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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
a2555879-c8de-4e98-a179-878e911fed53 | honestbait-headline-generation-via-faithful | null | null | https://openreview.net/forum?id=xfBbBwOkwgQ | https://openreview.net/pdf?id=xfBbBwOkwgQ | HonestBait: Headline Generation via Faithful Forward Reference | Current methods for generating attractive headlines often learn directly from data, which bases attractiveness on the number of user clicks and views. Although clicks or views do reflect user interest, they can fail to reveal how much interest is raised by the writing style and how much is caused by the event or topic itself. Also, such approaches can lead to harmful hallucinations by over-exaggerating the content, aggravating the spread of false information. In this work, we propose HonestBait, a novel framework for solving these issues from another aspect generating headlines using forward references(FR), a writing technique often used in clickbait. A self-verification process is also included in training to avoid harmful hallucinations. We start with a preliminary user study to understand how FR affects user interest, after which we present PANCO, an innovative dataset containing pairs of fake news with verified news020for attractive but faithful news headline generation. Automatic metrics and human evaluations show our framework yields better results in attractiveness while maintaining high veracity. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['headline-generation'] | ['natural-language-processing'] | [-1.03789873e-01 4.92946893e-01 -1.22007251e-01 -4.79107320e-01
-6.89859092e-01 -4.04655784e-01 8.93654048e-01 1.95129752e-01
-1.45066336e-01 1.03115213e+00 6.11166477e-01 1.29804939e-01
3.76140952e-01 -7.57920921e-01 -6.67617738e-01 -9.65382010e-02
1.79386407e-01 1.16021246e-01 2.71566898e-01 -6.10163510e-01
7.39388704e-01 -7.95324519e-02 -1.50564837e+00 7.05284059e-01
8.71457934e-01 9.89033401e-01 -2.46851444e-01 5.30083477e-01
-2.81265173e-02 1.31163657e+00 -1.20349813e+00 -8.66031468e-01
1.17500544e-01 -6.99936986e-01 -4.00887221e-01 -9.50909853e-02
7.69396961e-01 -6.61578715e-01 -2.94360757e-01 9.51827228e-01
5.02042711e-01 -2.40530908e-01 4.12392467e-01 -1.41717124e+00
-1.03991163e+00 8.85951519e-01 -4.57766145e-01 2.13703010e-02
7.78990984e-01 1.31841853e-01 1.19415069e+00 -8.26527894e-01
7.97996998e-01 1.22562647e+00 6.18011951e-01 7.01475024e-01
-1.24176955e+00 -7.50798821e-01 -1.62167519e-01 -3.11451387e-02
-9.83494282e-01 -6.04088604e-01 6.50513768e-01 -4.47844625e-01
2.21874908e-01 8.91431332e-01 6.59475744e-01 1.69119275e+00
-8.71502515e-03 1.02161229e+00 1.46011353e+00 -8.92213583e-02
2.77015209e-01 9.30094957e-01 1.29221290e-01 2.82122999e-01
3.82040381e-01 1.40946761e-01 -9.59713638e-01 -3.32971573e-01
4.51915085e-01 -1.82820469e-01 -5.27099490e-01 1.18377231e-01
-9.68023241e-01 1.09358954e+00 6.11113191e-01 3.77652720e-02
-3.45992148e-01 -8.08846056e-02 2.02614442e-01 5.24961114e-01
5.09016633e-01 1.06582940e+00 2.65825018e-02 -3.05830926e-01
-1.22976756e+00 6.80694997e-01 1.15656865e+00 8.86163771e-01
3.92992705e-01 -6.38698265e-02 -8.22108090e-01 8.26834917e-01
-2.44616042e-03 5.96931040e-01 7.04037905e-01 -7.19981134e-01
3.04896325e-01 3.22856754e-01 5.80084980e-01 -1.60568392e+00
1.19392611e-01 -6.72854543e-01 -3.32108617e-01 2.47966349e-01
4.14526165e-01 -1.40291527e-01 -4.57125008e-01 1.65930462e+00
-2.33594999e-01 -3.92202795e-01 -3.51685196e-01 9.50717747e-01
8.26162457e-01 4.97340083e-01 -2.21898332e-01 -2.88970768e-01
1.22896981e+00 -9.54038262e-01 -1.15254164e+00 -7.78298005e-02
5.88195741e-01 -1.10436249e+00 1.46167290e+00 6.51256680e-01
-1.43250406e+00 -1.20302409e-01 -9.30824578e-01 -8.19403455e-02
-5.04113019e-01 1.67878702e-01 3.96226197e-01 7.33278692e-01
-7.07111001e-01 6.71751320e-01 1.23328120e-01 -9.75341424e-02
4.68664825e-01 -2.25268215e-01 -1.15884982e-01 3.87038976e-01
-1.56503952e+00 1.06151652e+00 -1.59756020e-01 -4.68908727e-01
-4.84177440e-01 -7.69353211e-01 -3.37248683e-01 3.53873894e-02
2.27034673e-01 -3.07435334e-01 1.44240832e+00 -1.07455635e+00
-1.24220538e+00 6.14631414e-01 9.11223814e-02 -7.12191641e-01
1.24127865e+00 -6.48496628e-01 -4.50519860e-01 -3.33362482e-02
3.16747606e-01 6.94901764e-01 1.21374249e+00 -1.34313667e+00
-3.27657759e-01 -1.92296937e-01 3.86713110e-02 2.77884342e-02
-6.02130234e-01 -2.33832225e-01 3.95154068e-03 -1.08969510e+00
-1.81358621e-01 -8.32193077e-01 1.51681945e-01 1.39894769e-01
-9.51224685e-01 8.45519230e-02 9.18630183e-01 -8.97508800e-01
1.73108864e+00 -1.89842117e+00 -5.14607608e-01 1.90674528e-01
5.22756577e-01 1.74813151e-01 1.79402724e-01 6.97249591e-01
1.88268006e-01 4.12910759e-01 3.33216548e-01 -3.45383324e-02
3.74748446e-02 -3.16078067e-01 -7.34263897e-01 2.70599546e-03
-6.48613945e-02 8.54377627e-01 -8.07771087e-01 -3.22198659e-01
-4.26825762e-01 2.29459986e-01 -6.68222129e-01 2.24872857e-01
-1.83794141e-01 -8.46704585e-04 -2.29880273e-01 3.48968357e-01
7.43972778e-01 -3.49057823e-01 -6.03867173e-01 -6.99045137e-02
-1.06892347e-01 4.17364836e-01 -6.67688191e-01 8.66947353e-01
-3.89959216e-01 8.51768315e-01 -3.54920000e-01 -9.62305591e-02
9.43937480e-01 2.17829794e-01 1.29429540e-02 -8.66045237e-01
2.84947045e-02 2.36944258e-01 -3.00726444e-01 -5.59675574e-01
9.63840604e-01 -7.97961727e-02 5.67662753e-02 6.93869948e-01
-4.73417372e-01 -1.39096335e-01 1.50849596e-01 8.00912738e-01
9.99962628e-01 -3.45131457e-01 1.76364988e-01 1.80732161e-01
1.51040763e-01 -5.54273799e-02 -3.06352228e-02 1.07848620e+00
-1.26805559e-01 8.39234650e-01 8.83266449e-01 -1.34906366e-01
-1.22549987e+00 -8.71732891e-01 4.15342711e-02 9.94686782e-01
-2.43457779e-02 -6.23723328e-01 -8.29377353e-01 -7.57583737e-01
-1.88101586e-02 1.45806050e+00 -8.43735516e-01 -3.89405668e-01
2.26332005e-02 -4.36022580e-01 6.41542971e-01 1.05804138e-01
4.15224195e-01 -9.40391243e-01 -3.87218058e-01 2.22353682e-01
-6.35829806e-01 -8.04930210e-01 -7.20207810e-01 -5.03422260e-01
-5.21561623e-01 -7.60053098e-01 -1.19964278e+00 -5.87086529e-02
6.35942817e-01 3.82206351e-01 1.08028424e+00 1.62998110e-01
-1.05579413e-01 -3.39201689e-02 -4.34273601e-01 -6.20199502e-01
-5.73548257e-01 -1.84856683e-01 -1.66476279e-01 8.85441825e-02
3.08969140e-01 -2.15163574e-01 -6.11246705e-01 4.78648216e-01
-9.14663315e-01 1.21401496e-01 5.18710911e-01 9.47223246e-01
-2.54160613e-01 -4.17805374e-01 7.26311982e-01 -1.36074126e+00
1.38163459e+00 -6.12068892e-01 -3.12406242e-01 8.88259113e-02
-8.33978891e-01 2.98944302e-02 4.66795385e-01 -5.32253802e-01
-1.09116340e+00 -3.81004006e-01 6.34315759e-02 -3.16522419e-02
1.79276243e-01 1.98149338e-01 4.21395898e-01 6.41247779e-02
1.34207070e+00 2.04183966e-01 1.95596591e-01 -3.09015542e-01
4.81747836e-01 8.78472865e-01 1.69980988e-01 3.92569602e-02
8.58047009e-01 4.76473659e-01 -6.03143215e-01 -7.80188382e-01
-1.20837688e+00 -4.33143765e-01 7.41884187e-02 -4.15534496e-01
3.44561607e-01 -7.26062298e-01 -7.09172845e-01 2.25264624e-01
-1.22352505e+00 3.43716532e-01 -2.30247304e-01 2.24220514e-01
-5.50883353e-01 2.21798390e-01 -4.90841061e-01 -9.16618288e-01
-2.17881024e-01 -7.39499032e-01 6.98021770e-01 -4.61020187e-04
-7.73305178e-01 -6.86559379e-01 1.06514134e-01 5.85219443e-01
9.40065503e-01 2.20743477e-01 4.33218181e-01 -1.01359570e+00
-4.81330395e-01 -8.48826766e-01 -4.08337772e-01 3.32239449e-01
-1.17913976e-01 -2.51090024e-02 -1.11310959e+00 -1.60553623e-02
1.60808805e-02 -7.04414129e-01 6.92029357e-01 2.86449865e-02
9.57889497e-01 -1.26343560e+00 -3.40622962e-02 -6.62627444e-02
1.12217796e+00 -3.75639379e-01 9.34955895e-01 2.99055815e-01
3.45243037e-01 8.42567980e-01 6.66326821e-01 1.07655859e+00
1.84992999e-01 8.04300547e-01 3.89369488e-01 -1.28996328e-01
-1.45077273e-01 -8.61837387e-01 4.97651160e-01 2.78556615e-01
4.08576354e-02 -3.79013538e-01 -1.35661900e-01 9.42882448e-02
-1.49946141e+00 -1.58086169e+00 -3.98703128e-01 2.46154499e+00
1.10846937e+00 2.81007826e-01 3.40452015e-01 1.76913962e-01
6.85257733e-01 2.06063673e-01 -1.51352495e-01 -3.34062338e-01
-2.61050075e-01 -3.15416068e-01 5.39743602e-01 5.92422783e-01
-4.84973490e-01 7.44914711e-01 6.28038740e+00 6.87695086e-01
-1.29175532e+00 -5.25425933e-02 7.95773745e-01 -3.37158203e-01
-6.97814882e-01 -2.41801873e-01 -7.04749763e-01 8.67761135e-01
5.81663370e-01 -3.97849917e-01 2.67174214e-01 1.12337172e+00
7.62477994e-01 -8.46248195e-02 -9.31026816e-01 8.19022715e-01
4.55636770e-01 -1.31310129e+00 1.25943899e-01 -1.50562450e-02
6.39195740e-01 -3.80740911e-01 4.38344121e-01 2.73109466e-01
2.81446099e-01 -8.00837517e-01 9.56009567e-01 5.23730099e-01
4.88481194e-01 -2.50987172e-01 6.07680917e-01 3.30283463e-01
-1.05708698e-03 1.19383104e-01 -1.68972388e-01 -5.64132072e-02
3.12556922e-01 8.38970900e-01 -1.23155987e+00 -3.61003488e-01
4.03718770e-01 3.50468576e-01 -9.56516087e-01 1.09772897e+00
-2.77731687e-01 5.26428640e-01 -6.82889447e-02 -6.26691997e-01
1.24601528e-01 3.58734757e-01 8.06149900e-01 1.20606291e+00
3.91918808e-01 -1.80807397e-01 -3.81680965e-01 1.24076378e+00
-1.48335934e-01 3.50406498e-01 -7.87393749e-01 -3.45032603e-01
3.61653537e-01 1.29623353e+00 -1.94825411e-01 -5.34446776e-01
-1.38313994e-01 1.17181885e+00 1.49144933e-01 2.27320328e-01
-1.01798785e+00 -5.20966411e-01 -5.14592566e-02 8.57948184e-01
-3.26684229e-02 3.27266306e-01 -4.26472962e-01 -1.31555235e+00
1.04275338e-01 -1.17079246e+00 4.15962888e-03 -1.12025380e+00
-1.37738180e+00 6.00482106e-01 -3.15401703e-01 -1.42671418e+00
-1.46161556e-01 -9.13033932e-02 -6.67927146e-01 6.78688943e-01
-1.30135167e+00 -8.00869167e-01 -5.38158536e-01 2.50859201e-01
5.64202547e-01 -2.29434539e-02 4.54164982e-01 1.86816797e-01
-7.81567171e-02 8.56965125e-01 -1.78292856e-01 -2.26079211e-01
1.23278248e+00 -1.32525790e+00 1.86713159e-01 3.85400921e-01
5.57291582e-02 8.02838147e-01 1.37401009e+00 -7.50215232e-01
-6.87934339e-01 -6.77875221e-01 1.34732544e+00 -5.07693350e-01
8.01459849e-01 -2.46885031e-01 -9.22644377e-01 2.23689675e-01
3.35641593e-01 -3.76547426e-01 7.97591269e-01 5.30748516e-02
-6.16773069e-01 1.85762286e-01 -1.25456607e+00 9.14838314e-01
6.09209955e-01 -1.80611059e-01 -5.38917124e-01 7.06804872e-01
4.96776551e-01 -3.91178802e-02 -3.24279726e-01 -2.31110871e-01
6.34344697e-01 -1.58387995e+00 5.82078874e-01 -6.97358906e-01
1.05059493e+00 1.36814743e-01 2.37409696e-01 -1.55568838e+00
-2.49617368e-01 -9.49013948e-01 -8.69650673e-03 1.21140444e+00
9.09820974e-01 -3.84466022e-01 8.55616689e-01 8.18347514e-01
1.71271339e-01 -4.20952380e-01 -2.07335070e-01 -6.77852511e-01
-4.48979586e-01 -9.18611661e-02 2.24594772e-01 1.06742966e+00
3.83542567e-01 6.01456583e-01 -1.05566669e+00 -3.59664559e-01
5.64779401e-01 -3.95570010e-01 9.48894620e-01 -9.83066559e-01
-3.38073432e-01 -3.29349458e-01 -5.88840917e-02 -1.01544690e+00
-5.75515509e-01 -3.57402772e-01 -1.80300493e-02 -9.27772939e-01
1.78723380e-01 -2.21810862e-01 1.70782357e-01 2.34093294e-01
-5.38503267e-02 3.69467646e-01 2.83128977e-01 4.78381187e-01
-5.47190011e-01 4.92346823e-01 1.54596829e+00 1.44062981e-01
-1.79792389e-01 3.67793232e-01 -9.79627550e-01 5.99320471e-01
6.26376808e-01 -4.19118732e-01 -4.70873117e-01 -8.57023150e-03
6.38004184e-01 8.29593763e-02 5.00066221e-01 -9.30104315e-01
4.97550964e-02 1.55287851e-02 2.61339039e-01 -5.61125576e-01
5.51989436e-01 -5.48626482e-01 -2.09444538e-01 4.98170316e-01
-1.15934610e+00 1.52435660e-01 -4.36733305e-01 7.95001924e-01
-2.69367069e-01 -4.11869764e-01 7.88486838e-01 -4.41005975e-01
2.92899869e-02 -7.41007328e-02 -4.12104934e-01 2.74017572e-01
8.97557139e-01 -1.41144797e-01 -6.18803024e-01 -1.32406747e+00
-7.12958157e-01 -1.44623175e-01 4.56834406e-01 6.15576982e-01
7.15905547e-01 -1.35286474e+00 -8.15741599e-01 1.02982759e-01
4.93798822e-01 -1.09473574e+00 3.54104936e-02 8.36529315e-01
-4.92956758e-01 2.72393823e-01 -1.59835413e-01 -8.14798102e-02
-1.26355207e+00 4.80751395e-01 -1.34559542e-01 7.59012997e-02
-2.42848724e-01 9.43545401e-01 2.40865201e-02 2.12099031e-01
3.40875357e-01 9.39444751e-02 -3.44960004e-01 4.04095769e-01
8.52286756e-01 5.95118761e-01 -4.08915989e-02 -5.70615530e-01
2.20844179e-01 -3.11434478e-01 -5.98090291e-01 -5.03381252e-01
8.51443172e-01 -2.28843376e-01 1.34693205e-01 3.26510787e-01
1.16403484e+00 5.66418469e-01 -8.92542899e-01 1.62123032e-02
-3.34308855e-02 -9.56028163e-01 4.29021567e-02 -1.18033648e+00
-7.68440723e-01 7.62760699e-01 1.63119525e-01 8.35469007e-01
4.99281675e-01 -3.15425634e-01 1.13317204e+00 2.27749541e-01
7.46280923e-02 -1.36915302e+00 6.81803286e-01 6.29730225e-02
1.40393412e+00 -1.48744833e+00 -2.74533462e-02 -4.51673329e-01
-1.28452039e+00 1.05434763e+00 5.86135209e-01 1.44241825e-01
3.52948666e-01 -6.93727955e-02 2.45194957e-01 -2.15100780e-01
-8.44216168e-01 3.08296412e-01 2.78768390e-01 3.87284040e-01
7.68074572e-01 -1.84046060e-01 -8.20744812e-01 5.35274386e-01
-7.31700182e-01 8.29802230e-02 9.97590840e-01 5.92854381e-01
-6.78057313e-01 -6.77971840e-01 -3.03459316e-01 6.18500769e-01
-7.20275462e-01 -2.19935358e-01 -9.03957844e-01 4.90858823e-01
-3.31993729e-01 7.85892308e-01 -2.76839584e-01 -5.30836523e-01
2.98293442e-01 -2.98206687e-01 -4.28973958e-02 -4.87366349e-01
-6.59874499e-01 -1.00618318e-01 5.57527959e-01 -6.12871945e-01
9.84485224e-02 -3.71760011e-01 -6.44118309e-01 -5.17867267e-01
-7.62939155e-01 3.73491079e-01 7.22583652e-01 4.52301919e-01
2.94408202e-01 -1.92849839e-03 9.32571471e-01 -3.50026578e-01
-1.18242621e+00 -1.02790678e+00 -4.64107215e-01 1.05279267e+00
2.91219056e-01 -3.84494126e-01 -8.53034735e-01 -6.39222469e-03] | [12.082602500915527, 9.104093551635742] |
02a22902-03e2-44be-8ff2-35be27fe45c2 | give-me-more-feedback-annotating-argument | null | null | https://aclanthology.org/P18-1058 | https://aclanthology.org/P18-1058.pdf | Give Me More Feedback: Annotating Argument Persuasiveness and Related Attributes in Student Essays | While argument persuasiveness is one of the most important dimensions of argumentative essay quality, it is relatively little studied in automated essay scoring research. Progress on scoring argument persuasiveness is hindered in part by the scarcity of annotated corpora. We present the first corpus of essays that are simultaneously annotated with argument components, argument persuasiveness scores, and attributes of argument components that impact an argument{'}s persuasiveness. This corpus could trigger the development of novel computational models concerning argument persuasiveness that provide useful feedback to students on why their arguments are (un)persuasive in addition to how persuasive they are. | ['Nishant Gurrapadi', 'Vincent Ng', 'Zixuan Ke', 'Winston Carlile'] | 2018-07-01 | null | null | null | acl-2018-7 | ['automated-essay-scoring'] | ['natural-language-processing'] | [ 1.39524117e-01 7.27524817e-01 -5.75681448e-01 -4.67191398e-01
-6.12614870e-01 -8.90016317e-01 7.40109086e-01 1.14290607e+00
-5.12137890e-01 1.04771566e+00 9.96982932e-01 -1.19442141e+00
-7.00505674e-01 -7.09824204e-01 -3.89554381e-01 -3.13407071e-02
7.78503299e-01 5.08227706e-01 1.28279313e-01 -3.61731172e-01
1.00949073e+00 -2.76399385e-02 -1.39261127e+00 3.89377207e-01
1.39052010e+00 1.19320273e-01 -6.78572282e-02 8.16809118e-01
-4.15023059e-01 1.17586887e+00 -9.99339700e-01 -7.05053329e-01
-5.49173832e-01 -9.86109078e-01 -1.31693769e+00 -3.63114297e-01
5.57659388e-01 -1.16867296e-01 5.14919937e-01 1.04745865e+00
2.60537922e-01 -5.46818525e-02 6.96426928e-01 -7.69711137e-01
-8.36862206e-01 1.36332560e+00 -1.44333765e-01 7.58622408e-01
5.83557487e-01 -2.07283512e-01 1.59651041e+00 -4.30263400e-01
7.28469014e-01 1.23591352e+00 3.61408353e-01 5.59359193e-01
-1.05604601e+00 -3.36951971e-01 1.36339277e-01 2.61899680e-01
3.05091381e-01 -3.06366086e-01 9.23190236e-01 -4.72261310e-01
5.11108637e-01 4.21712488e-01 1.21566856e+00 1.07553911e+00
-1.28454655e-01 6.31644726e-01 1.45641649e+00 -8.65495920e-01
2.57683694e-01 -2.48178048e-03 1.15446424e+00 3.84601951e-01
8.06335330e-01 -2.81150758e-01 -4.80604947e-01 -4.20206964e-01
3.56338441e-01 -7.49288440e-01 -1.62961513e-01 6.79881126e-02
-1.06893563e+00 1.28445888e+00 1.67831540e-01 4.89107549e-01
-1.48635238e-01 -9.93954763e-02 5.00108123e-01 5.37950397e-01
3.23692352e-01 9.44284081e-01 -3.33539337e-01 -6.69362783e-01
-4.72457081e-01 7.66077459e-01 1.03387666e+00 -1.65015399e-01
1.10622849e-02 7.45707937e-03 -1.00349188e-01 9.33785200e-01
4.96129662e-01 3.93776476e-01 4.41748738e-01 -9.86728430e-01
4.77447271e-01 9.05246496e-01 2.73957133e-01 -7.49015093e-01
-2.75981963e-01 -2.61660933e-01 2.28137672e-01 3.90055656e-01
6.85628891e-01 -2.65685886e-01 7.77849182e-02 1.61043561e+00
2.79537171e-01 -7.86487818e-01 1.32046029e-01 8.50888729e-01
1.22436106e+00 2.11524278e-01 4.50659662e-01 -1.21637583e-01
1.61799622e+00 -7.43225634e-01 -7.59326577e-01 -2.23895699e-01
9.25237417e-01 -1.20337403e+00 1.43792129e+00 -4.62374091e-02
-1.47850394e+00 -7.91250020e-02 -1.20204830e+00 -2.39418268e-01
9.30095837e-02 -2.72221088e-01 8.84791672e-01 7.25599587e-01
-5.17685056e-01 3.83152276e-01 -1.68659464e-01 1.25216410e-01
5.46053767e-01 -2.58626223e-01 -1.70502625e-03 3.67993027e-01
-1.20323360e+00 1.42891407e+00 1.38042882e-01 -4.90620345e-01
1.88163176e-01 -7.33937740e-01 -6.99242949e-01 1.45135239e-01
3.35653096e-01 -7.75721550e-01 1.60943699e+00 -1.12323546e+00
-1.13886595e+00 8.59774232e-01 -9.05043446e-03 -2.30655760e-01
6.01967096e-01 -3.01338315e-01 -2.06306025e-01 1.94279432e-01
4.61449564e-01 6.89471811e-02 4.48682189e-01 -6.83627248e-01
-7.46497452e-01 -3.45251560e-01 3.68322521e-01 6.37233317e-01
-3.87698680e-01 4.83944058e-01 8.37968588e-01 -5.19989371e-01
-7.88855851e-02 -4.63916749e-01 2.79477030e-01 -1.59392744e-01
-2.22381607e-01 -1.12388372e+00 6.26954138e-01 -2.74973273e-01
1.19718385e+00 -1.62551594e+00 -3.78394216e-01 -9.26182568e-02
3.37090075e-01 2.71929324e-01 1.47445410e-01 5.11760950e-01
-1.01064973e-01 4.54617083e-01 2.63214588e-01 7.55169272e-01
2.98404872e-01 -1.86785813e-02 -3.71528983e-01 1.85667917e-01
-5.67652807e-02 9.21318531e-01 -1.05092072e+00 -5.66108823e-01
4.30846475e-02 7.62896240e-02 -4.15995061e-01 1.73925925e-02
-2.08636373e-01 -1.68942347e-01 -8.18628073e-01 3.21348220e-01
1.30017072e-01 -2.81971574e-01 2.49600057e-02 3.16177666e-01
-4.84049708e-01 1.66696787e+00 -7.92459428e-01 5.08855224e-01
2.34952550e-02 1.08537745e+00 -2.22488046e-01 -8.85183573e-01
8.13302815e-01 3.57786566e-01 -3.74071121e-01 -7.12359011e-01
2.75866210e-01 5.46550632e-01 7.62221336e-01 -4.58999693e-01
6.34979606e-01 -3.97282958e-01 9.77163613e-02 1.36823320e+00
-7.29520082e-01 -4.16053146e-01 7.26510048e-01 4.95828032e-01
8.90895605e-01 -3.06670964e-02 5.92231035e-01 -8.08940530e-01
5.47255635e-01 3.71005356e-01 1.65542498e-01 7.66602159e-01
8.04833509e-03 1.00305513e-01 7.17340589e-01 -4.99162048e-01
-1.28545558e+00 -6.53212011e-01 -4.71941829e-01 1.11093855e+00
-1.03462487e-01 -3.04593503e-01 -8.54860544e-01 -6.70597911e-01
-7.40052015e-02 1.24175763e+00 -7.29088187e-01 2.45900258e-01
-8.38360667e-01 -5.90206504e-01 5.25864720e-01 4.78859544e-01
1.86753124e-01 -1.25835347e+00 -1.11814153e+00 3.57932389e-01
-5.06882191e-01 -4.22064185e-01 3.65025252e-02 2.64102727e-01
-8.27269733e-01 -1.86473417e+00 -1.34233326e-01 -6.13593578e-01
5.88510275e-01 3.49569082e-01 1.29185295e+00 9.10805881e-01
9.33050662e-02 1.33516341e-01 -4.12262499e-01 -1.06049669e+00
-9.09383059e-01 -5.20107523e-02 -4.30566549e-01 -1.14320362e+00
4.32209283e-01 -2.52073616e-01 -3.74568015e-01 6.67886361e-02
-5.84371567e-01 1.46477684e-01 1.77956969e-01 1.00358367e+00
4.24902048e-03 -5.23897886e-01 1.03101254e+00 -1.26635838e+00
1.75870943e+00 -3.54029298e-01 -3.08480203e-01 1.82683349e-01
-1.04046690e+00 -3.58643457e-02 4.20272827e-01 -3.71586263e-01
-1.07832026e+00 -1.31728423e+00 -4.32438225e-01 1.08137619e+00
-1.67496242e-02 6.65812194e-01 3.80064428e-01 1.78724676e-02
1.15849948e+00 -4.24670368e-01 2.74996579e-01 -8.14558119e-02
1.00749880e-01 1.99889839e-01 1.24632329e-01 -1.03271818e+00
6.79732859e-01 -9.58875716e-02 -7.22007900e-02 -6.97001934e-01
-1.48071682e+00 -1.04993992e-01 4.38970365e-02 -2.33505517e-01
3.87063414e-01 -4.38775599e-01 -1.05416477e+00 -3.45140934e-01
-1.00505257e+00 -1.14390686e-01 -5.30576229e-01 7.18295336e-01
-4.55588907e-01 4.82001424e-01 -7.20439553e-01 -6.90528393e-01
-5.41748524e-01 -7.50287652e-01 1.85372159e-01 5.04213333e-01
-1.29852355e+00 -1.31376076e+00 3.71940374e-01 9.70129669e-01
3.50527108e-01 5.51137477e-02 1.66532516e+00 -1.09742141e+00
2.20508903e-01 -1.74092501e-01 -4.59201410e-02 3.84042375e-02
-1.35421664e-01 2.46127337e-01 -4.88501579e-01 5.22016525e-01
2.66321421e-01 -7.73889601e-01 4.29980814e-01 5.16896605e-01
5.64853966e-01 -6.88014269e-01 1.38786778e-01 -3.26453865e-01
8.76586616e-01 -1.70119345e-01 3.76962125e-01 9.70480621e-01
4.29974943e-02 7.64827907e-01 8.01301956e-01 7.73447976e-02
5.59220791e-01 2.92682528e-01 2.12085366e-01 1.71556666e-01
-2.95532554e-01 -2.24451676e-01 1.46982238e-01 6.73290372e-01
-2.99150683e-02 -2.10102156e-01 -8.47269595e-01 7.24406958e-01
-1.69659412e+00 -1.38342535e+00 -1.11090612e+00 1.49518335e+00
8.91210318e-01 4.70738590e-01 1.61421597e-01 8.13425899e-01
5.18298030e-01 3.22522879e-01 -6.75833449e-02 -9.75695491e-01
-4.88862842e-01 2.91949064e-01 -2.89016396e-01 9.36962724e-01
-5.23604989e-01 6.72795534e-01 6.85611725e+00 3.33680570e-01
-4.39954877e-01 -3.26813519e-01 8.02989066e-01 1.88553423e-01
-1.15181816e+00 3.08029413e-01 -6.52495801e-01 3.52767825e-01
5.84491551e-01 -5.58996797e-01 -2.56012112e-01 8.95834208e-01
4.16495681e-01 -4.42928433e-01 -1.00991488e+00 7.23113567e-02
2.06899419e-02 -1.63431156e+00 -1.38021469e-01 -1.06768683e-01
6.92148328e-01 -7.55056858e-01 1.81404036e-02 -4.62414557e-03
6.86143994e-01 -9.53866482e-01 9.97933567e-01 -1.43554479e-01
4.14799899e-02 -6.23408079e-01 9.23032105e-01 2.62364447e-01
-1.31355733e-01 -1.26408506e-02 -3.94667596e-01 -1.07094073e+00
1.22457132e-01 6.95875406e-01 -9.80637431e-01 -1.92079455e-01
2.72413015e-01 2.36245483e-01 -5.17094016e-01 5.61885238e-01
-1.07302773e+00 1.38581657e+00 5.79825565e-02 -8.90640914e-01
3.00547063e-01 -2.38636702e-01 6.75886273e-01 7.59791613e-01
-1.62153915e-02 4.92854804e-01 -2.07525998e-01 6.84353709e-01
-1.26053706e-01 4.12463546e-01 -3.48238081e-01 -2.39256859e-01
8.35227251e-01 1.23033595e+00 -8.29044640e-01 -3.00057620e-01
-8.76804590e-02 9.19677764e-02 4.54360366e-01 -1.69591367e-01
-4.15255964e-01 1.80786639e-01 5.28085470e-01 1.80034488e-01
-3.76118898e-01 2.51449585e-01 -1.09675169e+00 -5.79961419e-01
2.41146535e-02 -8.57659221e-01 5.24520397e-01 -6.85181439e-01
-1.35686803e+00 9.69662815e-02 -2.73253351e-01 -2.89573312e-01
-2.50956684e-01 -4.93538558e-01 -1.20691454e+00 1.00298762e+00
-1.54550362e+00 -7.91207552e-01 -5.64721040e-02 1.37315383e-02
6.13159657e-01 6.55639246e-02 9.00838375e-01 -5.14063656e-01
-4.82055992e-02 2.42872596e-01 -3.71156603e-01 5.21702915e-02
6.30668759e-01 -1.44303918e+00 2.79329717e-01 5.02237082e-01
1.16335377e-02 6.19374752e-01 1.11586690e+00 -6.77222550e-01
-7.33272970e-01 3.35204825e-02 1.34082282e+00 -7.22197413e-01
1.00920701e+00 4.45075363e-01 -8.93977404e-01 2.39686072e-01
4.94462967e-01 -6.63256884e-01 1.17474294e+00 6.74206793e-01
-4.34693456e-01 7.52595425e-01 -9.03216362e-01 9.68216419e-01
8.04864287e-01 -2.95885563e-01 -1.56204557e+00 2.96585917e-01
3.07589591e-01 -2.45460898e-01 -5.65925241e-01 4.05497961e-02
4.79677200e-01 -1.06872261e+00 7.91019678e-01 -9.90367591e-01
1.14879298e+00 1.07450724e-01 3.66100848e-01 -1.30439961e+00
-4.27539766e-01 -2.65829980e-01 1.74622178e-01 9.59153175e-01
6.55074239e-01 -5.38498759e-01 8.04638088e-01 6.99524641e-01
-3.61763984e-01 -7.91506886e-01 -5.84264517e-01 -1.36075139e-01
6.02853000e-01 -1.45113975e-01 6.43459797e-01 1.17623091e+00
6.20086849e-01 9.29671466e-01 3.37921977e-01 -8.15610170e-01
5.35577595e-01 5.32538772e-01 6.94106460e-01 -1.95664692e+00
4.29084413e-02 -1.16129148e+00 2.30610579e-01 -5.68807185e-01
2.68768281e-01 -8.50297689e-01 -2.90490657e-01 -2.01695728e+00
3.80971879e-01 -5.25987148e-01 4.46684241e-01 3.29480767e-01
-7.61942863e-01 -1.34897411e-01 2.24143304e-02 6.31146133e-02
-3.75498503e-01 2.01350778e-01 1.38632154e+00 2.14565694e-01
-3.55474427e-02 -1.67839617e-01 -1.47862422e+00 1.20408928e+00
1.02864027e+00 -3.63149285e-01 -4.94524777e-01 -3.02594423e-01
7.98692584e-01 5.34376875e-02 1.12043582e-01 -2.13947728e-01
5.59069961e-02 -7.65019536e-01 3.76953185e-01 -3.95889044e-01
-1.88547164e-01 -2.29017109e-01 -3.22894990e-01 5.33242226e-01
-1.10452771e+00 6.14871085e-01 -3.61396074e-02 3.19594592e-02
-4.40366045e-02 -8.25052679e-01 5.59322953e-01 4.65824567e-02
-3.79809700e-02 -6.40143633e-01 -6.33604169e-01 6.93334937e-01
7.14164495e-01 -4.78714883e-01 -1.04270542e+00 -3.78932297e-01
-2.31839672e-01 -1.62815571e-01 5.63559771e-01 3.09469521e-01
5.57904184e-01 -1.03704309e+00 -1.30039179e+00 -5.25932074e-01
-8.06374997e-02 -5.46022534e-01 -2.61386260e-02 6.76688910e-01
-5.79541922e-01 4.55283761e-01 -2.95395166e-01 1.20693281e-01
-1.78916895e+00 5.16218618e-02 -1.43499777e-01 -4.27639604e-01
-8.66231918e-01 9.12072539e-01 -2.11074919e-01 2.79640127e-02
-8.78204107e-02 2.54672199e-01 -7.49966979e-01 1.79935277e-01
8.87861371e-01 6.49572492e-01 -2.79781610e-01 -3.82211566e-01
-1.71226710e-02 -5.36950082e-02 -2.17682824e-01 -2.66691983e-01
1.38686514e+00 2.12828398e-01 -3.53731304e-01 5.30548930e-01
3.18109512e-01 7.30849028e-01 -5.79770684e-01 2.96391726e-01
2.48056486e-01 -6.67299151e-01 -3.24257672e-01 -1.29222798e+00
-5.98834753e-02 5.61184347e-01 -2.70360231e-01 6.37783527e-01
2.50500381e-01 7.70887546e-03 4.06067073e-01 3.44048411e-01
-3.18838000e-01 -1.45278764e+00 3.42032373e-01 6.43123746e-01
8.20389807e-01 -1.00856745e+00 3.62660825e-01 -5.31259835e-01
-5.22390842e-01 1.37455189e+00 5.12235284e-01 -1.50568625e-02
1.78862169e-01 1.71084344e-01 2.46444315e-01 -7.36371815e-01
-8.86569977e-01 2.23283529e-01 3.13725084e-01 2.65788198e-01
1.36791086e+00 2.70434409e-01 -1.78103650e+00 7.00145841e-01
-9.37686563e-01 -5.14324963e-01 1.04001451e+00 7.13455200e-01
-8.53508472e-01 -1.30621171e+00 -6.48180187e-01 6.94253027e-01
-7.69951940e-01 -3.22841018e-01 -1.16573238e+00 7.29253292e-01
-4.31189835e-01 1.18099856e+00 -2.51576215e-01 5.02809882e-01
-3.34903151e-02 1.25274107e-01 5.60164213e-01 -5.66626072e-01
-1.19175053e+00 -3.19118261e-01 9.39870417e-01 3.06043237e-01
-5.41712999e-01 -8.18680227e-01 -1.46532226e+00 -8.30654502e-01
-6.12254441e-01 8.95194471e-01 8.00019920e-01 1.26524591e+00
-1.08846456e-01 5.59570432e-01 2.56344397e-02 2.22027496e-01
-7.93908715e-01 -9.83963370e-01 2.72303857e-02 5.01386166e-01
-7.65576437e-02 -7.27724195e-01 -3.20879340e-01 -4.55599487e-01] | [11.239754676818848, 9.263590812683105] |
6c082703-f58d-4a46-bb08-d05f1dd6cb85 | structured-and-natural-responses-co | null | null | https://dl.acm.org/doi/abs/10.1145/3477495.3532063 | https://yecchen.github.io/paper/RERG_mm22.pdf | Structured and Natural Responses Co-generation for Conversational Search | Generating fluent and informative natural responses while maintaining representative internal states for search optimization is critical for conversational search systems. Existing approaches either 1) predict structured dialog acts first and then generate natural response; or 2) map conversation context to natural responses directly in an end-to-end manner. Both kinds of approaches have shortcomings. The former suffers from error accumulation while the semantic associations between structured acts and natural responses are confined in single direction. The latter emphasizes generating natural responses but fails to predict structured acts. Therefore, we propose a neural co-generation model that generates the two concurrently. The key lies in a shared latent space shaped by two informed priors. Specifically, we design structured dialog acts and natural response auto-encoding as two auxiliary tasks in an interconnected network architecture. It allows for the concurrent generation and bidirectional semantic associations. The shared latent space also enables asynchronous reinforcement learning for further joint optimization. Experiments show that our model achieves significant performance improvements. | ['Tat-Seng Chua', 'Wei Ji', 'Fuli Feng', 'Lizi Liao', 'Chenchen Ye'] | 2022-07-07 | null | null | null | acm-sigir-conference-on-research-and | ['conversational-search'] | ['natural-language-processing'] | [ 4.41424489e-01 5.44030011e-01 -3.31404805e-01 -6.09423041e-01
-7.79784620e-01 -4.36606973e-01 1.10447776e+00 -4.21561807e-01
-2.66982347e-01 1.01716602e+00 8.92944336e-01 -6.91188276e-02
1.43459998e-02 -6.99053705e-01 1.07825585e-02 -5.01477778e-01
4.05386150e-01 1.04141819e+00 -9.88413580e-03 -4.54902321e-01
2.99489230e-01 -3.43064852e-02 -1.18072772e+00 5.65332294e-01
1.11052096e+00 6.79678023e-01 5.68739533e-01 9.59875822e-01
-7.50202715e-01 1.48908293e+00 -6.33057296e-01 -2.22247109e-01
5.39351515e-02 -8.63755584e-01 -1.53872120e+00 1.73590309e-03
-3.88087839e-01 -4.47300434e-01 -4.04075384e-01 7.68663645e-01
4.60705996e-01 5.50691843e-01 4.23082739e-01 -1.24948072e+00
-6.63178980e-01 9.60326910e-01 1.52268812e-01 -2.41523623e-01
7.32914329e-01 6.56754851e-01 1.22294343e+00 -4.65473652e-01
4.34915990e-01 1.63735878e+00 1.38646945e-01 1.16915512e+00
-1.30665183e+00 -4.68576640e-01 3.08598638e-01 -1.27448395e-01
-7.71196723e-01 -6.65778100e-01 8.57254684e-01 -1.48962155e-01
1.16727662e+00 3.57113659e-01 5.76297939e-01 1.42801130e+00
-6.92671761e-02 1.12185848e+00 1.16654718e+00 -2.38469437e-01
2.47998253e-01 3.07396889e-01 1.75046682e-01 6.60731673e-01
-7.49984443e-01 2.05223650e-01 -8.71823549e-01 -4.71826732e-01
6.51373386e-01 3.31071839e-02 -2.06942588e-01 1.53138161e-01
-1.18793249e+00 1.08341539e+00 1.51928395e-01 1.47623301e-01
-6.00849986e-01 1.61576584e-01 1.82662457e-01 4.86544549e-01
3.16751570e-01 8.53195965e-01 -2.66796559e-01 -4.93603587e-01
-6.08630359e-01 4.79282022e-01 1.28709471e+00 1.01045072e+00
9.95160401e-01 1.83585986e-01 -6.59506977e-01 1.01297700e+00
3.61207873e-01 2.89801627e-01 7.42351592e-01 -1.10085285e+00
4.21042860e-01 7.45630741e-01 2.91343123e-01 -7.10760951e-01
-2.97351450e-01 1.26334369e-01 -6.89600408e-01 -3.99000317e-01
3.27706784e-01 -4.68307525e-01 -6.19750142e-01 1.99420309e+00
2.29695126e-01 -1.58714369e-01 4.00001764e-01 9.17886257e-01
8.40188444e-01 8.26299906e-01 2.36749604e-01 -1.78664878e-01
1.16939390e+00 -1.61226130e+00 -1.08271229e+00 -6.10575497e-01
2.17419922e-01 -7.42040753e-01 1.39330912e+00 -7.36806542e-02
-1.42301750e+00 -5.15137136e-01 -5.65093100e-01 -1.21333547e-01
-5.13697714e-02 -1.37131542e-01 8.25107396e-01 3.29470009e-01
-1.34868741e+00 2.31153712e-01 -5.76412380e-01 -3.83032672e-02
-2.36365795e-01 5.66472828e-01 1.02779195e-01 5.19545555e-01
-1.43219376e+00 8.87584329e-01 2.21572250e-01 -1.10674866e-01
-7.53139317e-01 -2.31969208e-01 -8.06289852e-01 4.25496548e-01
4.18499321e-01 -9.85020041e-01 1.84270132e+00 -9.23959374e-01
-2.58129382e+00 5.22405148e-01 -4.01094407e-01 -4.08541113e-01
3.19702059e-01 -1.90509349e-01 -8.98922384e-02 1.68424323e-02
6.87711909e-02 9.51075017e-01 6.34437501e-01 -1.16726279e+00
-5.83861411e-01 -1.80663735e-01 1.50373787e-01 6.74511254e-01
-3.65916997e-01 1.33396881e-02 -3.90712887e-01 -4.51326460e-01
-1.98181227e-01 -9.69073176e-01 -5.49045205e-01 -5.62558889e-01
-4.83229339e-01 -7.46696174e-01 5.24731874e-01 -2.95179546e-01
1.10545981e+00 -1.52538848e+00 3.65764260e-01 -9.68824476e-02
2.61815488e-01 -5.70763089e-02 -3.33543241e-01 7.17877984e-01
2.32874453e-01 -5.42718470e-02 1.85728893e-02 -6.06723487e-01
3.86634111e-01 1.43781707e-01 -5.48614025e-01 -1.88936576e-01
1.05626799e-01 1.25887549e+00 -1.08262014e+00 -4.58526731e-01
-7.48941377e-02 -6.64031059e-02 -7.19017744e-01 1.10246181e+00
-6.19866967e-01 4.99938995e-01 -7.47223854e-01 2.46144146e-01
1.07452571e-01 -5.43219388e-01 5.32414556e-01 2.63707727e-01
-5.77495657e-02 9.93311167e-01 -8.07628274e-01 1.80518448e+00
-7.00332701e-01 1.91449702e-01 4.30399001e-01 -7.12642789e-01
1.20402515e+00 5.69128096e-01 3.70067477e-01 -7.23661005e-01
-8.51572752e-02 -2.63184365e-02 -1.63766041e-01 -4.96452898e-01
8.64372969e-01 -2.65558779e-01 -4.58326072e-01 1.22483242e+00
8.84950534e-02 -5.16322315e-01 -4.94133346e-02 3.53306532e-01
8.97838891e-01 6.09399304e-02 1.15802079e-01 -1.29991084e-01
5.85006654e-01 -5.34439981e-02 5.24676859e-01 9.26694989e-01
-1.20196797e-01 1.90127045e-01 5.70243955e-01 -3.89489353e-01
-6.72958910e-01 -9.59563732e-01 6.94561601e-01 1.45807624e+00
2.71843672e-01 -4.52825278e-01 -6.48884475e-01 -7.41175175e-01
-3.69669855e-01 9.58194911e-01 -2.70350218e-01 -3.23685616e-01
-6.75952315e-01 -3.44213694e-01 5.96702039e-01 4.41285342e-01
5.88801146e-01 -1.65870297e+00 -5.36005497e-01 3.00266981e-01
-6.66450799e-01 -9.81579363e-01 -9.07847166e-01 1.80228263e-01
-6.05775356e-01 -7.03944981e-01 -3.49514931e-01 -8.44901860e-01
2.67367035e-01 1.69326499e-01 1.33961439e+00 1.26083493e-01
2.15764254e-01 3.44893754e-01 -2.18971997e-01 9.01454836e-02
-8.30443680e-01 2.66537100e-01 9.13436487e-02 -9.18236896e-02
3.68756741e-01 -5.51431954e-01 -5.23308635e-01 4.42944825e-01
-6.98342919e-01 4.87458408e-01 3.85293186e-01 1.28709567e+00
5.14771566e-02 -6.37515962e-01 8.80504906e-01 -9.40975547e-01
1.51591182e+00 -4.49259639e-01 -3.42790455e-01 4.25605625e-01
-8.21880698e-01 4.52619612e-01 7.79658258e-01 -3.40373188e-01
-1.62737393e+00 7.36684427e-02 -1.62295952e-01 9.28504616e-02
-3.78844500e-01 1.90303281e-01 6.48615882e-02 5.02861261e-01
5.38539827e-01 4.41624522e-01 2.83380568e-01 -2.42519215e-01
5.31989992e-01 9.18156505e-01 3.47038180e-01 -9.56111312e-01
3.23121071e-01 1.63683686e-02 -6.85875356e-01 -6.73189998e-01
-5.98380864e-01 -4.38145518e-01 -4.22966890e-02 -2.03734756e-01
9.35993075e-01 -6.33559406e-01 -1.05401206e+00 3.35234195e-01
-1.36731517e+00 -8.84159863e-01 -2.36697093e-01 3.31472635e-01
-9.90332365e-01 4.26189214e-01 -9.81678128e-01 -1.16935408e+00
-7.52173722e-01 -1.42612696e+00 1.07143104e+00 5.50915956e-01
-7.86985874e-01 -9.34578598e-01 3.15413326e-01 6.51113272e-01
7.13672042e-01 -3.78879189e-01 8.71665180e-01 -9.07597005e-01
-6.58420801e-01 1.26128644e-01 -4.92979139e-02 -1.45349786e-01
3.01717162e-01 -4.81172770e-01 -8.59961152e-01 2.15478316e-02
2.81570286e-01 -9.56597745e-01 4.94866580e-01 5.24518080e-02
9.28699434e-01 -8.22950363e-01 -1.36061087e-01 2.80462027e-01
6.92951083e-01 4.65747267e-01 3.32683533e-01 -1.22843340e-01
1.73084706e-01 1.04320550e+00 3.48965019e-01 4.59066927e-01
5.72483122e-01 8.22779953e-01 9.17074904e-02 -3.64068747e-02
1.58520132e-01 -6.31261766e-01 4.70738858e-01 9.19303656e-01
3.14051151e-01 -4.73675817e-01 -7.56642580e-01 4.26403970e-01
-2.18157339e+00 -1.06846905e+00 4.25586671e-01 1.63703835e+00
1.41929579e+00 -1.68603897e-01 1.54739588e-01 -6.42371237e-01
4.54112440e-01 4.99406993e-01 -5.31092346e-01 -5.41057765e-01
9.70552936e-02 3.37461621e-01 -1.90286502e-01 9.90554154e-01
-5.07129550e-01 1.29515755e+00 6.56876802e+00 4.95947957e-01
-8.41985703e-01 5.25136888e-02 6.37412488e-01 4.03094152e-03
-6.68007672e-01 2.00916529e-01 -7.26573408e-01 3.24414283e-01
8.72627079e-01 -1.72693834e-01 7.59783447e-01 8.71739089e-01
2.65225738e-01 8.01103339e-02 -1.07188976e+00 7.56129146e-01
-1.57203078e-01 -1.59862852e+00 6.02582609e-03 -2.78402030e-01
6.78149760e-01 -3.73701960e-01 -1.90094963e-01 6.13760769e-01
1.23244882e+00 -1.17415655e+00 2.99004734e-01 6.78250849e-01
5.73164165e-01 -4.70029294e-01 4.06462342e-01 8.53188097e-01
-9.90323186e-01 -5.33215888e-02 3.81969921e-02 -2.15778947e-01
4.89122421e-01 -2.21899450e-01 -1.22487903e+00 1.55136809e-01
1.89768776e-01 2.45820612e-01 -2.95151081e-02 1.76628292e-01
-4.73973632e-01 5.23657560e-01 -1.11579984e-01 -7.03640580e-01
5.69785774e-01 -3.22041124e-01 5.28614998e-01 1.14122534e+00
-2.19013855e-01 2.17088684e-01 6.72128022e-01 1.37359715e+00
1.46629978e-02 -1.35434568e-01 -4.27338690e-01 -3.53241324e-01
6.24222457e-01 1.07687509e+00 -3.22126657e-01 -4.04420137e-01
-8.93607885e-02 1.14065123e+00 4.69243735e-01 5.80130160e-01
-5.82517266e-01 -1.18377969e-01 6.50262594e-01 -2.70334959e-01
-4.54092532e-01 -5.99703528e-02 -4.03971165e-01 -1.23975623e+00
-2.35930771e-01 -1.32637548e+00 2.09216341e-01 -6.26416683e-01
-1.34022820e+00 1.01085031e+00 -5.95245063e-02 -5.64544499e-01
-1.20965362e+00 -3.87414135e-02 -8.67178500e-01 1.32813108e+00
-1.20510435e+00 -9.46552157e-01 -2.79564649e-01 7.17018366e-01
1.33224797e+00 -4.50345278e-01 1.30005443e+00 -2.72891849e-01
-4.60193068e-01 5.86096346e-01 -4.71125424e-01 -3.44797187e-02
7.41868734e-01 -1.39488041e+00 4.32176262e-01 4.56354350e-01
-1.58312172e-01 9.77740467e-01 5.36301374e-01 -6.54813230e-01
-1.11071002e+00 -4.61699992e-01 1.12357461e+00 -1.98217213e-01
4.28925812e-01 -4.29910630e-01 -6.63594306e-01 4.39276457e-01
9.46248293e-01 -9.44245934e-01 8.26467276e-01 2.93977290e-01
-1.44092999e-02 2.86304593e-01 -6.44039094e-01 1.02422047e+00
7.46229768e-01 -6.96631849e-01 -7.16333866e-01 6.04457557e-01
9.93502975e-01 -3.33186626e-01 -4.66578513e-01 3.91205288e-02
3.20984423e-01 -1.05221188e+00 8.28957677e-01 -9.93607700e-01
6.05656087e-01 1.32996172e-01 9.14362893e-02 -1.35192919e+00
-2.04874307e-01 -1.49887598e+00 -1.95927233e-01 9.20211256e-01
5.28031051e-01 -5.45012295e-01 1.07676351e+00 1.08494592e+00
-7.78824985e-02 -9.75103021e-01 -4.19099867e-01 -2.12524652e-01
-1.67185485e-01 8.24202970e-02 8.29338968e-01 7.49308884e-01
4.76384014e-01 1.08984900e+00 -6.92460060e-01 -4.69716430e-01
1.02710947e-01 5.75562179e-01 1.10570168e+00 -7.99473584e-01
-6.52506292e-01 -7.41714001e-01 6.93997979e-01 -1.91360700e+00
5.14225602e-01 -5.64426601e-01 4.59439725e-01 -1.40649056e+00
1.64547533e-01 -5.39967418e-01 2.72717774e-01 4.59448874e-01
-5.10625243e-01 -5.77784956e-01 1.69886142e-01 3.30641806e-01
-6.33891404e-01 1.06116199e+00 1.22744167e+00 -6.60091713e-02
-6.28441274e-01 3.68656933e-01 -7.26572454e-01 4.62867737e-01
9.74042535e-01 -1.70245379e-01 -8.04539502e-01 -2.91686088e-01
-3.73203307e-02 8.48026931e-01 2.11984739e-02 -3.45343262e-01
5.91050804e-01 -7.25974798e-01 -1.74323931e-01 -4.33268607e-01
6.90489531e-01 -2.68572807e-01 -2.39346012e-01 3.60149175e-01
-1.08147168e+00 2.61221826e-01 -2.66367614e-01 7.00257778e-01
-3.98446411e-01 -2.15781420e-01 5.83246052e-01 -5.01419783e-01
-4.12827462e-01 8.94907266e-02 -6.85729325e-01 8.73966217e-02
5.79510808e-01 -1.60420060e-01 -1.92370847e-01 -1.19843781e+00
-6.08864248e-01 7.22919643e-01 1.10937007e-01 6.27859175e-01
6.41783774e-01 -1.08040500e+00 -5.68370402e-01 1.07842945e-01
-2.84112662e-01 1.16417877e-01 1.48025915e-01 4.07666266e-01
-2.10216299e-01 6.04107797e-01 -5.85666858e-02 -3.55293423e-01
-1.07704341e+00 1.43051594e-01 3.58499944e-01 -9.72374856e-01
-2.06396207e-01 1.06826341e+00 1.92808032e-01 -8.56332541e-01
4.08917874e-01 2.52397269e-01 -3.25588197e-01 -2.25569561e-01
3.30694348e-01 -1.54292081e-02 -5.24437904e-01 -1.53799161e-01
1.12951569e-01 -2.64959604e-01 -2.33859777e-01 -6.76595449e-01
1.27439272e+00 -1.94753572e-01 -3.05225402e-01 2.10558876e-01
8.52698803e-01 -1.60042048e-01 -1.28192031e+00 -5.03078282e-01
1.13533318e-01 -4.19300467e-01 -4.17457700e-01 -8.28968704e-01
-6.46599770e-01 6.65869117e-01 -9.93121490e-02 3.54442626e-01
9.30113196e-01 -5.52927777e-02 9.95784700e-01 7.86608756e-01
6.37832358e-02 -1.38009703e+00 7.74476707e-01 9.45697665e-01
8.69178355e-01 -1.10811722e+00 -4.12169844e-01 -2.26230055e-01
-1.19848287e+00 1.11468482e+00 1.17365766e+00 1.25655472e-01
6.12281151e-02 2.99985975e-01 2.25401685e-01 -2.71135598e-01
-1.47065556e+00 -5.19395433e-02 1.18355174e-02 4.11424965e-01
6.58563495e-01 -1.02637202e-01 -3.16976011e-01 6.79739535e-01
-4.19272006e-01 -2.90246964e-01 2.08263099e-01 7.49832809e-01
-2.92706490e-01 -1.45868003e+00 -4.78871502e-02 2.26670578e-01
-2.03683097e-02 -1.86998695e-01 -9.79127884e-01 2.36514121e-01
-5.98313689e-01 1.26576257e+00 -2.20311768e-02 -4.91605163e-01
8.97131488e-02 5.27420759e-01 -3.86775918e-02 -7.36186147e-01
-1.13074648e+00 1.08989954e-01 3.74279797e-01 -6.85307860e-01
-2.31505990e-01 -2.93001682e-01 -1.21710241e+00 -1.43616959e-01
-3.45935047e-01 7.17295945e-01 2.45405257e-01 9.22144651e-01
5.36809802e-01 4.93019819e-01 8.70057225e-01 -5.42658091e-01
-1.17343354e+00 -1.23169672e+00 9.70644224e-03 4.38322216e-01
9.42637622e-02 -3.14826876e-01 -1.69684961e-01 -3.56716998e-02] | [12.784639358520508, 8.148518562316895] |
fa5ef8ad-bea8-4af3-a80d-0772d05d8bd6 | face-hallucination-using-linear-models-of | 1512.06009 | null | http://arxiv.org/abs/1512.06009v1 | http://arxiv.org/pdf/1512.06009v1.pdf | Face Hallucination using Linear Models of Coupled Sparse Support | Most face super-resolution methods assume that low-resolution and
high-resolution manifolds have similar local geometrical structure, hence learn
local models on the lowresolution manifolds (e.g. sparse or locally linear
embedding models), which are then applied on the high-resolution manifold.
However, the low-resolution manifold is distorted by the oneto-many
relationship between low- and high- resolution patches. This paper presents a
method which learns linear models based on the local geometrical structure on
the high-resolution manifold rather than on the low-resolution manifold. For
this, in a first step, the low-resolution patch is used to derive a globally
optimal estimate of the high-resolution patch. The approximated solution is
shown to be close in Euclidean space to the ground-truth but is generally
smooth and lacks the texture details needed by state-ofthe-art face
recognizers. This first estimate allows us to find the support of the
high-resolution manifold using sparse coding (SC), which are then used as
support for learning a local projection (or upscaling) model between the
low-resolution and the highresolution manifolds using Multivariate Ridge
Regression (MRR). Experimental results show that the proposed method
outperforms six face super-resolution methods in terms of both recognition and
quality. These results also reveal that the recognition and quality are
significantly affected by the method used for stitching all super-resolved
patches together, where quilting was found to better preserve the texture
details which helps to achieve higher recognition rates. | ['Christine Guillemot', 'Reuben Farrugia'] | 2015-12-18 | null | null | null | null | ['face-hallucination'] | ['computer-vision'] | [ 1.41122535e-01 1.44640788e-01 -6.97648749e-02 -2.17788950e-01
-9.86354828e-01 -1.97296217e-02 5.37198782e-01 -5.96920133e-01
1.50297970e-01 6.02035820e-01 3.92598718e-01 6.23394132e-01
-3.43295485e-01 -9.35952723e-01 -6.57564461e-01 -9.94583011e-01
-5.30213937e-02 3.43790889e-01 5.15153594e-02 -1.72809198e-01
1.44294947e-01 8.21426332e-01 -2.00182271e+00 6.42731190e-01
5.54862976e-01 6.52105212e-01 1.71135724e-01 4.74187344e-01
-7.76304528e-02 2.73286611e-01 -1.72832415e-01 -1.74040139e-01
2.49536961e-01 -4.71157312e-01 -5.74741840e-01 3.16074133e-01
1.04998052e+00 -1.96966931e-01 -1.74986258e-01 1.10562551e+00
1.34600371e-01 -4.59547304e-02 7.31774449e-01 -4.26748782e-01
-7.47861207e-01 -8.45211819e-02 -9.31325436e-01 6.97722062e-02
6.23652995e-01 -7.51760125e-01 6.68753982e-01 -1.55740690e+00
8.06297481e-01 1.65331137e+00 7.83025444e-01 4.38565016e-01
-1.63187516e+00 -4.40891325e-01 -2.35679328e-01 -1.63557287e-02
-1.84521222e+00 -7.47087061e-01 1.02445698e+00 -2.79056281e-01
5.14388561e-01 1.89401835e-01 1.46506354e-01 7.30270863e-01
3.21648031e-01 -2.16304734e-01 1.41902387e+00 -5.47703922e-01
3.62427123e-02 3.99935067e-01 -2.34701857e-01 1.02129161e+00
2.13404775e-01 3.22559744e-01 -7.81715512e-01 -3.40061426e-01
1.47762585e+00 -1.56430819e-04 -4.15058851e-01 -4.16020423e-01
-7.96300411e-01 8.41152668e-01 4.44548160e-01 8.76008868e-01
-6.22340381e-01 -5.27102292e-01 -4.36864704e-01 2.09037185e-01
6.38221264e-01 2.16349974e-01 7.92777985e-02 3.84394854e-01
-1.06540310e+00 -1.48806423e-01 6.26195788e-01 4.39365327e-01
1.35820949e+00 1.78038374e-01 4.71035719e-01 1.04881954e+00
3.41380686e-01 4.28938001e-01 2.95496196e-01 -1.01195431e+00
2.71670580e-01 3.05750817e-01 2.50473781e-03 -1.56351244e+00
-1.04360804e-01 -2.14756206e-01 -1.13198435e+00 3.41677964e-01
2.22161099e-01 4.33480471e-01 -6.18762374e-01 1.43133080e+00
5.02884388e-01 6.42188311e-01 1.60844177e-01 1.08996177e+00
5.89819431e-01 7.59525478e-01 -4.52320963e-01 -5.55311024e-01
1.28118479e+00 -4.22329098e-01 -8.07250857e-01 -1.01391422e-02
1.32408693e-01 -8.99849355e-01 6.44944429e-01 1.76946238e-01
-1.16590989e+00 -8.80789220e-01 -1.17503011e+00 -3.74832526e-02
-2.85490118e-02 1.93898827e-01 5.93601167e-02 5.41292846e-01
-1.16181231e+00 8.48871231e-01 -7.43004501e-01 -2.47057468e-01
3.27819496e-01 1.99600652e-01 -9.62170422e-01 -4.55808342e-01
-8.01097393e-01 9.79197681e-01 -2.41785228e-01 2.74996124e-02
-3.72719765e-01 -7.79083729e-01 -8.97605419e-01 -9.99349654e-02
-8.88886824e-02 -2.60970831e-01 1.64700434e-01 -8.19794059e-01
-1.47362220e+00 8.67089629e-01 -6.18974388e-01 2.02405199e-01
3.39253210e-02 -1.56305376e-02 -5.63081563e-01 6.00028515e-01
7.20520616e-02 2.70500958e-01 1.43846560e+00 -1.63284767e+00
-3.35684717e-01 -7.58749306e-01 -4.06229258e-01 1.41501799e-01
-1.76536918e-01 -4.93145622e-02 -1.43521413e-01 -5.87615192e-01
7.57784307e-01 -4.94041353e-01 8.22238699e-02 -2.95526418e-03
7.26131126e-02 2.01934397e-01 1.06545877e+00 -1.05386007e+00
9.33619261e-01 -2.29738545e+00 5.65597773e-01 3.27754050e-01
-4.34216745e-02 -5.27908616e-02 -3.01808715e-01 1.49493158e-01
-4.58737731e-01 -1.01537183e-02 -1.70267433e-01 -4.40214455e-01
-6.96856380e-01 1.66244209e-01 -1.89969420e-01 9.11553204e-01
4.56922024e-01 4.37696904e-01 -5.41510403e-01 -4.64771360e-01
2.67884940e-01 1.37111688e+00 -4.24878687e-01 3.48595262e-01
4.35919851e-01 7.23719954e-01 -3.23051691e-01 5.01808405e-01
9.72570896e-01 8.06979239e-02 1.23983007e-02 -6.92841709e-01
-2.08112225e-01 -1.30266786e-01 -1.54883540e+00 1.54329836e+00
-5.85274041e-01 3.23944628e-01 5.47442973e-01 -8.49083722e-01
1.36348581e+00 4.47962970e-01 5.81178606e-01 -6.03397489e-01
-2.37580895e-01 1.59071147e-01 -6.96012616e-01 -1.92187414e-01
2.87768722e-01 -5.21062851e-01 5.62747478e-01 1.58861786e-01
-1.24106789e-02 1.08402401e-01 -5.01129448e-01 -2.51593858e-01
3.98002267e-01 2.06352562e-01 2.53997892e-01 -4.01581526e-01
8.76157939e-01 -4.79678750e-01 4.59293664e-01 6.09738193e-02
4.08439875e-01 1.00309217e+00 1.06799856e-01 -4.42855120e-01
-1.18440783e+00 -1.06393468e+00 -5.66121042e-01 5.84334731e-01
7.09171444e-02 -3.91036421e-01 -9.65016603e-01 -1.19804397e-01
-2.06066906e-01 3.64822149e-01 -6.32758915e-01 -1.65403914e-02
-7.77571499e-01 -4.00501758e-01 -7.48549262e-03 1.38294816e-01
5.27190030e-01 -5.18777370e-01 -2.74127305e-01 -2.18474418e-02
-1.32558465e-01 -1.25848711e+00 -4.48721319e-01 -4.55009699e-01
-1.09973049e+00 -1.06830478e+00 -6.02953792e-01 -8.29246998e-01
9.32231307e-01 4.96593416e-01 7.37202168e-01 1.37010694e-01
-2.83786058e-01 5.35395861e-01 -1.22628063e-01 5.38582981e-01
-3.80585164e-01 -4.72444147e-01 3.17934096e-01 7.20171094e-01
1.67771056e-01 -8.59252334e-01 -1.09136961e-01 5.33417225e-01
-6.53484523e-01 -6.35442138e-02 6.51145935e-01 9.12433743e-01
1.05424988e+00 4.19608325e-01 3.02268893e-01 -5.63773453e-01
1.39354795e-01 -2.28864074e-01 -5.69060087e-01 1.74322754e-01
-4.05291766e-01 3.07638526e-01 6.20771706e-01 -5.42384565e-01
-1.22701383e+00 2.16588974e-02 8.54586959e-02 -8.69774878e-01
-2.13579372e-01 9.23843980e-02 -3.26681525e-01 -5.62414348e-01
7.06330657e-01 2.61724561e-01 2.61274129e-01 -9.25818384e-01
2.76580662e-01 4.85461682e-01 4.71853912e-01 -4.22036469e-01
1.23608720e+00 7.69898236e-01 3.51644605e-01 -1.56609905e+00
-7.83757806e-01 -2.93882132e-01 -1.14239812e+00 2.57566292e-02
8.68179500e-01 -9.37928855e-01 -3.07406515e-01 -7.46602118e-02
-8.77803802e-01 2.94777781e-01 -2.35484421e-01 5.98469138e-01
-5.70862830e-01 4.28846568e-01 -6.40976131e-01 -8.47764492e-01
2.49159057e-02 -9.15048301e-01 1.34793031e+00 1.89367607e-01
9.10644159e-02 -1.00248623e+00 1.18624888e-01 5.16757786e-01
4.86563593e-01 3.24932396e-01 8.31609130e-01 2.21929744e-01
-6.91623807e-01 -7.39382803e-02 -1.67908713e-01 3.69939476e-01
4.57088321e-01 -1.86549306e-01 -1.02767813e+00 -6.33961141e-01
4.67216283e-01 6.51920736e-02 5.59031963e-01 3.92239571e-01
5.82172573e-01 -4.97860402e-01 -2.53327399e-01 8.90962720e-01
1.65377367e+00 -4.26402390e-01 8.33801866e-01 -4.22363319e-02
9.28635418e-01 1.00919127e+00 5.75227380e-01 1.66887373e-01
1.13078028e-01 1.09872806e+00 6.51488006e-02 -1.75993368e-02
-4.51173186e-01 -4.03488219e-01 6.83691204e-01 7.24619269e-01
-4.00665641e-01 9.92113233e-01 -3.69582772e-01 2.35766158e-01
-1.49070168e+00 -1.27036881e+00 7.35632852e-02 2.58723426e+00
5.95042586e-01 -4.36204612e-01 -6.04718477e-02 1.25791073e-01
9.27780926e-01 2.02612281e-01 -1.29472300e-01 -3.09620798e-01
-2.94115305e-01 4.02553320e-01 6.58224747e-02 1.02760530e+00
-8.12154949e-01 8.47880721e-01 6.30127525e+00 7.48703480e-01
-1.06553984e+00 8.79667401e-02 4.17293638e-01 1.85690045e-01
-2.38202468e-01 -1.74360305e-01 -9.77325022e-01 7.90598020e-02
1.12051666e+00 6.09606318e-02 7.30710983e-01 5.32414258e-01
7.91464075e-02 1.77892804e-01 -1.10233951e+00 1.26600575e+00
4.17830706e-01 -1.37472129e+00 3.36995780e-01 3.58129710e-01
7.99735487e-01 -5.34231067e-01 2.16029048e-01 -1.29396543e-01
-3.74716371e-01 -1.35337138e+00 2.44850755e-01 8.15111637e-01
1.16796279e+00 -9.48678851e-01 5.18695951e-01 2.83717960e-01
-1.50384688e+00 1.42258316e-01 -9.25197721e-01 2.93469429e-01
-1.36431992e-01 4.91071463e-01 -3.88941348e-01 5.73186755e-01
7.69628346e-01 7.87686646e-01 -3.44824791e-01 2.33026102e-01
2.37494465e-02 1.36986062e-01 -2.25383401e-01 7.59745777e-01
-2.43751481e-01 -7.96424270e-01 7.08440781e-01 9.24204886e-01
4.54848588e-01 4.46281821e-01 1.34454221e-01 1.14111364e+00
1.87946349e-01 1.27329528e-01 -7.00064182e-01 2.25615248e-01
2.80406684e-01 1.43168235e+00 -3.54210138e-01 4.24570963e-02
-6.68043017e-01 9.25495565e-01 3.03046972e-01 5.29185474e-01
-3.20268124e-01 2.13062480e-01 7.56929219e-01 6.30428016e-01
5.38028181e-01 -3.23681653e-01 -1.34859532e-01 -1.31831515e+00
1.14739694e-01 -7.95467675e-01 2.32031062e-01 -6.35310411e-01
-1.19347739e+00 7.74231434e-01 -1.18442610e-01 -1.16190982e+00
-3.75292540e-01 -4.33859438e-01 -1.97520539e-01 1.34653437e+00
-1.58320642e+00 -1.12716496e+00 -1.80534273e-01 8.40337336e-01
3.05695951e-01 -2.31682196e-01 1.11947834e+00 5.38949221e-02
-1.59675643e-01 4.75155473e-01 1.26730287e-02 8.79231654e-03
4.68392164e-01 -8.05941403e-01 -3.28081638e-01 6.23474479e-01
2.98051149e-01 6.47841811e-01 5.92356920e-01 -5.21890819e-01
-1.71959281e+00 -9.65716183e-01 7.02119589e-01 -4.73050267e-01
2.31606483e-01 -1.64135978e-01 -1.49965346e+00 4.52951193e-01
-3.72570127e-01 2.29883149e-01 5.54637492e-01 7.33953416e-02
-7.37438321e-01 -2.38324016e-01 -1.66054928e+00 2.03439593e-01
7.28096426e-01 -9.96228456e-01 -6.89306974e-01 -9.66247544e-02
9.44460556e-02 -1.78748593e-01 -1.34051514e+00 2.37423345e-01
5.76139629e-01 -1.16088450e+00 1.25196958e+00 -3.09360951e-01
3.06572258e-01 -3.84427518e-01 -5.87697625e-01 -1.01438999e+00
-7.17358351e-01 -3.06119055e-01 -3.17695767e-01 1.17324769e+00
9.29752141e-02 -5.98734677e-01 9.06216562e-01 2.41928250e-01
2.84675330e-01 -6.93632543e-01 -1.29258990e+00 -6.80669963e-01
1.05377752e-02 2.73012251e-01 3.17677706e-01 1.24037039e+00
-2.33014405e-01 4.30797577e-01 -3.62487227e-01 8.88060749e-01
1.21462476e+00 4.12495643e-01 4.72872794e-01 -1.36489201e+00
-1.19768120e-01 -3.28705683e-02 -6.71078920e-01 -6.77938700e-01
3.09536576e-01 -7.65444696e-01 -2.51034230e-01 -1.09435213e+00
1.79646388e-01 -1.57608077e-01 -3.63270678e-02 3.07587199e-02
1.29831433e-01 5.34675479e-01 -8.30584466e-02 4.93997425e-01
1.02345429e-01 6.08982205e-01 1.27444518e+00 1.94583893e-01
-3.68323028e-01 -2.14292929e-01 -5.85707366e-01 6.90857351e-01
3.30319613e-01 -2.07536891e-01 -3.72057520e-02 -1.08660877e-01
-2.46054649e-01 3.69327694e-01 3.18826884e-01 -1.00043225e+00
1.20822154e-01 -4.32675965e-02 7.38645077e-01 -3.06048304e-01
8.52500141e-01 -9.93896306e-01 4.71483946e-01 3.86157967e-02
-1.37062818e-01 -2.73855150e-01 -4.80971411e-02 5.89069486e-01
-4.43782300e-01 6.21155612e-02 1.37959671e+00 8.51292387e-02
-3.10062677e-01 4.15998578e-01 2.17191696e-01 -4.29418951e-01
6.82474017e-01 -6.49156213e-01 5.95722273e-02 -3.19552064e-01
-9.09822583e-01 -5.78628182e-01 7.71418393e-01 4.04553175e-01
1.08576787e+00 -1.57620704e+00 -1.06429350e+00 7.76342213e-01
-1.09869704e-01 -1.84563965e-01 2.31113657e-01 6.53306484e-01
-2.35380977e-01 1.87540889e-01 -4.48819876e-01 -7.21515000e-01
-1.44876826e+00 7.18330085e-01 5.12883604e-01 7.34737590e-02
-9.67044652e-01 5.12641549e-01 2.25938320e-01 -1.42460451e-01
-1.37002766e-01 1.67405635e-01 -4.08148050e-01 -5.08380085e-02
1.12739253e+00 5.20690918e-01 -1.10845186e-01 -1.62354290e+00
-3.41160208e-01 1.59857547e+00 -4.42948230e-02 -2.44245175e-02
1.57565415e+00 -1.92363203e-01 -5.22735417e-01 4.09701258e-01
1.67090297e+00 2.81660497e-01 -1.20068491e+00 -5.33975422e-01
-7.12872148e-02 -1.06343174e+00 4.00368929e-01 2.19448879e-02
-1.20523524e+00 8.07764947e-01 6.91059113e-01 -5.73512577e-02
1.17335165e+00 1.36172250e-01 2.25707158e-01 6.49385825e-02
6.82050407e-01 -7.28099287e-01 1.26367241e-01 1.20711625e-01
1.29015303e+00 -9.83572483e-01 2.71802992e-01 -8.96829426e-01
-1.44966051e-01 1.32540655e+00 2.75070578e-01 -4.55567002e-01
7.75209367e-01 1.97831541e-01 -2.06697509e-01 -2.49542609e-01
-4.68819708e-01 1.56620845e-01 5.30605137e-01 8.80277514e-01
4.47039664e-01 -2.30948757e-02 1.29222199e-01 2.10593447e-01
-2.01337606e-01 -3.83398086e-01 3.72935861e-01 2.60560572e-01
-6.18961215e-01 -9.23446238e-01 -1.02090347e+00 9.06727761e-02
-2.54344553e-01 2.81836510e-01 -1.90613702e-01 5.04743218e-01
-4.08267155e-02 9.88375425e-01 1.89535022e-01 -3.09104383e-01
3.90431106e-01 -1.01830631e-01 7.78606772e-01 -5.53468883e-01
6.18877038e-02 3.42429608e-01 -2.33208761e-01 -1.06252575e+00
-4.71346706e-01 -6.49738789e-01 -9.74420667e-01 -3.62827986e-01
-1.82407826e-01 2.70426393e-01 5.04894137e-01 6.64830029e-01
4.43668306e-01 -7.13810548e-02 9.91966426e-01 -1.22277868e+00
-3.55328798e-01 -7.04156578e-01 -1.06246269e+00 4.31318671e-01
6.79080963e-01 -7.39557147e-01 -6.89100146e-01 9.48475078e-02] | [12.822114944458008, -0.01242329552769661] |
1cb35f3a-48f8-4d56-812c-2caa60124f74 | a-novel-dual-dense-connection-network-for | 2203.02723 | null | https://arxiv.org/abs/2203.02723v1 | https://arxiv.org/pdf/2203.02723v1.pdf | A Novel Dual Dense Connection Network for Video Super-resolution | Video super-resolution (VSR) refers to the reconstruction of high-resolution (HR) video from the corresponding low-resolution (LR) video. Recently, VSR has received increasing attention. In this paper, we propose a novel dual dense connection network that can generate high-quality super-resolution (SR) results. The input frames are creatively divided into reference frame, pre-temporal group and post-temporal group, representing information in different time periods. This grouping method provides accurate information of different time periods without causing time information disorder. Meanwhile, we produce a new loss function, which is beneficial to enhance the convergence ability of the model. Experiments show that our model is superior to other advanced models in Vid4 datasets and SPMCS-11 datasets. | ['Yonggui Zhu', 'Guofang Li'] | 2022-03-05 | null | null | null | null | ['video-super-resolution'] | ['computer-vision'] | [ 2.54795134e-01 -4.36734110e-01 -1.05165571e-01 -2.34115973e-01
-5.55037677e-01 1.47505373e-01 2.20563605e-01 -6.17593706e-01
-1.99263930e-01 1.01342881e+00 4.77833211e-01 3.34408820e-01
-1.86056271e-01 -8.55546594e-01 -5.31972885e-01 -5.43894708e-01
4.48418297e-02 -2.71280140e-01 6.22201681e-01 -2.21781477e-01
2.22411305e-02 4.64425772e-01 -1.48142934e+00 6.72755003e-01
8.80078495e-01 8.48579049e-01 7.11938798e-01 2.07820967e-01
-1.25979176e-02 9.71627295e-01 -4.57200974e-01 1.26605406e-01
1.73939660e-01 -5.82560182e-01 -6.78499043e-01 -2.85299663e-02
1.86557174e-01 -5.53356946e-01 -6.74472511e-01 1.03796327e+00
4.49944973e-01 3.19137216e-01 -3.62754688e-02 -8.03222418e-01
-9.80781674e-01 6.16921544e-01 -9.55382884e-01 9.59266901e-01
5.03509045e-01 -2.80759744e-02 4.98981893e-01 -7.96257734e-01
9.39049006e-01 1.49969053e+00 4.05948192e-01 7.31473744e-01
-1.15837431e+00 -9.32975531e-01 2.89809257e-01 6.30319715e-01
-1.43308496e+00 -4.05505329e-01 7.74443805e-01 -1.07857203e-02
5.21845877e-01 9.88082066e-02 5.18529475e-01 1.20699859e+00
1.64680585e-01 5.42791843e-01 1.19085848e+00 2.22977921e-01
-1.93786100e-01 -3.28359336e-01 -3.39339972e-01 1.99452966e-01
4.66424264e-02 1.68383032e-01 -8.02318871e-01 2.74643123e-01
1.51937747e+00 1.54450864e-01 -7.52443790e-01 2.55350262e-01
-1.39943278e+00 3.02792996e-01 3.86481732e-01 5.21470904e-01
-4.46493924e-01 -9.33881328e-02 2.44554803e-01 2.57517576e-01
6.97786033e-01 -6.01723641e-02 -1.81216761e-01 -3.72180976e-02
-9.71090794e-01 1.13051802e-01 -4.74457331e-02 9.07305121e-01
3.07046354e-01 4.32256132e-01 -3.54399562e-01 1.00671542e+00
6.64269328e-02 -1.36163621e-03 5.94504774e-01 -1.31523085e+00
5.45750380e-01 1.48767725e-01 3.41147751e-01 -1.03472149e+00
-2.00556919e-01 -5.34076750e-01 -1.40746129e+00 7.98782334e-02
8.84431750e-02 1.05501480e-01 -6.47201598e-01 1.75169754e+00
1.93143710e-01 7.71653295e-01 1.30177423e-01 1.34330916e+00
1.10739410e+00 1.06021273e+00 -1.12470023e-01 -8.88050079e-01
1.07673979e+00 -7.54859269e-01 -1.15018427e+00 2.21709445e-01
-5.14458716e-01 -6.11569107e-01 8.74787748e-01 5.39440930e-01
-1.52362263e+00 -9.23183739e-01 -9.50548410e-01 -2.67124712e-01
2.67359823e-01 -4.95633967e-02 1.83520988e-01 -1.88429132e-01
-1.13716006e+00 8.06614816e-01 -6.70197845e-01 8.66290107e-02
6.04761899e-01 8.21401402e-02 -3.15110952e-01 -2.49151289e-01
-1.40484977e+00 4.86088425e-01 3.12539101e-01 6.04218654e-02
-7.90311575e-01 -8.31974566e-01 -5.38128138e-01 -1.53536916e-01
3.78580511e-01 -5.62334299e-01 7.67026067e-01 -8.49327445e-01
-1.47076714e+00 6.14636004e-01 -3.56282979e-01 -4.11878318e-01
6.16529822e-01 -1.72248725e-02 -8.80267918e-01 5.88920891e-01
-1.27555709e-03 4.50027853e-01 1.06463575e+00 -1.23437023e+00
-8.86864662e-01 -2.54200250e-01 -8.03394690e-02 3.66413385e-01
-4.04988751e-02 3.85711044e-01 -5.71271539e-01 -9.05158877e-01
3.05830508e-01 -2.39839420e-01 -2.04484954e-01 -1.33064300e-01
-5.21297641e-02 -7.02001825e-02 9.45539892e-01 -1.11013758e+00
1.36049879e+00 -2.29810071e+00 4.83928353e-01 -2.72549003e-01
5.17940104e-01 2.15886071e-01 -1.43806010e-01 -2.45860860e-01
-3.43960136e-01 1.73692867e-01 -1.47682037e-02 -6.42666966e-02
-7.56363809e-01 1.59256935e-01 -4.43374485e-01 3.14551413e-01
8.96666646e-02 6.08138621e-01 -9.01128769e-01 -7.14700937e-01
1.95563525e-01 1.03551042e+00 -2.05374300e-01 2.79670477e-01
1.00472115e-01 9.46751058e-01 -4.48922396e-01 3.88490051e-01
8.76867056e-01 -3.61747861e-01 -1.23316094e-01 -6.32753253e-01
-3.56909782e-01 8.65498856e-02 -1.15678811e+00 1.76116192e+00
-3.95191610e-01 4.67074841e-01 8.48471150e-02 -5.09647012e-01
1.02350211e+00 3.10472459e-01 6.75109148e-01 -1.19759560e+00
-5.03925607e-02 -1.35240018e-01 -3.79106224e-01 -4.48114991e-01
7.26586998e-01 1.51157537e-02 6.19737506e-01 4.94455881e-02
-3.43700767e-01 5.68107188e-01 1.46517724e-01 1.97772875e-01
7.63436794e-01 2.72128493e-01 -6.52043894e-02 1.51764214e-01
6.31971061e-01 -5.61518312e-01 1.15941620e+00 1.19416907e-01
-1.73171952e-01 1.12884641e+00 3.02778035e-01 -4.10525739e-01
-1.36332715e+00 -1.27302563e+00 -2.20392168e-01 5.06868064e-01
6.79781675e-01 -3.04315656e-01 -4.77098227e-01 -1.74544677e-01
-4.94618088e-01 3.85836244e-01 -4.82711941e-01 -7.48059079e-02
-9.31438744e-01 -7.18796909e-01 8.61608833e-02 5.15148938e-01
9.74611044e-01 -1.11341918e+00 -3.97641361e-01 2.73580194e-01
-7.00903296e-01 -1.28166437e+00 -7.15981662e-01 -5.80852568e-01
-1.08998358e+00 -7.75959730e-01 -9.77980018e-01 -7.49679148e-01
4.49109435e-01 7.18175948e-01 1.11860573e+00 -4.67353947e-02
-1.46227837e-01 -2.06369534e-01 -5.21994233e-01 5.04000902e-01
-2.38379255e-01 -3.60904694e-01 4.33374196e-02 2.29614928e-01
-1.05155230e-01 -8.99690092e-01 -7.14774668e-01 4.83596951e-01
-9.37785268e-01 5.65279782e-01 4.53550220e-01 7.47400999e-01
1.26298189e+00 5.57142317e-01 7.35082746e-01 -5.14914155e-01
5.17182112e-01 -5.16345680e-01 -5.47786713e-01 1.26961507e-02
-5.80398381e-01 -2.55052507e-01 9.94980037e-01 -6.59018993e-01
-1.42264450e+00 -3.62546027e-01 2.21128613e-02 -9.73817706e-01
4.67859991e-02 1.59476265e-01 -3.15659463e-01 2.84927823e-02
2.00251475e-01 6.23885810e-01 -1.49448499e-01 -8.24705839e-01
1.92390501e-01 5.01461804e-01 8.97924125e-01 -3.54042441e-01
7.81129181e-01 6.76953673e-01 -4.61137034e-02 -6.05740547e-01
-8.02845120e-01 -9.25404727e-02 -3.67716134e-01 -2.58886069e-01
9.68098879e-01 -1.22145522e+00 -5.26369572e-01 3.62353086e-01
-1.04824042e+00 -7.58023560e-02 -2.31884569e-01 5.12595475e-01
-5.19393325e-01 4.46535408e-01 -9.68221247e-01 -5.86206079e-01
-3.86227161e-01 -1.02606094e+00 7.31778264e-01 6.38149559e-01
3.71484578e-01 -5.13246775e-01 -2.85962999e-01 3.38605046e-01
4.58935946e-01 3.71977210e-01 2.39208788e-01 2.55588323e-01
-8.78431976e-01 4.28132027e-01 -7.29688406e-01 4.31580096e-01
1.69091776e-01 -7.62916654e-02 -5.09158254e-01 -3.61114502e-01
1.02026597e-01 3.75050642e-02 9.51470435e-01 4.44922775e-01
1.75410903e+00 -3.30837309e-01 -2.97665931e-02 1.10724938e+00
1.59199607e+00 3.57720047e-01 1.17363894e+00 4.16465819e-01
8.60648692e-01 2.89491206e-01 9.02555227e-01 3.59323233e-01
4.51314121e-01 8.76973212e-01 2.95692295e-01 -1.20981157e-01
-4.54760700e-01 -2.19239026e-01 4.57411140e-01 7.44473636e-01
-7.24960744e-01 5.16020954e-02 -2.49049127e-01 4.73131090e-01
-1.86528599e+00 -1.60340738e+00 -2.96660781e-01 2.00253010e+00
9.82132673e-01 1.72237903e-02 1.14787504e-01 -8.41330886e-02
1.03711629e+00 4.59590256e-01 -6.91972911e-01 1.21821374e-01
-5.73632538e-01 8.15447196e-02 3.22243094e-01 2.98542082e-01
-7.09856272e-01 7.61030555e-01 5.91206121e+00 1.17064440e+00
-1.04085219e+00 3.67533088e-01 8.02518606e-01 -3.95490587e-01
-3.88943285e-01 -3.54835719e-01 -7.63480842e-01 8.99363041e-01
7.88014293e-01 -5.15941799e-01 8.23039412e-01 4.02851433e-01
7.37191498e-01 5.91428690e-02 -7.08128631e-01 1.40207708e+00
-4.64240424e-02 -1.48856580e+00 2.29253173e-01 -1.41583428e-01
7.20125437e-01 -2.41375506e-01 2.72577584e-01 -1.34013025e-02
4.68273088e-02 -1.12803435e+00 5.76829493e-01 8.23146582e-01
1.30670655e+00 -1.00947285e+00 4.27663505e-01 8.77367705e-02
-1.59879172e+00 -1.00963049e-01 -6.13368750e-01 2.96253979e-01
4.56514508e-01 5.15431881e-01 1.11359507e-01 9.55770195e-01
1.12206757e+00 1.28974450e+00 -3.89633536e-01 8.32468629e-01
-2.42839698e-04 2.87281841e-01 4.82901596e-02 7.97073364e-01
-1.97534278e-01 -4.88055617e-01 7.24388659e-01 7.81279504e-01
3.90270591e-01 5.52766979e-01 -8.01607445e-02 1.01792836e+00
-3.51067446e-02 -1.70306668e-01 -2.17027828e-01 2.67475277e-01
6.06263697e-01 1.23083854e+00 -5.59327185e-01 -3.18770170e-01
-4.24934834e-01 1.01872349e+00 1.81794390e-01 4.60115284e-01
-9.46889937e-01 -1.62864372e-01 5.67801178e-01 2.85374105e-01
2.05086440e-01 -1.36516124e-01 -8.14020485e-02 -1.39931476e+00
2.01350242e-01 -7.94406593e-01 4.48455036e-01 -1.26340079e+00
-1.11721432e+00 7.68186688e-01 -7.43803158e-02 -1.74264371e+00
-4.99821082e-03 2.33833984e-01 -3.79342347e-01 1.02248538e+00
-1.87159562e+00 -7.74885058e-01 -6.88636661e-01 9.14659262e-01
8.71677935e-01 -1.10660970e-01 1.41053572e-01 7.49063432e-01
-6.60649061e-01 4.33229417e-01 4.88179959e-02 -5.98427542e-02
6.45450711e-01 -7.86138892e-01 1.43545255e-01 1.10296881e+00
-5.04218757e-01 3.84834677e-01 6.69390678e-01 -7.54533410e-01
-9.44304049e-01 -1.39443028e+00 4.26498055e-01 6.11782856e-02
2.64115453e-01 2.30330676e-01 -1.34674001e+00 4.61729318e-01
1.99108887e-02 2.24444434e-01 1.52634755e-01 -4.68916923e-01
-2.78672546e-01 -4.64243680e-01 -1.40406811e+00 5.35245657e-01
1.32082081e+00 -3.52723897e-01 -4.44735467e-01 -4.04040627e-02
1.37692440e+00 -6.84582233e-01 -1.11274123e+00 4.82970655e-01
3.70422006e-01 -1.15064895e+00 1.37508404e+00 -2.41904125e-01
9.51760590e-01 -6.52513266e-01 -3.78669463e-02 -1.03986204e+00
-6.76048636e-01 -5.34439266e-01 -4.36630160e-01 1.22075868e+00
-2.96065688e-01 -3.91282111e-01 4.86277223e-01 2.12034628e-01
-1.82589591e-01 -8.08423877e-01 -8.90024126e-01 -8.52405071e-01
-2.85506815e-01 3.15389074e-02 7.42728353e-01 1.10371828e+00
-5.72680891e-01 1.76417783e-01 -7.96235144e-01 2.79930741e-01
9.88433719e-01 9.39650163e-02 1.63612932e-01 -1.09337163e+00
-6.31010458e-02 -2.90023357e-01 -8.29903930e-02 -9.31517065e-01
-3.95313762e-02 -5.97513437e-01 -3.15678984e-01 -1.64515316e+00
3.39443624e-01 -2.48549283e-01 -5.84657550e-01 9.80284140e-02
-1.88368440e-01 4.66875732e-01 4.48289253e-02 5.31528771e-01
-7.18825281e-01 6.08871341e-01 1.91589046e+00 2.33027965e-01
-3.44756097e-01 -1.94697246e-01 -5.13804555e-01 5.73064148e-01
7.71607518e-01 -4.96297330e-02 -4.12949681e-01 -4.79453057e-01
1.02594785e-01 6.68307424e-01 3.90155464e-01 -9.67957735e-01
-6.30663931e-02 -3.17214012e-01 7.67189860e-01 -9.04754817e-01
4.06914383e-01 -6.58773482e-01 6.17654324e-01 1.56295478e-01
-4.47877467e-01 1.28539830e-01 -9.44871753e-02 7.05947578e-01
-3.83553177e-01 3.29609275e-01 1.26607275e+00 -1.80593461e-01
-9.01118338e-01 8.48501503e-01 7.54741654e-02 5.40635251e-02
8.57118189e-01 -2.56086916e-01 -4.67262894e-01 -1.85797885e-01
-7.99857616e-01 3.14514756e-01 5.89459836e-01 6.11438394e-01
1.23589361e+00 -1.59198761e+00 -9.15834486e-01 -6.70301681e-03
-3.66367221e-01 3.66253227e-01 8.34250748e-01 6.93279445e-01
-4.53026444e-01 -4.48290743e-02 -6.07441902e-01 -3.27497751e-01
-1.34001994e+00 7.15513289e-01 2.57976562e-01 -2.40074903e-01
-1.31988370e+00 5.71537852e-01 2.39467502e-01 4.20491368e-01
5.24657592e-02 3.08920890e-02 -8.86156261e-01 -1.30104631e-01
1.17074072e+00 6.36919200e-01 -5.60349107e-01 -9.16763186e-01
-1.04372039e-01 6.63874686e-01 -2.36945733e-01 -6.18837867e-03
1.55424297e+00 -6.68473005e-01 -1.93169624e-01 4.38244730e-01
1.02267551e+00 -1.71540141e-01 -1.53982067e+00 -4.00783509e-01
-5.10732174e-01 -9.20643270e-01 2.42893592e-01 -4.69714493e-01
-1.59267759e+00 5.11213422e-01 7.57735789e-01 -1.40083989e-03
1.60420477e+00 -1.45473689e-01 1.26186442e+00 -4.53326613e-01
6.57254577e-01 -1.07429576e+00 1.51092216e-01 2.23522872e-01
9.83007908e-01 -1.00267065e+00 1.19661421e-01 -5.22458553e-01
-5.16383290e-01 1.00707066e+00 8.23222041e-01 -2.61336148e-01
2.02056289e-01 1.46775559e-01 -3.55028123e-01 9.36003700e-02
-9.68325198e-01 -9.01289657e-02 4.28833626e-02 7.29576409e-01
2.88704425e-01 -2.61398196e-01 -5.90006351e-01 7.75162697e-01
2.40569208e-02 3.75333130e-01 8.47131729e-01 3.20204556e-01
-3.07313085e-01 -7.67556846e-01 -3.51064175e-01 4.53256518e-01
-6.74757540e-01 -8.06485638e-02 3.53860289e-01 4.25769329e-01
1.88753679e-01 9.07272041e-01 2.22493544e-01 -5.75456738e-01
2.44239509e-01 -5.34357905e-01 5.42159498e-01 -2.56885260e-01
-1.12805299e-01 2.95550197e-01 -1.28411427e-01 -1.07770109e+00
-6.64668918e-01 -5.90374470e-01 -1.40968752e+00 -4.24546957e-01
1.40139356e-01 -2.71076206e-02 7.60041177e-02 5.93117774e-01
4.03981924e-01 1.06449878e+00 9.50982988e-01 -8.54826033e-01
1.56750219e-04 -6.17636085e-01 -8.64227295e-01 4.90913630e-01
4.70416754e-01 -4.87627089e-01 -3.10651928e-01 3.90915871e-01] | [11.0361328125, -1.8403072357177734] |
e78c07da-cebb-450d-9125-5648669b2831 | directional-skip-gram-explicitly | null | null | https://aclanthology.org/N18-2028 | https://aclanthology.org/N18-2028.pdf | Directional Skip-Gram: Explicitly Distinguishing Left and Right Context for Word Embeddings | In this paper, we present directional skip-gram (DSG), a simple but effective enhancement of the skip-gram model by explicitly distinguishing left and right context in word prediction. In doing so, a direction vector is introduced for each word, whose embedding is thus learned by not only word co-occurrence patterns in its context, but also the directions of its contextual words. Theoretical and empirical studies on complexity illustrate that our model can be trained as efficient as the original skip-gram model, when compared to other extensions of the skip-gram model. Experimental results show that our model outperforms others on different datasets in semantic (word similarity measurement) and syntactic (part-of-speech tagging) evaluations, respectively. | ['Yan Song', 'Jing Li', 'Shuming Shi', 'Haisong Zhang'] | 2018-06-01 | null | null | null | naacl-2018-6 | ['learning-word-embeddings'] | ['methodology'] | [ 8.46363083e-02 1.16128917e-03 -4.92690295e-01 -5.42847753e-01
-4.62420404e-01 -6.92765296e-01 6.91933751e-01 3.85650158e-01
-6.23550355e-01 4.71777856e-01 8.66937339e-01 -6.68384552e-01
2.03005895e-01 -6.46766484e-01 -1.26083091e-01 -6.18015110e-01
-1.25555873e-01 4.04987335e-01 3.60506743e-01 -4.30145055e-01
5.72459400e-01 1.26415849e-01 -1.31059337e+00 2.85300046e-01
6.46265328e-01 6.37063384e-01 6.74806833e-01 4.19887573e-01
-6.59841478e-01 4.25440282e-01 -2.25098863e-01 -5.19910932e-01
-1.33241177e-01 -2.87990063e-01 -8.78091872e-01 -1.60815299e-01
2.33060673e-01 2.23197713e-02 -1.55522019e-01 1.05915952e+00
5.95799446e-01 3.54709625e-01 5.02067745e-01 -4.48076606e-01
-7.65364468e-01 8.36301625e-01 -1.29066229e-01 2.57691205e-01
4.65384215e-01 -1.91781744e-01 1.97613931e+00 -1.42250681e+00
4.78445023e-01 1.40737689e+00 6.51799798e-01 5.77064037e-01
-1.02732110e+00 -3.02083284e-01 5.33925891e-01 3.99994850e-01
-1.10637999e+00 6.75434098e-02 5.95225513e-01 -3.77565116e-01
1.66523778e+00 2.00723588e-01 5.54683268e-01 9.93578136e-01
3.31787527e-01 8.48135173e-01 9.17142034e-01 -9.10182476e-01
2.66596407e-01 -1.13130987e-01 7.61290371e-01 4.75932896e-01
1.12599127e-01 1.19010374e-01 -3.79670948e-01 -4.48094547e-01
2.33081609e-01 -1.62676796e-01 8.47267807e-02 -2.45804146e-01
-1.11629486e+00 1.14566457e+00 -5.90482587e-03 6.35739267e-01
-3.93418461e-01 4.25051264e-02 4.30814415e-01 8.64019990e-02
6.38149798e-01 4.27820235e-01 -7.11709499e-01 -2.54553646e-01
-4.97440726e-01 2.14614555e-01 7.33108759e-01 8.55800152e-01
7.41944909e-01 5.88619197e-03 -4.18853968e-01 9.70474124e-01
3.85347217e-01 4.43470508e-01 1.07294273e+00 -4.60797757e-01
2.78285861e-01 4.47380006e-01 3.25629045e-03 -8.85809660e-01
-4.63704497e-01 -3.00724238e-01 -4.08982426e-01 -4.66536164e-01
-5.95213212e-02 1.38901034e-02 -9.53375578e-01 1.64966750e+00
2.10404381e-01 2.28137046e-01 -1.18777387e-01 5.86346805e-01
6.47120953e-01 8.04936349e-01 6.18182659e-01 -2.43153587e-01
1.67402637e+00 -1.16758573e+00 -9.91255760e-01 -6.63127661e-01
1.15738380e+00 -1.01916373e+00 1.23639774e+00 5.60704842e-02
-7.80667543e-01 -5.52283764e-01 -7.18310297e-01 -4.02638800e-02
-8.38415325e-01 1.31158875e-02 8.43001127e-01 6.67066693e-01
-1.02036405e+00 5.01371980e-01 -5.27992606e-01 -4.43718523e-01
-2.23175734e-01 9.48226377e-02 -3.94577533e-01 -1.98236138e-01
-1.68011785e+00 1.09324157e+00 5.35792768e-01 -3.41154277e-01
-2.55588174e-01 -4.66817558e-01 -1.12657487e+00 1.65113986e-01
2.31023908e-01 -4.99626935e-01 1.21385860e+00 -5.87817132e-01
-1.16847277e+00 7.55941927e-01 -6.51834190e-01 -3.32420319e-01
-3.19417953e-01 -3.39300752e-01 -6.26867175e-01 -2.06021711e-01
6.05443232e-02 4.49727446e-01 4.75827873e-01 -8.84197593e-01
-7.46877611e-01 -2.79691577e-01 1.92590624e-01 1.91792578e-01
-6.20899856e-01 3.08307320e-01 -4.05648053e-01 -1.04461479e+00
-8.06005765e-03 -9.83933032e-01 -7.63448954e-01 -4.92915541e-01
-1.76737055e-01 -7.64168918e-01 4.58566070e-01 -6.24663532e-01
2.05998349e+00 -2.23066831e+00 2.00613681e-02 1.13553189e-01
-1.26574397e-01 6.26013279e-01 -5.23912251e-01 9.30612504e-01
-2.56095320e-01 1.65109098e-01 -1.74936697e-01 -3.43691796e-01
1.16192788e-01 7.13358462e-01 -4.50380951e-01 6.99743330e-02
7.21360147e-02 9.31724072e-01 -1.19433486e+00 -3.60959172e-01
3.36378276e-01 3.99730593e-01 -6.60848975e-01 1.65183172e-01
-1.38654649e-01 -1.29127145e-01 -4.63823944e-01 2.66459763e-01
3.50142211e-01 -1.91107750e-01 7.57955074e-01 3.30838673e-02
4.70881760e-02 1.03440249e+00 -9.45296109e-01 1.46976507e+00
-1.00916326e+00 3.10412347e-01 -5.75745404e-01 -1.02454567e+00
8.02723527e-01 3.53764802e-01 1.48460656e-01 -7.23697543e-01
-1.01385852e-02 3.54331642e-01 1.11314707e-01 -5.48620224e-01
7.64030635e-01 -1.58419162e-01 -1.57435030e-01 2.71264821e-01
1.70304388e-01 2.29278386e-01 4.17279780e-01 1.87445059e-01
8.96172225e-01 -1.02925360e-01 8.59172523e-01 -4.67490047e-01
6.60911322e-01 -3.29477876e-01 4.25983131e-01 4.65579122e-01
-3.71734612e-02 3.12373310e-01 3.21510643e-01 -5.46958387e-01
-8.65479171e-01 -7.20431924e-01 -1.09585874e-01 1.65988100e+00
8.01799446e-02 -1.17262316e+00 -5.78867376e-01 -1.02097309e+00
-7.23832250e-02 1.06640708e+00 -7.19091296e-01 5.59781045e-02
-8.86905313e-01 -6.44195378e-01 2.97237068e-01 9.09898400e-01
-3.07418942e-01 -1.10672736e+00 -7.16265440e-02 4.83732879e-01
-8.99516344e-02 -1.18553865e+00 -7.31148899e-01 5.13992429e-01
-9.72940803e-01 -9.17363048e-01 -3.50205094e-01 -1.05269802e+00
5.71231127e-01 4.84711647e-01 1.25918341e+00 -1.11037539e-02
1.93672342e-04 5.64687736e-02 -9.70755339e-01 -3.51150259e-02
-1.23491198e-01 1.70668140e-01 1.97049193e-02 -1.77474365e-01
9.93001819e-01 -5.22286117e-01 -3.54904920e-01 3.23287845e-01
-9.42319870e-01 -4.10072744e-01 6.50456071e-01 1.20308793e+00
6.39147699e-01 -4.08323884e-01 3.41174304e-01 -1.00283408e+00
7.00422645e-01 -4.59132344e-01 -2.96968073e-01 4.07016784e-01
-1.03152633e+00 2.13079974e-01 5.31191349e-01 -3.32899868e-01
-6.96186543e-01 -1.45805866e-01 -6.08332574e-01 2.18752339e-01
5.05346544e-02 7.07130671e-01 -2.23447442e-01 2.44470894e-01
1.24148227e-01 3.73186260e-01 -4.86860335e-01 -8.51607740e-01
8.00725818e-01 5.99156320e-01 -2.52867043e-02 -4.18330073e-01
6.12823784e-01 -1.52448388e-02 -1.11126103e-01 -8.97464514e-01
-9.55053270e-01 -9.68277693e-01 -8.22331905e-01 2.45383903e-01
9.06022489e-01 -6.55107200e-01 -9.52231437e-02 -7.90482834e-02
-1.32849884e+00 9.04167667e-02 -1.16777457e-01 6.55452311e-01
-4.77972031e-02 7.38888264e-01 -4.88730848e-01 -8.17674220e-01
-1.90451384e-01 -8.97646368e-01 1.19791865e+00 -2.11247250e-01
-5.94946444e-01 -1.61437547e+00 2.55401254e-01 4.12546173e-02
4.41279978e-01 -5.35296321e-01 1.24585044e+00 -1.17700469e+00
5.57961762e-02 -3.69035721e-01 6.94596991e-02 6.10009968e-01
1.34224728e-01 -3.55589032e-01 -6.46966457e-01 -1.08080320e-01
-3.36164504e-01 5.62328286e-02 1.03388858e+00 2.71736234e-01
1.06630123e+00 -4.15931135e-01 -4.04871166e-01 3.38743985e-01
1.41474509e+00 1.14338338e-01 4.93609101e-01 2.84689486e-01
7.62423933e-01 6.45738780e-01 9.59738255e-01 2.60980248e-01
4.09564942e-01 8.49114776e-01 3.28166246e-01 2.94661909e-01
-1.20993041e-01 -4.79404628e-01 5.07725358e-01 1.45010889e+00
3.13500240e-02 -3.51433009e-01 -7.89720535e-01 8.03917706e-01
-1.67216623e+00 -7.90257096e-01 -3.91734064e-01 1.98641396e+00
6.17490113e-01 -2.66616493e-02 -6.39968067e-02 4.38053608e-02
5.70372164e-01 6.99292362e-01 1.52254431e-02 -7.77562559e-01
-2.81383365e-01 6.13557756e-01 4.93171841e-01 1.01245534e+00
-1.02341104e+00 1.31735229e+00 7.27747488e+00 1.11159694e+00
-7.31846750e-01 2.75332749e-01 3.03078920e-01 2.89025426e-01
-7.56538451e-01 1.82234585e-01 -1.14820325e+00 4.84956741e-01
9.90706861e-01 -6.86012357e-02 5.67260617e-03 1.14572442e+00
9.07891989e-03 1.66693956e-01 -8.49209785e-01 6.81828380e-01
3.45245570e-01 -1.10845900e+00 3.15645099e-01 -3.51611190e-02
5.52957654e-01 -3.96568030e-02 -1.74738675e-01 3.84989232e-01
2.79767990e-01 -8.66602480e-01 2.45908484e-01 -3.15762982e-02
3.77412885e-01 -6.77451432e-01 1.12602854e+00 2.44484812e-01
-1.40989125e+00 -7.02608600e-02 -4.33761567e-01 -1.96274757e-01
3.43195796e-01 7.91462123e-01 -7.30627120e-01 5.63140988e-01
2.37302527e-01 7.08719790e-01 -4.35372412e-01 5.07097721e-01
-6.36659205e-01 1.05527008e+00 1.73698038e-01 -5.31648755e-01
6.40360236e-01 -1.80492297e-01 5.37884414e-01 1.87858045e+00
3.26090634e-01 2.76358813e-01 3.20099413e-01 -1.71723857e-03
2.80757338e-01 6.84704721e-01 -3.97546798e-01 -2.76068263e-02
5.57076097e-01 1.08836591e+00 -4.27434295e-01 -5.58475912e-01
-6.71571672e-01 9.41273749e-01 3.17961544e-01 1.09915435e-01
-5.13167560e-01 -5.61520398e-01 1.11939549e+00 3.30901444e-02
7.80095160e-01 -2.93116122e-01 -2.93772727e-01 -9.42837954e-01
8.12860355e-02 -4.64825124e-01 6.22669935e-01 -4.75692689e-01
-1.58423841e+00 7.07273901e-01 -8.89316946e-02 -1.15961945e+00
-4.84327853e-01 -1.18073285e+00 -4.57648546e-01 9.60774839e-01
-1.81031644e+00 -1.08773386e+00 3.91608745e-01 1.45759866e-01
8.21087956e-01 -1.94861695e-01 1.25675404e+00 2.95306385e-01
-1.89712137e-01 7.33779609e-01 1.87168092e-01 1.24235772e-01
4.86716956e-01 -1.38333821e+00 8.17744732e-01 6.75995827e-01
4.47679609e-01 8.54950130e-01 7.23543346e-01 -5.21848381e-01
-1.17886031e+00 -1.04242086e+00 2.15353227e+00 -3.42737705e-01
1.08818603e+00 -3.42173100e-01 -9.04836535e-01 7.06051886e-01
2.21141532e-01 6.18013029e-04 1.14969885e+00 5.04155099e-01
-6.36065662e-01 2.76700556e-01 -5.92445731e-01 7.01400518e-01
1.22604239e+00 -5.13305902e-01 -1.19067228e+00 3.09854656e-01
1.16363657e+00 1.65144131e-01 -5.46000123e-01 2.35840738e-01
6.13900542e-01 -7.26242304e-01 1.03371274e+00 -1.04148161e+00
3.14277381e-01 -4.13995013e-02 -7.19578505e-01 -1.51858640e+00
-7.56325006e-01 -5.48992932e-01 -3.28738391e-01 1.09314609e+00
5.68542004e-01 -6.95939183e-01 4.20927286e-01 -3.32866702e-03
-3.94363284e-01 -1.16642463e+00 -8.13525021e-01 -1.03973281e+00
3.43639135e-01 -6.48509085e-01 6.06237590e-01 1.03084588e+00
3.50690395e-01 5.29560089e-01 -3.90051156e-01 3.45246159e-02
1.41868358e-02 -9.23345797e-03 6.28062263e-02 -9.21176672e-01
-4.70718801e-01 -4.44392145e-01 -5.88694870e-01 -1.63818383e+00
4.54787642e-01 -9.66698170e-01 -1.47290062e-03 -1.48842919e+00
4.93188389e-02 -2.27221474e-01 -5.68288267e-01 4.60891992e-01
-7.55055606e-01 1.99408323e-01 -1.45312748e-03 -1.34765208e-01
-4.99087960e-01 6.23982608e-01 9.10096288e-01 9.93000567e-02
1.28205881e-01 -2.42263556e-01 -4.96635169e-01 8.57289076e-01
6.17640316e-01 -4.75371093e-01 -2.60817498e-01 -6.21332586e-01
3.80821794e-01 -2.91470557e-01 -1.96454689e-01 -3.98097873e-01
-4.97733057e-02 -2.30769873e-01 -1.73061505e-01 -6.41113818e-01
2.75696993e-01 -7.22962379e-01 -5.29037774e-01 4.45848674e-01
-5.74491382e-01 4.71139371e-01 -1.18375763e-01 5.41541517e-01
-3.90906990e-01 -5.41746259e-01 3.00359368e-01 -1.05666302e-01
-1.05293620e+00 1.01158939e-01 -5.09674191e-01 -5.28438576e-03
7.68494487e-01 -8.06961060e-02 3.94418556e-03 -9.51861143e-02
-8.27320635e-01 5.60537614e-02 2.99820974e-02 7.32523441e-01
5.38345039e-01 -1.64341891e+00 -4.72055584e-01 1.40044779e-01
6.36115372e-01 -6.71621382e-01 5.55440262e-02 5.92798948e-01
-3.65156960e-03 8.68014216e-01 3.81142497e-01 -1.33012682e-01
-1.49410963e+00 8.18110287e-01 -2.36295611e-01 -7.19790041e-01
-3.47461045e-01 8.75100315e-01 2.51004577e-01 -6.29724085e-01
1.84377253e-01 -4.70557064e-01 -5.07737398e-01 2.24814743e-01
9.02235389e-01 1.02668382e-01 1.18720993e-01 -8.78432572e-01
-5.08441150e-01 7.48408139e-01 -4.39472962e-03 -9.65768471e-02
1.25437415e+00 -1.94879577e-01 -3.51837367e-01 5.73220730e-01
1.31924927e+00 2.54948586e-01 -3.30321044e-01 -5.34783423e-01
3.74281317e-01 -5.14357448e-01 -4.93741445e-02 -6.97048664e-01
-6.78578734e-01 1.06722891e+00 2.87458897e-01 2.60774136e-01
8.22480261e-01 7.21706375e-02 1.10934329e+00 4.40689623e-01
3.86657596e-01 -1.16778433e+00 -8.23751986e-02 1.07424402e+00
3.33396137e-01 -9.93378401e-01 -3.28423798e-01 -6.16546392e-01
-6.87575579e-01 1.12325692e+00 3.03087324e-01 -1.66623890e-01
1.02916706e+00 2.35964339e-02 1.10189952e-01 7.97800813e-03
-1.00787842e+00 -5.46376944e-01 4.73843187e-01 5.15056670e-01
9.17442918e-01 3.49072158e-01 -1.11678612e+00 8.19890618e-01
-2.42446512e-01 -6.38138771e-01 2.06873007e-03 9.47589934e-01
-8.50250781e-01 -1.73697758e+00 -4.72747982e-02 2.34175906e-01
-5.31307757e-01 -6.96429253e-01 -3.70966733e-01 5.73844075e-01
1.53956711e-01 9.05237675e-01 6.50125667e-02 -4.95730579e-01
2.77028739e-01 5.17628789e-01 1.59641609e-01 -9.54750836e-01
-5.74901938e-01 -1.23358294e-02 5.87998927e-01 -7.46034265e-01
-2.32134536e-01 -5.40367603e-01 -1.14129698e+00 1.51599543e-02
-3.48486990e-01 3.62537712e-01 8.07692349e-01 1.18865252e+00
2.73474604e-01 3.87545198e-01 7.74614990e-01 -5.50890565e-01
-7.89695680e-01 -1.19575977e+00 -5.63545883e-01 5.46454906e-01
-6.70324787e-02 -5.82824826e-01 -4.91387844e-01 -9.46933031e-02] | [10.50367259979248, 8.635720252990723] |
7da74691-b9d7-4bcf-b50a-1161b76d4ef4 | mfnet-towards-real-time-semantic-segmentation | null | null | https://ieeexplore.ieee.org/abstract/document/8206396 | https://ieeexplore.ieee.org/abstract/document/8206396 | MFNet: Towards real-time semantic segmentation for autonomous vehicles with multi-spectral scenes | This work addresses the semantic segmentation of images of street scenes for autonomous vehicles based on a new RGB-Thermal dataset, which is also introduced in this paper. An increasing interest in self-driving vehicles has brought the adaptation of semantic segmentation to self-driving systems. However, recent research relating to semantic segmentation is mainly based on RGB images acquired during times of poor visibility at night and under adverse weather conditions. Furthermore, most of these methods only focused on improving performance while ignoring time consumption. The aforementioned problems prompted us to propose a new convolutional neural network architecture for multi-spectral image segmentation that enables the segmentation accuracy to be retained during real-time operation. We benchmarked our method by creating an RGB-Thermal dataset in which thermal and RGB images are combined. We showed that the segmentation accuracy was significantly increased by adding thermal infrared information. | ['Tatsuya Harada', 'Yoshitaka Ushiku', 'Takumi Karasawa', 'Kohei Watanabe', 'Qishen Ha'] | 2017-12-14 | null | null | null | ieee-rsj-international-conference-on-6 | ['real-time-semantic-segmentation', 'thermal-image-segmentation'] | ['computer-vision', 'computer-vision'] | [ 5.40819049e-01 -1.57084465e-01 2.37516478e-01 -6.63351715e-01
-1.50521785e-01 -3.84523600e-01 3.96338195e-01 -4.16531771e-01
-7.55610406e-01 5.17933190e-01 -7.16868997e-01 -3.93338650e-01
4.03003842e-02 -1.08576691e+00 -6.22863591e-01 -7.12417364e-01
3.22729141e-01 9.33062583e-02 4.33206618e-01 -4.30156052e-01
5.99345705e-03 5.08536458e-01 -2.25625181e+00 1.28732957e-02
1.13835204e+00 1.14901865e+00 6.37235284e-01 6.54021740e-01
-4.49507564e-01 1.90344274e-01 -4.29722130e-01 2.55512334e-02
7.22282112e-01 -2.78629422e-01 -7.06785798e-01 3.20578426e-01
4.26015466e-01 -1.90933451e-01 -8.89230818e-02 1.00139773e+00
1.82009697e-01 2.73685604e-01 2.08665729e-01 -1.45337152e+00
-2.16250718e-01 -2.15337798e-02 -2.71167755e-01 8.22569206e-02
-1.08272597e-01 1.84109524e-01 4.06270981e-01 -2.34043688e-01
1.46783024e-01 7.57268548e-01 7.19739079e-01 3.15738231e-01
-8.35464597e-01 -6.24638557e-01 -7.09157586e-02 5.91079414e-01
-1.27403760e+00 -1.94012076e-01 9.38327074e-01 -1.57447204e-01
9.65740025e-01 4.38783556e-01 9.70253587e-01 6.98107898e-01
1.25900060e-01 5.91343939e-01 1.59217572e+00 -2.94034541e-01
3.55332792e-01 1.48135066e-01 7.91521147e-02 4.07735765e-01
3.34682614e-01 1.38222009e-01 -7.61919767e-02 6.82235539e-01
3.30446690e-01 -6.82181632e-03 8.91393200e-02 -2.31417239e-01
-8.17004740e-01 8.14659715e-01 5.84087789e-01 5.04241407e-01
-2.48352140e-01 1.58384532e-01 2.01386839e-01 8.92341360e-02
5.50114870e-01 1.38300642e-01 -6.61709607e-01 -4.84500863e-02
-1.19754279e+00 -1.46346390e-01 4.03920978e-01 8.28504205e-01
1.34561884e+00 4.87346202e-02 5.03258646e-01 7.97429979e-01
2.42863819e-01 9.35111046e-01 3.54666352e-01 -1.10187018e+00
8.82041827e-02 6.13977015e-01 -1.68662220e-02 -7.05393374e-01
-9.65995014e-01 -2.28446826e-01 -5.97112715e-01 3.85215610e-01
1.30870387e-01 -1.04720399e-01 -1.36441338e+00 1.18730080e+00
2.74289012e-01 -8.87441114e-02 2.82866359e-01 1.09249032e+00
9.15284276e-01 5.19965172e-01 1.57582313e-01 -1.08177029e-01
1.10673892e+00 -9.44770753e-01 -8.53165925e-01 -4.70060050e-01
7.12365150e-01 -6.06076062e-01 8.96737278e-01 3.55286241e-01
-3.66095990e-01 -8.45319390e-01 -1.12523282e+00 -4.26372094e-03
-1.05975878e+00 -6.20938931e-03 7.65648723e-01 1.25775766e+00
-1.38770974e+00 4.88264799e-01 -8.52749050e-01 -8.22912872e-01
3.90722126e-01 3.42538536e-01 -2.65968084e-01 1.76794022e-01
-1.38874722e+00 9.59297597e-01 6.65955126e-01 4.10532296e-01
-4.06413645e-01 -2.01163381e-01 -7.09432960e-01 -4.66045707e-01
4.53912795e-01 -3.20630819e-01 1.05478752e+00 -1.30839002e+00
-1.65082097e+00 9.15789783e-01 -1.42175823e-01 -4.81670260e-01
4.65270072e-01 -1.80586621e-01 -6.48742378e-01 3.47237319e-01
2.82809079e-01 8.34190309e-01 5.34412801e-01 -1.49159551e+00
-8.68845344e-01 -4.74908561e-01 4.83583622e-02 1.79159433e-01
-2.43357673e-01 -3.56193393e-01 -5.67244589e-01 -6.86107650e-02
2.63028622e-01 -1.09861052e+00 -3.57636273e-01 -3.99594665e-01
-2.21779674e-01 4.58288528e-02 1.48254061e+00 -6.34125590e-01
5.50247550e-01 -1.80953133e+00 -3.87017846e-01 1.78940862e-01
-3.36561292e-01 5.69750190e-01 2.68868394e-02 -3.93384369e-03
1.22660073e-03 2.15025306e-01 -7.45392263e-01 -8.14392641e-02
-2.40837738e-01 5.75291753e-01 1.05521455e-01 4.77436870e-01
-1.80823021e-02 8.48943293e-01 -4.90886301e-01 -6.61791384e-01
9.79629219e-01 4.09148395e-01 -1.51031882e-01 4.73338850e-02
-2.43712351e-01 6.55501425e-01 -3.26229841e-01 6.42026424e-01
1.18530607e+00 3.93445730e-01 -2.38849387e-01 -3.25547576e-01
-5.47640979e-01 -2.72433043e-01 -8.53145003e-01 1.78463089e+00
-5.98480880e-01 8.67942750e-01 1.96360201e-01 -1.10814393e+00
8.39233696e-01 -6.34811074e-02 9.48787391e-01 -1.55627918e+00
4.75784242e-01 1.40760079e-01 -3.46775502e-01 -8.12906682e-01
9.74391520e-01 -7.94906691e-02 7.17124268e-02 2.11966813e-01
-3.30579877e-01 -7.02715576e-01 2.78814703e-01 -2.87073463e-01
3.53204638e-01 7.05437839e-01 -3.68721962e-01 -1.17803320e-01
6.68172061e-01 6.82164013e-01 3.34956735e-01 5.64919472e-01
-4.92391706e-01 6.08422577e-01 -2.04892352e-01 -4.93735760e-01
-1.10001791e+00 -8.59214365e-01 -3.28102052e-01 9.06898201e-01
7.13329315e-01 1.17667101e-01 -1.12852049e+00 -3.32645923e-01
-2.08882093e-01 8.75135064e-01 -4.26388234e-01 1.33323878e-01
-4.86600220e-01 -1.19715631e+00 4.26229775e-01 2.80077547e-01
1.30233347e+00 -8.95679414e-01 -1.21111751e+00 1.30713224e-01
-3.46766531e-01 -1.52801335e+00 4.19425189e-01 3.46799999e-01
-9.70072150e-01 -1.20909274e+00 -3.18527550e-01 -4.25787091e-01
2.41632998e-01 6.79817080e-01 6.90875828e-01 -1.46788461e-02
-5.08862138e-01 4.23975080e-01 -6.08263552e-01 -5.93840659e-01
-2.23059803e-01 2.97413141e-01 -2.77662337e-01 3.20876576e-02
5.21285772e-01 -2.10451663e-01 -6.13635123e-01 3.77986044e-01
-1.10728037e+00 4.70270813e-01 4.90052700e-01 1.92417666e-01
4.51363444e-01 5.15350103e-01 9.73888114e-02 -6.82218134e-01
-2.47838888e-02 -1.30698934e-01 -6.86231494e-01 -3.46077718e-02
-5.62266171e-01 -2.24465519e-01 4.10249859e-01 3.01132619e-01
-1.43340600e+00 4.83781010e-01 -5.03141820e-01 -2.51216851e-02
-7.68419743e-01 6.06437884e-02 -1.33891672e-01 -3.43661964e-01
3.86115581e-01 1.97010547e-01 2.19750181e-01 -1.66834384e-01
5.17485201e-01 9.96945143e-01 5.52342355e-01 -1.21359639e-01
7.45543182e-01 9.64896917e-01 4.78390940e-02 -1.43609548e+00
-5.43424249e-01 -7.51188278e-01 -9.44423258e-01 -7.23258018e-01
1.32517064e+00 -7.49640584e-01 -5.96343875e-01 8.89435351e-01
-7.03648031e-01 -5.06177485e-01 7.35846981e-02 3.33530843e-01
-7.07055807e-01 6.76526487e-01 4.14429493e-02 -9.39281523e-01
-2.32214943e-01 -1.27528083e+00 9.25201774e-01 6.39391243e-01
3.69793504e-01 -9.49691117e-01 -2.67208636e-01 8.70950639e-01
6.92726791e-01 1.59392789e-01 3.83878559e-01 4.45145927e-02
-5.32925248e-01 3.15448567e-02 -5.06131709e-01 5.31805217e-01
1.62405670e-01 3.32832783e-02 -1.45682478e+00 1.70789868e-01
8.35080147e-02 -8.20998326e-02 1.12198365e+00 4.56816494e-01
1.32570088e+00 4.05875981e-01 -3.45246196e-01 9.29049194e-01
1.73516917e+00 4.51955289e-01 8.52925003e-01 1.06314421e+00
8.98177266e-01 8.87486100e-01 8.46162379e-01 3.85354124e-02
4.60706830e-01 4.57934260e-01 9.17603612e-01 -7.37946808e-01
2.13115923e-02 4.84938443e-01 -6.95103556e-02 3.77053738e-01
-2.60247052e-01 -1.56828731e-01 -1.04933381e+00 6.06562793e-01
-1.83684480e+00 -6.38393760e-01 -7.95893073e-01 1.93431890e+00
2.70688415e-01 1.83914769e-02 -2.80797970e-03 2.65050232e-01
6.47270858e-01 7.10653439e-02 -4.69579458e-01 -5.80784440e-01
-3.80441666e-01 4.19881672e-01 1.33114958e+00 2.79482543e-01
-1.61463678e+00 1.20061255e+00 6.61134005e+00 5.54364860e-01
-1.30970395e+00 2.38253698e-01 5.05341649e-01 2.93089986e-01
-2.88976561e-02 -4.78380322e-02 -3.71908545e-01 3.61132354e-01
1.27849400e+00 2.71800488e-01 2.99199402e-01 7.43743122e-01
5.69071233e-01 -8.41297984e-01 -9.35625359e-02 8.32559288e-01
5.56021929e-02 -8.25230062e-01 -3.09574366e-01 3.65567245e-02
7.47747719e-01 2.88951993e-01 -1.26441997e-02 -2.27867458e-02
1.46584986e-02 -9.40314829e-01 6.32913530e-01 5.44966459e-01
6.29970551e-01 -9.42332089e-01 9.10625339e-01 2.06588566e-01
-1.27840257e+00 -1.30280629e-01 -2.34334409e-01 -1.32385939e-01
3.84310149e-02 4.95704144e-01 -8.00809324e-01 8.54852617e-01
1.01157022e+00 6.44102275e-01 -9.56980288e-01 9.27749336e-01
-1.51086092e-01 4.40735042e-01 -4.19064403e-01 9.41038430e-02
5.79605579e-01 -7.18851864e-01 -8.14691302e-04 1.21524334e+00
2.59622097e-01 -1.60665110e-01 -1.49806822e-03 8.22012007e-01
6.86293840e-01 -7.43198991e-02 -6.43545985e-01 1.02952383e-01
-2.16561317e-01 1.55577111e+00 -1.29003561e+00 -2.92208731e-01
-4.88869637e-01 1.07502997e+00 -3.90399784e-01 5.39462745e-01
-1.10716784e+00 -3.93932819e-01 6.33153141e-01 -2.56567225e-02
2.14739084e-01 -5.86453080e-01 -8.09254467e-01 -6.33673668e-01
-2.61551589e-01 -2.45025363e-02 -1.05287507e-01 -1.00493765e+00
-5.68056881e-01 5.21437764e-01 2.30900422e-01 -9.83737051e-01
1.73670679e-01 -8.65159810e-01 -2.81613410e-01 7.70183265e-01
-1.90233552e+00 -1.20047760e+00 -8.65659058e-01 5.10987103e-01
5.72382033e-01 2.13030279e-01 5.89050174e-01 3.07027608e-01
-7.04482675e-01 -4.09186026e-03 2.51386255e-01 -1.51765317e-01
2.77890563e-01 -1.06479192e+00 2.89622724e-01 1.07348323e+00
-4.50516075e-01 -7.24483058e-02 7.66477406e-01 -4.65042621e-01
-1.26988089e+00 -1.45035446e+00 3.61629993e-01 -4.15134151e-03
2.78090149e-01 -8.82343724e-02 -7.32053220e-01 2.96821743e-01
3.37242037e-01 -3.16764683e-01 3.92955840e-01 -5.41493595e-01
1.62512824e-01 -4.19485688e-01 -1.26951551e+00 3.79863948e-01
8.67358744e-01 -4.14627820e-01 -4.09402817e-01 2.39670217e-01
6.56946540e-01 -2.41961032e-01 -4.72145826e-01 6.17658079e-01
4.15645659e-01 -1.39551234e+00 8.33537757e-01 2.17796922e-01
3.51496451e-02 -5.99788427e-01 -1.43713668e-01 -1.06760299e+00
1.16873145e-01 9.24718380e-02 6.35893106e-01 7.55934596e-01
1.81087941e-01 -7.60862172e-01 9.58675086e-01 5.26384175e-01
-6.00675762e-01 4.24724221e-02 -1.02748656e+00 -7.04569221e-01
-1.31753311e-01 -9.76702750e-01 4.88343328e-01 6.73256159e-01
-7.79230654e-01 -2.03553155e-01 -7.27316216e-02 2.35520467e-01
6.60349727e-01 1.65143251e-01 8.19093645e-01 -1.27142060e+00
5.43559015e-01 -4.51505274e-01 -5.43628275e-01 -4.03662771e-01
2.71304935e-01 -6.87768638e-01 5.03226161e-01 -2.00417423e+00
-6.91164806e-02 -4.56568599e-01 -2.77258251e-02 3.79006326e-01
1.35577228e-02 8.83551240e-01 -5.79349026e-02 5.08413427e-02
-4.92838979e-01 4.69157100e-01 1.15309000e+00 -2.26722807e-01
-1.81612238e-01 -7.03557953e-02 -2.42067322e-01 5.61728597e-01
1.26965809e+00 -9.88440588e-02 -3.24783862e-01 -2.41794840e-01
2.19251160e-02 -4.20483351e-01 3.55405807e-01 -1.49714017e+00
1.39370680e-01 -2.49151602e-01 2.41930038e-01 -1.04736662e+00
4.02751178e-01 -1.21509194e+00 3.81560713e-01 4.98055190e-01
3.11527699e-01 -3.85339648e-01 4.77912396e-01 3.48195106e-01
-1.73699409e-01 -8.48989487e-02 8.83248866e-01 -2.74739742e-01
-1.63007545e+00 3.04953475e-02 -6.50874317e-01 -5.29644489e-01
1.37677395e+00 -9.88108754e-01 -2.11960837e-01 3.68732470e-03
-5.63726306e-01 2.60186762e-01 7.18810141e-01 4.61757541e-01
4.31796402e-01 -9.22693968e-01 -2.21966907e-01 3.52165967e-01
2.82837033e-01 1.77795053e-01 6.18195355e-01 8.46904159e-01
-1.16679633e+00 6.63221896e-01 -5.66032767e-01 -9.81671631e-01
-1.18844843e+00 3.96976590e-01 6.16961658e-01 5.02091885e-01
-5.16303360e-01 4.94068950e-01 -1.22734047e-01 -6.27589226e-01
-1.50235325e-01 -3.56313825e-01 -4.02929902e-01 1.00249648e-01
-5.41443042e-02 4.97269511e-01 4.51577187e-01 -1.15539134e+00
-4.10073847e-01 8.72628331e-01 5.52114367e-01 -1.07813992e-01
1.17049801e+00 -5.75427830e-01 -2.74798423e-01 5.04657924e-01
1.19107103e+00 -6.81027651e-01 -1.30760789e+00 1.11720100e-01
-1.59155443e-01 -3.78503203e-01 4.94773000e-01 -6.76188052e-01
-1.36657774e+00 7.56819248e-01 1.11358857e+00 4.61582571e-01
1.60886002e+00 -3.52021128e-01 8.63755107e-01 3.80226076e-01
4.74651039e-01 -1.78109837e+00 -3.68655533e-01 6.58736110e-01
2.78849721e-01 -1.54323959e+00 -8.80395100e-02 -5.01980066e-01
-5.18239856e-01 1.38970065e+00 6.78594828e-01 2.58405685e-01
4.74728972e-01 5.31063937e-02 5.11736274e-01 -1.67336956e-01
7.87578821e-02 -1.00656652e+00 1.18950335e-02 9.71988678e-01
2.55917609e-01 2.46858567e-01 -2.26727515e-01 -1.33170143e-01
-3.98422480e-01 9.95160546e-03 6.00806892e-01 1.16123319e+00
-8.00621331e-01 -8.60658407e-01 -7.86679745e-01 2.72786885e-01
-9.49653834e-02 1.96050718e-01 -3.63125980e-01 9.34654355e-01
7.37442672e-01 1.50207734e+00 3.06435257e-01 -6.64735913e-01
2.41288334e-01 1.89465806e-01 2.23222613e-01 -1.78613082e-01
-3.55385184e-01 -9.56469402e-02 7.09302500e-02 -7.90617645e-01
-1.10086942e+00 -5.14398694e-01 -1.47159398e+00 -2.80176491e-01
-2.74105966e-01 -4.10259143e-02 1.62195039e+00 1.09433973e+00
-6.36136085e-02 8.28907609e-01 7.59243727e-01 -1.06835806e+00
4.68051434e-01 -8.70161951e-01 -8.12507153e-01 2.76520312e-01
3.09305042e-01 -6.39466047e-01 -3.44450384e-01 5.05910534e-03] | [9.003979682922363, -1.5247900485992432] |
aa100656-0afd-458d-8ea2-16818c140ef3 | combining-shallow-and-linguistically | null | null | https://aclanthology.org/W13-1726 | https://aclanthology.org/W13-1726.pdf | Combining Shallow and Linguistically Motivated Features in Native Language Identification | null | ['Sowmya Vajjala', 'Detmar Meurers', 'Serhiy Bykh', 'Julia Krivanek'] | 2013-06-01 | null | null | null | ws-2013-6 | ['native-language-identification'] | ['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] | [-7.335831642150879, 3.6701674461364746] |
6f65de30-a51a-4e7c-b1f6-fffe55d295a3 | vision-language-transformer-and-query | 2108.05565 | null | https://arxiv.org/abs/2108.05565v1 | https://arxiv.org/pdf/2108.05565v1.pdf | Vision-Language Transformer and Query Generation for Referring Segmentation | In this work, we address the challenging task of referring segmentation. The query expression in referring segmentation typically indicates the target object by describing its relationship with others. Therefore, to find the target one among all instances in the image, the model must have a holistic understanding of the whole image. To achieve this, we reformulate referring segmentation as a direct attention problem: finding the region in the image where the query language expression is most attended to. We introduce transformer and multi-head attention to build a network with an encoder-decoder attention mechanism architecture that "queries" the given image with the language expression. Furthermore, we propose a Query Generation Module, which produces multiple sets of queries with different attention weights that represent the diversified comprehensions of the language expression from different aspects. At the same time, to find the best way from these diversified comprehensions based on visual clues, we further propose a Query Balance Module to adaptively select the output features of these queries for a better mask generation. Without bells and whistles, our approach is light-weight and achieves new state-of-the-art performance consistently on three referring segmentation datasets, RefCOCO, RefCOCO+, and G-Ref. Our code is available at https://github.com/henghuiding/Vision-Language-Transformer. | ['Xudong Jiang', 'Suchen Wang', 'Chang Liu', 'Henghui Ding'] | 2021-08-12 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Ding_Vision-Language_Transformer_and_Query_Generation_for_Referring_Segmentation_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Ding_Vision-Language_Transformer_and_Query_Generation_for_Referring_Segmentation_ICCV_2021_paper.pdf | iccv-2021-1 | ['generalized-referring-expression-segmentation', 'referring-expression-segmentation'] | ['computer-vision', 'computer-vision'] | [ 0.132929 0.0541526 -0.31166387 -0.45031774 -0.9523509 -0.41902527
0.2513489 -0.32940218 -0.23258266 0.23399696 0.14480236 -0.11585448
0.14826262 -0.71958977 -0.7931515 -0.5152196 0.63574916 0.3656157
0.39320803 -0.35384807 0.31703204 0.26099378 -1.2658937 0.46654597
0.9239249 1.0339515 0.75834775 0.4881866 -0.42577514 0.88247716
-0.6623054 -0.5794572 -0.1050668 -0.7911618 -1.0385914 0.32251337
0.54562014 -0.16314115 -0.36642888 1.2982132 0.51970345 0.03350534
0.4006218 -1.0777541 -1.2230331 0.6279359 -0.7378025 0.53293175
0.50784487 0.35156587 1.2130843 -0.9259525 0.6110057 1.51519
-0.07167329 0.62427926 -1.1043139 -0.6271525 0.62487423 0.4361467
-1.5040243 -0.29810548 0.8313289 -0.34385788 0.7120874 0.22054668
0.6698007 0.8818937 -0.04221601 1.292383 0.7442243 -0.20200275
-0.12961023 0.0864475 0.16082631 0.8073513 -0.2869759 -0.35546878
-0.5763676 0.2902251 0.7723338 -0.01813721 -0.63541114 0.05205689
-1.1522275 0.70690393 0.9003573 0.46631902 -0.48886818 0.43110028
0.054523 -0.07701596 0.2376111 0.4377999 -0.4582298 0.15083812
-0.7839815 0.0442965 0.3539641 1.3412367 0.88577455 -0.2350244
-0.5809433 0.83392495 0.4719451 0.38429198 0.28721932 -1.1168342
0.6240966 0.87226164 -0.14501022 -0.8879313 -0.03480601 -0.576339
-0.65600735 -0.2245016 0.1545862 0.00731699 -1.021571 1.8015004
0.23306665 -0.07695106 -0.08876411 1.188528 1.2090106 0.81315154
0.23385718 0.14412823 1.6665548 -1.2404884 -0.74422574 -0.5786404
0.29031023 -0.75642955 1.4044905 0.06781141 -1.1382436 -0.75642085
-0.65859026 -0.659816 -0.31026986 0.3467101 0.44688225 -0.06064583
-1.0983824 0.06811661 -0.3699408 -0.3309336 0.67817646 0.26443547
-0.00738298 -0.1050219 -1.1490712 0.69729465 0.34659693 0.208045
-0.762664 -0.46413547 -0.7367155 0.19194394 0.6883493 -0.98647714
1.253857 -1.278622 -1.0733031 1.2733879 -0.6690398 -0.12970154
0.29041824 -0.17164965 -0.11677354 0.27808452 0.464298 1.2400767
0.7072982 -1.4945418 -0.6755834 -0.43530536 0.39144906 0.29579896
0.28649676 0.21528704 -1.3390096 -0.50989676 0.36761114 -0.529558
-0.03861364 0.07515302 -0.80313975 -0.51180017 0.73188114 -0.5617079
1.2687305 -2.3109145 0.45507076 -0.14904115 0.40548334 0.07153378
-0.15066536 0.14466289 -0.11486117 0.441832 -0.21336283 -0.16310515
-0.18948665 0.23337075 -0.32887882 0.02794873 0.4915977 1.2960286
-0.9030861 -0.83423644 -0.04629502 0.28033403 -0.33125493 0.41994607
-0.43815157 0.5934381 -0.8663977 0.91140723 0.37723893 -0.5470652
-0.4669131 -0.5401904 0.08101497 -0.10787956 -0.67171896 1.7560395
-0.34856635 0.5341603 0.12426212 -0.82541233 0.99274176 -0.05083398
0.10479563 -0.8979919 0.3456101 0.1254233 0.12379975 -0.88248515
0.30173564 0.12845805 -0.20703043 0.29044375 0.07515758 -0.25489226
0.2898297 0.38589618 0.61729085 0.143927 0.14910735 -0.04745815
0.7115024 -0.05246082 0.5550865 0.7283155 -0.29144552 0.72265404
0.84346575 -0.1677835 -0.5612837 -0.8021486 0.10095554 1.3869274
0.66249496 -0.06908084 -0.73124385 -0.54134446 -0.32857835 0.814472
-0.7539096 -0.09467697 -0.4992969 -0.29658076 0.2603922 0.58700347
0.8310316 -1.5223418 -0.77434117 -0.03870764 -0.37410727 -1.0952772
-0.9598605 0.15897532 -0.48538393 -1.0965458 -0.8837508 -1.0312937
0.78112584 0.26112658 1.3720115 0.34110078 -0.10281318 0.3330416
-0.31853482 -0.26953128 -0.06566086 0.12505502 -0.8253808 0.17305826
0.385067 -0.03764004 -0.7214886 0.261045 -0.82911193 0.3293936
0.5918126 0.83193964 1.0915945 -0.35665143 0.2963433 -0.72990346
0.6914744 -0.5067958 -0.42420918 0.6038824 -0.04485422 0.02511328
0.4567765 -0.35360786 -0.85080755 0.04249223 -0.16539066 -0.66695136
-0.15415233 0.33797777 -0.49515492 0.26556355 0.21333355 0.5741342
-0.3482139 -0.24967377 0.7269589 0.42354763 0.6043343 -0.68457747
0.34095758 0.2806114 -0.41479823 -0.5304489 -1.2247018 -0.47645834
-0.69488454 -0.25204796 1.3178326 -0.74462855 -0.6903291 0.3261419
-1.5627117 -0.39564687 -0.22143649 -0.11251416 -0.5073636 0.0189063
-0.35419932 -0.73791945 -0.35033128 -1.6310443 1.5170146 0.74041474
-0.05886613 -0.6555293 -0.54063374 0.43124127 0.2819076 0.03992832
1.042144 -0.5275065 -1.0721365 0.13612527 -0.8102131 -0.05346919
0.03898609 0.11914226 -0.98131895 0.18028185 -0.18400244 -0.25057665
0.9742209 0.43958893 1.6894752 -0.12263998 -0.29783502 0.8544256
1.3201284 0.5208379 0.5549779 0.18390727 0.8145931 0.7637266
0.65531105 -0.01923317 0.7113099 0.701533 0.6492016 -0.46516454
-0.3692145 -0.29902074 -0.06362718 0.37165043 0.31322595 -0.51553345
-0.8587114 0.71303374 -1.7941842 -0.8313122 -0.08429657 1.6755315
0.68804336 0.03710551 -0.17659022 -0.44657415 0.7927019 0.18947306
-1.0142856 -0.25982538 -0.21801043 0.02917108 0.19031759 0.4209373
-0.9875488 1.4229158 5.434303 0.9350448 -1.2049862 0.04996531
1.1100723 0.15432233 -0.4951124 -0.04821099 -0.8322963 0.34547132
0.15841019 -0.29916865 0.40059024 0.586191 0.36601883 -0.3681934
-0.9949521 1.066756 0.20793954 -1.0490612 0.2728045 -0.21861179
0.47214147 -0.03322073 0.18092263 0.20493606 0.01823586 -1.0846603
0.8914428 0.74936396 0.7731593 -0.4643708 0.64532137 0.2646289
-1.3132235 -0.04869537 -0.11598173 0.4612251 0.30482092 0.19371255
-0.33124095 0.44553375 0.76302445 0.77100444 -0.76042384 0.8888447
-0.59261566 0.33872116 -0.0585859 -0.08205187 0.47383642 -0.17961825
0.29212204 1.0984616 0.21292889 0.35635412 0.17510211 1.5580755
-0.28577906 0.21274826 -0.48838046 0.05724742 0.55050784 1.1839906
-0.7411413 -0.46154162 -0.38928655 1.1056005 0.3727082 0.65762985
-0.9165703 -0.31547064 0.33619866 0.0928756 0.3685546 0.09946448
-0.19796063 -1.0152034 0.0229347 -0.8057099 0.36323532 -1.3923621
-1.1641482 0.7368309 0.00949646 -0.9820116 0.10028163 -0.50212836
-0.8446536 1.2912692 -1.5755854 -1.250257 -0.45769706 0.6353561
1.0573285 0.08582928 0.410943 0.13453153 -0.8628638 0.41639763
-0.7552678 0.23012924 0.5040862 -1.1913004 0.27466393 0.78029877
0.25763932 0.70978516 0.38846552 -0.44087347 -1.1346111 -1.0248424
0.9093118 -0.18455377 0.67427677 -0.06534309 -1.1588764 0.6298498
0.4877316 -0.03678358 0.34007108 -0.1444641 -0.18934211 -0.04120266
-0.78635246 0.7600421 1.0709223 -0.5840558 -0.5963191 0.36850962
1.069746 -0.62087816 -0.51200026 0.05293968 0.18381666 -1.018825
0.84771514 -0.47082993 0.6015216 -0.44257644 -0.18650617 -0.9410712
-0.17404862 -0.3649864 0.19127673 1.3234576 0.56246096 -0.45179802
0.53708506 0.52824616 -0.21230237 -1.2617787 -0.7990261 -0.13970168
-0.07026295 -0.366539 0.62471443 0.6809117 -0.42564118 0.61867523
-0.02035187 0.14387003 0.3141505 0.41588172 0.5980346 -0.7139578
-0.23590317 -0.74816746 -0.1246687 -1.6470515 0.30498323 -0.9127798
0.10029585 -1.7888534 0.38552025 -0.16220725 -0.15165526 0.5837202
-0.5190187 -0.11700268 0.480325 0.24103808 -0.8393051 0.48136452
1.8423133 -0.37368134 -0.09466977 -0.06452057 -1.168106 0.6899137
0.6780972 -0.25787652 -0.4987507 -0.9639611 0.02160723 0.26387772
0.6287486 -0.48817214 0.36063057 -0.11761347 0.26387346 -0.9077517
0.35065645 -0.6552518 -0.26797408 0.22279438 -0.6038212 0.311916
0.01566353 0.27355424 -0.43292692 -0.2375861 0.68014914 -0.44331652
-1.0875809 0.53957367 -0.0375529 0.41319668 0.9166683 -0.24435796
-0.35646126 -0.51805305 -0.8789583 0.8252409 0.2882722 0.4733867
0.916961 -1.1425688 -0.6301767 0.04273 0.32744315 0.39324033
0.3451913 0.75425947 -0.30133113 0.29467452 0.14161626 -0.7293449
-1.142739 0.5631411 0.7536173 -0.04577574 -0.4131868 1.3563961
0.7169455 -0.09701555 0.19080834 -0.46425483 -0.41263616 0.07470886
0.30210617 -0.15295061 -0.47393304 -0.997112 -0.29205143 1.0211663
-0.04468745 0.08479814 0.81922495 -0.38534936 -0.21557598 0.50001556
1.1582689 -0.23366666 -1.1811204 -0.27925292 -0.06535154 -0.47144735
-0.06997503 -0.7700446 -1.4645644 1.1656164 0.3901935 0.12517299
1.4600853 0.69861585 0.6232843 0.10970958 0.15749575 -0.61690575
0.33350188 0.39699748 1.1715119 -1.3363177 -0.28073096 -0.5106177
-0.9463706 0.9645753 1.105746 -0.01996121 0.3140764 -0.05107831
0.33255777 -0.7069698 -0.7111156 -0.7950785 0.38652223 0.47010213
0.6330805 -0.08748037 -0.10620245 0.5383958 -0.0716917 -0.16283618
0.20070769 0.36290744 -0.5429889 -0.92705804 -0.35452175 0.37331375
-0.38367528 -0.24045952 -0.6526298 0.75172055 0.3424057 0.90614754
0.05682119 -0.17003398 0.6379425 -0.03156331 0.15176176 -0.8119032
-0.5177768 0.37205505 -0.24914813 -0.7132956 -0.4980167 -0.40572995
-1.4880105 0.22309315 -0.29049385 -0.15012902 0.19629422 0.9566424
0.38300893 0.9956127 0.352831 -0.665655 -0.13891563 -0.75001365
-0.3567344 0.363968 0.26279667 -0.5592056 0.03906657 0.03991944] | [10.277024269104004, 1.260948657989502] |
2800eaf7-81b7-4101-9860-df3f65dfaade | vqa-and-visual-reasoning-an-overview-of | 2212.13296 | null | https://arxiv.org/abs/2212.13296v1 | https://arxiv.org/pdf/2212.13296v1.pdf | VQA and Visual Reasoning: An Overview of Recent Datasets, Methods and Challenges | Artificial Intelligence (AI) and its applications have sparked extraordinary interest in recent years. This achievement can be ascribed in part to advances in AI subfields including Machine Learning (ML), Computer Vision (CV), and Natural Language Processing (NLP). Deep learning, a sub-field of machine learning that employs artificial neural network concepts, has enabled the most rapid growth in these domains. The integration of vision and language has sparked a lot of attention as a result of this. The tasks have been created in such a way that they properly exemplify the concepts of deep learning. In this review paper, we provide a thorough and an extensive review of the state of the arts approaches, key models design principles and discuss existing datasets, methods, their problem formulation and evaluation measures for VQA and Visual reasoning tasks to understand vision and language representation learning. We also present some potential future paths in this field of research, with the hope that our study may generate new ideas and novel approaches to handle existing difficulties and develop new applications. | ['Yuezhou Dong', 'Zaharaddeen Karami Lawal', 'Ke Qin', 'Hailin Wang', 'Jim Wilson Owusu', 'Rufai Yusuf Zakari'] | 2022-12-26 | null | null | null | null | ['visual-reasoning', 'visual-reasoning'] | ['computer-vision', 'reasoning'] | [ 2.46514797e-01 -8.48994777e-02 -4.01368737e-01 -4.15660203e-01
-3.45612466e-01 -4.76211101e-01 9.81671095e-01 1.93581983e-01
-4.23506230e-01 4.36604649e-01 2.84220070e-01 -2.47455016e-01
-4.90265936e-02 -8.51904988e-01 -3.30678314e-01 -5.47596216e-01
1.09848700e-01 4.49337602e-01 4.17777747e-02 -2.75415838e-01
3.57212275e-01 8.49058032e-01 -1.74919832e+00 6.28685355e-01
7.23242462e-01 1.15594053e+00 2.61978239e-01 5.42942703e-01
-6.09777570e-01 1.32904732e+00 -4.83136982e-01 -5.26892662e-01
1.25267595e-01 -3.32261890e-01 -1.06084883e+00 1.04682721e-01
5.44349074e-01 1.12915829e-01 -4.89667267e-01 8.11587691e-01
4.68800992e-01 1.32265896e-01 7.10896909e-01 -1.21970034e+00
-1.09316671e+00 7.90568665e-02 -5.77350199e-01 3.22798222e-01
1.66439980e-01 2.98593938e-01 9.97301936e-01 -9.23071027e-01
5.44067442e-01 1.35997283e+00 4.13839191e-01 6.84720635e-01
-9.42132533e-01 -3.52413774e-01 3.62458942e-03 6.68999910e-01
-1.06139326e+00 -2.92102128e-01 7.30533600e-01 -5.59039533e-01
1.22707486e+00 -8.72301236e-02 7.33565927e-01 1.07072282e+00
1.62005484e-01 1.26680899e+00 1.19593716e+00 -8.88166487e-01
1.56028748e-01 2.15971887e-01 2.66459018e-01 9.07437980e-01
1.28502935e-01 1.38780698e-01 -4.04638082e-01 2.68458158e-01
7.16139257e-01 -1.11888759e-02 -3.75072397e-02 -5.35233915e-01
-1.25720382e+00 1.26295769e+00 6.83021545e-01 6.86958194e-01
-4.27930832e-01 1.38870075e-01 5.79964578e-01 1.80959016e-01
8.94514397e-02 6.88557565e-01 -2.81933635e-01 -2.70485301e-02
-7.86918044e-01 3.05861145e-01 6.56574607e-01 6.52098119e-01
6.23529375e-01 4.71425533e-01 -3.01878095e-01 1.07741868e+00
2.93875813e-01 5.81635118e-01 4.47544158e-01 -1.24423480e+00
1.05558939e-01 7.11227119e-01 -3.58112335e-01 -1.01872659e+00
-3.33365977e-01 -2.50377119e-01 -9.88709152e-01 5.59808135e-01
2.93133050e-01 1.82979926e-01 -1.11872864e+00 1.42267537e+00
-8.62512738e-02 -2.06319883e-01 2.66973913e-01 7.25152850e-01
1.35002947e+00 8.41560245e-01 1.53093994e-01 -6.62871962e-03
1.49562073e+00 -1.17221069e+00 -6.95963621e-01 -3.95852685e-01
1.75743178e-01 -9.18988943e-01 9.21808302e-01 4.58264381e-01
-1.17645192e+00 -7.83466041e-01 -8.36321294e-01 -4.71955448e-01
-6.68320894e-01 2.44750874e-03 1.03055143e+00 4.49173033e-01
-1.16066909e+00 1.54990166e-01 -6.26359224e-01 -7.01667130e-01
9.23164368e-01 2.35660285e-01 -1.61093175e-01 -3.15242320e-01
-1.05665565e+00 1.17635703e+00 3.21401894e-01 1.46867428e-02
-8.29415321e-01 -4.55108583e-01 -9.24337149e-01 -1.59430951e-01
3.73370618e-01 -9.02326345e-01 1.23297822e+00 -1.24983668e+00
-1.13346982e+00 1.40105820e+00 -2.08931386e-01 -8.19347024e-01
4.39615995e-02 4.51839231e-02 -3.80191654e-01 2.24001005e-01
-3.07598375e-02 1.08036470e+00 8.51291239e-01 -1.24411142e+00
-7.88090467e-01 -5.35329521e-01 2.65153855e-01 1.44925013e-01
-2.62981296e-01 1.83157220e-01 -4.63553488e-01 -6.01570368e-01
-3.77948701e-01 -6.12139642e-01 -1.43242374e-01 2.77642459e-01
4.85159345e-02 -5.91213048e-01 5.62678218e-01 -2.35063657e-01
8.20783556e-01 -1.94354200e+00 2.37277731e-01 -1.78761274e-01
4.26368713e-01 8.03667963e-01 -2.52351701e-01 6.19886696e-01
2.42543593e-01 -1.04375802e-01 -1.88529044e-01 -2.07679674e-01
-2.23797649e-01 5.61307788e-01 -5.99398196e-01 1.34493202e-01
2.51714438e-01 1.21392739e+00 -7.98579812e-01 -5.74491858e-01
5.73069513e-01 6.40626669e-01 -1.50727451e-01 2.28933826e-01
-4.34811085e-01 2.71033496e-01 -4.77761596e-01 7.05641985e-01
3.64204854e-01 -3.86920720e-01 -2.29882374e-01 -2.95280933e-01
-4.21649754e-01 -1.27161425e-02 -6.13189638e-01 1.68723595e+00
-2.71816075e-01 1.12315929e+00 4.98619303e-02 -1.54084969e+00
8.89544606e-01 2.07683980e-01 4.85082477e-01 -1.10908222e+00
1.06467098e-01 -9.86193717e-02 -4.76178713e-03 -9.32441592e-01
3.61520767e-01 -1.09610438e-01 3.81649852e-01 1.40590459e-01
3.22002098e-02 -3.13506126e-01 3.30612242e-01 1.30681798e-01
7.16634393e-01 -4.96231876e-02 6.61074638e-01 9.15701687e-02
8.83487165e-01 4.30345535e-01 2.29835093e-01 7.66906202e-01
-6.28747284e-01 2.72533000e-01 -2.98866294e-02 -9.35175359e-01
-9.38495755e-01 -1.02590120e+00 -7.95106068e-02 1.00672090e+00
-8.76079351e-02 -1.91340595e-01 -5.54985464e-01 -2.41632760e-01
3.62794660e-02 5.23467004e-01 -6.81273222e-01 -6.05238043e-02
-4.74748015e-01 -4.67212617e-01 6.21994913e-01 7.79922903e-01
8.19790006e-01 -1.87399697e+00 -7.03350902e-01 -9.96954814e-02
-4.54053506e-02 -1.34531903e+00 4.36122000e-01 -5.89972585e-02
-9.17193770e-01 -1.02841175e+00 -8.66931081e-01 -1.30915427e+00
3.26556206e-01 6.22996032e-01 1.36825383e+00 2.06514285e-03
-5.85327744e-01 8.90220046e-01 -4.51670051e-01 -8.05606127e-01
-4.53974187e-01 -2.05250591e-01 -4.06075157e-02 -2.32550845e-01
9.38325644e-01 -1.41926110e-01 -5.26980758e-01 -2.76730418e-01
-1.04642189e+00 4.76436876e-02 6.88246727e-01 7.56006002e-01
5.56528747e-01 -1.60997316e-01 4.81465280e-01 -4.68708158e-01
6.66631818e-01 -2.87348926e-01 -6.15658998e-01 3.37991863e-01
-6.72479093e-01 2.81021688e-02 3.68627876e-01 -1.29284948e-01
-1.12220049e+00 -1.61147669e-01 -2.39740282e-01 -2.85093457e-01
-6.09527349e-01 5.92301309e-01 1.45977125e-01 -2.99938142e-01
7.45141983e-01 3.31473351e-01 1.96007758e-01 -3.26456428e-01
7.11432099e-01 7.69727290e-01 5.76690078e-01 -5.66292882e-01
5.98736227e-01 7.07905233e-01 7.33773038e-02 -1.29261601e+00
-9.20927048e-01 -4.76295322e-01 -5.32561779e-01 -2.32821286e-01
1.25590599e+00 -7.41835952e-01 -7.27243841e-01 3.46726269e-01
-1.36806691e+00 -1.56601369e-02 -4.16306764e-01 4.36214387e-01
-5.88199973e-01 4.32430774e-01 -4.82308894e-01 -9.00782049e-01
-4.78409410e-01 -1.24519372e+00 8.36654663e-01 6.08906329e-01
-9.21157673e-02 -1.15175796e+00 2.10479707e-01 8.25102746e-01
4.99301016e-01 1.42108619e-01 1.02306759e+00 -4.80629772e-01
-6.99793279e-01 4.58678827e-02 -6.19664490e-01 6.50384009e-01
-7.48165622e-02 -1.77700527e-03 -1.21394360e+00 -1.81066185e-01
-9.42467153e-02 -8.58115733e-01 1.15160441e+00 7.03165948e-01
1.12279880e+00 1.78178534e-01 -2.08688363e-01 5.14626741e-01
1.53257871e+00 4.97019231e-01 6.01762414e-01 5.73584259e-01
6.44315779e-01 7.65156388e-01 4.24788266e-01 1.48344338e-01
4.47075158e-01 5.31864405e-01 7.07364321e-01 -4.71110851e-01
-3.82996529e-01 1.85238034e-01 1.06925480e-01 6.58781946e-01
-5.10246813e-01 -1.54155254e-01 -1.22600102e+00 6.04426861e-01
-1.70078540e+00 -1.24055123e+00 5.12183122e-02 1.90202534e+00
6.02545500e-01 -1.66661993e-01 7.19090849e-02 8.56415182e-02
3.79495740e-01 3.53609741e-01 -4.37082797e-01 -7.15608835e-01
-3.04324806e-01 4.59174037e-01 -9.55120996e-02 2.74060935e-01
-1.10102546e+00 1.24594307e+00 6.82780933e+00 6.05967402e-01
-1.27543879e+00 -2.29466036e-01 6.02890909e-01 3.43966842e-01
-3.39257196e-02 -3.40796590e-01 -6.16900563e-01 -1.31803423e-01
5.52948534e-01 2.44700201e-02 5.21231949e-01 8.54701996e-01
6.81366697e-02 -4.04430963e-02 -9.75207746e-01 1.29473746e+00
4.51552659e-01 -1.69675362e+00 4.84025031e-01 -8.40453804e-02
7.22750425e-01 3.93874317e-01 4.10154670e-01 4.78595525e-01
1.12979703e-01 -1.25121045e+00 1.96855888e-01 4.16249901e-01
6.25106692e-01 -6.13805056e-01 5.66924036e-01 1.35658845e-01
-1.07993054e+00 -2.50574321e-01 -6.57566667e-01 -3.85874748e-01
-1.09149657e-01 3.54551375e-01 -7.71542788e-01 4.21531022e-01
7.83287466e-01 7.90037215e-01 -4.12104011e-01 1.18123996e+00
-8.35732147e-02 5.27270675e-01 2.56340712e-01 -2.22188145e-01
5.63781500e-01 -2.08069250e-01 4.42846507e-01 1.18745458e+00
-2.86155224e-01 3.34138945e-02 2.92559326e-01 1.08238292e+00
-1.01650901e-01 1.67093754e-01 -7.92260289e-01 -4.51115400e-01
1.81453496e-01 1.21461821e+00 -5.89596331e-01 -3.84349048e-01
-1.09883952e+00 7.08007693e-01 3.99535388e-01 5.88987708e-01
-5.60682118e-01 -1.98820502e-01 7.88988113e-01 -2.50574529e-01
2.66072005e-01 -3.08582276e-01 -2.81460464e-01 -1.07639694e+00
-2.54757196e-01 -1.04455674e+00 4.46214944e-01 -7.91035891e-01
-1.45961380e+00 6.56788409e-01 -1.20026972e-02 -8.96779418e-01
-2.73960412e-01 -1.02150261e+00 -3.04332554e-01 6.02539599e-01
-1.87953055e+00 -1.27484131e+00 -5.29066086e-01 6.04979217e-01
9.02527869e-01 -6.17858291e-01 9.96285200e-01 5.20234928e-02
-1.07304022e-01 2.73799062e-01 9.01379734e-02 4.11423951e-01
5.95827758e-01 -1.03279531e+00 3.77102643e-01 6.14575326e-01
7.36463130e-01 4.36436296e-01 5.54280818e-01 -1.41664580e-01
-1.48807406e+00 -8.48615229e-01 8.03669870e-01 -4.33151037e-01
5.69941223e-01 -9.61375758e-02 -8.46734285e-01 6.43891692e-01
5.79174280e-01 -1.33796250e-02 6.26554608e-01 -1.14018686e-01
-4.84607846e-01 -2.22024113e-01 -1.00302315e+00 5.31150758e-01
6.10237300e-01 -5.56633234e-01 -9.72179532e-01 2.90030360e-01
5.85868239e-01 -1.13274701e-01 -5.71092725e-01 5.39702475e-01
5.02807915e-01 -1.05024052e+00 1.28532410e+00 -7.82362640e-01
4.79144931e-01 -1.67400435e-01 -9.71371382e-02 -1.02593231e+00
-4.35065776e-01 -1.93697855e-01 -1.75206095e-01 9.05562937e-01
5.17420843e-02 -4.12340671e-01 7.42483079e-01 2.82045454e-01
-1.26296535e-01 -9.56161380e-01 -6.05401933e-01 -4.93297458e-01
2.79131740e-01 -6.41792297e-01 7.94007629e-02 9.75543678e-01
-2.82180250e-01 5.52741468e-01 -2.37759352e-01 -2.83798456e-01
5.90425849e-01 2.69974202e-01 6.65516496e-01 -1.49915457e+00
-7.50109479e-02 -7.70839393e-01 -7.68494844e-01 -1.03644776e+00
2.11629227e-01 -9.05110002e-01 -2.64368862e-01 -2.05516672e+00
2.86765158e-01 3.95393334e-02 -2.49214739e-01 5.21040797e-01
6.28315657e-02 5.23632586e-01 2.66529620e-01 1.26808554e-01
-5.69241107e-01 4.29601461e-01 1.44542325e+00 -5.21977782e-01
-1.92108732e-02 -1.23389669e-01 -7.45726287e-01 7.94698596e-01
8.34827721e-01 1.33337766e-01 -4.62420136e-01 -6.80416346e-01
2.35919788e-01 -4.15285259e-01 3.95383686e-01 -1.01006758e+00
1.84867352e-01 -2.45324507e-01 5.60951173e-01 -4.89604414e-01
5.24185121e-01 -8.81946027e-01 -5.55618405e-01 4.47457045e-01
-4.68218446e-01 -3.34899710e-03 3.94858569e-01 2.76532799e-01
-5.67772985e-01 -9.62965637e-02 9.48589385e-01 -3.47263604e-01
-1.38423753e+00 3.32139075e-01 -6.15181684e-01 1.00945987e-01
1.16809011e+00 -2.96606988e-01 -4.01328593e-01 -3.01569134e-01
-5.89199066e-01 3.73302042e-01 2.76691765e-01 7.06962287e-01
8.94366026e-01 -1.10851741e+00 -6.90849245e-01 7.22212940e-02
3.73701155e-01 -1.11339256e-01 1.24345899e-01 6.31777406e-01
-5.75180173e-01 8.26016486e-01 -4.62238759e-01 -6.19350016e-01
-1.37262428e+00 8.82578492e-01 2.79740632e-01 -9.07095969e-02
-6.42243445e-01 7.90347040e-01 4.02095020e-01 -1.08592905e-01
6.46040857e-01 -1.86641470e-01 -8.28378797e-01 -1.51907235e-01
8.50085139e-01 3.32146794e-01 -2.46858448e-01 -6.99320555e-01
-3.67912650e-01 7.70151854e-01 -2.30650648e-01 1.68163031e-01
1.29572034e+00 8.49324837e-02 -2.38069460e-01 4.69941080e-01
8.66467834e-01 -2.84911811e-01 -7.14417338e-01 -3.90213221e-01
-2.07878463e-02 -1.00225531e-01 1.14443794e-01 -9.72523987e-01
-1.04843664e+00 1.41791296e+00 7.52850175e-01 2.78130174e-01
1.23490989e+00 3.31806719e-01 7.66790688e-01 5.34906089e-01
3.23735088e-01 -1.06779063e+00 3.73272091e-01 6.91269815e-01
9.07796025e-01 -1.54386258e+00 -4.84723598e-02 -1.95391178e-01
-7.80451417e-01 1.34585333e+00 5.30449569e-01 -1.18185744e-01
4.91955698e-01 8.88278484e-02 3.31206679e-01 -4.14251268e-01
-5.82730055e-01 -7.32088625e-01 5.09669065e-01 1.15583754e+00
7.97569275e-01 -2.28307620e-01 -1.12140067e-01 1.17869265e-01
1.39415264e-01 2.59540975e-01 1.20514229e-01 9.08217728e-01
-7.80474007e-01 -1.20547163e+00 -4.14873779e-01 2.80087411e-01
-4.06191498e-01 -7.16692954e-02 -6.51925027e-01 8.59476209e-01
2.56474137e-01 9.64451849e-01 1.18083745e-01 -1.46215692e-01
1.36284575e-01 -2.05213875e-01 7.63441384e-01 -5.44104815e-01
-3.96035671e-01 -3.12657505e-01 -1.02292046e-01 -5.53302407e-01
-8.83859038e-01 -1.38930589e-01 -1.43621230e+00 -2.34437808e-01
8.05976912e-02 2.95839384e-02 6.67499423e-01 1.03974640e+00
2.88515031e-01 4.71128166e-01 1.04855247e-01 -7.53396273e-01
-2.85153806e-01 -7.15637207e-01 -2.59713203e-01 2.07751215e-01
3.25993687e-01 -6.06067777e-01 -3.52502130e-02 2.04221085e-01] | [10.32007122039795, 1.7171820402145386] |
d15aff47-3276-4432-b09a-dd60e27dcf44 | an-analysis-of-quantile-temporal-difference | 2301.04462 | null | https://arxiv.org/abs/2301.04462v2 | https://arxiv.org/pdf/2301.04462v2.pdf | An Analysis of Quantile Temporal-Difference Learning | We analyse quantile temporal-difference learning (QTD), a distributional reinforcement learning algorithm that has proven to be a key component in several successful large-scale applications of reinforcement learning. Despite these empirical successes, a theoretical understanding of QTD has proven elusive until now. Unlike classical TD learning, which can be analysed with standard stochastic approximation tools, QTD updates do not approximate contraction mappings, are highly non-linear, and may have multiple fixed points. The core result of this paper is a proof of convergence to the fixed points of a related family of dynamic programming procedures with probability 1, putting QTD on firm theoretical footing. The proof establishes connections between QTD and non-linear differential inclusions through stochastic approximation theory and non-smooth analysis. | ['Will Dabney', 'Marc G. Bellemare', 'Karl Tuyls', 'Anna Harutyunyan', 'Georg Ostrovski', 'Yunhao Tang', 'Mohammad Gheshlaghi Azar', 'Rémi Munos', 'Mark Rowland'] | 2023-01-11 | null | null | null | null | ['distributional-reinforcement-learning'] | ['methodology'] | [-2.67301977e-01 1.93351492e-01 -2.18940377e-01 1.23585150e-01
-1.06282711e+00 -5.95099926e-01 4.85954642e-01 2.66945362e-01
-5.64465046e-01 1.12603378e+00 -1.59877822e-01 -5.70030391e-01
-5.74499905e-01 -6.31138980e-01 -9.03675675e-01 -1.06946659e+00
-5.41645527e-01 5.11732161e-01 1.32929400e-01 -1.17163159e-01
2.00303361e-01 2.95378357e-01 -1.18423820e+00 -3.37736994e-01
9.29947495e-01 9.06522632e-01 -2.58640379e-01 8.17764938e-01
-1.35439664e-01 6.69242084e-01 -3.86520028e-01 -3.82587999e-01
3.86774927e-01 -5.30802369e-01 -6.31474257e-01 -4.73221868e-01
2.20869362e-01 -2.68482566e-01 -2.60388672e-01 1.27713442e+00
6.69532359e-01 4.58339304e-01 1.02978420e+00 -1.47794378e+00
-5.11327326e-01 5.04743159e-01 -8.59934747e-01 3.71310949e-01
1.05520785e-01 7.04162987e-03 1.00778925e+00 -6.71721399e-01
1.68002382e-01 1.22446585e+00 1.15196967e+00 5.63824773e-01
-1.64607418e+00 -3.11647594e-01 -8.81543010e-02 -1.65645465e-01
-8.71883571e-01 8.63867179e-02 3.05945963e-01 -5.66644669e-01
5.57589173e-01 -7.81935677e-02 7.05956638e-01 8.06007266e-01
6.28644764e-01 9.52151239e-01 1.30064547e+00 -4.82813388e-01
5.31788707e-01 2.60465425e-02 -1.36761814e-01 6.71653986e-01
8.93178210e-02 6.33568406e-01 -1.81002095e-01 -3.02789718e-01
1.05913019e+00 7.73857289e-04 -2.91516744e-02 -8.23968112e-01
-7.00981319e-01 1.30826986e+00 -1.85736105e-01 6.74554259e-02
-5.60323060e-01 4.74226236e-01 6.83905363e-01 7.30078876e-01
6.88911140e-01 -2.86834463e-02 -3.73492777e-01 -5.81818104e-01
-6.68595493e-01 8.62291157e-01 9.04515326e-01 7.26989269e-01
5.71545243e-01 2.35061944e-01 -2.49827281e-01 6.28563344e-01
1.56924143e-01 9.71311271e-01 2.83908397e-01 -1.28260863e+00
3.21869463e-01 -1.71769813e-01 4.12427217e-01 -5.41549385e-01
-3.93688262e-01 -3.23852807e-01 -7.71633744e-01 6.29388690e-01
9.23549533e-01 -4.80934232e-01 -1.03563122e-01 2.01249743e+00
3.70786875e-01 2.08481908e-01 -6.05763011e-02 4.16340709e-01
-3.96711260e-01 6.49443269e-01 6.75205663e-02 -6.24399900e-01
3.62775207e-01 -2.92518705e-01 -5.67082584e-01 4.32601422e-01
5.73615313e-01 -4.85933453e-01 1.19076455e+00 2.77462155e-01
-1.23399913e+00 -2.86218941e-01 -6.99560940e-01 4.58365679e-01
5.16594164e-02 -6.62366211e-01 3.45802277e-01 6.36884272e-01
-1.14420748e+00 1.00249887e+00 -8.42307568e-01 -1.40481368e-01
5.04317105e-01 1.63647369e-01 2.18865693e-01 2.59483904e-02
-1.10742569e+00 8.98776412e-01 1.15915872e-01 -2.06024319e-01
-8.91213238e-01 -1.23380721e+00 -5.68700254e-01 -2.04188779e-01
6.08631551e-01 -4.79161263e-01 1.75341451e+00 -8.16207230e-01
-1.71246624e+00 4.92388487e-01 4.96952049e-02 -8.77611279e-01
1.17954814e+00 -2.49286056e-01 4.12805229e-02 2.33914807e-01
1.42176479e-01 -1.07132174e-01 1.15261054e+00 -7.32691348e-01
-9.09427762e-01 -3.80009204e-01 2.03139093e-02 2.50791699e-01
-3.15651521e-02 -2.79924154e-01 4.99528795e-01 -6.24883354e-01
-5.53386331e-01 -7.91613400e-01 -3.02220374e-01 -7.53941312e-02
2.23592207e-01 -7.28289068e-01 4.70481783e-01 -2.32252017e-01
9.46049869e-01 -2.01570058e+00 1.87410474e-01 1.83808580e-01
4.10647728e-02 -4.10864167e-02 4.48346250e-02 7.61425793e-01
7.75716454e-02 3.00262198e-02 -5.37744999e-01 -1.26218170e-01
6.38907969e-01 2.72182226e-01 -8.05383801e-01 8.64887238e-01
2.33097486e-02 8.90914917e-01 -1.16359425e+00 -3.36132944e-01
3.04717273e-01 5.95137961e-02 -6.18582726e-01 2.15128623e-02
-4.61144298e-01 -2.50254888e-02 -4.79015648e-01 1.28750995e-01
5.21700501e-01 2.29713693e-01 -2.25819468e-01 5.02265155e-01
-3.26835841e-01 -2.30424181e-01 -1.30581403e+00 1.30430639e+00
-4.41669375e-01 5.28837502e-01 2.44728535e-01 -1.65464938e+00
4.29204851e-01 3.36969078e-01 1.02943301e+00 -5.88974476e-01
-9.97632816e-02 3.98753166e-01 -3.56882602e-01 -4.92054135e-01
2.48707175e-01 -9.16288555e-01 -7.35823140e-02 9.17831600e-01
-7.03910924e-03 -3.09718728e-01 1.75919458e-01 1.15320655e-02
9.39257681e-01 3.07880312e-01 1.15945667e-01 -8.19396675e-01
3.10276508e-01 -9.07791927e-02 3.51547539e-01 9.98928189e-01
-4.72926289e-01 9.85199884e-02 9.57234204e-01 -2.10539535e-01
-1.28649151e+00 -1.54418397e+00 -3.94712061e-01 1.13388216e+00
-1.36099219e-01 1.68323025e-01 -7.68561602e-01 -4.21294659e-01
5.51896274e-01 5.76233387e-01 -8.33358943e-01 -1.85674071e-01
-4.44939047e-01 -6.59119666e-01 4.77215916e-01 4.58844185e-01
3.17659020e-01 -9.81651425e-01 -6.39865994e-01 5.33580244e-01
3.48170757e-01 -5.19077063e-01 -6.65240586e-01 3.14513057e-01
-1.02951479e+00 -1.44477355e+00 -1.22785044e+00 -6.35697126e-01
1.02516316e-01 -6.31102622e-02 1.21249890e+00 -4.34170067e-01
1.62044796e-03 1.18240917e+00 1.99688613e-01 -8.15931797e-01
-5.28794110e-01 -7.10791796e-02 4.51583654e-01 -4.95486781e-02
2.96960980e-01 -6.72348917e-01 -3.53159159e-01 1.96132064e-01
-8.95464242e-01 -6.63174808e-01 1.30936339e-01 1.01040196e+00
8.22224081e-01 2.54515409e-01 1.03363752e+00 -7.67736793e-01
1.20674217e+00 -5.39009392e-01 -1.04095924e+00 8.04013014e-02
-7.25256681e-01 1.77183434e-01 6.90368116e-01 -5.05845189e-01
-8.38050127e-01 -4.69367474e-01 3.85046564e-02 -4.07357186e-01
1.69219002e-01 5.70758820e-01 4.32822794e-01 1.38909563e-01
6.44616067e-01 3.37912470e-01 5.57698786e-01 -1.62971452e-01
3.26110989e-01 2.84857363e-01 4.06731933e-01 -1.12559283e+00
9.76505280e-01 3.66337687e-01 2.51043260e-01 -1.01567614e+00
-8.46695960e-01 -2.47411549e-01 -3.76931012e-01 -2.58153349e-01
6.45338655e-01 -5.94228029e-01 -1.24833739e+00 4.75112200e-01
-5.95533431e-01 -1.06197047e+00 -1.01746488e+00 5.16788304e-01
-1.28587306e+00 2.11018965e-01 -6.52983606e-01 -1.48447001e+00
-7.25424886e-02 -5.43049037e-01 5.24885476e-01 3.04834634e-01
9.44974199e-02 -1.54665148e+00 6.62568867e-01 -4.62023079e-01
3.87313187e-01 4.38322455e-01 9.19577658e-01 -2.69534051e-01
5.68470210e-02 2.82500200e-02 2.08504319e-01 5.57097256e-01
-1.75550207e-01 1.27098128e-01 -5.77299595e-01 -3.83132845e-01
4.10514742e-01 -4.61903363e-01 6.14536166e-01 1.05179334e+00
7.93767333e-01 -2.15818599e-01 8.77696201e-02 2.89666891e-01
1.71138644e+00 1.28845870e-01 2.60222465e-01 2.71797001e-01
2.75877893e-01 5.71411490e-01 6.00190341e-01 6.20735943e-01
6.56650290e-02 1.54130310e-01 1.74132869e-01 4.37262386e-01
5.82312226e-01 -3.08011055e-01 6.87406421e-01 6.29950285e-01
5.62367365e-02 5.06549239e-01 -7.53584504e-01 5.33860207e-01
-2.00909615e+00 -1.54429138e+00 -7.90214762e-02 2.46359015e+00
9.94490266e-01 3.36297423e-01 9.89141762e-01 1.73702195e-01
7.43608177e-01 -2.87363678e-01 -1.06521523e+00 -8.33795011e-01
-1.20720610e-01 5.25485992e-01 7.11250186e-01 7.16712773e-01
-8.48288715e-01 4.26210195e-01 7.20665693e+00 9.84415233e-01
-6.90723002e-01 4.62603420e-02 4.77325737e-01 -2.56561860e-02
-4.15299058e-01 -3.96942645e-01 -3.14572424e-01 4.52736050e-01
1.00341332e+00 -7.38756597e-01 5.47857404e-01 1.05775893e+00
4.40572232e-01 -1.88292623e-01 -1.09024906e+00 8.51611972e-01
-5.40506840e-01 -1.25284803e+00 -6.07353747e-01 3.95722389e-01
1.04963946e+00 2.44486574e-02 4.79611635e-01 7.70386040e-01
8.22453082e-01 -1.05261672e+00 6.72761679e-01 4.66303468e-01
6.99848652e-01 -1.42791736e+00 5.94514370e-01 4.58872437e-01
-9.82080340e-01 -3.19335878e-01 -5.95677674e-01 -2.33980730e-01
9.74245071e-02 5.61725438e-01 -7.29711577e-02 3.37475061e-01
4.79992002e-01 7.88898647e-01 2.61293799e-01 1.05575442e+00
5.26909828e-02 8.48894298e-01 -4.30305630e-01 -2.21237063e-01
4.09701377e-01 -5.91992915e-01 4.79615211e-01 8.26979280e-01
2.38207310e-01 -2.85939034e-02 1.87919021e-01 7.87013888e-01
1.40195802e-01 1.17925584e-01 -7.34159768e-01 -1.27320841e-01
1.98502839e-01 7.41727948e-01 -4.11809593e-01 6.65628212e-03
-5.12443244e-01 5.18464983e-01 2.69511074e-01 4.71667379e-01
-8.69910419e-01 -5.17350376e-01 1.02846110e+00 5.17654419e-02
4.56222296e-01 -3.22399288e-01 -1.96201950e-01 -8.41643155e-01
2.39267349e-02 -7.07848012e-01 4.91267055e-01 -5.86110055e-02
-1.74618065e+00 -2.12982431e-01 1.18120719e-04 -9.67087209e-01
-5.17415643e-01 -6.03608966e-01 -6.83275700e-01 9.19581950e-01
-1.27920878e+00 -3.49673629e-01 6.60408318e-01 8.69571567e-01
3.32451671e-01 -1.16902731e-01 6.92508161e-01 4.53639366e-02
-4.33631003e-01 4.68606144e-01 8.77033234e-01 -1.08635165e-01
3.31920803e-01 -1.99103892e+00 -8.01011082e-03 5.90368330e-01
1.57566611e-02 1.81791082e-01 1.00098658e+00 -3.79745424e-01
-1.67477608e+00 -7.63157070e-01 2.86231637e-01 -3.46739173e-01
1.23989761e+00 -4.05182503e-03 -9.43473935e-01 5.09224057e-01
1.02660386e-02 -5.88412173e-02 4.41204458e-01 1.21898122e-01
8.22114795e-02 -1.63657889e-01 -1.14154530e+00 4.82033223e-01
6.65755630e-01 -7.09455311e-01 -5.78022301e-01 2.20809415e-01
5.29896557e-01 -3.01155388e-01 -9.79356050e-01 -1.30401030e-02
5.25365353e-01 -9.11843121e-01 8.27814519e-01 -8.47567737e-01
3.77574056e-01 1.83345616e-01 -8.04664940e-02 -1.43199003e+00
-1.07292514e-02 -1.40053916e+00 -3.30950648e-01 1.01684451e+00
-2.55059171e-03 -8.52280378e-01 7.18808055e-01 4.24353361e-01
-7.13848323e-02 -9.85668004e-01 -1.28708863e+00 -1.29718745e+00
1.11291742e+00 -6.59457624e-01 9.00453106e-02 6.28513932e-01
3.53536755e-01 -3.57953571e-02 -2.41418362e-01 -2.25141972e-01
1.12991750e+00 5.63593544e-02 5.24278343e-01 -1.22432709e+00
-4.43112016e-01 -9.19892967e-01 -6.71366230e-02 -8.65530908e-01
3.97086859e-01 -7.83845484e-01 1.98840514e-01 -1.14722121e+00
-1.41797766e-01 -5.57607651e-01 -5.28065741e-01 -8.83723721e-02
-2.68700439e-02 -9.00866911e-02 -1.81526408e-01 -5.60672395e-02
-2.79585749e-01 9.03201878e-01 1.14577222e+00 8.61325040e-02
-3.96949708e-01 6.23069108e-01 -3.86830449e-01 6.82896912e-01
9.20568883e-01 -5.62087893e-01 -7.37719476e-01 7.16394782e-02
4.62109327e-01 2.59214133e-01 2.16712043e-01 -6.97374940e-01
2.85198186e-02 -4.27459657e-01 -2.46605761e-02 -3.87480319e-01
-3.29180837e-01 -5.01386046e-01 -3.58336359e-01 8.52079868e-01
-6.14886701e-01 3.06015164e-01 1.91020966e-01 9.83105838e-01
7.21065998e-02 -3.36227506e-01 8.96642506e-01 1.37100175e-01
-2.87154496e-01 6.81445062e-01 -7.29046583e-01 9.92143929e-01
1.06754720e+00 6.81911632e-02 3.69577438e-01 -5.02634764e-01
-5.98746836e-01 2.40005314e-01 2.28212282e-01 -3.03988606e-01
4.09660101e-01 -1.45229268e+00 -6.91492498e-01 -2.44990796e-01
-4.94010925e-01 -1.14804998e-01 3.46954286e-01 1.01450598e+00
-2.59728342e-01 2.92456932e-02 -6.43163696e-02 -5.52680433e-01
-5.36198795e-01 6.93174839e-01 5.85840821e-01 -2.47697636e-01
-7.79854417e-01 5.22389829e-01 -7.92213008e-02 -2.00611711e-01
4.50195312e-01 -3.70860010e-01 2.81415731e-01 2.42029235e-01
5.71970880e-01 7.12368786e-01 -3.88283968e-01 2.82720570e-02
8.90567899e-02 5.08741677e-01 1.71681657e-01 -6.97254777e-01
1.60399687e+00 -2.68289715e-01 1.68890521e-01 1.05782771e+00
1.13688576e+00 -2.69398779e-01 -1.87904477e+00 -4.11225528e-01
2.49817148e-01 -1.07105859e-01 -3.48062068e-01 -2.10501656e-01
-7.65335083e-01 1.02590048e+00 5.45465708e-01 7.59204865e-01
7.39532471e-01 -2.52355903e-01 7.12841213e-01 4.34241742e-01
3.43170077e-01 -1.67964244e+00 2.23370358e-01 6.79516733e-01
7.79859841e-01 -1.10590780e+00 -2.30606183e-01 3.93114328e-01
-4.79902476e-01 1.24147618e+00 1.92476228e-01 -6.60357714e-01
9.24866319e-01 3.96493644e-01 -3.33115578e-01 3.07733059e-01
-6.16717696e-01 -9.07254219e-02 -1.82956189e-01 9.08977389e-01
1.17879540e-01 4.12050299e-02 -3.10755581e-01 2.61415273e-01
-9.13765505e-02 5.29194511e-02 5.67763805e-01 8.01141560e-01
-5.61038792e-01 -1.03045690e+00 -9.01803970e-02 3.96195620e-01
-7.48594880e-01 1.56676665e-01 4.22888160e-01 9.97275293e-01
-5.24247229e-01 4.52127397e-01 3.78368683e-02 1.59271345e-01
1.64885327e-01 2.29183495e-01 8.19118500e-01 -2.53882527e-01
-1.82518676e-01 -5.13477102e-02 -6.15909636e-01 -3.90781820e-01
-4.40105230e-01 -1.10469055e+00 -1.22343349e+00 -7.85479248e-01
2.44861051e-01 3.56618673e-01 3.04080725e-01 1.20104313e+00
-2.21892357e-01 1.97225317e-01 8.03088367e-01 -5.80343723e-01
-1.74062264e+00 -6.77530169e-01 -1.20712483e+00 1.15398422e-01
4.79168177e-01 -5.03967881e-01 -4.80648279e-01 -2.12833360e-01] | [4.158573627471924, 2.6037561893463135] |
82f7fa58-1b56-4122-9bd8-ec038197a9d4 | beyond-a-gaussian-denoiser-residual-learning | 1608.03981 | null | http://arxiv.org/abs/1608.03981v1 | http://arxiv.org/pdf/1608.03981v1.pdf | Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising | Discriminative model learning for image denoising has been recently
attracting considerable attentions due to its favorable denoising performance.
