036125e-02 -2.95496613e-01 -1.21358156e+00 6.32260084e-01
1.13071215e+00 1.72827184e-01 3.07872206e-01 1.11692429e+00
1.44792020e-01 -1.26563143e-02 -1.08347404e+00 1.00572121e+00
8.89509767e-02 -1.21305978e+00 -2.09023625e-01 -1.69532850e-01
9.89330828e-01 4.36370075e-01 1.20689794e-02 6.17555022e-01
5.54220200e-01 -1.12259161e+00 8.93686891e-01 4.49528068e-01
8.09263825e-01 -9.05497611e-01 1.10861433e+00 6.03460550e-01
-7.97468364e-01 -1.03445314e-01 -4.75281954e-01 1.40670195e-01
-8.87857005e-03 1.41181517e+00 -8.34091544e-01 3.89125615e-01
9.48578656e-01 5.98803818e-01 -5.64790964e-01 1.35070205e+00
-5.52929401e-01 1.30844462e+00 -4.36932921e-01 2.31756307e-02
9.69936326e-02 9.54136252e-02 2.59910852e-01 1.41831076e+00
1.91707537e-01 -1.35929108e-01 2.66135067e-01 7.04394400e-01
-5.77817917e-01 4.60135154e-02 -3.68231058e-01 3.40189219e-01
9.31394696e-01 1.33841109e+00 -5.32449663e-01 -6.09633684e-01
-1.11733742e-01 5.64037681e-01 7.81720817e-01 3.94278318e-01
-8.56230140e-01 -5.71068764e-01 2.79210895e-01 -1.96239799e-01
3.76222908e-01 3.42177719e-01 -3.93781483e-01 -1.10469067e+00
1.32060513e-01 -1.04207408e+00 3.04128289e-01 -4.72697467e-01
-1.68322766e+00 9.13797617e-01 -3.34589332e-01 -1.21559119e+00
-3.80838802e-03 -3.84202778e-01 -3.69200081e-01 7.69536793e-01
-1.69115424e+00 -1.05923617e+00 -3.91211778e-01 7.04431012e-02
5.80813885e-01 -1.22831769e-01 7.85073340e-01 5.68564475e-01
-7.16854453e-01 1.03375757e+00 5.04542887e-01 2.38000959e-01
9.91087496e-01 -1.33611774e+00 4.34870452e-01 9.83447254e-01
3.83049250e-01 4.18149650e-01 4.79071587e-01 -4.61751908e-01
-5.74424744e-01 -1.59662795e+00 1.11045873e+00 -4.87877101e-01
1.66476324e-01 -6.08828545e-01 -1.11689675e+00 5.20271719e-01
-5.52601218e-02 3.69963050e-01 6.06035650e-01 8.03418234e-02
-6.85379922e-01 -5.27506053e-01 -1.11044157e+00 3.76592606e-01
9.18944359e-01 -3.62200141e-01 -4.05637860e-01 5.28449774e-01
8.73162210e-01 -3.35135311e-01 -6.46907747e-01 4.87983972e-01
3.35390598e-01 -1.02419400e+00 8.30630124e-01 -3.20708126e-01
3.52849782e-01 -1.14567369e-01 -2.56708264e-01 -1.46328032e+00
-3.51014137e-01 -3.69632393e-01 -2.15451851e-01 1.43897963e+00
6.30822182e-01 -5.68616569e-01 7.97900736e-01 4.35739934e-01
-2.03544259e-01 -1.02667999e+00 -5.77836812e-01 -8.83160770e-01
-2.40076843e-04 -5.92478335e-01 5.48701882e-01 8.84287417e-01
-4.27411407e-01 2.86340743e-01 -6.40968084e-01 -9.39687621e-03
9.05385792e-01 5.41422293e-02 5.47716141e-01 -1.27136242e+00
-2.89644867e-01 -8.05090368e-02 2.45534524e-01 -7.52008259e-01
2.27326542e-01 -9.40431833e-01 5.00326633e-01 -1.29150891e+00
3.14472228e-01 -9.42188799e-01 -7.39135861e-01 8.67577791e-01
-5.12555420e-01 5.60690165e-01 2.10555762e-01 2.49945313e-01
-1.04802859e+00 6.02773607e-01 7.00244904e-01 -1.89976431e-02
1.43685909e-02 1.49228781e-01 -8.66602123e-01 6.51065528e-01
1.15786755e+00 -9.19880390e-01 -3.52058470e-01 -3.34699869e-01
7.53527954e-02 -5.31711876e-01 1.82956472e-01 -1.22672260e+00
-7.92647228e-02 1.29150689e-01 4.12857354e-01 -4.90189761e-01
2.11608678e-01 -5.88646889e-01 -6.10787943e-02 4.13304716e-01
-7.86257565e-01 -6.60648942e-01 -7.77587891e-02 5.70885539e-01
-2.40379244e-01 -5.08113205e-01 9.42378342e-01 -1.34088561e-01
-4.53850865e-01 3.93499047e-01 1.03473350e-01 3.98903430e-01
7.66642451e-01 3.98518950e-01 -5.14135718e-01 -2.46388718e-01
-4.91942942e-01 3.33350271e-01 2.34803811e-01 2.40204811e-01
3.79038960e-01 -1.35951233e+00 -8.83652151e-01 1.82340011e-01
1.71200022e-01 3.60142618e-01 1.37518615e-01 5.20384192e-01
-2.60228068e-01 3.31102848e-01 1.93954200e-01 -4.43983912e-01
-1.06701541e+00 4.24055576e-01 3.47561568e-01 -5.43572426e-01
-4.36871916e-01 1.17721903e+00 7.18570650e-02 -5.52586317e-01
6.50624871e-01 -1.97511092e-01 -5.66843860e-02 -1.38781175e-01
7.96883285e-01 4.23673868e-01 5.26217937e-01 -3.33414942e-01
-2.15851068e-01 4.98520657e-02 -1.49157047e-01 1.57006994e-01
1.40105844e+00 8.42152238e-02 1.41779557e-01 2.39479676e-01
1.35882473e+00 1.08431347e-01 -1.47802854e+00 -5.35733283e-01
-9.22800750e-02 -5.37151515e-01 2.83335447e-01 -1.12925744e+00
-1.13408744e+00 9.33177114e-01 6.60694659e-01 7.23427385e-02
1.11651695e+00 -8.52261186e-02 9.12812948e-01 5.74514925e-01
2.93931067e-01 -9.53440189e-01 -1.00305460e-01 4.39420223e-01
5.49870431e-01 -1.59679580e+00 -2.43344367e-01 -2.84170091e-01
-6.31709933e-01 5.59951603e-01 7.07386792e-01 -1.26269534e-01
5.53408146e-01 2.16621011e-01 2.53075480e-01 -5.64909242e-02
-8.50993574e-01 -7.51742125e-02 1.38850451e-01 5.57473123e-01
2.09854782e-01 3.29865783e-01 -1.17853098e-01 1.05467761e+00
-1.67420000e-01 1.21054333e-02 2.92810649e-01 7.31429040e-01
-7.11697459e-01 -9.40894365e-01 -3.27099115e-01 8.66740942e-01
-7.04757273e-01 -3.97294104e-01 4.56380509e-02 3.70404691e-01
3.23711932e-01 1.14086533e+00 -2.50252873e-01 -4.14832473e-01
2.97293276e-01 1.84701562e-01 -1.21035971e-01 -6.60853446e-01
-4.55363065e-01 8.63792822e-02 2.34366059e-01 -5.43125212e-01
-1.85718372e-01 -4.79388773e-01 -1.10710979e+00 -1.19332239e-01
-5.07103264e-01 3.71720612e-01 4.35884833e-01 8.35462511e-01
4.19367880e-01 3.36009473e-01 6.98354542e-01 -7.06440628e-01
-9.35164034e-01 -9.99696016e-01 -6.31569207e-01 5.58771610e-01
5.79061806e-01 -5.11453688e-01 -5.99930108e-01 2.67263036e-02] | [9.365472793579102, 3.885223150253296] |
b675c815-00bb-4194-9126-3cd032e2f622 | clustering-individuals-based-on-multivariate | 2212.01159 | null | https://arxiv.org/abs/2212.01159v1 | https://arxiv.org/pdf/2212.01159v1.pdf | Clustering individuals based on multivariate EMA time-series data | In the field of psychopathology, Ecological Momentary Assessment (EMA) methodological advancements have offered new opportunities to collect time-intensive, repeated and intra-individual measurements. This way, a large amount of data has become available, providing the means for further exploring mental disorders. Consequently, advanced machine learning (ML) methods are needed to understand data characteristics and uncover hidden and meaningful relationships regarding the underlying complex psychological processes. Among other uses, ML facilitates the identification of similar patterns in data of different individuals through clustering. This paper focuses on clustering multivariate time-series (MTS) data of individuals into several groups. Since clustering is an unsupervised problem, it is challenging to assess whether the resulting grouping is successful. Thus, we investigate different clustering methods based on different distance measures and assess them for the stability and quality of the derived clusters. These clustering steps are illustrated on a real-world EMA dataset, including 33 individuals and 15 variables. Through evaluation, the results of kernel-based clustering methods appear promising to identify meaningful groups in the data. So, efficient representations of EMA data play an important role in clustering. | ['Anne Roefs', 'Lourens Waldorp', 'Gerasimos Spanakis', 'Mandani Ntekouli'] | 2022-12-02 | null | null | null | null | ['clustering-multivariate-time-series'] | ['time-series'] | [-1.75801039e-01 -4.48137760e-01 -2.06584319e-01 -5.07822573e-01
-4.13616151e-01 -2.55802095e-01 4.03533399e-01 8.77422631e-01
-3.84587109e-01 3.12375218e-01 -5.02980836e-02 4.08981666e-02
-7.50632346e-01 -4.94256705e-01 2.85371393e-01 -8.67899477e-01
-7.83777893e-01 4.20967847e-01 -3.20101321e-01 1.36133954e-01
3.33948135e-01 5.97127140e-01 -1.41827977e+00 1.25045091e-01
1.28134906e+00 5.52333772e-01 2.97336996e-01 -1.66579895e-02
-1.17032923e-01 1.32848710e-01 -5.00906110e-01 -1.28952771e-01
-2.62903005e-01 -5.26133835e-01 -6.00324333e-01 4.41080600e-01
-3.41116101e-01 2.19939590e-01 3.05826634e-01 9.96899486e-01
3.32169771e-01 3.87451440e-01 6.51363671e-01 -1.34398973e+00
-4.59907681e-01 4.91395056e-01 -5.40162027e-01 2.07340702e-01
4.67602909e-01 -2.60283742e-02 7.00810254e-01 -6.13139570e-01
3.47640038e-01 1.31600869e+00 5.73966622e-01 1.85655728e-01
-1.66731727e+00 -4.63056356e-01 -7.12796077e-02 7.51390159e-01
-1.40575492e+00 -3.17207813e-01 9.37675357e-01 -7.22791791e-01
6.08115792e-01 2.05826178e-01 8.43982041e-01 9.86303985e-01
8.51256475e-02 2.26182610e-01 1.47232413e+00 -3.01569253e-01
6.03635132e-01 2.34211966e-01 5.26745141e-01 2.81858832e-01
2.28111267e-01 -2.41602048e-01 -2.20954150e-01 -3.85579795e-01
4.05598015e-01 1.39817998e-01 6.48494437e-02 -1.72824994e-01
-1.05154335e+00 7.91663289e-01 2.07360402e-01 8.57773244e-01
-7.05837131e-01 -5.11584818e-01 3.54508728e-01 1.55220151e-01
5.28485656e-01 2.84666240e-01 -5.36392406e-02 -4.65073347e-01
-9.98223424e-01 -2.44555939e-02 1.90371126e-01 9.38225165e-02
6.09933972e-01 -2.93090455e-02 1.17943235e-01 1.04331458e+00
1.11712970e-01 1.75850496e-01 8.95389557e-01 -8.81106675e-01
1.86718464e-01 1.01311767e+00 -1.39149994e-01 -1.64660418e+00
-8.37826431e-01 -2.31176376e-01 -1.19418228e+00 7.68037513e-02
3.64942938e-01 7.84320831e-02 -2.37061605e-01 1.75128341e+00
4.58643615e-01 1.00971825e-01 -2.02405885e-01 4.97796267e-01
8.65751356e-02 3.93519849e-01 3.08934361e-01 -8.18158686e-01
1.18215370e+00 -1.49840489e-01 -1.00724983e+00 1.00334652e-01
6.34333551e-01 -2.77294844e-01 8.86654854e-01 5.73380470e-01
-1.08240604e+00 -7.97672749e-01 -5.70162475e-01 4.32399839e-01
-2.57203341e-01 2.39729166e-01 8.39071870e-01 6.35305405e-01
-1.03587055e+00 8.01895678e-01 -1.31106699e+00 -5.24789274e-01
3.26177299e-01 3.99596125e-01 -6.10844493e-01 -1.36916209e-02
-7.87834466e-01 5.27480483e-01 5.11449039e-01 2.37969846e-01
-3.01650107e-01 -4.00654793e-01 -5.79371452e-01 1.32552058e-01
-1.50576353e-01 -1.77373543e-01 3.95695269e-01 -8.06457520e-01
-1.03263712e+00 7.08598316e-01 -3.87944937e-01 -1.21679701e-01
-6.05987245e-03 2.72077054e-01 -7.12678909e-01 3.86166751e-01
2.82066554e-01 1.00704573e-01 7.68964112e-01 -9.01101530e-01
-3.58637005e-01 -1.04145944e+00 -7.00651288e-01 -2.92176735e-02
-9.11559999e-01 2.74982631e-01 1.91258088e-01 -3.17591935e-01
6.00913346e-01 -6.78484917e-01 -3.15774918e-01 -4.66222823e-01
-2.51601100e-01 -5.92142642e-01 3.76246274e-01 -8.85905683e-01
1.52018237e+00 -2.41917467e+00 2.58649796e-01 4.11805958e-01
4.27256197e-01 1.48493964e-02 7.51575604e-02 6.87440038e-01
-3.80477369e-01 3.87118645e-02 -2.50714481e-01 -3.90650898e-01
-1.38336316e-01 -8.59738812e-02 2.28577822e-01 7.31892705e-01
1.07513405e-01 4.94959563e-01 -7.72438705e-01 -5.01220167e-01
2.01303825e-01 1.43358022e-01 -4.02445287e-01 3.80993605e-01
4.68469292e-01 7.02255189e-01 -3.34788740e-01 5.35884261e-01
5.74412465e-01 -2.31195211e-01 4.42851603e-01 1.51566833e-01
-2.17051983e-01 -1.72990635e-01 -1.08134544e+00 1.46934068e+00
1.24730155e-01 5.07750511e-01 -4.65029664e-02 -1.48821092e+00
1.07376504e+00 3.14582318e-01 9.46323872e-01 -7.47272968e-01
5.36901243e-02 1.05724037e-02 3.62523079e-01 -8.52015436e-01
1.42518625e-01 -7.97438398e-02 1.49482921e-01 4.79026258e-01
-2.05365941e-01 5.96423984e-01 1.57241210e-01 -2.58099101e-02
1.05060184e+00 -5.59204280e-01 4.19944346e-01 -2.95771152e-01
3.57153386e-01 -9.55490395e-02 7.17081487e-01 1.66781396e-01
-4.34955329e-01 8.90066624e-02 3.66903722e-01 -1.77229479e-01
-7.41162300e-01 -1.02960527e+00 -3.92331332e-01 5.10482311e-01
-2.27003306e-01 -3.53786618e-01 -6.78674698e-01 2.60449778e-02
-1.60116911e-01 5.65533757e-01 -5.68211436e-01 -3.73545229e-01
-1.40911981e-01 -1.16244483e+00 2.04174206e-01 3.71634245e-01
2.18923137e-01 -1.11280990e+00 -3.63060534e-01 3.39893550e-01
-3.16706985e-01 -9.43138659e-01 2.56713510e-01 -2.34171391e-01
-1.45774102e+00 -1.08126140e+00 -2.21875355e-01 -5.76407015e-01
7.63317108e-01 4.54400718e-01 5.71785152e-01 -5.52723892e-02
-4.98960227e-01 3.39582205e-01 -3.78435701e-01 1.63783908e-01
-1.55889750e-01 -1.75875559e-01 5.15972197e-01 4.25431192e-01
7.32022166e-01 -1.05424023e+00 -3.93011183e-01 6.66767433e-02
-7.26771355e-01 -2.61377960e-01 4.81402338e-01 5.06437957e-01
2.82019526e-01 7.80508876e-01 8.21142375e-01 -3.08106840e-01
1.12694085e+00 -9.27818418e-01 -2.96416044e-01 1.48402840e-01
-8.27325165e-01 -4.12669659e-01 6.21450305e-01 -7.72468567e-01
-8.15894902e-01 -3.49101126e-01 1.69712633e-01 -3.45977753e-01
-6.83247745e-01 8.26295972e-01 -8.81816000e-02 3.65151137e-01
6.18724346e-01 3.50547343e-01 3.25316101e-01 -6.29068851e-01
5.14508411e-02 8.20582926e-01 2.89216936e-01 -5.48470497e-01
5.28489113e-01 5.35645187e-01 -5.46621010e-02 -1.12648022e+00
-1.30822584e-01 -7.17027962e-01 -9.51100945e-01 -2.90615588e-01
8.78742039e-01 -5.45639217e-01 -1.04173076e+00 4.41773355e-01
-8.13658893e-01 -7.38878548e-03 5.74307516e-02 8.59113395e-01
-3.77654016e-01 5.52726567e-01 -5.27194142e-01 -1.16742647e+00
-1.05594888e-01 -9.16843712e-01 4.90223259e-01 1.71725437e-01
-5.92996716e-01 -1.23332119e+00 4.12999839e-01 4.17543739e-01
3.87827866e-02 3.83368909e-01 1.17612910e+00 -7.61070311e-01
1.33812338e-01 -3.20186615e-01 7.32260197e-02 7.89738968e-02
5.60812414e-01 7.47698471e-02 -6.03869319e-01 -4.66534704e-01
4.02074277e-01 -1.21248001e-02 2.78374851e-01 4.67442781e-01
1.14412439e+00 -1.32441908e-01 -2.94939190e-01 2.07261771e-01
9.85532641e-01 5.72156727e-01 4.82100815e-01 3.38119030e-01
6.02160275e-01 1.16822445e+00 5.89328647e-01 8.02325666e-01
4.60798502e-01 5.40895939e-01 1.34267062e-01 1.32718474e-01
5.72720647e-01 1.87890157e-01 2.87035197e-01 1.22888184e+00
-2.84098864e-01 3.67688656e-01 -9.08787608e-01 5.98290324e-01
-1.81455493e+00 -1.21462178e+00 -5.70046008e-01 2.30734038e+00
5.69757819e-01 -3.22274685e-01 5.94187915e-01 5.98019481e-01
9.15334404e-01 -1.97013482e-01 -4.58474785e-01 -2.58597702e-01
-3.33496975e-03 -4.91136685e-02 -5.48553169e-01 8.68675187e-02
-7.81656563e-01 5.16372323e-01 6.03480196e+00 7.17705309e-01
-8.62575471e-01 -2.73317434e-02 6.36649609e-01 1.39378950e-01
-6.77929521e-02 -1.01013578e-01 -2.80038685e-01 6.17369115e-01
1.21244609e+00 -3.42821836e-01 6.29994452e-01 5.32252133e-01
9.86729085e-01 -2.02698499e-01 -8.29567373e-01 1.17565525e+00
-2.78799325e-01 -5.88758290e-01 -3.51865768e-01 3.77803504e-01
4.23875600e-01 -6.57252610e-01 9.02274996e-02 7.89338350e-02
-1.48434564e-01 -9.38530087e-01 7.08151935e-03 7.17574298e-01
3.85806859e-01 -8.94604504e-01 3.91125917e-01 4.77590263e-01
-9.45793688e-01 -2.75271714e-01 -5.46752870e-01 -3.26887190e-01
-7.54286423e-02 1.02389336e+00 -7.42566407e-01 5.92465997e-01
6.44780278e-01 9.18889940e-01 -6.90167844e-01 1.06508505e+00
1.26938000e-01 7.03276217e-01 -1.08036054e-02 1.62192076e-01
-8.41526762e-02 -8.02349627e-01 2.82951146e-01 8.22251022e-01
4.26065832e-01 2.15096608e-01 1.34801522e-01 1.12124729e+00
5.60539067e-01 4.50758040e-01 -4.08880383e-01 -4.17639107e-01
5.79191089e-01 1.50066316e+00 -9.96476650e-01 -1.18804134e-01
-3.00797313e-01 7.82766938e-01 5.77092111e-01 2.45211095e-01
-4.06774968e-01 5.40720709e-02 7.66938329e-01 9.81419832e-02
-3.86850148e-01 -7.03691363e-01 -2.50834793e-01 -1.01927114e+00
-8.15445259e-02 -9.00673747e-01 5.69922507e-01 -3.32496643e-01
-1.46885240e+00 3.31510961e-01 1.35501295e-01 -1.07877636e+00
-3.26462656e-01 -6.01274557e-02 -7.46277034e-01 6.82924807e-01
-7.23470628e-01 -4.75636721e-01 -3.15593183e-01 8.77614021e-01
3.16927493e-01 -1.01420999e-01 8.98589551e-01 2.52154976e-01
-1.02467132e+00 1.97325870e-01 5.02506495e-01 -5.01075797e-02
5.33094287e-01 -1.06852269e+00 -3.35741669e-01 6.45914912e-01
-8.95388611e-03 8.07726800e-01 3.05093706e-01 -8.48416150e-01
-1.14809227e+00 -6.34181440e-01 8.19711387e-01 -1.25553265e-01
8.38769734e-01 -2.65036732e-01 -1.34054267e+00 2.99185127e-01
-6.04535965e-03 -5.21303713e-01 1.16827941e+00 5.15623987e-01
1.38757482e-01 -2.01458812e-01 -1.14997482e+00 5.70459068e-01
7.37215877e-01 -4.57901686e-01 -4.33364540e-01 1.73282236e-01
1.10800207e-01 6.06900632e-01 -1.41307712e+00 1.90470114e-01
2.41040826e-01 -1.19071794e+00 9.09012794e-01 -6.54665112e-01
1.91085652e-01 5.45307621e-02 2.06516236e-01 -1.44037187e+00
-6.89533114e-01 -2.54173249e-01 5.07976040e-02 1.46962917e+00
-1.88906878e-01 -6.16969109e-01 5.44873714e-01 9.76492107e-01
1.19057655e-01 -4.93855923e-01 -8.04530263e-01 -7.36626208e-01
-1.49371594e-01 -4.16059762e-01 3.38465661e-01 1.54526579e+00
7.40294635e-01 -4.05040793e-02 -7.14599490e-02 1.86627626e-01
9.51279759e-01 1.14377551e-01 3.62601936e-01 -1.65443170e+00
1.47340447e-03 -5.54514587e-01 -7.09472954e-01 1.87486202e-01
2.41361737e-01 -7.40281403e-01 -4.22454685e-01 -1.17657697e+00
3.15721720e-01 -4.53898221e-01 -2.88445681e-01 3.45850408e-01
-4.52116758e-01 -1.93093643e-01 -9.06763822e-02 4.56990540e-01
-4.59529340e-01 7.30114520e-01 7.90712655e-01 2.18376532e-01
-6.00474298e-01 6.25687838e-02 -4.14084226e-01 6.20821178e-01
1.16954863e+00 -3.36175859e-01 -5.73114514e-01 1.63888529e-01
-1.97197780e-01 1.80497989e-01 2.12588653e-01 -1.13764393e+00
2.60594666e-01 -3.34324896e-01 4.22734737e-01 -4.46796596e-01
3.55302334e-01 -7.69514084e-01 3.79623890e-01 4.11788076e-01
-1.75777778e-01 2.13075429e-01 4.40556109e-02 2.55367368e-01
-2.51318365e-01 -1.71257615e-01 6.23052239e-01 6.29211217e-02
-5.02500474e-01 1.75590605e-01 -6.37867808e-01 -2.38971949e-01
1.09548187e+00 -1.81759700e-01 2.03054637e-01 -2.45288849e-01
-1.21819699e+00 4.21646647e-02 1.63728580e-01 2.17226580e-01
7.72330582e-01 -1.39201999e+00 -5.13936937e-01 2.33484030e-01
-8.81066471e-02 -4.81381476e-01 6.18094981e-01 1.34287965e+00
7.11797625e-02 3.44541878e-01 -5.14632106e-01 -6.92536414e-01
-1.35842609e+00 8.18695664e-01 -2.39966109e-01 -1.21203423e-01
-5.30100763e-01 -3.88911143e-02 -9.59704369e-02 -9.41575319e-02
-1.40123172e-02 2.08477883e-04 -7.04718053e-01 5.65756202e-01
7.17173696e-01 9.48685944e-01 -1.55334443e-01 -5.40043712e-01
-1.82614058e-01 2.66387701e-01 1.82670116e-01 -1.38918594e-01
1.61756432e+00 -3.30297947e-01 -6.23397768e-01 8.50531578e-01
1.13430512e+00 -3.17763805e-01 -6.53867066e-01 4.59311791e-02
5.00825524e-01 -5.43365717e-01 -2.30199918e-01 -1.71452835e-01
-6.35254860e-01 9.41989064e-01 7.87708700e-01 6.42706990e-01
1.38369310e+00 -3.17340046e-02 1.47343427e-01 1.57646388e-01
3.44802171e-01 -1.08471084e+00 2.24175945e-01 -1.88048467e-01
5.41193664e-01 -1.16823685e+00 -2.55853385e-01 -1.43783122e-01
-4.31983054e-01 1.09201062e+00 3.68808091e-01 -5.74519001e-02
4.69218582e-01 -3.03824633e-01 -2.38682151e-01 -3.82042110e-01
-4.46251094e-01 -1.38349324e-01 1.99969858e-01 7.28636682e-01
4.29989994e-01 3.26548427e-01 -7.37001300e-01 8.00594449e-01
-1.54580921e-01 -4.18451041e-01 2.49082789e-01 5.28248131e-01
-4.37591136e-01 -1.12753987e+00 -7.82949984e-01 3.78661782e-01
-3.12527239e-01 2.01855421e-01 -3.78807068e-01 7.99672246e-01
-1.77722469e-01 1.30154026e+00 -1.08902708e-01 -4.08810169e-01
-1.21731395e-02 2.54471511e-01 2.06180483e-01 -4.35177058e-01
-2.45953009e-01 2.25350007e-01 -2.19396636e-01 -5.78996360e-01
-6.75623298e-01 -1.18437600e+00 -1.26573813e+00 -4.04515475e-01
-2.20780373e-02 4.61343020e-01 8.00638199e-01 9.85062897e-01
3.75867218e-01 4.00973827e-01 1.11185384e+00 -6.55745506e-01
-2.66335815e-01 -1.04741740e+00 -9.53239262e-01 7.40093648e-01
-2.15910092e-01 -6.49407089e-01 -3.55055660e-01 -3.27291302e-02] | [7.280033588409424, 3.449646472930908] |
b8414cf7-e66e-4bd8-bd9a-99de27d0b25e | self-supervised-one-shot-learning-for | 2303.05639 | null | https://arxiv.org/abs/2303.05639v2 | https://arxiv.org/pdf/2303.05639v2.pdf | Self-Supervised One-Shot Learning for Automatic Segmentation of StyleGAN Images | We propose a framework for the automatic one-shot segmentation of synthetic images generated by a StyleGAN. Our framework is based on the observation that the multi-scale hidden features in the GAN generator hold useful semantic information that can be utilized for automatic on-the-fly segmentation of the generated images. Using these features, our framework learns to segment synthetic images using a self-supervised contrastive clustering algorithm that projects the hidden features into a compact space for per-pixel classification. This novel contrastive learner is based on using a pixel-wise swapped prediction loss for image segmentation that leads to faster learning of the feature vectors for one-shot segmentation. We have tested our implementation on a number of standard benchmarks to yield a segmentation performance that not only outperforms the semi-supervised baseline methods by an average wIoU margin of 1.02% but also improves the inference speeds by a factor of 4.5. Finally, we also show the results of using the proposed one-shot learner in implementing BagGAN, a framework for producing annotated synthetic baggage X-ray scans for threat detection. This framework was trained and tested on the PIDRay baggage benchmark to yield a performance comparable to its baseline segmenter based on manual annotations. | ['Avinash C. Kak', 'Ankit Manerikar'] | 2023-03-10 | null | null | null | null | ['one-shot-segmentation', 'one-shot-learning'] | ['computer-vision', 'methodology'] | [ 7.68046796e-01 6.91951990e-01 -5.24063520e-02 -4.49769467e-01
-1.64706683e+00 -5.60039937e-01 6.99954271e-01 -2.87772894e-01
-3.33297640e-01 6.01758063e-01 -1.69810802e-01 -9.94623173e-03
3.03740531e-01 -9.56043601e-01 -1.15595150e+00 -1.07005310e+00
1.55187294e-01 8.66039515e-01 5.20901799e-01 -1.17178569e-02
1.10446863e-01 3.23090583e-01 -1.42441106e+00 4.52061921e-01
4.33310866e-01 1.11707532e+00 5.74297272e-02 1.10038257e+00
1.60873711e-01 7.44283140e-01 -7.93274760e-01 -5.33471763e-01
7.03892350e-01 -9.82745528e-01 -8.72958481e-01 3.62616539e-01
2.36678421e-01 -4.04710412e-01 -4.15437557e-02 8.18274081e-01
4.17862862e-01 -7.38341287e-02 8.01545262e-01 -1.13435662e+00
9.03732553e-02 8.24066460e-01 -7.09451318e-01 -5.01977243e-02
2.00207859e-01 2.95603305e-01 7.93443382e-01 -3.64145517e-01
8.75007987e-01 1.09574497e+00 5.85560679e-01 6.86730266e-01
-1.02351260e+00 -6.63348436e-01 -5.39696872e-01 -4.47573699e-02
-1.05044925e+00 -1.31072193e-01 6.11205816e-01 -4.30593729e-01
6.40013218e-01 4.75120366e-01 6.16909444e-01 1.11961126e+00
2.27853909e-01 8.87651026e-01 1.42228377e+00 -6.80653095e-01
4.05512780e-01 2.92161763e-01 -9.87007841e-02 1.00139463e+00
-1.28245831e-01 3.07278991e-01 -3.15892130e-01 -3.60776968e-02
6.93335354e-01 -3.03182960e-01 1.07858717e-01 -1.38145864e-01
-9.33539093e-01 1.26428890e+00 4.55091834e-01 1.25374809e-01
-3.15415412e-01 5.55042863e-01 3.75255078e-01 -7.02622458e-02
6.20521069e-01 5.79585254e-01 2.45313533e-02 3.18471752e-02
-1.43574202e+00 1.98150322e-01 6.81051135e-01 7.87671983e-01
8.25309634e-01 2.19357744e-01 -5.18526912e-01 5.51591396e-01
2.02950053e-02 5.52593350e-01 3.19939941e-01 -1.11777198e+00
8.96288231e-02 2.44684756e-01 -1.84266821e-01 -4.87181455e-01
-6.61858870e-03 -3.10376972e-01 -5.26403844e-01 5.30054986e-01
2.17585623e-01 -4.42665547e-01 -1.54809523e+00 1.39178169e+00
3.87118429e-01 4.17219669e-01 2.93256883e-02 7.07010984e-01
5.29635370e-01 8.48282933e-01 -8.97802263e-02 9.62961689e-02
1.28081667e+00 -1.12028086e+00 -3.20702016e-01 2.66137533e-02
5.57729781e-01 -7.26748168e-01 7.79308915e-01 5.47847033e-01
-1.25208080e+00 -4.36009467e-01 -1.27647722e+00 3.06780666e-01
-1.17376357e-01 -1.78639188e-01 3.77733171e-01 1.15204239e+00
-8.72601628e-01 5.85286915e-01 -8.43793571e-01 -2.06032306e-01
7.70862937e-01 2.98425257e-01 7.35967904e-02 1.92831177e-02
-9.35124278e-01 6.06284559e-01 7.96293557e-01 -4.87154216e-01
-1.37106550e+00 -5.86578012e-01 -7.34524250e-01 -1.03935279e-01
4.22429144e-01 -4.83123899e-01 1.20357215e+00 -1.14188218e+00
-1.61193264e+00 9.45347428e-01 2.70660996e-01 -1.09264696e+00
9.06482279e-01 2.91562006e-02 1.13818515e-02 8.14920723e-01
2.08180174e-01 1.24904740e+00 1.11959052e+00 -1.43479621e+00
-6.56754911e-01 1.08187143e-02 -3.72250788e-02 1.43205017e-01
2.13404343e-01 1.28713235e-01 -5.22170305e-01 -8.37234378e-01
-4.79228646e-01 -1.15508211e+00 -3.62724066e-01 -3.85117769e-01
-7.80167341e-01 3.36802632e-01 1.00966740e+00 -8.31693590e-01
5.33143520e-01 -1.87010646e+00 8.73814374e-02 4.23604041e-01
9.48084518e-02 6.09435849e-02 1.43984541e-01 1.42002851e-01
-3.98943648e-02 -1.01188086e-01 -9.90081429e-01 -3.29077899e-01
-2.64983177e-01 9.17899087e-02 -4.32083070e-01 4.31958020e-01
5.10899760e-02 1.06834507e+00 -7.13068664e-01 -7.10415244e-01
2.95924604e-01 1.15248017e-01 -4.15110260e-01 4.19706464e-01
-5.39748788e-01 5.10810971e-01 -1.33472383e-01 5.08011997e-01
5.43150306e-01 4.11645435e-02 -2.73426890e-01 9.79998335e-02
3.27490658e-01 -2.73480028e-01 -8.10529888e-01 1.80297363e+00
-2.71786362e-01 4.88519043e-01 -3.24410170e-01 -9.25167561e-01
7.69705653e-01 2.60407984e-01 4.58011627e-01 -3.51482689e-01
3.03156227e-01 3.07387114e-02 -3.07733536e-01 1.02037519e-01
2.87449092e-01 -2.56173581e-01 -4.04106319e-01 9.62121785e-01
2.95014977e-01 -7.38472462e-01 2.08125234e-01 4.59104836e-01
1.18063557e+00 2.79252887e-01 -4.52222768e-03 -1.51973814e-01
2.60451347e-01 3.02835464e-01 2.94401139e-01 9.82783794e-01
3.94129455e-01 9.73138034e-01 5.58816195e-01 -2.11980760e-01
-1.26605594e+00 -1.09216774e+00 9.38426778e-02 8.45148087e-01
-6.44320473e-02 -3.97791788e-02 -1.62141097e+00 -1.01004291e+00
-6.34823561e-01 1.17605948e+00 -7.19542205e-01 -7.41071925e-02
-5.20132124e-01 -9.51613963e-01 9.11571622e-01 4.85944182e-01
6.74686551e-01 -1.31601417e+00 -1.01016092e+00 7.07786605e-02
8.06642547e-02 -1.05952072e+00 -2.00183526e-01 3.39069873e-01
-5.13439298e-01 -1.00632823e+00 -7.76045501e-01 -7.14425862e-01
7.14662492e-01 -1.44257605e-01 9.34066176e-01 -9.90797952e-02
-6.54379666e-01 4.80153829e-01 -4.96227950e-01 -6.66562974e-01
-8.98004293e-01 9.97241437e-02 -5.17797470e-01 7.54781067e-02
-1.62791565e-01 -1.35753192e-02 -5.13777375e-01 2.39761725e-01
-1.23995531e+00 3.47111553e-01 6.11076355e-01 9.51302648e-01
8.23512256e-01 5.48833869e-02 4.61297154e-01 -1.44439471e+00
1.56500503e-01 -4.37009037e-01 -8.66107106e-01 3.10816318e-01
-3.05896252e-01 2.15068966e-01 5.20927429e-01 1.07213221e-02
-1.29874134e+00 3.61762643e-01 -2.99736559e-01 -3.21891963e-01
-2.04316407e-01 -2.80776799e-01 -3.55277285e-02 -1.90252468e-01
7.12577403e-01 2.32136548e-01 -7.53268301e-02 6.21220730e-02
8.33429158e-01 5.18783092e-01 7.33582258e-01 -3.62177581e-01
9.77661550e-01 5.65167606e-01 1.60558224e-02 -7.07900286e-01
-1.02780783e+00 -2.86757857e-01 -4.09525722e-01 -1.52259246e-01
1.41997015e+00 -7.79287279e-01 -1.71508419e-03 6.69401109e-01
-8.08339000e-01 -5.97405612e-01 -8.02335978e-01 1.20462514e-01
-1.10956037e+00 2.58040696e-01 -8.41264188e-01 -5.20247698e-01
-5.82064211e-01 -1.34277129e+00 1.28479815e+00 1.02119401e-01
-1.20049246e-01 -8.39874208e-01 6.65857494e-02 6.09879792e-01
6.28131181e-02 6.52611196e-01 8.08060706e-01 -7.56483912e-01
-7.71303236e-01 -2.28001624e-01 -8.65858197e-02 5.43224931e-01
-3.56893778e-01 -2.48605981e-01 -1.22429705e+00 -2.37265199e-01
3.23776603e-01 -6.69085443e-01 1.08787739e+00 4.85575289e-01
1.39525259e+00 -2.79544562e-01 -2.17120081e-01 7.93319941e-01
1.49775970e+00 2.32406229e-01 9.71864164e-01 8.39276090e-02
7.77934670e-01 3.51763994e-01 8.85378361e-01 1.88707218e-01
-2.14534432e-01 5.44653356e-01 4.52353984e-01 -3.85721862e-01
-3.16659272e-01 -2.42263615e-01 1.42562374e-01 2.85975605e-01
1.69877723e-01 -3.77169281e-01 -8.02269042e-01 5.34698606e-01
-1.50794709e+00 -1.04098809e+00 6.63322285e-02 1.92042613e+00
8.03834498e-01 3.31321478e-01 3.29165488e-01 2.00507209e-01
7.88777888e-01 2.51897991e-01 -3.43770176e-01 -5.65235078e-01
6.04361221e-02 8.50064337e-01 1.02234435e+00 4.28314477e-01
-1.21323621e+00 1.16109562e+00 6.33282137e+00 1.27719831e+00
-1.05259418e+00 4.01922911e-01 1.22354364e+00 6.95365593e-02
-2.41012126e-01 1.56159345e-02 -6.55319929e-01 4.91657019e-01
1.12419081e+00 1.10281527e-01 2.48664081e-01 9.24225688e-01
-3.48481506e-01 -3.61701965e-01 -8.02699685e-01 7.94268966e-01
4.92660820e-01 -1.76695454e+00 -1.95030317e-01 8.56286101e-03
1.20330453e+00 -7.93284969e-04 1.19693778e-01 8.71298537e-02
6.61962926e-01 -9.68107402e-01 8.63195598e-01 2.74610519e-01
1.10400081e+00 -1.12353599e+00 6.96817696e-01 2.71704704e-01
-6.70428753e-01 9.60863754e-02 -1.49511412e-01 5.61143458e-01
3.11203748e-01 4.87019330e-01 -1.59000742e+00 4.97220606e-01
3.28417867e-01 9.65431482e-02 -5.91658890e-01 9.38816249e-01
-4.90466952e-01 9.72488463e-01 -2.93535858e-01 3.28991890e-01
5.81987500e-01 3.10839489e-02 4.92034912e-01 1.26696217e+00
3.92498195e-01 -5.15578724e-02 9.81592610e-02 8.10346484e-01
-1.68935403e-01 -2.41202489e-01 -5.01662314e-01 9.66039449e-02
5.30219786e-02 1.29478753e+00 -1.30335772e+00 -5.76442719e-01
5.70353568e-02 1.30063701e+00 -3.36725861e-01 -7.11517259e-02
-1.25327301e+00 -6.30922496e-01 -1.29831001e-01 2.49130756e-01
7.41093457e-01 2.68870443e-01 -3.86100262e-01 -6.63278162e-01
-3.47361386e-01 -8.17311645e-01 4.07114834e-01 -6.22490287e-01
-8.66251409e-01 8.97765994e-01 3.46974075e-01 -1.03507113e+00
-9.15957391e-01 -2.79474229e-01 -9.15925920e-01 5.75751781e-01
-9.83302951e-01 -1.52804029e+00 -2.99812078e-01 3.36954802e-01
8.84843826e-01 -3.88443321e-01 9.65269148e-01 -2.93453097e-01
-2.13982731e-01 5.53285539e-01 -7.44610578e-02 2.07795769e-01
4.25414801e-01 -1.42069888e+00 7.58329570e-01 1.08231127e+00
3.21013272e-01 -2.27682099e-01 8.05370212e-01 -8.22803855e-01
-7.81696677e-01 -1.36909425e+00 7.19079971e-02 -3.62893373e-01
2.87488401e-01 -6.07604980e-01 -5.28343260e-01 8.39002430e-01
4.76394773e-01 -1.28951997e-01 6.47970617e-01 -7.62105942e-01
-1.74393043e-01 3.70397754e-02 -1.61431098e+00 1.65153563e-01
6.38984144e-01 -2.64844149e-01 -5.63538074e-01 5.87380052e-01
8.02028239e-01 -5.10746598e-01 -6.22621953e-01 7.88887367e-02
2.38099352e-01 -1.14355445e+00 8.06288064e-01 -1.71564981e-01
6.58931136e-01 -1.60812646e-01 7.55179115e-03 -1.32878792e+00
2.08270863e-01 -6.64125502e-01 3.44545066e-01 1.10884500e+00
3.79326522e-01 -3.57537061e-01 1.10272837e+00 1.84485719e-01
-7.90773332e-02 -6.09390557e-01 -7.39126325e-01 -6.34645402e-01
-1.21076241e-01 -4.30058807e-01 4.54517782e-01 3.99656415e-01
-6.81687891e-01 7.70260170e-02 -4.82445359e-01 -2.00547874e-01
1.14432526e+00 1.93477288e-01 7.89893031e-01 -7.25439847e-01
-8.31548274e-01 1.20106913e-01 -6.59477353e-01 -6.04777277e-01
1.98950455e-01 -8.67639780e-01 2.80808330e-01 -1.20744598e+00
3.09029847e-01 -4.87827390e-01 2.08304208e-02 3.25292826e-01
2.23465469e-02 8.28681052e-01 2.45628625e-01 -1.53410211e-01
-6.37618542e-01 2.70975344e-02 8.24422300e-01 -1.39184728e-01
1.19937107e-01 3.69807035e-02 -4.92541879e-01 7.97947586e-01
8.75683725e-01 -9.13845062e-01 -3.64422441e-01 -5.33625595e-02
-2.59259492e-01 2.09228560e-01 3.50184858e-01 -1.32916796e+00
-9.16772932e-02 1.00478046e-01 5.26976883e-01 -6.89504445e-01
5.77574253e-01 -4.71405834e-01 2.68690020e-01 7.05160558e-01
-3.67752254e-01 -5.54138124e-01 -7.10275695e-02 5.13848126e-01
-1.17933147e-01 -5.54113925e-01 1.14455998e+00 -4.67911750e-01
-6.37706399e-01 1.27647057e-01 -2.94150740e-01 2.20362172e-01
1.58560348e+00 -2.32110083e-01 -2.83273719e-02 -5.03144860e-01
-5.32702565e-01 -1.70395508e-01 7.35566437e-01 -1.17084319e-02
6.07935786e-01 -1.03261423e+00 -7.13322520e-01 3.57881606e-01
-5.82742617e-02 2.03148704e-02 2.18398452e-01 3.71588856e-01
-1.04174387e+00 2.15543225e-01 -5.31448424e-01 -7.77614832e-01
-1.27501833e+00 3.61088097e-01 1.72904551e-01 -5.42541802e-01
-6.82392240e-01 1.12904406e+00 3.35357040e-01 -1.72201365e-01
-1.80743393e-02 -7.39954179e-03 2.92210728e-01 -1.82526127e-01
5.28431594e-01 4.15092140e-01 6.14353307e-02 -6.06499076e-01
8.23853984e-02 3.87317091e-01 -5.26098311e-02 -5.86904228e-01
1.50654733e+00 3.39890093e-01 7.63763934e-02 2.98873752e-01
1.12479341e+00 -3.54981283e-03 -1.45761776e+00 1.33471131e-01
-9.28919464e-02 -4.16334331e-01 -3.78895202e-03 -9.01835799e-01
-1.24890411e+00 6.94821537e-01 6.94894075e-01 2.07631383e-02
1.19921637e+00 2.34236702e-01 1.16725957e+00 7.95971677e-02
4.99062508e-01 -1.07638943e+00 1.64480478e-01 -6.44717664e-02
5.46170533e-01 -1.02481318e+00 5.61934486e-02 -4.80293840e-01
-1.10557127e+00 9.94809747e-01 2.13157773e-01 -6.01825058e-01
2.10757345e-01 4.99755979e-01 4.12606709e-02 -2.68367141e-01
-3.68250191e-01 1.27574531e-02 2.04278186e-01 6.98833108e-01
-1.52354077e-01 2.69059449e-01 4.97643203e-02 2.22545311e-01
-3.93107086e-01 -2.01785892e-01 6.75704837e-01 8.30652833e-01
-5.27105987e-01 -1.27854252e+00 -5.97148955e-01 6.23830557e-01
-7.16345012e-01 1.85245305e-01 -4.49593633e-01 6.56587183e-01
2.15770841e-01 6.15112543e-01 4.46251919e-03 -2.33228624e-01
-3.18742067e-01 5.79218119e-02 7.13407457e-01 -7.69248962e-01
-7.45828748e-01 1.97067931e-01 6.33181259e-02 -5.61066508e-01
-2.91750193e-01 -5.34356177e-01 -1.34507680e+00 2.18382984e-01
-3.00063312e-01 1.25673339e-01 8.05362821e-01 9.12909329e-01
-2.28564903e-01 5.56287527e-01 6.99062765e-01 -9.12727177e-01
-4.50811207e-01 -6.22422040e-01 -5.74238420e-01 3.56185019e-01
-2.43320122e-01 -3.97963166e-01 -2.53746033e-01 2.47322515e-01] | [11.351344108581543, -0.31183692812919617] |
298e93bb-217b-400a-9891-8394b3403776 | stock-movement-prediction-based-on-bi-typed-1 | 2201.04965 | null | https://arxiv.org/abs/2201.04965v2 | https://arxiv.org/pdf/2201.04965v2.pdf | Stock Movement Prediction Based on Bi-typed Hybrid-relational Market Knowledge Graph via Dual Attention Networks | Stock Movement Prediction (SMP) aims at predicting listed companies' stock future price trend, which is a challenging task due to the volatile nature of financial markets. Recent financial studies show that the momentum spillover effect plays a significant role in stock fluctuation. However, previous studies typically only learn the simple connection information among related companies, which inevitably fail to model complex relations of listed companies in the real financial market. To address this issue, we first construct a more comprehensive Market Knowledge Graph (MKG) which contains bi-typed entities including listed companies and their associated executives, and hybrid-relations including the explicit relations and implicit relations. Afterward, we propose DanSmp, a novel Dual Attention Networks to learn the momentum spillover signals based upon the constructed MKG for stock prediction. The empirical experiments on our constructed datasets against nine SOTA baselines demonstrate that the proposed DanSmp is capable of improving stock prediction with the constructed MKG. | ['Ji Liu', 'Gang Kou', 'Fuzhen Zhuang', 'Qing Li', 'Xingyan Chen', 'Shaopeng Wei', 'Ying Liu', 'Huaming Du', 'Yu Zhao'] | 2022-01-11 | null | null | null | null | ['implicit-relations', 'stock-prediction'] | ['natural-language-processing', 'time-series'] | [-9.51033711e-01 5.23718037e-02 -7.45274186e-01 -5.48822880e-02
-6.03774562e-02 -6.54126644e-01 6.08272910e-01 -4.66664970e-01
5.71097806e-02 8.80299151e-01 5.09177923e-01 -6.18589342e-01
-1.91893354e-01 -1.34176600e+00 -6.61249518e-01 -2.30546266e-01
-1.42696396e-01 5.12844384e-01 2.85394341e-01 -4.93257165e-01
3.70023936e-01 2.53431916e-01 -6.54144228e-01 -3.69151831e-02
7.59603560e-01 1.26274526e+00 -2.17510849e-01 -2.20319450e-01
-5.80882311e-01 1.73509955e+00 -3.36759925e-01 -9.83835697e-01
6.37468338e-01 -1.36646062e-01 -4.92360950e-01 -3.14769894e-01
5.29122213e-03 -3.95425081e-01 -7.39772022e-01 1.27262354e+00
-2.03950587e-03 -2.78990537e-01 3.19429934e-01 -1.36638248e+00
-1.27645373e+00 1.57262003e+00 -7.32707083e-01 8.95953119e-01
-3.93885553e-01 3.34688127e-02 1.93763387e+00 -9.26768422e-01
6.28921449e-01 8.20275426e-01 8.17968547e-01 1.04791336e-02
-8.72473955e-01 -1.36201203e+00 7.52190411e-01 3.70627016e-01
-9.87317920e-01 5.81034087e-02 9.71458375e-01 -4.34548497e-01
1.02554023e+00 2.41857693e-02 1.10253727e+00 7.29310393e-01
4.11054432e-01 6.90379739e-01 1.00930130e+00 3.90685469e-01
-3.45078319e-01 1.81670994e-01 5.08599281e-01 2.19562799e-01
8.70000601e-01 1.32219106e-01 -8.80429983e-01 8.34125504e-02
1.06726062e+00 -3.06210667e-02 -3.50475848e-01 1.21703900e-01
-1.30924273e+00 8.94474447e-01 8.41329753e-01 5.63597560e-01
-5.87290704e-01 2.32358351e-01 3.03575426e-01 5.42728841e-01
8.75840485e-01 6.06705785e-01 -8.67530942e-01 -1.06265716e-01
-6.96383774e-01 2.72364885e-01 1.05881882e+00 9.23475325e-01
5.07081687e-01 3.17526847e-01 9.01577175e-02 1.54082164e-01
4.83611375e-01 2.98614174e-01 8.14590216e-01 -4.51357275e-01
1.01463056e+00 1.00176060e+00 8.01589116e-02 -1.28885758e+00
-5.55644870e-01 -1.18968630e+00 -8.15411389e-01 -1.67049527e-01
7.73441866e-02 -2.78262675e-01 -3.52371961e-01 1.39862192e+00
-2.14343965e-01 7.24659145e-01 2.02218577e-01 6.43906951e-01
8.81225824e-01 6.29184961e-01 -2.83302099e-01 -4.30179119e-01
1.08683622e+00 -1.29888892e+00 -1.00546741e+00 -1.92812145e-01
3.45098227e-01 -1.77501574e-01 4.96082842e-01 -7.09298402e-02
-8.72040927e-01 -1.78538233e-01 -1.04419863e+00 1.26320124e-01
-4.64298844e-01 -1.66238472e-01 1.04950655e+00 2.21661404e-01
-7.49892950e-01 6.83878064e-01 -7.23221540e-01 6.23468876e-01
5.69362640e-01 5.00000775e-01 1.09818771e-01 7.39781499e-01
-1.72207987e+00 8.60775113e-01 5.31980932e-01 5.86887300e-01
-1.15507096e-01 -9.87087309e-01 -5.07357299e-01 5.05359828e-01
6.01206899e-01 -5.57552338e-01 1.09276366e+00 -6.86460614e-01
-1.28718138e+00 3.53973627e-01 4.22488868e-01 -8.83200765e-01
5.40586948e-01 -3.85193199e-01 -6.83637738e-01 -2.88446844e-01
7.39499256e-02 -8.69773794e-03 1.74620762e-01 -5.91389239e-01
-7.20040739e-01 -1.50696814e-01 1.69507563e-01 1.69789508e-01
-5.10623932e-01 -1.70019999e-01 -2.01273501e-01 -1.11274076e+00
1.87184587e-01 -5.80476344e-01 -9.38756168e-02 -8.47107828e-01
-7.07297921e-01 -2.97316045e-01 5.22380352e-01 -8.29349756e-01
1.68232703e+00 -1.57380581e+00 -2.04965964e-01 4.06708956e-01
3.46702337e-01 3.57283559e-03 6.75660819e-02 4.11513150e-01
-3.37183863e-01 3.74082029e-01 2.51944691e-01 2.29296252e-01
3.16172153e-01 -8.92800316e-02 -1.01608706e+00 -1.26671150e-01
2.46190518e-01 1.67424858e+00 -6.93267882e-01 -9.52158645e-02
-4.94971216e-01 -1.18720762e-01 -3.05180192e-01 -1.70540914e-01
-4.50598657e-01 8.84158090e-02 -8.12603295e-01 7.00778961e-01
6.90532446e-01 -7.44710326e-01 2.35351518e-01 -1.87933043e-01
-2.12071255e-01 8.50397468e-01 -1.07574999e+00 8.00189018e-01
6.86537549e-02 5.10141075e-01 -5.38403571e-01 -6.87526941e-01
1.17950809e+00 2.54898548e-01 6.17634356e-01 -7.78781176e-01
2.81677721e-03 5.36692977e-01 4.76364791e-01 2.32203230e-02
6.13258481e-01 -3.68111759e-01 1.53391838e-01 5.93491495e-01
-3.19963902e-01 3.97606134e-01 2.34631836e-01 2.88115054e-01
1.01798272e+00 -5.58068557e-03 2.02745244e-01 -2.96204627e-01
4.79373842e-01 -4.72810715e-02 1.27018523e+00 2.30742708e-01
4.18602712e-02 7.54032061e-02 1.14832664e+00 -5.79639554e-01
-6.89084709e-01 -7.75813937e-01 -5.05690873e-02 3.71293426e-01
2.75225639e-02 -2.75473326e-01 -6.26778007e-02 -8.13982785e-01
5.56199968e-01 6.02342844e-01 -5.95574796e-01 7.20013827e-02
-5.53212464e-01 -1.16815674e+00 2.43116900e-01 9.32112575e-01
9.94890273e-01 -1.17374086e+00 4.24186289e-02 3.52829337e-01
1.64716735e-01 -1.01308143e+00 -5.43647826e-01 5.56506962e-02
-8.59031558e-01 -1.35673904e+00 -5.96900880e-01 -5.88889778e-01
2.78378308e-01 -2.02199429e-01 1.47255409e+00 -1.10474057e-01
5.45301616e-01 -2.72711128e-01 -6.30460456e-02 -5.06725848e-01
1.10552698e-01 4.38318670e-01 -1.20762035e-01 1.49319202e-01
6.78226650e-01 -7.23846734e-01 -5.55074215e-01 2.32345179e-01
-4.24332082e-01 1.93635032e-01 6.87647998e-01 8.69413316e-01
3.98703188e-01 5.85025787e-01 1.25981581e+00 -1.34325707e+00
8.36180687e-01 -7.65974700e-01 -1.02538872e+00 2.18495622e-01
-1.28179455e+00 -5.97341694e-02 2.00991154e-01 -1.99785694e-01
-1.06856275e+00 -5.54033339e-01 1.52242571e-01 -3.39904577e-01
9.50334132e-01 1.42546153e+00 -1.00699216e-01 2.39185765e-01
-1.69498861e-01 3.50454032e-01 -3.49349886e-01 -3.67375016e-01
9.11885723e-02 1.28107995e-01 4.28032249e-01 -1.10684276e-01
1.21698642e+00 2.25123897e-01 -8.52392521e-03 7.01675490e-02
-1.14328682e+00 -1.44121677e-01 -5.21348298e-01 1.30035207e-01
5.76880336e-01 -1.17142665e+00 -7.58994997e-01 8.02551210e-01
-9.16572273e-01 -6.49283007e-02 -1.58369482e-01 6.38551950e-01
-1.56249970e-01 -1.56586587e-01 -1.18383086e+00 -8.46067071e-01
-5.37453651e-01 -6.28122151e-01 2.17990175e-01 2.50136077e-01
1.81196168e-01 -1.30508602e+00 2.34600335e-01 2.97191769e-01
5.06140649e-01 1.81510717e-01 9.49203312e-01 -1.10647237e+00
-1.32519972e+00 -1.41054288e-01 -5.45414448e-01 2.91878074e-01
4.22667325e-01 -2.34047383e-01 -4.59488720e-01 1.59585163e-01