In this paper, we take one step forward by investigating the construction of
feed-forward denoising convolutional neural networks (DnCNNs) to embrace the
progress in very deep architecture, learning algorithm, and regularization
method into image denoising. Specifically, residual learning and batch
normalization are utilized to speed up the training process as well as boost
the denoising performance. Different from the existing discriminative denoising
models which usually train a specific model for additive white Gaussian noise
(AWGN) at a certain noise level, our DnCNN model is able to handle Gaussian
denoising with unknown noise level (i.e., blind Gaussian denoising). With the
residual learning strategy, DnCNN implicitly removes the latent clean image in
the hidden layers. This property motivates us to train a single DnCNN model to
tackle with several general image denoising tasks such as Gaussian denoising,
single image super-resolution and JPEG image deblocking. Our extensive
experiments demonstrate that our DnCNN model can not only exhibit high
effectiveness in several general image denoising tasks, but also be efficiently
implemented by benefiting from GPU computing. | ['Yunjin Chen', 'WangMeng Zuo', 'Lei Zhang', 'Kai Zhang', 'Deyu Meng'] | 2016-08-13 | null | null | null | null | ['color-image-denoising', 'image-deblocking', 'jpeg-artifact-correction'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 4.53795284e-01 -3.68793070e-01 3.35029751e-01 -3.97538632e-01
-7.62585402e-01 -2.33082417e-02 4.87093061e-01 -2.60162354e-01
-5.01586735e-01 4.53689843e-01 2.67389506e-01 -8.97173807e-02
6.53247237e-02 -7.44791448e-01 -7.24356115e-01 -1.31460726e+00
2.90277034e-01 -1.43386841e-01 -9.88763124e-02 -3.52791965e-01
1.38059095e-01 3.19568992e-01 -1.36147273e+00 9.26894024e-02
1.08768344e+00 9.75427926e-01 4.03583884e-01 8.26269686e-01
-5.91817088e-02 9.85951483e-01 -4.08818334e-01 -5.42256892e-01
1.31027013e-01 -4.78243321e-01 -2.79792666e-01 2.01727375e-01
2.51041681e-01 -7.00959027e-01 -6.15422249e-01 1.50859118e+00
7.00983465e-01 1.20252810e-01 3.66099328e-01 -7.01790333e-01
-8.82934690e-01 4.12021637e-01 -6.49609864e-01 1.04314975e-01
-2.66415328e-01 1.34341180e-01 6.62360847e-01 -8.49629641e-01
4.08472568e-01 1.38276839e+00 9.34666038e-01 6.26317024e-01
-1.25251889e+00 -5.11392772e-01 1.65163442e-01 2.63031870e-01
-1.07773888e+00 -5.84773064e-01 8.11338723e-01 1.58346724e-02
5.22766471e-01 1.25665501e-01 3.33276778e-01 1.44894207e+00
1.19081393e-01 7.79224336e-01 1.14645946e+00 -2.19653755e-01
2.23619267e-01 -3.32477510e-01 9.50259343e-02 2.16773286e-01
2.27579504e-01 1.66879490e-01 -3.39886248e-01 1.40625298e-01
1.01413453e+00 2.26471454e-01 -3.66599232e-01 -1.04939099e-02
-8.01286697e-01 8.51897359e-01 4.45660532e-01 3.39494437e-01
-5.61724663e-01 4.97978926e-01 7.13924110e-01 4.91300732e-01
7.57795870e-01 -2.30519339e-01 -2.48256698e-01 -1.48327788e-02
-1.01306093e+00 2.21665204e-01 6.26502097e-01 6.46373212e-01
7.40709364e-01 5.35039127e-01 -6.01116978e-02 1.02677214e+00
1.85729563e-01 5.18483818e-01 3.45361412e-01 -1.21466804e+00
4.02734637e-01 -4.21729535e-02 1.65766310e-02 -1.08240473e+00
-7.53834471e-02 -6.09324336e-01 -1.94436681e+00 4.62100297e-01
1.29493654e-01 -9.02193859e-02 -1.11948168e+00 1.56160378e+00
4.87163961e-02 3.80252689e-01 1.66716352e-01 9.46504593e-01
6.69924736e-01 7.11437464e-01 1.05534628e-01 -3.06064695e-01
1.20572305e+00 -9.40894604e-01 -1.14789295e+00 -3.27325255e-01
3.00277978e-01 -9.08194184e-01 6.13586307e-01 8.39943647e-01
-9.76413965e-01 -8.82836461e-01 -9.89141524e-01 -2.71454781e-01
-2.47050170e-02 8.48212987e-02 7.03695118e-01 7.40662456e-01
-1.23059702e+00 8.66966724e-01 -1.03707790e+00 -5.24515510e-02
6.20815754e-01 1.64651603e-01 -5.05713344e-01 -6.78135693e-01
-1.05224752e+00 6.75261319e-01 1.88319877e-01 9.06087995e-01
-1.12167132e+00 -3.15418154e-01 -8.14987183e-01 7.22125694e-02
2.49547720e-01 -7.76077747e-01 9.21699345e-01 -1.14266765e+00
-1.49586856e+00 5.41083395e-01 -1.91067412e-01 -5.27776361e-01
5.71657181e-01 -5.73475540e-01 -4.04599279e-01 1.32527158e-01
-2.00426728e-01 1.32326350e-01 1.50347054e+00 -1.48749697e+00
-3.24542254e-01 -4.19436216e-01 -2.42236033e-01 3.64306904e-02
-4.39742029e-01 6.93921074e-02 -5.22535563e-01 -1.07336438e+00
4.32969451e-01 -3.68986249e-01 -6.61615551e-01 -4.84882481e-02
-2.47315526e-01 1.14094488e-01 8.18478227e-01 -1.14763296e+00
1.11388087e+00 -2.36312485e+00 3.99161458e-01 4.43645418e-02
3.75834733e-01 4.31232840e-01 -3.89785588e-01 2.02597931e-01
-1.88087627e-01 -1.42816454e-02 -5.65347850e-01 -7.86971033e-01
-1.01995818e-01 6.50475919e-01 -3.06563556e-01 5.94083488e-01
2.04508841e-01 7.27278471e-01 -6.42977536e-01 -2.24742219e-01
3.54765892e-01 1.04109633e+00 -5.42720854e-01 4.55752850e-01
1.94041729e-01 6.81038678e-01 -3.15027058e-01 4.53773588e-01
1.33113980e+00 1.69603869e-01 -6.11504875e-02 -4.23677266e-01
5.31025371e-03 -2.89160699e-01 -1.42616582e+00 1.64959002e+00
-5.41321039e-01 3.36413413e-01 8.99685025e-01 -1.48431706e+00
9.71879423e-01 2.59033293e-01 2.15146452e-01 -7.38993287e-01
1.36809528e-01 3.70842040e-01 -3.00610632e-01 -6.30082190e-01
3.39781731e-01 -3.82228881e-01 3.18752378e-01 1.13929650e-02
2.60190457e-01 -5.50177693e-02 2.92246183e-03 1.29011810e-01
9.90715146e-01 8.93876031e-02 -5.21343313e-02 -5.63457161e-02
7.50529826e-01 -5.28250754e-01 6.20694578e-01 1.06201911e+00
-2.82163788e-02 9.74633873e-01 2.82997638e-01 -4.04042244e-01
-1.12189770e+00 -8.00074399e-01 3.33210710e-03 1.15418327e+00
9.25990045e-02 -1.95127144e-01 -9.45388258e-01 -2.61840850e-01
-3.78029525e-01 4.69358802e-01 -4.39338177e-01 -3.06207299e-01
-9.12038147e-01 -1.16771924e+00 4.44253117e-01 5.22497654e-01
9.06002522e-01 -1.00597370e+00 2.01067440e-02 3.33335966e-01
-2.09561452e-01 -9.95063841e-01 -2.69890338e-01 6.45957708e-01
-1.17409098e+00 -7.61001527e-01 -1.01844776e+00 -9.11224246e-01
4.64902729e-01 4.33336824e-01 1.04161203e+00 2.07627177e-01
-9.50163901e-02 3.02236855e-01 -3.51412088e-01 7.25901872e-02
-5.68673253e-01 -1.59751803e-01 -1.11039482e-01 2.24184602e-01
1.33971915e-01 -9.33056414e-01 -7.02860177e-01 -4.56302874e-02
-1.43417680e+00 -1.22783072e-01 7.36929536e-01 1.20939934e+00
6.06631994e-01 5.60338318e-01 3.71948153e-01 -8.97289097e-01
7.68966556e-01 -3.86820138e-01 -4.51611549e-01 2.94681434e-02
-4.95901912e-01 6.03180118e-02 8.18701625e-01 -3.30830991e-01
-1.46283376e+00 -1.11391194e-01 -8.57498407e-01 -4.76526648e-01
-2.45342135e-01 5.42104781e-01 -3.79245490e-01 -9.75503027e-02
4.43040669e-01 5.98864913e-01 2.30317771e-01 -1.04827642e+00
4.66376603e-01 5.49120843e-01 9.57013786e-01 -3.77406210e-01
1.03178942e+00 5.92937708e-01 -4.10832204e-02 -7.35555887e-01
-7.50844002e-01 -4.14714903e-01 -5.30657470e-01 9.72042009e-02
9.50094342e-01 -1.28366399e+00 -6.59145653e-01 1.02885401e+00
-1.24579501e+00 -1.90247163e-01 -1.20579693e-02 2.55735219e-01
-3.56132478e-01 8.27489734e-01 -1.06657434e+00 -8.66082191e-01
-5.58437526e-01 -1.18991268e+00 9.35109496e-01 9.17477682e-02
4.43092227e-01 -1.14303148e+00 -2.19401762e-01 1.91236362e-01
8.97898614e-01 6.18634075e-02 7.63013065e-01 -3.24157447e-01
-4.62533414e-01 -4.07767534e-01 -3.62937629e-01 1.32425678e+00
-1.05246855e-02 -4.67053562e-01 -1.03433478e+00 -5.63115597e-01
8.05379689e-01 -1.81581035e-01 1.45456457e+00 6.39604390e-01
1.41129971e+00 -2.44122043e-01 1.98370129e-01 1.18842840e+00
1.67632866e+00 -3.47506553e-01 1.13814640e+00 4.15451616e-01
9.53454792e-01 2.62014955e-01 1.27887964e-01 3.71567160e-01
-6.35802820e-02 2.76001960e-01 8.35197985e-01 -4.25162882e-01
-2.79105544e-01 1.95748191e-02 3.23018044e-01 1.06008434e+00
-4.86242354e-01 -1.74638510e-01 -3.77635866e-01 1.77897379e-01
-1.78689301e+00 -7.93106437e-01 -3.66311729e-01 1.89204931e+00
7.81020284e-01 3.63668986e-02 -4.71732557e-01 1.45383567e-01
7.53767788e-01 4.77259338e-01 -4.59044278e-01 -2.16775402e-01
-6.22216046e-01 4.65633124e-01 5.45903027e-01 4.50327456e-01
-1.34251010e+00 8.43121052e-01 5.66356659e+00 1.22451377e+00
-9.51614738e-01 1.76528215e-01 8.28583837e-01 4.54274416e-01
-1.44182667e-01 -2.30423748e-01 -5.29489398e-01 3.99008244e-01
7.49327898e-01 3.73106152e-01 8.01557004e-01 7.29436517e-01
4.37000513e-01 7.88791403e-02 -5.37093937e-01 1.22584867e+00
-2.94707585e-02 -1.04199266e+00 1.65113330e-01 -1.84189543e-01
6.50579154e-01 -6.62181377e-02 2.31987715e-01 2.97022551e-01
2.68352181e-01 -1.24707949e+00 4.18354332e-01 6.00880086e-01
6.00851655e-01 -8.37279916e-01 1.13746977e+00 4.40205306e-01
-8.42189670e-01 -1.03514709e-01 -7.82678545e-01 -2.48592403e-02
2.84805566e-01 9.86698687e-01 2.43533611e-01 6.70602739e-01
1.03689671e+00 1.01976967e+00 -3.12887758e-01 9.16003883e-01
-4.50059295e-01 8.52150142e-01 -2.30467692e-02 7.16985106e-01
2.87568569e-01 -7.77261198e-01 4.76425320e-01 1.39262521e+00
4.81370062e-01 -4.12443466e-02 -1.70657724e-01 7.01398253e-01
-1.91440493e-01 -1.42759100e-01 -1.99584767e-01 2.18191177e-01
-1.62116200e-01 1.37417471e+00 -4.13888693e-01 -2.94068098e-01
-3.93895119e-01 1.42867494e+00 6.74903840e-02 8.12235057e-01
-6.35461569e-01 -2.62184054e-01 8.01552296e-01 -2.07523599e-01
6.58298671e-01 -3.66718471e-01 -4.87220496e-01 -1.24291384e+00
-7.31496187e-03 -1.09817612e+00 5.27343750e-02 -5.28018355e-01
-1.52171957e+00 6.79100871e-01 -5.64382255e-01 -1.04158401e+00
1.71572328e-01 -7.02362597e-01 -6.66614354e-01 1.16806614e+00
-1.90292883e+00 -1.13728356e+00 -5.33934474e-01 7.13010252e-01
4.72110450e-01 8.39344598e-03 6.78256214e-01 5.77897012e-01
-7.04021275e-01 3.16936851e-01 5.95575511e-01 8.25398192e-02
8.80947828e-01 -1.17880964e+00 3.55342001e-01 1.12779713e+00
-3.03673387e-01 8.12426448e-01 1.00415599e+00 -5.56245089e-01
-1.50742519e+00 -1.14363515e+00 4.46109474e-01 1.62730306e-01
6.71835065e-01 -3.52627635e-01 -1.30719221e+00 4.36183780e-01
3.63288820e-01 3.42464119e-01 2.35300049e-01 -2.06601724e-01
-4.96948898e-01 -3.32223952e-01 -1.05487502e+00 4.47462857e-01
9.92242396e-01 -5.23243189e-01 -2.61127174e-01 2.22687274e-01
6.07386887e-01 -4.00280803e-01 -6.77917123e-01 2.81026810e-01
1.40720278e-01 -1.06360829e+00 1.37538290e+00 -5.38892627e-01
6.53245151e-01 -2.01187149e-01 -2.54380435e-01 -1.27376592e+00
-4.25172925e-01 -7.38765657e-01 -2.83022106e-01 1.39884448e+00
-2.66319305e-01 -5.05266726e-01 5.17048061e-01 2.79778481e-01
-2.55654216e-01 -2.99360543e-01 -1.04017830e+00 -5.08889258e-01
1.99639648e-01 -5.28283417e-01 1.61346033e-01 5.23732364e-01
-9.68210757e-01 3.84565070e-02 -1.16361833e+00 3.55274796e-01
1.10514927e+00 -3.74111027e-01 5.53208292e-01 -1.02399588e+00
-2.52632171e-01 -2.56486416e-01 -2.14631319e-01 -1.51897681e+00
1.45843238e-01 -5.84747553e-01 3.28869343e-01 -1.55001140e+00
2.24231586e-01 1.80522017e-02 -4.25006509e-01 1.69189170e-01
-4.79301035e-01 4.75597084e-01 -1.34232879e-01 1.58781603e-01
-5.62737346e-01 7.30461419e-01 1.24849439e+00 -2.09478229e-01
1.42759591e-01 2.17713207e-01 -9.56509113e-01 8.33230138e-01
5.87902188e-01 -3.95454377e-01 -9.02019069e-03 -8.34200323e-01
7.74576515e-02 -1.12772189e-01 6.12723589e-01 -8.50524008e-01
3.90565157e-01 2.19876722e-01 4.94070232e-01 -3.17177266e-01
2.46682301e-01 -1.00014985e+00 -2.29389239e-02 4.00480241e-01
-1.46084890e-01 -3.39639395e-01 -3.20337005e-02 9.45860445e-01
-6.68410242e-01 -3.74487430e-01 9.78598714e-01 -3.43005985e-01
-8.13207924e-01 2.55879402e-01 -4.27948356e-01 -2.90426821e-01
3.13159108e-01 -2.37363622e-01 -1.28344491e-01 -4.61323917e-01
-8.06836903e-01 -6.95946291e-02 1.50652364e-01 9.09628794e-02
7.17229366e-01 -1.10939658e+00 -1.00209129e+00 2.95288980e-01
-3.88599157e-01 1.66810751e-01 7.21881628e-01 9.54992115e-01
-5.85284710e-01 -1.46192431e-01 -9.50981863e-03 -5.09390950e-01
-1.10078335e+00 6.50684118e-01 3.50308061e-01 -3.28802556e-01
-8.93209517e-01 1.03156579e+00 2.45688885e-01 -3.38816017e-01
4.37320620e-01 -2.59100087e-02 -1.32631212e-01 -1.09718047e-01
7.79977083e-01 5.00025392e-01 1.39898822e-01 -4.88917112e-01
1.17437892e-01 5.43290496e-01 -1.31310865e-01 3.98005068e-01
1.85103321e+00 -4.75406617e-01 -6.13823593e-01 -7.76763484e-02
1.43814898e+00 -1.56448483e-01 -1.50924194e+00 -3.58211040e-01
-6.70576170e-02 -3.61395687e-01 5.83297670e-01 -4.58160669e-01
-1.49478030e+00 1.00930274e+00 8.64257157e-01 1.30548656e-01
1.56497717e+00 -3.60559434e-01 9.33133781e-01 2.14765310e-01
3.08507476e-02 -1.09361053e+00 7.06557631e-02 5.50118685e-01
9.82915282e-01 -1.37914145e+00 6.58491673e-03 -3.62731040e-01
-4.40616995e-01 1.38013673e+00 2.23501667e-01 -4.49344784e-01
5.71498215e-01 4.99906838e-01 1.75476551e-01 1.50358183e-02
-2.46973559e-01 -1.38380900e-01 -1.04700923e-01 6.61292195e-01
2.82064497e-01 -2.41556153e-01 -9.97801572e-02 7.62247682e-01
3.33400190e-01 6.66630790e-02 2.95190722e-01 6.41122282e-01
-5.36807835e-01 -1.09048533e+00 -5.56072533e-01 2.61342317e-01
-5.98268211e-01 -4.28423494e-01 3.30379099e-01 4.09464747e-01
1.46841213e-01 1.00367069e+00 -2.52558559e-01 -1.38373986e-01
3.00550967e-01 -3.09469342e-01 1.87204063e-01 -2.26468876e-01
-6.80628002e-01 4.84986782e-01 -3.65628034e-01 -5.66717923e-01
-5.04815102e-01 -2.84446359e-01 -6.98973298e-01 -3.79641086e-01
-2.52803892e-01 -9.43790153e-02 6.22955859e-01 8.52166772e-01
-5.62580004e-02 7.83394575e-01 5.56663692e-01 -1.17493415e+00
-9.59345460e-01 -1.11969960e+00 -8.40318680e-01 5.60999095e-01
6.11821413e-01 -1.60825238e-01 -7.82229424e-01 2.15341359e-01] | [11.454205513000488, -2.368751287460327] |
1e4e1aa2-11a9-4ab7-97d4-796e125b11b8 | red-deep-recurrent-neural-networks-for-sleep | 2005.07795 | null | https://arxiv.org/abs/2005.07795v2 | https://arxiv.org/pdf/2005.07795v2.pdf | RED: Deep Recurrent Neural Networks for Sleep EEG Event Detection | The brain electrical activity presents several short events during sleep that can be observed as distinctive micro-structures in the electroencephalogram (EEG), such as sleep spindles and K-complexes. These events have been associated with biological processes and neurological disorders, making them a research topic in sleep medicine. However, manual detection limits their study because it is time-consuming and affected by significant inter-expert variability, motivating automatic approaches. We propose a deep learning approach based on convolutional and recurrent neural networks for sleep EEG event detection called Recurrent Event Detector (RED). RED uses one of two input representations: a) the time-domain EEG signal, or b) a complex spectrogram of the signal obtained with the Continuous Wavelet Transform (CWT). Unlike previous approaches, a fixed time window is avoided and temporal context is integrated to better emulate the visual criteria of experts. When evaluated on the MASS dataset, our detectors outperform the state of the art in both sleep spindle and K-complex detection with a mean F1-score of at least 80.9% and 82.6%, respectively. Although the CWT-domain model obtained a similar performance than its time-domain counterpart, the former allows in principle a more interpretable input representation due to the use of a spectrogram. The proposed approach is event-agnostic and can be used directly to detect other types of sleep events. | ['Pablo A. Estévez', 'Nicolás I. Tapia'] | 2020-05-15 | null | null | null | null | ['k-complex-detection', 'spindle-detection', 'sleep-micro-event-detection'] | ['medical', 'medical', 'medical'] | [ 2.03901365e-01 -2.51599461e-01 2.29042277e-01 -1.37654379e-01
-3.11851025e-01 -5.38127244e-01 4.38677579e-01 3.38501811e-01
-6.32011890e-01 6.97650671e-01 -5.94154894e-02 -2.25211903e-01
-2.09185839e-01 -4.14033085e-01 -1.88729808e-01 -8.03755999e-01
-2.66715705e-01 -2.26630226e-01 2.40423143e-01 2.00486910e-02
1.93589568e-01 3.93872023e-01 -1.63779449e+00 2.02041104e-01
1.01184905e+00 1.20888543e+00 5.46205580e-01 3.88037026e-01
-3.34599875e-02 4.75745618e-01 -1.23189449e+00 -2.40544602e-02
-1.65300563e-01 -8.15931797e-01 -3.03502560e-01 1.29807949e-01
-1.46064028e-01 4.05973285e-01 -6.14671446e-02 1.14139009e+00
7.91177452e-01 8.13508704e-02 4.27810818e-01 -1.05119169e+00
-3.43263477e-01 1.60554707e-01 -4.71247375e-01 9.82604086e-01
6.08926713e-01 1.38675138e-01 6.08342767e-01 -8.54329586e-01
4.15669054e-01 5.01495481e-01 5.56479752e-01 2.54293948e-01
-1.37975633e+00 -7.82799423e-01 -1.91195890e-01 8.19579005e-01
-1.50022650e+00 -3.68134916e-01 8.57150137e-01 -2.98007160e-01
1.21777284e+00 1.08244367e-01 1.12511230e+00 1.24136436e+00
8.26081812e-01 6.49481714e-01 1.47486269e+00 -2.28250608e-01
5.84229767e-01 1.03225999e-01 8.43694508e-02 2.25955278e-01
5.13274483e-02 -2.87107192e-02 -6.47207201e-01 1.72545835e-01
4.83501285e-01 1.99309409e-01 -6.34224892e-01 -4.14841548e-02
-1.04388750e+00 5.41625202e-01 4.62002993e-01 6.42958403e-01
-7.69809544e-01 -1.83900118e-01 4.40420121e-01 3.31032187e-01
5.06773174e-01 5.42485714e-01 -7.41942301e-02 -4.40873563e-01
-1.60781252e+00 -1.58161491e-01 5.09825885e-01 4.12365168e-01
4.91789401e-01 2.14881867e-01 -3.08454365e-01 6.36859059e-01
-8.53446424e-02 3.47736478e-01 9.28394198e-01 -3.80620509e-01
7.54512548e-02 8.06018889e-01 8.40136781e-02 -7.96448410e-01
-1.13321364e+00 -7.83410192e-01 -7.43458748e-01 3.04429144e-01
4.09139246e-01 4.22158316e-02 -8.53454530e-01 1.36099052e+00
-2.06296489e-01 3.10595036e-01 -1.11150712e-01 9.65868354e-01
1.02745366e+00 5.63708961e-01 -5.88489883e-02 -4.80244786e-01
1.80632186e+00 -4.30443197e-01 -1.09710956e+00 -4.05745357e-01
4.88504358e-02 -5.83696723e-01 1.01946867e+00 5.67459643e-01
-1.03652430e+00 -4.08399224e-01 -1.43142486e+00 2.69914806e-01
-5.44485927e-01 5.64126730e-01 9.40874815e-02 6.95296407e-01
-1.12371898e+00 4.45263803e-01 -1.18750978e+00 -5.73507309e-01
4.95974720e-01 4.95560348e-01 -3.42941403e-01 5.61293662e-01
-9.81185615e-01 1.16590309e+00 1.63640380e-01 2.84970999e-02
-6.80153489e-01 -4.01282072e-01 -7.61702418e-01 2.87832409e-01
1.18627317e-01 -2.31470302e-01 9.86455858e-01 -8.34154844e-01
-1.38600099e+00 1.03111732e+00 -4.50582623e-01 -6.93516970e-01
1.35599047e-01 -2.22928599e-02 -9.39568639e-01 4.42592829e-01
-6.72636181e-02 1.87295198e-01 1.09277427e+00 -5.56488931e-01
-4.80538845e-01 -3.28426629e-01 -2.09922064e-03 -1.75920412e-01
-1.16165787e-01 2.13934600e-01 -2.72808084e-03 -7.98701286e-01
-1.40920252e-01 -7.51891851e-01 2.64898598e-01 -2.48962536e-01
-1.60055324e-01 -3.49516094e-01 6.61105871e-01 -5.07283211e-01
1.44563651e+00 -2.45434785e+00 5.97743392e-02 -9.83400345e-02
4.50162768e-01 2.39369586e-01 7.08627328e-02 3.51942688e-01
-5.33420444e-01 -2.91822761e-01 -2.73085684e-01 -2.47348651e-01
-1.68278620e-01 -1.51822135e-01 -1.87738150e-01 8.04087877e-01
3.93518537e-01 7.87376881e-01 -1.04925156e+00 -5.45584857e-02
3.04940313e-01 5.50505817e-01 1.58415690e-01 8.25710893e-02
3.25465560e-01 3.69174868e-01 2.26617709e-01 4.33727801e-01
4.03978258e-01 -3.25228542e-01 -1.08281747e-01 -2.24518448e-01
-3.69195700e-01 6.52812541e-01 -9.65883076e-01 1.65444720e+00
-4.51692134e-01 1.16416252e+00 -2.14629859e-01 -9.01139736e-01
9.40817714e-01 5.82460463e-01 1.77661255e-01 -1.05785608e+00
2.07708508e-01 1.17072895e-01 3.35731506e-01 -5.72512507e-01
5.81722595e-02 -1.58605903e-01 2.82356113e-01 4.85024601e-01
2.59383947e-01 4.02982794e-02 4.40612346e-01 -8.10727924e-02
1.30020177e+00 -1.85130686e-01 7.39025414e-01 -4.89162773e-01
3.78767878e-01 -5.18954396e-01 5.19073784e-01 4.60493565e-01
-1.06917538e-01 6.10863924e-01 5.85745811e-01 -4.96703386e-01
-4.13232625e-01 -1.17383635e+00 -3.51565093e-01 8.23901415e-01
3.05822968e-01 -7.14833677e-01 -6.40171945e-01 -5.29802501e-01
-4.81660992e-01 8.53072882e-01 -8.42963159e-01 -4.93718296e-01
-2.23360255e-01 -9.55726504e-01 3.13143253e-01 6.98740244e-01
3.12350899e-01 -1.42744839e+00 -1.56157160e+00 3.47926319e-01
-1.09488674e-01 -1.06180704e+00 -2.78498441e-01 8.43815804e-01
-6.49876714e-01 -1.05310822e+00 -7.22937882e-01 -3.87827218e-01
4.44163322e-01 6.41565993e-02 1.06333125e+00 -3.37136358e-01
-6.27866924e-01 3.13857049e-01 -4.96086121e-01 -6.39073253e-01
3.03022787e-02 -1.53814390e-01 9.41180736e-02 1.91523567e-01
7.41414487e-01 -9.46285248e-01 -9.56174254e-01 2.84749299e-01
-8.85618985e-01 -1.40971810e-01 7.83588111e-01 7.05825031e-01
3.77568394e-01 6.44524619e-02 8.43110979e-01 -2.00742945e-01
9.07346249e-01 -3.49033982e-01 -5.34050226e-01 8.00748989e-02
-6.23053849e-01 -1.75308362e-01 8.67372453e-01 -5.41140199e-01
-6.79373264e-01 -1.08083732e-01 9.85637829e-02 -4.26393062e-01
-3.29596162e-01 3.24427098e-01 1.38551384e-01 -2.76579391e-02
8.11279953e-01 5.70692420e-01 -2.13314876e-01 -1.73185587e-01
-3.94222915e-01 5.47826469e-01 5.65800250e-01 2.16233581e-01
3.68269116e-01 3.59477580e-01 -1.56868532e-01 -9.75832045e-01
-5.87947726e-01 -5.55205345e-01 -4.19082552e-01 -2.24863738e-01
1.05845535e+00 -7.99934447e-01 -6.08775675e-01 2.58550227e-01
-1.10555136e+00 -1.76996335e-01 -2.53290981e-01 7.00389802e-01
-3.47946346e-01 1.82584554e-01 -3.56610894e-01 -8.86768222e-01
-5.20216405e-01 -1.16683900e+00 1.00355232e+00 4.46022809e-01
-6.23885572e-01 -7.60565519e-01 6.37162179e-02 -3.79850119e-01
3.64132971e-01 1.31729931e-01 7.56315887e-01 -7.63077378e-01
-2.78623588e-02 -1.55012310e-01 2.77544130e-02 1.42708838e-01
4.60508406e-01 -2.50073940e-01 -1.30687392e+00 -2.69947320e-01
5.02626777e-01 2.69387886e-02 7.01166868e-01 4.39567685e-01
1.09356797e+00 2.22499460e-01 -2.52745152e-01 5.22739947e-01
1.10408127e+00 6.05672538e-01 8.47203255e-01 3.66251767e-01
-8.36723447e-02 3.42828691e-01 1.07876174e-01 5.57013690e-01
1.15971155e-01 6.26147270e-01 4.17627454e-01 -2.54439741e-01
-6.89141750e-02 3.79753977e-01 5.67542255e-01 4.07557845e-01
-2.67375827e-01 -3.82532477e-02 -9.30575252e-01 6.78884387e-01
-1.53835630e+00 -1.04271257e+00 -1.07662454e-01 2.21610928e+00
6.03678882e-01 3.06096554e-01 2.86065608e-01 5.46363294e-01
5.81685662e-01 4.34774272e-02 -6.25671506e-01 -3.72141272e-01
-2.75414288e-01 8.29632819e-01 5.40786535e-02 -3.48863095e-01
-8.29798996e-01 3.27567905e-01 6.31103611e+00 5.97971976e-01
-1.56041610e+00 2.54333556e-01 3.91643792e-02 -5.54781735e-01
2.81327397e-01 -5.64875007e-01 -3.81365657e-01 9.16506648e-01
1.20333087e+00 -3.62383485e-01 5.01892090e-01 3.99173379e-01
6.07295215e-01 -5.12167454e-01 -1.11598694e+00 1.51607502e+00
2.89461166e-02 -1.23539126e+00 -5.52185118e-01 -8.25102627e-02
2.35174701e-01 8.53592306e-02 6.25990704e-02 1.67283565e-01
-4.74903435e-01 -1.14704537e+00 7.94503748e-01 5.18874824e-01
9.75413501e-01 -7.82721043e-01 8.00003111e-01 1.20796911e-01
-1.39208710e+00 -1.94527075e-01 -1.38294682e-01 -9.15711001e-02
5.44410907e-02 7.20492661e-01 -6.79229438e-01 4.37520087e-01
1.11828017e+00 9.33867753e-01 -8.79781187e-01 1.40303624e+00
-4.82785910e-01 8.60900760e-01 -2.59611368e-01 -4.68625687e-02
5.03979586e-02 -7.26121888e-02 6.84450805e-01 1.35140872e+00
4.54890311e-01 -1.28756925e-01 -2.92394489e-01 1.11295235e+00
1.78372979e-01 -2.32216045e-01 -4.12395686e-01 -6.62428439e-02
3.15187067e-01 1.66009295e+00 -1.36240327e+00 -4.17681597e-02
-5.29970348e-01 9.76598024e-01 1.31228310e-03 4.49154556e-01
-7.68118501e-01 -7.87442267e-01 3.31799120e-01 -2.79286727e-02
3.63060564e-01 -5.62095363e-03 -2.78359830e-01 -1.04789007e+00
2.18461961e-01 -6.98170781e-01 1.90376714e-01 -8.84195030e-01
-1.13550627e+00 1.08975029e+00 -8.01874027e-02 -1.55526066e+00
-3.32128048e-01 -5.57488501e-01 -1.13583112e+00 7.40391731e-01
-1.39250767e+00 -4.61113036e-01 -4.58717614e-01 7.93760002e-01
5.55588484e-01 -4.72433902e-02 9.18894529e-01 2.80834138e-01
-6.26394510e-01 3.83300900e-01 -1.15555398e-01 -1.53688893e-01
6.45630419e-01 -1.54950798e+00 1.76024333e-01 9.54653621e-01
3.48723829e-01 4.81029063e-01 8.39099586e-01 -2.53943294e-01
-9.32301342e-01 -8.08281064e-01 5.80140591e-01 -1.15856469e-01
6.26445830e-01 -6.15318060e-01 -9.44668114e-01 2.63042927e-01
4.33840543e-01 -1.38330106e-02 8.30188453e-01 -1.05528094e-01
1.72258809e-01 -2.56675392e-01 -1.00799978e+00 3.18050921e-01
6.77614808e-01 -6.87962770e-01 -1.01884270e+00 2.54463982e-02
7.44203851e-02 -1.77350506e-01 -5.89423060e-01 1.44945085e-01
3.98078024e-01 -1.24958777e+00 4.83520061e-01 1.07790850e-01
1.15203582e-01 -5.14934838e-01 4.36847180e-01 -1.67448032e+00
-4.29872781e-01 -7.25583732e-01 -1.72644719e-01 6.74918175e-01
2.49217764e-01 -7.25035310e-01 3.29624861e-01 -1.69438526e-01
-3.61500502e-01 -8.72193575e-01 -1.13416970e+00 -8.99029434e-01
-5.72233081e-01 -5.03300011e-01 3.68177712e-01 5.09812236e-01
4.08673346e-01 5.51082194e-01 1.07504807e-01 8.41645077e-02
1.08521558e-01 2.68215239e-01 -1.46902371e-02 -1.32577956e+00
1.08729891e-01 -6.20968819e-01 -7.11184680e-01 -5.19425988e-01
5.72622009e-02 -8.40338588e-01 -7.71832280e-03 -1.49889779e+00
-4.25018780e-02 3.22533071e-01 -6.88878953e-01 5.41797101e-01
-1.07743228e-02 4.34130818e-01 -1.84862316e-02 -1.54556900e-01
-4.82674628e-01 4.80034083e-01 7.16166794e-01 -5.90586737e-02
-4.60512757e-01 9.10670683e-02 -2.77227819e-01 8.82837594e-01
7.77889132e-01 -4.79480147e-01 -4.61860329e-01 7.31027573e-02
3.72522891e-01 -1.44641660e-03 4.84421551e-01 -1.31444359e+00
5.64208686e-01 5.29626191e-01 7.25598276e-01 -6.39093757e-01
4.33430791e-01 -7.02847421e-01 1.50060087e-01 4.92477864e-01
1.22106321e-01 3.83331507e-01 4.47512120e-01 5.38107455e-01
-2.45444760e-01 -1.13898315e-01 7.74510324e-01 7.20333727e-03
-6.54782951e-01 -1.83671072e-01 -1.00645506e+00 -1.71758294e-01
1.00385845e+00 -6.79914296e-01 -2.98914164e-01 -2.80611038e-01
-8.14064801e-01 -1.14960030e-01 2.47909576e-01 3.08973163e-01
8.13816607e-01 -9.71393883e-01 -3.76436174e-01 4.92624879e-01
3.61169368e-01 -5.01636207e-01 1.96631938e-01 1.53345108e+00
-7.48975426e-02 3.13597828e-01 -6.16532505e-01 -9.05250430e-01
-1.05052233e+00 4.65804785e-01 4.12216485e-01 -8.56924653e-02
-9.73221719e-01 6.25752270e-01 3.07442099e-01 5.47820389e-01
2.99130470e-01 -6.06201589e-01 -6.41302228e-01 4.08800930e-01
9.52194870e-01 4.55225855e-01 6.38173044e-01 -3.82609725e-01
-7.68351138e-01 3.81097943e-01 1.16672158e-01 1.84919670e-01
1.43916059e+00 4.08366136e-02 -1.10836469e-01 8.10488522e-01
7.79319108e-01 -7.19886599e-03 -1.07182252e+00 1.97248697e-01
1.85519516e-01 4.19693626e-02 1.77843451e-01 -1.01271582e+00
-1.07559443e+00 9.26077724e-01 1.01191401e+00 6.95272982e-01
1.68638909e+00 -2.43528131e-02 5.87933242e-01 2.39749968e-01
3.46196085e-01 -8.57635796e-01 3.67735024e-03 7.91087374e-02
9.54668760e-01 -9.34357345e-01 -6.96597621e-02 1.10981055e-01
-4.31290358e-01 1.42280638e+00 2.43813813e-01 -2.21667528e-01
4.55984235e-01 3.66540968e-01 -4.16357480e-02 -4.86670077e-01
-7.35818326e-01 -3.95232707e-01 6.44095182e-01 5.07922530e-01
4.36398894e-01 -6.89324811e-02 -4.33251560e-01 9.02990222e-01
-3.49726915e-01 1.27308056e-01 5.73823273e-01 7.27790892e-01
-2.75769144e-01 -6.34205103e-01 -4.34435278e-01 7.39044726e-01
-6.82963431e-01 -4.84491363e-02 -2.08300993e-01 6.78297639e-01
2.55583078e-01 1.21352077e+00 3.49329442e-01 -3.68764758e-01
4.16853130e-01 2.43132517e-01 5.60004711e-01 -7.26559103e-01
-7.38140941e-01 1.38604730e-01 -3.43951255e-01 -7.73670495e-01
-6.46385908e-01 -4.99628007e-01 -1.24524724e+00 2.80740410e-01
-2.48183712e-01 1.50221065e-01 5.05976021e-01 1.10919905e+00
4.03258324e-01 8.96132231e-01 2.71641254e-01 -9.03322875e-01
1.46071479e-01 -1.22786248e+00 -7.84694970e-01 2.59000957e-01
6.87866628e-01 -1.00841081e+00 -4.20160294e-01 1.81978524e-01] | [13.47847843170166, 3.5158638954162598] |
fbf872f3-2252-42f0-8ea7-eda3a51428f0 | simcvd-simple-contrastive-voxel-wise | 2108.06227 | null | https://arxiv.org/abs/2108.06227v4 | https://arxiv.org/pdf/2108.06227v4.pdf | SimCVD: Simple Contrastive Voxel-Wise Representation Distillation for Semi-Supervised Medical Image Segmentation | Automated segmentation in medical image analysis is a challenging task that requires a large amount of manually labeled data. However, most existing learning-based approaches usually suffer from limited manually annotated medical data, which poses a major practical problem for accurate and robust medical image segmentation. In addition, most existing semi-supervised approaches are usually not robust compared with the supervised counterparts, and also lack explicit modeling of geometric structure and semantic information, both of which limit the segmentation accuracy. In this work, we present SimCVD, a simple contrastive distillation framework that significantly advances state-of-the-art voxel-wise representation learning. We first describe an unsupervised training strategy, which takes two views of an input volume and predicts their signed distance maps of object boundaries in a contrastive objective, with only two independent dropout as mask. This simple approach works surprisingly well, performing on the same level as previous fully supervised methods with much less labeled data. We hypothesize that dropout can be viewed as a minimal form of data augmentation and makes the network robust to representation collapse. Then, we propose to perform structural distillation by distilling pair-wise similarities. We evaluate SimCVD on two popular datasets: the Left Atrial Segmentation Challenge (LA) and the NIH pancreas CT dataset. The results on the LA dataset demonstrate that, in two types of labeled ratios (i.e., 20% and 10%), SimCVD achieves an average Dice score of 90.85% and 89.03% respectively, a 0.91% and 2.22% improvement compared to previous best results. Our method can be trained in an end-to-end fashion, showing the promise of utilizing SimCVD as a general framework for downstream tasks, such as medical image synthesis, enhancement, and registration. | ['James S. Duncan', 'Lawrence Staib', 'Ruihan Zhao', 'Yuan Zhou', 'Chenyu You'] | 2021-08-13 | null | null | null | null | ['semi-supervised-medical-image-segmentation'] | ['computer-vision'] | [ 4.58845824e-01 3.65842193e-01 -2.40800247e-01 -6.01922154e-01
-9.60614979e-01 -4.42126751e-01 3.56517345e-01 3.39293897e-01
-6.24725282e-01 6.45560324e-01 1.06616415e-01 -2.38987491e-01
1.61762848e-01 -4.80651826e-01 -7.23888636e-01 -8.01307499e-01
-3.33426520e-02 6.94203258e-01 2.73114353e-01 1.53128020e-02
-1.41737372e-01 3.91506463e-01 -9.00144458e-01 8.42488036e-02
1.08789504e+00 1.05718148e+00 7.81435594e-02 3.07555467e-01
2.47589173e-03 5.37767589e-01 -3.10397953e-01 -1.37390658e-01
4.48554814e-01 -5.07588923e-01 -1.01362407e+00 2.56862789e-01
5.73313177e-01 -2.42256939e-01 -2.32134283e-01 9.59741473e-01
6.96538091e-01 -7.51239955e-02 6.95491731e-01 -7.87823439e-01
-4.73703116e-01 7.42776573e-01 -7.09822655e-01 2.23426268e-01
-1.13147959e-01 2.54303306e-01 6.89787447e-01 -6.11933172e-01
6.98086560e-01 9.15686131e-01 8.26444864e-01 6.09507442e-01
-1.60637057e+00 -5.36922157e-01 4.52746078e-03 -3.74825269e-01
-1.28735995e+00 -1.49620503e-01 7.51914084e-01 -6.17303193e-01
5.35127997e-01 1.57844052e-01 4.95310366e-01 6.69245422e-01
4.73918878e-02 8.05724740e-01 1.39982033e+00 -2.48130649e-01
1.13480099e-01 7.19808266e-02 2.32167274e-01 8.43303502e-01
2.59292930e-01 -2.24710554e-02 3.88829075e-02 3.24456021e-02
8.76083612e-01 2.04414893e-02 -4.60498095e-01 -7.01339662e-01
-1.46595442e+00 7.04337478e-01 7.73314774e-01 3.69200885e-01
-2.72568464e-01 3.56027484e-02 4.77814436e-01 6.31738678e-02
6.36265993e-01 5.14192641e-01 -3.53248894e-01 2.21101686e-01
-1.24263763e+00 4.56854375e-03 5.02641559e-01 6.73605978e-01
5.33548951e-01 9.60675925e-02 -3.19573581e-01 8.54641557e-01
3.28882277e-01 2.76290894e-01 6.29069924e-01 -7.92595208e-01
3.82352948e-01 6.66986108e-01 -3.84827971e-01 -7.53443003e-01
-6.63067698e-01 -7.44101405e-01 -1.11704242e+00 4.40114439e-01
5.86006463e-01 -1.56272545e-01 -1.48522615e+00 1.82500184e+00
3.42247367e-01 3.09544593e-01 -7.12099969e-02 9.29082870e-01
1.12235081e+00 2.57316232e-01 -8.62439163e-03 -2.82809049e-01
1.21078670e+00 -1.05500066e+00 -6.24388576e-01 -1.69263512e-01
8.32002282e-01 -6.48554504e-01 1.17260909e+00 2.87801802e-01
-1.24252856e+00 -3.40049207e-01 -1.06345427e+00 -3.81863229e-02
-1.31735265e-01 8.81593674e-02 6.07678533e-01 6.83451056e-01
-1.05493081e+00 8.40290427e-01 -1.26225913e+00 -9.24482010e-03
9.47909772e-01 4.99868721e-01 -4.06190097e-01 -1.53376237e-01
-6.51678801e-01 7.88865447e-01 2.51652181e-01 1.38672546e-01
-8.98773789e-01 -1.13462543e+00 -9.07270193e-01 -1.74316183e-01
3.26913267e-01 -7.66572833e-01 1.04620922e+00 -7.99831748e-01
-1.26516867e+00 1.31705272e+00 1.21579334e-01 -6.76617503e-01
8.81545782e-01 -4.03969623e-02 8.39955360e-02 2.65951276e-01
1.72016129e-01 8.66055787e-01 5.67402184e-01 -1.26665020e+00
-1.57440647e-01 -5.63406348e-01 -2.98847020e-01 1.78083897e-01
-1.41295031e-01 -7.18447715e-02 -5.32922983e-01 -9.11440849e-01
3.67182255e-01 -9.29745078e-01 -6.73426449e-01 4.12753463e-01
-4.45425510e-01 1.57833114e-01 5.45280457e-01 -7.28794575e-01
8.96111965e-01 -2.07070303e+00 1.23898789e-01 2.06371799e-01
4.06718999e-01 4.05355483e-01 -6.41680583e-02 -3.83207202e-01
-2.45487839e-01 2.06924185e-01 -1.10257614e+00 -5.06488979e-01
-4.15443987e-01 1.78350806e-01 1.50741011e-01 6.05209291e-01
1.57748505e-01 1.03293550e+00 -8.54988575e-01 -5.47492862e-01
3.07788283e-01 5.76153815e-01 -6.57064080e-01 1.62991956e-01
1.13590034e-02 9.41938639e-01 -2.55069077e-01 4.50810313e-01
7.52341330e-01 -3.91188771e-01 8.86692479e-02 -4.28653449e-01
7.16223568e-02 1.84491098e-01 -1.05091703e+00 2.25184155e+00
-3.38025093e-01 3.78160208e-01 8.38380009e-02 -1.34115982e+00
7.64526725e-01 2.80144930e-01 8.91626596e-01 -7.07711458e-01
2.66283423e-01 4.18921143e-01 1.82413876e-01 -3.26824933e-01
-1.73105389e-01 -3.53322595e-01 1.33756310e-01 3.58036280e-01
1.69017211e-01 -3.11497599e-01 1.92581683e-01 1.94750562e-01
9.93242681e-01 9.44095105e-02 1.85020909e-01 -3.98296416e-01
3.90296310e-01 -7.07467198e-02 7.14561999e-01 5.43726385e-01
-2.13446170e-01 1.26863885e+00 4.48310226e-01 -3.55637789e-01
-8.95879865e-01 -1.02537692e+00 -3.80106926e-01 3.96852583e-01
1.13046475e-01 -1.57746643e-01 -9.79155242e-01 -1.03007102e+00
-1.30394042e-01 3.53919983e-01 -6.98585629e-01 -1.53159454e-01
-7.82715201e-01 -1.20166469e+00 5.63454092e-01 7.69188643e-01
5.75189292e-01 -8.88158262e-01 -5.48287928e-01 1.97080508e-01
-2.91490611e-02 -1.31113613e+00 -4.89332467e-01 1.89199775e-01
-1.35661685e+00 -1.08073545e+00 -1.11321199e+00 -9.70906615e-01
1.02986848e+00 4.09389138e-02 1.23202097e+00 2.44166091e-01
-6.39181077e-01 4.54596318e-02 -9.16595683e-02 -7.17120171e-02
-4.02494133e-01 1.33056000e-01 -2.51927972e-01 -1.03840835e-01
-2.44646728e-01 -7.12245107e-01 -8.73314440e-01 3.20775032e-01
-8.92132938e-01 1.66017041e-01 7.24381745e-01 9.87047553e-01
9.78287458e-01 -3.66891533e-01 4.65962082e-01 -1.30637741e+00
3.38463366e-01 -1.90307453e-01 -4.21363384e-01 1.75801948e-01
-8.86893213e-01 1.00573912e-01 5.29215038e-01 -3.18215996e-01
-8.71169925e-01 3.66449594e-01 -2.54723936e-01 -4.82683808e-01
-2.58257627e-01 5.06668448e-01 8.88130963e-02 -1.98359653e-01
6.93616509e-01 1.36871129e-01 3.67771566e-01 -5.92296720e-01
3.70544702e-01 2.81548083e-01 6.14140034e-01 -3.53034347e-01
7.40344286e-01 6.71135068e-01 1.31117433e-01 -5.25976062e-01
-7.31566846e-01 -4.64727521e-01 -9.24739599e-01 1.98699951e-01
9.23025727e-01 -7.67839193e-01 -2.87056863e-01 4.29577678e-01
-7.17018425e-01 -6.05747998e-01 -6.42032206e-01 6.07822180e-01
-5.41318834e-01 6.40258610e-01 -6.64289892e-01 -1.16267510e-01
-5.74511528e-01 -1.51161444e+00 8.69732559e-01 4.14899960e-02
-7.49451444e-02 -1.10392702e+00 1.46293426e-02 6.09171927e-01
5.75817406e-01 6.38307869e-01 8.79432559e-01 -7.92408288e-01
-3.27509969e-01 -7.68853053e-02 -3.80231440e-01 6.01891756e-01
3.17450285e-01 -4.06581283e-01 -7.90635705e-01 -4.44726110e-01
-3.51873636e-02 -4.33395207e-01 1.01719844e+00 7.27295578e-01
1.35144722e+00 -5.39153181e-02 -2.83564121e-01 9.31591094e-01
1.25774431e+00 -3.13752927e-02 5.22999704e-01 -1.50523670e-02
8.65602255e-01 4.21588689e-01 3.44944566e-01 1.26621827e-01
1.95524901e-01 5.63301027e-01 4.38843459e-01 -9.08146977e-01
-6.10665619e-01 3.62039208e-02 -1.26829714e-01 8.33686471e-01
-4.26005125e-02 2.02615455e-01 -1.03874063e+00 4.72475350e-01
-1.68794549e+00 -4.07400191e-01 -2.13749424e-01 2.19672489e+00
1.17322373e+00 1.65893883e-01 8.96556750e-02 8.16460475e-02
5.15537798e-01 6.63860738e-02 -5.93758643e-01 1.80431772e-02
-3.03433202e-02 5.88354707e-01 5.99312961e-01 4.92876977e-01
-1.38402081e+00 8.16900551e-01 6.16761446e+00 5.92404962e-01
-1.26947439e+00 2.80713350e-01 1.05012822e+00 -1.67821422e-02
-1.55847758e-01 -1.54772609e-01 -3.39934051e-01 3.81898582e-01
6.38709366e-01 1.43782318e-01 9.82354209e-02 6.26925468e-01
2.61232555e-01 4.36321050e-02 -1.15193880e+00 9.54970598e-01
2.01766044e-01 -1.44341385e+00 1.02785090e-02 -6.27928749e-02
9.14846599e-01 1.47396818e-01 6.65779635e-02 1.36036113e-01
3.47756557e-02 -1.28347850e+00 3.50409001e-01 2.21215352e-01
9.64223742e-01 -4.49889362e-01 8.80630851e-01 1.72325909e-01
-1.01077628e+00 3.86579603e-01 -1.13607250e-01 4.32634264e-01
3.42987388e-01 6.83323383e-01 -8.66726458e-01 4.95684087e-01
5.71002841e-01 8.27666998e-01 -6.47744119e-01 1.39369488e+00
-1.66683063e-01 7.97259986e-01 -3.50749701e-01 5.58734357e-01
2.90924251e-01 -2.91218698e-01 3.40070486e-01 1.22612703e+00
1.75421387e-02 5.04248254e-02 4.26299959e-01 1.04613805e+00
-4.20379341e-01 2.73820609e-01 -2.88708180e-01 2.36965820e-01
2.26369631e-02 1.33540571e+00 -1.14534140e+00 -4.28453863e-01
-3.48697454e-01 9.60499287e-01 8.46988857e-02 1.72856525e-01
-8.20600748e-01 -1.75544605e-01 5.18130027e-02 2.52978235e-01
7.96060786e-02 -5.56265973e-02 -6.04601502e-01 -1.18591642e+00
9.26958472e-02 -8.68958890e-01 3.99965316e-01 -2.84300268e-01
-1.18250692e+00 6.40706718e-01 -7.61762634e-02 -1.17778492e+00
-2.56805476e-02 -5.35727561e-01 -5.20044446e-01 6.68117046e-01
-1.78353751e+00 -1.19246149e+00 -5.19984722e-01 4.57722723e-01
4.70146060e-01 9.11858231e-02 7.77341902e-01 6.48863018e-01
-5.54590940e-01 7.74005830e-01 -2.50438545e-02 4.58114922e-01
7.53187716e-01 -1.41889894e+00 1.90805510e-01 7.97828734e-01
1.18739784e-01 4.05227154e-01 3.66156846e-01 -4.64848906e-01
-1.06745398e+00 -1.16287422e+00 4.89508629e-01 -1.61550134e-01
3.11779201e-01 -2.20315486e-01 -1.08088362e+00 6.15808964e-01
7.91366324e-02 7.14791834e-01 6.68848157e-01 -2.26171181e-01
-1.04329944e-01 -8.72613043e-02 -1.45213795e+00 4.65915412e-01
1.01970053e+00 -5.18282689e-02 -5.04594982e-01 5.01155138e-01
6.30135596e-01 -6.30075932e-01 -1.04179704e+00 7.65387297e-01
2.32758373e-01 -8.42965364e-01 1.08423543e+00 -5.87310433e-01
3.48014265e-01 -2.80482411e-01 9.37961787e-02 -1.11258709e+00
-5.56994714e-02 -5.51762104e-01 1.17697984e-01 1.11883938e+00
5.78250527e-01 -4.69118804e-01 9.77793038e-01 6.10812604e-01
-4.49156314e-01 -1.07764161e+00 -9.35811281e-01 -6.59119189e-01
5.16626358e-01 -2.40109637e-01 1.34335339e-01 1.05484188e+00
-3.18341166e-01 2.14750379e-01 -8.80554542e-02 -9.52067599e-02
7.75271416e-01 2.79684402e-02 4.69568282e-01 -1.23944223e+00
-1.80405840e-01 -6.22369170e-01 -4.67231303e-01 -1.01114476e+00
4.95150760e-02 -1.41836691e+00 -1.07608791e-02 -1.70420539e+00
3.48149627e-01 -7.24046767e-01 -3.88762027e-01 5.79427838e-01
-1.20888643e-01 6.01747692e-01 -5.05736796e-03 1.88637957e-01
-2.43285388e-01 4.91779953e-01 1.43011475e+00 -3.94191325e-01
-2.64151752e-01 9.61289927e-03 -6.69828594e-01 8.73826802e-01
8.21977198e-01 -4.61025327e-01 -3.93281966e-01 -4.93842125e-01
-3.41521710e-01 -4.75980006e-02 4.40637887e-01 -9.59954083e-01
4.41273227e-02 2.21250966e-01 3.14467758e-01 -2.94201672e-01
1.00492887e-01 -6.57965064e-01 -2.65409350e-01 6.70885324e-01
-5.01700282e-01 -1.73165679e-01 1.36212930e-01 3.57142568e-01
-2.99177736e-01 -9.22893509e-02 1.17205393e+00 -2.55050510e-01
-2.75185943e-01 6.06445849e-01 4.45853509e-02 5.11692226e-01
9.74120557e-01 -2.20209599e-01 1.37575105e-01 -5.77313788e-02
-1.00207269e+00 2.37748593e-01 2.91113436e-01 5.54158762e-02
6.94541335e-01 -1.12376690e+00 -8.96816373e-01 1.51909322e-01
-1.15765020e-01 4.70549196e-01 2.07766667e-01 1.29432416e+00
-6.53277814e-01 1.20422840e-01 -1.38515860e-01 -9.37014163e-01
-1.25009084e+00 3.22890788e-01 5.36314547e-01 -5.81162035e-01
-1.05353725e+00 8.78640652e-01 2.85508484e-01 -5.86682081e-01
3.08930367e-01 -7.42185235e-01 -1.08607695e-01 -1.19945735e-01
2.85176039e-01 3.82095575e-02 3.01123321e-01 -5.50521791e-01
-4.37805682e-01 7.65581548e-01 -2.03571439e-01 1.16603293e-01
1.46993828e+00 1.59847841e-01 2.07827799e-02 1.81449726e-01
1.41539431e+00 -3.38669509e-01 -1.18436897e+00 -4.24019933e-01
-3.33920754e-02 -3.25603843e-01 2.66845286e-01 -9.38303947e-01
-1.63950539e+00 1.12409294e+00 9.97891843e-01 -2.38709450e-01
9.92534161e-01 9.97816846e-02 9.00533497e-01 -4.17358465e-02
7.00299293e-02 -7.97829151e-01 1.02179512e-01 3.30475457e-02
8.13966036e-01 -1.49611652e+00 1.61996737e-01 -6.49182320e-01
-6.80173099e-01 8.39216411e-01 4.97217566e-01 -2.45822474e-01
6.79731190e-01 4.83905613e-01 1.89990729e-01 -1.78918287e-01
-2.59814024e-01 -1.12452827e-01 5.58938146e-01 4.94821340e-01
7.32837617e-01 -5.27024828e-02 -4.66050476e-01 5.03126144e-01
-1.87521037e-02 -7.23666453e-04 3.32958370e-01 9.06113744e-01
-1.91681325e-01 -1.04477477e+00 -7.15316534e-02 5.43480694e-01
-7.38460243e-01 -9.96687189e-02 -9.15300995e-02 7.28887677e-01
3.68043855e-02 4.81153190e-01 -2.80789137e-02 8.08591843e-02
4.36552554e-01 -1.15863301e-01 4.61278737e-01 -7.47244596e-01
-7.30931222e-01 2.20461264e-01 -2.60476500e-01 -5.79567134e-01
-5.14345706e-01 -5.08248448e-01 -1.72484100e+00 2.36186787e-01
-2.92572349e-01 -1.35765135e-01 4.88287508e-01 8.03934038e-01
2.71669179e-01 7.11448431e-01 4.89134550e-01 -7.77390480e-01
-5.59735000e-01 -7.65028656e-01 -3.60795379e-01 8.64419460e-01
2.00607702e-01 -5.86925685e-01 -2.15871125e-01 9.60811675e-02] | [14.567431449890137, -2.215947151184082] |
2aaba9e3-824c-4f0e-9a72-ceab746cc6aa | babyai-first-steps-towards-grounded-language | 1810.08272 | null | https://arxiv.org/abs/1810.08272v4 | https://arxiv.org/pdf/1810.08272v4.pdf | BabyAI: A Platform to Study the Sample Efficiency of Grounded Language Learning | Allowing humans to interactively train artificial agents to understand language instructions is desirable for both practical and scientific reasons, but given the poor data efficiency of the current learning methods, this goal may require substantial research efforts. Here, we introduce the BabyAI research platform to support investigations towards including humans in the loop for grounded language learning. The BabyAI platform comprises an extensible suite of 19 levels of increasing difficulty. The levels gradually lead the agent towards acquiring a combinatorially rich synthetic language which is a proper subset of English. The platform also provides a heuristic expert agent for the purpose of simulating a human teacher. We report baseline results and estimate the amount of human involvement that would be required to train a neural network-based agent on some of the BabyAI levels. We put forward strong evidence that current deep learning methods are not yet sufficiently sample efficient when it comes to learning a language with compositional properties. | ['Maxime Chevalier-Boisvert', 'Thien Huu Nguyen', 'Chitwan Saharia', 'Salem Lahlou', 'Yoshua Bengio', 'Lucas Willems', 'Dzmitry Bahdanau'] | 2018-10-18 | babyai-a-platform-to-study-the-sample | https://openreview.net/forum?id=rJeXCo0cYX | https://openreview.net/pdf?id=rJeXCo0cYX | iclr-2019-5 | ['grounded-language-learning'] | ['natural-language-processing'] | [ 3.37236464e-01 7.44435310e-01 1.42916217e-01 -3.20699245e-01
-6.96708620e-01 -7.77646720e-01 9.56915140e-01 1.20139077e-01
-6.55004263e-01 6.99997127e-01 2.12388709e-01 -9.86694694e-01
1.79259013e-02 -7.60438859e-01 -9.21691239e-01 -3.35434884e-01
1.99564472e-02 1.03587484e+00 5.88585734e-02 -4.03726071e-01
1.52086571e-01 5.30372381e-01 -1.43259382e+00 4.42408562e-01
8.92289460e-01 8.01770017e-02 3.73917550e-01 6.71373785e-01
-3.56014036e-02 1.26086903e+00 -5.25349796e-01 -3.34360093e-01
1.53331459e-01 -7.19092250e-01 -1.23329759e+00 3.16319875e-02
2.44550481e-01 -5.17742336e-01 -9.26214084e-03 6.52526081e-01
3.97990823e-01 1.25966743e-01 4.84781027e-01 -1.24550569e+00
-3.98161829e-01 1.19718432e+00 8.05452839e-02 8.12044367e-04
6.22663915e-01 7.53976643e-01 8.82457733e-01 -5.81441104e-01
7.13129520e-01 1.38302422e+00 6.31995976e-01 6.75956190e-01
-1.20726585e+00 -5.81475854e-01 1.88299254e-01 4.44329344e-02
-9.38916504e-01 -5.64717174e-01 6.20661318e-01 -4.59851354e-01
1.24872530e+00 -1.00009345e-01 1.03084648e+00 1.40059841e+00
-2.14051515e-01 1.10047853e+00 1.37607610e+00 -7.78267264e-01
2.57976860e-01 4.67754334e-01 -7.33988807e-02 1.10330546e+00
3.63157719e-01 3.89114201e-01 -7.61517823e-01 4.55884710e-02
6.92640960e-01 -8.29243183e-01 1.48272365e-02 -4.93580401e-01
-1.33554626e+00 8.92273724e-01 3.41356367e-01 4.65236723e-01
-2.36384511e-01 2.26118594e-01 4.83095825e-01 5.39230585e-01
-1.07979001e-02 1.32829368e+00 -6.05298281e-01 -1.60866931e-01
-6.71099424e-01 6.04151905e-01 9.01583850e-01 1.14494932e+00
4.68882799e-01 2.47552663e-01 1.43704876e-01 3.86105329e-01
4.37110275e-01 6.33701906e-02 2.53909141e-01 -1.36510956e+00
2.81944573e-01 7.27979362e-01 1.45806387e-01 -4.51977432e-01
-6.67736053e-01 -2.46706143e-01 -3.61962676e-01 4.82327729e-01
7.68089592e-01 -3.78903568e-01 -5.55384099e-01 1.69068849e+00
4.01052803e-01 2.32046060e-02 3.55841130e-01 6.91208482e-01
6.73382759e-01 8.43761623e-01 4.07120317e-01 -2.36979783e-01
1.41476238e+00 -8.77589703e-01 -3.58631015e-01 -4.94649351e-01
1.12793493e+00 -2.90333629e-01 1.64820492e+00 6.66037560e-01
-1.56770158e+00 -4.81416255e-01 -1.14842343e+00 -1.96980521e-01
-4.86444473e-01 -2.31192350e-01 1.16256046e+00 6.26571596e-01
-1.03704035e+00 3.56047422e-01 -8.69286537e-01 -1.86491102e-01
1.59746245e-01 4.47042853e-01 -2.72655040e-01 -1.39307112e-01
-1.23893189e+00 1.34938860e+00 7.64067054e-01 5.34042642e-02
-1.19451988e+00 -7.80197680e-01 -1.18125963e+00 -3.50561105e-02
4.89059329e-01 -8.23930979e-01 1.67508733e+00 -1.07504523e+00
-1.63367724e+00 9.34457600e-01 2.72142708e-01 -5.60791552e-01
8.67352307e-01 5.51372170e-02 2.03153282e-01 1.18067712e-01
-5.91307832e-03 1.04491401e+00 2.93509334e-01 -1.44600022e+00
-5.93940556e-01 3.97979356e-02 4.78174120e-01 3.60582501e-01
2.71479450e-02 1.31171584e-01 1.69489369e-01 -3.99120808e-01
-2.75503814e-01 -1.10395169e+00 -2.33293548e-01 -2.94042826e-01
-1.55214313e-02 -7.66126096e-01 1.68727040e-01 -3.17449182e-01
6.05754077e-01 -1.63329017e+00 2.31067047e-01 -2.62761880e-02
7.52426237e-02 2.82862097e-01 -3.30771178e-01 5.60505867e-01
1.45670846e-01 2.91741937e-01 -2.51145989e-01 -1.23949639e-01
2.85884708e-01 -2.57198643e-02 -1.23466894e-01 1.40038356e-01
3.43398869e-01 1.00501466e+00 -1.22186792e+00 -4.61079270e-01
1.81209609e-01 2.60029793e-01 -8.03504646e-01 5.05062222e-01
-7.00513601e-01 3.88590038e-01 -3.09450418e-01 2.41822690e-01
3.82592482e-03 1.83955382e-03 3.03885788e-01 1.59426734e-01
-1.56466290e-01 7.85726905e-01 -9.49896693e-01 1.86894870e+00
-7.28819788e-01 7.85293818e-01 5.68587333e-02 -7.84620404e-01
4.94108200e-01 4.55546111e-01 -9.94567480e-03 -5.44391453e-01
6.63073733e-02 2.55013376e-01 6.79545045e-01 -6.61593556e-01
3.17877650e-01 -4.12428290e-01 -3.85402925e-02 8.56013894e-01
1.36530653e-01 -7.13653445e-01 2.38602728e-01 8.91115516e-02
1.01809061e+00 6.16593242e-01 1.32270053e-01 -5.00173151e-01
3.77183288e-01 4.34754014e-01 1.07560888e-01 9.26942468e-01
-1.52630091e-01 -8.40117857e-02 3.54413807e-01 -7.30638146e-01
-1.28487742e+00 -6.77568257e-01 2.74586141e-01 1.47423434e+00
-2.86752969e-01 -3.03608090e-01 -9.45219398e-01 -5.59733570e-01
-5.67451477e-01 1.45890701e+00 -4.03288603e-01 -1.59263849e-01
-7.34462976e-01 -2.83648133e-01 7.87239909e-01 3.83500546e-01
4.85681385e-01 -1.70388782e+00 -1.27433670e+00 2.12561831e-01
2.68783011e-02 -9.54841077e-01 1.96149051e-01 4.74037379e-01
-5.46945989e-01 -7.84983814e-01 -2.90000796e-01 -1.00528860e+00
5.45295835e-01 -1.03529736e-01 1.44566846e+00 4.24580365e-01
1.85192991e-02 5.17033398e-01 -2.03164250e-01 -8.06197882e-01
-1.03952610e+00 3.61152977e-01 -1.01792909e-01 -9.40127730e-01
4.62497056e-01 -3.69140804e-01 -2.33742759e-01 -9.12254676e-02
-8.09193730e-01 7.26356745e-01 5.47145665e-01 7.45467722e-01
-1.59650728e-01 1.22012168e-01 5.74597776e-01 -1.01244831e+00
9.31226015e-01 -2.68720627e-01 -6.57758355e-01 3.26613396e-01
-4.46548581e-01 4.09118056e-01 6.47294879e-01 -5.19620001e-01
-1.17427123e+00 2.39866972e-01 -1.23652771e-01 2.97571421e-01
-4.96782184e-01 6.01761043e-01 -2.04731718e-01 1.50247797e-01
7.77875006e-01 8.38919058e-02 -1.11077972e-01 7.95312300e-02
4.97094423e-01 4.40083593e-01 2.21270129e-01 -1.29499888e+00
7.02700615e-01 -2.35758051e-01 -2.33891919e-01 -8.11466694e-01
-7.44248211e-01 2.42679089e-01 -5.23860157e-01 -5.32452762e-02
7.21410632e-01 -9.96324420e-01 -9.61918414e-01 2.28954896e-01
-1.22052455e+00 -1.22817802e+00 -1.99057892e-01 4.49291110e-01
-1.01422787e+00 -2.99061358e-01 -6.61280632e-01 -6.94822431e-01
-7.26974607e-02 -1.45534801e+00 6.72023118e-01 5.85096627e-02
-6.70650721e-01 -1.02746761e+00 -2.84881666e-02 5.89197874e-01
2.06698492e-01 8.17299485e-02 1.42887855e+00 -1.06636429e+00
-5.60155451e-01 1.91968501e-01 2.17278466e-01 -7.19017759e-02
-2.23207012e-01 5.36450222e-02 -8.75925422e-01 -2.69026935e-01
-1.17103057e-02 -1.17010486e+00 1.07824773e-01 -9.48437583e-03
8.09320688e-01 -4.33068812e-01 -6.71330169e-02 1.53879941e-01
9.87889469e-01 3.56256038e-01 3.37827832e-01 6.31013215e-01
4.80532855e-01 1.02389729e+00 2.38956735e-01 -2.33616158e-02
5.89244902e-01 3.49239677e-01 -2.98395548e-02 -5.03865518e-02
-1.99442785e-02 -5.54683208e-01 4.52780813e-01 7.54600167e-01
8.16785321e-02 -2.66912878e-01 -1.44939709e+00 7.09034145e-01
-1.48621941e+00 -9.34367776e-01 3.19616705e-01 1.60689116e+00
1.25795722e+00 4.23178822e-01 1.83583781e-01 5.52001446e-02
1.97264403e-01 -2.35673174e-01 -5.10692000e-01 -7.35930622e-01
3.37797612e-01 1.46164775e-01 -5.85741028e-02 8.64637911e-01
-5.29050887e-01 1.34431362e+00 6.88247681e+00 3.00970286e-01
-7.81679809e-01 -3.00393224e-01 6.25899315e-01 2.40652621e-01
-4.66403693e-01 -4.49056476e-02 -6.80052698e-01 -5.46990708e-02
1.34661722e+00 -1.81974232e-01 8.17315280e-01 6.52524173e-01
2.77421147e-01 -1.21428460e-01 -1.64890766e+00 4.06538814e-01
-5.48790488e-03 -1.34455061e+00 1.98719382e-01 1.39431376e-03
6.38807476e-01 -1.33364260e-01 -1.48832008e-01 7.56087005e-01
8.12288940e-01 -1.35994101e+00 9.30350959e-01 1.07904464e-01
2.99135059e-01 -9.71111178e-01 4.68071461e-01 9.33015466e-01
-6.59880996e-01 4.20626998e-02 -1.42264934e-02 -4.46751416e-01
-1.75306097e-01 -4.63848948e-01 -1.25886953e+00 -8.85974914e-02
2.84560829e-01 3.94722484e-02 -5.38275838e-01 5.40027618e-01
-5.80088735e-01 7.46765375e-01 -2.82154679e-01 -4.19359863e-01
7.00746894e-01 -3.74743789e-02 1.56956673e-01 1.22635007e+00
-6.14413582e-02 5.02154946e-01 3.38722616e-01 1.01861405e+00
-8.08057785e-02 -1.29427463e-01 -1.04841149e+00 -2.94771433e-01
6.38434350e-01 1.02255714e+00 -8.86030495e-01 -4.06032801e-01
-4.03964579e-01 5.99407315e-01 4.97459382e-01 1.83880448e-01
-5.12581468e-01 -2.27948174e-01 2.16125295e-01 1.08391039e-01
-1.42616570e-01 -4.53768939e-01 -4.44526374e-01 -7.73343146e-01
-3.80563587e-01 -1.77826154e+00 3.01239509e-02 -1.06495118e+00
-7.89252579e-01 5.51890254e-01 3.58555615e-01 -4.79238838e-01
-7.52140820e-01 -7.85030007e-01 -4.03214097e-01 6.75214231e-01
-9.98336136e-01 -1.29733586e+00 -1.38476565e-01 4.27753925e-01
8.31597328e-01 -4.39561218e-01 1.08124685e+00 -3.57276589e-01
-2.89983720e-01 3.86513144e-01 -7.06706107e-01 1.26163349e-01
1.26966432e-01 -1.31794024e+00 6.41170323e-01 6.62698627e-01
3.80379319e-01 9.58950281e-01 1.03775716e+00 -5.17750740e-01
-1.45569754e+00 -5.51629663e-01 8.69524062e-01 -6.22624934e-01
6.62303627e-01 -5.65478802e-01 -8.39925170e-01 1.01921189e+00
7.45584071e-01 -5.01357675e-01 5.14028966e-01 1.48352496e-02
-1.51067972e-01 4.13661212e-01 -1.08259034e+00 1.13967025e+00
1.06125367e+00 -6.16745353e-01 -1.06406355e+00 5.24544656e-01
9.14873004e-01 -4.14416552e-01 -6.71759129e-01 3.88342828e-01
2.97128499e-01 -8.63781214e-01 7.35082209e-01 -8.73995245e-01
5.44126570e-01 -2.52080023e-01 1.88539147e-01 -1.53546345e+00
-1.42860085e-01 -9.43684220e-01 2.60660708e-01 1.02503073e+00
6.62134588e-01 -2.74752468e-01 7.26299763e-01 9.03196216e-01
-2.63267189e-01 -6.21727705e-01 -4.29874748e-01 -4.59090084e-01
6.19972467e-01 -6.04888916e-01 5.12148023e-01 9.84042048e-01
2.73994952e-01 8.69590700e-01 4.88354340e-02 -2.75221076e-02
3.54550630e-01 -1.90368548e-01 8.72590959e-01 -1.05371678e+00
-2.59486258e-01 -6.66455090e-01 2.31019944e-01 -7.92782903e-01
6.06214404e-01 -1.00272477e+00 3.65341604e-01 -1.67345119e+00
-2.49194093e-02 -5.05980909e-01 1.83287233e-01 5.41088998e-01
5.77042662e-02 -3.03056419e-01 2.44376734e-01 -2.52305090e-01
-3.77016902e-01 3.48344207e-01 1.22309864e+00 -1.86061502e-01
-3.51220667e-01 -2.72847772e-01 -8.44674230e-01 1.02628827e+00
7.98281431e-01 -2.46056631e-01 -9.46734190e-01 -7.78480887e-01
6.28234625e-01 -1.18936658e-01 3.21706861e-01 -9.52491462e-01
3.56711745e-01 -4.97822165e-01 4.01043653e-01 -2.50459146e-02
1.13042397e-02 -8.68774772e-01 -1.78177636e-02 6.25275314e-01
-1.01637745e+00 2.61544496e-01 5.15536964e-01 -1.43516764e-01
1.49745911e-01 -4.45074081e-01 7.17653573e-01 -6.30872309e-01
-6.05027974e-01 -2.07127839e-01 -8.23277652e-01 2.10484087e-01
9.18635130e-01 -1.14969678e-01 -7.89108947e-02 -3.78212184e-01
-3.78662705e-01 3.25834632e-01 4.49676096e-01 2.20458001e-01
4.25158709e-01 -1.00180626e+00 -7.87850559e-01 4.59719449e-02
7.00529590e-02 3.40422451e-01 -4.03402269e-01 2.29642004e-01
-9.46139097e-01 4.19655323e-01 -3.33716631e-01 -2.37741694e-01
-9.57233608e-01 6.95608020e-01 6.76312089e-01 -1.92744628e-01
-6.93898916e-01 9.74081516e-01 -1.28402580e-02 -7.49668896e-01
5.05967796e-01 -4.37936842e-01 -4.21539880e-02 -8.39953721e-02
5.26864529e-01 1.24041468e-01 -6.38202131e-02 -2.75878102e-01
-1.13471322e-01 4.57019582e-02 -4.73474264e-02 -5.47782004e-01
1.42558420e+00 2.40144357e-01 6.86099827e-02 5.67064464e-01
6.12737119e-01 -1.78002089e-01 -1.30520022e+00 -4.19992842e-02
4.49289322e-01 2.20434323e-01 -9.45048705e-02 -1.07884634e+00
-2.58844703e-01 9.29208577e-01 1.11440778e-01 1.24835923e-01
6.26004338e-01 7.69188302e-03 3.98646593e-01 1.09160113e+00
5.32368064e-01 -1.14684141e+00 1.83399215e-01 7.20220149e-01
1.01205456e+00 -1.25248575e+00 9.29546654e-02 7.48834610e-02
-7.95157850e-01 1.05581141e+00 9.19212282e-01 -9.63549688e-02
1.22161787e-02 4.15169626e-01 1.01609804e-01 -2.83112258e-01
-1.17125630e+00 -9.38309580e-02 5.37736453e-02 7.66596854e-01
8.53209078e-01 -6.56574592e-02 1.15909269e-02 1.96548209e-01
-6.89153433e-01 -1.12811662e-01 5.97337663e-01 9.13627565e-01
-5.17437220e-01 -8.78920376e-01 -1.82143554e-01 -1.34592891e-01
-8.81730393e-02 -1.15822434e-01 -6.26140773e-01 1.18827748e+00
2.24143073e-01 1.00545895e+00 -1.31926164e-01 9.94062051e-02
3.72398585e-01 2.81280190e-01 7.96044707e-01 -7.35523045e-01
-1.04630125e+00 -2.85009414e-01 4.52753514e-01 -4.03817981e-01
-3.91585499e-01 -6.04372740e-01 -1.48868835e+00 -2.52045542e-01
1.12087496e-01 2.14015514e-01 5.99402905e-01 1.22794068e+00
-3.42679203e-01 3.70104432e-01 1.27697647e-01 -9.17592645e-01
-5.23984849e-01 -9.15081203e-01 2.00653866e-01 1.29606366e-01
1.80612788e-01 -3.31839591e-01 -1.12345509e-01 7.67920837e-02] | [4.154476165771484, 1.183275580406189] |
12eed9c5-9443-4fb0-b3b9-a37f0f8ecdd6 | design-considerations-for-hypothesis | 2211.09711 | null | https://arxiv.org/abs/2211.09711v1 | https://arxiv.org/pdf/2211.09711v1.pdf | Design Considerations For Hypothesis Rejection Modules In Spoken Language Understanding Systems | Spoken Language Understanding (SLU) systems typically consist of a set of machine learning models that operate in conjunction to produce an SLU hypothesis. The generated hypothesis is then sent to downstream components for further action. However, it is desirable to discard an incorrect hypothesis before sending it downstream. In this work, we present two designs for SLU hypothesis rejection modules: (i) scheme R1 that performs rejection on domain specific SLU hypothesis and, (ii) scheme R2 that performs rejection on hypothesis generated from the overall SLU system. Hypothesis rejection modules in both schemes reject/accept a hypothesis based on features drawn from the utterance directed to the SLU system, the associated SLU hypothesis and SLU confidence score. Our experiments suggest that both the schemes yield similar results (scheme R1: 2.5% FRR @ 4.5% FAR, scheme R2: 2.5% FRR @ 4.6% FAR), with the best performing systems using all the available features. We argue that while either of the rejection schemes can be chosen over the other, they carry some inherent differences which need to be considered while making this choice. Additionally, we incorporate ASR features in the rejection module (obtaining an 1.9% FRR @ 3.8% FAR) and analyze the improvements. | ['Shankar Ananthakrishnan', 'Rahul Gupta', 'Aman Alok'] | 2022-10-31 | null | null | null | null | ['spoken-language-understanding', 'spoken-language-understanding'] | ['natural-language-processing', 'speech'] | [ 4.76185411e-01 6.96416557e-01 6.90030605e-02 -6.85904801e-01
-1.01551747e+00 -6.17518723e-01 6.89420581e-01 3.20991367e-01
-2.24694908e-01 5.73078752e-01 2.75036037e-01 -4.77539361e-01
1.06238902e-01 -6.89519227e-01 -4.60394114e-01 -2.72175848e-01
1.13351539e-01 4.66897905e-01 4.70412344e-01 -4.30176646e-01
5.41200519e-01 2.45443329e-01 -1.92007935e+00 5.05763888e-01
7.71245182e-01 1.10186088e+00 2.00305641e-01 9.68132019e-01
-3.71791720e-01 7.74353385e-01 -7.46936023e-01 2.13019311e-01
1.68772146e-01 -9.48754549e-01 -1.07751238e+00 -1.40762078e-02
-6.68043569e-02 -2.94271797e-01 1.45206869e-01 6.87674463e-01
2.69835114e-01 2.89377332e-01 7.00937748e-01 -1.12610388e+00
-2.40457311e-01 5.67759395e-01 -6.54829219e-02 1.12202495e-01
1.13185596e+00 -4.38885354e-02 1.14666426e+00 -1.01010907e+00
4.93685961e-01 1.42428327e+00 3.50633472e-01 7.87048101e-01
-1.07591307e+00 -4.65833336e-01 1.58673868e-01 -1.52490839e-01
-1.33522427e+00 -7.15490758e-01 7.97800571e-02 -1.16369493e-01
1.28745997e+00 5.34652054e-01 1.39672309e-01 6.46143317e-01
7.73343220e-02 9.82354403e-01 1.26838243e+00 -8.29723060e-01
6.21319234e-01 4.11810666e-01 5.22953570e-01 4.51358974e-01
-2.71055341e-01 9.84737873e-02 -5.49944818e-01 -3.16827059e-01
3.43531162e-01 -6.49412394e-01 -4.28882748e-01 3.97321373e-01
-5.81928074e-01 9.43170726e-01 6.94226176e-02 3.02326888e-01
-1.63129181e-01 -3.43727767e-01 1.71441093e-01 6.49410486e-01
4.32707310e-01 3.49881023e-01 -6.21256948e-01 -1.24848001e-01
-8.23415279e-01 1.45462081e-01 8.89252424e-01 1.14529753e+00
8.61921906e-01 -5.40792160e-02 -3.81567813e-02 1.11800230e+00
6.19867563e-01 -3.28042805e-02 7.28807628e-01 -7.22365916e-01
9.09141377e-02 4.35973674e-01 1.88977852e-01 -3.62148672e-01
-3.71336043e-01 -3.96172926e-02 2.25204788e-02 2.16381088e-01
2.73969650e-01 -2.24729523e-01 -1.16493845e+00 1.56522000e+00
1.37857199e-02 1.70802042e-01 6.29050076e-01 9.06314135e-01
7.78040051e-01 9.54312384e-01 3.54195274e-02 -3.15421611e-01
1.13485980e+00 -7.59094715e-01 -6.79520488e-01 -4.45961803e-01
9.99227524e-01 -1.16688192e+00 8.59335482e-01 3.86952072e-01
-9.87689793e-01 -4.67231661e-01 -1.18800902e+00 2.40534559e-01
-1.63049579e-01 -1.69623196e-02 2.85342634e-01 6.25492275e-01
-1.45513260e+00 4.68747586e-01 -2.26086423e-01 -6.61743283e-01
-5.10018110e-01 4.27606493e-01 -2.27714568e-01 7.88012112e-04
-1.41835296e+00 9.58067656e-01 5.86740077e-01 -1.99082401e-03
-5.54539144e-01 -3.69508639e-02 -8.62724185e-01 2.43771166e-01
3.25328618e-01 -1.08803391e-01 1.79455352e+00 -1.08119726e+00
-1.83812451e+00 5.28694928e-01 -3.42873037e-01 -2.99298257e-01
4.32673037e-01 -2.70766884e-01 -4.99233723e-01 -9.11956280e-02
-3.58817764e-02 6.05732262e-01 4.80368614e-01 -1.25394356e+00
-1.19464314e+00 -5.50294630e-02 -8.70320126e-02 2.57547677e-01
4.37515408e-01 1.53799295e-01 -1.96337134e-01 -3.24451506e-01
5.81333160e-01 -7.20243037e-01 1.64683923e-01 -6.25268579e-01
-1.47675842e-01 -8.64080548e-01 8.12391460e-01 -4.73613232e-01
1.30115187e+00 -1.87580156e+00 -2.33443737e-01 3.85624498e-01
-3.82191807e-01 4.28475767e-01 -3.08062047e-01 7.17772186e-01
-2.48040825e-01 2.89244801e-01 -1.19177043e-01 -1.70671895e-01
-1.81016505e-01 2.20992729e-01 -3.44271779e-01 9.44048986e-02
4.57332760e-01 2.55054623e-01 -8.93508196e-01 -2.90046837e-02
2.66508222e-01 -1.19655520e-01 -6.02257848e-01 5.84950805e-01
-1.08875506e-01 1.59601718e-02 -4.46868747e-01 3.77689451e-01
4.62900996e-01 3.69385928e-01 2.81602114e-01 8.57902542e-02
-3.53739202e-01 7.25554526e-01 -1.37758541e+00 1.11411107e+00
-3.98883998e-01 5.22118628e-01 8.89596045e-02 -8.51877332e-01
1.45871472e+00 6.56495273e-01 -1.26291186e-01 -4.12531525e-01
2.41555616e-01 5.28961897e-01 1.87384993e-01 -2.74569869e-01
5.65993965e-01 -3.72559518e-01 -8.24035108e-02 5.94885230e-01
2.42722914e-01 -1.99938148e-01 -8.07013735e-02 3.06560516e-01
1.12481606e+00 2.03413472e-01 5.71536303e-01 -4.03636307e-01
7.90861666e-01 -2.04265311e-01 5.09425879e-01 1.16090667e+00
-4.26966876e-01 7.38289416e-01 4.09359902e-01 -1.28049850e-01
-8.19518209e-01 -8.26975703e-01 -3.76554243e-02 9.41857696e-01
5.80275841e-02 -4.66023326e-01 -7.67932296e-01 -8.26764345e-01
-2.69081533e-01 1.41441512e+00 -2.36547604e-01 -3.91194135e-01
-3.05571228e-01 -2.73350328e-01 6.19591713e-01 5.21535635e-01
1.42972469e-01 -1.00275505e+00 -4.40541774e-01 2.05348983e-01
-1.53287470e-01 -7.57929564e-01 -2.30938450e-01 3.16130906e-01
-6.48402154e-01 -9.22187865e-01 -3.18626881e-01 -6.66501045e-01
6.30916953e-01 4.15425211e-01 6.53378904e-01 3.20377797e-01
2.26062611e-01 4.37733591e-01 -1.11709285e+00 -3.61498922e-01
-1.09401500e+00 -2.27628514e-01 3.57039839e-01 1.05203930e-02
5.65020263e-01 1.09369434e-01 -2.22255662e-01 5.14011681e-01
-8.39050293e-01 -3.61021534e-02 4.21749294e-01 8.27654123e-01
2.03803509e-01 -1.64112762e-01 1.00946271e+00 -8.63110662e-01
6.71374500e-01 -4.65294689e-01 -3.16175401e-01 2.48526350e-01
-5.87289274e-01 8.64755958e-02 6.70329094e-01 -1.13082096e-01
-1.11361122e+00 2.61191633e-02 -7.62752891e-01 -1.15412332e-01
-6.08194947e-01 4.75122303e-01 -1.34209409e-01 3.08677018e-01
5.80644369e-01 2.16816157e-01 9.47133601e-02 -5.04373431e-01
1.76380932e-01 1.28396821e+00 1.37427881e-01 -2.10321426e-01
2.54715651e-01 -7.66761675e-02 -8.50389123e-01 -1.13189554e+00
-9.26553011e-01 -5.90763032e-01 -3.25294226e-01 -3.06485802e-01
5.21269202e-01 -5.43875992e-01 -4.32595998e-01 7.85122961e-02
-1.01023149e+00 -1.65231094e-01 2.07568482e-02 6.68622255e-01
-6.47758305e-01 4.44819957e-01 -4.88831013e-01 -1.42980611e+00
-2.38217682e-01 -1.13796651e+00 9.53118026e-01 3.49025786e-01
-8.78929436e-01 -4.27843809e-01 -3.13697577e-01 3.89079511e-01
2.16458395e-01 -5.10690689e-01 8.14577758e-01 -1.39815676e+00
2.06732880e-02 -3.58495981e-01 -9.34454724e-02 5.52492321e-01
2.51348466e-01 1.31274417e-01 -1.33796895e+00 -4.18261141e-01
2.32505739e-01 -3.40626001e-01 6.71101213e-01 2.16977194e-01
4.78327513e-01 -1.74747750e-01 -1.44465178e-01 -2.77166069e-01
1.02623141e+00 7.04173803e-01 6.47240341e-01 9.39259380e-02
-1.28712803e-01 1.08691013e+00 1.03344333e+00 2.15223730e-01
6.71636015e-02 4.45056498e-01 4.70044389e-02 2.21820071e-01
1.29173817e-02 -2.73792475e-01 6.71742797e-01 7.52904773e-01
4.55028564e-01 -5.62045217e-01 -8.52746665e-01 4.45229769e-01
-2.01718378e+00 -4.92278814e-01 -1.32139459e-01 2.33864403e+00
4.92929697e-01 4.20412630e-01 -8.41950923e-02 2.71497577e-01
7.81989574e-01 -1.39314309e-01 -8.94178599e-02 -1.01169944e+00
-7.73430569e-03 -1.40627462e-03 1.19837478e-01 1.07825649e+00
-7.59467363e-01 1.17190790e+00 6.79945564e+00 6.58563018e-01
-1.01356113e+00 -1.31466687e-01 6.52997196e-01 2.62673587e-01
-4.71715331e-01 3.38112354e-01 -8.83992910e-01 2.77320474e-01
1.24503756e+00 -8.42880681e-02 -1.27081471e-02 5.72098553e-01
4.17786539e-01 -7.39752650e-01 -1.24622488e+00 4.25350159e-01
1.34423494e-01 -9.24405336e-01 1.46708563e-01 -4.18147355e-01
2.19368666e-01 -1.62314475e-01 -5.43886900e-01 4.91735548e-01
2.92962462e-01 -8.61207366e-01 9.27194178e-01 2.11382672e-01
4.60134417e-01 -9.62473631e-01 1.07078683e+00 6.08357847e-01
-9.07081246e-01 2.31614947e-01 -3.24938387e-01 -3.61774147e-01
1.37344822e-01 2.03960523e-01 -1.29730272e+00 6.51078880e-01
5.11685848e-01 -4.04201709e-02 -1.99992526e-02 6.43556356e-01
-3.51635218e-01 8.65543365e-01 -4.08712715e-01 -2.82250732e-01
5.36602914e-01 -1.54480368e-01 7.13038981e-01 1.37989616e+00
3.47613126e-01 3.79191190e-01 3.29551637e-01 5.13107955e-01
3.22158873e-01 3.60151201e-01 -6.67905986e-01 3.00567061e-01
6.03162467e-01 8.08204174e-01 -7.34704792e-01 -4.14851308e-01
-4.95605022e-01 1.12118912e+00 1.65126801e-01 2.18670219e-01
-5.15597105e-01 -3.67791742e-01 5.18401504e-01 -1.95355803e-01
3.60492617e-02 1.28941149e-01 2.43989006e-02 -8.30116570e-01
-3.23304772e-01 -6.99783146e-01 4.17972714e-01 -7.98896968e-01
-1.01052940e+00 8.36506009e-01 -1.19143941e-01 -1.01200616e+00
-7.19568610e-01 -4.89661098e-01 -7.34846711e-01 1.08141327e+00
-1.31143677e+00 -6.98603690e-01 2.87243247e-01 -3.79058234e-02
1.20243883e+00 -1.08991014e-02 9.96162176e-01 1.39967846e-02
-4.42031860e-01 5.09553552e-01 -2.36168608e-01 -1.46946415e-01
7.51749516e-01 -1.11573327e+00 2.26789400e-01 7.44520068e-01
-1.97853327e-01 7.40937531e-01 1.08499408e+00 -7.20491052e-01
-9.77648258e-01 -8.37324202e-01 1.31027603e+00 -2.97038496e-01
2.89289713e-01 -1.36310369e-01 -1.11795270e+00 7.79257834e-01
1.84653744e-01 -5.74141145e-01 8.58689129e-01 5.42940162e-02
-7.85059780e-02 3.35155666e-01 -1.38101900e+00 5.16743839e-01
4.62492794e-01 -2.54337579e-01 -1.07111764e+00 -3.33014056e-02
6.34185195e-01 -1.60756156e-01 -7.15419352e-01 4.62133765e-01
5.42876482e-01 -1.08711886e+00 3.30040008e-01 -5.10501385e-01
1.07194267e-01 -4.70697343e-01 -3.79070133e-01 -1.30508971e+00
-2.36638978e-01 -5.86587608e-01 3.24734271e-01 1.31576538e+00
9.54715610e-01 -7.45593667e-01 3.79200608e-01 7.98701108e-01
-3.37791234e-01 -6.38436675e-01 -1.05692041e+00 -7.47801721e-01
-1.08601004e-01 -7.02340543e-01 1.17198840e-01 6.67580843e-01
2.47225925e-01 6.77295387e-01 -2.10281730e-01 2.45124564e-01
8.57719854e-02 -1.32489055e-01 5.24794698e-01 -8.96191061e-01
3.70050641e-03 -3.55454773e-01 -1.43353015e-01 -1.24401629e+00
8.70292038e-02 -5.85262895e-01 5.78694820e-01 -1.45208204e+00
-3.18368345e-01 -4.64925110e-01 -2.09517226e-01 5.70049345e-01
-1.73191458e-01 -1.71768963e-01 1.90156132e-01 2.13421151e-01
-9.65623930e-02 4.96515810e-01 4.72732186e-01 2.55665839e-01
-5.38776875e-01 2.57359505e-01 -6.33070111e-01 9.34218526e-01
9.81182039e-01 -1.95478573e-01 -4.05482858e-01 -2.94376574e-02
-4.20828998e-01 7.62004778e-02 -2.46468425e-01 -7.42020071e-01
1.30179584e-01 -1.94367945e-01 7.48470351e-02 -5.64263701e-01
3.32581252e-01 -4.87319887e-01 -3.06354910e-01 5.39644420e-01
-4.18291390e-01 -2.72981524e-01 2.57630378e-01 3.71383071e-01
-3.72384638e-01 -7.67325461e-01 9.22389925e-01 -1.43376127e-01
-1.08795881e+00 -6.19963884e-01 -1.09701800e+00 -4.69350696e-01
9.95248020e-01 -5.38717508e-01 7.09605515e-02 -8.09643745e-01
-7.29250252e-01 4.51920211e-01 1.47470728e-01 7.47989893e-01
8.61546457e-01 -7.55521834e-01 -7.98668265e-01 4.25886333e-01
2.00505495e-01 -3.66743267e-01 -4.06666845e-02 6.22394919e-01
-3.87721956e-01 6.42656982e-01 3.18190068e-01 -4.25050795e-01
-1.35751104e+00 5.87568842e-02 2.96623141e-01 2.26436079e-01
-4.54256564e-01 1.03011322e+00 2.62528628e-01 -6.27012789e-01
1.38865799e-01 -2.88972855e-01 -2.93717712e-01 2.20350489e-01
5.96558154e-01 2.61083394e-01 2.25926146e-01 -7.80544579e-01
-4.62455213e-01 6.50217384e-02 -2.35294878e-01 -3.99118513e-01
9.37159479e-01 -4.15615469e-01 8.26584846e-02 6.61115944e-01
9.38470364e-01 5.19914366e-02 -7.54349709e-01 -3.93499732e-02
3.54687870e-01 -2.37229899e-01 8.13993998e-03 -1.00968826e+00
-3.02701890e-01 3.68314117e-01 4.47508901e-01 6.42442226e-01
9.88090515e-01 1.84172109e-01 5.86525738e-01 2.34305054e-01
4.69475925e-01 -1.25637782e+00 -4.33241427e-01 8.36093187e-01
1.01619029e+00 -1.02445507e+00 -2.96245694e-01 -6.58502281e-01
-7.34173059e-01 1.16223323e+00 8.21190238e-01 -1.71039045e-01
3.23436588e-01 4.62661535e-02 4.61005867e-01 1.42344877e-01
-1.01942527e+00 -2.97242105e-01 1.43088877e-01 2.17959911e-01
8.10464323e-01 -1.50557663e-02 -1.00621128e+00 5.16931713e-01
-4.81103100e-02 -2.16356516e-01 7.63866603e-01 1.10798383e+00
-9.22591090e-01 -1.28751433e+00 -5.54620922e-01 5.34190595e-01
-8.01297873e-02 5.99613115e-02 -8.35584521e-01 7.46096373e-01
-1.58409879e-01 1.58818054e+00 1.04779065e-01 -5.44200182e-01
5.28044820e-01 4.81834799e-01 1.09267898e-01 -1.04298055e+00
-6.93661094e-01 5.30185938e-01 6.85688198e-01 -4.91042346e-01
-1.33197635e-01 -5.88642418e-01 -1.62154245e+00 6.89318702e-02
-8.65220904e-01 5.00462294e-01 4.08259332e-01 1.21274233e+00
2.77171522e-01 2.82948494e-01 8.07652056e-01 -5.48264802e-01
-5.06842852e-01 -1.22861421e+00 -4.75401461e-01 1.88340917e-01
1.27374962e-01 -4.51721698e-01 -5.70036471e-01 -3.01193744e-01] | [13.964408874511719, 6.940930366516113] |
2af0bdc0-46c4-45d0-be16-5e251b7dba4f | learning-based-symbolic-abstractions-for | 2004.01879 | null | https://arxiv.org/abs/2004.01879v4 | https://arxiv.org/pdf/2004.01879v4.pdf | Learning-based Symbolic Abstractions for Nonlinear Control Systems | Symbolic models or abstractions are known to be powerful tools for the control design of cyber-physical systems (CPSs) with logic specifications. In this paper, we investigate a novel learning-based approach to the construction of symbolic models for nonlinear control systems. In particular, the symbolic model is constructed based on learning the un-modeled part of the dynamics from training data based on state-space exploration, and the concept of an alternating simulation relation that represents behavioral relationships with respect to the original control system. Moreover, we aim at achieving safe exploration, meaning that the trajectory of the system is guaranteed to be in a safe region for all times while collecting the training data. In addition, we provide some techniques to reduce the computational load, in terms of memory and computation time, of constructing the symbolic models and the safety controller synthesis, so as to make our approach practical. Finally, a numerical simulation illustrates the effectiveness of the proposed approach. | ['Dimos Dimarogonas', 'Toshimitsu Ushio', 'Masako Kishida', 'Adnane Saoud', 'Kazumune Hashimoto'] | 2020-04-04 | null | null | null | null | ['safe-exploration'] | ['robots'] | [ 2.79545456e-01 4.47405308e-01 -3.52381200e-01 1.03878111e-01
-2.44993076e-01 -4.90907311e-01 4.90121484e-01 1.94908142e-01
1.54824346e-01 8.63426149e-01 -5.23732066e-01 -7.52855599e-01
-5.55964887e-01 -7.79376090e-01 -7.98862875e-01 -7.17869341e-01
-4.44229603e-01 9.05979201e-02 2.43307918e-01 -1.23678014e-01
1.07732052e-02 6.66128218e-01 -1.32095349e+00 -3.72107208e-01
8.90503407e-01 7.88551927e-01 -6.98973015e-02 3.66153568e-01
2.70369977e-01 6.06072724e-01 -1.90591112e-01 4.41733211e-01
1.66131854e-01 -5.13570070e-01 -6.33081794e-01 1.86279416e-01
-1.94338664e-01 -1.47159129e-01 -3.74679238e-01 1.37478030e+00
-2.02929616e-01 3.20354998e-01 4.44550663e-01 -1.57682765e+00
2.27170348e-01 5.56823552e-01 4.17244062e-02 -5.19845784e-01
1.73919365e-01 3.28553110e-01 5.46418786e-01 -2.25669518e-01
5.99898338e-01 1.06177104e+00 1.57671183e-01 3.55371147e-01
-1.33323789e+00 -6.81490839e-01 3.85365009e-01 9.69955549e-02
-1.72797310e+00 -1.69751436e-01 7.34786749e-01 -4.71692085e-01
5.20764291e-01 4.41612244e-01 7.89987803e-01 7.22024500e-01
5.13549626e-01 4.06438887e-01 9.83779609e-01 -6.34308994e-01
5.92235804e-01 1.76578313e-01 1.75522357e-01 7.91217685e-01
4.48686033e-01 6.52572811e-01 2.44041175e-01 -5.20843208e-01
6.59061968e-01 8.26910958e-02 -2.57229894e-01 -6.86412215e-01
-8.50862145e-01 6.63641810e-01 2.74728566e-01 1.50299475e-01
-2.38105729e-01 2.53627300e-01 3.48060757e-01 2.92081296e-01
-1.84034601e-01 6.18729770e-01 -2.87509948e-01 1.08138837e-01
-4.74176824e-01 5.25158048e-01 8.35480690e-01 1.09337234e+00
5.56068599e-01 4.72963780e-01 4.48724747e-01 -2.57722318e-01
3.25931370e-01 4.20406640e-01 -8.24663416e-02 -7.26759613e-01
1.41662523e-01 8.08487535e-01 5.15281260e-01 -7.39919543e-01
-1.95754513e-01 -1.18703119e-01 -7.13508070e-01 5.86776912e-01
3.30300957e-01 -2.79403657e-01 -6.65686309e-01 1.69089925e+00
4.79325980e-01 5.65992653e-01 2.14187384e-01 5.46041012e-01
-2.49704450e-01 1.06828380e+00 -5.66787980e-02 -5.79825521e-01
8.36462021e-01 -3.62243176e-01 -6.71372831e-01 1.26499042e-01
5.95780253e-01 6.05229521e-03 8.72274458e-01 5.87938190e-01
-9.21843112e-01 -3.53559166e-01 -1.45116436e+00 8.24009836e-01
-1.85215667e-01 1.69761807e-01 2.02619419e-01 3.51413995e-01
-6.35393202e-01 7.66214728e-01 -1.09016395e+00 -2.44840551e-02
-1.06577642e-01 5.88846862e-01 6.76406175e-02 2.66753912e-01
-1.12233543e+00 8.37203085e-01 1.03200185e+00 3.18164855e-01
-1.20615935e+00 -4.99523461e-01 -8.16167176e-01 2.19029203e-01
9.57331181e-01 -7.48286918e-02 1.05988514e+00 -8.10488641e-01
-1.57912672e+00 -8.88846889e-02 1.02711476e-01 -5.52476823e-01
3.10679913e-01 1.85288087e-01 -6.02607071e-01 -3.18927653e-02
-6.12241149e-01 -5.91070242e-02 1.00650609e+00 -1.23174834e+00
-4.92857903e-01 -3.75000238e-02 3.84685278e-01 -4.25074518e-01
-2.26657540e-01 -3.31356794e-01 6.21581450e-02 -2.35718817e-01
2.61169262e-02 -1.31137800e+00 -6.01395905e-01 -6.37362152e-02
-4.20788616e-01 -1.19828105e-01 9.58715260e-01 -5.46626627e-01
1.42712140e+00 -2.15725589e+00 4.43716437e-01 1.06092727e+00
-1.15708314e-01 6.35703325e-01 2.69357890e-01 6.42405391e-01
-9.52620506e-02 5.32251261e-02 -1.43834144e-01 3.64434838e-01
-4.86724637e-03 1.99236184e-01 -7.92258203e-01 5.06225705e-01
3.52501690e-01 3.85244519e-01 -7.87207365e-01 -3.34233969e-01
4.22864616e-01 9.03415754e-02 -2.94017643e-01 2.03591868e-01
-7.33602703e-01 6.62263513e-01 -1.00310135e+00 5.14876507e-02
3.47460955e-01 1.75984547e-01 8.39598060e-01 5.89225292e-02
-3.70058507e-01 1.41859651e-01 -1.39535868e+00 7.79681027e-01
-5.94415486e-01 -7.68045038e-02 2.02345699e-01 -9.62574184e-01
9.10283029e-01 5.19245684e-01 1.47519216e-01 -3.33345473e-01
4.84131187e-01 1.92560345e-01 3.88374962e-02 -7.48568177e-02
-6.89444393e-02 -5.69816977e-02 -3.07903290e-01 2.08986208e-01
-3.39178354e-01 -3.37581336e-01 1.80007175e-01 -2.02544004e-01
6.73947036e-01 3.71017270e-02 7.53749609e-01 -3.62070620e-01
1.19392705e+00 1.04404218e-01 6.69350266e-01 2.44478315e-01
2.03854784e-01 -7.34257758e-01 8.39043140e-01 -1.70676410e-01
-1.03921378e+00 -7.13282764e-01 2.04080373e-01 4.64214012e-02
1.86421037e-01 -3.63077104e-01 -7.83145487e-01 -3.78132105e-01
-2.82262236e-01 1.01345837e+00 -2.93328434e-01 -7.13033080e-01
-8.06852639e-01 2.20374823e-01 2.68777192e-01 2.84081310e-01
-1.46329418e-01 -6.72437668e-01 -8.80971313e-01 2.19729990e-01
4.65344340e-01 -8.48917246e-01 -6.49682134e-02 1.43325746e-01
-1.11243713e+00 -1.20263696e+00 1.78322375e-01 -8.20577502e-01
8.85014176e-01 -4.23575252e-01 1.01912171e-01 2.29992256e-01
-1.67245552e-01 3.06973271e-02 3.20420042e-03 -2.17248887e-01
-8.73624206e-01 -6.71687126e-02 2.14858964e-01 -1.76129311e-01
-4.00947362e-01 -2.95263052e-01 2.22200409e-01 2.47749716e-01
-9.57946777e-01 2.08935902e-01 2.39278316e-01 8.26253653e-01
6.92679882e-01 8.28552246e-01 6.47304416e-01 -4.76513177e-01
6.10509932e-01 -2.61517495e-01 -1.79946327e+00 4.50362265e-01
-5.53881288e-01 2.87330896e-01 1.32705557e+00 -6.43144011e-01
-7.87291408e-01 5.04237354e-01 3.01919729e-01 -6.36008859e-01
-5.51508293e-02 5.61764479e-01 -5.40897906e-01 -2.56915033e-01
3.23138326e-01 3.06995749e-01 3.62980485e-01 -1.26481801e-01
2.40503162e-01 3.95636171e-01 3.35454136e-01 -8.09357822e-01
1.08855057e+00 8.59159678e-02 5.25723279e-01 -7.14935780e-01
-2.16656655e-01 1.25107600e-03 -4.82918411e-01 -2.17187122e-01
3.58550102e-01 -3.43748569e-01 -1.37903774e+00 1.45018354e-01
-8.35460544e-01 -1.35944620e-01 -2.72436142e-01 4.27145660e-01
-7.99814641e-01 2.79115997e-02 -1.93995893e-01 -1.55001163e+00
-1.86945125e-02 -1.28840172e+00 4.32807624e-01 1.86854526e-01
-4.11266625e-01 -6.67855561e-01 9.55577418e-02 -6.03993773e-01
-6.18632622e-02 8.36147070e-01 1.38993824e+00 -4.99843508e-01
-8.12910497e-01 -5.13969839e-01 3.53348672e-01 1.50341660e-01
4.73522628e-03 1.68307453e-01 -4.65139955e-01 -7.14706361e-01
2.70104885e-01 -8.70613232e-02 -8.24662223e-02 1.62151113e-01
9.30386305e-01 -6.65187478e-01 -6.40377164e-01 7.58119449e-02
1.51549149e+00 9.74179685e-01 3.80521059e-01 -1.16035730e-01
4.20683682e-01 6.05895042e-01 1.02580643e+00 4.52239305e-01
-1.69426993e-01 7.04325318e-01 5.38508177e-01 3.13706875e-01
6.65053844e-01 -3.36346030e-01 3.81543368e-01 3.23186517e-01
2.01499939e-01 1.83057249e-01 -9.93206441e-01 5.38255155e-01
-1.88266790e+00 -6.76760316e-01 1.51817545e-01 2.61114597e+00
6.64473832e-01 3.46565127e-01 2.02193394e-01 5.78172088e-01
7.38204718e-01 -4.14634079e-01 -5.93894184e-01 -5.41249752e-01
5.02507508e-01 1.58537418e-01 3.75231087e-01 8.97016883e-01
-9.45591390e-01 6.27756476e-01 5.58515692e+00 6.98515892e-01
-1.32456493e+00 -5.56210458e-01 1.37613282e-01 2.62616694e-01
4.93613258e-02 3.57747644e-01 -6.98698819e-01 2.46855631e-01
1.28858316e+00 -6.86797917e-01 7.11192727e-01 1.08737743e+00
4.94381547e-01 -1.08965889e-01 -1.13265622e+00 2.58355200e-01
-5.47368884e-01 -1.06579745e+00 -1.37826368e-01 4.42375951e-02
6.15743458e-01 -8.82323742e-01 -5.09656146e-02 1.98248759e-01
1.26776874e-01 -7.38057375e-01 9.61842120e-01 4.32951182e-01
5.68666339e-01 -1.28610635e+00 5.07881463e-01 7.51931906e-01
-1.25195932e+00 -4.88316953e-01 1.57024965e-01 -8.95227641e-02
2.74406880e-01 1.20755725e-01 -1.04971504e+00 7.08642006e-01
-5.25584519e-02 4.07842427e-01 -2.34335706e-01 9.48877811e-01
-3.38199884e-01 5.04630983e-01 -3.11459005e-01 -5.72257936e-01
4.67333227e-01 -4.29305553e-01 6.26580775e-01 5.23865163e-01
2.43256018e-01 2.31572837e-01 6.58066273e-01 1.02617049e+00
7.12919176e-01 -3.49836946e-01 -8.58805776e-01 -4.82338637e-01
4.74791586e-01 7.83712804e-01 -5.23608685e-01 -3.43365580e-01
1.48396358e-01 4.73815762e-02 3.00786737e-02 3.66669685e-01
-1.14488101e+00 -6.07123494e-01 5.19362569e-01 1.25583991e-01
1.95696250e-01 -6.76113367e-01 -2.17277482e-01 -6.61794126e-01
-5.33358939e-02 -9.50749338e-01 3.42481472e-02 -3.01618516e-01
-5.04072487e-01 4.80291069e-01 4.57218349e-01 -1.54352558e+00
-8.56828034e-01 -4.03622806e-01 -5.40184140e-01 7.44218469e-01
-1.08955419e+00 -7.10837305e-01 1.30149916e-01 5.87891281e-01
5.70778456e-03 1.35434672e-01 8.36311877e-01 -5.28936237e-02
-5.12081683e-01 1.25213385e-01 -5.88895427e-03 -3.47568810e-01
-1.46064656e-02 -7.45478570e-01 1.08050629e-01 8.86453986e-01
-4.79769826e-01 8.39668155e-01 1.06361580e+00 -7.41572559e-01
-1.69984317e+00 -1.20587063e+00 5.11977196e-01 2.18066052e-01
9.32247043e-01 -2.43617579e-01 -9.78446245e-01 7.79785156e-01
-1.67304397e-01 -3.56648505e-01 -1.12780951e-01 -3.28620285e-01
6.76476657e-02 -2.21298009e-01 -1.12616718e+00 1.04384089e+00
4.51334059e-01 -3.35298836e-01 -5.35891533e-01 7.72573873e-02
9.10000980e-01 -3.86111259e-01 -7.98616290e-01 4.09051567e-01
3.58802766e-01 -5.92343993e-02 8.61478329e-01 -8.42366278e-01
-1.64502665e-01 -9.05513108e-01 1.50496811e-01 -1.24723172e+00
-1.66814685e-01 -9.08287883e-01 -5.53686440e-01 8.50091696e-01
1.41770616e-01 -7.17839479e-01 5.47935247e-01 4.62560892e-01
-8.29195976e-02 -8.13283026e-01 -1.00691545e+00 -1.17639947e+00
-6.25155047e-02 -2.32115641e-01 6.79155290e-01 5.46569109e-01
5.16046464e-01 4.59067263e-02 -3.84193212e-01 7.19015181e-01
4.21895087e-01 3.55735421e-01 5.01042604e-01 -9.31086719e-01
-3.28495055e-01 -1.55159071e-01 -2.86082208e-01 -2.97002852e-01
5.69963455e-01 -3.91896099e-01 2.49451384e-01 -9.75466669e-01
-4.71110702e-01 -5.02700627e-01 -1.30529657e-01 3.98197025e-01
3.68759274e-01 -7.62304783e-01 2.33217984e-01 -1.93289787e-01
-2.00724304e-01 5.77214718e-01 1.05505872e+00 -9.14974809e-02
-5.39619088e-01 4.74363446e-01 -4.59650112e-03 6.41750157e-01
8.59134853e-01 -2.02556059e-01 -9.53478634e-01 2.95962125e-01
-1.87293544e-01 7.89684415e-01 4.45308656e-01 -1.08706725e+00
1.40208239e-02 -7.79243290e-01 -3.00308764e-01 -6.85655832e-01
4.00553584e-01 -1.50519896e+00 7.17385054e-01 1.25359511e+00
-3.28742534e-01 -1.49972647e-01 3.46330166e-01 6.20369017e-01
-3.06138486e-01 -3.80581319e-01 9.11123335e-01 4.30362880e-01
-5.39722383e-01 1.09789195e-02 -5.46633363e-01 -4.22608435e-01
1.41915286e+00 1.78268313e-01 -2.46585906e-02 -2.26864204e-01
-8.35804343e-01 4.08742964e-01 3.19836140e-01 1.89307719e-01
3.35637808e-01 -1.11262178e+00 1.50858641e-01 2.82886714e-01
1.44588901e-02 -2.24463820e-01 1.32394046e-01 5.89984834e-01
-2.73622960e-01 8.78769577e-01 -2.99520552e-01 -3.37322444e-01
-1.12392068e+00 9.86639321e-01 2.35721529e-01 -1.99105814e-01
-5.69089293e-01 -1.50666773e-01 -9.44035351e-02 -3.21487963e-01
2.34041423e-01 -8.98758888e-01 -2.75065377e-02 -4.67980742e-01
5.00013113e-01 5.42197585e-01 -5.02716601e-02 -6.46816269e-02
-4.32688802e-01 3.80833983e-01 2.46906221e-01 -1.59902036e-01
1.01385546e+00 2.78042376e-01 -9.09697860e-02 5.02537787e-01
7.53204167e-01 -4.33877110e-01 -1.28342903e+00 -3.24606225e-02
2.41440058e-01 -2.53816754e-01 -8.88338313e-02 -4.52168345e-01
-5.38858414e-01 5.43044209e-01 4.35331285e-01 3.07485372e-01
1.14591372e+00 -4.85075474e-01 4.00198579e-01 7.29085088e-01
7.59657264e-01 -1.00534165e+00 -3.23017061e-01 5.23311913e-01
7.02008724e-01 -3.38421613e-01 -2.55923700e-02 -5.90502977e-01
-3.49359125e-01 1.22625649e+00 6.89212143e-01 -5.79748988e-01
5.69696844e-01 5.59765577e-01 -6.49137378e-01 2.60633200e-01
-7.54718542e-01 3.07804078e-01 1.90335035e-01 4.56079692e-01
-2.89753109e-01 1.67665273e-01 -4.08366531e-01 6.03163719e-01
2.72384793e-01 3.01616520e-01 5.97959459e-01 1.25037694e+00
-5.78711629e-01 -1.06886792e+00 -5.67438126e-01 -9.64765176e-02
2.77003378e-01 5.49076736e-01 -2.15079769e-01 1.15357804e+00
-1.22629441e-01 6.67103827e-01 -2.72409499e-01 -2.81107932e-01
5.67195892e-01 2.20146403e-01 4.80762899e-01 -6.08337343e-01
-1.53593525e-01 1.37580827e-01 2.44793639e-01 -6.61230922e-01
2.37776697e-01 -3.86896133e-01 -1.61337066e+00 -1.33879289e-01
-5.64803123e-01 3.60136926e-01 5.00019193e-01 1.05361617e+00
1.63171306e-01 6.95387006e-01 7.70773172e-01 -6.07425153e-01
-1.01198006e+00 -3.75130892e-01 -5.90430915e-01 -3.79599422e-01
3.68414968e-01 -7.81126201e-01 -7.37670437e-02 -1.56990010e-02] | [4.855257034301758, 2.2578728199005127] |
e9a9ea3f-1fe3-4dd3-b0d0-0f3cb9dad4d5 | sgdraw-scene-graph-drawing-interface-using | 2211.16697 | null | https://arxiv.org/abs/2211.16697v2 | https://arxiv.org/pdf/2211.16697v2.pdf | SGDraw: Scene Graph Drawing Interface Using Object-Oriented Representation | Scene understanding is an essential and challenging task in computer vision. To provide the visually fundamental graphical structure of an image, the scene graph has received increased attention due to its powerful semantic representation. However, it is difficult to draw a proper scene graph for image retrieval, image generation, and multi-modal applications. The conventional scene graph annotation interface is not easy to use in image annotations, and the automatic scene graph generation approaches using deep neural networks are prone to generate redundant content while disregarding details. In this work, we propose SGDraw, a scene graph drawing interface using object-oriented scene graph representation to help users draw and edit scene graphs interactively. For the proposed object-oriented representation, we consider the objects, attributes, and relationships of objects as a structural unit. SGDraw provides a web-based scene graph annotation and generation tool for scene understanding applications. To verify the effectiveness of the proposed interface, we conducted a comparison study with the conventional tool and the user experience study. The results show that SGDraw can help generate scene graphs with richer details and describe the images more accurately than traditional bounding box annotations. We believe the proposed SGDraw can be useful in various vision tasks, such as image retrieval and generation. | ['Haoran Xie', 'Xi Yang', 'Chia-Ming Chang', 'Xusheng Du', 'Tianyu Zhang'] | 2022-11-30 | null | null | null | null | ['scene-graph-generation'] | ['computer-vision'] | [ 2.77478158e-01 -2.29390338e-02 3.19939166e-01 -3.62675250e-01
1.09914154e-01 -4.82973456e-01 4.95598257e-01 6.23101830e-01
-3.73329408e-02 2.31792182e-01 -1.61013082e-02 -4.03622210e-01
-1.13009214e-01 -1.09354055e+00 -5.08328021e-01 -2.49905676e-01
4.63974983e-01 2.54075378e-01 6.02778614e-01 -2.54605085e-01
5.84217012e-01 7.58792281e-01 -1.70504391e+00 3.53295296e-01
9.35739815e-01 7.82201886e-01 7.47903585e-01 3.76038045e-01
-8.62989247e-01 7.64086843e-01 -8.28724325e-01 -3.50676835e-01
-6.75310194e-02 -3.56232435e-01 -6.11953855e-01 5.09661138e-01
6.70881033e-01 -5.67863286e-01 -1.30456209e-01 1.20058787e+00
4.25371438e-01 1.52865767e-01 4.83763963e-01 -1.55707657e+00
-8.40480447e-01 3.79395902e-01 -5.26030779e-01 -3.04807782e-01
5.78323305e-01 -1.86993495e-01 6.60387337e-01 -6.31429791e-01
7.84696519e-01 1.45547354e+00 2.40392894e-01 4.00965363e-01
-8.54433894e-01 -5.86964190e-01 2.59641528e-01 2.70442933e-01
-1.26910901e+00 1.02895228e-02 1.06370270e+00 -5.74757636e-01
7.50547349e-01 4.50112939e-01 8.57320666e-01 2.59223372e-01
-4.96926978e-02 7.94566453e-01 6.50573432e-01 -5.97723365e-01
1.42546117e-01 2.76850045e-01 3.67005110e-01 8.93967330e-01
4.06946391e-01 -4.50466871e-01 -3.14058959e-01 1.84334666e-01
1.29245770e+00 3.47108722e-01 -2.60941118e-01 -7.13797271e-01
-1.04840779e+00 6.50668800e-01 8.34621191e-01 1.14310458e-01
-1.24326505e-01 3.17390561e-01 5.24737954e-01 -2.39917994e-01
1.32534713e-01 3.73685509e-01 1.86972022e-01 2.07960442e-01
-6.36591077e-01 3.38255763e-01 4.85134423e-01 1.23869061e+00
7.32098639e-01 2.11125448e-01 -2.83930570e-01 8.52723002e-01
3.73636276e-01 4.37200844e-01 1.96251586e-01 -7.89401114e-01
3.80641907e-01 1.15892959e+00 4.11072001e-02 -1.63477266e+00
-2.73210913e-01 -6.30109161e-02 -8.45226407e-01 2.94372320e-01
2.13646963e-01 1.88196972e-01 -9.50033545e-01 1.11549544e+00
3.37320268e-01 -2.19697565e-01 -1.52844802e-01 8.69600117e-01
1.64656425e+00 7.69354105e-01 3.44908476e-01 1.87443897e-01
1.44646370e+00 -1.03817415e+00 -9.32150245e-01 -2.13009015e-01
6.24675333e-01 -7.70178795e-01 1.50165677e+00 2.29637086e-01
-8.09808016e-01 -7.04949617e-01 -1.02514446e+00 -4.33393180e-01
-6.54484689e-01 3.50193262e-01 8.33370090e-01 3.31808269e-01
-9.08843279e-01 2.14001417e-01 -4.23646808e-01 -5.98280489e-01
4.22546387e-01 -7.37275258e-02 -4.45503175e-01 -1.97774276e-01
-8.18680406e-01 6.88636720e-01 8.41682255e-01 1.11966819e-01
-4.69957680e-01 -3.83485824e-01 -1.16332650e+00 3.00054967e-01
3.67288411e-01 -6.83279276e-01 1.09968007e+00 -6.95062518e-01
-9.80300069e-01 7.85473824e-01 4.43993555e-03 -5.79500794e-02
3.50485325e-01 -3.27382416e-01 -4.87020761e-02 2.64439553e-01
1.36267528e-01 7.22024262e-01 5.39484024e-01 -1.52264225e+00
-3.93852562e-01 -2.95692503e-01 4.17564243e-01 5.55598438e-01
-5.20540215e-02 -4.85692061e-02 -7.35633969e-01 -5.61393499e-01
2.27779180e-01 -3.66662890e-01 -1.35775521e-01 4.29198354e-01
-7.41086245e-01 -2.33253658e-01 1.42486465e+00 -6.38962150e-01
1.28912663e+00 -2.11839128e+00 -3.47475529e-01 2.72315025e-01
3.12983125e-01 2.73664236e-01 2.58671343e-02 7.00892866e-01
-1.46992773e-01 2.11010352e-01 -6.60378858e-02 -7.16263279e-02
-1.61854208e-01 7.38756657e-02 -7.91085288e-02 -1.86382845e-01
-1.56941056e-01 6.75557315e-01 -8.90723109e-01 -7.91214406e-01
8.07020128e-01 4.87042397e-01 -5.69627821e-01 2.51879901e-01
-4.91066605e-01 1.65600225e-01 -5.69217086e-01 4.05302703e-01
6.78216815e-01 -2.77641773e-01 4.07598615e-02 -7.40769267e-01
6.73974678e-02 -1.56425193e-01 -1.41010177e+00 1.81481981e+00
-4.62283194e-01 8.99551928e-01 -3.85092944e-01 -7.52344012e-01
1.29281151e+00 -7.44419247e-02 1.12617508e-01 -7.48026788e-01
9.43272114e-02 -2.60438710e-01 -2.21318617e-01 -6.86517715e-01
6.54297769e-01 2.41994500e-01 1.41313598e-01 4.24854070e-01
-3.33522588e-01 -7.41703272e-01 4.41510320e-01 8.02274406e-01
4.73181784e-01 2.60859042e-01 4.25165325e-01 -4.65607829e-02
4.52786267e-01 2.52294570e-01 -1.43318065e-02 4.72902507e-01
4.87578243e-01 5.42012632e-01 4.90442485e-01 -5.00659287e-01
-1.06814897e+00 -9.76216316e-01 1.83188453e-01 6.17693126e-01
6.53143644e-01 -7.40246177e-01 -8.89579892e-01 -2.65547156e-01
-3.00988168e-01 9.36146379e-01 -1.06798373e-01 5.26397973e-02
-2.11289659e-01 -1.14360794e-01 3.27850282e-02 4.42462832e-01
1.04480243e+00 -1.39960432e+00 -9.43083167e-01 -8.26117843e-02
-8.22097808e-02 -9.74684536e-01 -4.41929311e-01 -5.95177174e-01
-7.43241489e-01 -1.21955013e+00 -5.86291909e-01 -1.11497784e+00
1.22809947e+00 6.81411028e-01 1.02731848e+00 5.60354710e-01
-6.03259325e-01 4.03077275e-01 -5.38016438e-01 -5.12072802e-01
-3.29977572e-01 -2.84014076e-01 -5.03599644e-01 -1.15944311e-01
-8.17026384e-03 -3.01180154e-01 -7.08253682e-01 1.39837369e-01
-1.23278797e+00 9.24622059e-01 2.17944875e-01 5.15139878e-01
5.75102627e-01 3.33956331e-01 2.32673898e-01 -1.25366187e+00
9.33381855e-01 1.07507266e-01 -6.38748884e-01 4.68816370e-01
-1.75658092e-01 3.78097296e-02 5.05468249e-01 -5.97980767e-02
-1.27015448e+00 1.15083702e-01 2.36179382e-02 -3.34522992e-01
-2.48989046e-01 5.71232140e-01 -5.22019148e-01 9.09212306e-02
4.96600270e-01 2.74206191e-01 -9.80951265e-02 -4.58251864e-01
8.35333586e-01 6.57414734e-01 4.58142430e-01 -4.04654533e-01
5.77543318e-01 1.19605370e-01 -3.93730886e-02 -1.00442791e+00
-5.14964640e-01 -5.04729211e-01 -7.35799789e-01 -4.48504806e-01
9.29865420e-01 -5.54826558e-01 -6.44755900e-01 3.32808524e-01
-1.60455203e+00 -2.67403215e-01 -9.40484703e-02 8.56493786e-02
-3.74702007e-01 6.53984487e-01 -2.00484917e-01 -7.11555600e-01
-5.03933370e-01 -1.08945692e+00 1.32775307e+00 4.74807054e-01
-9.47098956e-02 -8.17277730e-01 -4.66909885e-01 1.67159334e-01
1.55958399e-01 6.14740789e-01 1.20738935e+00 -2.79906958e-01
-8.15882146e-01 -2.29086772e-01 -8.41961384e-01 -3.68333235e-02
3.67783040e-01 2.48580784e-01 -6.94505692e-01 2.17591658e-01
-5.14464378e-01 -5.25648333e-02 3.35833251e-01 1.11137681e-01
1.62228048e+00 -4.17021155e-01 -5.02682090e-01 4.60356265e-01
1.77295446e+00 7.60721862e-01 9.07327950e-01 3.17446351e-01
1.18555260e+00 7.73884833e-01 6.46497965e-01 3.17877531e-01
1.50970399e-01 6.31583095e-01 4.64546770e-01 -4.99771923e-01
-2.04981983e-01 -6.14847302e-01 -3.07463109e-01 5.37141860e-01
1.08828843e-01 -4.14116472e-01 -8.75347435e-01 2.86524713e-01
-1.77509785e+00 -8.99024725e-01 -6.55110538e-01 2.07070708e+00
4.49021250e-01 -3.05560585e-02 -1.81035683e-01 1.54998258e-01
7.61113644e-01 -5.18246070e-02 -3.17282945e-01 -5.88296175e-01
2.10669473e-01 -4.60582040e-02 8.27320814e-02 3.87104064e-01
-7.85162628e-01 1.04301500e+00 5.51278973e+00 8.39449108e-01
-1.12695646e+00 -2.28380501e-01 4.22163934e-01 3.61396790e-01
-3.18633914e-01 -3.50238048e-02 -4.95376170e-01 2.14174569e-01
-9.21859443e-02 -3.91394943e-01 1.23909175e-01 1.16812277e+00
4.58509505e-01 -3.20025861e-01 -1.12987351e+00 1.56162620e+00
1.27418965e-01 -1.49763560e+00 9.02737081e-01 -4.06859189e-01
4.59972143e-01 -7.98748195e-01 -3.38877290e-01 -1.52888536e-01
1.02145538e-01 -9.55064118e-01 6.86448514e-01 5.10141551e-01
9.80697989e-01 -6.38579786e-01 5.47880232e-01 2.92640775e-01
-1.34241188e+00 3.73734474e-01 -5.12678385e-01 -4.42193821e-02
3.17584038e-01 4.38704073e-01 -1.10727894e+00 5.15834570e-01
5.54078162e-01 5.66185057e-01 -8.77887487e-01 1.45519960e+00
-1.94859296e-01 -2.05657780e-02 -7.30860531e-02 -2.87916899e-01
-2.15838645e-02 -3.69745880e-01 6.58065602e-02 1.07610369e+00
4.16593015e-01 1.49143159e-01 2.53021479e-01 1.02931213e+00
9.24683213e-02 6.40859365e-01 -1.15375149e+00 -1.86866924e-01
4.56856251e-01 1.27279222e+00 -1.21976089e+00 -5.11124849e-01
-2.21988529e-01 1.07801545e+00 7.29993433e-02 3.60687196e-01
-7.45984852e-01 -8.97622287e-01 -2.08700486e-02 4.40849155e-01
-2.00400323e-01 -4.58238631e-01 -2.84658492e-01 -8.37808132e-01
-1.73769802e-01 -5.94344854e-01 1.49519742e-01 -1.76683569e+00
-7.98288643e-01 7.00125515e-01 4.54517990e-01 -1.20626831e+00
-3.31996381e-02 -6.04342103e-01 -8.16302240e-01 8.09613228e-01
-7.60671318e-01 -1.40293491e+00 -1.06327963e+00 5.47195256e-01
8.01017880e-01 2.01003060e-01 6.12250745e-01 2.75233418e-01
-3.56853634e-01 1.03426494e-01 -1.86431393e-01 2.82630682e-01
2.30902746e-01 -1.16712749e+00 3.96679670e-01 7.12465107e-01
3.18012893e-01 7.04403758e-01 5.47199845e-01 -9.09306586e-01
-1.05699742e+00 -9.42716002e-01 3.98220062e-01 -6.63249791e-02
6.78882748e-02 -4.65180695e-01 -8.30657065e-01 5.38362205e-01
2.89246410e-01 -3.47990930e-01 2.99445748e-01 -3.96667838e-01
-7.84494504e-02 -1.07803561e-01 -8.95677149e-01 9.74161685e-01
1.09429860e+00 -3.58110577e-01 -4.46274847e-01 4.88123894e-01
8.71417284e-01 -3.70549679e-01 -4.05328840e-01 1.59718752e-01
5.44621825e-01 -1.01983130e+00 9.79570866e-01 -3.08850348e-01
3.69713813e-01 -6.20398283e-01 6.83085844e-02 -1.15433741e+00
-3.59275639e-02 -1.51955307e-01 4.70840901e-01 1.41725802e+00
1.21466242e-01 -2.75999397e-01 6.81338191e-01 7.43962526e-01
-1.89441130e-01 -3.62830013e-01 -1.17148362e-01 -3.74518424e-01
-7.12219954e-01 -4.78465259e-01 6.74564064e-01 7.95877099e-01
-7.84919262e-02 5.74611485e-01 -6.27150685e-02 -3.65467481e-02
4.32672232e-01 2.15413570e-01 1.21242630e+00 -1.30982220e+00
-6.86361734e-03 -4.41999733e-01 -6.05531275e-01 -1.06353974e+00
-4.38019603e-01 -7.43851006e-01 -1.16569161e-01 -2.54428482e+00
1.42848969e-01 -6.16292179e-01 4.15674448e-01 2.92813599e-01
-3.22337858e-02 4.93027866e-02 5.18408060e-01 6.24857582e-02
-4.72061396e-01 4.01165962e-01 1.66236639e+00 -3.04098934e-01
-2.88209260e-01 -3.34041923e-01 -6.34710193e-01 7.97650576e-01
7.35812783e-01 -3.52661982e-02 -1.08160543e+00 -5.82379460e-01
2.12795734e-01 -6.65659532e-02 4.59839910e-01 -8.70712876e-01
1.74564853e-01 -2.08078340e-01 3.58026892e-01 -9.79353368e-01
3.22219491e-01 -8.96375716e-01 4.97748673e-01 4.75264311e-01
-1.93181962e-01 1.67742580e-01 3.33523452e-01 3.04942429e-01
-1.70314148e-01 -3.90524596e-01 4.58002657e-01 -4.87300396e-01
-1.04917669e+00 1.92464322e-01 -5.56326620e-02 -3.24621558e-01
1.05133653e+00 -6.50352955e-01 -4.19276714e-01 -6.17072761e-01
-6.42845094e-01 9.61925387e-02 4.88410294e-01 3.90478253e-01
1.20190477e+00 -1.38310432e+00 -2.59798586e-01 2.79157907e-01
3.65056843e-01 5.74149966e-01 4.43605006e-01 7.56482184e-02
-1.47241604e+00 3.36227775e-01 -3.99156094e-01 -5.85069537e-01
-1.57402575e+00 6.82166398e-01 1.41319215e-01 1.91657186e-01
-8.36787164e-01 5.71985781e-01 1.03753757e+00 -1.83203429e-01
2.81220853e-01 -6.53265595e-01 -4.46612716e-01 -2.38162696e-01
4.85380232e-01 2.79924661e-01 -1.50462836e-01 -4.22655284e-01
1.95934176e-02 7.40554512e-01 1.84085444e-01 9.05973390e-02
8.14819276e-01 -1.32844314e-01 -1.90991089e-01 2.51083046e-01
7.67095506e-01 -3.76016758e-02 -7.66080797e-01 1.60094470e-01
-2.60832846e-01 -5.97590446e-01 -1.40625387e-01 -6.74249828e-01
-9.78542507e-01 1.26691735e+00 4.78616744e-01 5.17194092e-01
9.97782171e-01 -4.85815816e-02 5.54493845e-01 4.93410677e-01
5.25572360e-01 -6.56258821e-01 3.18764418e-01 -6.21301755e-02
1.31060791e+00 -9.30427790e-01 1.77686170e-01 -1.01268137e+00
-6.37925506e-01 1.22988307e+00 1.08503747e+00 1.99933480e-02
4.05936897e-01 -6.69683143e-02 -2.96566933e-02 -6.49640441e-01
-8.70522410e-02 -1.74527720e-01 4.56196159e-01 9.05240178e-01
4.35445309e-01 -1.06224241e-02 -2.92290270e-01 2.47777358e-01
-3.33293200e-01 -1.00015901e-01 5.35525262e-01 6.98243856e-01
-5.57114184e-01 -9.57524538e-01 -4.23156738e-01 5.94204426e-01
-1.95050531e-03 -1.81409478e-01 -5.15933752e-01 9.84502554e-01
8.58378112e-02 7.97619581e-01 2.34487236e-01 -7.50290155e-02
4.56910878e-01 -3.56576681e-01 4.06085521e-01 -9.82753575e-01
-3.83811183e-02 -1.25997275e-01 9.80447531e-02 -4.38011795e-01
-3.58384192e-01 3.22629176e-02 -1.53454256e+00 -1.03427239e-01
-3.88169587e-01 8.54877084e-02 8.40946734e-01 6.59413099e-01
2.13050678e-01 6.22184813e-01 8.64985213e-02 -7.31088042e-01
4.42234546e-01 -8.67908776e-01 -4.74773198e-01 7.52737164e-01
-2.18448624e-01 -7.57629871e-01 2.99077034e-01 4.34620887e-01] | [10.37491512298584, 1.4823120832443237] |
134394cb-6627-43fe-9772-4c26c56165e3 | palm-scaling-language-modeling-with-pathways-1 | 2204.02311 | null | https://arxiv.org/abs/2204.02311v5 | https://arxiv.org/pdf/2204.02311v5.pdf | PaLM: Scaling Language Modeling with Pathways | Large language models have been shown to achieve remarkable performance across a variety of natural language tasks using few-shot learning, which drastically reduces the number of task-specific training examples needed to adapt the model to a particular application. To further our understanding of the impact of scale on few-shot learning, we trained a 540-billion parameter, densely activated, Transformer language model, which we call Pathways Language Model PaLM. We trained PaLM on 6144 TPU v4 chips using Pathways, a new ML system which enables highly efficient training across multiple TPU Pods. We demonstrate continued benefits of scaling by achieving state-of-the-art few-shot learning results on hundreds of language understanding and generation benchmarks. On a number of these tasks, PaLM 540B achieves breakthrough performance, outperforming the finetuned state-of-the-art on a suite of multi-step reasoning tasks, and outperforming average human performance on the recently released BIG-bench benchmark. A significant number of BIG-bench tasks showed discontinuous improvements from model scale, meaning that performance steeply increased as we scaled to our largest model. PaLM also has strong capabilities in multilingual tasks and source code generation, which we demonstrate on a wide array of benchmarks. We additionally provide a comprehensive analysis on bias and toxicity, and study the extent of training data memorization with respect to model scale. Finally, we discuss the ethical considerations related to large language models and discuss potential mitigation strategies. | ['Noah Fiedel', 'Slav Petrov', 'Jeff Dean', 'Douglas Eck', 'Kathy Meier-Hellstern', 'Jason Wei', 'Michele Catasta', 'Orhan Firat', 'Mark Diaz', 'Brennan Saeta', 'Xuezhi Wang', 'Zongwei Zhou', 'Katherine Lee', 'Oleksandr Polozov', 'Rewon Child', 'Erica Moreira', 'Aitor Lewkowycz', 'Marie Pellat', 'Thanumalayan Sankaranarayana Pillai', 'Andrew M. Dai', 'Mark Omernick', 'Shivani Agrawal', 'David Dohan', 'Ryan Sepassi', 'Alexander Spiridonov', 'Barret Zoph', 'Hyeontaek Lim', 'David Luan', 'Daphne Ippolito', 'Denny Zhou', 'Liam Fedus', 'Kevin Robinson', 'Vedant Misra', 'Xavier Garcia', 'Henryk Michalewski', 'Sunipa Dev', 'Sanjay Ghemawat', 'Anselm Levskaya', 'Toju Duke', 'Pengcheng Yin', 'Guy Gur-Ari', 'Michael Isard', 'Jacob Austin', 'James Bradbury', 'Reiner Pope', 'Ben Hutchinson', 'Nan Du', 'Emily Reif', 'Vinodkumar Prabhakaran', 'Noam Shazeer', 'Yi Tay', 'Parker Barnes', 'Abhishek Rao', 'Joshua Maynez', 'Sasha Tsvyashchenko', 'Kensen Shi', 'Parker Schuh', 'Sebastian Gehrmann', 'Charles Sutton', 'Hyung Won Chung', 'Paul Barham', 'Adam Roberts', 'Gaurav Mishra', 'Maarten Bosma', 'Jacob Devlin', 'Sharan Narang', 'Aakanksha Chowdhery'] | 2022-04-05 | palm-scaling-language-modeling-with-pathways | https://storage.googleapis.com/pathways-language-model/PaLM-paper.pdf | https://storage.googleapis.com/pathways-language-model/PaLM-paper.pdf | google-research-2022-4 | ['multi-task-language-understanding', 'auto-debugging', 'known-unknowns', 'logic-grid-puzzle', 'hindu-knowledge', 'multiple-choice-qa', 'cross-lingual-question-answering', 'winowhy', 'strategyqa', 'novel-concepts'] | ['methodology', 'miscellaneous', 'miscellaneous', 'miscellaneous', 'miscellaneous', 'natural-language-processing', 'natural-language-processing', 'reasoning', 'reasoning', 'reasoning'] | [ 5.06676510e-02 1.36355637e-02 -3.48319530e-01 -1.80230886e-01
-1.21400595e+00 -5.61600447e-01 7.38347590e-01 6.87867776e-02
-4.05505151e-01 8.75557601e-01 3.21181476e-01 -4.21873391e-01
7.01257363e-02 -9.50211644e-01 -1.03737986e+00 -2.32740372e-01
8.74051377e-02 6.25864804e-01 2.13280976e-01 -5.73515654e-01
5.73535115e-02 -7.51291867e-03 -1.23443329e+00 4.65128183e-01
6.34755909e-01 4.72448021e-01 1.88746586e-01 4.97344434e-01
-2.75481880e-01 7.81415164e-01 -5.82847178e-01 -3.10858369e-01
2.63975328e-03 -2.85640866e-01 -6.95582151e-01 -4.90452766e-01
2.94543356e-01 -2.53655612e-01 -3.94616961e-01 5.62702119e-01
8.78022492e-01 2.26680651e-01 4.41180915e-01 -9.39291120e-01
-7.52320468e-01 9.65261161e-01 -4.23633039e-01 4.49297875e-01
1.14047378e-01 7.47595608e-01 1.08657026e+00 -8.60014260e-01
9.38596547e-01 1.13105357e+00 8.04160774e-01 9.39449370e-01
-1.43558085e+00 -1.07261753e+00 1.18146176e-02 -1.40133694e-01
-1.14450014e+00 -8.57603848e-01 7.45929480e-02 -4.20941770e-01
1.85283613e+00 -2.35807031e-01 4.40218896e-01 1.55701470e+00
5.49725235e-01 6.50564015e-01 1.04153097e+00 -2.84607798e-01
5.59403181e-01 -7.56028146e-02 4.63855684e-01 7.64335930e-01
3.04079920e-01 7.13552237e-02 -8.98704529e-01 -3.26932818e-01
3.99826109e-01 -2.90481925e-01 9.91929471e-02 -7.20297322e-02
-1.06334472e+00 8.10111761e-01 2.97677457e-01 3.98487031e-01
3.87593079e-03 6.50045514e-01 6.36760354e-01 2.23768830e-01
3.06933314e-01 9.33864355e-01 -6.17328286e-01 -2.06448913e-01
-8.92844856e-01 1.63197219e-01 5.89496076e-01 1.12244964e+00
5.89200079e-01 4.08919454e-01 -6.90690160e-01 9.61109936e-01
-5.51878922e-02 3.57627332e-01 6.74473584e-01 -7.80639291e-01
4.51254755e-01 3.12218249e-01 -1.38789341e-01 -5.59222847e-02
-4.11105692e-01 -5.21409631e-01 -5.18550813e-01 6.25902042e-02
2.11783141e-01 -3.22506726e-01 -1.18635559e+00 1.93405282e+00
-2.29850873e-01 7.04115927e-02 1.08307943e-01 3.35109353e-01
8.05371821e-01 7.73637712e-01 5.57735682e-01 2.20289856e-01
1.25768602e+00 -9.02350426e-01 -2.83859819e-01 -6.33721054e-01
8.23418558e-01 -4.59864885e-01 1.50333762e+00 4.25941110e-01
-9.74572361e-01 -4.02968466e-01 -1.27261019e+00 -2.06979126e-01
-5.57592034e-01 -2.99223036e-01 1.08820271e+00 7.56599009e-01
-9.32165265e-01 6.60671353e-01 -7.09334016e-01 -7.56861389e-01
6.34118259e-01 3.85596842e-01 -1.23207733e-01 -2.59883821e-01
-1.56846023e+00 9.20063257e-01 3.43694091e-01 -5.64816952e-01
-1.45272052e+00 -1.27114487e+00 -8.35656703e-01 3.62285823e-01
2.08408698e-01 -9.83672857e-01 1.34163094e+00 -1.62441179e-01
-1.56284177e+00 8.22544694e-01 6.61864877e-02 -8.01674545e-01
1.74720660e-01 -1.62748054e-01 -4.10331458e-01 -3.53656232e-01
8.14998671e-02 9.50965762e-01 5.09600520e-01 -6.32710397e-01
-2.97892779e-01 -1.20023293e-02 -7.66195208e-02 -2.42849246e-01
-4.98670042e-01 -2.40692161e-02 -3.89903128e-01 -4.83767062e-01
-5.51859319e-01 -7.28429794e-01 -1.68343037e-01 -1.93384439e-01
-5.46145849e-02 -2.26012498e-01 4.73013580e-01 -3.85290265e-01
1.16961145e+00 -1.86037457e+00 4.68835868e-02 -2.08467975e-01
1.18511647e-01 2.34402180e-01 -5.70717394e-01 5.95515132e-01
-3.43467519e-02 2.51178592e-01 -2.58201540e-01 -1.13019027e-01
8.32011923e-02 2.30857246e-02 -5.63065112e-01 8.83143544e-02
1.57260537e-01 1.36672342e+00 -9.81437564e-01 -2.00491235e-01
7.60377645e-02 2.81916261e-01 -6.19383693e-01 -7.46192783e-02
-5.29843032e-01 -1.96536668e-02 -2.63475597e-01 7.10429907e-01
2.25191772e-01 -4.14606959e-01 4.16980498e-02 1.74037486e-01
1.04148909e-01 3.09630215e-01 -4.22968537e-01 2.36592674e+00
-9.66902971e-01 4.32007790e-01 -2.62074679e-01 -4.63881344e-01
8.40859711e-01 1.95709646e-01 9.05223191e-02 -9.90691483e-01
-3.50524709e-02 1.57269329e-01 2.63643742e-01 -1.12935729e-01
2.66945213e-01 -4.24168676e-01 -6.07326448e-01 6.93394303e-01
6.36170805e-01 -2.87615508e-01 2.59972781e-01 5.19966722e-01
1.62269008e+00 3.69816646e-02 3.86169285e-01 -5.38215458e-01
-1.16587915e-01 -1.14724357e-02 4.85883594e-01 1.06301761e+00
-1.90429613e-01 3.91735494e-01 6.03657246e-01 -3.16247970e-01
-1.18065953e+00 -1.24007499e+00 -6.66183606e-02 1.53905928e+00
-3.69058728e-01 -6.46315277e-01 -6.24397516e-01 -4.41872060e-01
1.19519323e-01 1.34919441e+00 -6.83902383e-01 -5.79089284e-01
-1.48054749e-01 -1.18064940e+00 1.07282603e+00 6.88720167e-01
4.41490054e-01 -9.79395628e-01 -4.70452726e-01 3.21204662e-01
2.04707339e-01 -8.78381908e-01 -3.60005975e-01 5.31348526e-01
-8.03011119e-01 -7.86597371e-01 -6.62041008e-01 -5.05270839e-01
2.37621829e-01 -7.93937128e-03 1.26927245e+00 -1.37156948e-01
-5.41116655e-01 2.56813586e-01 -6.84879869e-02 -4.45175171e-01
-5.37323654e-01 5.28629899e-01 1.04013078e-01 -6.77189589e-01
3.19814056e-01 -5.76293528e-01 -2.49043033e-01 1.24344323e-02
-5.60763061e-01 -5.07931449e-02 7.93349147e-01 1.14026415e+00
2.81748444e-01 -3.56618971e-01 9.53814447e-01 -1.09000695e+00
9.20317531e-01 -6.68242335e-01 -4.85387623e-01 6.05335176e-01
-7.82937467e-01 4.68179882e-01 6.10009313e-01 -2.68206239e-01
-1.25577581e+00 -2.78086245e-01 -1.82904899e-01 -2.09910884e-01
2.75704443e-01 2.61313885e-01 3.95884216e-02 1.81342196e-02
1.25432348e+00 2.00168267e-01 -1.99789196e-01 -2.24465013e-01
5.48500359e-01 4.56501216e-01 2.05103084e-01 -8.79231513e-01
5.03755152e-01 1.13339663e-01 -2.58986801e-01 -7.33627558e-01
-7.46818542e-01 -2.10506290e-01 -2.39719555e-01 2.86241621e-01
7.08421290e-01 -1.13270462e+00 -4.30287570e-01 5.24298370e-01
-1.02510250e+00 -1.06479681e+00 -2.93315798e-01 -1.23677151e-02
-4.66713697e-01 -2.82306045e-01 -1.01206470e+00 -4.55308199e-01
-8.20254624e-01 -1.00984061e+00 1.02777326e+00 8.93783197e-02
-6.40389562e-01 -1.01017380e+00 3.16775501e-01 2.19788224e-01
7.09267855e-01 -1.88273638e-01 1.48789334e+00 -6.72949314e-01
-4.52995181e-01 1.08235188e-01 -1.39672473e-01 -1.50186559e-02
-2.31474578e-01 -3.02730024e-01 -1.13590789e+00 -3.42015564e-01
-3.42923135e-01 -9.80019152e-01 1.25778532e+00 1.79472834e-01
9.99635875e-01 2.69471705e-01 -6.21563971e-01 7.01073825e-01
1.26259947e+00 1.90705974e-02 6.16739750e-01 1.73917577e-01
3.80486995e-01 3.61766107e-02 3.73932004e-01 2.51707107e-01
-1.16997220e-01 4.51838672e-01 -1.62404194e-01 2.08007842e-01
-3.42760950e-01 -5.54908752e-01 4.80443150e-01 5.57062447e-01
3.91203672e-01 -2.65090346e-01 -1.26528704e+00 4.73353088e-01
-1.58263290e+00 -8.89928222e-01 4.45140898e-01 2.12857676e+00
8.52012157e-01 5.98353863e-01 -1.55559331e-01 -5.43779254e-01
4.23977017e-01 3.15155298e-01 -1.00915313e+00 -5.23423851e-01
-2.08805144e-01 7.94342995e-01 5.57206750e-01 3.63897085e-01
-6.53871894e-01 1.43028808e+00 8.13177395e+00 1.20322514e+00
-1.15980840e+00 5.75227141e-01 9.90134478e-01 -6.70385361e-01
-4.71695989e-01 -4.28276137e-02 -1.09209573e+00 1.88275918e-01
1.42866981e+00 -5.14493048e-01 6.33298397e-01 9.01264369e-01
-1.51595950e-01 -1.64118588e-01 -1.26673949e+00 6.99097395e-01
1.53144030e-02 -1.83107471e+00 5.96524589e-02 -4.98862341e-02
1.06934643e+00 7.29893506e-01 1.56933427e-01 1.11514318e+00
7.55543411e-01 -1.32920003e+00 4.18101370e-01 3.69370461e-01
1.14928520e+00 -7.24408627e-01 2.04667345e-01 2.50250369e-01
-7.92965949e-01 -2.83605278e-01 -5.62796593e-01 -1.25892073e-01
-1.74805559e-02 4.26617652e-01 -8.31957400e-01 -5.40224016e-02
5.08261442e-01 3.81884009e-01 -6.60497427e-01 6.67690933e-01
-2.35016972e-01 7.27410674e-01 -1.97470456e-01 -2.31597930e-01
2.69186914e-01 3.81559610e-01 1.78035185e-01 1.21681249e+00
2.99155980e-01 1.24540411e-01 -3.09815288e-01 1.29147363e+00
-4.91336673e-01 -1.33619100e-01 -7.94397116e-01 -3.82170439e-01
6.36336982e-01 1.29916084e+00 -5.00424743e-01 -5.15101910e-01
-4.30582166e-01 8.56428981e-01 6.34979725e-01 2.89877176e-01
-8.71916354e-01 -4.56320822e-01 6.26894116e-01 5.30564040e-02
-5.11951484e-02 -1.34747192e-01 -3.76514137e-01 -1.05188966e+00
-4.61606592e-01 -9.45370197e-01 4.40219998e-01 -8.45794141e-01
-1.16520274e+00 4.26981300e-01 -1.32799253e-01 -7.40008235e-01
-6.54052943e-02 -6.05097890e-01 -9.77746010e-01 8.79385352e-01
-1.01674974e+00 -1.08733845e+00 -6.24959245e-02 1.43667698e-01
7.90286005e-01 -5.53763509e-01 1.11469865e+00 2.08552212e-01
-7.51618087e-01 8.65333974e-01 2.16638476e-01 -1.27596438e-01
1.05850446e+00 -7.74342000e-01 1.15347922e+00 6.48215711e-01
-4.79414091e-02 1.03299057e+00 5.07161438e-01 -8.06807280e-01
-1.62614393e+00 -1.16645169e+00 6.44498944e-01 -7.19376564e-01
8.46620619e-01 -8.77339721e-01 -7.86261797e-01 9.58549619e-01
2.34531745e-01 -4.45445534e-03 8.26921403e-01 5.24415970e-01
-6.33552432e-01 -9.53008607e-02 -1.00226045e+00 6.58251226e-01
1.31833434e+00 -7.31562495e-01 -5.77833295e-01 5.18073797e-01
9.70519066e-01 -2.19998002e-01 -7.54404068e-01 1.50335312e-01
6.19100809e-01 -6.41876221e-01 8.30952346e-01 -9.64117885e-01
6.59735501e-01 3.12876791e-01 -1.32404476e-01 -1.47401023e+00
-6.66180968e-01 -6.03023589e-01 -1.43147409e-01 1.18421626e+00
8.15152287e-01 -5.18168747e-01 6.79341912e-01 6.73205435e-01
-1.22671284e-01 -8.30921948e-01 -8.10479879e-01 -8.71169508e-01
5.53362668e-01 -3.24460655e-01 5.22726119e-01 7.71736860e-01
3.77218783e-01 8.53952527e-01 -3.82614672e-01 -5.47065735e-01
4.93602186e-01 -3.23839426e-01 6.39198422e-01 -7.08531499e-01
-8.70291114e-01 -4.33971196e-01 2.02035401e-02 -5.95999599e-01
3.56651247e-01 -1.33893514e+00 -9.41491053e-02 -1.49027288e+00
5.46793699e-01 -3.20614457e-01 -4.48301643e-01 6.24015808e-01
-2.14663833e-01 1.29081188e-02 2.01689526e-01 7.35910684e-02
-7.31685638e-01 6.57772601e-01 8.37913215e-01 -5.20007133e-01
-1.93359017e-01 -3.67650986e-01 -1.13677752e+00 2.86520213e-01
7.32586682e-01 -3.89664292e-01 -6.73103571e-01 -6.55713916e-01
3.26470643e-01 -1.02153800e-01 3.54443453e-02 -1.33877409e+00
1.65854335e-01 -4.66163680e-02 5.49970210e-01 -1.12916477e-01
3.85896027e-01 -1.30402250e-02 -4.20567058e-02 7.33830035e-01
-5.53686142e-01 -2.12459609e-01 7.64487982e-01 3.97671491e-01
4.16527063e-01 -2.12713614e-01 8.61153483e-01 -3.37720007e-01
-1.14391434e+00 2.79478252e-01 -5.08815885e-01 3.55275720e-01
1.09738541e+00 3.34020227e-01 -7.81944215e-01 -1.02631427e-01
-4.84261304e-01 4.74982232e-01 4.26475644e-01 5.84807634e-01
3.54822546e-01 -1.13031816e+00 -5.36114633e-01 8.13507587e-02
4.61293370e-01 -3.92932206e-01 4.30793643e-01 5.74937165e-01
-3.78686190e-01 6.89943194e-01 -3.87424797e-01 -1.95169210e-01
-7.39730537e-01 6.39803052e-01 5.08650959e-01 -4.98812109e-01
-4.70147759e-01 1.03907275e+00 2.88443714e-01 -5.57925165e-01
-8.45039412e-02 -1.94441110e-01 4.05869007e-01 -1.46083072e-01
6.96480215e-01 4.00351495e-01 2.00588807e-01 2.00211152e-01
-3.10224205e-01 1.84645459e-01 -4.22782153e-01 -2.59333253e-01
1.22982156e+00 3.78780305e-01 1.15517996e-01 6.50027215e-01
9.75144088e-01 -3.71265680e-01 -1.29793060e+00 -3.56972367e-02
-1.81512237e-01 2.21216589e-01 -3.59099694e-02 -1.18319809e+00
-5.40202856e-01 1.14403379e+00 4.68132973e-01 -6.83568597e-01
5.99757910e-01 5.69549203e-02 1.05027294e+00 7.60905445e-01
7.86925197e-01 -1.34776080e+00 2.80131400e-01 8.10414732e-01
6.75950348e-01 -1.10418177e+00 6.99660778e-02 1.13449126e-01
-5.92803955e-01 7.91527867e-01 9.24438357e-01 -3.18898372e-02
3.70065510e-01 5.29064894e-01 -2.44939864e-01 -1.96188346e-01
-1.41393244e+00 1.82842910e-01 -1.82636380e-01 6.97329521e-01
5.97782314e-01 4.85491790e-02 -1.42345667e-01 9.12308574e-01
-2.05934748e-01 2.59606600e-01 3.36331278e-01 8.13212037e-01
-6.95353687e-01 -1.15707088e+00 -1.25861958e-01 7.36957371e-01
-1.15877256e-01 -5.56150854e-01 -2.91588515e-01 6.60226285e-01
-1.27749860e-01 6.32748008e-01 3.46906036e-02 -4.72974062e-01
2.01378569e-01 5.61522365e-01 6.88763320e-01 -1.03402281e+00
-7.61222899e-01 -3.11207175e-01 2.71190494e-01 -5.31433403e-01
5.41878343e-01 -4.15901959e-01 -1.39108407e+00 -5.50154388e-01
3.54638174e-02 -1.35838345e-01 3.69660586e-01 8.42283249e-01
5.96948385e-01 8.37545991e-01 -4.76177856e-02 -6.17444932e-01
-8.17897975e-01 -1.01978981e+00 -4.23039585e-01 1.43965676e-01
-2.96920776e-01 -4.40470457e-01 3.77210006e-02 -3.76873553e-01] | [10.752700805664062, 8.132765769958496] |
8b15db5d-089f-4f28-8f74-4c43198fff6c | joint-system-wise-optimization-for-pipeline | 2106.04835 | null | https://arxiv.org/abs/2106.04835v1 | https://arxiv.org/pdf/2106.04835v1.pdf | Joint System-Wise Optimization for Pipeline Goal-Oriented Dialog System | Recent work (Takanobu et al., 2020) proposed the system-wise evaluation on dialog systems and found that improvement on individual components (e.g., NLU, policy) in prior work may not necessarily bring benefit to pipeline systems in system-wise evaluation. To improve the system-wise performance, in this paper, we propose new joint system-wise optimization techniques for the pipeline dialog system. First, we propose a new data augmentation approach which automates the labeling process for NLU training. Second, we propose a novel stochastic policy parameterization with Poisson distribution that enables better exploration and offers a principled way to compute policy gradient. Third, we propose a reward bonus to help policy explore successful dialogs. Our approaches outperform the competitive pipeline systems from Takanobu et al. (2020) by big margins of 12% success rate in automatic system-wise evaluation and of 16% success rate in human evaluation on the standard multi-domain benchmark dataset MultiWOZ 2.1, and also outperform the recent state-of-the-art end-to-end trained model from DSTC9. | ['Tengyu Ma', 'Xiaodong He', 'BoWen Zhou', 'Jing Huang', 'Zichuan Lin'] | 2021-06-09 | null | null | null | null | ['goal-oriented-dialog'] | ['natural-language-processing'] | [-2.47561589e-01 3.53697658e-01 -1.83058053e-01 -5.19051909e-01
-1.08388853e+00 -9.19727087e-01 9.14456189e-01 -2.53977537e-01
-7.95670986e-01 7.91733444e-01 3.96784097e-01 -5.79441428e-01
2.08396778e-01 -1.74487770e-01 -3.19022357e-01 -3.65328461e-01
3.32988232e-01 1.03974378e+00 5.22931218e-01 -5.05661845e-01
2.29723886e-01 2.07627982e-01 -9.04515684e-01 1.84947208e-01
8.03453207e-01 8.37871909e-01 3.81175697e-01 9.61225569e-01
-1.53006956e-01 7.28605449e-01 -8.14655840e-01 -4.23038274e-01
1.27841771e-01 -1.81696251e-01 -1.19494700e+00 -2.89048459e-02
4.16539222e-01 -6.48390949e-01 -1.51780874e-01 7.92724192e-01
8.45275640e-01 4.16038394e-01 4.38024580e-01 -9.81970787e-01
-3.96066457e-01 1.01127934e+00 -2.20697507e-01 -1.25025362e-01
8.54984298e-02 7.12086380e-01 1.16207039e+00 -7.34714270e-01
5.71101606e-01 1.67100394e+00 1.81575000e-01 1.09929025e+00
-1.03897119e+00 -3.40678781e-01 3.53888065e-01 -1.04688935e-01
-8.14944804e-01 -6.76203609e-01 3.94740939e-01 -3.81725401e-01
1.31097996e+00 1.62018687e-01 9.59764719e-02 1.31878853e+00
-1.75897911e-01 1.20591414e+00 1.08419812e+00 -3.42704445e-01
2.56293833e-01 4.11335409e-01 1.45649686e-01 6.52733684e-01
-6.41333938e-01 7.41842166e-02 -4.59121883e-01 -2.23333955e-01
3.87613982e-01 -6.36473179e-01 2.63541117e-02 1.28336430e-01
-1.12832010e+00 8.37433577e-01 1.94316387e-01 2.67688215e-01
-2.22465232e-01 1.04823321e-01 4.37853992e-01 3.72386836e-02
4.48736161e-01 1.01337087e+00 -7.79199243e-01 -6.84920847e-01
-7.00895607e-01 5.88213563e-01 1.10917795e+00 9.64358926e-01
2.99671203e-01 -1.23875126e-01 -1.06671071e+00 9.57028508e-01
4.59310442e-01 4.25380498e-01 4.82349902e-01 -1.42058551e+00
5.36591053e-01 3.51726949e-01 4.93975312e-01 6.23849109e-02
-3.52476835e-01 -1.98081378e-02 -3.66765290e-01 1.47169143e-01
7.12347686e-01 -5.64858079e-01 -8.12496662e-01 1.72346044e+00
1.52812809e-01 -4.91919369e-02 2.58758247e-01 7.87773609e-01
6.66821420e-01 6.15782678e-01 3.58381689e-01 -3.74572724e-02
1.47842360e+00 -1.45282269e+00 -7.65340149e-01 -2.27577373e-01
5.82830787e-01 -9.88217711e-01 1.52280593e+00 4.49260890e-01
-1.35498500e+00 -6.03716135e-01 -5.88518202e-01 4.24330682e-03
-1.90498590e-01 5.26603222e-01 5.51396430e-01 5.80242991e-01
-1.35140264e+00 6.43563986e-01 -7.70592213e-01 -3.42393190e-01
3.73033322e-02 3.27587724e-01 1.41448185e-01 6.37965679e-01
-1.38501644e+00 1.28665137e+00 4.91039455e-01 -2.33771876e-01
-1.14242530e+00 -5.82024336e-01 -5.20809829e-01 2.17799112e-01
7.86887407e-01 -4.12651002e-01 2.36761093e+00 1.40650840e-02
-2.30222917e+00 5.81898153e-01 -2.00893700e-01 -8.26927364e-01
7.51370847e-01 -6.91307187e-01 4.96367812e-02 -5.49314432e-02
-1.79427683e-01 1.27776861e+00 4.99422610e-01 -1.02425528e+00
-9.63974535e-01 2.60106713e-01 3.37912977e-01 3.51595342e-01
-4.58671510e-01 1.80814326e-01 -5.51657557e-01 -1.34826705e-01
-6.84617639e-01 -1.08804786e+00 -5.79033971e-01 -5.61832964e-01
-6.05489433e-01 -9.66505945e-01 8.87101829e-01 -6.61644936e-01
1.36156392e+00 -1.80379081e+00 1.47540018e-01 -4.26771581e-01
1.94320977e-01 7.32763946e-01 4.03095149e-02 2.32583746e-01
5.94854593e-01 2.59925276e-01 -3.02081257e-01 -1.08820236e+00
3.00296366e-01 1.53804943e-01 -4.20835406e-01 -1.82556152e-01
5.52145898e-01 9.93651152e-01 -1.08268142e+00 -5.34541845e-01
4.67443109e-01 -3.13997753e-02 -6.34758711e-01 7.32227385e-01
-9.95552719e-01 7.61541188e-01 -5.01430571e-01 5.34980595e-01
2.28767991e-01 -1.70384720e-01 -5.74414507e-02 1.77566722e-01
-4.90392923e-01 6.85966492e-01 -5.23502290e-01 1.97096694e+00
-6.31226718e-01 5.37488759e-01 3.77867892e-02 -2.44854301e-01
9.09442246e-01 5.92732131e-01 2.97687620e-01 -2.61180848e-01
2.37752244e-01 7.38040954e-02 2.01021463e-01 -1.93120867e-01
9.75699246e-01 3.12910229e-01 -3.25986296e-01 3.16335261e-01
3.08375746e-01 -2.52257258e-01 3.01552206e-01 4.44018096e-01
9.38120842e-01 4.48316216e-01 1.57964753e-03 -4.04587150e-01
6.15429580e-01 4.60402779e-02 4.49093819e-01 9.35226679e-01
-6.31822944e-01 5.15212834e-01 7.66631007e-01 3.26235741e-02
-9.93317902e-01 -7.25979745e-01 8.79304633e-02 1.39842188e+00
-1.80963412e-01 -4.85296160e-01 -1.04372346e+00 -1.18677473e+00
-7.15494379e-02 1.08907843e+00 -4.01048630e-01 3.58432047e-02
-7.18782783e-01 -6.64426804e-01 7.71598458e-01 5.34701765e-01
7.62988031e-01 -1.31440353e+00 -5.69885969e-01 4.79985118e-01
-2.12316975e-01 -1.59459734e+00 -6.03132486e-01 5.40296882e-02
-5.84736288e-01 -5.62471092e-01 -9.04402435e-01 -2.77722895e-01
3.67583968e-02 -9.94766131e-02 1.21323454e+00 -3.96683067e-01
2.76447415e-01 3.34085226e-01 -4.39093888e-01 -3.51018757e-01
-7.88422287e-01 3.96897823e-01 2.03947853e-02 -4.41272974e-01
2.17269227e-01 1.06614949e-02 -5.95297635e-01 4.98545974e-01
-3.09073597e-01 2.67870110e-02 5.66822767e-01 8.78183007e-01
2.30955169e-01 -6.00236356e-01 6.58116102e-01 -8.76552641e-01
1.22987580e+00 -1.39121816e-01 -7.30355680e-01 5.17155170e-01
-9.34361756e-01 4.36664045e-01 5.04990339e-01 -4.36829865e-01
-1.54764140e+00 9.86027569e-02 -4.31196660e-01 -3.86375338e-01
-1.82778671e-01 -4.23504598e-02 -1.00961074e-01 2.66139358e-01
7.34191358e-01 -1.17164381e-01 -1.58668086e-01 -8.39123428e-01
8.39921653e-01 7.85619080e-01 4.99977767e-01 -8.87807965e-01
3.14074934e-01 -4.86137643e-02 -5.84742069e-01 -4.00117040e-01
-7.53686905e-01 -6.16404593e-01 -4.16667640e-01 -1.29538745e-01
1.17258716e+00 -8.30344021e-01 -1.01227295e+00 3.27436358e-01
-1.50714481e+00 -7.75155783e-01 -1.85593978e-01 2.24925295e-01
-5.45321465e-01 2.80959427e-01 -6.56870544e-01 -1.19766128e+00
-7.43656814e-01 -1.57543576e+00 1.36304879e+00 4.68493164e-01
-4.92398858e-01 -9.30085838e-01 1.77667409e-01 2.97368973e-01
6.27102017e-01 -3.06130975e-01 4.88791376e-01 -1.04754555e+00
-3.03455472e-01 1.17090292e-01 -1.13527693e-01 6.14329457e-01
-8.18165019e-02 1.09381765e-01 -1.22388363e+00 -2.20933959e-01
-1.05034813e-01 -5.47871649e-01 8.12790036e-01 3.19427013e-01
9.37404573e-01 -2.42085442e-01 -2.65686274e-01 8.79957452e-02
6.75162137e-01 4.21088874e-01 2.37411499e-01 1.87360197e-01
6.58757329e-01 7.88285255e-01 9.40137625e-01 3.98868173e-01
5.29127598e-01 9.09859061e-01 2.24761605e-01 9.77952853e-02
-1.78174883e-01 -1.85398430e-01 7.28917778e-01 4.86757129e-01
1.62255764e-01 -4.86884415e-01 -8.44889522e-01 5.16453087e-01
-2.20361733e+00 -4.73674208e-01 2.49539718e-01 2.00990033e+00
1.01309347e+00 4.98367906e-01 5.35998046e-01 -6.83357418e-01
4.76594090e-01 1.03713483e-01 -6.25889659e-01 -5.04396558e-01
2.99944561e-02 3.24605890e-02 4.50666815e-01 1.04000747e+00
-1.06039381e+00 1.81329966e+00 5.79151773e+00 8.95714641e-01
-8.27306867e-01 3.00927877e-01 9.09656048e-01 -4.88994941e-02
-6.19158754e-03 2.97708750e-01 -1.60935247e+00 3.87099296e-01
1.35655940e+00 -1.08553037e-01 2.62560844e-01 1.01166463e+00
2.43326381e-01 -2.25062966e-01 -1.06580961e+00 4.36556041e-01
-4.40651476e-01 -1.16159689e+00 -3.68890077e-01 3.22919227e-02
5.24074316e-01 1.71813563e-01 9.80568603e-02 8.33404660e-01
1.08710861e+00 -7.70431876e-01 5.41473329e-01 3.12143236e-01
4.27975595e-01 -4.15333152e-01 6.55645609e-01 4.50706303e-01
-6.83368862e-01 1.07824035e-01 -2.50407189e-01 1.74839452e-01
6.74975693e-01 1.34374619e-01 -1.76653492e+00 3.47856730e-01
4.96985108e-01 3.62022191e-01 -5.45188963e-01 5.83568394e-01
-4.99442250e-01 7.74182677e-01 -2.83037454e-01 -4.28324401e-01
6.52063668e-01 5.27951866e-02 6.96385503e-01 1.51429975e+00
1.47563703e-02 -1.48060337e-01 4.23249692e-01 8.04738820e-01
-1.20638490e-01 -1.88234493e-01 -2.64005899e-01 -1.70699000e-01
6.03066444e-01 1.55904567e+00 -2.69855261e-01 -5.95556796e-01
-4.27556559e-02 8.44498873e-01 3.86901140e-01 2.15021044e-01
-6.98073387e-01 1.12629764e-01 5.75541377e-01 -3.16844374e-01
-2.66299257e-03 -2.77023435e-01 -2.93413192e-01 -8.69702041e-01
-2.72020549e-01 -8.45215440e-01 2.03095868e-01 -3.85929763e-01
-1.09822500e+00 9.79054034e-01 3.51759613e-01 -1.03274965e+00
-8.36452663e-01 -7.13273942e-01 -5.81288040e-01 1.10284972e+00
-1.38872039e+00 -1.12662911e+00 1.38348520e-01 2.83232063e-01
9.92656291e-01 -4.86695975e-01 8.93846571e-01 -1.17208649e-04
-6.80710375e-01 7.52997339e-01 -1.38493955e-01 7.78848827e-02
1.19161701e+00 -1.84442461e+00 9.89150286e-01 6.87979102e-01
-1.94242522e-01 6.72991693e-01 8.55580270e-01 -5.94778419e-01
-9.46806014e-01 -8.41038585e-01 8.47521961e-01 -8.87444556e-01
7.75583923e-01 -3.85845780e-01 -7.39941478e-01 4.80701625e-01
8.08049738e-01 -5.94752550e-01 3.16113561e-01 3.00235838e-01
5.60052879e-02 2.90405184e-01 -1.18576503e+00 8.08569193e-01
6.63659811e-01 -2.47068226e-01 -5.48180640e-01 5.58971345e-01
1.33216441e+00 -8.29491675e-01 -8.80982101e-01 4.96501029e-01
4.09585446e-01 -7.14940727e-01 7.20218480e-01 -7.15524614e-01
3.77160043e-01 -2.38369545e-03 -4.93477993e-02 -1.36744416e+00
8.61646980e-03 -1.22982359e+00 -4.81546164e-01 1.42544103e+00
7.88291335e-01 -4.26523536e-01 6.90767467e-01 8.73716593e-01
-4.66543764e-01 -9.18943346e-01 -7.26699054e-01 -7.92623222e-01
3.43192339e-01 -1.50024608e-01 3.42703581e-01 4.39477444e-01
5.69529459e-02 7.34086514e-01 -6.49001896e-01 -2.74094999e-01
2.48952717e-01 -2.78945535e-01 8.07728171e-01 -8.79950404e-01
-6.04536653e-01 -8.46187532e-01 5.71177721e-01 -1.57838559e+00
2.47984782e-01 -4.62382048e-01 4.21915919e-01 -1.40443838e+00
-1.38670906e-01 -5.12988567e-01 -2.77584583e-01 5.49700320e-01
-5.11230171e-01 -5.65245509e-01 5.45071959e-01 2.85284191e-01
-8.20288897e-01 7.23640978e-01 1.30807722e+00 7.60706440e-02
-5.84839523e-01 2.63627857e-01 -7.91011274e-01 5.61813712e-01
9.25025105e-01 -2.57693380e-01 -4.12180573e-01 -3.90047491e-01
-3.73861194e-01 1.52125493e-01 4.35700454e-02 -7.24227488e-01
2.97886878e-01 -8.91067609e-02 -2.70200044e-01 -6.28001869e-01
5.09495080e-01 -2.46040538e-01 -7.40731418e-01 4.09404963e-01
-6.93713427e-01 9.82290208e-02 4.24531013e-01 3.63136113e-01
-4.51838970e-03 -4.22982752e-01 6.61745727e-01 -1.58574790e-01
-8.22083116e-01 1.19967222e-01 -1.90377921e-01 -5.81928268e-02
7.58332729e-01 5.79119861e-01 -6.81158006e-01 -4.38125849e-01
-5.94696820e-01 8.54643106e-01 5.90401888e-02 5.68674922e-01
1.69180095e-01 -8.03322196e-01 -8.43986094e-01 -5.16815662e-01
-1.02429844e-01 6.02837317e-02 -5.22037502e-03 6.88234210e-01
-1.31921768e-01 7.03438640e-01 6.74039274e-02 -5.53733706e-01
-1.12752676e+00 -6.46535978e-02 2.46722385e-01 -1.01158726e+00
-2.25290939e-01 1.34589541e+00 1.84035189e-02 -6.22192204e-01
4.82502073e-01 -3.31785351e-01 -3.07334691e-01 3.25576775e-02
4.06488985e-01 1.94538459e-01 -1.21262781e-01 -1.73573270e-01
-2.58731633e-01 -3.40455659e-02 -7.39429057e-01 -9.89800215e-01
8.96909297e-01 1.29685581e-01 3.63655627e-01 3.83856148e-01
6.71463370e-01 -2.20048159e-01 -1.82272124e+00 -4.68265563e-01
3.38351250e-01 -9.54050571e-02 -6.97911307e-02 -1.46147060e+00
-3.75235379e-01 1.11551178e+00 5.59460223e-01 3.28691393e-01
8.04725528e-01 5.05516864e-02 9.03739810e-01 6.65628076e-01
1.41279936e-01 -1.21337891e+00 1.88475937e-01 1.10667098e+00
9.07618225e-01 -1.69160140e+00 -2.54776418e-01 -6.81161582e-02
-1.29874742e+00 9.29712474e-01 1.07548070e+00 1.59173340e-01
1.77678123e-01 1.46562770e-01 1.12438574e-01 1.40324488e-01
-1.19909513e+00 -4.20534164e-01 2.99874216e-01 2.65434355e-01
6.80429161e-01 1.67477578e-01 -4.26206112e-01 6.14680231e-01
-2.06121728e-01 -2.93875724e-01 1.42691001e-01 7.99223781e-01
-6.98525488e-01 -1.62317324e+00 -4.89071123e-02 3.11448965e-02
-3.51206511e-01 -2.20448866e-01 -6.88447356e-01 6.78960264e-01
-3.63802671e-01 1.17109513e+00 -2.55332500e-01 -5.80418527e-01
5.65571904e-01 3.62341106e-01 2.03758270e-01 -7.41831362e-01
-1.12095785e+00 4.33961332e-01 6.03724301e-01 -4.77754116e-01
-9.58669335e-02 -6.45005882e-01 -1.25623977e+00 -4.26654704e-02
-2.40825251e-01 3.22707295e-01 7.17462838e-01 1.06993151e+00
5.02358556e-01 7.43582428e-01 5.54561496e-01 -7.98356831e-01
-1.20435607e+00 -1.69737601e+00 1.12247683e-01 9.14823934e-02
7.56377280e-02 -6.40058815e-01 -7.93220624e-02 -6.66592419e-02] | [12.774024963378906, 8.00940990447998] |
bd5e63a1-6b6c-487d-bbdc-6b36d2610f64 | improving-selective-visual-question-answering-1 | 2306.08751 | null | https://arxiv.org/abs/2306.08751v1 | https://arxiv.org/pdf/2306.08751v1.pdf | Improving Selective Visual Question Answering by Learning from Your Peers | Despite advances in Visual Question Answering (VQA), the ability of models to assess their own correctness remains underexplored. Recent work has shown that VQA models, out-of-the-box, can have difficulties abstaining from answering when they are wrong. The option to abstain, also called Selective Prediction, is highly relevant when deploying systems to users who must trust the system's output (e.g., VQA assistants for users with visual impairments). For such scenarios, abstention can be especially important as users may provide out-of-distribution (OOD) or adversarial inputs that make incorrect answers more likely. In this work, we explore Selective VQA in both in-distribution (ID) and OOD scenarios, where models are presented with mixtures of ID and OOD data. The goal is to maximize the number of questions answered while minimizing the risk of error on those questions. We propose a simple yet effective Learning from Your Peers (LYP) approach for training multimodal selection functions for making abstention decisions. Our approach uses predictions from models trained on distinct subsets of the training data as targets for optimizing a Selective VQA model. It does not require additional manual labels or held-out data and provides a signal for identifying examples that are easy/difficult to generalize to. In our extensive evaluations, we show this benefits a number of models across different architectures and scales. Overall, for ID, we reach 32.92% in the selective prediction metric coverage at 1% risk of error (C@1%) which doubles the previous best coverage of 15.79% on this task. For mixed ID/OOD, using models' softmax confidences for abstention decisions performs very poorly, answering <5% of questions at 1% risk of error even when faced with only 10% OOD examples, but a learned selection function with LYP can increase that to 25.38% C@1%. | ['Marcus Rohrbach', 'Matthieu Cord', 'Xinlei Chen', 'Stefan Scherer', 'Ramakrishna Vedantam', 'Rishabh Maheshwary', 'Spencer Whitehead', 'Corentin Dancette'] | 2023-06-14 | improving-selective-visual-question-answering | http://openaccess.thecvf.com//content/CVPR2023/html/Dancette_Improving_Selective_Visual_Question_Answering_by_Learning_From_Your_Peers_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Dancette_Improving_Selective_Visual_Question_Answering_by_Learning_From_Your_Peers_CVPR_2023_paper.pdf | cvpr-2023-1 | ['visual-question-answering-1', 'question-answering'] | ['computer-vision', 'natural-language-processing'] | [ 5.02831023e-03 6.44139051e-01 1.37029037e-01 -6.77200258e-01
-1.19376361e+00 -8.44419718e-01 2.00481847e-01 -1.76722482e-02
-3.65974456e-01 6.52009189e-01 -9.22735706e-02 -6.98440135e-01
9.47562754e-02 -4.47582126e-01 -8.13072145e-01 -1.75043374e-01
3.01103801e-01 6.66834772e-01 2.77462751e-01 -2.65257955e-01
-2.77587473e-01 3.04868698e-01 -1.68876350e+00 7.74923265e-01
1.32904899e+00 1.03505611e+00 -5.89685813e-02 1.07386065e+00
-2.32227117e-01 1.04252708e+00 -1.24024713e+00 -8.82666647e-01
1.26351580e-01 -3.53114963e-01 -8.51817548e-01 -2.23052531e-01
1.25003350e+00 -7.09400654e-01 -5.16041834e-03 8.51359725e-01
6.86298668e-01 -4.32893373e-02 7.32462168e-01 -1.60193825e+00
-9.19938624e-01 1.14914514e-01 -1.08390003e-01 -1.32494509e-01
6.43448412e-01 8.03582251e-01 1.10043061e+00 -9.11102653e-01
3.72248083e-01 1.44396245e+00 4.87279654e-01 1.09321976e+00
-1.51776743e+00 -6.44776344e-01 3.34643602e-01 1.46982297e-01
-1.10296953e+00 -4.30720806e-01 3.19595367e-01 -4.41823572e-01
1.06464934e+00 6.04390144e-01 2.49150574e-01 1.54055214e+00
-7.00567216e-02 1.11411297e+00 9.49843109e-01 -2.26406783e-01
3.69659722e-01 7.22290933e-01 1.39928088e-01 5.15060306e-01
-1.24550916e-01 -5.29377535e-02 -4.84618694e-01 -4.59902138e-01
1.26465475e-02 -4.85608667e-01 -3.78323257e-01 -2.50207514e-01
-4.09051627e-01 8.10801923e-01 5.69870889e-01 -2.79897004e-01
-2.02684617e-03 7.56762922e-02 -6.31522685e-02 7.02360451e-01
1.24767840e-01 5.59514523e-01 -4.51267302e-01 -7.09503517e-02
-6.63154364e-01 6.08723581e-01 8.42445076e-01 9.88265038e-01
5.64897656e-01 1.62920430e-01 -7.55490601e-01 1.01685405e+00
3.76609772e-01 7.82400906e-01 9.71408114e-02 -1.20277596e+00
5.54410636e-01 6.49746358e-01 4.86051649e-01 -4.68710482e-01
-2.81551123e-01 -3.03630352e-01 -3.59088033e-01 9.36892331e-01
9.16661620e-01 -3.90801728e-01 -1.42665184e+00 1.97742462e+00
2.73376286e-01 -4.38815415e-01 2.43782122e-02 8.64012957e-01
8.17289770e-01 6.15358531e-01 4.69583392e-01 7.71329254e-02
1.24270904e+00 -7.22661018e-01 -5.25150955e-01 -5.45582354e-01
5.57387173e-01 -5.74137509e-01 1.79699540e+00 6.04510546e-01
-1.04652894e+00 -5.01109898e-01 -8.49180579e-01 -1.74585372e-01
-1.64622515e-01 1.01825938e-01 -1.35124964e-03 1.12430203e+00
-1.34343696e+00 1.26218140e-01 -4.04004842e-01 -1.37477964e-01
6.11732185e-01 5.15844405e-01 -2.11789921e-01 -1.25210792e-01
-1.09215450e+00 1.08374119e+00 -3.70184153e-01 -1.40639201e-01
-1.26814616e+00 -7.97838867e-01 -6.78870618e-01 2.57836699e-01
2.92751521e-01 -6.57211125e-01 1.62894714e+00 -1.35210752e+00
-1.07595921e+00 7.82437325e-01 -2.94810116e-01 -4.18288678e-01
9.38878298e-01 -5.17758131e-01 -4.53647286e-01 1.30529568e-01
2.77289953e-02 9.44338560e-01 1.30290687e+00 -1.48896098e+00
-5.43562055e-01 -2.97646284e-01 4.45681542e-01 2.35018641e-01
-2.12576643e-01 -1.68578550e-01 -1.62723690e-01 -5.54884262e-02
-3.96778792e-01 -7.81287193e-01 1.21736407e-01 4.74017084e-01
-5.09646118e-01 -4.73319411e-01 8.67226243e-01 -7.74528146e-01
9.45683658e-01 -2.04451513e+00 -1.79480642e-01 2.51943588e-01
3.89363080e-01 4.45039004e-01 -2.83341140e-01 4.83647883e-02
1.15236737e-01 4.72954541e-01 -1.75661951e-01 -4.83944803e-01
2.07527414e-01 1.09702744e-01 -5.94692349e-01 -9.00259838e-02
6.36414707e-01 7.32240319e-01 -6.46731615e-01 -3.44566107e-01
-8.56629945e-03 3.61007571e-01 -6.66233838e-01 6.15865767e-01
-5.58331251e-01 2.87099063e-01 -7.99387023e-02 6.75825357e-01
5.99298179e-01 -2.83189625e-01 2.21956614e-02 9.90654379e-02
4.24239933e-01 1.58449382e-01 -8.86207879e-01 7.95878887e-01
-4.54301894e-01 6.24275625e-01 2.93758899e-01 -3.85402620e-01
7.21810341e-01 2.37618878e-01 -3.48173827e-01 -7.67087817e-01
-5.41144945e-02 2.37087682e-01 8.02151039e-02 -6.61166310e-01
2.14123994e-01 4.31779735e-02 7.49406889e-02 6.56996667e-02
1.99363157e-01 -2.02830657e-01 -2.55059034e-01 5.42194068e-01
1.34051526e+00 -4.67480347e-02 -1.19001202e-01 2.20052257e-01
3.43171358e-01 -8.24772716e-02 1.87898070e-01 1.24810386e+00
-6.28877640e-01 8.61609936e-01 8.79333198e-01 -7.49730542e-02
-8.83606255e-01 -1.50432038e+00 -2.70723775e-02 1.24682093e+00
-5.39015383e-02 -1.71583980e-01 -7.44893491e-01 -1.19132257e+00
2.36497313e-01 1.37161314e+00 -5.97414732e-01 -3.04325849e-01
-2.98564285e-01 -6.86523542e-02 6.98463857e-01 3.70410025e-01
2.54224181e-01 -1.02315640e+00 -4.33302075e-01 -2.04076305e-01
-2.17152447e-01 -8.07881296e-01 -2.62371868e-01 -4.88513969e-02
-4.94538188e-01 -9.75329936e-01 -8.30944479e-01 -4.02815491e-01
4.91226554e-01 -1.29661232e-01 1.31826484e+00 1.75594181e-01
6.84931502e-02 8.95766854e-01 -2.32765391e-01 -5.45880735e-01
-6.40363514e-01 -1.12571694e-01 -1.25625074e-01 -2.98014861e-02
3.98904800e-01 -2.02893659e-01 -6.50222957e-01 4.83985662e-01
-6.84162676e-01 -4.47962582e-01 4.77002800e-01 8.37371588e-01
3.84342462e-01 -6.18725479e-01 8.51049304e-01 -1.07807553e+00
9.43541408e-01 -6.56013727e-01 -4.17886585e-01 4.87799048e-01
-6.85540557e-01 -5.43925948e-02 7.92212427e-01 -7.30765879e-01
-1.07538581e+00 -1.75806075e-01 -4.44240242e-01 -6.01201653e-01
-4.21205163e-01 -3.27116847e-02 -2.55463868e-01 4.80913781e-02
1.27223587e+00 -1.67484969e-01 -9.28471610e-03 -2.92940229e-01
4.49300885e-01 8.87308717e-01 2.74382323e-01 -5.27790129e-01
7.50325799e-01 8.45860988e-02 -6.02868736e-01 -7.69761860e-01
-8.77689064e-01 -3.66125256e-02 2.01759711e-01 -3.59070748e-01
8.54054868e-01 -6.79603457e-01 -1.10939658e+00 2.18981057e-01
-1.04007840e+00 -5.33087909e-01 -3.01872343e-01 -6.58896044e-02
-3.29692930e-01 2.17151538e-01 -1.51487038e-01 -1.38072979e+00
-2.81534433e-01 -1.17933476e+00 8.63627017e-01 4.13180262e-01
-6.09799445e-01 -5.80894053e-01 -2.37805828e-01 8.33195210e-01
6.74700916e-01 -1.49661481e-01 1.29948640e+00 -1.03575635e+00
-4.76416141e-01 -1.38117433e-01 -6.79479912e-02 7.01313496e-01
-2.16422707e-01 -3.32084484e-02 -1.42500329e+00 -3.18819642e-01
-9.14246589e-02 -9.41462517e-01 6.45989776e-01 1.25152767e-01
1.26108396e+00 -4.19816792e-01 -4.14986350e-02 7.47130141e-02
8.43604743e-01 1.42415047e-01 7.18483627e-01 -1.91345923e-02
4.74477202e-01 7.91577637e-01 6.19574249e-01 1.08516797e-01
4.55995411e-01 6.37230992e-01 7.99534857e-01 1.10786110e-01
-3.37835908e-01 -3.87995601e-01 6.39208019e-01 -3.25989783e-01
4.38667655e-01 -6.78188086e-01 -1.07612193e+00 6.95830524e-01
-1.46464276e+00 -7.10402966e-01 -1.16974398e-01 2.43052530e+00
9.78581190e-01 2.48738483e-01 2.03309834e-01 -2.14644313e-01
4.48366553e-01 -1.50357455e-01 -9.65945780e-01 -6.58268452e-01
-1.63191147e-02 3.52472633e-01 -7.95364156e-02 8.83459568e-01
-9.03887093e-01 6.87480927e-01 6.35592079e+00 7.06683695e-01
-8.08746099e-01 6.88006133e-02 1.04874325e+00 -3.56007844e-01
-6.96534693e-01 -1.56074747e-01 -7.32798934e-01 5.13656914e-01
1.16028595e+00 4.24878299e-01 3.91612738e-01 9.68722045e-01
-7.71184191e-02 -2.25111902e-01 -1.21370554e+00 8.87652099e-01
1.36565939e-01 -7.14073420e-01 5.63866384e-02 -4.50373650e-01
4.45556819e-01 -3.26622799e-02 5.97478628e-01 7.98704684e-01
4.70912904e-01 -1.47160339e+00 9.67791259e-01 4.97621328e-01
9.78966296e-01 -5.87825894e-01 6.66853368e-01 5.05243301e-01
-3.86037529e-01 -2.65711546e-01 -2.48229459e-01 1.66296273e-01
3.99836451e-02 1.06683485e-01 -1.15238452e+00 -2.75653247e-02
1.08127403e+00 -3.67939591e-01 -8.46749008e-01 8.73401165e-01
-5.87152421e-01 9.59395468e-01 -4.33726251e-01 -3.03874493e-01
-1.27871362e-02 3.10020775e-01 5.35795093e-01 8.24358165e-01
1.73732966e-01 1.01196520e-01 -2.40028575e-01 1.02783191e+00
-2.39857242e-01 -3.98803912e-02 -5.05871058e-01 2.49010518e-01
4.75535214e-01 8.47119749e-01 3.09768528e-01 -2.72900313e-01
-3.34383488e-01 1.05422938e+00 6.94101930e-01 8.27715218e-01
-7.37804174e-01 -3.78632218e-01 6.54939055e-01 2.67059684e-01
1.19430654e-01 2.35477671e-01 -2.52412438e-01 -8.97030473e-01
2.28560016e-01 -1.44719672e+00 4.99105215e-01 -1.11836028e+00
-1.49064994e+00 7.77100325e-01 -1.22621357e-01 -9.59253669e-01
-3.84195983e-01 -7.82519042e-01 -4.89897311e-01 1.07107091e+00
-1.18935037e+00 -1.02917039e+00 -4.82339978e-01 6.01569116e-01
3.87036294e-01 -1.08090587e-01 8.47438872e-01 1.08332917e-01
-1.90513387e-01 1.17895555e+00 -2.24915549e-01 -7.10464269e-02
1.03490210e+00 -1.62103260e+00 1.64610505e-01 5.78492463e-01
1.22601822e-01 6.06646836e-01 8.93717170e-01 -5.95154285e-01
-9.02781308e-01 -9.03251648e-01 8.15162003e-01 -1.06826663e+00
1.86147317e-01 -5.12246013e-01 -1.27540004e+00 7.35678434e-01
1.96576774e-01 -1.37109190e-01 7.51441240e-01 1.46493465e-01
-6.69367611e-01 -2.31958270e-01 -1.60922337e+00 9.10684705e-01
8.54116142e-01 -7.04140484e-01 -3.87735963e-01 3.98733139e-01
8.24348450e-01 -3.91637117e-01 -5.27121067e-01 2.55227119e-01
5.31391084e-01 -1.18873072e+00 8.71811509e-01 -9.42788243e-01
2.72497624e-01 -1.89838246e-01 -1.77170709e-01 -1.18376589e+00
6.35403618e-02 -5.84029019e-01 -4.22970414e-01 1.18431032e+00
9.17073905e-01 -7.04369128e-01 9.37229753e-01 1.37479949e+00
2.98950691e-02 -8.41561675e-01 -1.05303395e+00 -5.64112067e-01
2.65632659e-01 -6.98395312e-01 3.38445544e-01 4.82253373e-01
-3.92518848e-01 3.24268192e-01 -5.77997446e-01 4.97057617e-01
5.25712073e-01 -4.72264141e-01 9.05156434e-01 -7.89160609e-01
-4.80171919e-01 -2.08347499e-01 -1.33111656e-01 -1.25671017e+00
-4.57961950e-03 -6.88739896e-01 -3.09736654e-02 -1.49628377e+00
-2.38316938e-01 -3.86069894e-01 3.75288241e-02 7.36712396e-01
-4.75310355e-01 -3.55804292e-03 3.63877833e-01 5.73195778e-02
-4.51259226e-01 6.32576406e-01 9.14813161e-01 -5.45573115e-01
-3.15422148e-01 3.60009462e-01 -7.92893410e-01 5.55517614e-01
7.57590175e-01 -3.98352206e-01 -7.63337672e-01 -6.09201431e-01
2.41207018e-01 -6.17547221e-02 6.61337614e-01 -8.58752966e-01
8.75696242e-02 2.34373901e-02 5.38086534e-01 -3.65855426e-01
7.58921027e-01 -8.46989095e-01 -2.53735602e-01 3.23912889e-01
-5.28546572e-01 -2.12731466e-01 3.47921520e-01 5.48096061e-01
6.37177899e-02 -4.18906331e-01 7.14030385e-01 1.19735524e-01
-4.39877898e-01 -5.18232882e-02 -4.26105708e-01 3.78629297e-01
8.70720446e-01 -2.98559014e-02 -8.69238496e-01 -1.22259343e+00
-1.14938843e+00 7.04333305e-01 3.29129994e-01 3.97262752e-01
9.94413435e-01 -9.71867323e-01 -7.89646149e-01 1.22966371e-01
3.67856771e-01 -2.48917446e-01 4.87730026e-01 6.05383217e-01
-4.13540244e-01 -1.65671334e-02 1.69568732e-01 -7.12362289e-01
-1.43962574e+00 2.62441516e-01 7.53543258e-01 1.20475546e-01
-8.57585296e-02 1.23132479e+00 3.25434834e-01 -6.92742467e-01
4.77895737e-01 -1.20923445e-01 -1.82017982e-01 -1.01589173e-01
5.28993607e-01 2.33845994e-01 5.86411799e-04 -2.49680832e-01
-3.89644295e-01 -3.19931284e-02 -3.58594447e-01 -2.84643292e-01
6.61029339e-01 7.75111616e-02 3.27327847e-01 3.49298328e-01
7.41993725e-01 2.75576741e-01 -1.34884989e+00 6.69778734e-02
-3.40556353e-01 -6.10990226e-01 -4.63770032e-01 -1.55930650e+00
-5.75306714e-01 1.21365452e+00 1.00441003e+00 4.39016581e-01
9.10983682e-01 3.56970876e-01 3.73153865e-01 5.07634759e-01
1.95852257e-02 -8.86807680e-01 3.74727815e-01 2.08035454e-01
1.24479091e+00 -1.42453289e+00 -4.29773211e-01 -1.65829092e-01
-1.17327321e+00 6.72387421e-01 1.30136979e+00 1.70933262e-01
2.05675960e-01 -2.25396872e-01 6.06160343e-01 -1.05486363e-01
-1.07620895e+00 1.56937074e-02 4.59780395e-01 1.06888604e+00