-4.37286869e-02 8.04824233e-02 1.22263646e+00 1.86923966e-01
3.38821650e-01 -4.73365813e-01 -9.50387940e-02 5.77788353e-01
1.30702662e+00 5.02023637e-01 4.50133413e-01 8.32392097e-01
8.80577207e-01 4.13830310e-01 5.31810641e-01 3.16116035e-01
7.91483164e-01 2.18404591e-01 2.89464563e-01 3.10035855e-01
4.70395625e-01 -4.84565645e-01 3.65441471e-01 1.37726724e+00
-3.23121428e-01 1.32194787e-01 -8.68943810e-01 2.89965570e-01
-2.03411627e+00 -1.25604331e+00 -2.75502264e-01 1.56044614e+00
8.96405220e-01 7.29723752e-01 1.05427764e-01 -1.26264527e-01
5.77188015e-01 5.04436970e-01 -9.77520704e-01 4.99399126e-01
-5.84582925e-01 -1.53897896e-01 7.72876740e-01 2.26605952e-01
-1.00147891e+00 9.01521742e-01 5.92329359e+00 3.33835244e-01
-1.12216496e+00 -2.20511764e-01 8.01191032e-01 -2.32048295e-02
-8.45412910e-01 9.97431483e-03 -1.19956720e+00 8.12827706e-01
6.87330604e-01 -9.69538808e-01 1.02070786e-01 7.92587519e-01
-5.94955757e-02 4.92303520e-01 -8.47359717e-01 7.29438066e-01
-4.56942141e-01 -1.95261979e+00 7.19035938e-02 3.48023862e-01
1.00474024e+00 1.18255578e-01 2.89956003e-01 5.36941230e-01
5.87774038e-01 -7.98863053e-01 7.98034668e-01 8.96173060e-01
7.43650924e-03 -8.61663640e-01 1.12723887e+00 2.83536822e-01
-1.57013035e+00 -3.31220955e-01 -3.09257030e-01 -2.26697415e-01
1.86855420e-01 6.58015192e-01 -3.76727700e-01 9.32563066e-01
6.29015207e-01 1.46127450e+00 -4.85977918e-01 7.43513525e-01
-3.47560048e-01 6.85896337e-01 -4.72585065e-03 -9.69412476e-02
3.33266944e-01 -7.35001683e-01 1.96470782e-01 4.99871314e-01
4.99073654e-01 1.11657932e-01 -2.71707982e-01 1.29738832e+00
-6.67650580e-01 5.80450855e-02 -4.51699048e-01 -7.82752156e-01
3.46345097e-01 1.13727987e+00 -7.16034889e-01 -2.87035435e-01
-8.78156841e-01 4.71106291e-01 3.21653157e-01 2.66938120e-01
-7.83749163e-01 -9.68146920e-02 6.05917633e-01 1.37407854e-01
5.15934169e-01 -1.90314334e-02 -5.03781974e-01 -1.65040874e+00
2.83642530e-01 -6.78207815e-01 5.03769636e-01 -6.44496739e-01
-1.67716026e+00 6.27356529e-01 -3.80443633e-01 -1.21207118e+00
-1.35334715e-01 -6.26030684e-01 -1.00701535e+00 1.16998661e+00
-1.99152756e+00 -1.03126276e+00 1.79521441e-01 2.18618736e-01
9.28353891e-02 -6.49510741e-01 1.77235767e-01 3.30812752e-01
-9.43010867e-01 1.33916020e-01 9.44272652e-02 6.11568928e-01
1.86622068e-01 -1.49712956e+00 1.01405454e+00 6.46324694e-01
4.56968755e-01 8.02068233e-01 2.67180521e-02 -1.26536286e+00
-9.75360990e-01 -1.09272528e+00 1.07701325e+00 -4.18143302e-01
1.68836701e+00 5.29899858e-02 -1.24168992e+00 1.31013405e+00
3.69823039e-01 -8.56625065e-02 4.05269444e-01 2.55386263e-01
-5.16626954e-01 -3.02469850e-01 -3.58984649e-01 3.39852780e-01
1.08811367e+00 -4.32645291e-01 -9.25154388e-01 1.91526771e-01
1.00853968e+00 -3.50453347e-01 -1.04416811e+00 4.84379113e-01
3.46903384e-01 -9.97997761e-01 7.62162328e-01 -6.15882516e-01
6.73521638e-01 -1.42194793e-01 3.22842002e-01 -1.21359992e+00
-5.63389421e-01 -4.36137110e-01 -5.60756743e-01 1.58762980e+00
8.78584862e-01 -1.20020807e+00 1.09039247e+00 6.77975655e-01
7.74071664e-02 -9.84732389e-01 -7.17988312e-01 -8.94275188e-01
2.16637045e-01 -1.09185323e-01 1.10140514e+00 1.43243361e+00
2.94152070e-02 5.84683776e-01 -4.66468841e-01 1.79206673e-02
2.15518191e-01 7.58949220e-01 3.94952416e-01 -1.56699049e+00
-5.24365544e-01 -9.01295543e-01 -2.43325263e-01 -9.17903006e-01
4.38614011e-01 -1.05600953e+00 -7.72068799e-01 -1.57649863e+00
1.97234660e-01 -5.11396468e-01 -8.91668320e-01 2.32122079e-01
-2.45749533e-01 -1.79876477e-01 2.36421943e-01 6.91350102e-01
-2.83603638e-01 6.82048619e-01 1.47401369e+00 -2.88102597e-01
-1.68427899e-01 3.08882236e-01 -1.01882660e+00 7.17301965e-01
7.49490440e-01 -1.99034005e-01 -3.44408214e-01 -2.30596103e-02
1.12745321e+00 2.20099054e-02 7.36103207e-02 -4.36598957e-01
4.20025885e-01 -2.41705865e-01 2.75056779e-01 -1.11288762e+00
7.67881749e-03 -6.46673262e-01 3.74671966e-01 6.46654367e-01
-1.94093674e-01 5.18738508e-01 -1.82466973e-02 6.76236928e-01
-7.67404556e-01 5.24174832e-02 5.71444295e-02 -2.34609917e-01
-8.07003438e-01 7.07218587e-01 2.64006525e-01 1.59079090e-01
8.92926455e-01 2.60492206e-01 -7.10753083e-01 -3.98179233e-01
-5.89344561e-01 6.60557151e-01 -1.25789270e-01 5.38840652e-01
3.55431467e-01 -1.71919096e+00 -7.29603827e-01 -9.10546705e-02
-3.15613866e-01 1.17354371e-01 -2.78723091e-02 1.03723693e+00
-6.03911817e-01 7.15388894e-01 -1.74923047e-01 3.72580826e-01
-6.16728067e-01 6.21940076e-01 5.63446045e-01 -9.87516999e-01
-7.53113627e-01 8.85014892e-01 3.22653055e-01 -9.60159078e-02
-1.18298940e-01 -6.77932024e-01 -6.23436928e-01 5.77043593e-01
2.33158991e-01 3.49435240e-01 -2.01635703e-01 -4.26310867e-01
-2.02186748e-01 4.78739440e-01 -2.99258798e-01 6.28655180e-02
1.68307281e+00 2.08002608e-02 -4.94629383e-01 7.63990760e-01
6.42290711e-01 3.22746247e-01 -1.14238477e+00 -5.41985333e-01
8.00764859e-01 -3.03465456e-01 -4.85793985e-02 -6.42547548e-01
-1.91320276e+00 3.88310313e-01 -3.44469905e-01 5.57524264e-01
8.73553574e-01 -2.41318852e-01 1.06507158e+00 6.18061483e-01
3.91036659e-01 -1.06046569e+00 -1.90797204e-03 6.72605157e-01
1.01759636e+00 -1.05518544e+00 1.04433507e-01 -6.37315214e-01
-7.11499929e-01 1.05985391e+00 8.42833638e-01 -2.86661565e-01
1.19766736e+00 2.26143762e-01 3.15662265e-01 -4.40786749e-01
-1.07554901e+00 -2.92728674e-02 3.59907359e-01 -4.20661904e-02
2.40008056e-01 -1.44395754e-01 -1.92517087e-01 1.33588898e+00
-6.95053995e-01 -2.59944685e-02 6.10041916e-01 5.87013245e-01
-1.12459123e-01 -1.07215273e+00 -9.94050410e-03 9.68016446e-01
-7.13416278e-01 -5.06594241e-01 -5.72258472e-01 1.06076407e+00
-8.23410749e-02 2.32481480e-01 3.54855180e-01 -6.23098314e-01
4.14771467e-01 1.07265890e-01 -2.10365444e-01 -5.58657169e-01
-9.27147508e-01 8.49568397e-02 1.26672477e-01 -2.53840655e-01
-4.02299047e-01 -7.76311159e-01 -1.35881484e+00 -4.88970190e-01
-4.55158919e-01 3.20037693e-01 -8.28448683e-02 7.01728940e-01
3.73932779e-01 9.10437047e-01 6.88513458e-01 -1.32559851e-01
-7.61143982e-01 -9.26395178e-01 -1.37456512e+00 2.41596729e-01
1.53697014e-01 -7.98203588e-01 -3.02980185e-01 -2.81579465e-01] | [4.30248498916626, 4.315986156463623] |
f14d4c04-472a-4b2f-9c2b-806994a799f6 | pyramid-region-based-slot-attention-network | 2206.10095 | null | https://arxiv.org/abs/2206.10095v1 | https://arxiv.org/pdf/2206.10095v1.pdf | Pyramid Region-based Slot Attention Network for Temporal Action Proposal Generation | It has been found that temporal action proposal generation, which aims to discover the temporal action instances within the range of the start and end frames in the untrimmed videos, can largely benefit from proper temporal and semantic context exploitation. The latest efforts were dedicated to considering the temporal context and similarity-based semantic contexts through self-attention modules. However, they still suffer from cluttered background information and limited contextual feature learning. In this paper, we propose a novel Pyramid Region-based Slot Attention (PRSlot) module to address these issues. Instead of using the similarity computation, our PRSlot module directly learns the local relations in an encoder-decoder manner and generates the representation of a local region enhanced based on the attention over input features called \textit{slot}. Specifically, upon the input snippet-level features, PRSlot module takes the target snippet as \textit{query}, its surrounding region as \textit{key} and then generates slot representations for each \textit{query-key} slot by aggregating the local snippet context with a parallel pyramid strategy. Based on PRSlot modules, we present a novel Pyramid Region-based Slot Attention Network termed PRSA-Net to learn a unified visual representation with rich temporal and semantic context for better proposal generation. Extensive experiments are conducted on two widely adopted THUMOS14 and ActivityNet-1.3 benchmarks. Our PRSA-Net outperforms other state-of-the-art methods. In particular, we improve the AR@100 from the previous best 50.67% to 56.12% for proposal generation and raise the mAP under 0.5 tIoU from 51.9\% to 58.7\% for action detection on THUMOS14. \textit{Code is available at} \url{https://github.com/handhand123/PRSA-Net} | ['Jun Hou', 'Lingbo Liu', 'Kunlin Yang', 'Rui Feng', 'Rui-Wei Zhao', 'Feng Zhang', 'Shuaicheng Li'] | 2022-06-21 | null | null | null | null | ['temporal-action-proposal-generation'] | ['computer-vision'] | [ 5.15230656e-01 6.17119335e-02 -5.24352014e-01 -2.20534623e-01
-7.86826849e-01 -1.77211687e-01 6.66563988e-01 -8.39962214e-02
-4.65086251e-01 6.94312274e-01 4.53220159e-01 5.12772202e-02
-1.22059602e-03 -6.66183472e-01 -7.80986249e-01 -6.95633352e-01
6.29314408e-02 -6.09537913e-03 7.67016649e-01 -2.24806681e-01
1.35993749e-01 9.20994431e-02 -1.51784599e+00 6.55269027e-01
6.12939775e-01 1.32906139e+00 5.55993915e-01 4.81716663e-01
-1.67281911e-01 8.83190215e-01 -5.66287935e-01 -1.02991082e-01
2.31999397e-01 -6.38162911e-01 -6.73922002e-01 7.35510420e-03
1.97270334e-01 -3.65305185e-01 -5.36677837e-01 8.45099926e-01
5.47285378e-01 4.15633649e-01 3.23809832e-01 -1.30022955e+00
-3.89680594e-01 5.95365822e-01 -7.53618658e-01 7.09625542e-01
3.43932092e-01 3.84528637e-01 1.08209157e+00 -9.97815669e-01
8.57230306e-01 1.18992794e+00 2.66584575e-01 4.32559490e-01
-8.52577031e-01 -8.15229535e-01 6.52929366e-01 5.60423315e-01
-1.44483984e+00 -1.80703357e-01 7.64215946e-01 -1.73077717e-01
1.05373168e+00 1.41816184e-01 6.63140357e-01 1.39569509e+00
2.55719423e-01 1.24287605e+00 9.62365448e-01 -1.58513322e-01
1.42148793e-01 -3.00171226e-01 -1.08976141e-01 5.95293045e-01
-2.86513299e-01 8.75843018e-02 -7.72289813e-01 2.54608661e-01
1.04040515e+00 2.34922349e-01 -1.49997756e-01 -1.32440701e-01
-1.59258842e+00 6.56691551e-01 6.05499744e-01 4.47671384e-01
-5.47931492e-01 5.24146497e-01 5.69692016e-01 -1.31669924e-01
3.21484774e-01 5.68526126e-02 -4.44141239e-01 -2.75070697e-01
-8.34869087e-01 2.07306743e-01 1.40707821e-01 9.95390236e-01
6.76374555e-01 9.51088294e-02 -8.13871086e-01 7.02820837e-01
2.08135754e-01 3.06627303e-01 6.51832163e-01 -8.91212642e-01
6.68178141e-01 6.57209575e-01 1.24744825e-01 -8.69676888e-01
-2.86137789e-01 -5.70412755e-01 -5.52038014e-01 1.25713833e-02
6.03382923e-02 -1.00829281e-01 -1.22262478e+00 1.71940315e+00
3.72657895e-01 5.29432893e-01 -7.71995028e-03 1.06342304e+00
9.10645962e-01 8.70271564e-01 6.05728090e-01 -2.08760753e-01
1.41214836e+00 -1.31987870e+00 -7.12680876e-01 -3.86689872e-01
3.33874762e-01 -6.72681868e-01 1.05998647e+00 9.15706009e-02
-9.66594458e-01 -8.81330788e-01 -9.34370697e-01 -3.74096930e-02
-2.99769372e-01 3.65368515e-01 4.75621819e-01 -1.09085031e-01
-8.26872528e-01 4.28396374e-01 -8.83937776e-01 -4.85458136e-01
6.57929420e-01 2.76013076e-01 -2.69327968e-01 1.33254332e-03
-1.18998671e+00 4.40687597e-01 6.28106475e-01 9.48748067e-02
-1.10337281e+00 -4.41310257e-01 -8.92330348e-01 -1.29297879e-02
9.29834843e-01 -5.59941828e-01 1.29106224e+00 -1.21290481e+00
-1.37011814e+00 4.98679221e-01 -2.01771289e-01 -5.86012840e-01
3.07355881e-01 -3.07804465e-01 -4.75031406e-01 4.09365833e-01
4.70792115e-01 1.05697048e+00 8.32837641e-01 -8.33003759e-01
-9.50659633e-01 -1.50807515e-01 1.84058130e-01 5.51293850e-01
-5.30642159e-02 1.45022586e-01 -9.53002632e-01 -1.01372361e+00
5.17081693e-02 -8.13678026e-01 -2.68620700e-01 4.22755592e-02
-3.14805597e-01 -3.05918843e-01 8.56903374e-01 -5.32609880e-01
1.45205343e+00 -2.21029663e+00 1.13278314e-01 -1.47596151e-01
-6.27297014e-02 3.84115100e-01 -1.51667759e-01 3.11000973e-01
-8.77133012e-02 -1.14400826e-01 -1.83593288e-01 -1.27636001e-01
2.77970135e-02 2.27967203e-01 -2.61524796e-01 2.08339110e-01
2.60917723e-01 1.05359435e+00 -1.02120042e+00 -5.86658895e-01
4.40742582e-01 4.23267066e-01 -5.62342525e-01 1.99098125e-01
-4.46540803e-01 4.16537136e-01 -6.97708607e-01 7.96336889e-01
1.74866110e-01 -2.49121353e-01 -1.02728583e-01 -2.81890810e-01
-1.07428111e-01 1.98723137e-01 -1.13910866e+00 2.03656888e+00
-2.79927284e-01 5.04201114e-01 -2.84030050e-01 -9.98557210e-01
8.51536989e-01 3.19292635e-01 7.40374923e-01 -9.86263990e-01
1.42508045e-01 2.82621440e-02 -6.72973990e-02 -4.35595602e-01
5.41799963e-01 5.43233044e-02 -2.50165135e-01 1.39705285e-01
2.73679197e-01 4.71806049e-01 2.25155845e-01 2.40927175e-01
1.10149062e+00 7.80980408e-01 3.96448553e-01 2.83702835e-02
5.74320912e-01 -8.68647620e-02 7.97516465e-01 5.98647118e-01
-5.00491679e-01 5.75245976e-01 4.46111411e-01 -4.81379300e-01
-7.92841554e-01 -1.13827479e+00 1.52761444e-01 1.35379124e+00
4.61871117e-01 -6.32488668e-01 -5.12868583e-01 -9.69290376e-01
-2.33471543e-01 5.70671320e-01 -7.43925452e-01 -2.48326361e-01
-5.58280468e-01 -4.59615588e-01 3.10635298e-01 8.74288976e-01
9.27084804e-01 -1.51210701e+00 -9.49400425e-01 4.10594404e-01
-3.25249314e-01 -1.14440644e+00 -7.18364537e-01 5.31610698e-02
-6.74078047e-01 -9.00921702e-01 -7.43430495e-01 -4.78972375e-01
3.82590860e-01 1.06826492e-01 9.03526545e-01 -2.07576126e-01
-3.18236738e-01 2.92254865e-01 -6.47066414e-01 -1.84061244e-01
1.94075644e-01 -3.59850377e-02 -2.97203362e-01 2.52352506e-01
4.31508660e-01 -5.53463757e-01 -1.05555630e+00 5.79608560e-01
-9.45932448e-01 4.20968354e-01 7.07454920e-01 7.70442903e-01
9.17399347e-01 -1.65781856e-01 7.61121094e-01 -4.47420329e-01
1.29024804e-01 -6.54243469e-01 -3.20828080e-01 8.10384378e-02
-2.95819163e-01 -4.53310981e-02 3.14487100e-01 -3.94107163e-01
-1.12731087e+00 2.07049951e-01 -1.42708048e-01 -6.48115098e-01
-2.37653807e-01 2.53464669e-01 -9.41659361e-02 5.02205193e-01
6.06683731e-01 5.45187175e-01 -3.49193990e-01 -3.26980233e-01
3.45710218e-01 3.57417911e-01 5.39396167e-01 -4.62475598e-01
4.28526133e-01 5.06305873e-01 -3.38757336e-01 -4.19553965e-01
-9.32688653e-01 -5.03181279e-01 -3.67324352e-01 -3.50420415e-01
1.20461953e+00 -1.06113803e+00 -5.39594710e-01 2.32184857e-01
-8.66766155e-01 -4.88117456e-01 -4.52473402e-01 3.94282907e-01
-6.91221118e-01 3.61791581e-01 -3.17499518e-01 -7.15353906e-01
-4.41927850e-01 -1.25084877e+00 1.46845758e+00 3.37356716e-01
-1.17010571e-01 -5.81718445e-01 -2.58975297e-01 5.07550895e-01
1.76555499e-01 3.16234827e-01 3.81857067e-01 -5.73840141e-01
-8.14718425e-01 1.44757405e-01 -3.71794850e-01 1.19172230e-01
6.33646250e-02 -3.00825119e-01 -7.24201441e-01 -8.64996687e-02
-4.64474261e-01 -1.58861294e-01 9.59093809e-01 4.32038844e-01
1.33755338e+00 -3.08998197e-01 -4.08849865e-01 3.96775603e-01
1.32518041e+00 5.86541533e-01 6.42809570e-01 3.70462626e-01
5.60522497e-01 1.72966048e-01 1.10908318e+00 6.99758828e-01
3.25432420e-01 1.04153347e+00 6.28756762e-01 1.52527578e-02
-2.89276838e-01 -4.00162041e-01 5.52060246e-01 1.52391359e-01
-1.36108205e-01 -2.29874253e-01 -6.34077013e-01 7.67513692e-01
-2.15304756e+00 -1.10708499e+00 2.53238738e-01 1.86475849e+00
7.58792639e-01 2.74352044e-01 2.01185286e-01 -1.03147134e-01
7.51532078e-01 5.41051686e-01 -6.64195538e-01 -1.40948787e-01
1.59471594e-02 3.11782479e-01 3.08388948e-01 1.56546518e-01
-1.35780776e+00 1.27285624e+00 4.21053648e+00 1.10661483e+00
-1.02220345e+00 3.13393891e-01 6.24916315e-01 -3.05632919e-01
6.69067027e-04 -5.97364530e-02 -9.32430744e-01 7.00016320e-01
6.19387031e-01 -3.84811051e-02 1.39069632e-01 8.71399641e-01
4.12365168e-01 -3.67903531e-01 -8.39232445e-01 1.04938960e+00
-4.40783566e-03 -1.28738415e+00 -2.55922023e-02 -2.08540156e-01
7.04782605e-01 1.65420340e-03 1.19082898e-01 6.27829075e-01
8.14067274e-02 -8.71288359e-01 9.07680094e-01 5.48819244e-01
7.37658381e-01 -5.63453197e-01 6.42244756e-01 -3.55870137e-03
-1.67266798e+00 -3.91122073e-01 -1.26371875e-01 6.62711859e-02
2.72740602e-01 2.56003618e-01 -7.27363408e-01 7.02554643e-01
8.89385402e-01 1.04040706e+00 -4.43433166e-01 9.20370162e-01
-3.33446473e-01 5.54594576e-01 -2.12425992e-01 8.71027336e-02
6.78378999e-01 1.91569120e-01 4.99587476e-01 1.20816994e+00
3.86027187e-01 2.23261565e-01 4.53868777e-01 7.41455674e-01
1.24307223e-01 1.50168404e-01 -3.50576609e-01 1.89641759e-01
3.21277410e-01 1.11725175e+00 -7.88078010e-01 -5.35851777e-01
-4.32732254e-01 1.19018304e+00 7.66395852e-02 4.95312363e-01
-1.28634846e+00 -2.65601605e-01 5.81718981e-01 9.09152254e-02
6.89999640e-01 -6.58248663e-02 1.01045251e-01 -1.07204962e+00
3.16251963e-02 -7.02861726e-01 7.00193763e-01 -1.02336574e+00
-6.85818315e-01 6.83767974e-01 3.37652117e-01 -1.44066787e+00
-1.51773974e-01 -2.99368322e-01 -6.24232173e-01 6.19875252e-01
-1.33733404e+00 -1.25755870e+00 -4.19449478e-01 8.17868054e-01
1.21171641e+00 -1.52252913e-01 3.65192801e-01 3.13705653e-01
-6.16921544e-01 4.83541638e-01 -3.62283051e-01 1.03481691e-02
4.54975367e-01 -1.05203640e+00 2.82192767e-01 8.10237825e-01
2.10808679e-01 2.10668758e-01 5.03274441e-01 -7.08048165e-01
-9.39919114e-01 -1.37807047e+00 6.33521199e-01 -2.19150245e-01
5.61235785e-01 -1.83297262e-01 -5.60656786e-01 8.06506872e-01
2.08670512e-01 3.67748648e-01 3.99053574e-01 -4.66244847e-01
-8.78776237e-02 -1.80158660e-01 -8.21710706e-01 8.37625802e-01
1.40686440e+00 -3.98604572e-01 -6.25882030e-01 2.71996200e-01
8.76379073e-01 -5.00400543e-01 -7.97751546e-01 4.96010035e-01
4.54423785e-01 -9.89121258e-01 9.77885664e-01 -2.96953738e-01
5.16853154e-01 -5.97513616e-01 -2.53235638e-01 -7.35656142e-01
-2.78102756e-01 -6.37841284e-01 -2.18720645e-01 9.70988512e-01
2.82669961e-01 -1.72813103e-01 7.34769166e-01 8.53813142e-02
-4.31498170e-01 -1.06348789e+00 -1.19161844e+00 -6.00320756e-01
-4.54201043e-01 -6.17837071e-01 3.22618306e-01 4.90676165e-01
-1.29914358e-01 3.42268378e-01 -4.70965624e-01 -1.05882785e-03
2.99800277e-01 1.18760712e-01 6.01923823e-01 -5.43743193e-01
-3.49463314e-01 -2.90520281e-01 -4.90851164e-01 -1.21230292e+00
-3.65560316e-02 -8.03082824e-01 2.97033470e-02 -1.57027733e+00
1.70290396e-01 -1.65486276e-01 -6.71144903e-01 8.44138622e-01
-2.65203863e-01 4.13466662e-01 4.32743609e-01 4.58822679e-03
-1.08352411e+00 8.47015083e-01 1.37102187e+00 -1.49153560e-01
-2.67699689e-01 -8.19921270e-02 -4.89710063e-01 5.76126575e-01
8.60267580e-01 -2.82641202e-01 -6.88441694e-01 -2.12309465e-01
-1.20489530e-01 1.83453813e-01 5.75845420e-01 -1.23686075e+00
-2.31710984e-03 -3.11521530e-01 4.39180404e-01 -7.97357202e-01
5.73914289e-01 -7.57219613e-01 8.99195373e-02 3.83467317e-01
-3.87844086e-01 -7.79360952e-03 2.84406751e-01 7.66545773e-01
-2.08347440e-01 4.35693227e-02 5.89198649e-01 -2.57928312e-01
-1.33636189e+00 4.63307589e-01 -2.94123799e-01 -8.11072960e-02
1.37088883e+00 -2.61735857e-01 -1.66850448e-01 -2.03514278e-01
-9.59486067e-01 2.15970859e-01 3.75268832e-02 6.30888820e-01
7.14104295e-01 -1.50117373e+00 -4.20370221e-01 1.22455113e-01
3.34045798e-01 -6.17944896e-02 6.34858370e-01 9.20846581e-01
-8.69652703e-02 4.30572122e-01 -3.48133177e-01 -6.76732421e-01
-1.13193560e+00 5.96224129e-01 3.18749994e-01 -3.04826826e-01
-7.17263520e-01 8.35954607e-01 5.89244306e-01 1.49562001e-01
2.33835205e-01 -5.69251299e-01 -3.46045285e-01 -8.66151899e-02
4.48352575e-01 1.52993724e-01 -3.68616223e-01 -8.86409044e-01
-4.25605178e-01 5.53736925e-01 -3.24424617e-02 -1.01682112e-01
1.08489585e+00 -3.01397387e-02 3.88684958e-01 9.27021354e-02
1.01936591e+00 -3.85486126e-01 -1.71802640e+00 -3.51097077e-01
-1.41842246e-01 -5.19928396e-01 -1.02099985e-01 -8.90727520e-01
-1.19921827e+00 6.42977118e-01 7.57594645e-01 -2.90373087e-01
1.40790665e+00 2.11152524e-01 7.99273908e-01 6.63783476e-02
2.72430599e-01 -1.16772103e+00 5.46118796e-01 3.36260229e-01
1.11827672e+00 -1.15251076e+00 -4.76059131e-02 -2.04461798e-01
-9.32708263e-01 7.09673643e-01 1.06590283e+00 -8.26503038e-02
3.74802798e-01 -1.92778006e-01 -2.48528078e-01 -1.84942558e-01
-8.53852749e-01 -4.55663413e-01 3.61801654e-01 3.68338495e-01
3.03495228e-01 -5.46130873e-02 -5.05311072e-01 8.43567073e-01
2.89098322e-01 2.08910123e-01 1.40026331e-01 1.04678988e+00
-4.96239305e-01 -1.03453207e+00 -1.65081620e-01 3.19767147e-01
-5.59786797e-01 -6.30588382e-02 7.53285661e-02 7.65193403e-01
5.56480944e-01 7.30281055e-01 7.25394189e-02 -4.33741003e-01
3.03623289e-01 5.37295230e-02 1.44427314e-01 -6.96073413e-01
-5.56705892e-01 4.43264157e-01 9.09641013e-02 -1.05867529e+00
-5.92539728e-01 -7.45201290e-01 -1.57042789e+00 2.69204587e-01
8.74769464e-02 -7.30430558e-02 1.92345724e-01 8.74662876e-01
5.73200285e-01 8.43066752e-01 4.03103858e-01 -8.79347801e-01
-1.32631004e-01 -9.13424134e-01 -3.14746380e-01 4.00613755e-01
1.00480229e-01 -9.46117997e-01 4.11400646e-02 1.26228049e-01] | [8.47105884552002, 0.47273802757263184] |
2065fb0a-f7c4-4523-a0e4-400f663cfacc | classify-respiratory-abnormality-in-lung | 2208.13943 | null | https://arxiv.org/abs/2208.13943v1 | https://arxiv.org/pdf/2208.13943v1.pdf | Classify Respiratory Abnormality in Lung Sounds Using STFT and a Fine-Tuned ResNet18 Network | Recognizing patterns in lung sounds is crucial to detecting and monitoring respiratory diseases. Current techniques for analyzing respiratory sounds demand domain experts and are subject to interpretation. Hence an accurate and automatic respiratory sound classification system is desired. In this work, we took a data-driven approach to classify abnormal lung sounds. We compared the performance using three different feature extraction techniques, which are short-time Fourier transformation (STFT), Mel spectrograms, and Wav2vec, as well as three different classifiers, including pre-trained ResNet18, LightCNN, and Audio Spectrogram Transformer. Our key contributions include the bench-marking of different audio feature extractors and neural network based classifiers, and the implementation of a complete pipeline using STFT and a fine-tuned ResNet18 network. The proposed method achieved Harmonic Scores of 0.89, 0.80, 0.71, 0.36 for tasks 1-1, 1-2, 2-1 and 2-2, respectively on the testing sets in the IEEE BioCAS 2022 Grand Challenge on Respiratory Sound Classification. | ['Xilin Liu', 'Chia-Hui Yeh', 'Hongliang Wang', 'Zizhao Chen'] | 2022-08-30 | null | null | null | null | ['sound-classification'] | ['audio'] | [ 3.90455395e-01 -2.94935405e-01 4.75000411e-01 -7.01750815e-02
-7.22169697e-01 -4.15674448e-01 2.36327380e-01 8.45985040e-02
-4.70018446e-01 4.86328632e-01 4.08124924e-01 -1.21791206e-01
-2.20905185e-01 -4.18842733e-01 -1.04570344e-01 -5.19829929e-01
-1.07668545e-02 6.94304109e-02 5.96046925e-01 -1.44436851e-01
3.71646360e-02 4.60796714e-01 -1.69157386e+00 6.45966232e-01
1.23835377e-01 1.18766224e+00 -3.15983891e-02 1.76184499e+00
-1.79954663e-01 1.06079662e+00 -6.94070816e-01 1.06531397e-01
5.62485009e-02 -6.22617662e-01 -8.97669733e-01 -6.89031065e-01
4.68253434e-01 1.18045777e-01 -2.93539584e-01 5.58821797e-01
1.21298599e+00 4.00299489e-01 4.75570112e-01 -1.01624894e+00
-2.23701254e-01 7.13310421e-01 2.77705729e-01 9.13219690e-01
4.61544216e-01 5.44530153e-02 1.09009993e+00 -8.74076188e-01
5.16182296e-02 8.50604773e-01 1.16196072e+00 8.64297867e-01
-7.56892741e-01 -8.64661872e-01 -7.17479944e-01 4.49093044e-01
-1.18548846e+00 -6.32653713e-01 6.49757683e-01 -5.98591685e-01
1.55332148e+00 7.01039076e-01 6.41800404e-01 1.12008214e+00
1.39340952e-01 1.76775545e-01 1.01162577e+00 -4.06415313e-01
7.91280717e-02 2.44250506e-01 1.30741403e-01 3.80074084e-01
3.69721069e-03 1.43044563e-02 -6.60947323e-01 -4.09208655e-01
2.95336872e-01 -1.34847611e-01 -3.28089923e-01 5.17932713e-01
-1.23375762e+00 6.16645873e-01 1.37629554e-01 9.99231279e-01
-6.11915410e-01 3.32831025e-01 5.79834402e-01 2.05761716e-01
3.24207962e-01 6.94522202e-01 -5.81553936e-01 -3.70397717e-01
-1.03770268e+00 1.17272235e-01 9.04251039e-01 3.56233090e-01
-6.98386878e-02 3.09361458e-01 -6.31840348e-01 1.27158916e+00
3.03575218e-01 5.07763088e-01 1.13617969e+00 -7.72926450e-01
1.64137453e-01 -1.46315217e-01 -3.09659570e-01 -9.49946582e-01
-9.61504459e-01 -4.29719359e-01 -8.01001489e-01 -4.65539515e-01
2.28154846e-02 -3.75270769e-02 -7.62652159e-01 1.39512837e+00
1.62973776e-01 6.96341217e-01 7.05503449e-02 4.66248512e-01
1.34583724e+00 4.70847636e-01 -1.12193851e-02 -1.54168934e-01
1.42583632e+00 -1.06891978e+00 -8.35313559e-01 2.45438725e-01
1.75894946e-01 -1.06365299e+00 1.09987843e+00 4.85592902e-01
-1.09542727e+00 -9.68441844e-01 -8.42775702e-01 7.21703395e-02
-3.74331802e-01 -5.75745618e-03 -1.38643667e-01 7.64554799e-01
-1.25426340e+00 9.09594297e-01 -7.95793951e-01 -3.91502053e-01
3.46318901e-01 4.55696344e-01 -8.43681470e-02 4.66048628e-01
-1.32923520e+00 6.78740025e-01 7.05988035e-02 -3.05550992e-01
-5.29019535e-01 -8.59095156e-01 -2.10508257e-01 2.18236551e-01
-2.39227101e-01 -7.56365359e-01 1.69671011e+00 -1.39741689e-01
-1.58098638e+00 6.06180310e-01 1.81355819e-01 -6.98946834e-01
3.64189118e-01 -2.70445704e-01 -8.42268407e-01 3.32939178e-01
-2.44272575e-01 2.53954321e-01 8.97127151e-01 -4.23889160e-01
-9.10952091e-01 2.28417665e-02 -5.80989063e-01 -1.44196525e-01
-5.11276960e-01 2.22441480e-01 2.26990551e-01 -7.90829718e-01
-2.11746350e-01 -8.21494520e-01 6.87064370e-03 -5.28687954e-01
-5.07925928e-01 -4.62262899e-01 6.12742066e-01 -7.66198814e-01
1.53614855e+00 -2.17992473e+00 -4.78011936e-01 4.54836972e-02
3.48413378e-01 6.13285482e-01 -1.43290684e-01 2.46314824e-01
-3.45519811e-01 2.26847783e-01 -5.37057705e-02 -1.11594573e-01
-9.69402269e-02 1.49669023e-02 -1.61752179e-01 2.18604177e-01
1.51115522e-01 5.99600613e-01 -8.44224751e-01 -2.77157933e-01
5.09045601e-01 9.53477025e-01 -5.47832251e-01 5.55826068e-01
1.77801773e-01 1.73233137e-01 -1.37439072e-01 1.72067583e-01
1.64214909e-01 -7.71357073e-03 -6.01879716e-01 -4.44831282e-01
-2.40564823e-01 9.37943757e-01 -1.03754187e+00 1.24737799e+00
-7.70270646e-01 6.81917250e-01 -3.68544221e-01 -6.41590655e-01
7.29257047e-01 1.09172893e+00 8.72733116e-01 -2.85808861e-01
1.97445422e-01 3.10931653e-01 2.15294272e-01 -1.03475881e+00
3.03052254e-02 -3.13028157e-01 3.08012933e-01 4.14946795e-01
3.59981626e-01 -3.24817806e-01 1.14097945e-01 -4.60941076e-01
1.51608539e+00 -5.86950779e-01 2.85327166e-01 -5.21162488e-02
8.89878750e-01 -3.99026662e-01 5.30903563e-02 7.72520125e-01
-3.61104071e-01 1.05164039e+00 -1.21482976e-01 -3.58921021e-01
-6.30427778e-01 -6.85662925e-01 -2.56469667e-01 1.30299282e+00
-8.22752357e-01 -4.86491293e-01 -8.00846934e-01 -5.85073948e-01
-8.01652968e-02 7.95968950e-01 -4.82942879e-01 -2.92519242e-01
-7.98065066e-01 -6.21230841e-01 1.22860885e+00 5.19438028e-01
1.32471338e-01 -1.57761848e+00 -9.11482930e-01 4.85643655e-01
-2.69834697e-01 -1.06800544e+00 -5.58860600e-01 4.85627621e-01
-5.11386096e-01 -1.05030179e+00 -6.71432734e-01 -5.94909430e-01
-2.39349082e-01 2.17134714e-01 1.00989187e+00 4.06867713e-02
-8.26574206e-01 6.03023946e-01 -3.16633761e-01 -7.74174750e-01
-6.09695911e-01 2.46671453e-01 2.96965748e-01 -1.64208338e-01
2.68233091e-01 -7.45813668e-01 -6.86102688e-01 -3.38337682e-02
-6.58691764e-01 -5.74080765e-01 4.08539385e-01 6.31783903e-01
7.02577531e-01 1.26108959e-01 8.48179698e-01 -7.38799334e-01
1.10109365e+00 -4.69998538e-01 3.11169233e-02 -8.45822990e-02
-6.56318903e-01 -2.82095969e-01 9.90725040e-01 -3.82985860e-01
-4.51373190e-01 -4.94903810e-02 -8.93190980e-01 -5.81395686e-01
-4.34270203e-01 -2.95364968e-02 5.16016960e-01 9.11677480e-02
1.13689435e+00 2.63092548e-01 -2.75703222e-01 -7.28725255e-01
2.32719108e-01 1.13748240e+00 7.32726336e-01 2.07707763e-01
8.04441512e-01 2.55776141e-02 -1.79954633e-01 -1.02040982e+00
-8.09320271e-01 -1.03101003e+00 -5.53756654e-01 -1.00560851e-01
1.20833242e+00 -6.02160752e-01 -3.87035817e-01 3.53282511e-01
-1.00943851e+00 -8.97688605e-03 -7.43532360e-01 8.33299220e-01
-3.62088144e-01 9.98141542e-02 -4.56062555e-01 -7.60733306e-01
-1.09614122e+00 -7.01906979e-01 6.73288107e-01 9.69756022e-02
-6.01183355e-01 -7.63306499e-01 7.19941735e-01 4.38506335e-01
1.10745299e+00 3.80739830e-02 6.73658192e-01 -1.37280917e+00
3.43956441e-01 -1.95433661e-01 1.40038699e-01 8.44965339e-01
6.54790640e-01 1.24091811e-01 -1.76045752e+00 -1.20324202e-01
3.78191173e-01 -2.81551152e-01 9.11744177e-01 3.12195778e-01
1.54155505e+00 -4.91004288e-01 1.77288145e-01 4.62735772e-01
7.20981658e-01 3.98215473e-01 2.78779805e-01 -1.69656008e-01
6.98403776e-01 3.02695870e-01 -1.55484816e-02 3.69564861e-01
1.07610919e-01 2.92137533e-01 1.04034714e-01 5.43747470e-02
-6.57141507e-01 -6.22909749e-03 2.07228661e-01 1.65020037e+00
-9.64552164e-02 -2.72048742e-01 -9.18093085e-01 5.94411731e-01
-1.01752651e+00 -9.97224033e-01 -2.35112727e-01 1.98289859e+00
8.49904299e-01 -3.86190973e-02 4.25586820e-01 9.02992725e-01
5.34809530e-01 8.13136771e-02 -2.32487723e-01 -5.35475433e-01
5.44575274e-01 1.02160561e+00 6.54128939e-02 1.64534956e-01
-1.37142682e+00 3.71393442e-01 6.47788143e+00 6.21182680e-01
-1.50143838e+00 2.70385325e-01 1.25405356e-01 -1.80112466e-01
1.49648875e-01 -1.00098300e+00 -5.47448814e-01 4.34594542e-01
1.71662104e+00 -1.95682738e-02 7.09805131e-01 6.54479563e-01
1.36494637e-01 6.10626340e-01 -9.44068968e-01 1.04030800e+00
8.41376111e-02 -1.00686073e+00 -2.65976816e-01 -4.88490015e-01
3.40936393e-01 8.00936639e-01 9.35424939e-02 2.58883238e-01
-3.23936909e-01 -1.22251475e+00 3.01031888e-01 4.49458539e-01
8.48983228e-01 -6.28854871e-01 8.39663327e-01 -1.90571304e-02
-1.44455481e+00 -1.88278019e-01 -3.60166430e-01 4.41203356e-01
-3.91531378e-01 6.06680930e-01 -1.49993920e+00 2.45095670e-01
9.74608839e-01 5.35578132e-01 -5.44852257e-01 1.17179704e+00
5.53304069e-02 1.19988906e+00 -4.04633075e-01 -2.40364894e-01
-1.29494980e-01 7.02403545e-01 6.08916819e-01 1.82326949e+00
5.71690202e-01 -5.21620084e-03 -2.53689438e-01 4.25498784e-01
-9.60852876e-02 4.23159897e-01 -4.62571949e-01 -3.18157315e-01
4.09579515e-01 1.48855746e+00 -4.84128833e-01 -2.06153944e-01
-2.99187630e-01 4.14265871e-01 -3.51916134e-01 -2.13100344e-01
-8.01043093e-01 -8.89112473e-01 5.27936220e-01 2.09731489e-01
2.58136392e-01 4.49229538e-01 9.46446583e-02 -6.46311879e-01
-2.76957273e-01 -1.09226537e+00 5.17839074e-01 -6.27111495e-01
-1.37674952e+00 1.01686704e+00 -3.27974737e-01 -1.09696937e+00
-3.73678416e-01 -4.72274899e-01 -8.04683447e-01 7.60343909e-01
-1.52103341e+00 -3.58680397e-01 -5.52217484e-01 6.62363768e-01
7.15902567e-01 -3.86253715e-01 1.12316144e+00 7.23617911e-01
-3.40215415e-01 6.67322099e-01 -2.27631584e-01 -5.27856275e-02
6.70089781e-01 -1.39200294e+00 5.75581729e-01 3.17864418e-01
4.40270990e-01 2.57713467e-01 4.16945606e-01 -1.23491615e-01
-1.13898826e+00 -1.58766699e+00 1.30070376e+00 -5.38855314e-01
5.46879351e-01 -1.06538467e-01 -1.17708874e+00 -4.68225777e-02
7.90101569e-03 3.90973777e-01 1.09180343e+00 -4.65625495e-01
-1.74421728e-01 -2.26020530e-01 -1.36742747e+00 -8.83175358e-02
7.76015401e-01 -8.73861969e-01 -7.65565217e-01 5.05461991e-01
9.37145591e-01 -2.55255520e-01 -1.13010371e+00 5.02428591e-01
6.84751332e-01 -8.25840831e-01 1.25585520e+00 -5.79854786e-01
1.36411101e-01 -1.23781756e-01 -4.18513387e-01 -1.21381295e+00
-5.15658438e-01 -6.42025590e-01 -1.37993902e-01 9.35683012e-01
3.95284981e-01 -6.04048193e-01 2.75461376e-01 -1.99860930e-01
-2.52739877e-01 -7.97647238e-01 -1.04339170e+00 -5.57957649e-01
-9.42091867e-02 -8.85807633e-01 4.95254248e-01 7.21597016e-01
-3.57784152e-01 5.20269454e-01 -1.99669972e-01 -1.42910138e-01
-1.97388921e-02 -4.99042928e-01 3.62650543e-01 -1.37308347e+00
-4.51715082e-01 -5.35000443e-01 -3.10189396e-01 -2.59363055e-01
-2.97652066e-01 -1.19134235e+00 2.89588094e-01 -1.52182245e+00
-1.28679648e-01 -9.79287252e-02 -1.23276520e+00 5.45649230e-01
-1.58363089e-01 4.75379169e-01 4.73523736e-02 5.82417250e-02
-2.24104360e-01 7.43806437e-02 9.71797228e-01 1.13274835e-01
-3.88074636e-01 5.74563265e-01 -5.32625079e-01 8.47581804e-01
1.19895041e+00 -8.87836695e-01 -5.39727390e-01 5.28555624e-02
-2.99704280e-02 -3.29860374e-02 3.02528620e-01 -1.48616469e+00
1.85452327e-01 2.15805158e-01 3.29129934e-01 -6.16853535e-01
3.45108032e-01 -6.67991281e-01 1.54778674e-01 7.57377982e-01
-7.80162275e-01 4.05670851e-02 3.07500035e-01 3.60807151e-01
-1.51299998e-01 -2.85424441e-01 1.01464045e+00 -1.29261643e-01
5.13331294e-02 8.53713974e-02 -5.82181454e-01 4.54230219e-01
4.19943333e-01 2.30550736e-01 -2.50912160e-01 -2.08826572e-01
-7.70102739e-01 -3.42211068e-01 -5.44035435e-01 5.23922026e-01
6.69628382e-01 -1.28265011e+00 -8.52755487e-01 5.04249871e-01
-1.69221908e-01 -2.56737679e-01 9.10810307e-02 9.63484466e-01
-4.64193851e-01 7.30133176e-01 -7.01169446e-02 -6.05655730e-01
-1.68462789e+00 1.24372773e-01 6.98710620e-01 -2.75191844e-01
-5.33398628e-01 1.29016840e+00 -1.76781490e-01 -5.47688246e-01
5.03373682e-01 -8.89053583e-01 -7.52156734e-01 1.80033728e-01
6.73760355e-01 7.98026919e-01 6.38289332e-01 -5.40211380e-01
-6.23233438e-01 6.96113706e-01 3.23918879e-01 3.61288153e-02
1.40045345e+00 3.60815942e-01 -8.07017609e-02 7.25244999e-01
1.31061876e+00 1.62728652e-01 6.28749561e-03 -1.70049090e-02
1.46002054e-01 -5.99755310e-02 1.49519965e-01 -9.41818655e-01
-1.08593512e+00 1.28599894e+00 1.12981510e+00 9.25757229e-01
1.36108375e+00 -2.49932960e-01 1.13768172e+00 7.00560868e-01
-3.57703984e-01 -9.64688361e-01 2.60130227e-01 5.82674026e-01
1.05543840e+00 -5.97347021e-01 -2.93375999e-01 1.23787662e-02
-2.41418093e-01 1.24038398e+00 3.93645048e-01 -3.54685001e-02
8.88676405e-01 1.92624599e-01 3.37959081e-01 -9.87133756e-02
-1.15393674e+00 -1.11184694e-01 8.92777026e-01 4.34123725e-01
9.15220082e-01 -2.15501003e-02 8.54516476e-02 8.60738754e-01
-7.91485548e-01 9.99572277e-02 1.75516367e-01 7.29703963e-01
-5.90291739e-01 -8.07752073e-01 -4.32414532e-01 8.20717514e-01
-1.35467303e+00 -8.61535594e-02 -5.80163777e-01 4.51614976e-01
3.71866137e-01 1.16479206e+00 -6.44595921e-02 -8.37016165e-01
7.04220533e-01 5.06037235e-01 2.40273148e-01 -8.25650632e-01
-1.21987844e+00 2.08813667e-01 1.34072065e-01 -4.66107816e-01
-3.55292112e-01 -4.11429584e-01 -1.32198632e+00 3.02870303e-01
-5.11496961e-01 1.70105234e-01 7.67631829e-01 5.65415740e-01
3.09006989e-01 1.34708714e+00 9.30888593e-01 -4.77002740e-01
-7.16000915e-01 -1.26975298e+00 -2.21895844e-01 1.74058408e-01
8.15620780e-01 -1.48416281e-01 -7.81503618e-01 6.47353232e-02] | [14.546807289123535, 3.912923574447632] |
728c8f28-bb58-490e-93d8-aabd1789c79d | toward-unsupervised-multi-object-discovery-in | 2007.02662 | null | https://arxiv.org/abs/2007.02662v2 | https://arxiv.org/pdf/2007.02662v2.pdf | Toward unsupervised, multi-object discovery in large-scale image collections | This paper addresses the problem of discovering the objects present in a collection of images without any supervision. We build on the optimization approach of Vo et al. (CVPR'19) with several key novelties: (1) We propose a novel saliency-based region proposal algorithm that achieves significantly higher overlap with ground-truth objects than other competitive methods. This procedure leverages off-the-shelf CNN features trained on classification tasks without any bounding box information, but is otherwise unsupervised. (2) We exploit the inherent hierarchical structure of proposals as an effective regularizer for the approach to object discovery of Vo et al., boosting its performance to significantly improve over the state of the art on several standard benchmarks. (3) We adopt a two-stage strategy to select promising proposals using small random sets of images before using the whole image collection to discover the objects it depicts, allowing us to tackle, for the first time (to the best of our knowledge), the discovery of multiple objects in each one of the pictures making up datasets with up to 20,000 images, an over five-fold increase compared to existing methods, and a first step toward true large-scale unsupervised image interpretation. | ['Patrick Pérez', 'Jean Ponce', 'Huy V. Vo'] | 2020-07-06 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/4433_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123680766.pdf | eccv-2020-8 | ['single-object-discovery', 'multi-object-colocalization', 'multi-object-discovery'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 4.97049838e-01 3.25912207e-01 -2.57665068e-01 -1.34068325e-01
-8.72668803e-01 -5.50910711e-01 6.95886075e-01 5.27479589e-01
-5.75459540e-01 6.22965276e-01 1.52653351e-01 2.00578012e-02
-1.13468260e-01 -5.47295392e-01 -9.27891314e-01 -5.85525870e-01
-9.93764102e-02 5.88223696e-01 9.01223958e-01 -1.94182009e-01
7.03617871e-01 6.34094477e-01 -2.01028681e+00 2.37898439e-01
6.70569956e-01 1.12551188e+00 4.19089645e-01 2.85248071e-01
6.27534240e-02 4.87765700e-01 -2.36655846e-01 -2.91774273e-01
3.11297745e-01 1.30265504e-02 -9.89125311e-01 4.78127807e-01
6.75974905e-01 -1.66974410e-01 -1.47371534e-02 9.89283264e-01
1.89623401e-01 1.40700057e-01 6.00609481e-01 -8.98528695e-01
-4.89495546e-01 6.12588644e-01 -8.60825121e-01 4.42949027e-01
7.82318786e-02 8.06943327e-02 1.37177849e+00 -1.12383103e+00
8.41747880e-01 9.08024967e-01 3.42617035e-01 1.80586755e-01
-1.40218914e+00 -3.09112012e-01 1.55054435e-01 2.83628315e-01
-1.54327667e+00 -5.20354271e-01 7.95531929e-01 -3.57781529e-01
6.87015533e-01 2.17836827e-01 6.94124579e-01 7.78445780e-01
-4.01867747e-01 1.11248696e+00 9.91303802e-01 -6.69933856e-01
4.29116428e-01 3.42172623e-01 1.40588254e-01 7.73503184e-01
2.91605890e-01 -1.62819698e-01 -4.29286510e-01 -1.41175926e-01
6.70295835e-01 -5.14650904e-03 -2.35615030e-01 -8.46529663e-01
-1.48475718e+00 7.60557175e-01 6.85710371e-01 4.68874097e-01
-4.71347064e-01 5.09212762e-02 2.35065848e-01 -3.04809153e-01
5.25478780e-01 7.55024016e-01 -3.92273426e-01 2.67505467e-01
-1.19024992e+00 4.38093245e-01 4.20538485e-01 7.02946246e-01
8.87281299e-01 -1.46448150e-01 -1.70756817e-01 6.28779173e-01
1.15624972e-01 1.39171362e-01 4.38784420e-01 -9.10824776e-01
2.68154174e-01 7.05264032e-01 2.65446007e-01 -1.03329432e+00
-3.73976767e-01 -7.61272073e-01 -5.60846865e-01 2.59021670e-01
4.03469473e-01 3.27429473e-01 -8.61157715e-01 1.36259007e+00
4.53277916e-01 3.78028639e-02 -2.32037097e-01 9.07225430e-01
5.54011285e-01 3.22053581e-01 -1.95876509e-01 -5.42687662e-02
1.45106566e+00 -1.07955062e+00 -1.79940075e-01 -1.84668541e-01
2.84572721e-01 -6.63806140e-01 9.03023422e-01 4.81388360e-01
-1.08716309e+00 -4.38686639e-01 -1.08189178e+00 -1.39067387e-02
-3.66611302e-01 3.59060258e-01 7.58382261e-01 2.98593998e-01
-1.24733484e+00 4.95016605e-01 -4.05340493e-01 -4.13297266e-01
8.78095865e-01 4.51060683e-01 -3.94665152e-01 1.57567265e-03
-5.37448883e-01 7.43270636e-01 6.27826154e-01 -6.78013489e-02
-9.29512143e-01 -6.66715503e-01 -6.98091328e-01 7.04833493e-02
9.87630785e-01 -3.84867162e-01 1.10582101e+00 -1.29179645e+00
-1.05993915e+00 1.19153798e+00 -2.48748332e-01 -7.09913492e-01
3.81724685e-01 -2.94112384e-01 4.86006886e-02 4.82030720e-01
3.64942461e-01 1.14730906e+00 9.14360762e-01 -1.54710925e+00
-8.49115610e-01 -2.69315362e-01 1.26546640e-02 7.89510384e-02
-2.84123808e-01 1.51226064e-02 -6.71783745e-01 -7.30889022e-01
2.19410896e-01 -7.39575565e-01 -4.57841098e-01 5.82468435e-02
-4.73653018e-01 -3.68321270e-01 5.60930192e-01 -2.73085684e-01
7.17280090e-01 -2.01796818e+00 2.23721087e-01 1.43708393e-01
5.64035416e-01 2.89389163e-01 -1.78048499e-02 2.14082554e-01
2.45330967e-02 8.75956845e-03 -4.83158886e-01 -4.18952584e-01
-1.55635417e-01 -7.74444593e-03 -5.79313278e-01 5.53818405e-01
4.55502331e-01 1.12520623e+00 -9.59182143e-01 -5.98122597e-01
2.27728456e-01 8.53976980e-02 -3.65521431e-01 -5.86754084e-02
-3.24894786e-01 3.04573894e-01 -4.27465379e-01 6.32278264e-01
4.32563424e-01 -5.10403633e-01 7.50320405e-03 -1.71072528e-01
-3.02973807e-01 1.54727131e-01 -1.12238061e+00 1.64353943e+00
2.07709819e-01 6.77668333e-01 -1.02677137e-01 -1.21994722e+00
8.44743431e-01 1.53197482e-01 6.26252711e-01 -5.60216188e-01
-2.32357979e-02 4.07713771e-01 -7.63201416e-02 -1.87582299e-01
6.51683748e-01 3.95795554e-02 9.79019478e-02 4.22664493e-01
1.79177955e-01 1.11840121e-01 3.28070581e-01 2.84737766e-01
9.60237324e-01 -4.90641519e-02 4.93692964e-01 -4.57041860e-01
3.81107450e-01 1.78044468e-01 3.62301648e-01 1.07778728e+00
-2.28904635e-01 8.74399006e-01 4.11729932e-01 -6.12631798e-01
-1.09484506e+00 -9.91398633e-01 -7.75255784e-02 1.04188609e+00
2.86106497e-01 -1.41674623e-01 -6.20786786e-01 -8.12592447e-01
-8.02876055e-02 4.74303573e-01 -9.68203962e-01 2.45520458e-01
-6.34480655e-01 -6.84772670e-01 2.57024020e-01 4.77225035e-01
4.31997627e-01 -1.23320985e+00 -9.13138211e-01 8.92991573e-02
-6.85553253e-02 -1.31321442e+00 -2.59413689e-01 2.61830926e-01
-9.41507399e-01 -1.11747563e+00 -7.16370761e-01 -7.64048755e-01
8.80562842e-01 5.77771246e-01 1.26799536e+00 1.82724535e-01
-5.87380171e-01 3.35858852e-01 -4.27076459e-01 -4.24536109e-01