1.73847303e-01 9.66825709e-02 2.72634357e-01 7.81328797e-01
-2.02496156e-01 -5.84486246e-01 5.25746822e-01 5.36636651e-01
-5.29969752e-01 -9.26406980e-01 -4.07520890e-01 7.55315006e-01
-3.43270123e-01 -5.60799986e-02 -7.32488811e-01 7.01946199e-01
7.41542205e-02 1.32902670e+00 -1.54878378e-01 -4.73230541e-01
6.65972590e-01 4.38933104e-01 2.92961866e-01 -5.61687946e-01
-7.74398863e-01 -5.04380703e-01 4.31863576e-01 -6.33257866e-01
2.51328796e-01 -4.41781074e-01 -9.87180591e-01 -1.21117905e-01
-2.03506976e-01 -1.34605274e-01 2.72936791e-01 7.58084416e-01
4.63416636e-01 2.50764549e-01 2.09215522e-01 -1.38359293e-01
-1.12122464e+00 -9.29381430e-01 -1.71317980e-01 5.33895016e-01
6.28999472e-01 -5.01569986e-01 -6.46728694e-01 -4.01024878e-01] | [11.243090629577637, 8.00083065032959] |
c81f6bbd-b9a6-407d-8508-28cccb77c493 | a-practical-system-based-on-cnn-blstm-network | null | null | https://ieeexplore.ieee.org/abstract/document/9420620 | https://ieeexplore.ieee.org/abstract/document/9420620 | A practical system based on CNN-BLSTM network for accurate classification of ECG heartbeats of MIT-BIH imbalanced dataset | ECG beats have a key role in the reduction of fatality rate arising from cardiovascular diseases (CVDs) by using Arrhythmia diagnosis computer-aided systems and get the important information from patient cardiac conditions to the specialist. However, the accuracy and speed of arrhythmia diagnosis are challenging in ECG classification systems, and the existence of noise, instability nature and imbalance in heartbeats challenged these systems. Accurate and on-time diagnosis of CVDs is a vital and important factor. So it has a significant effect on the treatment and recovery of patients. In this study, with the aim of accurate diagnosis of CVDs types, according to arrhythmia in ECG heartbeats, we implement an automatic ECG heartbeats classification by using discrete wavelet transformation on db2 mother wavelet and SMOTE oversampling algorithm as pre-processing level, and a classifier that consists of Convolutional neural network and BLSTM network. Then evaluate the proposed system on MIT-BIH imbalanced dataset, according to AAMI standards. The evaluations results show this approach with 50 epoch training achieved 99.78% accuracy for category F, 98.85% accuracy for category N, 99.43% accuracy for category S, 99.49% accuracy for category V, 99.87% accuracy for category Q. The source code is available at https://gitlab.com/arminshoughi/cnnlstmecg-classification. Our proposed classification system can be used as a tool for the automatic diagnosis of arrhythmia for CVDs specialists with the aim of primary screening of patients with heart arrhythmia. | ['mb dowlatshahi', 'armin shoughi'] | 2021-05-07 | null | null | null | 26th-international-computer-conference | ['ecg-classification'] | ['medical'] | [ 1.95559040e-02 -4.23351467e-01 3.84689495e-02 -2.27899656e-01
-4.22517031e-01 -2.19692990e-01 -4.57692266e-01 2.68762618e-01
-1.99501961e-01 7.27453113e-01 -2.05175400e-01 -5.40988088e-01
-2.95602888e-01 -5.83786666e-01 2.47124657e-02 -6.58462286e-01
-2.48951897e-01 4.29680318e-01 -3.32939804e-01 3.64321005e-03
1.38932213e-01 5.82120061e-01 -1.16141188e+00 5.07725656e-01
7.09810555e-01 1.23360217e+00 -3.37819189e-01 1.04289865e+00
2.75144011e-01 5.17018199e-01 -8.85163426e-01 1.20723553e-01
9.38516259e-02 -9.26633298e-01 -5.94180346e-01 -2.46762857e-01
-2.72667468e-01 -2.56090522e-01 9.83625129e-02 7.55583823e-01
1.29857075e+00 -5.09370625e-01 7.25993991e-01 -9.78618503e-01
-3.32292691e-02 4.10950571e-01 -3.70484054e-01 8.24778020e-01
4.48172390e-02 1.44670948e-01 3.56349289e-01 -6.14000320e-01
6.00285009e-02 6.41452432e-01 1.20082843e+00 4.23494369e-01
-8.48923385e-01 -9.46255922e-01 -7.39175260e-01 4.87649679e-01
-1.41392589e+00 -2.83781856e-01 8.26961339e-01 -5.74370921e-01
6.93177104e-01 4.39130366e-01 9.50134099e-01 6.34901345e-01
4.77654397e-01 2.00899899e-01 8.13536406e-01 -4.44808125e-01
-7.87380189e-02 5.55332713e-02 4.42406744e-01 4.52360570e-01
4.48323846e-01 1.85232282e-01 -1.73784465e-01 -2.63686508e-01
8.33714485e-01 1.56270191e-01 -2.77931809e-01 5.02363026e-01
-1.25737786e+00 5.05608857e-01 2.86479175e-01 7.13262022e-01
-5.30153215e-01 -4.81176265e-02 8.51977944e-01 7.40126491e-01
3.77751172e-01 3.87289375e-01 -8.28439295e-01 -2.98988938e-01
-8.90759289e-01 9.15377438e-02 5.13120592e-01 2.12302998e-01
-2.32528504e-02 3.72317791e-01 -3.19047809e-01 7.33707309e-01
1.89302042e-01 6.08405411e-01 1.00283432e+00 -6.68725491e-01
8.51228535e-02 5.52826703e-01 -1.15496412e-01 -1.06372559e+00
-7.13728964e-01 -1.04540050e+00 -1.63594794e+00 -3.41347884e-03
1.27758816e-01 -3.31245184e-01 -6.64461374e-01 1.37736070e+00
2.73241699e-01 2.48172820e-01 1.77082673e-01 8.63317072e-01
1.06470859e+00 5.07324398e-01 4.27790135e-02 -5.89909554e-01
1.48988652e+00 -1.52421162e-01 -9.35950279e-01 3.40211183e-01
6.91026688e-01 -7.51939952e-01 6.97752237e-01 7.41313040e-01
-8.66537809e-01 -8.76828730e-01 -1.02696645e+00 3.08300495e-01
4.34927084e-02 8.12438607e-01 2.37686455e-01 8.49981427e-01
-7.87186205e-01 7.94891357e-01 -7.41555631e-01 -3.47283959e-01
6.88792229e-01 4.09438848e-01 3.76306102e-02 3.65048707e-01
-1.47184849e+00 7.28735566e-01 3.30689758e-01 3.86603981e-01
-6.02643788e-01 -8.01857412e-01 -4.52272654e-01 -6.13533035e-02
-5.36350489e-01 -8.18826556e-01 9.19871092e-01 -9.40086424e-01
-1.06854117e+00 9.06009257e-01 3.63758877e-02 -7.12469161e-01
5.66937447e-01 -1.14401430e-01 -6.59119904e-01 5.84376454e-02
-1.02757821e-02 -6.49185777e-02 7.89884925e-01 -3.02103102e-01
-5.87590873e-01 -6.19226635e-01 -6.83865011e-01 1.66273654e-01
-1.75952405e-01 -1.80976803e-03 4.83785093e-01 -6.55879021e-01
2.63344616e-01 -5.26129425e-01 1.17553957e-01 -2.64833421e-01
-3.20401698e-01 -1.07957236e-02 9.27653015e-01 -1.15463054e+00
1.45313549e+00 -2.22302270e+00 -1.65288195e-01 1.46760166e-01
3.37046027e-01 7.18483865e-01 3.95120710e-01 1.46939620e-01
-3.96709174e-01 2.35781640e-01 -2.31012613e-01 3.48414838e-01
-4.36594456e-01 -1.73396215e-01 -1.01416111e-01 4.81000662e-01
2.10881025e-01 7.05383182e-01 -3.46932262e-01 -3.22260350e-01
2.89418489e-01 5.92301071e-01 -1.37110084e-01 3.44434232e-01
4.54183131e-01 7.81517744e-01 -3.58681232e-01 6.09796464e-01
3.99338186e-01 -1.11650825e-01 5.43241091e-02 -5.77284276e-01
1.44927993e-01 2.46558711e-03 -1.22913921e+00 1.33483636e+00
-3.33260372e-02 5.13460398e-01 -2.48377532e-01 -1.44301152e+00
1.22249663e+00 9.97278631e-01 6.35665417e-01 -4.26287562e-01
4.48298663e-01 5.63161194e-01 5.89349270e-01 -8.47738981e-01
-6.01876676e-01 -3.05421710e-01 2.05486014e-01 2.25432143e-01
-1.86382443e-01 1.63750127e-01 -5.34495525e-02 -2.37696379e-01
9.88539994e-01 -2.65528977e-01 5.88422716e-01 -2.90315747e-01
6.91830873e-01 -2.27287903e-01 6.78641200e-01 4.83673930e-01
-4.70887780e-01 6.34311557e-01 3.13768625e-01 -1.01534641e+00
-8.37689221e-01 -6.31762564e-01 -5.01748204e-01 1.09215535e-01
-3.18340361e-01 1.84146509e-01 -6.24865174e-01 -1.66832820e-01
-2.30970517e-01 2.47667972e-02 -2.72106886e-01 -4.69557941e-01
-5.24261355e-01 -1.23951745e+00 9.87631977e-01 4.06028301e-01
7.29634702e-01 -1.33759105e+00 -1.07688642e+00 4.07412767e-01
-4.08833623e-01 -6.24071956e-01 5.64450882e-02 7.94276446e-02
-1.19983506e+00 -1.33949018e+00 -7.89275229e-01 -8.46710086e-01
1.75293192e-01 -3.96807194e-01 1.14256048e+00 4.55106884e-01
-9.93834078e-01 -1.77891225e-01 -1.57666743e-01 -9.01708424e-01
-4.67328370e-01 1.90962285e-01 1.31246537e-01 -1.72479718e-03
2.75763273e-01 -8.02667201e-01 -1.01865900e+00 3.42875458e-02
-2.83126742e-01 -3.09587084e-03 5.53623259e-01 8.09138417e-01
3.79709929e-01 1.40101194e-01 1.22694135e+00 -8.69910657e-01
8.22798014e-01 -3.61052245e-01 -2.41736636e-01 -2.06705362e-01
-8.87762368e-01 -5.84952414e-01 5.93500316e-01 -7.07650781e-02
-2.57664502e-01 -8.59939307e-02 -5.23915231e-01 -4.39467281e-01
-3.72687221e-01 3.46073717e-01 3.79377860e-03 3.49004775e-01
8.20438445e-01 1.82265133e-01 1.55456752e-01 -3.55760396e-01
-5.94440401e-01 9.48158860e-01 3.73728603e-01 -1.82570234e-01
3.16633254e-01 -4.01641130e-02 1.84799016e-01 -7.71725476e-01
-6.03164911e-01 -4.88148659e-01 -5.45804083e-01 -1.89946368e-01
9.56432700e-01 -9.42585707e-01 -7.92668521e-01 8.82241488e-01
-1.09466243e+00 -2.39391858e-03 -3.52032989e-01 5.76292694e-01
-1.30920634e-01 2.87685156e-01 -8.36373270e-01 -9.86792266e-01
-1.33689344e+00 -7.49422312e-01 5.45001626e-01 6.36329576e-02
-5.48576891e-01 -6.58319175e-01 -3.33178937e-02 1.41284466e-01
5.42705595e-01 8.92935038e-01 9.44188595e-01 -8.01416993e-01
1.60051376e-01 -4.63578373e-01 -1.07540339e-02 6.81556940e-01
3.46588075e-01 9.52920504e-03 -1.03540945e+00 -2.39027575e-01
3.22664827e-01 6.08500540e-02 5.18379986e-01 8.03231597e-01
1.12695706e+00 -2.64244795e-01 -1.28288329e-01 6.45583451e-01
1.34983563e+00 6.29253507e-01 6.94544792e-01 -7.09679872e-02
6.32896006e-01 4.77874465e-02 4.54347461e-01 5.18717408e-01
5.76311760e-02 2.76256591e-01 2.15223089e-01 -5.01444221e-01
-2.25489900e-01 5.51386535e-01 4.72794324e-02 1.14262760e+00
-3.84679377e-01 1.20573305e-01 -1.11753547e+00 5.73121786e-01
-1.54918540e+00 -1.06551123e+00 -8.10527742e-01 2.23784113e+00
9.26759481e-01 2.97761764e-02 2.46919468e-01 1.23921299e+00
8.94648671e-01 -5.29764652e-01 -5.78320384e-01 -4.92671430e-01
8.70582014e-02 4.73789185e-01 -6.85122088e-02 1.78603441e-01
-1.17189050e+00 1.32269159e-01 5.13873291e+00 4.05048430e-01
-1.54158175e+00 2.37208337e-01 1.12309515e+00 1.13404386e-01
5.65324068e-01 -5.84351778e-01 -4.39737290e-01 7.58099735e-01
1.09904003e+00 1.17376685e-01 8.73783380e-02 3.69817376e-01
5.66902161e-01 3.92096728e-01 -8.61473620e-01 1.38616323e+00
-3.15857291e-01 -1.03898978e+00 -3.97104234e-01 -3.70853275e-01
2.36194313e-01 -1.57174483e-01 -1.85743317e-01 1.29579648e-01
-6.65920377e-01 -1.04652143e+00 3.34049582e-01 6.45936310e-01
1.24381089e+00 -8.15139055e-01 1.32935226e+00 3.59509945e-01
-1.05598021e+00 -1.10259138e-01 -1.52261868e-01 -3.82325292e-01
-1.61669537e-01 1.17646432e+00 -8.78660917e-01 4.47562963e-01
9.21945035e-01 8.98035228e-01 -3.31114501e-01 1.01885760e+00
6.26010075e-02 1.12409747e+00 -2.47936383e-01 2.33362302e-01
-5.93545020e-01 -6.12760894e-03 4.82316077e-01 1.08205140e+00
5.83041549e-01 4.03533340e-01 7.75254592e-02 5.22513390e-01
1.96055606e-01 2.53444284e-01 -4.24252599e-01 1.99609160e-01
3.94663036e-01 1.13854074e+00 -6.76029682e-01 -6.19433105e-01
3.20791692e-01 5.96499205e-01 -4.06317562e-01 3.06087248e-02
-9.14245725e-01 -7.93623686e-01 2.66651213e-01 4.18389708e-01
-2.99689740e-01 4.75641638e-01 -8.73271167e-01 -9.33526695e-01
1.45998433e-01 -1.17039955e+00 5.63683629e-01 -4.83835459e-01
-9.90565896e-01 7.68445909e-01 -6.38433516e-01 -1.49030375e+00
-2.15793163e-01 -3.10537368e-01 -8.42840493e-01 1.18024397e+00
-1.21074855e+00 -7.65932918e-01 -5.47500551e-01 5.07258534e-01
4.02258486e-01 -2.83615828e-01 1.25086653e+00 8.34954917e-01
-5.64907312e-01 4.40910995e-01 -2.64883578e-01 3.14135700e-01
6.60713911e-01 -1.11997068e+00 8.56865570e-02 7.20101416e-01
-2.25391209e-01 2.40056038e-01 5.61201572e-01 -4.99883801e-01
-7.78037906e-01 -1.25289464e+00 9.62937295e-01 -1.95073172e-01
-1.55891731e-01 1.87841535e-01 -7.80075014e-01 1.28218204e-01
2.73191975e-03 1.42875060e-01 7.82407284e-01 -1.54946938e-01
2.69139111e-01 -4.94230807e-01 -1.28918600e+00 -2.62552686e-03
5.33645630e-01 -2.96718448e-01 -3.67669493e-01 3.56395781e-01
1.81687087e-01 -4.58764613e-01 -1.28948855e+00 8.51847112e-01
8.53400171e-01 -9.57386434e-01 9.98199582e-01 -5.63922584e-01
4.03865933e-01 -3.56864870e-01 3.75524580e-01 -9.96630490e-01
-3.91719580e-01 -3.12804788e-01 -6.24355488e-02 1.02126348e+00
2.55595505e-01 -7.74433613e-01 3.89315128e-01 6.78062439e-02
-1.30953416e-01 -9.62458432e-01 -7.58979619e-01 -1.81910008e-01
-2.64878869e-01 -3.42607260e-01 2.61923909e-01 1.13646078e+00
-2.81205803e-01 4.60144043e-01 -3.64637405e-01 1.14153013e-01
5.15570641e-01 -5.94300888e-02 3.59978706e-01 -1.71286643e+00
-2.76214272e-01 -2.06213385e-01 -6.63169920e-01 1.95166692e-01
-6.78128898e-01 -7.86306202e-01 -5.30769229e-01 -1.54896665e+00
-7.05144629e-02 -5.25975943e-01 -7.10100949e-01 5.22909105e-01
-2.41451368e-01 7.36507177e-01 -1.50993735e-01 3.32537740e-01
9.84512866e-02 -8.28781351e-02 9.52791035e-01 -6.87519014e-02
-3.48031580e-01 4.74546790e-01 -5.18438816e-01 7.02127516e-01
1.31857944e+00 -5.51836550e-01 -5.19708157e-01 -3.36536281e-02
3.13003473e-02 4.00664687e-01 4.59532917e-01 -1.53801525e+00
-2.28474155e-01 4.18715417e-01 8.85660231e-01 -4.76762265e-01
5.35511263e-02 -7.20798373e-01 7.01367497e-01 1.33945739e+00
-1.42226204e-01 1.53876916e-01 2.37618729e-01 3.88404727e-02
-2.85459727e-01 6.44803122e-02 8.51892769e-01 -4.66806293e-02
1.71500596e-03 3.45265120e-01 -4.75306153e-01 1.66251063e-01
9.46148694e-01 -2.78583646e-01 3.36522087e-02 -9.48295668e-02
-1.05026841e+00 -1.79593652e-01 -3.30281824e-01 1.00961246e-01
6.58305168e-01 -1.31458938e+00 -1.07718062e+00 4.17728633e-01
-1.40131846e-01 8.20856392e-02 4.34511811e-01 1.32065666e+00
-9.56754088e-01 1.24668568e-01 -4.81006682e-01 -7.94157565e-01
-1.61299765e+00 9.16409343e-02 7.04645038e-01 -2.58747458e-01
-6.67668998e-01 5.73444545e-01 -3.02713782e-01 7.34292045e-02
2.13047236e-01 -6.60393775e-01 -5.44173837e-01 1.51318863e-01
5.18650532e-01 5.15862048e-01 5.85757375e-01 -2.44558573e-01
-4.44725215e-01 7.58140385e-01 2.89568990e-01 3.10930610e-01
1.08802629e+00 1.15340896e-01 -3.19565594e-01 6.59334719e-01
9.22475636e-01 -4.12494868e-01 -2.27778003e-01 3.09867442e-01
-3.03507477e-01 -5.91512062e-02 -3.47110443e-02 -1.13518953e+00
-1.25910509e+00 1.38394606e+00 1.51535106e+00 4.00992841e-01
1.42684948e+00 -7.24135697e-01 9.15376186e-01 5.93586266e-02
4.39882502e-02 -6.36241615e-01 -3.64484757e-01 -9.99971665e-03
6.61905825e-01 -1.06816804e+00 -5.18297292e-02 -7.33630732e-02
-4.28936511e-01 1.31178546e+00 1.64878488e-01 -2.53209323e-01
1.01444912e+00 9.70558673e-02 5.61604679e-01 -1.70560956e-01
-5.25490403e-01 3.40967745e-01 1.24586076e-01 6.03143692e-01
7.37601876e-01 2.62259543e-01 -7.47270882e-01 1.03151405e+00
-6.21291026e-02 5.14482856e-01 2.77376920e-01 6.12972319e-01
-3.83271366e-01 -8.14908922e-01 -4.16202813e-01 9.46452975e-01
-1.03089964e+00 -1.10451400e-01 -5.81356212e-02 2.74322033e-01
5.97881913e-01 8.52100492e-01 -9.60568264e-02 -3.26643556e-01
2.01075897e-01 4.47330356e-01 1.67953312e-01 -4.42416728e-01
-9.60723221e-01 2.91315373e-02 -1.90006774e-02 -8.13967288e-02
-2.27496654e-01 -4.29378062e-01 -1.15643275e+00 1.99185349e-02
-2.00517759e-01 2.75059074e-01 6.75481915e-01 7.10483789e-01
6.21971607e-01 1.15130949e+00 5.86752057e-01 -2.77333379e-01
-5.01081586e-01 -1.33033526e+00 -6.97362483e-01 3.56919855e-01
5.01776695e-01 -1.80417955e-01 -2.78625429e-01 5.32277882e-01] | [14.302411079406738, 3.285748243331909] |
59c8db77-3b45-4b90-ab56-2648000308ef | improved-multiple-image-based-reflection | 2208.04679 | null | https://arxiv.org/abs/2208.04679v2 | https://arxiv.org/pdf/2208.04679v2.pdf | Improved Multiple-Image-Based Reflection Removal Algorithm Using Deep Neural Networks | When imaging through a semi-reflective medium such as glass, the reflection of another scene can often be found in the captured images. It degrades the quality of the images and affects their subsequent analyses. In this paper, a novel deep neural network approach for solving the reflection problem in imaging is presented. Traditional reflection removal methods not only require long computation time for solving different optimization functions, their performance is also not guaranteed. As array cameras are readily available in nowadays imaging devices, we first suggest in this paper a multiple-image based depth estimation method using a convolutional neural network (CNN). The proposed network avoids the depth ambiguity problem due to the reflection in the image, and directly estimates the depths along the image edges. They are then used to classify the edges as belonging to the background or reflection. Since edges having similar depth values are error prone in the classification, they are removed from the reflection removal process. We suggest a generative adversarial network (GAN) to regenerate the removed background edges. Finally, the estimated background edge map is fed to another auto-encoder network to assist the extraction of the background from the original image. Experimental results show that the proposed reflection removal algorithm achieves superior performance both quantitatively and qualitatively as compared to the state-of-the-art methods. The proposed algorithm also shows much faster speed compared to the existing approaches using the traditional optimization methods. | ['Daniel P. K. Lun', 'Yuk-Hee Chan', 'Tingtian Li'] | 2022-08-09 | null | null | null | null | ['reflection-removal'] | ['computer-vision'] | [ 7.14687467e-01 -9.57661718e-02 5.49468100e-01 -1.18387423e-01
-3.91427875e-01 -1.89964920e-01 2.58339167e-01 -4.82459933e-01
-4.08332616e-01 6.07068658e-01 -2.11146876e-01 -8.02568346e-02
2.72873133e-01 -8.86508584e-01 -6.58531666e-01 -1.12730801e+00
5.35495162e-01 2.25908179e-02 3.27294886e-01 1.57087326e-01
3.29361528e-01 5.64998209e-01 -1.45672095e+00 1.47235155e-01
8.76260400e-01 1.23580945e+00 3.37333173e-01 4.39933360e-01
-1.60819098e-01 1.04746556e+00 -8.35249126e-01 -2.65313208e-01
6.60670936e-01 -6.57782435e-01 -1.25419497e-01 2.88103849e-01
5.11221826e-01 -8.18536460e-01 -5.47099292e-01 1.41018188e+00
4.10536915e-01 1.84914470e-01 4.95731711e-01 -9.10085976e-01
-4.92325604e-01 5.17329052e-02 -8.97463620e-01 7.47051612e-02
6.27953485e-02 -1.94026008e-01 1.62335351e-01 -8.20657492e-01
2.27511659e-01 8.50037754e-01 7.40059972e-01 6.23704910e-01
-6.41193211e-01 -7.36704946e-01 -6.42629415e-02 1.88074514e-01
-1.38203967e+00 -4.92166966e-01 1.20667195e+00 -2.52414346e-01
4.90743190e-01 2.57623851e-01 6.06452525e-01 7.92868435e-01
4.19096410e-01 4.04457271e-01 1.19542408e+00 -5.60522199e-01
-7.10352138e-02 1.72168791e-01 -8.11818615e-03 9.70423460e-01
3.18613887e-01 8.28847140e-02 -1.67651907e-01 2.35192150e-01
8.74238014e-01 3.32756460e-01 -6.87488198e-01 1.78359360e-01
-6.96082294e-01 4.04807061e-01 5.50137579e-01 1.60327613e-01
-4.07062382e-01 7.39327744e-02 -1.00464253e-02 9.15971920e-02
5.23105562e-01 4.45634797e-02 8.79514813e-02 4.29787934e-01
-9.56262887e-01 -1.58415601e-01 6.83873832e-01 6.06956661e-01
8.39129508e-01 2.77746797e-01 5.83191663e-02 8.07950854e-01
4.00423199e-01 4.04656857e-01 4.00537461e-01 -7.41706014e-01
2.75681317e-01 6.60832286e-01 3.00595284e-01 -1.32145333e+00
-2.31733114e-01 -4.44879949e-01 -1.17524910e+00 6.96626008e-01
4.55480397e-01 -1.51094615e-01 -1.08503175e+00 1.15385818e+00
3.08464706e-01 3.62184763e-01 4.01822686e-01 1.08707273e+00
1.11587954e+00 9.05635655e-01 -5.02753913e-01 -2.09803492e-01
1.06815124e+00 -1.15740621e+00 -7.80846000e-01 -5.68215907e-01
-1.73699901e-01 -9.55625653e-01 5.79265177e-01 7.25731730e-01
-1.05099010e+00 -4.44164395e-01 -1.16071200e+00 3.31729464e-02
2.72302218e-02 3.94496799e-01 4.39689606e-01 6.09251022e-01
-8.40224385e-01 3.76813173e-01 -7.14012206e-01 5.01812473e-02
2.82790393e-01 2.40862086e-01 -2.58061200e-01 -5.01013100e-01
-7.69179344e-01 7.16561139e-01 9.49479640e-02 7.33787537e-01
-8.02004993e-01 -3.18284363e-01 -7.24281907e-01 -4.89601754e-02
2.72485286e-01 -5.81265986e-01 8.10628533e-01 -1.52994931e+00
-1.72104919e+00 7.11728930e-01 -1.69223949e-01 -2.47269720e-02
6.62056386e-01 -1.60441115e-01 -3.80159914e-01 2.82251030e-01
-1.44557700e-01 5.85318953e-02 1.21018422e+00 -1.45763862e+00
-6.17697060e-01 -1.82829648e-01 6.94987774e-02 3.53140652e-01
-6.16811998e-02 -1.72375999e-02 -6.45877361e-01 -3.97243798e-01
4.58622307e-01 -6.70302927e-01 -2.08634019e-01 1.13584258e-01
-5.22553861e-01 4.48775291e-01 1.03250110e+00 -8.14010799e-01
8.04279566e-01 -2.24057841e+00 -3.18709910e-02 6.69538006e-02
2.65810430e-01 2.59621948e-01 4.69315518e-03 1.18362896e-01
-4.64434968e-03 -3.93361598e-01 -3.14088643e-01 -3.76578927e-01
-6.01424992e-01 3.95371858e-03 1.17171243e-01 7.82102048e-01
-1.19219264e-02 4.30658519e-01 -5.75540245e-01 -1.95161059e-01
4.12485182e-01 7.27280915e-01 -2.39528641e-01 4.66241449e-01
1.30386814e-01 7.86962867e-01 -2.23257497e-01 6.00583136e-01
1.08656693e+00 1.24321677e-01 -1.97624162e-01 -5.94203115e-01
-2.17077926e-01 -2.07274467e-01 -1.19668174e+00 1.03770769e+00
-6.59187019e-01 7.97362745e-01 2.76158035e-01 -1.03997219e+00
1.17562461e+00 4.78357710e-02 3.87633443e-01 -7.00028121e-01
2.06878722e-01 3.64889145e-01 1.09967917e-01 -6.61445856e-01
3.56498301e-01 -2.60435432e-01 4.01036829e-01 6.30523777e-03
-4.07618016e-01 -6.51539164e-03 -1.47826448e-01 -2.96026945e-01
1.08953345e+00 -2.87272930e-02 8.31381232e-03 4.43955325e-02
8.55581641e-01 -3.14161062e-01 8.09755504e-01 5.61040461e-01
1.57865927e-01 8.93053651e-01 -3.52848470e-02 -5.60344219e-01
-8.70801032e-01 -9.25131261e-01 -1.49145573e-02 1.60192385e-01
6.24737442e-01 1.56456098e-01 -7.13850737e-01 -2.76876599e-01
-4.99008447e-01 3.44991803e-01 -3.56475443e-01 -2.68221051e-01
-6.88408971e-01 -8.55893970e-01 2.50852317e-01 2.49526039e-01
1.17198443e+00 -1.00255144e+00 -7.12134302e-01 2.47791663e-01
-3.94131035e-01 -1.38974881e+00 -1.20225854e-01 5.65940933e-03
-7.01263309e-01 -1.10489953e+00 -9.49717462e-01 -8.78936231e-01
1.06183934e+00 5.31129122e-01 8.17087650e-01 2.45730430e-01
-4.36686277e-01 1.64271787e-01 -2.13212386e-01 -4.82467055e-01
-4.59563583e-01 -5.66321909e-01 -3.89103174e-01 5.15291512e-01
2.21987277e-01 -4.28295761e-01 -1.14466262e+00 2.61258781e-01
-9.35936809e-01 2.54420429e-01 6.81533396e-01 9.13474321e-01
4.01825070e-01 7.58638918e-01 2.59795845e-01 -8.43658984e-01
3.26878279e-01 -9.37599912e-02 -9.35251772e-01 -2.73647793e-02
-2.83261657e-01 -2.65985548e-01 8.11692238e-01 -3.75195593e-01
-1.40206254e+00 5.80158010e-02 -1.68606430e-01 -5.10167360e-01
-1.81482062e-01 3.07692647e-01 -2.52759665e-01 -4.35763657e-01
2.42268458e-01 4.65295196e-01 -3.93571518e-02 -2.73806334e-01
-2.97018260e-01 6.62832141e-01 5.51400423e-01 9.19203460e-02
8.47241640e-01 7.36882985e-01 2.07005665e-01 -1.14086413e+00
-9.25048292e-01 -3.94594640e-01 -1.01403534e-01 -5.62194467e-01
9.57764328e-01 -9.22006011e-01 -5.81804574e-01 1.08549142e+00
-1.30852139e+00 -3.37004602e-01 2.02591777e-01 4.85180557e-01
-9.06511694e-02 6.59497976e-01 -7.54419863e-01 -1.05943739e+00
-4.92190510e-01 -1.33955860e+00 9.14968908e-01 5.95671117e-01
3.52205247e-01 -9.36172783e-01 -3.06039512e-01 5.19638777e-01
4.19149220e-01 2.97176689e-01 7.45580196e-01 1.79461941e-01
-9.94329929e-01 -3.50187927e-01 -4.68369126e-01 7.35448956e-01
3.64825398e-01 -3.69622074e-02 -1.17457426e+00 -2.23737121e-01
6.29552007e-01 2.44656846e-01 9.62374687e-01 5.98774016e-01
1.23956406e+00 -3.12147260e-01 -2.43911967e-01 9.82285619e-01
2.02488613e+00 5.62722862e-01 1.22088063e+00 5.67283571e-01
9.54683304e-01 6.23534083e-01 5.39589763e-01 2.63994366e-01
-1.86998367e-01 4.89120692e-01 7.71941960e-01 -5.07892907e-01
-2.16712728e-01 3.17344934e-01 4.83362436e-01 6.83824003e-01
-1.69476703e-01 -6.97107375e-01 -5.13551831e-01 2.40279868e-01
-1.55112779e+00 -8.79310191e-01 -4.51229304e-01 2.30037498e+00
4.06547457e-01 -9.40733775e-02 -5.36783457e-01 2.27631718e-01
7.54058719e-01 -6.39823154e-02 -4.51873004e-01 -2.14433849e-01
-1.10809900e-01 3.94966871e-01 6.80728972e-01 6.02293909e-01
-9.27213669e-01 5.94106913e-01 5.45643759e+00 5.64162076e-01
-1.38365304e+00 3.00871097e-02 6.37035608e-01 4.13459614e-02
4.10021428e-04 -3.09778750e-01 -6.58466458e-01 5.20418465e-01
2.85688102e-01 3.27387512e-01 4.62629080e-01 4.41681802e-01
2.50011027e-01 -4.94200230e-01 -7.92529881e-01 1.35369945e+00
4.28430855e-01 -8.56455386e-01 -1.06356405e-01 -1.67420879e-01
7.77589679e-01 -1.67792916e-01 6.69183806e-02 -3.22631210e-01
-2.13005781e-01 -9.55047727e-01 3.99986774e-01 7.29765415e-01
6.02076650e-01 -9.47404325e-01 9.60027754e-01 3.21099848e-01
-8.26415777e-01 -1.62129104e-01 -6.71324968e-01 3.17910314e-02
1.15342505e-01 9.97876227e-01 -4.30231959e-01 5.88841200e-01
7.86587238e-01 5.33342838e-01 -2.20130488e-01 1.11041880e+00
-2.63536602e-01 3.58146966e-01 -2.67683387e-01 2.93346345e-01
1.83651343e-01 -8.07521999e-01 7.38725185e-01 9.05167818e-01
6.50870383e-01 -2.99748611e-02 -5.57909086e-02 9.10072327e-01
-1.24712568e-02 -1.42527327e-01 -7.11190701e-01 3.14424992e-01
-1.72279608e-02 1.44306540e+00 -8.85765076e-01 -2.65768737e-01
-7.64471352e-01 1.46788335e+00 -1.08244933e-01 5.16661942e-01
-8.48561585e-01 -4.78110909e-01 3.57459188e-01 2.56632805e-01
1.89343706e-01 5.93690500e-02 1.27524927e-01 -1.24385786e+00
2.75653541e-01 -8.10903311e-01 -6.08488880e-02 -1.01726162e+00
-1.28049028e+00 7.96133876e-01 -4.00790066e-01 -1.35400212e+00
1.58028960e-01 -7.19192207e-01 -8.10395718e-01 9.79217649e-01
-1.76852763e+00 -8.31275702e-01 -9.83660996e-01 5.65846622e-01
6.07214332e-01 -1.15030713e-01 5.12666523e-01 6.90074146e-01
-7.84152985e-01 3.01170379e-01 3.07699472e-01 2.86945134e-01
5.96185982e-01 -8.90630960e-01 -1.12087682e-01 1.33183289e+00
-2.92119712e-01 3.39164287e-01 6.22762382e-01 -5.51311493e-01
-1.23802888e+00 -1.13074863e+00 1.44421712e-01 2.66277969e-01
1.71442628e-01 -2.35246077e-01 -8.30283761e-01 4.19971496e-01
3.68562192e-01 -3.36174816e-02 4.47070301e-01 -6.95726335e-01
7.12841079e-02 -3.17953825e-01 -1.30936718e+00 6.09335542e-01
6.85752630e-01 -2.21528769e-01 -2.02463791e-01 1.25055909e-01
2.96578407e-01 -4.93541718e-01 -4.54587400e-01 4.04414743e-01
5.72787344e-01 -1.33434856e+00 1.02922022e+00 2.66392082e-01
7.23250389e-01 -4.43037897e-01 -1.14317290e-01 -1.25689769e+00
-1.68377355e-01 -3.49091232e-01 1.28245115e-01 1.17191494e+00
3.91041189e-02 -9.52029467e-01 9.71585810e-01 3.83065671e-01
-3.32785159e-01 -4.77635831e-01 -6.31240964e-01 -5.87658286e-01
-3.94072682e-01 6.72938973e-02 1.96670637e-01 8.75079095e-01
-8.49006593e-01 2.87563115e-01 -4.54145253e-01 4.93198901e-01
1.04877925e+00 2.28201792e-01 6.97267592e-01 -1.11992145e+00
-4.83589977e-01 -1.90605730e-01 -5.09175777e-01 -1.09257829e+00
1.36957727e-02 -5.77316999e-01 3.82700741e-01 -1.62120879e+00
1.30917028e-01 -4.42407370e-01 -1.12742729e-01 7.24122822e-02
-1.71249464e-01 6.24383628e-01 -1.10852979e-01 7.60581568e-02
-2.22391132e-02 4.68842953e-01 1.34034407e+00 -1.94925457e-01
-1.81367323e-01 1.61501989e-01 -5.03974617e-01 1.02438068e+00
8.72432172e-01 -4.48110521e-01 -3.50505322e-01 -6.02864683e-01
1.75525099e-01 1.08534239e-01 5.40566564e-01 -1.31389821e+00
9.87277105e-02 1.75644025e-01 6.35883033e-01 -5.24929762e-01
5.49014807e-01 -1.29552889e+00 3.50492388e-01 4.04094249e-01
2.56859511e-01 -2.49458730e-01 2.01591417e-01 5.86352706e-01
-3.68360430e-01 -7.81069636e-01 1.05295300e+00 -4.02995259e-01
-5.49502552e-01 1.47425294e-01 -6.38145328e-01 -3.78150344e-01
8.39975834e-01 -6.00684702e-01 -3.62598151e-01 -4.31740344e-01
-4.47548479e-01 -3.97767127e-01 4.81099695e-01 -9.84301195e-02
9.07068670e-01 -9.50451672e-01 -5.67472696e-01 3.10427397e-01
-1.16316713e-01 3.13065350e-01 3.50453377e-01 8.49420071e-01
-1.10920548e+00 -2.49435589e-01 -2.28144959e-01 -4.80247349e-01
-1.44025409e+00 3.47999990e-01 7.28476822e-01 -1.41221762e-01
-8.72395694e-01 6.87316656e-01 6.42722070e-01 6.31147027e-02
1.45571083e-01 -1.49804264e-01 -1.34595782e-01 -3.94671857e-01
6.01777196e-01 3.80783886e-01 8.12183470e-02 -4.58956063e-01
-1.06222443e-01 8.18721175e-01 6.24496937e-02 1.14872128e-01
1.38654494e+00 -1.81667149e-01 -3.69758636e-01 2.75373429e-01
1.31403899e+00 2.96779752e-01 -1.11168385e+00 -6.22915030e-02
-6.81618631e-01 -7.25674033e-01 4.03684586e-01 -5.12003362e-01
-1.66254866e+00 8.92422497e-01 7.74836123e-01 6.08473346e-02
1.56768239e+00 -5.54485500e-01 8.71527672e-01 2.69377798e-01
1.50305599e-01 -7.38990605e-01 2.27414727e-01 9.41410586e-02
6.58136249e-01 -1.26463532e+00 2.61870138e-02 -7.38351941e-01
-2.45250359e-01 1.41630435e+00 7.97743022e-01 -3.76711220e-01
4.03613031e-01 5.86447358e-01 4.05200094e-01 -1.32893026e-01
-1.18684523e-01 2.69107502e-02 8.83847624e-02 6.54954970e-01
2.37006977e-01 -5.55227399e-01 -9.25115943e-02 1.58542588e-01
1.48887664e-01 -1.86604455e-01 9.07813847e-01 6.91805303e-01
-3.71515453e-01 -8.03497553e-01 -5.80870688e-01 3.83572400e-01
-8.17287147e-01 -1.54200867e-01 -2.41022423e-01 6.26937866e-01
1.48936316e-01 9.66694236e-01 2.39515111e-01 -5.15401736e-02
1.47954628e-01 -2.57333487e-01 7.34244764e-01 -4.17812496e-01
-3.90377104e-01 2.38557354e-01 -2.80968752e-02 -3.33777130e-01
-6.30908966e-01 -3.11209708e-01 -1.07707965e+00 -5.49075799e-03
-6.29183829e-01 -1.58631369e-01 7.02615321e-01 6.33379877e-01
-1.16723217e-01 8.53122175e-01 7.06392288e-01 -9.31542099e-01
2.14429870e-02 -7.95425117e-01 -6.19728863e-01 3.22723895e-01
4.22834963e-01 -6.34202540e-01 -7.13508189e-01 -5.63857481e-02] | [10.48293685913086, -2.6911652088165283] |
80536980-29dd-49f5-a357-b1a4a36a47d7 | a-survey-of-quantum-cognitively-inspired | 2306.03608 | null | https://arxiv.org/abs/2306.03608v1 | https://arxiv.org/pdf/2306.03608v1.pdf | A Survey of Quantum-Cognitively Inspired Sentiment Analysis Models | Quantum theory, originally proposed as a physical theory to describe the motions of microscopic particles, has been applied to various non-physics domains involving human cognition and decision-making that are inherently uncertain and exhibit certain non-classical, quantum-like characteristics. Sentiment analysis is a typical example of such domains. In the last few years, by leveraging the modeling power of quantum probability (a non-classical probability stemming from quantum mechanics methodology) and deep neural networks, a range of novel quantum-cognitively inspired models for sentiment analysis have emerged and performed well. This survey presents a timely overview of the latest developments in this fascinating cross-disciplinary area. We first provide a background of quantum probability and quantum cognition at a theoretical level, analyzing their advantages over classical theories in modeling the cognitive aspects of sentiment analysis. Then, recent quantum-cognitively inspired models are introduced and discussed in detail, focusing on how they approach the key challenges of the sentiment analysis task. Finally, we discuss the limitations of the current research and highlight future research directions. | ['Dawei Song', 'Yazhou Zhang', 'Benyou Wang', 'Qiuchi Li', 'Yaochen Liu'] | 2023-06-06 | null | null | null | null | ['sentiment-analysis'] | ['natural-language-processing'] | [ 1.87242543e-03 -5.21874763e-02 2.18793988e-01 -3.18275452e-01
-1.83644250e-01 -4.99717861e-01 7.54525304e-01 4.61738199e-01
-4.36585486e-01 5.54071605e-01 3.14008221e-02 -1.30130902e-01
-3.99000674e-01 -1.20314884e+00 -2.70087928e-01 -8.27919245e-01
1.02041990e-01 3.72762412e-01 -2.21570760e-01 -6.91635489e-01
8.26055646e-01 1.13025587e-02 -1.95025432e+00 1.93674698e-01
8.86436284e-01 9.55573559e-01 3.67580727e-02 6.56431675e-01
2.39742314e-03 7.12657452e-01 -3.20641398e-01 -6.67012513e-01
-1.42934516e-01 -3.83816808e-01 -9.57213759e-01 -3.41912389e-01
-1.09677725e-02 2.91530579e-01 -4.30521846e-01 1.69677484e+00
2.86167532e-01 5.87693930e-01 7.88370311e-01 -8.84470165e-01
-1.39670014e+00 3.00946444e-01 5.94535135e-02 5.25493979e-01
4.22950894e-01 -1.95184693e-01 1.44917929e+00 -4.41090196e-01
2.16315135e-01 1.36726785e+00 5.75829625e-01 6.38379991e-01
-8.50910306e-01 -2.69057959e-01 -2.26120099e-01 8.76805067e-01
-1.35657990e+00 6.78602308e-02 4.55282927e-01 -3.64681989e-01
1.10126531e+00 -9.15047601e-02 6.19027019e-01 7.39423990e-01
1.13850999e+00 4.28885549e-01 1.50472760e+00 -7.10297823e-01
7.28783846e-01 1.81506917e-01 6.71196043e-01 4.43803132e-01
1.26456469e-01 5.79192579e-01 -1.04316890e+00 -2.86450167e-03
5.63997729e-03 -7.88604096e-02 3.31875116e-01 -3.52954399e-03
-1.10514295e+00 9.89840984e-01 4.83209044e-01 4.19103444e-01
-3.97790074e-01 1.20934226e-01 1.88958004e-01 1.04364734e-02
2.91403860e-01 4.80820149e-01 -5.03246307e-01 -1.77947670e-01
-5.47949314e-01 3.70409101e-01 7.87902176e-01 6.35277212e-01
1.05657196e+00 -2.40970775e-01 -9.26440135e-02 5.11596389e-02
5.17708182e-01 1.02421987e+00 3.46982092e-01 -9.59091604e-01
-4.25285667e-01 1.71904027e-01 2.50882447e-01 -8.78641665e-01
-8.31198215e-01 -2.17474908e-01 -7.16032565e-01 -2.29896288e-02
1.44908816e-01 -5.18574677e-02 -5.14759362e-01 1.55660439e+00
1.27345949e-01 8.39051753e-02 4.23576742e-01 7.69690454e-01
7.62107313e-01 6.03756189e-01 3.53084445e-01 -1.45109028e-01
1.88029635e+00 -5.78396916e-01 -9.11570966e-01 -1.05281830e-01
5.78848422e-01 -7.65295267e-01 8.22149038e-01 5.63564479e-01
-7.99225748e-01 -5.62781096e-01 -1.01778519e+00 -1.88273638e-01
-9.48310614e-01 -4.13117707e-01 1.13041174e+00 1.28849053e+00
-1.07756841e+00 9.90242243e-01 -9.46567416e-01 -5.62730253e-01
1.46429524e-01 3.48282009e-01 1.31775796e-01 1.59914866e-01
-1.85042286e+00 1.16789985e+00 4.56826985e-01 6.39118180e-02
-2.51187712e-01 -2.35935196e-01 -5.88450074e-01 1.55860871e-01
1.45539045e-01 -1.05928516e+00 1.40790963e+00 -4.64160830e-01
-1.90462601e+00 7.44846344e-01 -3.75260919e-01 -3.29633296e-01
-3.75088185e-01 -1.92575142e-01 -7.04752505e-01 1.16726741e-01
1.26931041e-01 2.06902340e-01 4.53991979e-01 -4.25322503e-01
-6.48667574e-01 -5.37130535e-01 4.33318824e-01 4.23088342e-01
-1.19862512e-01 -8.92335325e-02 1.77879184e-01 5.56182116e-02
9.09991860e-02 -1.25493407e+00 -2.50287443e-01 -9.46432889e-01
-2.27105543e-01 -6.33372486e-01 2.90348213e-02 3.92442644e-01
1.02958930e+00 -2.02944684e+00 1.14657663e-01 7.26125836e-02
1.01083949e-01 -2.17546914e-02 2.46971786e-01 7.46160746e-01
-1.53905163e-02 1.55197129e-01 -2.88787317e-02 -8.75726342e-02
5.79092681e-01 6.72964156e-02 -3.14389050e-01 5.12839437e-01
4.15285192e-02 1.15649366e+00 -1.12135947e+00 -7.02810958e-02
5.06540895e-01 4.00505513e-01 -5.12311161e-01 -4.21093017e-01
-1.55169442e-01 5.31484067e-01 -7.85347700e-01 3.83024454e-01
7.47546017e-01 -5.75592160e-01 1.45152211e-01 -7.99647421e-02
-4.03051943e-01 5.40452480e-01 -9.98861074e-01 1.48127997e+00
-3.90055656e-01 4.98422503e-01 -4.32790518e-01 -1.00076532e+00
2.92619646e-01 1.71390951e-01 1.05759956e-01 -9.60726380e-01
5.49787223e-01 8.64855424e-02 2.76638657e-01 -3.52919251e-01
8.70551586e-01 -8.53477716e-01 -4.40861702e-01 6.66495383e-01
2.19054222e-01 -3.91800433e-01 5.44993937e-01 1.35679573e-01
7.61228442e-01 1.39858201e-02 5.36737323e-01 -4.95329797e-01
9.31181908e-01 3.95960435e-02 2.97748506e-01 1.00308251e+00
-8.49036276e-01 1.71381906e-01 2.45559990e-01 -5.79417229e-01
-7.19712198e-01 -1.07384384e+00 -5.97780824e-01 1.25866413e+00
6.19974971e-01 -7.38272071e-01 -8.31496954e-01 1.55529678e-01
-3.29444259e-01 9.69795167e-01 -8.44328523e-01 -4.83991355e-01
9.85737666e-02 -1.66031444e+00 7.15711191e-02 1.38276771e-01
4.70813304e-01 -1.21935606e+00 -4.47453767e-01 2.10771814e-01
-1.57977954e-01 -1.39905453e+00 4.84550416e-01 1.72370553e-01
-7.38584578e-01 -8.93223166e-01 -2.08240673e-02 -3.44298631e-01
1.39790952e-01 2.88141996e-01 1.08626318e+00 -1.19243875e-01
-1.58421293e-01 6.06614530e-01 -6.78746760e-01 -8.38785112e-01
-5.08182228e-01 -1.45157889e-01 6.15113914e-01 -1.21810541e-01
1.40571833e+00 -3.17127794e-01 -6.94055319e-01 -3.31843764e-01
-9.99321342e-01 -2.74274349e-01 4.84539628e-01 6.17761433e-01
4.01034087e-01 5.62998056e-01 3.16164196e-01 -7.05196857e-01
8.43817890e-01 -3.15347761e-01 -3.98562998e-01 1.78119734e-01
-5.39981306e-01 1.17139257e-01 7.65591443e-01 1.89058214e-01
-1.08597732e+00 -6.57359719e-01 -2.34064832e-02 5.02722025e-01
-5.20026386e-01 8.47944736e-01 3.12510252e-01 -1.36193275e-01
7.62115598e-01 1.56151980e-01 -2.76071131e-01 1.17139623e-01
5.14539063e-01 5.84564567e-01 4.84619774e-02 -4.26958442e-01
6.69450402e-01 9.46499467e-01 5.46173513e-01 -1.00103545e+00
-1.64823020e+00 -4.73442465e-01 -8.60946298e-01 1.36024160e-02
1.35664392e+00 -6.71799719e-01 -1.25803673e+00 5.53079605e-01
-1.03121352e+00 3.60351503e-01 -1.71296820e-01 9.70347881e-01
-8.13819289e-01 5.30197203e-01 -8.25313509e-01 -1.14055896e+00
-2.83984840e-01 -1.13721371e+00 9.38556314e-01 8.77023935e-01
1.55146793e-02 -1.37478769e+00 3.18221599e-01 5.07779002e-01
3.68175000e-01 -3.26659083e-01 9.58071172e-01 -4.57555383e-01
-5.40336668e-01 -2.18587890e-01 3.03896256e-02 1.67459026e-01
-2.83717036e-01 -1.33522332e-01 -1.15152228e+00 1.40981540e-01
3.99178088e-01 -2.68784553e-01 7.36550212e-01 4.66832608e-01
5.36242366e-01 6.30894244e-01 -6.13419600e-02 1.76749960e-01
1.36622715e+00 7.19585493e-02 4.35546249e-01 6.03566051e-01
2.36454353e-01 6.42619908e-01 6.14745200e-01 4.51217145e-01
8.84935439e-01 1.81980774e-01 3.06382328e-01 7.52192438e-01
3.34794998e-01 2.80688673e-01 3.84948283e-01 1.33178198e+00
-4.51169699e-01 -1.59251809e-01 -8.36663485e-01 -6.16838746e-02
-1.83707881e+00 -1.35699356e+00 -2.60988802e-01 2.08622193e+00
2.21282229e-01 1.23676755e-01 -2.98797607e-01 -9.72444415e-02
8.91744614e-01 -7.37931281e-02 -4.70277280e-01 -8.31373394e-01
-3.91652793e-01 6.18460596e-01 3.03765923e-01 3.05684417e-01
-1.44547927e+00 1.31720376e+00 7.21101332e+00 5.35679698e-01
-9.00827050e-01 2.08742827e-01 1.61302418e-01 4.13324505e-01
-2.78932489e-02 2.66712248e-01 -7.76377618e-01 2.34803855e-01
1.66744423e+00 -3.50009382e-01 7.14141488e-01 4.40393955e-01
1.39401764e-01 -5.96408844e-01 -8.94939303e-01 9.74338889e-01
4.44079228e-02 -1.24246657e+00 -1.21894274e-02 1.22842252e-01
9.02717650e-01 4.46674615e-01 4.21016574e-01 4.38179642e-01
2.39667352e-02 -9.69449997e-01 6.31678820e-01 8.19917202e-01
9.03206319e-02 -9.26544964e-01 1.14897037e+00 4.00388062e-01
-8.60691607e-01 -1.61022589e-01 -8.08489919e-01 -8.85317862e-01
2.99382210e-01 5.98229766e-01 4.04623374e-02 9.60542321e-01
1.01284933e+00 8.75602484e-01 -4.48204964e-01 6.84262633e-01
-2.26708412e-01 3.14410955e-01 -1.38668984e-01 -7.19324887e-01
3.84218425e-01 -8.76318157e-01 3.20538133e-01 8.15582931e-01
2.29340762e-01 3.72389019e-01 -2.46434405e-01 1.08721840e+00
4.70412910e-01 1.20442256e-01 -2.70943612e-01 -4.27022398e-01
1.74926832e-01 1.35568810e+00 -1.00251091e+00 -2.87612379e-01
-7.88028419e-01 7.11603284e-01 1.45912200e-01 2.86388211e-02
-7.17941821e-01 -5.02020717e-01 5.50487578e-01 -6.57465219e-01
1.62339821e-01 -4.63601440e-01 -3.17805022e-01 -1.46633255e+00
-5.10381937e-01 -4.27509367e-01 9.80128068e-03 -6.65568888e-01
-1.61372578e+00 4.76960361e-01 -1.48367271e-01 -1.05010486e+00
-2.29087286e-02 -1.31868219e+00 -5.21181583e-01 8.78683627e-01
-1.67823434e+00 -6.96144044e-01 1.91849053e-01 3.89012814e-01
-2.72392213e-01 1.54925673e-03 1.42619073e+00 -1.74517650e-02
-4.76065308e-01 -1.09736957e-01 6.83114767e-01 -2.71227688e-01
5.39030313e-01 -1.47869611e+00 3.86904269e-01 5.84667325e-01
1.10252708e-01 1.17313159e+00 1.02634442e+00 -3.29998255e-01
-1.59799147e+00 -3.65677834e-01 1.08012950e+00 -9.33607578e-01
1.23659647e+00 -3.26600522e-01 -5.29211819e-01 2.48182759e-01
3.69618148e-01 -3.00934464e-01 1.43658566e+00 4.35082287e-01
-7.57365227e-02 2.77137697e-01 -7.91724563e-01 6.51749551e-01
5.70368409e-01 -1.02943361e+00 -1.09220147e+00 7.08658993e-01
2.69905418e-01 -8.44964758e-03 -8.30183387e-01 1.06651232e-01
5.58671951e-01 -1.50275397e+00 7.91812778e-01 -4.88423258e-01
1.44896343e-01 -3.50298136e-01 -2.82919139e-01 -1.16732597e+00
-6.36570275e-01 -6.87888920e-01 5.51802218e-02 2.39266500e-01
2.07007360e-02 -6.48822963e-01 7.66262531e-01 6.88538969e-01
-5.56560867e-02 -1.92549825e-01 -9.87508595e-01 -5.85813224e-01
6.70065582e-01 -8.15220356e-01 4.11678106e-01 5.39245903e-01
5.88356555e-01 6.20002866e-01 3.27127054e-02 1.53072864e-01
8.01157176e-01 2.66363978e-01 1.43551603e-01 -1.59105670e+00
1.10984087e-01 -4.35400099e-01 -7.37397850e-01 -8.11676264e-01
4.12025839e-01 -9.77695942e-01 5.38280932e-03 -1.33965683e+00
3.28204393e-01 2.46227413e-01 -7.27179408e-01 -4.04755473e-01
-1.93167433e-01 4.23280299e-01 1.00145467e-01 1.76306143e-01
-9.92064655e-01 6.76859081e-01 1.26856184e+00 2.06096560e-01
9.73132551e-02 -2.32746173e-02 -9.66913104e-01 1.05642569e+00
8.62184048e-01 -3.89681667e-01 -1.72122359e-01 -5.74960113e-02
1.27359724e+00 -2.96130747e-01 1.62391737e-01 -1.08905232e+00
4.06649858e-01 -1.80995941e-01 2.94437949e-02 -6.62187338e-01
4.95825648e-01 -3.25256854e-01 -6.52269423e-01 4.41490322e-01
4.48172800e-02 5.45700379e-02 -8.49670693e-02 8.59823644e-01
-1.88272864e-01 -4.42100227e-01 7.96798050e-01 -3.65514815e-01
-1.04472268e+00 2.11523727e-01 -9.48407352e-01 -9.25583541e-02
1.04644108e+00 -1.23339333e-01 -5.49254954e-01 -2.16348588e-01
-9.60549474e-01 -1.78196773e-01 2.37148449e-01 3.09201688e-01
8.60755965e-02 -9.09018457e-01 -1.90376773e-01 -2.70635843e-01
2.31888413e-01 -7.19138980e-01 9.15523410e-01 9.79098618e-01
-5.61945379e-01 1.32516861e+00 -2.96071917e-01 -3.51571232e-01
-3.70199025e-01 8.15645516e-01 1.54780105e-01 -1.05627507e-01
-1.70368791e-01 7.12255180e-01 4.24890846e-01 -7.42171824e-01
-5.42883813e-01 -2.68185943e-01 -5.89946449e-01 -1.51495934e-01
6.90648913e-01 5.00443041e-01 1.47820160e-01 -1.23594820e+00
-7.00532317e-01 8.81956160e-01 1.45037785e-01 -2.15413004e-01
1.03167725e+00 -7.62364447e-01 -5.35782456e-01 9.16928649e-01
8.56363475e-01 -3.20488632e-01 -3.59347671e-01 -1.48541451e-01
1.07895501e-01 3.04515272e-01 3.20399821e-01 -5.93352675e-01
-1.56643257e-01 1.23807108e+00 5.39327204e-01 7.97215939e-01
7.34985948e-01 3.43517475e-02 6.89965069e-01 9.21831071e-01
7.09090531e-01 -1.52672291e+00 -8.40545744e-02 1.21837592e+00
5.61392568e-02 -1.41304362e+00 8.37039500e-02 -3.02663863e-01
-5.59591055e-01 1.49809802e+00 1.92596793e-01 -4.20688897e-01
1.24469149e+00 -4.91337389e-01 -4.01652120e-02 -4.96303946e-01
-6.53321326e-01 -5.00639498e-01 3.58146340e-01 4.89192992e-01
8.69333267e-01 3.97296190e-01 -4.02074575e-01 5.97657442e-01
-8.56869221e-01 -1.20856784e-01 1.00404191e+00 1.13519597e+00
-6.53197527e-01 -1.12928081e+00 -5.73546827e-01 2.10197195e-01
-7.26325214e-01 -4.37079519e-01 -7.11866021e-02 2.11063296e-01
2.64580458e-01 1.51682436e+00 1.07372686e-01 -4.26913619e-01
-7.76518285e-02 1.67121187e-01 5.57047725e-01 -8.32914531e-01
-2.94922799e-01 -2.58378625e-01 -4.16349918e-01 -4.69004720e-01
-1.23862457e+00 -7.34240711e-01 -1.45876670e+00 -7.48836040e-01
-6.38846934e-01 5.09023190e-01 8.01842809e-01 1.60070276e+00
4.34877455e-01 6.19376719e-01 1.37033522e-01 -8.92701924e-01
-5.72885156e-01 -8.77712131e-01 -1.11579490e+00 1.08313709e-01
1.20658785e-01 -1.08410072e+00 -4.47033614e-01 -1.47008955e-01] | [5.585886001586914, 4.995668411254883] |
8b5f1138-b5ec-436d-8ee1-b58ba162c873 | is-bert-blind-exploring-the-effect-of-vision | 2303.12513 | null | https://arxiv.org/abs/2303.12513v1 | https://arxiv.org/pdf/2303.12513v1.pdf | Is BERT Blind? Exploring the Effect of Vision-and-Language Pretraining on Visual Language Understanding | Most humans use visual imagination to understand and reason about language, but models such as BERT reason about language using knowledge acquired during text-only pretraining. In this work, we investigate whether vision-and-language pretraining can improve performance on text-only tasks that involve implicit visual reasoning, focusing primarily on zero-shot probing methods. We propose a suite of visual language understanding (VLU) tasks for probing the visual reasoning abilities of text encoder models, as well as various non-visual natural language understanding (NLU) tasks for comparison. We also contribute a novel zero-shot knowledge probing method, Stroop probing, for applying models such as CLIP to text-only tasks without needing a prediction head such as the masked language modelling head of models like BERT. We show that SOTA multimodally trained text encoders outperform unimodally trained text encoders on the VLU tasks while being underperformed by them on the NLU tasks, lending new context to previously mixed results regarding the NLU capabilities of multimodal models. We conclude that exposure to images during pretraining affords inherent visual reasoning knowledge that is reflected in language-only tasks that require implicit visual reasoning. Our findings bear importance in the broader context of multimodal learning, providing principled guidelines for the choice of text encoders used in such contexts. | ['Hadar Averbuch-Elor', 'Michael Fiman', 'Morris Alper'] | 2023-03-21 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Alper_Is_BERT_Blind_Exploring_the_Effect_of_Vision-and-Language_Pretraining_on_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Alper_Is_BERT_Blind_Exploring_the_Effect_of_Vision-and-Language_Pretraining_on_CVPR_2023_paper.pdf | cvpr-2023-1 | ['visual-reasoning', 'visual-reasoning'] | ['computer-vision', 'reasoning'] | [ 2.12691814e-01 5.59381962e-01 -1.81580052e-01 -2.32745200e-01
-5.73431373e-01 -6.89527392e-01 9.57564771e-01 1.27176926e-01
-7.90101230e-01 3.90397668e-01 3.95534784e-01 -9.45628107e-01
1.47909522e-01 -4.49066073e-01 -1.10525596e+00 -3.83509606e-01
3.14583033e-01 5.85037172e-01 -9.24493223e-02 -1.70468107e-01
3.75848383e-01 4.05045301e-01 -1.48808265e+00 8.19498062e-01
5.82037032e-01 5.15408754e-01 5.87012410e-01 8.78785670e-01
-4.61149931e-01 1.38251328e+00 -2.17902213e-01 -5.72816014e-01
-9.31073725e-02 -5.98330140e-01 -8.94894004e-01 -6.62057549e-02
9.68966961e-01 -7.67822981e-01 -3.78488392e-01 6.30642653e-01
3.43468755e-01 1.54093251e-01 8.77093375e-01 -1.08651042e+00
-1.27282870e+00 4.38280135e-01 -4.79243100e-01 4.09586877e-01
5.73661089e-01 6.60469294e-01 8.86439383e-01 -1.15637457e+00
7.46775031e-01 1.58280838e+00 3.56524020e-01 8.09979022e-01
-1.52596414e+00 -4.72462386e-01 3.44819963e-01 5.17927229e-01
-9.11383271e-01 -8.59207094e-01 4.57495034e-01 -6.50114715e-01
1.51669359e+00 3.56403291e-02 6.57428443e-01 1.23148715e+00
2.32233658e-01 1.08878613e+00 1.51934052e+00 -6.59919679e-01
-1.31136015e-01 4.89172190e-01 1.70154795e-01 1.24858201e+00
1.64364263e-01 3.64858121e-01 -1.07988107e+00 3.58594954e-01
9.20313001e-01 -2.39219412e-01 -4.14308816e-01 -5.56644678e-01
-1.59417653e+00 8.22066367e-01 5.62413216e-01 3.19401085e-01
-4.12300557e-01 4.15255547e-01 3.56901646e-01 4.34549302e-01
1.02347009e-01 5.12387812e-01 -2.78025389e-01 -6.03096709e-02
-7.93718755e-01 -3.50975543e-01 5.02478004e-01 6.75168216e-01
9.38366890e-01 3.44028383e-01 -5.24562478e-01 7.38862097e-01
6.68907762e-01 5.62581837e-01 4.48114127e-01 -9.51830566e-01
5.81946909e-01 2.92062312e-01 -5.88287823e-02 -5.29754698e-01
-3.47772390e-01 -1.50053799e-02 -7.57168531e-02 5.11290550e-01
6.75593793e-01 -3.04894187e-02 -1.20089054e+00 1.95589554e+00
-8.57505947e-02 -5.15769243e-01 2.93502301e-01 9.93558705e-01
1.11442006e+00 6.07455909e-01 5.86930633e-01 -2.30626717e-01
1.63545144e+00 -1.00478911e+00 -7.96036720e-01 -7.61235297e-01
8.85711074e-01 -5.74082971e-01 1.64288390e+00 3.31712931e-01
-1.22562289e+00 -6.53783858e-01 -9.44907844e-01 -6.10181332e-01
-5.23993194e-01 -1.58570539e-02 1.00152862e+00 6.77190423e-01
-1.63263464e+00 6.56160116e-02 -5.45301914e-01 -8.87881875e-01
5.24402976e-01 2.79177111e-02 -5.85570574e-01 -2.59188831e-01
-1.00714910e+00 1.50310409e+00 3.66622537e-01 1.60065040e-01
-1.33041871e+00 -3.76017511e-01 -1.30318022e+00 1.18982330e-01
4.80308384e-01 -9.98029530e-01 1.39032006e+00 -1.31756890e+00
-1.12259531e+00 1.44507933e+00 -3.55698526e-01 -2.31444746e-01
1.51621327e-01 -1.16100267e-01 -5.00437580e-02 7.30514944e-01
8.36705416e-02 1.21857929e+00 8.64897668e-01 -1.72264099e+00
1.34716809e-01 -3.21221113e-01 1.73428595e-01 3.76704812e-01
1.10861056e-01 -2.13531867e-01 -1.65637463e-01 -4.12349910e-01
1.25342393e-02 -3.75018746e-01 4.20126140e-01 4.78093237e-01
-9.01767910e-02 -2.91829109e-01 5.17183900e-01 -8.26917470e-01
4.65417922e-01 -1.88728678e+00 2.06513330e-01 -2.76003689e-01
3.30778927e-01 5.78930303e-02 -5.69314122e-01 7.23564565e-01
-2.95208633e-01 9.29267332e-02 6.67928457e-02 -3.64503205e-01
3.02889258e-01 5.69408476e-01 -5.81095874e-01 3.58785897e-01
3.24580967e-01 1.60385668e+00 -7.07796097e-01 -7.05097318e-01
3.41494948e-01 3.77975434e-01 -4.37533975e-01 1.30674496e-01
-4.93598908e-01 4.00631279e-01 2.67673731e-01 5.48835933e-01
3.56609285e-01 -4.32241201e-01 1.80588290e-01 -2.14868888e-01
-1.64373413e-01 -1.72334909e-02 -2.70234019e-01 1.77560914e+00
-5.39599061e-01 1.18032491e+00 1.30095214e-01 -1.03355873e+00
2.87882358e-01 3.37067306e-01 -4.10821050e-01 -1.02962470e+00
2.60619491e-01 -1.88302264e-01 3.63318622e-01 -8.52687657e-01
-4.43469025e-02 -7.98301995e-01 4.55279619e-01 6.29061639e-01
5.17542720e-01 -1.47383466e-01 7.86529183e-02 6.21660650e-01
6.24151826e-01 2.07309648e-01 3.47643316e-01 -4.70586680e-02
2.44023189e-01 9.71887931e-02 -3.52317005e-01 1.21033049e+00
-3.49976718e-01 1.47789627e-01 6.45149589e-01 -1.38117641e-01
-8.51977408e-01 -1.40768921e+00 -7.06761405e-02 1.49268568e+00
-1.25549108e-01 1.25256712e-02 -5.00827253e-01 -4.51448977e-01
-5.97713776e-02 1.42502463e+00 -9.69228446e-01 -2.89688617e-01
-3.98498476e-02 -7.71285743e-02 4.56744730e-01 6.84170961e-01
3.38687390e-01 -1.43395495e+00 -7.63152242e-01 -2.26333261e-01
-6.04381412e-02 -1.31486940e+00 -8.63971636e-02 8.69531631e-02
-7.38168836e-01 -1.04803574e+00 -1.10691619e+00 -9.16684270e-01
7.43377805e-01 3.58512670e-01 1.01262426e+00 2.70661116e-01
-3.39027047e-01 1.47069049e+00 -1.78190663e-01 -4.08777982e-01
-4.37630296e-01 -6.42948747e-01 -2.73711562e-01 -2.39023045e-01
4.42234516e-01 -1.10121682e-01 -3.70128661e-01 -1.80021133e-02
-8.58251452e-01 7.22381413e-01 1.12008953e+00 7.92805552e-01
1.21873692e-01 -6.87965393e-01 2.66459525e-01 -4.23960835e-01
6.64750218e-01 -1.19420744e-01 -1.31391346e-01 4.84279662e-01
-2.88833976e-01 1.70535758e-01 9.29904077e-03 -6.50932193e-01
-1.40951765e+00 -3.80509347e-01 1.24332927e-01 -4.25344944e-01
-3.36000025e-01 5.64830005e-01 2.31018692e-01 -4.57883179e-01
7.75875568e-01 4.97595519e-01 2.55873799e-01 -5.85375577e-02
6.87914431e-01 1.59694239e-01 4.80843306e-01 -6.94681823e-01
7.04230666e-01 4.65428084e-01 -2.62150228e-01 -1.23519325e+00
-8.84950817e-01 -1.99746296e-01 -5.21675110e-01 -3.51857185e-01
1.49589491e+00 -1.12852156e+00 -9.97566342e-01 4.24611121e-02
-1.38747370e+00 -8.96994233e-01 -1.10000126e-01 5.82048833e-01
-9.23151135e-01 6.16042018e-01 -5.41681826e-01 -1.05725431e+00
8.98918323e-03 -1.03619742e+00 9.82896924e-01 -1.11562051e-01
-3.18539113e-01 -1.21968031e+00 -8.27326532e-03 7.73017347e-01
2.19813451e-01 -2.21264362e-01 1.44649684e+00 -4.78384197e-01
-7.49032676e-01 3.63370001e-01 -8.14481139e-01 2.16619726e-02
-5.08994937e-01 -5.01980186e-01 -1.34396601e+00 -1.34355232e-01
-2.59963851e-02 -1.02279747e+00 1.38957834e+00 2.56525397e-01
8.17666292e-01 -2.12926894e-01 -2.38688722e-01 3.74446452e-01
1.26499093e+00 5.95160089e-02 6.29519403e-01 -1.43311638e-02
5.80147743e-01 8.38170290e-01 1.64038569e-01 -2.07145303e-01
7.35395432e-01 3.93983126e-01 3.25552166e-01 -2.77354777e-01
-2.97281265e-01 -3.49718273e-01 6.35942459e-01 2.61384130e-01
2.32320372e-02 -1.38995618e-01 -1.20715833e+00 6.82956100e-01
-1.60143566e+00 -1.00518453e+00 1.39145598e-01 2.02122355e+00
8.99130523e-01 -6.60088956e-02 -2.05419451e-01 -2.22123936e-01
2.14893177e-01 2.30289608e-01 -4.31474268e-01 -7.39744246e-01
1.32241128e-02 1.01187959e-01 6.36739954e-02 7.96951294e-01
-6.41065419e-01 1.33621323e+00 6.53992367e+00 1.24872327e-01
-8.20387900e-01 3.95120412e-01 2.30227351e-01 8.52453709e-03
-5.62971413e-01 -2.30177660e-02 -2.89475411e-01 -3.39738101e-01
7.93507457e-01 2.68694431e-01 6.84618771e-01 3.47801983e-01
6.03315942e-02 -7.80500412e-01 -1.47980118e+00 1.05343568e+00
6.12866223e-01 -1.05434000e+00 4.60984707e-01 -8.46414790e-02
2.53323078e-01 -1.25164568e-01 3.21867138e-01 6.42596185e-01
-5.05645461e-02 -1.45517957e+00 7.07276642e-01 7.26236820e-01
9.12509978e-01 -1.51752159e-01 3.50791395e-01 4.04627502e-01
-6.60162985e-01 7.19402879e-02 -2.42616177e-01 -4.48509872e-01
1.68695807e-01 -1.91184044e-01 -1.02914238e+00 -1.67713180e-01
4.04062897e-01 2.52216101e-01 -8.87536228e-01 5.82502961e-01
-4.70692039e-01 3.78385603e-01 2.59685844e-01 -1.98263511e-01
2.97510207e-01 3.78623635e-01 2.79170454e-01 1.15227568e+00
-1.60654098e-01 1.10579379e-01 -2.50560850e-01 1.11286974e+00
2.51765519e-01 2.54904717e-01 -9.25983429e-01 -6.21461749e-01
-1.24905758e-01 8.20420682e-01 -5.99671006e-01 -5.15607536e-01
-8.94741654e-01 1.21743262e+00 6.97164953e-01 9.78779674e-01
-5.57846069e-01 -2.80890092e-02 1.96329951e-01 -2.11531781e-02
1.46970630e-01 -5.43383896e-01 -3.49468738e-01 -1.31785941e+00
-3.61555040e-01 -6.64330602e-01 2.32426152e-01 -1.77177107e+00
-1.03586853e+00 1.21519826e-01 2.02114806e-01 -4.34588075e-01
-2.54342079e-01 -1.39373052e+00 -4.29878920e-01 1.05605221e+00
-1.58824456e+00 -1.39647484e+00 -4.01908636e-01 8.16081524e-01
6.90607071e-01 -1.15265027e-01 8.19728374e-01 -4.68442947e-01
-9.45383757e-02 3.12884331e-01 -6.51438832e-01 1.78746060e-01
8.86726737e-01 -1.15907776e+00 -6.96644560e-02 5.89559734e-01
5.97738028e-01 9.45233285e-01 7.14342296e-01 -5.56018531e-01
-1.63004720e+00 -2.16860563e-01 6.97475791e-01 -6.49532378e-01
6.41322851e-01 -4.75012809e-01 -9.59018409e-01 1.21531165e+00
7.96954572e-01 -3.93375546e-01 7.79679954e-01 2.31248349e-01
-6.06477201e-01 4.61696208e-01 -8.73309672e-01 8.84431481e-01
9.69011068e-01 -1.14658332e+00 -1.35234320e+00 4.88302737e-01
6.21472299e-01 -4.48384471e-02 -3.18346649e-01 -2.00259034e-02
5.00402570e-01 -9.19200540e-01 1.14414191e+00 -9.80943978e-01
6.93883002e-01 -9.74553823e-03 -1.33688256e-01 -1.12524855e+00
-2.66207933e-01 1.77492239e-02 -1.45064294e-01 6.57973170e-01
4.63167787e-01 -7.82945931e-01 3.47464800e-01 4.48258877e-01
8.02722424e-02 -1.66309968e-01 -8.83422792e-01 -2.90828854e-01
1.38126060e-01 -5.45942128e-01 -5.38714468e-01 1.05080521e+00
1.40925929e-01 6.19830728e-01 -1.11005362e-02 1.23025760e-01
5.50576925e-01 -1.84410721e-01 7.88521707e-01 -9.90611315e-01
-4.07235324e-01 -4.76836771e-01 -2.16806680e-01 -8.80424619e-01
6.07838988e-01 -1.08640075e+00 -3.02281627e-03 -2.01049948e+00
4.53135401e-01 6.44479573e-01 1.09713219e-01 9.12421703e-01
1.10587962e-02 2.11796075e-01 5.35497904e-01 -1.65136270e-02
-5.22221029e-01 5.32740772e-01 1.66078079e+00 -8.30031410e-02
1.55522257e-01 -6.95530236e-01 -1.00056648e+00 7.93213487e-01
3.99086833e-01 2.33255371e-01 -6.72537386e-01 -7.06222951e-01
3.49973947e-01 2.64615983e-01 1.12837386e+00 -7.90681899e-01
1.26824424e-01 -1.01200759e-01 7.90831149e-01 -4.13297206e-01
5.44719994e-01 -6.81540906e-01 -4.81562525e-01 4.80831653e-01
-2.94719100e-01 -4.05003540e-02 9.49374557e-01 4.61709291e-01
1.37116805e-01 -2.72383273e-01 7.19136298e-01 -5.42807221e-01
-1.35725677e+00 -3.05346608e-01 -8.03309619e-01 1.85332283e-01
8.33865762e-01 -4.46363837e-01 -6.60561800e-01 -7.78068304e-01
-1.18074441e+00 2.87662864e-01 3.90801102e-01 3.24229777e-01
9.63400424e-01 -9.99867439e-01 -4.68034774e-01 7.26357698e-02
4.47958440e-01 -6.15118146e-01 3.80055815e-01 1.33185256e+00
-4.66501743e-01 8.42278659e-01 -2.81420320e-01 -5.39429605e-01
-1.09894621e+00 1.02117872e+00 4.29083675e-01 3.85543942e-01
-5.30062914e-01 1.07068050e+00 1.08834481e+00 -3.95032227e-01
1.37674659e-01 -1.84075281e-01 -1.89956710e-01 2.52028793e-01
6.00272894e-01 -1.36465892e-01 -5.36171913e-01 -6.24797702e-01
-3.19197804e-01 5.01991630e-01 -1.32209174e-02 -7.11136520e-01
7.40167558e-01 -2.46893689e-01 -5.87379299e-02 9.48586106e-01
9.03902769e-01 -2.37819359e-01 -1.09470761e+00 -1.16798192e-01
-2.84995198e-01 -3.14038038e-01 1.07655592e-01 -1.39203119e+00
-5.66881895e-01 1.65708828e+00 4.83329415e-01 -3.29388738e-01
8.20606053e-01 5.47865868e-01 -1.67415142e-01 7.58060038e-01
4.73112315e-02 -8.70963752e-01 5.02631962e-01 4.48880017e-01
1.29000759e+00 -1.55107677e+00 -6.97693042e-03 -6.84407540e-03
-1.03181779e+00 9.57878828e-01 8.38874876e-01 3.19732279e-01
-6.13176152e-02 -3.55274975e-01 3.04583192e-01 -4.79592085e-01
-9.70887125e-01 -7.56132722e-01 5.58132768e-01 8.67888689e-01
5.61158001e-01 -3.07316214e-01 -2.55262945e-02 2.92468071e-01
6.07520388e-03 -1.65014774e-01 2.75298983e-01 8.12547803e-01
-6.79171801e-01 -3.00740629e-01 -6.21634126e-01 2.96619654e-01
1.53988346e-01 -6.35532737e-01 -6.51209116e-01 1.20249498e+00
-1.01777725e-01 7.89556980e-01 2.46060342e-01 1.52437583e-01
1.36248678e-01 7.93533623e-01 1.02476180e+00 -7.86715090e-01
-7.58359656e-02 6.81757927e-03 3.34971905e-01 -4.98216212e-01
-3.76864821e-01 -6.41117156e-01 -1.17315888e+00 -4.49956059e-02
6.60725757e-02 -4.39568132e-01 4.08172905e-01 1.25415361e+00
-1.36501476e-01 5.45013070e-01 -4.92861927e-01 -9.48448539e-01
-1.79451093e-01 -9.67907608e-01 -3.74632329e-01 1.50509700e-01
5.72564840e-01 -9.66692626e-01 -4.09800440e-01 -1.26258293e-02] | [10.837738037109375, 1.7375761270523071] |
fa618b00-d035-466a-a13c-77dfbf1131d9 | spatio-temporal-point-processes-with-deep-non | 2211.11179 | null | https://arxiv.org/abs/2211.11179v1 | https://arxiv.org/pdf/2211.11179v1.pdf | Spatio-temporal point processes with deep non-stationary kernels | Point process data are becoming ubiquitous in modern applications, such as social networks, health care, and finance. Despite the powerful expressiveness of the popular recurrent neural network (RNN) models for point process data, they may not successfully capture sophisticated non-stationary dependencies in the data due to their recurrent structures. Another popular type of deep model for point process data is based on representing the influence kernel (rather than the intensity function) by neural networks. We take the latter approach and develop a new deep non-stationary influence kernel that can model non-stationary spatio-temporal point processes. The main idea is to approximate the influence kernel with a novel and general low-rank decomposition, enabling efficient representation through deep neural networks and computational efficiency and better performance. We also take a new approach to maintain the non-negativity constraint of the conditional intensity by introducing a log-barrier penalty. We demonstrate our proposed method's good performance and computational efficiency compared with the state-of-the-art on simulated and real data. | ['Yao Xie', 'Xiuyuan Cheng', 'Zheng Dong'] | 2022-11-21 | null | null | null | null | ['point-processes'] | ['methodology'] | [-2.93719135e-02 -3.53210092e-01 8.95149484e-02 -1.80266351e-01
-3.53507251e-01 -2.17278972e-01 7.26697206e-01 2.22136423e-01
-4.47550327e-01 6.41561985e-01 4.15807724e-01 -3.89871091e-01
-5.03116131e-01 -8.68339360e-01 -8.14919651e-01 -8.74375820e-01
-3.41128170e-01 4.99177039e-01 1.51891872e-01 1.36772245e-02
1.03963390e-01 6.26912177e-01 -1.22005033e+00 -1.84504732e-01
8.09867918e-01 7.72946000e-01 -3.80951017e-02 5.56869805e-01
-1.56457394e-01 8.54669094e-01 -3.52004409e-01 -1.18795857e-01
1.63522705e-01 -2.02070698e-01 -3.09307575e-01 -1.13147780e-01
-2.65770406e-01 -1.78486750e-01 -6.07562423e-01 7.66673625e-01
4.96829778e-01 3.14829946e-01 9.34375465e-01 -1.01526487e+00
-7.72889018e-01 2.89015800e-01 -9.69853401e-01 4.93367553e-01
-3.66302691e-02 -8.43071342e-02 9.28479314e-01 -6.95283175e-01
1.89586177e-01 1.40592575e+00 9.25833046e-01 1.32987157e-01
-1.29836190e+00 -4.55777586e-01 3.64033699e-01 8.38136524e-02
-1.28452146e+00 -1.58076793e-01 7.56791413e-01 -2.91167855e-01
8.21991265e-01 5.71414887e-04 7.10650206e-01 1.20111156e+00
4.90530521e-01 1.01736593e+00 7.96150327e-01 2.34631542e-02
2.81397998e-01 -6.12040162e-01 7.42975399e-02 4.96181905e-01
1.71489805e-01 -4.92980592e-02 -3.31663877e-01 -6.25639856e-01
1.30431116e+00 7.19848335e-01 -1.35259107e-01 -2.26426139e-01
-1.30304253e+00 7.16670692e-01 4.04526502e-01 3.43764812e-01
-8.35102499e-01 6.04299724e-01 3.08904588e-01 3.03733088e-02
7.35853851e-01 -1.68208644e-01 -3.66194516e-01 -2.58293182e-01
-8.60882580e-01 2.12966293e-01 8.79504442e-01 4.99609560e-01
5.06127357e-01 5.82641140e-02 -4.76249814e-01 7.91079402e-01
5.13423860e-01 6.62150443e-01 2.67567128e-01 -5.71885049e-01
5.06910324e-01 5.42732716e-01 2.23184720e-01 -9.54875171e-01
-4.53283250e-01 -6.47484541e-01 -1.54800546e+00 -3.13595757e-02
3.80664349e-01 -2.05487028e-01 -9.70583260e-01 1.68760169e+00
2.39437699e-01 7.93545187e-01 -1.56824440e-01 7.39363432e-01
1.30473718e-01 1.10600257e+00 2.51535326e-01 -2.80196816e-01
1.10339272e+00 -5.49827635e-01 -8.36148918e-01 1.36041284e-01
3.07386518e-01 -4.66302633e-01 7.62360811e-01 6.38191625e-02
-1.05198133e+00 -3.30466211e-01 -5.91432452e-01 -2.34531481e-02
-9.49514806e-02 -4.14783396e-02 7.78620720e-01 1.12736158e-01
-9.64620650e-01 7.01446235e-01 -1.31932044e+00 -1.77597523e-01
4.14489955e-01 3.57203573e-01 -1.75896689e-01 7.68671185e-02
-1.13447583e+00 3.29918474e-01 -9.85860676e-02 5.35265028e-01
-8.40737998e-01 -8.60825717e-01 -5.89246571e-01 2.67050773e-01
2.55455315e-01 -9.27246392e-01 9.47706878e-01 -7.86791980e-01
-1.53739202e+00 2.13947684e-01 -3.80368114e-01 -4.21743840e-01
6.21258378e-01 -5.46572149e-01 -1.94062948e-01 -7.52578676e-02
5.08187003e-02 -3.00683994e-02 9.81947362e-01 -9.29063737e-01
-4.66316700e-01 -5.15420914e-01 -2.18698561e-01 2.68664598e-01
-1.60387442e-01 -7.04536512e-02 -4.26259488e-01 -8.09343755e-01
1.89039245e-01 -9.70107138e-01 -6.07101560e-01 -3.54745463e-02
-2.19241038e-01 -3.91910344e-01 8.43296409e-01 -6.28281951e-01
1.10310805e+00 -2.12790298e+00 8.70701671e-02 2.40017846e-01
2.63278842e-01 1.99826047e-01 -1.53880909e-01 5.28144598e-01
-1.00947618e-01 9.23821777e-02 -6.00400031e-01 -6.62455022e-01
-1.00989811e-01 3.01465422e-01 -3.33730698e-01 8.08009923e-01
4.11050886e-01 8.63611758e-01 -1.02790916e+00 -2.21332088e-02
2.70878583e-01 9.35277820e-01 -3.30167294e-01 8.15334171e-02
1.76822729e-02 5.08420706e-01 -5.81006706e-01 3.48565370e-01
6.27902091e-01 -3.96975547e-01 -2.21353024e-01 1.59391820e-01
-1.15166582e-01 1.41601607e-01 -1.35388362e+00 1.38628674e+00
-4.92368013e-01 4.57867354e-01 1.84296757e-01 -1.01209903e+00
8.10421109e-01 4.31744158e-01 8.37733030e-01 -4.18064803e-01
-1.75511658e-01 4.08531465e-02 -5.14631495e-02 -1.81004614e-01
3.52495432e-01 -3.88388276e-01 6.62957504e-02 5.86689770e-01
-3.39310586e-01 4.74401861e-01 1.75196480e-03 -1.86231703e-01
1.37861025e+00 3.99493016e-02 -9.39987134e-03 -4.28704977e-01
6.16402388e-01 -5.96248209e-01 6.23988390e-01 7.05888271e-01
5.82360625e-02 4.88328248e-01 6.44071281e-01 -4.44842219e-01
-1.12481070e+00 -1.24568760e+00 5.85278943e-02 9.71824586e-01
-1.89352900e-01 7.21614584e-02 -2.78057516e-01 -2.45296910e-01
1.44044891e-01 4.10428226e-01 -7.18118131e-01 -8.47370028e-02
-7.83107102e-01 -1.19706321e+00 5.05206168e-01 6.83638811e-01
5.58672845e-01 -9.22396243e-01 -1.10917270e-01 5.56126297e-01
-7.29805529e-02 -7.90830016e-01 -2.82108814e-01 2.20023900e-01
-1.18851721e+00 -7.39532173e-01 -1.07045805e+00 -4.40662712e-01
5.27823508e-01 1.16771296e-01 9.15659249e-01 -1.89948007e-01
1.37002885e-01 2.82495379e-01 -7.12865889e-02 -4.42255080e-01
-1.39819711e-01 1.36090532e-01 1.38142481e-01 3.29600275e-01
3.70058805e-01 -8.57309282e-01 -9.36166167e-01 1.99710682e-01
-1.26608562e+00 -3.23521376e-01 7.76091814e-01 7.16051340e-01
6.80569291e-01 1.80573925e-01 5.04828334e-01 -5.41147053e-01
1.08330083e+00 -6.84323728e-01 -5.05415142e-01 1.44970529e-02
-2.21838921e-01 3.39572787e-01 6.47794127e-01 -5.06116569e-01
-1.16405475e+00 -1.25096351e-01 -1.30979732e-01 -7.50057757e-01
-3.36965434e-02 6.90758467e-01 1.82425469e-01 4.34953570e-01
2.35443816e-01 2.43075043e-01 -8.17362070e-02 -5.19220948e-01
1.59790397e-01 2.88109303e-01 3.32171172e-01 -5.64925075e-01
6.59345686e-01 9.78079319e-01 4.23191488e-01 -9.24307346e-01
-5.96404910e-01 -6.90436661e-01 -4.90342438e-01 1.35516942e-01
7.60064125e-01 -9.47722197e-01 -9.76774216e-01 7.78899252e-01
-1.36214948e+00 -2.48341367e-01 -3.03453922e-01 5.92725873e-01
-4.63540316e-01 3.63961011e-01 -1.07871938e+00 -1.16900551e+00
-3.81515652e-01 -6.32595658e-01 1.28921139e+00 -6.99031949e-02
2.24404156e-01 -1.30384302e+00 4.04937744e-01 -2.99468130e-01
3.77056688e-01 4.13392186e-01 8.60313594e-01 -4.19413596e-01
-5.18489003e-01 -3.17923635e-01 -3.93905520e-01 2.50726610e-01
2.70111561e-01 2.87321750e-02 -5.46949208e-01 -7.39878118e-02
2.34176740e-01 5.83573282e-01 1.04477751e+00 7.99985468e-01
9.51982081e-01 -1.65977031e-01 -4.30798978e-01 5.02889335e-01
1.34441304e+00 -5.37385605e-02 6.83951378e-01 6.68067187e-02
1.00893259e+00 4.61440861e-01 1.19045213e-01 5.94405472e-01
2.82152086e-01 2.80418456e-01 4.01412010e-01 -2.15222239e-01
5.06937981e-01 -2.35861510e-01 3.47941250e-01 1.13391399e+00
-5.81984937e-01 -2.94363737e-01 -1.10388958e+00 6.77640498e-01
-2.36166334e+00 -1.11909235e+00 -6.13399029e-01 2.27350068e+00
3.42348725e-01 9.53007564e-02 5.22480644e-02 7.63329566e-02
7.56590009e-01 2.66623974e-01 -5.04342079e-01 -2.24420011e-01
-1.09413564e-01 6.45427480e-02 6.78836405e-01 3.50302577e-01
-1.27985620e+00 5.73705196e-01 6.54728603e+00 6.38364315e-01
-9.65541065e-01 1.73592150e-01 5.59407890e-01 -1.16663307e-01
-1.65447593e-01 -4.02073324e-01 -6.81277812e-01 4.28845227e-01
1.08621383e+00 1.06818452e-02 2.14872107e-01 5.31609118e-01
6.53154612e-01 1.48985386e-01 -1.01888514e+00 9.85141337e-01
-3.70349944e-01 -1.18941772e+00 9.32080075e-02 2.74706334e-01
7.31941283e-01 2.88617164e-01 6.26258031e-02 3.41399521e-01
4.55317348e-01 -1.04473042e+00 3.84981096e-01 1.01928508e+00
2.81674445e-01 -1.01654720e+00 7.97514081e-01 4.51227099e-01
-1.33611381e+00 -2.26954669e-01 -5.60826302e-01 -3.59254122e-01
5.37474632e-01 1.06969011e+00 -3.29126120e-01 4.83883291e-01
7.05068290e-01 1.13581252e+00 -1.42477319e-01 9.27142799e-01
-1.32442504e-01 7.31895626e-01 -6.35170698e-01 -4.79474142e-02
5.78470230e-01 -5.75770617e-01 8.43223393e-01 1.17310214e+00
5.72139740e-01 -5.69807999e-02 -6.07073717e-02 9.08647180e-01
-4.20554075e-03 -1.16249315e-01 -7.23385572e-01 -6.51136637e-02
-8.05929974e-02 9.35630679e-01 -7.98807263e-01 -1.64411455e-01
-4.57932115e-01 1.02463007e+00 2.56809175e-01 7.26398349e-01
-8.47946465e-01 1.17789898e-02 1.03535819e+00 3.32182944e-01
4.31424260e-01 -6.53035998e-01 -2.06811041e-01 -1.09213841e+00
1.88462600e-01 -4.19914067e-01 3.22817445e-01 -5.08640289e-01
-1.78291571e+00 4.06171471e-01 -4.18577529e-03 -1.13562167e+00
-3.33556324e-01 -4.81451720e-01 -7.08203197e-01 1.12931097e+00
-1.50412953e+00 -1.13426757e+00 2.78189443e-02 8.93695951e-01
3.15730333e-01 1.98480591e-01 4.59201127e-01 3.90292794e-01
-5.79911768e-01 -1.61821768e-01 5.99105418e-01 4.72979620e-02
4.00721490e-01 -1.28169453e+00 6.97964787e-01 7.47374058e-01
-9.97062996e-02 8.24085653e-01 7.06981540e-01 -8.35092068e-01
-1.34194946e+00 -1.14664519e+00 6.94858074e-01 -3.97593468e-01
1.04888237e+00 -3.83864164e-01 -1.19752610e+00 5.94350874e-01
-5.01662605e-02 1.14977136e-01 5.41585803e-01 3.71397436e-01
9.31088507e-05 -8.52196664e-02 -7.49155939e-01 7.51376271e-01
1.05841732e+00 -4.98366684e-01 -5.52415967e-01 7.79483691e-02
5.98371565e-01 3.71374972e-02 -6.77640676e-01 5.94223678e-01
2.97403485e-01 -7.01729834e-01 9.68492627e-01 -4.94708329e-01
4.61162895e-01 -4.34911668e-01 7.18782172e-02 -1.23894238e+00
-6.99068367e-01 -6.68373346e-01 -4.75619137e-01 1.01949883e+00
1.73579067e-01 -7.28030920e-01 7.45423555e-01 6.01969600e-01
1.26621336e-01 -8.71867955e-01 -1.18982148e+00 -6.37536347e-01
-2.87582353e-03 -6.05541468e-01 4.79576170e-01 7.87491679e-01
-5.43654442e-01 4.59120780e-01 -5.36983192e-01 3.36178809e-01
7.18204558e-01 -1.71460778e-01 4.81484026e-01 -1.61949611e+00
-2.87596285e-01 -4.96546030e-01 -4.55896825e-01 -1.30711269e+00
2.03131661e-01 -4.50064659e-01 -2.24962272e-02 -1.83422935e+00
5.40486537e-02 -3.97058845e-01 -7.13992953e-01 1.41174555e-01
-2.73372620e-01 -7.48733506e-02 -1.38298213e-01 5.06548703e-01
-3.94717336e-01 1.03012562e+00 1.21669757e+00 -8.45011789e-03
-4.76381242e-01 3.03843230e-01 -3.15021873e-01 8.05074990e-01
6.38833404e-01 -5.95242560e-01 -4.52722907e-01 -4.90867436e-01
5.47179937e-01 9.20361951e-02 5.35445809e-01 -9.31277096e-01
3.22575480e-01 -6.73798844e-02 4.51089114e-01 -7.25651264e-01
4.03771311e-01 -9.38929558e-01 2.30996549e-01 3.07947487e-01
-1.20580480e-01 3.80307168e-01 2.55367488e-01 1.25378442e+00
-3.23007494e-01 1.13921672e-01 3.61109614e-01 -1.19644813e-01
-1.91727236e-01 6.87135220e-01 -6.74506903e-01 -3.54183376e-01
7.79457927e-01 1.41612114e-02 -3.31900530e-02 -6.73598289e-01
-6.72874510e-01 9.20029059e-02 1.61019817e-01 4.77910995e-01
5.41567147e-01 -1.39841962e+00 -7.01025665e-01 1.41376078e-01
-3.17093760e-01 2.21538678e-01 3.92350435e-01 1.03001750e+00
-3.94970477e-01 2.29983076e-01 1.12831376e-01 -7.85309732e-01
-7.43149817e-01 5.76868296e-01 3.41260850e-01 -8.20839047e-01
-8.22770298e-01 3.89070272e-01 5.44334412e-01 -4.04192358e-01
1.38040092e-02 -5.91474891e-01 -7.71309203e-03 -8.06225091e-02
4.66150671e-01 4.88861710e-01 -1.76300496e-01 -5.27188063e-01
-2.62666851e-01 4.57778335e-01 1.94373861e-01 -2.72975862e-01
1.56832683e+00 -1.50469661e-01 -3.29475433e-01 1.00251842e+00
1.08607233e+00 -3.28236639e-01 -1.40841365e+00 -2.42653370e-01
1.23555362e-01 -6.06002174e-02 8.58952329e-02 -1.40128911e-01
-8.51315975e-01 9.33786988e-01 4.59152877e-01 4.56373632e-01
9.18747604e-01 -1.63482770e-01 7.71298468e-01 5.11245072e-01
1.58602461e-01 -1.03982759e+00 -1.52355628e-02 9.55863535e-01
8.74471545e-01 -8.91480684e-01 -1.34297639e-01 -2.37318680e-01
-3.00501227e-01 8.14772308e-01 -3.83768207e-03 -6.19700730e-01
1.21091580e+00 1.54798791e-01 -2.19905183e-01 -1.75868958e-01
-7.79096305e-01 -1.07453175e-01 1.01290278e-01 4.02639925e-01
4.79609936e-01 8.06362033e-02 -1.63003907e-01 3.18822920e-01
3.30546677e-01 1.27675682e-01 2.85567909e-01 9.03760195e-01
-8.20290074e-02 -7.74012566e-01 -4.51401353e-01 5.65686285e-01
-8.05650651e-01 -1.51521742e-01 3.86799350e-02 5.62673509e-01
-3.84534955e-01 7.28251517e-01 3.74537885e-01 1.23167001e-01
3.87717843e-01 2.53278632e-02 1.20922141e-01 -3.31273973e-01
-5.54492056e-01 2.97513872e-01 -2.77786374e-01 -4.31020826e-01
-5.77936292e-01 -7.38218427e-01 -9.17959988e-01 -5.14248788e-01
3.16677280e-02 1.06132114e-02 7.11740911e-01 9.90113199e-01
4.34593409e-01 8.10027838e-01 4.70908344e-01 -9.58553255e-01
-4.04965401e-01 -1.12822914e+00 -6.96539104e-01 4.52465385e-01
5.36839783e-01 -6.99267566e-01 -3.38220149e-01 -1.23540699e-01] | [6.919344902038574, 3.3782198429107666] |
0a38a76a-be6c-4c2b-b119-52d758c59add | trueteacher-learning-factual-consistency | 2305.11171 | null | https://arxiv.org/abs/2305.11171v1 | https://arxiv.org/pdf/2305.11171v1.pdf | TrueTeacher: Learning Factual Consistency Evaluation with Large Language Models | Factual consistency evaluation is often conducted using Natural Language Inference (NLI) models, yet these models exhibit limited success in evaluating summaries. Previous work improved such models with synthetic training data. However, the data is typically based on perturbed human-written summaries, which often differ in their characteristics from real model-generated summaries and have limited coverage of possible factual errors. Alternatively, large language models (LLMs) have recently shown promising results in directly evaluating generative tasks, but are too computationally expensive for practical use. Motivated by these limitations, we introduce TrueTeacher, a method for generating synthetic data by annotating diverse model-generated summaries using a LLM. Unlike prior work, TrueTeacher does not rely on human-written summaries, and is multilingual by nature. Experiments on the TRUE benchmark show that a student model trained using our data, substantially outperforms both the state-of-the-art model with similar capacity, and the LLM teacher. In a systematic study, we compare TrueTeacher to existing synthetic data generation methods and demonstrate its superiority and robustness to domain-shift. Using the the mFACE dataset, we also show that our method generalizes to multilingual scenarios. Finally, we release a large-scale synthetic dataset with 1.4M examples generated using TrueTeacher. | ['Idan Szpektor', 'Chen Elkind', 'Roee Aharoni', 'Jonathan Herzig', 'Zorik Gekhman'] | 2023-05-18 | null | null | null | null | ['synthetic-data-generation', 'synthetic-data-generation'] | ['medical', 'miscellaneous'] | [ 3.73282395e-02 5.74668407e-01 -3.42828840e-01 -3.62571925e-01
-1.70848882e+00 -7.15634167e-01 1.30167150e+00 2.04032332e-01
-5.50782494e-02 1.54013956e+00 5.49858510e-01 -3.21588218e-01
2.38269404e-01 -6.80617511e-01 -1.14734972e+00 -2.49823332e-01
2.81864345e-01 1.07362092e+00 -9.51678604e-02 -1.10872760e-01
1.61090255e-01 -2.24161968e-01 -1.39768970e+00 3.94117892e-01
1.52644861e+00 2.40286946e-01 -6.63682446e-02 6.50505006e-01
-7.84141663e-03 1.02622974e+00 -1.40193617e+00 -7.48894513e-01
-2.12643206e-01 -7.05397785e-01 -7.25125611e-01 -6.47008568e-02
7.72417247e-01 -3.50127339e-01 -9.05731507e-03 8.10823798e-01
6.92902327e-01 2.29043998e-02 9.69588280e-01 -1.53129208e+00
-1.01789737e+00 1.19791567e+00 -1.18858246e-02 -1.90305755e-01
6.86400294e-01 2.60278523e-01 8.73214662e-01 -9.65911090e-01
9.42766011e-01 1.33260024e+00 7.88345397e-01 6.45745754e-01
-1.51432240e+00 -7.44241595e-01 -6.09595738e-02 -1.22342911e-02
-1.25194478e+00 -6.68265224e-01 5.53148568e-01 -3.41849148e-01
9.82166588e-01 2.25887567e-01 3.54789734e-01 1.89986145e+00
2.75057573e-02 1.03745103e+00 1.19534457e+00 -5.28312147e-01
2.58004636e-01 3.97688746e-01 -8.91479254e-02 3.34567696e-01
5.66529751e-01 6.35618642e-02 -6.27376556e-01 -4.05004472e-01
1.38762072e-01 -7.49765515e-01 -3.97827476e-01 -4.60192002e-02
-1.52878141e+00 8.31553340e-01 -1.31784350e-01 -4.19800244e-02
-1.79255441e-01 2.95856982e-01 4.60211366e-01 1.96596682e-01
7.06437886e-01 7.16518641e-01 -1.90195084e-01 -4.49959099e-01
-1.39650726e+00 1.02244449e+00 1.12005830e+00 1.47423720e+00
2.92856902e-01 2.98620641e-01 -7.07950532e-01 8.72823834e-01
2.10327357e-01 7.47823596e-01 6.17136300e-01 -1.23332632e+00
7.61552334e-01 2.06303522e-01 3.97014469e-01 -7.27561712e-01
6.67666495e-02 -3.48706335e-01 -8.08074772e-01 -1.99724957e-01
3.73218924e-01 -6.59099594e-02 -7.17297256e-01 1.87185800e+00
-3.20350123e-03 8.15529078e-02 5.03866434e-01 3.01005006e-01
1.11734319e+00 7.14056075e-01 9.33461078e-03 -9.14452970e-02
9.03097868e-01 -9.80472565e-01 -7.70974338e-01 -1.89371213e-01
6.56120062e-01 -8.27893257e-01 1.37958741e+00 5.46422601e-01
-1.34636438e+00 -4.30592299e-01 -9.68443096e-01 8.35795254e-02
-3.38540018e-01 2.31578931e-01 3.63224566e-01 5.14545143e-01
-1.07362294e+00 4.19017643e-01 -5.81895053e-01 -3.73677880e-01
1.84572577e-01 -3.17370832e-01 -1.23972163e-01 -1.46749094e-01
-1.51979530e+00 9.97832000e-01 6.60954416e-01 -4.62073207e-01
-1.22980499e+00 -9.67653215e-01 -1.21304476e+00 -1.14884719e-01
2.21125826e-01 -9.14810479e-01 1.83673370e+00 -2.66470253e-01
-1.36086559e+00 6.87027752e-01 -2.78655767e-01 -6.60195708e-01
8.27885628e-01 -1.22585095e-01 -5.18500030e-01 -1.63132980e-01
5.20436108e-01 7.81113863e-01 4.01150495e-01 -1.64035785e+00
-2.46891618e-01 4.23218310e-01 -1.08172767e-01 4.16634791e-02
-7.66041651e-02 -2.20725581e-01 -1.27659500e-01 -9.10037994e-01
-6.35374248e-01 -7.46163130e-01 -3.21001858e-02 -6.99620724e-01
-8.88423443e-01 -4.21314865e-01 4.92675841e-01 -5.46985090e-01
1.41979790e+00 -1.48178911e+00 -3.03340137e-01 -1.63857639e-01
-1.30575374e-01 3.89894098e-01 -2.80114084e-01 8.03080201e-01
2.92521358e-01 4.40371037e-01 -4.63095605e-01 -8.18406343e-01
3.77123058e-01 2.85621077e-01 -8.71894836e-01 -1.61996081e-01
2.37357393e-01 9.86231804e-01 -1.27268994e+00 -8.07717085e-01
-9.30321291e-02 1.36035040e-01 -6.55015707e-01 3.55357111e-01
-7.04943895e-01 3.21146965e-01 -9.66405869e-02 3.71705741e-01
3.87665838e-01 -3.79464060e-01 1.29797384e-01 1.54221430e-01
9.48573053e-02 8.43040943e-01 -8.88641834e-01 1.83776879e+00
-6.32920682e-01 7.20235407e-01 -5.61789334e-01 -7.13541269e-01
9.30687368e-01 6.07474208e-01 -2.11033642e-01 -4.53069866e-01
-3.70058835e-01 4.68629330e-01 -3.07562023e-01 -3.84275317e-01
9.90411937e-01 4.61929962e-02 -4.42838877e-01 8.42089653e-01
1.27385303e-01 -8.76187205e-01 8.14595282e-01 7.85840631e-01
9.57466483e-01 3.79791468e-01 2.64725626e-01 3.10345925e-02
1.58046499e-01 5.37237227e-01 4.30429041e-01 1.21452677e+00
1.93814933e-01 8.26929450e-01 4.30865556e-01 1.63369149e-01
-1.22852278e+00 -1.16704929e+00 -2.82906592e-01 4.90444571e-01
-2.74397582e-01 -7.01992810e-01 -8.49858403e-01 -7.45238006e-01
-1.00267888e-03 1.66802013e+00 -4.76628006e-01 -1.12173133e-01
-4.16329801e-01 -6.74307227e-01 1.10678172e+00 5.11308968e-01
4.39207584e-01 -1.18288100e+00 -3.43573183e-01 4.48534817e-01
-6.25030339e-01 -1.23616815e+00 -5.10536671e-01 -3.92236292e-01
-6.16575599e-01 -8.61703813e-01 -4.88956928e-01 -3.87842089e-01
5.03668785e-01 -2.15575233e-01 2.00944209e+00 -3.50500911e-01
8.47164765e-02 3.50095451e-01 -2.43822247e-01 -6.86160803e-01
-1.38614857e+00 2.53799826e-01 2.93877348e-02 -7.16319621e-01
1.35336757e-01 -4.34176773e-01 -1.54042408e-01 2.42259186e-02
-1.04430592e+00 3.48639786e-01 5.22024691e-01 1.11840820e+00
4.47142988e-01 -2.34557077e-01 1.21265423e+00 -1.27916920e+00
1.27541113e+00 -5.93972683e-01 -2.59834707e-01 5.24986446e-01
-9.22658920e-01 2.37571031e-01 9.90386844e-01 -4.11387742e-01
-1.42351019e+00 -7.82160103e-01 8.28788131e-02 -2.21066758e-01
-2.15788409e-01 6.55270278e-01 -1.74520835e-01 7.78555512e-01
9.89987075e-01 4.40959662e-01 -2.60849565e-01 -2.10502923e-01
6.82590306e-01 7.76555717e-01 7.83168435e-01 -1.06470704e+00
6.93389356e-01 7.39867464e-02 -4.76406932e-01 -3.06465775e-01
-1.16543651e+00 1.58558697e-01 -1.27779797e-01 2.11835001e-02
3.36559266e-01 -1.16189742e+00 -1.51643902e-01 3.62243682e-01
-1.42117906e+00 -6.97769582e-01 -4.72911149e-01 3.90991390e-01
-7.14345515e-01 2.33683020e-01 -5.37276208e-01 -6.90257668e-01
-5.42685986e-01 -9.70200896e-01 1.21169102e+00 -2.74014100e-02
-8.19966972e-01 -1.34520066e+00 4.91975605e-01 3.65613073e-01
4.84959424e-01 5.77207029e-01 9.26174521e-01 -8.34810436e-01
-3.98912251e-01 -1.44344896e-01 1.40460089e-01 3.85770172e-01
1.61014609e-02 3.74022096e-01 -8.49053979e-01 -1.21377423e-01
-3.88038099e-01 -9.38438237e-01 5.69012284e-01 1.47738189e-01
1.05486870e+00 -7.07504213e-01 -2.13489607e-01 1.61054283e-01
1.03405106e+00 -2.16166764e-01 4.92738396e-01 1.76982626e-01
4.71723825e-01 4.55934793e-01 8.60703588e-01 4.43341345e-01
9.71278131e-01 4.97265965e-01 5.62216379e-02 1.39800936e-01
-3.88872325e-01 -9.67704773e-01 5.56200862e-01 1.05161452e+00
3.08171391e-01 -6.97141111e-01 -8.89615715e-01 9.43167567e-01
-1.87515867e+00 -1.17333567e+00 -1.70145735e-01 2.19232655e+00
1.54038465e+00 2.93001205e-01 -1.12335108e-01 -1.04683883e-01
3.68243396e-01 2.45357901e-01 -4.35995638e-01 -4.50940102e-01
-3.94547045e-01 1.16310701e-01 7.66866840e-03 4.23507571e-01
-7.46810675e-01 8.10148299e-01 6.85533476e+00 9.84397292e-01
-5.64729452e-01 9.58359465e-02 5.31015277e-01 -1.99627116e-01
-9.77302670e-01 6.21589199e-02 -8.99596810e-01 6.51959896e-01
1.36298347e+00 -6.77012503e-01 1.48739770e-01 8.11650395e-01
2.36789286e-01 -1.06980257e-01 -1.50819027e+00 5.60953140e-01
3.58625263e-01 -1.45085156e+00 4.15161699e-01 -1.74003229e-01
1.31174517e+00 -1.39469326e-01 2.05542631e-02 7.71370173e-01
1.11788905e+00 -1.19084632e+00 1.04318118e+00 5.65210998e-01
9.93586183e-01 -6.68443501e-01 6.38711512e-01 6.92192197e-01
-6.17565632e-01 4.42915231e-01 -2.47929364e-01 1.24091074e-01
3.37136805e-01 8.45760465e-01 -1.25324929e+00 6.32598281e-01
3.59934062e-01 7.37247705e-01 -6.24104321e-01 6.78045809e-01
-6.44259512e-01 1.03310513e+00 -7.92889521e-02 -1.12426527e-01
1.13602191e-01 5.66862337e-02 5.73643327e-01 1.45003939e+00
5.12495577e-01 -3.76804352e-01 1.79326087e-01 1.33497250e+00
-4.07462686e-01 -2.21054450e-01 -9.65097547e-01 -2.20200464e-01
9.60915387e-01 9.47211921e-01 8.82934332e-02 -9.38246608e-01
-2.04289556e-01 5.93645215e-01 4.40864623e-01 4.59255666e-01
-8.80938411e-01 -3.76159638e-01 3.02257359e-01 3.55526321e-02
-1.51774585e-01 1.23669235e-02 -3.52557153e-01 -1.34663391e+00
7.84377456e-02 -1.37975538e+00 2.13601604e-01 -1.05965650e+00
-1.41883087e+00 6.21283293e-01 4.32149976e-01 -1.34386373e+00
-1.12044740e+00 2.18094718e-02 -7.62663603e-01 9.01162088e-01
-1.37168324e+00 -1.20251822e+00 -2.36675337e-01 1.41466320e-01
6.93464220e-01 -2.11266786e-01 1.08660352e+00 3.54177617e-02
-4.67382848e-01 8.84769082e-01 2.72940814e-01 -8.91728401e-02
1.23325944e+00 -1.54448164e+00 8.81516039e-01 9.47600901e-01
9.04327109e-02 8.47858131e-01 1.02424967e+00 -9.08068657e-01
-7.72126496e-01 -1.43837214e+00 1.28236997e+00 -9.69597757e-01
5.69888413e-01 -4.94736820e-01 -9.13979053e-01 8.60823750e-01
6.61742747e-01 -4.11339641e-01 8.17407846e-01 -1.23896725e-01
-3.50796074e-01 3.01292211e-01 -1.13013232e+00 7.99578667e-01
1.01692629e+00 -3.37565452e-01 -8.33292246e-01 5.96068263e-01
9.06349242e-01 -6.26886725e-01 -9.53581929e-01 3.46829265e-01
3.31965268e-01 -7.88357854e-01 7.09455967e-01 -7.81220436e-01
1.08388889e+00 -2.85267472e-01 -3.71384621e-02 -1.95880473e+00
1.91771835e-01 -6.79104984e-01 -4.63441461e-01 1.74647641e+00
7.32755065e-01 -4.91796911e-01 3.68145734e-01 7.11823285e-01
-4.27268088e-01 -5.87827027e-01 -5.54958642e-01 -1.02791488e+00
4.29262102e-01 -3.91887873e-01 8.87209773e-01 9.35547531e-01
-7.78345466e-02 3.85243624e-01 -4.51422155e-01 -1.93510324e-01
7.29022145e-01 2.43898377e-01 1.27897274e+00 -1.00445068e+00
-4.41105306e-01 -2.68174082e-01 3.16407770e-01 -9.29113448e-01
4.98841554e-01 -1.01400769e+00 2.35634744e-01 -1.84533572e+00
3.06482881e-01 -4.76543605e-01 4.02844161e-01 3.63423973e-01
-5.31463146e-01 9.24808383e-02 -7.63945431e-02 1.98017374e-01
-5.55181026e-01 8.17007780e-01 1.17775249e+00 -2.06441984e-01
3.21486257e-02 -2.21917972e-01 -8.34788501e-01 3.81611943e-01
7.15944469e-01 -5.91919303e-01 -7.39639461e-01 -4.57499951e-01
1.56623021e-01 5.35247400e-02 3.54719549e-01 -8.43452573e-01
1.16280138e-01 -1.00487754e-01 1.15387015e-01 -7.11449683e-01
6.84718713e-02 -4.47465815e-02 3.32208574e-01 2.15202883e-01
-7.60141551e-01 1.66002542e-01 1.75849661e-01 4.31249678e-01
-3.74843746e-01 -2.89972335e-01 3.44480038e-01 -2.46411175e-01
-1.41821250e-01 -1.10292919e-01 -3.27932626e-01 7.30319858e-01
6.94109738e-01 1.38028681e-01 -9.62193191e-01 -7.30369270e-01
-1.91426739e-01 3.17504644e-01 5.73042154e-01 2.89572865e-01
4.57408220e-01 -1.49719417e+00 -1.21562481e+00 -1.25755832e-01
3.75634044e-01 3.15050006e-01 -1.27796801e-02 5.67802310e-01
-2.51899838e-01 5.82458496e-01 3.58041435e-01 -5.40999651e-01
-7.71515131e-01 3.21080178e-01 4.76840921e-02 -7.93344736e-01
-2.97166884e-01 4.95704561e-01 -1.44515619e-01 -9.03163075e-01
7.83545300e-02 -3.68443757e-01 1.62710056e-01 -6.40812442e-02
4.02568758e-01 4.03415471e-01 1.53371915e-01 -2.61072308e-01
-4.45795208e-02 -6.46337941e-02 -5.79691119e-02 -5.80774426e-01
9.48264778e-01 1.70874447e-01 1.35311291e-01 6.35060012e-01
8.00304353e-01 2.06082717e-01 -1.09171236e+00 -3.38825405e-01
-5.32854274e-02 -3.02857012e-01 -5.72677433e-01 -1.04990995e+00
-3.99848789e-01 7.62941897e-01 -2.64686376e-01 1.44768953e-01
6.95310831e-01 -1.36259243e-01 7.96813011e-01 5.00096321e-01
4.57211167e-01 -1.02251434e+00 2.84291416e-01 4.92436349e-01
1.15046477e+00 -1.49026752e+00 2.68600434e-02 -9.55438688e-02
-9.37444627e-01 6.58985972e-01 7.42268026e-01 2.01678634e-01
1.51128443e-02 3.57223749e-01 1.25526667e-01 1.17579423e-01
-1.32536983e+00 2.96079516e-01 2.59881884e-01 6.89799547e-01
6.11661077e-01 1.74784899e-01 -2.96061009e-01 6.06139481e-01
-8.68998528e-01 2.12363265e-02 9.84526992e-01 6.71396255e-01
-3.18459161e-02 -1.26037431e+00 -3.13020945e-01 5.10626316e-01
-2.91425526e-01 -3.45646799e-01 -4.03725207e-01 9.87287641e-01
-2.46223733e-01 1.18445289e+00 -2.72356831e-02 1.14220101e-02
3.44765246e-01 2.55558848e-01 3.51323873e-01 -8.17112803e-01
-6.24428034e-01 -4.80036795e-01 5.38242519e-01 -7.36124143e-02
-3.37079644e-01 -7.40637422e-01 -9.27458227e-01 -4.01294947e-01
3.14397924e-02 5.20443380e-01 4.95838135e-01 7.03646064e-01
4.88793850e-01 5.36783636e-01 3.20211649e-01 -4.64085847e-01
-8.88511181e-01 -1.41965616e+00 -1.32861242e-01 6.52816176e-01
5.82764037e-02 -3.50864649e-01 -3.10806036e-01 3.42882037e-01] | [11.567138671875, 8.883318901062012] |
f06f6915-377d-48f9-919f-3e6dfa3e72c4 | pixel-objectness | 1701.05349 | null | http://arxiv.org/abs/1701.05349v2 | http://arxiv.org/pdf/1701.05349v2.pdf | Pixel Objectness | We propose an end-to-end learning framework for generating foreground object
segmentations. Given a single novel image, our approach produces pixel-level
masks for all "object-like" regions---even for object categories never seen
during training. We formulate the task as a structured prediction problem of
assigning foreground/background labels to all pixels, implemented using a deep
fully convolutional network. Key to our idea is training with a mix of
image-level object category examples together with relatively few images with
boundary-level annotations. Our method substantially improves the
state-of-the-art on foreground segmentation for ImageNet and MIT Object
Discovery datasets. Furthermore, on over 1 million images, we show that it
generalizes well to segment object categories unseen in the foreground maps
used for training. Finally, we demonstrate how our approach benefits image
retrieval and image retargeting, both of which flourish when given our
high-quality foreground maps. | ['Kristen Grauman', 'Suyog Dutt Jain', 'Bo Xiong'] | 2017-01-19 | null | null | null | null | ['image-retargeting', 'foreground-segmentation'] | ['computer-vision', 'computer-vision'] | [ 1.01047218e+00 5.45414865e-01 -1.57926697e-02 -4.88983065e-01
-1.13665593e+00 -8.10633659e-01 5.79750240e-01 -1.53553262e-01
-4.31391269e-01 5.67730784e-01 -2.55981803e-01 -3.16234112e-01
4.45679486e-01 -6.89079821e-01 -1.38473499e+00 -6.01362348e-01
6.53611273e-02 6.38009429e-01 8.44171107e-01 2.71132022e-01
1.86749741e-01 3.97538513e-01 -1.49452412e+00 6.23529553e-01
6.77283406e-01 7.88879395e-01 4.66732442e-01 1.01633835e+00
-2.91415751e-01 7.51300931e-01 -5.96619546e-01 -5.07854819e-01
5.03671527e-01 -3.22244912e-01 -1.29477775e+00 7.24189579e-01
1.01122510e+00 -5.60429811e-01 -2.27016047e-01 9.47308481e-01
-7.88489208e-02 2.50692666e-02 6.43067598e-01 -1.08176899e+00
-7.40494013e-01 5.95651865e-01 -7.67088473e-01 3.37988406e-01
-1.99593976e-01 5.97836435e-01 8.92713428e-01 -9.35967207e-01
9.66938972e-01 1.06296563e+00 5.27903497e-01 7.46810198e-01
-1.48233628e+00 -3.54152769e-01 5.52233458e-01 -1.65136352e-01
-1.10021937e+00 -3.24619830e-01 4.34628159e-01 -6.67224050e-01
5.70003986e-01 8.85749310e-02 5.98857164e-01 7.22419679e-01
-1.61203280e-01 1.34637308e+00 8.18677127e-01 -1.60640225e-01
1.68280542e-01 -1.84117518e-02 1.53730571e-01 6.37569189e-01
3.09577852e-01 -2.29315072e-01 -2.23765180e-01 5.54911308e-02
9.49128151e-01 -4.99569811e-02 -1.29600286e-01 -5.21111310e-01
-1.54081810e+00 4.93797749e-01 4.61865425e-01 1.38883054e-01
-2.09118530e-01 5.63994527e-01 7.26537080e-04 -1.82587817e-01
6.39434755e-01 3.09309006e-01 -6.30167723e-01 1.98237970e-01
-1.19483435e+00 3.77337098e-01 5.02103388e-01 1.14787805e+00
1.13786554e+00 -3.26622613e-02 -2.46665269e-01 5.84956169e-01
3.68547253e-02 5.74361384e-01 2.27014758e-02 -1.25248480e+00
1.72220334e-01 4.86562669e-01 2.52746165e-01 -4.42605883e-01
-1.24392487e-01 -5.41299582e-01 -4.44825113e-01 3.05028856e-01
6.48942351e-01 -1.26655281e-01 -1.62838137e+00 1.71338952e+00
4.65449512e-01 3.78111184e-01 4.58972007e-02 8.22558701e-01
5.59579015e-01 7.18563437e-01 2.38583073e-01 3.34535986e-01
1.22057045e+00 -1.22254205e+00 -3.33852857e-01 -6.79913402e-01
3.71452868e-01 -6.45898104e-01 1.04275608e+00 2.55795568e-01
-1.26796067e+00 -6.97711527e-01 -6.53105378e-01 -2.08409056e-01
-4.18402702e-01 6.94269016e-02 7.43593812e-01 5.85433304e-01
-1.37279689e+00 6.39805555e-01 -7.95230865e-01 -3.65429431e-01
1.06447899e+00 3.32475662e-01 -7.03361854e-02 -6.91153556e-02
-5.67012370e-01 4.63064045e-01 6.70791328e-01 -8.80230963e-02
-1.53681552e+00 -9.41825747e-01 -5.61201334e-01 -2.28004619e-01
3.79687518e-01 -7.83253014e-01 1.49374831e+00 -1.58914280e+00
-1.02731276e+00 1.34322214e+00 -2.51735568e-01 -5.47769725e-01
6.59394026e-01 -2.91054487e-01 -9.73581597e-02 4.23247725e-01
5.52391350e-01 1.56815851e+00 9.22145486e-01 -1.80966151e+00
-1.19437933e+00 5.26995063e-02 3.57746296e-02 2.80889636e-03
2.75875747e-01 -1.33027226e-01 -6.97681785e-01 -6.53113604e-01
1.24308094e-01 -6.98539317e-01 -5.03973603e-01 3.26747239e-01
-6.30792201e-01 -2.81031094e-02 1.02769756e+00 -6.78610086e-01
4.98151749e-01 -2.08454537e+00 -3.77528137e-03 1.48492455e-01
2.57384717e-01 1.47620201e-01 -3.69443446e-01 -2.39915803e-01
1.26946690e-02 4.02188331e-01 -6.29125059e-01 -2.96940327e-01
-1.46520630e-01 2.61179417e-01 -5.86995900e-01 3.60236138e-01
7.99171150e-01 1.28479719e+00 -1.11032403e+00 -4.79577273e-01
2.50347525e-01 1.80246457e-01 -5.85718393e-01 2.59646684e-01
-8.50511849e-01 5.19162595e-01 -2.10329920e-01 8.48912477e-01
7.28622377e-01 -6.87828004e-01 -9.42830071e-02 -1.62115798e-03
5.15106209e-02 -6.48071691e-02 -9.32397902e-01 1.72599864e+00
7.02318829e-03 7.78681755e-01 1.25569627e-01 -7.00825930e-01
4.07073855e-01 -9.21981484e-02 2.86378801e-01 -2.65623212e-01
-4.87643145e-02 1.85268849e-01 -1.00560889e-01 -3.78302187e-01
5.76157928e-01 1.75546199e-01 2.39344593e-02 5.23108602e-01
2.63214558e-01 -2.53551960e-01 3.47148418e-01 4.81533796e-01
1.00878692e+00 4.78442967e-01 -2.32363418e-01 -4.68785286e-01
1.96876749e-02 3.57472241e-01 4.26752567e-01 1.14942181e+00
-9.38168764e-02 9.84522581e-01 4.38193470e-01 -5.29145360e-01
-1.31488180e+00 -1.31773269e+00 -1.43620998e-01 1.28837860e+00
4.00512606e-01 1.72005057e-01 -1.19923842e+00 -8.49578381e-01
-5.18551469e-02 5.39317608e-01 -7.52346277e-01 2.46342123e-01
-6.88945293e-01 -6.55074060e-01 5.09488046e-01 5.20327151e-01
5.83942533e-01 -1.30304193e+00 -6.29587710e-01 1.15519024e-01
-1.13224998e-01 -1.23987210e+00 -4.92347747e-01 3.10181201e-01
-8.64439130e-01 -1.10145223e+00 -1.00940979e+00 -9.97636557e-01
9.48613107e-01 4.41080570e-01 1.46835661e+00 3.49606991e-01
-5.76347530e-01 4.29752320e-01 -1.45941556e-01 -2.48163179e-01
-5.46035051e-01 2.30024010e-01 -4.51295644e-01 2.38061607e-01
-9.77726746e-03 -3.41763377e-01 -6.64023042e-01 1.73753157e-01
-1.19391155e+00 4.11153197e-01 9.38209713e-01 5.08727789e-01
8.39721978e-01 -2.20555782e-01 3.76813531e-01 -1.37124133e+00
-1.82356820e-01 -4.38782871e-01 -6.20656967e-01 2.94285029e-01
1.67806130e-02 -1.27525628e-01 8.32849741e-02 -5.01190603e-01
-9.71062541e-01 3.76362115e-01 4.56361882e-02 -3.14264774e-01
-5.30418158e-01 -2.35208541e-01 -3.66268694e-01 -1.90606341e-01
6.55540109e-01 2.58001029e-01 -3.14625055e-01 -4.53400344e-01
9.16837156e-01 3.32394063e-01 1.09215021e+00 -7.64831722e-01
9.50361133e-01 7.77580202e-01 -4.27930236e-01 -6.38137996e-01
-1.08040047e+00 -4.91226912e-01 -9.83665287e-01 -1.59365967e-01
1.28548265e+00 -1.06882513e+00 -2.17457458e-01 5.95779836e-01
-1.25041831e+00 -1.06322086e+00 -5.76638460e-01 -2.65740812e-01
-5.96378446e-01 2.53035307e-01 -6.41547322e-01 -6.35852575e-01
-6.17931113e-02 -1.15655291e+00 1.50506997e+00 4.22855645e-01
8.22836980e-02 -6.39196634e-01 -4.29394752e-01 4.36843097e-01
1.81698516e-01 3.43867064e-01 8.49604666e-01 -5.52797616e-01
-1.50501025e+00 2.60963172e-01 -6.44421935e-01 2.98646063e-01
1.18720168e-02 1.17402293e-01 -1.28553939e+00 -1.08967036e-01
-5.72226226e-01 -2.99584419e-01 1.31470764e+00 3.52323413e-01
1.61285949e+00 -3.79617363e-01 -7.47658193e-01 6.49062276e-01
1.25534809e+00 9.81789678e-02 5.54911613e-01 1.67635277e-01
9.84941423e-01 5.39506674e-01 5.39657295e-01 -9.30730253e-02
8.29085931e-02 2.17055991e-01 4.97031003e-01 -7.48716712e-01
-4.41500306e-01 -7.58416429e-02 -6.06104452e-03 -3.00603211e-02
3.65638584e-01 -3.80925059e-01 -1.02458680e+00 1.17086065e+00
-1.89968336e+00 -7.63068438e-01 -2.57378072e-01 1.89159107e+00
1.08901656e+00 4.27739948e-01 2.05314189e-01 -4.88964707e-01
1.00371814e+00 3.42135355e-02 -8.87832344e-01 1.75184548e-01
-3.18651468e-01 3.86372805e-01 7.49442756e-01 6.64339960e-01
-1.56802452e+00 1.46303320e+00 6.98850918e+00 7.64973462e-01
-7.57909477e-01 1.88532472e-01 1.26139641e+00 3.77343968e-02
-4.50934917e-01 1.05664670e-01 -7.57586122e-01 4.02179331e-01
5.82206905e-01 2.76778609e-01 2.18759552e-01 9.97380972e-01
-1.18484877e-01 -3.20800245e-01 -1.24215889e+00 5.90537548e-01
-8.76215249e-02 -1.67776275e+00 2.80104548e-01 4.02723588e-02
1.31916249e+00 2.64148384e-01 1.87554419e-01 -3.16094724e-03
6.61392272e-01 -9.43107069e-01 1.07951641e+00 2.26353839e-01
7.33537316e-01 -4.37767237e-01 4.56189990e-01 8.70340019e-02
-9.09667790e-01 1.78041890e-01 -3.90817314e-01 2.51912683e-01
2.06965938e-01 6.84541345e-01 -9.60253537e-01 1.14633128e-01
6.70990825e-01 4.98828858e-01 -1.03216231e+00 1.10021603e+00
-2.69740492e-01 7.10455537e-01 -3.59173924e-01 3.02720457e-01
4.26884174e-01 6.74385875e-02 2.17085093e-01 1.45153534e+00
-1.46236658e-01 1.24314062e-01 3.64100307e-01 1.37304819e+00
-3.06228131e-01 -3.77056867e-01 -5.03406167e-01 -9.98188090e-03
2.51482368e-01 1.41956890e+00 -1.39590895e+00 -7.51352370e-01
-1.51033819e-01 1.25833964e+00 2.54417449e-01 6.44900680e-01
-8.14195752e-01 -3.05879742e-01 6.84741318e-01 2.25531489e-01
6.01763904e-01 4.43183444e-03 -4.10572529e-01 -9.36722934e-01
-1.61525905e-01 -7.08318293e-01 6.88490942e-02 -1.08382440e+00
-1.38480377e+00 3.52705449e-01 3.94727215e-02 -6.45168781e-01
-2.01841258e-03 -6.71877146e-01 -5.27710021e-01 6.19489014e-01
-1.42397010e+00 -1.37065911e+00 -3.16886932e-01 2.26300597e-01
7.17279077e-01 2.29699075e-01 2.70058632e-01 3.09162647e-01
-3.28898609e-01 1.82884634e-01 -8.57670903e-02 4.80019778e-01
4.61207062e-01 -1.51940751e+00 1.16938722e+00 1.30487788e+00
3.61824304e-01 3.76323164e-01 5.19156337e-01 -7.98446894e-01
-7.85165668e-01 -1.66997707e+00 6.20924592e-01 -8.37130427e-01
4.45774049e-01 -6.14564180e-01 -9.77479637e-01 9.85321701e-01
3.05182010e-01 1.83048457e-01 2.91892469e-01 -1.68440372e-01
-1.50191978e-01 2.73620129e-01 -1.02205729e+00 7.53501058e-01
1.30221891e+00 -3.30875874e-01 -5.10355055e-01 7.32827187e-01
1.06543386e+00 -5.99476874e-01 -3.12108815e-01 3.22899967e-01
1.45980954e-01 -8.06341588e-01 1.08485079e+00 -7.00044930e-01
4.91912425e-01 -5.98994195e-01 -6.98486790e-02 -7.91750133e-01
-2.87172437e-01 -6.64622128e-01 1.69469446e-01 1.27875543e+00
5.65361679e-01 -2.05095470e-01 1.13697672e+00 6.04939640e-01
-2.88385421e-01 -5.64314365e-01 -5.52627027e-01 -5.25538087e-01
1.24470212e-01 -3.70128483e-01 3.99717033e-01 5.98682165e-01
-9.76769745e-01 1.22979861e-02 -6.22562580e-02 2.94503778e-01
8.25689912e-01 2.83232749e-01 1.01280499e+00 -9.83458400e-01
-3.64008665e-01 -5.03598630e-01 -2.29535341e-01 -1.33942020e+00
1.24592677e-01 -8.44409943e-01 5.31024635e-01 -1.57690299e+00
5.14513969e-01 -5.58889747e-01 -7.79303983e-02 7.38987207e-01
-5.72527051e-01 8.83532584e-01 2.72288054e-01 1.87943831e-01
-1.06686521e+00 7.13664060e-03 1.29205406e+00 -3.48295748e-01
-5.92991523e-03 -5.79643846e-02 -7.05344856e-01 8.01384032e-01
6.17728472e-01 -5.53347170e-01 -2.79712200e-01 -6.84246480e-01
-2.85139591e-01 -4.30357099e-01 8.97835791e-01 -1.01880634e+00
-6.81640059e-02 -2.13438973e-01 7.83892095e-01 -8.58812094e-01
1.60182014e-01 -5.71089983e-01 1.69582099e-01 2.47631207e-01
-4.38169688e-01 -3.57786059e-01 3.76195759e-01 7.38174975e-01
2.92254329e-01 -1.96889728e-01 9.22607422e-01 -4.25283849e-01
-1.19259429e+00 3.81450921e-01 -4.74556446e-01 3.71767759e-01
1.00834572e+00 -3.48495811e-01 -4.72757101e-01 -4.06495705e-02
-7.69296944e-01 2.23787040e-01 8.10249746e-01 3.19121391e-01
3.24787855e-01 -9.66310143e-01 -5.78786433e-01 2.07190320e-01
1.25280112e-01 6.00979447e-01 1.43675968e-01 3.69947314e-01
-7.62055099e-01 4.00098860e-02 -9.44876522e-02 -1.11733675e+00
-8.89657021e-01 5.39518774e-01 4.23102409e-01 1.10493019e-01
-7.01376140e-01 1.34442878e+00 9.38535273e-01 -2.97661722e-01
2.61006236e-01 -6.42054498e-01 6.30335391e-01 -3.55501443e-01
4.39808935e-01 -1.61487237e-01 -1.83415696e-01 -4.31074470e-01
-2.47543946e-01 4.77320760e-01 -2.37620175e-01 -3.03717554e-01
1.07697463e+00 -1.78303629e-01 -6.51086271e-02 4.63873804e-01
1.01152718e+00 -2.71409452e-01 -1.91075969e+00 -2.07404479e-01
2.23335207e-01 -5.28464437e-01 -2.26183310e-01 -1.05269325e+00
-1.10105467e+00 9.26536381e-01 5.95741093e-01 -8.73829201e-02
9.02768672e-01 4.03386205e-01 8.53475273e-01 4.86532778e-01
2.98282206e-01 -8.52839649e-01 3.45241815e-01 2.56781012e-01
3.31449568e-01 -1.34943032e+00 -2.04087436e-01 -5.09296119e-01
-4.58744198e-01 8.41486216e-01 7.00203896e-01 -2.54558504e-01
3.03447306e-01 3.37547779e-01 3.01217526e-01 -4.11462076e-02
-5.23756087e-01 -5.64480245e-01 2.64668167e-01 9.15778756e-01
1.49735928e-01 -9.23630223e-02 4.21538621e-01 2.86218703e-01
1.88876480e-01 -1.81865558e-01 4.86242950e-01 9.63934600e-01
-7.92602777e-01 -1.05285430e+00 -2.17316642e-01 5.47331631e-01
-6.05819821e-01 -2.83404619e-01 -6.02228045e-01 8.92513931e-01
3.76437664e-01 7.66234040e-01 3.82086128e-01 7.22709149e-02
-1.01995759e-01 1.44846395e-01 4.75580692e-01 -9.41067636e-01
-5.35017967e-01 1.62182420e-01 -1.50594577e-01 -6.00267947e-01
-5.18949986e-01 -6.23236060e-01 -1.40760314e+00 -8.50584880e-02
-1.79533333e-01 -1.63864315e-01 4.40614134e-01 9.43150759e-01
3.39051425e-01 7.24522889e-01 1.82760537e-01 -1.25980067e+00
8.58516395e-02 -6.07663095e-01 -4.94610816e-01 5.91284811e-01
5.25544822e-01 -3.12222034e-01 -1.08530790e-01 8.49093020e-01] | [9.643096923828125, 0.4301162362098694] |
20e9f3ac-cf42-4256-be16-143fdc82c3e1 | medical-entity-corpus-with-pico-elements-and | null | null | https://aclanthology.org/L18-1044 | https://aclanthology.org/L18-1044.pdf | Medical Entity Corpus with PICO elements and Sentiment Analysis | null | ['Michael Andersson', 'Markus Zlabinger', 'Linda Andersson', 'Allan Hanbury', 'Vanessa Quasnik', 'Jon Brassey'] | 2018-05-01 | medical-entity-corpus-with-pico-elements-and-1 | https://aclanthology.org/L18-1044 | https://aclanthology.org/L18-1044.pdf | lrec-2018-5 | ['pico'] | ['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] | [-7.318605422973633, 3.6143624782562256] |
0aac044c-d62e-4586-9b32-6cab6e2cdf3f | training-neural-networks-to-have-brain-like | 1905.10679 | null | https://arxiv.org/abs/1905.10679v4 | https://arxiv.org/pdf/1905.10679v4.pdf | Improved object recognition using neural networks trained to mimic the brain's statistical properties | The current state-of-the-art object recognition algorithms, deep convolutional neural networks (DCNNs), are inspired by the architecture of the mammalian visual system, and are capable of human-level performance on many tasks. However, even these algorithms make errors. As they are trained for object recognition tasks, it has been shown that DCNNs develop hidden representations that resemble those observed in the mammalian visual system. Moreover, DCNNs trained on object recognition tasks are currently among the best models we have of the mammalian visual system. This led us to hypothesize that teaching DCNNs to achieve even more brain-like representations could improve their performance. To test this, we trained DCNNs on a composite task, wherein networks were trained to: a) classify images of objects; while b) having intermediate representations that resemble those observed in neural recordings from monkey visual cortex. Compared with DCNNs trained purely for object categorization, DCNNs trained on the composite task had better object recognition performance and are more robust to label corruption. Interestingly, we also found that neural data was not required, but randomized data with the same statistics as neural data also boosted performance. Our results outline a new way to train object recognition networks, using strategies in which the brain - or at least the statistical properties of its activation patterns - serves as a teacher signal for training DCNNs. | ['Haoyan Xu', 'Alona Fyshe', 'Joel Zylberberg', 'Callie Federer'] | 2019-05-25 | null | null | null | null | ['object-categorization'] | ['computer-vision'] | [ 3.53625447e-01 1.10857949e-01 6.11798130e-02 -5.18065393e-01
2.31135860e-01 -4.59815055e-01 7.87202001e-01 -4.19764183e-02
-6.69236958e-01 4.93581206e-01 -2.22460598e-01 -3.51692200e-01
-8.50489810e-02 -8.64472210e-01 -9.05755877e-01 -7.81528056e-01
-1.70420967e-02 2.16289520e-01 4.64228243e-01 3.27954739e-02
3.91187847e-01 9.26291585e-01 -1.99704313e+00 7.35437214e-01
4.61291760e-01 1.14582300e+00 3.93606961e-01 7.79007316e-01
-8.53230134e-02 8.59414279e-01 -7.06075549e-01 6.93686232e-02
2.40995720e-01 -4.28453743e-01 -9.71286356e-01 1.35281570e-02
7.14298844e-01 -5.47394529e-02 -2.69746423e-01 9.86861646e-01
2.03398004e-01 7.35679045e-02 8.74917567e-01 -1.12383807e+00
-1.10605204e+00 2.12373644e-01 3.63019705e-02 4.89510298e-01
2.26654802e-02 4.86630887e-01 6.95325553e-01 -8.36487353e-01
5.53786397e-01 1.24783742e+00 6.03580058e-01 8.98470759e-01
-1.78572536e+00 -5.27781844e-01 5.69930859e-02 2.91957170e-01
-1.00942743e+00 -4.70828593e-01 3.07863086e-01 -5.66397727e-01
1.19896877e+00 1.45076618e-01 8.22636664e-01 1.22393131e+00
7.60932788e-02 7.46798098e-01 1.56185842e+00 -6.22852683e-01
2.85636365e-01 2.66998559e-01 2.61931986e-01 5.87040722e-01
4.00415987e-01 5.39800763e-01 -1.19420007e-01 8.25144127e-02
8.62017572e-01 8.16884786e-02 -2.54006267e-01 -2.89179057e-01
-1.30267227e+00 8.00054371e-01 1.05514932e+00 8.39130759e-01
-4.22439754e-01 2.62385756e-01 2.54031777e-01 4.27772194e-01
-9.04735103e-02 8.81610453e-01 -4.73837435e-01 3.64994735e-01
-7.80446231e-01 -1.26702130e-01 6.66208625e-01 5.01551032e-01
7.81106293e-01 4.02952671e-01 -2.40482599e-01 8.17760050e-01
3.15189004e-01 1.35017782e-01 8.51865828e-01 -9.41907048e-01
-2.16944531e-01 8.21085095e-01 -4.90847260e-01 -8.32325339e-01
-5.77093780e-01 -4.81113940e-01 -8.91578317e-01 8.31232429e-01
7.13890553e-01 3.43156040e-01 -1.41196883e+00 1.72776544e+00
-3.24742854e-01 -1.72492206e-01 1.73294663e-01 9.52556133e-01
7.59591222e-01 6.29169047e-01 2.20100567e-01 4.20504436e-02
1.32277536e+00 -6.59300923e-01 -4.06702638e-01 -2.69098043e-01
6.79283738e-01 -3.75524044e-01 8.64041865e-01 4.66174990e-01
-7.67191112e-01 -1.03239250e+00 -1.12600565e+00 1.63828105e-01
-7.75438249e-01 3.38545963e-02 7.44393885e-01 7.14499772e-01
-1.15869951e+00 7.29232132e-01 -7.25017548e-01 -6.66781068e-01
8.27822566e-01 5.26433051e-01 -5.27722597e-01 1.05084315e-01
-6.82303667e-01 1.39298475e+00 7.86363423e-01 7.68200159e-02
-1.32008541e+00 -3.31656873e-01 -5.94909072e-01 1.88949421e-01
-5.90230562e-02 -3.10557693e-01 1.19990396e+00 -1.71195650e+00
-1.15391052e+00 1.29962647e+00 8.71254727e-02 -5.30998886e-01
-1.87695622e-01 3.31129491e-01 -3.78273547e-01 1.67878389e-01
-2.72128880e-01 1.08974600e+00 7.49386668e-01 -1.61675060e+00
-3.71287435e-01 -2.86623508e-01 -1.29616976e-01 -4.61101621e-01
-2.12211877e-01 -5.34904860e-02 2.71260440e-01 -5.53014874e-01
2.46346176e-01 -8.56041908e-01 -9.49236676e-02 2.75207072e-01
-1.20025508e-01 -4.88243043e-01 7.29015052e-01 -2.68249005e-01
6.49553716e-01 -2.25861335e+00 -5.62204942e-02 3.42144728e-01
3.56123656e-01 8.52243960e-01 -5.45955062e-01 1.20694250e-01
-3.69341969e-01 2.51670897e-01 -2.59667724e-01 1.94299191e-01
-1.22249082e-01 5.57234526e-01 -1.48314521e-01 4.30900067e-01
4.63298172e-01 1.07635009e+00 -5.90629518e-01 -1.24163143e-01
1.08474761e-01 1.70493782e-01 -4.53814954e-01 3.37572396e-01
-2.34068096e-01 4.06655192e-01 9.58943665e-02 5.33995986e-01
4.42000598e-01 -2.90306270e-01 1.56076863e-01 -1.43887237e-01
-1.56080589e-01 1.28740981e-01 -6.26488328e-01 1.16663229e+00
-2.27967009e-01 1.26337421e+00 -1.76680699e-01 -1.65225804e+00
1.04496408e+00 3.87038857e-01 1.46760596e-02 -1.08208323e+00
3.87715638e-01 3.44366103e-01 1.00646377e+00 -5.20598352e-01
-9.55069512e-02 -1.63241461e-01 4.71720099e-01 3.61546844e-01
6.22259378e-01 -9.24935937e-02 1.75910652e-01 -1.95395947e-01
1.02790630e+00 -8.26914385e-02 1.28931701e-01 -3.34997863e-01
4.01235104e-01 -3.06860246e-02 3.35148245e-01 1.08966458e+00
-1.83449462e-01 5.89177430e-01 3.86632025e-01 -8.40077996e-01
-1.15706873e+00 -1.03621864e+00 -5.86030543e-01 1.13859940e+00
-3.54046792e-01 2.08350435e-01 -7.29287803e-01 -4.80425060e-01
-1.01636678e-01 6.83320642e-01 -9.88289833e-01 -4.12419349e-01
-3.08246791e-01 -6.80238366e-01 7.24726617e-01 5.84543765e-01
3.79862517e-01 -1.76874614e+00 -8.87869656e-01 2.26707608e-01
5.23951530e-01 -8.79044771e-01 4.01685238e-01 7.52310097e-01
-1.05502391e+00 -1.20440710e+00 -7.46033728e-01 -1.15268481e+00
9.99992967e-01 3.08840182e-02 9.87035096e-01 5.80358088e-01
-5.09205759e-01 3.58662426e-01 -2.70171404e-01 -6.70951664e-01
-5.50757349e-01 -1.11658849e-01 1.81107536e-01 4.70085219e-02
5.56804955e-01 -6.73463643e-01 -3.16791117e-01 3.45766753e-01
-1.26842630e+00 -1.62763223e-02 9.30322111e-01 1.04481447e+00
1.60991505e-01 -2.01203018e-01 6.78043962e-01 -7.79604018e-01
4.22218531e-01 -3.28360021e-01 -5.72628081e-01 1.91901192e-01
-5.72725832e-01 1.45110294e-01 6.87670708e-01 -8.23766887e-01
-6.50023997e-01 1.20452493e-01 -2.48869315e-01 -1.46575272e-01
-7.80944467e-01 3.38691771e-01 3.15920800e-01 -4.34868455e-01
1.15249646e+00 4.52514440e-01 4.90250848e-02 -3.51673812e-01
1.07714860e-03 5.97394526e-01 4.56642032e-01 -4.04835433e-01
6.26038015e-01 2.18293369e-01 -1.37414977e-01 -9.56381798e-01
-6.40516162e-01 -1.91996232e-01 -7.01633394e-01 -2.13781431e-01
1.11753058e+00 -4.20223087e-01 -9.43007767e-01 5.58336318e-01
-1.38916671e+00 -6.18254840e-01 -4.20980483e-01 6.12689257e-01
-5.88448167e-01 -2.75279403e-01 -4.19584692e-01 -7.21837997e-01
2.81146973e-01 -9.32735980e-01 4.18983072e-01 3.70363533e-01
-1.23271435e-01 -9.06824231e-01 1.17415421e-01 -2.72623245e-02
7.00820804e-01 -1.11095767e-04 1.35132551e+00 -1.05332375e+00
-4.68248099e-01 1.51297089e-03 -5.95832825e-01 7.66052127e-01
1.26337543e-01 3.49126495e-02 -1.37615955e+00 -1.86019838e-01
-9.88411307e-02 -6.98046923e-01 1.18207645e+00 2.18836039e-01
1.47677040e+00 -3.62724096e-01 -2.40203694e-01 5.26601374e-01
1.42910230e+00 5.19341111e-01 9.75558937e-01 4.12760645e-01
2.00105697e-01 7.84993708e-01 -2.90092856e-01 -1.56131223e-01
-1.81665242e-01 5.43533564e-01 6.68477654e-01 -2.72432297e-01
-3.36953312e-01 6.44795820e-02 2.34225541e-01 6.07642412e-01
-3.15342963e-01 -1.54210150e-01 -9.79089737e-01 6.57905221e-01
-1.55984616e+00 -1.17048585e+00 -1.20574981e-01 1.86090875e+00
7.71387994e-01 3.01027268e-01 6.01568297e-02 3.84927601e-01
4.86079156e-01 -1.71989247e-01 -5.12956798e-01 -7.51248121e-01
-2.96084344e-01 5.74314535e-01 1.88209996e-01 -6.67776465e-02
-8.60174716e-01 7.57816553e-01 7.50085020e+00 3.27916473e-01
-1.34185576e+00 -1.21122979e-01 4.49881583e-01 3.05415154e-01
1.20526016e-01 -1.42750204e-01 -5.59557498e-01 3.22643042e-01
1.12477350e+00 3.16656828e-01 4.34776366e-01 8.34160149e-01
-2.11630762e-01 -2.70291150e-01 -1.34228051e+00 7.02919185e-01
1.00676529e-01 -1.48045301e+00 2.35894725e-01 1.01056151e-01
6.98952734e-01 9.66907591e-02 9.02476981e-02 5.50436676e-01
3.39826584e-01 -1.58339512e+00 6.43899858e-01 6.63169801e-01
3.59890133e-01 -9.19554308e-02 7.97872007e-01 4.78264898e-01
-5.95644534e-01 -2.90947407e-01 -7.73997664e-01 -2.25547731e-01
-5.08734941e-01 1.88372791e-01 -7.84473062e-01 -1.77748069e-01
9.13442731e-01 4.80393767e-01 -1.19339490e+00 1.41018236e+00
-3.63647826e-02 7.56596565e-01 -8.79450291e-02 -3.17730755e-01
3.20237070e-01 2.56118089e-01 1.95557967e-01 1.36748075e+00
5.71912527e-02 1.62902683e-01 -9.49559510e-02 1.06467736e+00
-2.65788864e-02 -1.31839395e-01 -9.11156178e-01 -3.16828817e-01
1.40420988e-01 1.20599747e+00 -8.79393101e-01 -4.69343185e-01
-4.04040992e-01 6.13134563e-01 5.59771597e-01 4.80571926e-01
-2.97177911e-01 -3.64153951e-01 5.32147527e-01 2.62291003e-02
4.13455218e-01 -2.18625695e-01 -3.24611336e-01 -9.10214841e-01
-3.66888732e-01 -1.04437745e+00 5.66363968e-02 -1.02992523e+00
-1.38089097e+00 8.76471460e-01 -2.41362542e-01 -1.05056655e+00
1.67028935e-04 -1.40662277e+00 -6.60343349e-01 6.82782650e-01
-1.27565372e+00 -8.36234987e-01 -3.29439670e-01 6.11207068e-01
2.73780316e-01 -4.49126750e-01 1.13687766e+00 -7.21296412e-04
-2.41581485e-01 2.66543388e-01 -3.99038084e-02 5.71773767e-01
2.68291771e-01 -1.03297567e+00 3.37872319e-02 5.46978533e-01
6.28310561e-01 7.88646877e-01 3.24970335e-01 -9.53346640e-02
-9.56886113e-01 -7.74535000e-01 5.18890619e-01 -2.96348691e-01
2.40455240e-01 -3.64663363e-01 -1.33986592e+00 6.72530890e-01
2.58399099e-01 3.60330582e-01 6.95481002e-01 5.48525266e-02
-6.19172871e-01 -4.93142754e-02 -1.06677771e+00 3.20742667e-01
8.69897842e-01 -6.28736377e-01 -1.13824129e+00 2.16935918e-01
2.77507067e-01 1.11222304e-01 -5.49302042e-01 4.45351809e-01
7.35140145e-01 -1.06817365e+00 8.69843543e-01 -1.22935820e+00
3.50874811e-01 -3.82712990e-01 -2.64296293e-01 -1.40573573e+00
-8.20020616e-01 2.50333071e-01 7.87525699e-02 8.01524639e-01
5.74694633e-01 -7.41941452e-01 5.95792115e-01 3.13090861e-01
-1.66156083e-01 -4.55149978e-01 -8.54348481e-01 -1.13338792e+00
1.67111650e-01 -3.52337301e-01 8.63903835e-02 8.72340560e-01
-3.86845231e-01 1.39341742e-01 2.04973236e-01 -5.28416708e-02
4.02737349e-01 -1.07413679e-01 4.47462142e-01 -1.60501277e+00
-2.61197351e-02 -8.32672358e-01 -9.50220585e-01 -6.27532601e-01
3.58633101e-01 -1.21785259e+00 2.30609342e-01 -1.54322672e+00
6.54446557e-02 -5.25422692e-01 -4.98989522e-01 9.38762665e-01
2.34009400e-01 6.48109734e-01 3.70208353e-01 2.84321100e-01
-2.22657427e-01 1.52572796e-01 1.15768063e+00 -1.86726764e-01
4.09774110e-02 -1.70880202e-02 -5.41960657e-01 7.47847736e-01
8.03258896e-01 -7.35709786e-01 -2.47023664e-02 -4.47116673e-01
-5.63129373e-02 -4.81235415e-01 8.28123569e-01 -1.37780905e+00
2.82760680e-01 -1.00158632e-01 9.98778224e-01 -6.66551217e-02
2.54082054e-01 -1.07321179e+00 4.93841581e-02 8.89758229e-01
-6.52405322e-01 -2.65144944e-01 3.89713317e-01 2.45097861e-01
-2.74588913e-01 -7.06145465e-01 1.27507985e+00 -3.23103547e-01
-9.51035619e-01 1.34234382e-02 -1.19729829e+00 -2.32684731e-01
8.81620646e-01 -5.76069236e-01 -4.86690372e-01 -2.24493295e-02
-8.23368967e-01 -2.86716342e-01 1.93974614e-01 4.28291976e-01
6.87540472e-01 -1.32330167e+00 -5.68155944e-01 4.86266226e-01
1.74721360e-01 -4.21028018e-01 -2.81982958e-01 7.07993567e-01
-5.16275227e-01 6.61802471e-01 -9.12721455e-01 -8.13702047e-01
-9.08473372e-01 6.75489247e-01 5.82571805e-01 2.44089335e-01
-2.33140171e-01 7.74734378e-01 3.97076935e-01 -5.88610768e-01
2.82108098e-01 -3.35861266e-01 -4.89107966e-01 8.45498815e-02
5.22143126e-01 -8.49222839e-02 1.06581941e-01 -4.31658596e-01
-3.62926215e-01 4.11147743e-01 -9.02312063e-03 1.32825822e-01
1.52309370e+00 4.52714026e-01 -1.91593111e-01 5.52864313e-01
1.17231977e+00 -5.93621194e-01 -9.89737988e-01 -2.73482263e-01
3.04062754e-01 -2.55927294e-01 -7.01679587e-02 -9.65382218e-01
-1.03275347e+00 1.17304504e+00 9.15811539e-01 6.33235455e-01
1.10599613e+00 -6.78407541e-03 7.24624768e-02 8.02139342e-01
3.31314862e-01 -9.81424212e-01 3.62502366e-01 6.56035721e-01
1.02850211e+00 -1.38041222e+00 -1.22187473e-01 1.63443610e-01
-3.05836797e-01 1.46484721e+00 8.60909939e-01 -4.25447315e-01
6.09639645e-01 -4.97114770e-02 1.03470646e-01 -2.81672955e-01
-9.96963918e-01 -5.11227727e-01 4.65668231e-01 9.55047607e-01
4.32443827e-01 -2.14880079e-01 1.79340333e-01 2.11786523e-01
2.78488081e-02 2.47964244e-02 4.26337004e-01 7.57797122e-01
-7.33786404e-01 -7.60486841e-01 -4.12372172e-01 6.62120223e-01
-2.65931100e-01 -1.13658197e-02 -4.92135197e-01 7.81777263e-01
4.43797708e-01 7.25788593e-01 3.18028361e-01 -3.70606244e-01
3.29728544e-01 3.28115880e-01 6.62922680e-01 -8.78103197e-01
-8.81971717e-01 -6.25035226e-01 -1.66440651e-01 -4.73396033e-01
-6.47581756e-01 -3.11130732e-01 -1.11370313e+00 -1.62280276e-01
-3.05249125e-01 -2.82293861e-03 7.95053661e-01 1.00301003e+00
-7.70203478e-05 5.69272220e-01 4.10827965e-01 -1.12786114e+00
-2.84947485e-01 -1.19328976e+00 -4.49690908e-01 3.81972402e-01
4.21243608e-01 -6.21007562e-01 -4.45092261e-01 3.09941322e-01] | [9.811328887939453, 2.3696019649505615] |
22562bc7-7783-48dc-94e2-da5e2f677757 | inter-patient-ecg-classification-with | 1810.04121 | null | http://arxiv.org/abs/1810.04121v1 | http://arxiv.org/pdf/1810.04121v1.pdf | Inter-Patient ECG Classification with Convolutional and Recurrent Neural Networks | The recent advances in ECG sensor devices provide opportunities for user
self-managed auto-diagnosis and monitoring services over the internet. This
imposes the requirements for generic ECG classification methods that are
inter-patient and device independent. In this paper, we present our work on
using the densely connected convolutional neural network (DenseNet) and gated
recurrent unit network (GRU) for addressing the inter-patient ECG
classification problem. A deep learning model architecture is proposed and is
evaluated using the MIT-BIH Arrhythmia and Supraventricular Databases. The
results obtained show that without applying any complicated data pre-processing
or feature engineering methods, both of our models have considerably
outperformed the state-of-the-art performance for supraventricular (SVEB) and
ventricular (VEB) arrhythmia classifications on the unseen testing dataset
(with the F1 score improved from 51.08 to 61.25 for SVEB detection and from
88.59 to 89.75 for VEB detection respectively). As no patient-specific or
device-specific information is used at the training stage in this work, it can
be considered as a more generic approach for dealing with scenarios in which
varieties of ECG signals are collected from different patients using different
types of sensor devices. | [] | 2018-09-27 | null | null | null | null | ['ecg-classification'] | ['medical'] | [ 2.39954606e-01 -1.21461906e-01 2.66393006e-01 -3.73630881e-01
-6.90069258e-01 -3.91638488e-01 -1.04741991e-01 3.42840016e-01
-3.11270088e-01 8.79772365e-01 -2.57312030e-01 -4.82898295e-01
-3.79321814e-01 -4.89118725e-01 -1.42780051e-01 -6.06519997e-01
-3.33565235e-01 4.49101567e-01 -3.12136531e-01 -8.25878158e-02
-2.56197542e-01 5.99995315e-01 -1.20983517e+00 3.26321423e-01
5.63913286e-01 1.23858619e+00 -2.10132957e-01 1.01900387e+00
4.29553062e-01 4.43872511e-01 -1.02872634e+00 -7.73249418e-02
1.69306919e-01 -8.33395302e-01 -5.61537087e-01 -1.19301721e-01
-1.43697876e-02 -1.49211571e-01 1.01686492e-04 5.28743148e-01
1.43054831e+00 -3.99540365e-01 4.77187663e-01 -6.70312703e-01
-1.34088159e-01 6.18701637e-01 -2.18805775e-01 6.31352484e-01
1.86486706e-01 -5.68192862e-02 4.98413742e-01 -6.35919750e-01
2.87227720e-01 4.02656466e-01 1.20542669e+00 4.33424234e-01
-1.08791697e+00 -4.67959166e-01 -5.88649631e-01 -1.10596074e-02
-1.59198689e+00 -2.02325448e-01 7.15349495e-01 -4.05999810e-01
1.02105153e+00 4.60896909e-01 8.23440075e-01 1.15220165e+00
3.24686646e-01 2.38371417e-01 8.68825555e-01 -4.51803654e-01
1.80534959e-01 9.30433944e-02 3.54318142e-01 3.99968565e-01
4.58083361e-01 1.21116243e-01 -2.28461415e-01 -4.15870428e-01
1.01633978e+00 1.36776581e-01 -4.89816636e-01 -5.64111322e-02
-1.21915507e+00 5.09644747e-01 1.54474244e-01 8.40393543e-01
-7.38453150e-01 -6.17895685e-02 8.91420782e-01 6.48615718e-01
2.09862590e-01 5.92575133e-01 -8.29936087e-01 -2.95656294e-01
-1.01151955e+00 2.36845110e-02 7.74322093e-01 6.52059317e-01
8.89798030e-02 5.23111284e-01 -3.28191727e-01 7.73508251e-01
7.76120648e-02 1.75050437e-01 8.43064129e-01 -3.82591903e-01
2.57414043e-01 5.03292680e-01 6.79867417e-02 -8.18109453e-01
-9.27292824e-01 -1.19715667e+00 -1.42671072e+00 -2.17290685e-01
2.25425541e-01 -5.12305737e-01 -9.62537527e-01 1.36225617e+00
-8.46242011e-02 2.31513053e-01 3.34254235e-01 7.42364824e-01
1.16287756e+00 2.99758315e-01 -1.48467552e-02 -4.79752392e-01
1.32655466e+00 -2.56156445e-01 -8.57287884e-01 2.08420798e-01
8.22062492e-01 -4.30259109e-01 6.79134011e-01 6.01283014e-01
-1.00915170e+00 -8.03541481e-01 -1.31387699e+00 3.67647737e-01
-1.47181213e-01 3.87370527e-01 3.39015603e-01 9.94798779e-01
-1.10213482e+00 8.90125453e-01 -7.62523949e-01 -4.81932431e-01
4.73882139e-01 6.65226638e-01 -2.35311225e-01 3.21294725e-01
-1.28570414e+00 6.63451672e-01 3.17036122e-01 3.17323744e-01
-7.66558170e-01 -5.32995284e-01 -5.51661491e-01 6.46097288e-02
-1.57484785e-01 -1.00256538e+00 9.09399390e-01 -7.72340298e-01
-1.38609922e+00 9.16143715e-01 2.92080283e-01 -8.00160706e-01
6.16778731e-01 -9.48189348e-02 -7.45624840e-01 -7.84997195e-02
-2.45080516e-01 2.42343172e-03 7.39881158e-01 -6.81386232e-01
-2.64869064e-01 -5.90528190e-01 -3.03155452e-01 3.31435874e-02
-5.93155995e-02 3.47566325e-03 -5.48202656e-02 -8.48915994e-01
2.55930036e-01 -7.44320154e-01 -2.30680808e-01 -5.43067515e-01
-4.07983810e-01 8.49822611e-02 7.17029035e-01 -7.36134887e-01
1.45229542e+00 -2.12526822e+00 8.17961618e-02 3.62504512e-01
4.78845179e-01 6.23530567e-01 2.78780043e-01 3.28305811e-01
-4.08819705e-01 7.20487908e-02 -3.65044594e-01 -1.57385334e-01
-5.76376975e-01 2.12458000e-01 1.31167367e-01 5.02243698e-01
-4.54355106e-02 8.94519806e-01 -4.52491760e-01 -7.19948187e-02
3.92458797e-01 7.32577562e-01 -1.02486476e-01 2.72495478e-01
5.87447345e-01 8.88365567e-01 -3.08929622e-01 6.06333494e-01
2.72122622e-01 -3.50907385e-01 2.89368987e-01 -2.83062577e-01
2.70648152e-01 8.50031301e-02 -1.15871215e+00 1.86767089e+00
-4.16250795e-01 3.02268714e-01 -2.23344773e-01 -1.32646322e+00
1.16759658e+00 1.04378688e+00 6.51837826e-01 -4.25091773e-01
5.96667111e-01 4.10278618e-01 5.41720629e-01 -5.48193753e-01
-2.44754195e-01 -1.85155421e-01 -3.66419437e-03 2.41068423e-01
2.73445517e-01 2.94750333e-01 -2.53646553e-01 -2.75927812e-01
1.11924934e+00 -2.72588193e-01 5.76901138e-01 -5.44814885e-01
6.21893346e-01 -5.70881426e-01 6.82489574e-01 1.03603709e+00
-1.81200564e-01 1.03749168e+00 1.68472379e-01 -9.94288504e-01
-7.69859731e-01 -8.53877902e-01 -5.24396896e-01 3.40318799e-01
-4.32914823e-01 -4.11825091e-01 -6.57724380e-01 -4.90522921e-01
-2.10237339e-01 6.28039613e-02 -5.21575630e-01 -1.55771792e-01
-6.17683768e-01 -1.12827325e+00 9.59110618e-01 7.72054136e-01
4.15412992e-01 -1.25726426e+00 -1.24875855e+00 7.18912244e-01
-6.21620752e-02 -8.41561794e-01 1.98043525e-01 5.54341376e-01
-1.27615833e+00 -9.28137422e-01 -1.05587065e+00 -5.82095325e-01
9.50148106e-02 -6.40938282e-01 1.28675961e+00 1.41426578e-01
-7.44146526e-01 1.64040238e-01 -2.67765135e-01 -6.98703289e-01
-3.19141328e-01 2.51950949e-01 1.64697617e-01 2.19984263e-01
2.02956572e-01 -8.28234434e-01 -6.87633336e-01 4.52248342e-02
-3.91010046e-01 -2.77285188e-01 3.31770092e-01 9.76889491e-01
5.49986303e-01 -2.46412262e-01 1.22038710e+00 -9.72971261e-01
9.38527048e-01 -3.56096357e-01 -2.32122228e-01 5.53768966e-03
-8.61993611e-01 -4.42424446e-01 5.54726124e-01 -1.53696880e-01
-2.23419368e-01 1.05796903e-01 -4.92619365e-01 -4.75869268e-01
-3.79302979e-01 5.04946530e-01 -1.21527418e-01 7.88630918e-02
8.66326094e-01 1.36121586e-01 -2.19016954e-01 -4.32199866e-01
-4.40320671e-01 9.04623568e-01 5.99370360e-01 -4.17535901e-02
1.40100732e-01 1.19940571e-01 -9.47609637e-03 -6.03623033e-01
-4.76639092e-01 -4.49258655e-01 -7.09696054e-01 -3.60127725e-02
1.09706080e+00 -9.01524782e-01 -5.80595195e-01 6.66585505e-01
-1.01756656e+00 8.67048558e-03 -5.05969703e-01 5.72450459e-01
-4.30650353e-01 2.71565259e-01 -6.69034660e-01 -9.22686219e-01
-1.11808527e+00 -1.01031506e+00 7.85506248e-01 -4.20873389e-02
-6.25501752e-01 -8.89060140e-01 -1.34276330e-01 -2.16249246e-02
7.62452900e-01 7.60639250e-01 6.64951861e-01 -9.42464232e-01
4.03585024e-02 -6.50491416e-01 1.76875219e-01 6.20652199e-01
4.08485413e-01 -5.16826808e-01 -1.23453319e+00 -4.98997122e-01
3.49407047e-01 4.82695624e-02 4.71186906e-01 5.65907836e-01
1.27834761e+00 8.62368867e-02 -1.12650990e-01 7.52739847e-01
1.31814706e+00 4.35647309e-01 7.91675627e-01 -3.98161337e-02
6.82906866e-01 -3.50880660e-02 3.04295924e-02 6.67198658e-01
-2.12380327e-02 4.23976153e-01 1.87608823e-01 -5.83258212e-01
1.44236952e-01 3.24086368e-01 -1.30411968e-01 6.94962919e-01
-5.07306695e-01 -2.11475357e-01 -1.11532128e+00 5.32088161e-01
-1.75965595e+00 -6.64636254e-01 -3.18895072e-01 2.35078430e+00
5.43731332e-01 1.16842248e-01 1.64631084e-01 8.78075361e-01
6.86445594e-01 -3.62430096e-01 -6.58845901e-01 -5.92422426e-01
-1.99595943e-01 7.94113815e-01 2.34195665e-01 -2.51271874e-02
-1.12213969e+00 1.68282017e-01 6.21320200e+00 1.71658859e-01
-1.33745468e+00 3.71938825e-01 8.99480760e-01 1.41796589e-01
5.28181553e-01 -5.62722862e-01 -3.97612661e-01 3.84970546e-01
1.40407395e+00 1.58018619e-01 -9.94779286e-04 4.93848681e-01
1.73248380e-01 3.34083647e-01 -1.11606479e+00 1.75554597e+00
-3.09259556e-02 -1.31770957e+00 -3.07450861e-01 -1.21340662e-01
3.86822611e-01 2.12393016e-01 -2.37973392e-01 2.57190704e-01
-6.45521045e-01 -1.24423110e+00 1.59246564e-01 5.79452455e-01
1.18968296e+00 -7.66138613e-01 1.36721039e+00 3.52958053e-01
-9.62510943e-01 -1.56334087e-01 -8.00709724e-02 -2.63565689e-01
7.09970295e-02 7.77743220e-01 -7.67207623e-01 7.85929799e-01
9.61061597e-01 8.42734635e-01 -4.22352463e-01 8.65464032e-01
2.04085827e-01 8.00017536e-01 -3.32216024e-01 3.76375467e-01
-2.29481131e-01 2.02947423e-01 4.73039389e-01 1.19228101e+00
4.25937027e-01 1.75081808e-02 8.32711458e-02 6.48104191e-01
6.78630592e-03 1.97236955e-01 -7.15230584e-01 2.84979075e-01
7.27233440e-02 1.27003789e+00 -6.37487888e-01 -5.61796010e-01
-7.13646561e-02 9.01983380e-01 -1.46401927e-01 2.19455838e-01
-7.63710320e-01 -6.11524701e-01 2.11612597e-01 2.40575776e-01
6.12657256e-02 2.86976010e-01 -6.96850240e-01 -1.12147033e+00
1.13572562e-02 -9.41234529e-01 6.42606258e-01 -4.52972680e-01
-1.19491100e+00 1.14317083e+00 -4.10466254e-01 -1.33045268e+00
-5.57437897e-01 -3.96883875e-01 -6.01301968e-01 1.17688751e+00
-1.09230292e+00 -7.27216959e-01 -2.81471431e-01 6.68259084e-01
2.13752866e-01 -4.47734326e-01 1.53004539e+00 7.03163326e-01
-4.28667605e-01 7.88698018e-01 -2.61472106e-01 3.32656562e-01
4.05308753e-01 -1.09484041e+00 2.30995953e-01 6.02895439e-01
1.60465971e-01 5.85855365e-01 4.29524094e-01 -3.53990465e-01
-9.58922625e-01 -1.12095094e+00 9.09649432e-01 -4.89744127e-01
-5.25075980e-02 -2.26414919e-01 -8.97440016e-01 4.09634084e-01
2.82859236e-01 4.24633831e-01 9.78793383e-01 1.23812087e-01
1.75365925e-01 -2.48105422e-01 -1.32489884e+00 -3.71603779e-02
8.24523926e-01 -4.20796931e-01 -4.14872319e-01 1.71280012e-01
-1.44189838e-02 -5.40958166e-01 -1.38230884e+00 1.00471842e+00
5.52216470e-01 -1.00310230e+00 7.53880203e-01 -7.24519432e-01
-6.75749406e-02 -2.25903809e-01 -2.17836048e-03 -1.10554397e+00
-2.85461694e-01 -8.09016049e-01 -2.71366775e-01 8.65633488e-01
5.58788955e-01 -7.65601039e-01 6.15123630e-01 2.39401296e-01
-2.18685910e-01 -1.18540096e+00 -9.68421817e-01 -4.15658534e-01
-2.96143234e-01 -3.43091339e-01 3.80622596e-01 1.08682120e+00
-1.56347573e-01 5.88754177e-01 -5.72230935e-01 1.20082252e-01
3.36912096e-01 2.21422482e-02 3.09225470e-01 -1.65225422e+00
-5.47176242e-01 -4.48328555e-02 -9.18520689e-01 -2.04314038e-01
-5.00328898e-01 -8.48225951e-01 -4.95611876e-01 -1.41570246e+00
-1.28632024e-01 -4.15145099e-01 -9.96177137e-01 3.47625047e-01
6.73538223e-02 6.28381371e-01 -6.85663447e-02 8.70344117e-02
-1.72187254e-01 -2.43458394e-02 7.31399238e-01 9.83900651e-02
-4.54310119e-01 4.55994368e-01 -5.82853019e-01 6.64417982e-01
1.16679013e+00 -4.41891491e-01 -4.74974155e-01 -1.94399133e-01
-8.98212194e-03 4.53199565e-01 3.15230697e-01 -1.51518834e+00
-1.36816725e-01 9.82407928e-01 8.22402179e-01 -4.59611148e-01
2.12271228e-01 -6.86264455e-01 3.92863750e-01 7.94351041e-01
-2.61303753e-01 5.03769338e-01 3.01777363e-01 3.90852034e-01
-1.77486986e-01 1.03644602e-01 7.16647625e-01 -1.98784575e-01
-5.14046997e-02 3.03609192e-01 -5.83144963e-01 -1.48359574e-02
9.62115824e-01 -3.21494818e-01 3.55892926e-01 -2.85610735e-01
-1.40527356e+00 -2.20347643e-01 -1.91284463e-01 4.09834802e-01
6.44036591e-01 -1.07466018e+00 -8.57803881e-01 5.01842856e-01
1.88482270e-01 -6.07296005e-02 4.48678464e-01 1.03282011e+00
-4.82103288e-01 3.95481229e-01 -3.41356128e-01 -9.67930198e-01
-1.24026644e+00 2.23713964e-01 8.76787424e-01 -3.44020098e-01
-9.58171129e-01 6.96073592e-01 -3.06639016e-01 -2.37383768e-01
3.84343147e-01 -5.22855639e-01 -4.07944053e-01 3.82120609e-02
3.95316243e-01 3.00764769e-01 7.80289769e-01 -2.66263336e-01
-4.59387571e-01 4.32057619e-01 9.80440825e-02 3.00837338e-01
1.38830495e+00 1.08994991e-01 1.54682457e-01 8.08554709e-01
9.74150956e-01 -5.14023364e-01 -3.89362276e-01 1.12550341e-01
-1.22883484e-01 1.62860766e-01 1.87652513e-01 -1.15090990e+00
-1.38512194e+00 1.11611700e+00 1.38766634e+00 3.09592038e-01
1.29752612e+00 -2.52355963e-01 7.01877236e-01 2.38362953e-01
4.16697323e-01 -7.89516747e-01 -5.23948014e-01 1.41106576e-01
7.40802109e-01 -1.00336528e+00 -2.14827970e-01 1.67532172e-02
-5.58271825e-01 1.15017748e+00 1.59303054e-01 -1.89382821e-01
9.77501512e-01 3.60432595e-01 4.07629162e-01 -3.39157194e-01
-4.20091450e-01 7.65253082e-02 1.67690471e-01 5.14591515e-01
9.05592740e-01 1.65325031e-01 -5.37575603e-01 8.94739807e-01
-1.11999223e-02 4.08656836e-01 4.74200904e-01 1.00694096e+00
1.61840305e-01 -1.11305296e+00 -6.93663657e-02 9.35550630e-01
-1.05846620e+00 -1.52027145e-01 -1.75212756e-01 5.62233508e-01
2.44879439e-01 8.52950990e-01 -2.25221649e-01 -3.37506890e-01
3.15954983e-01 3.78730178e-01 4.32638198e-01 -6.96323037e-01
-1.20803630e+00 1.62785321e-01 7.00086430e-02 -3.00592571e-01
-3.96498621e-01 -4.54656959e-01 -1.16207588e+00 3.23131174e-01
-2.95682400e-01 8.29448029e-02 5.75322807e-01 7.42575347e-01
8.53034437e-01 1.16147149e+00 2.95938790e-01 -5.73054194e-01
-5.35045266e-01 -1.22290313e+00 -8.18610907e-01 3.16150814e-01
5.83005309e-01 -3.15062910e-01 2.87408940e-02 3.13687511e-02] | [14.282831192016602, 3.281224012374878] |
37977ff4-14ef-4024-9ee5-668cb55f507a | chance-constrained-ac-optimal-power-flow-for | 2207.09520 | null | https://arxiv.org/abs/2207.09520v1 | https://arxiv.org/pdf/2207.09520v1.pdf | Chance-Constrained AC Optimal Power Flow for Unbalanced Distribution Grids | The growing penetration of distributed energy resources (DERs) is leading to continually changing operating conditions, which need to be managed efficiently by distribution grid operators. The intermittent nature of DERs such as solar photovoltaic (PV) systems as well as load forecasting errors not only increase uncertainty in the grid, but also pose significant power quality challenges such as voltage unbalance and voltage magnitude violations. This paper leverages a chance-constrained optimization approach to reduce the impact of uncertainty on distribution grid operation. We first present the chance-constrained optimal power flow (CC-OPF) problem for distribution grids and discuss a reformulation based on constraint tightening that does not require any approximations or relaxations of the three-phase AC power flow equations. We then propose two iterative solution algorithms capable of efficiently solving the reformulation. In the case studies, the performance of both algorithms is analyzed by running simulations on the IEEE 13-bus test feeder using real PV and load measurement data. The simulation results indicate that both methods are able to enforce the chance constraints in in- and out-of-sample evaluations. | ['Line A. Roald', 'Ashley M. Hou', 'Kshitij Girigoudar'] | 2022-07-19 | null | null | null | null | ['load-forecasting'] | ['miscellaneous'] | [-1.04260504e-01 -1.24500208e-01 6.40784651e-02 -4.38527167e-02
-4.27333474e-01 -1.03476703e+00 2.30099514e-01 4.91694510e-01
2.47531638e-01 1.43744826e+00 -1.24267772e-01 -2.62821436e-01
-7.76792407e-01 -8.16767991e-01 -8.14169422e-02 -1.04051149e+00
-4.50985730e-01 4.52361494e-01 -3.01204056e-01 2.29251832e-02
3.13565642e-01 5.98293364e-01 -1.09257042e+00 -3.28994721e-01
1.51342762e+00 1.09336305e+00 -3.45228873e-02 1.13205306e-01
5.38959801e-01 2.64111608e-01 -9.96679187e-01 2.80467927e-01
3.93394470e-01 -4.11469698e-01 -1.81082085e-01 7.69339725e-02
-4.57106739e-01 -5.19878149e-01 1.49039105e-01 1.42785621e+00
7.78167307e-01 3.34566534e-01 5.98277330e-01 -1.96483612e+00
-1.93400502e-01 7.04194963e-01 -7.55517602e-01 5.12994468e-01
2.92157501e-01 1.20577522e-01 9.71368790e-01 -5.06850898e-01
3.69651169e-02 6.81807578e-01 3.27607423e-01 -2.14068621e-01
-1.41760790e+00 -2.63004869e-01 2.32938126e-01 6.25019312e-01
-1.35213327e+00 2.02797979e-01 8.46149385e-01 -3.38465869e-01
1.17478776e+00 4.37595934e-01 1.10052550e+00 -1.38441809e-02
1.76796228e-01 4.53196704e-01 1.22205818e+00 -3.36963713e-01
7.66678274e-01 -2.22901165e-01 1.86620392e-02 -4.15381461e-01
8.32284927e-01 2.40959504e-04 -4.81236055e-02 -1.85993403e-01
3.60095277e-02 -5.29497087e-01 -9.82113302e-01 -5.29119909e-01
-7.74137259e-01 9.12219465e-01 5.32277465e-01 4.35488075e-01
-3.98363441e-01 -2.24185079e-01 2.56206304e-01 1.84484512e-01
2.73957610e-01 3.40184897e-01 -4.87939835e-01 -6.04845695e-02
-1.10094380e+00 1.96825758e-01 9.48624432e-01 8.86304379e-01
1.03412189e-01 7.66765356e-01 -3.68718728e-02 3.33747268e-01
4.27413881e-01 8.44818115e-01 2.48448089e-01 -6.74582601e-01
2.51682520e-01 4.14346576e-01 4.00807023e-01 -6.82984948e-01
-3.30180973e-01 -9.16327536e-01 -9.18919861e-01 7.62342453e-01
4.91321415e-01 -4.84240144e-01 -4.39613551e-01 1.33192039e+00
1.51957393e-01 -3.66629869e-01 -1.37329802e-01 6.30423963e-01
-2.75562286e-01 1.10796916e+00 -3.26019049e-01 -1.12444353e+00
1.19028187e+00 -5.49545348e-01 -1.01991260e+00 3.23019862e-01
1.29464164e-01 -6.07871890e-01 1.30432859e-01 6.74500942e-01
-1.35636675e+00 1.18399560e-01 -1.30855131e+00 5.69934011e-01
-2.90671676e-01 -1.93110555e-01 -1.29025310e-01 7.72509813e-01
-7.85320580e-01 5.17830491e-01 -6.32781565e-01 7.00613633e-02
4.09602284e-01 1.92746192e-01 6.05958179e-02 1.89198375e-01
-1.01156831e+00 1.40600359e+00 7.81680763e-01 7.08414853e-01
-3.95008594e-01 -1.12612939e+00 -6.53319418e-01 7.45151281e-01
4.80411947e-01 -5.75467110e-01 1.02060854e+00 -6.40821874e-01
-1.38006353e+00 -3.49938840e-01 6.48094118e-02 -5.71479261e-01
4.39485192e-01 2.81992793e-01 -3.51106763e-01 1.23131365e-01
-1.79066911e-01 -3.21024090e-01 2.91427135e-01 -1.30410790e+00
-7.94746459e-01 -1.72631666e-01 -2.99840719e-01 2.89357781e-01
3.04073980e-03 -3.39975089e-01 7.85866082e-01 -6.19050205e-01
-1.35202691e-01 -3.02120358e-01 -5.86079478e-01 -4.32594985e-01
-4.23077047e-01 -2.51212329e-01 7.82261610e-01 -9.48544919e-01
1.23162389e+00 -1.68872225e+00 8.31057653e-02 9.13969100e-01
-4.09668237e-01 5.04420213e-02 2.74744481e-01 9.64455903e-01
-4.38771486e-01 -9.60013568e-02 -2.83489615e-01 3.63745749e-01
6.09920740e-01 4.86051708e-01 -2.11315125e-01 4.90101874e-01
-5.85369095e-02 5.98786175e-01 -9.11205053e-01 3.49126101e-01
5.66896677e-01 2.25466162e-01 -2.95258194e-01 -6.73186257e-02
-3.09181988e-01 3.40823770e-01 -4.82445955e-02 4.16496158e-01
8.47605169e-01 -1.25172213e-01 5.30281186e-01 -3.58017474e-01
-3.12546194e-01 -8.41335580e-03 -1.67980850e+00 1.15630388e+00
-3.12237382e-01 5.06879032e-01 1.89575166e-01 -1.43410039e+00
4.84678686e-01 2.46908128e-01 6.98176980e-01 -9.15304244e-01
-1.16333447e-01 3.34991306e-01 3.94923896e-01 -1.66497231e-01
-2.06987951e-02 -2.12143529e-02 2.04395831e-01 4.38008130e-01
1.25592351e-01 -2.78132409e-01 1.02720439e+00 -6.44357800e-02
4.99951750e-01 -2.78889537e-01 7.23726332e-01 -9.38547075e-01
8.63225996e-01 -1.89846724e-01 1.15712106e+00 -1.46829318e-02
1.63653478e-01 1.36655658e-01 9.51166630e-01 2.52672344e-01
-9.35331583e-01 -1.05704379e+00 -4.06091839e-01 -1.68071434e-01
-1.26179427e-01 -2.38522962e-02 -3.66001338e-01 -4.05090541e-01
2.70376503e-01 1.68937027e+00 -9.33437720e-02 3.52096021e-01
-2.76898950e-01 -1.31546772e+00 -4.89790469e-01 4.07044023e-01
1.83798507e-01 -2.31298804e-01 -6.64006829e-01 4.12913650e-01
2.83896834e-01 -6.55575633e-01 -1.66365370e-01 5.19054890e-01
-5.59533060e-01 -1.24290466e+00 -9.43644643e-01 -3.76603872e-01
1.16813469e+00 -3.49996328e-01 1.08268797e+00 -1.25038654e-01
-2.18377650e-01 2.30378121e-01 -1.77285504e-02 -3.22334975e-01
1.24573465e-02 -2.68193960e-01 9.72727686e-02 -4.49565709e-01
-1.73044592e-01 -1.03019929e+00 -5.59139609e-01 2.26475775e-01
-6.65891647e-01 -3.84344429e-01 1.49535388e-01 1.04569304e+00
3.66158158e-01 1.21908033e+00 1.30516410e+00 -2.85138071e-01
8.08282912e-01 -6.50226116e-01 -1.72950518e+00 4.37781453e-01
-1.16869831e+00 -2.27798998e-01 8.04272234e-01 6.09107018e-02
-1.03242505e+00 -2.04907805e-01 8.79664272e-02 4.98720556e-02
8.28626156e-02 8.23661685e-01 -5.94321370e-01 -2.32906759e-01
-2.96918035e-01 1.05800210e-02 -3.83613914e-01 -3.66741031e-01
1.08662881e-01 3.44176412e-01 2.58328617e-01 -3.03690851e-01
1.28787541e+00 -8.34259912e-02 4.85614061e-01 -3.29527706e-01
-2.71349311e-01 -9.97040868e-02 -3.55087608e-01 -3.17338109e-01
1.83286697e-01 -7.23586679e-01 -1.32459092e+00 5.94295502e-01
-7.60517895e-01 -1.07287467e-01 -5.89751959e-01 3.99772108e-01
-1.91142306e-01 6.71551645e-01 -1.65041238e-01 -1.01066506e+00
-3.06141466e-01 -1.02832508e+00 -2.46680304e-02 7.08957970e-01
1.50132492e-01 -1.35708904e+00 -1.93531644e-02 -3.08145918e-02
6.00468397e-01 7.85972834e-01 1.19458926e+00 -6.13807976e-01
-7.25951433e-01 1.15931757e-01 8.63362942e-03 7.85824120e-01
1.44361943e-01 1.84560043e-03 -4.04007584e-01 -9.41562772e-01
1.21094368e-01 3.55311066e-01 -2.91829318e-01 4.28607136e-01
6.47451162e-01 -6.77812040e-01 -2.78199874e-02 2.58341014e-01
2.35248661e+00 6.57850921e-01 3.24954391e-01 1.40704647e-01
-3.74639519e-02 3.81429464e-01 5.47158778e-01 8.84354472e-01
4.36209708e-01 5.55960178e-01 7.35021651e-01 3.07974279e-01
5.44641256e-01 1.70366541e-01 1.59388170e-01 8.39333951e-01
1.22736804e-01 -6.79937303e-01 -5.98954082e-01 7.58798957e-01
-1.79910338e+00 -8.98903668e-01 -3.04867953e-01 2.18397045e+00
5.10824800e-01 -1.42987698e-01 -7.15970695e-02 6.90867543e-01
6.98492169e-01 -1.64957240e-01 -5.25086522e-01 -5.22407651e-01
-3.88407946e-01 3.01087618e-01 5.15534639e-01 6.38574839e-01
-3.10321480e-01 -6.15995169e-01 5.32022381e+00 7.04644322e-01
-8.60720098e-01 -1.25412002e-01 6.68497384e-01 -1.09811038e-01
-5.10106921e-01 1.04902409e-01 -3.61705720e-01 1.13773406e+00
9.20266211e-01 -1.44439447e+00 7.62044251e-01 5.21460235e-01
7.65642464e-01 -9.32334900e-01 -1.05880022e+00 6.08115137e-01
-3.25212888e-02 -1.02021945e+00 -4.37430739e-01 2.60539383e-01
1.38768017e+00 -2.15271637e-01 -5.15623212e-01 -7.03012273e-02
9.68574435e-02 -8.66084754e-01 6.59772813e-01 3.83319438e-01
-4.08759527e-02 -1.32314551e+00 1.10601521e+00 3.39461148e-01
-9.67142820e-01 -5.73881269e-01 4.93932664e-02 -1.45213857e-01
1.06993306e+00 1.33183968e+00 -4.31990862e-01 1.39349473e+00
6.63952291e-01 4.70004439e-01 4.89700735e-02 1.69866347e+00
-8.57826948e-01 4.59043443e-01 -6.92449212e-01 1.43264189e-01
-2.02794388e-01 -5.35153210e-01 6.63172603e-01 3.68366450e-01
5.44917643e-01 2.18381792e-01 6.26643747e-02 9.56393778e-01
1.53328910e-01 -2.34227955e-01 6.47241846e-02 2.33715475e-01
7.48469770e-01 1.40125823e+00 -5.01158595e-01 -1.96771055e-01
-3.33946019e-01 2.46731624e-01 -5.34251273e-01 7.44779646e-01
-6.19338691e-01 -3.61696094e-01 4.66973096e-01 -3.41148943e-01
3.31393629e-01 -2.17978120e-01 -5.59030473e-01 -1.05883205e+00
4.10982132e-01 -7.75150418e-01 6.19984686e-01 -8.60385656e-01
-1.68697000e+00 1.86478943e-02 3.02368999e-01 -1.10056734e+00
-9.77551222e-01 -3.84693921e-01 -9.21465576e-01 1.55101621e+00
-2.24703193e+00 -4.15325671e-01 1.29060568e-02 3.14781785e-01
2.90220320e-01 1.21728294e-01 6.52001441e-01 5.50540149e-01
-8.22027862e-01 1.85034368e-02 7.47112989e-01 -3.00372809e-01
-3.87196004e-01 -1.59183145e+00 -5.90411663e-01 1.34494352e+00
-3.87110263e-01 -8.92606080e-02 9.05272186e-01 -2.32290730e-01
-1.41439486e+00 -3.87071550e-01 7.23016083e-01 6.06831789e-01
9.24338758e-01 -8.54358301e-02 -7.76499271e-01 5.04670084e-01
1.19929194e+00 -1.99752208e-02 5.69909036e-01 -5.50695181e-01
3.67581576e-01 -3.46782476e-01 -1.64574349e+00 1.98535323e-01
-4.23907377e-02 2.57095098e-02 -5.86476445e-01 2.96122968e-01
-3.68640631e-01 -3.72647107e-01 -1.39644599e+00 6.55009747e-01
9.02544111e-02 -6.67360246e-01 6.13215387e-01 4.90863845e-02
-3.56248617e-01 -9.01859164e-01 3.78667302e-02 -2.21421647e+00
-5.38955070e-02 -8.56146097e-01 -4.43470716e-01 1.45770204e+00
2.40768507e-01 -1.33305693e+00 3.18032235e-01 5.23134649e-01
5.73948622e-02 -5.67797959e-01 -1.60852253e+00 -9.87477899e-01
1.94409400e-01 2.55477846e-01 7.86515713e-01 1.06255579e+00
7.19222009e-01 -5.13621904e-02 3.16226155e-01 7.70862579e-01
1.15508795e+00 3.41807246e-01 8.30575451e-02 -9.42023396e-01
-2.10038155e-01 -6.61581278e-01 -1.78320214e-01 -3.32777858e-01
-1.79139704e-01 -4.63107318e-01 -2.68262595e-01 -1.89006472e+00
-3.19082201e-01 -1.37073830e-01 -3.09855133e-01 3.58416498e-01
8.06935281e-02 -1.51494220e-01 4.63463634e-01 -2.58295715e-01
2.32898638e-01 8.51062417e-01 5.86143494e-01 -1.96786344e-01
3.03161681e-01 1.89321265e-01 5.44967018e-02 4.37757552e-01
9.81477261e-01 -1.26179894e-02 -6.61947548e-01 -1.41676828e-01
5.28212249e-01 5.22103250e-01 3.92638752e-03 -1.03067172e+00
2.25111216e-01 -8.47096518e-02 3.76539022e-01 -1.07087684e+00
-1.83017194e-01 -1.65856481e+00 6.86519265e-01 9.13986087e-01
2.87950575e-01 4.22029912e-01 1.21138580e-01 2.46294200e-01
-3.12811673e-01 -6.59813225e-01 6.86742246e-01 3.60151291e-01
-4.43611145e-01 -1.91927701e-01 -5.62637687e-01 -1.47881329e-01
1.57079613e+00 9.67134628e-03 -6.61668539e-01 -4.58581537e-01
-8.18813860e-01 1.24292803e+00 2.69397587e-01 1.42962381e-01
4.09753509e-02 -1.19593561e+00 -7.46240854e-01 2.71092094e-02
-5.18907666e-01 -3.67418081e-02 2.96986163e-01 8.93444598e-01
-3.47820610e-01 6.46962047e-01 -1.35260805e-01 -5.52131414e-01
-8.54222119e-01 3.64926726e-01 6.07673109e-01 -6.08979464e-01
-8.09387863e-02 1.84936505e-02 -6.75308347e-01 2.73965150e-01
8.89113098e-02 -4.15070146e-01 -8.63179639e-02 7.23140001e-01
2.91418910e-01 1.04108918e+00 3.62955987e-01 1.79192442e-02
-2.26452246e-01 8.96945074e-02 5.12061238e-01 9.01285633e-02
1.49054742e+00 -4.98545527e-01 -3.77244025e-01 2.75385529e-01
6.55452490e-01 -6.60074800e-02 -9.89496887e-01 3.53809774e-01
1.06757484e-01 -4.80155885e-01 1.09044179e-01 -1.49240732e+00
-1.59462416e+00 5.19170046e-01 3.54324490e-01 9.71608579e-01
1.43882096e+00 -9.76587713e-01 4.41191256e-01 -2.69371625e-02
6.31729126e-01 -1.16606951e+00 -7.26851702e-01 1.20863430e-01
7.40977049e-01 -4.97402698e-01 3.77673537e-01 -3.53480786e-01
4.95869331e-02 1.21778965e+00 7.79881924e-02 -2.04094812e-01
6.63478374e-01 6.35122657e-01 -4.52904522e-01 4.50635791e-01
-8.47997606e-01 1.30779356e-01 4.08071764e-02 4.47904468e-01
1.44439293e-02 9.87619087e-02 -1.00047219e+00 1.87180504e-01
1.85957968e-01 2.53817607e-02 9.52527642e-01 1.06798220e+00
-8.47734362e-02 -1.11589766e+00 -5.00812590e-01 4.62238044e-01
-3.71556461e-01 7.26193935e-02 6.14267230e-01 7.62300730e-01
3.81473079e-02 1.04468238e+00 1.54770851e-01 8.51098180e-01
5.69949210e-01 -1.82856262e-01 4.66728002e-01 -1.46427929e-01
-5.39403915e-01 1.74516663e-01 1.27453208e-01 -2.44720690e-02
-3.05175990e-01 -1.07309043e+00 -1.33822095e+00 -3.09619933e-01
-7.38804996e-01 7.39008069e-01 7.70613790e-01 9.47275400e-01
-8.73143747e-02 8.00312757e-01 1.04455674e+00 -7.60608137e-01
-1.13519835e+00 -7.61370242e-01 -1.03599489e+00 -2.31505677e-01
-3.97594273e-03 -3.95291954e-01 -7.84704268e-01 -3.32658648e-01] | [5.700668811798096, 2.55151629447937] |
70a98ab7-2245-4e76-98f1-5ba13cab5ba0 | bag-of-tricks-for-efficient-text | 1607.01759 | null | http://arxiv.org/abs/1607.01759v3 | http://arxiv.org/pdf/1607.01759v3.pdf | Bag of Tricks for Efficient Text Classification | This paper explores a simple and efficient baseline for text classification.
Our experiments show that our fast text classifier fastText is often on par
with deep learning classifiers in terms of accuracy, and many orders of
magnitude faster for training and evaluation. We can train fastText on more
than one billion words in less than ten minutes using a standard multicore~CPU,
and classify half a million sentences among~312K classes in less than a minute. | ['Edouard Grave', 'Armand Joulin', 'Piotr Bojanowski', 'Tomas Mikolov'] | 2016-07-06 | bag-of-tricks-for-efficient-text-1 | https://aclanthology.org/E17-2068 | https://aclanthology.org/E17-2068.pdf | eacl-2017-4 | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [-3.56484711e-01 -3.36783201e-01 -3.50413531e-01 -8.19586575e-01
-8.58839571e-01 -8.38765085e-01 7.91489065e-01 6.77897930e-01
-9.23440993e-01 8.13939273e-01 3.01914234e-02 -8.71640325e-01
2.86195278e-01 -8.75211060e-01 -5.20594299e-01 -3.52755964e-01
1.66539446e-01 7.89491892e-01 2.57950187e-01 -1.51309416e-01
2.90258586e-01 3.09597135e-01 -1.28524768e+00 8.02645504e-01
5.43944001e-01 1.01854384e+00 -1.76420003e-01 1.32609999e+00
-8.13325524e-01 8.92214298e-01 -1.08993173e+00 -7.52505124e-01
-1.10420763e-01 1.44148365e-01 -1.53107870e+00 -3.81835520e-01
8.00473154e-01 -6.79974914e-01 -7.32000470e-01 8.24652195e-01
4.93520975e-01 -3.36076654e-02 5.69392741e-01 -9.03998375e-01
1.69745591e-02 1.04972494e+00 -3.43000919e-01 6.21043026e-01
6.01677954e-01 -1.92147329e-01 1.26875782e+00 -7.14288235e-01
3.47135663e-01 1.19989979e+00 7.91184485e-01 2.76842147e-01
-1.09060872e+00 -9.85500872e-01 2.22326770e-01 1.38011733e-02
-1.20064247e+00 -5.13266802e-01 -2.88925916e-02 -2.62473017e-01
1.34995341e+00 3.06617141e-01 6.08150721e-01 1.39469099e+00
7.59213567e-01 1.19411194e+00 9.99603808e-01 -4.43400860e-01
2.48902682e-02 1.52728841e-01 1.05650091e+00 7.77827978e-01
4.22348887e-01 -4.63506877e-01 -5.51092148e-01 -5.52168846e-01
1.20196519e-02 -6.93808645e-02 -9.20716524e-02 7.07107484e-01
-1.26957119e+00 1.07363749e+00 8.92810598e-02 5.28575957e-01
2.15498507e-01 4.73704129e-01 1.02477157e+00 8.32686007e-01
9.02758837e-01 3.90871286e-01 -8.31626654e-01 -7.41984308e-01
-8.05739641e-01 2.34342635e-01 1.31830812e+00 9.70779359e-01
6.24550879e-01 -1.07665993e-01 2.89486974e-01 7.34517813e-01
-1.24029204e-01 5.65100312e-01 8.95418167e-01 -3.08711052e-01
8.15285504e-01 3.13885421e-01 -1.94743901e-01 -5.57073116e-01
-5.48462749e-01 -4.31439787e-01 -1.07796621e+00 -4.23401967e-02
2.94197083e-01 -4.41316664e-01 -8.35448503e-01 8.71692300e-01
8.24124441e-02 -1.69909105e-01 -2.22846389e-01 2.19531864e-01
1.08627129e+00 8.95061791e-01 8.67295489e-02 5.85015817e-03
1.33422434e+00 -8.34747851e-01 -4.38202709e-01 -2.76607871e-01
1.44023693e+00 -9.14606988e-01 1.09759510e+00 6.38290763e-01
-8.90897155e-01 -3.53527576e-01 -1.14682782e+00 -3.21484953e-01
-7.32657790e-01 -6.35962188e-02 1.04217052e+00 8.62510443e-01
-8.37109864e-01 7.60749102e-01 -7.96669722e-01 -4.02112126e-01
4.67778504e-01 5.29067397e-01 -3.84391218e-01 8.25771391e-02
-1.00834429e+00 7.25577593e-01 5.70200205e-01 -4.32338327e-01
-3.64501148e-01 -5.10488689e-01 -5.41914403e-01 6.43037409e-02
-2.12487638e-01 -5.97090602e-01 1.70350647e+00 -6.72455192e-01
-1.50051522e+00 8.90456378e-01 -2.37536103e-01 -4.99969482e-01
4.82732356e-01 -5.94088972e-01 -4.53823477e-01 1.00126810e-01
-7.95462355e-02 2.13632524e-01 6.22460902e-01 -3.94627005e-01
-1.11122453e+00 -2.09337875e-01 7.05563426e-02 1.34183854e-01
-9.47176158e-01 3.50978822e-01 -1.02223687e-01 -6.39992774e-01
-2.12828502e-01 -5.97694635e-01 1.44885987e-01 -1.55139327e-01
-5.11917531e-01 -7.15917289e-01 7.59549320e-01 -2.08274439e-01
1.15571785e+00 -1.78299022e+00 -3.39168191e-01 5.63027784e-02
4.82240111e-01 2.65678048e-01 -1.26884490e-01 4.42289799e-01
1.38799652e-01 2.99984783e-01 2.49361739e-01 -2.58258134e-01
1.25514537e-01 -4.08659466e-02 -4.35834378e-01 6.57712579e-01
-4.95854318e-01 9.18779671e-01 -7.57800758e-01 -5.86306512e-01
1.83990374e-02 1.01817966e-01 -1.05472498e-01 -5.09458929e-02
-4.93152104e-02 -4.83750761e-01 -5.81676185e-01 2.88888752e-01
4.34171259e-01 -4.79965597e-01 7.32978210e-02 2.25269839e-01
5.01266606e-02 1.05215931e+00 -8.93975496e-01 1.51071274e+00
-6.91309452e-01 1.40696526e+00 -5.16278446e-02 -1.11180258e+00
6.13279641e-01 4.22886491e-01 1.65013030e-01 -3.72988999e-01
5.65257668e-01 3.15211833e-01 -1.62188172e-01 -1.05091751e-01
4.70257163e-01 1.07374147e-01 -1.03797123e-01 1.11123180e+00
1.88364238e-01 -7.80566335e-02 2.80860066e-01 3.80806655e-01
1.31409359e+00 -7.53241718e-01 1.87097192e-01 -5.32077909e-01
1.29059240e-01 1.86461195e-01 -1.23769157e-01 9.91211653e-01
8.26467667e-03 -1.89062245e-02 4.56378430e-01 -1.24528980e+00
-9.45977807e-01 -3.72428894e-01 -3.51058751e-01 1.82936871e+00
-2.66939729e-01 -7.01130807e-01 -6.96212173e-01 -9.42893386e-01
2.56253541e-01 5.97469151e-01 -5.85427940e-01 2.02377319e-01
-5.81815660e-01 -1.16610944e+00 9.10205960e-01 7.93328464e-01
6.48245394e-01 -7.66862094e-01 -1.18217990e-02 2.53481925e-01
6.52158707e-02 -1.00973439e+00 -2.93134242e-01 7.15765059e-01
-1.00915301e+00 -9.55991089e-01 -4.37611282e-01 -1.01951551e+00
4.00672913e-01 4.32947010e-01 1.57256913e+00 4.04003054e-01
-3.70378464e-01 -1.91454068e-01 -4.98712897e-01 -4.99760479e-01
-6.33261323e-01 9.52136099e-01 7.00605363e-02 -6.47049189e-01
8.37817848e-01 -3.46391052e-01 -4.27341431e-01 -1.27760261e-01
-4.84075040e-01 -1.31111413e-01 6.56845510e-01 7.83611536e-01
-1.22959331e-01 2.59776950e-01 3.33967119e-01 -1.35590291e+00
7.66032636e-01 -4.92373526e-01 -4.77790475e-01 1.42289758e-01
-8.30418646e-01 1.25087798e-01 9.12970960e-01 -5.04455507e-01
-5.73289514e-01 -1.56729043e-01 -6.89291596e-01 3.71343285e-01
-1.78385898e-01 5.47475159e-01 6.23190820e-01 -7.05597624e-02
1.02546775e+00 2.88325369e-01 -4.59395409e-01 -5.59418261e-01
2.56489515e-01 1.29569530e+00 1.65631827e-02 -4.98500556e-01
6.47615135e-01 2.74041295e-01 -3.74189794e-01 -9.34972584e-01
-1.13352406e+00 -6.70258164e-01 -7.97088087e-01 3.04196864e-01
5.48357725e-01 -9.61608171e-01 -7.89832354e-01 6.17528319e-01
-1.04609919e+00 -6.52445614e-01 1.34410143e-01 4.57392573e-01
3.39319482e-02 3.44573617e-01 -1.30761516e+00 -1.77655295e-01
-1.19791436e+00 -5.96740305e-01 9.76957500e-01 -1.96011215e-01
-2.20035896e-01 -1.22937083e+00 6.10631332e-02 2.57496566e-01
3.25987816e-01 -4.95366812e-01 9.37481225e-01 -9.25758123e-01
1.40614346e-01 -8.95945668e-01 -4.70258147e-01 4.92747985e-02
-8.56025591e-02 1.00142337e-01 -1.00758088e+00 -5.44651568e-01
-2.61716545e-01 -8.73093784e-01 1.05428600e+00 1.26743898e-01
1.50352764e+00 -3.25419217e-01 -5.93124807e-01 5.99391878e-01
1.26873887e+00 3.33970524e-02 2.01890931e-01 4.24373806e-01
6.49513781e-01 1.40538648e-01 4.13718581e-01 2.60985553e-01
1.58883125e-01 2.70397037e-01 -9.83510390e-02 7.36745745e-02
1.07417934e-01 2.13886470e-01 1.73520416e-01 1.14791334e+00
3.99831027e-01 -7.23818064e-01 -1.23098183e+00 2.30184361e-01
-1.51196456e+00 -8.99874926e-01 -4.65077728e-01 1.67000103e+00
1.08308244e+00 5.68060637e-01 5.90092205e-02 4.75887030e-01
4.12182659e-01 1.70873314e-01 -2.50453591e-01 -6.43884778e-01
8.20778161e-02 6.13628626e-01 4.92107391e-01 6.90711379e-01
-1.34836137e+00 9.63930964e-01 8.32154179e+00 9.74117160e-01
-1.32817221e+00 8.30368623e-02 9.56151187e-01 -2.09994912e-01
-3.30352713e-03 -4.80563641e-01 -1.20845401e+00 4.32550073e-01
1.44255662e+00 -4.73029196e-01 1.00157745e-01 1.25176346e+00
-4.47591603e-01 -3.75652127e-02 -1.14007461e+00 1.12797201e+00
-3.84156853e-02 -1.45279825e+00 2.71114647e-01 1.17743827e-01
6.05670631e-01 7.17357576e-01 -2.18992382e-01 5.39632857e-01
8.04786742e-01 -1.11832511e+00 5.03385663e-01 -1.84220642e-01
8.52312326e-01 -9.03366029e-01 1.12426507e+00 5.48441589e-01
-1.03634405e+00 -5.06680198e-02 -6.60605609e-01 -1.76286533e-01
-5.31313837e-01 6.41224802e-01 -7.47077703e-01 -5.69689423e-02
8.62056613e-01 4.85293448e-01 -7.84572661e-01 7.24942982e-01
1.32821634e-01 9.73569453e-01 -4.74301279e-01 -8.69656503e-01
5.81198335e-01 4.80094492e-01 -1.70201734e-01 1.83661914e+00
4.11039330e-02 3.37490559e-01 2.77420163e-01 -3.73340815e-01
-4.69574481e-01 4.79031175e-01 -4.28177744e-01 8.50742683e-03
2.95192063e-01 1.16366100e+00 -9.39310551e-01 -1.08906698e+00
-4.20900166e-01 1.11037612e+00 5.59249818e-01 -2.33221184e-02
-4.10647750e-01 -1.22068274e+00 5.27996838e-01 -2.75484920e-01
8.35095868e-02 -1.71654284e-01 -4.38255310e-01 -1.46919322e+00
4.66231331e-02 -9.56428111e-01 3.45475852e-01 -3.50725770e-01
-1.47582662e+00 1.01276469e+00 -3.59489143e-01 -6.98505521e-01
-3.78432065e-01 -1.26435912e+00 -5.19850969e-01 6.85919166e-01
-1.18447292e+00 -6.51057124e-01 -5.39502203e-01 8.09199214e-01
6.68859303e-01 -1.64208949e-01 1.24116254e+00 1.82801753e-01
-6.64122820e-01 9.16755676e-01 7.70753324e-01 6.38222158e-01
7.06422210e-01 -1.70055985e+00 1.24250662e+00 2.23808885e-01
3.11488420e-01 3.89861345e-01 6.65342450e-01 -2.72378504e-01
-1.41948831e+00 -9.47818875e-01 1.26498914e+00 -5.39652646e-01
9.28236187e-01 -7.38229811e-01 -8.38109374e-01 6.56997204e-01
3.09575230e-01 1.34939060e-01 8.78768742e-01 7.73644269e-01
-7.28244483e-01 -2.11071119e-01 -7.43236780e-01 3.95128489e-01
8.28417301e-01 -8.34671438e-01 -3.16922218e-01 1.12791717e+00
5.87863147e-01 -2.97040582e-01 -8.68273795e-01 -1.69752032e-01
7.36915827e-01 -5.21988928e-01 5.09488940e-01 -8.24360967e-01
2.59947926e-01 6.70758307e-01 1.26917407e-01 -1.16754746e+00
-4.45726126e-01 -7.79783607e-01 -1.27790734e-01 7.41095304e-01
4.66808110e-01 -8.52047324e-01 1.11119819e+00 1.94102496e-01
-4.82708476e-02 -6.92562044e-01 -7.85776436e-01 -6.68509424e-01
7.63437152e-01 -6.56477809e-01 3.97454679e-01 1.05097425e+00
4.90138113e-01 6.74716115e-01 1.02414869e-01 -4.51652557e-01
2.44571060e-01 3.02145422e-01 7.62895763e-01 -1.57459712e+00
-2.30992138e-01 -7.14391112e-01 -4.18763578e-01 -1.51761127e+00
4.43941355e-01 -9.12506700e-01 -5.19086979e-02 -1.50196028e+00
4.31942284e-01 -4.99957055e-01 -2.54181087e-01 6.57769263e-01
-2.03832000e-01 4.12768930e-01 -4.27883655e-01 1.59951285e-01
-7.18049705e-01 1.60245433e-01 6.95159495e-01 -5.17841697e-01
2.88826346e-01 2.22226277e-01 -6.10471070e-01 7.72491693e-01
7.89577723e-01 -6.86535001e-01 -2.18469277e-01 -1.10704184e+00
5.84298968e-01 -1.78071380e-01 -4.57751244e-01 -1.08907807e+00
2.45982304e-01 3.06047220e-02 6.50794208e-01 -6.94791079e-01
7.80814216e-02 -3.04235220e-01 -7.39736855e-01 5.45830131e-01
-8.04878175e-01 2.71497697e-01 5.08701742e-01 3.91752779e-01
-8.60638246e-02 -5.28334618e-01 7.11380184e-01 2.82554030e-02
-8.15474987e-02 3.14913362e-01 -9.21460629e-01 9.72799882e-02
6.64639950e-01 2.62986600e-01 -7.71773756e-01 -3.94562334e-01
-2.84421563e-01 1.25737548e-01 1.18433438e-01 2.59731710e-01
2.08776191e-01 -8.53926778e-01 -8.21368098e-01 -1.36719272e-01
-6.15964690e-03 -2.43445367e-01 -2.15091020e-01 1.91011071e-01
-1.08182740e+00 9.45798993e-01 2.14148313e-01 -4.68458980e-01
-1.57399249e+00 2.49058440e-01 3.91989619e-01 -3.93959135e-01
-1.12448478e+00 1.07607591e+00 -1.23630673e-01 -1.93944931e-01
6.10518277e-01 -4.78854239e-01 9.67837796e-02 -6.13821484e-03
1.04596972e+00 2.14842811e-01 4.80434150e-01 -1.75530821e-01
-2.14539781e-01 2.19270661e-01 -7.97732174e-01 -2.46631831e-01
1.04075074e+00 3.44704002e-01 -4.40526940e-03 5.95184684e-01
1.70335197e+00 -2.28775755e-01 -6.38445973e-01 -2.73665428e-01
-6.05247915e-03 -3.02164346e-01 4.79161382e-01 -7.76739419e-01
-6.54404521e-01 1.13365948e+00 3.51024210e-01 5.71584046e-01
7.53094256e-01 -6.30368888e-02 1.13475466e+00 1.38556850e+00
4.00321543e-01 -1.11457825e+00 9.98620689e-02 7.98878372e-01
4.10117507e-01 -1.19806683e+00 4.45975065e-01 -1.84383571e-01
4.61176261e-02 1.59918833e+00 2.37065777e-01 -2.13233605e-01
1.05873978e+00 5.63824773e-01 1.91800341e-01 -2.48370200e-01
-1.38266802e+00 3.28802973e-01 1.17816515e-01 9.59155336e-02
8.82214665e-01 3.47688384e-02 -9.21231806e-02 -3.93049009e-02
-7.53712356e-01 -5.29601693e-01 3.46251845e-01 1.18295979e+00
-9.58598554e-01 -1.03830361e+00 9.76923630e-02 9.98333991e-01
-8.96481872e-01 -4.08116966e-01 -3.55478078e-01 5.20017803e-01
-4.03337538e-01 1.15706420e+00 4.64650363e-01 -6.75511420e-01
-5.92453890e-02 3.51701319e-01 2.67994583e-01 -9.21516180e-01
-9.97445881e-01 -3.12357515e-01 4.22215968e-01 -4.08073634e-01
8.10072273e-02 -4.84614760e-01 -9.87948835e-01 -1.36063409e+00
-6.89778447e-01 3.75276208e-01 1.12133133e+00 1.10192549e+00
1.32077247e-01 4.63172309e-02 8.04295719e-01 -4.60290074e-01
-7.37368286e-01 -1.32279301e+00 -3.94985586e-01 6.94335997e-03
2.45710909e-01 5.67777175e-03 -5.39235771e-01 -1.65705845e-01] | [10.66331958770752, 7.754453182220459] |
36e93352-3d49-4655-890d-76719a9d4c8f | robust-reflection-removal-with-flash-only | 2211.02914 | null | https://arxiv.org/abs/2211.02914v1 | https://arxiv.org/pdf/2211.02914v1.pdf | Robust Reflection Removal with Flash-only Cues in the Wild | We propose a simple yet effective reflection-free cue for robust reflection removal from a pair of flash and ambient (no-flash) images. The reflection-free cue exploits a flash-only image obtained by subtracting the ambient image from the corresponding flash image in raw data space. The flash-only image is equivalent to an image taken in a dark environment with only a flash on. This flash-only image is visually reflection-free and thus can provide robust cues to infer the reflection in the ambient image. Since the flash-only image usually has artifacts, we further propose a dedicated model that not only utilizes the reflection-free cue but also avoids introducing artifacts, which helps accurately estimate reflection and transmission. Our experiments on real-world images with various types of reflection demonstrate the effectiveness of our model with reflection-free flash-only cues: our model outperforms state-of-the-art reflection removal approaches by more than 5.23dB in PSNR. We extend our approach to handheld photography to address the misalignment between the flash and no-flash pair. With misaligned training data and the alignment module, our aligned model outperforms our previous version by more than 3.19dB in PSNR on a misaligned dataset. We also study using linear RGB images as training data. Our source code and dataset are publicly available at https://github.com/ChenyangLEI/flash-reflection-removal. | ['Qifeng Chen', 'Xudong Jiang', 'Chenyang Lei'] | 2022-11-05 | null | null | null | null | ['reflection-removal'] | ['computer-vision'] | [ 7.84501314e-01 -2.91538179e-01 5.20090640e-01 -8.77530798e-02
-1.02274168e+00 -5.10424614e-01 2.33843461e-01 -6.11459017e-01
-3.02509040e-01 4.68304217e-01 2.23451927e-01 -3.83259505e-01
3.34645629e-01 -4.07907218e-01 -8.97369146e-01 -9.72978830e-01
3.76009732e-01 -7.05771387e-01 3.88503551e-01 -1.22963481e-01
1.88150346e-01 9.64147672e-02 -1.59118509e+00 2.77587414e-01
7.88065195e-01 1.05722189e+00 5.85204661e-01 9.47103262e-01
4.28372413e-01 5.94433904e-01 -7.69990087e-01 2.63438020e-02
8.26627553e-01 -4.41933632e-01 -1.67695209e-02 9.20182467e-02
1.05197382e+00 -8.53340387e-01 -6.91056132e-01 9.17070508e-01
7.28135824e-01 -3.20173912e-02 9.07290801e-02 -1.05706894e+00
-6.69739962e-01 -3.08568686e-01 -8.78254950e-01 3.43195200e-02
6.74900711e-01 3.35282445e-01 3.01575392e-01 -9.97388721e-01
2.30464354e-01 8.84367406e-01 7.83585370e-01 2.39202663e-01
-1.00277770e+00 -8.47928345e-01 -2.25283861e-01 1.97830707e-01
-1.45424330e+00 -6.71758413e-01 7.59000659e-01 5.05269617e-02
7.72492707e-01 4.92176950e-01 5.02619028e-01 1.05002820e+00
2.65978932e-01 4.76862580e-01 1.59084690e+00 -7.02498436e-01
-8.76080096e-02 -3.13653350e-02 3.59948054e-02 5.80854177e-01
3.99686545e-01 3.51465315e-01 -6.20593607e-01 2.31842115e-03
6.71201468e-01 2.39035770e-01 -1.03026414e+00 1.77706003e-01
-1.13288057e+00 -8.94767642e-02 4.16451186e-01 -9.86210629e-02
-6.87725022e-02 2.08237991e-01 -3.75269145e-01 4.07446116e-01
2.91834503e-01 2.24285871e-01 -1.40160978e-01 9.72343385e-02
-8.90495718e-01 -1.87812984e-01 5.79925120e-01 1.10016525e+00
7.05355525e-01 9.40035880e-02 -1.65622160e-01 8.07445765e-01
1.40790060e-01 1.34602797e+00 -4.07071486e-02 -1.23776686e+00
6.76136613e-01 3.66961434e-02 5.83852768e-01 -9.59258378e-01
-2.35182181e-01 -2.13151127e-01 -7.64844358e-01 6.91906095e-01
3.98173660e-01 -1.47572666e-01 -1.01917160e+00 1.50231111e+00
8.75755548e-02 4.37516332e-01 5.28305247e-02 1.29856622e+00
7.78915703e-01 6.18418157e-01 -7.97374666e-01 -3.00578952e-01
1.13570595e+00 -9.64280844e-01 -8.19336653e-01 -3.42039317e-01
1.28718466e-01 -1.04608643e+00 1.40516901e+00 8.02513540e-01
-1.04296815e+00 -5.78080595e-01 -1.44888937e+00 -2.06658691e-01
2.07830369e-01 4.41576213e-01 1.24209575e-01 8.48307014e-01
-1.11795759e+00 1.59692451e-01 -5.08407712e-01 -2.78281718e-01
-1.10842437e-01 -2.26058923e-02 -3.04362506e-01 -6.88009501e-01
-9.49355125e-01 7.46774018e-01 -5.57087898e-01 2.93818831e-01
-4.25931990e-01 -7.53677905e-01 -5.15099466e-01 -1.61752999e-01
6.40468121e-01 -4.75610733e-01 1.01223314e+00 -1.08817053e+00
-1.50693130e+00 6.37631834e-01 -5.22924840e-01 4.79548164e-02
3.48539710e-01 -4.65302408e-01 -5.88123024e-01 4.20071602e-01
-2.41765812e-01 -9.11945105e-02 1.14816201e+00 -1.84916472e+00
-2.05598697e-01 -1.82744950e-01 6.98004151e-03 2.45468408e-01
-1.65018260e-01 -1.08287573e-01 -8.90062630e-01 -3.42897415e-01
9.69425738e-02 -9.79016602e-01 2.44336769e-01 1.83507949e-01
-5.91113389e-01 8.24721932e-01 9.65118408e-01 -7.58784473e-01
1.34339583e+00 -2.25414276e+00 -6.78582609e-01 -2.65855193e-02
1.77350745e-01 5.14836609e-01 -3.55383545e-01 5.11147857e-01
-2.48529762e-02 -2.13308811e-01 -2.48135880e-01 -5.30384839e-01
-4.45783466e-01 -1.35730386e-01 -4.69961792e-01 6.44824982e-01
-1.41026258e-01 4.28963751e-01 -7.67568886e-01 -9.29550901e-02
3.87342006e-01 7.81365216e-01 -8.41511637e-02 3.29574764e-01
4.49188352e-01 4.34237003e-01 -8.45256224e-02 8.00911069e-01
1.31386673e+00 -5.47996489e-03 1.01256259e-01 -6.29935145e-01
-4.04594123e-01 -2.23234490e-01 -1.14600658e+00 1.41308451e+00
-7.86111951e-01 1.02879167e+00 7.40722567e-02 1.61016844e-02
1.05290508e+00 -1.90928161e-01 1.45911291e-01 -1.35400188e+00
-1.11076169e-01 2.53179044e-01 -2.50926554e-01 -6.35759771e-01
7.16894567e-01 2.60057092e-01 1.87711269e-01 6.24472320e-01
-6.23653829e-01 -6.49419278e-02 -2.86856622e-01 2.36438811e-01
1.57557011e+00 1.69280231e-01 -6.76633045e-02 1.19017176e-01
2.91065395e-01 -6.01209641e-01 4.20832992e-01 9.70841706e-01
-8.91828910e-02 1.51622784e+00 -1.16439806e-02 3.85405272e-02
-8.93796086e-01 -1.30124307e+00 1.19294614e-01 6.69596970e-01
6.54669106e-01 -5.06177008e-01 -6.49218559e-01 -1.41449660e-01
-1.85162306e-01 7.50228465e-01 -4.14792478e-01 -1.42644733e-01
-5.29797137e-01 -4.34745133e-01 3.89588326e-01 2.66572386e-01
9.48436856e-01 -4.95096207e-01 -9.06109810e-01 -3.97182941e-01
-5.86661458e-01 -1.33734083e+00 -6.04475439e-01 -6.57509267e-02
-4.32994515e-01 -1.34764063e+00 -8.31927776e-01 -1.29173622e-01
7.20868349e-01 1.38518167e+00 1.09206283e+00 4.12570655e-01
-6.90726340e-01 8.56666207e-01 -3.02601844e-01 -3.46621156e-01
1.17376950e-02 -8.40182424e-01 -1.81732520e-01 2.10764199e-01
-3.30080539e-02 -4.82646823e-01 -1.35250616e+00 6.15849733e-01
-1.05288470e+00 2.33458281e-01 6.18319213e-01 4.90504473e-01
3.60987902e-01 -1.84377253e-01 -1.00207157e-01 -4.24788028e-01
3.47263098e-01 -6.73045292e-02 -5.78443408e-01 2.23364651e-01
-4.81767595e-01 -2.04414710e-01 5.79788387e-01 -3.00564051e-01
-1.51777029e+00 -2.79616490e-02 3.76527190e-01 -6.45834148e-01
7.14537874e-02 -3.27537239e-01 -2.75116622e-01 -1.95872098e-01
7.29408205e-01 9.67629254e-02 -1.43048123e-01 -4.26602513e-01
1.87455684e-01 9.39681888e-01 7.96998322e-01 -1.94751680e-01
8.44147861e-01 9.95199323e-01 2.42490605e-01 -1.32026231e+00
-7.70648718e-01 -6.17004395e-01 -3.15729022e-01 -4.11013007e-01
4.12665635e-01 -1.03014660e+00 -7.50400782e-01 8.58234167e-01
-9.83996630e-01 -8.36756170e-01 1.54038087e-01 3.45761031e-01
-3.84171098e-01 6.21056974e-01 -3.37116390e-01 -9.60836828e-01
-2.67092764e-01 -8.49146187e-01 1.32191086e+00 3.19398344e-01
1.92139804e-01 -5.62856972e-01 1.08518556e-01 3.41788113e-01
4.79503036e-01 -7.43345916e-02 3.39032799e-01 4.89728600e-01
-8.64779830e-01 -8.54945555e-02 -5.26708245e-01 6.59576893e-01
3.31937462e-01 -1.36456043e-01 -1.42462695e+00 -3.69552940e-01
5.48256040e-02 -2.02930108e-01 1.00216663e+00 2.92978078e-01
9.00044441e-01 4.87457775e-02 -1.25561386e-01 9.97211158e-01
1.83674765e+00 6.48701638e-02 1.45322752e+00 5.09037256e-01
7.87439466e-01 3.83470535e-01 1.01991522e+00 4.20459330e-01
1.22987680e-01 9.67643619e-01 5.06874919e-01 -6.09520912e-01
-8.04603934e-01 -1.81228071e-02 5.69464028e-01 3.49096149e-01
-2.05776766e-01 -6.51179373e-01 -5.11869729e-01 3.47718857e-02
-1.58913958e+00 -9.53338921e-01 -6.03140414e-01 2.75403905e+00
6.08778715e-01 -4.44515854e-01 -3.76346052e-01 1.14346214e-01
4.53054428e-01 2.88325042e-01 -4.84080166e-01 -9.12685767e-02
-6.01890564e-01 2.14253828e-01 9.64792490e-01 8.49732697e-01
-6.36814952e-01 5.57422638e-01 6.02457380e+00 3.40612739e-01
-1.27310550e+00 -3.87304761e-02 2.50818193e-01 -3.86626661e-01
-4.26511616e-01 -1.78013016e-02 -5.47449529e-01 6.93091810e-01
7.16178775e-01 2.69849569e-01 4.21208888e-01 1.08199967e-02
7.08850980e-01 -1.04213655e+00 -8.29944730e-01 1.33744526e+00
5.62462568e-01 -6.68023765e-01 -4.80043471e-01 1.11425705e-01
6.28761709e-01 -1.27482414e-01 4.39351022e-01 -4.12138313e-01
-6.74114525e-02 -8.65428686e-01 4.97261375e-01 9.84699249e-01
1.18768668e+00 -3.70180517e-01 3.79845977e-01 -1.39790982e-01
-9.50857997e-01 2.13001538e-02 -3.11607242e-01 1.45865381e-01
1.13258570e-01 7.24325478e-01 -4.54527736e-01 5.35529792e-01
1.02482200e+00 6.10510528e-01 -5.66685975e-01 1.01225162e+00
-4.23139095e-01 5.81904471e-01 -3.76809657e-01 5.03397107e-01
-3.32222253e-01 -4.75812584e-01 5.02698600e-01 1.17646646e+00
6.87183678e-01 2.84640461e-01 -1.69640362e-01 5.80310464e-01
2.74610847e-01 -5.26758432e-01 -5.63160539e-01 6.89551413e-01
5.30162573e-01 1.33932257e+00 -3.33070695e-01 3.56415659e-02
-6.58060193e-01 1.55535102e+00 -2.92012930e-01 1.03986156e+00
-8.54235888e-01 -6.30573988e-01 6.76429749e-01 3.75089914e-01
3.22339803e-01 -3.57240409e-01 -1.09806396e-02 -9.22666848e-01
4.90417331e-01 -7.92303503e-01 -1.07359633e-01 -1.67028964e+00
-8.79098892e-01 4.86634582e-01 -7.93694481e-02 -1.59289622e+00
2.41535768e-01 -7.45451748e-01 -5.74725151e-01 1.07627785e+00
-1.77421546e+00 -1.04493570e+00 -1.03960252e+00 9.15848434e-01
1.24082379e-01 3.81081462e-01 5.92145920e-01 3.64858359e-01
-3.71844381e-01 5.16773582e-01 5.97825348e-01 -1.55443132e-01
1.39391983e+00 -9.34098959e-01 1.10928141e-01 1.23842859e+00
-2.21247956e-01 6.48139477e-01 8.76902044e-01 -4.74318564e-01
-1.86264288e+00 -8.17074120e-01 1.76737383e-01 -3.23870271e-01
2.36577213e-01 -3.75530958e-01 -7.46261358e-01 4.69353050e-01
3.28670174e-01 1.39755592e-01 4.43721175e-01 -3.72360677e-01
-7.66246796e-01 -6.36491537e-01 -9.41384315e-01 8.24620247e-01
1.11378908e+00 -5.10145843e-01 -1.17548980e-01 1.41195850e-02
5.28346181e-01 -2.93829173e-01 -2.75879264e-01 6.77675158e-02
8.69817078e-01 -1.68172121e+00 1.25822043e+00 4.85020369e-01
2.90762901e-01 -5.79114616e-01 -4.38744426e-01 -1.19594431e+00
1.14473663e-01 -9.98476148e-01 3.42300013e-02 1.00036216e+00
1.48398988e-02 -9.60573852e-01 4.25933897e-01 5.36813676e-01
-2.34207332e-01 -3.44336629e-01 -3.49474907e-01 -8.51773500e-01
-5.50559759e-01 -5.33063233e-01 1.51323244e-01 4.34501022e-01
-4.99392569e-01 -1.50058633e-02 -8.48965406e-01 2.78128743e-01
1.07994795e+00 3.03623617e-01 1.22054684e+00 -5.56608915e-01
-4.81044292e-01 3.29701960e-01 -1.45278370e-03 -1.24130857e+00
-2.39190683e-01 -1.40305683e-01 3.76207232e-01 -1.59860444e+00
3.24621767e-01 -2.65320718e-01 -1.32317171e-01 3.51606041e-01
-3.74073505e-01 7.18942106e-01 4.25240368e-01 2.72076994e-01
-4.90297318e-01 3.18643719e-01 1.02481270e+00 1.75252691e-01
-3.22395533e-01 -7.50880465e-02 -8.49848986e-01 6.22835457e-01
6.77050829e-01 -1.61123618e-01 -4.96846050e-01 -5.82625747e-01
3.23442578e-01 -4.56536636e-02 6.64805174e-01 -1.04674101e+00
2.25746125e-01 1.47003174e-01 4.83950496e-01 -4.44840312e-01
7.98645973e-01 -9.07574117e-01 2.42002070e-01 8.88649523e-02
4.00970653e-02 -2.49259204e-01 1.88837424e-01 7.55916595e-01
1.06598839e-01 3.59144919e-02 6.17102504e-01 -4.32693912e-03
-6.10792041e-01 -2.73630098e-02 -1.67341426e-01 -4.55037355e-02
6.82085574e-01 -6.00710273e-01 -1.09437454e+00 -9.24000621e-01
-5.34162670e-02 -2.38452598e-01 8.58067930e-01 8.67870227e-02
9.66544271e-01 -1.02085316e+00 -4.34198022e-01 2.55046070e-01
1.37913451e-01 -4.32505041e-01 5.53865194e-01 1.12293720e+00
-4.61752832e-01 7.74532780e-02 -6.66583888e-03 -6.14051521e-01
-1.90994895e+00 4.20841545e-01 2.36926079e-01 1.91838890e-01
-7.11868644e-01 3.87515128e-01 4.10402179e-01 2.54850924e-01
1.34435996e-01 -3.06643069e-01 4.50740486e-01 -6.18619919e-01
9.48811829e-01 7.78784513e-01 1.24163523e-01 -5.25697947e-01
-2.16455862e-01 8.85437965e-01 3.66884589e-01 -3.49209726e-01
1.09758782e+00 -7.35625029e-01 2.09807083e-01 4.68462676e-01
1.41304338e+00 6.68857813e-01 -1.46417236e+00 -3.54427099e-01
-7.88101673e-01 -1.19203627e+00 1.79125294e-01 -9.86471176e-01
-1.02633536e+00 8.54053676e-01 7.52027810e-01 -2.33628482e-01
1.81042480e+00 -3.54682177e-01 6.26434386e-01 3.81528258e-01
3.70343775e-01 -7.57136643e-01 4.65207428e-01 2.40829736e-01
9.54009354e-01 -1.14340758e+00 3.30032647e-01 -6.03839219e-01
-5.33176959e-01 1.15355158e+00 4.83477384e-01 7.85920918e-02
3.41686964e-01 7.56442070e-01 5.16272068e-01 1.30452961e-01
-4.53598142e-01 -4.17635411e-01 1.88177004e-01 9.30894017e-01
1.54584825e-01 -4.19279695e-01 2.58391589e-01 2.15711176e-01
1.40253574e-01 -9.76200104e-02 8.12509298e-01 1.00152063e+00
-3.96121830e-01 -8.28911126e-01 -9.76938546e-01 -4.80719237e-03
-2.23003760e-01 -3.95303369e-01 -4.77672160e-01 7.82763064e-01
-2.46196374e-01 1.42165422e+00 2.83760671e-03 -4.70420897e-01
4.11534369e-01 -5.57970583e-01 8.60275984e-01 -1.02029905e-01
-1.65745273e-01 2.74397790e-01 3.68232876e-02 -1.16221142e+00
-5.93431771e-01 -4.13864702e-01 -9.71839070e-01 -2.12181568e-01
-2.19721898e-01 -4.18866694e-01 6.68130159e-01 2.67891407e-01
5.41522861e-01 3.41307759e-01 9.12213981e-01 -1.03362226e+00
-1.56724289e-01 -6.97254300e-01 -7.95504332e-01 4.36430812e-01
7.39397466e-01 -3.82522434e-01 -9.03103709e-01 4.82573360e-02] | [10.490581512451172, -2.7066938877105713] |
3c394843-17a7-4d0c-9881-7f920194b9c1 | variational-f-divergence-and-derangements-for | 2305.20025 | null | https://arxiv.org/abs/2305.20025v1 | https://arxiv.org/pdf/2305.20025v1.pdf | Variational $f$-Divergence and Derangements for Discriminative Mutual Information Estimation | The accurate estimation of the mutual information is a crucial task in various applications, including machine learning, communications, and biology, since it enables the understanding of complex systems. High-dimensional data render the task extremely challenging due to the amount of data to be processed and the presence of convoluted patterns. Neural estimators based on variational lower bounds of the mutual information have gained attention in recent years but they are prone to either high bias or high variance as a consequence of the partition function. We propose a novel class of discriminative mutual information estimators based on the variational representation of the $f$-divergence. We investigate the impact of the permutation function used to obtain the marginal training samples and present a novel architectural solution based on derangements. The proposed estimator is flexible as it exhibits an excellent bias/variance trade-off. Experiments on reference scenarios demonstrate that our approach outperforms state-of-the-art neural estimators both in terms of accuracy and complexity. | ['Andrea M. Tonello', 'Nicola Novello', 'Nunzio A. Letizia'] | 2023-05-31 | null | null | null | null | ['mutual-information-estimation'] | ['methodology'] | [ 2.39427909e-01 -2.74664372e-01 1.21388108e-01 -5.45778573e-01
-6.01452410e-01 -2.75727242e-01 6.97316527e-01 2.48988971e-01
-7.02599943e-01 9.17017579e-01 -1.98448658e-01 -2.84668915e-02
-5.47987580e-01 -6.72582090e-01 -7.78309882e-01 -9.45981622e-01
-2.01041520e-01 2.87128985e-01 3.89225155e-01 1.76891372e-01
3.21983129e-01 3.51679832e-01 -1.75943816e+00 -4.33540791e-01
1.17412233e+00 1.20463741e+00 3.97917747e-01 3.11939776e-01
-3.16375941e-02 3.90556119e-02 -5.15360892e-01 -2.50600338e-01
1.56494066e-01 -5.07691801e-01 -1.11902922e-01 -3.48007798e-01
8.34990293e-02 1.74284905e-01 1.75899081e-02 1.21711957e+00
5.78901947e-01 2.78317213e-01 1.18014431e+00 -9.97193813e-01
-1.07899338e-01 3.39579761e-01 -4.51468170e-01 1.79561853e-01
-1.63564354e-01 -1.98070168e-01 1.05248952e+00 -6.77845836e-01
3.46787035e-01 8.72248590e-01 6.35068536e-01 2.46306092e-01
-1.64899707e+00 -7.37121403e-01 -1.09503113e-01 2.63935924e-01
-1.59561098e+00 -3.69863659e-01 7.83721387e-01 -4.63152319e-01
5.01350403e-01 1.48339301e-01 2.65198648e-01 9.01996195e-01
3.45833153e-01 5.98399103e-01 9.87820327e-01 -2.92192817e-01
5.57164550e-01 3.29730392e-01 -2.75371242e-02 5.73669314e-01
4.06463951e-01 -2.57893559e-02 -3.63466293e-01 -6.12977073e-02
7.11797774e-01 -2.49968156e-01 -2.62955189e-01 -8.12622964e-01
-7.41561353e-01 8.97165537e-01 3.17377925e-01 5.10204852e-01
-1.89244837e-01 2.71026611e-01 1.20683782e-01 1.24948412e-01
5.37344098e-01 3.38311642e-01 -3.42060238e-01 -2.61926115e-01
-8.93113792e-01 2.84297228e-01 9.16820645e-01 5.36010861e-01
6.65628493e-01 -1.17694847e-01 3.03173065e-02 1.09995997e+00
4.59790289e-01 4.12584722e-01 2.07058534e-01 -5.94483078e-01
3.39489639e-01 4.48149353e-01 6.94911182e-02 -1.14967799e+00
-3.94647151e-01 -8.30040753e-01 -1.32350242e+00 3.00696343e-02
6.07607305e-01 -1.30250707e-01 -4.61834103e-01 2.03980041e+00
3.12097281e-01 -8.03164318e-02 -2.26324379e-01 8.12133253e-01
3.38868260e-01 4.03365761e-01 -1.65337086e-01 -3.16831619e-01
9.15817916e-01 -1.91939458e-01 -6.63527846e-01 -1.48620442e-01
3.38537872e-01 -5.21425843e-01 4.99986857e-01 2.60695457e-01
-8.12512934e-01 -3.81196260e-01 -1.09798074e+00 3.73993963e-01
-2.63051391e-01 8.61329511e-02 2.36389920e-01 7.86271214e-01
-7.21000433e-01 9.27392423e-01 -8.61652732e-01 -1.44619733e-01
3.08412075e-01 4.85021651e-01 -3.20768505e-01 2.73000956e-01
-1.00094080e+00 7.34527409e-01 3.34084511e-01 2.75357008e-01
-1.94426849e-01 -4.84680891e-01 -6.78170800e-01 2.67625302e-01
1.07234955e-01 -3.91507149e-01 6.78003311e-01 -7.81718552e-01
-1.50570416e+00 3.50977540e-01 -5.84977493e-02 -4.72181827e-01
7.78832436e-01 -1.77901104e-01 3.00818887e-02 -1.70622930e-01
-1.44902989e-01 2.11106494e-01 7.48570025e-01 -7.90995359e-01
-1.83781311e-01 -4.57379103e-01 -4.62747753e-01 -9.70617309e-02
-2.12600663e-01 -4.92249548e-01 -1.66712090e-01 -5.55782855e-01
3.37774575e-01 -8.55201244e-01 -1.41841412e-01 1.61700379e-02
-3.01291913e-01 7.46033294e-03 4.21392888e-01 -3.13597679e-01
1.17816317e+00 -2.08221269e+00 1.92484081e-01 4.63760227e-01
5.04953507e-03 1.60597935e-01 1.38267130e-01 3.92752558e-01
1.14267446e-01 -6.82070777e-02 -5.01001954e-01 -1.06823035e-01
1.59138247e-01 -8.08303803e-02 4.47357111e-02 6.42010987e-01
2.25566387e-01 2.35794917e-01 -5.90625644e-01 -2.97778189e-01
2.33297929e-01 7.46883690e-01 -6.56155229e-01 3.06387842e-01
-5.19265942e-02 4.24061835e-01 -3.46084744e-01 -1.56203926e-01
6.27855420e-01 -2.46969670e-01 4.05856878e-01 -3.32901418e-01
-1.34462684e-01 7.94697553e-02 -1.39189339e+00 1.38347590e+00
-4.66669172e-01 7.34903753e-01 -3.17526124e-02 -1.29082823e+00
1.07504058e+00 5.78271113e-02 3.86112541e-01 -5.34750521e-01
3.59117478e-01 3.15970421e-01 3.34466577e-01 -1.56668708e-01
2.45894238e-01 -1.61036476e-01 -5.45026995e-02 2.90361762e-01
1.26440734e-01 2.45290026e-02 2.03356430e-01 -2.07634598e-01
8.86907339e-01 9.91992950e-02 4.16888028e-01 -7.47079849e-01
5.04859269e-01 -9.32362616e-01 6.13333821e-01 7.53462970e-01
6.73923502e-03 4.27967757e-01 6.65047109e-01 3.92896608e-02
-9.10507381e-01 -1.25869989e+00 -5.53173363e-01 4.08046991e-01
1.03324823e-01 1.23345807e-01 -8.33627641e-01 -2.29474545e-01
7.71349594e-02 5.46515226e-01 -5.16100824e-01 -2.89807469e-01
-7.48394430e-02 -9.14351463e-01 2.60522306e-01 2.42996380e-01
5.46382010e-01 -4.80826885e-01 -8.21409941e-01 2.14925870e-01
2.53281347e-03 -9.76908743e-01 -2.74696648e-01 3.92079324e-01
-8.85425746e-01 -8.86445284e-01 -7.70699739e-01 -2.75241584e-01
4.94238198e-01 -1.62195891e-01 8.40473354e-01 -4.27227229e-01
-3.08139771e-01 1.99069291e-01 3.94774154e-02 -2.20099375e-01
-3.19196880e-01 2.28099942e-01 2.48570353e-01 3.08867157e-01
1.77909657e-01 -8.92327070e-01 -6.48330092e-01 5.66129804e-01
-8.38880837e-01 -2.26210147e-01 7.22488403e-01 9.38253284e-01
4.92806941e-01 9.22832564e-02 7.38832116e-01 -6.74782455e-01
5.91524065e-01 -4.61489469e-01 -1.02068830e+00 7.41135469e-03
-6.76157713e-01 7.68961191e-01 5.27632236e-01 -2.65965641e-01
-1.05591106e+00 -1.71373606e-01 -6.04903884e-03 3.54644209e-02
-7.40761533e-02 3.65014821e-01 -2.12538332e-01 -4.17978279e-02
3.35469276e-01 9.32762846e-02 1.28399685e-01 -5.03447950e-01
2.45639998e-02 6.67204738e-01 1.67250484e-01 -4.34695363e-01
3.53946477e-01 1.67766646e-01 5.37238061e-01 -1.21940267e+00
-6.38851464e-01 -4.82635528e-01 -5.05211473e-01 -2.97992408e-01
7.05482066e-01 -4.41734523e-01 -9.32128012e-01 5.26312768e-01
-9.95636642e-01 -4.15813625e-02 -6.80037541e-03 7.17090964e-01
-5.71740508e-01 3.50887746e-01 -2.41570607e-01 -1.10762537e+00
-2.92576551e-01 -1.28321981e+00 6.66870356e-01 2.60989606e-01
-3.70591767e-02 -1.07787192e+00 1.50546983e-01 -1.65161565e-01
5.61934471e-01 3.02555144e-01 1.08058584e+00 -6.33276880e-01
-4.55758899e-01 -8.66108909e-02 -3.22702259e-01 3.95672768e-01
1.82078660e-01 -7.50389472e-02 -8.08712602e-01 -1.30731076e-01
-6.66677132e-02 2.57449597e-02 9.58528280e-01 7.16349304e-01
9.77759600e-01 -9.71433148e-02 -2.73936629e-01 3.23747247e-01
1.37792516e+00 2.12055761e-02 3.32754225e-01 -1.71314344e-01
2.87799299e-01 6.77863538e-01 2.19216704e-01 6.58154607e-01
8.74606520e-02 7.55767822e-01 4.65122342e-01 5.53470969e-01
4.31109697e-01 7.61634037e-02 4.28831950e-02 9.86303091e-01
1.08053990e-01 -1.98203817e-01 -6.35872781e-01 4.85120177e-01
-1.82088208e+00 -9.99708295e-01 -1.32268101e-01 2.74733186e+00
7.91033149e-01 2.57422626e-01 -9.51344669e-02 3.06079835e-01
6.79819584e-01 1.13311134e-01 -5.59425175e-01 -2.15130702e-01
1.24507345e-01 1.50504827e-01 4.09646779e-01 4.69921708e-01
-9.75062430e-01 2.10079879e-01 5.95671558e+00 9.07242298e-01
-9.69674528e-01 -9.26171318e-02 4.41653609e-01 4.90908884e-02
-3.54988240e-02 -3.79572183e-01 -9.13625479e-01 7.39551425e-01
9.79821801e-01 1.01689026e-01 9.88138169e-02 4.70530272e-01
-6.59668893e-02 -5.75458586e-01 -1.02414811e+00 9.30585504e-01
-1.22019216e-01 -8.42909157e-01 -3.47906202e-01 1.99119091e-01
5.60386837e-01 1.10596515e-01 1.74828917e-01 -7.05526471e-02
-1.48060635e-01 -7.62319624e-01 4.52876806e-01 8.97332907e-01
3.71713251e-01 -9.38237429e-01 8.26965153e-01 6.02589428e-01
-9.36876774e-01 1.27281010e-01 -2.51427889e-01 1.97443981e-02
-7.23235011e-02 1.17154074e+00 -6.89247429e-01 1.19991206e-01
4.02886897e-01 3.25538188e-01 -1.84547663e-01 1.17401755e+00
2.75818817e-02 4.16021407e-01 -8.06084156e-01 -5.64293981e-01
-1.24923168e-02 -5.96022785e-01 6.74886346e-01 8.88369322e-01
6.07923925e-01 -3.70257527e-01 -2.95373172e-01 1.12676394e+00
-5.48202507e-02 1.07031107e-01 -4.57804978e-01 1.11315977e-02
4.88082975e-01 9.53379393e-01 -6.63917542e-01 1.72709823e-01
-8.16508308e-02 7.88075507e-01 3.38027567e-01 1.57417446e-01
-5.68820775e-01 -4.55619216e-01 7.61918247e-01 3.06888949e-02
7.06907094e-01 -3.98201376e-01 2.07390562e-02 -9.28315997e-01
1.83984622e-01 -2.96845019e-01 -5.40890433e-02 2.93230657e-02
-1.24193728e+00 4.39253449e-01 2.83662193e-02 -1.06321728e+00
-5.84924936e-01 -6.02278590e-01 -1.95862934e-01 6.23887241e-01
-1.17087245e+00 -2.75208890e-01 -1.56391859e-01 2.44997218e-02
1.05487786e-01 -2.91417390e-02 7.55560756e-01 4.43388939e-01
-5.67991316e-01 7.04101324e-01 9.16386724e-01 -1.34001806e-01
5.79645574e-01 -1.12220359e+00 6.11244515e-03 5.03723681e-01
1.62774950e-01 5.00574708e-01 1.15947390e+00 -1.72737584e-01
-1.14241135e+00 -6.29037321e-01 5.01537859e-01 -7.77399242e-02
4.74050969e-01 -4.62634802e-01 -8.08577120e-01 6.69000670e-02
-2.24622145e-01 -4.98589277e-02 8.31530631e-01 2.73266852e-01
-2.89831579e-01 -4.21400219e-01 -1.04836810e+00 4.59180981e-01
7.32576013e-01 -3.15748125e-01 -1.82029396e-01 -2.26443321e-01
6.98231310e-02 3.01571954e-02 -8.71528089e-01 4.33109403e-01
1.04892039e+00 -1.42057586e+00 5.68943679e-01 9.69823152e-02
1.21764705e-01 -1.20011680e-01 -3.39650810e-01 -1.36578107e+00
-9.12678149e-03 -3.25184315e-01 8.87139440e-02 1.15844643e+00
4.30201590e-01 -8.47914219e-01 6.12021804e-01 5.43106437e-01
3.83774459e-01 -9.13636148e-01 -1.28292835e+00 -8.17567229e-01
-1.80096582e-01 -3.86912137e-01 4.99649905e-02 5.37995696e-01
-1.51506498e-01 4.03385937e-01 -2.69219160e-01 -7.87347630e-02
9.81336713e-01 1.05868638e-01 6.47533119e-01 -1.61382926e+00
-6.72339261e-01 -6.50476933e-01 -7.64623404e-01 -1.09652400e+00
9.00624618e-02 -5.50103724e-01 3.25966328e-01 -1.00386870e+00
1.32168263e-01 -4.13990289e-01 -3.98031145e-01 -2.32196763e-01
-2.00915113e-02 1.68601751e-01 -2.79736519e-02 -4.79740277e-02
-3.81584734e-01 8.58189702e-01 5.76986432e-01 1.26456231e-01
-2.38975689e-01 1.61103562e-01 -1.79034546e-01 7.83228636e-01
6.75541759e-01 -5.37648439e-01 -1.72899768e-01 -2.14817122e-01
3.39595944e-01 -1.10088671e-02 1.77668527e-01 -1.30672348e+00
1.82694122e-01 2.99003989e-01 2.88660556e-01 -4.00208235e-01
5.64077735e-01 -8.78200352e-01 2.36402854e-01 3.94484073e-01
-2.80313164e-01 -2.45123729e-01 6.60080537e-02 8.45657647e-01
-2.73735076e-01 -3.36449206e-01 9.76711929e-01 2.89261281e-01
-1.84233412e-01 8.00924189e-03 -4.10372049e-01 3.24213691e-02
8.08010519e-01 -2.60897935e-03 1.44037738e-01 -5.28354704e-01
-2.76430428e-01 -5.12948781e-02 2.24767908e-01 1.46249995e-01
3.72735769e-01 -1.07171535e+00 -4.31551427e-01 3.03484648e-01
-1.25336528e-01 -3.07950884e-01 2.76074022e-01 1.06057680e+00
-1.64449334e-01 4.09652144e-01 -1.89857677e-01 -7.40527034e-01
-1.05899870e+00 -4.49729115e-02 3.31131697e-01 -4.42591548e-01
-8.28673095e-02 6.26277089e-01 3.28101844e-01 -3.28980029e-01
2.31686980e-01 -2.00814411e-01 -9.50816087e-03 2.07397655e-01
3.21645200e-01 4.63524252e-01 -1.06706368e-02 -5.16988754e-01
-3.57978165e-01 6.18708193e-01 1.93345193e-02 -1.54195637e-01
1.21911407e+00 -3.99141051e-02 3.39954719e-02 7.34142125e-01
1.40450418e+00 -3.17419648e-01 -1.29602945e+00 -3.48790139e-01
1.06842630e-01 -2.76849478e-01 1.87490642e-01 -3.21053445e-01
-7.69643068e-01 1.02542889e+00 7.79971242e-01 2.70366937e-01
7.83643961e-01 -2.16951504e-01 4.20223922e-01 4.66510922e-01
3.92170876e-01 -1.18356955e+00 -1.69501096e-01 4.58605587e-01
5.27720988e-01 -1.09084773e+00 1.14063494e-01 -1.93881631e-01
-1.46650756e-02 9.15831864e-01 6.82017654e-02 -2.62052208e-01
1.03318477e+00 1.48244262e-01 -2.97509432e-01 2.79458851e-01
-5.57685554e-01 -1.25320449e-01 4.74034667e-01 3.27731818e-01
6.57145858e-01 4.78422269e-02 -7.19809711e-01 3.91039670e-01
-1.73887648e-02 -3.84468168e-01 6.43310696e-02 4.68886197e-01
-4.37390745e-01 -1.06140196e+00 4.05894071e-02 5.85633576e-01
-6.46859467e-01 5.46582714e-02 -6.80872202e-02 6.58487976e-01
-2.36551806e-01 7.55134165e-01 2.22174197e-01 -1.00427143e-01
5.56345806e-02 6.53350726e-03 6.23038471e-01 -8.15366358e-02
-1.79678619e-01 -7.47750029e-02 -1.23980992e-01 -4.61324841e-01
-5.31692505e-01 -7.34965265e-01 -8.18946660e-01 -3.46723422e-02
-6.80693269e-01 4.14166272e-01 1.02904642e+00 1.17104316e+00
5.11646271e-01 3.96222234e-01 5.41250229e-01 -8.61475408e-01
-8.25649858e-01 -1.06936228e+00 -7.57064879e-01 2.16635272e-01
2.73144215e-01 -1.00110590e+00 -5.11618614e-01 -4.19950873e-01] | [7.280550479888916, 3.961782693862915] |
6588dabd-e8e3-4183-b50e-29838312c6f0 | natcat-weakly-supervised-text-classification | 2009.14335 | null | https://arxiv.org/abs/2009.14335v2 | https://arxiv.org/pdf/2009.14335v2.pdf | NatCat: Weakly Supervised Text Classification with Naturally Annotated Resources | We describe NatCat, a large-scale resource for text classification constructed from three data sources: Wikipedia, Stack Exchange, and Reddit. NatCat consists of document-category pairs derived from manual curation that occurs naturally within online communities. To demonstrate its usefulness, we build general purpose text classifiers by training on NatCat and evaluate them on a suite of 11 text classification tasks (CatEval), reporting large improvements compared to prior work. We benchmark different modeling choices and resource combinations and show how tasks benefit from particular NatCat data sources. | ['Karl Stratos', 'Zewei Chu', 'Kevin Gimpel'] | 2020-09-29 | null | https://openreview.net/forum?id=kmVA04ltlG_ | https://openreview.net/pdf?id=kmVA04ltlG_ | akbc-2021-10 | ['text-categorization'] | ['natural-language-processing'] | [-3.66016001e-01 -3.20101887e-01 -1.80107117e-01 -1.87340498e-01
-7.09243476e-01 -9.02173340e-01 1.27338624e+00 6.94803894e-01
-6.80187345e-01 6.48652136e-01 5.65598488e-01 -3.98503393e-01
-1.31599903e-01 -6.56952441e-01 -9.31932107e-02 4.94210832e-02
-1.26886874e-01 4.81518984e-01 4.46242094e-01 -2.61774033e-01
6.32681966e-01 -3.13878328e-01 -1.32613397e+00 9.75446999e-01
9.96805489e-01 7.25315094e-01 -6.07629344e-02 8.22195590e-01
-7.06353068e-01 1.00726378e+00 -7.62009084e-01 -7.79957891e-01
1.90266401e-01 3.46087996e-04 -1.33437002e+00 -4.42924947e-01
9.34905708e-01 3.19724046e-02 -2.46620268e-01 8.94695580e-01
2.03926340e-01 -3.03416938e-01 8.01721394e-01 -1.42898810e+00
-7.15606689e-01 1.42739594e+00 -2.44917899e-01 3.30141753e-01
3.42286259e-01 -8.03169683e-02 1.54060829e+00 -8.20339978e-01
9.80807245e-01 1.03840959e+00 1.22182775e+00 3.55238825e-01
-1.30718195e+00 -5.86178780e-01 -2.40411177e-01 -1.97839975e-01
-1.06357288e+00 -4.27570969e-01 2.31267527e-01 -9.82208252e-01
1.22355330e+00 4.50213283e-01 3.86658996e-01 1.13228238e+00
9.85837504e-02 8.51586044e-01 1.09626508e+00 -6.41152561e-01
-3.33181024e-03 2.74004132e-01 9.42892075e-01 7.67132878e-01
5.65588057e-01 -7.81456292e-01 -6.59646451e-01 -7.73755193e-01
-7.75446892e-02 -1.96740866e-01 -3.27843666e-01 -1.67961270e-01
-1.41048431e+00 6.42400801e-01 1.74640432e-01 4.78387654e-01
-2.34650038e-02 2.34106094e-01 8.01772118e-01 7.31173813e-01
9.50159371e-01 6.61125481e-01 -8.65040243e-01 -3.77289951e-01
-7.43476450e-01 4.72153664e-01 1.43425298e+00 1.07033193e+00
6.65037274e-01 -4.39621180e-01 -1.74581960e-01 1.27029419e+00
1.40205368e-01 2.62651652e-01 8.64568830e-01 -8.17183793e-01
1.13101065e+00 9.19837534e-01 1.28835559e-01 -7.05966949e-01
-5.62681854e-01 -4.29516822e-01 -3.01186830e-01 -1.52354077e-01
5.66286206e-01 -3.14587176e-01 -6.32408142e-01 9.64001238e-01
-6.70003295e-02 -4.59382951e-01 -1.07315861e-01 4.11495380e-02
1.14010024e+00 9.93215516e-02 2.52809256e-01 3.16103607e-01
1.14856839e+00 -1.26175034e+00 -6.86954081e-01 -1.42319351e-01
1.75512218e+00 -6.94698870e-01 9.16390717e-01 1.59883261e-01
-5.43657422e-01 5.21682985e-02 -6.59805775e-01 -1.82817459e-01
-1.05064976e+00 8.06048587e-02 5.73267639e-01 6.69404685e-01
-1.12800217e+00 6.41823232e-01 -6.59771144e-01 -8.52230489e-01
4.65997756e-01 -2.10884884e-01 -4.45452273e-01 2.69120902e-01
-1.05202878e+00 6.72322452e-01 4.05586660e-01 -5.11639237e-01
-3.03000599e-01 -8.76851380e-01 -4.02738839e-01 2.36247648e-02
1.14750467e-01 -5.63780904e-01 1.38967252e+00 -7.00049400e-01
-8.44646335e-01 1.16720188e+00 2.63740748e-01 -5.43247223e-01
6.69093907e-01 -2.38388151e-01 -4.50997114e-01 -9.24808979e-02
2.62472153e-01 3.45236808e-01 4.65719491e-01 -8.57372701e-01
-8.94870162e-01 -2.39465639e-01 3.45427096e-02 -8.73762593e-02
-1.13725591e+00 5.66850722e-01 -2.25438967e-01 -7.77230442e-01
-1.47871122e-01 -8.97247314e-01 1.65888384e-01 5.53647801e-02
-8.54212105e-01 -5.35801291e-01 7.98402905e-01 -7.30844676e-01
1.30663574e+00 -1.56921983e+00 -1.25080749e-01 -1.83070868e-01
7.74816930e-01 1.32287398e-01 -1.93221524e-01 8.69508624e-01
3.22244726e-02 8.91317964e-01 9.70859677e-02 -5.53087831e-01
2.58202463e-01 -3.30996752e-01 -2.08654851e-01 6.95665181e-02
-4.11278874e-01 7.28050053e-01 -7.75232852e-01 -7.74916768e-01
-3.54728907e-01 -1.46014825e-01 -6.36785805e-01 -1.75655708e-01
-5.49708009e-01 -4.44463551e-01 -6.38113678e-01 4.06238168e-01
2.54965186e-01 -6.39673531e-01 2.78000176e-01 -9.29018203e-03
-1.63056582e-01 8.69708061e-01 -4.60240304e-01 1.61373115e+00
-1.88093528e-01 1.01176131e+00 7.44682178e-02 -5.80100358e-01
6.35590851e-01 2.45676309e-01 4.69411522e-01 -2.41323084e-01
1.85993761e-01 2.37289026e-01 -3.25605124e-01 -5.37530124e-01
8.85477901e-01 6.95487022e-01 -1.68969065e-01 1.05169392e+00
1.88253805e-01 1.04554161e-01 5.23291707e-01 7.73425221e-01
1.49802768e+00 -2.27047369e-01 1.92436785e-01 -7.76945949e-01
2.33961895e-01 5.23042321e-01 -3.06749046e-02 1.01036239e+00
-1.24658778e-01 1.92734867e-01 6.17619157e-01 -5.96824110e-01
-1.02091503e+00 -4.34697568e-01 -3.70948315e-01 1.74003470e+00
-4.63211507e-01 -1.18569827e+00 -7.05065429e-01 -1.33600175e+00
5.16966045e-01 5.60554087e-01 -8.66598308e-01 4.14855063e-01
-2.79444784e-01 -8.16277802e-01 8.16596031e-01 3.65246147e-01
4.62430656e-01 -5.94095290e-01 2.00934514e-01 -3.01359296e-02
-3.61414254e-01 -1.04926336e+00 -6.07288063e-01 1.38492689e-01
-8.47586215e-01 -1.36531448e+00 -4.49247271e-01 -6.55726433e-01
1.55848950e-01 6.14404678e-01 1.67375112e+00 5.74995637e-01
-1.74859032e-01 5.73839068e-01 -5.84022939e-01 -2.68594116e-01
-7.16479957e-01 9.77101684e-01 -7.52187446e-02 -5.68359673e-01
6.85164332e-01 -3.38121831e-01 -8.68513715e-03 3.25219572e-01
-6.10147357e-01 1.80240199e-01 -1.01360969e-01 7.06821144e-01
-3.26862037e-01 -1.48786515e-01 6.01129830e-01 -1.34249628e+00
8.02604318e-01 -1.01103473e+00 -4.72951144e-01 3.74325544e-01
-8.23612750e-01 -5.27982153e-02 4.83864695e-01 7.13940803e-03
-8.92611086e-01 -4.23481137e-01 2.82523334e-01 8.82529989e-02
6.08617254e-02 9.29874420e-01 5.15624166e-01 5.69448061e-02
1.27063310e+00 -4.97197472e-02 -2.36985072e-01 -9.73789096e-01
2.41986200e-01 1.35370231e+00 1.84340328e-02 -7.70886958e-01
9.15752828e-01 1.01368561e-01 -6.58306301e-01 -5.64078033e-01
-9.08579767e-01 -4.95728284e-01 -8.57583582e-01 1.54230490e-01
5.17140210e-01 -1.02651393e+00 -4.00900304e-01 5.44734061e-01
-9.73140538e-01 -5.00292659e-01 1.36499822e-01 2.11737186e-01
9.02509093e-02 2.69791305e-01 -8.92326355e-01 -5.09650052e-01
-8.11900198e-01 -3.52245808e-01 7.12101519e-01 -2.27986678e-01
-2.94139594e-01 -1.50789714e+00 1.87828705e-01 5.18990934e-01
5.96349835e-01 -2.23276481e-01 9.47467983e-01 -1.44557893e+00
-3.48897606e-01 -2.75333464e-01 -4.69411224e-01 -6.72206953e-02
2.19840363e-01 4.39902157e-01 -9.05155182e-01 -3.20367783e-01
-7.14228630e-01 -6.04981542e-01 1.13694155e+00 -2.75574356e-01
1.14356649e+00 -4.80442852e-01 -7.63762355e-01 2.93002576e-01
1.25568879e+00 -4.09433514e-01 -9.69373435e-02 8.83534729e-01
9.89245415e-01 7.07290351e-01 7.55190756e-03 3.68842334e-01
7.27334559e-01 5.53117335e-01 7.76287690e-02 3.72544676e-01
-2.25794986e-01 -1.84136987e-01 -1.40713761e-02 1.18537903e+00
2.69582182e-01 -5.40889919e-01 -1.38748288e+00 6.68060720e-01
-1.90312111e+00 -8.18622172e-01 -6.69192314e-01 1.72281778e+00
1.25667191e+00 2.21294209e-01 2.47174338e-01 -1.13121018e-01
7.75325835e-01 3.63657512e-02 -2.25189209e-01 6.47015050e-02
-1.91460237e-01 -1.77890912e-01 5.84759176e-01 1.23263106e-01
-1.20549989e+00 6.63168132e-01 7.93789673e+00 5.57415903e-01
-8.51298153e-01 2.53940374e-01 2.37717181e-01 -2.22476631e-01
-2.46609837e-01 -1.17996439e-01 -1.01610351e+00 5.96104205e-01
1.09205794e+00 -6.97624266e-01 1.43701226e-01 9.08718526e-01
-3.83873284e-01 2.25073338e-01 -1.33909059e+00 5.47622800e-01
-6.12423234e-02 -1.60229957e+00 -1.35342376e-02 6.91970587e-02
8.45617890e-01 8.59427631e-01 -2.94412762e-01 4.88005102e-01
1.24599016e+00 -5.64233899e-01 7.28017509e-01 2.09058478e-01
5.01935124e-01 6.37178868e-02 5.16935825e-01 2.70854205e-01
-9.68601525e-01 -2.24655837e-01 -2.77223494e-02 1.55440524e-01
-5.65521359e-01 6.58132136e-01 -7.32945800e-01 4.14401054e-01
1.05965436e+00 1.10400116e+00 -1.30052412e+00 1.12338877e+00
2.53844023e-01 9.93229866e-01 -3.66465718e-01 -2.42526874e-01
-3.39735933e-02 3.39435637e-01 6.04210198e-01 1.82189453e+00
9.90417674e-02 -4.18809265e-01 1.73563346e-01 7.75984466e-01
-6.98310018e-01 2.76880860e-01 -4.66183513e-01 -2.33929634e-01
9.56174910e-01 1.60739267e+00 -5.90414047e-01 -5.27067482e-01
-6.70576751e-01 5.88823676e-01 7.05863297e-01 2.46380046e-01
-2.24093273e-01 -8.32324564e-01 7.40683675e-01 5.97239919e-02
5.69538958e-02 -2.88040578e-01 -2.32466534e-01 -1.66782534e+00
9.27075297e-02 -7.27248192e-01 8.58912051e-01 -8.68371964e-01
-1.81765366e+00 6.71760976e-01 8.42344612e-02 -1.08588815e+00
-4.44815308e-01 -6.06808782e-01 -5.41842461e-01 8.16577852e-01
-1.33818567e+00 -1.26396275e+00 -5.88280022e-01 4.46928650e-01
4.61785734e-01 -4.78072792e-01 8.71152341e-01 3.62889022e-01
-7.69435287e-01 7.43259847e-01 6.63725317e-01 8.28921914e-01
1.04226816e+00 -1.70813191e+00 1.04773319e+00 3.52565914e-01
-8.80810097e-02 7.12430120e-01 5.45570076e-01 -8.35448563e-01
-1.28760803e+00 -1.32515311e+00 1.01301515e+00 -1.20178437e+00
1.43236256e+00 -6.17251277e-01 -1.12581098e+00 1.22013259e+00
5.09439230e-01 9.89060923e-02 6.68376386e-01 6.20261848e-01
-9.20861661e-01 1.71635866e-01 -9.10088599e-01 5.86974800e-01
1.26895630e+00 -7.32842743e-01 -7.50193238e-01 6.72696412e-01
7.66817749e-01 -5.85914552e-02 -1.17428374e+00 -4.17176962e-01
5.81134379e-01 -6.87313914e-01 6.06942117e-01 -9.58289027e-01
6.50853932e-01 2.72270024e-01 -1.74365919e-02 -1.61569679e+00
-2.67813832e-01 -5.65499246e-01 -9.14501920e-02 1.62349546e+00
6.91984534e-01 -9.73321795e-01 8.47156644e-01 8.25710833e-01
-9.16138440e-02 1.53384104e-01 -6.94153130e-01 -8.95927370e-01
6.87358558e-01 -1.55382842e-01 5.44118583e-01 1.72840428e+00
7.29544520e-01 5.93190670e-01 7.69542456e-02 -5.43946862e-01
7.69860625e-01 -9.44065899e-02 8.79527926e-01 -1.89868271e+00
1.13862835e-01 -7.81651795e-01 -6.02557249e-02 -7.70339012e-01
2.06360921e-01 -1.31882763e+00 -2.69121647e-01 -1.61686516e+00
4.43326533e-01 -8.76918495e-01 -2.37665419e-02 1.02831721e+00
-3.32200378e-01 1.54602125e-01 1.09602772e-01 9.72622991e-01
-8.70827198e-01 2.63429612e-01 6.08853042e-01 -4.51190919e-01
7.69375935e-02 -5.38583159e-01 -7.56695628e-01 5.79946041e-01
6.24556005e-01 -7.38321304e-01 5.57804033e-02 -7.97202229e-01
5.27536273e-01 -1.40983000e-01 -4.76585887e-02 -9.68863189e-01
4.19401914e-01 -6.73657879e-02 1.01142943e-01 -5.03693044e-01
-1.82520136e-01 -5.34481823e-01 -2.36380026e-01 1.84971243e-01
-9.77915406e-01 -6.60554320e-02 3.85207795e-02 3.88440609e-01
6.14979938e-02 -5.26395977e-01 3.93781662e-01 -6.15796633e-02
-2.55146295e-01 5.26907369e-02 -6.53476238e-01 4.20606673e-01
4.18366104e-01 2.52288252e-01 -1.31370568e+00 -6.80469126e-02
-3.67615223e-01 3.40149075e-01 5.24097323e-01 7.44308650e-01
-2.96533525e-01 -1.24494553e+00 -1.04368389e+00 -1.90202981e-01
5.86748421e-01 -6.62819862e-01 -3.49910766e-01 6.13683581e-01
-4.57599789e-01 7.19011724e-01 -1.34902239e-01 -5.13601184e-01
-1.30754638e+00 1.91260725e-01 5.62562644e-01 -3.37729961e-01
-6.74614012e-01 4.45894569e-01 -3.27392578e-01 -1.01378620e+00
6.43154755e-02 -3.38151753e-01 -5.18860221e-01 1.06435761e-01
6.89680934e-01 4.84153807e-01 6.04934156e-01 -5.79830930e-02
-1.77289113e-01 8.96452293e-02 -4.81814355e-01 -4.91241813e-02
1.19200230e+00 -3.38096827e-01 -5.12842238e-01 5.43231905e-01
1.22251487e+00 1.51503071e-01 -5.21713555e-01 -7.56246448e-01
5.91068745e-01 -3.64946783e-01 2.05080435e-01 -1.02462280e+00
-9.11562562e-01 4.02967989e-01 2.91220397e-01 9.63991463e-01
4.14495915e-01 -2.21634701e-01 5.26177764e-01 1.15686190e+00
3.90271693e-01 -1.13231432e+00 -7.38215670e-02 9.52681839e-01
6.71283782e-01 -1.32725036e+00 1.20329544e-01 -4.24253911e-01
-8.13512206e-02 1.13702095e+00 6.01296782e-01 3.46850365e-01
1.10899723e+00 1.46691874e-01 9.95835885e-02 -4.81637388e-01
-1.42323613e+00 4.32758257e-02 2.74650753e-01 5.80618322e-01
9.65861440e-01 -2.38124132e-01 -3.53846371e-01 5.24070084e-01
-2.73879379e-01 2.71856748e-02 8.07438791e-01 1.17587256e+00
-5.67336082e-01 -1.07007635e+00 -1.91462591e-01 1.20899582e+00
-8.05285156e-01 -4.79643762e-01 -7.71417975e-01 7.12762952e-01
-4.98493671e-01 9.85752761e-01 2.93513447e-01 -5.32488167e-01
-1.04959652e-01 3.20426464e-01 -2.22575217e-02 -8.78394127e-01
-1.18596482e+00 -5.88525355e-01 1.06990397e+00 -1.58517152e-01
-2.54336208e-01 -7.28168666e-01 -9.20831144e-01 -6.78200424e-01
-4.60893959e-01 6.16819322e-01 9.29591656e-01 8.59961152e-01
6.06999755e-01 2.00331777e-01 4.34373647e-01 -4.45384473e-01
-3.97488534e-01 -1.45842528e+00 -4.09660637e-01 6.47650361e-01
-6.40360545e-03 -3.36096734e-01 -6.67042792e-01 3.45869005e-01] | [9.456633567810059, 9.122856140136719] |
205233cc-c18d-4565-9717-3232a8b442ec | hypothesis-testing-for-equality-of-latent | 2105.10838 | null | https://arxiv.org/abs/2105.10838v2 | https://arxiv.org/pdf/2105.10838v2.pdf | Hypothesis Testing for Equality of Latent Positions in Random Graphs | We consider the hypothesis testing problem that two vertices $i$ and $j$ of a generalized random dot product graph have the same latent positions, possibly up to scaling. Special cases of this hypothesis test include testing whether two vertices in a stochastic block model or degree-corrected stochastic block model graph have the same block membership vectors, or testing whether two vertices in a popularity adjusted block model have the same community assignment. We propose several test statistics based on the empirical Mahalanobis distances between the $i$th and $j$th rows of either the adjacency or the normalized Laplacian spectral embedding of the graph. We show that, under mild conditions, these test statistics have limiting chi-square distributions under both the null and local alternative hypothesis, and we derived explicit expressions for the non-centrality parameters under the local alternative. Using these limit results, we address the model selection problems including choosing between the standard stochastic block model and its degree-corrected variant, and choosing between the ER model and stochastic block model. The effectiveness of our proposed tests are illustrated via both simulation studies and real data applications. | ['Minh Tang', 'Xinjie Du'] | 2021-05-23 | null | null | null | null | ['stochastic-block-model'] | ['graphs'] | [ 1.54737264e-01 8.29199553e-02 -3.48149538e-01 -6.01997301e-02
-3.41947973e-02 -5.34619272e-01 3.58307093e-01 3.84086758e-01
-6.11705519e-02 6.45083070e-01 -2.80697376e-01 -6.35189652e-01
-9.08934295e-01 -1.07204843e+00 -5.47138572e-01 -9.91951525e-01
-6.92696810e-01 7.22205281e-01 2.40537927e-01 1.78065345e-01
1.86593652e-01 3.80910695e-01 -1.17163229e+00 -6.35391533e-01
9.05127704e-01 6.89799845e-01 -5.82255945e-02 7.65507162e-01
2.98955888e-01 3.72905880e-01 -3.15468192e-01 -4.95564967e-01
6.02310061e-01 -4.84127134e-01 -3.85163754e-01 1.24126434e-01
3.46286297e-01 2.09116891e-01 -5.66309512e-01 1.48718524e+00
3.57439697e-01 -6.22050054e-02 9.91717994e-01 -1.86263728e+00
-4.39925492e-01 9.45991099e-01 -1.02811599e+00 4.07007515e-01
2.53203779e-01 -1.70979291e-01 1.34320712e+00 -5.93425810e-01
6.22900128e-01 9.48710084e-01 4.13526922e-01 -2.44646594e-01
-1.74289441e+00 -8.24562907e-01 -2.47022390e-01 2.26665512e-01
-2.09974074e+00 -1.52843863e-01 5.64823568e-01 -6.56630993e-01
4.80173409e-01 2.08425179e-01 6.50268912e-01 3.97645265e-01
3.22389990e-01 1.54865861e-01 1.21886253e+00 -3.16983908e-01
2.30271980e-01 6.47353148e-03 3.63891035e-01 9.51081991e-01
1.04669869e+00 -1.41495049e-01 -3.68886560e-01 -7.08438098e-01
6.56264126e-01 -1.93365827e-01 -2.44175658e-01 -9.82268512e-01
-1.16407299e+00 9.70398486e-01 1.70685917e-01 4.28507537e-01
-3.81954193e-01 1.63580641e-01 -5.24913482e-02 4.39021498e-01
2.54195869e-01 -1.19851924e-01 3.25511582e-02 1.46518841e-01
-1.01586139e+00 -5.88276349e-02 9.22567070e-01 9.89887714e-01
9.41561699e-01 -1.39301583e-01 1.19830683e-01 2.32852191e-01
3.68197143e-01 9.30426776e-01 -2.45097652e-01 -5.21815181e-01
4.18647200e-01 2.87935555e-01 6.07915893e-02 -1.43999851e+00
-1.69685438e-01 -8.40185940e-01 -1.29080427e+00 -2.78720886e-01
4.95453179e-01 1.03273876e-01 -5.38067281e-01 2.08133483e+00
1.77979693e-01 4.60459054e-01 -2.80796021e-01 3.20546716e-01
2.54058599e-01 4.54986423e-01 -1.75657466e-01 -6.25320137e-01
8.24691951e-01 -3.24910402e-01 -4.90687221e-01 8.03613067e-02
5.60703218e-01 -4.85114157e-01 6.85684502e-01 1.88725486e-01
-1.14990664e+00 -2.55138397e-01 -1.07256806e+00 7.19670594e-01
1.20042920e-01 -3.56038690e-01 6.63744584e-02 8.12090158e-01
-1.32514155e+00 4.59457964e-01 -4.56958473e-01 -4.92520362e-01
-1.31692320e-01 5.03755748e-01 -4.19140160e-01 -3.77986491e-01
-1.09993601e+00 3.18684191e-01 -1.17021099e-01 4.26429994e-02
-3.98322225e-01 -2.37105697e-01 -5.61525941e-01 3.37716281e-01
2.33211920e-01 -5.96715868e-01 2.52083957e-01 -7.46087074e-01
-5.39121330e-01 7.78787076e-01 -2.66910531e-03 -2.32988104e-01
3.88550162e-01 5.55082381e-01 -4.07014936e-01 8.69725049e-02
4.41737920e-01 -2.04285514e-02 7.45301425e-01 -9.56553102e-01
-2.81483591e-01 -3.76673758e-01 -1.75113395e-01 2.88126655e-02
-2.10979655e-02 -4.01662767e-01 -3.38348031e-01 -4.40515488e-01
5.60605288e-01 -1.09688294e+00 -1.34705186e-01 -1.86624020e-01
-6.21993184e-01 -1.03158467e-01 4.41124558e-01 -3.08082312e-01
1.53046930e+00 -2.42152262e+00 2.10479632e-01 1.39409649e+00
3.16278279e-01 -4.28192079e-01 -2.58847624e-01 7.52916217e-01
-4.38524723e-01 3.56127352e-01 -6.26351163e-02 2.82202274e-01
1.85073048e-01 -1.15648098e-01 5.63545562e-02 1.13062406e+00
-4.07463908e-01 1.36507988e-01 -7.10656881e-01 -4.49026018e-01
-1.47383839e-01 5.42254262e-02 -4.59416777e-01 -2.06974939e-01
4.81026977e-01 -8.86730403e-02 -1.48494288e-01 1.56731695e-01
7.72549629e-01 -6.94419563e-01 5.62772155e-01 -1.07706986e-01
1.60886317e-01 -2.06746385e-02 -1.69809413e+00 7.67543316e-01
1.18798383e-01 6.54226422e-01 1.98871270e-01 -9.78982747e-01
7.31355309e-01 7.92420805e-02 4.53913808e-01 -2.99681276e-01
7.25798085e-02 1.40190616e-01 6.03796542e-01 -1.95457246e-02
3.04892026e-02 -1.36320993e-01 3.56487371e-03 6.40636504e-01
-3.15018803e-01 2.29918912e-01 5.31713128e-01 5.98365724e-01
1.62140346e+00 -6.98570430e-01 4.07522261e-01 -8.46920550e-01
3.49430263e-01 -5.09944618e-01 4.48191971e-01 1.10796118e+00
-1.76317722e-01 1.89837590e-01 1.06648219e+00 2.84554094e-01
-9.57899392e-01 -1.38535595e+00 -2.52339512e-01 5.61422288e-01
5.48916817e-01 -3.63013595e-01 -3.79758179e-01 -3.61125946e-01
3.58697236e-01 5.21360040e-01 -6.05523944e-01 -3.07699472e-01
-4.68882918e-02 -5.66475093e-01 1.62202314e-01 -7.68997893e-02
4.60700631e-01 -3.61035377e-01 6.61756843e-02 -1.23635620e-01
-9.84081626e-02 -6.65897846e-01 -5.96870720e-01 9.25093293e-02
-6.43546760e-01 -1.29839849e+00 -4.82882559e-01 -7.91905761e-01
8.19806874e-01 4.96549994e-01 8.18614125e-01 7.35656917e-02
-1.06711119e-01 4.39557523e-01 -2.06040084e-01 5.38117468e-01
-1.72130063e-01 6.24659611e-03 3.40705305e-01 2.08724573e-01
1.31360292e-01 -6.86891019e-01 -7.76662290e-01 7.73120701e-01
-7.89763868e-01 -1.16099656e-01 5.41099548e-01 5.17630816e-01
4.32770342e-01 7.15041399e-01 2.42509663e-01 -6.31494701e-01
5.31154573e-01 -8.37453544e-01 -5.85378051e-01 2.25925103e-01
-8.94574523e-01 1.68776721e-01 1.21585391e-01 -3.43285292e-01
-1.05389319e-01 -4.25528705e-01 4.63312864e-01 -2.86254346e-01
5.17790735e-01 9.50398088e-01 -2.94039369e-01 2.60104891e-02
2.36945003e-01 -1.02663971e-02 -9.77809653e-02 3.76255959e-02
2.34936669e-01 3.88148367e-01 2.87523493e-03 -3.34066749e-01
9.88333702e-01 4.55014676e-01 5.03689766e-01 -1.00489497e+00
-7.37303495e-02 -5.31694710e-01 -4.05281931e-01 -1.74635261e-01
5.98593891e-01 -6.79749370e-01 -7.40978837e-01 1.31009951e-01
-5.07156968e-01 -2.20892414e-01 -2.50830259e-02 4.58692521e-01
-4.43944335e-01 6.02327824e-01 -4.62534189e-01 -8.53486717e-01
1.54543668e-01 -9.50678170e-01 4.57632303e-01 -2.14382932e-01
-2.02663094e-01 -9.29330587e-01 3.07574183e-01 -1.76489174e-01
1.28065962e-02 8.96203443e-02 1.52235341e+00 -6.80970550e-01
-5.32975137e-01 -5.63652813e-01 -4.32614982e-01 -1.22575536e-01
3.49840485e-02 3.47525656e-01 1.01888694e-01 -8.21324408e-01
-3.73763353e-01 4.25191104e-01 5.21192253e-01 7.64687061e-01
7.10294664e-01 -3.81965011e-01 -6.33221686e-01 2.49167293e-01
1.45180249e+00 -7.88593292e-02 4.76210088e-01 -1.45806655e-01
2.52068430e-01 5.23804069e-01 5.23739696e-01 5.74674487e-01
-1.58026535e-02 6.47727668e-01 2.64838517e-01 2.03604817e-01
2.81896949e-01 -2.95165300e-01 3.01847011e-01 1.16532588e+00
2.09559590e-01 -7.75281310e-01 -9.90985751e-01 6.77219033e-01
-1.74601603e+00 -1.12253451e+00 -4.15021598e-01 2.79385543e+00
2.89684027e-01 3.15421194e-01 2.74006814e-01 1.15244187e-01
1.34526789e+00 8.46390501e-02 -3.99079561e-01 -9.16315094e-02
-2.78992087e-01 3.22816253e-01 9.72382009e-01 6.77449167e-01
-7.76247799e-01 2.96587855e-01 6.13518381e+00 1.01486957e+00
-5.91822326e-01 9.29679628e-03 4.94430065e-01 -1.75805241e-02
-4.05771464e-01 3.46509069e-01 -3.95232797e-01 5.72844982e-01
9.37096477e-01 -4.81786937e-01 1.16170868e-01 5.43340385e-01
2.20792443e-01 -3.66112024e-01 -8.22827697e-01 1.02361047e+00
-1.04701132e-01 -8.57331932e-01 -1.37401626e-01 8.52448821e-01
8.91182244e-01 -1.23139955e-02 2.87007213e-01 -2.28765070e-01
6.82588875e-01 -7.98974872e-01 3.39219183e-01 4.32973802e-01
1.07796252e+00 -7.37534404e-01 6.58975124e-01 1.80203736e-01
-1.35665667e+00 1.61157817e-01 -2.79878259e-01 7.23853707e-02
-2.48976368e-02 8.20195019e-01 -5.13003588e-01 3.60526472e-01
3.61703366e-01 5.64444780e-01 -5.23371935e-01 1.17259133e+00
4.26511690e-02 7.23878920e-01 -4.92487669e-01 -6.29681796e-02
-5.78759424e-02 -7.19185591e-01 7.67353773e-01 4.93961811e-01
6.51309490e-01 -8.90609622e-02 6.29840940e-02 4.38917518e-01
1.14114054e-01 2.47373894e-01 -7.37135708e-01 -1.07527234e-01
6.54421985e-01 8.79622281e-01 -1.15595043e+00 -9.82727930e-02
-3.43005896e-01 7.68395722e-01 1.12184800e-01 5.96980989e-01
-8.90549421e-01 -5.14624119e-01 6.71168387e-01 6.62558258e-01
4.64572251e-01 -3.76634419e-01 -1.20246680e-02 -1.15237129e+00
-2.97425121e-01 -5.90156555e-01 4.15050715e-01 -4.99230951e-01
-1.35708928e+00 2.21532464e-01 1.78929657e-01 -1.02648175e+00
-5.11229672e-02 -3.75370294e-01 -6.08608246e-01 6.95789278e-01
-6.33425891e-01 -4.10640925e-01 1.33433431e-01 5.73067844e-01
-3.70170861e-01 -1.98277668e-03 3.58587086e-01 3.19077015e-01
-6.47285044e-01 4.65653867e-01 7.61540353e-01 -1.19679570e-01
3.92467201e-01 -1.00162697e+00 -7.37091154e-02 9.43194032e-01
2.65274823e-01 6.75403297e-01 1.18070424e+00 -8.32240045e-01
-1.23133993e+00 -6.99229598e-01 7.79386699e-01 2.74246663e-01
9.15053606e-01 -4.33673441e-01 -4.88025248e-01 4.70892936e-01
1.40825421e-01 -1.26210243e-01 8.72371137e-01 2.77354091e-01
-2.29363739e-01 -1.35898992e-01 -1.16991901e+00 5.95543444e-01
1.15382349e+00 -4.03381854e-01 1.81920663e-01 3.61749053e-01
1.08337454e-01 5.65534413e-01 -8.40694785e-01 5.12537241e-01
5.66790402e-01 -9.63100135e-01 7.18166351e-01 -4.46176231e-01
6.26535192e-02 -3.90264571e-01 -4.43978041e-01 -1.06005538e+00
-8.14893126e-01 -4.77010489e-01 2.71133423e-01 8.16975236e-01
4.80729789e-01 -7.63535798e-01 8.57335329e-01 2.55712509e-01
7.14172900e-01 -3.97562474e-01 -1.36242080e+00 -9.04206514e-01
-1.34165749e-01 -1.22610480e-02 2.23752573e-01 1.00458503e+00
2.42454380e-01 4.88054127e-01 -2.09517464e-01 4.10613000e-01
8.58096302e-01 1.33660436e-01 8.49483550e-01 -1.40205491e+00
-4.74426836e-01 -6.93338752e-01 -8.60888243e-01 -8.69753182e-01
1.92390651e-01 -8.83317769e-01 -1.62448540e-01 -1.43742788e+00
5.71469963e-01 -6.30159676e-01 -3.08919251e-01 -2.33462736e-01
5.39803766e-02 2.82487631e-01 -1.79247215e-01 3.74005765e-01
-4.01231378e-01 3.71360004e-01 7.09727764e-01 -6.40981644e-02
-1.70300119e-02 1.98715836e-01 -3.08207870e-01 4.20468360e-01
4.76215184e-01 -5.16119421e-01 -5.40754497e-01 2.69555390e-01
8.90831947e-01 5.23972750e-01 2.45301843e-01 -8.25288415e-01
1.21746950e-01 -7.00028390e-02 -2.17141867e-01 -5.25862455e-01
5.87978885e-02 -8.88057709e-01 7.43219554e-01 7.87218451e-01
-2.50380844e-01 3.65339577e-01 -2.71728039e-01 1.12114453e+00
1.21110268e-02 -2.28663892e-01 7.41202116e-01 5.65048039e-01
-2.53447667e-02 4.74825770e-01 -3.90937954e-01 -2.22678438e-01
1.21841240e+00 -3.24289560e-01 -2.98732191e-01 -1.03034282e+00
-8.50428283e-01 1.87413961e-01 5.38836598e-01 -1.61650807e-01
4.27726477e-01 -1.34983754e+00 -7.40460336e-01 3.57743829e-01
1.17387779e-01 -7.65006065e-01 2.20037013e-01 1.29622281e+00
-5.18035531e-01 1.06468424e-01 4.07256819e-02 -6.75973356e-01
-1.48456633e+00 6.08861804e-01 8.97530019e-02 -2.52967179e-01
9.81457606e-02 5.38675249e-01 4.72751975e-01 1.65044907e-02
-1.91355690e-01 1.88185126e-01 2.58512795e-01 -1.06375761e-01
-1.09771587e-01 6.86031938e-01 -2.40953326e-01 -8.98795366e-01
-3.69784653e-01 3.64115417e-01 9.03928578e-02 -4.57670450e-01
8.74356449e-01 -3.40616196e-01 -6.03675544e-01 4.70918268e-01
1.19363916e+00 5.10008037e-01 -6.04854107e-01 -3.36320043e-01
-2.00022846e-01 -7.12556899e-01 -1.94545433e-01 -7.20291883e-02
-1.00283206e+00 5.84649086e-01 5.65090239e-01 9.58391309e-01
7.16180563e-01 3.33950043e-01 -1.64973900e-01 -3.32925841e-02
5.31260550e-01 -5.56534410e-01 -3.33032846e-01 2.21512035e-01
4.34682190e-01 -5.85057676e-01 1.94443956e-01 -7.71125376e-01
-1.25186266e-02 6.22626901e-01 3.26594621e-01 -5.07183731e-01
1.23217165e+00 -2.53367662e-01 -6.03343785e-01 -3.44938278e-01
-5.47253728e-01 -3.27885956e-01 2.06671745e-01 3.94848347e-01
3.43522727e-01 2.46661410e-01 -6.79540157e-01 -7.25968331e-02
-2.18258396e-01 -7.80687988e-01 5.28199673e-01 4.03096706e-01
-3.30290854e-01 -9.38860118e-01 -9.62349772e-02 9.26175892e-01
-1.03282832e-01 -2.58402318e-01 -3.37039262e-01 9.78003740e-01
-1.94012895e-01 9.41230357e-01 2.73309231e-01 -4.90978539e-01
-8.00243616e-02 -1.41563118e-01 5.39647102e-01 -5.02187908e-01
8.87808129e-02 2.49584958e-01 -1.62779957e-01 -4.54671979e-02
-3.12952548e-01 -7.89983869e-01 -4.51367319e-01 -9.75141168e-01
-7.80504704e-01 5.90800643e-01 3.16022962e-01 6.66772902e-01
3.79321575e-01 -8.70689750e-02 6.87605083e-01 -1.23043753e-01
-3.26393336e-01 -1.00123012e+00 -1.31553376e+00 2.38127381e-01
-3.70984152e-02 -7.84334898e-01 -7.54478335e-01 -5.07326365e-01] | [6.923585891723633, 5.172764778137207] |
e0161232-7989-414a-929c-9c0df8b0622f | social-processes-self-supervised-forecasting | 2107.13576 | null | https://arxiv.org/abs/2107.13576v3 | https://arxiv.org/pdf/2107.13576v3.pdf | Social Processes: Self-Supervised Meta-Learning over Conversational Groups for Forecasting Nonverbal Social Cues | Free-standing social conversations constitute a yet underexplored setting for human behavior forecasting. While the task of predicting pedestrian trajectories has received much recent attention, an intrinsic difference between these settings is how groups form and disband. Evidence from social psychology suggests that group members in a conversation explicitly self-organize to sustain the interaction by adapting to one another's behaviors. Crucially, the same individual is unlikely to adapt similarly across different groups; contextual factors such as perceived relationships, attraction, rapport, etc., influence the entire spectrum of participants' behaviors. A question arises: how can we jointly forecast the mutually dependent futures of conversation partners by modeling the dynamics unique to every group? In this paper, we propose the Social Process (SP) models, taking a novel meta-learning and stochastic perspective of group dynamics. Training group-specific forecasting models hinders generalization to unseen groups and is challenging given limited conversation data. In contrast, our SP models treat interaction sequences from a single group as a meta-dataset: we condition forecasts for a sequence from a given group on other observed-future sequence pairs from the same group. In this way, an SP model learns to adapt its forecasts to the unique dynamics of the interacting partners, generalizing to unseen groups in a data-efficient manner. Additionally, we first rethink the task formulation itself, motivating task requirements from social science literature that prior formulations have overlooked. For our formulation of Social Cue Forecasting, we evaluate the empirical performance of our SP models against both non-meta-learning and meta-learning approaches with similar assumptions. The SP models yield improved performance on synthetic and real-world behavior datasets. | ['Marco Loog', 'Hayley Hung', 'Chirag Raman'] | 2021-07-28 | null | https://openreview.net/forum?id=qcjOWDHAc4J | https://openreview.net/pdf?id=qcjOWDHAc4J | neurips-2021-12 | ['social-cue-forecasting', 'human-behavior-forecasting'] | ['time-series', 'time-series'] | [ 1.53224975e-01 1.96761806e-02 -1.80660367e-01 -5.51828325e-01
-7.02047721e-02 -4.51588511e-01 9.19571459e-01 1.26309380e-01
-2.12187096e-01 7.95427322e-01 7.61790574e-01 -1.92060247e-01
-1.71889246e-01 -5.17490029e-01 -6.61165595e-01 -6.15162790e-01
-2.36094818e-01 5.47621787e-01 1.95345283e-01 -4.39458162e-01
2.49927640e-01 1.67564616e-01 -1.72814023e+00 3.28037411e-01
9.41468358e-01 3.13642412e-01 2.03818336e-01 9.12192702e-01
-1.04779840e-01 7.70018637e-01 -4.11339998e-01 -5.27937174e-01
4.29383814e-02 -6.05516970e-01 -5.39413333e-01 4.36609328e-01
2.90329516e-01 -1.76614657e-01 -3.89291465e-01 5.27462006e-01
2.30541065e-01 4.36302871e-01 9.69842434e-01 -1.56929243e+00
-6.96037054e-01 7.27932394e-01 -3.64814758e-01 2.93030322e-01
5.33772826e-01 3.54804486e-01 8.85736883e-01 -1.32256851e-01
4.35750961e-01 1.54376149e+00 7.41783023e-01 8.22652459e-01
-1.35746062e+00 -5.96449435e-01 8.57896447e-01 1.64491504e-01
-7.56485879e-01 -4.35096830e-01 8.46193254e-01 -7.43986070e-01
5.19964635e-01 1.95017233e-01 7.91475415e-01 1.76244783e+00
1.51339516e-01 7.46315658e-01 9.20368850e-01 -3.98811586e-02
3.15439522e-01 1.24117427e-01 4.28959727e-01 2.38640472e-01
1.03398554e-01 1.33672684e-01 -7.50440121e-01 -4.45693970e-01
2.15696275e-01 2.72245646e-01 4.54428829e-02 -2.31844053e-01
-1.12916112e+00 7.62254357e-01 1.78178027e-01 3.05673838e-01
-3.74315232e-01 6.55725077e-02 1.18957557e-01 3.91903907e-01
7.31275380e-01 1.93512127e-01 -2.19943061e-01 -4.04425144e-01
-3.82168114e-01 6.98453188e-01 1.17854023e+00 8.46257925e-01
9.38357055e-01 -3.12508613e-01 -1.97193518e-01 7.32402623e-01
1.94947883e-01 3.08622211e-01 3.59335274e-01 -1.02634037e+00
3.95109177e-01 4.80346322e-01 5.09697855e-01 -1.36279571e+00
-4.13499117e-01 -2.33350918e-01 -5.82327068e-01 -3.06336761e-01
8.43365610e-01 -5.03492773e-01 -4.66646612e-01 2.23127580e+00
4.20402288e-01 7.26123929e-01 -1.62080377e-01 4.69396621e-01
3.58741045e-01 6.66247845e-01 2.97133297e-01 -3.69938344e-01
7.06746697e-01 -9.00805652e-01 -4.99448001e-01 -3.54880691e-01
5.56397319e-01 -2.96876550e-01 7.85619795e-01 8.14374834e-02
-9.25526857e-01 -6.69377208e-01 -6.77035511e-01 3.47687304e-01
-2.37252384e-01 -6.23696864e-01 8.78506243e-01 7.43179202e-01
-1.14581978e+00 5.98923922e-01 -7.75452495e-01 -8.07304323e-01
1.85936928e-01 3.44888717e-01 2.31331695e-04 1.98002413e-01
-9.44551766e-01 7.18662977e-01 -1.24787658e-01 -4.38010246e-02
-8.11292648e-01 -7.97938883e-01 -3.78291458e-01 -3.89286992e-03
4.13824946e-01 -8.95927370e-01 1.33322835e+00 -1.15720391e+00
-1.46320295e+00 5.88401914e-01 -6.85952485e-01 -4.92730886e-01
7.24938929e-01 6.36763200e-02 -4.79568660e-01 -3.53934348e-01
-2.83328649e-02 4.79782552e-01 8.25816691e-01 -1.49086726e+00
-8.76629114e-01 -3.88838530e-01 5.66154197e-02 2.79192239e-01
-3.43690306e-01 -2.24260896e-01 4.93963547e-02 -3.38195115e-01
-2.18166664e-01 -1.28429890e+00 -3.79811943e-01 -3.62110466e-01
-1.97323963e-01 -4.83472854e-01 5.49629867e-01 -1.54413000e-01
1.26513696e+00 -1.98754156e+00 2.96297997e-01 1.24213360e-01
3.47081453e-01 -7.17643276e-02 -1.83991536e-01 8.12713444e-01
3.27920049e-01 1.40041783e-01 -5.11380956e-02 -8.40924025e-01
4.17249091e-02 3.27223808e-01 -3.87110054e-01 3.87161762e-01
-2.21727446e-01 7.77002037e-01 -1.18746185e+00 -1.88604727e-01
2.89066546e-02 2.26591274e-01 -7.53984392e-01 3.08201462e-01
-2.76592225e-01 8.42158675e-01 -5.63688636e-01 2.07657337e-01
3.76235366e-01 -3.10954928e-01 4.06495929e-01 4.18543071e-01
-1.19788609e-01 1.19165525e-01 -9.13772106e-01 1.05707061e+00
-3.14983577e-01 5.83131909e-01 -6.81723729e-02 -1.10418904e+00
8.23597789e-01 5.16300276e-02 7.07235575e-01 -1.09386191e-01
-5.06576104e-03 -1.41134515e-01 3.41045290e-01 -6.73975229e-01
5.77753007e-01 -1.56903520e-01 -1.01686649e-01 8.91043127e-01
-2.08770752e-01 3.23858410e-01 6.62489310e-02 1.29930288e-01
9.87466931e-01 -1.41710430e-01 2.36900151e-01 -5.47819883e-02
3.47681075e-01 -2.82282978e-01 5.96776545e-01 1.16436803e+00
-4.85434264e-01 2.62693197e-01 4.84151632e-01 -5.14628649e-01
-8.87777448e-01 -1.10751045e+00 1.99795485e-01 1.53272998e+00
2.60158598e-01 -2.04213515e-01 -6.14170074e-01 -4.89567131e-01
3.14464509e-01 7.44374692e-01 -8.43622446e-01 -2.12444946e-01
-7.42628574e-01 -8.99529397e-01 5.45022711e-02 2.18188390e-01
1.49168670e-01 -9.55056190e-01 -1.70102015e-01 4.58075881e-01
-2.62319267e-01 -9.59623754e-01 -6.10290349e-01 -5.28463781e-01
-5.52081943e-01 -9.82279897e-01 -4.46089327e-01 -5.86232007e-01
5.24134219e-01 6.10226750e-01 9.90030229e-01 2.35916346e-01
3.49269927e-01 8.34044337e-01 -2.31083483e-01 -4.77874249e-01
-7.38502860e-01 1.86170965e-01 3.42856765e-01 6.29501820e-01
6.55058742e-01 -9.98397171e-01 -6.27799332e-01 6.64603710e-01
-4.86159652e-01 1.37499630e-01 -2.87883207e-02 4.97145593e-01
-2.18988389e-01 -2.13319406e-01 9.51495886e-01 -8.98459315e-01
8.73210609e-01 -1.06541860e+00 -1.13695100e-01 2.66894162e-01
-3.28210443e-01 -3.34923804e-01 4.74136531e-01 -1.01581168e+00
-1.19145417e+00 -1.74421281e-01 3.81740451e-01 3.41334902e-02
-3.26103956e-01 1.08709015e-01 -8.60552117e-03 3.07241589e-01
5.26176155e-01 1.39775574e-01 3.40077877e-01 -1.57528460e-01
2.82079846e-01 6.28300011e-01 3.31544429e-01 -8.80545616e-01
6.73320413e-01 6.56087339e-01 -1.69638425e-01 -9.54538882e-01
-8.89949322e-01 -4.92810726e-01 -6.32809341e-01 -7.21515179e-01
7.83149123e-01 -6.94348276e-01 -1.42952096e+00 6.50097609e-01
-1.12053657e+00 -5.78276694e-01 6.63865432e-02 3.85886371e-01
-7.57669628e-01 3.65640521e-01 -2.88593978e-01 -1.22122347e+00
4.26547527e-01 -9.71000254e-01 7.86637008e-01 3.01481307e-01
-6.49162173e-01 -1.52688658e+00 2.98497468e-01 4.33427185e-01
4.07175958e-01 3.10414374e-01 6.98249221e-01 -8.19365382e-01
-5.93181968e-01 5.12549579e-02 2.82181770e-01 4.53451648e-02
5.10179698e-01 2.02603135e-02 -8.66091311e-01 -1.76145360e-01
-5.91145866e-02 3.52137573e-02 6.94088221e-01 5.54384053e-01
8.35650027e-01 -4.09801632e-01 -8.31150651e-01 1.74397841e-01
6.05391622e-01 3.58761698e-01 9.55218524e-02 1.38984285e-02
6.77746892e-01 1.31979191e+00 1.55941427e-01 4.70853806e-01
9.74302948e-01 5.90786874e-01 1.08205006e-01 4.60165709e-01
1.79752603e-01 -7.57560313e-01 5.81592202e-01 6.12675369e-01
-1.40609324e-01 -6.41864479e-01 -9.47277486e-01 5.10545373e-01
-2.39529371e+00 -1.41450715e+00 4.94418746e-05 2.20124125e+00
4.89803702e-01 1.58712454e-02 7.46995032e-01 -3.11441571e-01
9.24436450e-01 3.33646685e-01 -7.48973012e-01 -1.46235913e-01
-6.54633790e-02 -6.72473848e-01 1.21761002e-01 9.11495984e-01
-7.92568028e-01 8.11786234e-01 6.63202000e+00 2.91598052e-01
-8.01193595e-01 9.26034302e-02 9.00606632e-01 -1.15535267e-01
-4.64040518e-01 7.32079819e-02 -1.09833801e+00 8.15325081e-01
1.18479502e+00 -5.36315501e-01 6.70145094e-01 4.12530690e-01
5.26030540e-01 -1.12305293e-02 -1.60406315e+00 8.08623552e-01
2.18039781e-01 -1.21885395e+00 -4.68336567e-02 3.47766846e-01
8.42560887e-01 -1.56234235e-01 2.27877468e-01 5.12905896e-01
7.86771297e-01 -9.25136983e-01 5.22903264e-01 8.61341834e-01
-2.18419835e-01 -5.83968818e-01 2.02787936e-01 8.31834555e-01
-9.09466326e-01 -5.06255329e-01 -1.83446512e-01 -6.11521363e-01
5.23319423e-01 2.09024698e-01 -9.73277450e-01 1.30009308e-01
3.10999244e-01 1.05513251e+00 -4.03807461e-01 6.26553357e-01
3.44003290e-01 8.60687912e-01 -1.65540636e-01 -3.01683217e-01
2.05227509e-01 -4.52708393e-01 8.94198656e-01 9.27901745e-01
2.10952118e-01 1.42060146e-01 4.35075045e-01 6.70712888e-01
2.20055178e-01 -1.26377851e-01 -8.55459273e-01 -1.05246119e-01
5.00972092e-01 8.24666202e-01 -3.77918690e-01 -2.01884031e-01
-3.67186755e-01 5.15961051e-01 5.31186521e-01 6.98805749e-01
-5.79415500e-01 6.37105823e-01 1.12509120e+00 3.67655814e-01
-3.54796983e-02 -2.87736803e-01 -1.76236734e-01 -1.21205497e+00
-1.82403654e-01 -7.51721621e-01 3.38239968e-01 -3.28241050e-01
-1.90570104e+00 3.13323110e-01 3.14032257e-01 -1.05810142e+00
-7.68267632e-01 -2.68493831e-01 -8.33519936e-01 6.84872985e-01
-1.17747211e+00 -1.18148494e+00 -2.67844409e-01 4.04801518e-01
6.55075371e-01 -3.10351819e-01 3.45363766e-01 -1.06499173e-01
-4.60014164e-01 3.87539983e-01 -3.63727659e-02 -2.11301968e-01
6.02645040e-01 -1.04264545e+00 5.38524330e-01 4.38433260e-01
-1.42005950e-01 7.79789567e-01 1.00734639e+00 -8.17080915e-01
-1.19162858e+00 -8.98493767e-01 1.05957067e+00 -9.60737765e-01
9.49107349e-01 -6.46965265e-01 -9.89455104e-01 9.09428298e-01
-2.72172503e-02 -3.80164623e-01 9.43134487e-01 5.72853863e-01
-1.43566698e-01 -3.41019258e-02 -9.19430733e-01 1.11592078e+00
1.64716637e+00 -4.57218826e-01 -4.63007748e-01 4.18392062e-01
6.38395429e-01 7.87714031e-03 -5.08743227e-01 3.56386527e-02
6.70960367e-01 -1.24849260e+00 9.35661197e-01 -9.88924682e-01
3.68560880e-01 4.74796407e-02 -2.05937028e-01 -1.55356789e+00
-3.09466392e-01 -1.08950043e+00 -1.54204607e-01 1.32729185e+00
2.11914375e-01 -1.00876129e+00 8.56634617e-01 1.04315424e+00
2.30815988e-02 -5.17074943e-01 -7.18746603e-01 -6.34275973e-01
3.74532610e-01 -4.12210166e-01 9.44194138e-01 8.20187390e-01
2.35027045e-01 2.97562867e-01 -7.53185034e-01 1.43444195e-01
6.62088454e-01 1.84132066e-02 1.28242433e+00 -1.43008316e+00
-6.27306461e-01 -5.58236420e-01 -1.85677096e-01 -1.30436110e+00
7.73469388e-01 -6.91149950e-01 -6.02878481e-02 -9.97157872e-01
3.38555455e-01 -5.43577373e-01 6.61841258e-02 -1.48967430e-01
-3.93534958e-01 -4.01370704e-01 3.95690829e-01 2.61301279e-01
-4.85738963e-01 5.92196524e-01 1.21753860e+00 -3.23369750e-03
-5.65566897e-01 6.56290174e-01 -9.84102726e-01 7.34069228e-01
7.07706630e-01 -1.78853229e-01 -8.22794676e-01 -7.86699057e-02
1.79463536e-01 1.96072012e-01 3.81027877e-01 -7.47733176e-01
4.41484779e-01 -6.78636789e-01 -5.45505919e-02 -3.17018569e-01
4.94477361e-01 -6.09508514e-01 3.19426805e-01 2.80788094e-01
-7.34714448e-01 8.14559162e-02 -2.06482589e-01 1.19562936e+00
2.37362340e-01 1.40862271e-01 3.47404033e-01 -1.86173677e-01
-3.07191074e-01 5.49188077e-01 -7.33414173e-01 4.67068776e-02
1.12427044e+00 -5.01347065e-01 -2.79614896e-01 -1.12327790e+00
-8.99276197e-01 4.95129853e-01 5.30191541e-01 7.41085410e-01
2.69919395e-01 -9.78471398e-01 -7.12317586e-01 -6.66665211e-02
2.62397118e-02 -4.01982695e-01 5.24722338e-01 6.87171638e-01
3.09317201e-01 7.78044090e-02 -9.72689986e-02 -7.65169263e-01
-1.13580096e+00 5.01514554e-01 3.93314332e-01 -4.05467153e-02
-4.14530784e-01 4.44273114e-01 5.23222268e-01 -5.84015548e-01
1.80899963e-01 -2.50036389e-01 -3.66179824e-01 2.22916603e-01
6.22053027e-01 6.54754519e-01 -6.46515429e-01 -8.11404467e-01
7.17852265e-02 2.92795420e-01 -2.74817199e-01 -1.72757819e-01
1.35587907e+00 -5.82074881e-01 1.85578153e-01 1.01479077e+00
9.96022046e-01 -1.67472228e-01 -1.58662462e+00 -3.51547837e-01
1.69245616e-01 -5.74415028e-01 -6.44590974e-01 -3.77514184e-01
-3.95716697e-01 5.51682770e-01 7.10100960e-03 4.87305790e-01
5.13454020e-01 1.08363740e-01 6.98743224e-01 2.41690993e-01
5.31068742e-01 -8.73152971e-01 3.67795795e-01 4.24434364e-01
7.27172375e-01 -1.30394852e+00 -2.82425553e-01 -3.51945490e-01
-9.09703076e-01 7.52346098e-01 6.23917520e-01 -2.12222133e-02
9.81833458e-01 -1.03596874e-01 -1.62987918e-01 1.57502159e-01
-1.21223116e+00 -1.06134554e-02 9.24476907e-02 8.98614764e-01
2.72482336e-01 2.83399880e-01 -3.80077064e-02 5.03710806e-01
-4.29175198e-01 -1.28511012e-01 6.28846884e-01 5.83184302e-01
-4.66916353e-01 -1.11053765e+00 -1.12218790e-01 4.80542928e-01
1.01191148e-01 4.48071450e-01 -4.04476136e-01 5.44202983e-01
8.24937299e-02 1.35193968e+00 4.09388006e-01 -4.90576953e-01
6.80983663e-02 1.12709157e-01 2.35806659e-01 -6.85746312e-01
-7.18673110e-01 -3.46049517e-01 1.95442274e-01 -2.55646974e-01
-7.82138050e-01 -1.41882396e+00 -7.12848246e-01 -8.29842567e-01
3.62546146e-01 5.09093963e-02 2.97823608e-01 1.18024921e+00
4.11296606e-01 5.06754555e-02 8.83174598e-01 -1.07004094e+00
-6.60111904e-01 -9.48453605e-01 -2.67106980e-01 7.01328754e-01
6.38595045e-01 -8.46543550e-01 -7.40196049e-01 9.33099836e-02] | [6.111873626708984, 0.8041769862174988] |
e7716f13-f84c-4396-a18e-69c6fa8c2d74 | few-shot-incremental-learning-in-the-context | 2207.00693 | null | https://arxiv.org/abs/2207.00693v1 | https://arxiv.org/pdf/2207.00693v1.pdf | Few-shot incremental learning in the context of solar cell quality inspection | In industry, Deep Neural Networks have shown high defect detection rates surpassing other more traditional manual feature engineering based proposals. This has been achieved mainly through supervised training where a great amount of data is required in order to learn good classification models. However, such amount of data is sometimes hard to obtain in industrial scenarios, as few defective pieces are produced normally. In addition, certain kinds of defects are very rare and usually just appear from time to time, which makes the generation of a proper dataset for training a classification model even harder. Moreover, the lack of available data limits the adaptation of inspection models to new defect types that appear in production as it might require a model retraining in order to incorporate the detects and detect them. In this work, we have explored the technique of weight imprinting in the context of solar cell quality inspection where we have trained a network on three base defect classes, and then we have incorporated new defect classes using few samples. The results have shown that this technique allows the network to extend its knowledge with regard to defect classes with few samples, which can be interesting for industrial practitioners. | ['Luka Eciolaza', 'Julen Balzategui'] | 2022-07-01 | null | null | null | null | ['defect-detection'] | ['computer-vision'] | [ 4.13817465e-01 2.60401130e-01 1.96430326e-01 -3.78161579e-01
1.57158062e-01 -1.21402703e-01 2.98284441e-01 5.10312259e-01
-1.61312670e-01 8.32506776e-01 -5.19473910e-01 -1.27134696e-01
-2.53083408e-01 -1.12203109e+00 -7.13021874e-01 -7.57525504e-01
1.42054632e-02 4.73408103e-01 3.31466049e-01 -2.90227860e-01
3.23138744e-01 6.07226014e-01 -2.07848001e+00 2.94072449e-01
1.13680220e+00 8.97203386e-01 5.43992937e-01 2.31417596e-01
-6.97106645e-02 2.87377864e-01 -9.86946166e-01 -3.84826176e-02
3.01581562e-01 -3.50869596e-01 -3.54249984e-01 5.32347798e-01
1.57534033e-01 -2.06803873e-01 2.22917616e-01 9.16773379e-01
3.45420659e-01 1.57732502e-01 7.29757309e-01 -9.85822618e-01
-5.02280533e-01 2.92144150e-01 -3.25141132e-01 -1.41854659e-01
1.37440383e-01 2.10781053e-01 6.61423624e-01 -4.60462213e-01
5.05809784e-01 8.22113335e-01 5.86693704e-01 6.35002017e-01
-1.09508765e+00 -2.69808441e-01 -1.82131499e-01 9.64287892e-02
-7.81842828e-01 -1.18596181e-01 1.10374641e+00 -4.38879371e-01
9.84476388e-01 -3.95014696e-03 9.87340093e-01 8.33172739e-01
2.38798827e-01 6.96671963e-01 9.51257408e-01 -7.24586308e-01
4.47066963e-01 4.71957326e-01 3.05943433e-02 4.80435014e-01
6.70059383e-01 1.11903235e-01 -1.48778260e-01 4.04009551e-01
8.00682425e-01 1.37367412e-01 -3.02860141e-01 -5.06082058e-01
-6.73886120e-01 8.71217191e-01 4.40955669e-01 9.11155820e-01
-3.92496765e-01 -2.84617931e-01 5.21991611e-01 6.32209659e-01
4.20384675e-01 7.23686159e-01 -5.88986516e-01 4.64008674e-02
-9.64415848e-01 -1.52404308e-01 7.59219110e-01 4.97584045e-01
7.61760771e-01 2.76150525e-01 2.56989360e-01 1.10415030e+00
4.60141823e-02 6.43371418e-02 5.38003683e-01 -6.03572905e-01
1.68908194e-01 1.04213989e+00 -1.29635399e-03 -9.46998894e-01
-3.62179607e-01 -6.13198757e-01 -1.02212703e+00 8.15303266e-01
5.48761189e-01 -6.18523685e-03 -1.11353314e+00 1.14635193e+00
1.49830088e-01 -3.33361953e-01 1.14823334e-01 6.97938025e-01
4.44359571e-01 5.72833121e-01 -4.30642366e-01 -2.26909146e-01
8.10221255e-01 -6.73900843e-01 -7.36988842e-01 -1.69054091e-01
7.07402647e-01 -4.98291641e-01 1.19303560e+00 1.16144931e+00
-7.90129125e-01 -8.51755500e-01 -1.43930984e+00 3.32365841e-01
-7.22340167e-01 3.11072707e-01 8.62674415e-01 9.22443748e-01
-8.61017942e-01 1.19982016e+00 -7.26520836e-01 -4.65803146e-01
6.02974534e-01 6.10650837e-01 -3.84952039e-01 -4.43961442e-01
-9.73941565e-01 7.88487852e-01 4.92413223e-01 4.55122679e-01
-7.36726761e-01 -2.89438128e-01 -6.58205330e-01 7.44971707e-02
2.98630089e-01 -3.20877075e-01 8.47179949e-01 -1.09306252e+00
-1.42658448e+00 4.85908002e-01 3.05656374e-01 -2.84097642e-01
3.55001688e-01 7.53466710e-02 -4.88467157e-01 -7.48475492e-02
-1.96750671e-01 4.03872848e-01 9.26541746e-01 -1.48707426e+00
-5.96328139e-01 -2.84547597e-01 3.32225680e-01 -3.11128646e-01
-6.77060485e-01 -4.51277584e-01 5.75591214e-02 -3.02845985e-01
9.94001627e-02 -5.61416030e-01 -1.16704471e-01 1.42302085e-02
-3.01202804e-01 -3.27714652e-01 8.45600486e-01 -5.17111719e-01
8.71268511e-01 -2.09132099e+00 -1.68424413e-01 3.23279917e-01
-1.32684987e-02 4.98752981e-01 1.55361826e-02 5.15659630e-01
-2.47792333e-01 -7.00568827e-03 -4.01409745e-01 -1.13520632e-02
-7.39947185e-02 3.72980744e-01 4.05754328e-01 1.21344663e-01
4.68143493e-01 2.51513779e-01 -6.76661730e-01 -7.24371374e-02
3.49596262e-01 3.94859940e-01 -5.97074032e-01 8.39630961e-02
-3.44529986e-01 3.70081723e-01 -2.08240047e-01 7.30896950e-01
6.56213820e-01 2.67465729e-02 -6.30171746e-02 -2.41350338e-01
8.81558005e-03 1.37306971e-03 -1.12917805e+00 1.44866979e+00
-7.14748800e-01 6.93074346e-01 -2.25685075e-01 -1.60181022e+00
1.30575132e+00 3.42488796e-01 5.80868542e-01 -6.10205710e-01
2.49418080e-01 3.46799821e-01 3.89354259e-01 -6.63285613e-01
3.10288072e-01 -2.20033273e-01 2.52074420e-01 2.67056912e-01
1.81675524e-01 -3.97789210e-01 7.16918111e-01 -3.94285858e-01
1.10764372e+00 -6.71751946e-02 -1.05075851e-01 -1.59031540e-01
4.90720481e-01 -6.92866817e-02 5.76485217e-01 4.56837058e-01
2.46420622e-01 5.24094045e-01 5.49101830e-01 -5.69210231e-01
-1.19172180e+00 -6.21226788e-01 -2.83595979e-01 3.62387061e-01
-2.31129855e-01 5.66344745e-02 -7.93086112e-01 -7.83163249e-01
-7.68460557e-02 6.41072392e-01 -6.37465596e-01 -3.98434103e-01
-2.74727285e-01 -8.69977713e-01 -1.35950029e-01 3.08715075e-01
4.62708265e-01 -1.45703316e+00 -6.92840159e-01 4.44298863e-01
4.83965486e-01 -6.21232808e-01 3.22454959e-01 6.72871768e-01
-1.29679692e+00 -1.25288403e+00 -6.01213276e-01 -8.14144850e-01
1.04543412e+00 -1.12371624e-01 1.00485790e+00 5.66060722e-01
-6.58720255e-01 6.61634952e-02 -6.53202832e-01 -7.71793544e-01
-6.81245804e-01 9.78015587e-02 -3.96886542e-02 -9.61547568e-02
4.55707043e-01 -6.63518667e-01 -3.89590085e-01 1.26741022e-01
-1.19796824e+00 -2.99403548e-01 8.27135623e-01 1.13583946e+00
1.21650457e-01 8.58657897e-01 1.05167449e+00 -9.34797645e-01
6.82098508e-01 -1.91596270e-01 -5.55502594e-01 8.49726051e-02
-7.22998798e-01 -4.89729606e-02 7.78953671e-01 -5.24847627e-01
-9.06041384e-01 -1.15218516e-02 -4.93979573e-01 -8.78127888e-02
-4.48503762e-01 5.23145080e-01 -3.43351722e-01 -1.20688386e-01
7.35565662e-01 -8.01442750e-03 2.46567085e-01 -5.01847386e-01
-3.47580492e-01 7.46400297e-01 -6.93852901e-02 -2.97978282e-01
7.73511112e-01 5.34950346e-02 -6.67345449e-02 -1.03841960e+00
-6.20446503e-01 -1.92058757e-01 -9.00864244e-01 -2.99461395e-01
5.46223402e-01 -2.89094865e-01 -4.47067678e-01 6.04629993e-01
-8.83034587e-01 -3.47502023e-01 -7.49018192e-01 3.56702894e-01
-2.07229406e-01 3.97483438e-01 -3.80482078e-01 -8.11114967e-01
-7.16451779e-02 -1.21372604e+00 6.22815728e-01 4.15065408e-01
1.17599174e-01 -1.17718661e+00 -3.02652091e-01 2.95269877e-01
4.42346811e-01 3.00245941e-01 1.19979286e+00 -5.49119830e-01
-4.01175052e-01 -5.63806653e-01 1.44968927e-01 1.01659441e+00
5.80577612e-01 1.04145385e-01 -9.69967604e-01 -4.31128472e-01
3.05129468e-01 -4.47732776e-01 7.67815173e-01 3.11669320e-01
1.25532651e+00 3.07505727e-01 -2.76593834e-01 -4.99014780e-02
1.42607713e+00 5.90327919e-01 7.43212163e-01 4.24359858e-01
2.86202312e-01 6.28885031e-01 7.74320364e-01 2.82904148e-01
-2.60692984e-01 3.62847507e-01 7.51370907e-01 -3.63167048e-01
-1.38691824e-03 2.23305538e-01 1.21922292e-01 8.97042334e-01
-2.94515371e-01 -3.36192966e-01 -6.06166959e-01 7.00588226e-01
-1.33644032e+00 -7.01170683e-01 -1.09530687e-01 2.15374446e+00
6.85470581e-01 5.50740957e-01 -1.38391599e-01 1.04913342e+00
5.29066324e-01 -3.47926378e-01 -5.53361952e-01 -6.18387818e-01
1.70230597e-01 4.76092696e-01 8.17121863e-02 -8.51228982e-02
-7.31512785e-01 3.84958297e-01 5.81478548e+00 5.64367473e-01
-1.31074750e+00 -9.72703397e-02 4.11726207e-01 1.72613755e-01
-2.12618873e-01 -1.87674925e-01 -4.95480329e-01 4.68756855e-01
7.25528061e-01 3.82207483e-01 2.18826905e-01 7.60322750e-01
1.26300767e-01 -5.62306345e-01 -1.23448336e+00 6.99566305e-01
2.64809094e-02 -9.23655689e-01 -1.22564539e-01 1.98371917e-01
7.73543298e-01 -5.05270243e-01 -2.62919754e-01 3.85530949e-01
-6.96629435e-02 -9.55782413e-01 1.34488970e-01 6.61220133e-01
3.80120665e-01 -8.55106771e-01 1.21900177e+00 5.68236470e-01
-5.65759301e-01 -5.00702798e-01 -8.91939402e-01 -1.58376008e-01
8.77083186e-03 1.33383834e+00 -1.09753811e+00 7.29083419e-01
6.90378487e-01 5.57125807e-01 -5.97318351e-01 1.36972690e+00
-3.11782155e-02 5.71073234e-01 -2.85930187e-01 -2.10117489e-01
-3.77720855e-02 -2.92706668e-01 1.49291918e-01 5.24663329e-01
8.43621373e-01 -4.84272867e-01 -8.96510556e-02 8.82449031e-01
1.55709371e-01 -1.95686799e-02 -8.48797858e-01 -9.89165902e-02
-2.63411377e-04 1.16312861e+00 -9.66747820e-01 2.32427623e-02
-4.33965921e-01 8.91944647e-01 -3.39402864e-03 2.79092528e-02
-4.16094452e-01 -5.18047452e-01 1.92053869e-01 3.74901891e-01
3.93332124e-01 -1.25396485e-02 -9.14246514e-02 -8.14004898e-01
9.34094116e-02 -8.80698144e-01 -4.05310048e-03 -6.51715457e-01
-1.33280528e+00 5.55024087e-01 -3.13095570e-01 -1.27535033e+00
-2.57635474e-01 -9.76664245e-01 -7.15934575e-01 5.88528693e-01
-1.31661034e+00 -9.04393911e-01 -2.81108081e-01 1.26603141e-01
6.86625600e-01 -4.82360095e-01 7.56351173e-01 6.13761485e-01
-5.60736835e-01 3.81210744e-01 2.41735667e-01 -1.48635179e-01
5.23012519e-01 -1.22802293e+00 -1.28100798e-01 7.22832382e-01
2.32120067e-01 3.09676230e-01 8.48797798e-01 -5.52387178e-01
-1.12946677e+00 -7.70271719e-01 5.38230956e-01 -5.98625615e-02
2.75042892e-01 -3.60041827e-01 -1.18380380e+00 1.48191854e-01
3.07157248e-01 -3.95246483e-02 4.11174148e-01 1.38710946e-01
4.51893002e-01 -4.64148879e-01 -1.37305403e+00 1.30309448e-01
6.66162670e-01 -1.18754499e-01 -4.30975229e-01 3.23832124e-01
2.60442793e-01 9.04822722e-02 -1.04505014e+00 5.68180323e-01
2.49685958e-01 -1.24194205e+00 5.14880002e-01 -1.63277328e-01
4.13355529e-01 -2.92952806e-01 2.42423683e-01 -1.71965325e+00
-1.93308368e-02 8.75974894e-02 1.19512074e-01 1.36750281e+00
6.15343392e-01 -7.12376356e-01 9.02500868e-01 3.62308592e-01
-4.04848099e-01 -9.62238193e-01 -7.01315165e-01 -9.15739477e-01
-4.26881295e-03 -2.54630983e-01 6.03730202e-01 9.19834197e-01
-2.12206170e-01 -1.01708490e-02 -8.74639228e-02 -2.73713693e-02
2.48464406e-01 -3.33157554e-02 5.71662724e-01 -1.86524379e+00
-3.29883665e-01 -1.61087349e-01 -6.68027878e-01 -5.71923912e-01
-1.28421813e-01 -6.65219307e-01 1.91784963e-01 -1.72429395e+00
-1.88690051e-01 -8.62067640e-01 -2.31834739e-01 5.33989668e-01
2.73723245e-01 2.27037951e-01 -1.17441230e-01 -2.86895663e-01
2.13015974e-02 5.32941818e-01 1.54601002e+00 -3.95152062e-01
-2.25209907e-01 4.20335382e-01 -4.68206227e-01 4.85507369e-01
1.02642751e+00 -4.32495743e-01 -6.62303567e-01 -2.11801946e-01
3.17986727e-01 -1.87587261e-01 1.13246113e-01 -1.51977170e+00
-1.26860857e-01 2.17789501e-01 6.26965702e-01 -5.01195848e-01
2.45396674e-01 -1.22508562e+00 1.30392715e-01 4.99438345e-01
1.44905970e-01 -4.41243261e-01 1.39173925e-01 3.00301433e-01
-5.01052737e-01 -1.04952955e+00 6.05484068e-01 -3.13295722e-01
-5.68343580e-01 -2.89834347e-02 -4.16397810e-01 -5.46550512e-01
9.74482358e-01 -8.48332405e-01 1.48035362e-01 -5.21420799e-02
-1.01480734e+00 -6.50009140e-02 5.99721849e-01 3.55078757e-01
5.35492063e-01 -1.08144569e+00 -2.38145813e-01 5.10017991e-01
2.86072250e-02 3.84313017e-01 3.36291969e-01 6.24704421e-01
-4.88420010e-01 4.00133021e-02 -6.53692007e-01 -7.35462964e-01
-1.02283978e+00 7.43549407e-01 3.09827268e-01 -2.68996328e-01
-6.04964197e-01 5.27713537e-01 -2.24616110e-01 -4.85014290e-01
4.73896153e-02 -6.19608581e-01 -6.24045491e-01 1.83054179e-01
2.02330068e-01 1.57127261e-01 6.35848105e-01 8.85398909e-02
1.94615871e-01 4.35584962e-01 -5.73956780e-02 5.61422467e-01
1.80451643e+00 2.84686714e-01 -1.44646510e-01 6.38616920e-01
8.38710964e-01 -2.11962298e-01 -1.22212529e+00 3.02446902e-01
1.32735699e-01 -4.32640284e-01 1.12202140e-02 -7.80239880e-01
-1.44628668e+00 1.15368927e+00 9.74919021e-01 7.34437406e-01
1.45145273e+00 -3.15353155e-01 6.58817649e-01 5.87042689e-01
5.63574135e-01 -1.44286013e+00 2.12439448e-01 2.22657546e-01
8.31836224e-01 -1.26220524e+00 -1.33862361e-01 -3.83732855e-01
-1.26310870e-01 1.45549035e+00 6.62801206e-01 -1.78013921e-01
5.92602730e-01 2.61903793e-01 -9.51887220e-02 -3.43656898e-01
-4.71791506e-01 -5.45097403e-02 7.56838992e-02 9.25450325e-01
4.53159750e-01 -3.06613505e-01 -3.56135130e-01 3.14811230e-01
-2.41960287e-02 2.39992440e-01 7.84897745e-01 1.07805967e+00
-6.26122713e-01 -1.61663902e+00 -3.29068512e-01 9.08380866e-01
-3.52035195e-01 3.15531790e-01 -1.10013902e-01 1.05912435e+00
7.01244414e-01 9.97867048e-01 -1.84355304e-02 -2.13659272e-01
4.65319812e-01 4.21022587e-02 6.95185483e-01 -8.71200681e-01
-4.60278064e-01 -2.80629903e-01 6.47731349e-02 -1.49510741e-01
-4.31265384e-01 -5.91865838e-01 -1.02827930e+00 1.61592528e-01
-7.62589693e-01 1.43381044e-01 8.95251513e-01 1.02598190e+00
-5.62413633e-02 1.05431831e+00 6.16824150e-01 -1.03141725e+00
-3.86107028e-01 -1.34340262e+00 -8.43586862e-01 4.16623026e-01
2.24342927e-01 -9.37841177e-01 -3.23088229e-01 1.44627005e-01] | [7.32025146484375, 1.989988088607788] |
0ae784d0-1e1a-45f8-9648-2ac32752ff2d | multiview-detection-with-cardboard-human | 2207.02013 | null | https://arxiv.org/abs/2207.02013v5 | https://arxiv.org/pdf/2207.02013v5.pdf | Multiview Detection with Cardboard Human Modeling | Multiview detection uses multiple calibrated cameras with overlapping fields of views to locate occluded pedestrians. In this field, existing methods typically adopt a ``human modeling - aggregation'' strategy. To find robust pedestrian representations, some intuitively incorporate 2D perception results from each frame, while others use entire frame features projected to the ground plane. However, the former does not consider the human appearance and leads to many ambiguities, and the latter suffers from projection errors due to the lack of accurate height of the human torso and head. In this paper, we propose a new pedestrian representation scheme based on human point clouds modeling. Specifically, using ray tracing for holistic human depth estimation, we model pedestrians as upright, thin cardboard point clouds on the ground. Then, we aggregate the point clouds of the pedestrian cardboard across multiple views for a final decision. Compared with existing representations, the proposed method explicitly leverages human appearance and reduces projection errors significantly by relatively accurate height estimation. On four standard evaluation benchmarks, the proposed method achieves very competitive results. Our code and data will be released at https://github.com/ZichengDuan/MvCHM. | ['Chuong Nguyen', 'Liang Zheng', 'Zicheng Duan', 'Jiahao Ma'] | 2022-07-05 | null | null | null | null | ['multiview-detection'] | ['computer-vision'] | [-2.35830247e-01 -2.44526327e-01 3.63219827e-02 -2.94838190e-01
-5.04765093e-01 -4.24028069e-01 4.52648252e-01 9.57200229e-02
-1.11548200e-01 4.18864071e-01 1.96813270e-01 2.43077464e-02
6.63836598e-01 -9.66641426e-01 -7.45337486e-01 -4.77587998e-01
4.57072914e-01 4.56782550e-01 6.64895594e-01 -2.90509276e-02
6.38171509e-02 2.17207089e-01 -1.64021671e+00 7.70626888e-02
7.04698980e-01 9.50009644e-01 7.50510842e-02 4.84068602e-01
1.54320210e-01 5.36321580e-01 -4.49934363e-01 -8.78918588e-01
4.61691737e-01 1.11911342e-01 -1.74479321e-01 5.50043404e-01
8.46190631e-01 -7.73699164e-01 -4.97371912e-01 1.04256523e+00
4.80268717e-01 3.25485431e-02 4.72744912e-01 -1.40614808e+00
-5.37601829e-01 -1.00682028e-01 -1.15712535e+00 -1.46501586e-01
8.16637397e-01 2.62065262e-01 6.67498350e-01 -1.21534777e+00
4.32810903e-01 1.57026052e+00 7.84804285e-01 3.04031283e-01
-8.50389719e-01 -7.82128870e-01 5.85608363e-01 2.40536973e-01
-1.67146158e+00 -4.54012334e-01 8.58750403e-01 -5.71505845e-01
4.64833051e-01 3.62856984e-01 1.05074859e+00 9.97457981e-01
-1.16576090e-01 8.75160158e-01 9.08164799e-01 -3.48839164e-01
4.44573984e-02 2.10323870e-01 2.44003221e-01 8.66787672e-01
7.75914907e-01 8.76388103e-02 -4.09092695e-01 -3.03440243e-01
8.86193633e-01 5.48719049e-01 -2.41150618e-01 -7.62188554e-01
-9.51541603e-01 6.05720341e-01 5.18534958e-01 -3.84107769e-01
-2.03381523e-01 2.72527654e-02 1.63979471e-01 -5.11931837e-01
4.59369332e-01 -5.56002617e-01 7.71454275e-02 2.80099690e-01
-7.32575893e-01 4.96867508e-01 3.06879997e-01 1.26303518e+00
8.04831862e-01 -1.75451130e-01 1.44720316e-01 6.35637462e-01
7.09104002e-01 7.49225855e-01 -6.12821206e-02 -1.21560967e+00
7.38315761e-01 8.09400976e-01 4.37850535e-01 -1.33467507e+00
-1.40760079e-01 -9.44340080e-02 -8.52399051e-01 3.07430804e-01
6.21129692e-01 -9.77041349e-02 -6.66872442e-01 1.29629803e+00
7.86940932e-01 1.52195156e-01 -3.46692771e-01 1.18979907e+00
8.75930190e-01 5.32655895e-01 -7.09171174e-03 1.82866767e-01
1.62203228e+00 -1.03040743e+00 -4.08984661e-01 -4.26030725e-01
6.40764087e-02 -7.58373499e-01 6.40002072e-01 4.98559564e-01
-1.22760081e+00 -6.42488182e-01 -1.02289212e+00 -2.53123790e-01
-1.49654061e-01 2.59794980e-01 5.09976923e-01 8.43049943e-01
-8.96831393e-01 1.28321856e-01 -8.08121979e-01 -5.05236328e-01
3.65904599e-01 8.84238631e-02 -3.31375778e-01 -4.81106669e-01
-8.04791868e-01 6.99709117e-01 3.63761634e-02 3.65935043e-02
-5.95890522e-01 -3.91841263e-01 -9.66027200e-01 -1.40275821e-01
3.35249513e-01 -1.22722077e+00 1.07060790e+00 -5.91085553e-01
-1.23377037e+00 8.59497249e-01 -5.18472672e-01 -2.76194632e-01
9.16208267e-01 -4.93343532e-01 -1.00585878e-01 1.28629372e-01
1.77495167e-01 4.89949167e-01 5.93758941e-01 -1.70625758e+00
-8.34805369e-01 -8.14310431e-01 2.42113784e-01 3.78625870e-01
1.36339469e-02 5.52795921e-03 -9.31865394e-01 -5.06940544e-01
4.58571941e-01 -8.69887352e-01 -3.54026854e-01 4.11300331e-01
-5.39284229e-01 7.29165524e-02 6.97391391e-01 -7.79509842e-01
9.13280249e-01 -1.94360793e+00 -5.97514957e-03 4.08606492e-02
5.25271297e-01 4.23374996e-02 4.06548798e-01 1.79097638e-01
3.47167462e-01 -2.01749012e-01 7.76731893e-02 -7.89614379e-01
5.16560022e-03 -2.81655751e-02 -1.05665170e-01 7.63113141e-01
-3.35454851e-01 5.15577435e-01 -7.65742183e-01 -6.99122548e-01
6.92200780e-01 7.82642424e-01 -5.02010643e-01 1.33896694e-01
2.77850181e-01 2.64607251e-01 -3.91160756e-01 9.83066320e-01
1.07730949e+00 -2.18952253e-01 1.60617009e-02 -2.92560041e-01
-2.29044244e-01 -1.71610758e-01 -1.45129228e+00 1.52658653e+00
-3.61838639e-02 2.58246899e-01 7.83234537e-02 -4.21811640e-01
8.01017642e-01 2.49899819e-01 5.27096152e-01 -3.32116693e-01
4.12445404e-02 -3.14332619e-02 -5.84320188e-01 -5.58229201e-02
4.99848753e-01 2.14289576e-01 1.68308690e-01 -7.52245188e-02
-3.80795777e-01 1.73120454e-01 -5.93529865e-02 1.58366278e-01
6.15891218e-01 2.40480155e-01 5.82528651e-01 1.77593574e-01
5.54464817e-01 -1.14764862e-01 8.89332235e-01 4.14915085e-01
-4.55959022e-01 1.02388144e+00 2.19808854e-02 -8.39717150e-01
-1.03958440e+00 -1.36189985e+00 4.41622138e-02 7.37143159e-01
7.23938286e-01 -7.16605544e-01 -7.98343778e-01 -4.01252121e-01
1.39373228e-01 4.28690702e-01 -3.51997018e-01 1.71349600e-01
-6.42372072e-01 -6.24686718e-01 1.69470772e-01 7.67938375e-01
6.61862671e-01 -2.04124451e-01 -8.66837800e-01 -6.85313940e-02
-5.03170192e-01 -1.28920341e+00 -5.18814266e-01 -6.88806295e-01
-6.42033637e-01 -1.22323465e+00 -1.05511975e+00 -4.62587327e-01
6.96831822e-01 9.79220688e-01 9.69144046e-01 1.23809181e-01
-2.52927452e-01 4.86843973e-01 -1.72881916e-01 -4.12947983e-01
3.72841448e-01 -5.00473082e-01 3.31064373e-01 1.05919667e-01
4.15749401e-01 -4.59898859e-01 -1.10114527e+00 4.99720156e-01
-3.37082632e-02 4.34833378e-01 3.73519182e-01 3.49144697e-01
7.13034093e-01 -8.18743557e-02 -3.19304913e-01 -5.27748048e-01
-2.33239859e-01 -4.32775944e-01 -7.58310854e-01 1.38758838e-01
-1.17185682e-01 -7.16172874e-01 3.36960226e-01 -1.10713571e-01
-1.01313281e+00 3.21064681e-01 1.24345161e-01 -7.59885371e-01
-4.30815071e-01 -4.33793217e-01 -6.32972777e-01 1.79442186e-02
2.81605899e-01 8.02434608e-02 -3.59086692e-01 -5.19378781e-01
2.90660560e-01 4.15790677e-01 6.11272871e-01 -6.71565056e-01
7.80686915e-01 8.57911527e-01 -1.20294079e-01 -7.93305695e-01
-6.95181429e-01 -6.45267963e-01 -8.91054153e-01 -5.61354518e-01
9.14011776e-01 -1.25503170e+00 -8.51401031e-01 3.86862636e-01
-1.47440100e+00 2.46847197e-01 1.29125491e-01 4.89748150e-01
-3.36505234e-01 8.02473903e-01 -6.13448858e-01 -1.12978113e+00
-1.73094437e-01 -1.29314363e+00 1.47879219e+00 2.21757993e-01
1.68847088e-02 -5.74156582e-01 -1.32756278e-01 6.86909497e-01
-1.09747916e-01 4.99508470e-01 3.38287920e-01 1.14641614e-01
-8.88100982e-01 -3.81951034e-01 -3.24051291e-01 -7.28669688e-02
-1.00978747e-01 1.07472651e-01 -1.06185567e+00 -2.35912845e-01
-2.14155346e-01 1.90668091e-01 6.61656618e-01 6.79750383e-01
8.36191118e-01 -2.12433472e-01 -6.73414528e-01 6.73166037e-01
1.28788197e+00 -6.03226833e-02 4.57971364e-01 3.77595127e-01
1.07475293e+00 7.52172709e-01 5.57676554e-01 6.96444333e-01
1.08058906e+00 1.00281632e+00 4.21269864e-01 -1.12038799e-01
-1.92081854e-01 -4.26923990e-01 4.48936164e-01 5.58724582e-01
-5.58226287e-01 -1.15767740e-01 -9.77266848e-01 1.98610365e-01
-1.90742922e+00 -1.01194656e+00 -5.48235536e-01 2.50758505e+00
4.57024127e-02 6.48432001e-02 5.61860681e-01 5.30114025e-02
1.01959205e+00 -3.04105300e-02 -4.37952220e-01 4.01092649e-01
-1.07768876e-02 -5.11333942e-01 5.51798820e-01 6.22172475e-01
-1.21548605e+00 7.41506219e-01 5.46977711e+00 3.43790442e-01
-5.58101475e-01 2.79431671e-01 5.18614590e-01 -1.13261819e-01
1.17136925e-01 1.34698004e-01 -1.14980423e+00 5.99151134e-01
2.75379807e-01 4.05377336e-02 -3.13922414e-03 9.74551916e-01
1.47557378e-01 -1.57258883e-01 -9.59578574e-01 1.43942761e+00
1.98848709e-01 -9.07697201e-01 3.72766592e-02 2.66732216e-01
5.26497841e-01 -2.88578361e-01 2.35931575e-02 3.89030799e-02
3.10303301e-01 -6.70563340e-01 1.17627001e+00 6.03057206e-01
3.88102740e-01 -7.46749699e-01 5.52786648e-01 4.47008371e-01
-1.87399268e+00 1.15876883e-01 -6.64206982e-01 -1.97567135e-01
3.72022927e-01 5.72818279e-01 -4.17154312e-01 6.38609052e-01
9.19751465e-01 6.86912715e-01 -8.34519267e-01 1.11284959e+00
-1.83628276e-02 6.66795149e-02 -3.28977317e-01 4.67536122e-01
-2.68098891e-01 -3.70547652e-01 5.63420415e-01 1.04109442e+00
4.22452003e-01 2.61147261e-01 6.56498313e-01 7.41125464e-01
3.08323681e-01 1.12361833e-01 -6.72719181e-01 8.01878929e-01
6.36164486e-01 1.15771985e+00 -7.01707661e-01 -5.50881386e-01
-8.22381318e-01 1.03963387e+00 2.80866086e-01 4.03421789e-01
-9.71787333e-01 2.60973305e-01 6.91833735e-01 6.03859127e-01
1.30130664e-01 -3.15676987e-01 -3.05513382e-01 -1.57915354e+00
2.90710896e-01 -6.18337870e-01 3.70487571e-01 -7.89398968e-01
-1.19772971e+00 4.44048464e-01 1.57659546e-01 -1.49702585e+00
-2.85691451e-02 -5.72010815e-01 -4.39199239e-01 7.72286177e-01
-1.11086059e+00 -1.52184629e+00 -7.12033153e-01 6.83019280e-01
6.98882282e-01 1.27253026e-01 4.42950636e-01 4.54236835e-01
-7.30807722e-01 4.28847045e-01 -3.37406904e-01 1.00816891e-01
5.84693491e-01 -1.16259992e+00 6.37331605e-01 9.39351141e-01
-1.90406233e-01 7.60120690e-01 6.68022215e-01 -6.79743886e-01
-1.29157519e+00 -1.14108217e+00 5.93946218e-01 -8.15899432e-01
-2.11587157e-02 -4.25021231e-01 -7.00012088e-01 9.05663967e-01
1.01822009e-02 1.48857430e-01 6.18936718e-01 -4.32969071e-02
-4.53319073e-01 -7.26404265e-02 -1.04727769e+00 6.05017483e-01
1.12769353e+00 1.18679740e-02 -3.75536442e-01 2.18039408e-01
4.25497591e-01 -5.93146265e-01 -6.46627188e-01 2.30743855e-01
8.05761099e-01 -1.27809930e+00 1.58407092e+00 -3.50556746e-02
1.18361719e-01 -6.55542970e-01 -6.06183648e-01 -7.73101270e-01
-3.90934050e-01 -5.09463698e-02 -4.61098969e-01 1.25672781e+00
-2.97079682e-01 -5.11237442e-01 8.95631313e-01 7.75708437e-01
3.61423343e-02 -6.45857036e-01 -8.00761163e-01 -4.87436026e-01
-2.03870296e-01 -2.95863092e-01 6.18131340e-01 6.11123383e-01
-2.84525424e-01 1.88520893e-01 -5.72823584e-01 6.76701903e-01
1.29575598e+00 1.44251794e-01 1.26419318e+00 -1.40195513e+00
-2.40294635e-01 -2.88969964e-01 -6.77778780e-01 -1.18909848e+00
-2.13876903e-01 -4.42268819e-01 -2.52210945e-01 -1.59697294e+00
4.99543846e-01 -1.98144838e-01 2.32769474e-01 6.64796084e-02
-3.70268941e-01 2.66217977e-01 5.23698747e-01 2.20491037e-01
-7.86573648e-01 4.66662318e-01 9.88022625e-01 4.13928889e-02
2.05477640e-01 1.52182981e-01 -5.90892434e-01 1.41618538e+00
6.49394274e-01 -9.85103771e-02 -1.52784482e-01 -6.91811740e-01
-2.38479048e-01 1.03519656e-01 9.97450292e-01 -1.33991778e+00
3.76854092e-01 -1.11547992e-01 1.01310837e+00 -1.02646792e+00
1.03252625e+00 -7.92275190e-01 3.37251872e-01 4.24693465e-01
3.84568691e-01 3.13544840e-01 -7.73425996e-02 7.88824141e-01
1.03912964e-01 1.50989607e-01 8.24043512e-01 -5.13671517e-01
-6.28310680e-01 5.85731626e-01 -2.11086161e-02 -2.66730368e-01
1.14045191e+00 -3.84827495e-01 -2.49972776e-01 -3.79556268e-01
-4.51396018e-01 2.18958989e-01 1.07267606e+00 3.18116009e-01
9.05188024e-01 -1.47811580e+00 -6.15813315e-01 1.29519165e-01
6.52573407e-02 2.27336362e-01 3.54194790e-01 7.27988601e-01
-5.44521987e-01 4.12268281e-01 6.88550919e-02 -9.00859594e-01
-1.51000619e+00 5.94988167e-01 3.53178859e-01 1.57642245e-01
-8.75365436e-01 6.28108919e-01 7.44525015e-01 -3.37201744e-01
2.27058932e-01 -2.05857992e-01 -2.73931753e-02 -1.85447335e-01
7.53485084e-01 8.07868600e-01 -2.86279202e-01 -1.17697263e+00
-4.24003839e-01 9.66430306e-01 6.42180815e-02 -1.34634674e-01
8.34945202e-01 -5.39996743e-01 2.10974097e-01 2.77935296e-01
8.52436841e-01 2.35475808e-01 -1.47099400e+00 -2.78830349e-01
-4.67095494e-01 -1.03396702e+00 -1.81836411e-01 -2.14361548e-01
-1.10801864e+00 9.55514014e-01 6.94069624e-01 -3.62043828e-01
8.09086144e-01 -1.27785310e-01 8.60008657e-01 5.30418493e-02
9.27555919e-01 -7.93162405e-01 -1.54361472e-01 5.43492399e-02
6.34757280e-01 -1.35561490e+00 3.23408484e-01 -9.11731720e-01
-6.64846420e-01 9.20946956e-01 1.02672887e+00 -6.39222339e-02
3.90814096e-01 5.64449169e-02 -1.58283353e-01 -2.79620737e-02
-3.64020824e-01 -2.32714564e-01 1.94314331e-01 7.97030628e-01
2.79358208e-01 2.99637228e-01 1.74695179e-01 7.20766604e-01
-2.26707742e-01 -2.02844769e-01 2.77647346e-01 8.14451694e-01
-5.67017615e-01 -8.99490952e-01 -1.09300673e+00 1.19891025e-01
-1.52443349e-01 2.38205925e-01 -3.68468672e-01 6.64074719e-01
4.32201087e-01 1.02320623e+00 -2.42953356e-02 -3.39117706e-01
6.63482606e-01 -2.40510553e-01 6.57349348e-01 -5.28475463e-01
-9.64150857e-03 2.68662155e-01 -1.74382225e-01 -6.31956756e-01
-2.87329555e-01 -9.83914495e-01 -9.64733422e-01 -6.77962661e-01
-2.14685932e-01 -2.72781402e-01 2.45838434e-01 6.13011897e-01
1.90000921e-01 9.20015797e-02 3.95163774e-01 -1.43486083e+00
-9.86037180e-02 -5.68098128e-01 -4.92156267e-01 3.99849683e-01
2.49527976e-01 -1.14617097e+00 -3.15964296e-02 1.57364070e-01] | [7.437475681304932, -1.1120645999908447] |
6e799f1f-ba76-4fab-a2bc-06ff5a479271 | bilingual-lexicon-induction-through | 1907.10761 | null | https://arxiv.org/abs/1907.10761v1 | https://arxiv.org/pdf/1907.10761v1.pdf | Bilingual Lexicon Induction through Unsupervised Machine Translation | A recent research line has obtained strong results on bilingual lexicon induction by aligning independently trained word embeddings in two languages and using the resulting cross-lingual embeddings to induce word translation pairs through nearest neighbor or related retrieval methods. In this paper, we propose an alternative approach to this problem that builds on the recent work on unsupervised machine translation. This way, instead of directly inducing a bilingual lexicon from cross-lingual embeddings, we use them to build a phrase-table, combine it with a language model, and use the resulting machine translation system to generate a synthetic parallel corpus, from which we extract the bilingual lexicon using statistical word alignment techniques. As such, our method can work with any word embedding and cross-lingual mapping technique, and it does not require any additional resource besides the monolingual corpus used to train the embeddings. When evaluated on the exact same cross-lingual embeddings, our proposed method obtains an average improvement of 6 accuracy points over nearest neighbor and 4 points over CSLS retrieval, establishing a new state-of-the-art in the standard MUSE dataset. | ['Mikel Artetxe', 'Gorka Labaka', 'Eneko Agirre'] | 2019-07-24 | bilingual-lexicon-induction-through-1 | https://aclanthology.org/P19-1494 | https://aclanthology.org/P19-1494.pdf | acl-2019-7 | ['unsupervised-machine-translation'] | ['natural-language-processing'] | [ 3.53877270e-03 -1.58605039e-01 -6.15500510e-01 -1.54350415e-01
-1.25409484e+00 -9.21615601e-01 9.32354510e-01 2.47987673e-01
-9.62516725e-01 7.82104850e-01 3.55914623e-01 -5.55945635e-01
2.23664761e-01 -8.23357105e-01 -7.82480597e-01 -5.55810034e-01
5.11460066e-01 8.88370275e-01 -2.43479405e-02 -4.55838978e-01
2.28895590e-01 1.85751230e-01 -1.00773871e+00 -5.62023371e-02
9.91845787e-01 3.66841912e-01 4.91339378e-02 1.32177114e-01
-3.65018874e-01 1.11252472e-01 -6.02461360e-02 -7.49615610e-01
4.07134175e-01 -5.85567236e-01 -9.17035043e-01 -2.17717707e-01
3.18555087e-01 2.76691169e-02 -9.65312496e-02 9.84802783e-01
7.04187393e-01 -2.01554462e-01 6.62232220e-01 -6.63512945e-01
-1.05199921e+00 7.30452359e-01 -5.47006726e-01 1.58214912e-01
5.50387025e-01 -2.04436243e-01 1.27694368e+00 -1.24778736e+00
8.55854094e-01 1.03287196e+00 6.01470053e-01 3.41030359e-01
-1.55901456e+00 -6.48515344e-01 -2.59019256e-01 2.69974232e-01
-1.59185898e+00 -4.86855477e-01 5.77919960e-01 -2.33885393e-01
1.10265338e+00 -1.91014424e-01 5.91517866e-01 1.13890529e+00
1.94940135e-01 3.93030226e-01 1.37152660e+00 -1.17530000e+00
-5.73936179e-02 6.17257297e-01 -1.04926899e-02 5.38021624e-01
2.40597144e-01 -1.53976621e-03 -5.12638092e-01 -3.91796865e-02
3.90864879e-01 -3.52550179e-01 -1.34206057e-01 -5.08960962e-01
-1.51004088e+00 1.00350630e+00 1.62465617e-01 7.96283245e-01
-1.74994156e-01 -2.65110761e-01 3.83165538e-01 5.19010484e-01
7.07923234e-01 6.76700294e-01 -3.09996307e-01 1.06793836e-01
-9.84327137e-01 1.14878165e-02 7.40616262e-01 8.88883889e-01
1.12740195e+00 -4.06540692e-01 -2.34583602e-03 9.80484128e-01
2.17436343e-01 6.79017782e-01 9.73182321e-01 -2.85399020e-01
6.17947638e-01 3.92574877e-01 -4.31547612e-02 -8.13654363e-01
-1.24111903e-04 -2.34349310e-01 -2.91815490e-01 -3.08830440e-01
2.85438955e-01 2.93568112e-02 -5.90897322e-01 1.78889644e+00
1.19185649e-01 4.14040536e-02 3.77680749e-01 6.44734025e-01
3.13126862e-01 6.88115120e-01 -1.51479021e-01 -4.81376462e-02
1.46245158e+00 -1.06702113e+00 -4.90642190e-01 -2.72488058e-01
8.98577631e-01 -1.32389486e+00 1.20252419e+00 -1.86387692e-02
-9.45822477e-01 -5.98012686e-01 -1.39818287e+00 -1.49842784e-01
-6.29760265e-01 1.47173226e-01 3.49441737e-01 4.01793480e-01
-1.22336459e+00 3.76756698e-01 -6.32637680e-01 -7.39588618e-01
-2.98706383e-01 3.43035579e-01 -7.06436038e-01 -4.10485387e-01
-1.51519179e+00 1.41862917e+00 5.18590033e-01 -2.17643023e-01
-4.18994933e-01 -4.23224628e-01 -9.43416774e-01 -4.11942720e-01
-6.91228509e-02 -5.77044189e-01 7.91696787e-01 -8.36131096e-01
-1.57916987e+00 1.29613829e+00 -2.27858067e-01 -2.99584657e-01
2.12080225e-01 -2.27812991e-01 -4.86167431e-01 -1.55121028e-01
5.41807353e-01 7.05104709e-01 3.55536014e-01 -8.96398544e-01
-3.84682387e-01 -3.88717502e-01 -7.12836683e-02 3.69239867e-01
-5.89550138e-01 2.41008073e-01 -7.30392158e-01 -5.51202059e-01
1.16748199e-01 -1.18804061e+00 -7.93671310e-02 -5.65751076e-01
-1.21092893e-01 -4.17992830e-01 2.72415698e-01 -6.59015417e-01
1.06233120e+00 -1.99165606e+00 4.39219356e-01 1.94700226e-01
-4.14234042e-01 3.31738532e-01 -6.22757256e-01 7.79767513e-01
-8.86062831e-02 1.48567066e-01 -2.32477665e-01 -3.76317352e-01
2.85624303e-02 3.13590258e-01 -3.68450105e-01 5.12569606e-01
3.46278608e-01 1.08622050e+00 -1.07053268e+00 -6.30702019e-01
4.89074364e-03 4.68031228e-01 -5.63333809e-01 1.26567021e-01
1.07241429e-01 3.49607348e-01 -2.49237925e-01 2.26800293e-01
5.32227695e-01 3.16412508e-01 6.47386611e-01 -9.11506340e-02
-1.28710508e-01 7.49685884e-01 -7.79479980e-01 2.27053928e+00
-1.08514500e+00 6.09348655e-01 -6.32495165e-01 -1.20288384e+00
1.07017231e+00 5.64800382e-01 1.94388226e-01 -9.89931643e-01
4.29720841e-02 8.52325857e-01 -6.73509985e-02 -3.56611758e-01
3.48142743e-01 -2.77905047e-01 -3.69916946e-01 8.04885030e-01
6.44048989e-01 -1.15490727e-01 3.46754581e-01 -9.68282148e-02
8.54584932e-01 5.43638289e-01 2.82726079e-01 -4.54432636e-01
7.47334361e-01 9.25205052e-02 3.47858191e-01 2.17790619e-01
3.24836522e-01 3.74897599e-01 9.25124139e-02 -4.69242066e-01
-1.33459854e+00 -1.29336500e+00 -3.94793808e-01 9.38652694e-01
9.10341293e-02 -4.62233752e-01 -8.53886724e-01 -7.61465490e-01
-2.43747711e-01 7.46975243e-01 -4.15570736e-01 -7.21827075e-02
-9.66680110e-01 -8.21031272e-01 7.51532733e-01 2.58448005e-01
1.68714717e-01 -1.00471830e+00 2.01308683e-01 4.49187189e-01
-4.77964908e-01 -1.35886776e+00 -6.77312195e-01 1.52500719e-01
-7.52714217e-01 -6.87323749e-01 -6.50321484e-01 -1.44350493e+00
7.05911219e-01 9.09985974e-03 1.31443393e+00 -2.97234505e-01
-3.28426138e-02 1.06372200e-02 -4.18193370e-01 -3.70976105e-02
-5.43277860e-01 5.01100719e-01 4.59175169e-01 1.06795672e-02
1.05880713e+00 -7.34203637e-01 -1.60745397e-01 1.41963542e-01
-9.07693267e-01 -2.25721225e-01 7.88597167e-01 1.06232941e+00
6.76067889e-01 -5.20437002e-01 6.03809714e-01 -6.49393678e-01
7.91816771e-01 -2.89976954e-01 -6.54895484e-01 3.67690891e-01
-8.43002379e-01 6.93591952e-01 6.94453239e-01 -4.19172704e-01
-5.91051996e-01 2.94245873e-02 -8.50138627e-03 -1.19801752e-01
5.31573966e-02 4.67986494e-01 -1.21813126e-01 2.84033492e-02
6.27797961e-01 5.05173266e-01 -1.99068040e-01 -5.65628767e-01
8.66903484e-01 1.05939579e+00 3.34873676e-01 -6.71831608e-01
1.07697177e+00 7.83380643e-02 -4.39191639e-01 -3.38078231e-01
-6.90738082e-01 -5.02921700e-01 -1.20955431e+00 4.24014837e-01
1.19601357e+00 -9.77953136e-01 1.29979372e-01 -6.04622290e-02
-1.50871789e+00 8.93018991e-02 8.22188891e-03 9.11336303e-01
-5.01774311e-01 1.74454212e-01 -4.35997963e-01 -1.62211642e-01
-3.15655082e-01 -1.29670978e+00 1.10558045e+00 -3.35774273e-01
-4.18606669e-01 -1.26510370e+00 8.17500710e-01 3.11108023e-01
2.74014205e-01 -2.36609250e-01 1.14025044e+00 -8.38205457e-01
-4.32934672e-01 -1.98510632e-01 -1.17219366e-01 4.21905994e-01
3.26517373e-01 -5.10903656e-01 -6.89252079e-01 -4.20018107e-01
-4.26083207e-02 -3.69677961e-01 5.25031328e-01 -3.09046060e-01
2.29068771e-01 -2.42279824e-02 -2.95131803e-01 5.72664201e-01
1.71902013e+00 -9.36840195e-03 4.72557515e-01 5.63567758e-01
5.58172941e-01 5.58897257e-01 4.44875956e-01 -3.93393695e-01
4.51355964e-01 1.07148576e+00 -3.57347190e-01 -1.86530754e-01
-2.17700854e-01 -5.01100302e-01 6.10413969e-01 1.71397996e+00
7.43910596e-02 1.89374909e-01 -9.51111555e-01 9.99699652e-01
-1.50705624e+00 -6.14647806e-01 2.42816344e-01 2.40094042e+00
1.35478449e+00 -2.21291155e-01 -1.65374994e-01 -1.06173858e-01
5.53219318e-01 -5.64487949e-02 4.46177572e-02 -6.95911109e-01
-5.33325551e-03 8.01194012e-01 6.69018388e-01 8.85484815e-01
-9.04025555e-01 1.54151905e+00 5.94663286e+00 8.76817167e-01
-1.33068347e+00 5.26349604e-01 1.71576843e-01 1.07448190e-01
-5.41485429e-01 2.26552352e-01 -9.62941647e-01 2.12181583e-01
1.04301119e+00 -2.95024544e-01 6.48438454e-01 4.37140137e-01
-9.33887716e-03 2.11369425e-01 -1.38772500e+00 8.67543638e-01
5.24048150e-01 -9.92030144e-01 1.60538688e-01 7.51491189e-02
8.53689969e-01 2.89730281e-01 -1.09233551e-01 3.01960021e-01
2.35516608e-01 -9.88682926e-01 3.76192331e-01 5.71130514e-02