4.27475683e-02 1.40462503e-01 -1.90382093e-01 3.02445889e-01
5.89276068e-02 -3.81498724e-01 -6.34762645e-01 2.82396764e-01
-8.33815753e-01 1.40032038e-01 1.02373183e+00 5.98019779e-01
8.46556246e-01 1.61999110e-02 5.11160433e-01 -7.97214210e-01
8.01033899e-03 -4.31433797e-01 -7.34560311e-01 7.98871592e-02
-5.98673105e-01 1.93899557e-01 2.43052930e-01 -2.19718441e-01
-5.99548221e-01 3.44520301e-01 2.36388266e-01 -4.08230007e-01
-3.09835404e-01 5.58523275e-02 6.38907775e-02 -2.25577384e-01
4.93084013e-01 3.90525520e-01 -2.23677948e-01 -6.00385427e-01
4.67922390e-01 2.10633501e-01 6.78908706e-01 -3.65884602e-01
8.05055976e-01 9.14187312e-01 5.47857098e-02 -6.29661322e-01
-1.04644179e+00 -9.05315399e-01 -1.02557731e+00 -6.62211999e-02
6.80053055e-01 -8.29810858e-01 -4.92637187e-01 1.11550905e-01
-1.05686533e+00 -1.23112291e-01 -3.85165542e-01 3.06440860e-01
-4.91845310e-01 3.89924526e-01 -6.91734999e-02 -7.58907437e-01
-2.51015157e-01 -1.05677497e+00 1.25368953e+00 1.18405722e-01
4.12060730e-02 -6.47462010e-01 -1.97739512e-01 5.00240982e-01
2.29585901e-01 2.68631101e-01 4.50675935e-01 -7.69353986e-01
-1.01969290e+00 -1.49531037e-01 -5.19539237e-01 5.33233397e-02
-1.34168699e-01 -1.43010959e-01 -1.06850362e+00 -1.78837910e-01
-7.53960684e-02 -1.92597955e-01 1.17785501e+00 3.94757777e-01
1.31722319e+00 -2.28940040e-01 -4.57260579e-01 2.37213105e-01
1.46649218e+00 -2.60250062e-01 4.68661219e-01 5.35694122e-01
5.74160635e-01 6.90060079e-01 6.63388073e-01 3.23655635e-01
3.18981618e-01 1.04677069e+00 7.81610429e-01 -2.63621867e-01
-2.62385141e-02 -7.15273544e-02 1.35019720e-01 2.05132708e-01
-1.20368816e-01 -1.48624061e-02 -8.05001974e-01 1.16152322e+00
-1.94686842e+00 -8.60530555e-01 -2.23117188e-01 2.34770703e+00
6.59438312e-01 2.48043373e-01 4.27913845e-01 6.90627918e-02
7.19278634e-01 2.61471063e-01 -3.95630509e-01 1.45514384e-01
-1.25311047e-01 1.70698091e-01 6.86569273e-01 1.24365300e-01
-1.37510288e+00 9.48452055e-01 6.27307320e+00 7.87355125e-01
-8.84097338e-01 5.99717237e-02 8.18424344e-01 -2.60064989e-01
-4.76885065e-02 1.57245487e-01 -9.05814350e-01 2.69022524e-01
4.53310221e-01 6.31048903e-02 2.92257369e-01 8.03189576e-01
1.34680107e-01 -4.49454516e-01 -1.00575423e+00 7.41579413e-01
4.41671997e-01 -1.58381820e+00 1.75450649e-02 3.04801166e-01
9.19246495e-01 3.04643035e-01 1.14458315e-01 -1.46078199e-01
2.32323229e-01 -9.56588149e-01 1.04807365e+00 3.42956185e-01
2.74690449e-01 -6.93081558e-01 7.04510152e-01 3.46548945e-01
-1.06786704e+00 -1.58383399e-01 -3.15453082e-01 1.64074734e-01
7.10232258e-02 7.98913777e-01 -9.53704834e-01 5.02637923e-01
7.98106551e-01 7.41948903e-01 -8.78645480e-01 1.43761885e+00
-1.07575879e-01 6.86418951e-01 -4.19411272e-01 2.02612597e-02
5.92041373e-01 1.81053549e-01 7.39786029e-01 1.06333756e+00
8.48774463e-02 -1.47012979e-01 2.41902605e-01 9.24769104e-01
-2.78927147e-01 2.81602919e-01 -5.20089209e-01 2.79756486e-01
2.25540265e-01 1.38575804e+00 -1.36097467e+00 -4.35507119e-01
-8.14770460e-02 7.93464124e-01 5.32176673e-01 1.89883232e-01
-5.61065435e-01 -1.62179500e-01 8.37116614e-02 3.28711987e-01
1.02735579e+00 -5.57211675e-02 -3.96826923e-01 -9.00366962e-01
3.82350355e-01 -6.52384639e-01 2.44498000e-01 -6.43986523e-01
-1.01764727e+00 4.01721954e-01 2.28555724e-02 -1.17679369e+00
6.99921371e-03 -5.36243677e-01 -5.33587039e-01 4.97786522e-01
-1.68709302e+00 -1.13540220e+00 -1.00025132e-01 2.22645536e-01
7.15639770e-01 -6.12619892e-02 4.50782329e-01 4.29082708e-03
-3.27771991e-01 2.02354431e-01 2.12819844e-01 1.18909292e-01
3.99236172e-01 -1.36806536e+00 3.22681904e-01 8.80886853e-01
6.68631911e-01 4.82544959e-01 6.01332664e-01 -3.34991664e-01
-1.14569807e+00 -1.05037177e+00 7.47488558e-01 -6.67304099e-01
6.44879997e-01 -4.91550773e-01 -9.44193184e-01 4.62741941e-01
1.52059108e-01 2.16066286e-01 3.54483306e-01 8.36888850e-02
-3.74813169e-01 -6.91597164e-02 -9.58626151e-01 5.27209222e-01
7.38054156e-01 -1.30719125e-01 -7.19650626e-01 3.58700812e-01
7.15725720e-01 -2.08842099e-01 -4.52380210e-01 4.13777918e-01
4.00576472e-01 -1.12926733e+00 1.18052924e+00 -2.74921685e-01
4.93224472e-01 -5.24515808e-01 -1.12669982e-01 -7.71229208e-01
-1.75749466e-01 -5.61245978e-01 -7.18293637e-02 1.06614280e+00
3.33653122e-01 -4.81303334e-01 7.59918749e-01 1.07680112e-01
-3.58310230e-02 -9.84403074e-01 -9.45478797e-01 -5.82838058e-01
-4.18461412e-01 -3.25616568e-01 4.03013885e-01 6.05533898e-01
-3.00467163e-01 2.02092230e-01 -3.06588858e-01 1.62866920e-01
8.62218082e-01 3.74267131e-01 6.85073912e-01 -1.33299482e+00
-3.58102351e-01 -6.37748361e-01 -5.89479089e-01 -1.05081296e+00
-2.95836013e-02 -7.87349164e-01 2.29257926e-01 -1.27224576e+00
3.90683770e-01 -5.01449406e-01 -5.34251809e-01 5.64908087e-01
-1.92755789e-01 7.26028085e-01 1.40296787e-01 4.36431259e-01
-1.10271561e+00 5.35708368e-01 1.03599787e+00 -1.41484320e-01
-5.53408004e-02 3.12085859e-02 -1.00801384e+00 7.24773943e-01
6.16215527e-01 -5.99571824e-01 -1.72920927e-01 -2.94162184e-01
-7.73705989e-02 -5.03822863e-01 8.81440401e-01 -1.04935014e+00
1.27994254e-01 -8.92876014e-02 4.08223152e-01 -7.16439068e-01
2.21717358e-01 -6.18422866e-01 -3.20798218e-01 3.21916729e-01
-4.06019837e-01 -2.57937819e-01 2.42868230e-01 7.95163333e-01
-9.01481211e-02 -3.75284493e-01 8.29205394e-01 -3.24227601e-01
-1.00315535e+00 2.67284717e-02 -1.54255956e-01 -3.80472317e-02
1.13314772e+00 -2.44044870e-01 -1.33749202e-01 -2.01504887e-03
-3.45567077e-01 2.06024535e-02 6.09949052e-01 3.59776884e-01
5.64680099e-01 -1.02415049e+00 -7.10085154e-01 1.50440391e-02
3.04119289e-01 3.39448243e-01 -9.86221656e-02 9.64551210e-01
-3.86248529e-01 5.49307406e-01 -1.93268061e-05 -8.13355267e-01
-1.26050186e+00 6.32897973e-01 -3.35290842e-02 -3.43203992e-01
-9.34987962e-01 7.97024429e-01 4.34661299e-01 -7.25291818e-02
6.00321777e-02 -1.72939941e-01 -3.67255539e-01 7.60360211e-02
4.96043891e-01 1.77021697e-01 9.68628079e-02 -7.74489284e-01
-3.92995447e-01 5.39721191e-01 -4.17356312e-01 -1.69591025e-01
1.55968297e+00 -1.05115138e-01 -2.01724112e-01 5.30810535e-01
9.86139834e-01 4.62274551e-02 -1.30081785e+00 -4.92529482e-01
3.41937125e-01 -6.13146365e-01 2.81221807e-01 -7.00382888e-01
-9.85679746e-01 5.97039819e-01 5.65091848e-01 1.15971155e-01
9.69937205e-01 4.59153533e-01 5.56100667e-01 4.84533429e-01
4.89382356e-01 -1.06065142e+00 3.08975399e-01 7.77964443e-02
8.03902924e-01 -1.48569751e+00 2.96479732e-01 -3.49460870e-01
-5.97027719e-01 9.62584198e-01 2.49036044e-01 -3.93510252e-01
2.99489647e-01 -1.77905038e-01 -3.32669884e-01 -4.57661569e-01
-6.96176827e-01 -6.31572306e-01 6.55632198e-01 4.33896691e-01
-8.54454562e-02 -4.78778109e-02 -1.35600522e-01 2.43456095e-01
3.46519828e-01 -2.81030256e-02 3.98705810e-01 8.31756592e-01
-8.43546510e-01 -8.58546197e-01 -4.10215497e-01 5.16715527e-01
-4.33743984e-01 -1.67574540e-01 -4.96302754e-01 7.40792871e-01
1.35301173e-01 7.12562799e-01 4.84765731e-02 1.13872685e-01
1.18055716e-01 -1.76334634e-01 2.72986472e-01 -7.44185030e-01
-4.35275525e-01 1.13094263e-01 -2.55383879e-01 -6.00302637e-01
-6.35196328e-01 -8.88706505e-01 -1.09437025e+00 4.19277817e-01
-4.44278240e-01 3.08126081e-02 6.47919655e-01 1.17136538e+00
4.41372484e-01 3.20257455e-01 6.44314706e-01 -1.31383836e+00
-2.09844872e-01 -6.85987055e-01 -4.30530936e-01 4.26878989e-01
2.80645221e-01 -9.63035941e-01 -3.54744136e-01 1.73117965e-01] | [9.404946327209473, 0.8826706409454346] |
f2fdb6b2-f109-4db8-a20e-1044a4c6bda4 | follow-us-and-become-famous-insights-and | 2301.06815 | null | https://arxiv.org/abs/2301.06815v1 | https://arxiv.org/pdf/2301.06815v1.pdf | Follow Us and Become Famous! Insights and Guidelines From Instagram Engagement Mechanisms | With 1.3 billion users, Instagram (IG) has also become a business tool. IG influencer marketing, expected to generate $33.25 billion in 2022, encourages companies and influencers to create trending content. Various methods have been proposed for predicting a post's popularity, i.e., how much engagement (e.g., Likes) it will generate. However, these methods are limited: first, they focus on forecasting the likes, ignoring the number of comments, which became crucial in 2021. Secondly, studies often use biased or limited data. Third, researchers focused on Deep Learning models to increase predictive performance, which are difficult to interpret. As a result, end-users can only estimate engagement after a post is created, which is inefficient and expensive. A better approach is to generate a post based on what people and IG like, e.g., by following guidelines. In this work, we uncover part of the underlying mechanisms driving IG engagement. To achieve this goal, we rely on statistical analysis and interpretable models rather than Deep Learning (black-box) approaches. We conduct extensive experiments using a worldwide dataset of 10 million posts created by 34K global influencers in nine different categories. With our simple yet powerful algorithms, we can predict engagement up to 94% of F1-Score, making us comparable and even superior to Deep Learning-based method. Furthermore, we propose a novel unsupervised algorithm for finding highly engaging topics on IG. Thanks to our interpretable approaches, we conclude by outlining guidelines for creating successful posts. | ['Ahmad-Reza Sadeghi', 'Mauro Conti', 'Marco Chilese', 'Pier Paolo Tricomi'] | 2023-01-17 | null | null | null | null | ['marketing'] | ['miscellaneous'] | [-3.37896198e-01 2.31520191e-01 -5.92553854e-01 -2.77590573e-01
-4.24060583e-01 -4.57790494e-01 8.89605165e-01 4.17311162e-01
-3.63147557e-01 7.67800152e-01 5.76270878e-01 -3.35183829e-01
7.77945593e-02 -1.19496500e+00 -7.77193427e-01 -2.95322537e-01
2.32210562e-01 3.91834021e-01 -3.41709256e-01 -1.98498473e-01
4.92264599e-01 -9.12624747e-02 -1.40874171e+00 1.69681802e-01
9.37982023e-01 7.43325651e-01 4.35712226e-02 3.03457975e-01
-3.70030373e-01 7.87179589e-01 -5.63703418e-01 -6.01999104e-01
-4.09523025e-02 -3.18823814e-01 -7.98640490e-01 -2.29372576e-01
8.39765817e-02 -6.00929022e-01 -3.41865510e-01 6.95186675e-01
1.18312821e-01 -2.09572390e-02 6.69552147e-01 -1.24186611e+00
-5.52810013e-01 1.19674838e+00 -6.43318236e-01 -1.03837019e-02
3.36413980e-01 -1.06789544e-01 1.44816756e+00 -7.11260200e-01
6.38813615e-01 1.04293346e+00 4.31048781e-01 1.39186949e-01
-1.01729608e+00 -7.18375802e-01 3.36732149e-01 -9.08077955e-02
-9.13721621e-01 -7.29063079e-02 8.96225393e-01 -5.82165062e-01
3.70300591e-01 2.87264079e-01 1.11970603e+00 1.44982672e+00
6.08754456e-02 1.08788240e+00 1.23117959e+00 -1.45270079e-01
7.42717311e-02 3.42743427e-01 2.32193813e-01 3.23157519e-01
3.03679436e-01 -3.59732628e-01 -5.49556613e-01 -2.17236504e-01
5.28968334e-01 4.25658673e-01 1.35576278e-01 3.11359227e-01
-1.08001566e+00 1.30652869e+00 6.46966517e-01 5.07083774e-01
-4.69197124e-01 1.75963551e-01 9.19372663e-02 1.73997208e-01
9.86333489e-01 7.13826835e-01 -2.88519740e-01 -5.62740862e-01
-1.04004812e+00 4.05288994e-01 9.60171044e-01 4.89295006e-01
9.65703547e-01 -3.21869940e-01 -8.48702788e-02 8.22001815e-01
6.25222266e-01 3.84031624e-01 3.27181607e-01 -5.85990608e-01
2.07710221e-01 9.00101900e-01 1.73488498e-01 -1.46448636e+00
-4.38321084e-01 -8.30889165e-01 -7.85999715e-01 -3.81961137e-01
4.96328950e-01 -4.80518878e-01 -5.76706946e-01 1.42881548e+00
-6.13734126e-02 -8.31703916e-02 -6.66703284e-01 6.49332762e-01
7.96256423e-01 6.97570682e-01 2.14049071e-01 -1.25934780e-01
1.09457922e+00 -8.10561240e-01 -7.11944461e-01 -1.43555373e-01
8.84957612e-01 -6.74164712e-01 1.22230685e+00 4.64657962e-01
-8.63599062e-01 -4.85713035e-01 -7.97698617e-01 1.64625391e-01
-5.55321217e-01 -1.26256794e-01 9.88952518e-01 5.48074424e-01
-8.47085178e-01 7.98019648e-01 -4.45301533e-01 -2.20977068e-01
6.94548666e-01 1.79733247e-01 4.19545770e-02 2.33228460e-01
-1.47152054e+00 5.02106607e-01 -1.00685120e-01 -2.35617653e-01
-6.25244856e-01 -6.21525824e-01 -1.53022632e-01 -3.20965238e-02
1.68023497e-01 -5.47011733e-01 1.21137452e+00 -1.24431479e+00
-1.30757940e+00 7.51100957e-01 -2.18365982e-01 -4.71059501e-01
5.31187594e-01 -6.43939435e-01 -2.50674576e-01 -3.07312399e-01
1.43222719e-01 4.38877225e-01 7.25377440e-01 -1.24556887e+00
-8.04725647e-01 -3.32108289e-01 4.19800460e-01 -2.50778869e-02
-9.67219949e-01 7.14463294e-02 -3.44681144e-01 -3.33844960e-01
-2.29999363e-01 -9.26088154e-01 -4.05887336e-01 -7.63509154e-01
-5.78467906e-01 -5.22210479e-01 5.33560991e-01 -6.29219174e-01
1.70089340e+00 -1.90404975e+00 -1.59530044e-01 4.23375607e-01
8.44150245e-01 2.31216960e-02 5.23042642e-02 6.25406504e-01
5.11417687e-01 8.31435144e-01 4.24455762e-01 -3.46069813e-01
1.99111044e-01 -2.28345349e-01 -4.63615030e-01 1.48976594e-01
-6.96780235e-02 1.02018726e+00 -9.67400610e-01 -1.40915930e-01
5.35360053e-02 4.48824823e-01 -6.78134561e-01 6.83481097e-02
-3.07358205e-01 4.79281127e-01 -8.82172287e-01 6.13303483e-01
4.81872082e-01 -6.21573448e-01 1.39633551e-01 8.97790715e-02
-3.14313531e-01 4.42241281e-01 -5.16040146e-01 8.82750034e-01
-7.63367474e-01 9.00029302e-01 -3.35205734e-01 -8.99272799e-01
1.15690613e+00 -1.25346869e-01 5.94138682e-01 -6.29242241e-01
3.84793788e-01 3.53678644e-01 -1.01961665e-01 -1.54928207e-01
6.35458052e-01 6.77877516e-02 -2.68240511e-01 6.77129388e-01
-4.96230453e-01 4.14100587e-01 1.02352239e-01 4.93325442e-01
1.18097281e+00 -2.55466968e-01 2.37529173e-01 -1.11133292e-01
-2.49977149e-02 -1.97738305e-01 1.80742264e-01 9.52401936e-01
-6.02467544e-02 2.39689693e-01 8.37707341e-01 -5.85363507e-01
-9.75645065e-01 -6.82058632e-01 -5.14347665e-02 1.25810730e+00
1.20885916e-01 -5.50498962e-01 -6.34082794e-01 -5.70356905e-01
7.00554624e-02 6.79354489e-01 -6.75273955e-01 2.23129794e-01
-5.51502705e-01 -7.23023415e-01 1.15045823e-01 2.00687945e-01
5.08122325e-01 -1.04992092e+00 -8.80762842e-03 4.24435824e-01
-5.09879768e-01 -8.47817838e-01 2.95617785e-02 -2.84670562e-01
-8.41700971e-01 -8.43191803e-01 -7.68690467e-01 -2.71976054e-01
5.45522630e-01 2.79672295e-01 1.32049489e+00 1.74531236e-01
3.14767927e-01 1.80276930e-02 -5.20269692e-01 -6.97480679e-01
-9.13095251e-02 5.80451965e-01 3.17077041e-02 2.14891270e-01
7.44145989e-01 -7.32893586e-01 -8.86188030e-01 3.03383738e-01
-5.89228094e-01 2.30952829e-01 8.27987552e-01 3.51643890e-01
2.41006240e-01 -1.35820851e-01 9.78988290e-01 -1.20839965e+00
9.53304827e-01 -1.03116679e+00 -8.13822076e-02 -3.69944423e-01
-8.55699122e-01 -4.15566891e-01 4.83907402e-01 -4.33756649e-01
-7.80745864e-01 -2.89948314e-01 -1.97279096e-01 1.36649117e-01
4.75498810e-02 8.55756164e-01 2.53539741e-01 4.25753176e-01
5.99918664e-01 -5.99427894e-02 -1.44818127e-01 -4.76794451e-01
3.47390890e-01 1.05047476e+00 -3.32432419e-01 -2.85280198e-01
8.64610434e-01 4.81519848e-01 -4.58036214e-01 -7.80616105e-01
-1.24364805e+00 -5.79056680e-01 -1.87766314e-01 -7.34906673e-01
6.24717712e-01 -1.05328131e+00 -8.99628460e-01 5.88450313e-01
-1.00092804e+00 -3.26384872e-01 1.06203854e-01 4.46144044e-01
-1.41448006e-01 -2.21326612e-02 -6.15783215e-01 -7.83434033e-01
-3.97586077e-01 -6.13502204e-01 6.49878502e-01 4.18658972e-01
-6.60903513e-01 -1.04520977e+00 1.39979094e-01 8.24034035e-01
7.60890543e-01 4.90663886e-01 4.47899252e-01 -8.60993862e-01
-6.50928676e-01 -4.29394662e-01 -2.86546677e-01 7.39252567e-02
2.35200161e-03 1.93380490e-01 -7.79598296e-01 -9.11158547e-02
-3.67814958e-01 -2.21084446e-01 7.41006553e-01 5.80277562e-01
1.38052607e+00 -6.42691314e-01 -6.29635930e-01 2.91091967e-02
1.02377856e+00 -7.00853989e-02 7.58655727e-01 5.89407504e-01
7.47394741e-01 6.33311808e-01 7.86030293e-01 8.61604035e-01
5.21290243e-01 4.71082181e-01 5.31691730e-01 -2.33822659e-01
5.22990584e-01 -6.83583736e-01 4.69364315e-01 9.99369621e-01
-5.37213802e-01 -3.89267266e-01 -8.20733130e-01 5.55515170e-01
-1.84294879e+00 -8.74078512e-01 -5.67407846e-01 1.88667154e+00
6.53041780e-01 4.74514812e-01 5.05658388e-01 -3.77163328e-02
6.27854764e-01 4.11939681e-01 -3.71066928e-01 -4.39268976e-01
8.24676454e-02 1.67797193e-01 4.07925069e-01 3.39497119e-01
-8.49932134e-01 8.09452951e-01 5.47763824e+00 8.14974487e-01
-1.03645813e+00 1.04110152e-01 1.16577768e+00 -2.43415058e-01
-7.01526284e-01 2.57935952e-02 -1.03371084e+00 7.49594390e-01
1.04285514e+00 -3.10747236e-01 2.49058872e-01 9.40115988e-01
6.43831730e-01 6.37817681e-02 -7.87614644e-01 7.26026952e-01
-1.22598805e-01 -1.41221499e+00 -8.21026787e-03 5.59776604e-01
1.06320620e+00 1.20918095e-01 1.23884715e-01 6.01962328e-01
3.19556415e-01 -1.02432954e+00 4.38494712e-01 7.05608070e-01
3.55570942e-01 -8.02186668e-01 8.25189054e-01 4.18235272e-01
-7.33883917e-01 -1.51258871e-01 -1.83035284e-01 -6.31974101e-01
2.35187262e-01 1.38693309e+00 -6.69446707e-01 2.25451872e-01
5.80350399e-01 9.88932908e-01 -3.57820600e-01 6.84361160e-01
-3.68522108e-01 1.05395806e+00 -2.42782757e-01 -7.40102828e-01
4.03689265e-01 -1.83353439e-01 3.31131548e-01 1.06425941e+00
5.08202136e-01 -4.94974107e-02 1.44318240e-02 8.07760596e-01
-5.37646055e-01 4.57679123e-01 -6.65043175e-01 -5.75682819e-01
4.40001696e-01 1.58099842e+00 -5.46315312e-01 -3.02724630e-01
-3.97032142e-01 4.28167671e-01 3.59682411e-01 2.13435352e-01
-1.02683663e+00 -2.53945947e-01 4.92100954e-01 9.39838052e-01
-1.96232066e-01 -2.59498390e-03 -4.46056634e-01 -9.43534732e-01
-2.47584477e-01 -8.90088618e-01 -9.53789949e-02 -4.23069924e-01
-1.32767713e+00 3.52391213e-01 -2.93231279e-01 -1.00135124e+00
-5.19275554e-02 -3.63113284e-01 -8.08761954e-01 6.20840788e-01
-1.30393934e+00 -1.09412658e+00 -5.04003942e-01 8.92770886e-02
3.35056275e-01 9.47795808e-02 2.98280805e-01 4.56923246e-01
-4.42487031e-01 3.96517724e-01 1.37894556e-01 7.85875469e-02
6.23586655e-01 -1.13077986e+00 3.22539568e-01 2.22509563e-01
1.83714733e-01 6.66171610e-01 6.05625689e-01 -6.15697145e-01
-1.08372080e+00 -9.29626286e-01 1.38868034e+00 -5.47349095e-01
1.03644705e+00 -2.97475487e-01 -5.33202350e-01 6.33075535e-01
1.65004998e-01 -7.03820944e-01 1.05074716e+00 8.92410338e-01
1.19675472e-01 -4.53846417e-02 -6.54867709e-01 8.33756566e-01
9.78838921e-01 -2.31786266e-01 -1.25032291e-01 4.33656216e-01
6.18663669e-01 8.60467181e-02 -9.30099607e-01 1.52687386e-01
8.44680429e-01 -1.06992328e+00 4.85737532e-01 -4.80300099e-01
1.19298899e+00 2.16237068e-01 1.63086176e-01 -1.27921498e+00
-4.94697601e-01 -6.61503911e-01 -3.80764991e-01 1.36492670e+00
6.40165508e-01 -6.88782811e-01 1.02683115e+00 6.67458832e-01
9.33690518e-02 -1.24887562e+00 -2.86879182e-01 -1.73453748e-01
-4.50297412e-05 -4.35238391e-01 5.32403529e-01 1.18177223e+00
6.07701624e-03 5.05627215e-01 -6.78913474e-01 -4.32903528e-01
2.73856848e-01 9.99577790e-02 1.25881803e+00 -1.54206872e+00
-3.13662976e-01 -6.19207263e-01 1.66728497e-02 -1.30365956e+00
-7.47774467e-02 -6.53158903e-01 -4.68826711e-01 -1.66901565e+00
5.75661182e-01 -6.27349079e-01 -4.86487001e-01 3.81389409e-01
-1.65785000e-01 3.14781576e-01 -3.16934660e-02 4.75682825e-01
-6.10774338e-01 3.69154602e-01 1.50596297e+00 -1.98316231e-01
-4.12543327e-01 3.28613520e-01 -1.33350122e+00 9.16522264e-01
1.23852730e+00 -3.31315428e-01 -1.73807621e-01 1.20480610e-02
1.10824955e+00 -4.03999597e-01 1.31936520e-01 -6.02683306e-01
-2.60397196e-01 -2.21025124e-01 2.77920663e-01 -6.36049986e-01
5.03945947e-02 -2.70903915e-01 9.06435251e-02 3.50566715e-01
-4.84066129e-01 -3.01609159e-01 -2.78149575e-01 6.36877179e-01
-3.28810871e-01 -7.20711127e-02 3.29055339e-01 -2.79003624e-02
-1.54806331e-01 4.81233060e-01 -5.42498589e-01 -3.61075712e-04
7.32056022e-01 -4.91469130e-02 -3.92954856e-01 -1.00040793e+00
-6.70897961e-01 1.45798862e-01 2.17324078e-01 5.87749481e-01
2.46067747e-01 -1.30781925e+00 -8.89742136e-01 -3.43602419e-01
-3.57677625e-03 -2.13174805e-01 1.93766072e-01 9.31769669e-01
-2.73683131e-01 3.93924057e-01 4.11275513e-02 -2.87105769e-01
-1.08175039e+00 6.88254386e-02 -2.23956361e-01 -6.94027007e-01
-1.38337448e-01 6.03650630e-01 2.57819444e-01 -2.49956727e-01
-5.89969456e-02 -8.50638151e-02 -6.68047190e-01 6.09286368e-01
4.99282539e-01 5.06170571e-01 -3.54414463e-01 -3.90357465e-01
-5.73220551e-02 1.82200834e-01 -3.85313511e-01 1.95064232e-01
1.45993674e+00 -9.93789583e-02 -1.15117371e-01 6.07225955e-01
1.27889454e+00 2.33504772e-01 -7.89644897e-01 -9.65897143e-02
-1.50610223e-01 -5.91854036e-01 1.90978482e-01 -5.43752670e-01
-1.01225162e+00 7.54955351e-01 1.40233964e-01 9.10006225e-01
6.89898491e-01 2.11833283e-01 1.01334965e+00 1.94954515e-01
1.81875244e-01 -1.29746222e+00 4.18808550e-01 4.01443332e-01
7.92341650e-01 -1.40995061e+00 7.23617747e-02 -1.57103255e-01
-4.07898009e-01 9.12416458e-01 5.29895604e-01 -2.78081931e-02
8.64524841e-01 -2.58167326e-01 -9.98736843e-02 -3.83809417e-01
-6.22135103e-01 1.70089621e-02 1.61140516e-01 5.10451496e-02
8.47814798e-01 4.92799073e-01 -8.46785903e-01 8.19849551e-01
-4.53831971e-01 -9.91146564e-02 6.00825965e-01 3.82036716e-01
-6.88833058e-01 -1.05197167e+00 5.20332567e-02 1.13097751e+00
-9.40718412e-01 -4.75598164e-02 -6.40365660e-01 7.07365513e-01
-1.21851293e-02 1.17713833e+00 3.63366154e-04 -6.76760316e-01
7.08340248e-03 -2.61637270e-01 -1.81201681e-01 -3.90792996e-01
-5.20353496e-01 3.59815720e-04 4.08961236e-01 -3.69536638e-01
-5.18867731e-01 -5.78503907e-01 -9.12735760e-01 -9.48741078e-01
-4.41829532e-01 2.02604920e-01 7.90067673e-01 9.02026474e-01
4.05674428e-01 4.07385617e-01 1.19122016e+00 -7.43464112e-01
-2.17406064e-01 -1.19513130e+00 -5.82754910e-01 5.19375563e-01
-2.09761962e-01 -5.06690264e-01 -7.52510190e-01 -3.11671317e-01] | [10.18502426147461, 6.530834197998047] |
c96c6d86-61ba-464c-ac40-4da90164eef3 | a-geometrical-imaging-of-the-real-gap-between | 1701.05114 | null | http://arxiv.org/abs/1701.05114v2 | http://arxiv.org/pdf/1701.05114v2.pdf | A geometrical imaging of the real gap between economies of China and the United States | GDP of China is about 11 trillion dollars and GDP of the United States is
about 18 trillion dollars. Suppose that we know for the coming years, economy
of the US will experience a real growth rate equal to \%3 and economy of China
will experience a real growth as of \%6. Now, the question is how long does it
take for economy of China to catch the economy of the United States. The early
impression is that the desired time is the answer of the equation
$11\times1.06^X=18\times1.03^X$. The correct answer however is quite different.
GDP is not a simple number and the gap between two countries can not be
addressed simply through their sizes. It is rather a geometrical object.
Countries pass different paths in the space of production. The gaps between GDP
of different countries depend on the path that each country passes through and
local metric. To address distance between economies of China and of the US we
need to know their utility preferences and the path that China passes to reach
the US size. The true gap then can be found if we calculate local metric along
this path. It resembles impressions about measurements in the General Theory of
Relativity. Path dependency of aggregate indexes is widely discussed in the
Index Number Theory. Our aim is to stick to the geometrical view presented in
the General Relativity to provide a visual understanding of the matter. We show
that different elements in the general relativity have their own counterparts
in economics. We claim that national agencies who provide aggregate data
resemble falling observers into a curved space time. It is while the World Bank
or international organizations are outside observers. The vision provided here,
leaves readers with a clear conclusion. If China keeps its growth rate, then
the economy of China should catch the economy of the United States sooner than
what we expect. | [] | 2019-04-24 | null | null | null | null | ['geometrical-view'] | ['computer-vision'] | [-7.18542099e-01 3.03005546e-01 -1.79528937e-01 1.72802750e-02
2.41389852e-02 -7.48597503e-01 6.20099664e-01 -3.52287382e-01
-5.57815254e-01 9.06023204e-01 3.31479132e-01 -1.10486269e+00
-1.18354581e-01 -1.14586353e+00 -1.63881376e-01 -6.41765296e-01
-1.28434934e-02 4.53157037e-01 -2.64252815e-02 -7.91667759e-01
4.37529534e-01 5.00222325e-01 -9.68200386e-01 -7.41775990e-01
8.15698564e-01 7.06524730e-01 1.38357475e-01 6.89022958e-01
-3.24031189e-02 4.64632094e-01 -4.66744512e-01 -3.86035800e-01
5.08647084e-01 -6.50589108e-01 -9.32114482e-01 -2.97585279e-01
-2.18920469e-01 -2.29125977e-01 -2.01151550e-01 1.12753189e+00
4.72440533e-02 8.41622278e-02 6.87937558e-01 -8.15936267e-01
-1.07332206e+00 7.02124357e-01 -5.78116417e-01 3.20773244e-01
5.13684809e-01 -1.82378381e-01 9.40121710e-01 -5.09049952e-01
6.38047278e-01 1.16949368e+00 3.44179541e-01 1.96009666e-01
-7.73365021e-01 -4.58957881e-01 5.99038377e-02 -2.21140906e-01
-1.25205922e+00 3.70756462e-02 3.48166138e-01 -4.27118450e-01
7.07445979e-01 6.05035961e-01 6.23205900e-01 3.50055695e-01
6.05184376e-01 -4.41485047e-01 1.07075071e+00 -5.42408109e-01
-1.70706457e-03 2.66058624e-01 -2.55638003e-01 5.19346118e-01
6.95834637e-01 3.63566965e-01 4.91735190e-02 2.47848466e-01
1.09314418e+00 -1.45268157e-01 -3.56352031e-01 1.11456946e-01
-1.55753338e+00 7.91228175e-01 5.50753832e-01 1.00278533e+00
-1.41600206e-01 1.39499128e-01 -2.52617262e-02 5.06890595e-01
2.49664918e-01 6.23452187e-01 -6.22069001e-01 -1.74628481e-01
-5.27757406e-01 2.85620689e-01 8.25727940e-01 6.99623108e-01
5.10288894e-01 -3.71076982e-03 9.30660605e-01 6.95871795e-03
3.26548338e-01 9.74032223e-01 4.99211282e-01 -1.16793954e+00
4.58654642e-01 6.36813760e-01 4.58439857e-01 -1.19927084e+00
-7.77202368e-01 -5.28371692e-01 -6.33606791e-01 6.42657638e-01
1.15120125e+00 -4.62928295e-01 -3.46768321e-03 1.73069632e+00
-2.60456968e-02 -4.49516118e-01 2.06340268e-01 8.97397101e-01
3.33088249e-01 8.40909421e-01 -1.52957246e-01 -4.29787576e-01
1.55135262e+00 -1.88242808e-01 -5.19917071e-01 2.06164569e-01
1.20455265e+00 -1.01228905e+00 1.04064989e+00 2.17501760e-01
-1.31680751e+00 -2.30640769e-01 -1.05320168e+00 1.89416140e-01
-5.29098451e-01 -1.96263820e-01 5.16368210e-01 7.99099088e-01
-9.31628525e-01 7.64563739e-01 -8.78023446e-01 -6.95675373e-01
-4.48980987e-01 1.69234425e-01 -3.02693546e-01 6.12366736e-01
-1.03146827e+00 1.64763737e+00 3.61814320e-01 -1.45488665e-01
6.31887764e-02 -2.97622025e-01 -7.38426983e-01 4.66855802e-03
-6.86964095e-02 -5.42250574e-01 8.89294982e-01 -7.51449466e-01
-1.34288919e+00 7.73152471e-01 4.73860167e-02 -1.57740533e-01
3.46788734e-01 2.45238096e-01 -9.26012218e-01 -1.03916451e-01
2.28177011e-01 -2.79045873e-03 -4.02697146e-01 -1.01236010e+00
-9.96114671e-01 -7.09300339e-01 3.20036709e-01 2.63767987e-01
1.38430391e-02 5.65936148e-01 3.90802294e-01 -2.80551225e-01
6.40838683e-01 -1.01827765e+00 -2.17569962e-01 -5.00084698e-01
-3.69483307e-02 -2.69682705e-01 6.84761763e-01 -7.11232185e-01
1.10363078e+00 -1.67718863e+00 -2.16745377e-01 1.88947290e-01
1.70306981e-01 -2.94170469e-01 3.80951345e-01 1.91968143e-01
-3.08901817e-01 3.79716963e-01 -9.04261842e-02 3.77981097e-01
2.83273101e-01 9.27428901e-02 -1.44735232e-01 1.00647020e+00
-2.51759827e-01 4.45490420e-01 -1.03817344e+00 -2.46724173e-01
-6.05625380e-03 2.85816729e-01 -3.89203429e-01 -3.44027847e-01
6.14686847e-01 4.25805330e-01 -3.46511692e-01 4.17619467e-01
9.59929168e-01 1.68010611e-02 1.61195934e-01 1.93128034e-01
-9.52261508e-01 3.46663862e-01 -1.13419926e+00 1.06628382e+00
-4.07893211e-01 4.51228052e-01 -1.32372633e-01 -1.05597949e+00
1.04248202e+00 6.15458727e-01 3.49212229e-01 -1.18593764e+00
3.02284390e-01 8.62855315e-01 5.33135712e-01 -2.29765818e-01
6.78607702e-01 -8.97961318e-01 -3.18547666e-01 7.17119694e-01
-3.54003489e-01 -3.81677538e-01 2.59567022e-01 1.43038630e-01
6.01528645e-01 4.22973074e-02 5.68019986e-01 -9.51995432e-01
5.15791297e-01 -1.16340086e-01 6.16009057e-01 -8.07119161e-03
-2.58508295e-01 2.07084715e-01 7.39559412e-01 -8.10082495e-01
-1.31203115e+00 -1.39462256e+00 -4.26622897e-01 5.80215156e-01
3.75832438e-01 -1.02947518e-01 -6.17895603e-01 -2.91696250e-01
-4.40600157e-01 1.03735113e+00 -4.19051588e-01 1.94424734e-01
-6.71255171e-01 -9.39963520e-01 -6.81143254e-02 3.47644389e-02
4.98196691e-01 -7.36690402e-01 -9.98381019e-01 -1.11090384e-01
-2.04546884e-01 -8.99066806e-01 -1.37588620e-01 -2.33552411e-01
-8.18505347e-01 -9.23554361e-01 -9.10518944e-01 -3.51155460e-01
6.36581957e-01 1.24760158e-01 1.10463750e+00 8.44706446e-02
4.10881728e-01 1.20442115e-01 -2.08121702e-01 -6.46523237e-01
-5.39220572e-01 -1.85199067e-01 5.92356563e-01 -6.21104777e-01
3.58453780e-01 -6.62163377e-01 -8.65047336e-01 5.45178592e-01
-3.85342687e-01 -3.56462538e-01 -9.17186663e-02 1.22342907e-01
-6.65812567e-02 3.88295352e-01 4.68451530e-01 -2.16539353e-01
1.88084006e-01 -5.61793208e-01 -8.10954332e-01 -7.07719475e-02
-5.29586554e-01 -2.27156937e-01 6.05554104e-01 1.15353994e-01
-1.12446451e+00 -6.68020189e-01 2.68867880e-01 8.48833263e-01
8.04019123e-02 4.95402485e-01 2.58332416e-02 1.42218664e-01
6.80135369e-01 -3.16156745e-01 -2.36981839e-01 -4.01476830e-01
2.42940053e-01 5.61899006e-01 6.29643559e-01 -6.49037898e-01
9.73388255e-01 6.82471335e-01 3.49765003e-01 -6.56213284e-01
-3.95255417e-01 -1.36388252e-02 -7.41634727e-01 -1.97165772e-01
9.21283126e-01 -8.00419867e-01 -9.17962849e-01 3.84749584e-02
-1.16887176e+00 -7.87836611e-02 -5.53702652e-01 1.11239433e+00
-6.60499692e-01 2.08976641e-01 -1.46926150e-01 -8.88257086e-01
2.00875208e-01 -9.81633186e-01 3.26531738e-01 3.56355220e-01
-2.31055096e-01 -1.59299660e+00 1.46606758e-01 1.88865233e-05
4.33327347e-01 4.39887077e-01 6.76691234e-01 -2.40990445e-01
-2.24152818e-01 -1.01163886e-01 -3.55292588e-01 4.94594388e-02
3.77938449e-01 2.15307668e-01 -3.32810432e-01 -2.36512944e-01
2.79063582e-01 4.50871766e-01 3.31320167e-01 5.13789058e-01
3.12768430e-01 -4.06772196e-01 -1.71977460e-01 4.76045012e-01
1.84264076e+00 6.18547857e-01 7.41490901e-01 8.48097563e-01
2.81476229e-01 7.36282468e-01 4.88410324e-01 3.24762762e-01
7.31858730e-01 6.43554926e-01 4.16119576e-01 -2.80252278e-01
1.95895687e-01 1.77927911e-01 3.14974070e-01 1.10508704e+00
-1.05590570e+00 1.81508407e-01 -1.14989567e+00 5.31787097e-01
-1.28975797e+00 -1.08583856e+00 -5.02805412e-01 2.57068825e+00
4.25677210e-01 2.75653303e-01 3.88166726e-01 1.30110279e-01
4.28700715e-01 -2.78873652e-01 2.01410159e-01 -1.01148653e+00
-1.91214472e-01 -2.22924039e-01 1.00177431e+00 9.41453397e-01
-5.14334559e-01 4.71142352e-01 7.41504145e+00 1.64956182e-01
-1.26336527e+00 -8.85271560e-03 6.03841007e-01 9.19677988e-02
-4.20627058e-01 6.58601642e-01 -5.34613669e-01 5.86822331e-01
1.14244771e+00 -1.10453844e+00 2.60344684e-01 3.68675917e-01
6.02368414e-01 -4.61073816e-01 -6.16491139e-01 6.25057399e-01
-4.73665267e-01 -9.14778829e-01 -2.21459106e-01 7.08645642e-01
7.61779726e-01 2.95015164e-02 1.90872923e-01 -1.27197415e-01
6.20531201e-01 -9.47406173e-01 9.60990250e-01 6.29812837e-01
7.13479459e-01 -9.21084404e-01 9.61594760e-01 4.48415101e-01
-1.26588643e+00 -2.29009137e-01 -7.24447846e-01 -8.87520373e-01
5.41535258e-01 6.01191163e-01 -5.22370338e-01 7.92113960e-01
4.43789244e-01 7.93270469e-02 -1.92429081e-01 6.68998480e-01
2.06100866e-01 2.06655964e-01 -4.33658302e-01 2.03523815e-01
5.86133301e-01 -8.63586485e-01 5.82995951e-01 8.21506023e-01
1.20416987e+00 5.83681107e-01 -5.74405551e-01 8.77461970e-01
2.04971150e-01 3.52919668e-01 -8.75033796e-01 1.07034661e-01
1.65986791e-01 8.49986196e-01 -9.17524338e-01 -2.20895141e-01
-6.86959386e-01 3.01576555e-01 -3.26561719e-01 2.10846841e-01
-7.68521607e-01 -5.21084487e-01 6.90067708e-01 3.10139239e-01
-3.05172086e-01 -5.70133150e-01 -4.02853996e-01 -1.08495510e+00
-1.82156786e-01 -2.83021748e-01 1.05551489e-01 -5.77048719e-01
-5.35986364e-01 3.65165740e-01 5.06726950e-02 -1.24030054e+00
-1.94216996e-01 -8.08817267e-01 -8.37182939e-01 1.08258569e+00
-9.80264723e-01 -7.88365364e-01 3.15784395e-01 1.88343033e-01
2.34992057e-02 -2.21580774e-01 7.70843446e-01 -5.78670613e-02
-3.81249100e-01 2.10732222e-01 4.34549421e-01 3.74185741e-02
1.70906231e-01 -1.35575235e+00 4.34396684e-01 1.03685915e+00
-1.81716532e-01 6.44084632e-01 1.23971379e+00 -3.36306721e-01
-8.27494383e-01 -2.62518257e-01 1.56167877e+00 -6.63317859e-01
1.03415632e+00 1.40720814e-01 -3.82048249e-01 8.83779585e-01
3.70534658e-01 -1.20142274e-01 2.71255910e-01 1.03384435e-01
-5.90256229e-02 -3.27115878e-02 -1.14062929e+00 6.48053229e-01
8.05555582e-01 -2.76801467e-01 -6.49558485e-01 1.57683596e-01
6.85546160e-01 -6.41393615e-03 -1.09315491e+00 3.93001847e-02
7.79457629e-01 -1.30631006e+00 9.14311469e-01 -3.16266418e-01
2.26053521e-01 -2.03357667e-01 -4.56181377e-01 -1.62304473e+00
-4.38620627e-01 -4.86599505e-01 7.07756281e-01 9.46203589e-01
5.69414377e-01 -1.31881845e+00 3.20585221e-01 6.43900990e-01
9.60304439e-02 -3.57457340e-01 -1.21769822e+00 -1.10806310e+00
1.11182439e+00 -1.56542405e-01 8.87357056e-01 1.25700581e+00
7.91696310e-01 3.76320817e-02 1.36830926e-01 8.03318918e-02
7.24524796e-01 -1.87828451e-01 4.98413712e-01 -1.37071943e+00
1.02810942e-01 -7.45486498e-01 -5.37283301e-01 -6.92124784e-01
-2.61970878e-01 -5.99324167e-01 -8.20741773e-01 -1.60163748e+00
-1.77923962e-02 -6.58618748e-01 -1.51697144e-01 -2.69471347e-01
3.53309691e-01 1.72495708e-01 5.26667058e-01 1.32439330e-01
1.22662902e-01 1.75380912e-02 1.71515715e+00 3.52285117e-01
-1.17726177e-01 -1.62945390e-01 -1.09522259e+00 1.16652882e+00
9.37670648e-01 -1.57724291e-01 -2.66539425e-01 -2.41589442e-01
8.69324565e-01 3.07720035e-01 1.60624862e-01 -1.06515968e+00
-5.58449477e-02 -5.87767303e-01 1.76684916e-01 -3.59058589e-01
2.12439895e-02 -9.00923550e-01 4.11562562e-01 6.68008089e-01
3.53519261e-01 4.40582603e-01 -1.49821013e-01 -1.74094886e-01
1.54381230e-01 -1.96961924e-01 8.91448081e-01 -5.60440361e-01
-1.47665083e-01 -5.51663004e-02 -3.53304207e-01 -1.55039832e-01
9.79220450e-01 -4.66046154e-01 -6.23716593e-01 -4.38601345e-01
-6.53875709e-01 -5.57036363e-02 9.75754321e-01 -1.25326738e-02
7.32029155e-02 -1.47800934e+00 -8.71759057e-01 -3.48369569e-01
-3.19150388e-01 -4.56017345e-01 -6.13317378e-02 1.19428599e+00
-9.50957000e-01 8.15178990e-01 -4.30249333e-01 -1.82578675e-02
-7.44119346e-01 6.33119702e-01 6.50372803e-01 -1.33624241e-01
-6.22646809e-01 4.97692376e-01 5.92881739e-01 -4.49709415e-01
-6.49178624e-01 -6.31288886e-01 -2.42989346e-01 5.72965071e-02
6.57644033e-01 6.85067654e-01 -3.78678799e-01 -1.29957855e+00
-3.16537917e-01 1.08381295e+00 4.42690700e-01 -3.94536555e-01
1.19334173e+00 -7.08817363e-01 -3.93067628e-01 6.65618598e-01
1.18033028e+00 4.23644871e-01 -6.30874455e-01 2.07805321e-01
-1.18355125e-01 -5.60437500e-01 -1.62105635e-01 -6.17072940e-01
-8.21136892e-01 5.97401083e-01 4.40168977e-01 7.93525279e-01
8.53938520e-01 1.94427609e-01 7.37589240e-01 -3.78379710e-02
6.61313295e-01 -1.09902775e+00 -4.80235159e-01 5.15180469e-01
9.44915771e-01 -9.97531354e-01 1.73215955e-01 -1.08576283e-01
-2.04186976e-01 1.07340944e+00 7.47498497e-02 -3.19506943e-01
8.14620435e-01 1.04714073e-02 1.01196788e-01 -2.45986953e-01
-3.29770237e-01 -8.36682767e-02 -1.31530285e-01 5.15808165e-01
7.88011312e-01 7.15522110e-01 -1.07360435e+00 2.92140156e-01
-1.05290616e+00 -3.39232296e-01 9.97651041e-01 5.59780419e-01
-9.92342472e-01 -9.65136826e-01 -9.87213552e-01 -1.02707259e-01
-5.47264993e-01 1.37339875e-01 1.34688362e-01 1.18281269e+00
5.63305795e-01 8.84931564e-01 5.32326221e-01 -5.49740857e-04
2.96743035e-01 -2.06475452e-01 5.48292160e-01 -3.84581871e-02
5.54440618e-02 -1.35335296e-01 1.00925416e-01 -3.47624511e-01
-5.64812005e-01 -7.57451594e-01 -1.70223022e+00 -1.15645278e+00
-2.16932222e-01 5.05178094e-01 8.63826156e-01 9.15132642e-01
-5.85544467e-01 6.41571209e-02 7.84323335e-01 -8.01904917e-01
-3.35847199e-01 -6.76673114e-01 -1.23241103e+00 2.82716118e-02
3.47558111e-01 -6.54513419e-01 -8.82604957e-01 -1.57860145e-01] | [5.7362895011901855, 4.111849308013916] |
fb03d5d8-afab-4058-8c8e-7fa9428d3a40 | machine-learning-in-and-out-of-equilibrium | 2306.03521 | null | https://arxiv.org/abs/2306.03521v1 | https://arxiv.org/pdf/2306.03521v1.pdf | Machine learning in and out of equilibrium | The algorithms used to train neural networks, like stochastic gradient descent (SGD), have close parallels to natural processes that navigate a high-dimensional parameter space -- for example protein folding or evolution. Our study uses a Fokker-Planck approach, adapted from statistical physics, to explore these parallels in a single, unified framework. We focus in particular on the stationary state of the system in the long-time limit, which in conventional SGD is out of equilibrium, exhibiting persistent currents in the space of network parameters. As in its physical analogues, the current is associated with an entropy production rate for any given training trajectory. The stationary distribution of these rates obeys the integral and detailed fluctuation theorems -- nonequilibrium generalizations of the second law of thermodynamics. We validate these relations in two numerical examples, a nonlinear regression network and MNIST digit classification. While the fluctuation theorems are universal, there are other aspects of the stationary state that are highly sensitive to the training details. Surprisingly, the effective loss landscape and diffusion matrix that determine the shape of the stationary distribution vary depending on the simple choice of minibatching done with or without replacement. We can take advantage of this nonequilibrium sensitivity to engineer an equilibrium stationary state for a particular application: sampling from a posterior distribution of network weights in Bayesian machine learning. We propose a new variation of stochastic gradient Langevin dynamics (SGLD) that harnesses without replacement minibatching. In an example system where the posterior is exactly known, this SGWORLD algorithm outperforms SGLD, converging to the posterior orders of magnitude faster as a function of the learning rate. | ['Michael Hinczewski', 'Deniz Yuret', 'Alexander Strang', 'Alkan Kabakçıoğlu', 'Shishir Adhikari'] | 2023-06-06 | null | null | null | null | ['protein-folding', 'navigate'] | ['natural-language-processing', 'reasoning'] | [ 1.00012682e-01 -6.71968609e-02 8.24246258e-02 -2.50777841e-01
-4.08614874e-02 -4.98209029e-01 8.88572216e-01 -1.17684975e-01
-8.69958401e-01 1.23871505e+00 -3.11285645e-01 -5.56782782e-01
-3.45505744e-01 -8.03606927e-01 -8.32456112e-01 -1.48894703e+00
-1.75559521e-01 4.79451835e-01 4.19474095e-01 -4.45570678e-01
3.34917039e-01 6.05493605e-01 -1.42423558e+00 -3.67853045e-01
7.84437716e-01 8.32032382e-01 1.66898757e-01 7.98830032e-01
-2.00547770e-01 4.14795011e-01 -3.58471870e-01 -1.08542621e-01
2.86956370e-01 -7.29646564e-01 -6.98341489e-01 -4.58315164e-01
-1.09214514e-01 1.60030022e-01 -2.62073040e-01 1.20602727e+00
4.91155028e-01 3.29768479e-01 1.12893808e+00 -6.01467371e-01
-5.50709546e-01 4.22896504e-01 -1.99227780e-02 4.18539912e-01
-2.35596716e-01 2.92501897e-01 7.50526309e-01 -3.98607939e-01
6.07336640e-01 1.08510470e+00 8.97622585e-01 8.92738223e-01
-1.66619253e+00 -2.43836984e-01 -9.39905643e-02 -1.03882343e-01
-1.14652455e+00 -1.65372312e-01 7.24344790e-01 -5.85402966e-01
8.10281456e-01 -1.84320901e-02 7.91492045e-01 1.13231814e+00
8.57022107e-01 3.16501945e-01 1.30989754e+00 -7.21172571e-01
8.19322109e-01 1.44961536e-01 3.43786687e-01 7.00163782e-01
4.14777964e-01 2.73002088e-01 -3.45134169e-01 -2.41747946e-01
7.23014176e-01 -2.77399391e-01 -3.73852313e-01 -6.48075223e-01
-8.62331152e-01 8.21074963e-01 8.58769938e-02 3.76543164e-01
-3.03299367e-01 3.56573492e-01 1.92196250e-01 4.31209415e-01
4.03971821e-01 4.60583061e-01 -6.96587145e-01 -1.85253412e-01
-7.43499756e-01 4.02468503e-01 1.47293782e+00 1.39648154e-01
8.07330489e-01 -1.10901870e-01 3.27774584e-02 4.96089607e-01
3.36680174e-01 5.54435730e-01 5.79665482e-01 -8.16842377e-01
-8.38922188e-02 1.72496021e-01 2.28273794e-01 -4.22313839e-01
-4.64986920e-01 -6.38274670e-01 -9.05729592e-01 5.69959223e-01
9.47292566e-01 -4.67533201e-01 -8.40185940e-01 2.04433250e+00
3.24419737e-01 -1.89235270e-01 7.28056058e-02 8.65713418e-01
5.92562594e-02 7.04271853e-01 3.63381952e-02 -4.36503291e-01
8.83523524e-01 -2.44746178e-01 -4.32343513e-01 5.20999096e-02
5.57277858e-01 -1.50763705e-01 1.05141246e+00 2.16034755e-01
-1.04726839e+00 -1.36121968e-02 -1.19646239e+00 2.88923174e-01
-5.42566895e-01 -4.71028000e-01 4.08448666e-01 7.38501847e-01
-1.09059238e+00 1.48872280e+00 -1.03426385e+00 -4.85563964e-01
9.65780765e-02 4.25892204e-01 -6.99657649e-02 5.34914434e-01