1.03256714e+00 -8.21325958e-01 9.15034294e-01 1.56199887e-01
-1.07146275e+00 5.79697609e-01 -5.12772381e-01 1.75361037e-01
9.48126242e-02 3.25382650e-01 -9.67942834e-01 7.67282903e-01
2.69413054e-01 6.06825411e-01 -5.11150420e-01 5.00355303e-01
-5.53555191e-01 4.63515401e-01 -2.09470749e-01 3.99996266e-02
3.91732693e-01 -5.48305631e-01 2.98837513e-01 1.30506814e+00
5.81586599e-01 -6.44659400e-01 1.65770948e-01 6.55826509e-01
-2.31381267e-01 8.31661463e-01 -1.02921069e+00 -1.08348884e-01
4.60485935e-01 1.08963490e+00 -5.42124867e-01 -4.12805200e-01
-5.67631900e-01 1.37962604e+00 6.35505617e-01 2.27448449e-01
-8.13513815e-01 -5.87561071e-01 6.20698929e-01 -9.10121650e-02
2.94393390e-01 -3.56999040e-01 -5.39497025e-02 -1.46666133e+00
3.63844812e-01 -1.09137058e+00 -2.40342781e-01 -3.62774342e-01
-1.36695874e+00 1.04127228e+00 -1.77834466e-01 -1.42156148e+00
-5.55260539e-01 -6.19071424e-01 -2.68709570e-01 1.26128173e+00
-1.63746250e+00 -1.29418468e+00 3.87019455e-01 4.54995960e-01
3.66226465e-01 -3.49272698e-01 1.42796731e+00 6.42412364e-01
-2.42081389e-01 8.50495934e-01 2.88475513e-01 3.92751426e-01
1.28310215e+00 -1.12958229e+00 5.53701699e-01 8.66988599e-01
8.97090554e-01 8.76145780e-01 3.91727805e-01 -2.80113995e-01
-1.27909720e+00 -1.05153596e+00 1.78138077e+00 -7.77944326e-01
1.07079864e+00 -7.55013406e-01 -6.54534578e-01 7.24817514e-01
8.93258333e-01 -5.20497523e-02 9.05240953e-01 2.00370356e-01
-6.01213634e-01 -2.61110365e-01 -7.43580937e-01 7.24527597e-01
1.06108654e+00 -8.69121432e-01 -9.79798615e-01 5.18436968e-01
8.39769483e-01 -3.28603834e-02 -9.25853670e-01 1.98982134e-01
6.06388152e-01 -4.82735634e-01 8.24197829e-01 -6.75416172e-01
4.47640151e-01 -3.99485528e-01 -3.27495605e-01 -1.67137921e+00
-1.15163311e-01 -3.15219194e-01 7.40060806e-01 1.34643137e+00
9.19802904e-01 -8.15392315e-01 3.22291672e-01 -1.85937911e-01
7.14181587e-02 -7.20639408e-01 -1.02717221e+00 -1.01120484e+00
6.34650707e-01 -2.07594588e-01 6.32197738e-01 1.11351562e+00
2.44551167e-01 7.61510432e-01 -1.81079850e-01 -6.08927868e-02
4.84375656e-01 2.91482389e-01 7.81779349e-01 -8.46647024e-01
-4.60170925e-01 -1.65679932e-01 -3.72250259e-01 -9.84615624e-01
7.00349927e-01 -1.63907814e+00 1.75893798e-01 -1.31992435e+00
1.32721215e-01 -5.82693338e-01 -4.58249003e-01 3.81472260e-01
-1.01850189e-01 7.82279849e-01 3.81292626e-02 2.76152432e-01
-2.67639011e-01 4.63478506e-01 9.10313129e-01 -2.54386038e-01
-6.75186962e-02 -4.91423160e-01 -6.75883174e-01 4.22786683e-01
6.36238575e-01 -8.80041778e-01 -3.06865960e-01 -8.57920587e-01
1.64697751e-01 -2.65089273e-01 -1.60258040e-01 -8.18884730e-01
2.30517667e-02 1.80065259e-01 6.68689907e-02 1.28187276e-02
2.97872927e-02 -6.83577120e-01 -2.11115703e-01 2.26106063e-01
-1.74777493e-01 5.19575775e-01 1.29360363e-01 1.08839437e-01
-5.43895125e-01 -4.36892450e-01 5.55106461e-01 4.61067371e-02
-4.11396354e-01 3.32589224e-02 -3.19047779e-01 8.06252509e-02
8.62104833e-01 1.47168919e-01 8.19739848e-02 1.22403055e-01
-6.40253186e-01 -1.61042601e-01 8.07552576e-01 7.40382910e-01
7.38310739e-02 -1.78062344e+00 -8.62010598e-01 4.65252042e-01
5.32114446e-01 -6.55371964e-01 -6.26633406e-01 8.46048594e-01
-3.96189630e-01 7.06315577e-01 -2.22853005e-01 -5.37912726e-01
-7.82397211e-01 6.70865238e-01 8.23311061e-02 -6.60936952e-01
-1.42730206e-01 4.76508319e-01 -1.20494805e-01 -1.02848029e+00
-2.83588201e-01 -3.47113274e-02 -6.18483499e-02 1.34709194e-01
2.00344354e-01 -2.71255374e-01 3.78121823e-01 -9.13719952e-01
-5.27106702e-01 1.24626887e+00 -1.63508244e-02 -7.41345763e-01
1.22435713e+00 1.16108665e-02 -3.76882017e-01 5.30771434e-01
1.59066689e+00 5.99792182e-01 -2.09498674e-01 -5.86452723e-01
3.37302476e-01 -2.56668389e-01 -7.17603341e-02 -5.11350214e-01
-6.87059402e-01 8.93015862e-01 5.83073318e-01 -2.08625585e-01
8.67915452e-01 5.64848445e-02 9.34848607e-01 3.95668209e-01
6.87695384e-01 -9.39559877e-01 -2.05635265e-01 6.47015095e-01
6.65128767e-01 -1.31617415e+00 -2.65339166e-01 -7.07753450e-02
-2.87708938e-01 1.02680695e+00 1.95703819e-01 -5.28411627e-01
5.05898714e-01 1.34106725e-01 5.38194239e-01 1.45326093e-01
-4.47046697e-01 -3.14914584e-01 3.71554881e-01 4.87953871e-01
8.87513459e-01 2.38052905e-02 -1.07721579e+00 1.21120356e-01
-4.11694050e-01 -4.03569080e-02 8.68378878e-02 5.23879051e-01
-2.91261263e-02 -2.31478667e+00 -1.12208605e-01 -9.83974561e-02
-4.30137902e-01 -6.68986201e-01 -3.02417219e-01 6.54701471e-01
1.84028134e-01 8.59277308e-01 2.04114109e-01 -4.36719418e-01
2.38391742e-01 5.07395983e-01 6.86655998e-01 -9.15593266e-01
-3.38815749e-01 1.55452294e-02 9.26427245e-02 -3.68541896e-01
-4.95807618e-01 -5.19962549e-01 -8.77527058e-01 -5.52711301e-02
-3.50619197e-01 5.57508469e-01 8.71625543e-01 1.15748119e+00
2.82980800e-01 3.65907960e-02 7.49924958e-01 -7.82343566e-01
-4.69147354e-01 -1.19852579e+00 -7.78162554e-02 4.65781808e-01
-1.23795658e-01 -4.80935186e-01 -2.16755152e-01 -6.24180119e-03] | [11.131728172302246, 10.046674728393555] |
e77075d3-a04a-41cb-b123-e2acc4d1ed09 | vad-vectorized-scene-representation-for | 2303.12077 | null | https://arxiv.org/abs/2303.12077v2 | https://arxiv.org/pdf/2303.12077v2.pdf | VAD: Vectorized Scene Representation for Efficient Autonomous Driving | Autonomous driving requires a comprehensive understanding of the surrounding environment for reliable trajectory planning. Previous works rely on dense rasterized scene representation (e.g., agent occupancy and semantic map) to perform planning, which is computationally intensive and misses the instance-level structure information. In this paper, we propose VAD, an end-to-end vectorized paradigm for autonomous driving, which models the driving scene as a fully vectorized representation. The proposed vectorized paradigm has two significant advantages. On one hand, VAD exploits the vectorized agent motion and map elements as explicit instance-level planning constraints which effectively improves planning safety. On the other hand, VAD runs much faster than previous end-to-end planning methods by getting rid of computation-intensive rasterized representation and hand-designed post-processing steps. VAD achieves state-of-the-art end-to-end planning performance on the nuScenes dataset, outperforming the previous best method by a large margin. Our base model, VAD-Base, greatly reduces the average collision rate by 29.0% and runs 2.5x faster. Besides, a lightweight variant, VAD-Tiny, greatly improves the inference speed (up to 9.3x) while achieving comparable planning performance. We believe the excellent performance and the high efficiency of VAD are critical for the real-world deployment of an autonomous driving system. Code and models will be released for facilitating future research. | ['Xinggang Wang', 'Chang Huang', 'Wenyu Liu', 'Qian Zhang', 'Helong Zhou', 'Jiajie Chen', 'Bencheng Liao', 'Qing Xu', 'Shaoyu Chen', 'Bo Jiang'] | 2023-03-21 | null | null | null | null | ['trajectory-planning'] | ['robots'] | [-1.26245946e-01 7.95010999e-02 -2.31781676e-01 -3.77802759e-01
-7.38977253e-01 -4.61732626e-01 7.87811995e-01 1.50931552e-01
-6.64707422e-01 6.42022192e-01 1.89562082e-01 -7.72558808e-01
1.73046999e-02 -1.29843819e+00 -8.08384597e-01 -4.57832664e-01
-4.17780519e-01 6.72023535e-01 7.96786427e-01 -5.97051382e-01
2.84824103e-01 4.51912045e-01 -1.88450038e+00 -3.11957866e-01
9.35630262e-01 8.34267318e-01 5.42845249e-01 6.64154172e-01
-2.48833634e-02 6.49671614e-01 -1.11366041e-01 2.27466211e-01
3.05590779e-01 1.48492023e-01 -5.55308819e-01 -2.50569701e-01
1.81058541e-01 -6.10951364e-01 -5.62763393e-01 6.03334665e-01
3.24431896e-01 4.52087700e-01 3.19738925e-01 -1.48768377e+00
2.20521539e-01 3.21656734e-01 -5.58591247e-01 1.46968570e-03
1.73070922e-01 5.72237372e-01 5.75483918e-01 -7.33251452e-01
5.86178720e-01 1.09661591e+00 4.09601718e-01 2.03518882e-01
-8.58254492e-01 -6.51352167e-01 3.77186775e-01 2.45611757e-01
-1.60949290e+00 -3.86668891e-01 4.62172747e-01 -3.74201864e-01
1.45209801e+00 2.96833456e-01 6.62132323e-01 6.10066533e-01
5.13086677e-01 6.35125577e-01 7.13886678e-01 1.14896201e-01
6.20089412e-01 -4.16365296e-01 1.44167081e-01 8.68257165e-01
3.23687941e-01 6.44909859e-01 -2.93682694e-01 -7.91301280e-02
3.44441295e-01 -1.55741081e-01 1.16507836e-01 -5.31522870e-01
-1.45050824e+00 8.30510736e-01 6.64965749e-01 -2.71697879e-01
-5.53103328e-01 5.41965067e-01 6.16074085e-01 -3.39130610e-01
2.42601007e-01 1.70099720e-01 -6.80169985e-02 -5.71863472e-01
-9.02327895e-01 8.46972883e-01 3.47477138e-01 1.35543549e+00
9.98441458e-01 1.56579331e-01 -2.03694999e-01 2.94718444e-01
2.76134372e-01 8.67015064e-01 -8.88548326e-03 -1.22647429e+00
6.88758790e-01 4.98955458e-01 3.88642818e-01 -1.01462770e+00
-8.42229903e-01 -1.12466313e-01 -7.48562753e-01 5.26594520e-01
6.14121258e-02 -3.18482041e-01 -1.05349469e+00 1.66758740e+00
6.46923840e-01 5.97888827e-01 3.22083086e-01 9.54407990e-01
5.57856977e-01 9.30225849e-01 1.55007804e-03 7.74556398e-02
1.23562241e+00 -1.29106033e+00 -6.98313653e-01 -6.08132601e-01
8.22426379e-01 -3.95280629e-01 9.26566243e-01 -6.16991008e-03
-9.25192118e-01 -5.65554440e-01 -1.26643622e+00 -1.24224178e-01
-2.73543000e-01 -2.62436241e-01 9.14873660e-01 3.13191444e-01
-1.03532004e+00 9.29606929e-02 -1.38600183e+00 -2.61355400e-01
4.83287960e-01 3.26214433e-01 -1.68268517e-01 -3.71549994e-01
-9.44922686e-01 9.34883118e-01 4.76261973e-01 -1.14440918e-01
-1.10880351e+00 -7.47815967e-01 -1.26860929e+00 -1.97670326e-01
6.22519493e-01 -6.16030753e-01 1.41406083e+00 2.52157390e-01
-1.45894361e+00 1.21246338e-01 -6.54142022e-01 -7.25061297e-01
5.36867321e-01 -1.38546035e-01 -1.95128590e-01 -2.84835964e-01
3.92693460e-01 8.69552255e-01 1.10481679e-01 -1.16841304e+00
-9.71750319e-01 -3.18150163e-01 1.09623395e-01 3.53694528e-01
3.24481964e-01 -6.40760005e-01 -8.92751396e-01 -2.70664934e-02
1.19145073e-01 -1.26871550e+00 -7.72134542e-01 -1.61254719e-01
-1.66915298e-01 -2.39545301e-01 8.63495171e-01 -1.57260731e-01
1.19459140e+00 -2.08293676e+00 -8.79444256e-02 1.12287052e-01
2.44148955e-01 9.98183042e-02 -1.05911419e-01 5.22636294e-01
6.34234905e-01 -2.76473433e-01 -4.63745832e-01 -3.77247602e-01
1.84905827e-01 4.91487861e-01 -4.13987190e-01 5.45147061e-01
3.90495434e-02 9.79801714e-01 -1.13880301e+00 -4.81819391e-01
7.98167408e-01 4.06792462e-01 -6.08923435e-01 -2.68670991e-02
-3.63964081e-01 2.32580483e-01 -6.97270155e-01 3.74083906e-01
1.00424027e+00 2.23227441e-01 1.82632767e-02 1.97485104e-01
-7.56397903e-01 3.87248933e-01 -1.10391247e+00 1.98448753e+00
-4.68564212e-01 6.83958948e-01 -1.09508917e-01 -5.15891075e-01
7.70990491e-01 -2.01216236e-01 4.82696533e-01 -1.19167447e+00
9.49339122e-02 4.71506715e-02 -3.53715777e-01 -1.13008261e-01
1.07076001e+00 3.50234598e-01 -5.97659767e-01 2.55542606e-01
-8.32449615e-01 -3.95576537e-01 2.17584074e-01 3.89153391e-01
1.26090157e+00 2.92921633e-01 3.02998483e-01 -3.51996958e-01
2.09652841e-01 9.05305505e-01 7.12363780e-01 6.58381283e-01
-3.38988692e-01 1.33763850e-01 2.23404005e-01 -6.07975841e-01
-9.12895977e-01 -8.22094142e-01 -1.32601306e-01 7.92920470e-01
8.36946011e-01 -6.55751348e-01 -5.72630763e-01 -2.28885815e-01
2.28010744e-01 1.28719914e+00 -4.68385756e-01 8.48236308e-02
-8.56775165e-01 -4.88209099e-01 5.96052468e-01 6.52024925e-01
6.40465438e-01 -6.72435164e-01 -1.20624542e+00 5.33184528e-01
-2.62106806e-01 -1.26111734e+00 -2.35520080e-01 -7.72133544e-02
-6.33840084e-01 -8.45245779e-01 -4.06348035e-02 -2.87825823e-01
4.78197426e-01 8.13683569e-01 9.62841988e-01 1.10812737e-02
4.69719544e-02 -1.29082710e-01 -2.72806585e-01 -5.10131896e-01
-1.56769082e-01 4.30172943e-02 1.41004011e-01 -6.09727025e-01
2.47862369e-01 -3.10745090e-01 -4.54350054e-01 4.86227065e-01
-6.14885032e-01 5.11019111e-01 4.10663217e-01 3.39024395e-01
9.94092226e-01 3.46177161e-01 1.71090290e-01 -6.08454585e-01
1.54851675e-01 -5.95024347e-01 -1.09316647e+00 -3.89771193e-01
-7.06318557e-01 1.85016975e-01 4.91884798e-01 7.00263605e-02
-9.51703727e-01 3.69456410e-01 -2.85738140e-01 -3.08485180e-01
-1.37009859e-01 5.59741259e-01 -5.73653774e-03 1.11319073e-01
5.58759391e-01 1.42006025e-01 -2.51866132e-02 -3.57064828e-02
4.83713388e-01 4.03324813e-01 7.80337393e-01 -3.42335165e-01
9.75713789e-01 7.50824809e-01 2.17067957e-01 -7.26324499e-01
-3.27829391e-01 -5.71719885e-01 -3.91532362e-01 -2.11101755e-01
8.62893343e-01 -1.16740870e+00 -8.83654058e-01 3.82104605e-01
-1.02149451e+00 -8.82591307e-01 -3.23263071e-02 6.03872597e-01
-6.57806993e-01 1.56084538e-01 -1.56213015e-01 -7.97028542e-01
-5.58079071e-02 -1.42029095e+00 1.26441050e+00 -3.50869261e-02
-2.33073048e-02 -5.47141492e-01 2.56813467e-01 1.71639383e-01
4.96205121e-01 5.67917347e-01 4.49716747e-01 1.51856765e-01
-9.27721143e-01 -7.82594681e-02 -3.67399812e-01 -6.00244403e-01
-2.49339625e-01 -1.06690735e-01 -7.66100883e-01 -2.66620040e-01
-5.75308681e-01 5.63008599e-02 9.25643861e-01 4.48026896e-01
9.03280377e-01 -2.56240200e-02 -8.42593729e-01 5.57452261e-01
1.52874613e+00 3.42082024e-01 7.64846742e-01 5.59288561e-01
7.36347914e-01 3.90652865e-01 1.37781549e+00 5.90071261e-01
1.13952410e+00 9.94513273e-01 9.36850131e-01 -1.56169072e-01
-9.96418297e-02 -3.45033348e-01 3.25678468e-01 5.15326262e-01
2.44267005e-02 -3.73746246e-01 -1.27235329e+00 7.30840027e-01
-2.24242043e+00 -9.62285638e-01 -4.37239677e-01 2.13462186e+00
8.55335966e-02 4.20176744e-01 1.24092609e-01 -3.38657666e-03
2.30297849e-01 4.16214705e-01 -8.63051414e-01 -3.51427436e-01
3.57386380e-01 -1.11849934e-01 1.17823899e+00 9.70018685e-01
-1.06512594e+00 1.37699342e+00 5.90775776e+00 7.26616681e-01
-9.09283757e-01 1.03677414e-01 -5.43372100e-03 -2.18653634e-01
-3.05667877e-01 1.13695137e-01 -1.05102110e+00 3.21836114e-01
1.05013716e+00 -2.23782122e-01 4.73455817e-01 1.13428867e+00
6.13394976e-01 -5.25584519e-01 -7.68795609e-01 7.45583534e-01
-3.17170113e-01 -1.67290676e+00 -4.13581282e-01 4.70054239e-01
4.99426335e-01 6.92426741e-01 -2.49540478e-01 6.07198119e-01
6.95649445e-01 -9.26925600e-01 9.87237871e-01 1.64254114e-01
7.80919194e-01 -1.13188088e+00 7.50184298e-01 6.04503393e-01
-1.74762690e+00 1.42355517e-01 -4.73226398e-01 -5.15498400e-01
7.88771331e-01 5.08621812e-01 -9.57397044e-01 7.18553305e-01
5.38462639e-01 6.31932795e-01 -2.75006056e-01 8.07082593e-01
1.26130804e-01 4.80890214e-01 -6.04302227e-01 -2.54220087e-02
8.79116654e-01 -7.33431727e-02 5.45522809e-01 1.06057835e+00
2.44824216e-01 2.64380544e-01 6.33884192e-01 6.19986832e-01
4.64201212e-01 -4.84513104e-01 -9.08735275e-01 3.41702074e-01
7.47433841e-01 1.00629401e+00 -5.82244575e-01 -3.39026392e-01
-3.55693221e-01 4.97882396e-01 3.17792743e-01 1.68867514e-01
-1.28656518e+00 -3.87258649e-01 1.15017807e+00 1.49161518e-01
3.79262835e-01 -8.90557349e-01 -4.65642780e-01 -5.83584487e-01
-3.90065350e-02 -3.68335456e-01 -1.44136742e-01 -5.24384379e-01
-3.87139648e-01 7.45218694e-01 2.63794631e-01 -1.25861502e+00
-4.03414816e-01 -1.55209526e-01 -4.98144954e-01 7.02726483e-01
-1.90546668e+00 -9.15622950e-01 -7.81052530e-01 4.30525273e-01
6.25679016e-01 1.15155317e-01 6.87241793e-01 2.11542502e-01
-6.36035979e-01 1.73953697e-01 -1.36220735e-02 -2.40529105e-01
1.49209723e-01 -8.82969737e-01 1.20836234e+00 1.03748417e+00
-3.34813446e-01 1.91130072e-01 7.78536201e-01 -7.95833826e-01
-1.91957092e+00 -1.66055274e+00 7.24354386e-01 -4.67327356e-01
3.47987115e-01 -2.17496380e-01 -5.94171107e-01 7.11938202e-01
4.01178055e-04 -1.18901320e-01 7.17824372e-03 -3.27118416e-03
5.71772568e-02 -5.05198576e-02 -1.01279819e+00 8.68459940e-01
1.30421972e+00 -1.59827936e-02 -2.86731839e-01 2.54541427e-01
1.04986632e+00 -8.84259939e-01 -5.02269208e-01 5.30654728e-01
4.41307455e-01 -9.63470161e-01 8.65482211e-01 -1.79756582e-02
9.26134526e-04 -7.91480362e-01 -4.11794901e-01 -1.06197560e+00
-5.42035103e-01 -3.87074560e-01 -3.71404886e-02 6.79661095e-01
3.61169130e-01 -7.41450727e-01 8.29594731e-01 6.76997006e-01
-6.65408254e-01 -9.04222608e-01 -9.65178907e-01 -8.35143209e-01
-1.99448824e-01 -1.17746603e+00 1.07858014e+00 4.58307803e-01
-1.61629781e-01 2.51630098e-01 -2.38750800e-01 6.49192691e-01
6.72312200e-01 1.20516501e-01 1.29274535e+00 -7.98681855e-01
2.92686850e-01 -2.45657548e-01 -5.71688473e-01 -1.29468310e+00
2.69132793e-01 -7.03184903e-01 4.46908474e-01 -1.91056573e+00
-3.07846993e-01 -9.63771701e-01 1.51007310e-01 5.53896606e-01
-4.92652319e-03 6.84425607e-02 7.03118965e-02 4.21157759e-03
-6.52260005e-01 8.31898808e-01 1.14799571e+00 -1.81890890e-01
-4.66551751e-01 -1.67293251e-01 -4.44374949e-01 5.65953135e-01
9.46817160e-01 -3.97652775e-01 -7.45485902e-01 -7.91644216e-01
-3.14372256e-02 1.79967552e-01 3.07769686e-01 -1.30348063e+00
4.62016732e-01 -6.08707309e-01 -2.28414446e-01 -1.31089544e+00
7.16841578e-01 -7.64605999e-01 2.33317390e-01 6.17226601e-01
1.72368243e-01 3.61171246e-01 7.02022314e-01 7.59125948e-01
-7.58633614e-02 2.71281034e-01 5.31808794e-01 7.61483982e-02
-1.33438456e+00 5.92016995e-01 -6.48020566e-01 1.79475243e-03
1.48286903e+00 -2.92266190e-01 -3.84939402e-01 -2.06180170e-01
7.89679885e-02 8.60018373e-01 7.35375047e-01 4.37440634e-01
6.51098132e-01 -1.36232412e+00 -7.57682562e-01 2.68928021e-01
2.75794178e-01 6.15704954e-01 4.67158169e-01 6.63157225e-01
-8.47263634e-01 7.13456988e-01 -2.92758550e-02 -7.93934226e-01
-1.02867723e+00 6.28076196e-01 -3.28387655e-02 -1.50234118e-01
-1.08843684e+00 5.32160878e-01 2.16967568e-01 -5.56960464e-01
2.41827639e-03 -4.83495831e-01 6.21983670e-02 -4.03023213e-01
7.11577356e-01 7.18293130e-01 8.82525742e-02 -8.98596525e-01
-7.03390121e-01 5.82967937e-01 1.61409870e-01 -3.23349267e-01
1.21810079e+00 -2.07535550e-01 2.58150280e-01 3.57861072e-02
8.33045781e-01 -4.47594910e-04 -1.52661860e+00 -9.92193967e-02
-2.95429140e-01 -5.85742593e-01 6.40345871e-01 -5.04149020e-01
-8.15611243e-01 6.25019550e-01 4.51263428e-01 -1.64109305e-01
7.59612322e-01 -1.95888564e-01 1.11890101e+00 4.63290244e-01
1.05772007e+00 -8.86589110e-01 -5.25315523e-01 8.68061841e-01
6.39941692e-01 -1.17699099e+00 1.57691598e-01 -5.12713492e-01
-8.84948850e-01 5.08153498e-01 7.03099847e-01 -1.18746765e-01
3.77955407e-01 5.33401847e-01 6.89788489e-03 -3.25908870e-01
-8.90960097e-01 -4.86239880e-01 -3.67759019e-02 6.62146986e-01
-3.12146991e-01 5.32210827e-01 -1.03400387e-01 1.66268528e-01
-4.67522413e-01 -5.88798858e-02 4.73020911e-01 1.08092844e+00
-6.45592809e-01 -7.45861530e-01 -2.48403072e-01 3.71628106e-01
4.44804251e-01 1.08547702e-01 2.60290384e-01 9.40184712e-01
6.46973699e-02 1.20974684e+00 3.88164520e-01 -6.34067953e-01
6.06950879e-01 -4.41176295e-01 7.42671415e-02 -4.15313721e-01
-1.16552062e-01 -2.99661875e-01 3.41195822e-01 -1.28537452e+00
-3.30157727e-02 -6.52477324e-01 -1.88548732e+00 -8.81637514e-01
-5.78154363e-02 -9.65113118e-02 9.38925505e-01 8.21986854e-01
8.62706900e-01 7.06996858e-01 4.23422694e-01 -1.40762734e+00
-1.46794081e-01 -4.56975132e-01 -2.00715691e-01 -2.43598491e-01
4.33937967e-01 -1.07837534e+00 6.71950132e-02 -5.31826019e-01] | [5.796955585479736, 0.8261553645133972] |
c36cf722-677c-4716-b86b-0a843e8d2301 | on-the-convergence-rates-of-policy-gradient | 2201.07443 | null | https://arxiv.org/abs/2201.07443v2 | https://arxiv.org/pdf/2201.07443v2.pdf | On the Convergence Rates of Policy Gradient Methods | We consider infinite-horizon discounted Markov decision problems with finite state and action spaces and study the convergence rates of the projected policy gradient method and a general class of policy mirror descent methods, all with direct parametrization in the policy space. First, we develop a theory of weak gradient-mapping dominance and use it to prove sharper sublinear convergence rate of the projected policy gradient method. Then we show that with geometrically increasing step sizes, a general class of policy mirror descent methods, including the natural policy gradient method and a projected Q-descent method, all enjoy a linear rate of convergence without relying on entropy or other strongly convex regularization. Finally, we also analyze the convergence rate of an inexact policy mirror descent method and estimate its sample complexity under a simple generative model. | ['Lin Xiao'] | 2022-01-19 | null | null | null | null | ['policy-gradient-methods'] | ['methodology'] | [-2.99467589e-03 4.70268875e-01 -5.27418017e-01 -7.68936649e-02
-8.64760458e-01 -7.10462928e-01 5.87615848e-01 -1.79290727e-01
-7.84444034e-01 1.21547318e+00 3.81560415e-01 -8.70637178e-01
-2.07149088e-01 -4.10758764e-01 -7.86533773e-01 -9.26090658e-01
-1.35936916e-01 4.74395156e-01 -4.35694866e-02 -2.09234640e-01
3.44792068e-01 3.14502507e-01 -1.01204062e+00 -4.46869284e-01
9.01004732e-01 9.72829700e-01 -9.93743911e-02 8.49272788e-01
1.82820186e-01 9.42184150e-01 -1.56881168e-01 -1.24065347e-01
6.83595300e-01 -7.29760468e-01 -8.66879940e-01 1.17917806e-01
2.44150221e-01 -8.45484555e-01 -5.14703751e-01 1.33264172e+00
5.74469268e-01 6.65803254e-01 7.27345765e-01 -9.58479345e-01
-4.38679069e-01 6.59355879e-01 -3.78783405e-01 3.76517624e-01
1.66387688e-02 1.80713043e-01 8.10915768e-01 -5.33683777e-01
4.91677850e-01 1.40710402e+00 4.98864442e-01 9.23021853e-01
-1.35121131e+00 -2.83786029e-01 5.61700344e-01 -1.94137067e-01
-6.38062775e-01 -3.04748595e-01 3.00021797e-01 -5.09346426e-01
7.14094937e-01 1.41345948e-01 5.32474339e-01 1.06062508e+00
1.32961810e-01 8.93266618e-01 1.46715653e+00 -4.01131600e-01
6.32488608e-01 1.31808758e-01 2.99690920e-03 9.01846409e-01
-8.93840753e-03 6.55570567e-01 9.56389979e-02 -5.95794559e-01
7.23279297e-01 5.09118922e-02 -3.20495218e-01 -4.69109505e-01
-9.24301028e-01 9.92287517e-01 -1.08056150e-01 -1.59905016e-01
-4.11016494e-01 3.73789191e-01 2.74814844e-01 5.03879368e-01
7.14239180e-01 2.60046303e-01 -3.47225636e-01 -3.11230540e-01
-6.85281575e-01 4.81511921e-01 1.21719110e+00 8.11209977e-01
3.38966429e-01 4.19655085e-01 -6.37551188e-01 3.95070314e-01
3.14058304e-01 8.59818995e-01 3.08521301e-01 -1.44039917e+00
6.30354881e-01 -6.45029321e-02 7.67845213e-01 -3.16825300e-01
-3.42983037e-01 -2.69079477e-01 -5.80279231e-01 6.17530882e-01
7.13011026e-01 -8.05436552e-01 -7.52609074e-01 2.05433106e+00
3.28925252e-01 -1.54497117e-01 8.02935287e-02 8.86251032e-01
-6.55936182e-01 7.65732944e-01 -2.63794303e-01 -9.48993146e-01
6.06168568e-01 -6.54704392e-01 -5.12721360e-01 -6.56777993e-02
5.18478870e-01 -3.47420663e-01 1.09246707e+00 3.47421318e-01
-1.31412232e+00 3.17750871e-02 -8.67292941e-01 2.71811694e-01
3.10016096e-01 2.78312014e-03 1.31798357e-01 6.09282255e-01
-9.60873187e-01 1.22432220e+00 -9.74385023e-01 -1.72764599e-01
9.46037322e-02 2.44941518e-01 2.69495726e-01 3.25522929e-01
-9.47002113e-01 1.04573047e+00 4.05757815e-01 1.70641784e-02
-1.47095263e+00 -5.63181698e-01 -4.10307288e-01 2.70932838e-02
7.31421292e-01 -5.78155577e-01 1.81306744e+00 -7.86039829e-01
-2.31635141e+00 2.54097313e-01 -3.73410434e-02 -6.72286689e-01
1.09314048e+00 -4.64577228e-01 5.69216833e-02 -3.13170739e-02
-1.20492674e-01 -1.38831645e-01 1.03935492e+00 -7.67829180e-01
-8.02596927e-01 -5.46280682e-01 3.63839447e-01 7.44486511e-01
-3.02604169e-01 -1.98275506e-01 4.05827194e-01 -2.10366324e-01
-3.56390715e-01 -1.31911278e+00 -7.38459051e-01 -6.03248738e-02
-2.20213637e-01 -1.80813655e-01 4.03706968e-01 -5.03329456e-01
1.02702403e+00 -1.90598774e+00 3.61795634e-01 7.13321939e-02
-5.28516695e-02 1.82018742e-01 1.01858214e-01 4.57081765e-01
2.68105447e-01 5.55686690e-02 -4.53297853e-01 -7.69100189e-02
3.19005907e-01 3.07702750e-01 -8.06006670e-01 8.96038592e-01
-2.72007972e-01 6.09023273e-01 -1.22763813e+00 -3.00293621e-02
1.02349102e-01 -1.84453189e-01 -5.90802729e-01 -6.78578243e-02
-4.24438685e-01 4.42435533e-01 -8.06239128e-01 -9.25820842e-02
3.06578219e-01 2.40093078e-02 3.44169825e-01 6.32328033e-01
-3.11055720e-01 2.16931298e-01 -9.74909067e-01 1.31822407e+00
-5.49135089e-01 3.15551817e-01 5.00931680e-01 -1.13677323e+00
3.81850809e-01 3.97501558e-01 5.72765410e-01 -1.69662446e-01
1.74361899e-01 2.83123881e-01 -5.57889581e-01 -9.54383388e-02
4.92844552e-01 -6.25181735e-01 1.02146104e-01 6.81628227e-01
-1.53091908e-01 7.99148306e-02 -5.30249253e-02 1.26792148e-01
7.13812351e-01 3.64856005e-01 1.85122281e-01 -6.93138003e-01
2.25347430e-01 -1.10378608e-01 6.66901052e-01 9.93928015e-01
-4.43146050e-01 -6.29634485e-02 9.00202215e-01 -2.80645430e-01
-1.23836231e+00 -1.01420307e+00 -7.16274381e-02 1.35767555e+00
-3.07618603e-02 -1.23870336e-01 -4.19204950e-01 -8.67753088e-01
4.23325866e-01 8.31202745e-01 -6.93665028e-01 -3.60057175e-01
-4.83326972e-01 -9.31166351e-01 2.24815905e-01 3.79817218e-01
4.57682937e-01 -5.46142876e-01 -6.48528159e-01 3.42329592e-01
1.81675360e-01 -6.21026278e-01 -8.37381303e-01 3.63492332e-02
-1.21292388e+00 -8.18846762e-01 -1.19239295e+00 -2.61215508e-01
5.20242214e-01 -2.89565384e-01 7.61636734e-01 -8.47433090e-01
5.71677446e-01 6.58413768e-01 1.89823970e-01 -3.86473596e-01
-8.02821875e-01 -2.26011068e-01 5.66418707e-01 -3.34677771e-02
-2.54362971e-01 -4.02452320e-01 -7.16062665e-01 3.11079204e-01
-5.16204119e-01 -1.86804250e-01 2.12763995e-01 8.12650502e-01
6.68185830e-01 -4.83118773e-01 6.99769408e-02 -6.37231588e-01
1.09794343e+00 -5.59714675e-01 -1.41702938e+00 2.01887935e-01
-1.13900900e+00 6.29920542e-01 7.51456797e-01 -5.66346228e-01
-1.16263974e+00 7.49464706e-02 1.83757216e-01 -5.45084417e-01
5.30126631e-01 2.75819361e-01 3.61894995e-01 1.48719028e-01
7.47498453e-01 4.91268933e-01 2.03976393e-01 -6.32998943e-01
7.00117946e-01 4.48304564e-01 4.88967985e-01 -8.52558851e-01
3.95518005e-01 8.20688367e-01 1.45548537e-01 -4.45340008e-01
-9.87902224e-01 -1.19805001e-01 7.06564076e-03 -1.19259566e-01
7.73563206e-01 -6.61268473e-01 -1.04822075e+00 2.25710347e-01
-7.79061079e-01 -7.79264450e-01 -7.96962738e-01 7.78502822e-01
-1.12610006e+00 4.91075814e-01 -7.64544010e-01 -1.36584616e+00
-2.76035666e-01 -8.12636554e-01 6.58321083e-01 6.14747368e-02
4.93108779e-01 -1.01338065e+00 6.85802758e-01 -4.15443122e-01
2.30294794e-01 2.63200313e-01 5.89205801e-01 -4.01048928e-01
1.55194206e-02 1.44220427e-01 2.27684215e-01 6.61415935e-01
9.97688714e-03 -1.46173865e-01 -3.43799174e-01 -7.25538373e-01
4.71000284e-01 -1.20701008e-01 9.24526870e-01 7.59655833e-01
7.27718890e-01 -1.02418077e+00 -1.51761532e-01 6.86517715e-01
1.45409405e+00 3.09550047e-01 3.64581347e-02 6.17873892e-02
4.39535677e-01 2.81671375e-01 4.17031527e-01 7.78259397e-01
5.80762364e-02 1.91273898e-01 4.81726766e-01 5.30216217e-01
6.76380754e-01 -5.50705969e-01 9.36816990e-01 5.94254494e-01
-3.17629874e-01 1.96849689e-01 -5.67771256e-01 4.34603482e-01
-2.22408247e+00 -1.17083299e+00 4.32838082e-01 2.72534060e+00
9.99784827e-01 2.08748672e-02 5.71858287e-01 -3.40388387e-01
7.97052383e-01 -1.45483706e-02 -1.16824198e+00 -8.61128569e-01
1.03510283e-01 -1.07883170e-01 1.18859136e+00 8.49419475e-01
-1.07542229e+00 7.89971113e-01 7.97626066e+00 6.60808563e-01
-8.76483858e-01 2.78072804e-01 5.64640939e-01 -4.10499930e-01
-1.86972484e-01 2.58822348e-02 -8.94763768e-01 5.53053617e-01
1.18830788e+00 -6.69245958e-01 8.51862729e-01 1.29969144e+00
4.12366390e-01 -1.76425222e-02 -8.60493720e-01 6.11329079e-01
-6.85165763e-01 -1.19911993e+00 -4.01898503e-01 5.77620864e-01
1.21858907e+00 3.14497709e-01 1.51957214e-01 3.97533774e-01
1.07568502e+00 -5.31812012e-01 7.31267691e-01 3.79305393e-01
6.84791625e-01 -8.82837892e-01 1.31308749e-01 4.84496295e-01
-7.03557014e-01 -7.25182295e-01 -7.61454523e-01 -4.08273101e-01
3.25391352e-01 2.63269275e-01 -2.94705063e-01 1.96351200e-01
1.78715169e-01 4.29957539e-01 4.21075732e-01 8.14781547e-01
-1.87099144e-01 6.86145842e-01 -6.58804417e-01 -1.35846287e-01
7.40574419e-01 -5.96763313e-01 9.60197389e-01 8.10396254e-01
2.43442610e-01 2.37996772e-01 3.60174417e-01 6.00585103e-01
7.59106949e-02 3.53425667e-02 -5.55790842e-01 -2.00195670e-01
4.59496416e-02 8.03181529e-01 -3.63441885e-01 -4.46593344e-01
-1.80294484e-01 1.07122314e+00 2.90174276e-01 6.19272292e-01
-7.23980308e-01 -1.28216464e-02 1.01117647e+00 -1.40672624e-01
4.80071515e-01 -4.44077969e-01 2.11332873e-01 -1.29568052e+00
1.40297383e-01 -4.83427465e-01 3.88234496e-01 -1.42778054e-01
-1.13087964e+00 2.75902092e-01 2.29022488e-01 -1.06517565e+00
-7.84644842e-01 -4.51507777e-01 -4.71931398e-01 8.01082075e-01
-1.05936778e+00 -1.33180380e-01 6.94346547e-01 8.44205022e-01
2.70241559e-01 -6.39474839e-02 4.61674988e-01 -1.51763037e-01
-5.37671506e-01 2.68291116e-01 1.33683407e+00 -2.14816853e-01
1.24253795e-01 -1.63012815e+00 1.55048057e-01 9.11038697e-01
-3.26027989e-01 2.90847987e-01 9.20385718e-01 -4.30445611e-01
-1.48465776e+00 -1.17131484e+00 1.89260721e-01 -1.91957727e-01
9.05277371e-01 -7.48945326e-02 -5.18976927e-01 9.04599965e-01
-1.89265283e-03 -5.12986891e-02 -6.57350756e-05 3.62694869e-03
-4.71125618e-02 1.32064253e-01 -1.06473780e+00 6.49695039e-01
9.56339955e-01 -6.93876863e-01 -6.09964192e-01 7.22023726e-01
6.84798896e-01 -3.78703654e-01 -6.15850389e-01 1.06899790e-01
5.46715021e-01 -6.88387752e-01 4.76467401e-01 -1.34353662e+00
1.13902479e-01 7.01062754e-02 -1.37134954e-01 -1.49295950e+00
-3.63630027e-01 -1.48199022e+00 -5.68144739e-01 5.52781940e-01
3.21169049e-01 -1.03440475e+00 7.02045381e-01 7.00771987e-01
3.83390523e-02 -9.40165281e-01 -1.14903426e+00 -1.45555413e+00
7.94996142e-01 -1.78243607e-01 5.27608320e-02 4.78184760e-01
5.46849549e-01 1.03542045e-01 -6.71154499e-01 -1.04951173e-01
7.11821854e-01 1.35377899e-01 1.33123487e-01 -6.24703467e-01
-7.04521418e-01 -6.32704139e-01 2.67950743e-01 -1.44107115e+00
2.54901141e-01 -8.22346270e-01 1.60941765e-01 -1.27413917e+00
1.29936589e-03 -1.62867129e-01 -6.76354170e-01 9.17030573e-02
-2.14976594e-01 -5.40984392e-01 4.00427341e-01 2.98537016e-01
-4.20539260e-01 1.01632571e+00 1.31683362e+00 2.17306167e-02
-6.01648271e-01 5.08500397e-01 -5.19514263e-01 7.50067949e-01
8.14329445e-01 -6.68413103e-01 -7.16806054e-01 -2.38006443e-01
3.17599356e-01 4.33818102e-01 1.67555436e-01 -5.62155247e-01
-1.70684591e-01 -6.66917026e-01 -2.23522186e-01 -1.39189571e-01
3.62246037e-02 -1.89162791e-01 -3.16254437e-01 9.54897583e-01
-8.13901901e-01 2.05943525e-01 -9.40606371e-02 9.85594213e-01
4.63700920e-01 -2.54058063e-01 8.74842763e-01 -1.66079387e-01
-2.52477050e-01 6.02023542e-01 -7.83580363e-01 4.70320731e-01
9.15020764e-01 1.94837093e-01 -1.82962194e-02 -7.46824980e-01
-1.19804466e+00 3.77139002e-01 2.28392124e-01 -6.34574983e-03
3.63286376e-01 -1.29169023e+00 -7.56242931e-01 -3.93756390e-01
-5.88700116e-01 -5.76570094e-01 -1.58375815e-01 9.61184978e-01
-1.18313786e-02 3.69281560e-01 -6.75029978e-02 -1.92517206e-01
-6.32476807e-01 7.71656156e-01 6.71206355e-01 -5.14334440e-01
-6.23611689e-01 6.83967650e-01 1.70300439e-01 -2.55706906e-01
2.45773315e-01 -5.12639821e-01 3.50848466e-01 -9.59156305e-02
6.17766261e-01 7.10874796e-01 -5.35327256e-01 -1.63961515e-01
-1.85565203e-01 2.30504751e-01 1.77157834e-01 -9.19544220e-01
8.83030236e-01 -2.63362855e-01 1.93959951e-01 4.98416334e-01
1.23288500e+00 -4.12618488e-01 -2.00514030e+00 -3.24050128e-01
1.12493858e-01 -2.64032662e-01 9.97989327e-02 -2.79552907e-01
-9.28204954e-01 7.66212344e-01 7.86353111e-01 3.08719009e-01
7.63737559e-01 -3.48776579e-01 4.63595927e-01 7.13862777e-01
5.10002255e-01 -1.47409034e+00 -4.40603256e-01 7.42645681e-01
7.55156338e-01 -1.01016366e+00 3.77456024e-02 4.11932349e-01
-6.89598918e-01 9.11619902e-01 -2.11548898e-02 -4.41180825e-01
8.64029706e-01 -8.31885710e-02 -3.28784108e-01 4.00293767e-01
-1.05354595e+00 -3.43290746e-01 -1.26584262e-01 4.10874337e-01
-6.15736917e-02 2.36747667e-01 -7.60405123e-01 6.44629002e-02
4.28320654e-02 2.66179860e-01 6.12512529e-01 9.10429835e-01
-5.66310585e-01 -7.53311872e-01 -2.91456655e-02 4.56629604e-01
-7.31210113e-01 -1.16035044e-01 -3.99636403e-02 3.14604044e-01
-7.16300428e-01 8.56536865e-01 -1.00977704e-01 -6.66608885e-02
1.74102291e-01 1.19301826e-01 8.42276871e-01 -2.28589237e-01
-3.45159799e-01 -5.92579506e-02 -2.04464234e-03 -7.99300432e-01
-5.45718521e-02 -6.40161157e-01 -9.90474284e-01 -5.17498314e-01
-2.60546327e-01 3.06948006e-01 6.37792349e-01 1.07479823e+00
2.39067823e-01 -1.14345640e-01 9.35953379e-01 -6.42260909e-01
-2.00805378e+00 -8.15497100e-01 -8.39803457e-01 1.56698629e-01
6.22679412e-01 -3.67654324e-01 -5.38991928e-01 -3.28997880e-01] | [4.286829471588135, 2.5814247131347656] |
6f8344c5-04fb-4908-afc2-2ca777a1f950 | prescriptive-pca-dimensionality-reduction-for | 2306.02223 | null | https://arxiv.org/abs/2306.02223v1 | https://arxiv.org/pdf/2306.02223v1.pdf | Prescriptive PCA: Dimensionality Reduction for Two-stage Stochastic Optimization | In this paper, we consider the alignment between an upstream dimensionality reduction task of learning a low-dimensional representation of a set of high-dimensional data and a downstream optimization task of solving a stochastic program parameterized by said representation. In this case, standard dimensionality reduction methods (e.g., principal component analysis) may not perform well, as they aim to maximize the amount of information retained in the representation and do not generally reflect the importance of such information in the downstream optimization problem. To address this problem, we develop a prescriptive dimensionality reduction framework that aims to minimize the degree of suboptimality in the optimization phase. For the case where the downstream stochastic optimization problem has an expected value objective, we show that prescriptive dimensionality reduction can be performed via solving a distributionally-robust optimization problem, which admits a semidefinite programming relaxation. Computational experiments based on a warehouse transshipment problem and a vehicle repositioning problem show that our approach significantly outperforms principal component analysis with real and synthetic data sets. | ['Ho-Yin Mak', 'Long He'] | 2023-06-04 | null | null | null | null | ['dimensionality-reduction', 'stochastic-optimization'] | ['methodology', 'methodology'] | [ 0.2142745 0.21819872 -0.03273612 -0.36545214 -0.85724413 -0.66222894
0.24029934 0.13278192 -0.3276058 0.4550105 0.45135847 -0.31756172
-0.7593534 -0.73404866 -0.5349427 -0.9334162 -0.03719681 0.87467074
-0.6653923 -0.14399347 0.11482874 0.5868443 -1.2225933 -0.2126238
0.9174971 0.82003117 0.06273422 0.34510636 -0.01957736 -0.021211
-0.2965523 -0.10319059 0.70117164 -0.08795943 -0.5268584 0.7042555
-0.08063665 -0.03932665 -0.16432972 1.2045215 0.36850438 0.35227114
0.87972945 -1.4412222 -0.41976467 0.40982425 -0.60051423 -0.1440754
-0.02299538 0.08261843 1.0392843 -0.7789232 0.41944894 1.4069434
0.23813324 0.33911926 -1.8082665 -0.01997322 0.43071264 -0.40717745
-1.1999936 -0.21378663 1.0074869 -0.7255131 0.61862475 0.32410008
0.38790593 0.7438083 -0.07874583 0.83749646 0.8333497 -0.03700651
0.6666958 0.20104924 0.30946228 0.23953454 0.57843715 0.05604413
-0.06918855 -0.46459273 0.08818962 0.09934834 -0.18365575 -1.1774021
-0.99031204 1.0697219 -0.15960437 -0.17987506 -0.7667254 -0.17600054
0.2968391 0.19673383 0.5684446 0.6538968 -0.50155973 -0.29224938
-0.7050539 0.4752845 1.0790836 0.8834029 0.64617807 0.05854945
0.07502156 0.7440126 0.5265953 0.56910485 0.23396704 -1.0348822
0.9069469 0.6040869 0.46551418 -0.8932035 -0.3799032 -0.00718345
-0.66827524 0.1516572 0.4468436 -0.583434 -0.6150814 1.724107
0.40142 -0.3714068 0.39771128 1.1877488 -0.24522781 0.78438485
-0.03319409 -0.5470608 1.0293112 -0.4645123 -0.6923137 -0.35847315
0.8614188 -0.33803108 1.0153623 0.40112138 -0.90092516 0.1461761
-0.9795168 0.3027903 -0.14190732 0.13855833 0.46603197 0.76601124
-0.62044895 0.46172062 -0.78698987 -0.18414912 0.15594509 0.48744172
-0.32678872 -0.2593834 -0.9355026 0.61503446 0.4394239 0.30604598
-0.854013 -0.8310701 -0.926185 0.21277195 0.61822015 -0.5978136
0.9197939 -0.647684 -1.4163393 0.6016794 0.05757217 -0.21876405
0.49800465 -0.22239338 -0.06725042 -0.21757315 0.13431314 -0.04142376
0.9480672 -1.3220341 -0.34034625 -0.7702021 0.16449599 0.32802132
-0.47686017 -0.27062258 -0.30409294 -0.57253385 0.14757034 -1.4026366
-0.6155363 -0.31480455 -0.50746804 -0.02986588 0.91399604 -0.43687975
1.3220097 -2.1888177 0.6963992 0.5289505 0.087263 -0.02747346
-0.208579 0.4436188 -0.37137392 0.14245367 -0.65679306 -0.52791744
0.41942167 0.36219507 -0.3132877 0.90031564 0.22711632 0.41955435
-0.8089862 0.03351569 -0.01450306 0.15755618 -0.76304185 0.09483918
-0.27602077 0.0360023 -0.79840267 0.37427408 0.68116456 0.05027658
0.39916977 -0.0827826 -0.04098792 -0.11907401 -1.6056194 1.5879272
-0.46368363 0.4229582 0.44671535 -1.4492191 0.79726243 -0.18192828
0.9994702 -0.30378047 -0.04444201 -0.06206968 -0.06503673 -0.57092434
0.60202515 -0.409654 -0.52865523 0.67926174 -0.41646975 -0.1500571
0.17543866 0.09470122 0.8673593 -0.11679614 -0.00930114 -0.528128
0.33056954 0.29668635 0.86793065 0.31664252 0.09060906 0.52515644
1.20527 -0.0462624 -1.14439 -0.9195718 0.01920544 0.6278523
-0.10825115 -0.30682704 -0.75728834 -0.6674376 0.39555877 1.0330325
-0.4282263 -0.3457394 -0.40054208 -1.2674259 -0.12234154 0.24654932
-0.11962273 -0.2007639 -0.40827665 0.21077618 0.16656621 -0.89710623
-0.55477107 0.27723938 -0.9783312 -1.0151894 -0.7003262 -0.42583004
1.0339607 0.09906302 0.75361073 -0.8558134 -0.12417916 0.7259222
0.12346664 -0.19866985 -0.14599887 0.04574107 0.5160412 0.3186713
0.5567793 -0.37612024 -0.37471294 0.21933584 -1.0786234 -0.33463877
0.43491933 0.7981031 0.72571003 0.4721546 0.47221756 -0.84062004
1.2164867 -0.69246227 -0.9522451 0.141655 -0.9642554 0.66748136
0.65928817 -0.39382866 -0.9275417 0.49113235 0.5702303 -0.6392808
0.10511867 0.8221263 -0.70544267 0.35471678 0.37804097 0.23547627
0.3817885 -0.6225941 0.41229346 0.60710925 -0.06517564 -0.9474341
0.87587684 0.42911017 0.39929304 -0.75776786 -0.4551839 -0.4859255
-0.5197561 0.17071384 0.63127124 -0.7490147 -0.80101484 -0.01695774
-0.83424234 -0.13383596 -0.60193413 0.55574507 -0.99204683 0.35699037
0.11874644 -0.9927309 0.3177366 -1.3378831 1.0490125 -0.1968335
-0.10760362 -0.9367209 0.21299793 0.16743286 0.06335128 0.5252347
1.2019457 -0.62268794 -0.28009215 -0.40458146 0.05426022 0.41036934
0.2573814 -0.1473534 -0.49769843 -0.33101645 0.29072648 0.02818723
0.6155893 0.55471325 1.0371532 -0.6765587 -0.21141422 0.58005047
1.377854 0.17576134 0.25542906 0.29536998 0.64767987 1.2593874
1.0046852 0.75529486 0.40516168 0.6042532 0.43912178 0.3202626
0.6072107 -0.11638473 0.2768988 0.34858984 0.38102835 -0.18817836
-1.183095 0.5161593 -2.228353 -0.7281479 0.03804747 2.362178
0.48779708 -0.12884757 0.40124655 -0.086747 0.62762636 0.09867416
-0.9170188 -0.6466003 -0.16052161 -0.65507054 0.6721849 0.47050294
-1.0641537 0.41306204 6.221689 0.32390448 -0.78069437 -0.3086554
0.602493 -0.47069255 -0.6413199 -0.12200394 -0.6603377 0.63599616
1.0972997 -0.6947462 0.5746541 1.0480701 0.811999 0.03329757
-1.446567 1.0746939 -0.09214886 -0.9664157 -0.04903961 0.6188886
0.78685224 -0.23239075 0.42405745 0.1526783 0.33854133 -0.8117151
0.46716747 0.51075745 0.36603868 -1.159027 0.3086025 0.42651957
-0.6567316 -0.4351357 -0.290507 0.15585403 0.30288056 0.82513434
-0.56597257 0.1620693 0.25320446 0.5882082 0.09889535 0.8445414
0.17073308 0.16163817 -0.5664352 0.06941148 0.38320246 -1.0248333
0.9568214 0.96285266 0.2707493 0.14972167 0.15674436 1.0139235
0.06803808 0.10784746 -0.89316535 -0.5567353 0.18197267 0.89754677
-0.38657996 0.07700153 -0.3219739 0.80955017 0.00960189 0.71677476
-0.5324128 -0.29824656 1.4580694 0.03349532 0.28545156 -0.60807186
-0.45185402 -1.2204192 0.29308742 -0.8875008 0.3711322 -0.30207843
-1.195619 0.03012249 0.22181883 -1.2723005 -0.5020731 -0.61103827
-0.31788242 1.048366 -1.3159984 -0.4748975 0.39660004 0.3778551
0.40855125 -0.19432092 0.5583933 0.24058594 -0.9620736 0.01326254
0.7164071 -0.317896 0.27112746 -1.4211617 -0.08182276 0.926452
-0.55166686 0.5646243 1.0486665 -0.5977366 -2.2226388 -1.3260396
0.50963205 -0.24907358 1.0115513 -0.42393956 -0.61234796 0.56759936
-0.53625447 -0.0881384 0.7631763 0.12394386 -0.08856499 -0.09738504
-1.2541492 0.579441 0.7342985 -0.5072987 -0.42095563 0.656465
0.64952147 -0.08188136 -1.074269 -0.00746215 0.19509575 -0.11747691
0.9818517 -1.1378863 0.32236046 -0.22675896 -0.45808202 -1.6236004
-0.16781364 -0.88623214 -0.34305972 1.2145114 0.3125493 -0.50054735
1.0830243 1.3494451 0.11962085 -0.78005284 -0.85980284 -1.0652102
0.2609733 -0.40447953 0.6180118 0.7415731 0.08900995 0.23843054
-0.3168721 0.4285327 1.0586852 0.34732875 0.65094763 -1.249336
-0.18742949 -0.35005453 -0.41537642 -0.9853303 0.5507101 -0.7538602
0.1210117 -1.2952713 -0.01548042 -0.56675434 0.08516622 0.07656042
0.02065788 -0.6740514 0.20030688 0.05076797 -0.30508295 0.84547246
0.7882852 -0.48441762 -0.61727124 0.12097637 -1.1324761 0.5703699
0.650455 -0.70896906 -0.76783365 -0.42050156 0.34782887 0.18841389
-0.06188729 -0.24933182 0.0342035 -0.6764566 -0.14710714 -0.31131184
0.4712491 -1.1652266 0.22432354 0.30257618 -0.34410518 0.10704643
0.02141846 0.88939714 -0.19902737 -0.1535776 0.55024046 0.3285621
-0.6261829 0.46744058 -0.43980533 0.12749319 1.3098512 -0.21817474
0.0092663 -0.30280906 -0.93568116 0.6885 0.4913127 0.5174204
0.6206838 -1.3769649 -0.70057726 0.1246027 0.11922469 0.07291674
0.06003543 0.88979197 -0.04990871 0.6803084 0.15775806 -0.42119274
-0.8367029 0.8251776 0.2747179 -0.24639404 -0.5050845 0.41505915
0.3197964 -0.43457252 0.08734883 -0.33423856 -0.06990167 0.47615117
0.15667097 0.62030995 -0.01005539 -0.8178501 -0.37145337 0.30860192
0.11328737 -0.22425815 1.7144824 -0.44848981 -0.01269289 0.2646077
1.6200486 -0.2563198 -1.3942249 -0.07105943 0.3172922 -0.43189237
0.38096172 -0.20924936 -1.0719018 0.45178282 0.32262397 0.35892007
1.1348222 -0.17425771 0.44001356 0.6183957 0.04269069 -1.485409
-0.2812085 0.4412949 0.90996635 -1.1588529 -0.09007131 -0.22804762
-0.8755285 1.1191607 0.2684181 -0.21534953 0.9024269 0.24633025
-0.33126393 -0.2764645 -0.74811125 -0.09767863 0.20442739 0.5791348
-0.0353772 0.37357816 -0.3760321 0.67890525 -0.10616393 -0.34976318
0.5598539 0.90642583 -0.07204518 -0.999246 -0.50824827 0.56587327
-0.19727503 0.26912317 -0.45366722 0.49241495 -0.4080183 0.8568783
0.12668131 -0.14606522 0.6395058 -0.05911905 0.02100775 -0.71991014
0.04127761 0.1170467 0.16221903 -0.79451096 -0.14181171 -1.1994277
-0.9245523 -0.16949053 0.03815763 0.3665883 1.0430893 0.81952703
0.47094575 0.38553095 0.86239815 -0.622764 -1.3012955 -0.25364763
-1.0667962 0.47469413 0.52050316 -0.5292505 -0.54987276 -0.1349016 ] | [7.124890327453613, 4.181656360626221] |
42d23a2b-2a01-47ff-872f-f5b4e9b8da85 | pointacl-adversarial-contrastive-learning-for | 2209.06971 | null | https://arxiv.org/abs/2209.06971v1 | https://arxiv.org/pdf/2209.06971v1.pdf | PointACL:Adversarial Contrastive Learning for Robust Point Clouds Representation under Adversarial Attack | Despite recent success of self-supervised based contrastive learning model for 3D point clouds representation, the adversarial robustness of such pre-trained models raised concerns. Adversarial contrastive learning (ACL) is considered an effective way to improve the robustness of pre-trained models. In contrastive learning, the projector is considered an effective component for removing unnecessary feature information during contrastive pretraining and most ACL works also use contrastive loss with projected feature representations to generate adversarial examples in pretraining, while "unprojected " feature representations are used in generating adversarial inputs during inference.Because of the distribution gap between projected and "unprojected" features, their models are constrained of obtaining robust feature representations for downstream tasks. We introduce a new method to generate high-quality 3D adversarial examples for adversarial training by utilizing virtual adversarial loss with "unprojected" feature representations in contrastive learning framework. We present our robust aware loss function to train self-supervised contrastive learning framework adversarially. Furthermore, we find selecting high difference points with the Difference of Normal (DoN) operator as additional input for adversarial self-supervised contrastive learning can significantly improve the adversarial robustness of the pre-trained model. We validate our method, PointACL on downstream tasks, including 3D classification and 3D segmentation with multiple datasets. It obtains comparable robust accuracy over state-of-the-art contrastive adversarial learning methods. | ['Chunming Qiao', 'Junsong Yuan', 'Bai Chen', 'Lu Cheng', 'Yatong An', 'Junxuan Huang'] | 2022-09-14 | null | null | null | null | ['3d-classification'] | ['computer-vision'] | [ 1.30431712e-01 3.14227223e-01 1.61760598e-01 -1.54960990e-01
-9.21557546e-01 -8.87251198e-01 8.26494992e-01 -3.79225522e-01
-2.91835368e-01 5.91662586e-01 -2.08916724e-01 -3.51129889e-01
2.42590010e-01 -1.18509519e+00 -1.37496269e+00 -6.04016185e-01
-8.43989626e-02 6.17273629e-01 2.26520211e-01 -5.71438134e-01
6.24835044e-02 1.23243546e+00 -1.34901822e+00 2.06050918e-01
1.01495481e+00 8.08092833e-01 -3.02830994e-01 5.85545123e-01
-2.04964757e-01 4.39536035e-01 -9.50229704e-01 -6.16509557e-01
8.58974218e-01 -1.92303151e-01 -7.46861994e-01 -2.95312554e-01
8.40248406e-01 -2.40580738e-01 -5.06579220e-01 1.03474867e+00
7.03762949e-01 1.70001820e-01 1.17312288e+00 -1.66359186e+00
-8.65110278e-01 1.74305841e-01 -3.07566732e-01 2.21050158e-02
2.17183277e-01 5.10522604e-01 4.76762682e-01 -9.97641385e-01
5.44908226e-01 1.38209856e+00 9.20026958e-01 9.53396738e-01
-1.15080023e+00 -8.92977476e-01 -2.50049215e-02 -1.93928108e-01
-1.18578815e+00 9.78583768e-02 1.31970274e+00 -5.21429062e-01
8.14019918e-01 5.87502658e-01 6.66626930e-01 1.33466685e+00
4.28464711e-01 5.93680441e-01 1.31784093e+00 -2.69445986e-01
-8.68432131e-03 3.86052459e-01 -4.60394949e-01 7.34631836e-01
-2.28665143e-01 9.26003754e-01 4.47899029e-02 -2.99062341e-01
9.50462162e-01 1.33986861e-01 -1.22732118e-01 -4.10692245e-01
-6.96007848e-01 9.03337300e-01 9.95007217e-01 -5.69792688e-02
-1.55612454e-01 1.19609572e-01 3.70105326e-01 5.77514827e-01
8.30375671e-01 6.66972160e-01 -4.31494743e-01 4.14053649e-01
-5.55633008e-01 6.04978085e-01 4.37339485e-01 1.10421348e+00
7.26594985e-01 4.30278569e-01 -3.77474219e-01 8.01217854e-01
2.45465532e-01 8.16325843e-01 3.13476354e-01 -8.14180851e-01
6.29033864e-01 5.86442411e-01 -2.45330349e-01 -7.82331944e-01
1.30424965e-02 -4.15129215e-01 -1.04406846e+00 1.28014684e+00
2.70480901e-01 -1.74422547e-01 -1.41189778e+00 1.54980946e+00
4.08476800e-01 4.03719068e-01 3.09281081e-01 6.89955115e-01
8.93593013e-01 7.46410728e-01 7.26359114e-02 7.53687993e-02
6.21256948e-01 -7.07152307e-01 -1.53374493e-01 3.72015648e-02
1.30863026e-01 -8.92592490e-01 1.28058445e+00 6.41184449e-02
-1.23919570e+00 -8.02773058e-01 -1.07333422e+00 1.20728165e-01
-6.05924249e-01 -7.25730300e-01 2.68540770e-01 6.18997872e-01
-6.53462887e-01 1.24254918e+00 -6.67336643e-01 3.33662271e-01
9.25674915e-01 3.31389874e-01 -4.30139989e-01 6.76059276e-02
-1.21081376e+00 1.08243096e+00 3.22965570e-02 -2.98589736e-01
-1.29527986e+00 -1.27915084e+00 -9.83596206e-01 -3.52055401e-01
-1.34234846e-01 -8.72953415e-01 7.24938512e-01 -9.09322500e-01
-1.51660526e+00 1.10405076e+00 5.31663001e-01 -4.41836506e-01
1.10001254e+00 -2.49599233e-01 -1.25507057e-01 1.09893575e-01
-5.54395691e-02 8.78779292e-01 1.41474748e+00 -1.78902793e+00
-4.27856259e-02 -2.75460273e-01 6.92102313e-02 3.07476580e-01
1.25504537e-02 -1.73509642e-01 1.75144468e-02 -1.15454960e+00
1.00032426e-01 -1.10331893e+00 -4.29877102e-01 6.00719273e-01
-4.61916029e-01 -8.86067078e-02 1.26174557e+00 -4.16808367e-01
2.98231393e-01 -2.04802823e+00 3.64456400e-02 3.00412744e-01
9.66429859e-02 5.58364928e-01 -1.51594251e-01 1.71098188e-02
-5.27119160e-01 4.40727144e-01 -6.68896794e-01 -2.31999919e-01
3.53341252e-02 3.12434405e-01 -6.88292861e-01 5.12717485e-01
7.80127287e-01 1.03779566e+00 -9.29515660e-01 -4.78563339e-01
6.21625781e-01 6.58639371e-01 -6.42468333e-01 6.58521652e-01
-1.09586880e-01 8.57651174e-01 -3.01086426e-01 6.68131948e-01
1.09709644e+00 5.30926466e-01 -8.89214754e-01 -1.21036671e-01
4.48714495e-01 -2.20395446e-01 -9.41459477e-01 1.48265874e+00
-5.70867538e-01 2.83116579e-01 -3.14579666e-01 -8.12087297e-01
1.10488558e+00 2.86996514e-01 3.03782731e-01 -3.52572680e-01
4.17185389e-03 8.51497948e-02 -3.01427186e-01 -1.95289060e-01
1.39023721e-01 -6.42946184e-01 -1.76391289e-01 -1.42624021e-01
2.20404714e-01 -1.17978418e+00 -6.67635560e-01 -7.13329529e-03
9.29719687e-01 5.62230110e-01 -1.40553802e-01 1.85141321e-02
5.85740983e-01 -1.00783296e-02 5.21466374e-01 5.51955581e-01
-3.02487046e-01 1.19852173e+00 2.02022031e-01 -3.84828836e-01
-1.23574948e+00 -1.75276220e+00 -2.91265219e-01 4.20702726e-01
4.79850508e-02 1.73628479e-01 -5.78765869e-01 -1.43037665e+00
4.14698988e-01 7.82640815e-01 -6.77329302e-01 -5.64730465e-01
-8.13378334e-01 -3.10791939e-01 9.75464582e-01 6.65274441e-01
5.62584877e-01 -1.25537086e+00 1.80750236e-01 -1.08906150e-01
4.12402689e-01 -7.12888777e-01 -2.42843881e-01 2.94125170e-01
-1.00367856e+00 -1.04458904e+00 -8.98401260e-01 -7.83215284e-01
9.74067748e-01 -2.39801347e-01 1.32350671e+00 3.34987305e-02
-1.60326406e-01 3.25751752e-01 -1.96617261e-01 -8.06101382e-01
-9.50063765e-01 -4.37296122e-01 1.10767640e-01 -5.23023665e-01
-2.22385287e-01 -9.09066379e-01 -4.30394381e-01 3.04238915e-01
-9.28543687e-01 -5.34002602e-01 3.56685191e-01 9.62384641e-01
8.17298651e-01 -9.19467732e-02 4.63799596e-01 -1.08701026e+00
4.45270419e-01 -4.62121636e-01 -4.34471011e-01 -2.42425755e-01
-4.00389105e-01 -5.18271476e-02 1.28658259e+00 -5.71469069e-01
-8.22725832e-01 -1.43738464e-01 -8.30030560e-01 -1.40767336e+00
-3.11703771e-01 -3.78019691e-01 -3.20654809e-01 -5.92277944e-01
1.17143190e+00 -6.05899654e-03 1.47387937e-01 -2.06483394e-01
4.71538454e-01 9.53222290e-02 4.60598379e-01 -7.71562576e-01
1.79494274e+00 3.87791395e-01 4.03526217e-01 -4.92853254e-01
-5.15459001e-01 9.66189876e-02 -6.53471947e-01 -5.44984080e-02
6.00746751e-01 -7.76616871e-01 -3.20369631e-01 6.85664296e-01
-1.05335367e+00 -2.82355487e-01 -9.48699951e-01 1.11046486e-01
-8.60103309e-01 2.71866769e-01 -4.02953327e-01 -5.41269898e-01
-5.60913503e-01 -1.21698642e+00 9.57927585e-01 -1.20661072e-01
1.98657420e-02 -1.11832857e+00 4.93466519e-02 2.36600652e-01
2.74572104e-01 1.08929920e+00 8.67229283e-01 -6.35587275e-01
-2.41679534e-01 -4.33222264e-01 1.94016591e-01 1.10062683e+00
3.95054556e-02 1.95969325e-02 -1.15289438e+00 -3.84757459e-01
-4.42846380e-02 -4.23021019e-01 5.63750207e-01 5.89570068e-02
1.46002889e+00 -4.65830415e-01 -2.12699860e-01 1.02246213e+00
1.41561592e+00 -1.48901433e-01 8.61212313e-01 1.31780341e-01
1.04409933e+00 3.35327655e-01 6.50391042e-01 -1.02988653e-01
-2.84364402e-01 3.32944483e-01 8.78212035e-01 -3.92268568e-01
-3.05814624e-01 -4.65722710e-01 2.28674337e-01 3.47849399e-01
-2.57520199e-01 -6.51055649e-02 -7.69778311e-01 3.05332124e-01
-1.24497199e+00 -9.94310379e-01 1.08283542e-01 2.08630300e+00
8.51325154e-01 5.92071533e-01 -9.20021012e-02 3.48711669e-01
6.27261817e-01 1.96471140e-01 -6.76750124e-01 -6.66350961e-01
-5.71568459e-02 8.44251990e-01 5.40011883e-01 4.84184146e-01
-1.29860842e+00 1.06400239e+00 5.80232716e+00 8.98102760e-01
-1.17264247e+00 1.18755519e-01 4.75329727e-01 1.04077533e-01
-5.29638886e-01 -2.86348760e-01 -3.59103531e-01 6.50582314e-01
4.58021969e-01 8.21266621e-02 1.72312576e-02 1.20528722e+00
-1.04659565e-01 6.42714143e-01 -1.11609697e+00 8.56445134e-01
-6.39443398e-02 -1.43116069e+00 6.05209649e-01 -6.31786585e-02
1.04349196e+00 -1.78566158e-01 4.57418174e-01 6.23245835e-01
4.79920477e-01 -1.24249768e+00 5.32510102e-01 5.96081495e-01
9.33989465e-01 -1.22824204e+00 6.53345823e-01 2.72694647e-01
-8.74712765e-01 2.07273334e-01 -4.52662140e-01 2.72172779e-01
-1.62291788e-02 4.50129479e-01 -8.41071248e-01 6.61620200e-01
7.70120203e-01 4.75465596e-01 -4.45281565e-01 6.73115194e-01
-5.04131854e-01 4.59869653e-01 -2.43113235e-01 4.86467481e-01
2.61765569e-01 -9.74779874e-02 1.14035916e+00 8.45616996e-01
1.13745965e-01 -5.74611942e-04 1.44053578e-01 1.11336517e+00
-2.48843148e-01 -5.48893400e-02 -1.25118566e+00 4.59873855e-01
4.43487227e-01 8.34439874e-01 -1.25061288e-01 -9.20702815e-02
-3.56718190e-02 1.17889106e+00 2.40419433e-01 1.35836497e-01
-1.02666664e+00 -5.12653053e-01 9.16580975e-01 3.10065061e-01
9.84357893e-02 -1.11760594e-01 -5.12745738e-01 -7.16089368e-01
1.33906916e-01 -8.43106449e-01 7.59036839e-02 -7.10742414e-01
-1.86598301e+00 7.71946013e-01 2.67794989e-02 -1.91799831e+00
-2.54542053e-01 -5.79272985e-01 -1.16088474e+00 1.15395796e+00
-1.64367628e+00 -1.67090189e+00 -2.66806215e-01 9.24614310e-01
4.80450511e-01 -5.15319526e-01 9.23797011e-01 3.94718610e-02
-8.77485201e-02 1.00373662e+00 -1.28421858e-01 2.29587644e-01
8.41383815e-01 -1.59715760e+00 6.87084377e-01 7.80828297e-01
-2.00318575e-01 2.50951707e-01 6.87697172e-01 -6.64190054e-01
-1.03388822e+00 -1.71687555e+00 1.44974872e-01 -8.38543415e-01
1.82491466e-01 -1.08897626e-01 -1.01763892e+00 7.40048647e-01
-2.10785363e-02 5.55724919e-01 5.42026460e-01 -4.43732351e-01
-2.82737195e-01 -5.29672205e-02 -1.97799635e+00 6.98376834e-01
1.23815978e+00 -4.30555642e-01 -9.93240118e-01 4.05636460e-01
1.23035610e+00 -8.22465360e-01 -1.21562827e+00 8.93930674e-01
2.52435710e-02 -7.49956310e-01 1.46868837e+00 -6.85938597e-01
7.36168563e-01 -2.35577509e-01 4.15113792e-02 -1.75030160e+00
-8.94130766e-02 -6.70639575e-01 -8.87399539e-02 1.29905415e+00
2.81569034e-01 -7.22054660e-01 1.11068630e+00 3.29763472e-01
-6.51066005e-01 -8.97672355e-01 -8.66154552e-01 -9.89069402e-01
1.07735097e+00 -3.83266091e-01 6.61598384e-01 1.11069489e+00
-7.47195005e-01 -1.55472115e-01 -1.67722493e-01 4.10867542e-01
8.17051053e-01 -3.09222192e-01 1.01285267e+00 -1.11175847e+00
-8.20371136e-02 -1.28346756e-01 -7.31283247e-01 -6.15237355e-01
5.28793275e-01 -1.21664631e+00 -9.11941379e-02 -1.12858415e+00
-5.29993653e-01 -9.14156020e-01 -8.86472240e-02 4.37791854e-01
-3.94890666e-01 5.15472293e-01 3.28702450e-01 4.13050391e-02
3.30426544e-01 7.59703577e-01 1.85064471e+00 -5.45296729e-01
-1.90613180e-01 4.09409672e-01 -3.65241140e-01 8.39792073e-01
7.19854236e-01 -6.89398885e-01 -4.68445122e-01 -1.09959312e-01
-2.54903495e-01 -3.02109182e-01 7.06974864e-01 -1.10300064e+00
-1.70388043e-01 -1.69489503e-01 9.39774752e-01 -6.30004346e-01
5.65539837e-01 -1.03137982e+00 -1.89537685e-02 4.81969744e-01
1.83201134e-02 -5.92376664e-02 6.02861941e-01 3.95948082e-01
-2.13994414e-01 -3.06456953e-01 1.22011507e+00 -4.04687226e-01
-3.92940432e-01 7.21707106e-01 1.73552260e-01 3.64264905e-01
1.15360487e+00 -3.60621929e-01 -7.61546791e-02 3.53505649e-02
-9.37658191e-01 -1.72599517e-02 6.68707967e-01 4.07680303e-01
9.74623322e-01 -1.64975381e+00 -8.34454834e-01 4.79865789e-01
-9.89971980e-02 6.68168902e-01 1.47343144e-01 -9.31241363e-02
-7.41940856e-01 -3.55640531e-01 -6.98523939e-01 -6.90094531e-01
-1.09725082e+00 8.76986444e-01 5.64434767e-01 -2.68653780e-01
-7.05806255e-01 1.04741919e+00 3.23808305e-02 -1.13124609e+00
2.00250745e-01 -2.06048004e-02 -1.94908883e-02 -4.14807469e-01
-7.69745782e-02 3.62742037e-01 1.46160007e-01 -3.93150419e-01
-2.23508343e-01 7.45647013e-01 -4.91855219e-02 8.94924179e-02
1.25987041e+00 4.98221606e-01 2.56386846e-01 1.11350909e-01
1.54168022e+00 1.68684006e-01 -1.42046010e+00 3.56101431e-02
-7.63051212e-01 -5.23113668e-01 -2.26015106e-01 -5.94581664e-01
-1.20684028e+00 8.56615484e-01 8.08638871e-01 -8.13069418e-02
8.55888486e-01 -1.96690977e-01 7.66656041e-01 1.22166149e-01
2.73907095e-01 -6.24467850e-01 4.58264619e-01 5.06750464e-01
1.45017612e+00 -1.43314230e+00 2.95352414e-02 -7.40200520e-01
-6.67947829e-01 8.04315746e-01 9.73412395e-01 -1.00520408e+00
8.94262016e-01 3.74232322e-01 2.82346338e-01 -6.05643280e-02
-6.66804910e-02 1.04447842e-01 5.70623696e-01 1.12006354e+00
-1.09020330e-01 -1.59689948e-01 7.86046386e-02 2.86101043e-01
-6.16075754e-01 -6.13236368e-01 1.00347616e-01 1.03991330e+00
1.35545880e-01 -1.17111337e+00 -6.35205030e-01 2.95835823e-01
-3.17438841e-01 4.93851863e-03 -4.08448815e-01 1.06050372e+00
3.46194625e-01 5.08533478e-01 1.92016006e-01 -5.01295984e-01
8.06707680e-01 5.28771020e-02 7.79827118e-01 -6.86749339e-01
-1.07008791e+00 -5.72243869e-01 -3.13282579e-01 -4.15765136e-01
5.27199619e-02 -2.76823014e-01 -1.27547586e+00 -3.49632442e-01
-1.45489767e-01 2.03058124e-02 4.70981807e-01 7.76369035e-01
1.69075564e-01 7.47942388e-01 1.31103146e+00 -1.24403465e+00
-8.61210644e-01 -7.68165767e-01 -2.28341535e-01 9.29451048e-01
3.58232170e-01 -7.73806691e-01 -7.48314798e-01 -1.36212319e-01] | [7.901355266571045, -4.263686656951904] |
ec611185-1838-40da-96d2-bcfa705d69e8 | asbert-siamese-and-triplet-network-embedding | 2104.08558 | null | https://arxiv.org/abs/2104.08558v1 | https://arxiv.org/pdf/2104.08558v1.pdf | ASBERT: Siamese and Triplet network embedding for open question answering | Answer selection (AS) is an essential subtask in the field of natural language processing with an objective to identify the most likely answer to a given question from a corpus containing candidate answer sentences. A common approach to address the AS problem is to generate an embedding for each candidate sentence and query. Then, select the sentence whose vector representation is closest to the query's. A key drawback is the low quality of the embeddings, hitherto, based on its performance on AS benchmark datasets. In this work, we present ASBERT, a framework built on the BERT architecture that employs Siamese and Triplet neural networks to learn an encoding function that maps a text to a fixed-size vector in an embedded space. The notion of distance between two points in this space connotes similarity in meaning between two texts. Experimental results on the WikiQA and TrecQA datasets demonstrate that our proposed approach outperforms many state-of-the-art baseline methods. | ['Olabanji Shonibare'] | 2021-04-17 | null | null | null | null | ['answer-selection'] | ['natural-language-processing'] | [ 4.15575117e-01 -1.04450680e-01 1.56857893e-01 -5.62610269e-01
-1.17775714e+00 -5.46609223e-01 4.71842378e-01 6.64240956e-01
-9.03557003e-01 3.56900930e-01 4.33633655e-01 -3.25811952e-01
-4.03003931e-01 -7.06629753e-01 -3.30106169e-01 -3.26936513e-01
1.37526989e-01 6.08886659e-01 3.72281700e-01 -5.76588929e-01
8.25644493e-01 4.93338518e-02 -1.62733090e+00 6.16887808e-01
9.47632551e-01 1.08314288e+00 1.79300904e-01 7.63603508e-01
-7.74327099e-01 7.63546705e-01 -6.39076710e-01 -8.10910881e-01
1.10248841e-01 -7.47436523e-01 -1.36116040e+00 -4.49449629e-01
6.34146154e-01 1.55872568e-01 -1.65017173e-01 1.07552779e+00
3.95202488e-01 3.45732778e-01 7.54907548e-01 -1.01799238e+00
-9.77485180e-01 4.07672197e-01 8.73934329e-02 6.33207917e-01
8.66363347e-01 -2.19693840e-01 1.55416024e+00 -1.33457696e+00
5.18710136e-01 1.38019431e+00 3.31393868e-01 4.36372042e-01
-1.11208749e+00 -2.15056896e-01 -9.89607871e-02 5.31256914e-01
-1.22493947e+00 -2.25993827e-01 7.96302617e-01 -1.48455992e-01
1.10220051e+00 4.73309427e-01 2.94208556e-01 5.93797684e-01
5.02239704e-01 9.65320408e-01 7.41953790e-01 -6.21433735e-01
4.04973865e-01 2.49891177e-01 6.44247293e-01 7.18936622e-01
-3.17966372e-01 -5.51766872e-01 -6.40172303e-01 -4.40964609e-01
-2.47555971e-01 -1.19468130e-01 -3.73095840e-01 -3.37948442e-01
-1.19401264e+00 1.35417688e+00 4.69935507e-01 4.10313487e-01
-3.01518232e-01 2.45778114e-02 4.85752940e-01 9.33379710e-01
3.23779196e-01 1.02269387e+00 -3.13024223e-01 -1.80718124e-01
-7.58707106e-01 5.71253359e-01 9.05566990e-01 5.11082172e-01
7.83586800e-01 -4.91775274e-01 -6.44832194e-01 6.74978256e-01
2.93051720e-01 2.64283985e-01 7.59977818e-01 -6.78774595e-01
9.92799938e-01 1.11541963e+00 9.82108712e-02 -1.44240987e+00
-4.99443226e-02 -1.48747608e-01 -4.09319550e-01 -1.82735801e-01
2.81192809e-01 9.33806691e-03 -2.75159687e-01 1.30841792e+00
4.87028569e-01 3.15967314e-02 4.01258320e-01 9.93705451e-01
1.17872465e+00 8.36263478e-01 -2.87152529e-01 1.38259828e-01
1.52353346e+00 -1.03148162e+00 -7.33112812e-01 -3.09265822e-01
6.37142241e-01 -7.80526698e-01 1.36232793e+00 -9.22445729e-02
-8.54054689e-01 -5.64441681e-01 -1.11281264e+00 -4.22465652e-01
-6.36651337e-01 1.83783889e-01 6.71178326e-02 3.92729491e-01
-9.17036712e-01 5.38918018e-01 -2.25982919e-01 -4.86209273e-01
-1.21618390e-01 3.02929431e-01 -3.40636343e-01 -6.14582784e-02
-1.53737199e+00 1.16211331e+00 4.75457162e-01 6.31388798e-02
-1.15538262e-01 -4.68831778e-01 -1.00677383e+00 2.69995123e-01
2.12240696e-01 -6.47604227e-01 1.17570364e+00 -7.07845986e-01
-1.47532713e+00 8.66731644e-01 -4.76443291e-01 -6.94104373e-01
-4.66155224e-02 -2.33369902e-01 -4.44077581e-01 4.69923735e-01
3.07007790e-01 5.26781619e-01 9.38241005e-01 -8.29414070e-01
-6.22579336e-01 -4.42065358e-01 1.60510436e-01 4.42671925e-01
-5.90228200e-01 2.50375479e-01 -2.73325592e-01 -3.93358231e-01
2.89748698e-01 -5.80017805e-01 1.30410837e-02 -1.56906247e-01
-2.32926354e-01 -1.07557380e+00 8.48061860e-01 -4.89940315e-01
1.69039583e+00 -1.84945869e+00 4.27987725e-01 1.33794233e-01
1.51151374e-01 3.22771788e-01 -4.71194267e-01 9.03361320e-01
2.44146842e-03 -4.05877866e-02 -3.75097036e-01 -1.82806924e-01
2.94612885e-01 8.74097645e-03 -6.28236949e-01 3.68198246e-01
5.87461531e-01 8.10525715e-01 -1.03869200e+00 -6.41943693e-01
-1.11792326e-01 1.42771557e-01 -4.07031924e-01 5.00327587e-01
-2.59467453e-01 -8.20781365e-02 -6.89457417e-01 2.42421746e-01
3.15030843e-01 -1.50640532e-01 -2.81287640e-01 -2.92973761e-02
1.43121958e-01 5.06775856e-01 -1.16374445e+00 1.73575592e+00
-3.06081682e-01 5.95671952e-01 -2.69210696e-01 -1.11058736e+00
1.30631268e+00 3.67793173e-01 5.25489450e-02 -7.69971371e-01
5.50959371e-02 5.12337387e-01 4.83571477e-02 -9.84883606e-01
8.75665128e-01 2.80435127e-03 -3.08079064e-01 7.13496387e-01
8.12673941e-02 -3.42080444e-01 4.82676178e-01 3.43793392e-01
1.35681915e+00 -1.38016269e-01 1.94578767e-01 -2.18675062e-01
1.20821321e+00 1.12844920e-02 1.82906225e-01 6.61369920e-01
-3.21896464e-01 6.11490190e-01 5.56501806e-01 -4.77169514e-01
-7.28095770e-01 -8.53951275e-01 2.30767578e-02 9.60348427e-01
1.21355295e-01 -4.28200632e-01 -7.36700118e-01 -9.81809139e-01
1.09376788e-01 8.77752960e-01 -8.81181896e-01 -2.46703580e-01
-8.46089184e-01 -8.27983022e-02 3.45102787e-01 1.45416379e-01
2.03076288e-01 -1.27722275e+00 -7.21318662e-01 2.82736450e-01
-3.55201602e-01 -8.34988773e-01 -7.70594895e-01 5.92875816e-02
-6.46018326e-01 -1.13400996e+00 -6.01320326e-01 -1.24955285e+00
6.04435444e-01 2.44233429e-01 1.35419464e+00 1.29640773e-01
1.08953737e-01 3.52680802e-01 -6.27434433e-01 -6.83443546e-02
-3.70751232e-01 3.09590220e-01 -1.92374900e-01 4.54249650e-01
9.71465290e-01 8.71799216e-02 -7.00045645e-01 7.84194320e-02
-1.13926232e+00 -5.69083273e-01 1.83931530e-01 1.01125276e+00
4.88076299e-01 -2.57633179e-01 8.24577272e-01 -6.85837567e-01
1.46329260e+00 -6.00071192e-01 -2.27896258e-01 6.71766579e-01
-4.57008451e-01 5.24156451e-01 8.02694321e-01 -2.62699798e-02
-7.15768993e-01 -1.66629210e-01 -1.02599896e-01 6.41584024e-02
1.72316030e-01 8.31501663e-01 1.97727323e-01 2.04972684e-01
6.47910893e-01 3.11347336e-01 4.97025140e-02 -4.01026696e-01
4.59740877e-01 9.31767642e-01 2.72003442e-01 -4.22516227e-01
7.12237418e-01 7.37878159e-02 -3.22516322e-01 -7.83610404e-01
-1.06646025e+00 -9.01453972e-01 -6.66564524e-01 6.25570044e-02
8.69308174e-01 -2.20877215e-01 -6.44548178e-01 -3.38337332e-01
-1.48021424e+00 5.66044390e-01 -2.55961806e-01 4.78250593e-01
-4.46503431e-01 3.86991978e-01 -2.27001071e-01 -7.58776844e-01
-7.43427932e-01 -1.14864981e+00 9.43538010e-01 2.83405870e-01
-6.17159367e-01 -9.86010313e-01 4.33409691e-01 5.51160634e-01
4.55025405e-01 -1.44052684e-01 1.29070628e+00 -1.15656793e+00
-3.04621935e-01 -5.96870840e-01 -1.54395690e-02 2.46924162e-01
-1.72505025e-02 -2.33005702e-01 -6.33838952e-01 -3.11143696e-01
4.10996318e-01 -6.15493774e-01 8.69215071e-01 -1.73998103e-01
8.46965611e-01 -3.00898105e-01 -3.95664060e-03 1.35599807e-01
1.35462034e+00 1.49663985e-01 3.86007190e-01 5.09667099e-01
1.96663022e-01 9.67745602e-01 7.34255791e-01 -5.39316505e-04
4.12719101e-01 5.61067760e-01 2.64300317e-01 3.33459735e-01
2.77431328e-02 -3.56522888e-01 3.07745427e-01 1.30291593e+00
8.86080503e-01 -3.59986156e-01 -7.76107609e-01 7.21671104e-01
-1.78407729e+00 -1.00837147e+00 -1.28858745e-01 2.13871789e+00
8.09623420e-01 6.01275377e-02 -3.06274265e-01 1.99985862e-01
5.68066776e-01 2.10777044e-01 -4.22335267e-01 -7.59943783e-01
-1.09455980e-01 5.78687668e-01 -1.42321989e-01 5.74940681e-01
-1.07966578e+00 7.33758628e-01 5.46915293e+00 7.29460180e-01
-7.54160225e-01 -7.65157565e-02 3.44631284e-01 1.07958637e-01
-4.86391097e-01 -7.07534477e-02 -7.10444987e-01 2.49960661e-01
1.01664364e+00 -3.71673733e-01 7.85138384e-02 4.82163668e-01
-6.32643094e-03 -4.65197973e-02 -1.40407956e+00 7.92951584e-01
6.83844864e-01 -1.30945659e+00 -2.71961908e-03 -6.45536482e-01
5.36631644e-01 -1.35784999e-01 1.32209405e-01 5.61186850e-01
-6.55919015e-02 -1.13294280e+00 3.82182211e-01 4.70314652e-01
3.11305434e-01 -7.50194490e-01 1.08327675e+00 2.68740296e-01
-1.17143810e+00 -1.60261974e-01 -6.40812874e-01 3.54344286e-02
-2.88778469e-02 2.67491606e-03 -8.16748321e-01 2.59744555e-01
5.89513481e-01 3.68932486e-01 -8.34895432e-01 1.02855015e+00
-3.42139840e-01 5.42648792e-01 -1.46265596e-01 -7.69700527e-01
6.69435740e-01 -2.03911677e-01 4.71288830e-01 1.15978515e+00
3.90391588e-01 -3.23098786e-02 1.16968714e-01 9.12052810e-01
-2.31827006e-01 7.26627052e-01 -8.25110137e-01 2.13128552e-02
5.40755749e-01 1.04231012e+00 -4.26590055e-01 -2.26994231e-01
-4.62962866e-01 1.23852813e+00 6.12476289e-01 3.76319259e-01
-3.43011230e-01 -1.16183519e+00 3.83533865e-01 -3.26157033e-01
3.84485960e-01 1.68994233e-01 -1.50709987e-01 -1.13531351e+00
5.78613341e-01 -1.05063021e+00 5.81592262e-01 -4.53679711e-01
-1.43363845e+00 8.70010197e-01 -3.93857718e-01 -1.38118041e+00
-2.89339870e-01 -5.44834316e-01 -8.68340671e-01 1.13152182e+00
-1.47839737e+00 -7.25302756e-01 9.24566984e-02 3.30638617e-01
8.53492737e-01 -4.26752120e-01 1.15599251e+00 5.75708412e-02
-3.08837414e-01 6.01564586e-01 3.90509397e-01 2.68644214e-01
4.74770457e-01 -1.45674324e+00 4.84812379e-01 7.28809118e-01
4.92200375e-01 7.97593892e-01 6.97895467e-01 -2.19186977e-01
-1.63600159e+00 -7.32867479e-01 1.78139555e+00 -7.80396283e-01
8.55872571e-01 -1.82769284e-01 -1.14553320e+00 1.70413122e-01
6.70714378e-01 -1.23803817e-01 8.50903690e-01 -7.02221990e-02
-3.11962813e-01 -2.27031231e-01 -1.08017063e+00 6.50766969e-01
2.05512598e-01 -7.83240974e-01 -1.24117410e+00 2.73937345e-01
8.85414779e-01 -6.97755814e-02 -8.28536034e-01 -5.72283864e-02
2.40926966e-01 -8.72942865e-01 8.03616345e-01 -9.59480941e-01
6.46189690e-01 -5.13682544e-01 -3.25287372e-01 -1.40470803e+00
-2.28077304e-02 -4.72436458e-01 -9.86746922e-02 1.12123585e+00
6.46144450e-01 -4.23202604e-01 9.07470703e-01 5.60639620e-01
2.27603447e-02 -1.17074347e+00 -1.32427454e+00 -4.66392010e-01
3.78474623e-01 -5.73051833e-02 7.83171952e-01 6.19883001e-01
2.74883002e-01 9.65299428e-01 7.32717589e-02 -6.70183674e-02
2.98494101e-01 3.57044846e-01 4.99473333e-01 -1.14661658e+00
8.38087574e-02 -3.52048129e-01 -5.11201799e-01 -1.19086969e+00
5.58087230e-01 -1.23400986e+00 1.33135438e-01 -1.75003779e+00
-7.87349939e-02 -1.38635226e-02 -3.77361566e-01 -1.64360523e-01
-5.39765000e-01 -8.99563730e-02 1.80975154e-01 -5.18856831e-02
-8.69875133e-01 1.02559125e+00 1.00542152e+00 -5.04021585e-01
-6.09475486e-02 -4.81369607e-02 -5.94342887e-01 4.86801058e-01
7.61774123e-01 -7.26934731e-01 -5.65469086e-01 -7.93671548e-01
4.19315100e-01 1.52141705e-01 3.25899236e-02 -7.15046704e-01
4.92439389e-01 4.57702577e-02 -4.89961207e-02 -7.90505409e-01
3.26810330e-01 -8.92650604e-01 -6.92894280e-01 4.59759086e-01
-1.02776754e+00 7.40036786e-01 -2.57490665e-01 5.52853405e-01
-7.91880846e-01 -8.92479599e-01 3.67968410e-01 6.65319934e-02
-5.76764643e-01 9.78604541e-04 -1.14400275e-01 5.47099888e-01
7.36754179e-01 -1.49897605e-01 -2.63822585e-01 -3.47630501e-01
-7.20795169e-02 6.69858277e-01 -6.89195022e-02 7.67249167e-01
1.29031038e+00 -1.52107847e+00 -1.00919664e+00 5.25209270e-02
6.36376917e-01 -3.39972854e-01 -7.73109049e-02 4.46318179e-01
-5.68398952e-01 7.51135170e-01 3.97841930e-01 -3.66743654e-01
-1.33406031e+00 3.42846394e-01 2.49361292e-01 -3.95388246e-01
-3.97027582e-01 1.14342785e+00 -4.12016243e-01 -8.76819611e-01
2.92556167e-01 -4.43735600e-01 -8.61804545e-01 1.22673728e-01
6.98180199e-01 2.12020203e-01 2.72373021e-01 -8.01616669e-01
-3.79612565e-01 5.59112668e-01 -1.68041050e-01 -1.63882494e-01
1.07869673e+00 -1.27919987e-02 -4.79888469e-01 4.94057655e-01
1.96171629e+00 -3.00743490e-01 -4.66463298e-01 -5.64082682e-01
6.43380940e-01 -5.53255796e-01 -2.19372258e-01 -3.65252376e-01
-4.95190710e-01 9.37609673e-01 7.39911318e-01 4.69074965e-01
6.95379674e-01 4.94063320e-03 1.13986981e+00 1.14402330e+00
-1.28392026e-01 -1.30058587e+00 4.07815069e-01 8.51134956e-01
1.10517919e+00 -1.40527618e+00 -3.54782343e-01 1.09890543e-01
-7.40354776e-01 1.46185029e+00 5.65544665e-01 -1.51652381e-01
5.80400944e-01 -5.50962269e-01 3.31253529e-01 -4.97040272e-01
-9.25501585e-01 -1.55039266e-01 5.89296401e-01 2.36417368e-01
7.56141841e-01 -3.06294352e-01 -6.44636571e-01 2.44076177e-01
-3.37806493e-01 -3.40688348e-01 2.93224305e-01 9.71800148e-01
-6.76568687e-01 -1.10366964e+00 -3.40607792e-01 5.93324423e-01
-4.38465804e-01 -1.47309005e-01 -5.67681134e-01 1.95329487e-01
-1.51164874e-01 1.35009539e+00 -5.05110994e-03 -2.98277706e-01
5.52311718e-01 3.19006354e-01 8.58853459e-02 -8.77525508e-01
-1.01703286e+00 -6.38877809e-01 9.75809097e-02 -2.59169608e-01
-2.70720869e-01 -5.30062854e-01 -1.27552903e+00 1.49477571e-01
-3.71850520e-01 8.69791329e-01 5.55192173e-01 9.29538488e-01
3.73724848e-01 1.51016518e-01 7.57097542e-01 -1.24529004e-02
-1.21860993e+00 -8.50514233e-01 -2.40973711e-01 7.84140646e-01
4.13315624e-01 -2.70580590e-01 -3.18623304e-01 -2.74968296e-01] | [11.248127937316895, 8.120339393615723] |
0af4f8d6-a14d-4909-bb23-eca99c73c017 | multi-agent-path-finding-with-prioritized | 2202.03634 | null | https://arxiv.org/abs/2202.03634v2 | https://arxiv.org/pdf/2202.03634v2.pdf | Multi-Agent Path Finding with Prioritized Communication Learning | Multi-agent pathfinding (MAPF) has been widely used to solve large-scale real-world problems, e.g., automation warehouses. The learning-based, fully decentralized framework has been introduced to alleviate real-time problems and simultaneously pursue optimal planning policy. However, existing methods might generate significantly more vertex conflicts (or collisions), which lead to a low success rate or more makespan. In this paper, we propose a PrIoritized COmmunication learning method (PICO), which incorporates the \textit{implicit} planning priorities into the communication topology within the decentralized multi-agent reinforcement learning framework. Assembling with the classic coupled planners, the implicit priority learning module can be utilized to form the dynamic communication topology, which also builds an effective collision-avoiding mechanism. PICO performs significantly better in large-scale MAPF tasks in success rates and collision rates than state-of-the-art learning-based planners. | ['Xiangfeng Wang', 'Hongyuan Zha', 'Wenzhe Tan', 'Bo Jin', 'Hongjun Chen', 'Wenhao Li'] | 2022-02-08 | null | null | null | null | ['pico', 'multi-agent-path-finding'] | ['natural-language-processing', 'playing-games'] | [-2.54995227e-01 4.06940371e-01 -1.89307213e-01 1.68608874e-02
-5.07771492e-01 -3.60856086e-01 4.89192516e-01 4.63371307e-01
-5.60715497e-01 1.30508101e+00 -4.79732193e-02 -2.90307641e-01
-7.33205140e-01 -1.12296557e+00 -5.83085477e-01 -7.97750294e-01
-6.00965798e-01 1.06082940e+00 5.68552256e-01 -6.80190146e-01
2.99180806e-01 3.43059033e-01 -1.20846224e+00 -2.69682080e-01
1.13505781e+00 4.69355345e-01 8.36118221e-01 2.81918883e-01
-3.76636803e-01 4.53318179e-01 -5.28983653e-01 2.06878945e-01
4.28974777e-01 1.76223703e-02 -9.13288832e-01 -1.46660283e-02
-6.50565267e-01 -2.72645801e-01 -2.35476762e-01 6.64399385e-01
5.97841501e-01 3.27326089e-01 4.19717461e-01 -1.77676761e+00
3.39072257e-01 1.03540480e+00 -9.61554885e-01 2.55437754e-02
3.91912818e-01 1.54610574e-01 7.21751153e-01 -2.64861912e-01
6.25543177e-01 1.47898984e+00 1.26828372e-01 6.79092556e-02
-1.02226043e+00 -6.53664708e-01 6.26713097e-01 4.35343146e-01
-1.16276908e+00 2.01740831e-01 6.77598894e-01 1.97286457e-01
9.82725084e-01 -3.18800099e-02 8.11880469e-01 6.00084901e-01
7.87169456e-01 7.55795181e-01 1.01668835e+00 -3.46471280e-01
4.01986867e-01 -4.87221032e-01 -4.18837488e-01 6.38940513e-01
4.05220360e-01 2.04844117e-01 -1.37369618e-01 -9.76784751e-02
8.96935642e-01 -8.82418156e-02 1.32988647e-01 -7.16138005e-01
-1.45671785e+00 1.00613558e+00 5.75726807e-01 4.49698046e-02
-7.21169889e-01 1.36450022e-01 2.90253043e-01 4.01953518e-01
-1.89474925e-01 8.23558748e-01 -4.85475808e-01 -1.11137986e-01
-2.33250991e-01 6.93720341e-01 9.07111585e-01 1.21590674e+00
1.06707358e+00 -8.46476182e-02 -1.26577599e-03 5.22763312e-01
3.87794048e-01 3.50723594e-01 -1.69927448e-01 -1.22007430e+00
8.07621539e-01 7.51730323e-01 4.56808388e-01 -9.97249365e-01
-1.08419526e+00 -7.21461028e-02 -7.64192522e-01 4.89349127e-01
2.66385913e-01 -4.74315554e-01 -5.60264409e-01 1.53495133e+00
6.62391365e-01 -3.42291147e-02 2.47642398e-01 8.79383922e-01
8.01139623e-02 9.77352977e-01 -3.29937153e-02 -5.47358155e-01
1.07450175e+00 -1.28567135e+00 -8.05347204e-01 -1.62906513e-01
5.49833298e-01 -6.00753129e-01 5.94798446e-01 4.43329006e-01
-1.02055323e+00 -6.86046109e-02 -1.03415644e+00 6.49734199e-01
-3.29420775e-01 -5.75459838e-01 9.18473959e-01 2.63776422e-01
-1.00671864e+00 4.68627602e-01 -7.89535761e-01 -4.72067654e-01
-1.44669153e-02 5.99086106e-01 -8.39011222e-02 -2.75645047e-01
-1.25503588e+00 1.18884480e+00 7.63371587e-01 -6.46577775e-02
-1.25782371e+00 -2.55932063e-01 -6.24972999e-01 -4.16158661e-02
1.39565909e+00 -5.21404862e-01 1.31088805e+00 -2.24394947e-01
-1.72375286e+00 -3.05461079e-01 3.14439207e-01 -6.87124878e-02
5.36869168e-01 1.61032662e-01 -1.40601933e-01 2.70052394e-03
2.34205619e-01 7.58076072e-01 4.20999110e-01 -1.67002273e+00
-1.19927156e+00 2.16094498e-02 6.25861466e-01 8.70773256e-01
-9.36419237e-03 -5.31865120e-01 -1.00044578e-01 -1.61465675e-01
-9.21592675e-03 -1.13181627e+00 -1.00594378e+00 -4.54675704e-01
-2.06152439e-01 -5.93914211e-01 6.77523017e-01 3.26976441e-02
1.14283788e+00 -1.66219366e+00 4.81455028e-01 4.78101581e-01
1.15258314e-01 2.81873764e-03 -3.95063758e-01 1.07821500e+00
5.80651641e-01 -3.34619343e-01 5.68337999e-02 2.53643006e-01
1.59140185e-01 7.80690908e-01 1.10433921e-01 1.30897179e-01
-4.29889038e-02 5.69302440e-01 -1.35335588e+00 -4.77395654e-01
2.70472974e-01 -3.53123397e-01 -4.75903213e-01 3.28836851e-02
-3.68184209e-01 2.94769228e-01 -9.32662070e-01 5.66111803e-01
6.80303574e-01 1.46485418e-01 6.23962581e-01 3.41674179e-01
-6.39588535e-01 -1.84757486e-01 -1.57855189e+00 1.84400630e+00
-3.62796485e-01 -1.25095561e-01 5.77564001e-01 -1.04004550e+00
9.22814310e-01 -1.21400738e-02 9.09494579e-01 -9.48403776e-01
-5.43753654e-02 2.47877136e-01 7.98321962e-02 -2.39019558e-01
7.03003407e-01 3.53650093e-01 -3.44833702e-01 4.67278033e-01
-3.56421828e-01 -1.87800229e-01 5.55050731e-01 2.33197421e-01
1.32771707e+00 1.35772303e-01 4.04397607e-01 -6.53245747e-01
5.53724289e-01 5.03400803e-01 8.35707247e-01 8.85556877e-01
-3.34483087e-01 -3.66394371e-01 4.92029786e-01 -6.10820889e-01
-7.19778955e-01 -9.68299210e-01 6.17449343e-01 9.52827096e-01
1.01464963e+00 -3.97350222e-01 -4.53814715e-01 -5.03991425e-01
2.04968691e-01 5.55857778e-01 -9.53379497e-02 3.95688340e-02
-8.40322673e-01 -7.68561482e-01 7.91618749e-02 1.29207432e-01
5.31198919e-01 -1.27693880e+00 -8.66139770e-01 9.39413249e-01
-2.92471915e-01 -8.13495874e-01 -3.85969043e-01 3.43251675e-01
-4.71166342e-01 -1.33646595e+00 -4.12024111e-01 -8.06288421e-01
5.84938824e-01 6.49051309e-01 6.68779910e-01 -1.19243249e-01
-5.65989502e-02 2.82150984e-01 -6.21185720e-01 -5.31669557e-01
-2.86865085e-01 3.48957449e-01 2.03297988e-01 -5.27439594e-01
-1.25847906e-01 -3.57477933e-01 -6.24567688e-01 6.45385265e-01
-6.28664613e-01 2.05652252e-01 9.57609892e-01 7.57429183e-01
5.42757332e-01 6.64608359e-01 9.82960582e-01 -4.25567299e-01
1.10963619e+00 -4.56945777e-01 -7.95509160e-01 3.53679597e-01
-9.98594761e-01 1.60610929e-01 7.76682019e-01 -2.20320225e-01
-1.00541067e+00 -1.10376418e-01 4.24610227e-01 1.77571595e-01
1.79731511e-02 8.11176121e-01 -8.26286748e-02 -2.39570335e-01
1.17934212e-01 -5.52579155e-03 2.34813124e-01 8.13024417e-02
1.48151115e-01 2.24862173e-01 2.04683896e-02 -8.28238070e-01
6.36378944e-01 5.02390154e-02 3.22433650e-01 -3.18825573e-01
6.58261627e-02 -3.44608247e-01 -3.02660763e-01 -3.92334461e-01
4.41780686e-01 -8.11933935e-01 -1.12066007e+00 3.46580297e-01
-1.11601436e+00 -7.11631954e-01 2.07237974e-01 3.14911872e-01
-8.64729166e-01 2.17910931e-01 -4.49072212e-01 -8.13020527e-01
-9.92894755e-04 -1.43847215e+00 5.98287940e-01 4.41905200e-01
5.67904562e-02 -6.48752213e-01 1.32130414e-01 -1.14306279e-01
5.39858878e-01 4.47797596e-01 9.80606735e-01 -1.91847220e-01
-9.26517308e-01 5.08719087e-01 -1.65991664e-01 -7.74012506e-01
4.25955579e-02 -2.43719146e-01 3.46763246e-02 -6.53998256e-01
-6.35105729e-01 -1.96427673e-01 1.80904746e-01 3.07923228e-01
5.63285768e-01 -3.66947144e-01 -8.18165779e-01 6.62914186e-04
1.52520621e+00 8.38203251e-01 5.10820985e-01 9.55462575e-01
2.08426893e-01 7.58045495e-01 1.32743907e+00 9.03256416e-01
1.02645135e+00 7.88519502e-01 8.55827928e-01 8.09339620e-03
3.78025353e-01 -1.55693009e-01 3.94383669e-01 7.78220534e-01
-1.83305904e-01 -4.21917111e-01 -1.02366281e+00 3.55852276e-01
-2.60660434e+00 -7.21134305e-01 8.13900959e-03 1.71146274e+00
5.42673767e-01 2.45912030e-01 1.86460838e-01 7.71446526e-02
7.31322587e-01 1.09343022e-01 -5.06791532e-01 -5.61167181e-01
2.00737894e-01 -3.76385391e-01 8.68077397e-01 6.19302332e-01
-9.20852602e-01 9.94785190e-01 5.83737659e+00 8.35391283e-01
-6.99537933e-01 -5.28451391e-02 7.25396816e-03 1.74744338e-01
-2.57261038e-01 9.88878459e-02 -5.82306147e-01 3.26965064e-01
6.49163723e-01 -4.34626132e-01 9.98546481e-01 6.94845557e-01
6.18553340e-01 -5.99716365e-01 -6.77695453e-01 7.13752151e-01
-5.50294816e-01 -1.23907840e+00 -6.33505955e-02 2.97329366e-01
7.53982842e-01 -1.14101410e-01 -5.45507848e-01 5.66495538e-01
9.03027892e-01 -5.93678117e-01 6.74569666e-01 1.38765603e-01
2.79014885e-01 -1.38954711e+00 8.16191256e-01 7.02214241e-01
-1.53814745e+00 -6.47017956e-01 -4.62706029e-01 -3.94789398e-01
7.39746809e-01 3.49708587e-01 -9.73125875e-01 1.19098926e+00
5.42757750e-01 1.71513945e-01 9.93863400e-03 1.23388720e+00
-1.68867245e-01 -2.51172483e-01 -5.04319012e-01 -5.21952808e-01
8.74997497e-01 -4.32246387e-01 6.58983171e-01 5.89137435e-01
1.62522644e-01 2.70552337e-01 1.25404239e+00 3.91732097e-01
5.08779287e-01 -4.04118299e-02 -5.82454681e-01 1.76554784e-01
7.93459058e-01 1.39086187e+00 -1.09175920e+00 2.54119020e-02
-2.15190083e-01 4.31977510e-01 3.67304713e-01 2.95253724e-01
-9.15519416e-01 -5.30617297e-01 4.59808439e-01 -1.42163709e-01
4.55904193e-02 -5.17519712e-01 -1.29480129e-02 -5.19110143e-01
-2.44610325e-01 -8.64405334e-01 2.42523938e-01 -3.52380633e-01
-9.14334655e-01 3.97212207e-01 3.75378549e-01 -1.00015271e+00
-2.17211038e-01 -2.79265285e-01 -7.30070293e-01 3.40143085e-01
-2.02743554e+00 -8.86460721e-01 -3.38989735e-01 7.10543156e-01
7.15284824e-01 -4.23700035e-01 8.05340171e-01 2.02560663e-01
-5.61649442e-01 9.46213827e-02 6.91750348e-02 -5.31870365e-01
5.59964061e-01 -1.16193771e+00 -2.34994739e-02 5.04728734e-01
-7.16877997e-01 1.82009086e-01 7.12027133e-01 -8.74866247e-01
-1.94060230e+00 -9.00884807e-01 3.18461806e-01 2.81881750e-01
6.49290323e-01 -1.85293909e-02 -3.03973168e-01 2.62325734e-01
3.89237255e-01 -5.58047533e-01 1.11830696e-01 3.76251005e-02
5.25931537e-01 -3.71764541e-01 -1.08069122e+00 8.33325326e-01
1.10672331e+00 5.31397104e-01 -3.16010147e-01 2.47616425e-01
8.23936880e-01 -4.39001203e-01 -6.84644639e-01 3.97520274e-01
2.88702369e-01 -5.72129369e-01 6.53391302e-01 -1.84452221e-01
6.49920478e-02 -5.11965036e-01 1.86192349e-01 -1.80967283e+00
-8.28518391e-01 -1.00399709e+00 2.99628209e-02 9.72313941e-01
4.26037967e-01 -8.81557941e-01 6.99314237e-01 3.81467998e-01
-4.32497740e-01 -8.89484406e-01 -1.11789906e+00 -8.72210979e-01
-1.92275763e-01 2.61328906e-01 8.06726038e-01 6.02685034e-01
3.47876400e-01 2.50398070e-01 -2.86333531e-01 3.24167669e-01
7.13293493e-01 -3.16648223e-02 9.01941478e-01 -1.15807438e+00
5.75377904e-02 -3.05220544e-01 1.10472664e-01 -7.71878242e-01
8.65809172e-02 -5.38576126e-01 3.06885988e-01 -2.01913619e+00
-3.00299793e-01 -1.20300436e+00 -2.93597281e-01 6.29703343e-01
1.27928674e-01 -8.03623676e-01 5.26138663e-01 2.18447119e-01
-1.00161946e+00 6.61583841e-01 1.75792873e+00 -2.46055424e-01
-5.85859835e-01 -9.05152336e-02 -6.02749467e-01 3.93427879e-01
1.11746705e+00 -2.95222878e-01 -7.75943398e-01 -6.00392401e-01
1.96866408e-01 7.43766189e-01 -2.96517342e-01 -7.12017000e-01
6.98902667e-01 -9.79238927e-01 -2.60118276e-01 -6.20925188e-01
9.88668650e-02 -1.02158999e+00 2.43204594e-01 1.02604330e+00
-1.10363495e-02 7.91512907e-01 1.10279955e-01 9.14177477e-01
-1.12803139e-01 -8.78597796e-02 3.04128230e-01 -4.81409490e-01
-1.21081400e+00 2.73941010e-01 -7.14288592e-01 -2.08340615e-01
1.71787560e+00 -6.21990189e-02 -4.77532119e-01 -3.02780896e-01
-3.04776669e-01 1.35098350e+00 1.37858391e-01 3.95657152e-01
6.03225708e-01 -1.14227891e+00 -5.72706342e-01 -4.77029085e-02
-1.72241673e-01 3.43379200e-01 2.00583249e-01 8.96991789e-01
-6.74407303e-01 4.90637183e-01 -6.48529232e-01 -2.27770552e-01
-6.64175332e-01 5.98400235e-01 -1.50295377e-01 -8.53797853e-01
-6.56792700e-01 2.13579014e-01 -2.17697993e-01 -6.04732156e-01
3.03806633e-01 -6.54586330e-02 -2.70239979e-01 2.17124656e-01
2.66362429e-01 7.52902806e-01 -2.68199533e-01 1.14905678e-01
-5.74011624e-01 3.16541404e-01 -1.25588357e-01 -2.46419594e-01
1.42852449e+00 -4.47604269e-01 2.50649806e-02 -2.40976870e-01
1.84233233e-01 -4.27730262e-01 -1.39593375e+00 6.06917925e-02
2.77424186e-01 -3.02247435e-01 -4.04770970e-02 -8.55160356e-01
-8.52227271e-01 1.01589173e-01 1.85048789e-01 3.44251961e-01
7.99624443e-01 -3.98423105e-01 8.14501822e-01 7.38442242e-01
1.48545182e+00 -1.51882756e+00 2.26350769e-01 7.04801857e-01
7.36940980e-01 -1.02750075e+00 2.79449299e-02 -5.65017998e-01
-8.64870608e-01 1.09822476e+00 1.01066220e+00 -2.18559623e-01
2.73495167e-01 5.28453767e-01 -3.67664918e-02 -8.42219517e-02
-9.32175338e-01 -2.13662639e-01 -8.29885542e-01 8.93351793e-01
-4.27563280e-01 3.44941676e-01 -8.42046022e-01 1.33241996e-01
9.63848904e-02 -1.45982906e-01 8.72969866e-01 1.36732900e+00
-8.45931709e-01 -1.35682166e+00 -4.39777911e-01 2.75771290e-01
3.43387574e-01 5.56584179e-01 5.58210574e-02 9.62371647e-01
4.39968407e-02 1.08157003e+00 5.99140003e-02 -1.69617251e-01
4.87450868e-01 -4.19167072e-01 4.60308641e-01 -3.54424119e-01
-3.71335626e-01 2.68543035e-01 4.06454086e-01 -7.49133945e-01
-5.26856124e-01 -5.19257843e-01 -1.71080685e+00 -4.22367156e-01
-1.73259720e-01 3.54518205e-01 3.73656124e-01 9.02716935e-01
3.60525101e-01 8.00773263e-01 7.60987699e-01 -1.07725549e+00
-6.23251259e-01 -5.28248489e-01 -4.54095155e-01 -2.31900975e-01
-6.42200112e-02 -1.21054530e+00 1.69902340e-01 -9.07571912e-01] | [4.725723743438721, 1.67091965675354] |
a713a258-b1e1-4f12-bf88-a7ed7e4a462c | text-gestalt-stroke-aware-scene-text-image | 2112.08171 | null | https://arxiv.org/abs/2112.08171v1 | https://arxiv.org/pdf/2112.08171v1.pdf | Text Gestalt: Stroke-Aware Scene Text Image Super-Resolution | In the last decade, the blossom of deep learning has witnessed the rapid development of scene text recognition. However, the recognition of low-resolution scene text images remains a challenge. Even though some super-resolution methods have been proposed to tackle this problem, they usually treat text images as general images while ignoring the fact that the visual quality of strokes (the atomic unit of text) plays an essential role for text recognition. According to Gestalt Psychology, humans are capable of composing parts of details into the most similar objects guided by prior knowledge. Likewise, when humans observe a low-resolution text image, they will inherently use partial stroke-level details to recover the appearance of holistic characters. Inspired by Gestalt Psychology, we put forward a Stroke-Aware Scene Text Image Super-Resolution method containing a Stroke-Focused Module (SFM) to concentrate on stroke-level internal structures of characters in text images. Specifically, we attempt to design rules for decomposing English characters and digits at stroke-level, then pre-train a text recognizer to provide stroke-level attention maps as positional clues with the purpose of controlling the consistency between the generated super-resolution image and high-resolution ground truth. The extensive experimental results validate that the proposed method can indeed generate more distinguishable images on TextZoom and manually constructed Chinese character dataset Degraded-IC13. Furthermore, since the proposed SFM is only used to provide stroke-level guidance when training, it will not bring any time overhead during the test phase. Code is available at https://github.com/FudanVI/FudanOCR/tree/main/text-gestalt. | ['xiangyang xue', 'Bin Li', 'jianqi ma', 'Haiyang Yu', 'Jingye Chen'] | 2021-12-13 | null | null | null | null | ['scene-text-recognition'] | ['computer-vision'] | [ 5.25788724e-01 -2.03147724e-01 3.76907596e-03 -3.41197848e-01
-3.52748841e-01 -3.65899444e-01 8.08180153e-01 -2.85114080e-01
-1.54571995e-01 3.17523062e-01 4.08187807e-01 -6.61253855e-02
-4.68962938e-02 -9.87561584e-01 -5.66817760e-01 -7.51439333e-01
6.73703432e-01 2.06730768e-01 3.43123853e-01 -3.46491009e-01
6.25415146e-01 7.00640261e-01 -1.59985566e+00 6.37906313e-01
1.04818547e+00 5.06452203e-01 6.59179807e-01 6.56413674e-01
-4.28204894e-01 8.18054497e-01 -5.44622362e-01 -3.33359003e-01
1.37773186e-01 -6.29980922e-01 -5.86154401e-01 4.75234389e-01
6.92093670e-01 -5.85508823e-01 -5.24998724e-01 1.18752491e+00
4.11639959e-01 7.40703046e-02 6.65508687e-01 -4.71469700e-01
-9.80793953e-01 7.45706260e-01 -7.62206912e-01 2.27965355e-01
2.97183633e-01 -2.99397372e-02 7.08779335e-01 -1.03478372e+00
5.90359688e-01 1.22929668e+00 2.65488297e-01 5.10582507e-01
-1.03233612e+00 -5.60098648e-01 2.70617187e-01 2.59084165e-01
-1.22878289e+00 -4.03153390e-01 9.36038136e-01 -2.36648828e-01
5.26501834e-01 3.68276805e-01 3.69635522e-01 1.19352496e+00
1.67777374e-01 8.73785138e-01 1.14074767e+00 -3.96328837e-01
-3.34486738e-02 2.37504646e-01 1.37699336e-01 6.11646175e-01
3.26103926e-01 -1.10310793e-01 -5.79596877e-01 5.12256384e-01
1.36260128e+00 1.27946228e-01 -5.24942040e-01 -1.68763265e-01
-1.19196916e+00 3.21576983e-01 4.47790474e-01 6.46341562e-01
-2.73557007e-01 -2.86383182e-01 6.96427524e-02 -1.41987592e-01
1.33997545e-01 2.79313117e-01 9.48353186e-02 1.20301478e-01
-1.01360834e+00 -7.08563849e-02 3.60465169e-01 7.87206411e-01
5.10084987e-01 4.08303112e-01 -2.53504515e-01 8.34726155e-01
-1.59626618e-01 3.38095009e-01 6.04042709e-01 -5.46972215e-01
6.11134887e-01 7.58982182e-01 -4.18172032e-02 -1.11075604e+00
-2.01590464e-01 -5.16357124e-01 -1.19891143e+00 3.44855934e-01
4.56520379e-01 2.04249710e-01 -9.11958873e-01 1.19500554e+00
1.30623192e-01 -1.16953611e-01 3.14246863e-02 1.28589988e+00
7.07982957e-01 6.92326427e-01 -2.90378749e-01 -1.67976506e-02
1.44846702e+00 -7.88055778e-01 -7.45192647e-01 -1.46099254e-01
-5.16716577e-02 -7.38413274e-01 1.49321866e+00 4.46438283e-01
-1.05580783e+00 -8.61412466e-01 -1.27407050e+00 -3.29055965e-01
-2.83655524e-01 4.51777399e-01 2.73597986e-01 4.85597938e-01
-6.72583222e-01 5.60114503e-01 -6.09434664e-01 -3.17771316e-01
4.70892072e-01 -1.42579734e-01 -2.35280648e-01 -1.13271810e-01
-8.58280122e-01 7.58792818e-01 3.80663276e-01 2.60206401e-01
-8.37304533e-01 -3.90704066e-01 -4.94810045e-01 2.49569312e-01
3.47558230e-01 -4.02387887e-01 6.84191287e-01 -1.17900336e+00
-1.64692652e+00 6.78031802e-01 -1.63020417e-01 -3.90642090e-03
7.04922140e-01 -1.89111754e-01 -4.09828305e-01 3.48423600e-01
-1.61102653e-01 3.09154898e-01 1.21099877e+00 -1.56086206e+00
-5.93565285e-01 -3.74654114e-01 -3.66908759e-02 6.14395320e-01
-3.24871242e-01 4.94934320e-02 -5.51174462e-01 -8.58303905e-01
3.43327194e-01 -2.79027253e-01 1.29033700e-01 -1.43106431e-01
-4.22874331e-01 5.72457202e-02 9.52173769e-01 -9.00544226e-01
1.02803695e+00 -2.01515794e+00 2.45857492e-01 -1.81966454e-01
1.80946365e-01 2.48024777e-01 -2.24352852e-01 1.07083969e-01
-2.45257933e-02 9.67327878e-02 -2.98230678e-01 -2.40578040e-01
-2.93538719e-01 -1.77018970e-01 -4.74429309e-01 2.75201201e-01
1.80891290e-01 8.28505874e-01 -5.91765463e-01 -5.27539253e-01
5.30920386e-01 8.20498526e-01 -1.38021320e-01 6.41360506e-02
-1.76114589e-01 4.85509157e-01 -5.46133161e-01 5.48840880e-01
9.07652557e-01 -2.23839968e-01 1.15949489e-01 -4.26454276e-01
-3.09513867e-01 -3.29998285e-02 -1.29716611e+00 1.47460532e+00
-2.65005767e-01 7.59312391e-01 1.32810995e-01 -9.01196837e-01
1.05947089e+00 -1.73097067e-02 1.44656733e-04 -9.96861160e-01
2.33679295e-01 -2.54361391e-01 -1.75788417e-01 -4.12287742e-01
7.96679854e-01 1.37243181e-01 2.24915877e-01 4.30556208e-01
-4.09710735e-01 -1.44535273e-01 1.59931872e-02 1.19012721e-01
4.65558738e-01 4.51481432e-01 5.61809540e-02 -7.31878728e-02
7.72201836e-01 2.06491109e-02 2.22887978e-01 5.99564552e-01
1.55299902e-01 1.02378333e+00 2.92524129e-01 -3.79904211e-01
-1.26284146e+00 -9.83716905e-01 -3.04166317e-01 1.14264810e+00
4.35528278e-01 -9.35937390e-02 -9.83516037e-01 -2.47266948e-01
-4.50502723e-01 7.01874256e-01 -5.98077953e-01 1.96863204e-01
-7.19987929e-01 -6.80143595e-01 5.37432611e-01 4.73667234e-01
1.09866393e+00 -1.14391911e+00 -8.36843014e-01 -9.62159336e-02
-2.00507253e-01 -1.26057386e+00 -5.04310966e-01 -1.09277323e-01
-8.19488347e-01 -9.35061336e-01 -8.92290413e-01 -8.13947976e-01
8.48934054e-01 4.64790404e-01 6.87296689e-01 1.10326953e-01
-4.83885765e-01 1.82448775e-02 -4.28282410e-01 7.29204193e-02
-2.94231594e-01 -7.09932372e-02 -2.50352710e-01 3.18457156e-01
1.37568504e-01 -4.29033160e-01 -5.58385611e-01 3.92932445e-01
-9.46716666e-01 7.66609490e-01 8.66802812e-01 7.10045040e-01
5.70483565e-01 4.24748421e-01 1.22277819e-01 -7.72195756e-01
5.58698893e-01 1.86327279e-01 -5.99803746e-01 3.57251078e-01
-3.30264956e-01 1.18658766e-01 9.65475500e-01 -5.12564301e-01
-1.50535703e+00 -1.29425600e-01 2.01870367e-01 -3.27634901e-01
-4.71072167e-01 1.16876654e-01 -4.43553180e-01 2.03303784e-01
6.90708339e-01 8.12226593e-01 -2.18993276e-01 -5.86311936e-01
4.00398284e-01 8.02704751e-01 6.74264014e-01 -5.35804212e-01
8.87566507e-01 6.87512279e-01 -1.23968616e-01 -1.14742029e+00
-7.60763407e-01 4.05015511e-04 -9.25250351e-01 -2.54411280e-01
1.08365428e+00 -6.98827446e-01 -4.86530483e-01 6.10297918e-01
-8.67144525e-01 -4.25775230e-01 -5.73644079e-02 1.36069059e-01
-3.59116226e-01 6.89955711e-01 -6.06724799e-01 -6.89392984e-01
-2.78344631e-01 -9.48713422e-01 1.00688982e+00 5.16735911e-01
3.44250977e-01 -5.87646008e-01 -3.53145778e-01 5.57437181e-01
3.89303178e-01 2.39778727e-01 1.00404978e+00 -1.78252995e-01
-9.04563308e-01 -2.54978687e-02 -7.03328729e-01 2.41024852e-01
3.05046201e-01 7.46968836e-02 -9.41456258e-01 -5.47960438e-02
-2.44869087e-02 -6.42575473e-02 7.81685054e-01 3.43268037e-01
1.14913678e+00 -2.18878582e-01 3.67359780e-02 7.62201309e-01
1.45722187e+00 6.00745939e-02 8.43025565e-01 2.75974959e-01
1.06381881e+00 6.77775919e-01 4.97187644e-01 3.22510451e-01
2.25632802e-01 7.08530247e-01 2.22483531e-01 -2.60614663e-01
-5.40013373e-01 -4.29214299e-01 3.44506860e-01 5.32888353e-01
-4.03241873e-01 -9.67853591e-02 -6.68151021e-01 1.51409656e-01
-1.57436740e+00 -1.12509537e+00 -2.14291885e-01 2.19656110e+00
6.93517745e-01 1.56810656e-01 -1.35013416e-01 2.00212732e-01
9.51122880e-01 2.21136957e-01 -6.17537618e-01 -2.92417467e-01
-5.75937390e-01 -1.31636225e-02 3.65889132e-01 4.67782468e-01
-8.00126016e-01 1.32552528e+00 4.96454191e+00 9.58522439e-01
-1.31689143e+00 -2.32941926e-01 5.77985466e-01 1.58756360e-01
-2.55102754e-01 -1.52228802e-01 -8.55899394e-01 2.18220130e-01
3.00074130e-01 -1.63032115e-01 6.92455471e-01 5.37455857e-01
4.04759616e-01 -1.52723834e-01 -7.84494340e-01 9.80476737e-01
2.81166524e-01 -1.21014559e+00 5.47979772e-01 -1.68787554e-01
7.64452219e-01 -4.61209565e-01 2.46000350e-01 -2.81103924e-02
-1.33246198e-01 -1.23288953e+00 7.76242733e-01 7.21791089e-01
1.00009680e+00 -7.74932384e-01 3.32018912e-01 4.76547003e-01
-1.38232446e+00 -9.17002037e-02 -8.16394687e-01 4.96070050e-02
-1.54582128e-01 3.99375975e-01 -5.54508567e-01 7.35903859e-01
7.01209843e-01 7.72639871e-01 -7.69544661e-01 5.93361080e-01
-4.44054991e-01 2.07922414e-01 1.81689650e-01 2.21809261e-02
-5.97019494e-03 -4.07769769e-01 4.90986824e-01 1.19875181e+00
1.22981749e-01 3.13369215e-01 -2.49031857e-01 1.16938353e+00
7.66333416e-02 2.29484946e-01 -2.97386408e-01 -1.25395954e-01
2.98278749e-01 1.40707922e+00 -1.04085326e+00 -3.88166934e-01
-3.05269241e-01 1.38371384e+00 1.17698111e-01 5.90650141e-01
-7.84453332e-01 -5.26832044e-01 2.51353413e-01 2.71516621e-01
3.79027545e-01 -2.69577205e-01 -7.26659417e-01 -1.30894375e+00
4.97281104e-02 -1.15209305e+00 -2.73935292e-02 -1.15517259e+00
-7.84019887e-01 8.39670122e-01 -2.31877521e-01 -1.16393769e+00
1.78485766e-01 -7.18624353e-01 -5.52013814e-01 1.02028823e+00
-1.33284461e+00 -1.16065466e+00 -7.12050557e-01 7.30601192e-01
1.04387629e+00 2.26371065e-02 5.07493317e-01 -9.12377685e-02
-7.09939420e-01 4.69393134e-01 1.58412308e-01 3.86757642e-01
6.02387905e-01 -1.08075225e+00 3.15869689e-01 1.18327296e+00
1.72850832e-01 5.37141621e-01 6.95216835e-01 -7.25317776e-01
-1.38808417e+00 -7.87786961e-01 3.32205415e-01 -5.23177803e-01
2.63688058e-01 -6.36134207e-01 -1.21023655e+00 3.02493572e-01
1.73322082e-01 -4.64848399e-01 -4.19305488e-02 -2.67724395e-01
-3.80205035e-01 -7.68009797e-02 -9.88053679e-01 8.53185296e-01
1.02914834e+00 -5.80294907e-01 -7.74813414e-01 -1.11905225e-01
4.54481840e-01 -4.43113059e-01 -5.48003495e-01 4.01240617e-01
6.16530418e-01 -1.26992273e+00 9.01861846e-01 -1.93167001e-01
6.28975511e-01 -5.76349616e-01 -1.20192589e-02 -7.62754560e-01
-3.43974948e-01 -1.28080860e-01 1.28474534e-01 1.22603738e+00
1.49685264e-01 -2.98061550e-01 7.61095345e-01 3.21557254e-01
-4.77935039e-02 -4.34676200e-01 -5.45132160e-01 -3.47202092e-01
3.81228104e-02 -7.17007965e-02 5.26860297e-01 9.94192719e-01
-1.63509950e-01 3.16386074e-01 -4.50859547e-01 3.27395350e-01
7.94217527e-01 4.10921782e-01 7.68380165e-01 -1.05197668e+00
-2.01619565e-01 -8.72243583e-01 -8.69198516e-02 -1.14721274e+00
-9.96874198e-02 -6.48953140e-01 9.96412151e-03 -1.73585224e+00
3.73377144e-01 -1.48385260e-02 -9.78364721e-02 1.94539174e-01
-2.96077341e-01 1.89911842e-01 2.80450761e-01 2.91261256e-01
-2.94445246e-01 4.66375172e-01 1.69577813e+00 4.70821336e-02
-2.03946516e-01 -1.51267409e-01 -7.43650258e-01 6.26494706e-01
7.83247054e-01 7.45482743e-02 -3.25077683e-01 -4.52408165e-01
6.00234829e-02 9.59612951e-02 4.10474807e-01 -8.97484958e-01
3.54110569e-01 -2.61360586e-01 8.62205505e-01 -8.17888021e-01
1.79298326e-01 -6.45219743e-01 -3.33243087e-02 3.06162447e-01
-3.64978164e-01 -1.19217247e-01 1.07004993e-01 3.23555887e-01
-1.41047537e-01 -1.70761704e-01 1.01414251e+00 -1.16750322e-01
-7.60967433e-01 3.71125787e-02 -4.07629848e-01 -2.38445088e-01
6.29324675e-01 -6.04587555e-01 -6.32415116e-01 -2.39849702e-01
-3.52753043e-01 -5.23255654e-02 8.06432188e-01 5.36084592e-01
9.35640812e-01 -8.80069554e-01 -8.43166530e-01 2.82021672e-01
-1.04424320e-01 4.92243879e-02 6.07360959e-01 5.65504551e-01
-6.20032668e-01 3.92671674e-01 -4.46401834e-01 -5.16212821e-01
-1.24596369e+00 7.01579213e-01 4.73621368e-01 1.68267358e-02
-1.07525730e+00 5.75686574e-01 5.62172711e-01 2.39587128e-02
3.32857281e-01 -3.94705147e-01 -4.29284990e-01 -6.87136576e-02
7.62875259e-01 3.35264593e-01 -2.58258820e-01 -6.82723463e-01
-5.78827374e-02 1.12476587e+00 -1.06740400e-01 -9.57428664e-02
1.12897861e+00 -4.06187415e-01 3.00480165e-02 4.22317743e-01
5.63869357e-01 2.17287987e-01 -1.51838994e+00 -3.33751261e-01
-2.05342650e-01 -6.17333233e-01 1.15159824e-01 -8.19524944e-01
-9.37962055e-01 1.11350465e+00 5.95830858e-01 -3.79688255e-02
1.35083318e+00 -2.04470232e-01 2.72366613e-01 2.97540724e-01
2.23812819e-01 -1.06247163e+00 4.47662294e-01 3.91407967e-01
9.62143838e-01 -1.04641545e+00 6.26103431e-02 -3.31054598e-01
-9.35373604e-01 1.32254159e+00 7.56046653e-01 -1.61368102e-01
3.92246172e-02 2.95518935e-01 2.78278464e-03 2.45488584e-02
-4.30207163e-01 -3.97226632e-01 3.32577050e-01 5.09931147e-01
4.83019531e-01 -1.70150444e-01 -3.34166847e-02 5.54796159e-01
-2.27417842e-01 -2.86216766e-01 7.72045970e-01 4.25328732e-01
-7.24901080e-01 -7.91204214e-01 -5.75344622e-01 1.50281057e-01
-2.24423945e-01 -2.71901965e-01 -4.42406863e-01 6.30398810e-01
-7.87172690e-02 6.98520422e-01 1.47038355e-01 -2.48745114e-01
3.43464673e-01 -2.72197612e-02 6.55098259e-01 -4.51998502e-01
-3.28008384e-01 2.54654169e-01 -2.81502575e-01 -3.33727300e-01
-2.80498892e-01 -5.22939682e-01 -1.39924037e+00 -3.44138861e-01
-1.40006229e-01 -1.28787365e-02 6.67469978e-01 6.90978467e-01
7.85006136e-02 7.55846739e-01 6.81257486e-01 -9.38345551e-01
-1.62837744e-01 -1.00791895e+00 -5.23178637e-01 3.45633298e-01
2.10680082e-01 -2.52928048e-01 -3.59562874e-01 2.18860656e-01] | [11.461540222167969, -1.6381542682647705] |
a77095eb-9491-45dc-be81-343031728d71 | procan-progressive-growing-channel-attentive | 2010.15417 | null | https://arxiv.org/abs/2010.15417v3 | https://arxiv.org/pdf/2010.15417v3.pdf | ProCAN: Progressive Growing Channel Attentive Non-Local Network for Lung Nodule Classification | Lung cancer classification in screening computed tomography (CT) scans is one of the most crucial tasks for early detection of this disease. Many lives can be saved if we are able to accurately classify malignant/cancerous lung nodules. Consequently, several deep learning based models have been proposed recently to classify lung nodules as malignant or benign. Nevertheless, the large variation in the size and heterogeneous appearance of the nodules makes this task an extremely challenging one. We propose a new Progressive Growing Channel Attentive Non-Local (ProCAN) network for lung nodule classification. The proposed method addresses this challenge from three different aspects. First, we enrich the Non-Local network by adding channel-wise attention capability to it. Second, we apply Curriculum Learning principles, whereby we first train our model on easy examples before hard ones. Third, as the classification task gets harder during the Curriculum learning, our model is progressively grown to increase its capability of handling the task at hand. We examined our proposed method on two different public datasets and compared its performance with state-of-the-art methods in the literature. The results show that the ProCAN model outperforms state-of-the-art methods and achieves an AUC of 98.05% and an accuracy of 95.28% on the LIDC-IDRI dataset. Moreover, we conducted extensive ablation studies to analyze the contribution and effects of each new component of our proposed method. | ['Maxine Tan', 'Kelvin Shak', 'Mundher Al-Shabi'] | 2020-10-29 | null | null | null | null | ['lung-nodule-classification'] | ['medical'] | [ 3.02675933e-01 1.49267375e-01 -3.01446229e-01 -1.88041613e-01
-9.26004708e-01 -3.23124081e-01 5.16944289e-01 4.98506092e-02
-5.61740279e-01 6.05786502e-01 3.53397392e-02 -4.63765800e-01
-1.73235223e-01 -6.74636424e-01 -5.25135100e-01 -8.73404562e-01
9.29526389e-02 7.20607579e-01 6.46883190e-01 2.25180343e-01
-2.88667083e-01 4.13549483e-01 -1.06330013e+00 4.83658850e-01
9.93002594e-01 1.04637313e+00 4.81376261e-01 6.21365070e-01
5.92551269e-02 8.75387192e-01 -1.86311752e-01 -2.96668530e-01
2.61851907e-01 -4.32674497e-01 -8.78894985e-01 3.77244949e-02
3.22534591e-01 -2.82407075e-01 -3.10680538e-01 9.55215156e-01
4.32078481e-01 -3.19917798e-01 8.26766610e-01 -6.87758446e-01
-3.43438238e-01 5.65975428e-01 -6.15499914e-01 3.08568269e-01
-3.88317704e-01 9.66280028e-02 9.21030223e-01 -6.57178462e-01
4.15156662e-01 5.71791589e-01 5.67004681e-01 7.29904711e-01
-8.00971627e-01 -7.43712068e-01 5.00082932e-02 1.18200801e-01
-1.13036156e+00 9.32388976e-02 3.88617009e-01 -5.55745959e-01
3.25749815e-01 3.41635495e-01 7.05063105e-01 1.11238563e+00
3.51728052e-01 7.51104057e-01 1.07297981e+00 -1.20301574e-01
-1.03117295e-01 3.97847027e-01 -8.06268454e-02 8.34672630e-01
4.54905093e-01 1.33592356e-02 2.04236001e-01 -1.36637330e-01
8.47951472e-01 2.45442629e-01 -1.77430585e-01 -3.21983993e-01
-1.21254992e+00 8.57466638e-01 9.32776272e-01 6.16504431e-01
-3.25437784e-01 2.61808276e-01 3.43133956e-01 -9.26045775e-02
3.43506426e-01 3.43190998e-01 -2.41341069e-01 2.69998640e-01
-8.02194059e-01 -8.83503333e-02 5.62660575e-01 4.46691394e-01
1.59006402e-01 -2.79594094e-01 -5.90120792e-01 7.56431580e-01
1.42615914e-01 1.47230700e-01 8.53156686e-01 -2.88055837e-01
2.88715154e-01 6.03239596e-01 -1.72146216e-01 -4.57035094e-01
-5.77811062e-01 -1.13741171e+00 -1.20089316e+00 -1.91342562e-01
3.24042708e-01 -4.51762341e-02 -1.17792284e+00 1.47031522e+00
1.99173227e-01 2.60881037e-01 -1.51606411e-01 7.61102259e-01
9.85638022e-01 3.12392116e-01 5.03961205e-01 -1.67503003e-02
1.50765741e+00 -1.17064846e+00 -3.14935565e-01 -2.45635584e-01
7.63748944e-01 -5.25994837e-01 9.14329231e-01 4.70121056e-02
-7.75007129e-01 -5.23985267e-01 -8.71038139e-01 2.25058496e-01
-2.27343179e-02 4.91727471e-01 6.03420913e-01 7.02799737e-01
-9.35772836e-01 3.39597344e-01 -1.00932837e+00 -4.60601360e-01
6.48403168e-01 4.96544600e-01 -4.79633361e-02 -2.67075241e-01
-1.13175821e+00 6.97991073e-01 3.91164362e-01 -6.48050979e-02
-1.10541928e+00 -7.23339438e-01 -2.25409135e-01 2.41275027e-01
5.89993834e-01 -8.90058875e-01 1.43784940e+00 -1.01975787e+00
-1.05958009e+00 7.44031668e-01 1.09245077e-01 -4.47493643e-01
8.04778755e-01 1.15731433e-01 -5.48029393e-02 2.25580364e-01
-5.23384772e-02 7.55711973e-01 7.28238046e-01 -1.09883809e+00
-7.52651632e-01 -8.49526599e-02 -9.26293880e-02 3.17366540e-01
-5.93089461e-01 -2.68049419e-01 -5.62554240e-01 -5.87773085e-01
-7.16442466e-02 -1.12151301e+00 -3.91833514e-01 1.84585318e-01
-3.70747894e-01 -3.49128306e-01 7.57633805e-01 -5.73008358e-01
1.18340349e+00 -2.01594281e+00 -6.41323403e-02 2.61976391e-01
6.12352192e-01 4.18605238e-01 1.29883979e-02 -1.00988142e-01
-9.71810818e-02 4.50673729e-01 -2.53036171e-01 -1.14861429e-02
-3.28127682e-01 2.05842592e-02 4.78731185e-01 2.90224701e-01
2.76296079e-01 1.10586464e+00 -7.51470923e-01 -8.02742898e-01
6.08240888e-02 4.36672598e-01 -5.69628179e-01 1.86053127e-01
-1.15552351e-01 6.49637938e-01 -6.16899729e-01 5.59896588e-01
4.23770010e-01 -6.76145315e-01 2.26629764e-01 -5.38775586e-02
1.22522503e-01 3.36279310e-02 -6.23295665e-01 1.20278633e+00
-4.90881354e-01 2.95995355e-01 -2.35771745e-01 -9.27852750e-01
5.66828668e-01 6.20587528e-01 7.16773212e-01 -6.34186029e-01
1.61619663e-01 3.48209828e-01 6.60631776e-01 -5.28530777e-01
-5.97203858e-02 -2.37867698e-01 1.59724340e-01 2.08677217e-01
-2.40872055e-01 6.33438081e-02 8.77553523e-02 1.08609013e-01
1.47491634e+00 -4.45522159e-01 5.19694328e-01 -3.26551139e-01
7.27583647e-01 -1.33536100e-01 4.21583503e-01 8.50948274e-01
-3.90822411e-01 5.70753694e-01 5.20830274e-01 -3.11862946e-01
-7.42269516e-01 -8.71432245e-01 -2.79520780e-01 8.15482557e-01
-2.11376846e-01 3.09495833e-02 -6.02320373e-01 -1.09976804e+00
-1.77199349e-01 3.14611286e-01 -9.51668143e-01 -1.49179935e-01
-6.24187469e-01 -1.05344641e+00 3.75072300e-01 7.80325353e-01
7.87638009e-01 -1.04652476e+00 -5.64620018e-01 1.15839317e-01
-2.49532551e-01 -1.06922734e+00 -3.48690122e-01 2.92991310e-01
-9.92731810e-01 -1.20166016e+00 -9.97625053e-01 -9.19876635e-01
7.44112253e-01 2.61375070e-01 1.08020306e+00 3.47293884e-01
-5.51896751e-01 2.80902326e-01 -3.32501471e-01 -4.25235629e-01
-5.10378897e-01 5.06349146e-01 -5.68046272e-01 -5.53786755e-02
1.77538231e-01 -3.08189094e-02 -8.66344273e-01 3.01314265e-01
-9.43365514e-01 6.19830005e-02 1.26063812e+00 9.46696162e-01
7.25057483e-01 1.92428485e-01 5.99133730e-01 -1.28032541e+00
4.42872554e-01 -7.31786072e-01 -2.12079421e-01 7.80841112e-02
-2.62965471e-01 -1.41593426e-01 6.87109590e-01 -5.26483953e-01
-1.04631639e+00 2.78709948e-01 -1.55798554e-01 -1.43949524e-01
-6.99641332e-02 6.86156929e-01 1.86868623e-01 -1.23089515e-01
6.51260912e-01 1.19545065e-01 -2.11540297e-01 -1.38176322e-01
-2.95812309e-01 5.69935203e-01 2.53090024e-01 -1.47211462e-01
9.79887962e-01 3.93552810e-01 1.23664163e-01 -6.75522804e-01
-1.03978050e+00 -6.69713914e-01 -5.85129082e-01 -1.55044824e-01
1.10944378e+00 -8.88853490e-01 -4.42011297e-01 3.71511102e-01
-5.95225215e-01 -4.12373871e-01 -2.18654394e-01 4.56541896e-01
-2.23833308e-01 1.38557196e-01 -5.48869550e-01 -4.64318305e-01
-4.14203227e-01 -1.22924042e+00 7.89440274e-01 2.30538875e-01
-9.01129469e-02 -9.37103033e-01 2.51874817e-03 3.28581661e-01
8.38423908e-01 1.53067976e-01 1.12028897e+00 -9.84677553e-01
-6.50003493e-01 -4.06544626e-01 -5.58691561e-01 2.46417969e-01
2.95342952e-01 -1.99002802e-01 -9.61529434e-01 -3.99649054e-01
8.89775809e-03 -3.95848066e-01 1.13219309e+00 4.46480244e-01
1.77764142e+00 -3.77406180e-02 -6.52361035e-01 4.35666859e-01
1.39758754e+00 1.32523999e-01 5.04700005e-01 3.27812791e-01
6.87598646e-01 2.78352946e-01 4.20213044e-01 8.85315239e-02
1.12610050e-01 4.20285553e-01 7.26743758e-01 -4.87019867e-01
-2.80677587e-01 7.93931037e-02 -7.98840895e-02 5.70603251e-01
-1.26996711e-01 -3.96399081e-01 -1.15709472e+00 5.28497100e-01
-1.57572222e+00 -6.29243135e-01 -9.69898105e-02 2.07096624e+00
6.26354694e-01 3.08933944e-01 -5.26485704e-02 -7.52908364e-02
5.43983936e-01 -6.35744631e-02 -5.52771688e-01 4.58949693e-02
4.42340702e-01 2.93034434e-01 4.56499249e-01 -6.09076209e-02
-1.56380618e+00 5.73377550e-01 5.41169405e+00 8.82531703e-01
-1.29013467e+00 2.97258377e-01 1.04943156e+00 1.06452465e-01
3.42358164e-02 -5.34484625e-01 -5.86815476e-01 2.69843698e-01
8.09123218e-01 1.61026314e-01 -1.90133974e-01 7.97721922e-01
4.62291995e-03 -1.13741405e-01 -1.19777441e+00 5.55927634e-01
-8.05153847e-02 -1.08097804e+00 -3.76457572e-02 1.28493577e-01
8.65223944e-01 3.33704025e-01 2.98348725e-01 4.56351697e-01
1.14980817e-01 -1.18541050e+00 6.88362047e-02 3.86245877e-01
9.19011772e-01 -6.29324794e-01 1.30475068e+00 4.78912234e-01
-1.20941079e+00 -2.60663301e-01 -2.65312642e-01 5.37029386e-01
-4.79626209e-01 5.28947473e-01 -1.23501122e+00 4.50801015e-01
7.25874543e-01 4.64383274e-01 -9.75892723e-01 1.44082701e+00
-4.86587547e-02 8.23723197e-01 -1.98632911e-01 -2.27855816e-01
4.96532202e-01 3.21717560e-01 1.80843756e-01 1.12237298e+00
3.80709618e-01 1.44325569e-01 3.05904925e-01 7.00460434e-01
-3.98312718e-01 1.99157178e-01 -2.74694890e-01 -6.94282800e-02
1.54979438e-01 1.52286375e+00 -9.56863940e-01 -3.54237348e-01
-5.83674788e-01 7.18179345e-01 1.98260158e-01 -4.55704741e-02
-1.03836584e+00 1.12389438e-02 -6.76870421e-02 4.04192924e-01
3.22191983e-01 3.68721306e-01 -7.33391102e-03 -8.75524342e-01
-1.81373432e-01 -8.52183342e-01 5.45296729e-01 -4.00698930e-01
-1.25316858e+00 8.63793671e-01 -1.37770444e-01 -1.37488461e+00
7.07138935e-03 -5.95121682e-01 -7.38130927e-01 4.78242368e-01
-1.68681037e+00 -1.15045667e+00 -7.65349627e-01 4.44437683e-01
6.23771667e-01 -2.35945149e-03 7.00558364e-01 3.37092131e-01
-4.37380046e-01 7.20554471e-01 1.66909128e-01 2.72008777e-01
8.43856990e-01 -1.27019048e+00 -6.49269670e-02 5.37756503e-01
-3.26224089e-01 1.79567263e-01 7.55822361e-02 -6.18456423e-01
-8.53772461e-01 -1.50719571e+00 6.16994798e-01 -2.98146904e-01
6.57466710e-01 -1.84391946e-01 -9.63901699e-01 5.46832442e-01
6.58614514e-03 3.29647481e-01 6.01843119e-01 -2.51056641e-01
-5.83791062e-02 -7.77059644e-02 -1.06350386e+00 3.68259877e-01
6.97764277e-01 -1.56313572e-02 -8.69036987e-02 4.43529069e-01
5.36506951e-01 -3.40421975e-01 -7.13975906e-01 7.17148542e-01
3.80968064e-01 -8.72780442e-01 8.96620214e-01 -2.44582236e-01
5.93301296e-01 1.94684826e-02 1.78819731e-01 -1.07003605e+00
-6.83346391e-01 3.81853729e-02 2.11320013e-01 6.43393874e-01
6.06851876e-01 -5.41683733e-01 1.02362216e+00 2.39688367e-01
-1.35970697e-01 -1.16426837e+00 -7.20877886e-01 -5.35189211e-01
3.75773132e-01 -3.69542167e-02 1.77313760e-01 7.85977840e-01
-5.50274014e-01 9.78639647e-02 -5.50361052e-02 -1.48327397e-02
4.03471500e-01 -1.36592150e-01 2.84120977e-01 -1.45818138e+00
-3.94526988e-01 -4.86125648e-01 -1.97809935e-01 -7.49256015e-01
-3.62811118e-01 -1.07938933e+00 -3.90152708e-02 -1.59104717e+00
9.73107517e-01 -5.60273647e-01 -5.79510212e-01 5.10935783e-01
-4.66404527e-01 3.20304304e-01 1.30949199e-01 3.50008756e-01
-5.86433768e-01 2.70212382e-01 1.58258092e+00 -2.75448889e-01
-3.74553949e-02 5.07320404e-01 -8.01405013e-01 7.47916996e-01
9.47413981e-01 -6.42962277e-01 -3.29492241e-01 -1.89520359e-01
-4.92601516e-03 2.33677309e-02 4.46590394e-01 -1.19931245e+00
2.14792296e-01 -4.81861969e-03 6.33313894e-01 -6.53886378e-01
8.20273161e-02 -9.39125657e-01 -1.58762783e-01 1.06815994e+00
-4.23303276e-01 -3.85422528e-01 1.51605859e-01 7.18207717e-01
-1.67511791e-01 -2.81813920e-01 1.08927178e+00 -3.81821275e-01
-2.74543434e-01 5.83128095e-01 -6.04480743e-01 -1.46495312e-01
1.24891806e+00 9.95882880e-03 -2.70525575e-01 -2.46818781e-01
-7.64911056e-01 3.04136723e-01 1.61524173e-02 1.87970206e-01
4.29898053e-01 -1.08749866e+00 -9.34877872e-01 1.72038060e-02
1.66761562e-01 1.95163056e-01 2.33690441e-01 1.27977097e+00
-5.28650999e-01 7.00933158e-01 3.51968743e-02 -7.62023032e-01
-1.35060132e+00 3.58432144e-01 7.21884131e-01 -1.00108850e+00
-4.47825164e-01 9.63489532e-01 6.71012759e-01 -4.26816255e-01
4.58697826e-01 -4.95525509e-01 -3.30100358e-01 -1.48887172e-01
1.40052065e-01 1.23297222e-01 2.18902156e-01 -9.96225402e-02
-2.86494404e-01 4.02523786e-01 -5.60754776e-01 3.84461254e-01
1.14951921e+00 3.21949303e-01 1.41125387e-02 4.06902134e-02
1.10176015e+00 -2.86232352e-01 -8.70552003e-01 -1.71511933e-01
-7.33107328e-02 -3.62263285e-02 1.50679097e-01 -9.75315392e-01
-1.23979008e+00 9.46049154e-01 1.02849066e+00 1.04013644e-01
1.25918519e+00 2.76618838e-01 5.68661511e-01 4.29275632e-01
-1.65234774e-01 -6.46264553e-01 4.66878623e-01 3.82733136e-01
6.60279751e-01 -1.66965115e+00 2.40062531e-02 -6.08139694e-01
-6.36188805e-01 9.73873258e-01 9.10605133e-01 -9.28848609e-02
7.33170211e-01 5.22686504e-02 4.81637046e-02 -1.45026773e-01
-8.35604191e-01 -2.05038950e-01 4.88211006e-01 2.87205487e-01
6.50931418e-01 7.93952271e-02 -3.57609808e-01 6.63852096e-01
2.21222997e-01 5.61469933e-03 4.11480993e-01 8.36367965e-01
-6.26651347e-01 -9.98063982e-01 -2.68088847e-01 9.41766620e-01
-7.13274479e-01 -8.29161778e-02 -5.64487994e-01 1.23988211e+00
1.07301973e-01 4.65978622e-01 -4.77791391e-02 -1.36540994e-01
1.96505319e-02 2.75653396e-02 2.02836305e-01 -8.01035047e-01
-7.63541460e-01 3.95336986e-01 -1.42670617e-01 -2.61093289e-01
-4.32338953e-01 -5.52043080e-01 -1.10438204e+00 2.29950786e-01
-5.21837592e-01 5.23514226e-02 4.18481320e-01 7.93329120e-01
-4.25585918e-02 1.18967557e+00 6.03284717e-01 -4.65461671e-01
-7.48472154e-01 -9.98684525e-01 -2.60954201e-01 1.49363607e-01
2.18841523e-01 -4.96418417e-01 -3.82727504e-01 -2.62634337e-01] | [15.39923095703125, -2.18733811378479] |
1aed10ce-a049-41e6-8d97-b858e5592593 | fact-check-worthiness-detection-as-positive | 2003.02736 | null | https://arxiv.org/abs/2003.02736v2 | https://arxiv.org/pdf/2003.02736v2.pdf | Claim Check-Worthiness Detection as Positive Unlabelled Learning | As the first step of automatic fact checking, claim check-worthiness detection is a critical component of fact checking systems. There are multiple lines of research which study this problem: check-worthiness ranking from political speeches and debates, rumour detection on Twitter, and citation needed detection from Wikipedia. To date, there has been no structured comparison of these various tasks to understand their relatedness, and no investigation into whether or not a unified approach to all of them is achievable. In this work, we illuminate a central challenge in claim check-worthiness detection underlying all of these tasks, being that they hinge upon detecting both how factual a sentence is, as well as how likely a sentence is to be believed without verification. As such, annotators only mark those instances they judge to be clear-cut check-worthy. Our best performing method is a unified approach which automatically corrects for this using a variant of positive unlabelled learning that finds instances which were incorrectly labelled as not check-worthy. In applying this, we out-perform the state of the art in two of the three tasks studied for claim check-worthiness detection in English. | ['Dustin Wright', 'Isabelle Augenstein'] | 2020-03-05 | null | https://aclanthology.org/2020.findings-emnlp.43 | https://aclanthology.org/2020.findings-emnlp.43.pdf | findings-of-the-association-for-computational | ['rumour-detection'] | ['natural-language-processing'] | [ 4.35780793e-01 3.31596732e-01 -4.58434939e-01 -9.66986716e-02
-1.27826107e+00 -8.16038489e-01 1.16189599e+00 1.05514538e+00
-2.86957979e-01 9.17278051e-01 4.08111721e-01 -7.54105747e-01
-2.28138939e-01 -7.52578974e-01 -6.16748512e-01 -2.26390138e-01
3.89854044e-01 7.11009800e-01 6.09073400e-01 -4.36485291e-01
9.93404031e-01 1.31665379e-01 -1.36742330e+00 6.57319307e-01
8.20509374e-01 8.99499834e-01 -2.73495466e-01 6.43009305e-01
-1.04818009e-01 1.61514235e+00 -8.59029293e-01 -1.09194446e+00
-2.93579966e-01 -4.71940666e-01 -1.40025556e+00 3.85771915e-02
7.41480589e-01 1.23111725e-01 1.14514567e-01 1.34602296e+00
1.25218078e-01 -4.82441127e-01 6.30011022e-01 -9.67222333e-01
-8.21658075e-01 8.34413826e-01 -4.95864749e-01 7.80739725e-01
7.08450317e-01 -3.22029710e-01 1.25536501e+00 -6.87152147e-01
9.97152388e-01 9.32728827e-01 7.89728701e-01 1.84473410e-01
-9.73606348e-01 -2.79209942e-01 -2.72717953e-01 3.46169949e-01
-7.47137308e-01 -6.05557919e-01 7.45813489e-01 -8.62030089e-01
6.79407001e-01 5.87086916e-01 3.97672266e-01 8.78435612e-01
1.72572955e-01 7.55016148e-01 1.70240641e+00 -8.57412875e-01
1.24567926e-01 4.40583616e-01 6.86296582e-01 5.76359570e-01
4.36981857e-01 -4.12676752e-01 -6.05263352e-01 -7.35189974e-01
-9.15368088e-03 -3.52953136e-01 -2.76976079e-01 5.51453568e-02
-1.21202028e+00 9.60750401e-01 -1.41383298e-02 6.64018214e-01
-5.16790032e-01 -3.18105280e-01 8.95377517e-01 5.02130210e-01
8.82988691e-01 5.76997876e-01 -3.30491304e-01 -1.26919240e-01
-1.27513778e+00 6.86574519e-01 1.05534804e+00 4.74746257e-01
3.05581242e-01 -6.46001995e-01 -4.53712583e-01 5.63295484e-01
-4.73858565e-02 1.16986118e-01 3.42285365e-01 -7.81803787e-01
7.62420416e-01 9.66912687e-01 3.86457443e-01 -1.55109549e+00
-2.44151741e-01 -4.29419875e-01 -4.39208746e-01 2.31290922e-01
5.71104586e-01 2.70764500e-01 -3.56854737e-01 1.24148679e+00
2.10916564e-01 -3.77584428e-01 -2.11861432e-02 7.36130893e-01
8.09452593e-01 2.12683037e-01 -3.95904928e-02 -3.06283981e-01
1.79931056e+00 -6.60020411e-01 -8.86290431e-01 -6.99374154e-02
6.85197949e-01 -1.09104598e+00 8.59789670e-01 2.77950048e-01
-1.09113955e+00 -4.30562943e-02 -1.10736060e+00 -1.30314246e-01
-4.98445690e-01 1.04546592e-01 5.09733796e-01 7.24286079e-01
-4.83046949e-01 7.09231436e-01 -3.58418614e-01 -1.79596990e-01
4.31173623e-01 -4.83687371e-01 -2.61562109e-01 9.23294425e-02
-1.40502846e+00 1.34904432e+00 2.28774950e-01 -2.25405827e-01
-3.15033436e-01 -3.50979686e-01 -4.87701625e-01 -2.43892357e-01
8.06765974e-01 -3.01467180e-01 1.37696385e+00 -9.52284813e-01
-6.57980978e-01 1.85307920e+00 -2.24956825e-01 -7.22979307e-01
7.80974507e-01 8.83814134e-03 -8.00076604e-01 1.28914312e-01
7.27848589e-01 -3.47313285e-01 8.73813510e-01 -1.22183728e+00
-5.83629012e-01 -4.28431034e-01 2.58947492e-01 -1.60055593e-01
5.00394702e-02 9.11715388e-01 2.40039721e-01 -7.57435143e-01
2.37350434e-01 -7.58801281e-01 2.71274477e-01 -3.53358656e-01
-7.34651208e-01 -6.84925258e-01 6.62506104e-01 -9.38935757e-01
1.55660236e+00 -1.62854517e+00 -5.25146425e-01 -2.52391025e-02
6.69753551e-01 4.35116500e-01 4.03955132e-01 7.68065155e-01
-3.71088996e-03 5.32850087e-01 -4.05682653e-01 8.30793753e-02
1.88004747e-01 -6.32481799e-02 -6.72086596e-01 6.58869028e-01
1.52865082e-01 8.46126616e-01 -1.17487574e+00 -6.27843857e-01
-3.54525059e-01 -2.52361428e-02 4.04349109e-03 -2.86983699e-01
-2.52474695e-01 1.19962640e-01 -4.69151407e-01 6.72564149e-01
5.74172378e-01 -5.20219684e-01 1.40986249e-01 4.15250659e-02
-3.75485688e-01 1.08206820e+00 -9.13173020e-01 7.27820635e-01
5.04858186e-03 8.88990879e-01 -2.07174290e-02 -1.03393149e+00
8.10510278e-01 3.49624753e-01 7.36043304e-02 -7.60954380e-01
5.70358075e-02 6.30165219e-01 -1.61841244e-01 -8.55927587e-01
7.99245000e-01 -4.06126618e-01 -1.12033561e-01 9.76319075e-01
-4.55467343e-01 1.17177844e-01 5.04756272e-01 6.03917599e-01
1.31914473e+00 -3.80751431e-01 5.66433847e-01 -1.68965876e-01
7.93082714e-01 4.98621702e-01 3.03782523e-01 1.00779068e+00
-3.42636764e-01 5.78960896e-01 9.42447901e-01 -4.96391922e-01
-1.03206241e+00 -4.19230580e-01 -1.66109771e-01 9.30426359e-01
-1.30603388e-01 -6.83723748e-01 -6.72873437e-01 -9.48462725e-01