-1.37078989e+00 1.17401230e+00 4.05971944e-01 1.50318518e-01
-5.22166789e-01 -6.41794503e-01 -2.92922318e-01 -2.15811178e-01
-1.47616044e-01 -8.27675521e-01 1.28639960e+00 -9.55346167e-01
-2.05353689e+00 7.08746195e-01 -1.33453771e-01 -6.07637107e-01
6.19893491e-01 1.47727519e-01 8.00118744e-02 -1.55044988e-01
-4.08092260e-01 2.76016116e-01 7.57903695e-01 -8.03534985e-01
2.23507851e-01 -3.53096724e-01 -2.03341544e-01 1.61226280e-02
-1.59657985e-01 -4.86988962e-01 2.64598101e-01 -1.60211086e-01
1.58548892e-01 -1.07479501e+00 -3.96255761e-01 -1.00481426e-02
-2.40379825e-01 -2.42255881e-01 5.07932544e-01 -1.39442414e-01
7.98313260e-01 -1.77801359e+00 4.17725086e-01 3.70023787e-01
4.72666360e-02 1.49206936e-01 2.77463377e-01 5.35908699e-01
3.92998867e-02 1.76742929e-03 -6.30835950e-01 8.86787623e-02
1.04169548e-01 3.05155665e-01 -1.90939143e-01 7.83372700e-01
1.77728832e-01 7.86160946e-01 -8.75327587e-01 -1.16410218e-01
-1.63904995e-01 6.85296416e-01 -4.34830785e-01 -1.56117782e-01
-2.80739009e-01 4.38382924e-01 -5.59774816e-01 -1.16192833e-01
4.64456618e-01 -2.93012649e-01 8.24361891e-02 7.47083724e-02
-2.64361471e-01 2.46698871e-01 -9.59565997e-01 1.35959506e+00
-2.43969083e-01 7.00453937e-01 -2.35065743e-02 -1.35673010e+00
8.57878685e-01 -3.40827256e-02 2.75496244e-01 -3.92631531e-01
1.27213821e-01 4.27174091e-01 3.03312361e-01 -4.08092082e-01
4.76067923e-02 -6.25600696e-01 2.84102410e-01 5.32859445e-01
2.86916107e-01 -1.94943249e-01 1.45613119e-01 1.12703562e-01
1.09952366e+00 3.32020551e-01 6.70622513e-02 -9.47868526e-01
4.05468494e-01 -1.91528738e-01 3.12723279e-01 1.03016210e+00
-5.85925095e-02 3.11583430e-01 8.98880959e-01 -4.24669087e-01
-1.32393992e+00 -1.14355278e+00 -5.39030612e-01 5.66732645e-01
7.39255846e-02 -1.77658275e-02 -1.07993555e+00 -2.17435583e-01
1.43013030e-01 6.29601955e-01 -7.07518041e-01 -4.70442772e-01
-4.65676188e-01 -1.47151840e+00 3.42179060e-01 -2.80679017e-01
3.70946258e-01 -1.16090500e+00 -8.15647662e-01 3.88050765e-01
4.95045692e-01 -6.86586142e-01 5.19428751e-04 6.96639717e-01
-1.16996288e+00 -8.65274966e-01 -7.60156512e-01 -1.73224553e-01
4.81786788e-01 -6.82973921e-01 1.12145770e+00 -8.31149444e-02
-3.74435365e-01 2.95789689e-01 3.25310141e-01 -1.82417542e-01
-7.53152668e-01 3.22868019e-01 3.58595073e-01 -2.16943979e-01
4.09762055e-01 -1.00903404e+00 -7.10504055e-01 1.36612905e-02
-9.06632781e-01 -1.69049665e-01 3.28468382e-01 7.43159354e-01
3.36935431e-01 -4.54561412e-02 4.58152503e-01 -6.93559110e-01
9.45405841e-01 -4.94558066e-01 -7.92980790e-01 1.01237640e-01
-8.52055013e-01 8.24513912e-01 7.31717348e-01 -6.42460108e-01
-8.91634524e-01 -9.01552215e-02 -3.82040218e-02 1.74286634e-01
2.99157551e-03 3.35294694e-01 2.37707913e-01 -2.23263264e-01
9.24571216e-01 3.47339779e-01 3.33467215e-01 -4.92228091e-01
1.43496871e-01 3.92553270e-01 2.49792546e-01 -9.25448835e-01
5.64231336e-01 4.04722720e-01 5.19262552e-01 -9.48387623e-01
-6.79137707e-01 1.74082682e-01 -7.25366175e-01 -2.65441656e-01
7.03453779e-01 -2.75023729e-01 -1.06094038e+00 8.08408380e-01
-1.01707971e+00 -7.07026541e-01 -6.54109776e-01 5.56309283e-01
-6.62076056e-01 4.40771803e-02 -7.24308372e-01 -1.27072167e+00
-1.61115929e-01 -8.31602097e-01 4.02544886e-01 5.25488317e-01
1.23515919e-01 -1.35456502e+00 5.61538398e-01 -6.24390543e-01
6.68959558e-01 2.31578216e-01 1.19796920e+00 -3.81325603e-01
-2.88870990e-01 -1.91555589e-01 2.48271868e-01 4.33258325e-01
-2.97975928e-01 1.91680297e-01 -7.08089173e-01 -1.59932122e-01
3.18748325e-01 2.02317555e-02 1.19594228e+00 6.55315936e-01
6.44999444e-01 -1.01229489e-01 -3.16687137e-01 4.92841095e-01
1.50855231e+00 1.64222836e-01 4.89816695e-01 2.61262804e-01
2.49400437e-01 5.90038538e-01 -3.75281781e-01 1.49055198e-01
-1.57487780e-01 4.22455877e-01 7.76091591e-02 3.48990202e-01
2.85744876e-01 -7.03709200e-02 5.97832322e-01 8.76958191e-01
-2.62246668e-01 9.45810527e-02 -9.01054025e-01 1.46146551e-01
-1.80235946e+00 -1.03912067e+00 -2.09992006e-02 2.45047855e+00
1.28155005e+00 4.45509344e-01 7.21674562e-02 -1.11618817e-01
5.16335785e-01 1.38661126e-02 -1.04511106e+00 -4.78939116e-01
-2.57918119e-01 3.14263105e-01 8.10620129e-01 7.62232482e-01
-6.14427328e-01 6.10888720e-01 6.60746574e+00 6.70620739e-01
-1.28644586e+00 2.56974667e-01 5.74922442e-01 -6.88395426e-02
-3.03061604e-01 1.05757490e-01 -1.00583422e+00 7.53057241e-01
1.41646695e+00 -1.18999302e-01 6.63079441e-01 5.14251411e-01
2.35019520e-01 -4.08217013e-01 -8.51746380e-01 5.77972353e-01
-4.55527216e-01 -1.42679346e+00 -3.69747549e-01 2.41379231e-01
6.31480575e-01 3.97187978e-01 -8.40597153e-02 5.95931523e-02
3.68354917e-01 -8.46611500e-01 5.66321671e-01 9.68195021e-01
5.16599953e-01 -6.12914562e-01 4.23431009e-01 6.13118291e-01
-4.83942389e-01 6.52282611e-02 -5.09352088e-01 -2.12230638e-01
1.59303740e-01 1.21034050e+00 -2.17264876e-01 -1.85928091e-01
5.10876775e-01 4.00738567e-01 -3.08054984e-01 9.01583731e-01
3.43567356e-02 7.37870693e-01 -8.51446629e-01 -8.18031192e-01
2.16114804e-01 -8.62397730e-01 7.42796183e-01 1.10557628e+00
1.75825611e-01 4.46680337e-02 -6.09541535e-01 1.19514716e+00
1.15546308e-01 -1.97744489e-01 -5.43436527e-01 -1.12736173e-01
1.56348586e-01 1.08014882e+00 -9.51062977e-01 -7.87581950e-02
1.51753768e-01 7.06387997e-01 3.27290207e-01 7.22057581e-01
-4.40312892e-01 -6.18392944e-01 7.03510463e-01 1.07788205e-01
4.77010459e-01 -5.23788929e-01 3.00026201e-02 -1.19077992e+00
1.06945060e-01 -2.91219592e-01 -2.72171140e-01 -3.78054678e-01
-1.25072300e+00 3.85217398e-01 1.02905817e-01 -4.20299351e-01
-3.55379403e-01 -8.35396349e-01 -7.23995745e-01 1.06283903e+00
-1.27564716e+00 -1.38806522e-01 3.51741731e-01 3.72522146e-01
-1.73346549e-01 -1.03887260e-01 7.38068402e-01 -1.52214214e-01
-6.44681096e-01 1.63828716e-01 8.84101510e-01 -1.83195695e-01
9.90717486e-02 -1.45343840e+00 4.94015247e-01 5.49123526e-01
-8.08980241e-02 7.79671490e-01 1.22628307e+00 -5.40854812e-01
-1.32278526e+00 -5.02538443e-01 6.00008309e-01 -3.06050837e-01
8.23493481e-01 -5.97600043e-01 -1.11354363e+00 1.21413529e-01
-7.45157003e-02 2.45149538e-01 2.15565532e-01 1.39915328e-02
-4.83125709e-02 -4.89807734e-03 -1.18149137e+00 6.09494269e-01
9.81837869e-01 -6.40532136e-01 -2.44963005e-01 4.98359501e-01
3.89869183e-01 -5.41406088e-02 -5.77313542e-01 1.58562601e-01
8.44783723e-01 -1.18221438e+00 6.09886825e-01 -8.60286593e-01
2.28256196e-01 -6.33071065e-02 6.70772493e-02 -1.31833136e+00
-2.08815411e-02 -1.07482255e+00 -2.36810848e-01 8.09819520e-01
5.37146270e-01 -1.04876637e+00 8.03930283e-01 6.67319655e-01
3.41695070e-01 -1.25761700e+00 -1.19690371e+00 -8.80442142e-01
7.05491781e-01 -1.41408309e-01 1.30860254e-01 4.18314666e-01
-8.54376182e-02 3.59699786e-01 1.66242465e-01 -1.57521695e-01
9.06546295e-01 -3.17185968e-01 1.16833478e-01 -1.47381842e+00
-4.49078649e-01 -5.91194332e-01 -3.32092941e-01 -9.73860025e-01
2.69326270e-01 -8.18047225e-01 1.24788269e-01 -1.03222835e+00
5.02384454e-02 -3.30190063e-01 -3.71014535e-01 -1.42931998e-01
2.00340077e-01 -2.25035444e-01 -2.02164799e-01 3.14495474e-01
-1.15130603e-01 5.31628549e-01 9.09668267e-01 3.71856749e-01
-4.94984567e-01 2.02494204e-01 -2.46713877e-01 6.45433664e-01
8.85691881e-01 -7.19295740e-01 -2.85429180e-01 1.51002228e-01
6.07386112e-01 -1.28304571e-01 3.68910551e-01 -1.00763822e+00
2.63929248e-01 -5.99424578e-02 2.62143403e-01 -5.49173281e-02
1.75709724e-01 -3.45403701e-01 1.10227399e-01 7.21067727e-01
-5.38669407e-01 -8.22238550e-02 -5.91381267e-02 6.64022744e-01
3.53343487e-01 -7.17843354e-01 1.07722819e+00 -2.38806471e-01
1.53447418e-02 1.31635800e-01 -8.06850851e-01 2.30263963e-01
5.82690299e-01 -1.57955915e-01 -2.22903356e-01 -3.18269789e-01
-6.59689963e-01 -1.87775597e-01 5.79287052e-01 -1.46976873e-01
1.83077216e-01 -1.02615380e+00 -3.82123798e-01 3.39192897e-01
-6.60916388e-01 -2.03238457e-01 -2.61839591e-02 9.36678469e-01
-4.93494660e-01 2.53337950e-01 -8.01695436e-02 -6.16565347e-01
-5.50403953e-01 1.77123010e-01 1.13772976e+00 -1.67198673e-01
-4.90184039e-01 6.73924863e-01 -2.03609332e-01 -4.75320607e-01
-8.24509710e-02 -3.08446378e-01 1.89214543e-01 -7.54271373e-02
3.25007558e-01 1.27572715e-01 3.23781185e-02 -1.80491135e-01
-1.35371402e-01 6.42200828e-01 -4.18951586e-02 -3.74480009e-01
1.41865182e+00 4.96943891e-02 -4.11224514e-01 1.01662004e+00
1.19719589e+00 -4.03917611e-01 -1.63719738e+00 -1.00334592e-01
2.07368538e-01 2.72780091e-01 2.03664184e-01 -5.75753748e-01
-7.95046866e-01 8.49067390e-01 7.36352623e-01 6.90545261e-01
4.43074346e-01 3.39290244e-04 5.26006699e-01 7.71611035e-01
2.50341594e-01 -1.14948487e+00 -4.69447911e-01 7.43102074e-01
4.24589545e-01 -8.25692236e-01 -5.61779179e-02 4.42211062e-01
-7.38029629e-02 1.43830395e+00 9.06271413e-02 -5.24590313e-01
1.07304943e+00 3.24007958e-01 -2.52254248e-01 -8.53868648e-02
-8.18429530e-01 1.61272138e-01 1.03444450e-01 4.16530848e-01
3.98253918e-01 -1.80193767e-01 -4.69635963e-01 9.28070694e-02
-1.21341847e-01 5.49741276e-03 4.77325171e-01 7.36783445e-01
-6.02747977e-01 -1.05705321e+00 5.51056676e-02 5.03074288e-01
-4.89421517e-01 4.99538668e-02 -1.95883796e-01 6.44766510e-01
-2.35895693e-01 3.51908833e-01 1.32484078e-01 2.05462929e-02
-9.15363710e-03 6.22617483e-01 8.13877702e-01 -2.80286342e-01
-2.67945141e-01 -4.57350791e-01 -2.38352239e-01 -3.44976902e-01
-5.27753472e-01 -8.47109318e-01 -1.30773747e+00 -4.81550753e-01
-5.19902050e-01 4.98685896e-01 1.07071257e+00 1.38177335e+00
1.15373917e-01 1.76680535e-01 2.46879950e-01 -1.11043489e+00
-1.02586615e+00 -8.09947848e-01 -9.06842470e-01 1.09846070e-01
6.87623739e-01 -5.52221298e-01 -9.11209285e-01 -2.94672310e-01] | [6.139919281005859, 4.214087963104248] |
b13ed99d-f158-4e35-b9f3-8a8099ec80ea | raft-stereo-multilevel-recurrent-field | 2109.07547 | null | https://arxiv.org/abs/2109.07547v1 | https://arxiv.org/pdf/2109.07547v1.pdf | RAFT-Stereo: Multilevel Recurrent Field Transforms for Stereo Matching | We introduce RAFT-Stereo, a new deep architecture for rectified stereo based on the optical flow network RAFT. We introduce multi-level convolutional GRUs, which more efficiently propagate information across the image. A modified version of RAFT-Stereo can perform accurate real-time inference. RAFT-stereo ranks first on the Middlebury leaderboard, outperforming the next best method on 1px error by 29% and outperforms all published work on the ETH3D two-view stereo benchmark. Code is available at https://github.com/princeton-vl/RAFT-Stereo. | ['Jia Deng', 'Zachary Teed', 'Lahav Lipson'] | 2021-09-15 | null | null | null | null | ['stereo-depth-estimation'] | ['computer-vision'] | [-2.86513597e-01 -2.38902956e-01 4.60136775e-03 -4.87169564e-01
-3.71903449e-01 -5.49407482e-01 4.75621939e-01 -5.29969752e-01
-5.75690091e-01 7.69770503e-01 5.37424386e-01 -3.88785750e-02
1.71218440e-01 -6.07579708e-01 -8.02318394e-01 -1.85651422e-01
-2.24962402e-02 2.86550224e-01 3.49191844e-01 -3.63671809e-01
4.08944756e-01 2.62910396e-01 -1.45932722e+00 5.30786395e-01
3.50886375e-01 7.11178005e-01 6.70721307e-02 9.67797875e-01
5.88943005e-01 9.71543431e-01 -2.89408658e-02 -4.72763598e-01
7.84274638e-01 -1.14230976e-01 -1.11887538e+00 -2.98880070e-01
1.58802104e+00 -1.05195045e+00 -9.28357899e-01 8.39490354e-01
6.40554488e-01 1.57122165e-01 2.56447643e-01 -1.27447438e+00
-4.11853909e-01 2.32359394e-01 -7.77565420e-01 8.81905258e-01
6.36559248e-01 4.73398447e-01 9.60447848e-01 -8.09502721e-01
1.03287578e+00 1.73599648e+00 8.17072332e-01 4.52161968e-01
-1.40962899e+00 -8.21599960e-01 9.51163545e-02 1.88231811e-01
-1.07791471e+00 -5.63609421e-01 7.62041658e-02 -5.62707663e-01
1.11796343e+00 -1.23407193e-01 8.98629248e-01 1.35592949e+00
5.25151014e-01 9.00850415e-01 1.28831005e+00 5.47022164e-01
-2.58558720e-01 -6.56530142e-01 1.50860846e-01 8.62742901e-01
1.98496789e-01 7.08952129e-01 -9.28551257e-01 1.19986907e-01
1.10937130e+00 -1.16143726e-01 -5.19385695e-01 -2.66938120e-01
-1.53040469e+00 6.44801855e-01 1.07837689e+00 -1.16562702e-01
-2.03609630e-01 5.65654516e-01 4.86935675e-01 2.19081610e-01
5.47209740e-01 2.63245851e-01 -4.73692745e-01 -1.75328612e-01
-8.66405010e-01 5.07025361e-01 4.42788094e-01 9.03196394e-01
8.26750636e-01 1.02400251e-01 -1.13688566e-01 3.16513151e-01
2.61116892e-01 4.89162594e-01 1.11948431e-01 -2.01105905e+00
4.75336045e-01 1.01004325e-01 -2.93129031e-02 -7.48377919e-01
-3.58751208e-01 -3.81344289e-01 -8.18422556e-01 5.28278172e-01
4.74300414e-01 -2.16055378e-01 -8.86053324e-01 1.32683539e+00
1.56001255e-01 4.04010504e-01 -2.10603941e-02 1.32000542e+00
1.13014746e+00 3.78609300e-01 -3.23871970e-01 6.37058198e-01
8.75994503e-01 -1.27035189e+00 -1.85606286e-01 -3.74507189e-01
4.01611775e-01 -7.08494544e-01 5.90901971e-01 4.72469807e-01
-1.01932538e+00 -5.67193747e-01 -8.65723670e-01 -7.44615018e-01
-2.60065556e-01 -2.20576108e-01 8.55991840e-01 -9.98672657e-03
-1.62787521e+00 7.37128735e-01 -7.82339156e-01 -4.90468621e-01
6.91441476e-01 2.62842596e-01 -8.54049683e-01 -3.81914884e-01
-9.81748402e-01 5.31163990e-01 1.70683414e-01 3.27061713e-01
-1.24925804e+00 -1.13537312e+00 -1.14664805e+00 -3.04411262e-01
-1.20684914e-01 -1.34417689e+00 1.54054153e+00 -5.60692549e-01
-1.46458626e+00 1.13957441e+00 -2.70800352e-01 -7.59639382e-01
1.08987272e+00 -6.33548260e-01 9.14548635e-02 2.07993343e-01
3.97613019e-01 1.44598067e+00 4.27375406e-01 -1.00997496e+00
-6.84699595e-01 -4.00012463e-01 5.51188827e-01 2.95987070e-01
7.29556978e-01 -4.03952867e-01 -3.47671986e-01 -2.27160037e-01
-1.00180209e-01 -9.71686006e-01 -2.64123231e-01 2.77442992e-01
-7.04705298e-01 1.75196931e-01 7.74471819e-01 -5.06542623e-01
3.69448066e-01 -2.01503634e+00 -7.64031569e-03 -3.82698327e-01
6.63080513e-01 1.13934405e-01 -2.50301152e-01 9.27038416e-02
-6.70707077e-02 -2.29504526e-01 -1.13522947e-01 -5.42713583e-01
-2.32516423e-01 8.37267414e-02 -2.24660799e-01 7.97746062e-01
-3.06733668e-01 9.26217020e-01 -1.16044307e+00 -2.85824478e-01
7.49609113e-01 7.99459457e-01 -9.73438084e-01 -3.34568024e-02
1.69713929e-01 1.06354392e+00 -1.45063847e-01 3.10788959e-01
9.55571532e-01 -2.73425907e-01 -3.29382151e-01 -2.42655247e-01
-4.49448436e-01 3.92314255e-01 -9.25703108e-01 2.13114166e+00
-3.97228897e-01 1.18243456e+00 3.15408915e-01 -4.00948346e-01
5.06540716e-01 -7.90878758e-02 3.80023986e-01 -7.70292103e-01
1.53117925e-01 9.79731083e-02 -1.83708712e-01 -9.12935734e-02
6.64499521e-01 2.77090579e-01 1.96764141e-01 -1.47475377e-02
3.70826215e-01 -2.46252105e-01 3.78022820e-01 4.69347298e-01
1.08806622e+00 4.90211368e-01 3.90739143e-02 -5.83880246e-01
5.06035566e-01 6.63351193e-02 6.89626813e-01 7.01673210e-01
-4.46377695e-01 8.82949531e-01 2.87339956e-01 -8.84635270e-01
-8.95575166e-01 -1.36651349e+00 -3.03708971e-01 7.90687799e-01
4.57501769e-01 -5.15718102e-01 -4.52546060e-01 -5.57509482e-01
4.02231663e-01 3.33156407e-01 -6.21928334e-01 4.50706542e-01
-3.91643375e-01 -9.64767709e-02 5.45543253e-01 4.93191481e-01
1.15334451e+00 -8.55299711e-01 -6.92861795e-01 -1.01952940e-01
-4.13530320e-01 -1.33964849e+00 -7.46525407e-01 -1.19321465e-01
-9.86080289e-01 -1.43123853e+00 -6.65410757e-01 -4.78938669e-01
3.60041738e-01 7.08197653e-01 1.46941996e+00 1.05143048e-01
-3.56842339e-01 3.91283125e-01 1.39053157e-02 4.52415161e-02
7.28618428e-02 2.02221990e-01 -2.03714937e-01 -4.08404887e-01
2.08375022e-01 -7.45720983e-01 -1.12182891e+00 3.52354556e-01
-5.31605840e-01 2.43572131e-01 7.27576464e-02 7.25215435e-01
6.30433321e-01 -5.36662698e-01 -3.43140453e-01 -8.30588758e-01
2.04409376e-01 -2.85235107e-01 -1.00044870e+00 -6.04786992e-01
-1.40537933e-01 6.48571998e-02 4.06546444e-01 4.15576905e-01
-1.07184541e+00 1.53144766e-02 -1.61098287e-01 -7.27547467e-01
-3.53830040e-01 -1.75058171e-01 5.51615119e-01 -2.07936496e-01
5.81466377e-01 -1.10323288e-01 -6.37989119e-02 -1.94287688e-01
3.92098695e-01 1.81500107e-01 9.13539052e-01 -1.86633930e-01
5.57107508e-01 1.11689496e+00 1.72828853e-01 -8.00771713e-01
-1.15988433e+00 -6.76756799e-01 -1.03549600e+00 -2.94615358e-01
1.15966177e+00 -1.42986226e+00 -1.11737955e+00 7.36853659e-01
-1.25047195e+00 -8.02502275e-01 4.44944762e-03 7.23910689e-01
-7.46543527e-01 1.55907288e-01 -8.73988509e-01 -2.83013266e-02
-3.84201348e-01 -1.30423021e+00 1.57908249e+00 1.64407045e-01
-2.47222349e-01 -1.16844165e+00 3.55095446e-01 6.75445199e-01
4.05413926e-01 3.81612539e-01 -1.45808920e-01 1.29430771e-01
-1.01081383e+00 3.46087754e-01 -4.58560288e-01 3.17406625e-01
-2.35358372e-01 -1.45973201e-04 -1.14728260e+00 -5.20184636e-01
-4.94427085e-01 -6.34708703e-01 1.51340318e+00 8.21627080e-01
1.04424012e+00 1.43006921e-01 -2.29363397e-01 1.37973821e+00
1.46653914e+00 -3.03328544e-01 7.24207759e-01 5.31035006e-01
8.35177124e-01 2.72893220e-01 3.35505724e-01 2.96505362e-01
8.05063009e-01 4.93760675e-01 8.07360470e-01 -1.92699686e-01
-5.12499034e-01 -3.63372803e-01 3.53461921e-01 4.99344409e-01
-3.38840693e-01 -4.56971563e-02 -8.97911310e-01 5.63780367e-01
-1.82364416e+00 -1.26627827e+00 -5.05222917e-01 1.89436030e+00
3.52783054e-01 1.54324085e-01 -9.57424790e-02 -2.31009424e-01
5.65589368e-01 4.93903100e-01 -6.93253517e-01 -3.06772351e-01
-1.67601556e-01 -6.45844918e-03 9.07781959e-01 1.04181373e+00
-1.26331520e+00 1.09996724e+00 6.62236500e+00 7.81985149e-02
-1.16365314e+00 2.65408698e-02 5.10492146e-01 -1.63036257e-01
4.61117662e-02 3.98314707e-02 -9.71333265e-01 2.37140477e-01
4.64042723e-01 -1.31397814e-01 6.19215965e-01 6.32082760e-01
3.38158906e-01 -4.36278015e-01 -1.06422937e+00 1.17681718e+00
-7.47605264e-02 -1.43565845e+00 -2.83374134e-02 3.62779737e-01
8.84513676e-01 1.15295482e+00 9.32119861e-02 1.16468705e-01
8.16433072e-01 -9.27820683e-01 6.90221608e-01 4.10698622e-01
7.23072648e-01 -4.29234505e-01 7.04831541e-01 7.65786320e-02
-1.23890638e+00 2.30948493e-01 -4.84722167e-01 -3.38529348e-01
1.64524078e-01 5.26501358e-01 -5.81357300e-01 5.38744390e-01
1.23201096e+00 1.73711181e+00 -5.52925885e-01 1.26656592e+00
-4.97706592e-01 1.29276857e-01 -3.61736625e-01 6.82551503e-01
4.95310992e-01 -1.56526864e-01 8.08725476e-01 1.14499032e+00
1.93277195e-01 -1.13436386e-01 2.04379007e-01 7.98854113e-01
-2.62086064e-01 -5.47563136e-01 -1.13018501e+00 5.99511027e-01
2.35717863e-01 1.14917123e+00 -3.32859755e-01 -3.31407070e-01
-3.93262565e-01 1.08162570e+00 3.76383662e-01 3.55398357e-01
-6.89228714e-01 -3.44233625e-02 1.26314402e+00 1.12925306e-01
3.48572820e-01 -3.74792129e-01 1.38780639e-01 -1.56077468e+00
-3.77555966e-01 -5.57422519e-01 4.37172115e-01 -1.33948922e+00
-1.26688361e+00 7.38282621e-01 -6.56057447e-02 -1.32993782e+00
-1.64026946e-01 -7.29205072e-01 -4.03711438e-01 7.43551850e-01
-1.91412485e+00 -7.60720134e-01 -7.40041614e-01 1.20253980e+00
5.46179056e-01 1.52018070e-01 5.40158570e-01 1.17568150e-01
-2.53954321e-01 9.17878523e-02 -1.02324113e-01 3.38521600e-01
9.55364406e-01 -1.39604688e+00 9.98257339e-01 9.17767107e-01
-1.04594223e-01 3.94074529e-01 6.77745879e-01 -4.20073539e-01
-1.18599963e+00 -1.17531168e+00 5.89838982e-01 -5.61076581e-01
6.70376301e-01 -2.16049254e-01 -3.05723816e-01 1.29683363e+00
7.00114429e-01 4.92455363e-01 1.03736937e-01 -3.66428308e-02
-7.03744352e-01 -2.17098624e-01 -1.11980081e+00 5.19313335e-01
1.51919341e+00 -5.37553191e-01 -5.12081981e-01 1.24248959e-01
5.23383498e-01 -9.64443088e-01 -7.96650767e-01 2.53188640e-01
5.96373916e-01 -1.80129373e+00 1.02665544e+00 -8.52141231e-02
6.35639489e-01 -3.98443252e-01 -1.78340867e-01 -1.62213063e+00
-4.52111959e-01 -7.36423433e-01 1.83394372e-01 3.06876332e-01
7.15505034e-02 -8.36823463e-01 8.05825055e-01 -1.33396059e-01
-2.58358777e-01 -2.58492053e-01 -1.05827737e+00 -7.74249732e-01
-2.31192596e-02 -2.45593175e-01 2.49508008e-01 6.46611750e-01
-4.85748589e-01 4.53203589e-01 -3.81508857e-01 8.17982182e-02
1.27407801e+00 1.16072074e-01 1.07201505e+00 -1.20661569e+00
-1.07545912e-01 -2.85301030e-01 -8.55151236e-01 -1.40224946e+00
2.11313471e-01 -9.86283362e-01 -6.87228516e-02 -1.62233007e+00
4.54821922e-02 2.63409555e-01 2.02539429e-01 4.97951180e-01
2.17473745e-01 6.57426894e-01 4.27238286e-01 7.32754022e-02
-6.83226585e-01 5.00369072e-01 1.78026974e+00 -2.49263108e-01
2.23630086e-01 -1.81854829e-01 -4.08289015e-01 8.92539561e-01
9.63196218e-01 -2.75195390e-01 -3.21333677e-01 -1.05700648e+00
5.73103242e-02 1.04183301e-01 7.83388793e-01 -1.31994867e+00
1.65179998e-01 2.56245971e-01 5.34122169e-01 -8.04115593e-01
5.18920362e-01 -4.95570064e-01 -3.90811116e-02 6.34984076e-01
-2.98595816e-01 3.78724366e-01 3.65971774e-01 4.67966437e-01
-3.66621226e-01 6.35204077e-01 1.02885103e+00 -4.98671561e-01
-1.16071522e+00 7.15203047e-01 -2.05940515e-01 4.86195743e-01
7.47854829e-01 -2.36613542e-01 -9.86549079e-01 -5.00118673e-01
-5.35838664e-01 5.81257820e-01 5.58995008e-01 5.36482632e-01
5.04343748e-01 -1.21937072e+00 -7.45521724e-01 2.07001522e-01
-1.67975612e-02 1.92349374e-01 3.93405795e-01 1.03103936e+00
-1.03741312e+00 6.96626186e-01 -6.89059615e-01 -9.65472400e-01
-1.10196388e+00 2.56917700e-02 8.33215952e-01 -8.70098127e-04
-1.00899553e+00 9.92375910e-01 5.35300851e-01 -6.32700443e-01
1.86501760e-02 -4.33320463e-01 4.98478636e-02 -3.55779737e-01
5.84906757e-01 6.14827633e-01 -8.34300555e-03 -8.69347155e-01
-4.22520459e-01 7.65891969e-01 1.41648218e-01 -1.51340542e-02
1.41253543e+00 -3.32259178e-01 -2.46295720e-01 2.06107393e-01
1.39819622e+00 -2.79856622e-01 -1.78656363e+00 3.64735201e-02
-4.83928293e-01 -7.59324431e-01 4.85854536e-01 -5.90694785e-01
-1.49646342e+00 9.86106396e-01 3.74239951e-01 -2.99009144e-01
8.47722888e-01 -2.90207148e-01 9.13119972e-01 3.35036516e-01
7.33061671e-01 -4.61343437e-01 -1.46010369e-01 1.00022280e+00
9.75689709e-01 -1.28606844e+00 7.18343034e-02 -2.81804383e-01
-4.25278813e-01 1.21530855e+00 9.00742948e-01 -6.80138767e-01
7.65552819e-01 3.88574958e-01 2.83053398e-01 -4.70701873e-01
-8.41972172e-01 -4.15513366e-01 1.29704088e-01 5.86594939e-01
5.01364529e-01 -1.36216953e-01 3.01946819e-01 -4.77235168e-01
-5.20568013e-01 3.82586062e-01 7.11254895e-01 7.17532575e-01
-1.46979734e-01 -6.36938572e-01 -1.07281022e-01 2.03153044e-01
-3.46493661e-01 -2.19325081e-01 -3.57205391e-01 8.82928729e-01
1.18921269e-02 7.87186801e-01 3.28286558e-01 -3.59592080e-01
4.89765912e-01 -5.92837572e-01 6.39488280e-01 -3.70893776e-01
-5.78193009e-01 -2.11296931e-01 2.00447232e-01 -1.59759581e+00
-8.76297176e-01 -6.45647705e-01 -9.13087666e-01 -1.01827943e+00
4.66900229e-01 -1.88717470e-01 1.61345109e-01 4.85068738e-01
5.72730958e-01 2.78015375e-01 4.23358709e-01 -1.44561136e+00
1.76787287e-01 -7.34338403e-01 -4.87189740e-01 2.00266808e-01
8.91506970e-01 -6.21471465e-01 -5.77602684e-01 -4.77480255e-02] | [8.615640640258789, -1.9343127012252808] |
c5ac6ecc-0e81-49bc-9e72-3dcd9eff9c02 | age-and-gender-classification-using | null | null | https://talhassner.github.io/home/publication/2015_CVPR | https://talhassner.github.io/home/projects/cnn_agegender/CVPR2015_CNN_AgeGenderEstimation.pdf | Age and Gender Classification using Convolutional Neural Networks | Automatic age and gender classification has become relevant to an increasing amount of applications, particularly since the rise of social platforms and social media. Nevertheless, performance of existing methods on real-world images is still significantly lacking, especially when compared to the tremendous leaps in performance recently reported for the related task of face recognition. In this paper we show that by learning representations through the use of deep-convolutional neural networks (CNN), a significant increase in performance can be obtained on these tasks. To this end, we propose a simple convolutional net architecture that can be used even when the amount of learning data is limited. We evaluate our method on the recent Adience benchmark for age and gender estimation and show it to dramatically outperform current state-of-the-art methods. | ['Tal Hassner', 'Gil Levi'] | 2015-10-26 | null | null | null | 2015-ieee-conference-on-computer-vision-and-1 | ['age-and-gender-estimation', 'age-and-gender-classification'] | ['computer-vision', 'computer-vision'] | [-7.24944174e-02 1.28922224e-01 -8.37620050e-02 -6.69168532e-01
-2.71296412e-01 -1.55956954e-01 8.46368372e-01 2.03469813e-01
-7.08279431e-01 7.18863428e-01 6.85794130e-02 -2.74972636e-02
5.61544113e-02 -8.18718970e-01 -4.62680161e-01 -5.30528903e-01
-2.76370287e-01 4.93347168e-01 -2.78595716e-01 -2.15931714e-01
1.39534369e-01 4.87363547e-01 -2.03003097e+00 -4.96443547e-02
6.67116463e-01 1.36488116e+00 -5.00191808e-01 4.41471010e-01
-4.37085479e-02 6.79031789e-01 -4.26683307e-01 -8.24883699e-01
1.28307909e-01 -5.93501180e-02 -7.67394781e-01 -3.01674396e-01
8.97224128e-01 -5.74342430e-01 -3.31987798e-01 7.75937378e-01
8.31809402e-01 -2.52931774e-01 5.87773263e-01 -1.28274500e+00
-5.51042378e-01 6.10932231e-01 -7.53339171e-01 4.61497046e-02
2.09930226e-01 -4.23968911e-01 8.00436676e-01 -8.14006090e-01
4.24170345e-01 1.45555663e+00 8.21004391e-01 8.87015224e-01
-9.50439870e-01 -1.01538467e+00 -5.50037287e-02 3.06248605e-01
-1.35568357e+00 -6.33292556e-01 5.38037717e-01 -3.13216686e-01
6.80914342e-01 -8.59010145e-02 5.29432535e-01 1.41974807e+00
-8.40523988e-02 6.87453389e-01 1.14745080e+00 -3.64296645e-01
4.18419614e-02 -1.01008885e-01 -2.66001727e-02 8.12997103e-01
2.64376640e-01 -8.65144581e-02 -7.48768568e-01 6.16236180e-02
4.08705175e-01 -1.35206327e-01 2.67242223e-01 -1.90302163e-01
-7.85162151e-01 8.83988917e-01 3.04298192e-01 3.01960379e-01
-2.15137884e-01 2.91900456e-01 6.17735028e-01 2.29116410e-01
9.98250008e-01 3.31326067e-01 -3.90634328e-01 -3.40124756e-01
-1.18072248e+00 5.91828465e-01 6.81095839e-01 3.07477623e-01
3.79190564e-01 3.35750803e-02 -8.60448852e-02 9.33939636e-01
3.23875472e-02 2.70421803e-01 4.27736849e-01 -9.08737183e-01
1.38986856e-01 8.32150757e-01 -2.34697446e-01 -8.58341575e-01
-5.89409888e-01 -4.75038201e-01 -9.41193640e-01 3.38952422e-01
9.55156207e-01 -1.37060553e-01 -9.39577699e-01 1.90561152e+00
2.23978668e-01 1.52685404e-01 -3.01648349e-01 4.25924718e-01
9.92709100e-01 1.90574422e-01 2.24100381e-01 -5.39512895e-02
1.40606928e+00 -5.41714430e-01 -5.74433804e-01 -3.14182192e-01
3.76323909e-01 -6.93906605e-01 5.51019371e-01 4.91196573e-01
-1.14048433e+00 -5.14295340e-01 -1.06165636e+00 -2.71212906e-02
-6.56471908e-01 1.78259403e-01 1.09812772e+00 1.13681304e+00
-1.13468409e+00 9.11478400e-01 -6.71073914e-01 -6.56658292e-01
1.04352832e+00 8.40541482e-01 -5.38683891e-01 -1.86922044e-01
-9.05198932e-01 6.88245475e-01 4.81415018e-02 1.20447084e-01
-5.66751778e-01 -9.03966725e-01 -7.33112037e-01 -2.04229504e-02
8.91946331e-02 -6.60587609e-01 1.22336376e+00 -1.07625723e+00
-1.18919587e+00 1.31586027e+00 1.36624649e-01 -5.02021909e-01
7.08925843e-01 -4.19341445e-01 -2.71970361e-01 -1.42111164e-02
-1.77744105e-01 1.07207739e+00 9.05133307e-01 -8.09845746e-01
-6.07700229e-01 -7.96148717e-01 1.94914788e-01 -1.04777224e-01
-1.19015169e+00 3.72308552e-01 -3.28670442e-01 -4.39837813e-01
-2.92051047e-01 -9.13923204e-01 -4.70594056e-02 1.65119365e-01
-5.36114462e-02 -5.55283129e-01 8.05106759e-01 -7.46024013e-01
1.24154449e+00 -1.89689898e+00 2.59937763e-01 -1.66829407e-01
4.84251499e-01 4.38520163e-01 9.26605985e-02 1.53201014e-01
-1.54498264e-01 5.32006845e-02 -9.16192830e-02 -8.41157019e-01
-5.18233143e-02 2.71181781e-02 1.71881318e-01 3.94587129e-01
2.52760708e-01 7.58057535e-01 -6.06132507e-01 -4.31523800e-01
4.73540509e-03 7.14693964e-01 -5.65607309e-01 2.35789850e-01
1.07063591e-01 4.15521204e-01 -1.15094170e-01 8.47807527e-01
7.99204767e-01 2.91176350e-03 1.31544143e-01 5.17916344e-02
-2.88289562e-02 -7.69857094e-02 -7.72116184e-01 1.41872692e+00
-5.63234806e-01 6.20664597e-01 7.21204877e-02 -1.01318860e+00
9.14897084e-01 1.72197074e-01 6.52706385e-01 -7.39328265e-01
5.48570037e-01 2.87466913e-01 1.76053748e-01 -1.95302099e-01
4.79983985e-01 1.11394495e-01 -1.83470838e-04 2.81129479e-01
3.11902672e-01 2.57288247e-01 3.49395514e-01 -1.52604179e-02
8.52747858e-01 -1.09679312e-01 9.62834433e-03 -3.30916017e-01
6.34339273e-01 -7.33701885e-01 3.37519348e-01 3.60740632e-01
-2.35336348e-01 7.37243772e-01 8.21222723e-01 -7.41888046e-01
-1.04822803e+00 -7.22680748e-01 -2.96131402e-01 1.39960921e+00
-5.27285218e-01 -4.79823977e-01 -1.17172348e+00 -8.04452240e-01
2.45183676e-01 -1.26084059e-01 -1.17051041e+00 1.51620358e-02
-6.63231313e-01 -1.05549943e+00 6.28806770e-01 7.31007278e-01
6.71707869e-01 -1.02395010e+00 -3.93597990e-01 -5.11272214e-02
3.91484857e-01 -1.41799438e+00 1.27772167e-01 -2.23191351e-01
-8.22203517e-01 -1.00863564e+00 -1.03142869e+00 -6.85386062e-01
4.84187126e-01 -3.38855833e-01 1.38003540e+00 3.06713700e-01
-3.41543078e-01 2.52743781e-01 -1.22359410e-01 -7.54433990e-01
-3.33025008e-02 7.88834810e-01 2.05208689e-01 1.42761603e-01
5.98589122e-01 -8.63368034e-01 -9.23151195e-01 -6.13231845e-02
-6.62594795e-01 -9.92404111e-03 3.32267702e-01 6.96067095e-01
-2.68293023e-01 -3.71226609e-01 1.15107071e+00 -9.95006621e-01
4.90024179e-01 -3.89204890e-01 -4.47002500e-01 -2.52265275e-01
-8.14989805e-01 -1.36971727e-01 3.77868116e-01 -3.31916839e-01
-7.74144828e-01 5.39071672e-03 -5.73515952e-01 -8.95644128e-02
-1.93821207e-01 4.48207468e-01 1.59269229e-01 -2.00306162e-01
5.19976318e-01 -2.79340267e-01 2.71902293e-01 -6.46623969e-01
7.14017227e-02 9.31569695e-01 4.39603150e-01 -5.49410164e-01
6.46346867e-01 4.75271493e-01 2.70798981e-01 -7.14822471e-01
-9.13214564e-01 -7.14594126e-02 -5.09282708e-01 -2.98046052e-01
8.32662106e-01 -9.33838308e-01 -1.09719431e+00 1.01046872e+00
-7.78237641e-01 -1.14748202e-01 1.96209043e-01 1.95468962e-01
-2.91915536e-01 1.62521794e-01 -7.29475915e-01 -8.54152739e-01
-5.45701265e-01 -1.00439334e+00 9.56780493e-01 4.64235187e-01
-2.25599468e-01 -1.04317153e+00 -1.19452156e-01 3.91676217e-01
6.55445635e-01 4.68587309e-01 7.60561287e-01 -5.48912764e-01
-8.36959109e-02 -3.48261833e-01 -5.12961745e-01 4.83957767e-01
-1.03183977e-01 3.83776389e-02 -1.41436982e+00 -3.57549012e-01
-7.03551173e-01 -6.59592569e-01 1.14831209e+00 3.70935857e-01
1.59580517e+00 1.24343596e-01 -1.99275479e-01 5.28118134e-01
1.07099080e+00 -2.65459299e-01 7.77297080e-01 2.42495209e-01
7.06786811e-01 8.61269653e-01 2.81535745e-01 6.99656308e-01
5.53113401e-01 6.90003395e-01 6.56207681e-01 -3.59230697e-01
-1.22486934e-01 -1.05094118e-02 -1.68701872e-01 3.69574606e-01
-6.12936676e-01 1.80337682e-01 -1.12835121e+00 6.80644751e-01
-1.77694166e+00 -6.35717630e-01 1.75266311e-01 2.13800621e+00
6.25426352e-01 1.39490753e-01 6.06107235e-01 3.48913610e-01
5.80136955e-01 2.34169409e-01 -4.43828344e-01 -5.82285225e-01
8.20265412e-02 7.66943276e-01 2.47914597e-01 -5.70700578e-02
-1.20768535e+00 6.67526841e-01 6.56211996e+00 6.88420296e-01
-1.24435723e+00 2.05658469e-02 1.23024607e+00 -1.73807472e-01
4.26199436e-01 -6.46885931e-01 -8.16201329e-01 3.49225104e-01
1.05507779e+00 -1.16678536e-01 3.49508762e-01 8.47412527e-01
-3.15671802e-01 -1.63156483e-02 -1.30887687e+00 1.32922482e+00
2.76976466e-01 -1.08451033e+00 -3.78740221e-01 2.94517666e-01
7.26839840e-01 -2.49352440e-01 6.53422058e-01 5.42392373e-01
-1.58913627e-01 -1.49659181e+00 4.31308091e-01 1.37498572e-01
8.21505189e-01 -1.17392719e+00 9.37644064e-01 -2.25140918e-02
-9.68199313e-01 -4.08673048e-01 -2.60762602e-01 -4.71521646e-01
-3.44719440e-01 5.69398582e-01 -5.26497602e-01 3.90451938e-01
1.01637948e+00 8.28005075e-01 -9.32762444e-01 9.02917266e-01
4.99123745e-02 5.42958915e-01 -1.13057569e-01 8.95799771e-02
2.32606996e-02 1.05355904e-01 -2.18304873e-01 9.85636055e-01
5.94756067e-01 -3.18257719e-01 -2.57089347e-01 3.56403708e-01
-6.21665776e-01 3.55791032e-01 -6.04412913e-01 -2.74683923e-01
1.18392199e-01 1.60719955e+00 -6.22919440e-01 -1.51863500e-01
-5.89578748e-01 6.21603847e-01 6.39696479e-01 -1.23657651e-01
-5.03236473e-01 -1.85975224e-01 8.44455063e-01 3.59701633e-01
1.61814719e-01 -2.14259058e-01 -2.23745972e-01 -8.39769423e-01
-1.44680470e-01 -8.11585844e-01 4.76644099e-01 -3.12022358e-01
-1.41974258e+00 3.90306145e-01 -5.00354506e-02 -7.18421459e-01
-4.49327111e-01 -9.40497875e-01 -5.13217986e-01 6.18047476e-01
-1.51039982e+00 -1.47326088e+00 -3.68297011e-01 3.16881150e-01
3.84306878e-01 -4.89769071e-01 1.01412487e+00 7.09916711e-01
-6.36731327e-01 9.60855663e-01 -7.05286441e-03 3.04423779e-01
9.73157108e-01 -1.26490581e+00 5.02494097e-01 3.78306985e-01
-2.01861173e-01 4.68657762e-01 6.63573444e-01 -2.28081837e-01
-1.12919855e+00 -8.56383681e-01 1.10811675e+00 -5.36261022e-01
5.70127785e-01 -8.93267155e-01 -5.67552269e-01 4.33371395e-01
3.70302439e-01 1.11099295e-01 6.42489672e-01 7.83663988e-01
-6.76497340e-01 -3.79603237e-01 -1.26482999e+00 6.02416575e-01
1.03386772e+00 -2.72360444e-01 -1.12647146e-01 1.66660011e-01
1.48376286e-01 -4.10789251e-01 -9.87289429e-01 7.49265611e-01
1.17226994e+00 -1.19876862e+00 1.15037036e+00 -4.74552542e-01
1.08702135e+00 4.05758470e-01 1.24230437e-01 -1.10939324e+00
6.18258938e-02 -2.78008282e-01 -4.43601549e-01 1.49165845e+00
1.47787124e-01 -3.56051117e-01 1.28655291e+00 5.43990970e-01
2.64250517e-01 -1.19622099e+00 -9.92528915e-01 -4.85232562e-01
4.88701373e-01 -1.48090288e-01 7.65969396e-01 9.20092404e-01
-2.01589704e-01 3.03709656e-01 -6.55294418e-01 -3.60857487e-01
6.56024814e-01 -2.09086388e-01 6.51198149e-01 -1.73260260e+00
3.26155797e-02 -5.46697676e-01 -8.02192509e-01 -2.77867913e-01
5.32193899e-01 -6.26705945e-01 -5.11797547e-01 -1.30062127e+00
4.68354315e-01 -2.66651601e-01 -4.56932276e-01 4.66866523e-01
-1.37774944e-01 1.02010059e+00 1.59919038e-01 -4.20655936e-01
-4.98375237e-01 4.11108285e-01 9.33329403e-01 -1.48121953e-01
1.51166588e-01 3.55592817e-02 -8.51070046e-01 7.62137890e-01
9.01287317e-01 -2.31756911e-01 -1.87261458e-02 -1.71205804e-01
5.79453826e-01 -3.17981362e-01 1.04827560e-01 -1.28707802e+00
-9.96787176e-02 3.38392854e-01 8.49850535e-01 -2.66105175e-01
5.62210441e-01 -4.00676221e-01 -3.02142054e-01 4.30130512e-01
-1.85517401e-01 9.97832194e-02 2.41263211e-01 6.96387142e-02
-1.26867443e-01 2.74132267e-02 7.08097816e-01 8.74031633e-02
-4.37490404e-01 6.20683670e-01 -1.60366371e-01 -1.78109035e-01
7.03558505e-01 -7.31671602e-02 -3.42209667e-01 -3.93724024e-01
-6.10847294e-01 6.32483959e-02 4.42072988e-01 7.01123357e-01
3.20645869e-01 -1.46597183e+00 -8.08320880e-01 1.36518180e-01
2.16778368e-01 -4.77372706e-01 3.54052991e-01 7.38394380e-01
-3.22166294e-01 1.01073615e-01 -7.35854089e-01 -5.25869906e-01
-1.67348099e+00 3.07841688e-01 7.44637325e-02 -3.17244381e-01
-1.14238642e-01 8.05801630e-01 1.19330890e-01 -5.06751239e-01
4.87427860e-01 2.23569080e-01 -7.51026809e-01 5.02353251e-01
7.57459879e-01 4.42200899e-01 3.24878961e-01 -5.53715885e-01
-3.13931942e-01 5.46213150e-01 -3.25408727e-01 1.91892490e-01
1.75451815e+00 1.62454262e-01 -1.97820097e-01 3.19352180e-01
1.19205999e+00 -1.91900000e-01 -1.01433337e+00 1.14227541e-01
-5.42060547e-02 -5.66222548e-01 -7.43360743e-02 -6.23065889e-01
-1.40041494e+00 1.08908749e+00 1.15372789e+00 2.12426022e-01
1.08467185e+00 2.76003266e-03 6.79894209e-01 3.10003996e-01
1.61745086e-01 -1.21606636e+00 3.10485780e-01 5.18177092e-01
5.46738386e-01 -1.50557911e+00 1.73249826e-01 -3.49376321e-01
-1.42284140e-01 1.05951440e+00 7.19447315e-01 -3.45665216e-02
6.99336290e-01 1.63961977e-01 -1.63239930e-02 -4.00266163e-02
-4.49278355e-01 -2.30645463e-01 1.93275228e-01 7.31203377e-01
9.79846537e-01 -2.10599191e-02 -5.34254253e-01 6.36804104e-01
-3.28180254e-01 1.30534634e-01 3.83775741e-01 6.92783594e-01
-1.99639946e-01 -1.49278378e+00 -1.75308242e-01 7.12625504e-01
-1.20349038e+00 5.71465865e-02 -4.33722347e-01 6.74071491e-01
3.45008492e-01 6.95252061e-01 2.38061726e-01 -2.78246880e-01
5.20260073e-02 9.97660831e-02 9.01032090e-01 -5.02535403e-01
-7.98781991e-01 -4.91523057e-01 3.31926852e-01 -5.32412767e-01
-4.65573400e-01 -6.94524467e-01 -5.97919345e-01 -6.97467387e-01
1.74941540e-01 -3.36328596e-01 9.55415189e-01 8.03175449e-01
1.53550655e-01 4.89993393e-01 4.46981221e-01 -1.13774383e+00
-2.35517919e-01 -1.05517161e+00 -6.64465070e-01 5.37842214e-01
2.03507743e-03 -9.57448661e-01 -1.20825037e-01 -2.44335324e-01] | [13.523804664611816, 0.9667657017707825] |
1a3361d5-862c-4c46-add9-49f0bc396f65 | efficient-linear-attention-for-fast-and | 2204.07731 | null | https://arxiv.org/abs/2204.07731v3 | https://arxiv.org/pdf/2204.07731v3.pdf | Efficient Linear Attention for Fast and Accurate Keypoint Matching | Recently Transformers have provided state-of-the-art performance in sparse matching, crucial to realize high-performance 3D vision applications. Yet, these Transformers lack efficiency due to the quadratic computational complexity of their attention mechanism. To solve this problem, we employ an efficient linear attention for the linear computational complexity. Then, we propose a new attentional aggregation that achieves high accuracy by aggregating both the global and local information from sparse keypoints. To further improve the efficiency, we propose the joint learning of feature matching and description. Our learning enables simpler and faster matching than Sinkhorn, often used in matching the learned descriptors from Transformers. Our method achieves competitive performance with only 0.84M learnable parameters against the bigger SOTAs, SuperGlue (12M parameters) and SGMNet (30M parameters), on three benchmarks, HPatch, ETH, and Aachen Day-Night. | ['Satoshi Komorita', 'Suwichaya Suwanwimolkul'] | 2022-04-16 | null | null | null | null | ['image-matching'] | ['computer-vision'] | [-2.18914151e-01 -2.40572289e-01 -1.75286591e-01 -2.24496752e-01
-1.08307338e+00 -2.12052941e-01 6.23215973e-01 8.80585462e-02
-3.84146631e-01 1.32440150e-01 1.00971349e-01 1.55925512e-01
-2.82370567e-01 -6.15085840e-01 -8.08744967e-01 -6.19779050e-01
-9.26126838e-02 7.24636376e-01 3.71959776e-01 9.68144508e-04
2.43892118e-01 7.22656608e-01 -1.96061718e+00 1.14556320e-01
7.99326122e-01 1.40284264e+00 4.04689610e-01 3.39166373e-01
-1.30018974e-02 6.48851454e-01 -3.22188944e-01 -4.98205930e-01
5.80496073e-01 1.32600948e-01 -4.86221135e-01 -1.03716105e-01
1.34221184e+00 -2.99277365e-01 -6.22032225e-01 1.01357329e+00
7.18509376e-01 1.33972138e-01 3.56224656e-01 -1.34070528e+00
-5.83353639e-01 3.70415330e-01 -5.68114877e-01 9.01182592e-02
3.71267974e-01 5.26478775e-02 1.34179556e+00 -1.24258852e+00
6.01601422e-01 1.23598611e+00 8.13069999e-01 3.60591084e-01
-9.65364873e-01 -9.64985967e-01 1.48298487e-01 6.99776292e-01
-1.53926241e+00 -6.82744324e-01 5.59626758e-01 -3.87469754e-02
1.25474179e+00 1.34842500e-01 9.77645040e-01 8.08249176e-01
-8.17374513e-03 7.73592412e-01 6.82320297e-01 -1.26108781e-01
6.27236292e-02 -1.40744328e-01 -2.11853068e-02 9.65186834e-01
4.23025876e-01 2.01947227e-01 -9.16175783e-01 -1.39754415e-01
7.75518000e-01 2.74636865e-01 -5.20000458e-02 -6.27432823e-01
-1.32030535e+00 7.82241583e-01 9.34745550e-01 1.19991608e-01
-3.40102226e-01 4.09336060e-01 2.08625495e-01 4.70883042e-01
3.23386550e-01 4.41584706e-01 -2.76749045e-01 -8.67178291e-02
-1.01600182e+00 3.05707186e-01 6.32491708e-01 1.27328837e+00
8.64742756e-01 -1.81528032e-01 -6.16148338e-02 7.92306066e-01
1.38037667e-01 9.54292476e-01 3.11491787e-01 -1.01981413e+00
4.97700036e-01 5.31795740e-01 -9.17558894e-02 -1.02085257e+00
-3.16044837e-01 -5.26355386e-01 -1.02532995e+00 1.48457721e-01
2.36768410e-01 6.64806426e-01 -1.09161305e+00 1.27279627e+00
3.46679032e-01 4.81537879e-01 -2.39516184e-01 1.12339604e+00
9.97810602e-01 3.28023344e-01 -2.06291094e-01 3.80814940e-01
1.25570571e+00 -1.39779615e+00 -2.68893450e-01 -1.31741703e-01
4.86482292e-01 -9.06302154e-01 8.00145924e-01 3.37689936e-01
-1.15876544e+00 -6.29899859e-01 -9.75592792e-01 -6.17238104e-01
-2.00812414e-01 2.89476942e-02 1.00562072e+00 2.19093531e-01
-9.01712179e-01 5.97915053e-01 -7.99816608e-01 -3.62416863e-01
6.58347607e-01 5.77473581e-01 -6.65719748e-01 -3.08893383e-01
-7.98544049e-01 8.55166912e-01 -2.77498096e-01 -4.35757861e-02
-8.67097378e-01 -1.23994458e+00 -9.46895242e-01 3.56845826e-01
2.58422673e-01 -9.86913800e-01 8.98748636e-01 -2.90248990e-01
-1.39561951e+00 1.22014511e+00 -1.19624041e-01 -6.53103292e-01
4.46011215e-01 -6.35233104e-01 1.55920953e-01 1.50509000e-01
1.68112621e-01 9.30128634e-01 9.71097827e-01 -6.54117286e-01
-6.35817409e-01 -5.30941427e-01 1.51460068e-02 8.88696834e-02
-2.39135697e-01 -2.48919472e-01 -6.80631638e-01 -4.89596277e-01
3.99486244e-01 -1.09590304e+00 -3.99125785e-01 3.81811887e-01
-4.26288359e-02 -4.26549733e-01 6.49231911e-01 -2.42176354e-01
7.70339787e-01 -2.30295110e+00 4.20237273e-01 2.44981810e-01
4.34799939e-01 2.41007373e-01 -4.46710259e-01 1.86831549e-01
2.61502922e-01 -5.73075235e-01 2.57717967e-02 -9.18085873e-01
2.19524890e-01 1.46131963e-01 -3.38033617e-01 7.05456436e-01
-1.06051557e-01 1.15440989e+00 -7.88494885e-01 -5.47930062e-01
3.84619087e-01 4.34069484e-01 -8.97564888e-01 2.02182978e-01
5.61996289e-02 -7.66530484e-02 -3.97707671e-01 9.48419094e-01
8.14725757e-01 -4.78968531e-01 -1.96498990e-01 -5.68320274e-01
2.78615393e-02 2.61001974e-01 -1.05173159e+00 2.36151290e+00
-6.11660242e-01 4.35046256e-01 1.85494542e-01 -9.87013578e-01
9.14345860e-01 -1.91518262e-01 6.23561084e-01 -8.75442684e-01
-4.67460342e-02 4.73936737e-01 -3.63183409e-01 -1.39995381e-01
5.53789496e-01 3.24618846e-01 -1.99692566e-02 2.14978516e-01
2.99446851e-01 -3.85677963e-01 1.04154669e-01 3.28108698e-01
1.21996868e+00 -8.91830549e-02 1.14219226e-01 -2.36639574e-01
4.38659549e-01 -2.13320389e-01 4.53709096e-01 8.20685267e-01
4.53031547e-02 6.70688093e-01 1.47123441e-01 -7.54916668e-01
-9.85575080e-01 -1.02251387e+00 -2.64543127e-02 9.60604012e-01
5.40675223e-01 -7.80835330e-01 -3.64532381e-01 -6.49657369e-01
5.43547750e-01 -9.85254124e-02 -5.38726270e-01 -1.56604603e-01
-6.26311660e-01 -1.69126615e-01 3.75689000e-01 5.72573900e-01
5.01355350e-01 -9.60138321e-01 -7.46911287e-01 7.11967656e-03
1.27840921e-01 -1.40656579e+00 -7.51077533e-01 1.14368379e-01
-8.21070433e-01 -1.00359571e+00 -7.71643579e-01 -6.56193078e-01
6.71253979e-01 6.59252524e-01 1.34934330e+00 2.25831971e-01
-6.44845605e-01 4.28472161e-01 -1.08111873e-01 -2.26861164e-01
3.63002688e-01 5.82384467e-01 3.67018348e-03 -1.75144359e-01
3.67874473e-01 -7.98796356e-01 -6.33841515e-01 2.41763100e-01
-5.26358366e-01 -3.97321992e-02 9.36896205e-01 9.32909429e-01
7.85396516e-01 -6.95670426e-01 6.55034557e-02 -6.78998947e-01
-1.96541212e-02 -2.46510059e-01 -9.92990017e-01 2.24058941e-01
-5.14597893e-01 3.01594108e-01 3.94351840e-01 -4.02079493e-01
-3.94038558e-01 4.18827802e-01 -1.30319789e-01 -1.01592362e+00
1.74075216e-01 1.60397053e-01 -3.26429047e-02 -8.54662061e-01
3.72940660e-01 2.01236099e-01 1.40276586e-03 -7.59902298e-01
4.15889949e-01 3.74998063e-01 5.69471180e-01 -7.10977137e-01
1.17892623e+00 6.06408298e-01 4.87016708e-01 -6.38972044e-01
-1.05926716e+00 -7.75825322e-01 -3.89928013e-01 3.32715809e-01
4.49696839e-01 -1.21922779e+00 -7.93842077e-01 4.75984901e-01
-1.15102971e+00 -7.60379657e-02 -5.03197372e-01 3.88571441e-01
-5.98202884e-01 3.01844835e-01 -4.32619363e-01 -3.15336555e-01
-7.59872794e-01 -1.11164403e+00 1.70047736e+00 1.24708116e-01
3.10271755e-02 -4.85466093e-01 1.86709929e-02 4.34350282e-01
6.99390411e-01 -8.58106539e-02 5.76983511e-01 -4.13384199e-01
-1.37158072e+00 -2.42421627e-01 -6.23555958e-01 -1.85898505e-03
-1.05274931e-01 -4.45147961e-01 -9.10207331e-01 -4.97925818e-01
-3.60505998e-01 -3.94328326e-01 1.16788304e+00 1.52150065e-01
1.17708147e+00 -7.08725303e-02 -3.39970529e-01 1.32653570e+00
1.35153162e+00 -4.53384519e-01 5.19467771e-01 2.77559549e-01
8.32765520e-01 2.67949283e-01 6.37094915e-01 5.35992384e-01
4.96960610e-01 1.00214040e+00 6.98936284e-01 -8.60142410e-02
-5.19652128e-01 -2.71051168e-01 2.13880345e-01 9.19117689e-01
-1.12884864e-01 2.78766751e-01 -5.72648883e-01 5.64095438e-01
-1.94673944e+00 -9.33084428e-01 1.48592874e-01 2.27895331e+00
3.91934156e-01 -3.61231714e-02 -7.39154369e-02 -9.04349312e-02
2.49852791e-01 3.21140587e-01 -4.99778777e-01 1.41511500e-01
-2.47205898e-01 7.35179603e-01 6.13931417e-01 5.04819870e-01
-1.01516545e+00 1.00326943e+00 5.93033600e+00 1.03904557e+00
-1.11565697e+00 9.35299546e-02 1.71354368e-01 -4.90387559e-01
-2.51669586e-01 -1.70730427e-01 -1.04340708e+00 3.55180770e-01
5.79443455e-01 -1.07842416e-01 6.49165511e-01 9.95898664e-01
-4.57600445e-01 6.51837885e-02 -1.28459716e+00 1.61376047e+00
2.42422625e-01 -1.66514170e+00 1.21062748e-01 1.35196567e-01
7.87012458e-01 4.63746786e-01 2.42135808e-01 2.18951151e-01
-2.23387256e-02 -9.82429743e-01 6.67460203e-01 4.86532092e-01
8.00537109e-01 -5.89179575e-01 7.61017859e-01 5.22943661e-02
-1.46596789e+00 -2.33061910e-02 -7.34659076e-01 1.69591069e-01
-8.15241113e-02 8.20194900e-01 -3.59363467e-01 6.45920694e-01
8.77071857e-01 1.11330938e+00 -6.65291727e-01 1.37421250e+00
-1.24491304e-01 -1.40339166e-01 -7.83194065e-01 6.07641675e-02
2.99323380e-01 -4.77873832e-02 5.65014243e-01 9.02692437e-01
5.50251484e-01 -1.52250782e-01 2.52947956e-01 6.08227551e-01
-3.67423058e-01 -5.46192229e-02 -5.38754702e-01 2.68649548e-01
5.52668035e-01 1.32675958e+00 -3.97151232e-01 -2.72901267e-01
-4.49013859e-01 9.62166071e-01 6.12206697e-01 -4.30455282e-02
-8.22860241e-01 -3.29431146e-01 9.77183938e-01 2.37898737e-01
5.55661440e-01 -2.60920137e-01 -1.35403022e-01 -1.33864927e+00
3.24328184e-01 -8.58684659e-01 3.34799767e-01 -4.87305194e-01
-1.28652942e+00 6.09776318e-01 -2.09750786e-01 -1.12015760e+00
1.09439597e-01 -5.37529469e-01 -3.13106090e-01 6.43492103e-01
-1.74503767e+00 -1.46837974e+00 -9.20292616e-01 6.44656062e-01
4.77379084e-01 -3.42618853e-01 7.66313851e-01 8.44256222e-01
-1.63638726e-01 8.65522921e-01 -2.83399045e-01 -1.46266177e-01
8.94112468e-01 -1.16374910e+00 7.36343384e-01 3.44019741e-01
5.98744154e-01 3.53527188e-01 3.44820946e-01 -1.48898423e-01
-1.81878257e+00 -8.79915416e-01 1.19713962e+00 -4.52396899e-01
6.55374050e-01 -3.83778334e-01 -7.57414103e-01 4.09938276e-01
-1.03149138e-01 6.15349591e-01 1.84118629e-01 2.08286598e-01
-8.76553774e-01 -6.24630511e-01 -1.10793483e+00 2.77446091e-01
1.67903745e+00 -6.61146045e-01 -4.75736469e-01 4.68026370e-01
8.45043659e-01 -6.64213181e-01 -8.92751515e-01 4.11336452e-01
8.13539386e-01 -1.12199867e+00 1.36885273e+00 -3.14468771e-01
1.80122733e-01 -1.96730852e-01 -4.32621837e-01 -9.46242452e-01
-5.08219302e-01 -7.70140409e-01 -3.71543765e-01 7.80711889e-01
-1.75274312e-02 -6.65872931e-01 1.22629189e+00 1.29423752e-01
-3.14726740e-01 -9.22487855e-01 -1.18080378e+00 -8.32831681e-01
-3.51597548e-01 -3.23075242e-02 8.70646715e-01 6.35335565e-01
-3.96001130e-01 1.95417807e-01 -4.41483885e-01 4.71768901e-02
1.11745012e+00 5.58434844e-01 1.02107346e+00 -1.36717582e+00
-2.45725185e-01 -4.88642275e-01 -9.00693774e-01 -1.44540167e+00
3.04436833e-01 -8.71378899e-01 -2.76006162e-01 -1.08943474e+00
3.67232263e-01 -7.73718536e-01 -3.79002154e-01 4.72576112e-01
2.46348232e-03 5.03561318e-01 5.56003511e-01 2.62111902e-01
-1.02609754e+00 6.63357139e-01 1.17594898e+00 -3.69073302e-01
1.49652138e-01 -1.46985531e-01 -2.97580868e-01 3.84997398e-01
2.75906712e-01 -5.45139611e-01 -1.74190417e-01 -7.61325538e-01
2.01074958e-01 -1.66166127e-01 6.08196974e-01 -1.26327729e+00
7.57940412e-01 1.55079603e-01 8.93941596e-02 -8.23108017e-01
6.74923837e-01 -1.11596859e+00 2.63082236e-01 5.86572647e-01
-1.46933258e-01 1.97590917e-01 1.52204400e-02 2.87918836e-01
-4.08166826e-01 5.88304587e-02 7.54938841e-01 -9.27329883e-02
-7.35749662e-01 9.76602554e-01 4.59870309e-01 5.95799871e-02
7.62840211e-01 -3.10984924e-02 -4.16840345e-01 -1.58411562e-01
-4.75893348e-01 3.75381559e-01 5.52582383e-01 3.83865356e-01
6.95810199e-01 -1.62822449e+00 -6.66402042e-01 5.65867186e-01
2.16785550e-01 2.33271688e-01 3.79979193e-01 1.02988756e+00
-6.16783082e-01 5.07657409e-01 -4.02056277e-01 -7.24616766e-01
-1.33211243e+00 4.69689041e-01 1.96017295e-01 -4.08981353e-01
-9.03599858e-01 9.66773391e-01 1.45972908e-01 -1.81517527e-01
5.67642391e-01 -3.69290441e-01 3.27626556e-01 7.57843852e-02
4.44319159e-01 4.69004363e-01 4.07312065e-01 -4.22137201e-01
-6.40514135e-01 1.25243318e+00 -6.16434440e-02 4.13248509e-01
1.46002662e+00 2.04113632e-01 -3.81339975e-02 -2.01083466e-01
1.37587917e+00 -2.71919672e-03 -1.21504903e+00 -5.40263832e-01
1.83671119e-03 -7.66352713e-01 9.06683505e-02 -2.32044518e-01
-1.35811627e+00 8.73710871e-01 5.74158728e-01 -1.20068140e-01
9.98288989e-01 1.76661864e-01 1.26636362e+00 7.02672660e-01
7.95407116e-01 -6.35361016e-01 6.31777570e-02 4.32528079e-01
1.00919926e+00 -1.41413057e+00 3.01457047e-01 -4.58815008e-01
-9.52822492e-02 8.61098051e-01 5.10931432e-01 -5.64014375e-01
5.86096764e-01 2.96210855e-01 -1.17942147e-01 -3.05622280e-01
-9.69698966e-01 -3.98821980e-01 6.45509899e-01 6.01913929e-01
2.04057433e-02 -2.25557223e-01 1.96827352e-01 2.04697385e-01
-2.41186544e-01 -2.22338378e-01 -1.59243315e-01 5.97204626e-01
-4.32490766e-01 -1.08792186e+00 -1.37512654e-01 3.48471701e-01
-2.93404132e-01 -2.14478433e-01 -2.17353702e-01 6.48924887e-01
4.03557271e-02 3.17054212e-01 1.73012242e-01 -2.60082752e-01
5.75555682e-01 -3.71330708e-01 9.66021717e-01 -4.18998837e-01
-7.87514746e-01 -2.35872522e-01 -1.39858395e-01 -1.21718454e+00
-5.33017874e-01 -5.23892462e-01 -7.82848001e-01 -5.80386639e-01
-2.42066562e-01 2.32154086e-01 6.01425231e-01 5.81558347e-01
7.43284523e-01 1.47159949e-01 5.65257668e-01 -1.18618345e+00
-7.49757528e-01 -6.58860028e-01 -2.92918742e-01 5.16811669e-01
3.87017816e-01 -8.73977304e-01 -4.85000521e-01 -4.02370334e-01] | [8.032221794128418, -1.9441312551498413] |
a3db831c-721b-48f2-b12f-07241a78842c | deep-no-reference-tone-mapped-image-quality | 2002.03165 | null | https://arxiv.org/abs/2002.03165v1 | https://arxiv.org/pdf/2002.03165v1.pdf | Deep No-reference Tone Mapped Image Quality Assessment | The process of rendering high dynamic range (HDR) images to be viewed on conventional displays is called tone mapping. However, tone mapping introduces distortions in the final image which may lead to visual displeasure. To quantify these distortions, we introduce a novel no-reference quality assessment technique for these tone mapped images. This technique is composed of two stages. In the first stage, we employ a convolutional neural network (CNN) to generate quality aware maps (also known as distortion maps) from tone mapped images by training it with the ground truth distortion maps. In the second stage, we model the normalized image and distortion maps using an Asymmetric Generalized Gaussian Distribution (AGGD). The parameters of the AGGD model are then used to estimate the quality score using support vector regression (SVR). We show that the proposed technique delivers competitive performance relative to the state-of-the-art techniques. The novelty of this work is its ability to visualize various distortions as quality maps (distortion maps), especially in the no-reference setting, and to use these maps as features to estimate the quality score of tone mapped images. | ['Sumohana S. Channappayya', 'Sathya Veera Reddy Dendi', 'Chandra Sekhar Ravuri', 'Shanmuganathan Raman', 'Rajesh Sureddi'] | 2020-02-08 | null | null | null | null | ['tone-mapping'] | ['computer-vision'] | [ 6.33441627e-01 -1.70958325e-01 3.07707429e-01 -5.43576598e-01
-7.24468052e-01 -4.57771331e-01 7.19792962e-01 -2.42351085e-01
-1.15310907e-01 5.02915442e-01 1.33920833e-01 -1.79032728e-01
2.65431292e-02 -9.50553060e-01 -7.84154117e-01 -5.97392440e-01
2.13586971e-01 6.90395460e-02 3.44411165e-01 -3.55661541e-01
4.99331772e-01 6.09913886e-01 -1.64709437e+00 2.72146642e-01
1.05922759e+00 1.27360380e+00 1.31277010e-01 9.71636772e-01
1.09902859e-01 7.92781472e-01 -9.31618154e-01 -4.82031375e-01
3.40744317e-01 -3.48104805e-01 -5.11857152e-01 -1.54982451e-02
5.96189260e-01 -6.74711525e-01 -3.98647428e-01 1.21720850e+00
4.21174496e-01 1.60162449e-01 7.97842443e-01 -1.13172257e+00
-1.23879910e+00 2.13872595e-03 -5.52283287e-01 2.48607904e-01
4.90543664e-01 5.56362756e-02 6.20845735e-01 -9.46767330e-01
5.44268072e-01 1.34664714e+00 1.64139464e-01 3.01345825e-01
-1.38420355e+00 -5.91967344e-01 -3.04054677e-01 2.01852202e-01
-1.33181107e+00 -4.67358828e-01 9.24498379e-01 -6.15475714e-01
6.41694605e-01 5.51643550e-01 4.91466016e-01 7.04851449e-01
5.34736753e-01 3.49126130e-01 1.53419924e+00 -3.18401843e-01
3.04383099e-01 2.27541193e-01 -4.20708954e-01 5.28835893e-01
-2.26554096e-01 4.83382046e-01 -4.13571388e-01 2.13918254e-01
1.05386233e+00 -2.49372333e-01 -5.58152974e-01 -4.31388885e-01
-1.09112692e+00 6.16843224e-01 7.42565453e-01 4.02363986e-02
-3.05542588e-01 1.86123569e-02 1.56829655e-02 4.02926892e-01
6.41340494e-01 4.08577383e-01 2.35524118e-01 -1.35896847e-01
-1.05385649e+00 1.51657820e-01 3.63218635e-01 4.64337379e-01
6.92604125e-01 2.52677321e-01 -2.78688461e-01 9.83915627e-01
2.04011053e-01 5.08567035e-01 4.06284541e-01 -1.07466102e+00
4.87931401e-01 3.86096090e-01 2.21700490e-01 -1.27161169e+00
-2.47448385e-01 -3.09806049e-01 -9.55880523e-01 9.91256118e-01
2.59492904e-01 3.07046473e-01 -1.11628902e+00 1.41636539e+00
4.11553010e-02 1.03818983e-01 -1.49168568e-02 1.17479122e+00
7.92541146e-01 1.00688398e+00 -3.19489658e-01 -2.12936234e-02
8.99454772e-01 -6.99231923e-01 -8.89987350e-01 8.61895084e-02
-8.16921815e-02 -8.48527431e-01 1.56760514e+00 5.74974895e-01
-1.38442540e+00 -8.41024399e-01 -1.56106043e+00 -3.21769118e-01
-4.73435253e-01 5.74155012e-03 -1.95005223e-01 7.64952660e-01
-1.42490685e+00 6.53398395e-01 -6.00858331e-01 1.50944725e-01
2.46377826e-01 6.88427761e-02 -1.97591498e-01 6.87319506e-03
-1.18442273e+00 1.19849396e+00 1.38669936e-02 -5.67165241e-02
-7.88354158e-01 -9.77706850e-01 -9.06173050e-01 2.67773475e-02
-5.09426137e-03 -2.73476213e-01 9.66324449e-01 -1.05289292e+00
-1.88666010e+00 8.13342810e-01 5.06701842e-02 -1.30201027e-01
8.14817846e-01 -9.61051583e-02 -6.49917543e-01 7.98198655e-02
-2.07335427e-01 5.57418466e-01 1.04409933e+00 -1.63837850e+00
-4.90414828e-01 1.32047702e-02 1.88869789e-01 2.24828884e-01
5.95154911e-02 7.78615326e-02 -5.68096638e-01 -7.29182839e-01
1.19158342e-01 -5.81180751e-01 2.06912413e-01 3.31139594e-01
-5.85094273e-01 3.79227698e-01 7.89626241e-01 -9.46603537e-01
1.17191780e+00 -2.16315269e+00 1.12086341e-01 3.74784350e-01
4.20571417e-01 2.30335623e-01 -7.10846633e-02 5.11809774e-02
-2.19193861e-01 1.60385340e-01 -2.76950032e-01 -2.47124746e-01
4.58063185e-02 -2.77976662e-01 -2.36291602e-01 4.24641669e-01
4.34261024e-01 9.15161669e-01 -6.90354288e-01 -1.01721749e-01
6.59115374e-01 8.77356231e-01 -2.01608688e-01 4.35333997e-01
1.23204269e-01 4.07098114e-01 2.31560230e-01 4.32117671e-01
1.07986140e+00 -1.85182929e-01 -2.13589162e-01 -5.63133419e-01
-2.94313341e-01 -3.54155595e-03 -1.07302451e+00 1.26922059e+00
-5.31689644e-01 1.06211352e+00 -3.16325396e-01 -5.70760429e-01
1.21182239e+00 -3.05174999e-02 1.79107323e-01 -1.30353141e+00
1.11387402e-01 1.00317128e-01 -9.77457389e-02 -2.81015873e-01
7.22618997e-01 -1.71854928e-01 6.54283538e-02 4.93086427e-01
-1.82634667e-01 -3.19576383e-01 -1.30597904e-01 -4.63119484e-02
6.15300655e-01 6.05327599e-02 1.20815553e-01 -3.71652767e-02
5.46013594e-01 -4.38557774e-01 5.34937009e-02 4.77443665e-01
-5.76483943e-02 1.21057153e+00 6.53524637e-01 -5.86666226e-01
-1.60619068e+00 -1.51841760e+00 -1.61055639e-01 6.92841291e-01
4.37867343e-01 7.13584293e-03 -7.82399297e-01 -1.81308120e-01
-3.00217479e-01 6.59419715e-01 -7.02294171e-01 -3.79354656e-01
-5.56662261e-01 -5.40938854e-01 2.33801380e-01 4.39031333e-01
6.47054553e-01 -9.15223777e-01 -5.99714100e-01 -2.70818044e-02
-1.95339322e-02 -8.84023130e-01 -5.25449216e-01 -5.73832691e-02
-4.14269865e-01 -7.18966901e-01 -1.10672057e+00 -3.68614912e-01
5.47481060e-01 4.45318259e-02 1.18095374e+00 -3.02062750e-01
-2.76475757e-01 -2.51611900e-02 -1.86608538e-01 -9.42921638e-02
-6.24394000e-01 -4.56657141e-01 -8.79573300e-02 3.79522115e-01
-2.12105215e-01 -4.31110531e-01 -9.16574180e-01 4.87556279e-01
-1.17368531e+00 2.89923817e-01 4.80408967e-01 5.48932195e-01
9.40219402e-01 1.62663877e-01 2.06913531e-01 -6.18950367e-01
9.50769901e-01 -9.34688225e-02 -8.06435585e-01 3.65244687e-01
-6.27823770e-01 3.76462638e-02 7.19281197e-01 -4.96532291e-01
-1.15856850e+00 -3.73399973e-01 -9.37898830e-02 -5.03192604e-01
4.84453663e-02 2.23068550e-01 -2.32478946e-01 -3.58178735e-01
8.55336845e-01 1.63567871e-01 -1.28685683e-01 -4.79089677e-01
4.74227428e-01 7.09426284e-01 9.79273379e-01 -3.64973024e-02
1.07493699e+00 4.02141452e-01 2.87652332e-02 -4.53993618e-01
-4.61708099e-01 1.63233414e-01 -5.90383053e-01 -4.73516434e-01
8.23411703e-01 -7.04848647e-01 -5.57152271e-01 5.36558688e-01
-7.52780318e-01 -5.00363171e-01 -1.10300586e-01 1.54817104e-01
-7.41777718e-01 1.00787781e-01 -4.06256646e-01 -6.82158411e-01
-1.93828836e-01 -1.30788243e+00 1.06660128e+00 5.06600261e-01
1.37768701e-01 -7.97096312e-01 1.22815609e-01 -8.03841278e-03
8.52704823e-01 5.53022563e-01 1.00982308e+00 4.55406457e-02
-7.15180039e-01 -1.30952716e-01 -6.32807136e-01 5.61612010e-01
4.27293420e-01 1.37691543e-01 -1.04197288e+00 -2.25205645e-01
-7.43151829e-02 -8.99855271e-02 5.69380343e-01 6.39486909e-01
1.32603312e+00 -1.79681793e-01 2.90170521e-01 1.08733404e+00
1.60248148e+00 4.66139466e-01 1.09233820e+00 5.09605169e-01
6.61514878e-01 3.65358531e-01 6.39325261e-01 1.82939678e-01
2.00950906e-01 1.11935318e+00 3.77533019e-01 -5.90170145e-01
-3.87181938e-01 -2.63191730e-01 1.90242186e-01 4.86847937e-01
-8.41256008e-02 -2.59331107e-01 -7.17705429e-01 9.89931449e-03
-1.26420021e+00 -7.43836939e-01 1.31615609e-01 2.29389977e+00
6.90716565e-01 2.28747696e-01 7.50041083e-02 3.88118535e-01
8.26681674e-01 2.42886797e-01 -5.94124138e-01 -8.41937184e-01
-1.42396435e-01 2.12647051e-01 2.55506039e-01 6.68215454e-01
-9.18779492e-01 5.99883139e-01 6.63508224e+00 5.59675395e-01
-1.63475680e+00 -1.53348804e-01 9.62462425e-01 -1.08669922e-01
-3.64150673e-01 -5.41987002e-01 -1.31319448e-01 5.98603666e-01
1.02322137e+00 -4.33252275e-01 8.04184556e-01 5.23246348e-01
4.31559414e-01 -7.64257982e-02 -8.70240331e-01 1.24181271e+00
3.05001140e-01 -1.13777781e+00 1.43921733e-01 -8.26191232e-02
8.49687755e-01 -4.52236682e-01 8.09461057e-01 -1.12525590e-01
5.09862937e-02 -1.27006984e+00 9.01008248e-01 9.92136896e-01
1.58452010e+00 -9.23735619e-01 6.02553606e-01 -1.48694187e-01
-9.41494703e-01 1.29032016e-01 -6.11932039e-01 2.66149879e-01
3.50071311e-01 4.89611864e-01 -5.73572099e-01 3.89577836e-01
7.65872002e-01 3.93144786e-01 -8.31564486e-01 1.10950863e+00
2.48037204e-02 7.77117163e-02 8.93386602e-02 4.34733868e-01
-1.61670387e-01 -2.27941275e-01 3.03176731e-01 9.82749343e-01
6.51024878e-01 -1.09784171e-01 -3.08514327e-01 1.24703002e+00
-1.27838105e-01 -1.19023226e-01 -3.48032534e-01 1.72145098e-01
2.11371094e-01 1.10810411e+00 -4.99013931e-01 -3.58178943e-01
-2.59945095e-01 1.22901309e+00 -3.10355029e-03 4.63299394e-01
-8.70157659e-01 -6.70998037e-01 5.11465311e-01 1.71521559e-01
4.45509478e-02 1.25353606e-02 -4.38728511e-01 -9.74981308e-01
3.00168574e-01 -7.90781498e-01 -1.94252014e-01 -1.41305399e+00
-1.00540590e+00 1.08065867e+00 1.51017457e-02 -1.56167531e+00
-2.84715950e-01 -6.08750522e-01 -5.39628744e-01 1.31883669e+00
-1.74457431e+00 -8.50239754e-01 -8.09753060e-01 5.58772802e-01
3.32915932e-01 1.78087465e-02 5.79773068e-01 3.94432902e-01
-2.03248352e-01 6.73386216e-01 2.78489321e-01 -5.85015751e-02
7.78840363e-01 -1.55281270e+00 7.55661488e-01 8.53358626e-01
-1.18938528e-01 1.79880589e-01 8.45191598e-01 -4.51638371e-01
-8.94532263e-01 -1.18229055e+00 4.43141073e-01 -3.58055979e-01
2.63254791e-01 -5.10306120e-01 -1.10115659e+00 2.24718809e-01
-3.07596270e-02 2.35844567e-01 4.00220096e-01 -5.88531911e-01
-3.53228509e-01 -3.65230262e-01 -1.33971453e+00 5.61634541e-01
5.68830967e-01 -7.41841972e-01 -3.10621411e-01 -2.43319333e-01
7.54741728e-01 -6.94650233e-01 -9.16286290e-01 3.13824832e-01
7.38923192e-01 -1.36511803e+00 1.12560380e+00 1.14090079e-02
8.60164464e-01 -5.60227573e-01 -2.10300535e-01 -1.70869291e+00
-4.62917119e-01 -4.03830945e-01 -1.29473299e-01 1.11262226e+00
5.06050527e-01 -4.35666740e-01 3.84034336e-01 5.39562762e-01
-5.76350689e-02 -6.04985595e-01 -8.41668785e-01 -7.01835811e-01
4.65775952e-02 -2.41678551e-01 8.86848092e-01 5.67894995e-01
-4.31266695e-01 -2.63802409e-01 -6.44506633e-01 1.44060209e-01
6.51007891e-01 -9.25143063e-02 4.68641549e-01 -9.95134950e-01
-1.29475966e-01 -4.27167684e-01 -5.88131487e-01 -7.35949099e-01
-2.99477577e-01 -5.50346673e-01 1.12672515e-01 -1.43331838e+00
3.89220379e-02 -3.06067526e-01 -4.69246000e-01 -1.10258326e-01
-2.11748451e-01 8.41000199e-01 3.67146403e-01 1.03148118e-01
-2.45472074e-01 5.50977468e-01 1.55950260e+00 -1.44712299e-01
-3.09747845e-01 -1.40472278e-01 -6.54889941e-01 3.29882175e-01
6.84890389e-01 -1.39392972e-01 -4.42541867e-01 -4.26336199e-01
3.05671662e-01 1.99857607e-01 4.23121572e-01 -1.24558103e+00
9.61164199e-03 -4.18517925e-02 8.98857176e-01 -4.98249739e-01
4.30501103e-01 -6.43644035e-01 4.50998545e-01 1.12509422e-01
-6.15114808e-01 2.59895951e-01 -4.61555868e-02 2.70642877e-01
-3.43339235e-01 1.41616419e-01 1.28970325e+00 2.45199025e-01
-3.63710970e-01 1.65002063e-01 -1.26247361e-01 -2.88121283e-01
8.28902841e-01 -3.90296906e-01 -5.03613830e-01 -7.12325394e-01
-5.70386827e-01 -4.29590046e-01 7.87290990e-01 5.83981812e-01
1.01429451e+00 -1.71609282e+00 -6.65610313e-01 4.82445538e-01
7.35819116e-02 -3.24223131e-01 4.59739476e-01 4.27216589e-01
-7.95151949e-01 1.37800723e-01 -6.45213664e-01 -6.19230688e-01
-1.04637277e+00 7.63939261e-01 5.33448756e-01 -4.60417755e-02
-6.10829353e-01 5.33858478e-01 4.74500805e-01 -8.26703608e-02
1.05148382e-01 -3.78412932e-01 -4.29134965e-01 -3.56875241e-01
9.21922147e-01 4.02282506e-01 2.43773237e-01 -8.50686431e-01
-6.90203831e-02 8.77641022e-01 9.02727023e-02 -4.52019840e-01
1.15265620e+00 -3.67263168e-01 2.08286777e-01 4.83929008e-01
1.45388103e+00 -2.17786372e-01 -1.52982068e+00 -9.30097327e-02
-3.61127704e-01 -1.07738161e+00 2.76591748e-01 -1.20266521e+00
-1.26505125e+00 9.76648211e-01 1.32756400e+00 3.67903054e-01
1.47331011e+00 -3.02547872e-01 4.51473802e-01 -2.28652120e-01
6.35460466e-02 -9.98938143e-01 2.49339849e-01 9.67222899e-02
1.25387585e+00 -1.12630808e+00 -2.24239931e-01 -3.33722085e-01
-6.89367771e-01 1.12162805e+00 4.97188330e-01 -3.16257149e-01
4.50074852e-01 2.77869940e-01 4.82482046e-01 -7.33879209e-03
-4.99849498e-01 1.32957473e-01 7.90027559e-01 9.19099927e-01
2.50177413e-01 1.46020958e-02 7.97071010e-02 1.25588313e-01
-6.34561479e-01 -2.01086123e-02 7.92588532e-01 5.47396243e-01
-3.30346912e-01 -7.45697439e-01 -5.22058904e-01 4.56355572e-01
-4.37886715e-01 2.02257317e-02 -2.15624899e-01 4.01431173e-01
1.45180421e-02 9.90487278e-01 2.74151325e-01 -7.17564642e-01
6.59037828e-01 -2.83630133e-01 3.42099994e-01 -8.10617059e-02
-2.59296209e-01 1.70462430e-02 -3.24370801e-01 -8.86611462e-01
-1.65115193e-01 -9.81955826e-02 -5.94629765e-01 -4.26483750e-01
7.62911106e-04 -4.17142063e-01 8.56383324e-01 4.95381236e-01
1.15897637e-02 8.89452636e-01 9.11700964e-01 -1.07332778e+00
-1.38844252e-01 -7.24205256e-01 -7.02461302e-01 6.35440230e-01
7.48338282e-01 -7.29934752e-01 -4.05705124e-01 1.90452233e-01] | [11.039560317993164, -2.2052266597747803] |
8731860f-b77c-4e04-b7e6-189b5a799c4b | codet5-identifier-aware-unified-pre-trained | 2109.00859 | null | https://arxiv.org/abs/2109.00859v1 | https://arxiv.org/pdf/2109.00859v1.pdf | CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation | Pre-trained models for Natural Languages (NL) like BERT and GPT have been recently shown to transfer well to Programming Languages (PL) and largely benefit a broad set of code-related tasks. Despite their success, most current methods either rely on an encoder-only (or decoder-only) pre-training that is suboptimal for generation (resp. understanding) tasks or process the code snippet in the same way as NL, neglecting the special characteristics of PL such as token types. We present CodeT5, a unified pre-trained encoder-decoder Transformer model that better leverages the code semantics conveyed from the developer-assigned identifiers. Our model employs a unified framework to seamlessly support both code understanding and generation tasks and allows for multi-task learning. Besides, we propose a novel identifier-aware pre-training task that enables the model to distinguish which code tokens are identifiers and to recover them when they are masked. Furthermore, we propose to exploit the user-written code comments with a bimodal dual generation task for better NL-PL alignment. Comprehensive experiments show that CodeT5 significantly outperforms prior methods on understanding tasks such as code defect detection and clone detection, and generation tasks across various directions including PL-NL, NL-PL, and PL-PL. Further analysis reveals that our model can better capture semantic information from code. Our code and pre-trained models are released at https: //github.com/salesforce/CodeT5 . | ['Steven C. H. Hoi', 'Shafiq Joty', 'Weishi Wang', 'Yue Wang'] | 2021-09-02 | null | https://aclanthology.org/2021.emnlp-main.685 | https://aclanthology.org/2021.emnlp-main.685.pdf | emnlp-2021-11 | ['code-translation', 'text-to-code-generation'] | ['computer-code', 'computer-code'] | [ 2.93867081e-01 1.30791426e-01 -5.54036021e-01 -2.38769144e-01
-1.17211306e+00 -6.73288703e-01 3.34919661e-01 2.68682867e-01
2.10227340e-01 1.43734753e-01 2.26116836e-01 -7.64140248e-01
4.99150395e-01 -6.23321354e-01 -9.60728884e-01 3.20196035e-03
2.33306676e-01 2.58323342e-01 1.42624183e-02 -1.83338076e-01
4.13164705e-01 -4.93998230e-01 -1.52313542e+00 7.74156868e-01
1.43486190e+00 5.74306786e-01 6.12607896e-01 6.83024049e-01
-4.10394281e-01 9.56464171e-01 -4.35956657e-01 -8.36088538e-01
1.46456853e-01 -2.60683745e-01 -9.18835282e-01 -1.88679352e-01
3.87237400e-01 -2.88911074e-01 -2.29902193e-02 1.08024466e+00
1.57133177e-01 -7.15748429e-01 4.03332651e-01 -1.32861245e+00
-1.15505743e+00 1.09890318e+00 -7.82733321e-01 -3.07868510e-01
5.81737220e-01 2.21446216e-01 1.37607074e+00 -9.75250661e-01
6.23602152e-01 1.09868753e+00 9.86907244e-01 6.99752986e-01
-1.26333296e+00 -7.38512278e-01 -9.81132761e-02 -3.76815014e-02
-1.13206935e+00 -3.44966799e-01 3.57486427e-01 -8.75811398e-01
1.31608796e+00 -1.93625063e-01 3.55911702e-02 1.23972154e+00
2.19208017e-01 9.60229993e-01 5.97155392e-01 -4.86995488e-01
-2.26365328e-01 1.73496500e-01 2.31672153e-01 1.05771983e+00
2.40735054e-01 -2.12202966e-01 -4.50212777e-01 -3.20304811e-01
3.27118367e-01 2.75519248e-02 -3.28328490e-01 -3.94212455e-01
-1.16661000e+00 9.32752192e-01 8.82391557e-02 5.41845933e-02
8.71195942e-02 3.64026427e-01 7.93863118e-01 4.68829513e-01
3.89574587e-01 5.75867414e-01 -8.13525736e-01 -5.12534559e-01
-9.56315279e-01 7.06409961e-02 9.04653311e-01 1.72544611e+00
1.10846233e+00 7.24093709e-03 -2.89283574e-01 9.73087788e-01
3.83514494e-01 4.23677742e-01 6.40121222e-01 -6.67415142e-01
1.03593707e+00 9.45210040e-01 -2.45398834e-01 -5.94501972e-01
1.23182341e-01 -5.19975424e-01 -3.74469936e-01 -2.45914549e-01
2.77143002e-01 -1.35068774e-01 -7.47926295e-01 1.74193847e+00
-3.96600574e-01 -2.75005605e-02 7.76829049e-02 1.88617364e-01
6.90323114e-01 4.05452251e-01 -4.78787422e-02 4.69054252e-01
1.43717873e+00 -1.25192213e+00 -2.04755083e-01 -6.52405798e-01
1.34969056e+00 -8.81919861e-01 1.28799558e+00 1.52428195e-01
-8.95983160e-01 -6.32520378e-01 -8.32445562e-01 -4.07557964e-01
-1.68138355e-01 7.71719933e-01 6.82452440e-01 7.49083757e-01
-1.23934329e+00 2.00728178e-01 -8.63763928e-01 -2.89446890e-01
5.47233760e-01 7.62484372e-02 -3.06128621e-01 -2.89913237e-01
-8.10744047e-01 6.41696155e-01 3.42220336e-01 -4.09168363e-01
-1.08002281e+00 -1.04128015e+00 -1.33757734e+00 2.57801741e-01
2.49810487e-01 -7.51746655e-01 1.66564572e+00 -1.06452966e+00
-1.17535412e+00 1.08281958e+00 -5.19123256e-01 -4.89238381e-01
2.96602547e-01 -3.60665470e-01 -1.21128149e-01 -1.90532088e-01
4.98050153e-01 5.71210265e-01 6.72201991e-01 -1.21863937e+00
-5.77646554e-01 6.96261600e-02 1.99409977e-01 -2.71426439e-01
-4.17851955e-01 2.52329648e-01 -5.01759648e-01 -6.72688961e-01
-2.81165332e-01 -8.91659200e-01 1.85391635e-01 -8.57689828e-02
-5.88873029e-01 -2.95451850e-01 6.40831709e-01 -1.04340315e+00
1.44990027e+00 -2.25941801e+00 6.56568408e-02 -2.25995839e-01
3.13215375e-01 3.22721958e-01 -5.36034226e-01 7.45783210e-01
-2.37783983e-01 3.69487435e-01 -5.38593531e-01 -7.59721339e-01
2.24690795e-01 1.48290440e-01 -5.96580148e-01 -1.83107164e-02
5.48292100e-01 1.11027479e+00 -9.76435542e-01 -3.10371876e-01
-1.95614710e-01 1.41808003e-01 -1.21902573e+00 2.71293402e-01
-6.28918827e-01 2.35998333e-01 -3.49214166e-01 7.70677626e-01
6.73281848e-01 -4.73563194e-01 1.19464122e-01 3.42495203e-01
-8.63705501e-02 8.11793745e-01 -5.43214321e-01 2.13829684e+00
-1.09752691e+00 6.65967643e-01 3.81580479e-02 -7.32114077e-01
8.88265252e-01 3.69042933e-01 -8.19042772e-02 -4.66012657e-01
-2.21867993e-01 5.68847954e-01 -4.49255332e-02 -7.15618849e-01
7.29036570e-01 2.56505996e-01 -3.72496396e-01 7.77545154e-01
2.88373291e-01 8.20520669e-02 2.51598358e-01 5.53148925e-01
1.41763842e+00 3.86136860e-01 2.51724899e-01 -8.59575048e-02
4.39728767e-01 -7.99152553e-02 4.94019002e-01 8.25027525e-01
3.70009422e-01 7.72263885e-01 8.81968439e-01 1.25456408e-01
-1.02478945e+00 -8.21685016e-01 1.47128981e-02 1.24975109e+00
-6.17052913e-02 -9.73282337e-01 -7.68742561e-01 -1.05558491e+00
3.50412950e-02 8.94755602e-01 -4.32124406e-01 -4.33731705e-01
-6.45041525e-01 -4.59554523e-01 8.04755569e-01 7.16648579e-01
1.66420832e-01 -7.75879383e-01 -1.88782424e-01 1.88655555e-01
-4.50762630e-01 -9.90050435e-01 -6.98880911e-01 1.67493597e-01
-5.77636898e-01 -1.09842825e+00 -2.68506467e-01 -9.80300128e-01
8.76569033e-01 1.34630144e-01 1.42953646e+00 6.28747284e-01
-1.31736979e-01 2.76904166e-01 -6.69390857e-01 -9.32162330e-02
-1.06946969e+00 5.34503162e-01 -6.85945213e-01 -3.45646292e-01
4.30904031e-01 -5.65242946e-01 -2.53403991e-01 1.68542668e-01
-7.84949958e-01 3.63582253e-01 8.79911780e-01 9.23079669e-01
-1.18788525e-01 -4.21430111e-01 5.46619892e-01 -1.21578705e+00
4.46588635e-01 -9.08574820e-01 -5.55750012e-01 5.01583874e-01
-5.24141967e-01 4.42567289e-01 7.37509251e-01 -2.79451221e-01
-1.19814074e+00 -1.46714836e-01 -4.48527873e-01 -1.20468549e-01
1.03279287e-02 6.76077724e-01 -2.32119203e-01 5.37646748e-02
5.65958142e-01 3.55114043e-01 -1.81654379e-01 -6.63200021e-01
3.61325562e-01 9.76079345e-01 5.16931951e-01 -1.07025635e+00
9.48949575e-01 2.71017980e-02 -7.62223661e-01 -1.47781670e-01
-5.78630209e-01 -5.10778844e-01 -4.85603571e-01 4.44323778e-01
6.23374283e-01 -1.30288661e+00 -2.43648469e-01 6.38486564e-01
-1.67716098e+00 -4.36336875e-01 2.20532194e-02 4.61877286e-02
-4.51864690e-01 5.15375495e-01 -9.12740350e-01 -3.99069220e-01
-3.09639871e-01 -1.48187852e+00 1.55206156e+00 -1.87828988e-01
-3.28272790e-01 -1.00634623e+00 -8.19304958e-02 4.85433310e-01
5.59418976e-01 -2.55838513e-01 1.52361059e+00 -6.71247303e-01
-9.90987420e-01 -5.19298874e-02 -5.94299912e-01 4.39737588e-01
1.07737437e-01 8.92877281e-02 -9.22470033e-01 -2.70009011e-01
-1.71247214e-01 -4.15063411e-01 9.78197813e-01 -2.36157835e-01
1.16248536e+00 -4.23617154e-01 -4.69188333e-01 8.41167510e-01
1.54024851e+00 -8.65762904e-02 6.56914353e-01 1.17766209e-01
9.22067821e-01 4.07636046e-01 2.08582103e-01 4.27779496e-01
8.95995557e-01 5.50572455e-01 5.29475152e-01 1.74734011e-01
-3.61616373e-01 -7.61029959e-01 7.66535044e-01 1.14066160e+00
4.32240158e-01 -2.99428165e-01 -1.11824751e+00 8.89396131e-01
-1.66871691e+00 -6.12307310e-01 -3.49226683e-01 2.05734873e+00
1.32001364e+00 -1.23151325e-01 -2.63432294e-01 -3.04226279e-01
6.76432848e-01 -1.22199781e-01 -2.94604152e-01 -4.42856699e-01
1.43673822e-01 3.35253596e-01 4.42229301e-01 3.91852438e-01
-7.68898845e-01 9.01055872e-01 5.69762135e+00 9.78270590e-01
-8.56376529e-01 5.32658458e-01 1.81165084e-01 4.81651038e-01
-8.13977897e-01 5.15440643e-01 -9.87447977e-01 7.08895445e-01
8.80502760e-01 -3.52876067e-01 4.09112066e-01 1.16613591e+00
-3.20335627e-01 2.64717396e-02 -1.52913749e+00 6.91289723e-01
1.57954112e-01 -1.20499492e+00 8.93573835e-02 -4.16967757e-02
8.90752792e-01 2.75118560e-01 8.67308825e-02 7.66734004e-01
6.49176836e-01 -8.99011850e-01 1.07221913e+00 1.16311878e-01
1.05205894e+00 -1.88921720e-01 7.02256978e-01 3.39591682e-01
-1.21153605e+00 -3.36379647e-01 -2.77786702e-01 -1.71900600e-01
-8.64390209e-02 7.11594522e-01 -9.99535918e-01 7.67549753e-01
2.45654583e-01 1.25262868e+00 -9.22465026e-01 1.06704009e+00
-6.69550180e-01 6.88806713e-01 2.17784658e-01 3.30321878e-01
1.32759679e-02 1.84222728e-01 3.46875846e-01 1.49029458e+00
8.15851212e-01 -6.90740049e-01 1.14557818e-01 1.52155924e+00
-4.20898736e-01 -1.14341378e-01 -5.41095257e-01 -2.05912456e-01
6.00884795e-01 1.09216034e+00 -3.49854529e-01 -3.45728099e-01
-9.86366689e-01 9.33620274e-01 4.32770073e-01 3.07450771e-01
-8.57337356e-01 -5.23908257e-01 8.45119834e-01 8.48731920e-02
4.66422856e-01 -1.97323650e-01 -2.66935736e-01 -1.52263045e+00
2.42012352e-01 -1.13593888e+00 1.49588212e-01 -7.39923239e-01
-1.13841534e+00 5.38393080e-01 -1.21205464e-01 -1.22816598e+00
-4.00335670e-01 -6.41666353e-01 -7.12334633e-01 8.59986067e-01
-1.72408593e+00 -1.29887736e+00 -1.00029811e-01 1.50886923e-01
6.61914706e-01 -1.23684503e-01 6.80956483e-01 5.72899044e-01
-4.99766380e-01 9.93406594e-01 8.05750303e-03 4.88924354e-01
7.57480502e-01 -1.38774371e+00 8.51144135e-01 1.25823021e+00
-2.56671645e-02 1.08949888e+00 3.97340000e-01 -7.32871771e-01
-1.44081974e+00 -1.36805201e+00 1.10801363e+00 -8.36887598e-01
8.36428821e-01 -7.98580408e-01 -9.68688369e-01 1.05674553e+00
3.03406596e-01 -3.31527323e-01 5.00305593e-01 1.60956725e-01
-1.01625633e+00 3.35053295e-01 -6.01417124e-01 3.54783684e-01
1.06550348e+00 -1.08962965e+00 -5.03110111e-01 2.92745233e-01
1.03514254e+00 -6.21120393e-01 -6.91392601e-01 6.89911246e-02
2.83951521e-01 -9.18246984e-01 6.89189017e-01 -1.87685817e-01
1.07708168e+00 -3.13036501e-01 -2.17611697e-02 -1.38582361e+00
6.49955869e-02 -6.71879888e-01 -1.81242675e-02 1.48466587e+00
6.92608476e-01 -7.22722411e-01 4.69110757e-01 4.08718288e-01
-7.17296779e-01 -4.65927184e-01 -4.77653891e-01 -7.82202303e-01
1.21802062e-01 -5.55413008e-01 7.65828073e-01 8.56186569e-01
3.01106423e-01 1.17771968e-01 -3.16568613e-01 1.17834672e-01
3.06491852e-01 3.45767945e-01 8.09400678e-01 -1.07039237e+00
-5.99089324e-01 -4.15510535e-01 -1.01428360e-01 -1.39033020e+00
5.65098166e-01 -1.50642049e+00 2.08923042e-01 -1.46190941e+00
5.48742294e-01 -6.48028135e-01 2.64939010e-01 9.91326451e-01
-3.33303481e-01 -1.62036404e-01 -5.55032566e-02 2.21335873e-01
-5.12354910e-01 4.60069031e-01 7.46355236e-01 -1.76609084e-01
2.26989314e-01 2.43742000e-02 -1.03675842e+00 4.08696622e-01
5.87940931e-01 -7.59339929e-01 -3.32183897e-01 -1.12178290e+00
6.36957169e-01 7.08468780e-02 4.13008600e-01 -7.63761163e-01
1.20584100e-01 3.40777814e-01 -3.93414825e-01 -1.34249762e-01
-2.57928669e-01 -4.79498833e-01 -9.01133046e-02 4.19179142e-01
-4.71924245e-01 1.05566271e-01 2.19149992e-01 4.28126007e-01
-2.88633168e-01 -7.36757576e-01 4.30319488e-01 -2.50235349e-01
-6.04142010e-01 2.11064085e-01 -2.56039232e-01 4.44839329e-01
6.17596805e-01 -1.52905390e-01 -8.91005814e-01 -1.85928389e-01
-3.94217893e-02 2.90866077e-01 9.37321782e-01 7.76990235e-01
3.45563829e-01 -9.51624215e-01 -6.96110249e-01 4.11945164e-01
6.87414885e-01 -6.16505519e-02 1.66081920e-01 9.48133171e-01
-4.22801942e-01 3.85878325e-01 -5.97383566e-02 -5.23111701e-01
-1.13319850e+00 4.93477553e-01 1.32090822e-01 -4.94620323e-01
-4.13527220e-01 8.84689510e-01 5.19973099e-01 -8.48364174e-01
-9.39413458e-02 -7.40208030e-01 3.31924498e-01 -3.38967711e-01
3.99953574e-01 -2.11217976e-03 1.83594003e-01 -4.08891797e-01
-1.62210822e-01 4.39518452e-01 -2.99623966e-01 4.84534532e-01
1.33710206e+00 3.11135575e-02 -5.97762167e-01 3.17970440e-02
1.39067531e+00 1.98398232e-01 -1.12249660e+00 -4.40847844e-01
3.68874311e-01 -4.96133566e-01 -4.73601013e-01 -8.16375315e-01
-1.01166725e+00 1.13659990e+00 -1.32248461e-01 -9.38126631e-03
9.38736618e-01 1.98925972e-01 9.83340561e-01 1.69685930e-01
5.95401645e-01 -5.65617561e-01 2.30932131e-01 7.80133367e-01
5.76647937e-01 -1.09235156e+00 -5.96832931e-01 -6.79023385e-01
-4.67337728e-01 1.13545966e+00 7.74969041e-01 2.81254590e-01
1.90292239e-01 6.99142694e-01 -1.86713040e-01 5.95154278e-02
-1.05246198e+00 -9.75401774e-02 1.09345987e-01 6.74604535e-01
7.99207330e-01 -1.50291741e-01 5.01531437e-02 8.53821814e-01
-2.37789959e-01 -5.19127911e-03 8.02437544e-01 9.59641039e-01
-1.57281265e-01 -1.69032931e+00 -1.89875942e-02 5.94293118e-01
-4.31266159e-01 -7.49013424e-01 -9.21726450e-02 4.77447361e-01
3.18272710e-01 8.40029955e-01 -1.95672363e-01 -3.92551541e-01
-1.76168326e-02 3.04726541e-01 2.17156500e-01 -1.43760324e+00
-8.28873158e-01 -4.17550832e-01 7.45373741e-02 -4.81996864e-01
-5.68412393e-02 -6.31833851e-01 -1.13035917e+00 -2.08941400e-01
-4.67104912e-01 1.71698019e-01 4.23946023e-01 7.13354111e-01
8.03599298e-01 5.59634566e-01 3.42766851e-01 -3.88528585e-01
-4.07900333e-01 -8.95480752e-01 -2.02050120e-01 4.11612570e-01
3.66319746e-01 -3.77011150e-01 -3.54726076e-01 3.66244078e-01] | [7.671043395996094, 7.9166951179504395] |
fae3dbb0-205d-4e5d-95a3-e098de1a7dda | unsupervised-domain-adaptation-by-learning | 2303.09350 | null | https://arxiv.org/abs/2303.09350v2 | https://arxiv.org/pdf/2303.09350v2.pdf | Unsupervised domain adaptation by learning using privileged information | Successful unsupervised domain adaptation (UDA) is guaranteed only under strong assumptions such as covariate shift and overlap between input domains. The latter is often violated in high-dimensional applications such as image classification which, despite this challenge, continues to serve as inspiration and benchmark for algorithm development. In this work, we show that access to side information about examples from the source and target domains can help relax these assumptions and increase sample efficiency in learning, at the cost of collecting a richer variable set. We call this domain adaptation by learning using privileged information (DALUPI). Tailored for this task, we propose a simple two-stage learning algorithm inspired by our analysis and a practical end-to-end algorithm for multi-label image classification. In a suite of experiments, including an application to medical image analysis, we demonstrate that incorporating privileged information in learning can reduce errors in domain transfer compared to classical learning. | ['Fredrik D. Johansson', 'Anton Matsson', 'Adam Breitholtz'] | 2023-03-16 | null | null | null | null | ['multi-label-image-classification'] | ['computer-vision'] | [ 9.39936399e-01 2.58951753e-01 -5.39952874e-01 -6.29505932e-01
-9.76473927e-01 -7.74417400e-01 5.61615646e-01 4.19764876e-01
-6.86468720e-01 1.05966806e+00 -3.97675633e-02 -3.39053184e-01
-2.57168680e-01 -4.10562247e-01 -8.44574630e-01 -7.60465682e-01
9.60085019e-02 7.57093906e-01 3.17834839e-02 2.45177642e-01
-1.80780366e-02 4.91883695e-01 -1.24443793e+00 3.56213897e-01
8.56875837e-01 6.48996174e-01 -1.43937441e-02 2.83412367e-01
3.99836823e-02 4.52719837e-01 -3.02872241e-01 -1.74770892e-01
2.44980440e-01 -4.16246086e-01 -1.00789201e+00 1.34742156e-01
3.55023444e-01 -2.77088612e-01 1.92032605e-01 9.02305543e-01
6.49129629e-01 2.81178206e-01 1.07767475e+00 -1.30171525e+00
-2.10379109e-01 3.27197999e-01 -4.53218758e-01 -1.22519195e-01
-1.42275146e-03 -6.22521602e-02 9.64832425e-01 -6.18479967e-01
8.49451780e-01 9.08237994e-01 7.91841090e-01 9.35313404e-01
-1.63942099e+00 -7.54998147e-01 6.36669248e-02 -1.05309598e-01
-1.06378186e+00 -3.79600644e-01 8.12793195e-01 -6.87331259e-01
4.48174566e-01 1.18412234e-01 -1.61968160e-03 1.27388763e+00
-1.95656091e-01 1.04650068e+00 1.35065150e+00 -6.06005609e-01
5.50963342e-01 7.38774896e-01 2.56171823e-01 3.91452342e-01
1.74457639e-01 1.22530892e-01 -3.38810742e-01 -5.50438404e-01
4.23892409e-01 -1.11201212e-01 -1.36208415e-01 -1.10011256e+00
-9.51736867e-01 1.08812845e+00 2.86601782e-01 3.92854884e-02
-2.53398865e-01 -4.55532670e-01 5.26451886e-01 5.19766748e-01
5.95050335e-01 4.61273909e-01 -7.87675261e-01 1.10930882e-01
-6.77511275e-01 3.38103235e-01 7.16484368e-01 7.00502157e-01
7.52211690e-01 -4.11253452e-01 -4.87783141e-02 1.05150998e+00
5.92235178e-02 3.48018080e-01 6.02189958e-01 -7.11148202e-01
1.58024326e-01 4.38575298e-01 8.15260857e-02 -1.89736709e-01
-5.66400409e-01 -4.39610809e-01 -7.77713656e-01 4.23701018e-01
7.98930705e-01 -1.64312258e-01 -1.00737154e+00 2.13564539e+00
6.32638037e-01 1.62266329e-01 9.26669315e-02 6.17278278e-01
1.77953675e-01 5.33751883e-02 4.46056873e-01 -4.08815980e-01
1.22317159e+00 -4.71300751e-01 -4.27638501e-01 -2.42594227e-01
1.08841515e+00 -4.80488837e-01 1.17728126e+00 4.54430878e-01
-7.60124981e-01 -2.91893035e-01 -9.01905060e-01 -8.71412829e-02
-4.67442065e-01 -2.30826035e-01 6.06463492e-01 7.78490186e-01