4.50809598e-02 7.17593133e-01 -8.32314551e-01 3.85382295e-01
-4.45294946e-01 -6.39661252e-01 5.62258780e-01 -1.07422039e-01
3.82086426e-01 -1.13053143e+00 -9.53181863e-01 1.91978738e-01
-6.22388601e-01 -9.97305989e-01 2.62833126e-02 7.61664882e-02
-3.23774546e-01 -1.89106512e+00 -4.38072622e-01 -4.63861793e-01
2.03237683e-01 2.95032144e-01 1.51328862e+00 6.68964803e-01
1.38538554e-01 -5.88093475e-02 -5.82875550e-01 -5.88539243e-01
-1.11821640e+00 -6.23845942e-02 -2.99693018e-01 4.87818234e-02
4.80368018e-01 -1.42029123e-02 -5.49352989e-02 -1.21886497e-02
-9.24630582e-01 -3.69293660e-01 3.19276273e-01 6.85383379e-01
2.50963718e-01 1.38296276e-01 4.75339293e-01 -1.49817705e+00
1.28908873e+00 -5.70759833e-01 -9.32112038e-02 4.78034258e-01
-8.01055133e-01 -1.95506603e-01 2.26384670e-01 -4.32249866e-02
-6.40772998e-01 -6.17973089e-01 -2.96399683e-01 2.60303557e-01
-1.37638256e-01 7.82988667e-01 4.75842714e-01 1.25165135e-01
9.53786850e-01 -1.35295093e-02 -4.58291061e-02 -3.61106575e-01
1.89043686e-01 9.16968226e-01 6.28893912e-01 -2.84968287e-01
8.29263985e-01 5.19979000e-01 -3.01775098e-01 -4.84330148e-01
-1.73410094e+00 -8.32806289e-01 -5.87633431e-01 -3.14386755e-01
5.37581623e-01 -4.75113988e-01 -7.04343319e-01 3.65170658e-01
-1.39483953e+00 2.79083878e-01 -3.42445038e-02 6.90122396e-02
-3.54480475e-01 7.44022727e-01 -7.91334152e-01 -1.04588950e+00
-3.42790872e-01 -6.84560180e-01 8.01086962e-01 -2.58695304e-01
-7.43661821e-01 -8.04058433e-01 2.77757257e-01 9.17887151e-01
3.51748854e-01 4.20407981e-01 1.03107429e+00 -9.54046428e-01
7.22383410e-02 -7.20098615e-01 -3.20512444e-01 1.46098584e-01
-2.94465553e-02 -8.66768956e-02 -8.56780410e-01 1.65543392e-01
3.84215117e-01 -5.85263252e-01 1.00453687e+00 7.43118078e-02
5.32929122e-01 -8.94128978e-01 -3.01250488e-01 -4.62852210e-01
1.21868551e+00 -3.80849689e-01 6.71768904e-01 1.03622699e+00
1.62012503e-01 7.76484191e-01 7.32212305e-01 2.02132121e-01
4.88471359e-01 8.02737653e-01 5.17192125e-01 7.53487870e-02
-4.93863560e-02 -2.20710635e-01 1.37446746e-01 5.47899604e-01
-3.08242530e-01 -1.55496031e-01 -9.33326364e-01 6.30669653e-01
-1.74489605e+00 -1.81608689e+00 -1.07465744e+00 1.97927070e+00
1.00490963e+00 5.59967935e-01 3.07845086e-01 8.92992795e-01
9.82794642e-01 4.05014098e-01 2.17262670e-01 -4.08422172e-01
-3.05133075e-01 -2.78287768e-01 2.14027509e-01 5.25251508e-01
-1.36615360e+00 6.02876961e-01 6.01809978e+00 6.70749545e-01
-6.89325035e-01 5.95890284e-01 5.85805595e-01 3.95730287e-01
-3.24496686e-01 2.43258864e-01 -5.24019957e-01 5.89192808e-01
6.43316805e-01 5.86932786e-02 -2.06634849e-01 7.22885847e-01
2.09658965e-01 -4.13652867e-01 -6.79420769e-01 6.06182516e-01
6.33456767e-01 -1.69996750e+00 -2.32594863e-01 -4.91794199e-02
6.44857883e-01 -1.38512924e-01 -3.47417474e-01 1.80648267e-01
5.53546567e-03 -5.68847775e-01 1.39112818e+00 4.07387793e-01
3.19729716e-01 -2.93358147e-01 1.03168488e+00 6.33102596e-01
-3.39433461e-01 2.90145539e-02 -2.06333935e-01 -3.96784306e-01
4.68637764e-01 1.13175786e+00 -7.39200294e-01 3.43249887e-01
4.10669446e-01 4.69316036e-01 -7.62944579e-01 9.85789597e-01
-6.37811601e-01 8.25525761e-01 2.14817822e-01 -1.18103206e-01
1.96695432e-01 3.07003617e-01 8.22096825e-01 1.35134983e+00
-5.01441211e-02 3.45224701e-02 1.78544819e-01 6.67955935e-01
-2.67193973e-01 9.96025056e-02 -5.58379710e-01 -7.50856772e-02
2.91651130e-01 1.11893702e+00 -8.63214076e-01 -5.96309423e-01
-1.74714506e-01 8.08694065e-01 4.95415688e-01 -3.96197468e-01
-6.13506258e-01 -5.97439185e-02 4.11003008e-02 5.13927162e-01
1.05819345e-01 1.29931822e-01 -4.88870919e-01 -1.14147151e+00
5.24065912e-01 -9.39181387e-01 5.66773057e-01 -8.09680820e-01
-1.60850823e+00 6.67904973e-01 -3.72672290e-01 -1.19096577e+00
-5.71638197e-02 -3.90038282e-01 -5.98524213e-01 5.87464154e-01
-1.83921874e+00 -8.82334054e-01 -1.58084128e-02 2.17074290e-01
2.76108831e-01 5.22163510e-02 6.98917925e-01 3.02645326e-01
-2.40224436e-01 8.96393806e-02 -3.39140475e-01 2.63125867e-01
7.11209476e-01 -1.19204175e+00 1.84273034e-01 1.00169182e+00
3.97618800e-01 5.52456975e-01 1.01154864e+00 -9.13340032e-01
-6.90655529e-01 -6.90806746e-01 2.11855435e+00 -1.07941723e+00
1.23495972e+00 7.82499388e-02 -1.12521851e+00 3.06486338e-01
2.33810559e-01 -2.39839375e-01 5.04129887e-01 4.62097287e-01
-8.20471346e-01 4.85024035e-01 -8.56541097e-01 7.26160333e-02
6.44481063e-01 -8.68821800e-01 -1.24239504e+00 8.53225827e-01
6.46247342e-02 -3.47680509e-01 -4.90608901e-01 1.02370545e-01
2.38017008e-01 -1.14584339e+00 5.79882860e-01 -7.94392586e-01
9.79355752e-01 -3.52680176e-01 1.00496046e-01 -8.29905689e-01
-3.54541034e-01 -5.02839088e-01 -1.44899815e-01 1.48407114e+00
7.56228447e-01 -5.29025972e-01 4.66738135e-01 2.44485736e-01
-1.29146099e-01 -5.99413693e-01 -8.94840896e-01 -6.51627004e-01
-6.59787729e-02 -4.04726774e-01 3.01804692e-01 1.61846256e+00
4.11231905e-01 5.77558160e-01 -3.34842443e-01 -1.29296094e-01
4.89578575e-01 5.23363471e-01 4.98528183e-01 -1.60482144e+00
-1.39188357e-02 -6.68226302e-01 -3.75435024e-01 -4.62239385e-01
7.90020749e-02 -1.01230121e+00 -6.28109202e-02 -1.77081966e+00
7.35303760e-01 -5.20535171e-01 2.26134062e-02 3.81774455e-01
-4.12491173e-01 3.78398508e-01 6.37041330e-02 7.74350166e-01
-8.13097537e-01 -1.46316156e-01 8.95820200e-01 -3.24919224e-01
2.52184451e-01 2.27565572e-01 -8.85882318e-01 9.09617782e-01
7.62109637e-01 -9.03660357e-01 2.28335768e-01 -1.87291741e-01
1.00450146e+00 -5.86034022e-02 7.68334150e-01 -7.54823506e-01
3.93204004e-01 -7.57574290e-02 -2.29323789e-01 -6.68136418e-01
-1.83016673e-01 -2.49922231e-01 -2.33397141e-01 3.70219409e-01
-5.81668079e-01 3.39663267e-01 -4.55292940e-01 7.18896985e-01
-3.06820452e-01 -7.48711884e-01 6.70008421e-01 -4.39995497e-01
-4.25284266e-01 -3.81845623e-01 -5.40845513e-01 5.42256296e-01
6.81782424e-01 -4.31002229e-02 -1.00244927e+00 -3.42037797e-01
-4.69090134e-01 -3.44090283e-01 3.35784376e-01 1.72871128e-01
2.38313079e-01 -1.01212204e+00 -1.30307305e+00 -5.48451960e-01
3.38885814e-01 -7.95508206e-01 -4.47199605e-02 1.36636579e+00
-4.02438670e-01 4.01777625e-01 1.48209244e-01 -1.20372012e-01
-1.44789326e+00 8.65572274e-01 4.00132015e-02 -6.19067013e-01
-5.87794363e-01 5.12112856e-01 -9.18665648e-01 -1.94880337e-01
-6.74555898e-02 1.36377543e-01 -6.17443562e-01 4.67921257e-01
8.21000755e-01 4.38674241e-01 5.27206600e-01 -9.75508034e-01
-3.75568360e-01 1.36758015e-01 -7.10169151e-02 -2.22175606e-02
1.23857617e+00 -1.25197992e-01 -6.49166107e-01 6.34832561e-01
6.82982504e-01 5.78807652e-01 -3.36562425e-01 -1.96317613e-01
6.34442806e-01 -5.63963532e-01 -6.85750172e-02 -1.08168149e+00
-4.75980163e-01 5.46782196e-01 -2.69357294e-01 1.37438965e+00
4.63440478e-01 4.64694768e-01 5.57144105e-01 2.91852117e-01
2.07847461e-01 -1.32714701e+00 -1.34749860e-01 6.75155461e-01
1.17553759e+00 -1.51431108e+00 2.96457320e-01 -5.98536909e-01
-6.44521117e-01 9.93685067e-01 -1.48027956e-01 -7.21502006e-02
3.86955291e-01 5.25979958e-02 7.83487931e-02 -1.06143773e+00
-6.46154702e-01 -2.38894239e-01 3.09999108e-01 2.03247905e-01
7.50780165e-01 1.50372937e-01 -1.11434221e+00 3.68537873e-01
-3.37684780e-01 -1.67528406e-01 7.92073190e-01 9.59176123e-01
-5.99052250e-01 -8.03583503e-01 -6.14552617e-01 7.58103907e-01
-1.11071086e+00 -1.34318948e-01 -9.76666629e-01 6.19096756e-01
7.57098719e-02 1.34258354e+00 -4.48144674e-01 -1.47369042e-01
2.76043415e-01 1.74628153e-01 3.23493838e-01 -6.87656522e-01
-1.03038096e+00 -5.38192332e-01 9.18389559e-01 -1.08438432e-01
-1.02469170e+00 -8.72074008e-01 -8.04379463e-01 -4.61726367e-01
-6.36433721e-01 4.36034501e-01 5.57425022e-01 1.44332838e+00
-2.40697060e-02 9.35521424e-02 2.76642948e-01 -2.25684822e-01
-6.81580007e-01 -9.61296976e-01 -4.12613302e-01 8.00180078e-01
4.49466020e-01 -6.28527582e-01 -4.46259767e-01 2.90878136e-02] | [8.61080265045166, 9.874982833862305] |
65c51ae3-5dc4-4d51-b8df-738a64e83bcf | unleashing-realistic-air-quality-forecasting | 2306.13948 | null | https://arxiv.org/abs/2306.13948v1 | https://arxiv.org/pdf/2306.13948v1.pdf | Unleashing Realistic Air Quality Forecasting: Introducing the Ready-to-Use PurpleAirSF Dataset | Air quality forecasting has garnered significant attention recently, with data-driven models taking center stage due to advancements in machine learning and deep learning models. However, researchers face challenges with complex data acquisition and the lack of open-sourced datasets, hindering efficient model validation. This paper introduces PurpleAirSF, a comprehensive and easily accessible dataset collected from the PurpleAir network. With its high temporal resolution, various air quality measures, and diverse geographical coverage, this dataset serves as a useful tool for researchers aiming to develop novel forecasting models, study air pollution patterns, and investigate their impacts on health and the environment. We present a detailed account of the data collection and processing methods employed to build PurpleAirSF. Furthermore, we conduct preliminary experiments using both classic and modern spatio-temporal forecasting models, thereby establishing a benchmark for future air quality forecasting tasks. | ['Hakim Hacid', 'Michele Baldo', 'Wenbin Li', 'Jingwei Zuo'] | 2023-06-24 | null | null | null | null | ['spatio-temporal-forecasting'] | ['time-series'] | [-1.07140034e-01 -8.57413173e-01 -1.78493395e-01 -4.66726214e-01
-7.37579048e-01 -4.02624756e-01 6.35267556e-01 5.37904799e-01
-2.84418106e-01 6.93778038e-01 3.53731334e-01 -3.84086907e-01
-4.80254292e-01 -1.15362942e+00 -4.05760825e-01 -6.95232153e-01
-5.23469597e-02 1.45412043e-01 -1.95548341e-01 1.07362799e-01
-2.75298595e-01 5.52399039e-01 -1.50905383e+00 2.23911285e-01
9.55888629e-01 9.62571621e-01 8.29135180e-02 4.04955357e-01
-7.16777593e-02 2.74437845e-01 -4.28806663e-01 -1.10500515e-01
2.86643326e-01 -1.18916027e-01 -5.61015189e-01 -4.41174835e-01
3.34961802e-01 1.39041275e-01 -3.47621948e-01 5.70513368e-01
6.03224218e-01 5.29285312e-01 1.84679806e-01 -1.11654282e+00
-3.90407115e-01 6.16722666e-02 -7.79600348e-03 5.21217704e-01
1.27096310e-01 4.70820189e-01 8.81602108e-01 -3.49750638e-01
8.99020135e-02 9.02094662e-01 8.44885886e-01 2.97420055e-01
-9.96743321e-01 -5.74591696e-01 2.83907115e-01 2.18769744e-01
-1.22010624e+00 -2.03065515e-01 2.52879053e-01 -6.75567865e-01
9.81452763e-01 7.09394217e-01 8.55426967e-01 1.04187405e+00
1.80372432e-01 3.10171247e-01 1.22955406e+00 2.69597173e-01
2.08703935e-01 3.64669748e-02 2.52624124e-01 2.49192238e-01
1.14907578e-01 5.99508345e-01 -1.86821088e-01 -2.42417485e-01
4.51241732e-02 7.04680860e-01 -1.92656532e-01 6.05861247e-01
-1.04780054e+00 5.37477553e-01 1.03462660e+00 3.61701846e-01
-7.22067058e-01 -4.43964414e-02 1.33208120e-02 3.94556709e-02
8.92616987e-01 4.65418071e-01 -6.39231682e-01 -9.12689716e-02
-9.76336539e-01 4.71396476e-01 2.68056095e-01 4.63502973e-01
5.46520174e-01 -8.33929405e-02 -2.82976359e-01 7.95378327e-01
3.85015130e-01 1.11515880e+00 -1.69304537e-03 -6.37818336e-01
4.46072429e-01 6.26084089e-01 1.48535460e-01 -1.25580966e+00
-7.68539667e-01 -5.39560258e-01 -1.04786706e+00 -2.81447828e-01
3.62496734e-01 -2.14059383e-01 -4.12099481e-01 1.31766868e+00
3.86922389e-01 2.41755798e-01 -4.27810490e-01 8.75969172e-01
7.62000978e-01 1.20457244e+00 4.56949502e-01 2.06434920e-01
1.43151677e+00 -7.47449458e-01 -4.71797198e-01 1.11450173e-01
3.95718336e-01 -2.03229487e-01 9.61639345e-01 2.76801765e-01
-4.89421129e-01 -6.28180504e-01 -3.78462762e-01 2.68433154e-01
-1.12256587e+00 -1.72688931e-01 4.96390671e-01 6.91574335e-01
-5.06865501e-01 5.49025774e-01 -5.80933928e-01 -2.88210452e-01
7.52453566e-01 -1.45618096e-02 -2.09232286e-01 -3.42134863e-01
-1.51554334e+00 7.10749805e-01 -1.59453824e-02 3.93387377e-01
-8.05488408e-01 -1.29335678e+00 -6.21207297e-01 3.89016643e-02
-3.34139653e-02 -8.42561424e-01 1.01889527e+00 -7.81952068e-02
-7.36330450e-01 2.13229328e-01 -2.60030359e-01 -4.59389299e-01
1.62127435e-01 -2.58195072e-01 -1.19690859e+00 -2.55332351e-01
1.28370151e-01 4.00076807e-01 4.07342732e-01 -7.12747753e-01
-8.10943604e-01 -5.76365888e-01 -1.67573020e-01 -2.69168317e-01
-3.76071632e-01 2.06924811e-01 -1.39088593e-02 -4.29158777e-01
-7.04285383e-01 -8.77180576e-01 -6.76429331e-01 -1.24362178e-01
-2.08784327e-01 -3.78939480e-01 4.01409149e-01 -8.04089487e-01
1.52002943e+00 -2.15813708e+00 -4.14261669e-01 1.94050193e-01
1.77352190e-01 6.23579323e-01 -2.15836748e-01 5.09925246e-01
-8.30232650e-02 2.27409869e-01 -3.35642517e-01 -2.60414332e-01
-2.67229415e-02 2.20192418e-01 -3.01095128e-01 5.44458270e-01
2.85443485e-01 8.56818080e-01 -1.19485319e+00 1.08858697e-01
4.78054404e-01 7.41307676e-01 -2.36361280e-01 2.97431380e-01
-4.38471496e-01 8.16630781e-01 -5.64424336e-01 4.11915928e-01
5.78639269e-01 -1.47535086e-01 -3.77211720e-01 6.69669211e-02
-8.08100760e-01 6.25995398e-01 -7.91502059e-01 1.13329244e+00
-8.94728899e-01 5.93555093e-01 -7.69004077e-02 -5.10084987e-01
5.71458459e-01 3.45458180e-01 5.76851726e-01 -1.25484097e+00
-8.73092860e-02 -4.01679017e-02 -2.26911128e-01 -7.71031380e-01
1.77311078e-01 -1.34513497e-01 2.20888361e-01 3.88644814e-01
-8.28859389e-01 1.38858646e-01 1.07230633e-01 -3.06565583e-01
6.93130136e-01 -2.75010437e-01 -3.79545182e-01 -2.03302562e-01
5.18860996e-01 2.22010016e-01 4.09138352e-01 2.22771600e-01
-1.59379944e-01 4.77670640e-01 -4.59238812e-02 -8.68555546e-01
-6.36640310e-01 -7.17963815e-01 -6.01983249e-01 8.85508537e-01
-3.99454325e-01 -4.99365419e-01 -3.46245438e-01 -4.75801766e-01
2.15447053e-01 6.36853397e-01 -5.98685086e-01 2.15358809e-02
-4.64916617e-01 -1.31101453e+00 5.38760722e-01 3.87486845e-01
6.10439599e-01 -8.34475696e-01 -4.26709145e-01 1.14893571e-01
-2.72929341e-01 -6.96359336e-01 -1.46515265e-01 -1.15571029e-01
-6.20626926e-01 -1.25086880e+00 -5.87427735e-01 -1.67166665e-01
4.49484885e-01 6.43112957e-01 1.25353467e+00 -1.54951155e-01
-3.96580368e-01 1.74283788e-01 -2.18948931e-03 -8.03799272e-01
-1.21125653e-01 3.08232516e-01 6.15572790e-03 2.99235173e-02
3.26781690e-01 -2.49178946e-01 -8.42589617e-01 4.69777137e-01
-1.02917337e+00 -2.74802357e-01 2.96901494e-01 1.10811323e-01
6.12998545e-01 5.18357396e-01 8.82086754e-01 -6.07386053e-01
5.40280342e-01 -9.87426043e-01 -8.09346199e-01 9.25028175e-02
-7.93598711e-01 -3.25439304e-01 7.46944189e-01 3.45452785e-01
-9.89372969e-01 -2.93165594e-01 -5.74249029e-01 1.89604118e-01
-7.04478979e-01 5.28391480e-01 -5.25101051e-02 4.74944174e-01
5.37276745e-01 -1.31269805e-02 -2.31161401e-01 -1.01738203e+00
2.75495321e-01 8.22768271e-01 3.60540859e-02 -2.15955585e-01
8.46739292e-01 4.22275752e-01 -7.44211376e-02 -1.10150158e+00
-1.32274711e+00 -8.65175664e-01 -5.74161589e-01 -2.37090006e-01
1.11534178e+00 -1.15370357e+00 -5.66301405e-01 5.73648453e-01
-9.11789358e-01 -3.22489411e-01 -1.46924987e-01 4.24977243e-01
4.64170188e-01 -2.57539451e-01 -1.18113920e-01 -8.10336828e-01
-4.38913226e-01 -8.69482279e-01 9.69860315e-01 7.75791556e-02
-2.52841741e-01 -1.36789656e+00 6.00654662e-01 6.85713053e-01
9.47582424e-01 4.70825106e-01 9.00203466e-01 -2.56127775e-01
-5.83222926e-01 -2.54026324e-01 -3.85199785e-01 2.77189761e-01
3.97210360e-01 2.50477856e-03 -1.16418135e+00 -5.01656383e-02
-2.59751469e-01 3.77550796e-02 9.09494638e-01 2.51879394e-01
1.55955791e+00 -3.21900994e-01 -3.21369976e-01 5.03478050e-01
1.05919218e+00 6.39794469e-02 3.00214499e-01 3.16281140e-01
6.00099146e-01 7.19686508e-01 3.40104520e-01 2.27055535e-01
5.58546126e-01 5.99637687e-01 5.85935652e-01 -3.97721589e-01
-4.72137965e-02 -1.14086038e-02 8.41963664e-03 6.99649453e-01
-8.80988464e-02 -3.04531306e-01 -1.07993102e+00 6.33731902e-01
-1.51559412e+00 -1.14695489e+00 -6.77472830e-01 2.09052515e+00
6.04791701e-01 -3.84741902e-01 4.94267851e-01 -4.27929237e-02
3.17400783e-01 4.78015363e-01 -3.08684736e-01 -1.30989224e-01
-1.14721097e-02 1.89778239e-01 3.19296628e-01 1.67297408e-01
-1.33164692e+00 2.56618530e-01 6.82307053e+00 2.78616279e-01
-1.29666221e+00 1.76385015e-01 6.57223463e-01 -3.45238984e-01
-3.47308666e-01 -6.09870970e-01 -8.51106703e-01 7.86345124e-01
1.60685396e+00 7.71212652e-02 5.13593435e-01 5.20124137e-01
9.30652261e-01 1.40958473e-01 -7.15225816e-01 6.71408832e-01
-5.64430356e-01 -1.44289434e+00 -2.33308926e-01 1.55654892e-01
8.59772444e-01 8.27720284e-01 2.26495475e-01 1.03603072e-01
-1.53577719e-02 -1.21176291e+00 1.96756318e-01 8.33228886e-01
5.08504927e-01 -6.39601290e-01 8.47424507e-01 3.85845840e-01
-1.20105433e+00 -1.90293044e-01 -3.94149333e-01 -2.77117848e-01
1.47069886e-01 1.15608358e+00 -3.21162254e-01 8.22715938e-01
9.15720761e-01 9.98079896e-01 -6.06741011e-01 1.25301516e+00
-1.62592288e-02 9.69457924e-01 -4.64098454e-01 -2.92578358e-02
3.66503924e-01 -3.20163757e-01 2.63766080e-01 1.28681159e+00
4.19468284e-01 1.69064235e-02 1.53036743e-01 8.92546475e-01
-9.62520838e-02 5.82010746e-02 -5.91284633e-01 -2.04794794e-01
1.74722478e-01 1.29778087e+00 -3.47550660e-02 -1.33301824e-01
-7.09348202e-01 6.82502910e-02 -1.37429908e-01 2.71422774e-01
-1.02975619e+00 -1.26976907e-01 1.33447623e+00 5.01914203e-01
-2.33362332e-01 -3.72450262e-01 -1.99721202e-01 -9.09004629e-01
-2.10829273e-01 -7.92598665e-01 3.89900476e-01 -3.29110980e-01
-1.39481390e+00 5.12971222e-01 1.28038928e-01 -1.02804279e+00
2.87812024e-01 -3.80717695e-01 -9.79443371e-01 1.11034513e+00
-2.00126338e+00 -6.97293162e-01 -5.40182412e-01 4.71781850e-01
2.14385659e-01 7.55640417e-02 8.80590200e-01 8.16575348e-01
-1.20125568e+00 -1.39114693e-01 4.08384889e-01 -1.17377326e-01
4.13968325e-01 -9.16204929e-01 7.93025017e-01 6.75274968e-01
4.40938435e-02 7.04665244e-01 3.01431388e-01 -4.30457443e-01
-1.25486457e+00 -2.08649755e+00 1.12445199e+00 -6.05280161e-01
6.65890813e-01 -2.29478791e-01 -9.26875114e-01 2.17218891e-01
1.41232476e-01 -2.30759904e-02 1.15007961e+00 2.85465717e-01
-1.84720144e-01 -6.83127880e-01 -1.02377212e+00 2.16217220e-01
7.54457533e-01 -4.87025946e-01 -9.93793383e-02 7.28976309e-01
5.22375345e-01 1.77288339e-01 -1.05685377e+00 3.45811099e-01
4.73729253e-01 -7.14787483e-01 9.49255109e-01 -7.23098338e-01
2.67568409e-01 -4.16209728e-01 5.25772758e-03 -1.57211876e+00
-6.65447891e-01 1.45817604e-02 2.23305464e-01 1.11199129e+00
6.46109819e-01 -7.52440155e-01 2.85829633e-01 9.66744244e-01
-1.80223152e-01 -6.93932354e-01 -8.30024898e-01 -7.25136757e-01
1.45951942e-01 -9.06211853e-01 1.18698692e+00 8.55092168e-01
-5.33105075e-01 5.85642979e-02 -3.74250889e-01 4.06890690e-01
3.54005933e-01 4.29770589e-01 5.28971732e-01 -1.47062600e+00
1.64534107e-01 -4.37646389e-01 2.70119369e-01 -6.88516021e-01
-5.38068265e-02 -9.60432291e-01 -1.26676545e-01 -1.81617785e+00
-1.61249563e-01 -7.90559649e-01 -5.68335712e-01 6.34699643e-01
-2.20964745e-01 3.23277086e-01 -4.89956401e-02 -4.23465371e-02
-4.41364199e-01 6.79071248e-01 1.02416003e+00 -5.22052646e-01
-1.66360840e-01 4.17200714e-01 -6.23771131e-01 3.70814800e-01
1.09545374e+00 -5.07048607e-01 -2.50694901e-01 -8.45377982e-01
4.91387933e-01 -4.71196920e-01 5.84359586e-01 -1.20675766e+00
-1.17169268e-01 -4.81459379e-01 5.86260080e-01 -7.83008635e-01
7.79189691e-02 -1.10040021e+00 6.72389448e-01 5.38243234e-01
-3.15373182e-01 3.14433396e-01 3.54967713e-01 5.14322519e-01
-1.55648768e-01 3.19367409e-01 6.30104661e-01 -1.22336097e-01
-4.01525319e-01 8.95315647e-01 -5.93102574e-01 -1.28151268e-01
8.69874835e-01 5.07599890e-01 -4.85039920e-01 9.95825976e-02
-6.49957895e-01 6.49833143e-01 1.45799279e-01 6.66460872e-01
2.99619377e-01 -1.38636792e+00 -7.92374909e-01 5.08481205e-01
1.64453268e-01 1.58311397e-01 5.49886286e-01 7.28388488e-01
-3.40079725e-01 9.09877837e-01 1.73448548e-01 -5.23181319e-01
-9.81377542e-01 5.72819889e-01 4.20499504e-01 -2.80808330e-01
-5.70149004e-01 3.43042254e-01 -2.65356656e-02 -6.74256682e-01
-1.08109768e-02 -5.80318332e-01 -1.65501744e-01 2.29046822e-01
9.02404547e-01 1.02480102e+00 3.88830215e-01 -5.80565453e-01
-4.74739432e-01 2.75612235e-01 6.44444883e-01 5.77351689e-01
1.66305506e+00 -6.24145232e-02 -1.99218690e-01 5.38804710e-01
1.17396128e+00 2.76377867e-03 -7.29614258e-01 -1.16663210e-01
-4.00116034e-02 -5.27085125e-01 2.61025637e-01 -9.35511827e-01
-1.20474041e+00 1.17411304e+00 6.40588284e-01 6.05386913e-01
9.01672006e-01 -1.68148324e-01 1.37768745e+00 2.95293152e-01
1.84884039e-03 -6.89179420e-01 -4.16303337e-01 5.18052995e-01
8.38736951e-01 -1.30164421e+00 -6.80882484e-02 4.35565747e-02
1.64500810e-02 6.10740483e-01 2.98093110e-01 3.75454217e-01
1.10235906e+00 -1.57457590e-01 1.99722692e-01 -3.93304706e-01
-8.09965253e-01 -2.56826371e-01 5.78327358e-01 6.59657776e-01
5.00302255e-01 3.62470239e-01 3.09346437e-01 5.30995190e-01
-9.70317274e-02 -2.18985640e-02 -4.00718212e-01 2.75440872e-01
-4.71656621e-01 -1.07700241e+00 -4.51527834e-01 6.84092581e-01
-5.02476156e-01 -1.59906089e-01 -1.53705284e-01 4.11134362e-01
4.00369763e-01 1.33838499e+00 3.86190750e-02 -1.57678962e-01
5.83517849e-01 1.22454688e-01 -2.68633336e-01 -4.63746488e-01
-8.08861315e-01 -4.48894292e-01 1.42714009e-01 -5.35772383e-01
-5.09820342e-01 -7.55946875e-01 -8.38639379e-01 -5.77993274e-01
7.07208738e-02 3.89529288e-01 8.82718265e-01 9.33640957e-01
6.14792109e-01 7.38169432e-01 9.03173387e-01 -5.84037840e-01
-6.39197081e-02 -7.86167085e-01 -3.07706803e-01 2.71766782e-01
5.83988726e-01 -2.77347237e-01 -2.34425724e-01 -3.11664730e-01] | [6.260385990142822, 2.542649745941162] |
b5ab33e5-06ec-4759-94c1-a530bbb84001 | a-large-scale-multimodal-dataset-of-human | 2303.08295 | null | https://arxiv.org/abs/2303.08295v1 | https://arxiv.org/pdf/2303.08295v1.pdf | A large-scale multimodal dataset of human speech recognition | Nowadays, non-privacy small-scale motion detection has attracted an increasing amount of research in remote sensing in speech recognition. These new modalities are employed to enhance and restore speech information from speakers of multiple types of data. In this paper, we propose a dataset contains 7.5 GHz Channel Impulse Response (CIR) data from ultra-wideband (UWB) radars, 77-GHz frequency modulated continuous wave (FMCW) data from millimetre wave (mmWave) radar, and laser data. Meanwhile, a depth camera is adopted to record the landmarks of the subject's lip and voice. Approximately 400 minutes of annotated speech profiles are provided, which are collected from 20 participants speaking 5 vowels, 15 words and 16 sentences. The dataset has been validated and has potential for the research of lip reading and multimodal speech recognition. | ['Muhammad Imran', 'Qammer H. Abbasi', 'Daniele Faccio', 'Kevin Chetty', 'Wenda Li', 'Zikang Zhang', 'Haobo Li', 'Chong Tang', 'Yao Ge'] | 2023-03-15 | null | null | null | null | ['motion-detection'] | ['computer-vision'] | [ 3.34020793e-01 -2.97307402e-01 -2.29832958e-02 -4.01606500e-01
-1.22887504e+00 -2.74393052e-01 4.67316777e-01 -7.55237281e-01
-5.26622951e-01 5.46104491e-01 6.10608101e-01 -3.72917235e-01
-2.58098751e-01 -5.34098744e-01 8.59270245e-02 -1.19395578e+00
-3.90015962e-03 -2.67327458e-01 -2.55200595e-01 9.26032364e-02
2.66725063e-01 7.19710648e-01 -1.87548792e+00 1.26249924e-01
7.54769862e-01 9.65430319e-01 7.69411027e-01 7.51456320e-01
1.65237740e-01 2.47753799e-01 -6.74890757e-01 6.94297776e-02
-1.58928279e-02 3.72767411e-02 -2.54015714e-01 -2.86594555e-02
5.40894449e-01 -6.49625421e-01 -6.38122857e-01 1.02900207e+00
1.16147220e+00 1.57832772e-01 6.82027876e-01 -9.15891528e-01
-5.15293896e-01 2.48315006e-01 -6.74967527e-01 2.28007272e-01
6.47734582e-01 9.10779163e-02 2.65031308e-01 -1.06880057e+00
3.63168150e-01 1.09815538e+00 5.02934337e-01 7.95976818e-01
-1.01288982e-01 -7.33865976e-01 -6.67603493e-01 6.42881274e-01
-1.63486648e+00 -1.04982281e+00 6.85725033e-01 -3.54612887e-01
7.14906752e-01 3.54723781e-01 3.19652617e-01 1.24298680e+00
1.26383036e-01 2.79819429e-01 8.78679216e-01 -4.85321045e-01
2.46878237e-01 3.08553036e-02 -3.08170207e-02 4.45520103e-01
9.75463688e-02 3.70450974e-01 -7.70800591e-01 -7.73194283e-02
2.45648876e-01 -2.77258724e-01 -6.43925488e-01 4.02163655e-01
-8.93734574e-01 5.80163002e-01 -8.82314667e-02 4.73104596e-01
-5.87426901e-01 -8.16918090e-02 -4.00961608e-01 1.40473396e-01
6.53987885e-01 -1.23756729e-01 -1.99217170e-01 -2.89533377e-01
-7.76143312e-01 -1.44807026e-01 5.71892083e-01 8.07723999e-01
5.89833140e-01 4.02846545e-01 3.88103455e-01 8.32294285e-01
9.84791279e-01 1.58596563e+00 4.67352629e-01 -8.14253509e-01
4.26749825e-01 -3.70419621e-01 2.23625079e-02 -8.37908328e-01
-4.72806692e-01 -3.59154582e-01 -8.82681310e-01 6.37201639e-03
1.68191046e-01 -5.33197761e-01 -9.39412355e-01 1.42791176e+00
4.50866848e-01 4.37599331e-01 3.60164016e-01 1.04614449e+00
1.31691897e+00 6.40609860e-01 -2.19749674e-01 -6.24426663e-01
1.65237570e+00 -3.05436194e-01 -9.87261772e-01 -4.73155320e-01
1.67063400e-01 -1.03094339e+00 6.79660618e-01 3.54405433e-01
-6.48210645e-01 -4.87844318e-01 -7.62149811e-01 3.18788588e-01
-7.24660680e-02 2.06845224e-01 4.19380963e-01 1.23694849e+00
-7.70786285e-01 -1.49178490e-01 -5.96388876e-01 -1.67902797e-01
4.57812279e-01 -1.53358877e-01 -2.25656584e-01 -4.37267840e-01
-1.12797153e+00 6.56995833e-01 -8.10922205e-01 4.80079979e-01
-4.63125080e-01 -8.15553427e-01 -8.10736477e-01 -3.77107620e-01
-3.18327159e-01 -4.85401720e-01 1.15816259e+00 2.18401439e-02
-1.70821261e+00 8.23173821e-01 -4.66391504e-01 -5.54119721e-02
-8.49029943e-02 -7.41092861e-02 -1.12918353e+00 5.86275339e-01
-2.60315798e-02 2.30226114e-01 1.22888660e+00 -7.85690665e-01
-3.71945560e-01 -7.82850623e-01 -8.04662287e-01 7.62296468e-02
-5.75618818e-02 2.71521449e-01 4.09968108e-01 -3.39413524e-01
5.35555899e-01 -4.65451181e-01 1.35563880e-01 -1.58387244e-01
-3.36977690e-01 -4.75805402e-02 1.18013501e+00 -9.38042164e-01
7.79666483e-01 -2.45115185e+00 -4.80240613e-01 -1.24883354e-01
-2.47512296e-01 3.23996365e-01 -3.02587062e-01 3.49149466e-01
3.35062802e-01 -1.97572947e-01 -3.03675532e-01 -2.67691612e-01
-2.02101290e-01 -1.58362582e-01 -5.88045597e-01 9.38870907e-01
-4.94473219e-01 6.02342010e-01 -4.45520997e-01 -2.65332580e-01
2.59282321e-01 7.68641412e-01 -6.25267401e-02 -5.79362921e-02
3.26203585e-01 6.64849937e-01 -5.60506463e-01 1.17899644e+00
1.54132843e+00 6.87621474e-01 -4.92238939e-01 -2.74404973e-01
-5.53385973e-01 -1.63881201e-02 -1.07421088e+00 1.66365755e+00
-4.55810279e-01 9.50077474e-01 6.11030161e-01 -6.38354063e-01
1.03893888e+00 6.34876311e-01 4.24270451e-01 -8.11922371e-01
5.97204231e-02 1.70103312e-01 -3.55595559e-01 -1.54120207e+00
4.16826904e-01 -3.31900954e-01 1.49030194e-01 3.08080465e-01
-3.68368745e-01 -3.77963543e-01 -5.19450426e-01 -4.10198838e-01
9.58808601e-01 -5.46576321e-01 6.37129843e-02 1.30270988e-01
6.15078509e-01 -2.04327345e-01 1.54141456e-01 5.91023684e-01
-3.85217547e-01 4.99213398e-01 -6.46014869e-01 1.82539567e-01
-4.34565276e-01 -1.32438445e+00 -5.71617723e-01 7.32212484e-01
-2.44215690e-02 2.88707167e-01 -6.04822576e-01 1.77316546e-01
-2.16033116e-01 5.93306959e-01 1.46657288e-01 1.77807629e-01
-3.93682063e-01 -6.63057685e-01 9.80171621e-01 6.23451844e-02
9.57740307e-01 -9.51692939e-01 -7.91459382e-01 -3.22011858e-02
-6.39635265e-01 -1.37666690e+00 -3.31665874e-01 -5.37289560e-01
-5.67278922e-01 -8.98787200e-01 -8.65024388e-01 -8.99339616e-01
2.89385587e-01 9.55446005e-01 3.50639820e-01 -5.02215028e-01
-8.44316244e-01 8.16996336e-01 -3.18734914e-01 -5.71793497e-01
-4.93625812e-02 -3.63501400e-01 4.68106389e-01 1.79727644e-01
4.06037062e-01 -6.39577270e-01 -5.68047881e-01 2.34784469e-01
-3.62467140e-01 -5.85152745e-01 7.03287065e-01 1.89185828e-01
3.48531902e-02 1.50969744e-01 9.53243136e-01 3.45637798e-02
5.73233068e-01 -4.16362554e-01 -3.99200767e-01 -1.20935865e-01
-1.48799643e-01 -5.52945435e-01 5.50709525e-03 -4.08656627e-01
-1.46604073e+00 -1.65583968e-01 -5.40900350e-01 8.44156444e-02
-7.74567723e-01 5.79762280e-01 -5.31289577e-01 -1.95921272e-01
7.55477071e-01 3.75978947e-01 2.46837497e-01 -6.13554418e-01
5.10750651e-01 1.82043457e+00 1.09517550e+00 1.12688079e-01
7.60239601e-01 7.39944994e-01 -1.08756766e-01 -2.14109182e+00
-3.15060109e-01 -8.54150355e-01 -2.33401537e-01 -4.19777066e-01
9.85235989e-01 -1.03834319e+00 -8.82306397e-01 1.04653037e+00
-1.03504848e+00 6.92154467e-02 2.61507332e-01 1.33297825e+00
-3.16455990e-01 6.22397006e-01 -4.00262713e-01 -1.61159003e+00
-3.92145246e-01 -5.39475381e-01 1.27437282e+00 5.59641242e-01
2.04740331e-01 -4.46138382e-01 9.40186083e-02 8.46358955e-01
6.53573990e-01 -2.68130481e-01 3.25752884e-01 1.74470186e-01
-6.50796950e-01 -3.42732430e-01 1.64389610e-01 1.52875155e-01
4.32984143e-01 -2.01485112e-01 -1.46529579e+00 -5.08288592e-02
4.67097759e-01 -2.48916214e-03 8.86440814e-01 1.02165401e+00
5.56346416e-01 -1.37205094e-01 -3.69506091e-01 7.85378575e-01
1.02671385e+00 4.27249193e-01 1.04852295e+00 -2.26443350e-01
2.26628006e-01 8.57623994e-01 6.34415984e-01 6.78556621e-01
2.17057392e-01 4.99544442e-01 4.42129076e-01 2.08118781e-01
-1.23121046e-01 2.92426229e-01 5.55274487e-01 4.99533921e-01
-1.03394084e-01 -3.14795673e-01 -7.06042647e-01 2.48574495e-01
-9.83561516e-01 -1.57501733e+00 -4.52515185e-01 2.01116729e+00
2.17631862e-01 -9.51488554e-01 -2.39992857e-01 2.84019798e-01
7.66274631e-01 3.62369239e-01 -3.58649015e-01 3.75169590e-02
-3.22505981e-01 4.02264088e-01 2.79291362e-01 8.41347158e-01
-1.05905521e+00 6.16261303e-01 5.94437933e+00 5.33131599e-01
-1.37996876e+00 -4.10384275e-02 -1.52438805e-01 2.01830238e-01
-3.33607078e-01 -6.04761541e-01 -6.69753015e-01 1.68682426e-01
9.55176234e-01 -3.61300856e-02 3.79446983e-01 4.88075763e-01
6.22894764e-01 -5.31031311e-01 -2.32772410e-01 1.31632090e+00
2.16833964e-01 -1.06381989e+00 -6.73013449e-01 3.82485658e-01
1.85329303e-01 2.90230840e-01 4.70365316e-01 -5.86740077e-02
-3.82111639e-01 -8.66561413e-01 1.95729420e-01 9.00887847e-01
1.02801073e+00 -7.30851173e-01 4.25614595e-01 9.35452759e-01
-1.13676393e+00 -9.34526250e-02 -6.06780469e-01 1.65672880e-02
3.80920768e-01 7.71678925e-01 -9.28469658e-01 5.58150709e-01
5.76203108e-01 2.29560286e-01 1.30201578e-01 9.79593039e-01
-1.28647715e-01 6.85018599e-01 -2.93679595e-01 -1.24297865e-01
-9.78907868e-02 -1.60379678e-01 8.16729546e-01 9.59551990e-01
1.01215518e+00 4.96471345e-01 -2.99508154e-01 4.52363968e-01
2.60601580e-01 -1.74272716e-01 -9.04666007e-01 9.73495096e-02
8.58528435e-01 1.50903308e+00 2.80232336e-02 1.30897686e-01
-3.04956257e-01 4.01280105e-01 -9.40745294e-01 5.45130491e-01
-6.19154334e-01 -4.77320820e-01 1.03974712e+00 -7.71545339e-04
5.94001636e-02 -8.33103955e-01 9.69737843e-02 -8.92043233e-01
-7.89099559e-02 -4.04817581e-01 -9.99445654e-03 -1.16362500e+00
-1.02365625e+00 3.59265298e-01 -2.48362541e-01 -1.22356105e+00
-4.56353784e-01 -4.09411132e-01 -5.55005848e-01 1.15899134e+00
-1.77696359e+00 -9.48327243e-01 -7.14242339e-01 6.22307539e-01
3.15291435e-01 -4.24432844e-01 1.03303921e+00 5.05417585e-01
-2.23472625e-01 2.47011796e-01 1.24350242e-01 -4.96130483e-03
5.53472281e-01 -3.69867116e-01 1.09493233e-01 6.16683602e-01
-8.16467404e-03 4.14182574e-01 7.39097118e-01 -6.74727678e-01
-1.90537262e+00 -1.08515477e+00 9.52841759e-01 -7.36735947e-03
1.35187373e-01 -3.85042056e-02 -5.26100039e-01 1.14295766e-01
4.13546599e-02 2.62778113e-03 9.41532135e-01 -5.31525552e-01
-9.86791179e-02 -1.71914741e-01 -1.56836903e+00 2.22053617e-01
9.98559833e-01 -7.05479980e-01 -5.22744358e-01 3.74576241e-01
3.61061603e-01 -2.72365421e-01 -4.07180250e-01 4.39583778e-01
8.52016151e-01 -6.32408082e-01 1.37640512e+00 -9.22340602e-02
-3.24916840e-01 -1.48833930e-01 -7.98825026e-01 -1.13097441e+00
1.88945219e-01 -6.25966251e-01 1.77695081e-01 1.06400335e+00
6.88291490e-02 -9.17973697e-01 8.22505534e-01 4.05067392e-02
-3.96530330e-01 -6.57280758e-02 -1.50061822e+00 -8.39882672e-01
-1.67370439e-01 -8.76057148e-01 4.32061344e-01 6.19011045e-01
-4.51735109e-02 2.20311210e-01 -5.74611962e-01 7.77610362e-01
1.13303816e+00 1.56072956e-02 6.24724507e-01 -1.37898171e+00
2.08282039e-01 -6.47470132e-02 -3.47822189e-01 -1.28350556e+00
3.06080639e-01 -7.17760026e-01 2.67536461e-01 -1.76271462e+00
-4.05470490e-01 4.42208946e-02 5.80778956e-01 -9.14934725e-02
5.05242825e-01 1.42414004e-01 -3.06238055e-01 1.40833229e-01
3.73226076e-01 6.88712478e-01 1.11158633e+00 -2.19096065e-01
3.97409516e-04 6.94503129e-01 -4.08424735e-01 8.16944242e-01
8.06020975e-01 -2.53982037e-01 -4.06500161e-01 -2.85551667e-01
-3.17526639e-01 6.39435351e-01 5.36716282e-01 -1.08427513e+00
7.88185179e-01 -3.31427276e-01 1.28011197e-01 -9.53960657e-01
8.59342396e-01 -5.74017763e-01 -3.96459393e-04 5.14972329e-01
2.82435924e-01 -5.89820743e-01 -2.13312760e-01 5.80132782e-01
7.23509956e-03 -2.94325560e-01 9.05513346e-01 4.92643751e-02
-6.80424750e-01 5.49217165e-01 -9.38281238e-01 -1.49575129e-01
7.61574864e-01 -2.12511137e-01 -9.00665939e-01 -7.77426720e-01
-5.38576841e-01 -3.28914285e-01 -3.95411640e-01 2.48161286e-01
1.20114982e+00 -1.14574409e+00 -9.36934710e-01 5.75545788e-01
-1.76787063e-01 -4.81049687e-01 7.85056651e-01 8.11223567e-01
-2.01573819e-01 6.06154323e-01 4.74546812e-02 -8.13012242e-01
-1.50844002e+00 -1.00134499e-01 3.42173964e-01 8.55338931e-01
-7.43099570e-01 8.05729330e-01 -3.73228014e-01 -1.85065627e-01
1.56683937e-01 -1.23305865e-01 -4.07043397e-01 2.71103859e-01
9.60836649e-01 7.38291204e-01 -3.91018391e-02 -7.88612425e-01
-8.35776865e-01 1.00621736e+00 4.91000980e-01 -7.17654645e-01
1.17266536e+00 -5.23487866e-01 1.28231138e-01 8.02778974e-02
1.15442562e+00 7.11887181e-01 -7.41218507e-01 -2.07366437e-01
-2.50752926e-01 -4.70465153e-01 7.87211806e-02 -5.21598756e-01
-6.83617055e-01 1.18741786e+00 9.53205228e-01 6.44593090e-02
1.15649104e+00 2.18311474e-01 8.02731276e-01 7.13446736e-01
6.01774454e-01 -8.86189520e-01 4.94302139e-02 4.06570762e-01
8.75190258e-01 -9.32165384e-01 -2.45693639e-01 -5.20663857e-01
-3.09235543e-01 1.27481878e+00 -1.67914834e-02 6.88781559e-01
9.63754416e-01 7.01337695e-01 7.44437218e-01 -3.46374437e-02
-3.35558146e-01 -5.70669293e-01 -1.00909397e-01 1.37928689e+00
1.68464974e-01 9.36048329e-02 1.87406182e-01 4.75064933e-01
-5.25673509e-01 -1.63025856e-01 4.56507236e-01 6.81332409e-01
-1.19102025e+00 -6.76407456e-01 -6.88985467e-01 6.47033751e-02
-1.14670448e-01 -6.04716316e-02 6.43360093e-02 4.20056015e-01
-2.50620782e-01 1.76597476e+00 -1.35509476e-01 -4.38548297e-01
2.58861303e-01 1.43915161e-01 4.25075561e-01 -2.02923968e-01
3.28079760e-01 1.46690100e-01 1.82442605e-01 -2.35625535e-01
-4.48736459e-01 -9.24311042e-01 -1.47419548e+00 -3.03431123e-01
-2.79149950e-01 3.92824523e-02 1.48709762e+00 8.48719954e-01
3.62540960e-01 1.19419284e-01 1.16139424e+00 -6.10886216e-01
-6.95419192e-01 -1.30342257e+00 -1.08615386e+00 -4.30724472e-01
6.89939559e-01 -4.71782982e-01 -9.70613837e-01 -1.47402644e-01] | [15.059425354003906, 5.765749454498291] |
a57d0c10-4c35-4d1b-9a4d-a32919705bb4 | walking-your-lidog-a-journey-through-multiple | 2304.11705 | null | https://arxiv.org/abs/2304.11705v1 | https://arxiv.org/pdf/2304.11705v1.pdf | Walking Your LiDOG: A Journey Through Multiple Domains for LiDAR Semantic Segmentation | The ability to deploy robots that can operate safely in diverse environments is crucial for developing embodied intelligent agents. As a community, we have made tremendous progress in within-domain LiDAR semantic segmentation. However, do these methods generalize across domains? To answer this question, we design the first experimental setup for studying domain generalization (DG) for LiDAR semantic segmentation (DG-LSS). Our results confirm a significant gap between methods, evaluated in a cross-domain setting: for example, a model trained on the source dataset (SemanticKITTI) obtains $26.53$ mIoU on the target data, compared to $48.49$ mIoU obtained by the model trained on the target domain (nuScenes). To tackle this gap, we propose the first method specifically designed for DG-LSS, which obtains $34.88$ mIoU on the target domain, outperforming all baselines. Our method augments a sparse-convolutional encoder-decoder 3D segmentation network with an additional, dense 2D convolutional decoder that learns to classify a birds-eye view of the point cloud. This simple auxiliary task encourages the 3D network to learn features that are robust to sensor placement shifts and resolution, and are transferable across domains. With this work, we aim to inspire the community to develop and evaluate future models in such cross-domain conditions. | ['Laura Leal-Taixé', 'Elisa Ricci', 'Aljoša Ošep', 'Cristiano Saltori'] | 2023-04-23 | null | null | null | null | ['lidar-semantic-segmentation'] | ['computer-vision'] | [ 1.36416599e-01 3.98684561e-01 3.03770248e-02 -5.65114915e-01
-6.86784863e-01 -8.24937761e-01 3.62876981e-01 -7.98485056e-02
-6.71231866e-01 6.27823949e-01 -3.55524749e-01 -1.24658316e-01
8.57878104e-03 -8.53860438e-01 -1.22018862e+00 -2.35951126e-01
-1.73729107e-01 9.30474162e-01 5.33445597e-01 -3.67881238e-01
8.16618800e-02 6.28951490e-01 -1.70628977e+00 -2.88247298e-02
1.00776660e+00 1.29680943e+00 4.73368466e-01 2.02010021e-01
1.82474583e-01 2.37269238e-01 -7.04531133e-01 -8.00693706e-02
6.33995950e-01 9.54815298e-02 -6.75598860e-01 -3.01045668e-03
5.80896080e-01 -4.69087899e-01 -9.80458781e-02 8.70289683e-01
5.08315742e-01 1.66335940e-01 7.40246892e-01 -1.58345044e+00
-5.99349678e-01 2.45074183e-01 -3.95235777e-01 -1.34480566e-01
3.24669868e-01 2.90775627e-01 7.96236873e-01 -6.58346474e-01
8.40290546e-01 1.51369011e+00 8.25560689e-01 6.54554248e-01
-1.16175163e+00 -7.57683992e-01 4.11756366e-01 -3.31026427e-02
-1.31459844e+00 -2.27616087e-01 6.44244850e-01 -3.30856025e-01
1.18864763e+00 -3.18980128e-01 5.08083820e-01 1.29463220e+00
3.55593413e-02 9.36814308e-01 9.62985516e-01 1.57305747e-01
5.63022375e-01 3.13901789e-02 -1.52697369e-01 6.73558772e-01
5.53868592e-01 2.00127587e-01 -6.00075841e-01 2.73187935e-01
6.21763110e-01 -3.29546064e-01 1.13723852e-01 -7.04389095e-01
-9.92461801e-01 8.33208144e-01 8.33795369e-01 9.55976695e-02
-1.76708803e-01 2.36246645e-01 3.32079917e-01 1.81052938e-01
4.83482778e-01 5.65914094e-01 -7.08864033e-01 -9.47559476e-02
-6.96022928e-01 5.60363352e-01 8.52150500e-01 1.50810933e+00
9.17613208e-01 -1.27086803e-01 4.72559273e-01 6.04453146e-01
2.84929365e-01 8.40452313e-01 1.63697600e-01 -1.22710526e+00
5.14010549e-01 6.78357899e-01 1.38198957e-01 -4.39001262e-01
-8.58855665e-01 -1.10512920e-01 -2.95133114e-01 5.19915879e-01
3.72087359e-01 -4.54635412e-01 -1.15384901e+00 1.80710244e+00
3.66358638e-01 -3.24990630e-01 1.89029932e-01 9.45624113e-01
7.65445232e-01 2.39327997e-01 4.62221093e-02 6.55597508e-01
1.23497224e+00 -8.82664680e-01 -6.81019574e-02 -8.13324749e-01
7.16553926e-01 -3.35867733e-01 1.22451890e+00 2.86196023e-01
-8.29853654e-01 -5.58989227e-01 -1.28340471e+00 -3.46093804e-01
-6.69133723e-01 5.32938354e-03 5.95552802e-01 4.53714162e-01
-1.21709228e+00 4.05949563e-01 -1.07220876e+00 -9.75013077e-01
7.97608972e-01 5.46221972e-01 -4.08428431e-01 -1.23834647e-01
-9.24393296e-01 1.19098282e+00 5.00037491e-01 -3.78941774e-01
-1.15678287e+00 -7.32600868e-01 -1.11129749e+00 -4.38055485e-01
3.15396011e-01 -7.21414447e-01 1.35255110e+00 -7.86942482e-01
-1.35638714e+00 1.05608845e+00 7.71277472e-02 -6.35977626e-01
5.07689655e-01 -3.50448161e-01 -1.12616710e-01 1.66076556e-01
6.38146520e-01 1.51293743e+00 4.86412644e-01 -1.50251091e+00
-8.06118071e-01 -5.98545134e-01 3.42141658e-01 3.01137477e-01
1.16775416e-01 -5.16712189e-01 -1.89161301e-01 -2.77027190e-01
1.39870182e-01 -1.17366183e+00 -2.20992520e-01 3.35170209e-01
-2.23022446e-01 -4.02861178e-01 1.04398644e+00 -3.11044484e-01
1.51107088e-01 -2.26997328e+00 5.79901896e-02 -1.90623686e-01
-1.22064404e-01 -7.20257312e-02 -1.94983408e-01 1.55175194e-01
3.52013171e-01 -6.69311136e-02 -5.07879436e-01 -5.39203823e-01
3.46301317e-01 4.88345176e-01 -7.98373073e-02 3.81178975e-01
4.82657969e-01 8.53823483e-01 -7.64498413e-01 -2.53506422e-01
9.00412947e-02 2.89625794e-01 -7.53831804e-01 5.23029827e-02
-6.59230947e-01 7.09209144e-01 -5.91002762e-01 8.12961519e-01
9.80457902e-01 -8.91962200e-02 -1.04889363e-01 -1.17822878e-01
-2.11358145e-01 2.60463446e-01 -7.21810281e-01 2.54653263e+00
-4.35802162e-01 6.39311016e-01 2.56376624e-01 -1.08108318e+00
1.06237268e+00 -2.37452552e-01 5.31710327e-01 -1.01023853e+00
2.73800403e-01 5.76938152e-01 -2.12956876e-01 -4.82635051e-01
5.72025120e-01 -2.25504681e-01 -5.72158217e-01 2.07731321e-01
2.10172594e-01 -7.18883336e-01 9.08653662e-02 3.84418517e-02
1.15438783e+00 4.83078957e-01 -3.42224926e-01 -3.82979333e-01
-5.17357364e-02 5.10716081e-01 2.89529890e-01 5.23322344e-01
-3.99319351e-01 5.57973802e-01 2.94947356e-01 -2.73895741e-01
-9.55775261e-01 -1.37347019e+00 -2.45222554e-01 1.09670877e+00
8.33017349e-01 1.92926198e-01 -8.86569321e-01 -9.17466581e-01
4.54843014e-01 9.04115260e-01 -4.45488453e-01 -1.53888389e-01
-5.21859467e-01 -2.51362264e-01 7.88688719e-01 9.43578303e-01
1.04310811e+00 -7.81103790e-01 -1.34338140e+00 3.35559957e-02
-6.78571090e-02 -1.50382924e+00 1.21112252e-02 4.99652088e-01
-7.52232909e-01 -9.18498755e-01 -6.31368279e-01 -9.25119460e-01
3.71586919e-01 2.13267207e-01 1.12173402e+00 -4.39540476e-01
-8.23965445e-02 6.48373902e-01 -4.62408125e-01 -7.99217701e-01
-9.50825214e-03 5.15490711e-01 1.60835773e-01 -7.48419881e-01
6.18956089e-01 -6.32061005e-01 -6.76759362e-01 4.11984682e-01
-6.97927713e-01 -1.42548710e-01 5.94808519e-01 2.92289227e-01
5.21200359e-01 -2.78581917e-01 5.53497434e-01 -4.63998139e-01
3.50868255e-01 -5.85790217e-01 -6.74993038e-01 -8.82444531e-02
-3.44018787e-01 9.68574360e-02 4.49012458e-01 -2.59288400e-01
-9.17954803e-01 1.72094226e-01 -3.01895708e-01 -1.08974718e-01
-4.90603685e-01 2.39492044e-01 -3.96449238e-01 -1.08947001e-01
8.28009307e-01 -2.67441005e-01 5.52088805e-02 -4.67648536e-01
5.55182874e-01 6.97903514e-01 5.82182705e-01 -8.53443503e-01
7.51256764e-01 9.12704885e-01 -1.27536267e-01 -7.66593814e-01
-6.79609358e-01 -2.17837527e-01 -7.90803432e-01 6.18644641e-04
1.03678751e+00 -1.21626616e+00 -5.21486282e-01 3.47919136e-01
-1.22027051e+00 -9.24846292e-01 -4.71585929e-01 3.03068906e-01
-8.39105844e-01 -1.19713088e-02 -8.62734541e-02 -4.42021936e-01
-9.43964627e-03 -1.26229131e+00 1.64251840e+00 2.45071307e-01
-3.22075665e-01 -8.21418166e-01 -1.23750336e-01 3.33378911e-01
2.39567146e-01 5.41756749e-01 6.56140387e-01 -7.13298142e-01
-6.84725463e-01 8.93511921e-02 -4.96276677e-01 3.44299972e-01
-6.18825667e-03 -5.44160008e-01 -1.17167628e+00 -3.65650475e-01
-1.16552509e-01 -6.98024452e-01 7.89175570e-01 2.07661435e-01
9.77999389e-01 2.59021729e-01 -6.62686527e-01 6.35431349e-01
1.15143633e+00 2.94394076e-01 3.86453032e-01 6.66742504e-01
5.46201408e-01 4.98042315e-01 8.47781539e-01 1.80218205e-01
9.64218318e-01 5.47071755e-01 7.60884345e-01 -8.87217745e-02
-1.47584736e-01 -3.39454830e-01 3.81168157e-01 8.41947794e-02
1.40347287e-01 -2.11411640e-01 -1.01969028e+00 6.98320448e-01
-1.65674365e+00 -3.27494740e-01 2.64562279e-01 1.77511108e+00
3.81177187e-01 3.23326379e-01 2.02818304e-01 -2.14727238e-01
3.09067518e-01 1.42292619e-01 -1.16863453e+00 -3.45223457e-01
-1.17869385e-01 3.81239980e-01 9.24130499e-01 1.70561939e-01
-1.14481843e+00 1.22382617e+00 5.96807194e+00 5.18376112e-01
-1.20313954e+00 1.96035445e-01 1.23342648e-01 -1.70936689e-01
-2.44229123e-01 -2.63447881e-01 -6.76385522e-01 3.02774549e-01
8.64611506e-01 8.68908316e-02 3.52391750e-01 1.22153115e+00
-4.09630463e-02 -4.45294976e-01 -1.46601355e+00 7.85062850e-01
9.49136913e-02 -9.57905710e-01 -2.25735918e-01 4.36874442e-02
7.29256153e-01 6.14131987e-01 3.16462457e-01 5.01412153e-01
5.43671846e-01 -1.05747998e+00 1.16799867e+00 -3.45620587e-02
9.24890041e-01 -6.12781763e-01 4.62795526e-01 4.75072175e-01
-1.12950981e+00 -4.39293906e-02 -4.35006082e-01 -1.87310770e-01
9.73931774e-02 1.97470710e-01 -1.03334713e+00 5.35173237e-01
1.06737685e+00 8.26161444e-01 -4.19655174e-01 6.69765413e-01
-1.18794762e-01 1.28208576e-02 -7.89232254e-01 6.85534254e-02
3.86093915e-01 2.55234204e-02 4.53137338e-01 9.38129425e-01
5.47228575e-01 -4.47101146e-02 1.82184339e-01 1.00127733e+00
-1.70270696e-01 -4.72125202e-01 -9.90944386e-01 5.41976504e-02
5.30563116e-01 8.03332090e-01 -8.27954710e-01 -6.85881078e-02
-2.76728570e-01 1.17392850e+00 3.64560813e-01 3.14767957e-01
-1.00915825e+00 -2.87240237e-01 9.59851980e-01 1.54152945e-01
5.08785486e-01 -6.39575779e-01 -7.40730822e-01 -8.15857768e-01
1.30206347e-01 -6.11349344e-01 1.69456899e-01 -8.56206834e-01
-1.26174378e+00 4.82436538e-01 1.46708995e-01 -1.10760260e+00
-5.90413734e-02 -9.23147976e-01 -1.16419666e-01 4.34793830e-01
-1.75621665e+00 -1.40049911e+00 -5.17660677e-01 5.22699833e-01
6.31047666e-01 5.01534902e-02 6.44845784e-01 3.44864637e-01
-1.76012382e-01 5.45224130e-01 -7.59591311e-02 -1.05909761e-02
6.22320592e-01 -1.07728910e+00 6.67217433e-01 5.02048910e-01
-1.26579210e-01 3.39350581e-01 6.24611020e-01 -6.62325025e-01
-1.31294394e+00 -1.34392560e+00 3.12163860e-01 -8.88806581e-01
4.13315773e-01 -5.87397695e-01 -5.83279371e-01 1.03133655e+00
9.55239311e-02 -2.34752581e-01 3.66840839e-01 -7.43022710e-02
-3.79713774e-01 -6.79499954e-02 -1.64905024e+00 3.70745838e-01
1.64541709e+00 -4.89193171e-01 -6.56369984e-01 2.80591667e-01
1.07293928e+00 -7.64054596e-01 -7.95577586e-01 6.16255999e-01
4.86325473e-01 -1.01709700e+00 9.97961640e-01 -1.33394077e-01
3.88598740e-01 -2.83098817e-01 -4.97772664e-01 -1.33138120e+00
2.61507958e-01 -1.83109060e-01 3.40433061e-01 8.14064562e-01
4.76756245e-01 -8.09875846e-01 9.37987089e-01 7.11853802e-01
-5.49359679e-01 -4.61829126e-01 -1.22984624e+00 -1.18749952e+00
4.85047936e-01 -6.79914832e-01 8.21347177e-01 4.71125036e-01
-2.29647294e-01 1.81042016e-01 3.86667728e-01 4.83838886e-01
6.27027035e-01 8.43523666e-02 9.33366597e-01 -1.32943010e+00
2.58586377e-01 -2.57324964e-01 -5.64888895e-01 -1.59499013e+00
4.60467160e-01 -8.61091077e-01 4.25076842e-01 -1.88197803e+00
-3.35528731e-01 -8.32778454e-01 1.90357670e-01 4.41125274e-01
4.24692005e-01 2.58406699e-01 1.73917681e-01 7.40288291e-03
-7.75157273e-01 7.41633415e-01 1.16980362e+00 -2.46777281e-01
-5.96216246e-02 -2.62280554e-01 -7.74840951e-01 9.35813665e-01
8.33028972e-01 -3.52220565e-01 -5.35858750e-01 -1.08443594e+00
-1.14756107e-01 -6.23187423e-01 6.41014874e-01 -1.49087512e+00
1.80390328e-01 -6.37770370e-02 2.33869016e-01 -7.39185035e-01
6.58358335e-01 -1.02566433e+00 -1.53923228e-01 3.19313467e-01
1.45455927e-01 -1.12197474e-01 5.45803130e-01 4.60663021e-01
1.44286349e-01 1.85702590e-03 7.27854133e-01 -1.34528413e-01
-1.17192650e+00 2.30967522e-01 -3.19660634e-01 3.12093437e-01
1.22420382e+00 -5.91445446e-01 -3.64099562e-01 3.53345983e-02
-4.71802771e-01 4.70982939e-01 9.78967726e-01 5.19836664e-01
4.83313441e-01 -1.09974921e+00 -2.83261001e-01 2.64763415e-01
2.95686156e-01 8.40563416e-01 -6.24636710e-02 4.54098672e-01
-6.08303010e-01 4.23378915e-01 -3.22891027e-01 -8.87721717e-01
-6.16781533e-01 2.85747111e-01 3.19616377e-01 3.34062517e-01
-5.47138155e-01 1.18559611e+00 2.16422334e-01 -9.72103357e-01
2.34576881e-01 -8.17547560e-01 3.37910891e-01 -9.09890048e-03
-1.44574001e-01 3.30712438e-01 5.58423139e-02 -6.09461427e-01
-5.74539363e-01 7.16227889e-01 2.50404745e-01 -1.95390984e-01
1.36757720e+00 -1.91457465e-01 2.19711185e-01 3.74688596e-01
1.25766206e+00 -4.33923841e-01 -1.91569734e+00 8.69441405e-02
8.94915611e-02 -1.65992096e-01 -2.66336739e-01 -9.54906464e-01
-8.91945004e-01 9.10711646e-01 7.23873615e-01 2.24913154e-02
9.29341197e-01 4.41548020e-01 1.04585111e+00 4.89096671e-01
1.05260050e+00 -1.28826475e+00 2.77616888e-01 7.52841473e-01
8.59165013e-01 -1.48299551e+00 -2.25751787e-01 -2.51465231e-01
-7.63458133e-01 7.70115435e-01 8.35415900e-01 -3.76822472e-01
5.38941741e-01 3.72079104e-01 1.37294948e-01 -5.18744111e-01
-2.15596706e-01 -5.52294433e-01 -1.84869990e-01 1.21405935e+00
-4.44718041e-02 1.85462341e-01 3.36452425e-01 5.89604020e-01
-5.04753530e-01 1.07309490e-01 8.77057686e-02 1.16803002e+00
-7.29387760e-01 -8.75922382e-01 -1.89927980e-01 1.45953760e-01
2.87899464e-01 3.39578688e-01 -5.25095344e-01 1.09213507e+00
6.31688118e-01 1.12188554e+00 3.10347348e-01 -3.66742939e-01
8.58457863e-01 -7.78001025e-02 6.27093494e-01 -7.76472688e-01
-1.95828915e-01 -3.66952807e-01 1.46897316e-01 -6.75692439e-01
-4.47940558e-01 -8.18894625e-01 -1.68255389e+00 -1.92741424e-01
1.66552532e-02 -2.89580852e-01 8.50803733e-01 9.50312853e-01
5.36004484e-01 4.80387241e-01 1.49850264e-01 -1.29658830e+00
-3.47057283e-01 -7.06619859e-01 -4.50546712e-01 3.37919414e-01
1.45192206e-01 -1.14656866e+00 -1.77293658e-01 -1.42037988e-01] | [8.124849319458008, -2.566190242767334] |
cc2d8d09-0868-445f-b24a-3c72e926b835 | retrieval-of-boost-invariant-symbolic | 2306.13496 | null | https://arxiv.org/abs/2306.13496v1 | https://arxiv.org/pdf/2306.13496v1.pdf | Retrieval of Boost Invariant Symbolic Observables via Feature Importance | Deep learning approaches for jet tagging in high-energy physics are characterized as black boxes that process a large amount of information from which it is difficult to extract key distinctive observables. In this proceeding, we present an alternative to deep learning approaches, Boost Invariant Polynomials, which enables direct analysis of simple analytic expressions representing the most important features in a given task. Further, we show how this approach provides an extremely low dimensional classifier with a minimum set of features representing %effective discriminating physically relevant observables and how it consequently speeds up the algorithm execution, with relatively close performance to the algorithm using the full information. | ['Francesco Romeo', 'Christoph Ortner', 'Ilyes Batatia', 'Jose M Munoz'] | 2023-06-23 | null | null | null | null | ['jet-tagging', 'retrieval'] | ['graphs', 'methodology'] | [-4.57289040e-01 -4.41538393e-02 -9.31185856e-02 -2.90930718e-01
-8.59507263e-01 -6.32851124e-01 8.35354626e-01 4.05093908e-01
-4.16640341e-01 8.12423348e-01 -6.71785846e-02 -4.05095726e-01
-6.50142610e-01 -7.38059938e-01 -4.12111759e-01 -1.11461616e+00
-4.60282117e-01 8.91524553e-01 3.37025493e-01 -2.07769871e-01
2.08390895e-02 1.09926212e+00 -9.88813937e-01 1.56639352e-01
1.64009243e-01 1.30705380e+00 -4.32014018e-01 9.34172034e-01
3.56355458e-02 7.68961132e-01 -4.07531917e-01 -5.12303531e-01
5.34712374e-01 -2.69244403e-01 -9.58564281e-01 -8.73618960e-01
1.95933267e-01 5.55316694e-02 -5.86002767e-01 1.01317692e+00
6.12234175e-01 4.17028755e-01 8.35925162e-01 -9.80931520e-01
-3.26376528e-01 3.37842852e-01 7.47760236e-02 8.10436785e-01
-1.05574364e-02 9.94596537e-03 1.28896892e+00 -7.92672098e-01
5.68928838e-01 1.27468181e+00 8.05323899e-01 2.35025927e-01
-1.31817114e+00 -4.73384559e-01 -6.18146896e-01 1.75981522e-01
-8.41952980e-01 -3.61342907e-01 1.00735116e+00 -4.84199524e-01
1.07937753e+00 4.90419358e-01 6.61261022e-01 9.92566288e-01
9.04782295e-01 3.83095562e-01 1.14357197e+00 -5.45881748e-01
3.43509346e-01 2.68054843e-01 4.67117429e-01 8.12444389e-01
1.80963814e-01 8.09306324e-01 -5.67356050e-01 -4.93798763e-01
6.67695045e-01 -3.26740801e-01 3.12871218e-01 -7.04525173e-01
-1.19225299e+00 1.11389434e+00 4.97550249e-01 5.46231091e-01
-2.05503017e-01 2.20282465e-01 7.85498083e-01 5.94737649e-01
4.20356780e-01 9.22512650e-01 -8.90501142e-01 -7.34734461e-02
-6.56298041e-01 5.45173943e-01 9.25472558e-01 5.69481492e-01
7.01426208e-01 1.94925353e-01 -3.80793989e-01 2.12234378e-01
-1.16927221e-01 4.20777500e-01 2.20575705e-01 -7.86302865e-01
-6.72464026e-03 1.90872967e-01 1.35434777e-01 -7.02129722e-01
-1.11555672e+00 -5.64616501e-01 -1.00366473e+00 4.32399064e-01
4.29016024e-01 -3.05659771e-01 -6.73895657e-01 1.37003505e+00
2.43135616e-01 -4.25954640e-01 1.99879557e-01 6.86295331e-01
9.73084152e-01 6.35778368e-01 1.09524183e-01 -5.06459922e-02
1.41128182e+00 -7.17527330e-01 -5.92221439e-01 1.78780958e-01
2.98833370e-01 -6.99532509e-01 3.95882130e-01 2.62660742e-01
-9.81425643e-01 -8.32874596e-01 -9.31195319e-01 -4.46052164e-01
-6.77209675e-01 -2.20932458e-02 1.62045670e+00 6.06983483e-01
-1.10914350e+00 1.19100130e+00 -8.35306883e-01 2.75228452e-02
2.01248713e-02 7.95437038e-01 -3.81511539e-01 7.03278363e-01
-1.31655371e+00 9.71236706e-01 7.04132318e-01 -1.21908206e-02
-7.21537828e-01 -9.44300532e-01 -7.58987308e-01 3.07192326e-01
1.17816487e-02 -7.93701947e-01 1.56239295e+00 -4.12905365e-01
-1.69623876e+00 5.90785921e-01 -1.10357419e-01 -6.50871873e-01
3.67229015e-01 -9.73458067e-02 -3.58954012e-01 2.52158731e-01
-8.56012106e-02 4.83075231e-01 1.02888608e+00 -7.97079980e-01
-3.81835103e-01 -7.76553452e-02 2.32174948e-01 -3.32078189e-01
1.53817534e-01 5.11726141e-01 1.71691805e-01 -4.68955129e-01
6.32873848e-02 -9.38516021e-01 -4.94155735e-01 -4.26169842e-01
-2.18115866e-01 -3.41110677e-01 4.72647846e-01 -4.20034021e-01
5.69146991e-01 -1.83266389e+00 4.68551069e-01 1.52473003e-01
5.81706703e-01 1.37623385e-01 2.28784636e-01 5.74823558e-01
-2.47023195e-01 -2.33292822e-02 1.37757197e-01 -3.86480205e-02
3.23012352e-01 9.96150896e-02 -1.92184836e-01 4.28173393e-01
4.21853662e-01 1.31877100e+00 -8.94845903e-01 -2.29210168e-01
4.14001554e-01 2.11029902e-01 -3.69101465e-01 1.03685565e-01
6.79546669e-02 8.50852251e-01 -7.17169523e-01 4.77761149e-01
5.13354599e-01 1.01207420e-01 -1.03320554e-01 -1.20067619e-01
-2.68635064e-01 3.94647002e-01 -9.26188350e-01 1.30658257e+00
-3.91082495e-01 8.44915092e-01 1.30279690e-01 -1.27180290e+00
9.31502163e-01 4.49682713e-01 9.13608730e-01 -5.91660976e-01
4.41917002e-01 1.12539090e-01 3.24332088e-01 -2.11154908e-01
4.37965572e-01 -8.10763717e-01 -5.29665470e-01 1.25559330e-01
6.28800869e-01 -2.17309102e-01 4.67517525e-02 -2.15791017e-02
1.36290777e+00 -3.05298299e-01 5.70278764e-01 -7.84588039e-01
6.01760983e-01 3.50376256e-02 5.37374675e-01 1.01312363e+00
-2.91062117e-01 2.70618517e-02 8.66005003e-01 -1.04127514e+00
-1.25004113e+00 -1.05217552e+00 -5.99034488e-01 7.50248134e-01
-2.46404603e-01 -6.52323425e-01 -2.56397724e-01 -5.29178560e-01
1.53223023e-01 3.72783810e-01 -5.72848141e-01 -2.98987448e-01
-6.48405194e-01 -9.17453527e-01 4.05713499e-01 5.46981037e-01
1.39945596e-01 -1.25189948e+00 -5.05432844e-01 -7.12968921e-03
4.35253561e-01 -1.03767836e+00 3.42605174e-01 9.42549884e-01
-8.77008379e-01 -8.20730150e-01 -2.59336352e-01 -5.36745787e-01
2.49557257e-01 -2.40583897e-01 1.26354063e+00 -4.24185693e-01
-5.05772352e-01 -7.09733590e-02 -1.79132670e-01 -5.77117264e-01
-7.32951820e-01 -4.23527649e-03 3.77491921e-01 -3.45626116e-01
5.14871776e-01 -4.85934436e-01 -8.84642377e-02 -4.51327473e-01
-4.65170592e-01 -3.55092973e-01 7.56556928e-01 1.10299885e+00
3.19474667e-01 3.75583798e-01 2.01861665e-01 -7.44853199e-01
4.84633744e-01 -3.79508138e-01 -9.82696176e-01 -1.06161565e-01
-3.48498315e-01 5.74905932e-01 1.08304346e+00 -1.00476734e-01
-7.60126531e-01 -2.45259460e-02 -3.00683677e-01 -2.56414533e-01
-2.26306573e-01 5.92539087e-02 1.62517518e-01 -6.70013905e-01
4.81143564e-01 -1.38908058e-01 -6.14952147e-01 -9.12311316e-01
3.18657845e-01 1.10503420e-01 5.23646533e-01 -7.52161205e-01
1.28952718e+00 3.25191230e-01 1.03005195e+00 -7.56107628e-01
-9.07706678e-01 -6.89267933e-01 -1.42362189e+00 8.40155184e-02
9.79567647e-01 -5.32583296e-01 -8.93725097e-01 -1.23279030e-03
-1.21841502e+00 1.82616383e-01 -6.98808551e-01 8.19825530e-01
-6.37701094e-01 2.29101136e-01 -7.90736735e-01 -8.98515642e-01
-3.82703573e-01 -1.09131157e+00 1.18500066e+00 5.51391914e-02
8.54590349e-03 -1.20837367e+00 4.66294616e-01 -1.99180990e-01
2.11629868e-01 4.40813512e-01 1.40074968e+00 -8.99696648e-01
-4.52082187e-01 -2.78752267e-01 -2.71548569e-01 3.31266344e-01
-3.38832766e-01 -1.94706410e-01 -1.06810439e+00 -4.48167473e-01
3.20514649e-01 -2.45640650e-01 1.06085765e+00 4.75853086e-01
1.06956220e+00 2.09235802e-01 -2.00431868e-01 8.96705031e-01
1.23807228e+00 2.61004418e-01 3.59968841e-02 1.85221657e-01
5.13736546e-01 2.50094324e-01 3.95146966e-01 1.80462852e-01
-4.33299720e-01 7.77521253e-01 -8.27364810e-03 -4.19021875e-01
1.31418154e-01 8.46256465e-02 1.47674039e-01 1.00783622e+00
-2.87928015e-01 2.16676623e-01 -7.41187215e-01 3.69102582e-02
-1.76115882e+00 -1.03391373e+00 -4.03288662e-01 1.92886400e+00
5.91562271e-01 4.13335025e-01 1.67437494e-02 2.52732914e-02
9.13572684e-02 1.66249573e-01 -2.72590041e-01 -9.64843035e-01
1.81551173e-01 1.07529724e+00 6.67383909e-01 5.44079185e-01
-1.56151319e+00 8.82859588e-01 8.24885559e+00 7.15787113e-01
-8.38655591e-01 4.50723886e-01 2.43157908e-01 4.11331765e-02
-6.20439537e-02 2.03152299e-01 -8.22144985e-01 2.22163975e-01
1.29432106e+00 -3.51103097e-01 2.28983805e-01 9.06833172e-01
2.44550761e-02 1.06485151e-01 -1.45976448e+00 8.59511197e-01
-4.84702498e-01 -1.36255419e+00 -1.48100749e-01 -8.26692954e-03
5.86921751e-01 1.66989207e-01 -1.53540343e-01 7.34966516e-01
3.59245628e-01 -8.48640323e-01 4.97084707e-01 6.59253180e-01
4.85333353e-01 -9.07251179e-01 8.40096891e-01 2.56900162e-01
-9.51010942e-01 -2.27941483e-01 -1.00404072e+00 -3.35300416e-01
-8.33999068e-02 7.43998945e-01 -6.96019948e-01 9.97256756e-01
5.67433178e-01 2.75249064e-01 -4.97992128e-01 9.95023012e-01
-6.45968169e-02 3.69407624e-01 -1.65938884e-01 -2.09038630e-01
4.07390594e-01 -3.29197854e-01 8.99309993e-01 1.51238012e+00
1.15046032e-01 5.58383651e-02 2.70910352e-01 8.43501866e-01
6.16564974e-02 -5.88036375e-04 -7.74973214e-01 4.54156147e-03
-1.57346949e-01 1.62461841e+00 -8.12120318e-01 -2.64669150e-01
-3.46896887e-01 5.47721207e-01 5.21746874e-01 1.37546629e-01
-6.76001787e-01 -2.34861851e-01 9.17544246e-01 -4.34290886e-01
2.54944175e-01 -6.94031775e-01 5.25637567e-02 -1.15401304e+00
-2.69405127e-01 -5.16763806e-01 3.74596238e-01 -2.73192793e-01
-1.37953770e+00 5.46678543e-01 1.29145101e-01 -9.54188764e-01
-2.99742490e-01 -1.30423188e+00 -7.25279391e-01 9.65829015e-01
-1.61431336e+00 -8.16923976e-01 2.23469839e-01 1.94548026e-01
2.82183439e-01 -4.49258685e-01 1.14643073e+00 1.22234553e-01
-4.08248514e-01 2.02529237e-01 7.47141421e-01 1.90883413e-01
3.83201778e-01 -1.80003393e+00 6.73609257e-01 5.88175416e-01
4.86642241e-01 5.21877825e-01 8.63881230e-01 -4.14045066e-01
-1.68949628e+00 -6.38827801e-01 7.84384131e-01 -6.94292724e-01
9.92152750e-01 -8.33294809e-01 -5.69694579e-01 5.97084999e-01
-1.24610979e-02 3.43140483e-01 6.57376766e-01 5.08453131e-01
-9.32780618e-04 -1.68277055e-01 -8.81940663e-01 1.14860617e-01
7.61622488e-01 -7.38334954e-01 -7.32895672e-01 9.26692307e-01
5.61163664e-01 -4.20180589e-01 -1.06731892e+00 5.18873870e-01
3.47199947e-01 -1.01448584e+00 1.22777653e+00 -1.42569935e+00
-2.50917487e-02 1.41219512e-01 2.64220566e-01 -1.24547231e+00
-1.05739486e+00 -9.79707003e-01 -1.94563404e-01 7.13661790e-01
2.37286270e-01 -5.87643385e-01 4.98398006e-01 4.55494255e-01
-1.15841724e-01 -6.61116183e-01 -1.39368391e+00 -9.24452960e-01
5.37714660e-01 -4.07153279e-01 3.10382456e-01 7.20870972e-01
-8.35883170e-02 3.21177214e-01 -2.95828074e-01 2.75223732e-01
4.63837445e-01 4.22859013e-01 4.11267370e-01 -1.60644531e+00
-4.08780009e-01 -3.57982725e-01 -8.43137205e-01 -4.77677226e-01
5.30317485e-01 -1.38298738e+00 -3.09664518e-01 -9.15553749e-01
3.36960822e-01 -5.70157230e-01 -8.61188471e-01 1.83806911e-01
1.96244717e-01 -7.84461945e-03 -3.28479744e-02 -5.31213917e-02
-4.67746496e-01 4.94579136e-01 1.14978635e+00 -1.42228141e-01
7.13371709e-02 2.09517807e-01 -1.36899084e-01 8.64322305e-01
7.40921080e-01 -6.60943449e-01 1.00560255e-01 7.02321623e-03
1.94661364e-01 2.19268218e-01 6.84487998e-01 -1.15736425e+00
-2.51360778e-02 -7.33473450e-02 1.00287163e+00 -4.87553895e-01
5.68454623e-01 -5.59341431e-01 -9.30174366e-02 4.68778580e-01
-3.27219993e-01 1.05960995e-01 3.54517967e-01 2.74464577e-01
-2.03756928e-01 -7.43988693e-01 8.23636472e-01 -3.60142440e-01
-6.52505815e-01 3.25257748e-01 -2.46830955e-01 -6.60694838e-02
9.32119787e-01 5.78892231e-01 -1.37379169e-02 -1.30451649e-01
-8.77057195e-01 -2.05602273e-01 4.16117199e-02 3.35367799e-01
1.22723974e-01 -1.49191344e+00 -5.37186861e-01 2.27621600e-01
-3.84120405e-01 -6.10793293e-01 -5.40020913e-02 6.96594417e-01
-5.38193107e-01 1.12022519e+00 -4.44903493e-01 -4.29151505e-01
-1.13375545e+00 9.93151724e-01 5.18831968e-01 -6.18191421e-01
-1.01432240e+00 9.06224608e-01 1.12253800e-01 -3.25870097e-01
-2.79787272e-01 -7.00918436e-01 -1.13695659e-01 -5.41683920e-02
2.87329048e-01 2.50855654e-01 4.46980000e-01 -6.39256120e-01
-2.99125016e-01 4.82149184e-01 1.28546119e-01 1.00773983e-01
1.21379268e+00 5.37065744e-01 -3.13710451e-01 5.65502763e-01
1.29121852e+00 3.45343709e-01 -8.69880497e-01 -1.26986668e-01
3.17848265e-01 -3.30500752e-01 4.31162626e-01 -5.92142880e-01
-6.54972851e-01 1.12609553e+00 4.78357077e-01 6.77488923e-01
7.60592461e-01 2.45535240e-01 8.10503066e-01 8.84261608e-01
3.23750734e-01 -7.84270644e-01 -3.61520678e-01 6.44386113e-01
6.70059443e-01 -1.28522575e+00 2.46484324e-01 -3.08375567e-01
-3.57286967e-02 1.71127009e+00 1.39399096e-01 -1.84297994e-01
7.26775408e-01 6.02059484e-01 -1.60216048e-01 -5.38584173e-01
-8.19930553e-01 -3.75044554e-01 7.17852950e-01 5.38293004e-01
3.26072186e-01 1.99022353e-01 -3.27508122e-01 6.63685620e-01
-6.14623249e-01 -6.95090055e-01 7.44529217e-02 5.25452554e-01
-4.51214790e-01 -1.36994982e+00 -1.84148788e-01 4.30728793e-01
-7.22331285e-01 -3.91154699e-02 -3.75367373e-01 1.04895985e+00
6.67493939e-02 5.93045652e-01 -1.73339874e-01 -1.54605091e-01
1.83241770e-01 4.71912861e-01 6.46705687e-01 -5.82594275e-01
-6.61085665e-01 -3.31571847e-01 1.29581422e-01 -6.17262363e-01
-8.04026797e-02 -7.41509676e-01 -1.15369964e+00 -1.90896168e-01
-1.53058007e-01 5.63320398e-01 6.59544468e-01 1.11846745e+00
-1.24138772e-01 8.45999837e-01 7.01882243e-01 -1.17058480e+00
-9.89821374e-01 -8.11398268e-01 -6.84043288e-01 6.97260126e-02
6.29022062e-01 -8.70443642e-01 -3.50806266e-01 -5.01140237e-01] | [15.69823932647705, 2.91747784614563] |
0b6ee4f3-1a0d-4613-b8f4-62c713f3f282 | boost-video-frame-interpolation-via-motion | 2306.13933 | null | https://arxiv.org/abs/2306.13933v1 | https://arxiv.org/pdf/2306.13933v1.pdf | Boost Video Frame Interpolation via Motion Adaptation | Video frame interpolation (VFI) is a challenging task that aims to generate intermediate frames between two consecutive frames in a video. Existing learning-based VFI methods have achieved great success, but they still suffer from limited generalization ability due to the limited motion distribution of training datasets. In this paper, we propose a novel optimization-based VFI method that can adapt to unseen motions at test time. Our method is based on a cycle-consistency adaptation strategy that leverages the motion characteristics among video frames. We also introduce a lightweight adapter that can be inserted into the motion estimation module of existing pre-trained VFI models to improve the efficiency of adaptation. Extensive experiments on various benchmarks demonstrate that our method can boost the performance of two-frame VFI models, outperforming the existing state-of-the-art methods, even those that use extra input. | ['Yanfeng Wang', 'Ya zhang', 'Weidi Xie', 'Xiaoyun Zhang', 'HaoNing Wu'] | 2023-06-24 | null | null | null | null | ['video-frame-interpolation', 'motion-estimation'] | ['computer-vision', 'computer-vision'] | [ 2.08065420e-01 -4.94324863e-01 -4.91671413e-01 -2.14869514e-01
-5.74074388e-01 -3.64171863e-01 5.29179275e-01 -5.51769376e-01
-1.58285841e-01 7.29572952e-01 1.48404226e-01 -3.31001163e-01
2.55475849e-01 -4.08671409e-01 -9.64077830e-01 -5.11130869e-01
-8.57656524e-02 -6.66007400e-02 6.74746215e-01 2.27027182e-02
1.84018150e-01 7.50645697e-02 -1.44423425e+00 3.97260994e-01
7.81189859e-01 9.48158622e-01 2.52828449e-01 7.22500861e-01
1.17157593e-01 1.13105583e+00 -4.23851550e-01 -1.29732974e-02
2.48480350e-01 -6.43655419e-01 -7.55548060e-01 -4.62076925e-02
4.23806906e-01 -6.59190118e-01 -6.00500464e-01 5.53082287e-01
3.57694566e-01 3.48991096e-01 4.94258165e-01 -1.25765848e+00
-5.21729529e-01 2.52389848e-01 -4.01640356e-01 3.87493700e-01
5.16314685e-01 3.34651738e-01 5.74002206e-01 -1.16171718e+00
7.66630173e-01 1.20927942e+00 7.19663978e-01 9.43027079e-01
-1.14811683e+00 -6.94314659e-01 3.63745600e-01 7.34271467e-01
-1.38099658e+00 -5.60479462e-01 7.92526305e-01 -4.93525416e-01
9.94624615e-01 1.18093818e-01 7.13019252e-01 1.15713370e+00
1.36888698e-01 1.05847752e+00 5.88849425e-01 -2.16401249e-01
1.96311861e-01 -5.15363872e-01 -4.73951370e-01 5.50126970e-01
-8.83457363e-02 1.89490303e-01 -7.51058280e-01 8.98186043e-02
1.12822938e+00 -1.62850782e-01 -6.44599140e-01 -4.26714003e-01
-1.58875513e+00 6.69984162e-01 2.11880788e-01 1.64268576e-02
-8.17232728e-02 2.12692827e-01 3.89372677e-01 1.56810507e-01
3.95489752e-01 7.33713731e-02 -4.11812216e-01 -2.76054829e-01
-1.23528504e+00 4.44805980e-01 5.63857853e-01 1.00737834e+00
7.76296616e-01 3.31809044e-01 -6.23169839e-01 6.46053433e-01
2.87430227e-01 1.50624141e-01 6.04471028e-01 -1.28122306e+00
5.06711483e-01 3.06358188e-01 2.22874165e-01 -8.13163340e-01
-1.21575080e-01 -8.90261754e-02 -7.93937683e-01 2.08085269e-01
3.54044408e-01 5.85890375e-04 -9.35441732e-01 1.65293741e+00
4.95177656e-01 1.07921886e+00 -2.07969278e-01 8.84900510e-01
8.40185344e-01 8.02994490e-01 9.30594727e-02 -2.10648924e-01
5.08896887e-01 -1.51978016e+00 -6.02923810e-01 -1.78663149e-01
3.99248153e-01 -7.62382746e-01 1.04880500e+00 2.72329211e-01
-1.21802795e+00 -9.24467444e-01 -1.06545687e+00 3.29869576e-02
3.36277783e-01 -2.59320345e-03 5.17896593e-01 4.48542446e-01
-1.02171397e+00 6.21092558e-01 -9.80510592e-01 1.30870953e-01
5.10828376e-01 4.69053775e-01 -2.05299065e-01 -2.35263303e-01
-8.87788951e-01 3.55887264e-01 5.17514706e-01 6.95021749e-02
-1.11973262e+00 -9.59352314e-01 -1.03047335e+00 -8.95464122e-02
3.77541989e-01 -9.87084031e-01 1.33388531e+00 -1.13971186e+00
-1.86386096e+00 1.78536013e-01 -6.80014849e-01 -4.38434064e-01
7.73972511e-01 -2.82923758e-01 -3.58911812e-01 1.89092785e-01
-5.61797246e-02 1.05373406e+00 1.31647694e+00 -1.21936178e+00
-8.64022613e-01 3.07480276e-01 8.28754306e-02 2.51084846e-02
-2.79032528e-01 -2.03493640e-01 -8.99199843e-01 -1.07917047e+00
-2.47330144e-01 -1.13456678e+00 -2.90052623e-01 7.77437091e-02
-1.35656267e-01 -1.71955585e-01 1.12150753e+00 -5.17053366e-01
1.72014701e+00 -2.05685949e+00 2.88105935e-01 -1.38038248e-01
-2.63622440e-02 6.84192955e-01 -2.37034321e-01 9.25527215e-02
1.54211685e-01 -7.81634003e-02 -1.31325558e-01 -4.06820536e-01
-3.70937914e-01 4.05098319e-01 -1.51622474e-01 1.56609669e-01
3.96368086e-01 8.35619867e-01 -1.10106373e+00 -6.29872739e-01
3.84069026e-01 5.41742325e-01 -1.00525129e+00 4.58957791e-01
-3.16780537e-01 1.10453510e+00 -9.81954560e-02 5.18649340e-01
5.95494628e-01 -4.36449021e-01 1.10281296e-01 -3.64385456e-01
8.01719204e-02 1.35558173e-01 -1.08688128e+00 2.01742458e+00
-4.02248412e-01 6.49813056e-01 -3.92786354e-01 -9.77716208e-01
4.96842831e-01 4.99391764e-01 6.42897010e-01 -2.57224172e-01
-1.64051905e-01 1.29120678e-01 7.12971552e-04 -4.26777810e-01
4.11324471e-01 4.11097646e-01 5.16333640e-01 1.17222160e-01
-3.11159100e-02 2.06177965e-01 4.16312456e-01 -8.84222686e-02
9.36212182e-01 8.27559710e-01 2.33405784e-01 -9.95213818e-03
9.43129420e-01 -1.59601644e-01 1.08365643e+00 7.39028156e-01
-2.00777784e-01 9.64093089e-01 -1.19993821e-01 -7.53173828e-01
-1.11825848e+00 -9.58063126e-01 7.11793154e-02 8.56130958e-01
3.05919856e-01 -5.15293479e-01 -6.91161394e-01 -7.53621042e-01
-2.42792323e-01 3.67918879e-01 -4.43463087e-01 -1.10650852e-01
-1.04846931e+00 -4.87831831e-01 7.68458843e-02 1.00681031e+00
5.85056722e-01 -8.10853004e-01 -5.98907709e-01 5.08293152e-01
-5.97172856e-01 -1.39903343e+00 -8.52519393e-01 -5.43241858e-01
-9.98857319e-01 -7.91493893e-01 -9.36623693e-01 -9.96057808e-01
5.72551668e-01 5.49578130e-01 1.29705250e+00 3.79239708e-01
-1.22589627e-04 3.40959579e-01 -3.91683817e-01 -1.33561358e-01
-6.01760626e-01 2.05981940e-01 -1.04788683e-01 9.15404931e-02
1.57938972e-01 -5.39836943e-01 -1.11065412e+00 7.40669489e-01
-9.71261561e-01 3.74459088e-01 2.19874486e-01 1.07005620e+00
7.68870473e-01 -3.28383684e-01 5.29522955e-01 -6.22589707e-01
-4.42335643e-02 -5.39275050e-01 -4.92527574e-01 1.28929526e-01
-4.46437627e-01 9.85511113e-03 7.03622580e-01 -7.84244955e-01
-1.19898582e+00 2.25715086e-01 -1.16757326e-01 -7.33925700e-01
3.42660621e-02 3.14443350e-01 -8.05120096e-02 -4.11647052e-01
3.26511413e-01 1.24430820e-01 -2.46660650e-01 -2.86431551e-01
1.25732109e-01 2.63607949e-01 8.41456115e-01 -4.66899574e-01
8.89787257e-01 4.99378383e-01 7.47590885e-02 -6.61641598e-01
-6.63148642e-01 -4.48168874e-01 -6.03420556e-01 -3.93636167e-01
6.59136117e-01 -1.24900591e+00 -5.58200717e-01 5.11796832e-01
-1.06330609e+00 -8.36146414e-01 6.35673329e-02 5.54635644e-01
-8.23487878e-01 6.83965385e-01 -6.38161659e-01 -4.42275256e-01
-2.17589900e-01 -1.37492788e+00 9.72229362e-01 1.95667744e-01
-2.87127256e-01 -1.01248109e+00 4.24005613e-02 3.69057983e-01
4.88450319e-01 2.01052651e-01 6.83358788e-01 6.31132349e-02
-1.01619804e+00 1.76041231e-01 9.47730839e-02 3.26408982e-01
2.59263843e-01 2.30511829e-01 -6.43961906e-01 -5.47906458e-01
-3.35799396e-01 -2.21436247e-01 9.74962056e-01 5.39658546e-01
1.49453473e+00 -3.74890119e-01 -3.83236498e-01 1.04139221e+00
1.31254125e+00 2.58042991e-01 9.03737128e-01 6.72138155e-01
9.58842158e-01 2.70685051e-02 7.89306879e-01 6.50599718e-01
5.55873156e-01 9.70747828e-01 4.58601117e-01 1.10099487e-01
-4.04097378e-01 -2.75389850e-01 5.06372273e-01 8.31376851e-01
-5.08115232e-01 -2.01427653e-01 -6.57593846e-01 7.73652434e-01
-2.20288777e+00 -1.11116040e+00 6.38638716e-03 2.39325261e+00
8.69408965e-01 3.38013843e-02 2.78113008e-01 1.92940563e-01
5.09417117e-01 2.23118857e-01 -6.43247783e-01 1.98031273e-02
1.16394171e-02 9.49961543e-02 3.12295437e-01 4.97091919e-01
-1.29593909e+00 8.84078562e-01 6.92938805e+00 7.88780630e-01
-1.18684375e+00 -6.03817403e-02 6.60159588e-01 -1.82622895e-01
-2.63820291e-01 3.11303623e-02 -8.94424438e-01 6.80537164e-01
8.44699383e-01 -2.01588385e-02 5.21337926e-01 7.07308531e-01
4.41820174e-01 1.46364570e-01 -1.20404530e+00 1.07498491e+00
-4.23295377e-03 -1.69498563e+00 1.98924407e-01 -2.82855153e-01
1.17007482e+00 -1.99254930e-01 1.17189214e-01 2.48604044e-01
9.32957418e-03 -9.47751462e-01 7.89173543e-01 3.92286539e-01
8.58983099e-01 -5.92053115e-01 4.00996387e-01 1.71486557e-01
-1.57487500e+00 -1.35288045e-01 -3.55175853e-01 -1.24287814e-01
4.29656714e-01 1.77720368e-01 -7.10041821e-01 6.21969700e-01
7.60765314e-01 9.73672986e-01 -4.94378984e-01 1.26518548e+00
-3.45791489e-01 1.09428036e+00 -1.46136746e-01 5.00229716e-01
2.20298633e-01 -3.04394681e-03 4.63163048e-01 1.17968142e+00
5.33119440e-01 -1.59641042e-01 3.79481405e-01 4.99512434e-01
-4.88230288e-02 4.07722853e-02 -4.27607775e-01 4.44513738e-01
4.90595490e-01 9.68695819e-01 -3.33407640e-01 -4.79021460e-01
-9.58466113e-01 1.20372689e+00 1.33965760e-01 5.56546807e-01
-1.10430908e+00 1.20718725e-01 7.83835948e-01 1.33916378e-01
8.66363704e-01 -4.57335770e-01 5.46989217e-02 -1.44532621e+00
1.47191465e-01 -9.79606569e-01 3.73348892e-01 -5.84419072e-01
-9.76207495e-01 4.72323358e-01 8.47188607e-02 -1.72121644e+00
-8.43652844e-01 -3.61826956e-01 -7.06102848e-01 4.51065928e-01
-1.80935621e+00 -1.27106333e+00 -6.03330135e-01 8.68025422e-01
1.00534225e+00 -1.64365619e-01 4.85517442e-01 4.11185443e-01
-6.40662730e-01 8.20151389e-01 2.52109051e-01 1.61300659e-01
8.75121117e-01 -8.43818069e-01 6.92561865e-01 1.07293665e+00
1.06580310e-01 4.05694574e-01 4.70976919e-01 -5.13850391e-01
-1.34787464e+00 -1.50925314e+00 6.14194036e-01 -1.85379818e-01
2.97378540e-01 9.26957875e-02 -1.13822556e+00 8.59760523e-01
2.30080128e-01 4.93192285e-01 6.40513778e-01 -4.82259542e-01
-1.86667472e-01 -4.82087806e-02 -6.30877674e-01 8.51422548e-01
1.31225336e+00 -1.34038568e-01 -1.84180558e-01 8.81682411e-02
5.72166264e-01 -7.31332719e-01 -9.44576919e-01 6.97155476e-01
6.98779225e-01 -1.07508183e+00 1.44614267e+00 -6.87485158e-01
6.83135331e-01 -6.12365544e-01 1.03261530e-01 -1.12677085e+00
-6.45449042e-01 -1.05642009e+00 -7.06639469e-01 1.07956338e+00
1.09368466e-01 -2.53089756e-01 8.79856765e-01 5.97115278e-01
-6.59755766e-02 -7.38017142e-01 -8.33134949e-01 -1.04411900e+00
-1.30782932e-01 -2.61535108e-01 6.15408182e-01 6.83529317e-01
-3.52746010e-01 1.64577544e-01 -8.51210833e-01 7.82547817e-02
4.82508957e-01 8.93576592e-02 1.11781466e+00 -7.02242017e-01
-6.71144485e-01 -2.41016492e-01 -5.52354276e-01 -1.48343241e+00
3.49341393e-01 -5.55178583e-01 4.77481708e-02 -1.33790267e+00
1.31634682e-01 -2.05375120e-01 -2.50608891e-01 3.21250707e-01
-8.04184556e-01 5.20714104e-01 4.67881531e-01 5.50740004e-01
-6.74436629e-01 5.79780459e-01 1.31096411e+00 -9.33656767e-02
-3.35755795e-01 -5.41086644e-02 -6.60882965e-02 9.26522076e-01
6.63364768e-01 -2.48244628e-01 -7.21319437e-01 -7.73945630e-01
-2.32915819e-01 7.90106431e-02 3.50366861e-01 -1.48195887e+00
1.96644709e-01 -2.81354755e-01 5.41237772e-01 -5.87499321e-01
2.49056429e-01 -5.41826069e-01 4.35542583e-01 3.54075998e-01
-1.56649232e-01 1.93814754e-01 2.15804592e-01 7.30683506e-01
-2.78109759e-01 -1.03535205e-01 7.94364691e-01 4.09775451e-02
-1.20733380e+00 6.95174813e-01 -4.92277473e-01 1.18194118e-01
8.65806222e-01 -2.17796057e-01 6.24340102e-02 -6.08244359e-01
-3.12857300e-01 7.47232661e-02 6.60961926e-01 4.16033208e-01
7.57348239e-01 -1.64497364e+00 -7.56962478e-01 1.16566584e-01
3.67820524e-02 1.98647007e-01 1.92700744e-01 8.14134598e-01
-5.78214586e-01 1.28132626e-01 -2.06933036e-01 -8.09935749e-01
-1.19013977e+00 8.18162918e-01 -2.61262152e-02 -2.97826171e-01
-7.23804593e-01 6.43423915e-01 4.44835275e-01 2.03152552e-01
9.87797081e-02 -3.34624618e-01 -6.88943788e-02 -6.14634693e-01
7.88515270e-01 4.94861513e-01 -1.80938631e-01 -7.36707151e-01
-2.36157030e-01 6.68032825e-01 -5.54005094e-02 7.66257048e-02
1.08490932e+00 -3.05653006e-01 3.80784929e-01 2.55468726e-01
1.15749550e+00 -2.86379814e-01 -1.89801908e+00 -2.66223103e-01
-1.55238956e-01 -9.87733245e-01 -2.53319353e-01 -2.81054318e-01
-1.17827320e+00 7.33770072e-01 4.23163444e-01 -4.08795238e-01
1.34872091e+00 -4.71428096e-01 1.24759400e+00 1.55601397e-01
3.36945534e-01 -8.94381225e-01 1.59629643e-01 3.59292090e-01
5.80419898e-01 -1.35680437e+00 -6.53327182e-02 -6.09685779e-01
-5.49127102e-01 1.31096900e+00 6.85227394e-01 -3.77727263e-02
4.07059252e-01 9.07016695e-02 9.47874114e-02 6.25024140e-01
-8.84972036e-01 5.20391576e-02 5.51952600e-01 8.14738452e-01
5.16784549e-01 -5.09533346e-01 -3.20113361e-01 1.77258000e-01
3.51691157e-01 6.00266933e-01 3.32594812e-01 9.67507124e-01
-1.59926131e-01 -1.32091904e+00 -2.04692885e-01 9.62879062e-02
-4.21086639e-01 -5.73639479e-03 3.30626130e-01 6.73400283e-01
-4.44344804e-02 9.91481185e-01 -1.48279175e-01 -3.76908064e-01
7.61057734e-02 -1.49768844e-01 5.91348827e-01 -1.71217740e-01
-2.61366665e-01 1.29686251e-01 -1.18385412e-01 -9.63760018e-01
-9.18255687e-01 -6.19768560e-01 -1.01285398e+00 -2.50010580e-01
-1.33017767e-02 -1.72713086e-01 4.81368154e-02 8.88261616e-01
5.43542206e-01 4.25685674e-01 6.72775090e-01 -1.44385326e+00
-3.50185513e-01 -6.08130276e-01 7.40434006e-02 6.68461800e-01
4.49602664e-01 -8.17886114e-01 -1.66765191e-02 7.31751800e-01] | [10.701504707336426, -1.3455008268356323] |
c8317cb7-4113-42b1-a7b6-b504b850e887 | moving-beyond-word-lists-towards-abstractive | 2211.05599 | null | https://arxiv.org/abs/2211.05599v1 | https://arxiv.org/pdf/2211.05599v1.pdf | Moving beyond word lists: towards abstractive topic labels for human-like topics of scientific documents | Topic models represent groups of documents as a list of words (the topic labels). This work asks whether an alternative approach to topic labeling can be developed that is closer to a natural language description of a topic than a word list. To this end, we present an approach to generating human-like topic labels using abstractive multi-document summarization (MDS). We investigate our approach with an exploratory case study. We model topics in citation sentences in order to understand what further research needs to be done to fully operationalize MDS for topic labeling. Our case study shows that in addition to more human-like topics there are additional advantages to evaluation by using clustering and summarization measures instead of topic model measures. However, we find that there are several developments needed before we can design a well-powered study to evaluate MDS for topic modeling fully. Namely, improving cluster cohesion, improving the factuality and faithfulness of MDS, and increasing the number of documents that might be supported by MDS. We present a number of ideas on how these can be tackled and conclude with some thoughts on how topic modeling can also be used to improve MDS in general. | ['Domenic Rosati'] | 2022-10-28 | null | null | null | null | ['topic-models', 'document-summarization'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.60041824e-01 8.01480651e-01 -2.40743518e-01 -5.33726990e-01
-8.35732341e-01 -5.56249678e-01 1.02809179e+00 8.52715313e-01
-7.33026043e-02 5.27164042e-01 1.10632432e+00 -4.24957484e-01
-4.33035284e-01 -7.93981493e-01 -1.22302130e-01 -4.40775841e-01
1.25756919e-01 9.12506759e-01 3.36601466e-01 -8.15053433e-02
8.60571921e-01 6.49749413e-02 -1.57028532e+00 4.14895982e-01
9.53666866e-01 1.06032044e-01 2.11822629e-01 4.46500450e-01
-5.81761062e-01 6.27469599e-01 -1.24042344e+00 -1.10882998e-01
-4.35352743e-01 -5.91413140e-01 -1.21626198e+00 5.87043703e-01
2.38590047e-01 -1.16904438e-01 4.30629134e-01 6.37092829e-01
4.35756415e-01 2.96231300e-01 8.98761511e-01 -1.11849833e+00
-3.10187727e-01 9.76177275e-01 -4.14330244e-01 1.52823016e-01
6.32700682e-01 -2.19187483e-01 8.71748805e-01 -3.62240344e-01
8.92237306e-01 1.70829904e+00 4.19071078e-01 1.29274562e-01
-1.26920724e+00 -2.95006782e-01 2.72918999e-01 -1.58571631e-01
-1.09347260e+00 -3.19391787e-01 7.67111063e-01 -5.93784869e-01
8.73858094e-01 4.86666858e-01 6.21209383e-01 7.49831080e-01
1.51005462e-01 3.86418641e-01 9.95755672e-01 -6.45676017e-01
3.21998805e-01 5.70576549e-01 6.30124986e-01 1.67695940e-01
6.84974253e-01 -6.55507207e-01 -4.67418224e-01 -5.05851269e-01
2.41545692e-01 -4.69787866e-01 -1.13285176e-01 -2.22595781e-01
-1.42077982e+00 1.22549748e+00 -8.47320631e-02 8.97517920e-01
-4.57074732e-01 -3.69074978e-02 6.64214432e-01 -1.06272727e-01
1.02433133e+00 9.58181500e-01 4.42880914e-02 -2.86449105e-01
-1.49730337e+00 5.67891359e-01 1.00564945e+00 7.02764332e-01
5.62769294e-01 -1.63927272e-01 -2.75406361e-01 7.44763136e-01
4.46666718e-01 1.53612614e-01 5.32029986e-01 -1.26529419e+00
1.81000710e-01 6.56738520e-01 8.57537538e-02 -1.19928133e+00
-5.12946010e-01 -1.57740757e-01 -1.84669390e-01 -1.13792874e-01
2.79222475e-03 -1.53736547e-01 -4.96241599e-01 1.31328797e+00
5.94915310e-03 -2.44190007e-01 -1.39391236e-02 3.53522986e-01
8.82372618e-01 1.03111744e+00 2.35389456e-01 -5.26214600e-01
1.69587100e+00 -5.35533071e-01 -9.39480722e-01 -2.37013679e-02
1.08309126e+00 -9.50741053e-01 1.01058757e+00 4.60711151e-01
-1.26179087e+00 -4.25302833e-01 -1.03232348e+00 -1.45017922e-01
-3.47989827e-01 -2.58152131e-02 6.16517544e-01 9.18501675e-01
-1.22133207e+00 4.68171030e-01 -9.34122503e-01 -1.00782180e+00
-3.90805155e-02 1.55524552e-01 4.26479056e-02 2.35107645e-01
-9.83294368e-01 1.12409723e+00 7.66117990e-01 -7.00904787e-01
-2.33409390e-01 -5.31569064e-01 -7.45541394e-01 6.97623938e-02
3.24483246e-01 -7.50075459e-01 1.28549504e+00 -4.13015813e-01
-9.54615116e-01 8.07228506e-01 -4.53296363e-01 -3.15112770e-01
-1.00349002e-01 1.38014987e-01 -6.16057701e-02 4.86173004e-01
5.39163351e-01 8.17191303e-01 6.28936738e-02 -1.71850455e+00
-7.76663423e-01 -3.63554448e-01 1.93296839e-02 3.85867715e-01
-6.26367450e-01 5.50704718e-01 -1.02241181e-01 -6.98275864e-01
2.35880852e-01 -7.98491299e-01 -1.52951302e-02 -6.11897647e-01
-5.15923262e-01 -7.96165287e-01 8.23583543e-01 -5.07883787e-01
1.60919440e+00 -1.71772087e+00 -6.50267601e-02 1.77849770e-01
3.41697037e-01 -1.61674637e-02 2.86227137e-01 1.16105890e+00
5.67500815e-02 7.75380254e-01 -1.37732774e-01 -3.97439063e-01
2.15656340e-01 1.60752796e-02 -6.31822765e-01 2.53781945e-01
-1.40060466e-02 4.30121720e-01 -9.50623095e-01 -7.86570191e-01
1.43160939e-01 4.22913879e-01 -3.04220408e-01 -3.50392044e-01
-2.64334232e-01 -3.46216664e-04 -4.44035649e-01 1.51022688e-01
2.63288856e-01 -1.25856593e-01 1.10942513e-01 2.17999727e-01
-4.33328360e-01 8.65165710e-01 -1.27314723e+00 1.45235312e+00
-6.27377555e-02 1.02390194e+00 -3.89287680e-01 -9.48941529e-01
1.16313136e+00 5.38712740e-01 5.82701743e-01 1.66190732e-02
-2.99904495e-02 5.08384667e-02 -1.52125567e-01 -4.29225713e-01
1.15029275e+00 -4.25144404e-01 -1.00287430e-01 1.12904811e+00
-1.08062085e-02 -6.65571809e-01 6.05635285e-01 6.77541912e-01
6.15776896e-01 -2.96745956e-01 3.64418864e-01 -6.90950751e-01
8.98232460e-02 5.18401325e-01 2.39696056e-02 7.88177431e-01
3.12494606e-01 4.21844274e-01 7.37515271e-01 1.32489018e-02
-1.09679186e+00 -6.02992594e-01 -4.06136811e-01 1.00701284e+00
-1.57251582e-01 -1.03979647e+00 -9.56292331e-01 -3.49515915e-01
-4.28308040e-01 1.47367966e+00 -5.24216473e-01 3.02122217e-02
-3.53778958e-01 -9.71893728e-01 5.09823024e-01 3.12867582e-01
-1.53208605e-03 -8.21489453e-01 -9.79773641e-01 3.32182646e-01
-4.37365294e-01 -4.93667006e-01 -3.93250864e-03 6.89389359e-04
-1.17429698e+00 -7.47315228e-01 -8.07762384e-01 -6.13523424e-01
5.16130328e-01 5.48952103e-01 9.10173595e-01 1.94457937e-02
3.82319450e-01 6.45216703e-01 -6.77129388e-01 -7.97401011e-01
-7.69506395e-01 3.32439929e-01 -1.23456113e-01 -8.33538771e-01
6.70966387e-01 -3.28475118e-01 -3.10587376e-01 1.85671344e-01
-1.28256750e+00 -1.80707555e-02 1.72982112e-01 2.93722153e-01
-1.10749982e-01 4.13434535e-01 6.59113109e-01 -1.06404018e+00
1.34378922e+00 -5.84975779e-01 2.17979893e-01 2.15656415e-01
-7.57797658e-01 -1.87395569e-02 -1.66869685e-01 -2.45242953e-01
-1.21711457e+00 -6.93004787e-01 2.63435133e-02 5.92810392e-01
-3.16400737e-01 7.67736614e-01 1.87600493e-01 4.55685526e-01
9.05912876e-01 -8.51343423e-02 1.65780589e-01 -3.08250129e-01
5.41683614e-01 8.74920666e-01 1.17515028e-01 -6.77789330e-01
3.41116279e-01 5.96026361e-01 -3.31690758e-01 -1.10960281e+00
-4.91300523e-01 -8.11386585e-01 -5.28911173e-01 -2.94674873e-01
6.56708658e-01 -8.03059161e-01 -2.18138441e-01 -2.46770561e-01
-1.22624326e+00 -1.34903565e-01 -5.27625799e-01 5.86063802e-01
-7.51758158e-01 6.08434498e-01 -2.85006613e-01 -7.54356623e-01
-2.15690255e-01 -8.98599029e-01 1.28190923e+00 -4.53224266e-03
-1.23557067e+00 -1.32957530e+00 2.68969297e-01 3.32642913e-01
2.94954628e-01 3.08991671e-01 1.15712082e+00 -1.09304631e+00
2.23405603e-02 -1.01592697e-01 1.25647396e-01 -1.49651811e-01
2.65221208e-01 3.86016071e-01 -6.70740366e-01 -3.13843369e-01
5.46047926e-01 -5.85837141e-02 7.99262106e-01 5.86172581e-01
4.45502192e-01 -4.73539233e-01 -8.09891820e-01 -3.31486523e-01
1.15456438e+00 5.00794888e-01 6.43124819e-01 5.93566477e-01
1.73897937e-01 1.29400182e+00 7.34918237e-01 3.88493031e-01
7.24186361e-01 5.82907319e-01 -2.73486257e-01 -1.44464867e-02
-1.86461672e-01 3.46795842e-02 1.37231067e-01 1.01653874e+00
1.20965235e-01 -5.52037835e-01 -1.28096187e+00 7.65973210e-01
-1.60041070e+00 -1.17136431e+00 -5.30668676e-01 1.73668540e+00
7.13403046e-01 3.59358221e-01 4.68212634e-01 3.42758507e-01
7.49355853e-01 2.53172636e-01 1.36562139e-01 -6.03393674e-01
3.18432576e-03 -2.46597633e-01 -1.25546500e-01 6.24404728e-01
-7.29645252e-01 7.19053447e-01 6.80996799e+00 6.23554409e-01
-8.19021463e-01 3.03610843e-02 6.81850672e-01 9.86567810e-02
-8.23816717e-01 4.59457457e-01 -8.84094656e-01 3.74768019e-01
1.06522477e+00 -6.40892684e-01 -5.40166080e-01 7.16501415e-01
6.72776818e-01 -5.45739233e-01 -7.67810941e-01 1.51302487e-01
4.40179497e-01 -1.24796951e+00 1.85790524e-01 3.35930586e-01
7.18736887e-01 -6.05529904e-01 -2.26788253e-01 1.59212559e-01
3.05833220e-01 -6.16996109e-01 5.08344293e-01 4.19762641e-01
2.54200578e-01 -5.44790983e-01 8.13960493e-01 2.92099029e-01
-6.56726420e-01 2.43512601e-01 -5.59947968e-01 -1.80654138e-01
5.08611858e-01 5.94300091e-01 -1.06114757e+00 6.43158555e-01
3.69887739e-01 4.28656131e-01 -6.88988507e-01 1.22357762e+00
7.70079568e-02 8.14552665e-01 -1.75964609e-01 -3.13973010e-01
3.34186941e-01 -3.38340411e-03 6.91021085e-01 1.41233695e+00
4.13250864e-01 6.36013299e-02 7.29099885e-02 6.05023026e-01
6.10874653e-01 3.79365325e-01 -7.09813595e-01 -1.64681431e-02
6.39939070e-01 9.56485868e-01 -1.51101303e+00 -6.86711907e-01
-1.72680184e-01 4.17224437e-01 -3.75338316e-01 1.14806652e-01
-3.15722734e-01 -3.80688578e-01 2.28546038e-02 5.59788346e-01
-1.48063406e-01 -2.51324296e-01 -6.62864447e-01 -7.50390232e-01
-4.05793875e-01 -8.23330700e-01 2.62994587e-01 -9.53625500e-01
-9.30588663e-01 4.98614997e-01 1.00890446e+00 -7.87360311e-01
-5.78223586e-01 3.02250385e-02 -1.06602132e+00 5.41900754e-01
-7.76851833e-01 -8.73779535e-01 -8.17972198e-02 -1.49038687e-01
8.01469922e-01 1.96504503e-01 7.93839157e-01 -3.51248622e-01
-1.01931110e-01 -1.25103056e-01 8.52942765e-02 -5.43039918e-01
8.91186118e-01 -1.42953622e+00 4.18598711e-01 6.01425648e-01
5.85205853e-02 9.93755817e-01 1.19901288e+00 -1.02774167e+00
-4.10049409e-01 -6.26321197e-01 1.38306344e+00 -5.45350909e-01
5.91792524e-01 -2.08003670e-01 -1.03129292e+00 6.34959757e-01
8.86853755e-01 -1.50532639e+00 1.21669853e+00 5.17382205e-01
3.19468379e-01 4.02981013e-01 -8.74797583e-01 5.15849590e-01
4.97172117e-01 -7.10268989e-02 -1.28976393e+00 6.50949836e-01
8.95764709e-01 8.42875615e-02 -1.02174938e+00 7.85685107e-02
5.43993637e-02 -7.14705765e-01 6.34405851e-01 -2.33594999e-01
4.01767671e-01 -3.15776199e-01 1.39299184e-01 -1.36833882e+00
-2.47352824e-01 -4.75483537e-01 3.78347993e-01 1.94828343e+00
3.44595402e-01 -4.01743352e-01 6.59033000e-01 6.12816989e-01
-3.34735662e-01 -5.14255404e-01 -5.71928859e-01 -5.38691938e-01
5.82156420e-01 -4.54514295e-01 4.76788878e-01 1.05163431e+00
7.71764576e-01 5.56153476e-01 8.38322118e-02 -3.99236292e-01
2.54881650e-01 -1.48812428e-01 7.04887927e-01 -1.69543195e+00
4.26776618e-01 -5.70861816e-01 -3.14235747e-01 -5.92307150e-01
-5.48684187e-02 -6.64422274e-01 -5.79652786e-02 -2.34324980e+00
3.98371398e-01 -1.71371073e-01 4.31066215e-01 1.24955244e-01
-1.49799138e-01 -2.46946633e-01 3.82307649e-01 4.50330734e-01
-7.17374742e-01 2.47193307e-01 8.56470644e-01 1.32525533e-01
-5.18526077e-01 -9.17160586e-02 -1.35954154e+00 7.66112506e-01
9.69548106e-01 -7.75473058e-01 -7.11854100e-01 -8.07494298e-02
2.49825567e-01 -4.21466827e-02 -2.98871547e-02 -8.99114013e-01
5.70639670e-01 -6.09413497e-02 3.93524654e-02 -9.70168173e-01
1.66225955e-01 -2.18790919e-01 1.82774588e-01 4.14310008e-01
-6.91791177e-01 2.43566930e-01 3.04522842e-01 2.86355764e-01
-2.84122795e-01 -7.30617285e-01 1.75784409e-01 -3.98057163e-01
-2.94428200e-01 -5.61693549e-01 -1.05355620e+00 -1.45216420e-01
1.05536115e+00 -4.26173538e-01 -5.29176116e-01 -8.49865198e-01
-6.19632363e-01 3.57326537e-01 8.32917929e-01 2.85065472e-01
3.48288625e-01 -9.74824548e-01 -8.34650457e-01 -4.81581509e-01
-1.26702497e-02 -9.15169194e-02 -1.64666437e-02 4.89929885e-01
-5.06341577e-01 1.08091962e+00 -2.01876927e-03 -4.92030919e-01
-1.13313949e+00 6.32086456e-01 -3.34048629e-01 -3.04883897e-01
-5.33077657e-01 3.72403711e-01 -5.70019567e-03 -6.38595074e-02
2.24317580e-01 -2.25551814e-01 -6.67578161e-01 7.57233024e-01
5.19184053e-01 7.20407307e-01 -1.22508861e-01 -7.11879194e-01
-1.80917770e-01 4.86501068e-01 -1.36982068e-01 -7.40714371e-01
1.25081396e+00 -4.40548271e-01 -4.42630976e-01 8.45478535e-01
7.91209996e-01 2.87648402e-02 -7.36741543e-01 1.99734256e-01
6.71893835e-01 -1.36959672e-01 7.84749463e-02 -6.95887208e-01
-9.70819890e-02 7.41289318e-01 5.87242730e-02 1.07271075e+00
9.25113440e-01 4.61923063e-01 2.75220633e-01 2.59735942e-01
-5.69278933e-02 -1.22460544e+00 2.02564567e-01 4.97093275e-02
1.16082478e+00 -8.34909379e-01 4.99083489e-01 -4.51971471e-01
-8.99759471e-01 1.29024231e+00 1.69371903e-01 1.09711498e-01
6.21598959e-01 -1.03022411e-01 -5.55521026e-02 -7.84151673e-01
-8.00145388e-01 -1.63804576e-01 3.23264211e-01 6.04795575e-01
8.67455125e-01 -4.94554043e-02 -1.02289915e+00 3.35880905e-01
-6.87550306e-01 -2.81346917e-01 1.02710807e+00 9.36848581e-01
-1.04858136e+00 -1.18127751e+00 -7.90334582e-01 7.32597768e-01
-5.98786831e-01 1.16404325e-01 -7.94920087e-01 1.02747846e+00
-3.62454611e-03 1.53894532e+00 3.26714218e-01 -1.42558888e-01
4.24556397e-02 2.55608976e-01 4.95136762e-03 -1.29597688e+00
-5.26843250e-01 4.88671601e-01 4.57755148e-01 1.17440417e-01
-8.75596583e-01 -1.13403225e+00 -1.09998739e+00 -3.31546992e-01
-5.35396039e-01 7.39415109e-01 9.63806510e-01 9.04184520e-01
2.54421443e-01 5.51468790e-01 1.30089551e-01 -6.59352005e-01
-2.34598443e-02 -1.39825356e+00 -3.92267555e-01 -9.27431975e-03
-1.96097031e-01 -5.65302432e-01 -2.54489303e-01 1.08458839e-01] | [10.456543922424316, 7.219030380249023] |
1a6ba7f2-4f17-4885-999e-b1e687f8bdd7 | another-dead-end-for-morphological-tags | 2305.15119 | null | https://arxiv.org/abs/2305.15119v1 | https://arxiv.org/pdf/2305.15119v1.pdf | Another Dead End for Morphological Tags? Perturbed Inputs and Parsing | The usefulness of part-of-speech tags for parsing has been heavily questioned due to the success of word-contextualized parsers. Yet, most studies are limited to coarse-grained tags and high quality written content; while we know little about their influence when it comes to models in production that face lexical errors. We expand these setups and design an adversarial attack to verify if the use of morphological information by parsers: (i) contributes to error propagation or (ii) if on the other hand it can play a role to correct mistakes that word-only neural parsers make. The results on 14 diverse UD treebanks show that under such attacks, for transition- and graph-based models their use contributes to degrade the performance even faster, while for the (lower-performing) sequence labeling parsers they are helpful. We also show that if morphological tags were utopically robust against lexical perturbations, they would be able to correct parsing mistakes. | ['David Vilares', 'Alberto Muñoz-Ortiz'] | 2023-05-24 | null | null | null | null | ['adversarial-attack'] | ['adversarial'] | [ 3.08672160e-01 5.93467772e-01 2.35967934e-01 -1.74324781e-01
-1.03001010e+00 -1.05896854e+00 3.91419351e-01 3.61624569e-01
-6.16447628e-01 7.22328782e-01 3.10334712e-01 -1.04947710e+00
4.47287977e-01 -8.08056653e-01 -9.60775256e-01 -5.97276330e-01
8.76460969e-02 4.17850405e-01 6.76537514e-01 -3.00992578e-01
8.74194801e-02 3.01538378e-01 -1.05388236e+00 1.60698116e-01
6.04771614e-01 1.93266466e-01 1.75721571e-01 7.55789638e-01
-2.92669624e-01 7.19793975e-01 -8.37853432e-01 -8.96309912e-01
2.88554102e-01 -3.69123816e-01 -1.00013673e+00 -2.77675420e-01
2.32415259e-01 -1.03328198e-01 3.62073407e-02 1.43962920e+00
5.06592035e-01 -1.67518094e-01 4.19434994e-01 -6.41367078e-01
-4.43296045e-01 1.15258372e+00 -1.22300338e-03 3.82554412e-01
3.12411457e-01 1.21934153e-01 1.18921745e+00 -2.55235076e-01
9.61547136e-01 1.40290773e+00 7.69340038e-01 8.25643361e-01
-1.34650791e+00 -4.02823359e-01 2.02798441e-01 -3.17835808e-01
-6.80026948e-01 -5.27019441e-01 5.18017471e-01 -3.04027766e-01
1.17327964e+00 9.64403376e-02 1.23407893e-01 1.37728071e+00
3.33688378e-01 4.97401416e-01 1.21841013e+00 -7.33309269e-01
2.21837416e-01 2.65010670e-02 1.64843664e-01 7.76414037e-01
5.40777683e-01 2.26378620e-01 -2.73550510e-01 -2.44382679e-01
3.78777862e-01 -8.76810014e-01 -5.24944961e-02 8.42920914e-02
-7.35151470e-01 9.89114106e-01 -8.50475281e-02 7.77072728e-01
-1.51938140e-01 1.06487975e-01 6.58547223e-01 5.62521100e-01
3.87172312e-01 8.22979391e-01 -8.13037157e-01 -4.09859568e-01
-6.29855573e-01 5.37770381e-03 8.28016579e-01 6.85558856e-01
5.79399884e-01 2.86699444e-01 1.46514073e-01 6.65137649e-01
1.20202832e-01 3.63304377e-01 6.16696000e-01 -7.32133389e-01
6.77516937e-01 3.48934591e-01 3.46037000e-02 -8.24759483e-01
-5.55946350e-01 -1.37666881e-01 -7.94715360e-02 1.66560978e-01
1.04378951e+00 -4.45951343e-01 -9.52914000e-01 2.09971142e+00
1.82750896e-01 -3.56863290e-01 3.57360393e-02 5.72378218e-01
4.74348307e-01 4.24379945e-01 5.25441885e-01 -1.35318428e-01
1.46953976e+00 -3.09936404e-01 -7.06471503e-01 -6.42877281e-01
1.19964302e+00 -1.05221593e+00 1.20224273e+00 1.33612111e-01
-1.15179658e+00 -2.97853440e-01 -1.05786383e+00 5.80241531e-02
-3.41790766e-01 -3.70261729e-01 5.49114943e-01 1.37149549e+00
-9.75665212e-01 8.11353326e-01 -1.13167453e+00 -4.83229548e-01
8.22072849e-03 4.35418189e-02 -4.64708298e-01 1.03220388e-01
-1.39014447e+00 1.29693699e+00 2.05462262e-01 -1.11109108e-01
-6.16235435e-01 -2.34693915e-01 -9.28081989e-01 -8.63898844e-02
2.98520118e-01 -2.50185907e-01 1.35512304e+00 -1.00848317e+00
-1.38641417e+00 9.30002809e-01 -2.03361847e-02 -4.27519113e-01
5.88979244e-01 -6.95678964e-02 -1.86548978e-01 2.68694311e-02
2.63384342e-01 1.97754964e-01 5.46249092e-01 -1.16942692e+00
-6.53036654e-01 -3.91449600e-01 1.19029574e-01 -1.06640816e-01
-1.71095163e-01 5.80053210e-01 8.03942606e-02 -6.97955549e-01
1.59120128e-01 -1.12364066e+00 -2.33471930e-01 -6.48915052e-01
-6.20542765e-01 -1.27124816e-01 4.40909922e-01 -1.06044364e+00
1.17252755e+00 -2.00679493e+00 -1.34126619e-01 4.69054207e-02
-3.06678534e-01 5.87981462e-01 -2.24886328e-01 6.59662127e-01
-1.65819049e-01 8.56179357e-01 -1.75127655e-01 -3.19278747e-01
1.07917920e-01 5.77826262e-01 -2.36105934e-01 3.53069544e-01
2.96172321e-01 7.62392402e-01 -6.96359456e-01 -2.36921266e-01
-1.05403453e-01 1.92985997e-01 -4.70339626e-01 -8.73775855e-02
-1.56783819e-01 2.71371484e-01 -4.85209048e-01 4.45029855e-01
1.66269183e-01 4.03987646e-01 5.62113643e-01 2.76904076e-01
-3.18509340e-03 9.71680403e-01 -9.15970922e-01 1.13251293e+00
-4.81962621e-01 3.36124659e-01 1.83822930e-01 -8.21151555e-01
5.93185902e-01 5.07959366e-01 -2.42686540e-01 -7.89489746e-01
1.78025693e-01 4.70279187e-01 5.75902760e-01 -2.74391383e-01
5.22775054e-01 -3.93666238e-01 -3.35934997e-01 4.95870888e-01
-5.83988316e-02 7.13985367e-03 6.73306212e-02 3.10329676e-01
1.69275403e+00 2.65343606e-01 2.04188496e-01 -2.61619657e-01
1.48850366e-01 9.77656245e-02 9.60387707e-01 8.20810795e-01
-2.70592660e-01 5.30773640e-01 8.94458950e-01 -1.76884025e-01
-1.18362057e+00 -8.05427015e-01 -6.91112056e-02 1.26437807e+00
-3.15710485e-01 -3.48495007e-01 -1.07238805e+00 -1.01331031e+00
-1.93532825e-01 1.14579499e+00 -4.64243352e-01 -3.15202922e-01
-7.89497197e-01 -8.09082210e-01 1.16897905e+00 5.33411205e-01
1.49550363e-02 -1.29229248e+00 -5.73968112e-01 5.73994219e-01
-1.31782383e-01 -1.14020419e+00 -2.91825354e-01 6.59392774e-01
-8.93364966e-01 -1.19139028e+00 -1.70320883e-01 -7.45696723e-01
5.10748982e-01 -3.39403659e-01 1.12907040e+00 3.64664614e-01
1.78729698e-01 1.75302744e-01 -7.78243423e-01 -4.51233447e-01
-1.35068595e+00 2.17468768e-01 -6.71453476e-02 -4.55430627e-01
2.80668885e-01 -4.78800982e-01 3.14872563e-01 1.96419701e-01
-8.38818610e-01 -6.09352052e-01 3.93185288e-01 9.08361793e-01
4.87316633e-03 -6.03106841e-02 4.35972184e-01 -1.64755750e+00
6.53805196e-01 -9.30071846e-02 -6.00832701e-01 1.31146699e-01
-5.42864740e-01 3.24261576e-01 7.00926960e-01 -2.51625091e-01
-1.13013434e+00 -1.01826631e-03 -7.17113554e-01 1.87654540e-01
-4.82636273e-01 3.16994041e-01 -4.64753240e-01 -1.02378568e-02
7.51285315e-01 -8.90650377e-02 -2.71775931e-01 -7.36557007e-01
1.66291624e-01 3.59884113e-01 1.84463203e-01 -8.97779047e-01
1.04090297e+00 -8.23782906e-02 -1.62983879e-01 -8.01121473e-01
-4.84786123e-01 -7.42935389e-02 -4.65332121e-01 2.58536458e-01
1.06120539e+00 -6.06785297e-01 -1.93296313e-01 5.16338050e-01
-1.37687337e+00 -5.15221596e-01 -1.28224477e-01 1.93747923e-01
-3.64676267e-01 6.58319414e-01 -1.13586140e+00 -8.33012581e-01
-1.16783299e-01 -1.35933828e+00 5.85375130e-01 -1.78786799e-01
-3.41500998e-01 -1.06688547e+00 -2.28678472e-02 4.23849851e-01
1.27873957e-01 2.49284767e-02 1.50119781e+00 -1.28581536e+00
-1.56500787e-01 -1.85755864e-01 2.34903067e-01 4.90760207e-01
-5.72142266e-02 1.57475039e-01 -1.13552511e+00 -6.34738011e-03
-2.45121360e-01 -1.61933467e-01 5.89403629e-01 4.06626649e-02
2.03861624e-01 -4.60209757e-01 -1.25175238e-01 1.65299937e-01
1.27336311e+00 2.74660200e-01 7.33239412e-01 5.12710631e-01
5.75682878e-01 7.88594007e-01 4.66740161e-01 -2.01067477e-01
7.71563798e-02 6.79353774e-01 4.06298280e-01 2.54370421e-01
-3.11036594e-02 -5.28547704e-01 7.03863502e-01 8.35107744e-01
3.75329726e-03 -3.98346961e-01 -9.85463202e-01 7.29609907e-01
-1.49860370e+00 -8.58552575e-01 -3.77954900e-01 2.01844311e+00
1.00953352e+00 6.87596202e-01 -3.59683521e-02 2.15843901e-01
8.91874611e-01 1.76193088e-01 2.43239645e-02 -1.01092148e+00
-1.16601825e-01 5.08765578e-01 6.96249425e-01 6.54947221e-01
-8.59313786e-01 1.43743587e+00 6.22405005e+00 5.59123874e-01
-7.14051306e-01 4.10365582e-01 4.51714873e-01 4.08439338e-01
-4.26892936e-01 1.26062199e-01 -7.97502637e-01 5.23071051e-01
1.16065788e+00 3.45242471e-01 2.06091747e-01 7.61048675e-01
2.10573170e-02 -2.37272307e-01 -9.14918482e-01 6.57436997e-02
-4.16650534e-01 -7.54089177e-01 -3.65178853e-01 9.54956263e-02
3.87917429e-01 1.13736875e-02 -2.67415255e-01 4.28291351e-01
1.05250442e+00 -7.90247262e-01 1.11205173e+00 -1.45536304e-01
2.57073045e-01 -7.60586977e-01 1.14891148e+00 5.62472939e-01
-7.01167464e-01 3.40306103e-01 -6.12519205e-01 -1.96232662e-01
2.81681031e-01 5.10861278e-01 -1.02398837e+00 2.54102588e-01
4.09089297e-01 -1.32583499e-01 -6.36461854e-01 5.53648293e-01
-8.48471522e-01 1.42921960e+00 -3.61332953e-01 -7.74958655e-02
3.82982999e-01 1.98612064e-01 6.90058172e-01 1.31674850e+00
9.63955969e-02 -3.12571339e-02 6.59825280e-02 3.10437202e-01
-2.79731695e-02 1.99306294e-01 -5.83491862e-01 -3.17964226e-01
6.74475610e-01 1.00879920e+00 -9.38180029e-01 5.09129018e-02
-4.63123709e-01 9.35355663e-01 5.12216151e-01 3.35388035e-02
-4.57024485e-01 -1.92235500e-01 8.71453285e-01 1.39333487e-01
4.01383996e-01 -1.88532129e-01 -3.92941147e-01 -8.80283594e-01
-2.60574482e-02 -1.14064479e+00 3.62773120e-01 -5.35732150e-01
-1.07466435e+00 5.42313635e-01 -5.50942540e-01 -3.07724774e-01
-4.01031852e-01 -7.53818274e-01 -6.46753311e-01 9.31060493e-01
-1.29730594e+00 -8.99348974e-01 6.26068711e-01 2.06106752e-01
2.81057596e-01 -9.46879983e-02 1.16337788e+00 1.19591482e-01
-5.35385072e-01 7.82789111e-01 -1.55430868e-01 6.32080019e-01
6.24082446e-01 -1.39456296e+00 9.51366842e-01 1.47363234e+00
4.47784662e-01 7.62223363e-01 9.48080957e-01 -8.61062288e-01
-9.90869820e-01 -8.85514259e-01 1.36138856e+00 -8.45017254e-01
6.95984960e-01 -3.97608846e-01 -9.95996118e-01 9.03823256e-01
1.96040384e-02 -1.19270399e-01 3.68385911e-01 3.45358014e-01
-6.36093080e-01 4.87501562e-01 -1.29222298e+00 5.15925348e-01
1.12864244e+00 -5.33541203e-01 -8.52159977e-01 1.04951821e-01
7.96505213e-01 -2.61039436e-01 -7.01345325e-01 6.20488040e-02
2.14236468e-01 -1.12540638e+00 5.20528972e-01 -1.06049025e+00
3.88507575e-01 9.07605290e-02 -2.58247465e-01 -1.55066693e+00
-3.61978501e-01 -5.80350876e-01 6.18995070e-01 1.73134351e+00
9.22076762e-01 -8.32025647e-01 6.60163105e-01 5.95388770e-01
-4.02835429e-01 -2.12201223e-01 -1.17881465e+00 -8.54932189e-01
6.28088057e-01 -6.32947505e-01 4.41121668e-01 9.97526467e-01
-1.84126161e-02 3.45332205e-01 -1.04242809e-01 3.79890949e-01
6.95365518e-02 -6.17893457e-01 4.40489471e-01 -1.10159302e+00
-4.39774752e-01 -1.91083759e-01 -5.86615443e-01 -3.49665105e-01
2.87514567e-01 -7.55935490e-01 3.67478132e-01 -1.24805140e+00
-3.87089163e-01 -7.58296430e-01 7.83410966e-02 7.39635766e-01
-4.68739420e-01 1.89125970e-01 4.53234851e-01 -7.35779032e-02
-9.70979035e-02 -1.82509691e-01 7.57725835e-01 2.75074601e-01
1.46412000e-01 -4.44114348e-03 -6.70629025e-01 9.34389591e-01
7.51687825e-01 -9.98553395e-01 3.44722755e-02 -4.48729187e-01
4.95934308e-01 2.42964029e-02 7.48916417e-02 -8.07568431e-01
-1.65854827e-01 1.57084540e-01 -9.63946730e-02 2.80894458e-01
-1.16462004e-03 -7.12176025e-01 -1.18241094e-01 6.06273651e-01
-2.84738243e-01 3.78705740e-01 3.13463390e-01 4.15453970e-01
7.88555369e-02 -9.03198302e-01 7.91658580e-01 -4.76975411e-01
-5.84133387e-01 -2.15248808e-01 -8.88753295e-01 5.41439295e-01
5.00904143e-01 -9.23445448e-02 -1.70426071e-01 -2.63145834e-01
-7.49071062e-01 -5.07739842e-01 7.66849697e-01 2.14687735e-01
-2.43493780e-01 -7.07524300e-01 -4.94759589e-01 2.02307120e-01
-7.94687644e-02 -3.93860728e-01 -2.17535257e-01 3.30636472e-01
-6.29231513e-01 2.26232961e-01 5.08401133e-02 -5.02962954e-02
-1.37231994e+00 4.15212512e-01 2.84270585e-01 -5.31891584e-01
-4.53007787e-01 1.16561341e+00 -1.49383441e-01 -5.60489833e-01
1.20000944e-01 -2.81842023e-01 1.54578462e-01 1.42165765e-01
2.08250359e-01 1.82861388e-01 5.72721839e-01 -7.16291070e-01
-4.93453234e-01 1.76071912e-01 -2.00841978e-01 -1.85523525e-01
1.16766167e+00 1.02132037e-01 2.10032508e-01 2.25803107e-01
6.42497063e-01 6.86191022e-01 -9.96635437e-01 -6.12214729e-02
3.83188605e-01 -1.23732939e-01 -1.13899097e-01 -9.25184906e-01
-8.55437934e-01 1.07329714e+00 6.81167170e-02 5.85918486e-01
5.45932293e-01 -1.08035088e-01 8.31583083e-01 2.55647272e-01
7.50356376e-01 -1.11193705e+00 -4.08848554e-01 7.34729350e-01
1.92760333e-01 -9.24235284e-01 -4.17653024e-01 -6.12017870e-01
-6.23624861e-01 9.47768211e-01 2.36164987e-01 3.10770925e-02
2.95246780e-01 5.46384215e-01 4.31789249e-01 1.92592233e-01
-6.36787236e-01 -2.81235397e-01 -2.52986908e-01 7.89561689e-01
6.39510334e-01 1.54046044e-01 -5.66960990e-01 6.92650378e-01
-4.96404916e-01 -6.99582219e-01 8.55761647e-01 8.66515398e-01
-5.79562426e-01 -1.57753086e+00 -3.49356502e-01 2.40669429e-01
-1.11924291e+00 -3.82961422e-01 -5.60145080e-01 1.02876747e+00
9.63209495e-02 1.22344327e+00 -1.64996833e-01 -2.76864171e-01
3.15101445e-01 7.44911253e-01 4.45610464e-01 -1.07935393e+00
-1.07412839e+00 -1.32666007e-01 9.15560603e-01 -4.79138583e-01
-6.11874424e-02 -1.06921196e+00 -1.20717204e+00 -3.80854815e-01
-4.24351782e-01 1.48156837e-01 7.00876653e-01 1.09534955e+00
-1.10499345e-01 3.65263611e-01 9.28909555e-02 -2.35130191e-01
-9.03982222e-01 -1.08386123e+00 -6.53310359e-01 5.14637470e-01
-1.30332872e-01 -3.47965956e-01 -4.98904526e-01 -4.88662682e-02] | [10.497384071350098, 9.651650428771973] |
10b6621a-f0f6-4bc2-9499-d378e8f7d3b7 | sketch-specific-data-augmentation-for | 1910.06038 | null | https://arxiv.org/abs/1910.06038v2 | https://arxiv.org/pdf/1910.06038v2.pdf | Sketch-Specific Data Augmentation for Freehand Sketch Recognition | Sketch recognition remains a significant challenge due to the limited training data and the substantial intra-class variance of freehand sketches for the same object. Conventional methods for this task often rely on the availability of the temporal order of sketch strokes, additional cues acquired from different modalities and supervised augmentation of sketch datasets with real images, which also limit the applicability and feasibility of these methods in real scenarios. In this paper, we propose a novel sketch-specific data augmentation (SSDA) method that leverages the quantity and quality of the sketches automatically. From the aspect of quantity, we introduce a Bezier pivot based deformation (BPD) strategy to enrich the training data. Towards quality improvement, we present a mean stroke reconstruction (MSR) approach to generate a set of novel types of sketches with smaller intra-class variances. Both of these solutions are unrestricted from any multi-source data and temporal cues of sketches. Furthermore, we show that some recent deep convolutional neural network models that are trained on generic classes of real images can be better choices than most of the elaborate architectures that are designed explicitly for sketch recognition. As SSDA can be integrated with any convolutional neural networks, it has a distinct advantage over the existing methods. Our extensive experimental evaluations demonstrate that the proposed method achieves the state-of-the-art results (84.27%) on the TU-Berlin dataset, outperforming the human performance by a remarkable 11.17% increase. Finally, more experiments show the practical value of our approach for the task of sketch-based image retrieval. | ['Sicheng Zhao', 'Xiaoshuai Sun', 'Ying Zheng', 'Fatih Porikli', 'Hongxun Yao', 'Shengping Zhang'] | 2019-10-14 | null | null | null | null | ['sketch-based-image-retrieval', 'sketch-recognition'] | ['computer-vision', 'computer-vision'] | [ 2.43243709e-01 -3.49825174e-01 -1.50918603e-01 -2.23956838e-01
-6.11588717e-01 -5.93710661e-01 1.00117362e+00 -3.48769516e-01
-3.36076826e-01 4.35025394e-01 -4.96650152e-02 7.86603093e-02
-2.68807739e-01 -7.97110379e-01 -6.64261937e-01 -6.21819556e-01
1.77022457e-01 4.25824702e-01 1.68020502e-01 -4.33354080e-01
2.89161146e-01 1.19417715e+00 -1.70283270e+00 2.89172709e-01
6.27715945e-01 1.28012490e+00 1.72534317e-01 3.78163278e-01
-4.82538402e-01 2.30206102e-01 -4.13364351e-01 -6.16639733e-01
5.83004236e-01 -2.31540878e-03 -2.57141799e-01 8.59995708e-02
1.00238681e+00 -8.06486368e-01 -5.43427408e-01 5.84131300e-01
6.51040852e-01 5.81682548e-02 5.00608921e-01 -1.26468873e+00
-6.87332034e-01 1.65679753e-01 -3.68355393e-01 -3.74448746e-01
1.47661060e-01 8.39840770e-02 8.99891734e-01 -1.42580509e+00
9.71370041e-01 1.07230043e+00 5.95685661e-01 6.21417999e-01
-1.15849376e+00 -8.90652299e-01 1.96872875e-01 -8.97203013e-02
-1.35028040e+00 -4.28235590e-01 1.31528842e+00 -3.14803213e-01
4.49957788e-01 3.85274917e-01 6.33424580e-01 1.47336328e+00
-6.80671275e-01 1.12763619e+00 9.48643684e-01 -3.44414055e-01
5.84698804e-02 7.63309246e-04 -2.90579349e-01 6.43410981e-01
7.77799264e-02 2.73294032e-01 -6.23748600e-01 -9.17328745e-02
1.28645587e+00 3.89276594e-01 -1.73789158e-01 -9.18962896e-01
-1.41077089e+00 5.71309030e-01 4.33796257e-01 4.09271300e-01
-4.00657862e-01 3.25543642e-01 4.12939519e-01 1.97014540e-01
2.72188067e-01 4.50313777e-01 -2.76033700e-01 1.07891364e-02
-1.32993424e+00 4.86626804e-01 5.94728649e-01 1.16083264e+00
5.20314276e-01 1.95915714e-01 -3.34862560e-01 9.06458974e-01
5.49256504e-02 7.42280364e-01 1.81440517e-01 -7.99531758e-01
4.63961452e-01 6.44186080e-01 2.57777363e-01 -1.12351859e+00
1.14623062e-01 -4.27272767e-01 -1.09157193e+00 4.25278187e-01
6.16268694e-01 4.84442145e-01 -9.81586814e-01 1.58545303e+00
1.34306058e-01 6.99874982e-02 -2.73651809e-01 1.00553107e+00
8.67182255e-01 3.66626292e-01 -1.05307199e-01 2.49124274e-01
1.14439058e+00 -9.10743058e-01 -7.76174068e-01 1.65548161e-01
-4.15409207e-02 -9.80008662e-01 1.25247800e+00 5.60729861e-01
-1.00561774e+00 -8.59407365e-01 -1.17432249e+00 -7.44432881e-02
-6.28257036e-01 6.84290767e-01 7.34930158e-01 6.32902324e-01
-8.75510812e-01 8.77231658e-01 -4.58192140e-01 -2.63271183e-01
6.46713257e-01 2.59828627e-01 -6.03092015e-01 -2.22498730e-01
-8.85171175e-01 5.87128460e-01 4.43871766e-02 2.27945343e-01
-5.97952068e-01 -8.05849254e-01 -5.69857776e-01 2.18173280e-01
4.02411819e-01 -4.66720104e-01 6.90787613e-01 -9.15583730e-01
-1.68697190e+00 6.90531313e-01 2.07602143e-01 -1.92582645e-02
1.01388299e+00 -3.61849248e-01 -7.57620782e-02 2.95132875e-01
-3.37176949e-01 8.47243607e-01 1.24363256e+00 -1.49200416e+00
-2.34047815e-01 -1.58420414e-01 -1.66514125e-02 -9.02024806e-02
-6.25577867e-01 -2.80295491e-01 -7.36793697e-01 -1.20230496e+00
5.26813343e-02 -8.72631490e-01 -5.68781793e-02 8.09816301e-01
-6.77519292e-02 -1.41999304e-01 9.38779294e-01 -5.46629131e-01
9.64783490e-01 -2.14889812e+00 1.69045508e-01 3.29007059e-01
-2.46068928e-02 6.75057352e-01 -4.95141983e-01 6.77454710e-01
7.39445388e-02 -1.13715855e-02 -3.14115733e-01 -5.28880239e-01
2.71247715e-01 2.83499241e-01 -8.21653128e-01 1.63472697e-01
6.19488537e-01 1.06910765e+00 -7.97658265e-01 -4.08799082e-01
4.84131098e-01 7.52189577e-01 -3.34836692e-01 3.56109500e-01
-1.54049769e-01 4.21882957e-01 -3.47879976e-01 8.05762827e-01
8.90340805e-01 9.88333393e-03 1.30712926e-01 -4.29283768e-01
6.27626851e-02 -1.65312335e-01 -1.41564536e+00 2.15987587e+00
-3.49677444e-01 4.30849791e-01 5.19029051e-02 -7.76730418e-01
1.23096490e+00 2.16713414e-01 4.65428531e-01 -6.40784323e-01
-1.40151814e-01 4.30639625e-01 -3.56143892e-01 -2.24367082e-01
5.81131220e-01 -6.37956262e-02 1.21811301e-01 4.28813189e-01
2.55539596e-01 -2.78450608e-01 9.22488645e-02 1.77027613e-01
7.76077151e-01 5.33458352e-01 -6.34784326e-02 -8.04621428e-02
5.57198882e-01 -5.70002079e-01 4.32076082e-02 9.04859781e-01
3.29558551e-02 9.99513745e-01 2.01300293e-01 -7.19778955e-01
-1.36467004e+00 -1.09157395e+00 -1.24044538e-01 8.53161871e-01
6.86451867e-02 -3.03464770e-01 -3.11616659e-01 -7.10168362e-01
2.21364349e-01 1.22120284e-01 -6.19985342e-01 2.23699346e-01
-7.76625216e-01 -1.73309833e-01 7.00087845e-01 7.48563349e-01
4.47967827e-01 -1.13300145e+00 -4.60358918e-01 -7.77399447e-03
1.30725116e-01 -1.27655053e+00 -3.78626436e-01 -3.37591916e-01
-8.95703137e-01 -8.61576557e-01 -1.33901203e+00 -4.46939796e-01
7.08299756e-01 1.86508670e-01 8.96133244e-01 4.76400971e-01
-4.36628461e-01 5.90136409e-01 -3.21297646e-01 -1.01828389e-01
-1.60423577e-01 6.38142154e-02 -1.10186912e-01 3.51208687e-01
-2.20005736e-01 -8.48432600e-01 -7.05899417e-01 1.72279924e-01
-1.13956833e+00 1.32289648e-01 9.70193744e-01 1.15900052e+00
4.86654580e-01 -5.27351499e-01 5.77037275e-01 -6.55244470e-01
5.57212591e-01 2.05057282e-02 -5.58339953e-01 2.90766597e-01
-6.09877169e-01 2.01878503e-01 5.52415669e-01 -8.32250893e-01
-1.00363028e+00 3.17957461e-01 -1.21368058e-01 -8.51387918e-01
-2.73722142e-01 1.25275850e-01 -2.25955650e-01 -2.96919763e-01
3.80884647e-01 3.09843183e-01 2.14908481e-01 -7.53323138e-01
7.13003099e-01 2.89793164e-01 5.38458765e-01 -9.31061268e-01
9.80383039e-01 6.88025296e-01 3.03121567e-01 -8.05466712e-01
-3.21069986e-01 -3.45733255e-01 -8.46892238e-01 -2.88830817e-01
2.67310649e-01 -6.65993273e-01 -6.01253569e-01 5.24553061e-01
-1.18514204e+00 -2.53259122e-01 -3.79421562e-01 1.08712919e-01
-5.33384979e-01 7.18668520e-01 -4.18990105e-01 -9.54634368e-01
-5.93956530e-01 -1.04794407e+00 1.39782310e+00 1.84657623e-03
1.47356361e-01 -4.75565970e-01 -3.10213953e-01 5.84345534e-02
7.34588504e-01 2.78588653e-01 7.37110198e-01 -6.27493382e-01
-9.70491946e-01 -5.20025611e-01 -5.79090476e-01 3.90459418e-01
5.69155887e-02 1.10663518e-01 -9.99921858e-01 -3.28869462e-01
-4.95435685e-01 -3.15173805e-01 8.18746567e-01 -1.03928879e-01
1.34299660e+00 -1.74002841e-01 -2.52536535e-02 5.24768233e-01
1.46367264e+00 -1.10200532e-01 7.36398578e-01 6.46281168e-02
7.92073429e-01 6.12085521e-01 5.62535644e-01 5.28879762e-01
-7.78037608e-02 9.57560062e-01 4.33782548e-01 -1.78682223e-01
-5.13841569e-01 -4.55085129e-01 3.91744152e-02 6.87268972e-01
-4.00314152e-01 -4.87649441e-02 -6.92611098e-01 5.92881203e-01
-1.85582328e+00 -9.90336835e-01 1.25395447e-01 2.30966353e+00
6.17788792e-01 -5.54136224e-02 5.93637861e-02 2.90209830e-01
4.30537730e-01 2.22793385e-01 -3.60501707e-01 2.81113461e-02
-2.46913686e-01 6.27627552e-01 2.07573757e-01 1.81327432e-01
-9.49222445e-01 8.53505909e-01 5.61909389e+00 1.12983930e+00
-1.20455098e+00 -2.62325257e-01 2.81277180e-01 1.95170157e-02
-2.67633766e-01 -2.60857224e-01 -4.35844421e-01 3.32839310e-01
2.72908777e-01 4.23844546e-01 6.58983231e-01 8.29323471e-01
-3.89307350e-01 2.53896594e-01 -1.17627132e+00 1.22360981e+00
8.14106762e-02 -1.50341916e+00 3.84705752e-01 4.66123084e-03
6.20639801e-01 -4.33243006e-01 1.84944063e-01 2.70807743e-01
-3.22230488e-01 -8.57448399e-01 7.69072592e-01 8.30872178e-01
1.14967263e+00 -6.69127941e-01 5.93601406e-01 1.75392807e-01
-1.10756707e+00 1.43620074e-01 -2.82560915e-01 7.85773844e-02
-8.11678619e-05 4.71686006e-01 -5.96274137e-01 8.15982878e-01
4.67026234e-01 5.36062419e-01 -6.67612433e-01 8.43421876e-01
-8.89535993e-02 1.16130464e-01 -4.99609828e-01 -5.80376340e-03
1.48874387e-01 -2.04326048e-01 5.53659499e-01 1.19614112e+00
3.48270625e-01 -3.53205353e-02 -1.54784977e-01 1.14568245e+00
-1.83285490e-01 1.47788152e-01 -8.07963789e-01 -2.57148683e-01
4.68240827e-01 1.51480198e+00 -7.09908068e-01 -5.42237401e-01
-1.94919512e-01 1.18259990e+00 2.80363441e-01 3.90557319e-01
-6.25159204e-01 -5.11965930e-01 4.77009267e-01 2.50487030e-02
5.77926159e-01 -3.49310666e-01 -2.96652883e-01 -1.17549896e+00
4.89220202e-01 -7.62871683e-01 -1.54811861e-02 -7.80473471e-01
-1.66663921e+00 6.89588010e-01 -1.62506074e-01 -1.37049532e+00
-1.10732570e-01 -8.87017906e-01 -4.62991655e-01 8.20475757e-01
-1.65978110e+00 -1.69077218e+00 -6.06341720e-01 5.45017123e-01
5.51163554e-01 -3.53012353e-01 9.37030792e-01 6.48087323e-01
-4.18176353e-02 8.10380757e-01 -7.53015454e-04 2.98923820e-01
8.87638450e-01 -1.08021653e+00 4.91100430e-01 6.45842910e-01
4.72643465e-01 6.96269035e-01 2.94614434e-01 -4.84141558e-01
-1.69274163e+00 -7.23912418e-01 5.84287047e-01 -4.15500522e-01
5.15451550e-01 -7.00859666e-01 -9.47228611e-01 2.60135025e-01
-5.01518250e-02 3.02816987e-01 3.17582399e-01 -6.25742376e-02
-6.61725223e-01 -2.27053598e-01 -1.01773119e+00 6.94297194e-01
1.23940003e+00 -6.57737613e-01 -3.52891177e-01 -5.14037497e-02
3.45596254e-01 -4.17075723e-01 -9.50160325e-01 6.34692907e-01
1.17581594e+00 -9.49682355e-01 1.47278905e+00 -6.87210202e-01
6.64609969e-01 -2.06304297e-01 -2.38239124e-01 -7.69775450e-01
6.19297428e-03 -4.27078158e-01 -2.52834737e-01 1.38969231e+00
-1.96073040e-01 -2.74061084e-01 8.32171381e-01 7.48079181e-01
1.70419604e-01 -9.33886468e-01 -8.73557329e-01 -8.71894419e-01
-1.58520683e-03 -3.51295233e-01 8.19509923e-01 1.01718366e+00
-3.81465971e-01 -5.29007167e-02 -6.17930889e-01 -1.75118372e-01
5.81865847e-01 4.13000703e-01 1.13120949e+00 -1.23639095e+00
-3.42007205e-02 -7.22016215e-01 -4.10688221e-01 -1.06766558e+00
1.66188702e-01 -7.48093128e-01 -1.04213193e-01 -1.28071296e+00
-4.74948473e-02 -8.65337133e-01 -2.96443224e-01 4.16478187e-01
-8.69998187e-02 4.45845008e-01 6.75480962e-01 2.86562532e-01
-2.82592744e-01 8.29406321e-01 1.41158986e+00 -2.30047867e-01
4.46756370e-02 -1.08583011e-01 -1.18774585e-01 4.49338496e-01
4.63165045e-01 -7.44330361e-02 -2.22240865e-01 -3.33928376e-01
1.99092347e-02 2.88496315e-02 6.78693771e-01 -9.02956843e-01
1.51886374e-01 5.25104851e-02 5.67526817e-01 -7.96561062e-01
7.24784493e-01 -1.14178813e+00 2.32171327e-01 2.86333412e-01
-3.01015109e-01 -7.54749551e-02 3.08549255e-01 6.38940811e-01
-1.85966954e-01 -7.43667558e-02 5.47035873e-01 -3.61819901e-02
-4.85029221e-01 4.67604339e-01 4.81891297e-02 -4.99956459e-01
5.50609231e-01 -1.56890720e-01 -1.41665921e-01 -3.71371359e-01
-5.15594006e-01 -2.92260140e-01 3.85966927e-01 6.86269462e-01
8.40797007e-01 -1.83638656e+00 -5.26558578e-01 4.29563880e-01
1.87642798e-01 -1.35130763e-01 3.54168147e-01 7.33579040e-01
-3.00524086e-01 3.67769510e-01 -6.17130220e-01 -4.47025180e-01
-1.01297939e+00 7.33143985e-01 -2.62144059e-02 -2.77552485e-01
-8.55001450e-01 5.02616525e-01 2.09315717e-01 -5.60777724e-01
3.80618155e-01 -2.53099233e-01 1.34962365e-01 1.20530777e-01
3.85051787e-01 2.29145050e-01 9.44527760e-02 -4.34544027e-01
-2.05341607e-01 5.51399410e-01 8.49513188e-02 -2.24845886e-01
1.48851085e+00 3.99179459e-01 6.99399710e-02 1.51033953e-01
9.68020856e-01 1.37945011e-01 -1.41462600e+00 -3.12950850e-01
-4.84346449e-02 -8.31875145e-01 -2.45652214e-01 -1.06807232e+00
-1.30178428e+00 1.17982614e+00 5.83786666e-01 -1.00312494e-01
1.01953101e+00 -3.00834775e-01 7.26297200e-01 4.07788277e-01
4.53907460e-01 -8.01177144e-01 4.08541888e-01 7.51398802e-02
1.55686975e+00 -1.25330067e+00 7.81089216e-02 -3.09828013e-01
-4.13147271e-01 1.47315133e+00 3.52635205e-01 -3.55245799e-01
3.94153148e-01 1.24164812e-01 -1.25022322e-01 -1.07829891e-01
-3.42749745e-01 -9.94982421e-02 6.45088136e-01 4.16923076e-01
1.69357255e-01 -1.10221937e-01 -2.08843037e-01 7.11231530e-01
2.64710546e-01 1.78410143e-01 8.12489912e-02 8.46909821e-01
1.81498930e-01 -1.52236176e+00 -2.72566825e-01 2.86892742e-01
-4.09189761e-02 -3.07896119e-02 -5.23265541e-01 1.09734845e+00
-1.10718114e-02 2.58461922e-01 -1.61822259e-01 -2.24222198e-01
4.89453256e-01 1.62773177e-01 8.12470913e-01 -2.02111527e-02
-5.12187600e-01 4.61351499e-02 -1.84015274e-01 -6.06438041e-01
-5.19905329e-01 -4.07936871e-01 -7.50660479e-01 -1.42835200e-01
-2.97879010e-01 -4.34198648e-01 8.58604312e-01 6.53197348e-01
5.82234502e-01 2.62276947e-01 3.54811400e-01 -1.39127505e+00
-6.01259768e-01 -8.82921934e-01 -3.96407992e-01 7.51290560e-01
2.19247460e-01 -9.02983189e-01 -6.44415766e-02 3.17854322e-02] | [11.720267295837402, 0.4515213072299957] |
0e1601e7-2be1-43f0-8742-9dc521a81b3e | situational-object-boundary-detection | 1504.06434 | null | http://arxiv.org/abs/1504.06434v1 | http://arxiv.org/pdf/1504.06434v1.pdf | Situational Object Boundary Detection | Intuitively, the appearance of true object boundaries varies from image to
image. Hence the usual monolithic approach of training a single boundary
predictor and applying it to all images regardless of their content is bound to
be suboptimal. In this paper we therefore propose situational object boundary
detection: We first define a variety of situations and train a specialized
object boundary detector for each of them using [Dollar and Zitnick 2013]. Then
given a test image, we classify it into these situations using its context,
which we model by global image appearance. We apply the corresponding
situational object boundary detectors, and fuse them based on the
classification probabilities. In experiments on ImageNet, Microsoft COCO, and
Pascal VOC 2012 segmentation we show that our situational object boundary
detection gives significant improvements over a monolithic approach.