-5.27390122e-01 6.60022497e-01 -6.14730716e-01 -4.75263655e-01
1.01600373e+00 5.45249104e-01 -5.32703817e-01 -3.74191523e-01
-1.15282440e+00 8.31141055e-01 2.96457052e-01 -6.75393462e-01
-6.95965946e-01 -1.08960879e+00 -7.99363017e-01 -1.40108794e-01
5.50279021e-01 -8.06025922e-01 1.30842006e+00 -1.10892689e+00
-1.37151945e+00 1.25940239e+00 7.19360784e-02 -6.48350537e-01
7.82934487e-01 -3.21607202e-01 -6.47755563e-02 -1.78443536e-03
9.54776853e-02 6.76375031e-01 9.79675472e-01 -1.06411195e+00
-5.96469462e-01 -6.10232830e-01 -1.48003578e-01 2.83528179e-01
-4.42824662e-01 1.27114495e-03 1.53068770e-02 -5.65388978e-01
-3.16267461e-01 -1.02760279e+00 -3.03627461e-01 2.39956513e-01
-1.29108712e-01 -1.13942884e-01 8.03086996e-01 -4.18890059e-01
8.02662492e-01 -2.13338780e+00 1.73718244e-01 4.99821573e-01
1.26483083e-01 4.04114097e-01 1.63508505e-02 3.84931341e-02
-4.31955844e-01 -1.09983571e-01 -7.84689128e-01 -2.99552500e-01
2.70758616e-03 3.10142964e-01 -3.74019921e-01 7.03205824e-01
2.42730275e-01 6.64107859e-01 -9.56848800e-01 -5.35173357e-01
9.95771810e-02 2.73742288e-01 -6.46973312e-01 1.67367235e-01
-2.92542279e-01 8.01589608e-01 -3.70985270e-01 2.64719367e-01
5.78559935e-01 -4.63271588e-01 2.60843009e-01 2.20292509e-01
4.69938099e-01 4.45933677e-02 -1.07650042e+00 1.76591814e+00
-4.25274372e-01 2.05810219e-01 9.46364254e-02 -1.24786162e+00
5.29958308e-01 2.80472845e-01 6.69834077e-01 -4.97948945e-01
8.85917917e-02 2.27754086e-01 3.01322471e-02 -2.30439395e-01
-6.61640093e-02 -6.95628643e-01 -1.87919617e-01 5.62538326e-01
2.06217825e-01 7.98380598e-02 -2.09364757e-01 1.44344553e-01
8.25257778e-01 1.31464019e-01 7.01192200e-01 -3.58968854e-01
5.50002575e-01 6.25829622e-02 4.88391072e-01 7.09037602e-01
-3.39591324e-01 4.79159832e-01 5.66250563e-01 -1.94819421e-01
-1.05254710e+00 -1.17260587e+00 -5.39181769e-01 1.35643065e+00
-2.50539720e-01 1.19150272e-02 -6.55370533e-01 -1.22669125e+00
3.21852386e-01 6.79127514e-01 -7.92052448e-01 -3.18235695e-01
-4.11817193e-01 -8.32624912e-01 4.12697345e-01 4.52611774e-01
2.43399832e-02 -8.34199846e-01 -4.76115108e-01 7.74191543e-02
2.75499165e-01 -1.05070293e+00 -4.43047941e-01 5.41102767e-01
-8.13402355e-01 -1.15298223e+00 -1.03930390e+00 -7.16984630e-01
7.61616886e-01 -1.96690381e-01 1.07113492e+00 -4.46114212e-01
-3.43266666e-01 5.69976449e-01 -1.45940766e-01 -6.72386289e-01
-6.95369899e-01 3.94759476e-01 1.40078828e-01 7.27447793e-02
4.57587808e-01 -6.75851226e-01 -3.86184365e-01 2.75381982e-01
-1.16400039e+00 -3.62154134e-02 6.69211149e-01 1.11305225e+00
6.06316566e-01 -3.28716487e-01 8.93750966e-01 -1.75667000e+00
5.90536833e-01 -6.47752523e-01 -5.97044289e-01 1.95780441e-01
-9.24332738e-01 2.27109894e-01 6.91733360e-01 -6.76224411e-01
-1.05624080e+00 4.47482169e-01 -5.11021027e-03 -3.92388672e-01
-4.07942384e-01 3.35555613e-01 -3.07239562e-01 -9.40216333e-02
9.79429424e-01 7.09225163e-02 4.89196479e-01 -5.91098964e-01
4.13079679e-01 7.09921420e-01 4.86393064e-01 -6.54761851e-01
7.00090945e-01 5.95142365e-01 1.81786925e-01 -6.69939339e-01
-9.41035092e-01 -6.19463563e-01 -8.75314593e-01 2.23233461e-01
6.66251540e-01 -9.06576216e-01 -3.99362326e-01 2.77167439e-01
-5.51353276e-01 -6.28219604e-01 -4.59236115e-01 4.63868469e-01
-8.67966771e-01 3.15401673e-01 -1.30608529e-01 -5.36500216e-01
-7.97700733e-02 -9.61768627e-01 7.93250084e-01 -5.13825752e-02
-3.07178438e-01 -1.34310615e+00 2.61883914e-01 2.77977407e-01
2.65283793e-01 2.89994478e-01 1.11940312e+00 -1.23183453e+00
-5.27240708e-02 -3.55439000e-02 -5.53801609e-03 5.00409365e-01
2.97929287e-01 -8.71866524e-01 -1.25564504e+00 -5.32636464e-01
5.79507761e-02 -6.35054231e-01 8.67356956e-01 4.37267452e-01
1.25392926e+00 -9.13067311e-02 -4.58015412e-01 5.52095652e-01
1.25349712e+00 -4.12068926e-02 2.60110945e-01 2.18569234e-01
5.88490129e-01 8.44856322e-01 7.62212098e-01 3.49297494e-01
-1.32361893e-02 6.60958886e-01 -2.90744398e-02 -2.11585954e-01
-1.00259058e-01 -4.66882996e-02 7.79766440e-02 1.58975348e-01
4.16450113e-01 1.69529662e-01 -1.01335001e+00 6.36141837e-01
-1.71112025e+00 -6.45538807e-01 2.67082393e-01 2.51207900e+00
1.27597296e+00 1.34113580e-01 5.00362158e-01 -3.15373540e-02
5.77544570e-01 -2.76188672e-01 -9.91084337e-01 -2.01926813e-01
2.88953837e-02 4.10854638e-01 6.93534672e-01 5.74890614e-01
-1.33053184e+00 7.14156210e-01 6.33767462e+00 7.87055910e-01
-1.09015763e+00 1.74890563e-01 7.30281472e-01 -1.78391058e-02
-1.26755148e-01 -2.58544415e-01 -6.55492425e-01 3.31780165e-01
9.59939182e-01 -1.97262540e-01 3.10708344e-01 1.04025340e+00
-2.55222142e-01 8.17814916e-02 -1.59086037e+00 8.32837701e-01
-3.01112719e-02 -1.04504645e+00 -1.10278249e-01 1.06961191e-01
6.72257304e-01 -8.83071050e-02 3.46007079e-01 3.94930661e-01
4.70351994e-01 -1.00442791e+00 6.72087260e-03 -5.28766550e-02
1.04089868e+00 -6.92826927e-01 5.22760510e-01 4.31159317e-01
-4.68640059e-01 -5.07771857e-02 -5.77820726e-02 2.02760786e-01
-1.75679043e-01 4.83241022e-01 -1.22303462e+00 4.55160677e-01
2.11182937e-01 5.56077719e-01 -3.39090288e-01 8.45831871e-01
1.08549580e-01 6.56256258e-01 -2.83805996e-01 4.40849543e-01
5.28181531e-02 -2.91894153e-02 4.04845953e-01 1.17259145e+00
-1.40801162e-01 2.89247390e-02 5.19197062e-02 5.95817268e-01
-2.36376241e-01 2.28946909e-01 -7.48360813e-01 1.26443505e-01
2.53977329e-01 9.72026408e-01 -4.85720694e-01 -3.80135506e-01
-4.08455759e-01 1.03615475e+00 4.10826147e-01 2.77086735e-01
-5.90284348e-01 -2.37865761e-01 7.70452440e-01 1.14910625e-01
3.12656492e-01 1.52319819e-01 -4.49276835e-01 -9.99177098e-01
-1.79047137e-01 -1.06765568e+00 8.77887487e-01 -1.52362436e-01
-1.58669889e+00 -5.65905608e-02 2.69836813e-01 -1.23183072e+00
-3.85407001e-01 -7.41715908e-01 -6.82058334e-02 9.21609938e-01
-1.78592467e+00 -1.16885006e+00 5.73993959e-02 7.67504096e-01
2.01524869e-01 -2.39321217e-01 1.04704499e+00 3.90894353e-01
-2.00741291e-01 9.65190291e-01 5.40120959e-01 1.40087595e-02
1.41146100e+00 -1.40441597e+00 -7.43829235e-02 3.79536182e-01
-3.99655551e-02 5.07064998e-01 7.40354598e-01 -5.06665111e-01
-9.73510563e-01 -1.27892947e+00 5.15994728e-01 -6.37629211e-01
4.48663443e-01 -4.84383464e-01 -1.26622307e+00 9.09135163e-01
-2.61665314e-01 1.66108474e-01 1.14081383e+00 3.76906276e-01
-5.79517663e-01 -2.77379565e-02 -1.49768865e+00 3.54183227e-01
7.66051590e-01 -5.10344684e-01 -5.38189590e-01 4.59501594e-01
5.38708448e-01 -4.01740640e-01 -9.81386185e-01 2.58011073e-01
5.32057345e-01 -5.01188099e-01 1.12872052e+00 -1.10237992e+00
3.16919625e-01 3.19425799e-02 -4.25167754e-02 -1.32696152e+00
-2.49195993e-01 -6.04910791e-01 -1.45807788e-02 1.02330673e+00
2.98929989e-01 -7.33138680e-01 9.49268103e-01 9.18877125e-01
1.93961382e-01 -4.99530107e-01 -8.82308602e-01 -8.10451210e-01
6.74822509e-01 -2.24756747e-01 3.18843365e-01 1.30519414e+00
-5.22155240e-02 4.53206331e-01 -3.40100050e-01 1.87615231e-01
9.65331376e-01 -3.03044375e-02 8.18358541e-01 -1.49104655e+00
-4.15655345e-01 -3.08926493e-01 -2.79305339e-01 -7.54482388e-01
3.96458954e-01 -1.09656692e+00 -9.86193791e-02 -9.69050646e-01
4.41790134e-01 -7.42224693e-01 -5.70498943e-01 5.68326950e-01
-3.35153282e-01 1.25816777e-01 -5.68202585e-02 1.93555281e-01
-4.05314803e-01 2.19815150e-01 8.72205555e-01 1.01093128e-02
-2.34109730e-01 1.90097168e-01 -9.15485501e-01 5.31376600e-01
7.42552519e-01 -6.85073435e-01 -6.47584975e-01 4.60535698e-02
-7.84370154e-02 -1.71559051e-01 1.83064654e-01 -7.04600751e-01
-7.36390650e-02 -2.05100447e-01 3.34592044e-01 -1.62386969e-01
2.29477331e-01 -1.05311728e+00 -1.43810034e-01 5.07568717e-01
-7.74738073e-01 -7.27152407e-01 2.78124750e-01 7.62487531e-01
1.00360081e-01 -2.18308166e-01 1.21470118e+00 4.30243202e-02
-6.17599308e-01 2.63449073e-01 2.08945535e-02 2.74166554e-01
1.21310210e+00 -2.10404582e-03 -8.16681087e-02 -1.53440148e-01
-9.62456346e-01 2.99487233e-01 5.60221493e-01 1.33763090e-01
2.31958151e-01 -1.21527457e+00 -7.80331612e-01 3.69442999e-01
3.72919172e-01 1.16031446e-01 1.97660044e-01 8.15818906e-01
1.10966958e-01 3.14513266e-01 -2.80284882e-01 -6.99188828e-01
-1.66107678e+00 7.71253347e-01 1.43144861e-01 -5.62679827e-01
-4.68054861e-01 6.43983126e-01 6.79982901e-01 -6.00930691e-01
2.45873526e-01 -4.43634763e-02 -8.65684729e-03 4.29718457e-02
5.05596280e-01 9.27690044e-02 4.07110751e-02 -2.33530715e-01
-3.76419067e-01 3.00112456e-01 -5.23459494e-01 -8.73175561e-02
1.40486419e+00 -1.38325328e-02 3.92211586e-01 5.09497166e-01
1.39270532e+00 -1.68531269e-01 -1.33242118e+00 -6.50087714e-01
2.02313662e-01 -4.00455415e-01 -1.85253680e-01 -1.08394921e+00
-4.89886761e-01 8.82577837e-01 8.04082394e-01 -4.76296321e-02
1.10347426e+00 1.57789096e-01 4.47043031e-01 3.78007114e-01
1.23525955e-01 -1.13852131e+00 -7.92476237e-02 1.86284378e-01
4.87158775e-01 -1.64880455e+00 1.55299306e-01 -3.31175953e-01
-7.63840497e-01 7.15642393e-01 3.00369143e-01 5.49295656e-02
6.53232515e-01 3.70909125e-02 7.52163678e-02 1.91029802e-01
-5.32253683e-01 -6.03131168e-02 2.11409777e-01 9.15226102e-01
1.86278820e-01 -1.53665217e-02 -1.88025177e-01 5.64475536e-01
2.36067563e-01 1.33899376e-01 2.63720483e-01 1.12400687e+00
-3.31886083e-01 -1.46018958e+00 -3.68696779e-01 5.62863231e-01
-7.07966447e-01 6.73166886e-02 -2.83341467e-01 9.10180748e-01
-1.25507563e-01 3.48008513e-01 -9.53697786e-03 2.08298519e-01
1.68635711e-01 5.75527728e-01 6.03822470e-01 -6.55387402e-01
-3.66594583e-01 -7.56513327e-02 -1.67725105e-02 -2.71017313e-01
-5.12876153e-01 -8.92165720e-01 -1.08533001e+00 2.09499553e-01
-1.13861792e-01 2.37651002e-02 5.84357917e-01 9.39988077e-01
4.13318247e-01 3.31656188e-01 5.35538077e-01 -3.95585030e-01
-1.02579701e+00 -8.14510465e-01 -6.45962298e-01 9.89623368e-01
5.58571219e-01 -7.52183616e-01 -2.69022912e-01 3.82279605e-01] | [10.328304290771484, 3.246525287628174] |
7efa04dd-d2f6-42f7-8483-d2da0bd7aeb9 | accurate-and-robust-pulmonary-nodule | 1907.11704 | null | https://arxiv.org/abs/1907.11704v1 | https://arxiv.org/pdf/1907.11704v1.pdf | Accurate and Robust Pulmonary Nodule Detection by 3D Feature Pyramid Network with Self-supervised Feature Learning | Accurate detection of pulmonary nodules with high sensitivity and specificity is essential for automatic lung cancer diagnosis from CT scans. Although many deep learning-based algorithms make great progress for improving the accuracy of nodule detection, the high false positive rate is still a challenging problem which limits the automatic diagnosis in routine clinical practice. Moreover, the CT scans collected from multiple manufacturers may affect the robustness of Computer-aided diagnosis (CAD) due to the differences in intensity scales and machine noises. In this paper, we propose a novel self-supervised learning assisted pulmonary nodule detection framework based on a 3D Feature Pyramid Network (3DFPN) to improve the sensitivity of nodule detection by employing multi-scale features to increase the resolution of nodules, as well as a parallel top-down path to transit the high-level semantic features to complement low-level general features. Furthermore, a High Sensitivity and Specificity (HS2) network is introduced to eliminate the false positive nodule candidates by tracking the appearance changes in continuous CT slices of each nodule candidate on Location History Images (LHI). In addition, in order to improve the performance consistency of the proposed framework across data captured by different CT scanners without using additional annotations, an effective self-supervised learning schema is applied to learn spatiotemporal features of CT scans from large-scale unlabeled data. The performance and robustness of our method are evaluated on several publicly available datasets with significant performance improvements. The proposed framework is able to accurately detect pulmonary nodules with high sensitivity and specificity and achieves 90.6% sensitivity with 1/8 false positive per scan which outperforms the state-of-the-art results 15.8% on LUNA16 dataset. | ['YingLi Tian', 'Jingya Liu', 'Oguz Akin', 'Liangliang Cao'] | 2019-07-25 | null | null | null | null | ['lung-cancer-diagnosis'] | ['medical'] | [ 7.36228302e-02 1.42144235e-02 -2.83731908e-01 -1.94300354e-01
-8.17941070e-01 -2.55544603e-01 2.30297029e-01 -2.09447965e-02
-3.96432430e-01 3.29665482e-01 -6.43261820e-02 -2.17392430e-01
-3.92094910e-01 -8.98591280e-01 -3.50061089e-01 -8.54455233e-01
-8.17196295e-02 6.24426961e-01 9.34309483e-01 2.08643079e-01
-3.91926974e-01 6.65668011e-01 -1.21909833e+00 3.96740586e-01
7.76769698e-01 1.02301455e+00 6.78531051e-01 3.83428603e-01
-1.33531436e-01 7.83429027e-01 -3.07620987e-02 -3.96193331e-03
3.71368229e-01 -4.44891334e-01 -5.25568485e-01 2.64026850e-01
1.76331606e-02 -4.42244709e-01 -4.85327125e-01 9.82380211e-01
4.36444700e-01 -4.37916219e-01 6.58668339e-01 -6.48897588e-01
-2.98692018e-01 4.02484447e-01 -5.30378163e-01 4.31597203e-01
-4.30457622e-01 3.85207355e-01 6.94415748e-01 -9.57464576e-01
4.02302533e-01 5.13796806e-01 6.89905882e-01 3.65500808e-01
-6.09538078e-01 -6.00290656e-01 -3.92859876e-01 1.56599060e-01
-1.33428800e+00 1.08045168e-01 3.32390606e-01 -3.77589703e-01
2.97789872e-01 4.40909952e-01 9.59169984e-01 6.81204557e-01
3.95002693e-01 4.61021692e-01 9.10096884e-01 -6.19895346e-02
-9.12506059e-02 2.41845563e-01 -3.86998564e-01 1.32125020e+00
6.79331064e-01 2.05106154e-01 1.48036271e-01 -4.89268228e-02
1.22133982e+00 6.12911344e-01 -1.04377449e-01 -6.82586670e-01
-1.56176591e+00 7.32938349e-01 1.15070689e+00 7.97869623e-01
-5.02342820e-01 -2.34805932e-03 3.04948777e-01 -3.85802150e-01
-1.38870366e-02 2.61327773e-01 -9.02022794e-02 4.02621090e-01
-7.74302125e-01 -3.12024295e-01 5.53910017e-01 5.45584917e-01
3.65978718e-01 -4.10340242e-02 -7.00358570e-01 7.12910354e-01
2.28412941e-01 6.59370303e-01 1.02145004e+00 -3.78429592e-01
1.88203137e-02 9.08327758e-01 -6.21259362e-02 -7.43260443e-01
-8.04737628e-01 -1.02686167e+00 -1.13447690e+00 -5.69588803e-02
2.56325394e-01 1.96658403e-01 -1.30927420e+00 1.10602856e+00
4.10176545e-01 2.46200234e-01 -2.28236914e-01 1.09294283e+00
9.50782359e-01 2.11119011e-01 1.16996929e-01 -2.47865111e-01
1.70454407e+00 -1.06674850e+00 -5.04981160e-01 -2.43243426e-02
8.18178117e-01 -5.53866386e-01 9.88213956e-01 -1.68398827e-01
-7.52649724e-01 -5.78304112e-01 -9.85758901e-01 3.95816535e-01
5.18089645e-02 7.66945541e-01 5.13238728e-01 6.33386314e-01
-5.75051785e-01 3.40929389e-01 -1.37048650e+00 -4.10009146e-01
7.05570459e-01 7.16648638e-01 -2.25243926e-01 -2.66996235e-01
-9.76374328e-01 8.37815225e-01 4.75177974e-01 5.50522329e-03
-9.61093307e-01 -9.14791167e-01 -4.48853195e-01 2.49663383e-01
7.40054846e-01 -7.88154721e-01 1.30499494e+00 -6.98141992e-01
-1.22151756e+00 6.12832546e-01 1.20667301e-01 -5.27090251e-01
7.48521805e-01 1.85971513e-01 -4.12965536e-01 3.49847347e-01
2.42409512e-01 4.09066737e-01 6.99399710e-01 -6.74223781e-01
-9.13442910e-01 -2.76421040e-01 -5.97833157e-01 3.28267545e-01
-3.91629457e-01 -3.56837928e-01 -6.87864065e-01 -4.32414263e-01
3.94812077e-01 -1.15778470e+00 -4.80044186e-01 2.51586348e-01
-2.11582333e-01 -7.75484070e-02 1.03064084e+00 -5.26506364e-01
9.96804059e-01 -2.14937973e+00 -4.38764125e-01 3.26925039e-01
4.81519431e-01 5.02379358e-01 1.97919026e-01 -3.96574557e-01
9.53483135e-02 9.47360098e-02 -1.45362139e-01 4.99416113e-01
-4.89151508e-01 2.22911000e-01 5.66193223e-01 4.23889756e-01
3.37373704e-01 1.37598145e+00 -8.68820131e-01 -8.74007463e-01
5.97646773e-01 3.64061445e-01 -3.16489130e-01 2.94561824e-03
7.45147094e-02 6.11793160e-01 -7.65795112e-01 6.44831061e-01
4.33217585e-01 -9.04858530e-01 3.19469243e-01 -4.79598820e-01
-5.05903438e-02 -6.69363560e-03 -9.87027884e-01 1.42211688e+00
-4.50787514e-01 8.23048875e-02 -1.43561006e-01 -6.21502757e-01
5.81464708e-01 6.23517990e-01 9.55921948e-01 -6.57599092e-01
1.40043259e-01 4.90631849e-01 6.38399959e-01 -7.18874991e-01
-2.66370505e-01 -1.11591637e-01 2.20305249e-01 2.44359821e-01
-1.40017018e-01 9.75136296e-04 5.99849857e-02 2.25689113e-02
1.37439728e+00 -3.94218981e-01 6.64327204e-01 -3.35639745e-01
6.81591034e-01 2.78622895e-01 5.31153560e-01 7.41613567e-01
-3.47374827e-01 5.04017353e-01 3.02973688e-02 -4.07667816e-01
-9.46662366e-01 -1.12950170e+00 -2.78406024e-01 5.35353839e-01
1.04340568e-01 2.07026109e-01 -3.04015994e-01 -1.14909124e+00
-5.54960668e-02 1.07702091e-01 -6.40895605e-01 -1.31262004e-01
-6.45501256e-01 -9.39991355e-01 3.60371768e-01 7.32922971e-01
8.33932757e-01 -8.74406993e-01 -8.76372397e-01 2.96904325e-01
-3.01351875e-01 -9.92325544e-01 -3.01497787e-01 3.82004231e-01
-1.03433192e+00 -1.21386755e+00 -8.05362225e-01 -9.52750027e-01
8.09138834e-01 5.06116807e-01 8.39052975e-01 1.79423094e-01
-8.55453610e-01 1.26891630e-02 -1.12634584e-01 -3.48552734e-01
-4.11473751e-01 3.03214490e-01 -2.73986876e-01 -1.05432503e-01
1.59845859e-01 -1.07910618e-01 -8.77088130e-01 5.92711508e-01
-9.61650908e-01 6.77613169e-02 1.51538801e+00 9.84683216e-01
9.42888081e-01 3.43678325e-01 4.53348935e-01 -1.10857487e+00
2.67060380e-02 -4.67758417e-01 -5.23991287e-01 1.85041562e-01
-4.20965314e-01 2.74364725e-02 5.85795581e-01 -3.98639321e-01
-1.11761272e+00 5.63189924e-01 2.92213038e-02 -3.64423960e-01
-3.98392454e-02 4.19326454e-01 2.66627640e-01 -3.24408710e-01
9.50691819e-01 2.95555323e-01 9.24672335e-02 -6.36749491e-02
-1.65818617e-01 3.73514265e-01 4.38661337e-01 1.55604839e-01
1.06636620e+00 6.63186848e-01 4.24177647e-01 -6.23187780e-01
-1.17293358e+00 -8.43671560e-01 -7.53368616e-01 -1.66700229e-01
9.23051834e-01 -1.00072610e+00 -2.36661464e-01 4.84477525e-04
-3.86700541e-01 4.50282544e-02 -4.56231445e-01 8.74172688e-01
-1.32760391e-01 2.15967253e-01 -6.14842474e-01 -2.51712412e-01
-4.74836826e-01 -1.25791728e+00 7.63372004e-01 3.53742152e-01
1.22129478e-01 -6.94166839e-01 -1.88834488e-01 2.22574517e-01
7.35379398e-01 6.16638623e-02 8.32119584e-01 -8.53686750e-01
-8.69675934e-01 -3.58544558e-01 -5.48659742e-01 1.22490413e-01
6.98794901e-01 -2.13866651e-01 -5.88409245e-01 -2.12217420e-01
1.76799908e-01 -6.68492988e-02 7.57784784e-01 6.56941473e-01
1.24757361e+00 1.41797876e-02 -7.68279970e-01 5.49899518e-01
1.47849309e+00 1.82935923e-01 1.78926915e-01 2.09743932e-01
7.85563290e-01 1.69767048e-02 5.73316574e-01 2.96458900e-01
-2.10135877e-01 3.41462582e-01 7.16990411e-01 -4.48963732e-01
-4.91365105e-01 -4.38509807e-02 -1.84389725e-01 6.33175850e-01
-2.02062368e-01 5.38275354e-02 -1.06436169e+00 6.25407875e-01
-1.63125765e+00 -8.10957730e-01 -3.97007972e-01 2.14200878e+00
4.04244751e-01 1.55528545e-01 -1.11740507e-01 2.44764052e-02
7.04576850e-01 -2.07710341e-01 -7.41844058e-01 5.90400994e-01
2.42134765e-01 3.15519452e-01 8.33719671e-01 -6.05063997e-02
-1.44417012e+00 3.81775081e-01 5.43434095e+00 8.90675545e-01
-1.39925921e+00 3.76681268e-01 5.02474070e-01 -5.77060394e-02
1.38509318e-01 -4.78584856e-01 -5.20848453e-01 2.25080550e-01
5.31522334e-01 -5.16716763e-02 -1.06056131e-01 9.90472138e-01
2.67864347e-01 -1.12174630e-01 -8.73264074e-01 7.80628145e-01
-1.86135516e-01 -1.37423062e+00 -8.87014717e-02 5.07176407e-02
8.15816164e-01 3.50945145e-01 1.48476437e-01 1.82487383e-01
1.31124456e-04 -9.88679349e-01 -5.92566766e-02 2.66333818e-01
9.27256227e-01 -4.81157780e-01 1.10306656e+00 4.36037838e-01
-1.43506086e+00 -9.42655504e-02 -2.45974720e-01 5.26742876e-01
-2.40466118e-01 6.14432454e-01 -1.70983422e+00 7.09677637e-01
7.73888171e-01 5.70887744e-01 -9.24203634e-01 1.38282323e+00
-1.35909513e-01 7.48821914e-01 -5.46158433e-01 -1.67943493e-01
3.13616365e-01 3.16681862e-01 2.90104687e-01 9.95048046e-01
6.61785185e-01 2.21036464e-01 2.80152053e-01 6.74557209e-01
-9.31117088e-02 2.82579601e-01 -4.12106991e-01 1.02416649e-01
3.31477523e-01 1.64720023e+00 -1.12783480e+00 -3.11662972e-01
-5.91735840e-01 7.65305936e-01 -1.40536845e-01 -1.25253707e-01
-1.07137609e+00 8.52713659e-02 -2.97579378e-01 6.16323233e-01
5.00375748e-01 1.69126406e-01 -4.12901212e-03 -9.11184907e-01
-1.37020007e-01 -5.51720321e-01 5.79316556e-01 -4.44057584e-01
-1.20525765e+00 6.21146977e-01 -3.58996153e-01 -1.66322041e+00
-1.39421612e-01 -5.66173434e-01 -5.96201301e-01 4.66588587e-01
-1.43837726e+00 -1.19615459e+00 -7.91999638e-01 6.48625493e-01
4.87351209e-01 -5.40281385e-02 6.47507012e-01 2.67289132e-01
-3.56833965e-01 3.41247350e-01 1.65309221e-01 3.20344925e-01
6.85182810e-01 -1.09101355e+00 -1.71304882e-01 6.67304933e-01
-7.21361563e-02 -3.13292369e-02 -2.47598290e-02 -6.60349011e-01
-1.19074357e+00 -1.59641802e+00 3.26759130e-01 -1.16205335e-01
4.92406577e-01 2.08856151e-01 -9.45407927e-01 5.94965100e-01
-3.48059475e-01 7.60120571e-01 4.49065715e-01 -6.07130647e-01
1.45166352e-01 -1.51189700e-01 -1.28767681e+00 3.51748317e-01
7.90825427e-01 -5.22485152e-02 -2.39593595e-01 5.82466543e-01
3.96317691e-01 -5.61082959e-01 -9.08460379e-01 9.95212495e-01
4.46048319e-01 -9.49364424e-01 1.00985658e+00 -7.29082376e-02
1.40463263e-01 -4.95184034e-01 2.88494319e-01 -8.17694724e-01
-9.08058345e-01 3.44312876e-01 1.35294378e-01 4.07931507e-01
3.62689584e-01 -4.32764798e-01 1.22070420e+00 1.80672079e-01
-2.00315267e-01 -9.20389593e-01 -8.70417774e-01 -6.46469712e-01
-2.52284527e-01 -2.88882125e-02 1.58213317e-01 6.15369081e-01
-5.09075522e-01 -1.44593678e-02 1.06792957e-01 3.88304889e-01
5.23864627e-01 1.30801827e-01 2.13092625e-01 -1.33166873e+00
-3.42122078e-01 -4.29770440e-01 -5.05022764e-01 -5.83587050e-01
-6.50749326e-01 -1.14524567e+00 -7.88706318e-02 -1.64547014e+00
7.04391360e-01 -3.57896775e-01 -6.30329907e-01 4.13627476e-01
-1.73180312e-01 4.49319750e-01 -1.15930013e-01 5.04315913e-01
-6.32321477e-01 3.38491172e-01 1.77342105e+00 -1.67766884e-01
-1.41284421e-01 4.17020112e-01 -2.86102802e-01 8.95604968e-01
8.05515230e-01 -7.01926410e-01 -3.28852147e-01 -1.34936899e-01
-3.89494479e-01 2.67419338e-01 5.39761007e-01 -1.38574398e+00
3.00860792e-01 -7.65875056e-02 9.02268767e-01 -7.99493492e-01
7.89685100e-02 -1.07436645e+00 2.11323857e-01 1.32122207e+00
7.22087687e-03 -3.96976352e-01 8.28258693e-02 7.76169538e-01
-2.86360592e-01 -2.06062347e-01 1.09167922e+00 -5.95499098e-01
-6.32369578e-01 6.42899156e-01 -4.06968117e-01 -3.06449324e-01
1.35274541e+00 -7.44054839e-02 3.51322852e-02 4.51091863e-02
-8.04263294e-01 5.65723628e-02 1.36872530e-01 1.05552606e-01
4.91147816e-01 -1.41342950e+00 -6.86205804e-01 2.37454966e-01
2.23564193e-01 3.23859662e-01 4.98900294e-01 1.30078018e+00
-6.75135553e-01 7.11091220e-01 -6.68746457e-02 -1.02587616e+00
-1.19876707e+00 3.78010273e-01 7.76675284e-01 -8.79600704e-01
-7.47614086e-01 6.43590748e-01 4.60514128e-01 -2.92652041e-01
1.01222523e-01 -5.00132740e-01 -9.72269773e-02 -4.81338143e-01
2.32360661e-01 8.14399719e-02 4.17261779e-01 -2.63826251e-01
-5.00682831e-01 3.93306941e-01 -2.70704865e-01 2.80886620e-01
9.78267431e-01 1.58254012e-01 2.21748278e-01 1.17301725e-01
8.18354130e-01 1.85827760e-03 -9.80627716e-01 -4.77476358e-01
3.64900865e-02 -1.95340484e-01 2.26041213e-01 -1.00997949e+00
-1.22927058e+00 6.33803308e-01 1.13820887e+00 -9.69845895e-03
1.05403674e+00 1.21641003e-01 6.10498905e-01 4.61106181e-01
2.24109142e-04 -4.52832282e-01 3.97986919e-01 1.35612831e-01
4.49508160e-01 -1.67525756e+00 1.86300740e-01 -7.01940656e-01
-7.42260516e-01 1.20407462e+00 8.52166951e-01 -6.63149282e-02
7.04600573e-01 2.82630056e-01 1.53792277e-01 -3.33658606e-01
-4.37102705e-01 -3.11240613e-01 5.42717099e-01 3.22597086e-01
4.70889181e-01 1.96176678e-01 -2.12962076e-01 6.57219768e-01
2.56758064e-01 1.44561514e-01 2.62414575e-01 8.68854642e-01
-6.79492474e-01 -8.09237301e-01 -3.36632282e-01 1.05954504e+00
-5.93855381e-01 1.39130577e-01 -3.12969796e-02 1.27108276e+00
1.83123767e-01 4.41193700e-01 -1.66235626e-01 -8.81017447e-02
2.19133496e-01 -9.09570456e-02 2.68965036e-01 -8.47663999e-01
-5.75617671e-01 3.54838192e-01 -3.50910276e-01 -2.87020177e-01
-4.26033497e-01 -5.73254347e-01 -1.51184237e+00 3.46504331e-01
-4.30402666e-01 -2.57842299e-02 4.35666412e-01 8.27320576e-01
2.35604778e-01 1.00714433e+00 6.94767416e-01 -3.71687382e-01
-7.31694162e-01 -8.48332405e-01 -4.56936091e-01 3.64470065e-01
8.17219261e-03 -5.98681450e-01 -1.39820442e-01 -4.70647141e-02] | [15.369974136352539, -2.1442806720733643] |
2a6c5222-7d3c-457c-9cc3-4546d4736f95 | revise-self-supervised-speech-resynthesis | 2212.11377 | null | https://arxiv.org/abs/2212.11377v1 | https://arxiv.org/pdf/2212.11377v1.pdf | ReVISE: Self-Supervised Speech Resynthesis with Visual Input for Universal and Generalized Speech Enhancement | Prior works on improving speech quality with visual input typically study each type of auditory distortion separately (e.g., separation, inpainting, video-to-speech) and present tailored algorithms. This paper proposes to unify these subjects and study Generalized Speech Enhancement, where the goal is not to reconstruct the exact reference clean signal, but to focus on improving certain aspects of speech. In particular, this paper concerns intelligibility, quality, and video synchronization. We cast the problem as audio-visual speech resynthesis, which is composed of two steps: pseudo audio-visual speech recognition (P-AVSR) and pseudo text-to-speech synthesis (P-TTS). P-AVSR and P-TTS are connected by discrete units derived from a self-supervised speech model. Moreover, we utilize self-supervised audio-visual speech model to initialize P-AVSR. The proposed model is coined ReVISE. ReVISE is the first high-quality model for in-the-wild video-to-speech synthesis and achieves superior performance on all LRS3 audio-visual enhancement tasks with a single model. To demonstrates its applicability in the real world, ReVISE is also evaluated on EasyCom, an audio-visual benchmark collected under challenging acoustic conditions with only 1.6 hours of training data. Similarly, ReVISE greatly suppresses noise and improves quality. Project page: https://wnhsu.github.io/ReVISE. | ['Yossi Adi', 'Jacob Donley', 'Bowen Shi', 'Tal Remez', 'Wei-Ning Hsu'] | 2022-12-21 | null | null | null | null | ['video-synchronization', 'text-to-speech-synthesis', 'audio-visual-speech-recognition'] | ['computer-vision', 'speech', 'speech'] | [ 2.90184617e-01 -2.33800814e-01 1.73212335e-01 -6.94436282e-02
-1.22966313e+00 -3.71195287e-01 4.15485620e-01 -3.08126807e-01
-1.52866185e-01 4.35381383e-01 5.99419534e-01 -3.24868381e-01
3.17958981e-01 -9.19867754e-02 -8.97643745e-01 -7.47873545e-01
2.39853874e-01 -3.54991078e-01 8.26958716e-02 -2.32763246e-01
-1.61359638e-01 2.28669107e-01 -1.66192448e+00 4.87913400e-01
6.04750454e-01 9.33410108e-01 6.73981488e-01 1.28611338e+00
2.34555051e-01 7.78752446e-01 -8.66040170e-01 -2.31191918e-01
1.97351903e-01 -5.50023675e-01 -3.89285415e-01 3.09488416e-01
3.96126807e-01 -5.32920480e-01 -7.53655076e-01 1.14028323e+00
9.34146643e-01 2.22742721e-01 4.90228027e-01 -1.42993689e+00
-7.74280608e-01 4.34568495e-01 -4.61630642e-01 2.33204290e-01
5.13976455e-01 3.83026332e-01 8.98346066e-01 -1.38418436e+00
3.78635198e-01 1.22507203e+00 4.27345932e-01 6.47919297e-01
-1.09937060e+00 -6.61381364e-01 -3.56055498e-02 5.11955857e-01
-1.36542070e+00 -1.20986843e+00 6.47877097e-01 -2.11830527e-01
1.03057277e+00 4.66226906e-01 4.97064978e-01 1.31293881e+00
-1.03307642e-01 9.47484434e-01 9.78624284e-01 -3.35633039e-01
2.61067331e-01 -5.23372032e-02 -1.76108092e-01 1.30091801e-01
-3.18915665e-01 4.67682183e-01 -7.88999617e-01 2.37047195e-01
4.74289536e-01 -2.86308259e-01 -8.44199479e-01 9.47973356e-02
-9.90149736e-01 3.30495507e-01 5.06631136e-02 8.49726796e-02
-3.20475072e-01 1.65192127e-01 4.14319664e-01 5.46288133e-01
4.74013627e-01 -1.32429019e-01 -2.53172606e-01 -2.32672170e-01
-1.31680465e+00 2.99129616e-02 3.64456385e-01 1.15710485e+00
1.34691566e-01 8.18147302e-01 -3.83182198e-01 1.25520039e+00
4.75494087e-01 8.10179412e-01 4.77916062e-01 -1.11222327e+00
5.90181351e-01 -6.84279382e-01 5.42913377e-02 -5.12652576e-01
3.89143303e-02 -4.63286132e-01 -1.04953110e+00 4.14334923e-01
-1.39540449e-01 -1.80555999e-01 -1.00541162e+00 1.63082504e+00
1.97002113e-01 4.96485084e-01 2.46571168e-01 1.04211974e+00
1.20701361e+00 1.27743423e+00 -1.74363330e-01 -5.68887413e-01
1.17932117e+00 -1.37748384e+00 -1.13422978e+00 -1.69895068e-01
-9.22739804e-02 -1.18950927e+00 1.13952577e+00 6.88068509e-01
-1.59039438e+00 -9.60348487e-01 -1.14129376e+00 -2.05677405e-01
1.41800538e-01 2.47360796e-01 -3.55983913e-01 5.90024054e-01
-1.55984724e+00 3.75250101e-01 -6.88132048e-01 3.55722867e-02
8.91159847e-02 -6.26499802e-02 -3.18097472e-01 4.09918167e-02
-1.21027195e+00 5.49034894e-01 -7.62199834e-02 -2.77837999e-02
-1.47837031e+00 -6.27985716e-01 -9.84373808e-01 1.80370942e-01
4.40540880e-01 -7.23238111e-01 1.58986104e+00 -1.00346828e+00
-1.86302757e+00 4.21410918e-01 -5.19721508e-01 -5.53559482e-01
3.63643914e-01 -2.06203073e-01 -7.92373955e-01 3.10944825e-01
-1.47008136e-01 5.42657554e-01 1.72799027e+00 -1.50913942e+00
-6.12098813e-01 1.03050947e-01 -3.82573992e-01 3.70467722e-01
-2.77110457e-01 3.08186889e-01 -8.85052621e-01 -1.27812374e+00
-4.68513146e-02 -5.94378710e-01 2.05662951e-01 -9.71934795e-02
-3.19039047e-01 2.30509847e-01 8.67558181e-01 -1.27796543e+00
1.36823058e+00 -2.49577546e+00 3.02459985e-01 -2.86897063e-01
2.49898776e-01 6.51528299e-01 -5.01102448e-01 2.95233965e-01
-4.31245923e-01 -3.12299258e-03 -3.01317900e-01 -9.88329887e-01
1.08037386e-02 -8.86119381e-02 -5.10222256e-01 4.98118699e-01
1.73349539e-03 5.62797248e-01 -7.24543631e-01 -4.24318105e-01
4.62877184e-01 7.94100225e-01 -5.79209268e-01 4.47973222e-01
2.20048189e-01 3.83825690e-01 1.54937029e-01 7.06087649e-01
8.35223973e-01 1.45927325e-01 -1.83227062e-01 -2.77696282e-01
-1.46327347e-01 3.23710173e-01 -1.26046216e+00 1.51049399e+00
-6.93199337e-01 9.49790835e-01 7.17982709e-01 -6.12184227e-01
5.86524129e-01 7.30275571e-01 1.71917602e-01 -8.40521932e-01
-8.26046988e-03 2.18743443e-01 -3.16720039e-01 -5.30350864e-01
6.45925224e-01 2.02897619e-02 5.35016537e-01 2.64686365e-02
3.00623000e-01 -3.80281061e-01 9.87188797e-03 3.89849335e-01
9.06411529e-01 -2.72489861e-02 2.52802163e-01 2.86819428e-01
5.41319668e-01 -6.79819942e-01 2.64147043e-01 4.79999602e-01
-3.95552576e-01 1.08755279e+00 4.40296456e-02 6.87081993e-01
-1.34333932e+00 -1.58316529e+00 2.64647994e-02 1.13523889e+00
5.04101515e-02 -5.90794146e-01 -9.08521950e-01 -1.52520344e-01
-5.17137468e-01 8.69997561e-01 -1.16496071e-01 1.82550400e-02
-2.95846254e-01 -2.39795581e-01 5.40908515e-01 3.68112475e-01
3.15401822e-01 -9.93988037e-01 -1.01991765e-01 1.30377188e-01
-5.36328077e-01 -1.29670310e+00 -1.08206403e+00 -1.00958431e-02
-3.70053709e-01 -5.53947866e-01 -1.26388347e+00 -9.36401904e-01
2.36892343e-01 7.25230753e-01 9.05370176e-01 -1.24647595e-01
5.11776879e-02 6.27131522e-01 -5.57328165e-01 -2.71197438e-01
-7.51916766e-01 -5.11455715e-01 1.85219511e-01 3.06471735e-01
-3.86262715e-01 -7.23407269e-01 -5.31406641e-01 3.09720010e-01
-9.48659241e-01 1.01237960e-01 4.11849111e-01 1.06205249e+00
7.20422685e-01 1.43694431e-01 6.16275966e-01 1.57430079e-02
7.01302290e-01 -2.43648484e-01 -4.41333324e-01 1.06679052e-01
-2.91950732e-01 -4.20349568e-01 7.09968865e-01 -5.36600053e-01
-1.30939984e+00 -1.94071189e-01 -6.81020617e-01 -9.66027617e-01
-2.06012830e-01 1.45064041e-01 -4.13670897e-01 1.77063718e-01
4.73116010e-01 6.78010881e-01 -8.25755596e-02 -5.95865548e-01
4.71123457e-01 1.19600725e+00 9.72423077e-01 -9.66902152e-02
9.01598692e-01 2.06169620e-01 -3.40327680e-01 -1.45221627e+00
-2.72235215e-01 -5.43877780e-01 3.58291678e-02 -3.32291901e-01
7.68634737e-01 -1.34700751e+00 -4.81159925e-01 7.17416465e-01
-1.18939781e+00 -6.34272635e-01 -3.25995803e-01 5.35353601e-01
-6.60563409e-01 7.50449240e-01 -8.88091862e-01 -8.57239962e-01
-5.06412923e-01 -1.48599267e+00 1.15042782e+00 -7.77378753e-02
1.70362070e-01 -4.53370959e-01 -2.89073177e-02 3.84089291e-01
4.30762947e-01 -4.76413012e-01 1.52222931e-01 -1.73353061e-01
-2.85031617e-01 3.51811588e-01 -1.00319661e-01 1.05730617e+00
1.01710156e-01 2.06368137e-02 -1.44003475e+00 -5.90297937e-01
2.83853024e-01 -2.08334640e-01 9.33634162e-01 7.12895095e-01
1.14317906e+00 -3.23180169e-01 2.38761112e-01 9.61985290e-01
8.38918686e-01 5.06089091e-01 8.46750081e-01 9.21273790e-03
4.62655187e-01 3.61170858e-01 6.93595588e-01 4.61873174e-01
1.50024801e-01 1.04251063e+00 3.94493103e-01 -3.49845320e-01
-1.00089538e+00 -3.50743979e-01 9.49176967e-01 1.34312403e+00
1.61834329e-01 -4.96107727e-01 -4.33690071e-01 5.32442570e-01
-1.36394560e+00 -9.76703405e-01 1.93831120e-02 2.01707840e+00
8.95243943e-01 -9.10114646e-02 7.45295659e-02 4.90981042e-01
8.66913140e-01 4.27169710e-01 -3.70644003e-01 -2.13280901e-01
-3.95017326e-01 3.98816466e-01 2.10755050e-01 8.63776028e-01
-8.17884445e-01 7.81633615e-01 5.50985432e+00 1.29117262e+00
-1.17748833e+00 5.10923266e-01 5.84412575e-01 -3.77473563e-01
-5.35086505e-02 -3.59069973e-01 -4.76819098e-01 4.44055736e-01
1.03174984e+00 -1.41057730e-01 8.97113442e-01 5.25303423e-01
7.29490519e-01 2.25288853e-01 -8.94280493e-01 1.35265315e+00
2.08246261e-01 -1.10455513e+00 5.03723510e-02 -3.29250276e-01
5.19611597e-01 4.50570732e-02 3.78733069e-01 2.60478884e-01
-1.28633901e-01 -8.68285537e-01 1.33311403e+00 2.14884564e-01
1.39873326e+00 -5.71133673e-01 1.87930644e-01 7.90072754e-02
-1.51986408e+00 2.69734859e-02 -4.37608883e-02 4.27363902e-01
5.85710168e-01 3.13786119e-01 -7.27922380e-01 6.86510861e-01
9.76535618e-01 7.53446639e-01 -2.04610899e-01 1.06465161e+00
-5.39070547e-01 8.36906910e-01 -1.45487990e-02 4.46742088e-01
-7.67198429e-02 9.19578224e-02 1.11297870e+00 1.41265833e+00
4.83761698e-01 1.36456609e-01 -2.68855184e-01 4.86220807e-01
-1.45289898e-02 5.44390418e-02 -2.86799043e-01 6.20866977e-02
5.02139151e-01 9.28972185e-01 -1.25044048e-01 -4.66457635e-01
-4.62960660e-01 1.09284222e+00 -3.38333547e-01 6.81900442e-01
-1.11818707e+00 -5.02330720e-01 8.05000544e-01 7.48648122e-02
4.53858733e-01 -8.31645727e-02 6.48435578e-02 -1.11733913e+00
6.93662167e-02 -1.39645696e+00 -4.63666096e-02 -1.45477378e+00
-8.70245159e-01 9.30390418e-01 -2.01915100e-01 -1.44978285e+00
-1.60448298e-01 -4.06953692e-01 -4.21993315e-01 9.66752291e-01
-1.61211002e+00 -8.41846108e-01 -3.30508411e-01 9.30426121e-01
1.22638631e+00 -2.46380940e-01 3.06341588e-01 7.49194741e-01
-5.17615080e-01 8.32521260e-01 1.86995640e-01 -7.92042613e-02
9.73901391e-01 -9.31299329e-01 7.18583465e-01 1.28257549e+00
1.25920355e-01 1.51491061e-01 1.02037883e+00 -5.56805611e-01
-1.30137587e+00 -1.23028100e+00 6.66941047e-01 1.21551387e-01
5.32062709e-01 -4.32884514e-01 -8.66282761e-01 4.72363651e-01
4.96221006e-01 2.76367087e-02 2.28090033e-01 -7.57105827e-01
-2.00553939e-01 -1.51764184e-01 -8.81857932e-01 7.80696511e-01
1.02573752e+00 -8.73383760e-01 -4.49613839e-01 9.37465206e-02
1.26295173e+00 -6.89899862e-01 -4.75599468e-01 6.52078986e-02
1.88449740e-01 -9.80184734e-01 1.32237899e+00 -1.37697682e-01
4.01593179e-01 -6.34035647e-01 -4.90014255e-01 -1.61490619e+00
-1.93836205e-02 -1.16994214e+00 -1.85429499e-01 1.46500218e+00
2.65318155e-01 -5.11196256e-01 2.63190120e-01 -2.29763493e-01
-6.84311032e-01 -2.96417326e-01 -1.00145316e+00 -8.88831973e-01
-3.68276149e-01 -8.39155376e-01 5.04950941e-01 7.15082884e-01
-2.30469942e-01 2.60901123e-01 -9.51938450e-01 4.50643092e-01
7.79402018e-01 -4.27474201e-01 8.05354118e-01 -3.56024206e-01
-7.75248289e-01 -3.01791608e-01 1.62414145e-02 -1.50626731e+00
1.50885172e-02 -6.96715057e-01 3.36974293e-01 -1.42862201e+00
-2.06567183e-01 2.49269873e-01 -1.41637502e-02 1.43813059e-01
-5.22119850e-02 1.48929238e-01 5.23483813e-01 2.10386496e-02
-3.69821638e-01 9.41974878e-01 1.34736145e+00 -3.55073541e-01
-3.45172435e-01 1.28734991e-01 -4.27147210e-01 4.03649539e-01
5.68838060e-01 -3.39242458e-01 -5.43914437e-01 -4.86309916e-01
-3.36667597e-01 7.22425103e-01 4.08469886e-01 -9.60795820e-01
3.95203054e-01 9.08816457e-02 1.76842585e-02 -5.99657595e-01
8.58036280e-01 -6.41389132e-01 2.05534711e-01 2.19781339e-01
-3.59473586e-01 -7.05722496e-02 2.68893421e-01 5.79396188e-01
-6.32631898e-01 -3.83841656e-02 1.07429266e+00 2.69791335e-01
-5.68197131e-01 2.14261189e-01 -4.75125313e-01 6.46402761e-02
6.82626128e-01 -6.66157454e-02 -3.51927608e-01 -8.85480165e-01
-1.05261374e+00 -2.63417810e-02 1.36075690e-01 3.80696326e-01
1.13091004e+00 -1.26462984e+00 -1.09541202e+00 2.22536772e-01
-1.39521435e-01 -3.23221534e-01 7.99483716e-01 6.76409423e-01
-3.41621071e-01 2.10857213e-01 1.18852794e-01 -6.36581123e-01
-1.72275615e+00 7.38833606e-01 3.91160697e-01 1.88725352e-01
-6.81596935e-01 8.76393437e-01 4.59836513e-01 1.95636213e-01
6.96668804e-01 -2.42080644e-01 -8.31308365e-02 -1.29537791e-01
7.76132584e-01 5.09868622e-01 1.72756702e-01 -8.49982738e-01
-8.69971216e-02 3.06396991e-01 3.08944196e-01 -5.95804870e-01
1.24159729e+00 -5.37833273e-01 1.85504913e-01 2.60561168e-01
1.24682879e+00 2.77581573e-01 -1.37598145e+00 -3.55274796e-01
-6.21125877e-01 -5.21380126e-01 5.06956160e-01 -6.82166457e-01
-1.26549852e+00 1.14482939e+00 7.28768408e-01 4.06574644e-02
1.67266345e+00 -1.77858055e-01 8.83888245e-01 -1.05883077e-01
-2.49051638e-02 -9.73591805e-01 4.23548520e-01 3.50509048e-01
1.46346879e+00 -9.49184477e-01 -2.63466805e-01 -3.84414703e-01
-8.26199710e-01 8.79836619e-01 2.60832131e-01 4.57556218e-01
6.03434920e-01 6.69245958e-01 9.68241990e-02 4.12155241e-01
-8.92855942e-01 -2.28240713e-01 3.77358079e-01 7.57385671e-01
1.30471185e-01 -3.32929450e-03 2.47110188e-01 6.76115751e-01
-3.09380025e-01 -1.30774274e-01 5.87496758e-01 6.40846074e-01
-3.71112376e-01 -7.68316269e-01 -8.01336646e-01 -7.59156123e-02
-4.15212125e-01 -5.53357959e-01 -1.45465694e-02 2.72791386e-01
-2.11209610e-01 1.55235922e+00 -9.16813165e-02 -6.05875194e-01
4.83452350e-01 -1.84975907e-01 2.59155065e-01 -2.91886836e-01
-5.38909733e-01 8.29913557e-01 6.93957582e-02 -4.78943795e-01
-1.86719403e-01 -5.97626626e-01 -9.98075664e-01 -3.26813549e-01
-1.76242992e-01 -9.31164715e-03 7.72731543e-01 4.77028698e-01
2.76118219e-01 1.09273088e+00 8.87525737e-01 -1.06287944e+00
-5.79253972e-01 -9.05346155e-01 -6.97593510e-01 2.41071269e-01
9.58225429e-01 -3.38369727e-01 -7.99132288e-01 3.27245891e-01] | [14.586451530456543, 5.479820251464844] |
9d79bffa-d83c-4079-97c6-536e31889997 | social-media-medical-concept-normalization | null | null | https://aclanthology.org/2020.knlp-1.3 | https://aclanthology.org/2020.knlp-1.3.pdf | Social Media Medical Concept Normalization using RoBERTa in Ontology Enriched Text Similarity Framework | Pattisapu et al. (2020) formulate medical concept normalization (MCN) as text similarity problem and propose a model based on RoBERTa and graph embedding based target concept vectors. However, graph embedding techniques ignore valuable information available in the clinical ontology like concept description and synonyms. In this work, we enhance the model of Pattisapu et al. (2020) with two novel changes. First, we use retrofitted target concept vectors instead of graph embedding based vectors. It is the first work to leverage both concept description and synonyms to represent concepts in the form of retrofitted target concept vectors in text similarity framework based social media MCN. Second, we generate both concept and concept mention vectors with same size which eliminates the need of dense layers to project concept mention vectors into the target concept embedding space. Our model outperforms existing methods with improvements up to 3.75% on two standard datasets. Further when trained only on mapping lexicon synonyms, our model outperforms existing methods with significant improvements up to 14.61%. We attribute these significant improvements to the two novel changes introduced. | ['Sivanesan Sangeetha', 'Katikapalli Subramanyam Kalyan'] | null | null | null | null | aacl-knlp-2020-12 | ['medical-concept-normalization'] | ['medical'] | [ 2.72689581e-01 3.77091169e-01 -3.32607925e-01 -1.40446827e-01
-2.58123428e-01 -3.44961852e-01 6.50206029e-01 1.04361534e+00
-7.43217528e-01 4.22870785e-01 6.98112309e-01 -1.15248822e-01
-2.03631729e-01 -1.00632513e+00 -1.42000020e-01 -1.42034367e-01
4.24659438e-02 4.49930012e-01 2.02156916e-01 -7.13528156e-01
1.92334756e-01 1.22850776e-01 -1.10232449e+00 2.25260079e-01
7.35276401e-01 2.70359874e-01 -9.65011045e-02 5.68525553e-01
-4.32149202e-01 4.07165378e-01 -4.62791920e-01 -5.53863883e-01
5.41936047e-02 -2.46281520e-01 -8.22965920e-01 -4.11542475e-01
3.76286656e-01 1.04260214e-01 -4.79536146e-01 1.18028140e+00
6.22723699e-01 1.88808963e-01 8.10254276e-01 -1.22109365e+00
-1.00920212e+00 8.30079079e-01 -5.00298858e-01 2.22812071e-02
4.02722061e-01 -5.05012929e-01 1.31016922e+00 -8.83666039e-01
8.75168860e-01 1.19629419e+00 9.33937669e-01 7.23248661e-01
-1.13824165e+00 -6.96676731e-01 1.04759827e-01 2.84341007e-01
-1.55860472e+00 8.10622144e-03 5.89122117e-01 -3.97762746e-01
1.39909256e+00 4.17957455e-01 6.75307751e-01 9.31910872e-01
1.51574478e-01 3.81562889e-01 4.15214360e-01 -7.13445485e-01
6.57306015e-02 1.66920692e-01 4.47553068e-01 6.61736429e-01
6.79279685e-01 -3.09499353e-01 -8.20009038e-02 -4.68607128e-01
4.68115091e-01 4.48179454e-01 -1.71357170e-01 -4.26524460e-01
-1.43312478e+00 1.27595937e+00 6.64791286e-01 6.69347644e-01
-3.22742760e-01 6.68833638e-03 6.34636104e-01 1.64469197e-01
4.50188279e-01 8.37274849e-01 -1.83316559e-01 1.55420423e-01
-7.10976720e-01 1.19800143e-01 6.00265920e-01 1.10299492e+00
4.68386799e-01 -2.24056751e-01 -2.62042195e-01 9.43695724e-01
3.14973980e-01 3.54488164e-01 1.08159316e+00 -4.44126487e-01
4.32511449e-01 9.41037357e-01 -3.63856018e-01 -1.35534370e+00
-6.57377660e-01 -5.95323384e-01 -7.48591781e-01 -5.63166082e-01
-2.56078482e-01 5.29871807e-02 -9.91341233e-01 1.63302374e+00
2.08066285e-01 3.31651896e-01 2.62302876e-01 4.80044663e-01
1.19588017e+00 3.51979077e-01 2.23707184e-01 4.09535393e-02
1.67379820e+00 -9.80407417e-01 -9.43190336e-01 -6.31488636e-02
1.25893104e+00 -9.42137778e-01 9.49108839e-01 -2.28509679e-01
-4.73859936e-01 -1.15824960e-01 -1.24611199e+00 -1.63227886e-01
-8.12135935e-01 -2.94445872e-01 8.90430689e-01 7.78718531e-01
-1.09769857e+00 6.19965136e-01 -6.24360323e-01 -9.77090776e-01
3.42447042e-01 4.06024069e-01 -6.50256395e-01 -1.65166199e-01
-1.41854131e+00 8.92479420e-01 7.22676456e-01 -7.22329676e-01
6.78892434e-02 -1.03255892e+00 -1.13580644e+00 1.50955781e-01
2.26327986e-01 -1.01260459e+00 6.21881187e-01 -5.15740156e-01
-9.96978819e-01 6.82024419e-01 1.05622508e-01 -4.51340705e-01
7.96828419e-03 -1.60807744e-02 -7.94003785e-01 2.10461140e-01
3.57529640e-01 6.29350245e-01 1.10169485e-01 -8.31098378e-01
-4.28993076e-01 -3.29881161e-02 1.78506210e-01 4.22016948e-01
-9.56820846e-01 -1.47341937e-01 -5.76764643e-01 -9.93681908e-01
3.22131276e-01 -8.66510093e-01 -3.93458456e-01 -5.73222153e-02
-3.93351585e-01 -1.61629766e-01 3.05132002e-01 -4.50174361e-01
1.49478996e+00 -2.00968242e+00 2.50900611e-02 3.80062252e-01
7.86548436e-01 2.54529029e-01 -5.05347073e-01 7.78291464e-01
-4.79212254e-01 3.02145898e-01 -4.66595411e-01 -3.08660597e-01
2.59245038e-02 3.09541583e-01 -1.16670681e-02 2.49780327e-01
2.67910391e-01 1.03970432e+00 -1.31433189e+00 -6.04717016e-01
1.40145063e-01 7.18897939e-01 -9.58019853e-01 -2.85481811e-01
2.59058565e-01 -2.37955004e-01 -3.75436217e-01 4.65239942e-01
5.55143178e-01 -2.03231275e-01 5.60921371e-01 -5.37420213e-01
3.31472278e-01 2.67629862e-01 -9.92059469e-01 1.85986650e+00
-3.31675977e-01 3.63764942e-01 -6.93610787e-01 -1.03607464e+00
5.97391427e-01 5.18673301e-01 9.63838816e-01 -4.39440638e-01
1.37382686e-01 7.25863948e-02 -2.13823654e-02 -2.63137162e-01
7.43403077e-01 -5.01929283e-01 -1.79906577e-01 4.25677210e-01
3.14316809e-01 7.38824606e-02 2.83443362e-01 7.87119210e-01
1.28200853e+00 -3.53371918e-01 8.43994975e-01 -4.01316851e-01
5.30794799e-01 -5.80329895e-02 4.66020167e-01 6.15281224e-01
-8.93493593e-02 8.29394400e-01 5.59664369e-01 -1.67384446e-01
-9.11670566e-01 -1.02575600e+00 -2.74896204e-01 9.16407883e-01
-6.26687333e-02 -1.19598043e+00 -4.49914575e-01 -8.84473264e-01
1.63597852e-01 6.84008420e-01 -1.00644946e+00 -4.39630628e-01
-4.31959182e-01 -9.77132857e-01 7.40589082e-01 5.64968705e-01
-2.19697461e-01 -5.57311833e-01 -1.97128393e-02 4.58689719e-01
-2.25689158e-01 -1.03462148e+00 -7.45475054e-01 -2.30740771e-01
-9.19748485e-01 -1.17805755e+00 -7.31854737e-01 -7.84886479e-01
7.69761503e-01 1.99704662e-01 1.13489139e+00 2.73521751e-01
-5.70351362e-01 6.70894682e-01 -7.50316858e-01 -4.75153357e-01
-3.42732936e-01 1.86608881e-01 2.88646728e-01 -2.11220637e-01
7.36942172e-01 -4.89601195e-01 -5.23530662e-01 -2.37101451e-01
-1.20329893e+00 4.67603700e-03 4.23117369e-01 1.01615083e+00
4.81910020e-01 -3.68654072e-01 5.00554621e-01 -1.24521172e+00
7.94908941e-01 -7.59301603e-01 6.66920096e-02 2.88538307e-01
-1.14642739e+00 2.52878308e-01 3.38107586e-01 -3.93720299e-01
-2.90637910e-01 -2.19272524e-01 -2.74680763e-01 -1.79875150e-01
2.75366277e-01 5.96575081e-01 2.34200165e-01 2.02460792e-02
8.08917820e-01 -4.09036390e-02 -1.10029034e-01 -4.77345377e-01
7.31892109e-01 6.46899760e-01 1.58701807e-01 -9.39067379e-02
7.16284096e-01 3.86126012e-01 -5.56573123e-02 -6.50160253e-01
-5.64363956e-01 -1.01250410e+00 -7.28040159e-01 4.65384752e-01
9.33395386e-01 -9.52501714e-01 -4.31175441e-01 -4.12595779e-01
-1.00413573e+00 6.25253320e-01 -3.41725171e-01 8.15310597e-01
-1.38847977e-01 7.68926680e-01 -5.27619362e-01 -1.37576714e-01
-6.65730715e-01 -6.87291086e-01 8.62353146e-01 -1.65133253e-01
-5.79452813e-01 -1.30668902e+00 4.11237031e-01 1.09740406e-01
4.69078213e-01 2.93864101e-01 1.16322637e+00 -9.97761905e-01
2.55032748e-01 -6.45200610e-01 -3.28067601e-01 -2.01413985e-02
6.68837309e-01 -3.64449829e-01 -6.68737888e-01 -6.25223160e-01
-2.08311871e-01 2.47646585e-01 8.79309237e-01 1.45410537e-03
6.93121910e-01 -2.87523389e-01 -5.93936324e-01 6.44116879e-01
1.66285062e+00 -1.63611427e-01 4.41862047e-01 3.69241774e-01
1.04442942e+00 5.14867544e-01 2.90830642e-01 2.76211500e-01
6.30941272e-01 6.90705836e-01 9.55976471e-02 -8.26283023e-02
-2.44196862e-01 -2.43691757e-01 -1.29396934e-03 1.51740134e+00
2.96577476e-02 -1.98812336e-01 -9.79086399e-01 8.48099589e-01
-1.76895249e+00 -8.11541021e-01 -1.03937976e-01 2.07036304e+00
9.37387407e-01 -6.97941333e-02 -2.43452951e-01 7.97693729e-02
5.93891501e-01 -9.68334898e-02 -9.45031270e-02 -3.76964539e-01
-1.28375068e-01 5.74392915e-01 6.05155885e-01 6.81838036e-01
-1.04464912e+00 9.41826284e-01 5.57422543e+00 7.02565670e-01
-8.39127183e-01 3.98510158e-01 -2.46414587e-01 -3.30594629e-01
-6.26570702e-01 5.08562289e-02 -5.33642411e-01 2.58844048e-01
9.11488473e-01 -6.34112597e-01 -4.60132733e-02 4.59543288e-01
-4.39816445e-01 5.15182018e-01 -1.04252899e+00 1.15228331e+00
5.52991927e-01 -1.42124760e+00 6.03969634e-01 -8.84609893e-02
6.47506595e-01 6.31042942e-02 -1.92365393e-01 4.73237127e-01
1.87306926e-01 -1.07267249e+00 -2.36394256e-02 4.12480921e-01
7.45598257e-01 -7.47303665e-01 1.19478035e+00 -4.06693310e-01
-1.34478545e+00 1.83556229e-01 -6.69124246e-01 1.82597518e-01
3.33749205e-02 5.12298763e-01 -1.23009074e+00 1.01949859e+00
3.12052161e-01 1.15635777e+00 -7.33077168e-01 8.58183086e-01
-2.45471187e-02 4.15827781e-01 -1.59005970e-01 -3.03096008e-02
3.27791393e-01 4.05864604e-03 4.73999441e-01 1.50146616e+00
2.62220740e-01 1.70764536e-01 2.00728774e-01 4.61379468e-01
-1.32594079e-01 8.36490154e-01 -6.62169337e-01 -2.69851863e-01
4.62630481e-01 1.27692688e+00 -7.98297703e-01 -5.64599156e-01
-5.06043375e-01 1.04584062e+00 1.12892218e-01 1.05492450e-01
-6.20289922e-01 -8.98398757e-01 8.40165257e-01 1.25768229e-01
-1.40552921e-02 2.51849983e-02 -1.24802038e-01 -1.34095597e+00
-2.31310859e-01 -4.80074048e-01 7.09102273e-01 -2.74971217e-01
-1.42814565e+00 7.26688862e-01 9.02756117e-03 -1.44159746e+00
1.93573209e-03 -6.26960278e-01 -2.53384620e-01 7.71579683e-01
-1.45317292e+00 -1.30752146e+00 -6.77414984e-02 5.40649772e-01
1.97337762e-01 -3.67620945e-01 1.33671427e+00 6.91033006e-01
-3.38137597e-01 9.58719850e-01 3.31983089e-01 3.66268963e-01
9.78299677e-01 -1.37114239e+00 5.51658452e-01 5.30210972e-01
2.25039706e-01 1.12020707e+00 6.33391678e-01 -8.43047798e-01
-9.11472440e-01 -1.16594732e+00 1.42029190e+00 -7.72813857e-01
8.41167092e-01 -3.92159998e-01 -9.69509304e-01 6.16518021e-01
2.32995838e-01 -1.35509789e-01 1.32677388e+00 1.27565861e-01
-5.83186626e-01 1.86761171e-01 -1.05176508e+00 8.49983335e-01
1.03993106e+00 -5.85993528e-01 -9.92682934e-01 6.66166604e-01
1.18622172e+00 -8.19887966e-02 -1.22567964e+00 3.73415112e-01
3.94689918e-01 -1.55795276e-01 1.20980299e+00 -9.67056155e-01
2.28337884e-01 -2.54753321e-01 -3.71127903e-01 -1.23523760e+00
-5.44244170e-01 -1.57366663e-01 1.21246703e-01 9.20892239e-01
5.45993090e-01 -8.36657345e-01 4.27247018e-01 2.84471959e-01
-6.60817549e-02 -6.14708960e-01 -8.27569962e-01 -7.42776394e-01
2.91163236e-01 -3.52076918e-01 6.99988961e-01 1.64168561e+00
5.09514928e-01 5.20752072e-01 -1.55246153e-01 -2.75213365e-02
4.22916770e-01 -1.56635046e-01 2.14268997e-01 -1.36423945e+00
1.25957774e-02 -4.56167519e-01 -1.06169558e+00 -2.46070236e-01
-3.61254662e-02 -1.59297574e+00 -4.84855950e-01 -1.84733903e+00
4.50725645e-01 -3.05629015e-01 -8.98787081e-01 7.71887898e-01
-3.49484980e-01 4.56675440e-01 2.48748422e-01 9.69046727e-02
-4.28649902e-01 5.95953405e-01 7.53815532e-01 -2.64695644e-01
-7.51801208e-02 -5.31607926e-01 -1.06278181e+00 3.86349082e-01
8.39833975e-01 -8.30468893e-01 -6.01158202e-01 -3.64769667e-01
3.05590868e-01 -4.96248811e-01 3.31264511e-02 -8.11161995e-01
2.98396260e-01 6.82202652e-02 3.97079848e-02 -1.72964752e-01
2.30209887e-01 -6.53827608e-01 -9.81480181e-02 7.20203638e-01
-2.96373695e-01 3.90694678e-01 3.57688755e-01 8.20510507e-01
-1.35535359e-01 -1.96643278e-01 4.94048119e-01 -1.00648245e-02
-6.25394404e-01 1.77963391e-01 -2.52269536e-01 1.16726667e-01
8.72610569e-01 -3.15232754e-01 -2.26286054e-01 -1.59379363e-01
-7.40727067e-01 1.53537154e-01 4.44838226e-01 8.03404331e-01
7.68313944e-01 -1.63610899e+00 -8.26745033e-01 1.32702380e-01
6.49199963e-01 -5.37945986e-01 1.46761417e-01 7.95541525e-01
-6.60768926e-01 7.38203764e-01 -2.18383417e-01 -2.38079697e-01
-1.37377894e+00 6.82448804e-01 7.16273766e-03 -2.38469809e-01
-8.84589791e-01 7.74919331e-01 3.32997292e-01 -9.00841475e-01
-1.55749083e-01 -2.83066779e-01 -7.88563073e-01 2.93268919e-01
5.25839806e-01 1.25929862e-01 8.61729458e-02 -7.15936720e-01
-6.60208523e-01 7.63870299e-01 -4.20981109e-01 -2.22513010e-03
1.47101951e+00 9.33780521e-02 -3.46951127e-01 2.12897316e-01
1.52597821e+00 1.98887587e-01 -3.88385653e-02 -5.24249136e-01
2.41018549e-01 -2.43195966e-01 4.68975008e-02 -6.76151097e-01
-8.93590331e-01 6.41134202e-01 8.50970864e-01 -3.46782625e-01
9.81077552e-01 -1.08399346e-01 6.94076657e-01 5.06950855e-01
4.76746112e-02 -9.09987271e-01 -3.03680543e-02 5.35107732e-01
6.85130000e-01 -1.06576121e+00 3.66297513e-01 -5.96655905e-01
-6.63025379e-01 1.03767741e+00 2.56618410e-01 3.79957445e-02
8.51617992e-01 -2.46698529e-01 1.81008279e-01 -5.07969737e-01
-5.63420057e-01 -3.16592187e-01 7.38543749e-01 6.52049303e-01
8.66948545e-01 2.80321032e-01 -8.80984426e-01 6.66181028e-01
-2.88941950e-01 -2.89298952e-01 7.08048582e-01 9.58727300e-01
-1.00441054e-01 -1.46710980e+00 5.16418256e-02 8.19087029e-01
-3.89987379e-01 -8.39357972e-01 -1.76107123e-01 7.17899799e-01
1.44139245e-01 7.68283486e-01 3.24640870e-02 -5.88837206e-01
4.66779798e-01 1.92124426e-01 7.36738443e-02 -1.29249668e+00
-7.31340408e-01 -2.17842758e-01 -1.37026072e-01 -4.66631085e-01
-2.69042432e-01 -3.76780003e-01 -1.43638194e+00 -1.55515254e-01
-3.83508921e-01 2.67392904e-01 4.72945064e-01 7.47266948e-01
6.10489190e-01 7.74533987e-01 9.74948183e-02 -9.64897126e-02
-2.50425994e-01 -1.01008415e+00 -5.47748685e-01 8.01172256e-01
-4.70589427e-03 -6.45867944e-01 -1.28132626e-01 -1.83047608e-01] | [8.545105934143066, 8.51929759979248] |
d2d1034f-4142-4d35-8d99-0936aa7de338 | trajectoryformer-3d-object-tracking | 2306.05888 | null | https://arxiv.org/abs/2306.05888v1 | https://arxiv.org/pdf/2306.05888v1.pdf | TrajectoryFormer: 3D Object Tracking Transformer with Predictive Trajectory Hypotheses | 3D multi-object tracking (MOT) is vital for many applications including autonomous driving vehicles and service robots. With the commonly used tracking-by-detection paradigm, 3D MOT has made important progress in recent years. However, these methods only use the detection boxes of the current frame to obtain trajectory-box association results, which makes it impossible for the tracker to recover objects missed by the detector. In this paper, we present TrajectoryFormer, a novel point-cloud-based 3D MOT framework. To recover the missed object by detector, we generates multiple trajectory hypotheses with hybrid candidate boxes, including temporally predicted boxes and current-frame detection boxes, for trajectory-box association. The predicted boxes can propagate object's history trajectory information to the current frame and thus the network can tolerate short-term miss detection of the tracked objects. We combine long-term object motion feature and short-term object appearance feature to create per-hypothesis feature embedding, which reduces the computational overhead for spatial-temporal encoding. Additionally, we introduce a Global-Local Interaction Module to conduct information interaction among all hypotheses and models their spatial relations, leading to accurate estimation of hypotheses. Our TrajectoryFormer achieves state-of-the-art performance on the Waymo 3D MOT benchmarks. | ['Hongsheng Li', 'Simon See', 'Ka Chun Cheung', 'Qiang Wang', 'Benjin Zhu', 'Chao Zhang', 'Shaoshuai Shi', 'Xuesong Chen'] | 2023-06-09 | null | null | null | null | ['object-tracking', 'multi-object-tracking', '3d-object-tracking', '3d-multi-object-tracking'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [-3.95091206e-01 -2.71843761e-01 -2.12709799e-01 -1.31377637e-01
-5.14989138e-01 -3.28925103e-01 4.80520368e-01 1.37153685e-01
-3.45520973e-01 3.41407180e-01 -2.58943766e-01 7.08285943e-02
8.39956999e-02 -6.79294705e-01 -8.14011931e-01 -7.91532457e-01
-2.96307802e-01 6.13846481e-01 1.19357538e+00 6.08507954e-02
1.16471397e-02 7.58912563e-01 -1.83434868e+00 7.02399015e-02
4.27925915e-01 1.10010731e+00 5.53351641e-01 5.03063262e-01
-1.89693600e-01 5.39779902e-01 -6.43122911e-01 1.32283881e-01
3.42193425e-01 2.03360040e-02 1.24916390e-01 -1.20606028e-01
5.01441419e-01 -5.61382592e-01 -7.28745162e-01 9.39041495e-01
3.42078388e-01 2.31659979e-01 4.87427950e-01 -1.70494175e+00
-3.31580102e-01 1.36474252e-01 -6.47306144e-01 3.81366581e-01
1.40059233e-01 4.22853678e-01 5.04117846e-01 -1.13104546e+00
6.54244006e-01 1.57213032e+00 6.66696131e-01 7.03478515e-01
-8.53246152e-01 -1.04312396e+00 3.17244023e-01 5.39224029e-01
-1.59446418e+00 -4.13047940e-01 6.18718088e-01 -5.12926340e-01
9.62960064e-01 1.44631490e-01 7.46136010e-01 6.56101704e-01
5.22965729e-01 9.17085648e-01 4.95029062e-01 2.28412107e-01
-1.86342709e-02 4.98941839e-02 1.37371927e-01 6.80022836e-01
3.75793755e-01 5.94776809e-01 -5.00990212e-01 3.21263298e-02
6.23829544e-01 4.59572881e-01 3.05262841e-02 -4.71153796e-01
-1.52913952e+00 5.34471393e-01 6.89344823e-01 1.26813963e-01
-2.53828019e-01 4.86271590e-01 1.92378372e-01 1.08215541e-01
4.09582615e-01 -4.29710597e-01 -9.31432545e-02 1.17673062e-01
-9.00559425e-01 4.13141936e-01 2.83715844e-01 1.51481843e+00
9.73311901e-01 -3.09741795e-02 -4.62513685e-01 2.17181966e-01
7.04712629e-01 1.05345845e+00 -1.78880859e-02 -7.17700720e-01
2.57122159e-01 8.01795304e-01 3.10211092e-01 -1.13870406e+00
-4.56394166e-01 -2.64444798e-01 -6.46117449e-01 5.01662374e-01
1.14591114e-01 3.35369587e-01 -8.63632619e-01 1.42584717e+00
9.41849709e-01 8.04303288e-01 3.19001004e-02 1.12611222e+00
9.14963245e-01 9.68300641e-01 -1.51754439e-01 -2.57437587e-01
1.18431246e+00 -9.64595258e-01 -7.72903919e-01 -1.84179887e-01
5.75832188e-01 -5.14876187e-01 2.12070227e-01 -6.79220334e-02
-9.97972250e-01 -8.22546661e-01 -1.01139021e+00 5.50332293e-03
-3.33088219e-01 1.06661394e-02 1.72341287e-01 2.92175919e-01
-8.71705651e-01 2.85081387e-01 -1.23222351e+00 -1.84809044e-01
5.21937430e-01 5.57227552e-01 -2.10212559e-01 -3.38719398e-01
-7.75802732e-01 1.08196557e+00 5.39062381e-01 9.86249149e-02
-1.22703111e+00 -5.94671965e-01 -7.51473367e-01 -1.98786035e-01
2.29970813e-01 -5.60814261e-01 8.62592220e-01 -4.22609486e-02
-9.15473282e-01 5.50405502e-01 -6.25636816e-01 -5.37861049e-01
4.52145845e-01 -1.99412294e-02 -6.68564796e-01 -1.00228630e-01
2.88946092e-01 1.02967870e+00 9.21751142e-01 -1.27417660e+00
-1.22052085e+00 -3.39851707e-01 -2.94873476e-01 9.97569188e-02
-5.15207089e-02 -9.57739074e-03 -9.12660360e-01 -1.55910194e-01
4.14312333e-01 -1.07778823e+00 -1.35933846e-01 6.62153244e-01
4.76084370e-03 -6.32699430e-01 1.51661658e+00 -5.22139259e-02
9.35016692e-01 -2.47897053e+00 -8.11498240e-02 -1.39976263e-01
5.46721578e-01 2.43430510e-02 -6.79756999e-02 5.07271513e-02
6.14302039e-01 -4.86681610e-01 2.68872291e-01 -7.75775254e-01
1.17199242e-01 2.90630460e-01 -3.95523578e-01 8.86663258e-01
2.45395184e-01 8.78533661e-01 -1.09982038e+00 -7.36841023e-01
7.25322306e-01 6.64300919e-01 -3.65849644e-01 2.40421500e-02
-2.86597550e-01 4.55342829e-01 -6.47137761e-01 8.29920232e-01
1.06887269e+00 -1.20871373e-01 -4.97439951e-01 -1.37993649e-01
-4.82909441e-01 1.03579476e-01 -1.12096548e+00 1.59344149e+00
-2.44313069e-02 7.64100552e-01 -1.62575543e-01 -3.36379826e-01
9.55232501e-01 1.49024129e-01 8.24423850e-01 -5.50063133e-01
-2.81100366e-02 2.71711230e-01 -2.88174450e-01 -1.54037267e-01
9.25194144e-01 3.03331643e-01 -1.49205312e-01 -1.99410017e-03
-2.69003570e-01 1.02887377e-01 3.05190519e-03 2.16656789e-01
1.32049990e+00 2.04978764e-01 -1.58037648e-01 2.03935713e-01
4.42149252e-01 4.44784075e-01 9.67268944e-01 6.42611861e-01
-6.73515856e-01 3.38450581e-01 -3.56824249e-01 -5.09035707e-01
-9.15029466e-01 -1.06231046e+00 -2.82991558e-01 6.62208021e-01
1.11418355e+00 -1.49919420e-01 4.71887290e-02 -7.32342601e-01
6.03913546e-01 4.90300924e-01 -3.70172888e-01 -3.71731579e-01
-9.09924626e-01 -1.97290614e-01 3.11390519e-01 4.58214283e-01
2.20365599e-01 -9.04808581e-01 -9.68278050e-01 5.83262384e-01
-2.78439969e-02 -1.17731917e+00 -6.50190771e-01 2.25314572e-02
-8.50021720e-01 -8.48163307e-01 -4.61901516e-01 -5.66844523e-01
5.90789735e-01 1.00195694e+00 6.34763360e-01 2.72858918e-01
-6.94748759e-02 1.46370113e-01 -3.80220830e-01 -4.02866960e-01
-3.20739269e-01 -4.99189794e-01 4.53913122e-01 -2.63613999e-01
6.81281388e-01 -1.93416268e-01 -7.10700214e-01 7.62710214e-01
-5.18844187e-01 2.11785007e-02 5.98340511e-01 4.94417161e-01
8.88577580e-01 4.57115658e-03 3.10819983e-01 5.65569289e-02
-3.55430245e-01 -7.75652289e-01 -9.42090869e-01 -1.72343731e-01
-2.02510402e-01 -2.58006960e-01 2.03836486e-01 -6.76305115e-01
-4.29665685e-01 2.59321660e-01 1.34147018e-01 -1.31955099e+00
-9.56348777e-02 -1.34946421e-01 -9.02865082e-02 -2.78066933e-01
2.14433670e-01 3.14868838e-01 -3.43988806e-01 -3.88147563e-01
2.41437331e-01 5.20927370e-01 5.98058581e-01 -7.28447139e-02
1.18747294e+00 9.37868834e-01 1.78260535e-01 -6.38245344e-01
-4.73525912e-01 -9.06316578e-01 -5.53004622e-01 -8.41446161e-01
8.58789802e-01 -1.32862425e+00 -9.76578474e-01 1.52362227e-01
-1.60734642e+00 4.06256365e-03 -1.39867112e-01 6.65295064e-01
-2.86147624e-01 3.47336307e-02 -1.95441425e-01 -9.91588414e-01
-2.32476369e-02 -1.19848740e+00 1.51095057e+00 3.03970098e-01
3.29227656e-01 -4.81526405e-01 -6.01677643e-03 -2.31259927e-01
1.43744081e-01 2.62098610e-01 2.15555564e-01 -2.99130887e-01
-1.58049548e+00 -3.31546992e-01 -4.55056697e-01 -3.33501428e-01
-8.22096616e-02 -1.87197462e-01 -8.39113832e-01 -6.17448092e-01
-1.16998911e-01 2.37028658e-01 1.05057609e+00 4.38000470e-01
6.92157388e-01 7.42772222e-02 -1.33078480e+00 5.64145207e-01
1.26265490e+00 3.30535114e-01 3.27390343e-01 9.85806435e-02
7.53003716e-01 2.00736329e-01 1.21440971e+00 2.59868324e-01
6.47115707e-01 9.36296165e-01 8.22523713e-01 3.31328124e-01
-4.66075808e-01 -2.50199527e-01 6.47022605e-01 8.15732121e-01
3.77086073e-01 -5.70699096e-01 -6.42320275e-01 8.55274916e-01
-2.24128842e+00 -1.04163790e+00 -4.64720309e-01 2.09180140e+00
9.93586853e-02 5.29956877e-01 1.25677079e-01 -1.75324991e-01
8.61873865e-01 1.85010448e-01 -8.18312049e-01 4.10184532e-01
-4.48614508e-02 -6.48242116e-01 7.75623143e-01 3.63883734e-01
-1.15811849e+00 9.25995708e-01 5.24758339e+00 8.43833745e-01
-8.12307239e-01 4.88731951e-01 -2.97497839e-01 -3.25264364e-01
-1.21583425e-01 8.91240016e-02 -1.69100106e+00 6.71288967e-01
8.21795344e-01 3.13058496e-02 -6.30895719e-02 9.08414662e-01
2.93696910e-01 -1.08116269e-01 -1.17095053e+00 9.21787143e-01
5.51381521e-03 -1.42846990e+00 -1.74402282e-01 2.19106287e-01
6.07976437e-01 4.48823184e-01 -1.43067434e-01 3.81077111e-01
1.84188440e-01 -4.23700720e-01 1.15785110e+00 5.65107584e-01
5.94536602e-01 -5.21891713e-01 4.45298314e-01 5.81880927e-01
-1.97037637e+00 -1.74390852e-01 -7.01761425e-01 1.44426078e-01
5.13354063e-01 6.19074047e-01 -8.20351362e-01 5.84626794e-01
8.51726353e-01 1.14540374e+00 -2.71252602e-01 1.64246833e+00
3.13708276e-01 1.86964482e-01 -8.32176626e-01 -2.41589591e-01
3.11100274e-01 9.56209600e-02 1.27806115e+00 9.40272868e-01
5.89596808e-01 1.11850694e-01 6.57987595e-01 1.00620580e+00
1.34434044e-01 -4.48023289e-01 -7.38395095e-01 3.99911582e-01
9.56775069e-01 9.81672287e-01 -7.19097674e-01 -4.20543015e-01
-5.70188522e-01 6.34294689e-01 1.56225577e-01 -1.14274081e-02
-1.17317653e+00 -1.24282509e-01 1.03326297e+00 3.39072160e-02
6.94549322e-01 -5.58090806e-01 1.30460775e-02 -8.23395371e-01
9.33390707e-02 -5.98856173e-02 6.42273948e-02 -6.75457239e-01
-8.73554409e-01 3.71152073e-01 2.44656135e-03 -1.82494736e+00
2.83210754e-01 -3.03132176e-01 -4.59392399e-01 6.01677358e-01
-1.69509912e+00 -1.20498049e+00 -5.29028714e-01 5.22983909e-01
6.04021192e-01 2.08475038e-01 2.68569976e-01 8.32927525e-01
-4.76820499e-01 3.74107122e-01 1.85106620e-01 -1.80466115e-01
6.10361040e-01 -7.17048347e-01 5.18404245e-01 8.20708692e-01
-1.23585835e-02 1.88838884e-01 6.85030222e-01 -1.15351081e+00
-1.66076863e+00 -1.66576040e+00 8.51068318e-01 -6.65533245e-01
4.24104095e-01 -4.39244300e-01 -9.11020458e-01 7.03259170e-01
-2.72046566e-01 4.40148324e-01 -9.06444490e-02 -6.59498334e-01
-3.87462117e-02 -2.36714065e-01 -1.08384860e+00 4.11137223e-01
1.28810728e+00 -1.43222511e-01 -4.76436228e-01 3.69263291e-01
1.12069392e+00 -7.67941892e-01 -6.73483670e-01 5.42264581e-01
5.22489011e-01 -5.85351527e-01 1.11705256e+00 9.16748308e-03
-3.55604827e-01 -1.19919884e+00 -7.08370954e-02 -6.96407795e-01
-4.94988710e-01 -4.02191818e-01 -7.89551616e-01 1.03184080e+00
5.42357797e-03 -4.06082511e-01 9.15837109e-01 1.21184163e-01
-8.28334451e-01 -4.28642988e-01 -1.45328987e+00 -1.06980932e+00
-3.92008901e-01 -5.26534438e-01 8.97223413e-01 4.96288002e-01
-4.19441730e-01 -2.39926428e-01 -4.23032016e-01 7.97646821e-01
8.63493800e-01 3.02222878e-01 1.05642426e+00 -1.31962597e+00
2.92475402e-01 -3.37660789e-01 -9.11126137e-01 -1.70524228e+00
7.15391040e-02 -9.20614302e-01 6.10128999e-01 -1.38251817e+00
8.07816759e-02 -8.96176696e-01 -3.31629336e-01 3.45720828e-01
-2.06110924e-02 3.02373201e-01 3.33964199e-01 5.60144305e-01
-1.14698350e+00 8.18047881e-01 1.19355845e+00 -3.52005243e-01
-2.51592845e-01 1.23088568e-01 2.60264605e-01 3.99988681e-01
2.77152091e-01 -1.04643989e+00 4.23848964e-02 -5.65905213e-01
-1.47159353e-01 1.95466414e-01 7.82249749e-01 -1.48687482e+00
7.46505320e-01 -1.30011857e-01 4.03360516e-01 -1.78327644e+00
1.03213263e+00 -1.26548076e+00 3.13257039e-01 5.69823742e-01
2.80328035e-01 1.78769790e-02 5.29071450e-01 1.13152575e+00
1.08183891e-01 1.09425679e-01 6.77776277e-01 1.74000114e-01
-1.01157606e+00 7.69510448e-01 -3.61966580e-01 -4.97053921e-01
1.49041271e+00 -5.85534275e-01 -4.65879411e-01 2.01323971e-01
-2.48285428e-01 8.05191994e-01 7.31958747e-01 7.18902707e-01
1.02735496e+00 -1.63839984e+00 -6.56251907e-01 1.50561512e-01
4.21014369e-01 2.78486639e-01 3.21586728e-01 1.05116522e+00
-2.19153672e-01 4.89152700e-01 6.13058731e-02 -1.30267334e+00
-1.27795887e+00 7.57696450e-01 1.81553945e-01 3.14813733e-01
-1.14037991e+00 8.39049637e-01 2.13917673e-01 -1.00994773e-01
3.62548888e-01 -5.89309454e-01 1.12594012e-02 -3.71208400e-01
6.23295367e-01 3.93325299e-01 -2.55082726e-01 -1.05421197e+00
-6.99523926e-01 7.29761183e-01 -1.39186457e-02 1.76973343e-01
9.52394128e-01 -3.84443253e-01 2.47895762e-01 5.74980080e-01
8.74103844e-01 -3.07297945e-01 -1.59974563e+00 -3.12923461e-01
-1.92072287e-01 -7.69562900e-01 2.43210390e-01 -1.78833365e-01
-8.30281556e-01 7.39509165e-01 9.45763171e-01 -6.70266757e-03
5.41071713e-01 2.50446051e-01 1.11442077e+00 3.84594887e-01
8.29675734e-01 -5.62324405e-01 -1.64078236e-01 6.01597726e-01
3.68995488e-01 -1.25176632e+00 -2.85372138e-02 -2.10090071e-01
-1.15266047e-01 8.06389630e-01 8.88147533e-01 -3.96967754e-02
5.78176081e-01 1.36634141e-01 -2.54626602e-01 -2.81333148e-01
-8.70330811e-01 -4.06051069e-01 2.77089655e-01 7.67230392e-01
-4.37976480e-01 -4.44116108e-02 2.22384244e-01 1.63125828e-01
3.61165017e-01 -2.68837214e-01 1.78984672e-01 9.80502486e-01
-9.58709240e-01 -7.25541115e-01 -7.05445409e-01 4.41274315e-01
2.25175366e-01 2.92829990e-01 7.56243765e-02 8.02868605e-01
5.20846128e-01 9.67153132e-01 4.20987546e-01 -6.25286520e-01
2.53062546e-01 -2.65113711e-01 2.99216658e-01 -5.39091289e-01
-2.01885164e-01 1.33149713e-01 -2.24557385e-01 -6.46120012e-01
-4.57053304e-01 -9.71457899e-01 -1.67668819e+00 -4.18741375e-01
-9.05332923e-01 -3.81349660e-02 7.10623860e-01 8.88532937e-01
6.23162329e-01 5.82271338e-01 5.75402737e-01 -1.41076624e+00
-7.69483000e-02 -7.50047147e-01 -3.15585792e-01 3.06782103e-03
8.92327249e-01 -1.30265713e+00 -2.42358476e-01 -2.61157364e-01] | [6.63966703414917, -2.21549129486084] |
0ffc687a-f468-4bdc-a900-945021e9a475 | starcraft-micromanagement-with-reinforcement | 1804.00810 | null | http://arxiv.org/abs/1804.00810v1 | http://arxiv.org/pdf/1804.00810v1.pdf | StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer Learning | Real-time strategy games have been an important field of game artificial
intelligence in recent years. This paper presents a reinforcement learning and
curriculum transfer learning method to control multiple units in StarCraft
micromanagement. We define an efficient state representation, which breaks down
the complexity caused by the large state space in the game environment. Then a
parameter sharing multi-agent gradientdescent Sarsa({\lambda}) (PS-MAGDS)
algorithm is proposed to train the units. The learning policy is shared among
our units to encourage cooperative behaviors. We use a neural network as a
function approximator to estimate the action-value function, and propose a
reward function to help units balance their move and attack. In addition, a
transfer learning method is used to extend our model to more difficult
scenarios, which accelerates the training process and improves the learning
performance. In small scale scenarios, our units successfully learn to combat
and defeat the built-in AI with 100% win rates. In large scale scenarios,
curriculum transfer learning method is used to progressively train a group of
units, and shows superior performance over some baseline methods in target
scenarios. With reinforcement learning and curriculum transfer learning, our
units are able to learn appropriate strategies in StarCraft micromanagement
scenarios. | ['Yuanheng Zhu', 'Kun Shao', 'Dongbin Zhao'] | 2018-04-03 | null | null | null | null | ['real-time-strategy-games'] | ['playing-games'] | [-1.75785825e-01 2.42572464e-02 -1.36323586e-01 1.74453735e-01
-4.15589333e-01 -5.71921170e-01 3.44475329e-01 -2.09323376e-01
-9.39456105e-01 1.11992621e+00 -2.07211196e-01 -2.61470765e-01
-2.28986382e-01 -1.00211251e+00 -6.22317195e-01 -8.66615951e-01
-3.59701872e-01 5.29199481e-01 7.54276514e-01 -1.07276070e+00
3.44464630e-01 -1.53319226e-04 -1.06797755e+00 -6.62532225e-02
1.08435297e+00 6.46537364e-01 5.31980991e-01 8.74205112e-01
2.64972806e-01 1.28501499e+00 -1.01178896e+00 -3.21392082e-02
4.59410310e-01 -5.77332616e-01 -7.15072155e-01 -9.56184193e-02
-3.55054349e-01 -5.34098685e-01 -3.87045801e-01 9.70248938e-01
8.61195743e-01 6.67965114e-01 4.33399945e-01 -1.28126299e+00
9.28941146e-02 7.79173136e-01 -6.82673812e-01 1.61569819e-01
6.55303597e-02 3.88466239e-01 8.38550985e-01 -1.72386229e-01
3.59526992e-01 1.21457338e+00 3.63464952e-01 1.01585150e+00
-8.46428692e-01 -9.62245524e-01 2.99262136e-01 2.06832856e-01
-1.00836205e+00 4.20177430e-01 7.10915685e-01 -2.07886528e-02
9.77912843e-01 -2.27714051e-02 1.25075448e+00 1.01365626e+00
5.20207465e-01 9.84561086e-01 1.01088881e+00 -1.09872527e-01
7.27385759e-01 -2.79298484e-01 -6.13338172e-01 6.85119808e-01
1.82907254e-01 6.72990263e-01 -1.26706630e-01 -1.61327660e-01
1.14095783e+00 -1.68524683e-01 1.34143755e-01 -6.62633479e-01
-7.04311252e-01 1.29421473e+00 6.57207727e-01 3.95546760e-03
-4.11880910e-01 4.23192859e-01 5.73545158e-01 6.84996784e-01
6.27269689e-03 1.14025557e+00 -5.42306244e-01 -4.71341133e-01
-2.76534826e-01 7.33771861e-01 7.72103608e-01 4.97813672e-01
5.62486172e-01 6.99930787e-01 6.31800219e-02 8.07920694e-01
-7.06195235e-02 2.90519834e-01 6.12987638e-01 -1.06965256e+00
4.48239833e-01 6.62586331e-01 1.51444286e-01 -6.96313679e-01
-3.44665408e-01 -4.97909784e-01 -4.41400409e-01 9.05354321e-01
-8.74392241e-02 -9.18899655e-01 -7.29207397e-01 1.65906334e+00
3.78028929e-01 5.40215135e-01 4.01023149e-01 9.55286026e-01
1.21718593e-01 8.63683701e-01 3.96321118e-02 -2.65510917e-01
8.72993112e-01 -9.87901628e-01 -4.27774906e-01 -4.46758598e-01
6.84253097e-01 -2.26654951e-02 8.44814539e-01 5.57238400e-01
-1.23234665e+00 -3.62149566e-01 -1.21671009e+00 8.04910004e-01
-9.83180478e-02 -4.51224267e-01 7.41815090e-01 4.60343540e-01
-8.81402850e-01 8.32628012e-01 -9.45222914e-01 -1.44672081e-01
2.14346915e-01 8.89615774e-01 1.26870766e-01 4.37654257e-01
-1.47079742e+00 1.07082534e+00 1.05351114e+00 -2.64180988e-01
-1.40506363e+00 -3.43283594e-01 -7.69995987e-01 8.66314769e-02
9.66747522e-01 -3.44969779e-01 1.53330374e+00 -9.91867185e-01
-2.22989011e+00 2.58112431e-01 1.11033642e+00 -5.52130044e-01
3.81325066e-01 -8.99662077e-02 -6.29226342e-02 2.16085277e-02
-1.13446295e-01 6.82168126e-01 7.28226781e-01 -1.17153811e+00
-1.10273635e+00 3.25690299e-01 6.20387375e-01 8.27451050e-01
-4.56617773e-01 -5.23110330e-02 7.37314522e-02 -6.21659219e-01
-5.28397441e-01 -8.66353571e-01 -7.23482072e-01 -6.81139648e-01
4.70220536e-01 -1.85392097e-01 7.42278397e-01 -2.02312440e-01
1.10724127e+00 -1.82782257e+00 7.70818293e-01 1.52661145e-01
1.51835471e-01 6.54021740e-01 -3.01072955e-01 5.35574257e-01
2.69362748e-01 -3.25855821e-01 -1.75105810e-01 3.40080053e-01
-1.06121190e-01 6.23761952e-01 -1.14249453e-01 3.92935378e-03
7.76140951e-03 9.88523364e-01 -1.27546465e+00 -2.78325528e-01
-2.18381230e-02 -1.95286512e-01 -1.02264833e+00 6.73041761e-01
-3.61476958e-01 4.51386094e-01 -8.21270049e-01 2.49781862e-01
3.12435180e-01 2.85672009e-01 4.16563809e-01 4.86791342e-01
-1.13229088e-01 -1.87798128e-01 -1.24420536e+00 1.44160128e+00
-4.62109953e-01 -5.79080451e-03 2.84555078e-01 -1.36205125e+00
1.21557868e+00 9.22133029e-02 6.41432166e-01 -8.37757647e-01
6.54868841e-01 7.73850679e-02 5.72329521e-01 -2.52337664e-01
4.27924484e-01 -1.82960793e-01 -5.69846094e-01 3.44573617e-01
1.90351501e-01 -8.80997956e-01 2.86635339e-01 7.46082701e-03
1.17991531e+00 3.09104979e-01 4.04479384e-01 -2.98822433e-01
4.92431223e-01 2.58506596e-01 9.18742120e-01 8.14807951e-01
-3.45613420e-01 -9.41910073e-02 6.98813319e-01 -6.23532176e-01
-9.35805023e-01 -9.43243504e-01 6.72465086e-01 1.47403479e+00
3.28266233e-01 -3.97842288e-01 -7.49100924e-01 -6.77445531e-01
-9.51256827e-02 5.69050252e-01 -7.67050087e-01 -9.02654529e-01
-1.01457822e+00 -7.74370015e-01 3.02059114e-01 6.31135702e-01
7.24486053e-01 -1.54083109e+00 -1.14006245e+00 5.98145843e-01
1.51312292e-01 -5.90301216e-01 -5.60011625e-01 5.16247749e-01
-7.27230549e-01 -1.13366318e+00 -5.91760576e-01 -1.05749369e+00
3.93401802e-01 -2.95909550e-02 5.92609644e-01 1.08805001e-01
-1.73369557e-01 2.07663491e-01 -3.55467319e-01 -4.69832718e-01
-3.72001678e-01 8.40503350e-02 3.45540226e-01 -3.88898432e-01
-6.98002055e-02 -6.33630872e-01 -5.69992304e-01 5.22928298e-01
-7.60667086e-01 1.00558817e-01 6.50795102e-01 1.42772746e+00
-4.31781821e-02 6.44190460e-02 6.79942489e-01 -4.68646288e-01
1.31012976e+00 -3.03118885e-01 -9.89575267e-01 1.01678021e-01
-3.90604824e-01 2.05020159e-01 9.16520953e-01 -1.08986807e+00
-8.53004932e-01 -1.24774903e-01 1.89931110e-01 -4.15286869e-01
5.09336650e-01 3.66779476e-01 7.41003156e-02 -3.15222293e-01
7.40778029e-01 2.59400427e-01 2.82330692e-01 1.32300705e-01
1.50939882e-01 3.36784542e-01 3.45210999e-01 -1.05835629e+00
8.57051492e-01 -2.87131488e-01 -1.40176192e-01 -1.79304063e-01
-3.90160143e-01 -5.87756485e-02 -9.77341384e-02 -3.58809352e-01
6.46243811e-01 -7.60970056e-01 -1.30002499e+00 5.91570079e-01
-6.39662385e-01 -9.69228804e-01 -4.60026711e-01 5.39087415e-01
-1.03443134e+00 -9.68430042e-02 -7.78208256e-01 -7.52912998e-01
-2.57281870e-01 -1.08130097e+00 3.23424637e-01 7.43078232e-01
3.01334292e-01 -7.92879760e-01 4.67798412e-01 -1.28498062e-01
4.93558764e-01 2.79350013e-01 5.25617540e-01 -4.34494853e-01
-9.83181447e-02 1.14765435e-01 3.56032312e-01 2.58483231e-01
6.79215193e-02 -3.27298641e-01 -4.49267700e-02 -8.35333526e-01
-9.20531452e-02 -9.83685255e-01 4.29977089e-01 2.41216019e-01
8.64058614e-01 -3.34161520e-01 -2.04742476e-01 5.23464501e-01
1.15330327e+00 9.55409110e-01 5.16918480e-01 7.53361523e-01
3.28234792e-01 2.83399254e-01 9.55938041e-01 5.94339311e-01
2.61217088e-01 5.54613709e-01 8.20983529e-01 -1.08728465e-02
4.40381527e-01 -4.75368202e-01 6.50303781e-01 7.19554961e-01
-4.73249733e-01 -1.23946031e-03 -6.88602805e-01 2.37049475e-01
-2.39660764e+00 -1.01613784e+00 7.09201574e-01 1.87101448e+00
9.82922256e-01 4.54615772e-01 7.09984183e-01 -2.09644079e-01
6.10515535e-01 -8.90042633e-02 -8.70469928e-01 -6.47660077e-01
1.00326337e-01 4.10850227e-01 5.57256877e-01 5.97839415e-01
-9.17335629e-01 1.57964993e+00 6.39910078e+00 1.00341523e+00
-1.16126430e+00 -1.40647367e-01 3.76647115e-01 -1.47774771e-01
1.26833662e-01 -1.01446867e-01 -4.87607777e-01 4.26707536e-01
7.02735722e-01 -5.18682003e-01 8.18444967e-01 1.08208704e+00
1.94394320e-01 -1.28081352e-01 -4.63443309e-01 7.19507158e-01
-1.39402658e-01 -1.17152739e+00 -2.99169689e-01 -1.64072923e-02
9.52126563e-01 -2.06270441e-01 -8.63785297e-02 1.15037596e+00
1.30524909e+00 -9.67336357e-01 4.23284888e-01 -3.54323983e-02
4.69154954e-01 -1.43531621e+00 7.76103079e-01 5.58728933e-01
-1.23461890e+00 -8.23051333e-01 -6.56491160e-01 -5.16323030e-01
4.70139179e-03 -6.92704201e-01 -8.53627861e-01 3.95844609e-01
4.86312270e-01 2.92139858e-01 -2.11269855e-01 1.07006061e+00
-4.20416862e-01 5.00625253e-01 -9.54387933e-02 -8.52651358e-01
8.38524520e-01 -5.36808491e-01 4.38292146e-01 5.98101735e-01
1.23679616e-01 4.77555811e-01 9.27898109e-01 4.42855239e-01
3.82869542e-01 -1.12018059e-03 -5.12130201e-01 7.91496336e-02
1.91348508e-01 1.32670403e+00 -7.46952951e-01 -1.90439418e-01
1.52324751e-01 8.47093761e-01 7.74828851e-01 1.68223947e-01
-1.03113616e+00 -5.83317101e-01 6.87520027e-01 -1.62377387e-01
3.40908200e-01 -3.00286293e-01 2.65062094e-01 -1.03433943e+00
-6.65130138e-01 -1.09929049e+00 4.52180266e-01 -5.05055606e-01
-8.95101130e-01 6.08267784e-01 1.70574427e-01 -1.37110269e+00
-7.41913259e-01 -6.28983438e-01 -1.14841712e+00 3.53300154e-01
-1.12784505e+00 -6.41443789e-01 8.81913900e-02 7.89758444e-01
5.52049279e-01 -9.01151896e-01 6.62316322e-01 -2.95268930e-03
-5.66318274e-01 5.93473852e-01 2.05863807e-02 1.41380012e-01
4.23848838e-01 -1.40289640e+00 2.27571577e-01 5.07014215e-01
-5.29895663e-01 1.31969035e-01 7.31021345e-01 -5.87356687e-01
-1.42049718e+00 -8.21000576e-01 -4.03217405e-01 1.57896072e-01
1.06865096e+00 -5.74917078e-01 -5.24567008e-01 3.46518636e-01
3.12344521e-01 -1.06013440e-01 1.91297933e-01 -9.33967233e-02
2.40581200e-01 -2.14515895e-01 -9.74999487e-01 1.05938745e+00
8.86422336e-01 1.43824399e-01 -6.66682422e-01 -2.02470906e-02
8.55371296e-01 -8.23628664e-01 -4.95166689e-01 2.93782562e-01
2.56304592e-01 -4.18619543e-01 7.56904483e-01 -1.12990057e+00
2.76454777e-01 -1.67475626e-01 2.33950883e-01 -2.05612159e+00
-5.44578135e-01 -9.90794659e-01 -2.18618155e-01 5.91601491e-01
2.29233075e-02 -5.88206708e-01 1.14399016e+00 7.62855038e-02
-1.42383069e-01 -8.36821258e-01 -9.34697628e-01 -8.75585496e-01
5.82576036e-01 2.27265999e-01 2.71290272e-01 8.40130329e-01
4.47958529e-01 4.77545798e-01 -7.56806374e-01 -7.89865032e-02
4.30437595e-01 -1.31311074e-01 7.40646780e-01 -9.04134333e-01
-7.73921072e-01 -4.21655089e-01 -2.89142936e-01 -1.04433680e+00
2.03181073e-01 -5.59282005e-01 1.74229607e-01 -1.15067458e+00
-1.65423244e-01 -4.04544324e-01 -4.88613337e-01 6.60721183e-01
-3.78397852e-01 -2.34738648e-01 4.64087784e-01 -2.74288684e-01
-7.95587420e-01 8.96101594e-01 1.62372959e+00 -2.36613542e-01
-5.36783993e-01 2.16494918e-01 -7.84317911e-01 5.62505662e-01
1.21589386e+00 -3.54417831e-01 -7.50543296e-01 -1.04423910e-01
9.44605917e-02 4.89725679e-01 -4.47026901e-02 -1.19594765e+00
4.20771927e-01 -8.50264013e-01 8.51006731e-02 -7.37066939e-02
3.64162356e-01 -6.19864643e-01 -3.18681121e-01 1.37745607e+00
-1.46464050e-01 4.06889498e-01 3.80846918e-01 3.84467840e-01
-7.84929842e-02 -2.37209603e-01 8.26463223e-01 -3.35744977e-01
-8.47347438e-01 3.75085711e-01 -7.35938609e-01 1.95884734e-01
1.43979537e+00 -2.36009821e-01 -4.85954843e-02 -6.12373173e-01
-4.11420673e-01 1.03994513e+00 -5.40682953e-03 5.37720799e-01
6.79860890e-01 -1.27667975e+00 -8.54204655e-01 -1.92932226e-02
-2.58560717e-01 -1.19524814e-01 1.71052739e-01 1.88789129e-01
-7.30945826e-01 -2.77080268e-01 -9.95401382e-01 -2.65343606e-01
-9.44919229e-01 4.59437460e-01 8.01459968e-01 -6.27911985e-01
-4.44639683e-01 6.95773005e-01 1.70145243e-01 -8.48459661e-01
1.65781334e-01 4.78303097e-02 -3.26080412e-01 -3.00784975e-01
5.11060238e-01 2.70729065e-01 -4.44700301e-01 1.27612099e-01
-1.19246222e-01 3.16658497e-01 -3.46905142e-01 -4.04601008e-01
1.52188885e+00 4.74966586e-01 3.88729066e-01 -7.91908354e-02
5.32613933e-01 -3.84287268e-01 -1.87224984e+00 -2.06295913e-03
-2.31734976e-01 -2.15265945e-01 -1.02236249e-01 -8.64644110e-01
-1.06914711e+00 4.48876321e-01 4.96169865e-01 3.55960280e-02
1.11449182e+00 -5.77527404e-01 4.84228492e-01 6.31153405e-01
6.87231719e-01 -1.50549507e+00 7.99776912e-01 9.54763710e-01
8.44112754e-01 -9.71210241e-01 -2.02675551e-01 6.43317997e-02
-1.14397609e+00 1.11214590e+00 1.52838922e+00 -8.02082479e-01
1.56334922e-01 5.89167774e-01 2.17433795e-02 -4.46467055e-03
-9.36990261e-01 -3.11442763e-02 -4.03999150e-01 8.70236993e-01
-2.61173397e-01 -2.93234189e-04 -5.77029586e-01 4.70540494e-01
-2.83363581e-01 -3.01807433e-01 7.09251761e-01 1.23987687e+00
-9.36228991e-01 -1.40751684e+00 -2.12461665e-01 8.55578780e-02
2.18856167e-02 3.12971473e-01 -1.98115587e-01 9.52663720e-01
4.84799333e-02 7.22215116e-01 1.54202534e-02 -5.85184813e-01
4.43755448e-01 -5.95233262e-01 5.51117718e-01 -5.65139949e-01
-1.05445695e+00 9.30294916e-02 -1.27832815e-01 -5.29842794e-01
4.11610417e-02 -2.76535332e-01 -1.46947968e+00 -3.77434760e-01
-1.84798077e-01 6.64421320e-01 1.56781927e-01 6.70629799e-01
-2.61671282e-03 9.05446112e-01 9.88563001e-01 -7.97301471e-01
-1.08523178e+00 -8.02103221e-01 -6.12824857e-01 1.10959515e-01
-4.99304570e-02 -9.76475716e-01 4.13393602e-02 -6.81840241e-01] | [3.6259562969207764, 1.6431585550308228] |
b3dca677-345c-4494-8424-27a3eb805e13 | rcsearcher-reaction-center-identification-in | 2301.12071 | null | https://arxiv.org/abs/2301.12071v1 | https://arxiv.org/pdf/2301.12071v1.pdf | RCsearcher: Reaction Center Identification in Retrosynthesis via Deep Q-Learning | The reaction center consists of atoms in the product whose local properties are not identical to the corresponding atoms in the reactants. Prior studies on reaction center identification are mainly on semi-templated retrosynthesis methods. Moreover, they are limited to single reaction center identification. However, many reaction centers are comprised of multiple bonds or atoms in reality. We refer to it as the multiple reaction center. This paper presents RCsearcher, a unified framework for single and multiple reaction center identification that combines the advantages of the graph neural network and deep reinforcement learning. The critical insight in this framework is that the single or multiple reaction center must be a node-induced subgraph of the molecular product graph. At each step, it considers choosing one node in the molecular product graph and adding it to the explored node-induced subgraph as an action. Comprehensive experiments demonstrate that RCsearcher consistently outperforms other baselines and can extrapolate the reaction center patterns that have not appeared in the training set. Ablation experiments verify the effectiveness of individual components, including the beam search and one-hop constraint of action space. | ['Fei Ma', 'Zhenfu Liu', 'Binjie Hong', 'Zuo Zeng', 'Zixun Lan'] | 2023-01-28 | null | null | null | null | ['retrosynthesis'] | ['medical'] | [ 4.36738253e-01 2.75581509e-01 -8.26250136e-01 2.09596425e-01