Additionally, our method substantially outperforms [Hariharan et al. 2011] on
semantic contour detection on their SBD dataset. | ['Vittorio Ferrari', 'Jasper Uijlings'] | 2015-04-24 | situational-object-boundary-detection-1 | http://openaccess.thecvf.com/content_cvpr_2015/html/Uijlings_Situational_Object_Boundary_2015_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2015/papers/Uijlings_Situational_Object_Boundary_2015_CVPR_paper.pdf | cvpr-2015-6 | ['contour-detection'] | ['computer-vision'] | [ 5.42595923e-01 1.20729059e-01 -3.55486929e-01 -3.18713665e-01
-8.26987922e-01 -8.62786472e-01 5.09112000e-01 8.51365998e-02
-4.25001711e-01 3.32354814e-01 -4.51671839e-01 -4.55002904e-01
4.08941537e-01 -5.42969584e-01 -7.57739544e-01 -4.36773330e-01
2.76595265e-01 7.05777705e-01 9.14343178e-01 2.62883097e-01
4.23630685e-01 3.22767407e-01 -1.35024691e+00 2.54576325e-01
8.66161406e-01 1.22820258e+00 3.22860181e-01 9.49939728e-01
-2.19681099e-01 4.02081430e-01 -5.36576331e-01 -1.99188590e-01
3.31707656e-01 -3.86684984e-01 -1.11228168e+00 6.03535235e-01
1.04777789e+00 -1.82390466e-01 2.31110618e-01 1.30121839e+00
9.55729485e-02 1.67586524e-02 8.51850331e-01 -1.02438259e+00
-4.36336249e-01 2.80873895e-01 -8.90796125e-01 2.55042762e-01
1.54431909e-01 -2.00799350e-02 1.04252434e+00 -9.77270901e-01
8.30171525e-01 9.33819950e-01 7.38392711e-01 5.64948559e-01
-1.49725378e+00 -3.17786366e-01 6.04644716e-01 1.11420430e-01
-1.11699402e+00 -1.79827079e-01 6.85585201e-01 -5.67380071e-01
6.66015267e-01 1.10715769e-01 4.93829042e-01 6.14189744e-01
-1.62933782e-01 1.17227674e+00 1.06251633e+00 -6.11335456e-01
2.75968552e-01 1.79771781e-02 5.04272699e-01 7.53566146e-01
4.37674522e-01 -2.30172470e-01 -2.65247934e-02 -4.53469157e-02
7.14715004e-01 -3.15095752e-01 -2.38708198e-01 -6.33596122e-01
-8.99595678e-01 5.48699319e-01 3.07956010e-01 1.62808210e-01
-1.44313410e-01 1.19631305e-01 1.55111775e-01 3.88927199e-02
4.83128071e-01 8.52975771e-02 -4.71142948e-01 2.80921549e-01
-1.07643223e+00 2.58228213e-01 8.80866885e-01 9.24142480e-01
9.32959616e-01 -3.47640127e-01 2.23254219e-01 8.71692777e-01
2.52131194e-01 2.98636138e-01 2.05228344e-01 -1.18165493e+00
2.28609532e-01 4.60540950e-01 2.96496123e-01 -7.32599556e-01
-4.20749575e-01 -2.53567882e-02 -2.43358612e-01 5.12139916e-01
8.37672472e-01 -5.65486662e-02 -1.62290764e+00 1.48428333e+00
5.31932712e-01 3.29760253e-01 9.12280567e-03 9.61015522e-01
6.30559504e-01 4.37445104e-01 2.67003268e-01 1.75690949e-01
1.22504449e+00 -1.13824570e+00 -4.37422991e-01 -7.31552958e-01
5.35810113e-01 -9.73251104e-01 8.79391372e-01 3.91980439e-01
-1.00749445e+00 -5.48462391e-01 -1.09329176e+00 -1.16030611e-02
-1.31408229e-01 3.77002567e-01 4.97369081e-01 5.86476266e-01
-1.10510790e+00 6.23471022e-01 -8.03611815e-01 -3.55528146e-01
4.43509638e-01 3.39642465e-01 -8.60089660e-02 1.10538535e-01
-7.53887296e-01 7.67403364e-01 4.81783599e-01 1.16620675e-01
-8.58179808e-01 -3.21410090e-01 -8.14207137e-01 -2.26882264e-01
6.78132415e-01 -5.91475248e-01 1.52884114e+00 -1.38773882e+00
-1.23050249e+00 1.24757564e+00 -1.54017761e-01 -3.94619584e-01
5.92956722e-01 -3.09628248e-01 6.15215413e-02 4.56386477e-01
2.80738026e-01 1.05063546e+00 9.31370616e-01 -1.72963679e+00
-1.00768590e+00 -2.78669566e-01 -7.56080896e-02 2.99793929e-01
2.39728361e-01 -5.00112586e-02 -8.51982236e-01 -4.84519511e-01
4.95408773e-01 -8.71756732e-01 -2.39307821e-01 2.16493160e-01
-4.84176219e-01 -3.80819589e-01 9.86476541e-01 -6.48187876e-01
7.87104189e-01 -2.12605667e+00 -1.17234467e-02 2.55533755e-01
1.61709830e-01 2.65155375e-01 1.98971648e-02 -3.20433557e-01
1.15451783e-01 3.39355648e-01 -6.48601413e-01 -4.98315364e-01
-2.20757693e-01 1.89365238e-01 -5.98283894e-02 5.07518351e-01
2.85994440e-01 7.81868875e-01 -6.21611953e-01 -8.98499846e-01
1.95786297e-01 1.84064224e-01 -5.54964066e-01 1.70270987e-02
-4.56350446e-01 1.02179594e-01 -2.82259583e-01 8.16428781e-01
7.12582946e-01 -2.92821407e-01 2.02644140e-01 -4.79545742e-02
-1.03947230e-01 -1.43330554e-02 -1.26119304e+00 1.28762937e+00
-1.85176313e-01 6.79087162e-01 3.69735211e-01 -1.03105152e+00
7.87534356e-01 2.30565995e-01 2.69620955e-01 -3.49439561e-01
2.19157919e-01 2.95215786e-01 -1.79946154e-01 -3.57271194e-01
3.08333516e-01 -1.72003329e-01 7.80712366e-02 2.49179259e-01
4.73185219e-02 -2.25518152e-01 2.69178510e-01 1.41099870e-01
8.48317027e-01 3.79646003e-01 1.12835906e-01 -3.04106325e-01
5.11291385e-01 2.32728288e-01 6.05461717e-01 1.02309644e+00
-7.48678088e-01 1.03396749e+00 5.38200080e-01 -1.71609774e-01
-9.46918070e-01 -1.33660686e+00 -3.63004088e-01 9.40628111e-01
6.66089177e-01 7.01000467e-02 -1.05565310e+00 -8.73260498e-01
1.54038325e-01 5.48074841e-01 -8.04751396e-01 5.56425489e-02
-6.66535139e-01 -3.67194086e-01 2.45054424e-01 6.78472996e-01
6.63900077e-01 -1.16219831e+00 -6.43094718e-01 1.03817657e-01
-1.04054958e-01 -1.10398340e+00 -6.63855791e-01 2.48030901e-01
-9.69611585e-01 -1.07777810e+00 -8.22441876e-01 -1.17700613e+00
6.46234453e-01 3.89078945e-01 1.04049027e+00 1.10361941e-01
-4.99027699e-01 4.12211925e-01 -1.81479350e-01 -6.09660856e-02
-5.24116695e-01 -3.31324190e-02 -3.61544639e-01 1.79717559e-02
1.11954853e-01 -4.71327230e-02 -6.94990337e-01 3.19437444e-01
-6.42683089e-01 1.77812520e-02 5.01452148e-01 7.24320054e-01
7.25238740e-01 -2.52721936e-01 4.17762429e-01 -1.12454987e+00
1.14845790e-01 -1.91262424e-01 -7.11508811e-01 4.63091314e-01
-5.89565992e-01 -2.83800095e-01 1.53848037e-01 -3.70229721e-01
-1.27891028e+00 4.85317528e-01 3.40345912e-02 -4.21583712e-01
-5.22135675e-01 9.75849405e-02 -1.68888092e-01 -1.30750537e-01
4.08163965e-01 -3.82949114e-02 -1.96206346e-01 -6.79938734e-01
4.80706304e-01 5.77112317e-01 8.70252430e-01 -6.75726235e-01
4.63028342e-01 5.96679866e-01 -4.45003420e-01 -9.27154362e-01
-9.43265319e-01 -8.50877285e-01 -7.35330403e-01 -3.24275911e-01
1.31724107e+00 -4.79338884e-01 -4.07465458e-01 6.29673958e-01
-1.17425501e+00 -7.23898947e-01 -1.05968609e-01 1.31102517e-01
-7.19145060e-01 6.23085618e-01 -7.15345263e-01 -7.58481383e-01
7.45011680e-03 -1.14826238e+00 1.27047276e+00 5.31988621e-01
-1.15426101e-01 -1.12957942e+00 -5.55026717e-03 4.32508320e-01
-4.69625182e-02 9.42609012e-02 7.28879988e-01 -6.50301456e-01
-5.22314012e-01 -1.11910343e-01 -5.62461197e-01 3.26688617e-01
9.04735699e-02 2.49286637e-01 -9.99452174e-01 -6.05236329e-02
-2.76347622e-02 -1.95456415e-01 1.21665227e+00 5.82314491e-01
1.02296734e+00 8.98583010e-02 -6.09103322e-01 4.67690855e-01
1.57730389e+00 4.02797341e-01 2.65810072e-01 3.64615262e-01
6.61825359e-01 5.84021091e-01 6.19656980e-01 -1.02459811e-01
1.04442887e-01 5.12769818e-01 2.96719104e-01 -5.22233129e-01
-3.27957332e-01 8.64486471e-02 2.09844187e-01 9.90524329e-03
1.40246004e-01 -1.98282957e-01 -1.10781729e+00 6.61778033e-01
-1.67970598e+00 -7.07216501e-01 -7.46053532e-02 1.86223626e+00
8.28109384e-01 5.36665678e-01 2.79675543e-01 -1.25147358e-01
1.15205562e+00 -1.91088572e-01 -5.88660181e-01 -2.95030296e-01
-1.31901249e-01 1.20632425e-01 6.37121499e-01 7.17334330e-01
-1.54893434e+00 1.26850235e+00 6.57500601e+00 6.09835267e-01
-1.09109139e+00 5.59925810e-02 9.56540108e-01 4.45653081e-01
2.71709412e-02 2.44924992e-01 -9.00400102e-01 7.93440044e-02
4.20840323e-01 5.95252961e-02 8.77845846e-03 9.79876757e-01
-3.66399378e-01 -6.90367281e-01 -1.06209791e+00 6.35597587e-01
1.74773961e-01 -1.13879967e+00 -2.61565089e-01 -2.28480622e-01
8.61194909e-01 6.10716157e-02 2.05355659e-02 8.59209709e-03
4.72844809e-01 -7.53441989e-01 1.05163419e+00 1.97395056e-01
3.33046615e-01 -2.40807116e-01 5.85134447e-01 2.94928133e-01
-1.14471340e+00 1.46965072e-01 -1.97011560e-01 1.84387252e-01
1.56250983e-01 2.53224939e-01 -8.25976431e-01 1.65227294e-01
7.67807901e-01 4.15078282e-01 -6.55063927e-01 1.25632429e+00
-1.34714797e-01 8.10015321e-01 -6.27809048e-01 3.13227445e-01
3.27304453e-01 -3.02791387e-01 4.01238829e-01 1.20166814e+00
-2.12529883e-01 8.59735459e-02 4.88663137e-01 7.93816328e-01
1.36532530e-01 1.80551559e-02 -4.46209818e-01 3.21820855e-01
2.51522362e-01 1.27559328e+00 -1.57689130e+00 -6.29503608e-01
-6.30729496e-01 1.16893625e+00 2.31385157e-01 5.05592406e-01
-6.77083254e-01 -1.96748868e-01 3.78378958e-01 -8.07275325e-02
8.03194165e-01 -1.18885927e-01 -5.85926414e-01 -1.00203097e+00
-3.91594693e-02 -4.66030836e-01 5.57725251e-01 -6.02631927e-01
-1.13102627e+00 4.23604667e-01 1.29053861e-01 -7.66559362e-01
3.18471551e-01 -7.75170982e-01 -7.05740154e-01 5.53195596e-01
-1.39410746e+00 -9.89557385e-01 -2.10716143e-01 1.06238343e-01
6.72667861e-01 5.57407260e-01 3.22749197e-01 1.16093144e-01
-6.12042427e-01 2.94543862e-01 -4.67981994e-02 3.87711793e-01
5.52648485e-01 -1.48242188e+00 6.47957444e-01 8.70530486e-01
2.54603714e-01 3.01276088e-01 6.37048244e-01 -8.95793319e-01
-6.78733051e-01 -1.02932870e+00 5.39496660e-01 -3.82315099e-01
6.33384228e-01 -2.81865060e-01 -1.14034450e+00 8.84412408e-01
2.39349455e-01 4.38806415e-02 1.67040676e-01 -9.92266536e-02
-2.18881041e-01 2.30058014e-01 -9.31060135e-01 6.49419367e-01
9.80102360e-01 -2.84076542e-01 -7.81202197e-01 2.11853087e-01
5.47074139e-01 -3.44608217e-01 -5.30407608e-01 5.90298951e-01
2.60291159e-01 -1.05935907e+00 8.78879726e-01 -4.88950461e-01
4.00823019e-02 -3.76544267e-01 -6.05241284e-02 -8.65071952e-01
-3.34604383e-02 -3.53428066e-01 3.90350163e-01 1.29886770e+00
4.33773756e-01 -6.14329875e-01 1.02738512e+00 6.05485260e-01
-2.21270680e-01 -7.21173227e-01 -9.76032555e-01 -7.76248455e-01
2.52312422e-01 -3.41554552e-01 -2.80850939e-02 5.63211560e-01
-2.38929480e-01 2.36043543e-01 2.61109382e-01 1.00136384e-01
7.71378458e-01 3.89278978e-01 6.66404545e-01 -1.08673048e+00
-1.82850033e-01 -7.47837067e-01 -4.23692882e-01 -1.36924040e+00
1.08844735e-01 -8.67542207e-01 5.80462933e-01 -1.51868546e+00
2.61113554e-01 -7.61187136e-01 -2.04576254e-01 4.68633175e-01
-4.48195547e-01 5.28273880e-01 2.08057925e-01 3.13431084e-01
-8.13149869e-01 6.12057336e-02 1.22028160e+00 -1.80104703e-01
-7.46132061e-02 1.59553200e-01 -3.03986132e-01 1.02738106e+00
7.49884963e-01 -3.60719562e-01 -4.69258800e-02 -1.94201648e-01
-3.19201142e-01 6.62883818e-02 5.37012994e-01 -9.51615214e-01
1.22512922e-01 -1.21938296e-01 3.41843307e-01 -5.73530078e-01
1.22733019e-01 -7.20800042e-01 -2.58467108e-01 4.56063330e-01
-2.86948293e-01 -3.58459771e-01 3.24887365e-01 7.03937829e-01
-4.60046381e-02 -6.08298481e-01 1.09081852e+00 -3.08539812e-02
-9.41412747e-01 1.35544345e-01 -4.27046776e-01 4.30100769e-01
1.31425071e+00 -5.92434347e-01 -2.51962453e-01 -1.02008201e-01
-1.07562888e+00 3.04376036e-01 7.82429278e-01 8.75136256e-02
4.16463375e-01 -8.78743768e-01 -2.79224902e-01 1.45253316e-01
-1.01723887e-01 2.96117574e-01 -3.41172591e-02 7.91009367e-01
-6.99378848e-01 1.08205274e-01 -2.47355141e-02 -1.03906190e+00
-1.30898058e+00 5.11590660e-01 6.13174915e-01 7.12354779e-02
-6.92848742e-01 1.01273906e+00 7.52741814e-01 -1.21224366e-01
2.46434316e-01 -4.88340616e-01 -9.70150977e-02 9.61595774e-03
5.60336560e-02 2.84307189e-02 -1.99262518e-02 -5.69706082e-01
-3.83648634e-01 9.93884087e-01 -1.50879264e-01 -2.81525910e-01
7.79057741e-01 -1.04865298e-01 1.65624842e-01 6.11615658e-01
1.16433930e+00 -2.43856668e-01 -1.72117519e+00 -1.68519393e-01
4.84041065e-01 -2.73589373e-01 -1.64011136e-01 -7.54646003e-01
-9.99234378e-01 8.04972231e-01 5.25037646e-01 2.61242032e-01
1.08521867e+00 4.59632695e-01 5.99490225e-01 2.42705733e-01
9.46776792e-02 -1.28502357e+00 2.63549741e-02 4.14346009e-01
4.64150101e-01 -1.36941075e+00 -1.06945552e-01 -7.62444139e-01
-5.93285203e-01 1.16871309e+00 6.50065780e-01 -3.72569859e-01
7.05897152e-01 3.24022651e-01 4.08993006e-01 -1.86517179e-01
-5.08921623e-01 -5.40183485e-01 2.27939814e-01 4.62121278e-01
6.86448365e-02 5.00009097e-02 -2.56831318e-01 2.14682490e-01
2.47669950e-01 -1.89986050e-01 3.70054066e-01 9.63565171e-01
-8.25906157e-01 -1.06985307e+00 -5.00704467e-01 3.27128798e-01
-4.33300525e-01 6.96886480e-02 -6.46820426e-01 1.10486853e+00
1.12855852e-01 7.54691541e-01 3.73743027e-01 -5.15416153e-02
2.49753445e-01 2.22572699e-01 5.90925455e-01 -8.11360240e-01
-4.00385410e-01 4.35488671e-01 9.54320133e-02 -2.64061153e-01
-2.97237754e-01 -9.86863017e-01 -1.66790557e+00 2.68411487e-01
-6.62530422e-01 2.11863709e-03 6.37286186e-01 1.02354228e+00
-6.00978881e-02 1.96497306e-01 2.45097503e-01 -8.66051316e-01
-5.03266394e-01 -6.64349914e-01 -3.87015641e-01 5.69723129e-01
2.67876834e-01 -8.29712749e-01 -5.95443368e-01 4.00000066e-01] | [9.491044044494629, 0.34098103642463684] |
566d41a6-b755-4c3a-985c-c8ca21762875 | factuality-enhanced-language-models-for-open | 2206.04624 | null | https://arxiv.org/abs/2206.04624v3 | https://arxiv.org/pdf/2206.04624v3.pdf | Factuality Enhanced Language Models for Open-Ended Text Generation | Pretrained language models (LMs) are susceptible to generate text with nonfactual information. In this work, we measure and improve the factual accuracy of large-scale LMs for open-ended text generation. We design the FactualityPrompts test set and metrics to measure the factuality of LM generations. Based on that, we study the factual accuracy of LMs with parameter sizes ranging from 126M to 530B. Interestingly, we find that larger LMs are more factual than smaller ones, although a previous study suggests that larger LMs can be less truthful in terms of misconceptions. In addition, popular sampling algorithms (e.g., top-p) in open-ended text generation can harm the factuality due to the ''uniform randomness'' introduced at every sampling step. We propose the factual-nucleus sampling algorithm that dynamically adapts the randomness to improve the factuality of generation while maintaining quality. Furthermore, we analyze the inefficiencies of the standard training method in learning correct associations between entities from factual text corpus (e.g., Wikipedia). We propose a factuality-enhanced training method that uses TopicPrefix for better awareness of facts and sentence completion as the training objective, which can vastly reduce the factual errors. We release our code and FactualityPrompts benchmark at: https://github.com/nayeon7lee/FactualityPrompt. | ['Pascale Fung', 'Bryan Catanzaro', 'Mohammad Shoeybi', 'Mostofa Patwary', 'Peng Xu', 'Wei Ping', 'Nayeon Lee'] | 2022-06-09 | null | null | null | null | ['misconceptions'] | ['miscellaneous'] | [-1.98595092e-01 6.33395910e-01 -2.81692356e-01 -1.64233416e-01
-9.70219851e-01 -6.44444585e-01 9.17881787e-01 -5.59463874e-02
-2.61163235e-01 1.49192345e+00 5.95660388e-01 -5.37144899e-01
-7.05553219e-02 -1.32757509e+00 -1.18946898e+00 -3.14616382e-01
1.49035841e-01 3.67193609e-01 -1.26518488e-01 -4.14279014e-01
4.18292552e-01 -2.94662535e-01 -1.38260615e+00 6.09120309e-01
1.70798135e+00 3.37843448e-01 2.57998616e-01 5.07710218e-01
-3.27581257e-01 7.54731774e-01 -1.19890726e+00 -1.20492876e+00
2.02888876e-01 -4.45001423e-01 -9.18892324e-01 -4.40990090e-01
5.47241390e-01 -2.82324314e-01 -6.23476915e-02 1.11316979e+00
5.56350589e-01 1.17475100e-01 8.17378759e-01 -1.15161920e+00
-9.73112702e-01 1.47912133e+00 -3.30996245e-01 2.48175576e-01
5.11589944e-01 3.76693815e-01 1.11728013e+00 -8.62694502e-01
7.57955611e-01 1.41482091e+00 6.37585104e-01 6.47183955e-01
-8.45348895e-01 -9.32001531e-01 5.03177717e-02 -3.09504513e-02
-1.27527750e+00 -4.16184753e-01 6.49077773e-01 -2.71972746e-01
7.82025456e-01 4.69765931e-01 5.40471315e-01 1.59044194e+00
4.64111835e-01 7.50508904e-01 9.22825754e-01 -5.06474197e-01
1.91234484e-01 4.24879730e-01 -2.05028638e-01 4.33253527e-01
1.01005757e+00 2.02949911e-01 -6.09349191e-01 -3.03676009e-01
4.71751302e-01 -3.62062305e-01 -5.20078778e-01 2.77601570e-01
-1.15558875e+00 1.05198932e+00 1.88401282e-01 3.74003798e-01
-3.76118541e-01 8.74632001e-02 2.44858757e-01 2.16405228e-01
6.15619421e-01 1.11830735e+00 -6.92114949e-01 -2.34490946e-01
-9.21605408e-01 7.40593553e-01 8.95199180e-01 8.66638482e-01
4.08994853e-01 5.43661863e-02 -5.84245384e-01 9.15626645e-01
6.08086139e-02 7.84986138e-01 9.79618132e-01 -6.88325763e-01
9.38665569e-01 4.06447202e-01 2.38533318e-01 -1.08377719e+00
6.85586557e-02 -6.01900578e-01 -8.17198992e-01 -3.33324343e-01
3.38032037e-01 -6.90626621e-01 -4.42523688e-01 1.90697181e+00
8.13105553e-02 -6.77870438e-02 2.95355320e-01 5.04942298e-01
8.05675387e-01 8.25479865e-01 4.77275550e-02 -3.14234376e-01
1.12345004e+00 -6.60206378e-01 -8.47501278e-01 -1.30428314e-01
8.61596882e-01 -8.28765690e-01 1.48960567e+00 2.84571588e-01
-9.53137994e-01 -3.39290172e-01 -9.83777285e-01 2.62138933e-01
-5.36626279e-01 1.40054226e-01 7.48868227e-01 9.04211223e-01
-6.23219192e-01 8.33885193e-01 -2.67369777e-01 -3.38836089e-02
5.13805211e-01 -2.96351045e-01 8.45520645e-02 1.91494644e-01
-1.93533933e+00 7.76872277e-01 9.03003514e-01 -4.69464511e-01
-7.00869620e-01 -1.10640681e+00 -6.98937476e-01 -2.28282828e-02
6.91378117e-01 -8.69828701e-01 1.34599423e+00 -6.62212610e-01
-1.33190453e+00 3.23526770e-01 2.25616749e-02 -8.02431703e-01
6.74498379e-01 -3.84289712e-01 -4.10031736e-01 -1.69073075e-01
2.82729506e-01 5.58747053e-01 6.03235364e-01 -1.39494240e+00
-5.90536892e-01 1.59790516e-01 2.09403738e-01 3.35840315e-01
-4.73502696e-01 -4.99967128e-01 1.99613705e-01 -1.01283693e+00
-3.49502593e-01 -5.16015410e-01 -3.71349417e-02 -7.88445473e-01
-9.30045784e-01 -4.00679410e-01 2.20810950e-01 -5.67658186e-01
1.65680623e+00 -1.63911271e+00 -6.20858967e-01 -6.38870522e-02
1.33647129e-01 3.45083952e-01 -1.22738726e-01 6.21872365e-01
3.06609031e-02 8.96352947e-01 1.58210546e-02 1.39938638e-01
1.46566540e-01 -1.48048684e-01 -6.88897908e-01 -4.32459503e-01
2.73015685e-02 1.00133884e+00 -1.07767665e+00 -4.90210474e-01
-1.60285085e-01 -6.96073771e-02 -7.22283244e-01 4.76847310e-03
-6.10046804e-01 -2.62723684e-01 -3.29499036e-01 2.10527346e-01
5.29891789e-01 -2.82473862e-01 2.27698963e-02 6.61625192e-02
1.11586899e-01 1.00993276e+00 -9.78560686e-01 1.29478514e+00
-7.01771259e-01 4.55258220e-01 -8.87167752e-01 -3.17339182e-01
8.01829517e-01 3.43628019e-01 -1.52710527e-01 -4.80590731e-01
-3.48831825e-02 3.13331753e-01 3.42273004e-02 -5.24790347e-01
1.04733038e+00 -2.20191628e-01 -5.99617325e-03 6.86020494e-01
-1.00362621e-01 -3.77687901e-01 5.40401697e-01 6.29676700e-01
8.40915740e-01 -1.57654300e-01 4.58094895e-01 -2.08230659e-01
9.02101025e-02 -1.87424254e-02 4.47288215e-01 1.12962663e+00
2.24934489e-01 4.24077421e-01 6.30830586e-01 -1.62319764e-01
-1.04310369e+00 -1.16458154e+00 -1.85833797e-01 6.28290176e-01
-1.88392833e-01 -6.50563300e-01 -8.03626835e-01 -1.10302675e+00
-4.05943468e-02 1.79025531e+00 -4.84592795e-01 -3.24431062e-01
-2.49634147e-01 -1.06119716e+00 7.34813690e-01 1.23567991e-01
7.53677785e-01 -1.12442112e+00 -4.87451732e-01 1.52617842e-01
-7.67472804e-01 -6.48147523e-01 -4.74526465e-01 -3.92302245e-01
-8.05511177e-01 -9.61263657e-01 -5.59054434e-01 -6.90811127e-02
5.87994277e-01 -3.81765217e-02 1.38822842e+00 -5.04660010e-02
3.13821286e-01 -6.76957592e-02 -5.45992970e-01 -8.73569131e-01
-8.25791955e-01 2.47664899e-01 -1.30383344e-02 -4.50652570e-01
2.26587549e-01 -4.22817320e-01 -5.93669355e-01 -3.65608968e-02
-9.57703233e-01 2.18821049e-01 5.98135650e-01 9.58292842e-01
9.29826573e-02 5.22838771e-01 1.01050866e+00 -1.19673049e+00
1.36562276e+00 -7.55927622e-01 -1.46463007e-01 4.36910719e-01
-8.30089152e-01 2.11464837e-01 9.16338682e-01 -5.43336153e-01
-1.56272221e+00 -9.06880915e-01 -1.06372431e-01 8.30560401e-02
1.61523856e-02 6.54126883e-01 -8.69271904e-02 5.13029158e-01
1.22347343e+00 2.99305350e-01 -4.18007612e-01 4.55050655e-02
6.17847264e-01 6.21041417e-01 1.92170382e-01 -7.82632768e-01
8.86300027e-01 1.55386478e-01 -5.53023458e-01 -3.64437670e-01
-1.20698786e+00 2.43172556e-01 1.60956547e-01 -1.11825958e-01
2.44774655e-01 -7.61608064e-01 -3.84100199e-01 2.16372684e-01
-1.41761577e+00 -3.33392143e-01 -4.85433459e-01 4.29766029e-01
-2.59978712e-01 2.34643891e-01 -3.79021704e-01 -8.49618018e-01
-5.86383045e-01 -4.12644953e-01 7.00989485e-01 3.14789027e-01
-4.85539258e-01 -1.09367979e+00 2.03653768e-01 2.66775668e-01
3.38230848e-01 2.92520404e-01 9.62186694e-01 -7.75823772e-01
-4.37284231e-01 -9.72766653e-02 6.47249967e-02 3.23986471e-01
1.27511442e-01 3.04141231e-02 -7.52695978e-01 8.65508318e-02
1.27527639e-01 -3.94841760e-01 7.85409153e-01 3.87401670e-01
8.97925794e-01 -1.30740941e+00 -2.66416460e-01 1.11100845e-01
1.40436137e+00 2.09033728e-01 8.71949434e-01 2.49789074e-01
3.76256615e-01 6.92423224e-01 8.08600247e-01 6.04876161e-01
5.55979371e-01 2.52268076e-01 -3.36921699e-02 3.16912681e-01
-1.14397861e-01 -1.02913344e+00 4.41781819e-01 7.37077117e-01
1.77160855e-02 -7.32294500e-01 -7.82373548e-01 7.43429244e-01
-1.52474189e+00 -1.41252685e+00 -8.04038271e-02 2.32141876e+00
1.50638425e+00 5.20558655e-01 -2.61325330e-01 1.03417505e-02
6.82789922e-01 2.28229731e-01 -1.28061980e-01 -4.44671720e-01
-3.08959186e-01 -1.12922236e-01 2.55818546e-01 5.21915317e-01
-5.98137379e-01 1.03172529e+00 5.56028843e+00 1.44176137e+00
-7.12184191e-01 2.05004334e-01 9.05456960e-01 -2.85673946e-01
-1.07347512e+00 -4.05105092e-02 -9.90563750e-01 9.24807131e-01
9.45695102e-01 -8.90289426e-01 1.49659961e-01 8.42406213e-01
2.49897465e-01 -2.73153275e-01 -9.48018193e-01 6.41056299e-01
1.30319238e-01 -1.54194725e+00 7.05047965e-01 -1.30545393e-01
1.24397945e+00 -4.78731394e-01 1.08386181e-01 5.72172284e-01
6.51740611e-01 -9.54351723e-01 8.78244221e-01 5.00206172e-01
6.16088092e-01 -6.99259162e-01 9.02985811e-01 6.06457949e-01
-5.27661562e-01 -1.11256922e-02 -5.15074670e-01 -1.14953093e-01
1.23611882e-01 1.25262368e+00 -1.21619654e+00 6.90343857e-01
3.43106538e-01 9.94356796e-02 -6.59306765e-01 6.90331697e-01
-5.72894394e-01 8.06809664e-01 -1.81176320e-01 -4.83291119e-01
-3.42890322e-02 1.45483613e-01 6.44105196e-01 9.86822486e-01
5.32395244e-01 5.76346405e-02 -5.10277450e-01 1.30571568e+00
-4.31837201e-01 1.68038055e-01 -8.21824729e-01 -9.97364298e-02
9.48945105e-01 8.07512283e-01 -2.36699849e-01 -5.39110541e-01
2.67580990e-02 5.43853164e-01 3.78323168e-01 3.56643319e-01
-8.41466784e-01 -5.09843230e-01 1.71593368e-01 9.20609161e-02
2.21372359e-02 3.36405486e-01 -5.63885629e-01 -1.21205032e+00
2.16687143e-01 -1.11804175e+00 3.14820498e-01 -7.25422263e-01
-1.42522466e+00 3.95393968e-01 1.53974503e-01 -9.48539913e-01
-4.54864293e-01 -1.25886396e-01 -9.21739995e-01 6.48654401e-01
-1.30020308e+00 -5.51291585e-01 1.34477928e-01 1.26772746e-01
6.53615296e-01 -6.44706562e-02 5.03707886e-01 -6.45528659e-02
-4.07955348e-01 9.24678147e-01 -8.54715109e-02 -7.44729415e-02
7.26540208e-01 -1.23732030e+00 5.82457304e-01 8.93555105e-01
2.36440524e-01 1.13601255e+00 8.94138813e-01 -1.20389342e+00
-7.81201363e-01 -1.17691064e+00 1.49570823e+00 -7.19640076e-01
4.75884765e-01 -3.11460584e-01 -6.83082402e-01 5.27519405e-01
3.39598894e-01 -8.70175600e-01 5.37737012e-01 3.76219749e-01
-3.13832104e-01 2.13626921e-02 -1.19079971e+00 1.17318785e+00
1.06004822e+00 -1.48200959e-01 -9.70299065e-01 5.60816944e-01
1.04264247e+00 -3.48383486e-01 -5.42306423e-01 2.72025019e-01
2.92940289e-01 -9.37605381e-01 7.59974897e-01 -5.62917411e-01
1.11136746e+00 6.54290617e-02 2.73403257e-01 -1.79510379e+00
-2.91122571e-02 -8.79774570e-01 -1.60862222e-01 1.53379941e+00
9.82054651e-01 -8.55889797e-01 4.82978910e-01 4.74407196e-01
2.87353899e-02 -9.28657234e-01 -7.43270278e-01 -1.05306661e+00
4.54605609e-01 -3.65609795e-01 9.94793832e-01 1.04247463e+00
2.08339348e-01 3.98440808e-01 -2.96162009e-01 -1.47153020e-01
4.29890692e-01 2.21998561e-02 7.90679514e-01 -6.67352021e-01
-3.13046277e-01 -3.91218781e-01 3.07117224e-01 -7.72916377e-01
5.02695628e-02 -9.10196960e-01 -2.47534234e-02 -1.44997728e+00
3.44385028e-01 -3.29267442e-01 1.28828183e-01 2.66862780e-01
-7.31673062e-01 -3.72593790e-01 1.78746566e-01 -3.17664555e-04
-3.82203609e-01 7.44408071e-01 1.51071882e+00 6.02774061e-02
-2.89039433e-01 -1.42209470e-01 -1.11071646e+00 6.34915113e-01
1.12598717e+00 -6.31902695e-01 -5.99169791e-01 -2.37366185e-01
6.78310812e-01 -6.10541217e-02 3.95739585e-01 -9.20275271e-01
-1.15452982e-01 -2.28181452e-01 2.29937181e-01 -4.18927580e-01
-1.87908605e-01 -1.69787094e-01 8.40622932e-02 4.64707643e-01
-6.63050056e-01 -2.19854712e-02 4.20112424e-02 3.86520296e-01
-4.56840582e-02 -5.72981954e-01 3.26869786e-01 -3.84866357e-01
-9.51149911e-02 -8.13779160e-02 -2.62423158e-01 7.03701913e-01
6.91894114e-01 9.70224515e-02 -9.29907858e-01 -5.20069540e-01
1.51556194e-01 1.65024891e-01 1.76830515e-01 3.90730858e-01
5.58373034e-01 -1.35001910e+00 -9.83437896e-01 -3.12428385e-01
-2.00600363e-02 -1.45531669e-01 3.39507610e-01 4.88395393e-01
-2.82110900e-01 7.02470303e-01 1.04031883e-01 2.51492336e-02
-6.76066458e-01 3.14798534e-01 2.56329626e-01 -7.92647183e-01
-1.21270373e-01 1.00923395e+00 8.44390243e-02 -5.32471359e-01
-9.15083811e-02 -2.17448115e-01 -8.92469436e-02 -4.18968834e-02
5.12802124e-01 5.83968997e-01 -2.24116333e-02 2.25867644e-01
1.25819549e-01 -1.71256512e-01 -4.11270618e-01 -3.23268801e-01
7.57101178e-01 3.76554020e-03 3.00363358e-02 3.47113431e-01
7.34498799e-01 4.66987967e-01 -8.46613467e-01 7.54724741e-02
-3.35106216e-02 -5.70721626e-01 -2.44360879e-01 -1.23805988e+00
-6.09005749e-01 4.61751103e-01 -8.97979960e-02 4.64741200e-01
7.97261178e-01 -2.26934761e-01 1.01403093e+00 4.28531736e-01
4.94742543e-01 -1.23953998e+00 2.10058257e-01 4.65757966e-01
1.19482660e+00 -1.13374436e+00 1.12764724e-02 -4.63237911e-01
-8.55630815e-01 7.00159907e-01 8.93289804e-01 1.97995156e-01
2.07861081e-01 1.95228420e-02 -1.13641880e-02 -8.07213858e-02
-1.13174987e+00 1.69933274e-01 1.79718345e-01 3.93572390e-01
6.66144550e-01 2.98185706e-01 -1.04361272e+00 1.02530980e+00
-1.12966812e+00 -1.15821905e-01 7.37124920e-01 3.32431704e-01
-2.44290665e-01 -8.80824566e-01 -4.03650820e-01 8.55814695e-01
-5.77270806e-01 -6.15936220e-01 -4.79824334e-01 6.38977766e-01
3.37894678e-01 1.18741369e+00 -2.76705414e-01 -3.75076443e-01
1.16233103e-01 2.27403924e-01 2.08075315e-01 -6.66970849e-01
-5.99708021e-01 -5.47049165e-01 5.73335171e-01 -2.30622515e-01
3.08855940e-02 -6.19882524e-01 -1.07207990e+00 -5.76316774e-01
-5.21593809e-01 4.13871944e-01 4.06166404e-01 8.44943821e-01
4.70377505e-01 4.95931596e-01 6.65939391e-01 4.89284210e-02
-9.11703169e-01 -1.20585120e+00 -3.80802304e-01 5.55743456e-01
-1.50879383e-01 -4.65917915e-01 -6.88172460e-01 -7.65931606e-02] | [11.749735832214355, 8.900555610656738] |
7535c2ec-b958-4b11-b687-be650ecfe5be | binary-classifier-calibration-non-parametric | 1401.3390 | null | http://arxiv.org/abs/1401.3390v1 | http://arxiv.org/pdf/1401.3390v1.pdf | Binary Classifier Calibration: Non-parametric approach | Accurate calibration of probabilistic predictive models learned is critical
for many practical prediction and decision-making tasks. There are two main
categories of methods for building calibrated classifiers. One approach is to
develop methods for learning probabilistic models that are well-calibrated, ab
initio. The other approach is to use some post-processing methods for
transforming the output of a classifier to be well calibrated, as for example
histogram binning, Platt scaling, and isotonic regression. One advantage of the
post-processing approach is that it can be applied to any existing
probabilistic classification model that was constructed using any
machine-learning method.
In this paper, we first introduce two measures for evaluating how well a
classifier is calibrated. We prove three theorems showing that using a simple
histogram binning post-processing method, it is possible to make a classifier
be well calibrated while retaining its discrimination capability. Also, by
casting the histogram binning method as a density-based non-parametric binary
classifier, we can extend it using two simple non-parametric density estimation
methods. We demonstrate the performance of the proposed calibration methods on
synthetic and real datasets. Experimental results show that the proposed
methods either outperform or are comparable to existing calibration methods. | ['Gregory F. Cooper', 'Milos Hauskrecht', 'Mahdi Pakdaman Naeini'] | 2014-01-14 | null | null | null | null | ['classifier-calibration', 'classifier-calibration'] | ['computer-vision', 'miscellaneous'] | [ 2.95847416e-01 -3.44866887e-02 -1.87257469e-01 -7.87175834e-01
-9.85439420e-01 -5.21452844e-01 5.84006310e-01 3.97572607e-01
-3.25660527e-01 9.86176729e-01 -3.85644197e-01 -4.77228522e-01
-3.60244840e-01 -1.10955203e+00 -7.51397133e-01 -9.28962529e-01
1.14326596e-01 7.11605012e-01 5.83390832e-01 1.37255535e-01
4.67477828e-01 7.00740278e-01 -1.82464290e+00 -4.37735766e-02
8.67341220e-01 1.15083730e+00 -2.22100332e-01 8.22095096e-01
-1.22760803e-01 3.76870483e-01 -5.41712642e-01 -4.16889876e-01
6.31702319e-02 -2.75162131e-01 -6.52631044e-01 -1.58902198e-01
1.69438273e-01 2.56203990e-02 2.09612548e-01 1.16169190e+00
9.33241323e-02 1.36515871e-01 1.39619160e+00 -1.59997976e+00
-5.13419747e-01 6.51715815e-01 -5.92955530e-01 -1.59537084e-02
2.19135478e-01 -6.06744707e-01 6.05949640e-01 -5.68792403e-01
-4.10436355e-02 1.37480199e+00 7.63547122e-01 9.27725211e-02
-1.26356900e+00 -8.60068440e-01 -3.85045230e-01 3.21177877e-02
-1.51651073e+00 -2.92952359e-02 5.64249933e-01 -7.66560316e-01
2.58101404e-01 2.43139356e-01 1.54130146e-01 5.32724023e-01
4.32196558e-01 1.34672403e-01 1.59920931e+00 -8.04100156e-01
5.40589392e-01 5.34863055e-01 4.95895892e-01 5.09885669e-01
6.37472749e-01 2.04872414e-01 -1.55546561e-01 -3.67878854e-01
4.64966625e-01 -8.47646445e-02 -1.13390602e-01 -5.86457968e-01
-9.18819547e-01 1.00577199e+00 2.20537037e-01 4.18053865e-01
-1.37035027e-01 3.12045030e-02 6.52360246e-02 -1.00278981e-01
5.03713906e-01 1.83177814e-01 -1.92581385e-01 1.22170679e-01
-1.13099635e+00 2.12269679e-01 9.49512541e-01 9.73182678e-01
9.50601399e-01 -2.67437160e-01 -1.21039219e-01 6.46452129e-01
3.10748726e-01 6.24001265e-01 5.40992677e-01 -6.11852467e-01
-2.62759961e-02 3.77662838e-01 3.37020159e-01 -8.73976886e-01
-1.07759878e-01 -1.90724526e-02 -7.34010279e-01 3.25718850e-01
4.51399475e-01 1.19776681e-01 -9.13793027e-01 1.65093803e+00
2.84313202e-01 1.79053664e-01 1.59034848e-01 2.92334199e-01
4.09072101e-01 7.46360540e-01 3.36481452e-01 -3.44284892e-01
1.16447234e+00 -3.56116772e-01 -4.53389168e-01 3.46410096e-01
2.53315032e-01 -9.71321583e-01 9.67581809e-01 5.13754845e-01
-5.28638542e-01 -6.01895988e-01 -1.37160003e+00 2.57480741e-01
-6.84247375e-01 4.03767750e-02 5.47525525e-01 1.18100929e+00
-7.21737027e-01 6.15017533e-01 -7.62490153e-01 -5.52557290e-01
3.06095760e-02 4.16622907e-01 -4.39941406e-01 -3.17330137e-02
-9.80918288e-01 1.06472802e+00 8.92400861e-01 -3.11502278e-01
-4.89107937e-01 -3.91164601e-01 -7.75103092e-01 1.90845668e-01
-1.28452182e-01 -2.48088256e-01 1.33807063e+00 -8.17549765e-01
-1.48736668e+00 5.45559883e-01 -1.94864407e-01 -2.94917554e-01
3.78566772e-01 -2.26535983e-02 -3.83355588e-01 -4.53872196e-02
2.94242911e-02 4.28704560e-01 6.66909635e-01 -1.53850102e+00
-8.75949562e-01 -2.49001384e-01 -1.17191091e-01 -1.38766333e-01
-2.66885847e-01 -1.44860879e-01 -9.05746128e-03 -3.41799498e-01
3.22257757e-01 -8.47211003e-01 -2.53994279e-02 -1.84688285e-01
-5.04678786e-01 -1.81522146e-01 9.59505856e-01 -3.94372016e-01
9.29232657e-01 -1.94487393e+00 -2.26461902e-01 7.66971350e-01
-2.31790721e-01 -4.01307680e-02 3.37836117e-01 2.72069395e-01
-2.11229801e-01 1.30801290e-01 -6.10721707e-01 -7.53781796e-02
-3.55859324e-02 3.62401634e-01 -4.12105203e-01 6.17513478e-01
9.74459127e-02 2.63994962e-01 -5.55328906e-01 -9.07963872e-01
4.06816572e-01 6.24801993e-01 -1.44786671e-01 3.60365599e-01
5.12872972e-02 2.09968790e-01 -1.26757741e-01 6.14687502e-01
9.07446921e-01 1.75179079e-01 1.36897817e-01 -2.88021833e-01
-1.14564095e-02 -1.33887410e-01 -1.39810109e+00 7.82747328e-01
-3.76065671e-01 3.47694159e-01 -5.25474072e-01 -1.38171971e+00
1.33661950e+00 1.33307725e-01 2.65127838e-01 -2.58548018e-02
3.57598752e-01 3.32699418e-01 -4.23383266e-01 -5.29128965e-03
4.13623899e-01 -6.92767799e-01 -2.90960759e-01 2.15769798e-01
2.46907279e-01 -3.66012424e-01 2.66341001e-01 -1.41138330e-01
4.55896109e-01 1.44423127e-01 7.43340015e-01 -5.89981437e-01
7.57838607e-01 6.91577345e-02 3.62279236e-01 6.83445871e-01
6.09389283e-02 2.51660913e-01 6.20772779e-01 3.89478332e-03
-1.16055167e+00 -1.40595436e+00 -6.08509243e-01 1.07818484e+00
3.95397469e-02 1.37237042e-01 -9.57092822e-01 -3.79586220e-01
1.62440106e-01 9.24153149e-01 -6.34217620e-01 -2.09996372e-01
-1.37717828e-01 -1.01667678e+00 4.87068921e-01 7.03895092e-01
4.97798592e-01 -4.09693152e-01 -3.11011255e-01 1.08847931e-01
1.83339580e-03 -7.90976107e-01 4.69012335e-02 5.46363115e-01
-1.01641846e+00 -1.09748316e+00 -6.13253474e-01 -8.11161578e-01
5.73679686e-01 1.22363761e-01 9.02566731e-01 -1.34853140e-01
1.69066131e-01 2.92014241e-01 -2.89363116e-01 -5.45916915e-01
-7.47952998e-01 6.50876239e-02 2.74087161e-01 -9.71327722e-02
5.26086688e-01 -5.50693691e-01 -1.67117670e-01 5.39288044e-01
-9.03616250e-01 -2.71749198e-01 8.31537068e-01 8.22579920e-01
9.19595301e-01 4.21948910e-01 6.66970015e-01 -1.16361618e+00
5.57242513e-01 -4.46252197e-01 -9.83734488e-01 6.42393351e-01
-9.42824185e-01 4.20918941e-01 5.12544930e-01 -5.22311687e-01
-1.05012572e+00 4.65910584e-01 -7.52232894e-02 -4.60503437e-02
-3.03721666e-01 4.20761436e-01 -3.54154468e-01 -2.56954014e-01
8.06909382e-01 1.03009373e-01 -3.32325958e-02 -2.47495383e-01
3.32032472e-01 9.61270273e-01 9.07500863e-01 -9.17023420e-01
1.10171211e+00 9.57970470e-02 4.07440096e-01 -6.99835658e-01
-8.63558948e-01 -5.81195712e-01 -9.46612060e-01 -1.49516806e-01
8.83341730e-01 -3.78457338e-01 -7.36064672e-01 3.09450060e-01
-7.23762631e-01 2.03234013e-02 8.15976113e-02 6.71089768e-01
-5.91915190e-01 4.63782609e-01 -8.16790164e-02 -1.10558546e+00
-1.79194927e-01 -9.40420866e-01 8.19722474e-01 4.54376668e-01
-6.46038214e-03 -1.10228837e+00 2.76633650e-01 -5.46140596e-02
2.19230190e-01 4.68973488e-01 1.22228825e+00 -1.10353506e+00
1.22432353e-03 -5.77965617e-01 -2.65490621e-01 5.62588751e-01
9.09297392e-02 4.33836550e-01 -1.08000636e+00 -1.69983134e-01
-4.83434089e-02 -1.85357720e-01 6.18314981e-01 2.74247497e-01
1.22477365e+00 -2.24623606e-01 -5.71131349e-01 3.67557466e-01
1.56722915e+00 4.32214439e-01 6.66184366e-01 2.47758701e-01
3.04095864e-01 3.79264861e-01 8.26142550e-01 9.96022075e-02
1.85750172e-01 4.85453367e-01 5.08527756e-02 3.19667836e-03
4.70615238e-01 -3.27765405e-01 -3.05654015e-02 5.78211725e-01
-2.23743170e-03 -6.74666092e-02 -9.83380497e-01 1.37102589e-01
-1.76299560e+00 -9.51568425e-01 -1.64982319e-01 2.74516916e+00
8.64868701e-01 2.95494974e-01 2.10800022e-01 6.60635531e-01
1.08997893e+00 -6.95824623e-01 -2.26264447e-01 -4.78295654e-01
6.94638342e-02 5.76724946e-01 6.61071777e-01 6.82696700e-01
-1.40338945e+00 5.06940365e-01 6.97996521e+00 8.27637315e-01
-1.00747716e+00 -1.40447598e-02 8.68959963e-01 7.42843807e-01
-8.42438415e-02 1.32920340e-01 -8.99271667e-01 3.81068856e-01
1.18761253e+00 -3.76907676e-01 -2.65260916e-02 1.26032722e+00
-2.22057745e-01 -7.40303218e-01 -1.21395338e+00 8.41816306e-01
-1.83370393e-02 -8.89695883e-01 2.79466547e-02 -7.93549940e-02
5.43644726e-01 -8.20849240e-01 -1.09405471e-02 5.16167402e-01
3.56131107e-01 -1.20944035e+00 5.15118778e-01 6.82706714e-01
7.54350483e-01 -1.11326444e+00 1.23943567e+00 3.59249145e-01
-1.14927089e+00 1.10768810e-01 -5.74417770e-01 1.45295948e-01
-2.84363449e-01 8.07387412e-01 -1.02276003e+00 4.13094342e-01
4.90661621e-01 -1.09292911e-02 -6.97814941e-01 1.33168805e+00
-7.63805583e-02 7.53674150e-01 -5.33787251e-01 -1.15691818e-01
-1.61233500e-01 -1.77048504e-01 -4.91707139e-02 1.38402510e+00
6.30568445e-01 2.65003741e-02 -7.17455447e-02 6.52635396e-01
2.70282149e-01 1.80699736e-01 -5.06478131e-01 1.97829753e-01
7.61061311e-01 1.11427426e+00 -9.61104512e-01 -4.84110326e-01
-1.09423600e-01 5.05875528e-01 1.37300149e-01 5.44590577e-02
-1.02265239e+00 -5.96775770e-01 -4.39579301e-02 6.40308782e-02
2.74166372e-02 -1.13687083e-01 -4.47982073e-01 -7.81538427e-01
-3.29796940e-01 -5.02137899e-01 5.50135970e-01 -7.68134177e-01
-1.41491318e+00 5.08877516e-01 7.69319117e-01 -1.10516047e+00
-3.76522928e-01 -8.14225674e-01 -4.78128642e-01 9.45571899e-01
-1.27149296e+00 -1.14888573e+00 -5.00065863e-01 6.50299072e-01
-3.49082313e-02 -5.76152727e-02 9.54582036e-01 1.99500825e-02
-1.96852833e-01 6.40625775e-01 2.86800623e-01 -3.84802185e-02
8.92103255e-01 -1.54495633e+00 -4.45287436e-01 9.10475850e-01
-3.14335376e-02 4.63055193e-01 9.56030428e-01 -4.80947107e-01
-7.82220781e-01 -1.03512740e+00 7.19852209e-01 -1.65419236e-01
5.38173735e-01 -2.56066501e-01 -1.12144566e+00 5.89247584e-01
-2.14703336e-01 -1.56332955e-01 9.04683471e-01 1.45325541e-01
-4.86635566e-01 -4.11811650e-01 -1.47095752e+00 -1.05855362e-02
1.10831521e-01 -2.45768115e-01 -1.00713253e+00 3.73556197e-01
2.32174933e-01 -2.22480208e-01 -1.07105720e+00 5.44411719e-01
7.05960453e-01 -1.09751391e+00 8.48542929e-01 -2.39180505e-01
1.39359936e-01 -4.67329562e-01 -4.30886894e-01 -1.17523527e+00
-1.73048422e-01 -1.60694033e-01 2.46747047e-01 1.53503287e+00
3.95669192e-01 -8.68221581e-01 5.27832270e-01 7.43105233e-01
2.43186027e-01 -5.36812067e-01 -9.06481802e-01 -9.09099042e-01
4.86777395e-01 -3.69095504e-01 5.41946948e-01 6.76811814e-01
-5.95863117e-03 1.20817356e-01 -8.61688852e-02 3.28765780e-01
7.77805567e-01 4.45560031e-02 7.86979198e-01 -1.56237233e+00
-2.91860908e-01 -2.09609613e-01 -8.27712536e-01 -4.50477332e-01
1.15908980e-01 -5.75086057e-01 3.74283850e-01 -1.31582606e+00
4.29815590e-01 -7.49166131e-01 -4.05387729e-01 4.55671996e-01
-2.05084965e-01 2.04768300e-01 -2.70927668e-01 3.09795350e-01
3.96909192e-02 1.12822063e-01 6.04140341e-01 2.10385129e-01
-1.00720845e-01 3.89817506e-01 -6.00819588e-01 7.66234696e-01
9.27601039e-01 -7.05446839e-01 -3.84404331e-01 5.14493763e-01
-1.53827548e-01 7.34051764e-02 3.67831945e-01 -1.40176272e+00
1.52804971e-01 -2.58222073e-01 6.43736005e-01 -6.96136177e-01
1.79052651e-01 -8.19551826e-01 2.62542814e-01 3.51591140e-01
-2.17268452e-01 4.79178205e-02 1.64536402e-01 5.81788540e-01
-3.47549111e-01 -5.99282503e-01 1.38076389e+00 2.77261823e-01
-5.74190795e-01 -1.33933589e-01 -3.12738746e-01 -2.56438971e-01
1.40316403e+00 -2.82127053e-01 -2.71049857e-01 -3.33000124e-01
-4.96998072e-01 -2.31758237e-01 4.74184394e-01 -1.60959378e-01
3.17977041e-01 -1.28427505e+00 -3.41316909e-01 -1.10711139e-02
1.70202434e-01 -2.36201614e-01 -2.90674269e-01 5.38698614e-01
-6.98376298e-01 3.90942156e-01 -3.95198524e-01 -7.93389320e-01
-1.22712195e+00 7.75396705e-01 3.39562863e-01 3.00697144e-02
-3.21120359e-02 4.61615175e-01 -1.57118831e-02 -3.73018295e-01
1.34813428e-01 -5.51363826e-01 -2.05386385e-01 -2.36317307e-01
2.94474244e-01 3.34142417e-01 1.05214149e-01 -7.35758007e-01
-3.60180199e-01 7.90456116e-01 3.18603128e-01 -3.59025240e-01
9.94443238e-01 2.22316653e-01 -1.59222201e-01 6.75106525e-01
1.04222202e+00 -1.56534687e-01 -8.02000046e-01 9.22972932e-02
1.54465571e-01 -3.78764689e-01 -3.30130495e-02 -6.74416482e-01
-5.31553626e-01 8.48338604e-01 7.34407306e-01 4.57343727e-01
1.23070085e+00 -1.07647218e-01 1.09663524e-01 5.78038871e-01
3.94308537e-01 -8.38658273e-01 -3.36670846e-01 1.99015230e-01
5.48018694e-01 -1.15876651e+00 2.11454794e-01 -7.16221154e-01
-5.32391369e-01 1.54333735e+00 4.82109457e-01 -1.64624229e-01
1.03054035e+00 4.14666742e-01 -1.65258616e-01 3.74587476e-01
-2.46798083e-01 8.41281656e-03 4.84307528e-01 8.01253915e-01
6.11598313e-01 3.05562645e-01 -4.17161971e-01 6.63196683e-01
-4.93051857e-01 1.25720173e-01 5.54039717e-01 8.82963955e-01
-8.40372801e-01 -1.14355910e+00 -8.43301356e-01 4.73755747e-01
-1.95164159e-01 2.28801057e-01 -1.01805776e-01 1.13670182e+00
7.33078420e-02 1.04252779e+00 1.07255086e-01 -4.34300512e-01
1.74214512e-01 4.04080689e-01 4.81031567e-01 -4.14059073e-01
7.79309571e-02 -1.62865683e-01 -1.32791862e-01 3.35835107e-03
-5.95219016e-01 -5.34408450e-01 -1.07417393e+00 -4.51131105e-01
-6.26346290e-01 4.86553669e-01 9.17573690e-01 9.99998987e-01
-2.84387678e-01 1.40702739e-01 5.64417601e-01 -7.63861835e-01
-8.34579706e-01 -9.73200440e-01 -7.27537453e-01 6.34495988e-02
-1.39485881e-01 -1.33127809e+00 -5.35754263e-01 3.14732343e-01] | [8.074762344360352, 4.195536136627197] |
2396cd92-21cc-48f5-9381-06243089be4f | towards-reliable-online-clickbait-video | 1907.07604 | null | https://arxiv.org/abs/1907.07604v2 | https://arxiv.org/pdf/1907.07604v2.pdf | Towards Reliable Online Clickbait Video Detection: A Content-Agnostic Approach | Online video sharing platforms (e.g., YouTube, Vimeo) have become an increasingly popular paradigm for people to consume video contents. Clickbait video, whose content clearly deviates from its title/thumbnail, has emerged as a critical problem on online video sharing platforms. Current clickbait detection solutions that mainly focus on analyzing the text of the title, the image of the thumbnail, or the content of the video are shown to be suboptimal in detecting the online clickbait videos. In this paper, we develop a novel content-agnostic scheme, Online Video Clickbait Protector (OVCP), to effectively detect clickbait videos by exploring the comments from the audience who watched the video. Different from existing solutions, OVCP does not directly analyze the content of the video and its pre-click information (e.g., title and thumbnail). Therefore, it is robust against sophisticated content creators who often generate clickbait videos that can bypass the current clickbait detectors. We evaluate OVCP with a real-world dataset collected from YouTube. Experimental results demonstrate that OVCP is effective in identifying clickbait videos and significantly outperforms both state-of-the-art baseline models and human annotators. | ['Dong Wang', 'Daniel Zhang', 'Lanyu Shang', 'Shuyue Lai', 'Michael Wang'] | 2019-07-17 | null | null | null | null | ['clickbait-detection'] | ['natural-language-processing'] | [-8.25945288e-02 -6.48746669e-01 -7.12769270e-01 -8.60580243e-03
-9.51899111e-01 -9.97998178e-01 5.82885981e-01 1.09611504e-01
-1.61199495e-01 3.42347145e-01 8.22833329e-02 -3.80431086e-01
3.64285350e-01 -1.99145511e-01 -7.73380876e-01 -2.47147828e-01
1.59576505e-01 -3.32751632e-01 9.84889865e-01 2.56803602e-01
6.35512114e-01 -2.30775848e-01 -1.67382681e+00 6.93049312e-01
7.33118773e-01 1.66978884e+00 1.33367985e-01 7.84463644e-01
-1.77312806e-01 8.85538518e-01 -8.35568368e-01 -4.47457761e-01
3.23337018e-01 -2.26217434e-01 -2.76859850e-01 9.45766717e-02
1.23129737e+00 -9.93483901e-01 -9.19696867e-01 1.21604586e+00
-6.02563433e-02 -4.28827077e-01 4.45618443e-02 -1.60370696e+00
-6.81005478e-01 2.66479522e-01 -6.24653816e-01 8.53952050e-01
7.44715750e-01 1.05290115e-01 1.15818965e+00 -8.41263890e-01
7.54869163e-01 8.74165773e-01 4.38652426e-01 2.99816400e-01
-8.11594665e-01 -8.28357100e-01 -5.57893515e-02 5.42387843e-01
-1.54245579e+00 -1.76909938e-01 7.19269753e-01 -8.00296247e-01
2.41775542e-01 7.76783824e-01 7.31720030e-01 1.41662991e+00
1.75154477e-01 1.62538624e+00 8.42457056e-01 -6.64078221e-02
7.35883340e-02 1.94645509e-01 3.00880909e-01 2.53122628e-01
3.42387646e-01 -1.94050461e-01 -8.68716180e-01 -3.65049243e-01
1.94836155e-01 2.68073291e-01 -3.64095360e-01 -2.91369677e-01
-7.48977780e-01 8.56825411e-01 2.02993348e-01 8.60905945e-02
-1.20602071e-01 2.50105590e-01 6.09780610e-01 1.04872666e-01
5.28384984e-01 -3.04285288e-02 -4.38139439e-02 -4.79486823e-01
-1.36025977e+00 5.03672719e-01 7.67734766e-01 1.15163553e+00
4.44603503e-01 -4.40714180e-01 -2.72177339e-01 6.32984638e-01
3.15997481e-01 5.14508009e-01 3.94319773e-01 -6.57934666e-01
9.41299260e-01 6.50994420e-01 3.87821525e-01 -1.29416311e+00
5.03278553e-01 -2.59962417e-02 9.77992713e-02 -3.00924003e-01
3.94161671e-01 4.58886512e-02 -7.91739285e-01 7.86647201e-01
1.98055118e-01 5.30579314e-02 -1.09695613e+00 9.72105682e-01
6.13750994e-01 6.00944519e-01 9.77133214e-02 1.53024383e-02
1.59406400e+00 -1.05662179e+00 -9.67092514e-01 -2.73658693e-01
4.03732091e-01 -1.04796791e+00 1.05585396e+00 3.95820737e-01
-9.20645833e-01 -4.20065403e-01 -1.11875224e+00 -8.84908587e-02
-5.42758644e-01 1.46323264e-01 2.35840917e-01 7.74374843e-01
-3.27809334e-01 2.66816944e-01 -5.06747305e-01 -2.86237448e-01
6.86169326e-01 -3.97356421e-01 -7.78345391e-02 -9.13174450e-02
-1.08256102e+00 1.37612462e-01 3.35422784e-01 -2.87204474e-01
-9.58070517e-01 -1.03149498e+00 -3.13863158e-01 1.11625165e-01
1.02889109e+00 2.18878716e-01 1.56383097e+00 -1.32227325e+00
-7.57000864e-01 7.66999006e-01 6.66182414e-02 -3.95212948e-01
8.99158120e-01 -8.08282077e-01 -6.52614236e-01 6.46100461e-01
4.55714166e-01 1.94706798e-01 1.58098531e+00 -7.39846230e-01
-9.18262839e-01 -2.44902685e-01 -1.11055449e-02 -4.45139170e-01
-6.26403809e-01 4.24380809e-01 -8.95154119e-01 -8.57793808e-01
-1.91603065e-01 -8.99235427e-01 8.13018084e-01 2.83577383e-01
-8.73833716e-01 -3.97756398e-01 1.54873705e+00 -1.07597828e+00
2.13523769e+00 -2.42836833e+00 -6.44449949e-01 2.11821094e-01
4.78920907e-01 7.35551417e-01 2.21757844e-01 7.03033507e-01
1.24917023e-01 6.95816875e-01 7.43065655e-01 3.69222045e-01
5.64088523e-02 -4.11933482e-01 -6.33391678e-01 4.57453370e-01
-3.05203050e-01 8.09984267e-01 -8.04903448e-01 -5.36972880e-01
-1.61191866e-01 1.18430220e-01 -5.83290100e-01 2.25965172e-01
-3.56993049e-01 -1.50383309e-01 -8.57376099e-01 8.30627382e-01
7.58776426e-01 -5.42107105e-01 -1.05202213e-01 -2.59920746e-01
-4.16557640e-01 4.55915064e-01 -7.77958393e-01 9.67037082e-01
5.89904562e-02 1.31827760e+00 7.91744366e-02 -2.67005563e-01
2.88019896e-01 3.02369595e-01 2.54862875e-01 -7.61734307e-01
8.27673972e-02 2.61395454e-01 -4.28584903e-01 -1.01583982e+00
4.97195870e-01 8.19724619e-01 -4.60823961e-02 4.29798335e-01
-2.66224772e-01 5.45227945e-01 6.01041913e-01 6.98523462e-01
1.32689607e+00 -6.62351102e-02 2.83914775e-01 1.48935944e-01
6.06905043e-01 4.91965562e-03 1.91188812e-01 1.16803217e+00
-7.37884581e-01 5.92614710e-01 7.03701735e-01 -3.40856940e-01
-1.26427495e+00 -9.65171278e-01 1.27597839e-01 1.55969238e+00
1.34708136e-01 -9.62797046e-01 -9.72866058e-01 -8.97044659e-01
2.08064184e-01 4.39175844e-01 -3.89706522e-01 2.06861585e-01
-5.26781023e-01 1.05475575e-01 4.88731563e-01 3.23558390e-01
9.96138692e-01 -4.77046251e-01 -3.82793933e-01 4.75009046e-02
-4.76730943e-01 -1.28570354e+00 -1.26423538e+00 -8.15611243e-01
-2.54200369e-01 -1.32823694e+00 -7.12134123e-01 -5.02708256e-01
3.14356089e-01 8.66245389e-01 7.13829100e-01 4.74964887e-01
-6.70902610e-01 5.54480493e-01 -5.01454711e-01 -3.90882120e-02
-1.33810341e-01 -5.32618649e-02 -4.28345561e-01 5.05500674e-01
6.52933478e-01 2.06683710e-01 -8.96262467e-01 7.49142408e-01
-1.18500876e+00 -4.19337749e-01 -1.30481943e-01 3.46078992e-01
1.52131673e-02 1.63334116e-01 6.07104339e-02 -8.18928659e-01
6.89362943e-01 -7.76260078e-01 -5.10952592e-01 3.01958621e-01
-1.14895739e-01 -7.47724354e-01 3.74337494e-01 -6.28313065e-01
-7.84198999e-01 -2.59221852e-01 1.15613356e-01 -4.61255670e-01
-1.13788553e-01 2.84650803e-01 5.83288558e-02 -2.52647195e-02
4.39722836e-01 3.01199943e-01 -2.83181161e-01 -7.10875928e-01
-8.61172155e-02 1.12947929e+00 1.92332610e-01 1.61913246e-01
8.92423153e-01 5.06430984e-01 -7.27907121e-01 -9.08487320e-01
-9.79906738e-01 -1.13315725e+00 -1.56055346e-01 -6.76817715e-01
8.29739213e-01 -9.82195914e-01 -9.35438991e-01 5.85646749e-01
-1.14264357e+00 5.44990718e-01 5.29436171e-01 2.31031507e-01
-1.50252655e-01 8.77621353e-01 -7.54507899e-01 -7.19818711e-01
6.25843331e-02 -8.41488898e-01 9.66401160e-01 2.54391164e-01
-2.80375302e-01 -5.64719081e-01 -2.19067171e-01 9.76344109e-01
5.20752430e-01 -1.58369735e-01 5.82629859e-01 -1.12288523e+00
-1.16056681e+00 -9.56912935e-01 -5.67983091e-01 3.82328987e-01
3.82935256e-02 4.34308231e-01 -7.70520687e-01 -2.31622979e-01
-1.59195945e-01 -3.04684967e-01 6.67497158e-01 -1.05828561e-01
1.43042696e+00 -7.60194480e-01 -6.01782143e-01 1.82758290e-02
1.09337676e+00 1.86907575e-01 7.32601583e-01 4.92795974e-01
4.41054046e-01 2.52755195e-01 7.89766729e-01 6.39407456e-01
-1.11458153e-02 8.16696227e-01 5.67999661e-01 6.74381316e-01
-1.18347369e-02 -7.83314824e-01 5.80303967e-01 4.73714143e-01
1.93195492e-01 -4.94859248e-01 -6.48889899e-01 3.02177310e-01
-1.68623316e+00 -1.10079205e+00 -1.71922430e-01 2.18215370e+00
3.81058395e-01 3.09589833e-01 4.56224054e-01 1.11535475e-01
1.17091310e+00 5.15324831e-01 -3.62434238e-01 5.05543686e-02
3.60273778e-01 -3.44710827e-01 6.45781875e-01 -1.54632047e-01
-1.33077466e+00 6.15874231e-01 5.72418928e+00 1.24221265e+00
-1.00171936e+00 2.64483750e-01 2.91924149e-01 -3.37141097e-01
5.69689795e-02 -2.46139258e-01 -1.13432598e+00 1.39511752e+00
5.54904044e-01 6.74247891e-02 4.09765124e-01 1.21603692e+00
5.14626980e-01 -2.73229748e-01 -1.00655222e+00 1.09973192e+00
4.91490483e-01 -1.54540849e+00 3.76445195e-03 8.67593959e-02
4.61439073e-01 -1.83385074e-01 2.67944127e-01 3.60340148e-01
-4.75426942e-01 -3.59934330e-01 9.74691153e-01 -5.12153171e-02
3.32770973e-01 -3.52756493e-02 3.37940484e-01 9.93194580e-02
-9.98431027e-01 -2.87448734e-01 3.99262942e-02 3.80073011e-01
3.22453499e-01 3.60227853e-01 -4.75424618e-01 -2.51052111e-01
1.06780684e+00 9.35955048e-01 -1.11557710e+00 1.62428486e+00
-9.20894071e-02 8.27676594e-01 -1.20641641e-01 -4.00410265e-01
2.62091637e-01 2.45555788e-01 8.72115552e-01 9.99315798e-01
3.35102856e-01 -3.71373177e-01 1.23152800e-01 6.99423790e-01
-2.49543577e-01 3.15242186e-02 -4.39180136e-01 -8.44964385e-01
4.30399835e-01 9.93915796e-01 -5.02211452e-01 -4.74694848e-01
-8.87388885e-01 9.69697297e-01 -2.25927513e-02 3.58271599e-01
-1.29200590e+00 -6.21716559e-01 5.88727951e-01 8.82186413e-01
9.46600437e-01 -1.35741234e-01 5.43109953e-01 -1.31037056e+00
5.68742335e-01 -9.07309353e-01 3.10558826e-01 -1.00694561e+00
-1.25480151e+00 3.15071464e-01 -1.84643026e-02 -1.56293380e+00
2.09985510e-01 -4.77857590e-01 -7.01357782e-01 9.25568417e-02
-1.20663714e+00 -8.29926670e-01 -6.97298110e-01 4.83415186e-01
8.27717543e-01 1.97165795e-02 -3.46190065e-01 9.28935409e-01
-5.37770927e-01 4.87536520e-01 2.88421027e-02 5.33722162e-01
6.94650292e-01 -5.91449559e-01 1.85935587e-01 7.57337153e-01
-7.79215917e-02 7.91221797e-01 7.56585717e-01 -1.10446024e+00
-1.56220067e+00 -8.81074071e-01 8.42399359e-01 -4.31174606e-01
1.50716186e+00 -4.47980881e-01 -7.79661894e-01 6.69283628e-01
2.66877860e-01 -9.48941633e-02 8.39919448e-01 -3.11777890e-01
-6.88307822e-01 -1.24173842e-01 -6.53988123e-01 6.60586774e-01
7.48196065e-01 -8.40948403e-01 -6.06787384e-01 6.88693762e-01
2.56468743e-01 -3.90723646e-02 -4.15102154e-01 -1.93279639e-01
8.48526835e-01 -1.03515816e+00 8.71087670e-01 -7.96653092e-01
3.78525615e-01 -1.70481533e-01 -1.22404099e-02 -6.83357894e-01
-6.04689531e-02 -7.52125621e-01 -2.44967684e-01 1.52118969e+00
5.56148849e-02 -2.39095375e-01 9.23103571e-01 8.05572093e-01
2.35347539e-01 4.74391654e-02 -1.05099046e+00 -9.24529195e-01
-8.20723236e-01 -3.53382826e-01 2.52531797e-01 4.10692871e-01
7.34941438e-02 2.66655460e-02 -7.97986746e-01 -2.24961668e-01
6.03256524e-01 -1.42761230e-01 8.81091237e-01 -8.36612761e-01
-1.75317109e-01 -3.16234916e-01 -4.86218899e-01 -1.57061958e+00
-3.15867126e-01 -4.63520020e-01 -7.92622492e-02 -8.19987774e-01
3.43994856e-01 -4.34530759e-03 -6.76375674e-03 -4.67612920e-03
-3.75204510e-03 5.05078495e-01 4.98837888e-01 5.32573164e-01
-1.23378682e+00 -5.64201735e-02 8.98326278e-01 -3.55236351e-01
2.97687091e-02 1.23965316e-01 -2.22637355e-01 8.22165012e-01
4.78791803e-01 -5.86062431e-01 2.80186720e-02 -1.02331312e-02
6.63544655e-01 -1.76963508e-01 5.35481632e-01 -8.12307000e-01
3.05450827e-01 5.72990738e-02 1.27893642e-01 -9.38368499e-01
2.07415119e-01 -8.83685112e-01 -8.38126987e-02 3.85719270e-01
-5.43058693e-01 -9.22127739e-02 -2.51118869e-01 1.10624313e+00
-3.33606362e-01 -6.48142993e-01 4.39750493e-01 -1.46554202e-01
-7.41495967e-01 1.14030018e-01 -1.04029763e+00 2.13480800e-01
1.26529217e+00 -3.41055393e-01 -9.28841412e-01 -4.73077804e-01
-4.00497854e-01 2.33299375e-01 1.82223484e-01 8.13374043e-01
4.96081978e-01 -1.22913897e+00 -1.99532822e-01 2.51647055e-01
3.36764038e-01 -1.12150955e+00 3.18310857e-01 7.73406506e-01
-8.29871416e-01 8.14810932e-01 -1.01119839e-02 -5.18613279e-01
-1.72799361e+00 7.92721272e-01 -1.94432124e-01 2.70963460e-01
-3.70193064e-01 5.12350857e-01 9.37252045e-02 5.54185927e-01
5.72209954e-01 -3.09625119e-01 8.28994662e-02 3.49412054e-01
1.10692465e+00 8.42933834e-01 -1.49314910e-01 -4.26624179e-01
-3.67405027e-01 1.01686507e-01 -5.68230689e-01 4.13317800e-01
7.19258726e-01 -4.58980143e-01 1.42432556e-01 1.48567379e-01
1.59684074e+00 2.86108613e-01 -1.24977684e+00 -4.06577885e-02
2.46412203e-01 -1.34333932e+00 2.44448278e-02 -4.26071584e-01
-1.10924184e+00 6.58178985e-01 2.01859623e-01 9.09903646e-01
5.37265480e-01 1.66270509e-01 1.43422687e+00 8.94903317e-02
2.51539230e-01 -1.59701777e+00 6.61949575e-01 -6.67527318e-02
8.29399049e-01 -1.41346979e+00 2.59784818e-01 -7.08622932e-01
-4.27324682e-01 1.08002782e+00 6.38560772e-01 -1.71649769e-01
8.68260145e-01 -3.21486264e-01 -2.61111408e-01 -7.42967501e-02
-6.44920528e-01 2.59439260e-01 4.60345834e-01 -3.81912962e-02
-2.75865421e-02 -1.25372604e-01 -3.92343462e-01 3.15776318e-01
3.85100156e-01 7.96516687e-02 6.07488513e-01 1.10597348e+00
-7.34636724e-01 -7.42773175e-01 -4.45013076e-01 8.06790292e-01
-1.15487802e+00 -8.96261483e-02 -5.61427116e-01 8.10025036e-01
-1.78721681e-01 8.82225692e-01 2.22988397e-01 -2.98423469e-01
-1.45799473e-01 8.26344267e-02 -2.74752472e-02 -3.52014482e-01
-6.54005706e-01 4.06058460e-01 2.15779826e-01 -9.20378804e-01
-2.58505076e-01 -7.36028671e-01 -2.80534595e-01 -2.83592701e-01
-4.48865056e-01 2.02321485e-01 8.26751351e-01 7.02164173e-01
5.02210915e-01 -2.27444366e-01 6.00734472e-01 -3.68944168e-01
-3.85098130e-01 -5.84545791e-01 -5.45615971e-01 6.04398310e-01
5.23552179e-01 -6.96402788e-01 -7.83177018e-01 1.25253692e-01] | [7.733821868896484, 9.7260160446167] |
e37e4031-e1b0-4a89-9b17-66c1b3dd5ced | m3ke-a-massive-multi-level-multi-subject | 2305.10263 | null | https://arxiv.org/abs/2305.10263v2 | https://arxiv.org/pdf/2305.10263v2.pdf | M3KE: A Massive Multi-Level Multi-Subject Knowledge Evaluation Benchmark for Chinese Large Language Models | Large language models have recently made tremendous progress in a variety of aspects, e.g., cross-task generalization, instruction following. Comprehensively evaluating the capability of large language models in multiple tasks is of great importance. In this paper, we propose M3KE, a Massive Multi-Level Multi-Subject Knowledge Evaluation benchmark, which is developed to measure knowledge acquired by Chinese large language models by testing their multitask accuracy in zero- and few-shot settings. We have collected 20,477 questions from 71 tasks. Our selection covers all major levels of Chinese education system, ranging from the primary school to college, as well as a wide variety of subjects, including humanities, history, politics, law, education, psychology, science, technology, art and religion. All questions are multiple-choice questions with four options, hence guaranteeing a standardized and unified assessment process. We've assessed a number of state-of-the-art open-source Chinese large language models on the proposed benchmark. The size of these models varies from 335M to 130B parameters. Experiment results demonstrate that they perform significantly worse than GPT-3.5 that reaches an accuracy of ~ 48% on M3KE. The dataset is available at https://github.com/tjunlp-lab/M3KE. | ['Deyi Xiong', 'Qun Liu', 'Xiaowen Su', 'Qingqing Lyu', 'Peiyi Zhang', 'Jianxiang Peng', 'Shuting Zhang', 'Xiaohan Peng', 'Tianyu Dong', 'Linhao Yu', 'Yuqi Ren', 'Renren Jin', 'Chuang Liu'] | 2023-05-17 | null | null | null | null | ['multiple-choice-qa', 'instruction-following'] | ['natural-language-processing', 'natural-language-processing'] | [-6.41936421e-01 -5.35781682e-01 -2.18911752e-01 -6.55868500e-02
-1.20329309e+00 -7.04921722e-01 3.10703337e-01 -3.25020310e-03
-7.86395967e-01 9.82029319e-01 7.37814531e-02 -4.65989560e-01
-1.26229748e-01 -6.42161369e-01 -6.43610835e-01 -3.53477061e-01
2.89074183e-01 3.13137680e-01 4.42967325e-01 -3.85450006e-01
3.01486254e-01 8.40965584e-02 -1.17397749e+00 3.87834340e-01
1.54773247e+00 6.77117765e-01 2.77241498e-01 5.70643246e-01
-3.28507096e-01 6.47721887e-01 -7.37151325e-01 -8.22468877e-01
-3.82475406e-01 -2.10006028e-01 -1.05726731e+00 -3.87307256e-01
4.92867887e-01 1.07812665e-01 -4.69098873e-02 1.12112999e+00
8.27073097e-01 4.18150663e-01 3.60245496e-01 -9.04043078e-01
-1.37002754e+00 7.49627948e-01 -4.59980935e-01 4.33201820e-01
3.71177316e-01 -5.22094173e-03 8.39072287e-01 -1.33307016e+00
2.67888010e-01 1.49849498e+00 3.98041785e-01 5.83986878e-01
-8.78583252e-01 -1.11417496e+00 1.70215905e-01 4.25747097e-01
-1.75419152e+00 -2.90288657e-01 2.89359719e-01 -4.76596951e-01
9.21561897e-01 1.56525634e-02 2.44319558e-01 9.74107146e-01
3.57963979e-01 1.04249012e+00 1.47786701e+00 -5.13122797e-01
-1.22877821e-01 4.67311680e-01 5.09259105e-01 7.26240218e-01
9.00758952e-02 -5.33398628e-01 -7.49344409e-01 -1.15492558e-02
4.50841308e-01 -2.30403036e-01 -3.54171097e-01 2.16919243e-01
-1.34688330e+00 7.47351646e-01 3.53067219e-02 6.47661507e-01
6.21926151e-02 -1.03455409e-01 1.80984154e-01 5.73554635e-01
5.25522172e-01 3.77828449e-01 -1.04553270e+00 -3.32910180e-01
-4.87984985e-01 1.80242285e-01 7.24291682e-01 1.29882240e+00
6.67845070e-01 -6.78910092e-02 -4.39988405e-01 1.07434011e+00
5.19551411e-02 6.14060462e-01 1.05371928e+00 -5.35944104e-01
7.34867454e-01 6.99109077e-01 -1.90510660e-01 -5.83609879e-01
-2.72319674e-01 -4.58315909e-01 -9.86789048e-01 -4.93247569e-01
2.42451683e-01 -2.82694846e-01 -5.02173364e-01 1.79338992e+00
1.43795818e-01 4.27557617e-01 2.72878021e-01 4.93852109e-01
1.36374819e+00 6.15308464e-01 4.58409101e-01 -3.11936941e-02
1.64053869e+00 -1.14891291e+00 -7.49422669e-01 -2.06314772e-01
9.14717674e-01 -1.07384074e+00 1.72372949e+00 4.99748588e-01
-1.16212690e+00 -8.33243310e-01 -5.96645534e-01 -3.27886671e-01
-9.33479011e-01 2.55738080e-01 3.65880430e-01 8.35680008e-01
-1.01511514e+00 1.39994532e-01 -2.56263822e-01 -2.94348866e-01
1.12660259e-01 -6.72668964e-02 -1.70268178e-01 -2.87427276e-01
-1.69715834e+00 9.46484864e-01 6.60887420e-01 -2.83356458e-01
-7.55288005e-01 -9.33649361e-01 -6.61617100e-01 3.51348162e-01
6.03201270e-01 -2.49414757e-01 1.31032228e+00 -1.55475408e-01
-1.41799760e+00 9.46529269e-01 -1.84039384e-01 1.25988841e-01
3.13211203e-01 -6.22641385e-01 -7.91426003e-01 -3.02654922e-01
1.47408128e-01 3.98572475e-01 -1.59463789e-02 -6.29987895e-01
-5.41884840e-01 -1.96112335e-01 -3.67708243e-02 1.10490635e-01
-8.88267398e-01 3.73779595e-01 -8.12226593e-01 -6.13701403e-01
-3.77249241e-01 -8.81898761e-01 -6.39419258e-02 -7.34319031e-01
-2.71868825e-01 -7.55024970e-01 5.35698414e-01 -6.87410116e-01
1.80339289e+00 -1.86658061e+00 -1.53649583e-01 -1.93311021e-01
-9.55401175e-03 5.40501356e-01 -2.95375109e-01 2.98601836e-01
2.97912300e-01 5.97111523e-01 1.75281286e-01 -3.05698365e-01
5.77337295e-03 -1.50745556e-01 -1.03046626e-01 3.04771233e-02
-9.14452225e-02 1.31505001e+00 -9.11797404e-01 -6.90133393e-01
-1.82116494e-01 1.92729816e-01 -2.72769898e-01 9.60365906e-02
-1.26130074e-01 2.56301075e-01 -8.47708404e-01 7.19470620e-01
5.07006705e-01 -6.57697141e-01 -2.02192411e-01 4.80746686e-01
-1.04845956e-01 3.03207934e-01 -1.10524511e+00 2.08688903e+00
-6.67886853e-01 5.73307931e-01 -2.40090311e-01 -5.85338950e-01
9.24813867e-01 5.52081704e-01 -3.53002734e-03 -6.83794320e-01
1.20000862e-01 2.20659107e-01 1.87587872e-01 -4.73830491e-01
7.08187342e-01 2.13409394e-01 -4.19935286e-01 4.52769518e-01
3.03154796e-01 -1.36431204e-02 3.88350964e-01 3.52189004e-01
6.98408306e-01 -3.69776152e-02 3.81159782e-01 -7.82621264e-01
9.64882493e-01 -2.62246996e-01 4.47987586e-01 8.45174968e-01
-2.99291968e-01 8.99897888e-02 1.27806351e-01 5.30687235e-02
-3.84245664e-01 -1.01133323e+00 -2.38279730e-01 1.79553425e+00
-5.26598468e-02 -4.78811681e-01 -6.58249974e-01 -5.24586022e-01
-5.09745069e-02 8.66658151e-01 -3.93685967e-01 -3.00169766e-01
-4.36785132e-01 -7.33573079e-01 6.99868739e-01 2.85355747e-01
9.24196899e-01 -1.28166032e+00 -1.70665875e-01 1.24446787e-01
-4.25368488e-01 -1.33934426e+00 -7.70111561e-01 -1.59861550e-01
-5.42380273e-01 -8.35081935e-01 -7.85498500e-01 -1.14456296e+00
1.94122776e-01 4.03822601e-01 1.51348364e+00 9.48160589e-02
-5.46407588e-02 6.31239355e-01 -4.86188591e-01 -8.04956138e-01
-3.10136050e-01 4.95130688e-01 1.61703065e-01 -3.50452393e-01
9.31669652e-01 -1.50095105e-01 -1.34319231e-01 4.96452063e-01
-6.42515421e-01 -8.20817947e-02 7.58321822e-01 5.97732127e-01
5.20946205e-01 1.16832204e-01 1.04233336e+00 -8.26401770e-01
1.07748091e+00 -5.52958310e-01 -6.06308162e-01 1.02454734e+00
-5.60623467e-01 -2.49998033e-01 6.38768494e-01 -8.00738871e-01
-1.23076046e+00 -7.08661973e-01 -7.85170346e-02 -1.88328758e-01
-2.78219461e-01 9.91715312e-01 -2.63950914e-01 -1.66308716e-01
5.56600988e-01 5.63467920e-01 -6.54308558e-01 -7.49290347e-01
2.54483372e-01 8.65491033e-01 4.65454847e-01 -1.19429183e+00
4.23046559e-01 -2.44948208e-01 -5.38850427e-01 -9.41324294e-01
-1.25009561e+00 -6.26953542e-01 -6.67083681e-01 -2.07390502e-01
8.09412658e-01 -1.23445773e+00 -1.04975212e+00 6.55773759e-01
-9.73243952e-01 -5.38568020e-01 3.01963896e-01 6.18084371e-01
1.13382496e-01 1.94651797e-01 -8.25320780e-01 -4.73011434e-01
-5.22988796e-01 -1.25410700e+00 7.32500434e-01 5.10902405e-01
2.77348250e-01 -1.25636792e+00 -9.88897234e-02 6.88200116e-01
5.04694879e-01 -3.78117293e-01 1.00004888e+00 -8.99864256e-01
-5.66057920e-01 3.12799774e-02 6.97311712e-03 4.67151761e-01
-9.16603729e-02 -3.60194027e-01 -9.41865563e-01 -4.42844272e-01
-2.06322163e-01 -1.07352030e+00 8.95878732e-01 1.10438652e-01
1.55938649e+00 4.23687100e-02 -8.15038532e-02 3.19576919e-01
1.33457530e+00 1.83539510e-01 5.41626334e-01 4.23479885e-01
6.71880960e-01 2.90493280e-01 4.87249464e-01 5.12400195e-02
7.04110980e-01 5.16938925e-01 -4.03354049e-01 3.63781810e-01
-6.58204630e-02 -9.01418179e-02 5.91323435e-01 1.53392029e+00
1.05080999e-01 -2.84411281e-01 -1.34299684e+00 6.63400710e-01
-1.49280834e+00 -7.39829004e-01 -1.83100820e-01 2.20644784e+00
1.33547604e+00 1.66350439e-01 -4.17554796e-01 -5.05816996e-01
4.69170243e-01 -1.58368781e-01 -4.05251414e-01 -3.09821546e-01
-3.15708399e-01 2.56186575e-01 1.12318367e-01 5.16126633e-01
-9.32316720e-01 1.46954954e+00 5.56585836e+00 1.57816219e+00
-8.84785235e-01 4.34124619e-01 7.21879303e-01 7.66583681e-02
-1.71346173e-01 -2.27860421e-01 -1.45010436e+00 4.29774731e-01
1.04264271e+00 -7.72432923e-01 1.03302568e-01 7.57010460e-01
-4.28069122e-02 -1.65487140e-01 -6.63009048e-01 8.02532196e-01
1.45260543e-01 -1.08088851e+00 2.09541440e-01 -7.91878626e-02
1.28191733e+00 1.90191135e-01 1.79823175e-01 1.09341979e+00
5.12343228e-01 -9.99769151e-01 4.42364454e-01 4.77556467e-01
7.33274996e-01 -7.28834093e-01 5.03403544e-01 8.30363989e-01
-1.23684990e+00 1.37397513e-01 -4.85418797e-01 -5.59607521e-02
-1.75065413e-01 2.91379064e-01 -3.69348049e-01 5.78557312e-01
9.23311114e-01 4.30479735e-01 -9.67372119e-01 8.19976985e-01
-4.12983090e-01 1.01861441e+00 9.00871456e-02 -4.14710909e-01
2.74585724e-01 -1.79468900e-01 -9.87756401e-02 1.32638109e+00
4.39772159e-01 3.80190909e-01 4.93213147e-01 5.68664074e-01
-3.92212510e-01 5.34574509e-01 -2.12060899e-01 8.43228996e-02
7.09320009e-01 1.23060429e+00 -3.13099086e-01 -5.79560280e-01
-9.06798065e-01 5.99544764e-01 6.55536294e-01 5.91924131e-01
-7.10888386e-01 -3.38832170e-01 4.48257476e-01 -5.59534013e-01
-9.70684886e-02 -1.31692573e-01 -1.77067697e-01 -1.51963437e+00
-7.85242468e-02 -1.02377355e+00 6.98516130e-01 -7.37777412e-01
-1.37957633e+00 3.24741602e-01 2.67553367e-02 -9.07322168e-01
1.27037361e-01 -6.27717257e-01 -6.40735626e-01 1.27778852e+00
-1.65320587e+00 -1.17954612e+00 -3.32067609e-01 9.32850122e-01
7.83498704e-01 -5.43293297e-01 8.24366927e-01 5.84110022e-01
-9.05398130e-01 1.01708877e+00 3.39522064e-01 3.91857237e-01
1.01195931e+00 -1.13642752e+00 3.14999402e-01 7.74134994e-01
5.79913147e-02 8.87290537e-01 1.35424539e-01 -6.48300588e-01
-1.12721395e+00 -1.07373023e+00 1.35118032e+00 -7.03030765e-01
8.06688011e-01 -4.83327329e-01 -1.32763290e+00 7.52064049e-01
3.11172545e-01 -2.48644277e-01 1.00162506e+00 4.34206873e-01
-1.05825260e-01 -8.98840185e-03 -6.42607868e-01 6.29969835e-01
6.92936957e-01 -5.52775562e-01 -5.90273559e-01 3.93922031e-01
9.15628016e-01 -4.83455539e-01 -1.35117304e+00 1.95015058e-01
3.71249884e-01 -6.29841745e-01 9.97215152e-01 -1.00592887e+00
4.16372389e-01 2.59786278e-01 8.06914060e-04 -1.33629692e+00
-5.69349468e-01 -2.73946643e-01 -1.20218664e-01 1.42504764e+00
6.06570244e-01 -6.91431940e-01 6.10168517e-01 4.63736594e-01
-3.82697195e-01 -9.82202530e-01 -8.76709640e-01 -9.50684726e-01
9.30821240e-01 -4.31464463e-01 5.41224599e-01 1.43653691e+00
-7.83099607e-02 5.80249548e-01 -3.33183438e-01 1.12500973e-03
4.22941446e-01 1.87781185e-01 6.79682076e-01 -1.23221695e+00
3.27907503e-02 -7.64234483e-01 1.70452252e-01 -1.26905715e+00
4.56562400e-01 -9.65954423e-01 -4.62505311e-01 -1.42353356e+00
4.38501716e-01 -2.72175461e-01 -5.54013550e-01 4.93426830e-01
-8.28203678e-01 -1.80536270e-01 -1.97126810e-02 -9.56262574e-02
-1.03635657e+00 6.89200819e-01 1.69836605e+00 -2.52327342e-02
-6.37093782e-02 -1.24029420e-01 -8.32610071e-01 7.50688612e-01
1.02579117e+00 -2.00455278e-01 -4.03854191e-01 -8.27596545e-01
6.66357055e-02 -5.04449271e-02 -5.60192615e-02 -8.86438906e-01
5.09307981e-01 -4.14280921e-01 3.78327519e-01 -4.55946594e-01
1.20946415e-01 -4.58107263e-01 -5.75161874e-01 4.15377855e-01
-5.46068072e-01 3.40394557e-01 4.74947870e-01 9.79858935e-02
-3.89209002e-01 -1.63334489e-01 7.32585728e-01 -5.27661026e-01
-1.35576272e+00 4.44351584e-01 -9.61351842e-02 7.39443660e-01
8.31228793e-01 2.68937916e-01 -6.19626105e-01 -1.71721891e-01
-5.28706133e-01 8.00517201e-01 -2.05872372e-01 8.91874373e-01
6.11610234e-01 -1.47232366e+00 -1.08874547e+00 -6.02541566e-02
3.70419413e-01 -1.42106146e-01 5.75682759e-01 7.73706257e-01
-3.88592072e-02 1.04312301e+00 1.32998705e-01 -2.83220172e-01
-1.28132975e+00 4.02103424e-01 1.57068938e-01 -6.51878238e-01
-7.80764744e-02 1.02279556e+00 1.46126211e-01 -8.32146168e-01
2.03178883e-01 -3.17096561e-01 -4.77002710e-01 -4.13667671e-02
7.15806901e-01 5.56488216e-01 1.06229708e-01 -4.82273161e-01
-1.83884725e-01 6.28253937e-01 -2.12935269e-01 1.73824102e-01
8.29590619e-01 -1.19753569e-01 -1.04356460e-01 6.10886455e-01
1.16712642e+00 1.30100265e-01 -3.64298910e-01 -6.86295509e-01
2.33207762e-01 -2.05767334e-01 -6.43037558e-02 -1.07443452e+00
-6.97533846e-01 8.77590537e-01 3.24653357e-01 -1.10579088e-01
1.02837276e+00 6.90324232e-02 7.41855085e-01 8.96764159e-01
6.12297237e-01 -1.22697556e+00 3.35681587e-01 1.09159064e+00
8.07849824e-01 -1.63090646e+00 -3.71840298e-02 -1.47807717e-01
-5.89376092e-01 7.14986980e-01 1.34996164e+00 4.84198660e-01
8.47926736e-01 -1.27447993e-01 2.82670707e-01 -1.03069372e-01
-1.05932260e+00 -1.90069959e-01 7.51330733e-01 -4.12520021e-04
1.08386624e+00 1.94041267e-01 -4.61100847e-01 9.35452819e-01
-3.81112546e-01 -1.98324211e-02 5.76520622e-01 8.00403595e-01
-7.04395235e-01 -9.61354136e-01 -4.38604623e-01 4.26968902e-01
-7.30997443e-01 -4.70003694e-01 -1.51593328e-01 9.68487501e-01
6.40952736e-02 1.07718945e+00 -3.66748929e-01 -2.05008537e-01
3.71764332e-01 3.54854494e-01 1.41548038e-01 -6.63206458e-01
-6.60863161e-01 -6.02301769e-02 -1.16509281e-01 -1.94846481e-01
-1.34373531e-01 -3.94824356e-01 -9.41138327e-01 -3.58613282e-01
-3.56426924e-01 5.03859401e-01 3.76086175e-01 8.86540830e-01
2.33343676e-01 4.86869335e-01 3.81196551e-02 -6.19952567e-02
-6.49521947e-01 -1.41805303e+00 -5.62470376e-01 2.45672807e-01
-4.17970978e-02 -5.63270986e-01 -1.00537345e-01 -1.49993941e-01] | [10.70329475402832, 8.774886131286621] |
df8b0144-c252-4aec-922f-7878c12061be | dry-focus-and-transcribe-end-to-end | 1904.09049 | null | http://arxiv.org/abs/1904.09049v3 | http://arxiv.org/pdf/1904.09049v3.pdf | An Investigation of End-to-End Multichannel Speech Recognition for Reverberant and Mismatch Conditions | Sequence-to-sequence (S2S) modeling is becoming a popular paradigm for
automatic speech recognition (ASR) because of its ability to jointly optimize
all the conventional ASR components in an end-to-end (E2E) fashion. This report
investigates the ability of E2E ASR from standard close-talk to far-field
applications by encompassing entire multichannel speech enhancement and ASR
components within the S2S model. There have been previous studies on jointly
optimizing neural beamforming alongside E2E ASR for denoising. It is clear from
both recent challenge outcomes and successful products that far-field systems
would be incomplete without solving both denoising and dereverberation
simultaneously. This report uses a recently developed architecture for
far-field ASR by composing neural extensions of dereverberation and beamforming
modules with the S2S ASR module as a single differentiable neural network and
also clearly defining the role of each subnetwork. The original implementation
of this architecture was successfully applied to the noisy speech recognition
task (CHiME-4), while we applied this implementation to noisy reverberant tasks
(DIRHA and REVERB). Our investigation shows that the method achieves better
performance than conventional pipeline methods on the DIRHA English dataset and
comparable performance on the REVERB dataset. It also has additional advantages
of being neither iterative nor requiring parallel noisy and clean speech data. | ['Shinji Watanabe', 'Dung Tran', 'Aswin Shanmugam Subramanian', 'Toru Taniguchi', 'Yuya Fujita', 'Xiaofei Wang'] | 2019-04-19 | null | null | null | null | ['noisy-speech-recognition'] | ['speech'] | [ 2.03406155e-01 -2.16917440e-01 8.50088775e-01 -6.26790583e-01
-1.48735452e+00 -5.55064082e-01 4.18140858e-01 -4.66270864e-01
-5.72786808e-01 4.55477893e-01 6.40837193e-01 -6.78523362e-01
-1.42201185e-01 7.98630640e-02 -6.25547647e-01 -6.73081458e-01
-1.17897578e-01 -2.04899758e-02 -1.66740939e-01 -7.55422533e-01
-3.46683830e-01 5.55550933e-01 -1.29842019e+00 5.05709648e-01
4.21039611e-01 8.35265160e-01 6.17388070e-01 1.45171440e+00
4.35383618e-01 8.43759537e-01 -1.12621868e+00 -8.84714723e-02
2.83723056e-01 -1.99776188e-01 -4.15243715e-01 -1.34132475e-01
5.98740578e-01 -3.41188610e-01 -6.63507819e-01 7.46426105e-01
1.30296171e+00 6.42597139e-01 9.80313793e-02 -5.91682553e-01
-5.39545536e-01 7.19415903e-01 -2.05972135e-01 5.90510607e-01
4.40616697e-01 2.78974891e-01 9.28124845e-01 -1.22150910e+00
1.13476187e-01 1.45411325e+00 9.97202337e-01 4.10499990e-01
-1.17483854e+00 -6.33655965e-01 7.49181211e-02 2.85037518e-01
-1.32108867e+00 -1.30337059e+00 4.51241612e-01 -8.35337490e-02
1.76452708e+00 5.47843874e-01 1.94017559e-01 1.31834006e+00
-2.72325397e-01 8.91411364e-01 1.03910840e+00 -2.66820222e-01
1.36085868e-01 -1.69094265e-01 2.71312118e-01 5.08362390e-02
-4.52770412e-01 5.67852199e-01 -8.50914657e-01 2.06115022e-01
2.27272943e-01 -6.70245647e-01 -6.66634083e-01 8.24353099e-01
-9.99644816e-01 3.45309556e-01 3.56479436e-01 5.15822351e-01
-4.20878917e-01 3.76014411e-01 3.21872681e-01 6.92736089e-01
8.31114888e-01 2.25599945e-01 -7.96399653e-01 -3.03918093e-01
-1.36879313e+00 1.57949299e-01 7.60317147e-01 6.16568089e-01
3.22490185e-01 8.35491538e-01 -3.96989495e-01 1.40480840e+00
5.19500136e-01 5.79488575e-01 3.96118432e-01 -7.86775470e-01
6.53869987e-01 -7.32274890e-01 -3.45810084e-03 -4.86403167e-01
-3.25589210e-01 -1.02038157e+00 -8.73692274e-01 2.67113775e-01
1.20719798e-01 -4.48119342e-01 -1.09813881e+00 1.78062785e+00
-4.45179082e-02 2.99408287e-01 1.65926322e-01 9.50890303e-01
9.80460525e-01 1.08540273e+00 -8.93293694e-02 -7.39559904e-02
1.02746665e+00 -9.29063916e-01 -1.19954860e+00 -4.11192328e-01
2.90108055e-01 -1.23572636e+00 6.95947707e-01 6.29899561e-01
-1.37567997e+00 -8.21132004e-01 -1.07058930e+00 -3.34175617e-01
-2.61022002e-01 2.53524601e-01 6.23468123e-02 1.00674212e+00
-1.84723103e+00 5.36677063e-01 -8.39534163e-01 -9.24178306e-03
1.28925964e-01 2.58839786e-01 -4.01744813e-01 -1.02367423e-01
-1.28911126e+00 1.12302721e+00 1.63008794e-01 7.45380998e-01
-1.02834046e+00 -9.16802585e-01 -9.45447147e-01 2.98773438e-01
1.42405882e-01 -3.17935288e-01 1.68879008e+00 -6.85118735e-01
-1.82822382e+00 3.24431866e-01 -4.53393966e-01 -6.05486929e-01
3.43329877e-01 -5.86615980e-01 -8.54045033e-01 -2.16465667e-01
-5.00666082e-01 3.41664404e-01 9.42073047e-01 -1.08151996e+00
-3.56720179e-01 -2.34760135e-01 -4.59496230e-01 3.39344710e-01
-3.59962881e-02 4.75894272e-01 -1.02872998e-01 -1.14151931e+00
4.67565954e-02 -3.84321868e-01 -1.38535962e-01 -5.49601376e-01
-3.46289188e-01 -3.88149694e-02 8.51397872e-01 -1.67646253e+00
1.26538289e+00 -2.25089264e+00 3.11837494e-02 2.58308090e-02
-1.26314089e-01 7.65786529e-01 -5.45046926e-01 4.63430047e-01
-6.19881809e-01 -5.07368669e-02 -3.39311540e-01 -7.84465194e-01
8.50010365e-02 -5.80986515e-02 -4.17474240e-01 5.57727695e-01
3.08041841e-01 6.04629040e-01 -7.40084887e-01 1.97803885e-01
4.15889621e-01 1.14290798e+00 -4.80493844e-01 4.16364998e-01
2.67855108e-01 3.29234540e-01 3.19068044e-01 3.71896774e-01
7.77033746e-01 4.37770337e-01 -1.53065994e-01 -2.34052330e-01
-2.58996546e-01 6.84591055e-01 -1.36090481e+00 1.61481524e+00
-8.77871156e-01 1.08628595e+00 1.13653672e+00 -8.11950028e-01
9.01094854e-01 8.88195217e-01 1.35557741e-01 -5.97743392e-01
-3.70230898e-02 6.57999218e-01 2.00092673e-01 -3.30240250e-01
4.15490150e-01 -4.35770273e-01 5.87661147e-01 7.16362447e-02
5.32713771e-01 -4.02095973e-01 -6.56691268e-02 -7.59261986e-03
1.39305973e+00 -1.61825418e-02 2.64320690e-02 -5.50298318e-02
5.93536139e-01 -5.33332407e-01 9.83323231e-02 9.05027449e-01
-2.58450806e-01 1.10642588e+00 -3.80313933e-01 2.76789844e-01
-9.38185394e-01 -1.23498476e+00 -6.00474626e-02 1.19517291e+00
-6.82652533e-01 -2.86866546e-01 -8.72773945e-01 -1.01590909e-01
-4.02239293e-01 1.09712243e+00 -6.69221357e-02 2.01127216e-01
-9.10909772e-01 -6.64033294e-01 8.07391763e-01 4.13245022e-01
2.50306755e-01 -9.45058763e-01 2.13201702e-01 4.80581850e-01
-3.58265251e-01 -1.07629991e+00 -7.11565673e-01 7.36628175e-01
-3.31807166e-01 -4.26133573e-01 -1.09888768e+00 -6.86695576e-01
-1.18065916e-01 5.18277764e-01 1.02706921e+00 -2.15441719e-01
-7.06656575e-02 4.21191812e-01 -3.31724644e-01 -2.70489663e-01
-7.29468942e-01 -1.59594908e-01 5.67361712e-02 5.66646680e-02
1.20072141e-02 -7.09945381e-01 -3.82262349e-01 1.60811916e-01
-8.54053319e-01 -4.29518878e-01 4.68386203e-01 9.92584646e-01
2.27607131e-01 1.05069458e-01 7.97709823e-01 -2.69645452e-01
8.85975778e-01 -4.18806911e-01 -4.33183372e-01 -1.90867893e-02
-1.49592146e-01 -2.32654095e-01 6.15468919e-01 -1.91552028e-01
-1.44823492e+00 -2.57947177e-01 -1.24314106e+00 -3.22883576e-01
-3.54520679e-01 4.09206569e-01 -3.60577703e-01 2.22722620e-01
7.62944996e-01 2.15247318e-01 -8.91344696e-02 -8.91505420e-01
3.72309923e-01 1.28605354e+00 7.06312835e-01 -1.29043847e-01
7.57741928e-01 1.49346460e-02 -4.24114883e-01 -1.61218750e+00
-6.33019507e-01 -8.70641768e-01 -1.17506318e-01 -1.10974424e-02
6.37939572e-01 -1.25042522e+00 -5.85265160e-01 7.39167988e-01
-1.43173862e+00 -4.68163699e-01 -2.15715989e-01 6.32164180e-01
-4.41158533e-01 3.31590414e-01 -7.57409871e-01 -1.11176980e+00
-5.45675993e-01 -1.29539049e+00 1.10211110e+00 -1.75061986e-01
-1.12399466e-01 -8.32507491e-01 1.12109249e-02 5.12955070e-01
1.03306854e+00 -5.37473857e-01 2.69244850e-01 -6.65309966e-01
-1.94044948e-01 -1.84950575e-01 1.01104319e-01 1.11073720e+00
-5.65803200e-02 -4.08398986e-01 -1.64918888e+00 -5.58049440e-01
4.88970309e-01 -6.76835552e-02 9.27542984e-01 7.29283392e-01
9.06143785e-01 -2.75760114e-01 4.13504273e-01 7.91887462e-01
1.00452852e+00 3.37697417e-01 8.04700434e-01 1.76656656e-02
4.54280436e-01 5.08223116e-01 1.18167579e-01 -6.18783683e-02
-9.97034833e-02 8.03466439e-01 1.42259663e-02 -4.32413697e-01
-9.64556515e-01 5.40011190e-02 9.54390764e-01 1.29027390e+00
4.04536039e-01 -6.84477866e-01 -6.57853901e-01 6.62768781e-01
-1.22689760e+00 -1.13621128e+00 -2.52263993e-01 1.97870374e+00
8.18875253e-01 -2.45097712e-01 -2.49019310e-01 3.97305667e-01
7.13989258e-01 4.55296546e-01 -2.04481527e-01 -6.96863055e-01
-5.67134976e-01 7.49328136e-01 2.75836259e-01 1.05447865e+00
-8.50004673e-01 6.58193707e-01 6.58991194e+00 1.22720015e+00
-9.27505553e-01 5.20144999e-01 8.11053336e-01 -3.30812752e-01
-8.59503970e-02 -4.15391803e-01 -6.83582306e-01 1.86114851e-02
1.41214633e+00 4.90483552e-01 1.03295505e+00 3.20907354e-01
7.95969009e-01 2.27689818e-01 -1.05393219e+00 1.00503504e+00
-8.12395588e-02 -9.10109520e-01 -4.84086245e-01 -4.67289627e-01
4.64142174e-01 5.98797381e-01 1.78472757e-01 3.10741633e-01
2.40989327e-01 -1.33026981e+00 1.01014233e+00 1.31865546e-01
1.00214005e+00 -4.69462961e-01 6.08704686e-01 2.83381194e-01
-1.06167412e+00 -2.04307035e-01 8.96546841e-02 1.07810758e-01
5.13827503e-01 5.68430424e-01 -1.04243708e+00 4.97903049e-01
7.71019697e-01 4.14100438e-01 -7.54634365e-02 9.07422721e-01
-3.35899323e-01 1.22338736e+00 -4.86583471e-01 2.71475971e-01
1.89890817e-01 7.03593716e-02 1.06062579e+00 1.83255076e+00
4.12454993e-01 6.15052395e-02 -4.19146180e-01 5.18110573e-01
3.37123871e-02 -2.59413898e-01 -2.04510435e-01 6.86774999e-02
2.91006565e-01 9.75600362e-01 -1.30845979e-01 -4.30425741e-02
-2.75854707e-01 8.74496102e-01 -4.82093841e-02 8.31471086e-01
-4.57904398e-01 -5.58961868e-01 9.46015418e-01 1.22012207e-02
5.98563373e-01 -3.80272746e-01 -2.93964058e-01 -8.59299421e-01
8.55384469e-02 -1.29979312e+00 -5.74732460e-02 -1.15861118e+00
-1.24658835e+00 9.05491292e-01 -3.17812771e-01 -5.74181318e-01
-1.62849918e-01 -8.01434278e-01 -4.83807355e-01 1.69661438e+00
-1.76312184e+00 -9.82744753e-01 1.04693219e-01 5.54784358e-01
9.56256270e-01 -1.39166906e-01 6.41004026e-01 7.39475906e-01
-4.30562824e-01 6.28214777e-01 3.64942670e-01 -1.76436990e-01
6.11041605e-01 -1.35527337e+00 7.97242820e-01 1.38944697e+00
2.52573818e-01 7.20090389e-01 9.37410653e-01 -4.65334773e-01
-1.35714996e+00 -1.01656079e+00 1.02294791e+00 -1.96201950e-01
6.11296415e-01 -6.36711538e-01 -9.05337214e-01 5.90238452e-01
5.23013890e-01 -1.45419300e-01 3.18998784e-01 5.81929125e-02
-2.11771414e-01 -3.85293096e-01 -9.11570787e-01 4.81634051e-01
1.03269994e+00 -8.55338097e-01 -5.52373707e-01 3.01278323e-01
9.53538775e-01 -5.04299700e-01 -5.78039706e-01 1.84644595e-01
1.08625174e-01 -9.60906625e-01 1.22397733e+00 -2.08120689e-01
-8.97259079e-03 -5.16015172e-01 -7.04475880e-01 -1.78237772e+00
-1.45026579e-01 -1.34606612e+00 -8.24860334e-02 1.35531819e+00
5.74443340e-01 -5.94336867e-01 1.52023509e-01 4.82857600e-02
-8.47695172e-01 -2.04266891e-01 -1.24049759e+00 -7.32059121e-01
-1.71820581e-01 -1.07802653e+00 2.25192696e-01 3.49888325e-01
-6.81082487e-01 4.32473719e-01 -5.34233749e-01 5.61535358e-01
3.95299554e-01 -7.60716200e-01 4.64747727e-01 -5.28872669e-01
-8.08433831e-01 -4.87748682e-01 4.08330783e-02 -1.33807218e+00
1.70030706e-02 -9.72488880e-01 7.08088577e-01 -1.52256227e+00
-5.00196099e-01 -1.13596991e-01 -3.14875931e-01 2.64010519e-01
-3.26530278e-01 -2.70750988e-02 1.22273378e-01 -3.38040322e-01
1.31723424e-02 5.60812593e-01 9.50847149e-01 -6.78050444e-02
-1.19389482e-01 2.66114861e-01 -5.85515201e-01 3.78444582e-01
4.49371487e-01 -2.91224450e-01 -2.54981756e-01 -7.89750218e-01
-2.21085310e-01 3.36986691e-01 3.94398242e-01 -9.63692725e-01
2.52581686e-01 6.40207648e-01 2.37352788e-01 -5.95668733e-01
7.84091413e-01 -7.07855284e-01 2.27019057e-01 1.04869783e-01
-4.56654757e-01 -8.50345269e-02 3.84266585e-01 4.11080390e-01
-5.54176509e-01 -6.02304675e-02 1.00488234e+00 4.97549661e-02
-5.07388532e-01 -1.62584767e-01 -6.08331025e-01 -2.77689248e-02
1.34236902e-01 8.10505543e-03 -1.52212217e-01 -7.57141471e-01
-8.79882276e-01 -1.66129410e-01 -3.48928869e-01 2.95207322e-01
6.57378376e-01 -7.01296806e-01 -1.30231583e+00 3.13066006e-01
-5.89654744e-01 -1.99125305e-01 6.30423188e-01 7.35786498e-01
-2.88523704e-01 4.56258416e-01 4.67250973e-01 -5.12075663e-01
-1.31624794e+00 1.86937466e-01 7.64784098e-01 -7.46481940e-02
-4.71083879e-01 1.44383371e+00 1.88368913e-02 -6.13473952e-01
6.44700885e-01 -2.40415350e-01 -1.07694473e-02 -1.87344074e-01
7.09399581e-01 7.54291713e-01 8.23961377e-01 -7.54883587e-01
-3.21111768e-01 -2.86049093e-03 1.21823229e-01 -6.74590647e-01
1.77970159e+00 -2.90022373e-01 1.08228564e-01 2.86233604e-01
1.49094343e+00 2.71879196e-01 -1.15265369e+00 -7.30220675e-02
-2.97473907e-01 -6.87440932e-02 7.57609606e-01 -1.38296247e+00
-8.97835910e-01 1.10949028e+00 7.40911663e-01 1.69257045e-01
1.51273417e+00 -3.99170876e-01 8.13320756e-01 3.70082408e-01
-2.45355919e-01 -9.15191352e-01 -1.69488996e-01 9.11585867e-01
1.38748181e+00 -1.00206101e+00 -4.66128975e-01 -1.28197417e-01
-3.19849998e-01 1.08663249e+00 7.81634897e-02 1.62601218e-01
8.35829556e-01 7.70692348e-01 4.58213389e-01 5.49391024e-02
-5.83210945e-01 -7.24462718e-02 3.43964815e-01 7.06878304e-01
7.06135392e-01 1.20401976e-03 8.04711357e-02 4.96875823e-01
-5.16362369e-01 -3.54994088e-01 1.87208042e-01 6.21630907e-01
-2.80165792e-01 -9.68487680e-01 -7.66888261e-01 4.35945988e-02
-7.24148691e-01 -6.33344710e-01 -8.30867216e-02 3.58608752e-01
-2.09743053e-01 1.78388345e+00 -2.36464143e-01 -2.63909191e-01
5.69910824e-01 8.16219002e-02 1.70894250e-01 -5.00884235e-01
-1.26525319e+00 6.05068743e-01 4.41996902e-01 -4.30255741e-01
-4.83200736e-02 -4.96898144e-01 -7.77922630e-01 -1.29561543e-01
-5.08855164e-01 1.34949526e-02 1.11737931e+00 8.35386753e-01
3.18842769e-01 1.06277430e+00 6.37051463e-01 -1.12196767e+00
-9.47782576e-01 -1.44414282e+00 -5.94285369e-01 7.30611309e-02
1.04224098e+00 6.35389611e-02 -5.53327739e-01 -5.78581840e-02] | [14.952703475952148, 6.027982711791992] |
3e0306a2-9d29-4e41-b3ed-e2af8618509a | multi-scale-knowledge-distillation-for | 2204.09931 | null | https://arxiv.org/abs/2204.09931v2 | https://arxiv.org/pdf/2204.09931v2.pdf | Learning to Purification for Unsupervised Person Re-identification | Unsupervised person re-identification is a challenging and promising task in computer vision. Nowadays unsupervised person re-identification methods have achieved great progress by training with pseudo labels. However, how to purify feature and label noise is less explicitly studied in the unsupervised manner. To purify the feature, we take into account two types of additional features from different local views to enrich the feature representation. The proposed multi-view features are carefully integrated into our cluster contrast learning to leverage more discriminative cues that the global feature easily ignored and biased. To purify the label noise, we propose to take advantage of the knowledge of teacher model in an offline scheme. Specifically, we first train a teacher model from noisy pseudo labels, and then use the teacher model to guide the learning of our student model. In our setting, the student model could converge fast with the supervision of the teacher model thus reduce the interference of noisy labels as the teacher model greatly suffered. After carefully handling the noise and bias in the feature learning, our purification modules are proven to be very effective for unsupervised person re-identification. Extensive experiments on three popular person re-identification datasets demonstrate the superiority of our method. Especially, our approach achieves a state-of-the-art accuracy 85.8\% @mAP and 94.5\% @Rank-1 on the challenging Market-1501 benchmark with ResNet-50 under the fully unsupervised setting. The code will be released. | ['DaCheng Tao', 'Jing Zhang', 'Xiang Zhang', 'Xiao Teng', 'Long Lan'] | 2022-04-21 | null | null | null | null | ['unsupervised-person-re-identification'] | ['computer-vision'] | [ 1.05612926e-01 -2.76541591e-01 1.11858353e-01 -4.84607309e-01
-5.35247982e-01 -4.50798333e-01 4.48968738e-01 2.07929853e-02
-6.98790014e-01 5.70454061e-01 1.79388061e-01 3.69646460e-01
-8.91362876e-02 -5.85114360e-01 -4.77863461e-01 -1.00696313e+00
4.16097105e-01 5.58854818e-01 -1.70096800e-01 1.26525819e-01
1.24898124e-02 1.95894971e-01 -1.67258847e+00 -2.17777744e-01
9.58060205e-01 6.71637595e-01 4.39858027e-02 1.93132401e-01
1.54538661e-01 4.13619518e-01 -3.83831680e-01 -4.88537014e-01
4.56553847e-01 -2.85663337e-01 -6.17827117e-01 4.01589900e-01
4.92186010e-01 -2.89763033e-01 -3.79092664e-01 1.44452536e+00
8.42075944e-01 4.39986825e-01 4.49723333e-01 -9.32208717e-01
-5.58256149e-01 5.87541759e-01 -8.09661567e-01 2.54424512e-01
2.52125293e-01 4.90603633e-02 7.64141083e-01 -1.07175732e+00
2.61054009e-01 1.05268908e+00 6.16668940e-01 8.31055343e-01
-1.37466371e+00 -1.09345114e+00 4.20880765e-01 1.91039145e-01
-1.66789341e+00 -4.68404651e-01 7.61846840e-01 -4.60246146e-01
3.96745354e-01 2.36547455e-01 3.76178950e-01 9.88309324e-01
-6.49311185e-01 8.01807225e-01 1.29667115e+00 -1.94559097e-01
-2.00226113e-01 6.05241776e-01 7.70961821e-01 6.54433072e-01
2.27249086e-01 8.04033279e-02 -4.29085791e-01 -2.02977329e-01
4.08559769e-01 4.58865881e-01 -2.67938673e-01 -1.93639934e-01
-1.07768083e+00 4.70888019e-01 3.83791894e-01 1.63164064e-01
-1.47138238e-01 -1.12325862e-01 3.35579485e-01 1.95090622e-01
3.52647662e-01 1.74404547e-01 -2.83491939e-01 4.65354770e-02
-6.97816491e-01 4.40498479e-02 4.23156679e-01 8.54272842e-01
1.02678299e+00 -2.53316939e-01 -2.82850534e-01 1.16932857e+00
9.63940173e-02 6.42955363e-01 7.31964767e-01 -5.48971951e-01
3.59169811e-01 6.21116042e-01 1.99027732e-01 -8.63313377e-01
-4.84554350e-01 -8.97306561e-01 -1.13646042e+00 -2.38338649e-01
5.99882126e-01 -1.49539009e-01 -8.52496505e-01 1.75987136e+00
4.55685198e-01 5.54453135e-01 -1.03327081e-01 9.84924316e-01
7.67313361e-01 2.74754822e-01 1.39772281e-01 -2.22728714e-01
1.46245468e+00 -1.08891857e+00 -4.09129828e-01 -5.97862639e-02
6.34020090e-01 -4.84979868e-01 6.68112576e-01 3.94645035e-01
-5.33122778e-01 -9.27540720e-01 -8.36633801e-01 2.88905472e-01
-2.68121958e-01 5.31813741e-01 4.06983107e-01 8.39084089e-01
-6.39631093e-01 5.58001459e-01 -5.87020993e-01 -3.27938884e-01
4.10884082e-01 7.17461407e-01 -6.46826327e-01 -1.85201302e-01
-1.00066352e+00 3.24045479e-01 3.50033820e-01 3.55094701e-01
-7.33220160e-01 -5.45888722e-01 -6.09210432e-01 1.14399560e-01
6.03745282e-01 -6.37369454e-01 8.21298122e-01 -9.10339773e-01
-1.43812287e+00 8.98448050e-01 -2.69375950e-01 -2.21241370e-01
4.56135571e-01 -2.76980966e-01 -3.96524876e-01 -4.72360216e-02
3.09212089e-01 3.14740717e-01 8.19148183e-01 -1.34937751e+00
-8.79380822e-01 -6.35410845e-01 -7.65331089e-02 2.82688916e-01
-7.20641553e-01 -1.00205205e-01 -5.38174629e-01 -6.34708047e-01
5.26191555e-02 -1.14794290e+00 -2.25054488e-01 -5.87645531e-01
-4.57816631e-01 -4.90847141e-01 3.22805375e-01 -6.46661103e-01
9.51422572e-01 -2.27469039e+00 8.10766220e-02 5.26305974e-01
4.44263816e-01 3.14789861e-01 -1.85347795e-02 4.48851846e-02
-2.00297207e-01 2.81234756e-02 -6.73392639e-02 -6.62687480e-01
-1.30747780e-01 -1.85200185e-01 -1.03439212e-01 6.47268713e-01
-2.85134137e-01 6.32827401e-01 -9.14701700e-01 -2.66620129e-01
2.27663189e-01 3.29725891e-01 -4.50229049e-01 3.39161366e-01
4.85788524e-01 7.65682578e-01 -6.31289244e-01 4.10731107e-01
8.19232523e-01 -2.44361848e-01 8.97601768e-02 -2.49282017e-01
1.06674522e-01 -1.43521130e-01 -1.57917607e+00 1.52045107e+00
-1.42536283e-01 -1.01895951e-01 -4.37600203e-02 -1.09620297e+00
8.78367126e-01 1.22861788e-01 5.04117548e-01 -4.82932657e-01
1.04029074e-01 8.11708272e-02 -1.46801874e-01 -3.93969774e-01
1.90342680e-01 -1.31634831e-01 -3.14232297e-02 4.55256492e-01
1.51869908e-01 7.39719272e-01 5.35339974e-02 1.42605618e-01
5.19842982e-01 -7.94997662e-02 5.73842647e-03 -3.08705509e-01
9.69348967e-01 -4.92711216e-01 7.32835948e-01 9.29521739e-01
-2.88412929e-01 6.00080311e-01 -9.69967693e-02 -3.98574203e-01
-6.40009522e-01 -6.76361382e-01 -1.07988156e-01 1.27790976e+00
4.81818914e-01 -5.28797269e-01 -7.37773895e-01 -9.53610957e-01
-6.72139972e-02 2.07844526e-01 -6.37295187e-01 -9.41590890e-02
-5.20252824e-01 -1.11992228e+00 4.03174579e-01 5.14912486e-01
6.24243021e-01 -7.28523016e-01 1.02193981e-01 8.89395848e-02
-2.82422513e-01 -1.05807388e+00 -8.35933566e-01 -2.27753315e-02
-4.71938759e-01 -1.08088183e+00 -9.80356157e-01 -9.45250392e-01
9.78819013e-01 6.50184691e-01 5.55623114e-01 2.85890847e-01
-4.28052321e-02 5.18249691e-01 -2.26000294e-01 -2.05934905e-02
9.18151736e-02 1.60070434e-01 5.92311800e-01 6.48467362e-01
6.93240881e-01 -6.47896051e-01 -6.40408218e-01 5.19365966e-01
-5.09148240e-01 -4.85480204e-02 2.64481127e-01 1.17689145e+00
6.64428830e-01 3.06968570e-01 5.42682528e-01 -1.05762339e+00
3.26539040e-01 -2.92579055e-01 -4.86287892e-01 2.31285274e-01
-7.20217764e-01 -4.44869697e-03 8.22569549e-01 -6.10157788e-01
-1.07953393e+00 3.67840052e-01 -2.73235813e-02 -3.22671950e-01
-3.11517775e-01 1.53611213e-01 -5.29626191e-01 -1.00101866e-02
3.07158262e-01 4.92150068e-01 -3.00723851e-01 -8.22620869e-01
2.37979189e-01 7.84154773e-01 5.79945445e-01 -7.13192642e-01
1.19187284e+00 6.20130360e-01 -2.98023999e-01 -5.98321676e-01
-9.49770153e-01 -9.20924604e-01 -6.93761289e-01 -3.73900548e-04
6.42263532e-01 -1.27920163e+00 -9.33405995e-01 7.11200118e-01
-6.96762860e-01 1.73219651e-01 -2.06503898e-01 4.38442916e-01
-4.07371037e-02 6.71971083e-01 -3.86779189e-01 -7.35862017e-01
-5.02209187e-01 -1.22634578e+00 8.00815642e-01 7.99993277e-01
1.47381797e-01 -6.49455070e-01 2.36402338e-04 5.17141044e-01
1.29568741e-01 -2.86974311e-01 3.44213754e-01 -1.05586839e+00
-2.53058195e-01 -2.99497664e-01 -3.39256823e-01 2.76860207e-01
3.48833293e-01 -7.57821977e-01 -1.27407777e+00 -6.44528806e-01
-1.18655097e-02 -3.19470197e-01 1.09505045e+00 -1.10690847e-01
1.31803834e+00 -4.57156375e-02 -5.37287116e-01 8.27093482e-01
1.21868706e+00 -1.41375318e-01 1.03282839e-01 3.41370314e-01
1.20708382e+00 6.84780121e-01 2.70035684e-01 5.65100670e-01
5.01072049e-01 6.80026889e-01 -2.66458336e-02 6.74374178e-02
6.71549365e-02 -4.16503936e-01 2.25963399e-01 7.77781367e-01
-3.04904699e-01 1.09965235e-01 -6.51472330e-01 3.51691693e-01
-1.92488456e+00 -9.44421351e-01 8.75291675e-02 2.39505553e+00
7.54753590e-01 -5.57129011e-02 4.40635443e-01 8.39501098e-02
1.03980505e+00 -1.46602675e-01 -5.02511621e-01 4.48407501e-01
-7.32065663e-02 -1.63347825e-01 6.23727262e-01 4.51286912e-01
-1.42996895e+00 9.81340349e-01 4.38128519e+00 8.92736077e-01
-8.97703290e-01 1.82507306e-01 6.72135830e-01 4.54874486e-02
3.35677043e-02 -8.22751522e-02 -1.32972431e+00 6.89582586e-01
6.80931985e-01 -4.99861455e-03 6.10025346e-01 7.59429157e-01
-4.83057648e-03 1.17522940e-01 -1.09403515e+00 1.52396023e+00
2.35851273e-01 -6.50485873e-01 -6.04656041e-02 1.12179399e-01
6.80714250e-01 -2.12296039e-01 7.08448589e-02 3.62373710e-01
2.03909725e-01 -9.53995109e-01 4.58722800e-01 6.42846107e-01
6.48279071e-01 -9.29958463e-01 8.46134722e-01 4.53036457e-01
-1.19350910e+00 -1.91393778e-01 -6.20786071e-01 7.11659566e-02
-3.49668711e-02 6.36598885e-01 -4.08085197e-01 6.63864911e-01
8.66240323e-01 9.87034798e-01 -8.41195583e-01 1.11190844e+00
-5.52981012e-02 6.32906437e-01 -3.75395238e-01 2.74268717e-01
-2.46161699e-01 -3.33631039e-01 4.72336859e-01 1.10626745e+00
8.17183033e-02 2.26528808e-01 6.53072596e-01 5.73486567e-01
-1.89522162e-01 1.93435460e-01 -1.70139506e-01 2.09972903e-01
2.36327484e-01 1.37820911e+00 -5.89521825e-01 -4.73517865e-01
-3.69184196e-01 1.20331085e+00 5.36361337e-01 4.31386411e-01
-6.41716063e-01 -1.40852198e-01 5.64280450e-01 -9.44465473e-02
2.41775721e-01 1.80314183e-02 5.90002611e-02 -1.53858113e+00
5.57909831e-02 -8.83997142e-01 5.11121333e-01 -6.21506050e-02
-1.75615168e+00 5.77792585e-01 -1.25253126e-01 -1.26427507e+00
6.40889704e-02 -4.04781401e-01 -2.42747322e-01 9.98481929e-01
-1.52910292e+00 -1.08458126e+00 -6.14813566e-01 7.24557817e-01
3.29347879e-01 -3.99427146e-01 7.44305730e-01 6.28799319e-01
-1.20095599e+00 1.12072003e+00 3.36239278e-01 5.06921053e-01
1.09239209e+00 -1.21868217e+00 -9.36357155e-02 8.43837976e-01
1.51847735e-01 9.19823885e-01 3.58454823e-01 -5.91214061e-01
-1.19511902e+00 -1.14107990e+00 6.32859230e-01 -4.20920908e-01
3.53289336e-01 -4.31618899e-01 -8.53048861e-01 6.41900837e-01
-2.14060426e-01 8.97275731e-02 9.50584650e-01 2.56472290e-01
-5.05554974e-01 -4.58836526e-01 -9.86675441e-01 3.41455191e-01
1.06586266e+00 -5.62550604e-01 -4.00493115e-01 3.33027154e-01
4.16978240e-01 -1.92868054e-01 -7.32664108e-01 3.12523693e-01
5.51641464e-01 -6.58840835e-01 1.14726233e+00 -4.96024162e-01
-2.91387230e-01 -5.90084434e-01 1.36578470e-01 -1.17429745e+00
-6.86637998e-01 -5.52860975e-01 1.50627136e-01 1.76762950e+00
4.83421199e-02 -7.90925026e-01 7.92516887e-01 7.62712657e-01
3.22807491e-01 -3.81023526e-01 -7.12872386e-01 -5.95106483e-01
-2.05127373e-01 -3.31530510e-03 6.21185422e-01 9.17982578e-01
-1.16284378e-01 4.63354617e-01 -7.70566821e-01 4.02014256e-01
9.07690465e-01 1.50197282e-01 9.70834911e-01 -1.41805518e+00
-3.69753063e-01 -2.04905853e-01 -2.50958353e-01 -1.26808441e+00
4.02840078e-01 -1.07500482e+00 -1.04547039e-01 -9.78355110e-01
8.38420510e-01 -6.86745286e-01 -7.03284025e-01 4.99035895e-01
-7.74256706e-01 4.03479367e-01 2.90607214e-01 5.90675712e-01
-7.91897714e-01 7.24660575e-01 8.71060669e-01 -3.92656595e-01
-2.14869946e-01 2.71913946e-01 -1.00412822e+00 6.98421061e-01
7.33204722e-01 -5.78608334e-01 -2.93383032e-01 -3.10191184e-01
-1.95051089e-01 -5.26660800e-01 4.28086072e-01 -1.02504063e+00
4.38460439e-01 1.55764103e-01 6.20587528e-01 -3.62272054e-01
2.68072754e-01 -8.91099155e-01 -1.79947391e-02 2.66280115e-01
-2.38810927e-01 -2.30716959e-01 -1.45505756e-01 8.12936723e-01
-6.14415482e-02 -2.78856575e-01 8.14698935e-01 -1.25059113e-01
-6.43378973e-01 5.58861554e-01 4.15591523e-02 -1.03196763e-01
6.99279070e-01 -6.69111907e-02 -1.74736291e-01 -1.29356593e-01
-9.64820325e-01 5.53959012e-01 4.31750953e-01 3.34779799e-01
4.20099705e-01 -1.25134492e+00 -7.32537389e-01 3.14022928e-01
1.78584129e-01 -1.79676458e-01 5.21511436e-01 8.73832941e-01
2.17084318e-01 2.06689030e-01 3.71042378e-02 -6.08924389e-01
-1.22778368e+00 8.04876864e-01 4.91447747e-01 -3.18858832e-01
-6.73216820e-01 8.03194761e-01 5.86209059e-01 -5.47425866e-01
4.95886773e-01 3.20876956e-01 -6.25082016e-01 1.74563870e-01
8.79881799e-01 4.92129207e-01 -2.28529721e-01 -9.69900250e-01
-4.71116334e-01 8.10756803e-01 -4.26333368e-01 1.83099076e-01
1.23624110e+00 -4.66495395e-01 -7.88156390e-02 1.77137509e-01
1.13509846e+00 1.68837234e-01 -1.02959383e+00 -7.17632174e-01
2.19754614e-02 -3.41160446e-01 -3.32009614e-01 -5.63809693e-01
-1.05653930e+00 6.47144318e-01 1.01063263e+00 -1.17927872e-01
1.03800714e+00 -2.07693398e-01 8.88960779e-01 4.01475430e-01
3.42468888e-01 -1.07947516e+00 -2.70310119e-02 2.64835447e-01
3.37946504e-01 -1.48440742e+00 -1.16001153e-02 -4.93758261e-01
-4.41676944e-01 8.20072472e-01 8.13983560e-01 -1.20557711e-01
7.54445314e-01 -2.13648632e-01 1.02452859e-01 1.72818378e-01
4.69129197e-02 -4.39838469e-01 3.85217726e-01 4.30173576e-01
8.42445940e-02 2.08282381e-01 -2.06120849e-01 1.25353146e+00
-1.29737645e-01 -1.56654075e-01 1.29559144e-01 3.41028452e-01
-2.74705261e-01 -1.19003940e+00 -3.64557981e-01 3.66338879e-01
-5.72411895e-01 -5.10574728e-02 -2.24396497e-01 2.73430437e-01
4.97859478e-01 1.05518281e+00 -2.66240269e-01 -6.44589007e-01
2.98833370e-01 1.38036564e-01 3.09050024e-01 -6.59495294e-01
-7.43924737e-01 2.48981774e-01 -1.54363587e-01 -3.85835558e-01
-5.82830906e-01 -6.24251604e-01 -1.15016782e+00 -9.27627981e-02
-5.11701882e-01 4.64822114e-01 1.46475241e-01 1.05974269e+00
3.07445198e-01 1.88383922e-01 8.52457345e-01 -7.30684400e-01
-7.39324808e-01 -1.00906050e+00 -7.08408594e-01 8.10120761e-01
2.28580758e-01 -6.40644729e-01 -4.10443008e-01 4.62764949e-02] | [14.80541706085205, 1.0621559619903564] |
0130168f-e37f-4a21-b382-7399f7f3995c | an-approach-to-solving-the-abstraction-and | 2306.03553 | null | https://arxiv.org/abs/2306.03553v1 | https://arxiv.org/pdf/2306.03553v1.pdf | An Approach to Solving the Abstraction and Reasoning Corpus (ARC) Challenge | We utilise the power of Large Language Models (LLMs), in particular GPT4, to be prompt engineered into performing an arbitrary task. Here, we give the model some human priors via text, along with some typical procedures for solving the ARC tasks, and ask it to generate the i) broad description of the input-output relation, ii) detailed steps of the input-output mapping, iii) use the detailed steps to perform manipulation on the test input and derive the test output. The current GPT3.5/GPT4 prompt solves 2 out of 4 tested small ARC challenges (those with small grids of 8x8 and below). With tweaks to the prompt to make it more specific for the use case, it can solve more. We posit that when scaled to a multi-agent system with usage of past memory and equipped with an image interpretation tool via Visual Question Answering, we may actually be able to solve the majority of the ARC challenge | ['Tan John Chong Min'] | 2023-06-06 | null | null | null | null | ['visual-question-answering-1'] | ['computer-vision'] | [ 1.79084376e-01 8.34147274e-01 5.10946155e-01 -1.10855468e-01
-9.60973322e-01 -9.22167480e-01 8.16491961e-01 -6.28667325e-02
-2.54883289e-01 5.64948857e-01 1.14877202e-01 -8.15154552e-01
-6.02030575e-01 -6.43942416e-01 -5.69655955e-01 -3.50511193e-01
-1.37119606e-01 1.16993070e+00 3.72658163e-01 -2.10553020e-01
4.80060846e-01 2.85979867e-01 -1.25779128e+00 6.04680777e-01
6.48386776e-01 6.99550569e-01 6.12248361e-01 1.09840250e+00
-1.95975840e-01 1.16874528e+00 -8.86627018e-01 -2.29474500e-01
2.48264760e-01 -2.17967585e-01 -1.40051734e+00 1.54971793e-01
9.56914276e-02 -3.39293927e-01 7.89112672e-02 6.42833352e-01
3.71227533e-01 3.58705312e-01 5.34698069e-01 -1.45214593e+00
-7.60617793e-01 5.15672088e-01 -6.12899028e-02 2.58742899e-01
8.69053543e-01 8.72559547e-01 7.04589903e-01 -7.38174736e-01
6.11795902e-01 1.57447243e+00 3.89872015e-01 2.89192438e-01
-1.22377491e+00 -5.58095835e-02 1.40009657e-01 1.87076896e-01
-1.09322238e+00 -2.24178359e-01 6.40535429e-02 -2.95503438e-01
1.70333540e+00 4.71758366e-01 5.11178553e-01 1.08120739e+00
2.10944220e-01 6.26016498e-01 1.50648701e+00 -5.81435204e-01
4.24735397e-01 3.04418653e-01 -3.59341241e-02 8.03324699e-01
-4.93075430e-01 7.67739117e-02 -2.20518440e-01 -3.67890626e-01
1.05719781e+00 -6.29153788e-01 8.38764459e-02 1.61741376e-02
-1.28549886e+00 6.86287165e-01 2.20756873e-01 1.70342401e-01
-2.35284552e-01 1.33392006e-01 -5.23082800e-02 6.10078990e-01
2.18075797e-01 7.98935890e-01 -5.93618631e-01 -1.47455156e-01
-4.70571041e-01 5.35645127e-01 1.12433100e+00 1.03459001e+00
7.53782570e-01 -1.78272098e-01 -4.24012303e-01 6.91779375e-01
4.11638170e-01 1.43253252e-01 4.74988371e-01 -1.51754797e+00
4.40378606e-01 6.03246808e-01 3.61142665e-01 -6.78450167e-01
-6.37893498e-01 -3.55793573e-02 2.06152499e-01 5.09089649e-01
6.60917342e-01 -3.24799091e-01 -1.13549948e+00 1.52794051e+00
2.72422016e-01 5.24099283e-02 1.91740215e-01 6.27904236e-01
7.26036429e-01 7.92199671e-01 4.31754440e-01 2.01427862e-01
1.68392730e+00 -1.07234120e+00 -3.33943248e-01 -8.67265344e-01
9.06397283e-01 -7.67642617e-01 1.51467276e+00 3.85452569e-01
-1.38577461e+00 -4.76436645e-01 -9.69027460e-01 -1.48946106e-01
-7.39885449e-01 -1.57592565e-01 5.64908147e-01 3.83326828e-01
-1.58318985e+00 3.99579048e-01 -7.45779932e-01 -7.07296610e-01
-1.30249754e-01 4.17697430e-01 -1.33668467e-01 -2.01200351e-01
-1.05338514e+00 1.62442052e+00 5.82856715e-01 2.76668295e-02
-1.07397461e+00 -4.50655550e-01 -9.03166175e-01 1.92601737e-02
6.12031698e-01 -1.02204561e+00 1.61722469e+00 -7.28841126e-01
-1.52587557e+00 7.62432039e-01 2.04901665e-01 -2.48560421e-02
4.81838971e-01 -5.91182336e-02 -1.73019916e-01 2.55064875e-01
2.13266462e-01 1.11910212e+00 6.20681286e-01 -1.20880890e+00
-3.59646797e-01 -1.70797870e-01 6.47678494e-01 5.13961494e-01
4.13089395e-01 2.81837046e-01 -5.77403963e-01 -4.09637988e-01
-1.88808709e-01 -9.66604352e-01 -3.55789423e-01 -3.59625489e-01
-2.26608679e-01 -4.46155906e-01 7.27334678e-01 -7.33425558e-01
7.97208548e-01 -1.93198717e+00 4.46545891e-02 1.89738616e-01
-5.41840456e-02 5.83655722e-02 -5.80232918e-01 7.76796460e-01
-1.05796002e-01 1.93488032e-01 -8.21238235e-02 -2.76409715e-01
4.80675757e-01 4.91506666e-01 -1.55232713e-01 8.26620013e-02
4.18886423e-01 1.03826547e+00 -7.99409330e-01 -5.44506192e-01
2.41389960e-01 7.69208819e-02 -3.52404863e-01 5.33961117e-01
-9.80960250e-01 2.98208147e-01 -5.87209344e-01 3.57005596e-01
3.92660558e-01 -5.04672766e-01 1.16912305e-01 2.82925844e-01
-7.33248666e-02 2.41788626e-01 -1.35351872e+00 1.70971775e+00
-5.60331225e-01 4.80496466e-01 3.64196151e-01 -4.74306822e-01
4.51799065e-01 3.98793161e-01 -2.94724315e-01 -6.63476825e-01
-3.46198641e-02 8.47865120e-02 5.66545920e-03 -1.18509817e+00
2.08483115e-01 -2.37924065e-02 3.19856824e-03 1.00688970e+00
1.15578361e-01 -5.49867511e-01 4.06797200e-01 4.44270581e-01
1.35196507e+00 5.98092139e-01 4.92413901e-02 -2.23695293e-01
2.01856822e-01 4.58712608e-01 -2.10175082e-01 1.06233120e+00
1.88232183e-01 5.91861248e-01 7.28400588e-01 -2.35528409e-01
-1.09103811e+00 -9.75004554e-01 2.87074476e-01 1.33296299e+00
-2.87679911e-01 -4.20126468e-01 -8.47240925e-01 -4.56952363e-01
-3.39141071e-01 1.23833680e+00 -5.20768464e-01 1.89312965e-01
-4.85093206e-01 -6.35602653e-01 4.91370231e-01 4.62399274e-01
4.57877070e-01 -1.80099189e+00 -1.04584920e+00 1.55890867e-01
6.30711466e-02 -9.21080947e-01 -9.20907930e-02 9.45944563e-02
-5.46782911e-01 -9.09603059e-01 -4.85576212e-01 -1.03948152e+00
6.44672751e-01 -3.88187766e-01 1.29366875e+00 4.58995551e-01
-1.74766347e-01 1.04782712e+00 -3.30390096e-01 -2.70756453e-01
-6.42039895e-01 -3.72387171e-02 -6.34758711e-01 -6.06444776e-01
2.61469577e-02 -4.60245162e-01 -4.26999032e-01 1.70786828e-01
-1.06591427e+00 5.60696244e-01 6.63097978e-01 4.56084698e-01
3.51245468e-03 -7.52599388e-02 4.38787043e-01 -9.04968798e-01
1.19599557e+00 -4.95190144e-01 -5.01286864e-01 5.74798763e-01
-4.90649790e-01 6.04353100e-02 3.73301148e-01 -4.01052147e-01
-1.19070339e+00 -2.38188967e-01 -1.32767603e-01 7.95820951e-02
-5.01284599e-01 6.84261143e-01 -4.40104045e-02 1.96377009e-01
7.45069087e-01 1.50698796e-02 2.69129537e-02 -2.96571672e-01
5.48683941e-01 3.62885445e-01 3.56634200e-01 -1.14514041e+00
4.15880799e-01 -1.02902286e-01 -3.21413636e-01 -1.73487887e-01
-4.89754498e-01 2.39984259e-01 -3.00161332e-01 -2.23457873e-01
1.02702594e+00 -5.49849391e-01 -9.48541224e-01 2.20448330e-01
-1.27139354e+00 -1.23794663e+00 -1.29183009e-01 8.32871497e-02
-9.18551803e-01 1.43645995e-03 -7.84002662e-01 -7.96053469e-01
-5.05645052e-02 -1.38521481e+00 1.01965892e+00 2.93454915e-01
-5.42129695e-01 -1.21253645e+00 9.09787193e-02 4.10042584e-01
6.59081757e-01 4.76813130e-02 1.51608741e+00 -9.99260783e-01
-6.34429514e-01 1.29094809e-01 -3.12748224e-01 3.50996815e-02
-3.52863044e-01 -2.04557091e-01 -8.92306566e-01 -1.97386533e-01
1.12668365e-01 -6.94100380e-01 1.14368901e-01 7.97762070e-03
7.73344994e-01 -4.36796874e-01 -3.13241541e-01 -2.72042360e-02
1.24127531e+00 2.88819820e-01 7.22610474e-01 6.90025389e-01
2.55921602e-01 9.13927615e-01 5.45335650e-01 5.43640032e-02
6.23273671e-01 4.40142304e-01 3.31078887e-01 -6.04921207e-02
-2.34983619e-02 3.93289253e-02 4.83115345e-01 2.53483832e-01
2.37820983e-01 -2.74311453e-01 -1.27603638e+00 5.75300157e-01
-2.00314713e+00 -6.15691304e-01 7.51429200e-02 1.56904805e+00
7.72458613e-01 2.21589759e-01 -1.58279032e-01 -4.63052720e-01
4.07803237e-01 1.10768788e-01 -5.64847827e-01 -7.01034606e-01
3.26305360e-01 3.20806265e-01 -7.93961808e-02 8.88911545e-01
-4.85729635e-01 9.23074126e-01 7.79546070e+00 5.63786149e-01
-4.47173327e-01 7.46913850e-02 7.27139175e-01 1.95480660e-01
-3.22019607e-01 4.37092304e-01 -3.55777174e-01 4.80180755e-02
1.26676655e+00 -9.88191292e-02 8.95284951e-01 4.54274118e-01
2.08888888e-01 -6.60272181e-01 -1.42967999e+00 4.29161847e-01
3.11862994e-02 -1.07369244e+00 3.30294594e-02 -2.47997046e-01
2.95629144e-01 -5.67685924e-02 4.40514274e-02 7.97231197e-01
8.88659775e-01 -1.47372997e+00 6.74731731e-01 5.32152712e-01
4.52109367e-01 -3.92400026e-01 4.33249801e-01 8.07107627e-01
-5.80064774e-01 -1.84425950e-01 -6.99191689e-02 -3.23206574e-01
9.66987610e-02 -2.30631113e-01 -1.26826036e+00 4.28163081e-01
7.11233914e-01 -1.06590949e-01 -8.80527020e-01 7.60193467e-01
-3.20145279e-01 2.32995585e-01 -4.43483829e-01 7.15814531e-02
3.30926538e-01 -2.50509113e-01 3.48486841e-01 1.11489427e+00
3.80682975e-01 3.50769430e-01 4.15322632e-01 1.02165127e+00
5.18337131e-01 -1.29868314e-01 -4.18858469e-01 3.36287208e-02
2.95810670e-01 1.36263418e+00 -8.02132368e-01 -5.96952736e-01
-3.35146517e-01 8.03124487e-01 4.85341787e-01 6.90910578e-01
-7.80515254e-01 -2.80675918e-01 1.39348209e-01 6.78307489e-02
1.42948762e-01 -2.25789517e-01 -1.20751046e-01 -6.76932454e-01
-1.52336791e-01 -1.45453966e+00 5.96601903e-01 -2.03858614e+00
-1.08507037e+00 7.97272027e-01 5.53029537e-01 -3.78202647e-01
-7.11882532e-01 -8.45967472e-01 -8.71871352e-01 1.18713558e+00
-9.58562970e-01 -1.19672811e+00 -1.23342484e-01 6.29492223e-01
6.89175665e-01 -3.47889438e-02 1.21340108e+00 -1.38314858e-01
-2.98103750e-01 -1.15421779e-01 -7.31987953e-01 -1.42624453e-01
4.43400174e-01 -1.42499030e+00 4.77439463e-01 6.75630748e-01
-1.44192139e-02 8.17401052e-01 9.92543697e-01 -7.57081330e-01
-1.23341203e+00 -5.87517738e-01 5.73706806e-01 -1.01798797e+00
9.92502093e-01 -2.87309051e-01 -9.04560566e-01 1.29627001e+00
9.54977930e-01 -4.89571691e-01 3.49235356e-01 -2.23342121e-01
8.79289061e-02 4.04310077e-01 -1.24412715e+00 9.66820359e-01
8.62660587e-01 -6.11535370e-01 -1.11109233e+00 8.12763035e-01
6.63762510e-01 -8.29388857e-01 -7.95953155e-01 -2.90623810e-02
1.47711262e-01 -8.91372502e-01 8.76299798e-01 -6.04855716e-01
1.77218974e-01 -5.12351274e-01 7.65529787e-03 -1.50455558e+00
-3.80451858e-01 -7.32016802e-01 2.07742304e-01 1.16119707e+00
7.36108363e-01 -8.35270405e-01 3.90776426e-01 1.11693501e+00
-2.22537801e-01 -5.04616618e-01 -7.47913301e-01 -3.71232778e-01
1.30643651e-01 -7.42504776e-01 5.52897274e-01 5.27681589e-01
3.07056665e-01 3.14359158e-01 7.89541751e-02 5.21036804e-01
5.55703565e-02 -8.25070441e-02 5.45317352e-01 -8.34678173e-01
-8.16968799e-01 -2.11910695e-01 3.00864279e-01 -9.73657072e-01
-7.74149671e-02 -7.77034104e-01 7.55606443e-02 -2.04075146e+00
-9.75299999e-03 -3.79571050e-01 1.11538656e-01 7.66862631e-01
4.04236503e-02 -4.23433818e-02 4.03426647e-01 1.07171215e-01
-6.54073417e-01 -6.55616820e-02 1.18774664e+00 5.39764315e-02
-3.09510380e-02 -2.90070295e-01 -7.93247163e-01 6.84925377e-01
6.97894037e-01 -5.11670351e-01 -7.73525953e-01 -8.29284966e-01
5.90274274e-01 3.47423971e-01 6.03469133e-01 -9.38264251e-01
5.53510785e-01 -3.88712883e-01 6.20020151e-01 -2.15854466e-01
4.98876095e-01 -8.25248897e-01 4.45417643e-01 1.82478279e-01
-5.34187198e-01 6.45874262e-01 5.93817532e-01 5.19746691e-02
1.10606380e-01 -5.95451593e-01 2.73409337e-01 -7.28857338e-01
-7.72942603e-01 -3.27181011e-01 -7.94432044e-01 -1.84931811e-02
1.11579096e+00 6.82775676e-03 -8.73047054e-01 -5.49629033e-01
-1.32231784e+00 7.12672710e-01 4.07197505e-01 3.95237774e-01
3.06105137e-01 -9.47009146e-01 -4.81502920e-01 1.73098341e-01
-1.36293858e-01 -6.12814315e-02 8.15704837e-02 7.09247947e-01
-6.25607789e-01 4.63654965e-01 -1.56295538e-01 -3.96057129e-01
-8.29915762e-01 6.57229900e-01 4.59111482e-01 -4.43513542e-01
-6.40033782e-01 7.00886667e-01 2.32879192e-01 -5.33743680e-01
6.64254948e-02 -3.09641361e-01 -2.13234261e-01 -8.46600980e-02
5.81498682e-01 7.14424551e-02 4.43654023e-02 1.43496422e-02
-2.70324945e-01 2.81926662e-01 2.79755481e-02 -7.35232115e-01
1.37232018e+00 -1.82862401e-01 -2.30235517e-01 3.19901466e-01
6.58833444e-01 -2.25931510e-01 -1.48206401e+00 1.38935953e-01
2.12531537e-01 -1.25300854e-01 -4.14794087e-01 -1.37860262e+00
-3.79782319e-01 5.01988053e-01 3.90895039e-01 7.36716628e-01
8.59151602e-01 3.77820015e-01 1.65748402e-01 4.66566682e-01
3.47084343e-01 -1.15905178e+00 1.89806387e-01 6.57295823e-01
1.41474187e+00 -8.89002860e-01 7.12706149e-02 -1.26277089e-01
-7.54288971e-01 1.15435779e+00 8.30412805e-01 2.19922457e-02
2.10491106e-01 5.02893329e-01 3.79446626e-01 -6.21150017e-01
-1.41915858e+00 2.87453458e-03 -9.76364240e-02 8.03037524e-01
1.00406744e-01 -2.68901676e-01 1.29087463e-01 2.79796749e-01
-3.14287990e-01 -7.28760958e-02 5.11575997e-01 1.09006858e+00
-3.61321509e-01 -1.07066166e+00 -6.20669186e-01 2.26384968e-01
-1.76753849e-01 -2.71733254e-01 -3.85464549e-01 1.09145689e+00
-2.71536373e-02 1.16776013e+00 1.60077363e-01 4.41643298e-02
5.14419734e-01 4.27919865e-01 4.88859028e-01 -8.42952371e-01
-9.48084176e-01 -7.73831010e-02 6.88225329e-01 -5.66057384e-01
-1.57175899e-01 -4.72470075e-01 -1.43421340e+00 -1.82750419e-01
2.45569304e-01 3.11438926e-02 5.21614313e-01 1.16270542e+00
4.72832471e-01 3.42016071e-01 -1.67482257e-01 -9.62210119e-01
-4.63619262e-01 -1.22812676e+00 -2.32718229e-01 2.69283831e-01
-5.37824929e-02 -3.00377846e-01 -3.95224899e-01 1.16632596e-01] | [4.274390697479248, 1.1087110042572021] |
f14e95c9-2740-4107-91c5-f6a5f8950f8b | time-to-die-death-prediction-in-dota-2-using | 1906.03939 | null | https://arxiv.org/abs/1906.03939v1 | https://arxiv.org/pdf/1906.03939v1.pdf | Time to Die: Death Prediction in Dota 2 using Deep Learning | Esports have become major international sports with hundreds of millions of spectators. Esports games generate massive amounts of telemetry data. Using these to predict the outcome of esports matches has received considerable attention, but micro-predictions, which seek to predict events inside a match, is as yet unknown territory. Micro-predictions are however of perennial interest across esports commentators and audience, because they provide the ability to observe events that might otherwise be missed: esports games are highly complex with fast-moving action where the balance of a game can change in the span of seconds, and where events can happen in multiple areas of the playing field at the same time. Such events can happen rapidly, and it is easy for commentators and viewers alike to miss an event and only observe the following impact of events. In Dota 2, a player hero being killed by the opposing team is a key event of interest to commentators and audience. We present a deep learning network with shared weights which provides accurate death predictions within a five-second window. The network is trained on a vast selection of Dota 2 gameplay features and professional/semi-professional level match dataset. Even though death events are rare within a game (1\% of the data), the model achieves 0.377 precision with 0.725 recall on test data when prompted to predict which of any of the 10 players of either team will die within 5 seconds. An example of the system applied to a Dota 2 match is presented. This model enables real-time micro-predictions of kills in Dota 2, one of the most played esports titles in the world, giving commentators and viewers time to move their attention to these key events. | ['James Alfred Walker', 'Ryan Spick', 'Simon Demediuk', 'Anders Drachen', 'Adam Katona', 'Victoria Hodge', 'Florian Block'] | 2019-05-21 | null | null | null | null | ['dota-2'] | ['playing-games'] | [-2.78861791e-01 -5.71170300e-02 -9.93572772e-02 -1.78955719e-02
-9.83287096e-01 -6.57739997e-01 2.54268050e-01 4.34050679e-01
-8.19650292e-01 7.23316371e-01 4.56294775e-01 5.62485354e-03
-7.67044052e-02 -1.08077371e+00 -5.41314900e-01 -2.08722383e-01
-3.70466381e-01 6.89027071e-01 6.65200830e-01 -7.69290328e-01
1.84606299e-01 3.07159305e-01 -1.46549237e+00 5.26037931e-01
-2.54444145e-02 8.83896887e-01 -2.48139247e-01 8.00586224e-01
4.18004453e-01 1.15788925e+00 -9.97547030e-01 -9.81494784e-01
4.09248292e-01 -3.10937464e-01 -7.20831692e-01 -5.84744632e-01
3.60391378e-01 -3.47864360e-01 -8.19246709e-01 2.89558381e-01
7.33757913e-01 3.82220000e-01 2.14976221e-01 -1.10321522e+00
4.10970658e-01 8.31380367e-01 -5.22456467e-01 8.64210963e-01
5.10314405e-01 4.15670693e-01 1.03531647e+00 -3.77845317e-01
6.75527871e-01 7.54451573e-01 1.13313234e+00 1.22908272e-01
-8.67171228e-01 -9.99621928e-01 -2.96671182e-01 3.92416388e-01
-1.27963650e+00 -8.26512501e-02 4.31755513e-01 -5.71797550e-01
7.30210245e-01 2.70988762e-01 1.14245641e+00 1.12877476e+00
6.27001762e-01 5.42426348e-01 4.94952887e-01 3.14160317e-01
8.58561695e-02 -3.55279416e-01 -2.89505005e-01 8.99371952e-02
-2.14705303e-01 3.55490327e-01 -1.42971408e+00 -1.25638366e-01
4.84890759e-01 1.82154715e-01 1.70433015e-01 4.95467931e-01
-1.09538841e+00 9.13189828e-01 3.21107030e-01 -9.95733216e-02
-3.92384648e-01 4.59395975e-01 8.96772325e-01 1.69548362e-01
5.79304874e-01 3.93036485e-01 -9.49174687e-02 -1.24663436e+00
-1.10532904e+00 9.26931083e-01 7.74998665e-01 1.44230455e-01
2.51567155e-01 -5.32667749e-02 4.23565172e-02 6.65180385e-01
-4.57825482e-01 1.25851175e-02 3.65400463e-01 -6.88597500e-01
5.41514754e-01 6.24517560e-01 2.77830753e-02 -1.37573814e+00
-7.05955327e-01 -4.42351162e-01 -2.78576583e-01 4.85744268e-01
8.93839478e-01 -3.98142517e-01 -2.64802545e-01 1.50698841e+00
2.44708955e-01 2.39563465e-01 -4.52375323e-01 1.14792526e+00
7.92299628e-01 5.82360387e-01 1.12432107e-01 2.16182724e-01
1.26793945e+00 -2.80740470e-01 -2.04069465e-01 -4.30085927e-01
5.82539558e-01 -6.69560611e-01 6.97208583e-01 4.69590008e-01
-1.38722670e+00 -3.86674434e-01 -8.45257580e-01 1.30533829e-01
-1.34000540e-01 -5.50306380e-01 6.82205558e-01 1.94238484e-01
-2.70537585e-01 1.03750408e+00 -8.12769055e-01 -4.66492742e-01
3.64951462e-01 4.34288234e-01 -4.69375432e-01 2.53335536e-01
-1.65419161e+00 1.00172961e+00 2.77469188e-01 -2.48491108e-01
-8.85682166e-01 -1.06884277e+00 -5.80273509e-01 -4.95167263e-02
6.60361648e-01 -3.07253543e-02 1.30947423e+00 -7.30342329e-01
-8.76517415e-01 1.14691198e+00 4.78030533e-01 -6.78496301e-01
9.55476224e-01 -3.52481276e-01 -6.99620605e-01 -1.43702537e-01
4.81745303e-01 2.00693056e-01 8.58574286e-02 -2.75981784e-01
-1.18593955e+00 -2.19141871e-01 3.15059423e-01 4.46093082e-01
4.97821569e-02 4.73354101e-01 -2.70685941e-01 -5.06393015e-01
-2.30827659e-01 -6.76934063e-01 -1.68420240e-01 -1.34377912e-01
-2.96183646e-01 -1.24268681e-01 4.31008667e-01 -5.95797718e-01
1.52098393e+00 -2.24722481e+00 -2.66682386e-01 -1.64925437e-02
3.27141255e-01 -2.69381702e-02 4.42035109e-01 1.07773912e+00
-7.00020716e-02 -8.82968828e-02 3.85773659e-01 1.32177770e-01
-8.21225345e-02 -2.79565784e-03 -3.67051333e-01 5.42192578e-01
-4.26822424e-01 6.67192340e-01 -8.01882982e-01 -1.64282739e-01
6.10759743e-02 -3.09303645e-02 -4.98199195e-01 -1.98496021e-02
2.40408659e-01 1.55175358e-01 -1.39034331e-01 4.59478676e-01
4.51158583e-02 2.22915992e-01 -8.32074657e-02 1.60218418e-01
-3.94226283e-01 4.75615948e-01 -1.29782188e+00 1.42640162e+00
4.62888311e-05 7.37650692e-01 -1.28547952e-01 -7.38286018e-01
7.74101794e-01 2.74556398e-01 7.62253225e-01 -7.59721160e-01
3.12876433e-01 1.01457246e-01 8.00528526e-02 -3.56140435e-01
9.52949405e-01 -8.04304600e-01 -8.92886221e-01 8.67567003e-01
-1.64538309e-01 1.16700627e-01 3.98338765e-01 3.96464705e-01
1.63211799e+00 -2.59159684e-01 9.12520662e-02 1.62709951e-01
-3.86384428e-01 5.20115256e-01 9.51933026e-01 8.33521008e-01
-3.88426602e-01 7.55818665e-01 7.64254570e-01 -1.16441154e+00
-9.87637877e-01 -1.17665863e+00 3.53950381e-01 1.43676186e+00
4.01091307e-01 -7.53831923e-01 -1.70545965e-01 -1.72174856e-01
6.08082525e-02 4.31429803e-01 -8.46168399e-01 -4.81246650e-01
-6.17276192e-01 -4.35550153e-01 8.30932200e-01 5.57182312e-01
3.67009938e-01 -1.17342806e+00 -9.53185499e-01 6.54357314e-01
-4.56780404e-01 -7.83421218e-01 -5.56171060e-01 1.53081641e-01
-2.25948036e-01 -1.29735363e+00 -2.94288665e-01 -4.54256952e-01
-3.52121919e-01 -2.24840119e-01 1.34552991e+00 -4.63846922e-02
-7.32607484e-01 1.21643260e-01 -2.69680887e-01 -5.84907651e-01
-1.86979651e-01 1.46681368e-01 4.18837607e-01 4.47796099e-03
5.75473964e-01 -8.79183948e-01 -5.74177623e-01 5.40701091e-01
-5.59245229e-01 2.00654954e-01 -4.12756987e-02 5.32876849e-01
3.32082093e-01 -5.12410849e-02 5.42117476e-01 -6.64823711e-01
5.17036855e-01 -1.02299535e+00 -1.10550888e-01 -3.83328706e-01
1.98390931e-02 -7.77338743e-01 3.41276228e-01 -8.40360999e-01
-4.08870608e-01 -2.92994261e-01 -1.75663501e-01 -1.63965166e-01
-3.26292217e-02 6.21446371e-01 3.72830153e-01 4.60784316e-01
1.13070536e+00 -2.51844414e-02 -2.18313634e-01 -3.90479922e-01
-2.37486109e-01 3.19053501e-01 1.09542763e+00 -4.33466583e-01
7.55007505e-01 4.35961753e-01 -2.15549931e-01 -5.11785924e-01
-7.56685257e-01 -5.21640956e-01 -3.35069209e-01 -1.08780050e+00
7.19558716e-01 -1.13572609e+00 -1.43891656e+00 4.33206111e-01
-6.68241441e-01 -5.11646569e-01 -7.23915696e-01 7.37852693e-01
-5.09378433e-01 -3.72987956e-01 -6.63172722e-01 -4.87799019e-01
1.36562167e-02 -5.99729538e-01 5.57368577e-01 3.97780329e-01
-8.72990847e-01 -5.41531980e-01 4.39442664e-01 5.98898292e-01
8.52723271e-02 5.74357271e-01 3.42816591e-01 -8.18451703e-01
-1.49517387e-01 -7.40108132e-01 1.93911582e-01 -3.17259878e-01
-1.53293788e-01 -3.07268739e-01 -5.10173738e-01 1.54781500e-02
-6.28851593e-01 -6.00430489e-01 5.36330760e-01 3.86969894e-01
6.06161237e-01 1.13852128e-01 -2.86688924e-01 4.36092913e-01
8.79988790e-01 4.61048633e-02 5.09227455e-01 5.69952607e-01
2.90814400e-01 4.36745346e-01 8.40359867e-01 8.21558356e-01
4.12224829e-01 9.40302432e-01 5.34694672e-01 3.14730592e-02
-3.35564055e-02 -5.34555852e-01 3.93066227e-01 2.48385414e-01
-5.08198380e-01 -1.42257720e-01 -8.90490472e-01 9.39119339e-01
-1.78879964e+00 -1.60818660e+00 -4.30245042e-01 2.21919084e+00
6.26423538e-01 6.36652946e-01 8.64952028e-01 1.91825211e-01
7.63899505e-01 2.69404352e-01 -6.39714837e-01 -6.64900303e-01
3.17992903e-02 2.62903720e-01 6.89833879e-01 1.26205266e-01
-8.02200735e-01 8.69819939e-01 6.57679272e+00 1.15622056e+00
-1.00113499e+00 6.28784224e-02 3.80489826e-01 -1.04836261e+00
1.23150617e-01 2.41926778e-02 -5.39733469e-01 6.07566833e-01
9.86086071e-01 -5.45430124e-01 1.93930194e-01 6.64646149e-01
4.70189333e-01 -4.79186535e-01 -8.49429727e-01 1.11544645e+00
2.09616702e-02 -1.63927925e+00 -5.06699443e-01 2.31367722e-01
2.86965221e-01 1.27696544e-01 -2.07642719e-01 6.11489415e-01
6.55194521e-01 -1.18908274e+00 1.16618550e+00 5.15235245e-01
8.23214352e-01 -1.17894328e+00 6.53131306e-01 4.85933244e-01
-1.10388041e+00 -2.98961997e-01 -7.59161040e-02 -1.04060066e+00
7.93513596e-01 2.90900975e-01 -6.18336022e-01 1.87453665e-02
1.14427209e+00 5.66996813e-01 -1.72116324e-01 1.23728657e+00
9.22098383e-03 9.35738385e-01 -6.16219938e-01 -7.83202797e-02
2.73992002e-01 2.05846950e-01 6.41439021e-01 7.20460355e-01
3.88727993e-01 4.93779778e-01 1.24515459e-01 3.97470742e-01
-1.31437898e-01 1.07829258e-01 -5.10872781e-01 4.12235707e-02
3.13042223e-01 1.08248258e+00 -6.83028281e-01 8.04794673e-03
-9.53227058e-02 6.81567609e-01 3.26191515e-01 -3.08488786e-01
-9.87606466e-01 -2.37021714e-01 1.18945920e+00 9.96923268e-01
5.91211505e-02 -2.45619495e-03 -2.62632936e-01 -7.71302223e-01
-2.13842005e-01 -7.88667560e-01 7.84319937e-01 -7.72828341e-01
-1.12258625e+00 3.09732944e-01 -2.09891468e-01 -1.36568975e+00
-2.59318978e-01 -1.25019521e-01 -1.13387072e+00 5.58412194e-01
-2.80961722e-01 -9.77064073e-01 -1.65391732e-02 5.89061618e-01
6.93006039e-01 -9.50068161e-02 4.25606877e-01 4.44200993e-01
-3.27628195e-01 5.50944209e-01 -2.42751777e-01 3.89539540e-01
8.36977363e-01 -8.97514641e-01 3.89888167e-01 4.39571440e-01
2.08117515e-01 -2.12974265e-01 1.00414658e+00 -9.24059272e-01
-6.84771538e-01 -6.01421654e-01 1.08507156e+00 -5.86541712e-01
1.01185036e+00 -3.81530851e-01 -5.02273977e-01 7.87891924e-01
-3.57196718e-01 -3.33941996e-01 1.08389711e+00 6.28207982e-01
1.04367934e-01 -2.43124366e-01 -7.38552392e-01 7.49981344e-01
1.00174010e+00 -5.41474938e-01 -5.90485930e-01 3.29282343e-01
3.75573263e-02 -7.22954273e-01 -8.98371994e-01 3.04971635e-03