-2.71885693e-01 -9.25316334e-01 6.79452181e-01 2.12575167e-01
-2.65309196e-02 7.28416145e-01 2.21244916e-01 -7.49137461e-01
-2.89022103e-02 -9.42677081e-01 -7.53433645e-01 -1.14232838e+00
3.91816050e-02 4.59682673e-01 1.35442942e-01 -3.93262029e-01
4.14023936e-01 9.07019556e-01 -8.35855067e-01 7.11958557e-02
5.34327567e-01 5.84916830e-01 1.07036062e-01 4.22059327e-01
-2.98914053e-02 8.33963513e-01 -6.31858766e-01 -1.91847518e-01
3.86261195e-01 -6.39686704e-01 -9.19536710e-01 -1.53523266e-01
6.77988976e-02 -1.98761925e-01 -4.30898428e-01 1.04481232e+00
4.89132643e-01 3.76860291e-01 7.24420607e-01 -1.21889746e+00
-3.69736850e-01 1.17954588e+00 -4.17520314e-01 9.07214507e-02
5.62075853e-01 3.30208242e-02 1.36635149e+00 -1.13964522e+00
1.04403651e+00 9.57963526e-01 3.00295144e-01 4.08353567e-01
-1.07933533e+00 -8.16871524e-01 3.19108993e-01 2.00974748e-01
-1.31435144e+00 -4.54797536e-01 1.02215636e+00 -1.69301167e-01
1.60264957e+00 9.64752808e-02 7.66796112e-01 1.06792974e+00
3.48952860e-01 4.12087619e-01 3.20920348e-01 -4.81818169e-01
3.73290241e-01 -3.38815004e-01 -2.71822631e-01 8.99469018e-01
2.41674960e-01 4.58440334e-01 -5.75833797e-01 -5.40769398e-02
6.95298910e-01 4.84003723e-02 -1.04126863e-01 -6.38650894e-01
-1.28899765e+00 9.99352217e-01 9.09431100e-01 4.26076531e-01
-4.77619022e-01 2.77752876e-01 2.89507449e-01 -1.73588291e-01
-1.06750481e-01 1.10047209e+00 -2.17688024e-01 3.51056129e-01
-6.15049005e-01 1.64876729e-01 8.39868009e-01 1.24017060e+00
9.11525667e-01 1.88947126e-01 -1.24901630e-01 1.09897718e-01
4.48283255e-02 -1.45487189e-01 9.52182859e-02 -7.58297801e-01
4.40560460e-01 9.19948637e-01 1.47393659e-01 -6.03405356e-01
-8.24366450e-01 -4.86447126e-01 -7.03944504e-01 -2.74140202e-03
2.48447239e-01 -3.60583574e-01 -9.55620646e-01 1.66319966e+00
7.04971313e-01 -1.10076390e-01 1.62364826e-01 7.89176524e-01
9.62561846e-01 9.43901420e-01 1.12597764e-01 -5.12519002e-01
9.19720829e-01 -1.49151766e+00 -6.99747503e-01 5.91243356e-02
1.09283113e+00 -6.19410813e-01 2.84447253e-01 3.26130658e-01
-1.08913028e+00 -2.70185143e-01 -1.47733164e+00 1.27207115e-01
-8.26046646e-01 -5.16118035e-02 1.04022992e+00 4.29615796e-01
-6.91425443e-01 1.05615830e+00 -4.25974071e-01 -6.39124513e-02
-2.29892433e-01 6.27909780e-01 -4.24012721e-01 6.44656792e-02
-1.43653643e+00 8.46964359e-01 1.02066183e+00 1.97433069e-01
-1.23427939e+00 -7.86163032e-01 -9.07518506e-01 1.26306593e-01
1.10629284e+00 -5.62481821e-01 1.47731030e+00 -7.95680881e-01
-1.49178720e+00 2.53215343e-01 -6.61752373e-02 -1.27429500e-01
2.23726958e-01 4.34673667e-01 -5.23121715e-01 1.73509240e-01
1.37825862e-01 7.03910708e-01 6.14417374e-01 -1.15474975e+00
-5.94282091e-01 -5.02667092e-02 1.27120361e-01 4.13079977e-01
1.99363768e-01 -6.59563392e-02 -2.74740547e-01 -3.81229788e-01
2.89039105e-01 -9.17368293e-01 -4.02068317e-01 -7.18840063e-01
-1.07554030e+00 -5.88442624e-01 3.79247308e-01 -6.55485243e-02
1.20617700e+00 -1.67707992e+00 2.22515002e-01 6.02162421e-01
2.68051267e-01 -2.30358937e-03 -1.12531058e-01 1.04530561e+00
-8.73503149e-01 4.25745636e-01 3.16900581e-01 3.23261678e-01
2.14904863e-02 -2.83961117e-01 -2.05174521e-01 6.14209771e-01
9.82700661e-02 8.36752474e-01 -1.26493859e+00 -1.83429644e-01
8.46804902e-02 3.52408737e-02 -3.28512758e-01 -9.85366404e-02
-5.19969583e-01 3.19085330e-01 -5.63831747e-01 9.21827555e-01
4.10362393e-01 -3.88067454e-01 5.98115504e-01 -5.71315527e-01
-4.17538673e-01 5.60924649e-01 -1.07987046e+00 1.75057197e+00
1.94834396e-01 -1.46539241e-01 -4.06502992e-01 -7.16989160e-01
8.43842924e-01 4.57029402e-01 7.06891775e-01 -7.59268224e-01
1.38866818e-02 2.41032198e-01 3.79263461e-01 3.50084417e-02
6.60090625e-01 -1.66164264e-01 1.39687136e-01 5.15356779e-01
2.86213279e-01 -7.71864364e-03 8.15828085e-01 3.86210799e-01
9.60385740e-01 2.10831180e-01 6.28969491e-01 -2.27546707e-01
6.09044313e-01 9.77171808e-02 6.51114285e-01 8.32889915e-01
-1.06918752e-01 1.52380735e-01 7.23623753e-01 -5.33371449e-01
-1.10380113e+00 -7.16217756e-01 2.68618643e-01 1.23670340e+00
3.20950896e-01 -8.20780635e-01 -5.71816027e-01 -7.97389150e-01
-2.99430281e-01 6.89196527e-01 -6.92621768e-01 -5.82940698e-01
-4.94045883e-01 -3.25332016e-01 2.45103240e-01 5.58147430e-01
1.82432994e-01 -1.11258936e+00 2.09607948e-02 4.20546472e-01
2.18699425e-01 -7.22240567e-01 -9.05261278e-01 6.28242731e-01
-4.95341063e-01 -1.42057657e+00 -2.63307363e-01 -7.45719850e-01
6.87481642e-01 4.60832596e-01 9.40624356e-01 4.47213203e-02
-2.43383408e-01 -6.79470450e-02 -3.17189723e-01 -3.63324344e-01
-6.63196027e-01 2.82617718e-01 -2.04917360e-02 -2.63890594e-01
8.57062340e-02 -1.98752403e-01 -9.13474023e-01 3.53301853e-01
-5.03118932e-01 2.53495853e-02 5.12172818e-01 6.05348110e-01
5.72017789e-01 2.90661871e-01 4.15588647e-01 -7.87632704e-01
4.90658045e-01 -5.28340816e-01 -6.48983717e-01 4.21006233e-01
-9.30731058e-01 2.83285528e-01 7.69914687e-01 -3.58360559e-01
-8.63088191e-01 3.36994797e-01 1.53232023e-01 -2.95751542e-01
2.07167625e-01 7.45194137e-01 -3.28741789e-01 7.91622996e-02
6.76338255e-01 1.40920416e-01 -3.35359067e-01 -2.20944807e-02
5.00380397e-01 -1.47674978e-01 2.83473641e-01 -6.60363078e-01
5.28045177e-01 1.54468521e-01 5.17425835e-01 -4.53150213e-01
-6.19757771e-01 -5.69482505e-01 -6.03745043e-01 -2.31848836e-01
8.60107064e-01 -7.81177282e-01 -1.16372550e+00 -9.86392945e-02
-1.22025728e+00 -3.24557751e-01 -2.07725212e-01 3.09986115e-01
-3.56584966e-01 5.56926966e-01 -5.51199377e-01 -5.88957489e-01
-4.80395496e-01 -1.44030213e+00 7.80712187e-01 1.82005808e-01
-3.15842509e-01 -9.01459396e-01 1.29506871e-01 -5.96495606e-02
-2.47352690e-01 1.90816700e-01 1.19802475e+00 -1.06889129e+00
-6.69000328e-01 -4.03442889e-01 2.79741138e-01 -4.19538349e-01
3.39155763e-01 4.32004988e-01 -5.90317547e-01 -3.93111646e-01
-6.59875393e-01 -2.25412786e-01 9.26430047e-01 3.01933795e-01
8.53100300e-01 -1.29784316e-01 -6.28129303e-01 3.53175282e-01
1.06275833e+00 9.00005341e-01 3.09597969e-01 3.51641119e-01
8.71598899e-01 5.37924528e-01 4.98026729e-01 4.17047292e-01
-1.46273717e-01 4.50975209e-01 9.37121630e-01 -2.09897637e-01
1.04840212e-01 -6.99837923e-01 5.90247750e-01 2.84884900e-01
-3.24529618e-01 -6.11970365e-01 -6.37436450e-01 2.20215283e-02
-1.72273552e+00 -1.07186604e+00 9.74511728e-03 2.05539441e+00
6.74555480e-01 2.42852703e-01 3.70233357e-01 1.53542206e-01
9.18089271e-01 2.01722011e-01 -9.95077372e-01 -6.34553969e-01
-5.05706901e-03 1.59928188e-01 7.61423647e-01 4.74250555e-01
-1.03519464e+00 1.28830826e+00 7.41024542e+00 7.06650615e-01
-9.67396915e-01 -4.36458588e-01 3.57792705e-01 1.86690167e-02
-3.35083991e-01 3.92238259e-01 -1.02922928e+00 -1.06270418e-01
6.91620290e-01 -1.27353206e-01 6.37895226e-01 8.56379926e-01
7.47932270e-02 2.03464240e-01 -1.61383569e+00 4.99698639e-01
-2.88759023e-01 -1.97108603e+00 2.99011379e-01 1.95138544e-01
8.05459917e-01 -3.95909518e-01 -1.09701529e-01 2.45299652e-01
6.21375918e-01 -9.46422160e-01 9.10529375e-01 -5.94472550e-02
5.64325094e-01 -1.05694830e+00 1.84044868e-01 -3.44687179e-02
-1.36659348e+00 -1.11880571e-01 -1.68918744e-01 1.89529717e-01
-1.87290579e-01 1.74976423e-01 -1.44297922e+00 8.68711889e-01
1.37339801e-01 8.80618513e-01 -1.23040341e-01 6.47268772e-01
-6.46097481e-01 2.41470531e-01 -3.10117584e-02 -4.09500122e-01
5.86374402e-01 -6.39900506e-01 3.30012113e-01 8.95305693e-01
3.38714480e-01 -1.09226279e-01 3.91938627e-01 1.22134924e+00
-3.34910274e-01 1.64658070e-01 -8.47608805e-01 -5.23157597e-01
6.48559272e-01 1.37263775e+00 -1.01073349e+00 -3.44342500e-01
-4.02678490e-01 6.32483959e-01 3.15694243e-01 6.56055927e-01
-8.14485848e-01 -4.19690162e-01 3.97858202e-01 -2.81474501e-01
5.08157849e-01 -3.40443105e-02 2.48536989e-01 -5.32746375e-01
-5.26744485e-01 -8.15266371e-01 6.74024224e-01 -7.41006792e-01
-8.27296317e-01 2.00011328e-01 -1.68050259e-01 -8.78280640e-01
-1.12473451e-01 -7.65656352e-01 -9.40516531e-01 5.85979164e-01
-1.16817141e+00 -9.09657657e-01 1.66695625e-01 6.28250241e-01
4.93197471e-01 -2.07290396e-01 8.11729014e-01 -9.83028039e-02
-1.17425299e+00 4.67578381e-01 3.98198627e-02 -1.18339196e-01
6.58959150e-01 -1.39300108e+00 2.83284903e-01 8.76857221e-01
7.15177953e-02 7.85088599e-01 5.76811254e-01 -8.45646322e-01
-1.88180137e+00 -9.00088966e-01 4.95439887e-01 -1.96637679e-02
8.51879418e-01 -5.49984515e-01 -3.46970230e-01 7.93223321e-01
4.08599138e-01 -2.72342443e-01 6.56790018e-01 7.10007027e-02
-3.51840407e-01 -5.99039160e-03 -6.52716100e-01 1.09703898e+00
1.09565151e+00 -4.22011375e-01 3.71549977e-04 8.58513176e-01
1.04538405e+00 -4.52318370e-01 -1.05439854e+00 2.16213793e-01
1.21671632e-01 -7.54732847e-01 1.22840142e+00 -9.70129371e-01
4.96518970e-01 -3.18082035e-01 2.52947301e-01 -1.28202510e+00
-7.83010542e-01 -1.12869120e+00 -3.48835826e-01 7.87893653e-01
9.36348796e-01 -3.87411892e-01 9.81365621e-01 5.16446054e-01
-6.61223948e-01 -4.75766927e-01 -7.46329546e-01 -6.19012773e-01
7.16064349e-02 2.36508474e-01 9.50642347e-01 1.02610457e+00
5.69995999e-01 9.14908469e-01 -2.66431123e-01 7.51793459e-02
2.95242339e-01 4.41252947e-01 4.80030268e-01 -9.35475647e-01
-1.88419208e-01 -7.12850034e-01 3.90148580e-01 -7.85440147e-01
2.67495453e-01 -1.38009799e+00 3.72494869e-02 -1.41621482e+00
1.07336894e-01 -2.38684073e-01 -5.95755696e-01 5.67233682e-01
-2.72184517e-02 -4.82955247e-01 -8.05480182e-02 -1.88392121e-02
-6.63490593e-01 4.58099872e-01 1.66904593e+00 -4.83665228e-01
-4.12167311e-01 -2.62259722e-01 -8.53355825e-01 4.19306576e-01
9.49913859e-01 -4.29466039e-01 -4.88513589e-01 3.51231843e-01
8.00973117e-01 3.79734069e-01 -2.85150319e-01 -5.38218319e-01
5.36697447e-01 -4.96692598e-01 3.68247867e-01 -8.14424038e-01
2.62837380e-01 -6.30767107e-01 3.41905832e-01 8.15007269e-01
-4.68066096e-01 9.42853913e-02 -6.95143417e-02 7.13886023e-01
2.25999862e-01 -4.41983849e-01 4.41689044e-01 -5.21950245e-01
-6.47801280e-01 6.15276039e-01 -5.74991286e-01 -5.51930428e-01
1.21626651e+00 -3.39801282e-01 -5.15084743e-01 -6.96641058e-02
-6.52754962e-01 1.91241518e-01 1.45034283e-01 3.15833122e-01
5.25667489e-01 -1.29051054e+00 -1.43178657e-01 -3.03568095e-02
2.24655151e-01 2.05503464e-01 -2.02820659e-01 7.59350896e-01
-6.32706583e-01 6.81067586e-01 8.44916403e-02 -1.14589758e-01
-9.36076403e-01 1.39272904e+00 5.30455351e-01 -3.82248729e-01
-3.77760440e-01 7.46290565e-01 2.86400020e-01 -3.56293738e-01
3.23214591e-01 -4.81723726e-01 -1.58411235e-01 1.06801607e-01
2.06258282e-01 3.63197535e-01 1.67591169e-01 -3.19788694e-01
-3.62969071e-01 1.91775233e-01 -4.87984717e-01 3.58548880e-01
9.63762343e-01 4.82588798e-01 -1.42198220e-01 1.21373326e-01
9.76108968e-01 -1.58960119e-01 -9.51997519e-01 -9.69660729e-02
-1.74227972e-02 1.44372031e-01 7.88160637e-02 -6.73389852e-01
-9.97245729e-01 4.61547911e-01 4.60313484e-02 1.85734287e-01
5.67995727e-01 2.12648045e-02 3.15114766e-01 9.95112479e-01
-1.67559423e-02 -1.28807580e+00 4.26496863e-01 3.54292274e-01
8.30870330e-01 -1.01756549e+00 3.45602453e-01 -5.89297175e-01
-4.22072947e-01 1.46794689e+00 6.57240808e-01 4.71087620e-02
2.76221246e-01 -1.20095059e-01 -5.15805542e-01 -4.29257631e-01
-6.54369414e-01 -7.85377324e-02 -4.20056731e-02 1.41022608e-01
4.79842424e-01 -9.95842814e-02 -1.74108818e-01 1.13870211e-01
-9.19177663e-03 -4.79872674e-01 4.48677778e-01 1.03779662e+00
-4.15446788e-01 -1.28897858e+00 -2.31764972e-01 -9.34124142e-02
-1.48618475e-01 -2.33237863e-01 -9.77426291e-01 8.97733212e-01
-1.03399493e-02 1.08423591e+00 -1.92794099e-01 -4.36961114e-01
2.76837349e-01 1.87579915e-01 6.78939283e-01 -4.55908209e-01
-6.67650759e-01 3.26129973e-01 3.97477269e-01 -7.00040877e-01
-3.67551565e-01 -3.95123154e-01 -1.67691267e+00 -3.90860409e-01
-6.65903866e-01 4.05590475e-01 6.83587730e-01 7.59219408e-01
2.89395869e-01 6.96839511e-01 8.33145678e-01 -8.63083303e-01
-4.72526282e-01 -7.04795182e-01 -6.46188378e-01 1.14118256e-01
2.14234322e-01 -8.01553309e-01 -8.37501585e-02 -2.95850158e-01] | [4.496026515960693, 6.1087965965271] |
c6bfb9df-4517-44ca-b81c-a3715a49c923 | brain-anatomy-prior-modeling-to-forecast | 2306.11837 | null | https://arxiv.org/abs/2306.11837v2 | https://arxiv.org/pdf/2306.11837v2.pdf | Brain Anatomy Prior Modeling to Forecast Clinical Progression of Cognitive Impairment with Structural MRI | Brain structural MRI has been widely used to assess the future progression of cognitive impairment (CI). Previous learning-based studies usually suffer from the issue of small-sized labeled training data, while there exist a huge amount of structural MRIs in large-scale public databases. Intuitively, brain anatomical structures derived from these public MRIs (even without task-specific label information) can be used to boost CI progression trajectory prediction. However, previous studies seldom take advantage of such brain anatomy prior. To this end, this paper proposes a brain anatomy prior modeling (BAPM) framework to forecast the clinical progression of cognitive impairment with small-sized target MRIs by exploring anatomical brain structures. Specifically, the BAPM consists of a pretext model and a downstream model, with a shared brain anatomy-guided encoder to model brain anatomy prior explicitly. Besides the encoder, the pretext model also contains two decoders for two auxiliary tasks (i.e., MRI reconstruction and brain tissue segmentation), while the downstream model relies on a predictor for classification. The brain anatomy-guided encoder is pre-trained with the pretext model on 9,344 auxiliary MRIs without diagnostic labels for anatomy prior modeling. With this encoder frozen, the downstream model is then fine-tuned on limited target MRIs for prediction. We validate the BAPM on two CI-related studies with T1-weighted MRIs from 448 subjects. Experimental results suggest the effectiveness of BAPM in (1) four CI progression prediction tasks, (2) MR image reconstruction, and (3) brain tissue segmentation, compared with several state-of-the-art methods. | ['Mingxia Liu', 'Guy G. Potter', 'Shijun Qiu', 'David C. Steffens', 'Li Wang', 'Lihong Wang', 'Jinjian Wu', 'Lintao Zhang'] | 2023-06-20 | null | null | null | null | ['trajectory-prediction', 'image-reconstruction', 'mri-reconstruction', 'anatomy'] | ['computer-vision', 'computer-vision', 'computer-vision', 'miscellaneous'] | [ 4.16244507e-01 1.92740753e-01 -1.30100548e-01 -6.81854367e-01
-1.06390870e+00 -1.46337256e-01 3.49023938e-01 -1.98476896e-01
-5.61700940e-01 7.62757599e-01 5.97227633e-01 -5.57733119e-01
-1.05093040e-01 -6.75601542e-01 -8.15210402e-01 -4.37365413e-01
-4.79664207e-01 8.73232961e-01 3.76354784e-01 3.28699499e-01
3.56078856e-02 1.57653578e-02 -9.90142643e-01 5.94386458e-01
1.07833433e+00 1.12080014e+00 7.27402091e-01 4.61294830e-01
5.12538990e-03 9.54422951e-01 -3.25521648e-01 -1.55569062e-01
4.18794043e-02 -2.00890049e-01 -1.16964734e+00 1.23828033e-03
1.68959111e-01 -4.84083891e-01 -4.48535949e-01 9.46787357e-01
7.28050530e-01 -1.34183347e-01 6.35325134e-01 -9.40364778e-01
-7.70837247e-01 7.30736077e-01 -6.04712307e-01 7.19341695e-01
-1.80065319e-01 2.74673611e-01 6.70713603e-01 -6.63032472e-01
6.03123784e-01 9.23638284e-01 6.86384141e-01 5.65134525e-01
-1.12771499e+00 -8.25767219e-01 2.81568289e-01 6.91295743e-01
-1.30858970e+00 -2.40191981e-01 5.87791324e-01 -8.93481374e-01
7.59215474e-01 8.58036987e-03 7.91662335e-01 9.36351240e-01
4.70247269e-01 6.63315594e-01 1.41111720e+00 7.51980841e-02
1.09335810e-01 -3.52000237e-01 3.77208859e-01 7.47477531e-01
-1.80631623e-01 1.91129059e-01 6.14902154e-02 -2.12735776e-02
7.82260954e-01 -1.35886818e-01 -3.91614556e-01 6.28011599e-02
-1.32586372e+00 5.83887994e-01 8.36522698e-01 1.34429380e-01
-6.62163734e-01 -1.23278737e-01 3.83066267e-01 -4.88912798e-02
6.05816901e-01 -3.62393558e-02 -5.80098867e-01 3.53705615e-01
-1.16500962e+00 -2.43993998e-02 3.53367507e-01 7.07371175e-01
4.12921429e-01 -2.26851955e-01 -5.19445658e-01 1.09587455e+00
2.74313182e-01 1.90138757e-01 9.08993185e-01 -6.82397008e-01
7.31356800e-01 5.50494969e-01 -5.07522881e-01 -3.66002321e-01
-8.20078552e-01 -7.98458397e-01 -9.75562274e-01 5.61157055e-02
3.55493039e-01 -3.47233713e-01 -1.32915831e+00 1.80029070e+00
5.56378327e-02 3.28929365e-01 -4.20593083e-01 1.12017179e+00
7.49146521e-01 3.43280494e-01 3.50510120e-01 8.18517245e-03
1.55516338e+00 -1.24145329e+00 -2.29422003e-01 -5.24443030e-01
7.85296917e-01 -2.67958909e-01 1.08807850e+00 1.33608371e-01
-1.02940702e+00 -4.65622544e-01 -7.05544651e-01 -1.41526461e-01
3.89731303e-02 3.20541292e-01 5.74234307e-01 4.21337008e-01
-1.33423650e+00 1.63671389e-01 -9.75211561e-01 1.40049577e-01
9.34223711e-01 5.26202381e-01 -3.69224608e-01 -3.18015963e-01
-1.15981579e+00 1.07838416e+00 4.34489399e-01 8.48813206e-02
-1.31036079e+00 -1.26576185e+00 -4.41190302e-01 1.01458602e-01
3.08008373e-01 -9.80113804e-01 1.11716032e+00 -7.59011149e-01
-1.23823035e+00 1.14977705e+00 -2.39790946e-01 -5.93150079e-01
5.72350144e-01 -2.31722347e-03 -4.60085452e-01 2.08874166e-01
4.52555418e-01 8.72677684e-01 6.32131457e-01 -1.01590002e+00
-5.98707438e-01 -6.91231430e-01 -1.10569209e-01 1.15706302e-01
2.13664188e-03 9.52686220e-02 -3.28764558e-01 -6.61218464e-01
-7.86274076e-02 -9.07894969e-01 -5.30734122e-01 -4.16580066e-02
-5.76311767e-01 2.18153913e-02 3.50323290e-01 -1.45851648e+00
9.63786244e-01 -1.64132929e+00 -2.57938192e-03 2.07153946e-01
6.63374543e-01 -9.90734342e-03 -1.49674460e-01 -5.69630504e-01
-4.86836284e-01 1.57597691e-01 -5.11677623e-01 -2.91123420e-01
-3.28705251e-01 -5.77938072e-02 2.36805663e-01 3.67627472e-01
-8.79358873e-02 1.13879323e+00 -6.34219289e-01 -5.80235064e-01
-1.09722726e-01 3.36962879e-01 -8.21246922e-01 1.59732401e-01
1.28101170e-01 1.08335662e+00 -4.84597027e-01 4.85802174e-01
4.73578393e-01 -5.88794053e-01 1.32369950e-01 -3.25620234e-01
-2.31130868e-02 4.62162048e-01 -4.25889641e-01 1.71265352e+00
-4.14695144e-01 1.28264725e-01 1.82710290e-01 -1.48863351e+00
5.54929793e-01 4.30759698e-01 7.76985288e-01 -9.38878298e-01
1.03598043e-01 1.70441166e-01 5.84748209e-01 -5.57670295e-01
-5.20762920e-01 -2.21819922e-01 2.61467934e-01 6.28018081e-01
5.62833957e-02 3.25018436e-01 1.01504333e-01 -7.83795118e-02
1.35865462e+00 1.16622243e-02 -6.74306452e-02 -4.05684739e-01
8.97536933e-01 9.67589319e-02 7.75006354e-01 5.09085357e-01
-4.29050416e-01 4.97248620e-01 2.89697498e-01 -4.35500741e-01
-1.08929956e+00 -1.04115283e+00 -3.51704210e-01 9.71135914e-01
-3.79392594e-01 -2.87707746e-01 -1.00604463e+00 -7.49531865e-01
-3.24044883e-01 5.60714006e-01 -7.09921241e-01 -2.50439227e-01
-8.14570546e-01 -1.28572094e+00 5.69064736e-01 9.03450310e-01
7.52833843e-01 -1.06750596e+00 -3.43232274e-01 2.75288224e-01
-6.34245634e-01 -9.66370761e-01 -7.03562617e-01 3.42260897e-02
-1.12629616e+00 -1.17281938e+00 -1.03896725e+00 -9.16569173e-01
7.50824273e-01 -1.49173401e-02 1.00514603e+00 9.19794515e-02
-3.36259931e-01 2.47352690e-01 6.83552632e-03 -2.30885983e-01
-3.22760940e-01 1.49347171e-01 -1.57496914e-01 -4.78779711e-02
8.80233198e-02 -8.68354738e-01 -9.74371552e-01 2.04005033e-01
-5.54663777e-01 6.17041290e-01 1.01471388e+00 8.71176958e-01
9.98949826e-01 -1.02412574e-01 8.06037664e-01 -7.97141433e-01
4.00682658e-01 -7.18255222e-01 -2.67149061e-01 5.09347141e-01
-6.83652461e-01 -5.84841929e-02 4.86493468e-01 -4.37659353e-01
-1.27030933e+00 -1.32038100e-02 -4.70791519e-01 -9.97560248e-02
-2.75783509e-01 9.85940456e-01 -2.52768517e-01 1.83253381e-02
4.59146917e-01 4.60472882e-01 4.43286728e-03 -7.72553265e-01
1.05499662e-01 5.36362946e-01 8.36510122e-01 -7.97854245e-01
2.13094071e-01 2.41812661e-01 -1.56064210e-02 -3.72899085e-01
-9.89191055e-01 -1.88608766e-01 -9.75751817e-01 -2.50439763e-01
1.00695288e+00 -9.41399038e-01 -2.87691206e-01 3.42360884e-01
-9.03288960e-01 -8.22486639e-01 8.38885363e-03 6.94518209e-01
-6.07014060e-01 2.15082765e-01 -8.45975280e-01 -1.46111026e-01
-7.60451138e-01 -1.56551731e+00 7.65693128e-01 -1.43532753e-01
1.00572161e-01 -9.59723353e-01 -1.18612453e-01 9.09198582e-01
4.37584609e-01 -6.24145754e-02 1.55061698e+00 -5.06382108e-01
-4.48200285e-01 6.41680509e-02 -5.52165568e-01 3.06055993e-01
5.91158606e-02 -9.28864062e-01 -7.64242351e-01 -1.69499755e-01
9.67777818e-02 -1.59504473e-01 8.91709030e-01 8.38243127e-01
1.72162306e+00 -1.45472601e-01 -3.80714893e-01 7.88509190e-01
1.04415607e+00 3.78857762e-01 7.12004602e-01 4.28378195e-01
8.08196306e-01 5.18785536e-01 2.80771405e-03 1.57361124e-02
9.76509035e-01 5.42052150e-01 3.00078064e-01 -2.48566605e-02
-6.21546984e-01 7.96742961e-02 3.34152639e-01 1.00349510e+00
-4.60760027e-01 2.52438813e-01 -1.41270471e+00 5.41513741e-01
-1.35130036e+00 -5.84712803e-01 -2.92171001e-01 2.08271575e+00
1.12749851e+00 3.95984426e-02 5.23005500e-02 -1.33605465e-01
7.81167388e-01 -2.02976093e-01 -8.53481293e-01 3.97865981e-01
1.46251604e-01 3.87938261e-01 3.99187505e-01 3.22081894e-01
-1.05581391e+00 7.21301436e-01 5.93252850e+00 5.68028331e-01
-1.07909524e+00 8.42739582e-01 9.44699585e-01 4.71181832e-02
-9.31258053e-02 2.67665926e-02 -6.24244213e-01 6.48427784e-01
1.10680306e+00 -5.48790209e-03 4.43023950e-01 5.23003340e-01
3.02434742e-01 3.68011110e-02 -9.29920554e-01 6.02951050e-01
-1.90676883e-01 -1.26585650e+00 4.15194817e-02 5.17919026e-02
4.53376859e-01 6.27014279e-01 -6.03604391e-02 5.49954414e-01
8.73675756e-03 -1.21870160e+00 7.33840227e-01 8.45524848e-01
1.17649126e+00 -4.27614331e-01 6.88952804e-01 4.53903139e-01
-1.14055312e+00 -1.72006071e-01 -2.08336964e-01 2.28623778e-01
3.74643773e-01 6.21011078e-01 -6.33465350e-01 5.04741967e-01
7.20463693e-01 6.07232034e-01 -8.38165462e-01 1.35144973e+00
-3.77641469e-01 1.13733101e+00 -7.69831240e-02 7.25313842e-01
1.74673378e-01 -1.42116547e-01 3.62088472e-01 1.12632561e+00
2.50498533e-01 2.89883137e-01 5.38211823e-01 1.15546024e+00
-5.83076477e-02 1.42608970e-01 1.35158882e-01 3.03363502e-01
1.37308195e-01 1.10658908e+00 -7.43911386e-01 -5.31967700e-01
-6.71822131e-01 8.08472812e-01 4.46587592e-01 3.22935760e-01
-7.70967245e-01 2.11338118e-01 1.03518322e-01 3.56188297e-01
-1.87743232e-01 -7.67964125e-02 -5.98100066e-01 -1.09296525e+00
-6.82340786e-02 -7.07595050e-01 5.54975688e-01 -8.12200010e-01
-1.39040434e+00 7.17424631e-01 -5.54554462e-02 -7.23096609e-01
1.86995976e-02 -4.02008891e-01 -7.65338421e-01 1.24073780e+00
-1.59777439e+00 -1.42249858e+00 -3.18002790e-01 7.95781076e-01
4.94383782e-01 -1.87032461e-01 7.64945030e-01 7.39117980e-01
-9.08960402e-01 3.44362795e-01 -2.26962924e-01 3.05909514e-01
6.31669760e-01 -1.22269857e+00 1.21983707e-01 6.77339792e-01
-3.26653928e-01 5.47672153e-01 -3.83890197e-02 -1.13493514e+00
-7.04698622e-01 -1.58703411e+00 7.76345015e-01 -2.54564404e-01
6.63252890e-01 -6.87484741e-02 -1.20275390e+00 1.01787961e+00
-2.94569343e-01 2.16463104e-01 7.66189694e-01 3.93372215e-02
-7.68986419e-02 8.38154331e-02 -9.65380847e-01 4.59724098e-01
1.36273491e+00 -4.48101193e-01 -8.45745504e-01 5.42438269e-01
5.77304542e-01 -4.12773460e-01 -1.21839726e+00 4.72593725e-01
4.87502694e-01 -3.42952013e-01 1.34129977e+00 -7.45223165e-01
5.98142982e-01 -8.09273496e-02 5.07400744e-02 -1.27727938e+00
-7.38699555e-01 2.05223680e-01 -1.87588051e-01 9.39975619e-01
4.07723725e-01 -5.39718628e-01 8.34095359e-01 7.81223118e-01
-6.93746924e-01 -9.31301415e-01 -9.64237273e-01 -6.82809055e-01
5.05909145e-01 -6.09955609e-01 5.96281588e-01 7.03948677e-01
-2.64640391e-01 4.18227375e-01 -1.59750208e-01 3.02321732e-01
6.27690375e-01 -7.80032799e-02 1.18666947e-01 -1.35708129e+00
-1.65645719e-01 -6.30483985e-01 -1.75742641e-01 -9.39876854e-01
4.14811194e-01 -1.64705372e+00 -4.54901271e-02 -1.85013926e+00
7.05092788e-01 -7.15359330e-01 -6.38092995e-01 6.79009378e-01
-3.85723889e-01 1.34273738e-01 -4.21328358e-02 4.60248947e-01
-3.20611477e-01 5.23664057e-01 1.61597621e+00 -3.12125385e-01
-1.49339318e-01 2.74153855e-02 -6.38949215e-01 8.91493201e-01
9.00654435e-01 -5.69162667e-01 -6.06890798e-01 -5.20257652e-01
-4.14512038e-01 2.31089592e-01 8.21977258e-01 -1.14547837e+00
2.77696580e-01 4.67942134e-02 7.31783748e-01 -4.80686486e-01
2.11794749e-02 -4.88781601e-01 -7.26520568e-02 6.49136841e-01
-4.64107454e-01 -9.07823443e-02 -7.33854547e-02 3.87258500e-01
5.49102724e-02 -1.83560118e-01 7.99520016e-01 -3.59464943e-01
-6.80875301e-01 9.64876473e-01 -5.40194154e-01 3.15029532e-01
6.32743299e-01 5.10589704e-02 -1.81589663e-01 -9.21903327e-02
-1.24118519e+00 3.51528049e-01 -2.63744295e-01 3.06641400e-01
5.01617849e-01 -1.13279140e+00 -8.49509954e-01 5.64851575e-02
-1.75590336e-01 1.07548960e-01 6.36882365e-01 1.51567471e+00
-1.55555084e-01 6.26438737e-01 -5.35456598e-01 -5.92746258e-01
-8.92006040e-01 5.39725184e-01 4.80433881e-01 -7.41244495e-01
-1.21310675e+00 7.49123394e-01 6.89413428e-01 -5.48801303e-01
1.05214439e-01 -4.01066184e-01 -3.77266794e-01 -2.61779040e-01
7.40867019e-01 2.18006983e-01 2.02524588e-01 -6.80567384e-01
-3.75187606e-01 2.42316604e-01 -3.19436401e-01 -9.23828855e-02
1.59204781e+00 -2.07067877e-01 -2.75337487e-01 6.86145425e-02
1.04412162e+00 -4.94813085e-01 -1.38836670e+00 -4.34240580e-01
2.72890925e-01 8.82754922e-02 4.18291032e-01 -1.32005465e+00
-1.43530226e+00 1.10398281e+00 9.07087266e-01 -7.42464006e-01
1.23406100e+00 4.89187762e-02 1.06356812e+00 -5.14568053e-02
7.38816500e-01 -7.53553152e-01 -1.36109278e-01 4.40147907e-01
1.04821312e+00 -1.09139335e+00 -3.58728319e-01 -3.80513221e-01
-6.98342741e-01 9.46853518e-01 8.43283176e-01 2.26007458e-02
9.13538158e-01 2.93309242e-01 -2.01868922e-01 -3.44545752e-01
-5.65527856e-01 -1.07311942e-01 6.47970140e-01 7.87889898e-01
4.19342458e-01 3.72619212e-01 -3.68490487e-01 1.51973557e+00
-3.02443713e-01 3.33553761e-01 8.71389881e-02 6.05137587e-01
-3.66315603e-01 -1.04185963e+00 -2.39750117e-01 1.22712600e+00
-5.89617550e-01 -3.75361145e-01 -2.62518786e-03 4.07777637e-01
4.96713161e-01 4.88908350e-01 -9.20301229e-02 -3.14360410e-01
1.42206952e-01 3.65637541e-01 4.80321854e-01 -7.45989740e-01
-5.21905065e-01 -8.91593620e-02 4.28021066e-02 -4.69817549e-01
-1.02873079e-01 -7.43856609e-01 -1.40828788e+00 7.84331486e-02
9.20391157e-02 -1.83678001e-01 3.68930906e-01 1.32162893e+00
3.68728936e-01 1.06129336e+00 3.95567436e-03 -6.59801900e-01
-2.73311496e-01 -1.16382968e+00 -4.35037792e-01 1.49563462e-01
-5.59621677e-02 -8.33506644e-01 2.32836753e-01 3.18767607e-01] | [14.364457130432129, -1.919914722442627] |
9f6d4ae9-5941-4310-a5b6-dc8518854c51 | wikisqe-a-large-scale-dataset-for-sentence | 2305.05928 | null | https://arxiv.org/abs/2305.05928v1 | https://arxiv.org/pdf/2305.05928v1.pdf | WikiSQE: A Large-Scale Dataset for Sentence Quality Estimation in Wikipedia | Wikipedia can be edited by anyone and thus contains various quality sentences. Therefore, Wikipedia includes some poor-quality edits, which are often marked up by other editors. While editors' reviews enhance the credibility of Wikipedia, it is hard to check all edited text. Assisting in this process is very important, but a large and comprehensive dataset for studying it does not currently exist. Here, we propose WikiSQE, the first large-scale dataset for sentence quality estimation in Wikipedia. Each sentence is extracted from the entire revision history of Wikipedia, and the target quality labels were carefully investigated and selected. WikiSQE has about 3.4 M sentences with 153 quality labels. In the experiment with automatic classification using competitive machine learning models, sentences that had problems with citation, syntax/semantics, or propositions were found to be more difficult to detect. In addition, we conducted automated essay scoring experiments to evaluate the generalizability of the dataset. We show that the models trained on WikiSQE perform better than the vanilla model, indicating its potential usefulness in other domains. WikiSQE is expected to be a valuable resource for other tasks in NLP. | ['Mamoru Komachi', 'Satoshi Sekine', 'Kenichiro Ando'] | 2023-05-10 | null | null | null | null | ['automated-essay-scoring'] | ['natural-language-processing'] | [-4.79985058e-01 3.42087179e-01 -2.26514101e-01 -2.45815158e-01
-1.04769516e+00 -4.85483915e-01 3.13494802e-01 7.78219819e-01
-5.88390350e-01 1.14955378e+00 4.44693297e-01 -1.04145892e-01
-2.02074185e-01 -8.39797914e-01 -6.34998918e-01 9.00347456e-02
2.74702638e-01 4.32118654e-01 2.54191369e-01 -4.85045463e-01
8.20468962e-01 -3.80558997e-01 -1.33318341e+00 3.42581898e-01
1.83826983e+00 6.08747303e-01 4.55039352e-01 6.82013214e-01
-4.36520308e-01 8.53699744e-01 -1.11068749e+00 -1.02205884e+00
-1.00366883e-01 -3.04160327e-01 -9.86507475e-01 -2.40559384e-01
7.34078765e-01 -6.50421306e-02 -6.69384971e-02 1.30097950e+00
2.58955270e-01 -2.17583254e-01 7.82023311e-01 -1.07447600e+00
-9.54003930e-01 9.21111166e-01 -1.46056607e-01 2.73899287e-01
7.80000627e-01 -1.66787565e-01 1.48225176e+00 -9.67224956e-01
9.09406126e-01 9.99548495e-01 5.97716391e-01 1.91209123e-01
-6.05859637e-01 -4.97717947e-01 -2.33465821e-01 2.91349649e-01
-8.74906361e-01 -2.96247661e-01 5.57223678e-01 -4.88530964e-01
7.01994658e-01 1.38766348e-01 6.70863628e-01 9.76066828e-01
3.89728367e-01 8.29206526e-01 1.18402147e+00 -7.21099436e-01
3.46887521e-02 5.08438587e-01 6.87469363e-01 8.44922841e-01
6.40500426e-01 -6.36322439e-01 -6.89248681e-01 -5.01076356e-02
-8.21899995e-02 -4.60662037e-01 -6.30193233e-01 1.43316224e-01
-1.17598486e+00 7.38812268e-01 1.61674544e-01 2.46676609e-01
-2.51706690e-01 -3.88429224e-01 6.25504971e-01 7.39875436e-01
5.97532868e-01 1.14945662e+00 -5.64811468e-01 -5.82832098e-01
-1.04450297e+00 2.52345711e-01 1.13209105e+00 9.26432729e-01
4.38165158e-01 -5.01850843e-01 -2.83075809e-01 1.10291636e+00
2.95565069e-01 6.21660292e-01 5.45208454e-01 -1.10908461e+00
1.01395309e+00 9.45063829e-01 3.01831365e-01 -1.35733330e+00
-3.53011280e-01 -3.36270958e-01 -6.12234652e-01 -1.71492338e-01
4.81165618e-01 -6.71405196e-02 -2.27221742e-01 1.22604311e+00
-1.43168807e-01 -8.41655254e-01 -6.52035847e-02 5.63055456e-01
1.21418250e+00 8.24323177e-01 -8.53769034e-02 -4.01158035e-01
1.30771112e+00 -1.03150606e+00 -1.09380460e+00 -6.33307472e-02
8.19853067e-01 -9.04410779e-01 1.46934044e+00 7.44819701e-01
-1.22409844e+00 -3.91943067e-01 -1.16950488e+00 -5.02947383e-02
-3.92890841e-01 4.18471694e-01 3.71282965e-01 5.95771015e-01
-8.34318042e-01 7.10077703e-01 -2.82545775e-01 -2.94917643e-01
3.09052080e-01 -3.97214532e-01 -5.03643632e-01 -7.59281814e-02
-1.68768072e+00 1.27896798e+00 1.67148620e-01 -1.64253801e-01
-3.59318554e-01 -3.90375793e-01 -9.59242582e-01 9.48072299e-02
3.45043361e-01 -3.41975629e-01 1.29453266e+00 -6.10769451e-01
-1.05286467e+00 9.20390248e-01 -2.46345773e-01 -9.05596018e-02
3.13821554e-01 -2.09133983e-01 -5.84020793e-01 1.66083530e-01
7.15247691e-01 3.85999195e-02 4.11536425e-01 -8.27407539e-01
-7.19958901e-01 -2.61625290e-01 1.34330571e-01 1.95123553e-01
-8.11190486e-01 1.49358124e-01 -6.45996153e-01 -5.30557513e-01
-1.07685782e-01 -5.07481098e-01 2.78587133e-01 -2.28222147e-01
-4.23468083e-01 -6.15579188e-01 -8.94954950e-02 -1.16262650e+00
1.94264281e+00 -1.68650687e+00 -7.42176250e-02 6.41684160e-02
3.74477953e-01 2.31451645e-01 -6.50376678e-02 6.08220637e-01
6.11906767e-01 6.13033414e-01 1.11282347e-02 -3.21853422e-02
2.04910859e-01 -2.69150168e-01 1.50143728e-01 1.20585404e-01
-3.34237777e-02 7.11428523e-01 -1.18742394e+00 -8.28206420e-01
-6.08205855e-01 -3.52097332e-01 -2.30418101e-01 7.76287317e-02
-7.69693777e-02 -3.43799353e-01 -3.66698176e-01 7.60201275e-01
3.85128766e-01 -3.57421845e-01 -3.91321555e-02 1.44990578e-01
-1.44992813e-01 9.44502652e-01 -1.02337635e+00 1.33156323e+00
-3.91895533e-01 9.43386316e-01 -6.85040876e-02 -4.84684378e-01
1.02692342e+00 1.20364718e-01 -1.90834895e-01 -1.00830936e+00
-1.50853992e-01 4.18122172e-01 -3.69414054e-02 -8.39839220e-01
1.14981949e+00 2.98644394e-01 -3.88884068e-01 6.36223137e-01
5.28457537e-02 -3.23846698e-01 1.03134072e+00 8.61962259e-01
1.19174886e+00 -3.90920162e-01 4.90532666e-01 -2.55743921e-01
4.37576413e-01 3.61538231e-01 6.49399519e-01 7.85495758e-01
-4.07507896e-01 3.31763923e-01 8.23127806e-01 2.01879948e-01
-1.12670922e+00 -7.18506277e-01 -3.17737043e-01 8.03850412e-01
-1.14085181e-02 -1.07120264e+00 -7.99260914e-01 -9.34350908e-01
-5.62032424e-02 7.56056368e-01 -3.82523626e-01 8.07192549e-03
-1.30969971e-01 -5.01911044e-01 5.22363722e-01 7.10075200e-02
4.70943421e-01 -1.11692917e+00 -9.94452909e-02 2.64820009e-01
-7.53190279e-01 -7.15815008e-01 -5.79008460e-01 -1.00543648e-01
-5.86410582e-01 -1.55801666e+00 -4.66402024e-01 -8.38056028e-01
4.98986542e-01 2.26781800e-01 1.47801077e+00 2.60899097e-01
1.23856679e-01 2.83486098e-01 -8.88793707e-01 -6.06448829e-01
-7.59120584e-01 2.78220654e-01 1.42537402e-02 -8.06812286e-01
4.67046857e-01 2.89324433e-01 -1.04663901e-01 1.22500829e-01
-6.11493170e-01 -3.04960817e-01 3.47782969e-01 1.02061450e+00
7.09524825e-02 3.39267880e-01 1.00087655e+00 -1.21221721e+00
1.38486886e+00 -5.13787389e-01 -1.97845235e-01 5.10852098e-01
-9.53974247e-01 -9.63687301e-02 6.51856303e-01 1.23076729e-01
-9.65953410e-01 -8.54209363e-01 -1.80249125e-01 6.81622684e-01
2.99153298e-01 1.06962740e+00 -8.52708519e-02 1.94807827e-01
7.53463984e-01 -1.61044568e-01 5.18660527e-03 -4.00232464e-01
-2.72991583e-02 1.10144734e+00 1.06729887e-01 -4.04914767e-01
6.10352993e-01 -3.55497837e-01 -6.01165652e-01 -8.86316061e-01
-1.21698761e+00 -4.76867735e-01 -3.97065729e-01 -5.36231518e-01
4.17370021e-01 -8.43037307e-01 -2.78618932e-01 4.37133312e-01
-1.19222391e+00 1.95340682e-02 6.77851662e-02 5.25843561e-01
1.24207385e-01 7.68075347e-01 -9.24899578e-01 -5.71614444e-01
-4.23556477e-01 -8.67663383e-01 4.70140278e-01 3.04185838e-01
-5.64386904e-01 -9.23849165e-01 2.32411996e-01 9.08163488e-01
2.12703168e-01 -3.71725529e-01 9.57109690e-01 -6.26635075e-01
-1.49183333e-01 -3.69995952e-01 -1.43978328e-01 6.48876488e-01
-1.04813449e-01 5.92767715e-01 -4.28583443e-01 8.60863850e-02
-1.89548343e-01 -7.09659338e-01 9.13437605e-01 1.44737110e-01
8.91666174e-01 -4.15732086e-01 3.75714339e-02 -2.32139304e-01
1.09029830e+00 -1.19836174e-01 7.01595902e-01 9.05452132e-01
2.77356803e-01 6.59434736e-01 1.19508493e+00 4.70695078e-01
8.09220493e-01 3.95019323e-01 1.81223899e-01 4.31170136e-01
6.45152703e-02 -2.80955940e-01 7.57041514e-01 1.80102170e+00
1.72474861e-01 -3.32958639e-01 -1.00387943e+00 6.02674365e-01
-1.35020494e+00 -1.09994805e+00 -6.86274230e-01 1.95104671e+00
1.31844461e+00 4.59983766e-01 -1.28387511e-01 3.86430532e-01
7.02516317e-01 4.02602516e-02 5.37101515e-02 -5.28494775e-01
-3.06474864e-01 6.56122044e-02 1.16488077e-01 4.08278197e-01
-9.89230156e-01 5.85066080e-01 6.40046549e+00 7.12038875e-01
-4.07422006e-01 1.21728323e-01 3.55189681e-01 1.00189678e-01
-5.71075559e-01 -7.70550817e-02 -9.79729593e-01 8.55044246e-01
8.52226317e-01 -4.21254277e-01 -5.19798733e-02 8.19474101e-01
2.99450696e-01 -5.27201116e-01 -6.60485685e-01 6.49667084e-01
3.89381111e-01 -1.23377597e+00 -1.31823972e-01 -1.88259110e-01
8.44682693e-01 -1.35918662e-01 -2.71969855e-01 8.08633924e-01
3.21735218e-02 -7.48112738e-01 8.03766787e-01 7.03990281e-01
5.10915875e-01 -6.18778110e-01 1.14221609e+00 6.94066882e-01
-6.18656814e-01 8.81166160e-02 -7.88594067e-01 -2.72836536e-01
3.02674919e-02 1.16693032e+00 -4.79543149e-01 2.89362788e-01
6.89618170e-01 8.17528188e-01 -1.15355515e+00 1.25608170e+00
-6.14993632e-01 6.99360251e-01 2.67062336e-01 -7.73064792e-01
6.64979126e-03 -2.00291082e-01 4.25012052e-01 1.28269911e+00
4.59567070e-01 -2.89335579e-01 1.81032978e-02 5.55929124e-01
-6.56419039e-01 6.16411746e-01 -4.88112181e-01 -3.11009258e-01
7.47071445e-01 1.33623040e+00 -3.88084054e-01 -3.74701977e-01
-5.41139662e-01 8.25371385e-01 7.16404319e-01 -2.29473878e-02
-6.48681164e-01 -9.96601224e-01 2.47031614e-01 4.43305857e-02
-1.05227210e-01 -1.57781281e-02 -3.04810733e-01 -1.43956745e+00
4.23932076e-01 -1.03162944e+00 2.03387275e-01 -8.73405159e-01
-1.56653643e+00 4.01202261e-01 -5.90262413e-01 -1.20414984e+00
2.53466725e-01 -6.29321754e-01 -5.62128067e-01 6.68536127e-01
-1.68409085e+00 -3.32058340e-01 -4.48025584e-01 -3.39604393e-02
5.92761397e-01 -2.57025629e-01 5.36602318e-01 4.20733869e-01
-6.04395509e-01 6.05755925e-01 3.18399489e-01 3.15122634e-01
1.35725129e+00 -1.63017774e+00 -3.71368639e-02 7.70433426e-01
3.53595056e-02 7.04553187e-01 5.61727345e-01 -9.63467777e-01
-1.04527533e+00 -1.02626300e+00 1.73565149e+00 -6.81789875e-01
7.80019045e-01 1.06486462e-01 -1.02944720e+00 2.57185429e-01
4.78449225e-01 -8.47458184e-01 8.27637911e-01 5.54734111e-01
-2.39694342e-01 -9.11506191e-02 -8.93230081e-01 3.61687452e-01
5.66255212e-01 -5.80473423e-01 -1.11145902e+00 5.25619328e-01
5.02563000e-01 -2.59605885e-01 -1.16194761e+00 -3.21151204e-02
2.26833060e-01 -7.13867068e-01 3.56874645e-01 -4.25730169e-01
1.17337286e+00 -1.19232506e-01 2.41156012e-01 -1.86367035e+00
-4.98426080e-01 -1.18498653e-01 -1.31749243e-01 1.56826496e+00
1.06670249e+00 -2.64481515e-01 5.88801026e-01 6.64362311e-01
-2.91680008e-01 -5.99874854e-01 -6.33884907e-01 -7.63335049e-01
1.36812970e-01 -3.59867513e-01 2.34610051e-01 1.12399828e+00
7.46439338e-01 6.10712409e-01 -2.33125046e-01 -3.12411517e-01
4.51731503e-01 -1.79398090e-01 6.16305590e-01 -1.46054137e+00
6.71068057e-02 -6.42754078e-01 -8.48957598e-02 -8.19106758e-01
1.87040776e-01 -1.00717771e+00 1.06374227e-01 -2.10655570e+00
5.83068788e-01 -3.63761783e-01 2.21278593e-01 1.33935839e-01
-8.32659543e-01 -1.11638233e-01 6.04991913e-02 2.55855024e-01
-1.00688899e+00 6.09985650e-01 1.52181005e+00 -4.23621982e-01
5.28292544e-02 -1.18643723e-01 -9.97135222e-01 5.33186436e-01
8.52925777e-01 -5.12707353e-01 -2.35398356e-02 -2.24797264e-01
9.53621864e-01 -6.08819164e-02 -2.28705555e-01 -8.24812591e-01
3.67291987e-01 -4.86774649e-03 2.39647210e-01 -4.52159673e-01
-1.33324698e-01 -2.16140747e-01 -4.02052045e-01 1.19844802e-01
-6.00697100e-01 1.95415482e-01 -2.85919696e-01 3.05427194e-01
-5.33928514e-01 -1.08967268e+00 5.38933039e-01 -4.30605978e-01
-4.93966252e-01 -8.18949789e-02 -6.21724844e-01 6.59303069e-01
7.19219327e-01 6.22473955e-02 -8.22683632e-01 -4.15078759e-01
-2.98008174e-01 3.95737141e-01 7.09123969e-01 2.25047126e-01
7.03154087e-01 -1.13205981e+00 -9.22420144e-01 -4.23972130e-01
4.87920016e-01 -4.03789103e-01 1.19094737e-01 7.63580024e-01
-7.20600784e-01 5.04291236e-01 -1.60671562e-01 -3.66670527e-02
-1.17289829e+00 2.22789899e-01 -2.37762347e-01 -4.98114526e-01
-3.80440176e-01 6.77396238e-01 -7.08081067e-01 -7.38413453e-01
3.52050662e-01 -2.96765000e-01 -9.29098368e-01 5.63290834e-01
7.71052361e-01 6.77288830e-01 1.82492778e-01 -3.09175253e-01
5.21645769e-02 1.43539652e-01 -1.72052756e-01 8.92622843e-02
1.15936923e+00 -3.10077161e-01 -6.20410800e-01 7.33707488e-01
9.46106195e-01 4.79371876e-01 -3.29531282e-01 -9.98863354e-02
4.23063010e-01 -4.78096098e-01 1.14108548e-01 -1.01989150e+00
-6.46740198e-01 6.40121639e-01 -2.15052158e-01 4.92392421e-01
6.49601460e-01 -1.92954004e-01 8.14948380e-01 6.72512949e-01
4.77451414e-01 -1.77713335e+00 2.72538185e-01 1.00535309e+00
9.10803914e-01 -1.62041533e+00 1.19107313e-01 -3.27214450e-01
-9.23189461e-01 1.19187999e+00 8.26887429e-01 3.04225773e-01
5.06171346e-01 -2.54730076e-01 -2.40447093e-02 -3.58110428e-01
-9.14129436e-01 2.21826568e-01 3.63319337e-01 2.40962192e-01
8.31474483e-01 1.46493986e-01 -1.08121824e+00 9.99625385e-01
-3.22837591e-01 -2.36917257e-01 1.38419688e+00 6.78463161e-01
-8.62448156e-01 -1.07240641e+00 -2.89722830e-01 1.13757861e+00
-5.34182668e-01 -2.57913589e-01 -4.23417181e-01 6.15120173e-01
-2.78281569e-01 1.19966137e+00 -4.22193050e-01 -1.56612873e-01
3.35489541e-01 8.17245618e-02 2.81113744e-01 -7.94064999e-01
-7.69328952e-01 -5.43886721e-01 7.72896886e-01 -2.59352535e-01
-3.34703416e-01 -7.02309728e-01 -9.22096014e-01 -5.11996627e-01
-5.91720521e-01 7.21696794e-01 7.35866785e-01 6.91290438e-01
1.17929809e-01 4.57814783e-01 5.58454514e-01 -1.59914326e-02
-7.03216136e-01 -1.33922410e+00 -8.18473876e-01 4.66977447e-01
-1.35618880e-01 -6.45380735e-01 -5.57020664e-01 -8.74756947e-02] | [12.018891334533691, 9.402193069458008] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.