9.33622718e-01 -1.26145923e+00 8.51905048e-01 -7.76154280e-01
8.34471345e-01 7.85822049e-02 2.06708074e-01 -1.25150108e+00
-1.95347778e-02 -6.50874138e-01 4.65183765e-01 1.08594966e+00
2.49279812e-01 -5.26618995e-02 9.52655673e-01 8.88535261e-01
-4.15930822e-02 -7.44107068e-01 -1.14534342e+00 -5.13146937e-01
1.54515162e-01 -1.17717910e+00 3.79837722e-01 7.58093476e-01
5.69365263e-01 1.55293852e-01 -9.09581840e-01 -1.25833392e-01
3.40091467e-01 -3.73426266e-02 1.10814095e+00 -1.33260369e+00
-5.19971788e-01 -3.14649194e-01 -1.03594923e+00 -6.01500928e-01
-5.32649636e-01 -6.68972790e-01 -8.06878135e-02 -1.14070296e+00
1.49935961e-01 -2.64837801e-01 7.55627751e-02 5.27227640e-01
1.51546270e-01 9.41644192e-01 2.47845262e-01 1.23547710e-01
-6.29833102e-01 2.89383996e-02 8.49920630e-01 1.67188630e-01
-1.29097015e-01 4.72566932e-01 -5.60265243e-01 8.42633486e-01
5.94813466e-01 -7.88120151e-01 7.01342225e-02 -1.00579500e-01
9.25153255e-01 5.50042689e-01 4.64817166e-01 -1.36333621e+00
4.18028265e-01 -3.17911834e-01 2.25441515e-01 -5.91127455e-01
6.22059822e-01 -2.47503564e-01 7.14880824e-01 4.44775969e-01
-3.48386675e-01 8.10543075e-02 3.13516021e-01 6.52603626e-01
-3.18862647e-01 1.14795342e-01 4.19912457e-01 -1.63589537e-01
-6.48515940e-01 4.54552591e-01 -8.50135863e-01 4.34566498e-01
1.34252751e+00 -5.51745296e-01 7.01986719e-03 -9.40546215e-01
-1.18572009e+00 2.00296015e-01 2.09247291e-01 4.74665523e-01
1.94782555e-01 -1.29050207e+00 -9.93499994e-01 -2.09957138e-01
3.89995277e-02 -3.78882140e-01 8.37629259e-01 7.58928537e-01
-8.13172936e-01 -4.42705631e-01 -4.34626251e-01 -1.79208562e-01
-1.34850001e+00 1.00456454e-01 3.32693607e-01 -4.31876779e-01
-9.79959369e-01 1.07641077e+00 -5.53047024e-02 -1.88653320e-01
-1.58786587e-02 1.84761465e-01 -1.19163536e-01 4.07908320e-01
7.59858727e-01 4.86080348e-01 -9.76380184e-02 -8.92935693e-01
-4.39735472e-01 1.60985395e-01 2.68034916e-02 -1.74983233e-01
1.53745401e+00 1.52279541e-01 5.27654707e-01 8.97069633e-01
6.71905577e-01 2.16068253e-01 -1.40707350e+00 -5.70962988e-02
-5.28975248e-01 -5.69355547e-01 -1.79307193e-01 -8.94840658e-01
-1.08119547e+00 8.82493079e-01 2.53244370e-01 3.01108479e-01
6.06146097e-01 2.25910977e-01 1.27920508e+00 -2.68272907e-01
5.84506094e-01 -1.24696124e+00 2.00523920e-02 4.38979834e-01
5.68714321e-01 -9.72146988e-01 -6.01394959e-02 2.36197948e-01
-9.15516734e-01 9.78136301e-01 7.00335920e-01 -4.30825651e-01
4.54552710e-01 3.65883619e-01 1.94510594e-01 -5.43623269e-01
-8.84309351e-01 -3.74926329e-02 -1.68012187e-01 3.18517536e-01
1.01396833e-02 1.60037756e-01 -5.32960951e-01 1.01605952e+00
-8.66316140e-01 -9.96509269e-02 7.41612434e-01 7.48918831e-01
-3.85912836e-01 -6.75689161e-01 -3.52626979e-01 7.56901920e-01
-8.87086332e-01 1.15067728e-01 -3.07686359e-01 8.90890241e-01
4.80651826e-01 8.98668230e-01 4.73109722e-01 -8.51356506e-01
5.64986646e-01 -1.01470806e-01 -6.74050376e-02 -5.43594182e-01
-1.31291699e+00 -3.62947881e-01 4.30352479e-01 -8.64123821e-01
1.74747527e-01 -9.08269405e-01 -1.20464778e+00 -1.28520596e+00
-2.04211008e-02 2.62088895e-01 2.31627300e-01 1.05182219e+00
-2.19750181e-02 3.69211853e-01 2.54004806e-01 -9.02693510e-01
1.02072157e-01 -8.55694652e-01 -1.24042785e+00 5.66705227e-01
-1.75823599e-01 -6.74805403e-01 -2.69981205e-01 -1.39971852e-01] | [6.631248950958252, 0.35377177596092224] |
7ee0fbb4-96cc-4581-a600-1b2d86fd4c69 | rotated-object-detection-via-scale-invariant | 2204.00840 | null | https://arxiv.org/abs/2204.00840v2 | https://arxiv.org/pdf/2204.00840v2.pdf | Rotated Object Detection via Scale-invariant Mahalanobis Distance in Aerial Images | Rotated object detection in aerial images is a meaningful yet challenging task as objects are densely arranged and have arbitrary orientations. The eight-parameter (coordinates of box vectors) methods in rotated object detection usually use ln-norm losses (L1 loss, L2 loss, and smooth L1 loss) as loss functions. As ln-norm losses are mainly based on non-scale-invariant Minkowski distance, using ln-norm losses will lead to inconsistency with the detection metric rotational Intersection-over-Union (IoU) and training instability. To address the problems, we use Mahalanobis distance to calculate loss between the predicted and the target box vertices' vectors, proposing a new loss function called Mahalanobis Distance Loss (MDL) for eight-parameter rotated object detection. As Mahalanobis distance is scale-invariant, MDL is more consistent with detection metric and more stable during training than ln-norm losses. To alleviate the problem of boundary discontinuity like all other eight-parameter methods, we further take the minimum loss value to make MDL continuous at boundary cases. We achieve state-of-art performance on DOTA-v1.0 with the proposed method MDL. Furthermore, compared to the experiment that uses smooth L1 loss, we find that MDL performs better in rotated object detection. | ['Yi Liu', 'Ruijie Wu', 'Wei Guo', 'Siyang Wen'] | 2022-04-02 | null | null | null | null | ['object-detection-in-aerial-images'] | ['computer-vision'] | [-2.28224173e-01 -3.02195400e-01 -5.01793111e-03 -1.95684910e-01
-4.62994188e-01 -5.13523519e-01 1.23486131e-01 -2.35059544e-01
-5.79373360e-01 4.82985258e-01 -3.37832659e-01 -1.53544417e-03
-3.73150259e-01 -6.49809122e-01 -5.64599574e-01 -6.59448564e-01
-3.93361688e-01 1.82437360e-01 7.32655346e-01 -1.89104825e-01
2.23134175e-01 9.59368348e-01 -1.35615790e+00 6.27767369e-02
5.58591366e-01 1.24749875e+00 1.03912294e-01 5.80128789e-01
3.65612954e-01 3.03253591e-01 -7.97241330e-01 -3.32167804e-01
7.96682596e-01 -1.54718444e-01 -5.73958933e-01 3.28296423e-02
5.10390759e-01 -5.17767847e-01 -2.80695498e-01 1.27952981e+00
5.55321217e-01 2.27164909e-01 9.22081351e-01 -1.39358282e+00
-4.28891867e-01 1.95544913e-01 -1.17795384e+00 2.03426614e-01
1.91740572e-01 1.61847826e-02 9.64872420e-01 -1.25430357e+00
5.07055759e-01 1.46340740e+00 1.01220036e+00 2.16317594e-01
-1.08693480e+00 -6.50439262e-01 4.05365080e-01 -1.34768009e-01
-1.79647553e+00 3.47605020e-01 6.00001633e-01 -5.03904223e-01
6.40669167e-01 5.30173659e-01 4.37146306e-01 5.75236976e-01
2.60358214e-01 9.35885787e-01 5.42562306e-01 -3.34856719e-01
-2.47444153e-01 2.23776966e-01 -2.73322314e-01 9.01054144e-01
4.88895983e-01 4.26249318e-02 2.40116403e-01 3.87973920e-03
1.07704437e+00 1.77602455e-01 -2.75330782e-01 -7.67922699e-01
-1.25239420e+00 9.47282314e-01 8.56509030e-01 5.87103665e-02
4.98962589e-03 8.09840187e-02 4.29289579e-01 8.27278420e-02
3.93324435e-01 5.39341390e-01 -5.10142863e-01 4.71965671e-01
-8.99662197e-01 4.16054606e-01 2.01875746e-01 1.12103474e+00
4.20492649e-01 -9.12117772e-03 -3.52452010e-01 9.19352829e-01
5.61567962e-01 6.42011642e-01 3.74603570e-01 -5.58169425e-01
4.64033455e-01 7.80551612e-01 2.34708950e-01 -1.43628383e+00
-5.37206650e-01 -5.38320184e-01 -8.78628492e-01 4.70128715e-01
5.29917538e-01 8.50060508e-02 -7.61285067e-01 1.37973070e+00
3.57849181e-01 -1.71156690e-01 -2.43416622e-01 1.28269041e+00
8.76077712e-01 6.17697775e-01 -2.65587628e-01 -1.51176332e-02
1.30830503e+00 -7.51641214e-01 -3.76157343e-01 -8.11607111e-03
6.37205303e-01 -1.02505422e+00 1.20557415e+00 4.19754058e-01
-1.03522372e+00 -7.20258117e-01 -1.24854434e+00 5.84377442e-03
-4.35892582e-01 6.93944454e-01 5.72417736e-01 4.08923984e-01
-4.07187283e-01 4.73397732e-01 -6.47293329e-01 -2.62203455e-01
4.10789251e-01 2.10817084e-01 -3.05179119e-01 1.23541251e-01
-7.49795198e-01 8.26851904e-01 3.26416552e-01 9.88043919e-02
-5.69269001e-01 -5.99405766e-01 -8.22286904e-01 -1.39814854e-01
4.35464174e-01 -3.70642871e-01 8.60753179e-01 -3.90634745e-01
-1.16323280e+00 9.56899822e-01 3.87905508e-01 -3.89228344e-01
9.62121904e-01 -3.53467494e-01 -2.20298365e-01 2.23974511e-02
1.06331393e-01 8.75469863e-01 7.99104154e-01 -1.17056990e+00
-7.43924499e-01 -5.26513159e-01 1.26581818e-01 3.61258328e-01
-1.36752605e-01 4.62887548e-02 -3.68384987e-01 -8.75731766e-01
6.36315823e-01 -7.30846286e-01 -3.19827199e-02 7.48843968e-01
-3.51294845e-01 -3.25402439e-01 1.27843940e+00 -4.51207280e-01
1.38336408e+00 -2.29592586e+00 -7.37961158e-02 -1.08122025e-02
-1.28304318e-01 4.10986036e-01 -1.12134703e-01 -9.32409689e-02
-1.24743052e-01 4.40098457e-02 -4.41623658e-01 -5.73260412e-02
-1.89268485e-01 -2.32918367e-01 -3.34990829e-01 8.91815960e-01
4.61626887e-01 5.21410584e-01 -9.06123698e-01 -6.27360642e-01
4.34740961e-01 5.11208177e-01 -4.39499706e-01 -8.04707184e-02
2.34520331e-01 -1.38760339e-02 -3.67632091e-01 8.13621163e-01
9.87111330e-01 1.00103363e-01 -6.30252779e-01 -6.24426901e-01
-2.71093100e-01 -4.21693146e-01 -1.59373045e+00 1.35645676e+00
-1.37997419e-01 7.71881759e-01 3.77125740e-02 -8.82662237e-01
1.31248629e+00 -1.01713024e-01 4.51915264e-01 -1.59879297e-01
2.31420040e-01 -5.18557131e-02 -1.94030359e-01 -2.80498624e-01
4.97422189e-01 -6.81149261e-03 1.65728912e-01 -3.38395238e-01
-3.01841348e-01 -5.83565176e-01 2.21965030e-01 -3.96161303e-02
5.58604300e-01 2.80243874e-01 4.36216086e-01 -2.00658292e-01
6.26706004e-01 -4.98643480e-02 7.63234258e-01 4.34869140e-01
-3.65364432e-01 1.00136876e+00 5.12651145e-01 -5.78843176e-01
-1.00574279e+00 -1.00868762e+00 -7.84896731e-01 7.51463294e-01
5.90137839e-01 -2.18408220e-02 -3.85944933e-01 -8.32646489e-01
3.00900698e-01 6.39085233e-01 -4.51123208e-01 -1.50201738e-01
-5.71019769e-01 -9.13652956e-01 5.92649639e-01 6.82000816e-01
6.68404639e-01 -6.53471947e-01 -7.55920172e-01 -1.42948646e-02
9.09690037e-02 -8.90108645e-01 -6.28296137e-01 2.77530812e-02
-9.02941704e-01 -1.31920028e+00 -1.08116853e+00 -8.21976602e-01
9.04911578e-01 6.11081839e-01 8.83811355e-01 -2.33677968e-01
-1.03057516e+00 2.88804676e-02 -5.09734511e-01 -4.71996635e-01
3.02722573e-01 -3.06012452e-01 6.30251169e-02 -2.66809165e-01
2.53671885e-01 9.39299986e-02 -9.00616646e-01 9.98542905e-01
-8.28281879e-01 -2.77049869e-01 4.86089230e-01 7.97111213e-01
6.84869766e-01 3.13968390e-01 2.00080782e-01 -3.60868752e-01
4.11513001e-01 -3.57876979e-02 -1.12351012e+00 3.27109307e-01
-5.09185195e-01 -3.34216624e-01 4.15268898e-01 -6.09784007e-01
-6.11190796e-01 1.19461402e-01 3.00537348e-01 -6.61868513e-01
2.20847696e-01 9.07079689e-03 5.66301905e-02 -4.03991729e-01
8.23300719e-01 -7.86426216e-02 -2.44096607e-01 -2.49581546e-01
2.28085384e-01 6.43227398e-01 2.51845509e-01 -2.54843742e-01
8.45470011e-01 6.74386799e-01 2.60703355e-01 -9.48188722e-01
-7.85097241e-01 -8.30245495e-01 -6.11874759e-01 -2.76834786e-01
8.53850305e-01 -8.29310656e-01 -9.08649862e-01 3.17686915e-01
-1.27865469e+00 1.43576294e-01 -3.95688683e-01 8.52779448e-01
-2.39598572e-01 5.27034640e-01 -3.33118111e-01 -9.82676923e-01
-2.56986409e-01 -1.24226844e+00 1.21404731e+00 1.65795013e-01
2.44002074e-01 -6.27238214e-01 -1.69686973e-01 -7.22159073e-02
1.54267177e-02 4.92866069e-01 4.13609743e-01 -1.37081116e-01
-3.68112862e-01 -7.07551718e-01 -6.88429713e-01 6.95386410e-01
-1.97073016e-02 3.62904996e-01 -6.25894368e-01 -6.23625159e-01
1.14008971e-01 -3.10159057e-01 1.01787400e+00 6.83217287e-01
1.46361804e+00 -8.28037411e-02 -4.41043526e-01 8.93403947e-01
1.33188796e+00 2.85651535e-01 6.34430587e-01 4.10130858e-01
6.88150525e-01 5.70995033e-01 1.24602127e+00 5.22118390e-01
-5.65082431e-02 7.59102762e-01 6.73542202e-01 -2.77593464e-01
-5.69878034e-02 -3.39416303e-02 3.09150964e-01 3.48893583e-01
-2.46812567e-01 -2.46542946e-01 -4.75762486e-01 2.60854244e-01
-1.76271427e+00 -9.27012026e-01 -2.63291359e-01 2.45307446e+00
3.82431656e-01 4.02241677e-01 2.40159556e-01 1.35759890e-01
8.27316940e-01 1.90098196e-01 -7.47700036e-01 -6.64292276e-02
-3.90095860e-01 -4.05381322e-01 1.04016495e+00 4.11856323e-01
-1.60380423e+00 9.68358874e-01 5.94431877e+00 1.09129655e+00
-1.06271470e+00 -2.42308542e-01 4.47310835e-01 7.54229575e-02
3.97561461e-01 -1.90724909e-01 -1.18580973e+00 3.92734200e-01
-2.10325554e-01 2.75045723e-01 1.17196336e-01 1.36095428e+00
1.56484619e-01 -8.33579525e-02 -8.45577836e-01 1.08724344e+00
8.84433910e-02 -7.34682322e-01 -1.54659981e-02 -1.90974399e-01
7.20141411e-01 -1.92407295e-01 2.34975785e-01 2.37512231e-01
-8.38605464e-02 -9.38129961e-01 7.60851741e-01 2.53413647e-01
9.84445155e-01 -8.77980590e-01 8.22740912e-01 1.94794700e-01
-1.64939237e+00 -2.18056634e-01 -1.06512463e+00 2.62628168e-01
-3.91515996e-03 4.47483599e-01 -9.26395178e-01 3.65276098e-01
9.06298399e-01 6.36575103e-01 -5.98689854e-01 1.42995822e+00
-1.08978167e-01 -2.10753572e-03 -4.16795880e-01 -6.42036870e-02
3.56019437e-01 -5.11696577e-01 8.84367704e-01 1.28293872e+00
2.99163043e-01 -1.60617158e-02 3.90632033e-01 8.02347183e-01
2.13628456e-01 6.10137843e-02 -7.03052342e-01 4.47108030e-01
3.25402945e-01 1.36251128e+00 -1.04825759e+00 -6.06094673e-02
-3.03172827e-01 8.14885020e-01 -1.83997750e-02 1.04574278e-01
-1.11831558e+00 -9.19188201e-01 5.39031923e-01 3.73211145e-01
3.90548825e-01 -3.14779937e-01 -1.35287270e-02 -8.91453445e-01
1.50532424e-01 -6.62874043e-01 3.84282351e-01 -7.77153611e-01
-1.32601762e+00 6.04580343e-01 3.26902121e-01 -1.72704327e+00
4.03442264e-01 -1.15627015e+00 -5.81561744e-01 5.00089407e-01
-1.32853317e+00 -1.09502363e+00 -5.34803033e-01 2.35170245e-01
7.77145147e-01 -1.70210496e-01 3.93612593e-01 4.18689460e-01
-4.54105198e-01 7.72561193e-01 1.14226632e-01 2.69172609e-01
9.08094585e-01 -1.28434324e+00 2.40375921e-01 7.11440921e-01
-9.04176608e-02 5.01069129e-01 5.68156064e-01 -5.19616902e-01
-9.96290743e-01 -1.27838492e+00 1.00918107e-01 -3.69863868e-01
3.30670029e-01 -2.29294047e-01 -7.36482084e-01 5.03612161e-01
-5.15927255e-01 3.88306350e-01 1.22785941e-01 -2.91946799e-01
-1.81328312e-01 -2.95349061e-01 -1.42757630e+00 4.30859417e-01
1.08199346e+00 -3.75193432e-02 -3.46733809e-01 6.27913773e-01
8.82772684e-01 -4.19042915e-01 -7.89287508e-01 9.69168603e-01
5.77346504e-01 -9.28915262e-01 1.44029784e+00 -5.46186149e-01
-6.54593036e-02 -9.12666440e-01 -1.40559331e-01 -8.76867473e-01
-4.96007651e-01 -3.38526100e-01 2.68667489e-01 8.32033396e-01
3.36964279e-01 -5.66821396e-01 7.35539019e-01 2.00173128e-02
-1.27090454e-01 -9.72989202e-01 -8.50330412e-01 -1.24011242e+00
-4.94414531e-02 -1.94766775e-01 3.01753551e-01 8.20195913e-01
-3.93088102e-01 1.71394348e-02 -3.46496254e-01 2.87569374e-01
7.06225932e-01 -7.74362460e-02 7.63522685e-01 -1.25836515e+00
3.21612954e-02 -6.43233955e-01 -8.06259394e-01 -1.39318395e+00
-3.90748650e-01 -7.70119607e-01 7.92701989e-02 -1.34182096e+00
3.17126103e-02 -5.42601466e-01 -2.36875132e-01 1.76452443e-01
1.07699726e-03 4.22972262e-01 7.95307457e-02 1.53026566e-01
-4.85120237e-01 6.61594570e-01 1.62382364e+00 -1.17426626e-01
-4.50791031e-01 1.80096284e-01 -3.63664217e-02 1.11203027e+00
5.22594750e-01 -5.60066938e-01 -2.34418035e-01 -3.38403493e-01
1.37778819e-01 -2.69324809e-01 4.69190568e-01 -1.12338352e+00
2.26852968e-01 -1.25676125e-01 7.63628781e-01 -1.11368275e+00
4.55234766e-01 -8.71225476e-01 -3.16898733e-01 5.69540024e-01
-1.30467877e-01 2.81942338e-02 1.12538576e-01 4.44240749e-01
-2.01375484e-01 -3.22520077e-01 1.20556700e+00 -7.85267055e-02
-4.96271729e-01 3.92775476e-01 7.48196170e-02 9.51197371e-03
1.38638055e+00 -6.43575609e-01 -1.54157832e-01 1.82645425e-01
-4.99821633e-01 3.73446226e-01 3.58007997e-01 2.47983560e-01
9.25854802e-01 -1.34529710e+00 -7.37047255e-01 2.44072780e-01
1.33624464e-01 6.41546786e-01 -1.60629079e-01 8.41582656e-01
-1.04339910e+00 3.72508645e-01 -7.72475004e-02 -7.91697502e-01
-1.42841077e+00 4.95075852e-01 3.81255150e-01 6.63756812e-03
-6.12302899e-01 1.30767477e+00 5.14780164e-01 -4.20341700e-01
4.99714196e-01 -7.06489980e-01 -1.19076945e-01 1.02234751e-01
3.36899400e-01 8.30132306e-01 -1.90435797e-01 -5.10280490e-01
-5.14754951e-01 9.80395973e-01 -9.30686146e-02 1.35226011e-01
1.08474541e+00 2.43835330e-01 8.51620585e-02 2.15735421e-01
1.19326615e+00 -1.88227549e-01 -1.47495222e+00 1.43016502e-02
-2.12467924e-01 -8.46215427e-01 9.07526016e-02 -4.16411102e-01
-8.84343445e-01 8.28079820e-01 9.55394328e-01 2.99746424e-01
8.62972379e-01 -2.02986896e-01 5.64390719e-01 4.68466341e-01
1.04693778e-01 -1.14924037e+00 4.11455691e-01 4.06136513e-01
1.29264402e+00 -1.53287578e+00 6.73216462e-01 -6.28923535e-01
-6.03648365e-01 1.18067610e+00 9.32874680e-01 -4.62117314e-01
6.49905980e-01 -3.20301927e-03 -1.24541067e-01 -1.32125005e-01
-4.74566929e-02 -1.63021207e-01 7.38614202e-01 4.30022329e-01
3.46417278e-01 -1.18939131e-02 -4.80947167e-01 3.14745456e-01
-6.97543770e-02 -4.97570783e-01 -1.16539998e-02 7.79568613e-01
-6.57375872e-01 -4.63739365e-01 -5.97851276e-01 3.59093279e-01
-1.49482712e-01 3.47972184e-01 -2.28727311e-01 1.18355179e+00
2.84487188e-01 5.69980562e-01 1.57201976e-01 -2.10107297e-01
6.93375647e-01 -5.73538959e-01 5.74261665e-01 -4.57180947e-01
-2.87445430e-02 9.39732343e-02 -5.15174091e-01 -5.05101681e-01
-2.48883978e-01 -3.46456259e-01 -1.30294359e+00 -9.16054025e-02
-1.02549827e+00 -1.83892742e-01 8.20591927e-01 4.00164694e-01
-2.88663898e-02 4.20270920e-01 7.60201395e-01 -8.46430540e-01
-1.06024671e+00 -8.40267360e-01 -5.93662679e-01 2.65875846e-01
1.28946066e-01 -9.91934121e-01 -6.49631381e-01 -1.25145987e-01] | [8.684552192687988, -0.7592396140098572] |
4b1a8e6b-d76f-45c7-81d8-77b37a18f322 | concra-a-convolutional-neural-network-code | 2009.01959 | null | https://arxiv.org/abs/2009.01959v1 | https://arxiv.org/pdf/2009.01959v1.pdf | CoNCRA: A Convolutional Neural Network Code Retrieval Approach | Software developers routinely search for code using general-purpose search engines. However, these search engines cannot find code semantically unless it has an accompanying description. We propose a technique for semantic code search: A Convolutional Neural Network approach to code retrieval (CoNCRA). Our technique aims to find the code snippet that most closely matches the developer's intent, expressed in natural language. We evaluated our approach's efficacy on a dataset composed of questions and code snippets collected from Stack Overflow. Our preliminary results showed that our technique, which prioritizes local interactions (words nearby), improved the state-of-the-art (SOTA) by 5% on average, retrieving the most relevant code snippets in the top 3 (three) positions by almost 80% of the time. Therefore, our technique is promising and can improve the efficacy of semantic code retrieval. | ['Marco A. Gerosa', 'Marcelo de Rezende Martins'] | 2020-09-03 | null | null | null | null | ['code-search', 'code-search'] | ['computer-code', 'computer-vision'] | [-2.47521177e-01 -8.96399692e-02 -1.99024335e-01 1.28721505e-01
-8.76132607e-01 -6.62365496e-01 2.30120748e-01 5.93370140e-01
-2.22574770e-01 -1.53501019e-01 3.73224378e-01 -6.58672452e-01
-3.97879988e-01 -6.80420995e-01 -6.66039646e-01 6.32793754e-02
-1.18867114e-01 -2.07969710e-01 4.97300982e-01 -4.04976845e-01
1.06249237e+00 -1.47526145e-01 -1.91314197e+00 6.91175580e-01
1.09365833e+00 6.06825709e-01 9.01675463e-01 5.36307335e-01
-9.74506319e-01 1.14223623e+00 -7.89348543e-01 -4.02460843e-01
-3.35256308e-01 -3.36011738e-01 -1.29169691e+00 -9.75649118e-01
5.04623532e-01 1.16309844e-01 -2.47984473e-02 1.41639268e+00
2.98526376e-01 -1.49646357e-01 1.25800878e-01 -8.14970315e-01
-1.06099677e+00 8.79511118e-01 -4.51571703e-01 5.75522363e-01
8.29244316e-01 -4.96712416e-01 1.43889594e+00 -1.10241807e+00
7.85525382e-01 7.89239585e-01 9.36148465e-01 5.64755797e-01
-6.83250368e-01 -4.23489094e-01 -3.48957658e-01 1.22317120e-01
-1.47560656e+00 -9.04471874e-02 6.49282455e-01 -8.43354464e-01
1.87966096e+00 3.30093861e-01 3.99997741e-01 5.41286767e-01
2.73015559e-01 5.31283200e-01 -7.27293268e-02 -9.54895198e-01
2.25400731e-01 -1.18511096e-01 5.71691513e-01 1.14601922e+00
3.40983242e-01 -4.95958775e-01 -4.85431790e-01 -8.23222458e-01
9.71418768e-02 6.22305386e-02 -3.51244628e-01 -1.23049142e-02
-7.04519093e-01 8.24392319e-01 3.76267761e-01 9.24331486e-01
-6.81917667e-02 6.39685869e-01 6.91190779e-01 2.73555070e-01
3.73204380e-01 1.11807144e+00 -6.21914625e-01 -6.51584387e-01
-9.42865074e-01 4.73462552e-01 1.01190054e+00 1.29301465e+00
9.34307456e-01 -7.31024563e-01 -1.33811712e-01 1.16615760e+00
6.25764668e-01 1.83147818e-01 7.96450496e-01 -8.08258891e-01
6.85763955e-01 1.21199286e+00 -5.77727854e-02 -1.27370048e+00
-1.35267839e-01 -4.92944837e-01 3.90910059e-01 3.38827334e-02
-4.37458716e-02 1.67412579e-01 -3.81875962e-01 1.15763950e+00
-4.98418659e-01 -3.35210472e-01 -1.81822449e-01 4.04016376e-01
1.07714677e+00 2.86673248e-01 3.76934893e-02 4.00457352e-01
1.45874059e+00 -1.06114852e+00 -6.25987172e-01 -5.58537960e-01
1.20008588e+00 -1.03270280e+00 1.21306074e+00 -6.84673190e-02
-8.06491554e-01 -3.17967236e-01 -9.77460086e-01 -1.16976149e-01
-5.91698349e-01 3.99072379e-01 5.86770356e-01 5.41408479e-01
-1.68424153e+00 4.52399522e-01 -6.68198049e-01 -6.49063110e-01
2.86710590e-01 -2.72948015e-02 -5.03887385e-02 3.73711088e-03
-9.22152042e-01 7.04241693e-01 8.57536048e-02 -7.02066481e-01
-7.20922768e-01 -9.62746680e-01 -9.05554116e-01 6.97995543e-01
2.04835594e-01 -2.76781440e-01 1.52225280e+00 -1.00686598e+00
-5.44484079e-01 1.04203486e+00 -3.62479001e-01 -1.33440793e-01
-3.57627183e-01 -4.40426052e-01 -3.81871074e-01 6.15416393e-02
5.58704197e-01 2.84054399e-01 2.74866760e-01 -8.00239623e-01
-5.12871802e-01 -8.44183005e-03 5.08663595e-01 -3.15860420e-01
-8.40408087e-01 8.35643470e-01 -7.22438872e-01 -4.80552554e-01
-2.78335720e-01 -6.48703933e-01 -3.11929882e-02 1.23795018e-01
1.46349147e-01 -6.82054222e-01 6.31222308e-01 -7.95719445e-01
2.04350710e+00 -2.41364646e+00 -1.38850048e-01 -4.80037630e-02
5.28453231e-01 2.56373197e-01 -3.68650705e-01 9.32971239e-01
-3.39021772e-01 6.73903584e-01 -2.60578930e-01 8.27913359e-02
-5.76100266e-03 -3.69944334e-01 -1.93482906e-01 -1.24716640e-01
1.26677498e-01 1.13113606e+00 -1.23890674e+00 -4.04646963e-01
-3.45864475e-01 3.92015785e-01 -9.76685703e-01 1.10885093e-03
-3.46111387e-01 -5.64816892e-01 -7.84801781e-01 6.61822796e-01
1.59430265e-01 -7.82230318e-01 -1.59864172e-01 5.13321638e-01
-3.17268252e-01 6.52712464e-01 -1.75728276e-01 2.05436802e+00
-9.30772483e-01 1.31045413e+00 -4.43438768e-01 -3.77953857e-01
1.02911758e+00 2.91625381e-01 3.32400262e-01 -9.22931433e-01
-4.89586920e-01 5.55553138e-01 -3.20248425e-01 -1.10862386e+00
3.98591459e-01 6.76462948e-01 -1.53985098e-01 5.96286297e-01
-1.56968206e-01 3.19732487e-01 1.85255725e-02 3.64552796e-01
2.09827518e+00 3.02482327e-03 3.75480235e-01 -8.60399842e-01
5.86397886e-01 2.36169368e-01 -1.52920261e-01 1.00052726e+00
1.50062844e-01 4.34922248e-01 6.56388342e-01 -6.84418559e-01
-6.52431786e-01 -3.77668500e-01 1.77677929e-01 1.48712230e+00
1.76630005e-01 -1.09407544e+00 -1.27813530e+00 -9.95924771e-01
-1.13458000e-01 6.28390014e-01 -8.31237376e-01 -3.14623415e-01
-6.12621367e-01 -9.14200991e-02 7.32070982e-01 4.55392659e-01
2.51661420e-01 -1.19309282e+00 -1.06097555e+00 1.99837402e-01
-2.74344683e-01 -5.02141476e-01 -6.90962195e-01 9.55821946e-03
-5.93891740e-01 -1.37121367e+00 -7.59885371e-01 -9.33777571e-01
8.25354755e-01 5.66439569e-01 1.73685288e+00 1.28977430e+00
-4.77606624e-01 3.41203749e-01 -9.75186288e-01 2.48095747e-02
-5.56890488e-01 3.79244953e-01 -8.75537574e-01 -9.31649268e-01
1.02954245e+00 -2.61519045e-01 -5.91860712e-01 -4.94151600e-02
-1.15145552e+00 -4.60628867e-01 2.69788116e-01 3.56270790e-01
-3.62926960e-01 -1.46236345e-01 2.77175516e-01 -6.73676491e-01
1.04406381e+00 -8.02459657e-01 -7.68696666e-01 5.04317105e-01
-8.41997802e-01 3.52531016e-01 3.54640931e-01 -4.91062701e-02
-8.91244948e-01 -3.47007699e-02 -2.39793181e-01 -7.72279277e-02
1.15979441e-01 1.27357733e+00 6.53046370e-01 -5.97816825e-01
1.34495199e+00 9.50834975e-02 -3.85906577e-01 -8.96021485e-01
9.44588557e-02 8.54788303e-01 2.69727483e-02 -7.32353806e-01
4.16173309e-01 2.47312319e-02 -5.82898498e-01 -3.33979756e-01
-3.54292214e-01 -9.89539683e-01 -2.09586799e-01 -3.56030434e-01
9.11204457e-01 -7.58382916e-01 -3.56330812e-01 1.28820315e-02
-1.62864363e+00 1.90971717e-01 3.30493838e-01 -2.18791943e-02
-1.17208876e-01 5.45352995e-01 -3.49380761e-01 -3.95734638e-01
-4.03634757e-01 -1.33415675e+00 1.32300985e+00 1.71746071e-02
-6.02440417e-01 -8.73242915e-01 5.90729892e-01 2.20905825e-01
1.00803816e+00 -2.51624703e-01 1.47485697e+00 -7.92713523e-01
-6.58677518e-01 -2.47939780e-01 -3.17919582e-01 -1.85826272e-01
2.42667079e-01 1.18943125e-01 -7.93244421e-01 -1.63813516e-01
-3.23482931e-01 1.25115484e-01 6.93575382e-01 -1.59527555e-01
1.34527135e+00 -1.16768293e-01 -6.30206525e-01 1.19801357e-01
1.93661845e+00 5.28645992e-01 6.42786741e-01 8.34400237e-01
5.91156840e-01 4.07273293e-01 2.63749331e-01 6.08324349e-01
1.83549479e-01 8.27604830e-01 8.94174814e-01 5.36136687e-01
-1.94855601e-01 -3.38067375e-02 3.54589343e-01 1.09185302e+00
1.79375768e-01 -1.52662367e-01 -1.77429044e+00 1.13843358e+00
-1.72354364e+00 -5.82270145e-01 -1.58281401e-01 1.91987538e+00
8.48725796e-01 -2.35902458e-01 -3.85752827e-01 -4.81987655e-01
6.87490582e-01 -4.06071432e-02 -1.20014600e-01 -3.34416002e-01
4.71331060e-01 3.96414846e-01 3.23987067e-01 3.11802208e-01
-7.01500893e-01 7.29849637e-01 6.39942646e+00 8.29911053e-01
-6.84767485e-01 2.90343881e-01 -1.11353852e-01 2.92310148e-01
-7.03337610e-01 2.02449068e-01 -6.24115169e-01 5.69102943e-01
1.04896498e+00 -4.40336585e-01 6.99151814e-01 1.51902115e+00
-2.92079091e-01 1.41020147e-02 -1.03683424e+00 7.88533986e-01
3.51810873e-01 -1.78065419e+00 -1.96975499e-01 -2.29911074e-01
8.14020038e-01 3.02981019e-01 -2.60034084e-01 4.58365917e-01
2.78911680e-01 -8.11812520e-01 6.44242644e-01 4.14737135e-01
4.35201764e-01 -4.89236295e-01 6.93706632e-01 8.76403302e-02
-1.22899723e+00 -4.14741337e-01 -2.83103466e-01 3.20495702e-02
-6.58365905e-01 4.88012940e-01 -5.73974013e-01 2.36757472e-01
1.18552184e+00 8.67068708e-01 -1.19254959e+00 1.48251462e+00
3.08314543e-02 3.03775638e-01 2.75449216e-01 -7.41195500e-01
3.97819936e-01 4.83819932e-01 3.80209148e-01 1.67010689e+00
6.53836966e-01 -3.90150517e-01 -3.47141325e-01 1.36499524e+00
-2.50642270e-01 2.65625805e-01 -8.52543890e-01 -2.95556277e-01
6.00865722e-01 7.96318114e-01 -7.97429204e-01 -1.56965703e-01
-9.47342753e-01 7.35839069e-01 3.38716865e-01 3.11880350e-01
-7.20197976e-01 -1.35242581e+00 7.46803522e-01 -1.46398172e-01
5.54255009e-01 2.11279616e-01 8.60314742e-02 -9.48528349e-01
5.18918693e-01 -9.38601017e-01 2.85993516e-01 -1.09385788e+00
-9.40400720e-01 7.72118568e-01 -7.18950108e-02 -1.21557760e+00
-3.27134967e-01 -4.59014475e-01 -6.17947102e-01 7.43393660e-01
-1.24342299e+00 -4.78177398e-01 -4.64260995e-01 -3.24861482e-02
7.65611947e-01 -3.67732704e-01 8.50768447e-01 8.81877005e-01
6.90138638e-02 5.01810372e-01 1.84773520e-01 3.24084401e-01
2.64513969e-01 -1.14518976e+00 9.48194504e-01 8.02360952e-01
3.04007232e-01 1.43131328e+00 5.50836623e-01 -6.44087017e-01
-1.30415118e+00 -8.91913831e-01 1.37173522e+00 -7.53129065e-01
9.14754212e-01 -6.09566346e-02 -1.04283381e+00 3.93200636e-01
5.04352450e-01 -1.30652934e-01 5.38926899e-01 1.02049284e-01
-8.07418108e-01 4.12309140e-01 -7.48861790e-01 5.54609776e-01
1.01761770e+00 -1.30588531e+00 -6.82853341e-01 4.00982857e-01
1.14756298e+00 -2.99647480e-01 -7.41574407e-01 -6.06036074e-02
2.15583563e-01 -8.46729755e-01 8.52583408e-01 -4.63072538e-01
1.02463865e+00 -3.39118928e-01 -1.48070887e-01 -8.72802913e-01
-1.56279504e-01 -5.50603807e-01 -8.41388479e-02 9.08882916e-01
7.32337952e-01 -3.21594656e-01 4.44094032e-01 5.35740018e-01
-3.65226924e-01 -5.54272771e-01 -7.44761109e-01 -6.89219594e-01
-1.15938317e-02 -3.55447501e-01 6.94585621e-01 1.11112261e+00
5.88393331e-01 -2.24210933e-01 5.14464796e-01 -2.80917529e-02
-1.34438956e-02 8.70926753e-02 2.21889708e-02 -1.22970021e+00
-3.39904904e-01 -9.54123259e-01 -4.94128764e-01 -8.73690546e-01
4.69932526e-01 -1.15388668e+00 2.70737559e-01 -1.36111867e+00
6.91934407e-01 -7.54335597e-02 -4.84383821e-01 7.36515880e-01
-2.91415304e-01 -1.34804800e-01 -2.07725421e-01 2.20658466e-01
-9.21829641e-01 -1.33980855e-01 5.52874267e-01 -4.39564049e-01
1.90517142e-01 -3.41917306e-01 -9.36946750e-01 4.02160048e-01
4.28780377e-01 -9.69357550e-01 -4.45672035e-01 -9.53928888e-01
1.29384542e+00 -5.04206754e-02 2.96944469e-01 -1.05841184e+00
5.58146358e-01 3.87035251e-01 -4.63387221e-01 -1.94839284e-01
-5.35108924e-01 -7.97445416e-01 2.07080781e-01 5.71996689e-01
-9.46813345e-01 4.51626480e-01 5.40695608e-01 3.18460435e-01
-3.48276228e-01 -1.41605592e+00 1.90705165e-01 -4.74276423e-01
-8.88691008e-01 -2.49399722e-01 -6.02630973e-01 3.12974423e-01
4.67544645e-01 1.12920910e-01 -7.19228446e-01 -3.89115848e-02
-1.27087459e-01 -1.83853790e-01 6.50984883e-01 1.13449430e+00
6.77322209e-01 -1.05856478e+00 -2.08665401e-01 -1.20837644e-01
9.29932833e-01 -6.88414454e-01 -3.20296735e-01 2.71861941e-01
-1.01968849e+00 7.77845562e-01 1.26925588e-01 -2.84178525e-01
-1.41816282e+00 4.79699939e-01 4.05045748e-01 1.17034525e-01
-6.65155590e-01 1.05640662e+00 -1.00892246e-01 -3.94351661e-01
2.04230577e-01 -5.22169828e-01 -5.19551992e-01 -2.22440824e-01
8.98636580e-01 2.19224378e-01 5.55572271e-01 -2.18626127e-01
-7.85869956e-01 7.67052829e-01 -6.24770038e-02 2.53294349e-01
1.24733651e+00 2.86495239e-01 -9.68081713e-01 5.60038425e-02
1.78823376e+00 2.00292259e-01 -4.88703609e-01 -9.10939872e-02
9.34754252e-01 -8.73769701e-01 1.56742498e-01 -8.22263241e-01
-1.07436824e+00 7.18622446e-01 5.47058642e-01 6.33652329e-01
7.94087768e-01 5.38073182e-01 5.48550189e-01 1.02904093e+00
6.37297630e-01 -6.59871995e-01 2.50650465e-01 7.92711735e-01
7.87909031e-01 -8.84335399e-01 -3.13711971e-01 -1.83191538e-01
1.83298260e-01 1.37487626e+00 5.80158532e-01 9.39602405e-02
7.62553513e-01 2.72053629e-01 2.85439610e-01 -1.04260147e+00
-9.32591796e-01 4.93749566e-02 3.22855026e-01 3.68778020e-01
1.05079412e+00 -5.36928058e-01 -3.05221349e-01 3.05213988e-01
3.64249229e-01 -2.44030863e-01 3.79280180e-01 1.23529518e+00
-7.50397503e-01 -1.03469813e+00 6.54143840e-02 5.30956924e-01
-8.64951313e-01 -8.81640851e-01 -5.64553857e-01 4.20152724e-01
-1.32614866e-01 8.85251701e-01 -8.28178674e-02 -4.16991830e-01
1.95117280e-01 1.55095205e-01 -2.47516096e-01 -1.07740390e+00
-1.17535341e+00 -5.81650794e-01 -8.34548473e-03 -1.05935538e+00
-1.96166903e-01 -5.68908751e-01 -1.24463046e+00 1.99321121e-01
-6.49996102e-01 4.33282048e-01 9.20557439e-01 9.03066933e-01
8.34018826e-01 7.79872179e-01 3.16361248e-01 -1.37292966e-01
-1.50723085e-01 -6.98916674e-01 3.08900684e-01 1.60422534e-01
4.82415110e-01 -4.71467406e-01 -4.32514906e-01 2.07543969e-01] | [7.527677059173584, 8.083995819091797] |
e4d5e4d1-9d09-485b-a19d-2ebd02fd9425 | learning-spatiotemporal-frequency-transformer | 2208.03012 | null | https://arxiv.org/abs/2208.03012v1 | https://arxiv.org/pdf/2208.03012v1.pdf | Learning Spatiotemporal Frequency-Transformer for Compressed Video Super-Resolution | Compressed video super-resolution (VSR) aims to restore high-resolution frames from compressed low-resolution counterparts. Most recent VSR approaches often enhance an input frame by borrowing relevant textures from neighboring video frames. Although some progress has been made, there are grand challenges to effectively extract and transfer high-quality textures from compressed videos where most frames are usually highly degraded. In this paper, we propose a novel Frequency-Transformer for compressed video super-resolution (FTVSR) that conducts self-attention over a joint space-time-frequency domain. First, we divide a video frame into patches, and transform each patch into DCT spectral maps in which each channel represents a frequency band. Such a design enables a fine-grained level self-attention on each frequency band, so that real visual texture can be distinguished from artifacts, and further utilized for video frame restoration. Second, we study different self-attention schemes, and discover that a divided attention which conducts a joint space-frequency attention before applying temporal attention on each frequency band, leads to the best video enhancement quality. Experimental results on two widely-used video super-resolution benchmarks show that FTVSR outperforms state-of-the-art approaches on both uncompressed and compressed videos with clear visual margins. Code is available at https://github.com/researchmm/FTVSR. | ['Dongmei Fu', 'Jianlong Fu', 'Huan Yang', 'Zhongwei Qiu'] | 2022-08-05 | null | null | null | null | ['video-super-resolution', 'video-enhancement'] | ['computer-vision', 'computer-vision'] | [ 6.58754051e-01 -4.31876361e-01 -3.99347186e-01 -8.64779055e-02
-1.13978851e+00 -1.34664282e-01 2.95127988e-01 -4.35519427e-01
1.17108300e-01 7.60241389e-01 7.01855123e-01 1.32504106e-01
-1.11388145e-02 -7.87377954e-01 -9.74756718e-01 -7.46905863e-01
1.89469494e-02 -5.97464383e-01 4.27958459e-01 -3.83989513e-01
2.35224113e-01 9.23587084e-02 -1.60542297e+00 7.85630345e-01
9.01949704e-01 9.57483530e-01 7.08355248e-01 7.40437329e-01
1.69449925e-01 9.69668388e-01 -2.95126259e-01 3.35359573e-02
2.78673172e-01 -6.64508939e-01 -8.41037035e-01 2.60837257e-01
5.90844810e-01 -6.73566639e-01 -6.69928253e-01 1.44039369e+00
4.02759731e-01 1.88030571e-01 1.85458332e-01 -5.85421681e-01
-1.29372621e+00 6.44662321e-01 -1.01584172e+00 9.48903263e-01
6.55684352e-01 3.33517580e-03 7.73437262e-01 -1.06117773e+00
6.34980917e-01 1.38694131e+00 4.18222010e-01 4.48698461e-01
-1.29024994e+00 -7.00693190e-01 1.95365354e-01 5.05563557e-01
-1.44104576e+00 -5.97883940e-01 8.57246459e-01 -2.24325538e-01
6.91540718e-01 2.17248440e-01 4.33776855e-01 1.06613660e+00
3.03761870e-01 6.07835948e-01 1.13817608e+00 -9.81758386e-02
6.31003408e-03 -4.48294640e-01 -1.86839953e-01 2.76226014e-01
9.90875158e-03 1.22752734e-01 -7.30279326e-01 1.59172356e-01
1.40790820e+00 1.99648291e-01 -1.05770350e+00 9.44966376e-02
-1.35015655e+00 4.45501983e-01 2.95652002e-01 5.81398964e-01
-6.53528810e-01 -1.14277033e-02 3.11247319e-01 2.85729468e-01
6.01745427e-01 -1.01109646e-01 -2.24634096e-01 4.75111566e-02
-9.97387230e-01 8.02526027e-02 -9.43562984e-02 8.36922944e-01
6.65518582e-01 3.19773942e-01 -4.34790343e-01 1.00246596e+00
-4.05286700e-02 4.48551685e-01 7.15171933e-01 -1.14173436e+00
4.54349875e-01 -1.06065646e-01 2.53691614e-01 -1.12669051e+00
2.51121730e-01 -3.29342782e-01 -1.30104232e+00 4.89493571e-02
-4.08277586e-02 2.54366636e-01 -6.80611372e-01 1.46965981e+00
1.24698021e-01 7.07189918e-01 6.53079599e-02 1.45805931e+00
8.87860298e-01 9.61565852e-01 -8.71726125e-02 -6.46936238e-01
1.51740885e+00 -8.82522941e-01 -1.07334220e+00 -1.46031994e-02
-3.41584355e-01 -7.60915101e-01 1.06707954e+00 4.43278164e-01
-1.51406360e+00 -1.14725077e+00 -1.16796541e+00 -3.71178985e-01
2.41969079e-01 -1.58241466e-01 1.35414481e-01 9.83271822e-02
-1.14191866e+00 7.58622885e-01 -7.44351804e-01 8.39114189e-02
5.74306726e-01 4.09638658e-02 -4.18410987e-01 -3.68871689e-01
-1.43717468e+00 2.49580994e-01 2.71308720e-01 8.42549093e-03
-9.07669604e-01 -7.74658561e-01 -8.33734512e-01 8.30749124e-02
4.55841929e-01 -5.07551908e-01 9.29092407e-01 -1.22242308e+00
-1.31777287e+00 5.89862347e-01 -5.13289034e-01 -4.27797645e-01
2.87988454e-01 -2.33453080e-01 -9.09785509e-01 5.20269156e-01
1.65653482e-01 2.55697787e-01 1.51119590e+00 -1.31291497e+00
-9.31930900e-01 -2.61122793e-01 6.27799779e-02 1.96342304e-01
-3.03264081e-01 2.25730374e-01 -6.22924268e-01 -1.17440665e+00
1.85385838e-01 -2.93649971e-01 7.34333321e-02 -2.56578803e-01
-6.22677580e-02 1.55023396e-01 9.64534938e-01 -1.12961578e+00
1.47181249e+00 -2.35907507e+00 5.04623711e-01 -3.86969596e-01
4.38996851e-01 2.75259495e-01 -2.37892702e-01 -7.22117350e-02
-4.09166902e-01 1.01452470e-01 -1.52596429e-01 3.29409540e-02
-5.62940121e-01 -1.18818723e-01 -5.75283706e-01 4.11887974e-01
2.89557487e-01 6.40109837e-01 -9.88837600e-01 -4.16677028e-01
2.53781021e-01 1.15388584e+00 -6.50004089e-01 2.74485618e-01
1.24512285e-01 6.72810256e-01 -5.51222324e-01 6.96826398e-01
9.30689871e-01 -4.85492617e-01 1.32978514e-01 -7.17709959e-01
-2.54148483e-01 -3.74912843e-02 -1.09846306e+00 1.72518897e+00
-4.05096978e-01 6.67019427e-01 1.30348891e-01 -8.17263067e-01
8.58316362e-01 3.27947259e-01 5.44646561e-01 -1.03984642e+00
-7.05507491e-03 1.20251387e-01 -4.51820850e-01 -5.89926839e-01
8.11412632e-01 -5.37078045e-02 3.95945817e-01 -5.82331792e-02
-6.96223825e-02 4.29980278e-01 1.49588883e-01 6.02117106e-02
9.10974741e-01 2.66904593e-01 3.27255458e-01 -1.71391830e-01
8.56857181e-01 -4.79590446e-01 8.18280935e-01 4.30136800e-01
-2.88405776e-01 1.13637578e+00 5.61429970e-02 -3.17893833e-01
-1.20218766e+00 -1.16729116e+00 -1.29099265e-01 1.26825857e+00
4.90682870e-01 -3.90031755e-01 -7.72778511e-01 -1.96930587e-01
-3.12519014e-01 2.47747794e-01 -6.63572252e-01 -4.42078188e-02
-8.05310786e-01 -5.57305515e-01 -7.84344152e-02 4.86146003e-01
8.20271254e-01 -9.38503563e-01 -4.89271641e-01 2.70771295e-01
-7.65063286e-01 -1.23999059e+00 -8.64779949e-01 -4.20127183e-01
-7.59592295e-01 -7.90829182e-01 -1.27358639e+00 -9.46606219e-01
2.59708494e-01 9.53674734e-01 1.06246614e+00 1.27689019e-01
-1.27407894e-01 1.53066322e-01 -6.70419633e-01 4.72884148e-01
-1.93338081e-01 -2.96968251e-01 -4.21717614e-02 4.99618322e-01
8.26737806e-02 -7.28536725e-01 -9.53057349e-01 3.43937099e-01
-1.07189989e+00 2.02663451e-01 5.49597681e-01 9.66931462e-01
1.16288435e+00 3.68158311e-01 6.32517755e-01 -6.64202273e-01
4.24919844e-01 -6.11211419e-01 -4.05544102e-01 4.08353060e-02
-2.03147978e-01 -2.57346421e-01 8.79515707e-01 -4.44050282e-01
-1.29254103e+00 -4.15084422e-01 5.38808946e-03 -8.86513770e-01
-8.71676132e-02 1.23862259e-01 -2.30840608e-01 1.33161275e-02
4.96743619e-01 7.24988341e-01 -1.80647492e-01 -6.37422860e-01
1.57564312e-01 6.95411980e-01 9.59943593e-01 -4.87244427e-01
8.15589130e-01 6.06695354e-01 -3.95066798e-01 -8.43975604e-01
-7.90487945e-01 -2.55662590e-01 -2.80593395e-01 -1.89021125e-01
1.05427408e+00 -1.28377259e+00 -3.70570153e-01 3.20982128e-01
-7.33635008e-01 -3.80357087e-01 -1.89342558e-01 3.85865808e-01
-6.33026004e-01 6.21806741e-01 -8.58246684e-01 -3.81016076e-01
-2.51689762e-01 -1.31924176e+00 1.17870843e+00 3.72549266e-01
2.30623752e-01 -6.79346979e-01 -2.40061611e-01 4.63787317e-01
6.98586226e-01 1.07124195e-01 4.74760145e-01 3.06738704e-01
-8.66611481e-01 4.19114798e-01 -5.91934443e-01 4.35910344e-01
4.17341143e-01 -1.94643378e-01 -7.49117851e-01 -6.88949585e-01
1.43213645e-01 -8.70690346e-02 1.03533363e+00 5.75283885e-01
1.65164185e+00 -3.75982046e-01 -4.30467390e-02 9.75602269e-01
1.66593194e+00 9.45914015e-02 1.06583822e+00 4.01043504e-01
7.83693731e-01 9.92102176e-02 7.98523068e-01 4.31448042e-01
2.46751547e-01 7.96867549e-01 2.44529784e-01 -1.72339618e-01
-5.81905723e-01 -1.82021737e-01 4.94639993e-01 6.88569307e-01
-6.72458649e-01 -6.27167746e-02 -3.86887431e-01 5.31840563e-01
-1.63399291e+00 -1.27363384e+00 9.27792713e-02 2.14472055e+00
9.75431621e-01 -1.06408201e-01 -4.52889241e-02 2.90414542e-01
1.08233023e+00 5.91117680e-01 -5.49463809e-01 1.36498287e-01
-4.89909381e-01 2.61156738e-01 3.65616828e-01 5.70292711e-01
-1.17124104e+00 7.31419921e-01 5.26519346e+00 1.27784085e+00
-1.16202378e+00 2.89898098e-01 9.30701375e-01 -1.52402028e-01
-2.43527830e-01 -3.00315738e-01 -4.54500169e-01 7.54142404e-01
6.88905180e-01 -3.24613988e-01 8.46239507e-01 4.70039815e-01
5.42751074e-01 1.71972334e-01 -6.50557697e-01 1.23656213e+00
-3.03427055e-02 -1.43135405e+00 2.31889516e-01 -1.06136888e-01
9.73513186e-01 -2.83036470e-01 3.71258736e-01 4.62113731e-02
-5.74224629e-02 -1.04004717e+00 7.22483218e-01 5.54280341e-01
1.35429871e+00 -8.28157723e-01 5.41193008e-01 -2.39173636e-01
-1.74873471e+00 -2.19954833e-01 -5.42119741e-01 2.94348210e-01
1.40519351e-01 5.55578530e-01 1.96081147e-01 8.90267849e-01
1.29737186e+00 1.28645515e+00 -2.49853998e-01 5.94415247e-01
1.08702146e-01 5.54360211e-01 3.67817074e-01 8.59483063e-01
-1.82591736e-01 -1.75183788e-01 7.40002692e-01 1.22000289e+00
4.44243371e-01 6.76789582e-01 1.18891269e-01 7.80022025e-01
-2.88701989e-02 -8.41893256e-03 -1.06464915e-01 1.97686657e-01
3.49837780e-01 1.03834963e+00 -4.93312031e-01 -5.06088495e-01
-6.29754186e-01 1.28313470e+00 -8.73338804e-03 6.48036957e-01
-9.79526520e-01 -2.33211517e-01 7.83122301e-01 3.23209643e-01
7.20953107e-01 6.21806011e-02 1.05200067e-01 -1.52052879e+00
1.51543975e-01 -1.26855075e+00 3.39944482e-01 -1.07961977e+00
-1.11922848e+00 8.57946813e-01 -3.45097303e-01 -1.66552055e+00
3.45029160e-02 -5.19169196e-02 -8.70059356e-02 9.30404782e-01
-1.88275635e+00 -8.06283057e-01 -6.45652652e-01 9.16630745e-01
1.06574678e+00 -4.10238393e-02 2.82818347e-01 5.84083259e-01
-4.70205784e-01 4.63196665e-01 1.08652279e-01 5.53549640e-02
7.25302041e-01 -8.52305591e-01 1.54280007e-01 1.09611714e+00
-3.25327694e-01 4.91850585e-01 7.22349584e-01 -7.45702863e-01
-1.59324455e+00 -1.37414527e+00 3.01269710e-01 1.34784311e-01
4.43837553e-01 1.09756880e-01 -1.47326744e+00 5.73673368e-01
2.44605556e-01 5.18835366e-01 2.01767206e-01 -5.25197923e-01
-4.36103165e-01 -2.19959006e-01 -1.13981903e+00 4.57147747e-01
1.19074190e+00 -7.42255449e-01 -4.90314186e-01 5.18171676e-02
9.74845052e-01 -4.79894400e-01 -1.21855283e+00 4.74438906e-01
3.22582513e-01 -1.35305774e+00 1.35488343e+00 -1.79338500e-01
1.05536020e+00 -6.96297824e-01 -5.02407730e-01 -1.13726950e+00
-8.43654692e-01 -7.26283252e-01 -2.74262339e-01 1.20933390e+00
-2.49593943e-01 -4.66490835e-01 2.95953482e-01 -9.18143839e-02
-3.90336439e-02 -6.69809997e-01 -5.99362671e-01 -6.03539765e-01
-2.33363613e-01 -1.62833054e-02 6.31427467e-01 1.13908267e+00
-3.62267643e-01 4.21302617e-02 -7.49907970e-01 4.21289831e-01
9.62913871e-01 2.42420286e-01 2.94983238e-01 -6.74648464e-01
-4.67415899e-01 -4.19023365e-01 -3.15102607e-01 -1.26519644e+00
-1.15210362e-01 -5.10640204e-01 -1.11913495e-01 -1.37418365e+00
5.22813261e-01 1.41126543e-01 -5.53894460e-01 1.43006057e-01
-4.54443336e-01 7.49061704e-01 8.61750841e-02 3.61818969e-01
-6.46768928e-01 5.56136489e-01 1.54570961e+00 -5.30391000e-02
-6.64360300e-02 -4.56392944e-01 -8.40086818e-01 6.57513738e-01
6.23327017e-01 3.25093716e-02 -2.97771335e-01 -4.85851467e-01
-1.94263890e-01 6.10035956e-01 4.87074226e-01 -9.18648362e-01
-1.10988744e-01 -2.53510237e-01 6.96966946e-01 -4.32783782e-01
1.90708548e-01 -5.78093827e-01 3.31427425e-01 1.44744799e-01
-3.55739027e-01 -1.56992197e-01 8.09720159e-02 7.98456669e-01
-5.94619870e-01 4.22903150e-01 1.20271730e+00 -7.76174664e-02
-1.03943217e+00 5.05848527e-01 -1.62976265e-01 8.85482654e-02
6.41716123e-01 -2.95921266e-01 -3.39025587e-01 -2.93957829e-01
-7.13776648e-01 -1.40497848e-01 6.58193052e-01 5.75775623e-01
1.01073730e+00 -1.44553697e+00 -1.13782489e+00 2.45897904e-01
-2.68696278e-01 -1.29963174e-01 1.07832825e+00 6.60988092e-01
-3.03513259e-01 6.93352073e-02 -5.46770513e-01 -6.04169309e-01
-1.33800638e+00 8.88136089e-01 2.79982269e-01 -1.35377022e-02
-1.14888346e+00 6.34400666e-01 6.11326516e-01 6.55766308e-01
-9.69226584e-02 -2.72015303e-01 -4.28654909e-01 -2.27576807e-01
1.34000981e+00 4.38963026e-01 -2.10854545e-01 -1.03259492e+00
-4.39790413e-02 1.02324474e+00 -1.39196053e-01 2.50699639e-01
1.34817636e+00 -7.18479633e-01 1.95736252e-02 1.37794852e-01
1.23313892e+00 9.80540887e-02 -1.74110234e+00 -5.60760677e-01
-6.23678386e-01 -1.16291380e+00 3.47951561e-01 -3.76695424e-01
-1.61053681e+00 5.79213619e-01 7.33980000e-01 3.15545678e-01
1.90229034e+00 -1.55659825e-01 1.08326399e+00 -4.12875623e-01
4.61183220e-01 -7.96285927e-01 1.69490129e-01 2.29875579e-01
1.20405865e+00 -1.19781351e+00 8.28036219e-02 -5.78769803e-01
-5.00974357e-01 9.82092559e-01 4.47868884e-01 -3.67729306e-01
3.45728040e-01 2.52392083e-01 -2.22302660e-01 2.19314307e-01
-6.66700959e-01 -1.58272475e-01 3.27993542e-01 5.88211060e-01
5.41228533e-01 -8.09306800e-02 -1.27954692e-01 7.82013297e-01
4.08187658e-02 3.32521588e-01 6.65722609e-01 5.71686864e-01
-6.22872829e-01 -6.01412594e-01 -6.61717951e-01 1.59959063e-01
-8.21109116e-01 -2.01666042e-01 3.36732507e-01 3.33776474e-01
1.13230355e-01 9.48249459e-01 1.08611606e-01 -4.36638743e-01
2.04067662e-01 -4.85809773e-01 4.85415459e-01 -1.54151037e-01
-1.68690145e-01 5.26591837e-01 -3.32480162e-01 -9.67835784e-01
-5.74118555e-01 -4.86853957e-01 -1.12312126e+00 -3.72085392e-01
1.34499177e-01 -8.98500606e-02 1.04329223e-02 4.50956315e-01
4.08240348e-01 1.04964662e+00 8.07364464e-01 -1.22247362e+00
-2.04820484e-01 -7.08421648e-01 -7.45085001e-01 6.10973120e-01
8.38990688e-01 -6.04657769e-01 -4.03962851e-01 5.30754864e-01] | [11.097932815551758, -1.9728692770004272] |
68fbaca1-68ac-421b-a8a3-965191ee440d | comparison-of-the-performance-of-machine | null | null | https://www.jphres.org/index.php/jphres/article/view/1677 | https://www.jphres.org/index.php/jphres/article/view/1677 | Comparison of the performance of machine learning algorithms in breast cancer screening and detection: A protocol | BACKGROUND:
Breast Cancer (BC) is a known global crisis. The World Health Organization reports a global 2.09 million incidences and 627,000 deaths in 2018 relating to BC. The traditional BC screening method in developed countries is mammography, whilst developing countries employ breast self-examination and clinical breast examination. The prominent gold standard for BC detection is triple assessment: i) clinical examination, ii) mammography and/ or ultrasonography; and iii) Fine Needle Aspirate Cytology. However, the introduction of cheaper, efficient and non-invasive methods of BC screening and detection would be beneficial.
DESIGN AND METHODS:
We propose the use of eight machine learning algorithms: i) Logistic Regression; ii) Support Vector Machine; iii) K-Nearest Neighbors; iv) Decision Tree; v) Random Forest; vi) Adaptive Boosting; vii) Gradient Boosting; viii) eXtreme Gradient Boosting, and blood test results using BC Coimbra Dataset (BCCD) from University of California Irvine online database to create models for BC prediction. To ensure the models’ robustness, we will employ: i) Stratified k-fold Cross-Validation; ii) Correlation-based Feature Selection (CFS); and iii) parameter tuning. The models will be validated on validation and test sets of BCCD for full features and reduced features. Feature reduction has an impact on algorithm performance. Seven metrics will be used for model evaluation, including accuracy.
EXPECTED IMPACT OF THE STUDY FOR PUBLIC HEALTH:
The CFS together with highest performing model(s) can serve to identify important specific blood tests that point towards BC, which may serve as an important BC biomarker. Highest performing model(s) may eventually be used to create an artificial intelligence tool to assist clinicians in BC screening and detection. | ['Yashik Singh', 'Zakia Salod'] | 2019-12-04 | null | null | null | journal-of-public-health-research-2019-12 | ['breast-cancer-detection', 'breast-cancer-detection'] | ['knowledge-base', 'medical'] | [ 2.94108123e-01 -1.46663964e-01 -5.87997496e-01 -2.51906812e-01
-6.58340096e-01 -1.97802514e-01 5.92042923e-01 7.00529456e-01
-3.58804673e-01 9.59283650e-01 1.03545338e-01 -1.10697651e+00
-3.08054179e-01 -1.04561496e+00 -1.80195093e-01 -8.55871201e-01
-3.13172162e-01 5.51509738e-01 3.26993585e-01 -1.17761679e-01
1.88108176e-01 8.94722342e-01 -1.37106907e+00 4.89324927e-01
9.03516114e-01 1.26962149e+00 4.48708571e-02 9.86310124e-01
-2.14520637e-02 7.97836602e-01 -1.74568459e-01 -2.47896835e-01
1.65871322e-01 -8.03231001e-01 -5.61997533e-01 -2.92926848e-01
-4.40247416e-01 -3.03748190e-01 3.87921810e-01 5.03394246e-01
5.33633649e-01 -5.47919273e-01 8.59306931e-01 -1.11642516e+00
-1.21130571e-01 1.15237482e-01 -6.53289557e-01 2.84728914e-01
3.53144556e-01 1.14674129e-01 1.14272095e-01 -6.26993716e-01
4.68363315e-01 1.06878257e+00 9.38821793e-01 4.93260235e-01
-8.80667388e-01 -7.01754153e-01 -5.75471044e-01 4.71064240e-01
-9.94744301e-01 -3.80000412e-01 2.35827103e-01 -3.25631827e-01
5.44167340e-01 8.84521067e-01 1.27403426e+00 6.00355268e-01
5.03104448e-01 3.79402846e-01 1.68008149e+00 -8.39879215e-01
3.96747470e-01 4.99348968e-01 1.24347553e-01 8.11926365e-01
5.05929649e-01 3.63154709e-01 -3.50982338e-01 -7.31678307e-01
2.61525273e-01 1.54879346e-01 5.93109988e-02 -8.67274553e-02
-5.64813733e-01 1.06227922e+00 2.57419020e-01 2.93265939e-01
-4.60363179e-01 -3.10792655e-01 5.69660604e-01 2.47773349e-01
2.56295383e-01 8.41464177e-02 -4.78964925e-01 1.87902808e-01
-7.26208568e-01 6.56830519e-02 7.60715485e-01 2.44973466e-01
4.81777102e-01 -4.78878051e-01 -1.56493381e-01 9.68917131e-01
3.89771730e-01 5.62547982e-01 7.83732474e-01 -4.13576484e-01
-1.06781922e-01 8.20396662e-01 8.05577543e-03 -1.06875575e+00
-5.35960317e-01 -3.27215075e-01 -9.35091734e-01 -6.10329509e-02
2.19143167e-01 -1.42574131e-01 -1.02796781e+00 1.03834081e+00
6.06005490e-01 -1.01275019e-01 1.45789022e-02 7.67192364e-01
7.66828299e-01 2.62255073e-01 2.74335653e-01 -4.98326182e-01
1.34500015e+00 -4.24575627e-01 -4.30414647e-01 1.60826802e-01
8.85029614e-01 -4.65326279e-01 2.50479072e-01 6.44945323e-01
-8.11907649e-01 -2.57598430e-01 -9.94814456e-01 3.54370475e-01
-6.17099762e-01 2.33825549e-01 9.63712156e-01 1.26060903e+00
-8.22121739e-01 2.70219177e-01 -9.81673300e-01 -7.50504255e-01
4.40669537e-01 4.62252200e-01 -4.40307975e-01 -4.16372806e-01
-1.02709317e+00 1.08770931e+00 -1.81973483e-02 1.63450718e-01
-7.18309820e-01 -5.84331393e-01 -7.17869222e-01 -4.03983146e-01
-1.52646020e-01 -7.23530948e-01 5.82753241e-01 -1.05463970e+00
-1.14289522e+00 1.00043881e+00 -5.49097419e-01 -4.78084385e-01
6.40959740e-01 -1.36277014e-02 -3.46161395e-01 2.23585948e-01
2.40881160e-01 2.12538108e-01 1.81019947e-01 -1.06102085e+00
-1.18113971e+00 -9.28095877e-01 -8.60880911e-01 -6.36158586e-02
-4.11625624e-01 2.57027835e-01 -2.44878829e-01 -1.13158211e-01
3.25496376e-01 -8.05608511e-01 -3.22127283e-01 -1.60301313e-01
-3.46282363e-01 -1.82986140e-01 7.51857996e-01 -1.30457246e+00
1.19026589e+00 -1.57826340e+00 -3.50068659e-01 5.02075016e-01
-1.54593006e-01 3.54820788e-01 3.44699174e-01 2.54142106e-01
1.62222415e-01 2.42461622e-01 6.77077696e-02 4.22252774e-01
-7.08282232e-01 -1.57216609e-01 4.20968950e-01 8.01989436e-01
4.56904471e-01 6.44216359e-01 -7.25716770e-01 -7.22494006e-01
2.59218693e-01 4.05217171e-01 -1.97164252e-01 -3.35550308e-03
4.21215296e-01 4.23939198e-01 -4.36316818e-01 1.11211312e+00
6.09096169e-01 3.55727002e-02 4.08277601e-01 -1.13832392e-01
5.57381287e-02 -4.41976152e-02 -9.32701349e-01 6.42375290e-01
-2.90936362e-02 1.87911570e-01 4.42496091e-02 -1.04633474e+00
1.28115678e+00 1.03024758e-01 5.57693720e-01 -7.54112542e-01
-5.03522046e-02 3.84111106e-01 2.19804570e-01 -5.56164265e-01
-3.72527868e-01 -2.03762993e-01 3.53443623e-01 -6.86757714e-02
-4.32352513e-01 7.68943802e-02 1.79047137e-01 -4.29896265e-02
1.33882809e+00 -9.95085835e-02 8.90130460e-01 -3.67384076e-01
7.56007254e-01 4.38559026e-01 6.45587087e-01 6.05453491e-01
-2.96938390e-01 -6.03959933e-02 5.24838686e-01 -5.70202887e-01
-5.03119946e-01 -7.71168530e-01 -6.39654934e-01 7.88205385e-01
-2.54337490e-01 2.43218705e-01 -2.23171219e-01 -4.29766059e-01
4.55994427e-01 4.12074655e-01 -8.31753552e-01 -1.75903752e-01
-2.43754268e-01 -1.42720187e+00 5.25853276e-01 2.74997622e-01
5.09993792e-01 -5.95956802e-01 -7.23006964e-01 2.12998003e-01
2.18205646e-01 -1.63816437e-01 5.31647325e-01 5.04454076e-01
-1.17836177e+00 -1.47369790e+00 -7.38214493e-01 -5.04836977e-01
5.60671568e-01 -3.43401991e-02 7.06254244e-01 3.35223675e-01
-7.86327541e-01 4.16838750e-02 -4.75990236e-01 -8.56311262e-01
-6.19966924e-01 -3.33457023e-01 -9.72439796e-02 -1.10714905e-01
7.55939126e-01 -7.68599883e-02 -8.20721686e-01 2.57565826e-01
-5.87288678e-01 2.46345047e-02 1.08525038e+00 9.65417385e-01
6.27434790e-01 -1.63958728e-01 6.40337527e-01 -1.16895998e+00
3.93567890e-01 -8.46706092e-01 -3.49507511e-01 2.30942070e-01
-8.30486238e-01 -7.33005822e-01 2.64707953e-01 -2.58007050e-01
-9.70450282e-01 1.41549915e-01 1.19796032e-02 3.53851348e-01
-1.78805441e-01 6.63456321e-01 5.60374893e-02 -8.83347094e-02
9.12800252e-01 2.45458841e-01 2.87798345e-01 -2.99506664e-01
-5.02984226e-01 9.82107937e-01 1.55308455e-01 -1.73739955e-01
3.53463411e-01 5.88421702e-01 5.51478446e-01 -9.89435613e-01
-3.54061723e-01 -7.18105435e-01 -5.40223539e-01 -3.41812819e-01
6.74109280e-01 -8.10358703e-01 -5.47845185e-01 4.19552803e-01
-4.60040480e-01 -2.73208499e-01 5.17936885e-01 4.72888142e-01
-2.34363079e-02 9.82636437e-02 -6.17543995e-01 -1.13564003e+00
-5.53962588e-01 -7.46933103e-01 8.12176287e-01 5.25455356e-01
-2.72275627e-01 -8.28966618e-01 1.11085981e-01 5.55101931e-01
4.35724318e-01 8.63093793e-01 1.09405088e+00 -6.40786290e-01
1.52639553e-01 -5.30386567e-01 -3.84677291e-01 1.84049189e-01
4.67489660e-01 1.90014869e-01 -7.99590886e-01 -1.33934557e-01
-2.49716774e-01 -3.77441525e-01 7.58795500e-01 6.17032886e-01
1.14169133e+00 -1.18197659e-02 -9.31395411e-01 5.41008830e-01
1.58635306e+00 6.70670807e-01 6.46323800e-01 5.70105553e-01
7.72075504e-02 7.36126423e-01 7.41175473e-01 3.97601038e-01
1.73660859e-01 3.08365852e-01 3.20188046e-01 -2.59811193e-01
8.49635899e-03 -2.45860014e-02 -2.84970105e-02 1.91170022e-01
-4.48403001e-01 3.49221349e-01 -1.46315563e+00 6.27538979e-01
-1.27984130e+00 -6.54009283e-01 -6.19687140e-01 2.16695189e+00
7.60649860e-01 1.01942189e-01 3.10949922e-01 5.96460700e-01
4.61423516e-01 -9.68688428e-01 -4.06548053e-01 -7.52753556e-01
-7.54956156e-02 3.84893388e-01 4.70043570e-01 -9.31091234e-02
-9.52640951e-01 2.87258714e-01 5.88872623e+00 4.49270904e-01
-1.45564044e+00 2.16596830e-03 1.48473847e+00 1.59024268e-01
2.37781703e-01 4.21957448e-02 -8.01968217e-01 4.28977937e-01
1.05351150e+00 5.81815764e-02 -1.05747059e-01 7.56953776e-01
2.75491863e-01 -8.23309124e-01 -7.44059861e-01 5.01495898e-01
-2.25448236e-01 -1.37011194e+00 -2.07033902e-01 1.85874179e-01
5.28367281e-01 -1.41221195e-01 -5.30621469e-01 2.77606755e-01
2.75206596e-01 -1.13189685e+00 -2.87821796e-02 6.36809111e-01
1.05772436e+00 -6.75847173e-01 1.16285062e+00 4.11093086e-01
-7.50928283e-01 -2.44422778e-01 -1.30079985e-01 -1.91415306e-02
-4.02294189e-01 8.67345572e-01 -1.14802682e+00 6.18391156e-01
1.15999770e+00 1.84163049e-01 -1.02825940e+00 1.19023514e+00
2.87834197e-01 9.95952189e-01 -2.28860751e-01 -5.78479230e-01
-8.67883265e-02 -2.37700287e-02 -5.83554395e-02 1.41575634e+00
4.12538022e-01 1.16381019e-01 -1.11575723e-01 2.45637596e-01
7.50528812e-01 4.58780259e-01 -3.38362068e-01 2.13310614e-01
3.64139169e-01 1.14162898e+00 -9.80669916e-01 -3.93075645e-02
-3.93585891e-01 5.94506443e-01 5.88320233e-02 -4.32427302e-02
-3.34294498e-01 -3.00255150e-01 7.02536628e-02 4.84560013e-01
-3.32866795e-02 7.55316734e-01 -4.91837204e-01 -2.58334517e-01
-4.95948911e-01 -8.42433631e-01 8.09210956e-01 -3.77365142e-01
-1.24080563e+00 8.67005885e-02 2.11281348e-02 -6.66268170e-01
-2.19719738e-01 -7.08235443e-01 -4.94023860e-01 1.10626864e+00
-1.39115214e+00 -1.12433338e+00 -5.59334517e-01 2.47213602e-01
-9.85318348e-02 -1.88502774e-01 1.09029555e+00 8.33174661e-02
-4.32030648e-01 4.39331234e-01 2.54125744e-01 1.63756222e-01
5.91566741e-01 -1.10686302e+00 -6.93541348e-01 2.69104779e-01
-9.34910953e-01 3.75678509e-01 4.22512561e-01 -9.94821072e-01
-1.46886945e+00 -1.16334248e+00 9.61340725e-01 -2.37742122e-02
3.10647368e-01 1.07174315e-01 -4.12284821e-01 3.05042356e-01
-5.05864859e-01 1.21797979e-01 1.24414492e+00 -3.15121748e-02
2.75439292e-01 -5.24801075e-01 -1.69341433e+00 2.32299343e-01
2.97580957e-01 2.17952669e-01 -2.99735442e-02 3.97987902e-01
-1.84144855e-01 -1.68672457e-01 -1.00314188e+00 6.94543362e-01
1.04427993e+00 -1.20256639e+00 6.81948960e-01 -7.03247130e-01
7.29999781e-01 8.11230019e-02 -5.79529889e-02 -7.68364131e-01
-5.11989832e-01 -2.35167459e-01 1.54707521e-01 1.02180564e+00
4.94192183e-01 -7.76046336e-01 8.80560219e-01 6.08425379e-01
3.02355796e-01 -1.41974890e+00 -9.59265232e-01 -3.13643873e-01
2.85592616e-01 -1.62914097e-01 4.24927533e-01 8.48047137e-01
9.25469920e-02 -2.23439842e-01 5.65479808e-02 -1.05816267e-01
3.68939430e-01 -2.34106138e-01 9.03699398e-01 -1.29870558e+00
-2.05164358e-01 -2.32181728e-01 -6.27957582e-01 1.97580621e-01
-5.69718301e-01 -5.93302965e-01 -5.06268561e-01 -1.50814509e+00
8.50915313e-01 -7.86590338e-01 -1.87940747e-01 4.45913464e-01
-3.21947664e-01 3.86596859e-01 -3.35263938e-01 -2.59862728e-02
1.10350735e-01 -2.30494916e-01 1.00484002e+00 8.44912156e-02
-3.09208542e-01 2.22466186e-01 -8.66605639e-01 4.69217330e-01
8.29680800e-01 -5.34109592e-01 6.60971403e-02 3.64049315e-01
2.14714725e-02 5.69659650e-01 4.44404274e-01 -8.14809620e-01
1.05555877e-01 -5.61079025e-01 1.30972123e+00 -5.35119355e-01
1.06109068e-01 -6.48877501e-01 4.83630657e-01 1.53804600e+00
1.91232860e-02 -4.57329392e-01 2.85993606e-01 4.82723475e-01
1.13359936e-01 -1.98825806e-01 9.02635872e-01 -1.42324626e-01
-2.96918482e-01 -1.52221099e-01 -5.99705458e-01 -7.57880688e-01
1.40593696e+00 -5.56488931e-01 -3.16375554e-01 -3.14078569e-01
-6.14638388e-01 2.19195142e-01 2.90337920e-01 -1.35921717e-01
4.34267014e-01 -1.14909017e+00 -1.06651628e+00 1.26171276e-01
1.01915561e-01 -3.66761327e-01 -1.21897392e-01 1.16040337e+00
-9.26074266e-01 5.65011621e-01 -4.09183949e-01 -6.58583462e-01
-1.62684822e+00 2.18729258e-01 4.04431194e-01 -6.53941453e-01
-1.71107888e-01 7.75457025e-01 -4.68135029e-01 -2.38161355e-01
6.54987991e-03 1.47872686e-01 -5.18108904e-01 2.31849588e-02
6.31994247e-01 8.24020505e-01 3.62118423e-01 -5.10992348e-01
-6.10852480e-01 8.41441453e-02 1.21361844e-03 9.01996121e-02
1.43663418e+00 2.25108698e-01 -4.73898828e-01 2.37554461e-01
9.31493402e-01 -2.26832688e-01 -3.36980790e-01 1.09854110e-01
1.77093074e-01 -3.98795277e-01 2.03111604e-01 -1.35962892e+00
-6.24119580e-01 3.60968590e-01 1.07569599e+00 1.95866793e-01
1.46600640e+00 -1.62953496e-01 2.64981568e-01 1.11072987e-01
3.28275234e-01 -9.22421634e-01 -5.09917855e-01 -3.65760922e-01
8.47977221e-01 -1.46814144e+00 4.36728179e-01 -3.99779350e-01
-3.33929271e-01 1.27310777e+00 2.77103275e-01 -1.02179917e-02
1.06138003e+00 2.35394597e-01 1.24345869e-01 3.05737667e-02
-8.50736856e-01 5.70288971e-02 2.10041225e-01 8.84586334e-01
7.45573878e-01 2.86606491e-01 -1.00242269e+00 1.00806820e+00
-1.94216341e-01 4.57066029e-01 1.33366182e-01 1.04343176e+00
-4.10315961e-01 -8.88103962e-01 -8.36525321e-01 1.47173250e+00
-7.08074093e-01 3.02414715e-01 -7.72273839e-01 9.77248609e-01
3.93768340e-01 1.21367466e+00 -1.33864656e-01 -4.04169634e-02
-8.38149413e-02 4.32149582e-02 3.81150514e-01 -2.72489280e-01
-7.25409806e-01 1.56433582e-02 3.21055889e-01 -3.87290716e-01
-4.42177445e-01 -9.91138399e-01 -1.02118003e+00 -2.38398567e-01
-5.36222994e-01 1.96355298e-01 1.07789421e+00 8.67464304e-01
1.53247584e-02 1.96922630e-01 5.52953243e-01 -4.45395261e-01
-3.96896809e-01 -1.25343728e+00 -6.59704506e-01 2.30808809e-01
1.48146331e-01 -4.36577678e-01 -3.65246236e-01 1.51043907e-01] | [15.285818099975586, -2.7199935913085938] |
35ff9556-95af-499d-a20b-3916a8b933cd | improving-automatic-skin-lesion-segmentation | 1807.08392 | null | http://arxiv.org/abs/1807.08392v2 | http://arxiv.org/pdf/1807.08392v2.pdf | Improving Automatic Skin Lesion Segmentation using Adversarial Learning based Data Augmentation | Segmentation of skin lesions is considered as an important step in computer
aided diagnosis (CAD) for automated melanoma diagnosis. In recent years,
segmentation methods based on fully convolutional networks (FCN) have achieved
great success in general images. This success is primarily due to the
leveraging of large labelled datasets to learn features that correspond to the
shallow appearance as well as the deep semantics of the images. However, the
dependence on large dataset does not translate well into medical images. To
improve the FCN performance for skin lesion segmentations, researchers
attempted to use specific cost functions or add post-processing algorithms to
refine the coarse boundaries of the FCN results. However, the performance of
these methods is heavily reliant on the tuning of many parameters and
post-processing techniques. In this paper, we leverage the state-of-the-art
image feature learning method of generative adversarial network (GAN) for its
inherent ability to produce consistent and realistic image features by using
deep neural networks and adversarial learning concept. We improve upon GAN such
that skin lesion features can be learned at different level of complexities, in
a controlled manner. The outputs from our method is then augmented to the
existing FCN training data, thus increasing the overall feature diversity. We
evaluated our method on the ISIC 2018 skin lesion segmentation challenge
dataset and showed that it was more accurate and robust when compared to the
existing skin lesion segmentation methods. | ['Jinman Kim', 'Lei Bi', 'Dagan Feng'] | 2018-07-23 | null | null | null | null | ['melanoma-diagnosis', 'skin-lesion-segmentation'] | ['computer-vision', 'medical'] | [ 7.30084002e-01 1.71461791e-01 9.87238213e-02 -2.96761185e-01
-8.53780508e-01 -5.52293837e-01 5.25577426e-01 -1.01396419e-01
-3.12923580e-01 5.66522360e-01 1.22630961e-01 -5.53734116e-02
-7.53955767e-02 -8.76662254e-01 -5.43915927e-01 -7.68660188e-01
2.67039150e-01 3.46165806e-01 2.46740475e-01 -3.61235917e-01
1.82381775e-02 6.47955656e-01 -1.21756685e+00 3.83951038e-01
1.01043355e+00 1.01290023e+00 -1.85720369e-01 9.24449146e-01
-1.94533914e-01 4.83236462e-01 -5.09236038e-01 -4.12428558e-01
3.51111025e-01 -6.43525720e-01 -8.84236395e-01 2.34501734e-01
2.47669876e-01 -3.58964019e-02 -1.14870846e-01 9.44656372e-01
5.23756802e-01 -2.93795466e-01 8.44290733e-01 -1.12031138e+00
-5.21773636e-01 2.02050939e-01 -5.24699748e-01 -1.37745932e-01
1.68075964e-01 2.98586369e-01 3.81193548e-01 -3.24295908e-01
7.01336384e-01 8.44342828e-01 1.08960307e+00 8.45575929e-01
-1.06557119e+00 -5.09142458e-01 -2.85578519e-01 -3.35048400e-02
-1.18766916e+00 -1.97520524e-01 6.65950000e-01 -3.60760659e-01
5.44678271e-01 3.84008914e-01 9.26630318e-01 1.19887865e+00
2.85232395e-01 5.54230690e-01 1.39462340e+00 -6.25906944e-01
2.75276989e-01 9.00195464e-02 -3.36923659e-01 7.25012541e-01
-5.87022714e-02 1.35879472e-01 -8.49377885e-02 7.30053931e-02
9.65699315e-01 1.30287632e-01 -3.11307847e-01 -2.54758984e-01
-7.20185339e-01 9.94842291e-01 8.10436130e-01 2.66080022e-01
-3.37702602e-01 3.19403231e-01 4.06143397e-01 1.91545978e-01
4.91434306e-01 4.56814528e-01 -2.74283379e-01 -4.15375419e-02
-1.02314818e+00 3.10325017e-03 6.68816566e-01 3.78165513e-01
5.71919739e-01 -1.22947693e-01 -3.33158493e-01 7.62233496e-01
1.07487291e-01 2.30402261e-01 6.11500680e-01 -7.38122702e-01
2.83807032e-02 1.00228059e+00 -3.74786556e-01 -6.80413842e-01
-3.41423690e-01 -4.80743408e-01 -1.03013396e+00 4.38000023e-01
4.82726067e-01 -1.34436116e-01 -1.81297266e+00 1.53860795e+00
3.95585001e-01 1.64883494e-01 8.33974555e-02 6.33682668e-01
5.65689802e-01 1.84826523e-01 2.63373673e-01 3.59885931e-01
1.01537585e+00 -9.08634424e-01 -3.97792935e-01 -1.38111904e-01
6.00469947e-01 -7.40936339e-01 9.05453801e-01 2.02609822e-01
-8.41231227e-01 -4.08940196e-01 -9.40520167e-01 5.06405123e-02
-5.35517335e-01 -1.76287904e-01 8.27514470e-01 9.97786164e-01
-1.19297349e+00 6.10960841e-01 -1.07216191e+00 -5.82490444e-01
1.05908680e+00 5.95705926e-01 -4.90809411e-01 -4.10254985e-01
-9.74787235e-01 8.74503791e-01 2.83860624e-01 1.74785152e-01
-8.40814531e-01 -8.94112706e-01 -9.21087861e-01 -3.10940832e-01
9.23772007e-02 -9.60120380e-01 8.13191652e-01 -1.53174651e+00
-1.54956210e+00 1.01216149e+00 1.70266926e-01 -5.18854976e-01
7.17339277e-01 7.78778195e-02 -2.73859560e-01 4.24785078e-01
-1.61275014e-01 1.01207256e+00 9.81163442e-01 -1.15047896e+00
-4.02505100e-01 -2.41854370e-01 1.24239670e-02 5.58563285e-02
-1.67829216e-01 -2.41422594e-01 -3.69912326e-01 -6.21186256e-01
-1.85141161e-01 -1.01771367e+00 -5.78623235e-01 2.48610735e-01
-4.94079381e-01 2.46195793e-01 8.03501487e-01 -7.51342475e-01
7.07631707e-01 -2.06696463e+00 8.07803571e-02 4.59415436e-01
1.66957542e-01 5.34085870e-01 -2.05682233e-01 3.42639357e-01
-2.79647801e-02 4.24844235e-01 -6.20037854e-01 -1.93045616e-01
-2.88400769e-01 2.98970222e-01 2.39142612e-01 4.96027052e-01
6.39840066e-01 1.23413253e+00 -5.81274748e-01 -5.82384348e-01
4.06447917e-01 7.42208421e-01 -6.08698249e-01 1.36318237e-01
-1.62145063e-01 7.79520988e-01 -4.08734083e-01 7.83216417e-01
5.13539135e-01 -1.12186790e-01 -2.43067443e-01 -3.05732220e-01
3.95060539e-01 -3.77388149e-01 -7.75196612e-01 1.83229887e+00
-5.04782796e-01 4.16559309e-01 2.88189929e-02 -9.88272727e-01
5.46893775e-01 3.46924692e-01 6.77158237e-01 -4.95768189e-01
2.55509853e-01 1.76813141e-01 1.50443256e-01 -4.44447488e-01
-5.82145818e-04 -3.40306133e-01 1.25900328e-01 9.14439335e-02
8.22856501e-02 -4.91263360e-01 -7.25366622e-02 -2.54553147e-02
1.19424653e+00 1.77258953e-01 3.34715486e-01 -5.46723530e-02
4.53840941e-01 3.40003163e-01 3.17679793e-01 4.53513652e-01
-2.71872103e-01 1.00338399e+00 3.74442279e-01 -1.80271000e-01
-1.00333035e+00 -9.61266220e-01 -3.41714710e-01 4.81397957e-01
-1.66533113e-01 5.27556427e-02 -1.20973468e+00 -8.98740590e-01
-1.46690533e-01 2.51318902e-01 -1.26233375e+00 -3.54343891e-01
-3.31698835e-01 -7.91832924e-01 8.32255423e-01 6.58591926e-01
6.82796597e-01 -1.20269299e+00 -5.16749263e-01 1.01221725e-01
3.65184963e-01 -9.00454938e-01 -3.68947595e-01 2.27343068e-01
-5.79466045e-01 -1.30921483e+00 -9.91145968e-01 -6.89788997e-01
1.03158069e+00 -3.00157875e-01 8.81001353e-01 3.54417652e-01
-9.09364700e-01 4.22220558e-01 -5.28750479e-01 -5.44035673e-01
-7.61962891e-01 1.79405764e-01 -4.94360298e-01 5.48963137e-02
1.80230573e-01 -4.43988264e-01 -7.92175412e-01 6.92294315e-02
-1.43121910e+00 1.08878575e-01 8.85662317e-01 1.15618658e+00
6.96871459e-01 1.82536051e-01 6.00384533e-01 -1.37507915e+00
4.63219076e-01 -3.11677992e-01 -1.50181219e-01 2.36310884e-01
-5.20944238e-01 -7.26098418e-02 6.18469059e-01 -3.86242032e-01
-8.53517771e-01 2.54975200e-01 -5.24026394e-01 -3.21753442e-01
-3.20238292e-01 4.66239214e-01 -4.35997695e-02 -6.17483079e-01
7.15550244e-01 5.10014184e-02 3.57685357e-01 4.18617502e-02
3.42281997e-01 4.96840447e-01 4.69631851e-01 -2.14070737e-01
8.63102615e-01 5.22332370e-01 3.02513540e-01 -5.06748319e-01
-7.96908438e-01 -2.33518824e-01 -7.27709413e-01 -4.23413841e-03
9.61653888e-01 -6.03523672e-01 -1.05881892e-01 7.68770397e-01
-5.95631480e-01 -5.80111325e-01 -5.12916684e-01 -2.59073800e-03
-4.67211694e-01 2.66871303e-01 -6.55629516e-01 -3.47694546e-01
-6.25292301e-01 -1.19086266e+00 1.05161107e+00 4.86857891e-01
-1.24195956e-01 -1.42137706e+00 3.59990522e-02 4.07317191e-01
8.28416049e-01 1.03027630e+00 8.75285804e-01 -5.32804132e-01
-2.07636714e-01 -5.28823733e-01 -2.56460041e-01 6.63270533e-01
4.72971946e-01 1.57017767e-01 -9.62639272e-01 -3.35284978e-01
-3.11017454e-01 -5.45509458e-01 9.65000689e-01 5.80407739e-01
1.20100009e+00 -7.64478669e-02 -4.37576920e-01 9.59699810e-01
1.76653993e+00 -8.45538303e-02 8.77579510e-01 2.61019677e-01
8.26759279e-01 5.36058724e-01 2.95696706e-01 -4.33708429e-02
5.15706353e-02 2.84630418e-01 5.55958569e-01 -6.87605083e-01
-5.61040580e-01 -2.56470442e-01 -8.48105326e-02 2.49563769e-01
-2.98473500e-02 -1.52468562e-01 -9.52647924e-01 6.50508881e-01
-1.47713256e+00 -6.35593414e-01 1.23361669e-01 1.91215360e+00
1.01328313e+00 1.45027831e-01 -7.42687732e-02 2.18027994e-01
5.47998369e-01 -2.12343290e-01 -6.87922120e-01 -4.70487654e-01
1.73235405e-02 9.55563784e-01 5.87097526e-01 2.32429281e-01
-1.07356119e+00 9.77197289e-01 6.46713352e+00 9.73219752e-01
-1.43603981e+00 -6.83254525e-02 8.28171492e-01 1.89594463e-01
-1.92623690e-01 -1.87641621e-01 -3.99112999e-01 3.79570425e-01
8.13712716e-01 2.42371455e-01 2.53474504e-01 5.38576841e-01
-1.28487125e-01 -1.66617319e-01 -9.29303527e-01 7.02375114e-01
2.46736333e-02 -1.36577928e+00 1.85611069e-01 1.83129042e-01
9.89078403e-01 -1.56645939e-01 2.40724295e-01 1.59638166e-01
2.91986793e-01 -1.77585816e+00 1.55526951e-01 6.26860917e-01
1.10803747e+00 -8.69773209e-01 1.06077993e+00 6.08915612e-02
-7.79952347e-01 1.94287404e-01 -4.07020710e-02 3.47069055e-01
-2.33118422e-02 3.95533293e-01 -1.13379025e+00 5.73737800e-01
4.32230741e-01 5.49288094e-01 -9.69584346e-01 9.86708701e-01
-1.31601498e-01 8.04059148e-01 -3.40840459e-01 2.37304613e-01
4.05702829e-01 9.89891440e-02 9.63022485e-02 1.11433244e+00
9.31716785e-02 -1.95490375e-01 9.14345980e-02 7.87583590e-01
-1.16273917e-01 1.46235479e-03 -4.45750326e-01 -1.19484387e-01
1.58270746e-01 1.46026373e+00 -7.78658628e-01 4.06650789e-02
-1.73727095e-01 1.09010994e+00 1.60336062e-01 1.55526683e-01
-7.58724749e-01 -2.57530212e-01 4.02750731e-01 2.50682294e-01
2.68198282e-01 2.14252934e-01 -3.81095856e-01 -6.99753881e-01
-2.53827602e-01 -9.83249009e-01 2.75037438e-01 -4.31388140e-01
-1.45777977e+00 7.61763871e-01 -4.10478741e-01 -9.57912743e-01
-2.69160897e-01 -6.20765686e-01 -8.25766861e-01 9.49645340e-01
-1.65991759e+00 -1.87246835e+00 -6.14954233e-01 8.16692650e-01
3.75342906e-01 -6.71186224e-02 1.21732688e+00 1.70157459e-02
-5.26982963e-01 7.64393389e-01 -1.11358039e-01 3.23452055e-01
6.87595546e-01 -1.43265796e+00 2.08918780e-01 7.91586637e-01
-8.07531774e-02 2.54349858e-01 3.64227444e-01 -4.37777609e-01
-9.96455193e-01 -1.37272334e+00 4.46136072e-02 -4.32111412e-01
4.06879038e-01 -1.17419437e-01 -8.13286245e-01 4.52872187e-01
3.90724897e-01 4.00525570e-01 8.72793138e-01 -2.74252594e-01
1.24499351e-02 2.84367781e-02 -1.66443980e+00 4.57557678e-01
7.56034315e-01 -3.19073558e-01 -2.20692873e-01 2.02895030e-01
4.23854381e-01 -5.96188843e-01 -9.99658227e-01 6.81206942e-01
4.01401460e-01 -9.09120917e-01 8.93988609e-01 -6.00649714e-01
6.90714180e-01 -8.11300427e-02 4.89827991e-02 -1.48707306e+00
-1.69366807e-01 -3.46841455e-01 3.19861323e-01 1.19670653e+00
3.10458213e-01 -6.31730080e-01 1.00429130e+00 6.77223325e-01
-1.53936148e-01 -1.17372370e+00 -8.82509530e-01 -2.59509653e-01
2.45197847e-01 -1.74138248e-01 4.16585535e-01 7.98217356e-01
-6.12024605e-01 -1.09175943e-01 -7.56508261e-02 3.87436189e-02
5.94976425e-01 -1.59274921e-01 6.80970430e-01 -9.35162544e-01
-2.08259836e-01 -4.40252244e-01 -7.53274858e-01 -2.16612414e-01
6.09657019e-02 -9.83881712e-01 -1.95405856e-01 -1.60807669e+00
1.20898180e-01 -6.62677586e-01 -4.08495933e-01 7.61067808e-01
-3.83086711e-01 9.08441603e-01 -6.27105832e-02 -8.10593367e-03
-1.39215231e-01 1.90334186e-01 1.56406164e+00 -2.51382023e-01
-1.31811261e-01 -2.03777291e-03 -7.71566570e-01 6.46065295e-01
9.22676086e-01 -3.42000157e-01 -2.91702867e-01 -1.71864361e-01
-1.82441160e-01 -2.06592381e-01 6.12110198e-01 -1.15502584e+00
1.85726643e-01 -1.17112203e-02 8.76693189e-01 -2.25698262e-01
3.08109969e-01 -7.67036676e-01 2.36866817e-01 5.81510425e-01
-2.07747459e-01 -4.70138073e-01 3.87414694e-01 5.44191241e-01
-3.68256032e-01 -2.26262629e-01 1.09231853e+00 -2.58541852e-01
-5.91304243e-01 4.15851504e-01 -5.06219566e-02 5.15326075e-02
1.23620713e+00 -5.90779662e-01 -2.04445682e-02 -1.90024093e-01
-6.19852424e-01 1.01587689e-02 8.26662242e-01 2.03852683e-01
5.06970584e-01 -1.00881720e+00 -7.63663948e-01 2.92585492e-01
6.37798086e-02 4.16665316e-01 3.33672374e-01 8.85291636e-01
-9.02386844e-01 1.12281166e-01 -4.58291292e-01 -6.84863091e-01
-1.01836312e+00 3.28905195e-01 6.54843330e-01 -5.88302493e-01
-5.74256659e-01 8.49199891e-01 -8.34094286e-02 -3.16354901e-01
-2.14803731e-03 -1.68693304e-01 -9.89089385e-02 -3.21870744e-01
1.98022559e-01 -8.59335363e-02 1.24388188e-01 -3.29952121e-01
-2.45960891e-01 7.30141103e-01 -2.63252646e-01 1.39030650e-01
1.33462214e+00 2.44450375e-01 2.50543971e-02 -1.61698252e-01
1.16332245e+00 -5.17003685e-02 -1.37663329e+00 1.87985554e-01
-2.43672237e-01 -2.98289597e-01 2.32247114e-01 -1.24661183e+00
-1.48659980e+00 6.20925665e-01 9.21020508e-01 1.96969043e-02
1.40128243e+00 -2.48772547e-01 9.24536109e-01 -3.04849088e-01
2.81620651e-01 -8.61472070e-01 -1.62559096e-02 -4.05856185e-02
6.65893853e-01 -1.32340288e+00 -3.52722257e-02 -6.69334114e-01
-6.22631729e-01 1.18283784e+00 4.72537667e-01 -4.83385295e-01
5.80248833e-01 4.35637712e-01 4.23925400e-01 -1.50873765e-01
-2.11685911e-01 -2.30614528e-01 3.98558527e-01 8.79827917e-01
2.63622940e-01 2.78224312e-02 -1.19308010e-01 3.90794754e-01
-1.54944673e-01 1.58961132e-01 3.39140326e-01 8.85970116e-01
3.54067683e-02 -1.46845031e+00 -1.41676128e-01 6.77470207e-01
-7.43525624e-01 -7.24683776e-02 -4.55183953e-01 9.92595375e-01
3.49929094e-01 7.34650254e-01 -2.60112464e-01 -2.01383531e-01
9.69445854e-02 -6.41426668e-02 7.63561130e-01 -6.65095329e-01
-9.22269106e-01 -2.08706915e-01 -2.20115781e-01 -5.01786709e-01
-5.44130921e-01 -4.04726803e-01 -1.20548904e+00 3.32838558e-02
-2.71860242e-01 -1.97860762e-01 7.83243477e-01 9.74639118e-01
2.98763007e-01 8.40472221e-01 5.76314986e-01 -6.67373896e-01
-4.82014894e-01 -1.05279195e+00 -4.86883014e-01 6.94572985e-01
2.07105830e-01 -4.45357293e-01 -1.92596585e-01 2.05680817e-01] | [15.27465534210205, -2.7162668704986572] |
8e4ff511-f664-417a-94fa-3975f237d268 | uncertainty-aware-deep-co-training-for-semi | 2111.11629 | null | https://arxiv.org/abs/2111.11629v2 | https://arxiv.org/pdf/2111.11629v2.pdf | Uncertainty-Aware Deep Co-training for Semi-supervised Medical Image Segmentation | Semi-supervised learning has made significant strides in the medical domain since it alleviates the heavy burden of collecting abundant pixel-wise annotated data for semantic segmentation tasks. Existing semi-supervised approaches enhance the ability to extract features from unlabeled data with prior knowledge obtained from limited labeled data. However, due to the scarcity of labeled data, the features extracted by the models are limited in supervised learning, and the quality of predictions for unlabeled data also cannot be guaranteed. Both will impede consistency training. To this end, we proposed a novel uncertainty-aware scheme to make models learn regions purposefully. Specifically, we employ Monte Carlo Sampling as an estimation method to attain an uncertainty map, which can serve as a weight for losses to force the models to focus on the valuable region according to the characteristics of supervised learning and unsupervised learning. Simultaneously, in the backward process, we joint unsupervised and supervised losses to accelerate the convergence of the network via enhancing the gradient flow between different tasks. Quantitatively, we conduct extensive experiments on three challenging medical datasets. Experimental results show desirable improvements to state-of-the-art counterparts. | ['Chiu-Wing Sham', 'Xingwei Wang', 'Jialei Chen', 'Haoyu Xie', 'Chong Fu', 'Xu Zheng'] | 2021-11-23 | null | null | null | null | ['semi-supervised-medical-image-segmentation'] | ['computer-vision'] | [ 3.52937490e-01 4.33413118e-01 -5.31731248e-01 -8.17543983e-01
-7.70977199e-01 -1.12190358e-01 1.17774479e-01 9.74999517e-02
-5.86759090e-01 1.03181696e+00 2.11687712e-03 -3.57085876e-02
-1.32081807e-01 -6.11817360e-01 -6.33382440e-01 -9.63427186e-01
4.27876115e-01 3.84019673e-01 1.64420947e-01 2.62007505e-01
-1.54245242e-01 9.27107260e-02 -9.42305505e-01 4.22336301e-03
1.43994284e+00 1.06157804e+00 5.45266271e-01 -2.53064007e-01
-2.15925679e-01 8.06736588e-01 4.36470099e-02 -1.52843729e-01
5.41820098e-03 -4.68758702e-01 -6.59823656e-01 5.54239035e-01
-1.90094814e-01 -3.41707736e-01 -3.47880214e-01 1.40998137e+00
3.55430394e-01 1.05696963e-02 5.59007764e-01 -9.94654834e-01
-4.13119078e-01 5.71484625e-01 -8.29772949e-01 -8.05076882e-02
-4.42104638e-01 9.16657373e-02 8.91383290e-01 -7.12285101e-01
5.58817863e-01 7.90491164e-01 4.38536912e-01 5.95465302e-01
-9.91809309e-01 -6.07511938e-01 3.62526804e-01 -8.37127939e-02
-1.32124150e+00 -2.68522918e-01 1.04226148e+00 -2.42048725e-01
-3.02330479e-02 -8.49446654e-02 5.55132210e-01 6.34979367e-01
-2.37296253e-01 1.43025446e+00 9.55699861e-01 -2.89290428e-01
2.58564442e-01 5.25390327e-01 -1.63379312e-02 9.20017958e-01
1.49377078e-01 -1.75509264e-03 -2.80141950e-01 1.06475502e-02
8.51867676e-01 2.38907486e-01 -4.23647702e-01 -7.46003389e-01
-1.01084423e+00 7.12501168e-01 7.92381465e-01 9.81363207e-02
-4.67510313e-01 -2.23461390e-01 4.44590032e-01 -1.19025581e-01
7.93254316e-01 1.87013417e-01 -4.82385784e-01 2.27043971e-01
-1.11117208e+00 -2.01713040e-01 2.07211822e-01 7.65225947e-01
9.08999324e-01 -2.56972998e-01 -1.87902689e-01 8.58355820e-01
5.61061978e-01 1.38084278e-01 4.61368740e-01 -6.80934072e-01
5.89548826e-01 8.72716129e-01 1.01697870e-01 -6.45063102e-01
-2.74373591e-01 -7.21523583e-01 -9.33270574e-01 -1.43273324e-01
7.16449022e-01 -3.37574303e-01 -1.03014314e+00 1.61803865e+00
6.04596734e-01 7.64237717e-02 -1.78253934e-01 1.14919770e+00
4.58592147e-01 4.63371903e-01 1.01840883e-01 -4.56767440e-01
9.91189301e-01 -1.10700679e+00 -7.46865332e-01 -2.27099210e-01
6.48378134e-01 -3.92490804e-01 1.08390915e+00 2.28701994e-01
-7.67911017e-01 -4.57410395e-01 -9.64983344e-01 2.42032439e-01
2.38034964e-01 3.87283057e-01 6.73216581e-01 4.30206925e-01
-4.90662873e-01 5.61158121e-01 -1.16125762e+00 1.76415518e-01
1.13626063e+00 1.89907312e-01 -5.24226986e-02 -2.86885262e-01
-1.17802155e+00 4.18448269e-01 5.62094688e-01 4.19471323e-01
-8.94402444e-01 -6.00803256e-01 -8.15699399e-01 -1.48984134e-01
6.76015079e-01 -3.38313013e-01 1.04378283e+00 -1.14515877e+00
-1.13456130e+00 7.78645158e-01 -1.20089119e-02 -3.76031816e-01
9.14478481e-01 -1.96300924e-01 -8.07253197e-02 2.47916728e-01
2.65737504e-01 8.50814402e-01 7.04897523e-01 -1.29294443e+00
-7.32815146e-01 -4.92729872e-01 -1.90931067e-01 2.39890531e-01
-5.16076148e-01 -4.06649828e-01 -6.73879862e-01 -6.20803118e-01
3.02763999e-01 -7.51023352e-01 -6.69680417e-01 4.21352357e-01
-5.75147092e-01 -3.34559456e-02 6.87033355e-01 -5.59379280e-01
1.04751265e+00 -2.19147325e+00 7.75309578e-02 3.06011230e-01
2.58442074e-01 2.44588196e-01 2.16754645e-01 -2.66483694e-01
3.12024653e-01 -1.08441330e-01 -7.61058450e-01 -3.09588701e-01
-2.91980147e-01 4.17494357e-01 -1.74609497e-01 5.67528784e-01
3.37163299e-01 8.33462715e-01 -1.11154997e+00 -1.05891442e+00
1.73449561e-01 3.39869380e-01 -4.46258634e-01 2.61684448e-01
-2.70867288e-01 9.30754125e-01 -1.15185440e+00 5.65748930e-01
6.78048611e-01 -5.80621600e-01 9.74904075e-02 -3.35525647e-02
2.43163988e-01 4.68844064e-02 -8.93582463e-01 1.97976720e+00
-4.73298013e-01 5.28236330e-02 1.41046986e-01 -1.29908550e+00
7.62086034e-01 2.45846570e-01 6.22024000e-01 -4.63866025e-01
1.36621803e-01 3.11004162e-01 -2.60047138e-01 -4.87350285e-01
-1.03981361e-01 -3.20946097e-01 1.47491485e-01 2.16290489e-01
-7.17587173e-02 -7.43910968e-02 5.91399893e-02 1.57816038e-01
4.42961663e-01 3.37790757e-01 7.42504001e-02 -2.55545676e-01
5.69488525e-01 6.83187693e-02 1.10676026e+00 3.89785767e-01
-4.17675823e-01 6.05494201e-01 4.55551565e-01 -1.26972333e-01
-7.42339194e-01 -1.00049806e+00 -4.28711683e-01 6.21614277e-01
4.43551749e-01 9.35138315e-02 -8.03218782e-01 -1.32466054e+00
-1.94508523e-01 5.46541691e-01 -5.40557921e-01 -3.07546705e-01
-3.65960717e-01 -9.84841585e-01 1.71437860e-01 7.11139321e-01
6.51354551e-01 -1.06415021e+00 -4.53587413e-01 2.90721297e-01
-3.29973936e-01 -9.49859023e-01 -6.06892169e-01 1.89458176e-01
-1.23648012e+00 -1.06151354e+00 -1.05780804e+00 -8.92252207e-01
1.29974604e+00 3.24549787e-02 8.16899478e-01 8.51779506e-02
-3.14610928e-01 -2.56427675e-01 -2.42661238e-01 -3.66805196e-01
-8.62977728e-02 6.14206828e-02 -2.55752146e-01 1.96540728e-01
2.35291049e-01 -3.65479380e-01 -8.08136702e-01 4.38366979e-01
-8.50509763e-01 2.44352221e-01 7.34081030e-01 1.29110074e+00
9.70806479e-01 2.34993011e-01 7.64949143e-01 -1.16234875e+00
1.61811933e-01 -3.73396248e-01 -5.97862959e-01 2.89577723e-01
-8.05211544e-01 1.84229329e-01 6.66425943e-01 -2.84222633e-01
-1.47151995e+00 4.71692622e-01 -8.85468721e-02 -3.47661585e-01
1.31067529e-01 5.40101051e-01 -2.93484271e-01 1.50993481e-01
2.14016050e-01 3.80043298e-01 1.68576375e-01 -3.20496142e-01
2.26564422e-01 7.06991196e-01 2.93653429e-01 -5.72384477e-01
6.03253365e-01 7.26330340e-01 -3.26411963e-01 -3.00053895e-01
-1.43949187e+00 -5.49799442e-01 -5.23687601e-01 -1.49696663e-01
7.29489625e-01 -9.63989198e-01 -2.10214466e-01 4.00510609e-01
-5.79358757e-01 -1.84529141e-01 -5.37090182e-01 7.10439861e-01
-4.28538084e-01 5.33080339e-01 -4.99769568e-01 -8.25812399e-01
-3.38714302e-01 -1.13527417e+00 9.70139802e-01 5.87553024e-01
1.64035246e-01 -9.61404204e-01 -2.24254027e-01 5.58145344e-01
1.00675464e-01 6.05021417e-02 6.98533773e-01 -3.92626733e-01
-4.57441360e-01 -1.83088645e-01 -5.64892054e-01 5.54394722e-01
3.39028984e-01 -4.21302557e-01 -8.59423518e-01 -2.02751204e-01
1.15204789e-01 -8.74228656e-01 1.19129980e+00 6.74309731e-01
1.58232415e+00 -1.62550911e-01 -5.81370592e-01 5.59352160e-01
1.18557405e+00 4.55489801e-03 2.81137764e-01 5.63036464e-02
7.90077031e-01 7.46159852e-01 1.06427288e+00 5.00664115e-01
2.43602961e-01 3.26760381e-01 3.97403061e-01 -2.93815017e-01
1.47832125e-01 -5.43207049e-01 -2.09077194e-01 6.58927977e-01
3.02084565e-01 8.84895176e-02 -7.06164479e-01 7.31339395e-01
-2.06301379e+00 -3.40335757e-01 2.51576364e-01 2.09296489e+00
1.26129341e+00 4.11255091e-01 -1.38394490e-01 -6.56946152e-02
8.38980258e-01 1.33264005e-01 -9.70378160e-01 6.09934092e-01
1.34580031e-01 -2.54758000e-01 4.55250740e-01 1.92354083e-01
-1.26497734e+00 8.44126821e-01 4.90600348e+00 1.16497743e+00
-1.02379215e+00 1.25480011e-01 1.15472400e+00 -7.48495758e-02
-3.56076002e-01 -1.24004990e-01 -5.96461177e-01 6.56644762e-01
2.36290604e-01 8.24771300e-02 7.56397396e-02 9.64085340e-01
4.31713015e-01 -1.23060688e-01 -9.31021631e-01 7.87970662e-01
-3.86393130e-01 -1.11267781e+00 -1.48508281e-01 -7.31306076e-02
1.00914013e+00 5.90007147e-03 3.27234082e-02 1.57812208e-01
2.35637993e-01 -7.71263599e-01 5.10946631e-01 4.75717425e-01
7.18217313e-01 -7.96483040e-01 8.78036916e-01 7.34014988e-01
-9.03441310e-01 5.24804257e-02 -6.00815892e-01 2.24234000e-01
2.95543492e-01 1.11822116e+00 -6.60738349e-01 5.26724160e-01
3.88267934e-01 8.21011841e-01 -1.44772574e-01 9.87682462e-01
-4.11557436e-01 7.86225736e-01 -3.41838181e-01 9.41163152e-02
4.61937785e-01 -3.64868641e-01 6.71921447e-02 6.61175847e-01
-5.58990091e-02 8.30516890e-02 5.45575380e-01 1.00662887e+00
-2.99910188e-01 2.32641071e-01 -1.07971653e-01 -4.45232913e-02
4.40520197e-01 1.22451842e+00 -8.74012411e-01 -1.30931586e-01
-2.63232112e-01 8.16351652e-01 4.91287261e-01 2.66480803e-01
-8.14806700e-01 -2.50954092e-01 -2.00674981e-02 5.36345430e-02
1.01823635e-01 1.58420533e-01 -5.52654445e-01 -1.17261398e+00
2.11121634e-01 -3.42346907e-01 3.87851745e-01 -3.65645289e-01
-1.42490757e+00 3.89500469e-01 -2.49185219e-01 -1.35082376e+00
-2.83430684e-02 -1.71772823e-01 -5.24347723e-01 7.20130503e-01
-1.85903049e+00 -9.69802976e-01 -2.29278088e-01 3.82693797e-01
5.08040130e-01 3.56120653e-02 3.08529049e-01 3.93267930e-01
-6.92257881e-01 5.86314499e-01 2.93259084e-01 3.32829922e-01
7.03408897e-01 -1.12919915e+00 -1.52123496e-01 6.81419432e-01
5.15872389e-02 3.46433550e-01 3.39706331e-01 -5.35163164e-01
-7.04378545e-01 -1.31484747e+00 4.55311894e-01 1.10109128e-01
4.38490301e-01 -4.85140271e-02 -1.01641476e+00 2.90838212e-01
-5.43207109e-01 5.14217913e-01 5.73267579e-01 -1.23607136e-01
9.68242139e-02 -2.20324114e-01 -1.24988043e+00 3.82782698e-01
9.16332066e-01 -3.76889229e-01 -4.14237082e-01 4.36285496e-01
7.02479601e-01 -3.27037007e-01 -7.33692348e-01 5.06811500e-01
2.79724926e-01 -6.04388773e-01 6.56592309e-01 -4.15146589e-01
5.29659629e-01 -3.52000833e-01 2.07735285e-01 -1.27305603e+00
1.58434466e-01 -2.70310789e-01 3.66234817e-02 1.31155336e+00
5.79136610e-01 -5.45348942e-01 1.12109506e+00 6.52133763e-01
-1.36078775e-01 -1.19420385e+00 -8.16206396e-01 -4.59900349e-01
-3.20263393e-02 -3.02496016e-01 3.41618776e-01 8.66118491e-01
-1.12058725e-02 5.36837466e-02 -4.65859711e-01 4.94959988e-02
8.84832084e-01 2.52670765e-01 2.43697181e-01 -1.01981294e+00
-3.99963856e-01 -1.14486381e-01 -8.83742869e-02 -1.21509063e+00
1.22281991e-01 -7.41272271e-01 2.88470685e-01 -1.53893590e+00
5.56516171e-01 -9.21801388e-01 -6.90551043e-01 5.64816117e-01
-6.52845085e-01 9.51278210e-02 -8.06403682e-02 5.12860596e-01
-7.92452991e-01 9.96483564e-01 1.75372195e+00 -2.05890894e-01
-1.86993778e-01 2.72814214e-01 -7.47019172e-01 9.42133009e-01
7.05632985e-01 -6.24099255e-01 -8.18289816e-01 -2.99068898e-01
-6.79483935e-02 2.01367483e-01 2.25707933e-01 -7.85373628e-01
9.40070599e-02 -1.91832125e-01 4.89448756e-01 -4.63360071e-01
-2.22893991e-02 -8.16240728e-01 -4.67302650e-01 4.22873318e-01
-5.63579142e-01 -1.02887154e+00 -1.80614293e-01 9.14873362e-01
-4.82039154e-01 -2.29369253e-01 8.63239586e-01 -5.20762838e-02
-5.40267527e-01 6.44703209e-01 2.70253032e-01 3.00059766e-01
9.58332241e-01 -2.88159028e-02 1.47534326e-01 -2.08249733e-01
-7.85073638e-01 8.55976760e-01 3.68246466e-01 1.96082190e-01
5.84375978e-01 -1.10580504e+00 -5.83473146e-01 8.10582563e-02
1.84628159e-01 6.63723290e-01 5.29053092e-01 8.99321795e-01
-3.54915887e-01 2.76676655e-01 -3.96868028e-02 -6.36415064e-01
-7.42500186e-01 6.43011808e-01 3.36580426e-01 -5.87884009e-01
-6.03571713e-01 8.87378812e-01 5.22532225e-01 -6.30952001e-01
3.21598709e-01 -9.34271440e-02 -2.14502990e-01 -1.53912067e-01
3.92474055e-01 7.61109591e-02 -1.43588260e-01 -2.40932569e-01
-3.44109952e-01 2.22167790e-01 -5.15095949e-01 1.70367867e-01
1.22207832e+00 -3.40751797e-01 2.45240420e-01 3.13712388e-01
1.17167544e+00 -3.55387568e-01 -1.98048937e+00 -6.11875534e-01
-1.46879971e-01 -3.79666775e-01 4.77956682e-01 -6.96065187e-01
-1.46768606e+00 9.18637931e-01 5.93639493e-01 -3.46884876e-01
1.17230797e+00 1.28511101e-01 1.03284013e+00 2.55361885e-01
3.50700825e-01 -1.36107028e+00 -1.06113657e-01 -1.37951657e-01
2.08592221e-01 -1.62299740e+00 -2.40621828e-02 -7.15438724e-01
-9.46838677e-01 7.47652471e-01 6.47604704e-01 3.88685651e-02
6.92002237e-01 1.09626956e-01 7.64909163e-02 -1.05325907e-01
-3.49983126e-01 -6.90411627e-02 2.62374789e-01 3.33095551e-01
3.29580158e-01 5.80627471e-02 -4.03495312e-01 8.10516894e-01
4.23384935e-01 3.75404507e-01 7.20395520e-02 9.14714515e-01
-5.15608728e-01 -8.64335358e-01 -1.10937329e-02 6.58574283e-01
-5.56696713e-01 4.03487869e-02 9.31607187e-02 5.03913879e-01
3.55884954e-02 7.55214155e-01 -1.97997808e-01 9.13279578e-02
-1.37729729e-02 -2.83630848e-01 2.74588466e-01 -5.01932979e-01
7.14681298e-02 3.12178135e-01 -1.99140295e-01 -2.56640345e-01
-3.62948626e-01 -5.83145022e-01 -1.67082500e+00 4.77339447e-01
-6.53168023e-01 3.89079601e-01 3.09956312e-01 1.18602669e+00
6.82057142e-02 6.82194054e-01 8.98507178e-01 -4.25134391e-01
-9.17555094e-01 -8.49364400e-01 -7.36625135e-01 4.11779195e-01
8.65349695e-02 -6.17027164e-01 -9.95409191e-02 8.70554447e-02] | [14.699736595153809, -2.010460376739502] |
472c8657-e046-4f37-aa1d-488a5e24dc95 | reactive-exploration-to-cope-with-non | 2207.05742 | null | https://arxiv.org/abs/2207.05742v2 | https://arxiv.org/pdf/2207.05742v2.pdf | Reactive Exploration to Cope with Non-Stationarity in Lifelong Reinforcement Learning | In lifelong learning, an agent learns throughout its entire life without resets, in a constantly changing environment, as we humans do. Consequently, lifelong learning comes with a plethora of research problems such as continual domain shifts, which result in non-stationary rewards and environment dynamics. These non-stationarities are difficult to detect and cope with due to their continuous nature. Therefore, exploration strategies and learning methods are required that are capable of tracking the steady domain shifts, and adapting to them. We propose Reactive Exploration to track and react to continual domain shifts in lifelong reinforcement learning, and to update the policy correspondingly. To this end, we conduct experiments in order to investigate different exploration strategies. We empirically show that representatives of the policy-gradient family are better suited for lifelong learning, as they adapt more quickly to distribution shifts than Q-learning. Thereby, policy-gradient methods profit the most from Reactive Exploration and show good results in lifelong learning with continual domain shifts. Our code is available at: https://github.com/ml-jku/reactive-exploration. | ['Sepp Hochreiter', 'Hamid Eghbal-zadeh', 'Angela Bitto-Nemling', 'Vihang Patil', 'Marius-Constantin Dinu', 'Fabian Paischer', 'Thomas Schmied', 'Christian Steinparz'] | 2022-07-12 | null | null | null | null | ['policy-gradient-methods'] | ['methodology'] | [-3.56868565e-01 -2.70497948e-01 -4.20944154e-01 9.10092071e-02
-3.68163794e-01 -6.22418940e-01 5.82333028e-01 1.01955451e-01
-5.98934591e-01 1.44542575e+00 -1.80466250e-01 -1.78073272e-01
-3.93609852e-01 -6.78626895e-01 -6.27701879e-01 -9.80545163e-01
-4.95378882e-01 6.81115150e-01 3.07565272e-01 -4.23704565e-01
3.50012243e-01 1.82299078e-01 -1.52993369e+00 -4.87372607e-01
1.36067426e+00 4.56162244e-01 4.40131694e-01 6.38097584e-01
-1.24358959e-01 4.63998765e-01 -5.32091856e-01 1.55030772e-01
2.85071999e-01 -6.58333421e-01 -5.78362226e-01 -2.04833075e-02
-2.86222488e-01 -1.61314696e-01 -9.66807231e-02 1.08461404e+00
7.12406933e-01 5.31824946e-01 4.66383398e-01 -1.30510521e+00
-5.34059703e-01 6.92453146e-01 -6.49131775e-01 4.91242915e-01
3.18338782e-01 5.11439383e-01 6.08358860e-01 -6.33548915e-01
3.90446872e-01 1.35732532e+00 4.24748391e-01 7.78943419e-01
-1.16110981e+00 -6.15040064e-01 5.02289295e-01 1.93520874e-01
-8.35327208e-01 -2.81261176e-01 7.52683103e-01 -1.12317987e-01
5.65665781e-01 -1.21521531e-02 9.45016861e-01 1.21806753e+00
4.96580958e-01 9.07076657e-01 1.37961996e+00 -3.22957337e-01
7.31099904e-01 1.23001941e-01 -3.07778001e-01 3.13746363e-01
3.47348303e-01 6.54656827e-01 -4.33380842e-01 -8.68673101e-02
7.57857919e-01 1.84483215e-01 -9.42844376e-02 -8.30920756e-01
-9.75089133e-01 6.62922740e-01 2.75745362e-01 2.83967346e-01
-6.28626585e-01 3.49992871e-01 2.75012791e-01 9.00271297e-01
4.19287890e-01 7.89808333e-01 -4.85948890e-01 -7.30062723e-01
-3.51108134e-01 5.46860039e-01 9.50775027e-01 8.39566231e-01
8.12717140e-01 1.68233901e-01 -1.42524675e-01 7.87721574e-01
-1.71664823e-02 6.23981178e-01 9.64778721e-01 -1.09873319e+00
1.71366259e-01 2.36306548e-01 5.55733025e-01 -3.95159513e-01
-4.69583303e-01 -8.37775588e-01 -5.92212081e-01 4.14776593e-01
5.86821496e-01 -6.35008574e-01 -5.74880362e-01 1.82330179e+00
5.89327335e-01 1.09231815e-01 2.24428207e-01 6.77252054e-01
-1.11988574e-01 5.45580387e-01 -8.80015194e-02 -7.05274999e-01
8.51271868e-01 -7.94125736e-01 -6.98722005e-01 -2.98838645e-01
5.46923816e-01 -4.77098703e-01 1.56095088e+00 4.05065954e-01
-1.22938061e+00 -2.88719118e-01 -7.37819314e-01 7.72580624e-01
-1.56297982e-01 -7.67781198e-01 3.27438802e-01 3.53028774e-01
-8.91053736e-01 8.02480638e-01 -9.46506560e-01 -3.62923145e-01
3.06433260e-01 6.70873374e-02 4.49113220e-01 1.64156005e-01
-1.32755744e+00 6.72940731e-01 5.57065010e-01 -1.87771827e-01
-9.99221981e-01 -5.68695784e-01 -4.39715385e-01 -2.49394625e-02
9.27435160e-01 -4.08610582e-01 1.77062488e+00 -1.12562883e+00
-1.75880694e+00 3.33699971e-01 2.44232893e-01 -5.34640789e-01
9.46388662e-01 -2.71515578e-01 -1.31221846e-01 -2.04297975e-01
6.24645837e-02 3.93117368e-01 9.87586915e-01 -1.23343873e+00
-8.01510990e-01 -2.54649848e-01 4.40577641e-02 6.41639590e-01
-5.42295516e-01 -4.40375119e-01 2.66104247e-02 -5.73195815e-01
-2.75721818e-01 -1.19264400e+00 -5.96682727e-01 -4.16243911e-01
-4.37305047e-04 -4.15250868e-01 7.14230835e-01 2.61029154e-01
1.17769873e+00 -2.15428495e+00 -5.53508699e-02 7.00892583e-02
3.13531458e-02 1.95880458e-01 -1.63241744e-01 6.69532061e-01
3.22756737e-01 -1.91591844e-01 -1.69766918e-01 -7.99885318e-02
1.18546702e-01 4.41615790e-01 -2.22113386e-01 3.14159960e-01
-2.11974949e-01 9.49404418e-01 -1.37694490e+00 -2.00402230e-01
-1.73647344e-01 -8.91571715e-02 -4.64682221e-01 2.90451586e-01
-6.72999382e-01 6.37176394e-01 -6.45676553e-01 5.37173152e-01
3.78714502e-01 -3.78823072e-01 2.82677382e-01 9.68582451e-01
-2.40587890e-01 -7.01811463e-02 -1.05777395e+00 1.36430323e+00
-4.20684904e-01 2.99109340e-01 -5.49827591e-02 -1.10029936e+00
1.02033675e+00 9.55849439e-02 4.85126257e-01 -1.08010089e+00
-2.59362794e-02 4.33351040e-01 9.21171382e-02 -4.55527782e-01
3.84387434e-01 -7.23726079e-02 -1.34573216e-02 6.83772147e-01
-4.32980150e-01 -1.55905157e-01 4.39376771e-01 -8.01069662e-02
9.74876344e-01 5.11469617e-02 3.69724184e-01 -5.64276338e-01
3.04424137e-01 -1.06184937e-01 8.87327611e-01 1.05211556e+00
-4.12247390e-01 -1.83888763e-01 5.83690345e-01 -5.62271714e-01
-7.50702381e-01 -9.91498947e-01 1.06498934e-01 1.31833947e+00
4.72689092e-01 1.14461407e-01 -4.91268307e-01 -6.01202965e-01
3.36400688e-01 7.29117393e-01 -3.78211975e-01 -3.52503896e-01
-8.05765092e-01 -6.68536961e-01 -1.69061482e-01 1.81560025e-01
5.65841496e-01 -1.49423361e+00 -1.22043788e+00 5.85802674e-01
2.59355098e-01 -4.11716223e-01 -5.42183101e-01 4.74274278e-01
-1.08154786e+00 -1.02458286e+00 -1.09494758e+00 -6.73111856e-01
6.03556633e-01 1.72541216e-01 1.07507050e+00 -8.67619589e-02
-9.90562961e-02 5.43680906e-01 -3.54693502e-01 -4.76067781e-01
-5.26487052e-01 3.35208923e-01 3.44890594e-01 -4.19648379e-01
1.50362104e-01 -8.38664770e-01 -8.89804006e-01 5.66981077e-01
-9.12962675e-01 -4.06453907e-01 6.09059632e-01 1.15650654e+00
6.28861964e-01 2.29153737e-01 1.21256638e+00 -9.67528999e-01
1.22357070e+00 -7.62630522e-01 -1.02118528e+00 2.21541524e-01
-1.25556445e+00 2.12260827e-01 9.32931840e-01 -9.13505614e-01
-1.02296221e+00 -2.79431850e-01 2.71748066e-01 -2.95913726e-01
-9.09278840e-02 4.49529201e-01 2.14838162e-01 2.24416777e-01
9.02159095e-01 3.89460981e-01 1.79585874e-01 -3.97646487e-01
3.72596204e-01 4.34448451e-01 2.33781964e-01 -8.81541371e-01
8.52925062e-01 1.20128751e-01 -1.79876685e-01 -5.07274747e-01
-4.76634622e-01 -1.25346363e-01 -1.81590244e-01 -3.33071560e-01
1.14214592e-01 -5.21337748e-01 -6.79131448e-01 6.10844851e-01
-4.95654374e-01 -1.08060515e+00 -8.94548953e-01 3.20475966e-01
-9.33357179e-01 1.91156060e-01 -3.03251058e-01 -1.01696408e+00
-1.42654449e-01 -7.75486350e-01 2.62824923e-01 8.71290565e-01
-4.15650718e-02 -1.16805542e+00 5.80365181e-01 -4.65064734e-01
6.09977067e-01 1.19820446e-01 6.15395010e-01 -4.41716164e-01
-5.44327676e-01 3.38203877e-01 3.81555647e-01 -1.09857038e-01
4.11237538e-01 -3.65654469e-01 -4.10761237e-01 -8.95155907e-01
-6.40819445e-02 -5.09103477e-01 6.46156669e-01 4.45180207e-01
9.79891539e-01 -5.32942295e-01 -1.87284514e-01 5.50147891e-01
1.12345898e+00 8.12633038e-01 1.42363831e-01 7.69683540e-01
7.58676603e-02 6.36797190e-01 1.07728720e+00 1.00524831e+00
2.85338163e-01 4.22668248e-01 6.07629061e-01 1.66660398e-01
3.58635366e-01 -4.15829509e-01 3.73393029e-01 4.30940002e-01
1.95087418e-01 -1.82457343e-01 -1.02109396e+00 7.38662422e-01
-2.15280724e+00 -8.11952055e-01 5.34169257e-01 2.27951980e+00
1.28951609e+00 3.75811219e-01 6.02691352e-01 -1.96990162e-01
5.58875918e-01 2.94880718e-02 -1.45291770e+00 -3.33116472e-01
3.94872110e-03 -1.31095260e-01 4.13267821e-01 4.19261903e-01
-6.91216409e-01 1.07105827e+00 6.00517130e+00 6.09994292e-01
-1.18294656e+00 -4.92797345e-02 6.81162059e-01 -3.07545871e-01
-3.74694973e-01 -9.22362134e-02 -6.82977378e-01 8.01528811e-01
9.83670235e-01 -6.80118620e-01 6.84083939e-01 9.26049829e-01
2.94157058e-01 -2.26404473e-01 -8.58884394e-01 8.71043265e-01
-5.78440189e-01 -1.01528251e+00 -5.75404942e-01 2.42935047e-02
1.25293863e+00 2.12123711e-02 2.79921144e-01 5.69312155e-01
8.15712154e-01 -7.93594360e-01 3.22229505e-01 4.97602433e-01
5.44289470e-01 -1.13278317e+00 3.40900958e-01 7.30179250e-01
-8.89809787e-01 -6.00158036e-01 -4.42099839e-01 -1.57896936e-01
-8.70659947e-02 5.15997410e-01 -6.59937143e-01 3.51244062e-02
6.83509767e-01 6.20005369e-01 -1.80259675e-01 1.27853799e+00
-3.04286987e-01 5.59696615e-01 -1.72746554e-01 -6.31946802e-01
2.70311415e-01 -2.12953418e-01 7.44852901e-01 7.66509533e-01
4.44885790e-01 -1.95629254e-01 5.06429434e-01 7.25598216e-01
6.33912086e-02 5.67304529e-02 -5.80355763e-01 -1.67886876e-02
7.69132674e-01 7.87478507e-01 -6.82958007e-01 -2.59864628e-01
8.44240114e-02 6.57258987e-01 2.23175034e-01 6.45924509e-01
-6.22122884e-01 -4.76912916e-01 6.83535039e-01 8.27305540e-02
1.96846098e-01 -3.30336243e-01 6.80159926e-02 -9.32414591e-01
-1.14209689e-01 -9.92946804e-01 4.57507789e-01 -5.58263548e-02
-1.35251665e+00 4.62806970e-01 1.34921700e-01 -1.18543828e+00
-7.53647804e-01 9.52375829e-02 -6.49920046e-01 3.76459509e-01
-1.87031567e+00 -2.13070840e-01 -2.05668926e-01 7.47277319e-01
8.19442630e-01 -2.03675061e-01 4.54307288e-01 -1.01861276e-01
-4.59806472e-01 5.72789967e-01 7.74378836e-01 -4.61498231e-01
7.56910741e-01 -1.57002807e+00 3.04457128e-01 5.15828013e-01
-1.48252919e-01 4.09538329e-01 9.07362700e-01 -6.80431604e-01
-1.51911402e+00 -8.97859156e-01 1.85333177e-01 2.01500610e-01
8.29968452e-01 -2.73393273e-01 -1.11348414e+00 1.89166650e-01
1.11348890e-01 -4.59212326e-02 1.49163947e-01 8.53304788e-02
2.71437019e-01 -3.07418346e-01 -9.47837114e-01 7.12196290e-01
1.15881264e+00 -1.40295118e-01 -2.23279119e-01 3.01318198e-01
5.87041199e-01 -4.49904382e-01 -5.59756041e-01 1.56926855e-01
3.50237995e-01 -9.45121050e-01 6.40927076e-01 -3.93307567e-01
2.76032812e-03 1.61828287e-02 4.61598754e-01 -1.97055268e+00
-2.01483741e-01 -1.28795791e+00 -4.96308863e-01 9.24630642e-01
1.92078233e-01 -1.33067584e+00 9.71547484e-01 9.82918665e-02
2.19225213e-01 -9.41973269e-01 -9.97491360e-01 -1.46887040e+00
4.20592099e-01 2.79732347e-01 7.43569613e-01 7.86618114e-01
7.77263716e-02 -9.97119471e-02 -4.70659882e-01 -2.79727131e-01
8.09535921e-01 3.67265344e-01 7.59621978e-01 -1.14131987e+00
-5.16001105e-01 -5.45905769e-01 2.79713005e-01 -1.23848355e+00
6.09998368e-02 -3.58057350e-01 3.72856319e-01 -1.11395931e+00
-1.43243531e-02 -7.69931078e-01 -5.87427258e-01 2.58861333e-01
-3.43344927e-01 -3.49373460e-01 1.02248423e-01 5.71638942e-01
-8.19612503e-01 1.04202843e+00 1.47494888e+00 8.21168795e-02
-8.54451895e-01 4.15346116e-01 -7.48736620e-01 5.16141295e-01
1.33272994e+00 -5.41216850e-01 -8.75814676e-01 -5.75329848e-02
3.27579707e-01 3.60744774e-01 -1.29686788e-01 -9.65874851e-01
1.96064338e-01 -6.90692306e-01 -1.07937902e-01 -2.75093138e-01
-9.30267423e-02 -5.14199793e-01 -1.09845653e-01 1.05078089e+00
-3.44605356e-01 3.40053111e-01 6.23579957e-02 6.69634640e-01
-3.68451402e-02 -1.14061296e-01 9.05826509e-01 -1.95285708e-01
-7.46074736e-01 4.83261526e-01 -5.72459698e-01 6.33669317e-01
1.25084722e+00 -2.79947966e-01 -3.12540412e-01 -6.46688461e-01
-6.06834352e-01 9.14027035e-01 5.92628419e-01 3.91260892e-01
5.00803530e-01 -1.13976538e+00 -5.59356213e-01 1.53991625e-01
-1.12098582e-01 1.05337322e-01 8.59501436e-02 6.43873632e-01
-1.10260732e-01 1.71414897e-01 -4.40857649e-01 -5.66962600e-01
-8.89649391e-01 5.29977381e-01 4.83559191e-01 -4.23595965e-01
-5.54132402e-01 6.83144510e-01 1.90585516e-02 -4.01384890e-01
5.52645624e-01 -2.29834229e-01 -4.24249582e-02 1.59188733e-01
4.03122604e-01 4.99254197e-01 -4.69842017e-01 2.96288013e-01
-1.19242959e-01 3.85612011e-01 -8.28194544e-02 -4.57907543e-02
1.35244429e+00 -3.27772975e-01 2.78021455e-01 6.29690409e-01
5.89823306e-01 -3.30921441e-01 -1.94583941e+00 -5.49835682e-01
3.79913926e-01 -6.48750484e-01 -3.38665962e-01 -6.96468115e-01
-6.92461431e-01 4.90440637e-01 7.94498920e-01 6.03194773e-01
1.07331455e+00 -1.66875750e-01 7.05963492e-01 6.10864460e-01
6.21112347e-01 -1.47203267e+00 6.82329476e-01 7.01971114e-01
7.36260772e-01 -1.11255074e+00 -2.44487762e-01 4.21855450e-01
-6.13005936e-01 9.34923708e-01 7.48497725e-01 -2.54438490e-01
4.78273451e-01 2.03299418e-01 1.17535822e-01 7.37015009e-02
-1.12952399e+00 -2.14168042e-01 -3.13045144e-01 7.11506069e-01
2.79245758e-03 1.34885460e-01 -3.65589112e-01 -2.24883631e-01
-7.80294612e-02 -7.28607625e-02 4.89488661e-01 1.14043868e+00
-7.91674495e-01 -1.43427324e+00 -3.91855359e-01 1.84419677e-01
-1.54363856e-01 3.76051992e-01 -1.86084017e-01 7.19931841e-01
-2.36362830e-01 7.29431987e-01 -8.72801542e-02 1.47693962e-01
2.28293568e-01 -2.34880880e-01 2.51884043e-01 -4.52992976e-01
-2.87033051e-01 2.07112834e-01 -2.34436065e-01 -6.23910427e-01
-1.34175420e-01 -9.67236161e-01 -1.46434057e+00 -3.11576545e-01
-8.64454359e-02 5.32095969e-01 4.28015739e-01 5.35429180e-01
2.47881964e-01 5.05813003e-01 1.10755432e+00 -4.09388661e-01
-1.34251595e+00 -6.66978121e-01 -6.28270090e-01 2.13868201e-01
5.18721700e-01 -8.45079422e-01 -3.22735488e-01 -5.67600489e-01] | [4.048673152923584, 2.231920003890991] |
c5cca409-4b6d-48e2-845a-6e56daa0d505 | robust-monopoly-regulation | 1910.04260 | null | https://arxiv.org/abs/1910.04260v1 | https://arxiv.org/pdf/1910.04260v1.pdf | Robust Monopoly Regulation | We study the regulation of a monopolistic firm using a robust-design approach. We solve for the policy that minimizes the regulator's worst-case regret, where the regret is the difference between his complete-information payoff minus his realized payoff. When the regulator's payoff is consumers' surplus, it is optimal to impose a price cap. The optimal cap balances the benefit from more surplus for consumers and the loss from underproduction. When his payoff is consumers' surplus plus the firm's profit, he offers a piece-rate subsidy in order to mitigate underproduction, but caps the total subsidy so as not to incentivize severe overproduction. | ['Eran Shmaya', 'Yingni Guo'] | 2019-10-09 | null | null | null | null | ['robust-design'] | ['miscellaneous'] | [ 1.04174607e-01 8.86084735e-01 -5.65902472e-01 1.14517421e-01
-8.48719239e-01 -7.34413743e-01 -2.85727262e-01 1.69635177e-01
-4.70359355e-01 7.90824354e-01 4.21688765e-01 -6.18908405e-01
-4.82509732e-01 -6.58353686e-01 -3.59510869e-01 -8.92816484e-01
8.33097771e-02 -6.13186276e-03 -5.28929830e-01 1.12640135e-01
2.84577996e-01 2.07180485e-01 -9.00560498e-01 -3.88532504e-02
8.51691604e-01 1.21107209e+00 2.14257464e-01 2.94726759e-01
6.04175180e-02 9.00081158e-01 -5.48055530e-01 -4.19085115e-01
1.21494222e+00 -3.37799728e-01 -2.53301293e-01 3.50672603e-01
-5.21315336e-01 -8.50255191e-01 -2.51832426e-01 1.27037919e+00
2.82359064e-01 -7.44565800e-02 2.40735725e-01 -1.23641455e+00
-4.72093403e-01 9.98966873e-01 -8.92324448e-01 1.31736621e-01
7.21657276e-02 3.78077596e-01 1.54053020e+00 2.11752996e-01
5.05229592e-01 1.22821331e+00 -1.14713848e-01 5.31262457e-01
-1.38845170e+00 -3.82275343e-01 3.86673540e-01 -8.98657858e-01
-8.11836004e-01 -3.60745370e-01 5.16875088e-01 -3.38272750e-01
5.68622649e-01 2.81777710e-01 6.54623508e-01 2.08026469e-01
5.03455222e-01 6.28693998e-01 7.91189849e-01 -2.57584304e-01
5.73229849e-01 1.87000334e-01 -5.39840817e-01 -4.14288193e-01
8.92676651e-01 3.46550822e-01 3.82224955e-02 -2.41858363e-01
8.44678402e-01 4.28124107e-02 -4.93119180e-01 -3.98268789e-01
-6.96331501e-01 6.12946033e-01 7.55112022e-02 -1.35614529e-01
-9.33995843e-01 5.84561646e-01 2.05495074e-01 9.50396597e-01
2.57732332e-01 6.63687646e-01 -5.63666523e-01 -1.42948747e-01
-2.46226817e-01 2.42532700e-01 7.67014921e-01 1.05241871e+00
3.52612048e-01 -6.59369752e-02 -2.77274191e-01 3.27407181e-01
1.08108021e-01 5.24189472e-01 1.51035506e-02 -1.70018232e+00
1.05263209e+00 4.48709428e-01 1.13679039e+00 -5.81131339e-01
2.48835087e-02 -5.95084012e-01 -1.09419122e-01 4.02422249e-01
5.53409755e-01 -9.42265689e-01 -2.47802824e-01 1.73092175e+00
-3.64089340e-01 -8.46974909e-01 2.22197890e-01 1.04112351e+00
-3.07267368e-01 8.05587769e-01 -2.02844277e-01 -1.11286402e+00
8.84374261e-01 -5.78599453e-01 -7.46627450e-01 -2.77221411e-01
8.42908859e-01 -4.74727869e-01 4.85136926e-01 3.12122971e-01
-1.60115862e+00 4.75517839e-01 -5.63807070e-01 5.81423163e-01
3.72640729e-01 -2.17788771e-01 3.26504320e-01 4.14606810e-01
-5.58804452e-01 5.72170019e-01 -3.96144569e-01 2.90811449e-01
4.17114079e-01 5.60081840e-01 9.78873596e-02 3.89027804e-01
-7.54864097e-01 9.26060975e-01 6.35688230e-02 1.80831492e-01
-7.60845959e-01 -8.01330328e-01 -5.34920514e-01 1.03371763e+00
1.20391941e+00 -5.03781378e-01 1.86853695e+00 -1.18173814e+00
-1.34825504e+00 4.04328614e-01 4.51301903e-01 -5.05501270e-01
8.25490892e-01 2.89844990e-01 2.77479410e-01 -2.07919478e-01
2.69060403e-01 1.63604856e-01 7.11410046e-02 -1.22122550e+00
-8.96786571e-01 -4.14341122e-01 5.48499703e-01 6.15391254e-01
-2.40417309e-02 -1.34112000e-01 4.91388083e-01 -3.73736680e-01
6.71636909e-02 -1.13046181e+00 -7.68785954e-01 -3.23377609e-01
-4.30384874e-01 1.52016208e-02 1.26911864e-01 -2.78074622e-01
8.15962374e-01 -2.14790297e+00 -4.73788261e-01 1.57763422e-01
2.63993293e-02 -2.61683643e-01 -2.87677884e-01 5.70163667e-01
4.92759570e-02 3.56930017e-01 -1.97506789e-02 2.98096854e-02
4.57210213e-01 1.83961228e-01 3.06524318e-02 4.50145215e-01
-1.82816282e-01 6.27127767e-01 -5.99110007e-01 6.47212327e-01
-1.84966549e-01 -7.38128304e-01 -4.71262753e-01 1.89983174e-02
-1.99649438e-01 6.38357876e-03 -7.88088202e-01 3.98953676e-01
8.59359980e-01 1.54270023e-01 4.55692858e-01 8.77926767e-01
-5.32237887e-01 5.11660159e-01 -1.30703938e+00 9.31204736e-01
-6.23474419e-01 3.02157074e-01 1.03801680e+00 -8.52133155e-01
4.19943720e-01 2.76790440e-01 6.17574215e-01 -4.59212989e-01
5.99715948e-01 2.23351210e-01 2.90156037e-01 -4.54196692e-01
2.65730411e-01 -4.81734842e-01 -1.99249610e-01 7.70264447e-01
-7.45921850e-01 3.57176620e-03 -7.27722496e-02 5.16050279e-01
1.08061230e+00 -2.87208021e-01 5.62788785e-01 -5.56387901e-01
-1.74650699e-01 -1.19319975e-01 1.16735172e+00 6.77210450e-01
-3.50397944e-01 -1.04916155e-01 1.52301276e+00 2.39005148e-01
-9.35440838e-01 -8.27734649e-01 4.68921155e-01 7.64811695e-01
7.59744272e-02 3.68893296e-01 -5.61310887e-01 -4.88018811e-01
1.03468418e+00 7.71631479e-01 -5.25580347e-01 3.86013314e-02
2.17492715e-01 -3.17677736e-01 -5.73787332e-01 4.62426245e-01
4.24105018e-01 -5.59540451e-01 -6.34442747e-01 1.50998637e-01
1.65075168e-01 -5.84426105e-01 -1.09900248e+00 1.77182451e-01
-5.60051739e-01 -7.57575512e-01 -6.73413754e-01 1.12609547e-02
8.69427621e-01 8.07142377e-01 4.24108565e-01 -3.76348674e-01
-3.76006514e-02 2.38308206e-01 7.89714977e-02 -5.65149844e-01
-2.28677318e-01 -3.19970876e-01 -1.42511591e-01 -1.53595895e-01
-4.93604923e-03 -1.32064939e-01 -6.49915695e-01 1.39145494e-01
-5.95248282e-01 -5.05079389e-01 3.92470300e-01 5.58638453e-01
2.13957429e-01 4.52020437e-01 1.26056468e+00 -8.39328885e-01
9.35875595e-01 -6.61903024e-01 -1.50759721e+00 1.89194962e-01
-7.12177336e-01 -5.72138838e-02 4.86549199e-01 -1.79829866e-01
-1.66651917e+00 1.44588009e-01 5.44827580e-01 -2.28388995e-01
9.98356864e-02 3.66866529e-01 -7.33865082e-01 7.80794770e-02
1.75959900e-01 -2.72189111e-01 2.78291926e-02 -5.94629228e-01
3.84297550e-01 7.50164568e-01 4.32537019e-01 -4.27315116e-01
5.67027509e-01 -6.60327598e-02 1.32539168e-01 -9.06650573e-02
-7.06284046e-01 -4.08892244e-01 3.48527431e-02 8.72390866e-02
3.37297648e-01 -1.25814080e+00 -1.67453897e+00 -2.66057998e-01
-8.76658976e-01 -5.94983809e-02 -1.04990292e+00 8.24124396e-01
-8.58637989e-01 1.59413904e-01 -1.20662659e-01 -1.55578983e+00
-3.09829600e-02 -7.18467832e-01 8.06514397e-02 4.86674130e-01
3.49774629e-01 -4.40505981e-01 -5.33261657e-01 4.95865762e-01
4.67279196e-01 2.51325786e-01 3.09637070e-01 -3.57071251e-01
-1.21971583e+00 -2.04105496e-01 -1.46073624e-01 4.85551119e-01
4.80738759e-01 -2.69973606e-01 -3.14626366e-01 -5.70678234e-01
8.10234249e-02 -8.22256431e-02 3.98809880e-01 6.48205876e-01
6.55155838e-01 -9.80489612e-01 -3.13487738e-01 1.78443238e-01
1.59729695e+00 7.43385792e-01 2.58874774e-01 3.85475457e-01
-1.07199013e-01 6.91721618e-01 8.24723303e-01 9.38461721e-01
1.90528393e-01 2.67977476e-01 6.49450123e-01 4.82560664e-01
6.78346634e-01 -4.59550142e-01 1.77046478e-01 -1.49391174e-01
8.74639899e-02 -5.26339233e-01 -2.48605698e-01 7.18091071e-01
-2.07421923e+00 -1.12944007e+00 1.43849239e-01 2.62375879e+00
4.10138816e-01 3.44743401e-01 3.64624739e-01 -4.69417311e-02
7.40286410e-01 -1.11830585e-01 -9.30433214e-01 -1.22945046e+00
3.68403979e-02 -7.33703494e-01 1.43963385e+00 3.13466191e-01
-1.97243690e-01 4.69783574e-01 6.47686243e+00 5.72093606e-01
-4.75163400e-01 -9.46199074e-02 8.68747056e-01 -9.09538031e-01
-7.96306133e-01 6.68620586e-01 -5.85294962e-01 6.74786747e-01
7.56820202e-01 -1.24572694e+00 6.99926019e-01 1.02792406e+00
9.89114523e-01 -6.96529150e-01 -9.89995837e-01 2.32480407e-01
-6.66050434e-01 -1.18996990e+00 -8.27399373e-01 7.84461439e-01
1.01138353e+00 -4.40368891e-01 -2.29085889e-02 6.40621558e-02
9.01463330e-01 -4.11473721e-01 7.67276347e-01 1.39895812e-01
4.26835328e-01 -1.24671078e+00 7.23922193e-01 9.03288543e-01
-9.21739936e-01 -9.39251065e-01 -3.54780704e-01 -8.70256186e-01
6.62334979e-01 8.32397282e-01 -7.43395627e-01 2.29682565e-01
8.32591057e-02 -1.87320769e-01 7.41895378e-01 1.13883317e+00
-4.90837485e-01 5.08001633e-02 -1.72752962e-01 8.75180885e-02
3.12658876e-01 -8.66518021e-01 4.59070444e-01 5.60203731e-01
2.67618388e-01 7.44129956e-01 2.30010018e-01 8.77440572e-01
-5.81822217e-01 1.78975567e-01 -7.49164760e-01 -3.55526924e-01
8.11040223e-01 9.99178946e-01 -4.43093091e-01 -2.47638062e-01
-3.24933112e-01 5.24890482e-01 -1.72660992e-01 2.39619449e-01
-2.01790109e-01 -5.71412861e-01 8.09646070e-01 1.54210553e-01
2.78097838e-01 2.06976116e-01 -8.12061846e-01 -6.37460172e-01
2.30179667e-01 -2.15547353e-01 3.81671786e-01 -7.70599365e-01
-9.26833510e-01 -4.25620466e-01 -2.13571116e-01 -1.04925394e+00
-3.88016164e-01 -7.02529922e-02 -8.39268386e-01 1.12243998e+00
-1.35056019e+00 -1.34176955e-01 6.73919618e-01 -3.10786009e-01
2.08026782e-01 2.64970452e-01 4.95451391e-02 -1.15785472e-01
-6.04062974e-01 3.64667684e-01 6.18545830e-01 -3.71333718e-01
-4.53832559e-02 -1.22277939e+00 -2.16524750e-01 7.28761375e-01
-9.36175644e-01 6.03892744e-01 7.27406442e-01 -5.25227666e-01
-1.53692365e+00 -6.42606258e-01 8.76845300e-01 2.99071997e-01
8.40254843e-01 2.62190315e-05 -4.19947118e-01 7.64218569e-01
3.43663543e-01 -3.65388483e-01 3.19178075e-01 -2.37985954e-01
-1.44669199e-02 -5.40403843e-01 -1.76976573e+00 4.24985558e-01
9.60003793e-01 -7.54371956e-02 -3.93989086e-01 3.12586069e-01
8.01117063e-01 2.89948583e-02 -9.03845251e-01 -2.02461928e-02
3.94998878e-01 -4.01746571e-01 2.20703512e-01 -4.03067529e-01
1.56204730e-01 -6.91230148e-02 -1.22815669e-02 -1.71264982e+00
-7.12837160e-01 -1.31858695e+00 7.02825785e-01 1.09134185e+00
6.10265613e-01 -8.70306730e-01 7.98858583e-01 1.21495903e+00
-2.97575397e-03 -3.75794709e-01 -1.16730404e+00 -1.20248938e+00
2.63273686e-01 3.32259953e-01 6.66153908e-01 5.72296262e-01
8.94114971e-01 1.01133771e-01 -3.88618022e-01 9.91857573e-02
7.44589627e-01 3.02773356e-01 4.78754163e-01 -7.53898203e-01
-2.53094912e-01 -6.20654523e-01 3.66178185e-01 -1.16468239e+00
-4.09371732e-03 -3.86161178e-01 8.54080990e-02 -1.59257865e+00
3.41833174e-01 -2.38342255e-01 -2.08042935e-01 3.17066729e-01
4.24353033e-01 -7.61395931e-01 8.87491226e-01 1.33532460e-03
-1.73460424e-01 2.87229598e-01 1.58224237e+00 -1.51512757e-01
-6.96922719e-01 4.93963629e-01 -1.69320321e+00 3.64232242e-01
8.26072454e-01 -4.32775378e-01 -4.88137871e-01 -3.55013192e-01
3.41015041e-01 9.61747110e-01 -1.98225915e-01 3.03293109e-01
1.09247908e-01 -1.20835900e+00 -2.04301015e-01 -6.44781590e-01
-5.65857394e-03 -8.52516949e-01 6.41132712e-01 7.16164768e-01
-6.79118037e-01 -1.66587949e-01 -3.86536479e-01 7.69324005e-01
3.18739474e-01 -6.14508927e-01 9.52057421e-01 -4.29474562e-01
4.63249773e-01 -3.14597376e-02 -8.74790668e-01 -1.95409894e-01
1.38588738e+00 1.29723445e-01 -5.98785698e-01 -8.90060544e-01
-6.21901155e-01 9.20161009e-01 8.16077650e-01 3.60389709e-01
-7.97641873e-02 -9.37790990e-01 -4.45903540e-01 -3.91140044e-01
-3.09298486e-01 -1.36082664e-01 4.08529013e-01 4.65310246e-01
1.39689431e-01 6.26203060e-01 5.29303700e-02 2.04143524e-01
-7.96551347e-01 9.05354977e-01 5.99620938e-01 -1.85025230e-01
-2.38419592e-01 6.42588198e-01 3.14671069e-01 5.13369918e-01
6.75573498e-02 -3.97359461e-01 4.17362869e-01 9.77719501e-02
3.90053004e-01 7.80738533e-01 -2.55953610e-01 -1.53714254e-01
-1.77594855e-01 1.89273339e-02 -8.04547369e-02 -3.02016169e-01
1.19744134e+00 -5.67097485e-01 9.85904410e-02 -5.08208126e-02
9.96255994e-01 1.99770227e-01 -1.73270941e+00 -4.05388884e-03
4.92237248e-02 -1.21600974e+00 3.01890463e-01 -1.08470845e+00
-1.40576708e+00 2.22797439e-01 -2.10001349e-01 9.26752687e-01
1.04841208e+00 -2.56652117e-01 4.34426516e-01 2.53850967e-01
5.37389219e-01 -1.45916522e+00 -3.51423711e-01 -1.83561012e-01
7.19928086e-01 -7.87327468e-01 9.52063650e-02 -5.20613015e-01
-1.04241931e+00 2.46627197e-01 2.23560080e-01 -1.89680755e-01
5.08213818e-01 1.48000792e-01 -2.80413091e-01 3.82051319e-01
-1.11084127e+00 -3.11219335e-01 -4.17293549e-01 2.26342395e-01
3.36724892e-02 7.26306558e-01 -1.22526109e+00 9.13837612e-01
2.61575989e-02 1.47632912e-01 1.34377062e+00 1.07641220e+00
-8.78284514e-01 -8.55690658e-01 -3.67241532e-01 8.58939409e-01
-7.49371052e-01 3.77725661e-01 -3.39953423e-01 3.45388323e-01
2.44921491e-01 1.28077829e+00 3.07453543e-01 5.44475317e-01
7.78599679e-01 -3.01641226e-01 1.65515393e-01 -7.32581556e-01
-1.12868138e-01 7.05256164e-01 2.70039290e-01 -2.40265682e-01
-5.83042242e-02 -6.33181274e-01 -1.28684139e+00 -4.17607874e-01
-5.81835628e-01 2.46893212e-01 6.93668783e-01 8.10338497e-01
2.61474878e-01 1.11448944e-01 1.63384700e+00 -1.53877333e-01
-1.46465552e+00 -5.25996804e-01 -1.42346096e+00 1.23758808e-01
3.95710707e-01 -2.65962750e-01 -7.96629071e-01 -5.45970261e-01] | [4.429364204406738, 3.1739656925201416] |
13978621-ce8c-43d3-b535-9e8c2115efc3 | converging-measures-and-an-emergent-model-a | 2303.13799 | null | https://arxiv.org/abs/2303.13799v1 | https://arxiv.org/pdf/2303.13799v1.pdf | Converging Measures and an Emergent Model: A Meta-Analysis of Human-Automation Trust Questionnaires | A significant challenge to measuring human-automation trust is the amount of construct proliferation, models, and questionnaires with highly variable validation. However, all agree that trust is a crucial element of technological acceptance, continued usage, fluency, and teamwork. Herein, we synthesize a consensus model for trust in human-automation interaction by performing a meta-analysis of validated and reliable trust survey instruments. To accomplish this objective, this work identifies the most frequently cited and best-validated human-automation and human-robot trust questionnaires, as well as the most well-established factors, which form the dimensions and antecedents of such trust. To reduce both confusion and construct proliferation, we provide a detailed mapping of terminology between questionnaires. Furthermore, we perform a meta-analysis of the regression models that emerged from those experiments which used multi-factorial survey instruments. Based on this meta-analysis, we demonstrate a convergent experimentally validated model of human-automation trust. This convergent model establishes an integrated framework for future research. It identifies the current boundaries of trust measurement and where further investigation is necessary. We close by discussing choosing and designing an appropriate trust survey instrument. By comparing, mapping, and analyzing well-constructed trust survey instruments, a consensus structure of trust in human-automation interaction is identified. Doing so discloses a more complete basis for measuring trust emerges that is widely applicable. It integrates the academic idea of trust with the colloquial, common-sense one. Given the increasingly recognized importance of trust, especially in human-automation interaction, this work leaves us better positioned to understand and measure it. | ['Karen M. Feigh', 'Yosef S. Razin'] | 2023-03-24 | null | null | null | null | ['common-sense-reasoning'] | ['reasoning'] | [-6.45559192e-01 3.99367251e-02 -3.81158710e-01 -3.82865608e-01
2.34727059e-02 -4.05731887e-01 1.59340709e-01 5.28444946e-01
-3.40728253e-01 3.88905525e-01 2.17274457e-01 -5.04181087e-01
-2.76306242e-01 -1.90366834e-01 -4.39357847e-01 3.38509604e-02
2.02623338e-01 -1.42481595e-01 -4.00811166e-01 -4.01822835e-01
4.45029289e-01 1.30070210e-01 -1.30442536e+00 -3.34308058e-01
1.00673759e+00 8.97023737e-01 2.32209653e-01 1.72156513e-01
7.07739711e-01 1.15425336e+00 -4.80969340e-01 -9.95622098e-01
-3.91146034e-01 -3.69989902e-01 -9.20383692e-01 -1.34299651e-01
-3.26545477e-01 -6.45149708e-01 3.12179416e-01 5.63158691e-01
3.07251811e-01 -5.09338498e-01 1.94511145e-01 -1.87937212e+00
-1.01856613e+00 6.23622596e-01 2.57808536e-01 -4.81354475e-01
1.02057517e+00 7.48692006e-02 1.09614837e+00 -8.64271641e-01
4.08940285e-01 9.06638324e-01 1.17555118e+00 -3.57387550e-02
-9.44523215e-01 -8.73782992e-01 -1.51190773e-01 -8.09746459e-02
-1.18892074e+00 -2.57000297e-01 4.52577710e-01 -6.55369997e-01
1.11902857e+00 2.16473550e-01 1.60497200e+00 1.21791637e+00
8.38811159e-01 3.80937487e-01 1.65507412e+00 -4.62640882e-01
3.52640182e-01 1.00409973e+00 4.64214504e-01 5.04000187e-01
1.11452544e+00 2.13216037e-01 -4.25243676e-01 -6.16422832e-01
5.94794869e-01 1.66625932e-01 -2.74016768e-01 -4.89940315e-01
-1.17607009e+00 7.50220895e-01 1.74879938e-01 7.77159512e-01
-5.45654118e-01 -1.17987417e-01 5.85189641e-01 7.07574844e-01
-7.40650520e-02 9.02558744e-01 -6.17975712e-01 -8.62244129e-01
-5.32797396e-01 -1.65146798e-01 1.45242035e+00 7.90821135e-01
5.05187035e-01 -1.07578710e-01 3.72208178e-01 2.12166369e-01
8.93513143e-01 3.19261998e-01 2.43190289e-01 -1.11452973e+00
-5.49032986e-01 5.66088498e-01 8.00666690e-01 -1.24841964e+00
-4.04265344e-01 -5.07903934e-01 -2.10348293e-01 3.99590850e-01
-1.38949960e-01 -1.83835462e-01 3.11300486e-01 1.52625108e+00
8.79793912e-02 -4.66449499e-01 2.11997315e-01 8.01351488e-01
2.95662522e-01 -1.87891483e-01 2.65369475e-01 -3.77062619e-01
1.34384370e+00 -9.58481431e-01 -1.09685695e+00 -7.71356076e-02
6.75641596e-01 -7.88772583e-01 1.40621138e+00 4.44277138e-01
-9.42966104e-01 -5.58696568e-01 -1.55901766e+00 2.48084128e-01
-1.06654786e-01 4.89019342e-02 1.07459950e+00 1.27094853e+00
-1.20049131e+00 4.50561136e-01 -6.55913353e-01 -8.25418591e-01
-3.06306809e-01 2.07833767e-01 -3.65561098e-01 6.05586581e-02
-1.07727492e+00 1.60497451e+00 -1.67471215e-01 2.32740790e-01
-6.28436148e-01 -1.59041882e-01 -8.40503156e-01 -2.67026693e-01
-4.31918837e-02 -1.09048736e+00 1.38697124e+00 -1.38134515e+00
-1.58608162e+00 4.02392507e-01 3.53299201e-01 -1.95815101e-01
4.71458351e-03 -4.64278936e-01 -5.70348442e-01 -8.50947350e-02
2.61935622e-01 -1.10311434e-01 3.89442831e-01 -1.61665475e+00
-4.37915891e-01 -3.23142678e-01 2.08586201e-01 4.99858782e-02
-5.78722000e-01 4.04743105e-01 5.09400427e-01 -1.36871397e-01
-8.17759708e-02 -9.69265103e-01 -9.71406028e-02 -3.32444102e-01
4.78815109e-01 7.34807849e-02 4.58921224e-01 -4.56480384e-01
1.56529999e+00 -2.22734404e+00 -4.40839738e-01 1.87423617e-01
5.06308913e-01 -3.20652843e-01 1.93951011e-01 1.20920074e+00
1.41716629e-01 2.65196681e-01 1.50394663e-01 -5.73823676e-02
4.36959028e-01 1.28971264e-01 1.27165958e-01 6.20427668e-01
-2.23880351e-01 8.49459946e-01 -1.10493219e+00 -2.34426811e-01
2.17874795e-01 5.40404797e-01 -1.49947003e-01 4.00903195e-01
7.00412691e-01 1.21555790e-01 -3.91669691e-01 6.80885017e-01
3.75842422e-01 -3.33933443e-01 4.72223550e-01 -9.46223959e-02
-4.34423864e-01 2.88849205e-01 -6.71324253e-01 1.57976019e+00
-5.00251651e-01 1.36515692e-01 3.56274903e-01 -1.54324815e-01
1.16426539e+00 8.47480297e-01 6.69375539e-01 -2.90882021e-01
4.95386928e-01 5.24356604e-01 1.32372439e-01 -4.79488224e-01
6.49056256e-01 -1.98962897e-01 -3.86725903e-01 6.85122132e-01
-1.94231165e-03 -8.04292738e-01 -5.87513626e-01 2.52005845e-01
1.10269284e+00 -2.66152639e-02 7.72700071e-01 -4.42030907e-01
1.63486138e-01 -6.79752380e-02 3.73659402e-01 4.08917397e-01
-1.01299620e+00 -2.24886611e-01 4.45302546e-01 -3.19318146e-01
-6.51725709e-01 -5.17003894e-01 -1.60490833e-02 5.35613596e-01
1.74714729e-01 -8.73774350e-01 -8.12349737e-01 -5.58676481e-01
1.53512731e-01 1.16736376e+00 -5.92185736e-01 -3.79461437e-01
5.40496111e-01 2.02867225e-01 4.58462149e-01 1.04884289e-01
3.23619843e-01 -7.87305832e-01 -1.39291120e+00 1.62151411e-01
-8.60589966e-02 -9.61466908e-01 -4.12689112e-02 1.12789519e-01
-7.69312441e-01 -1.13196087e+00 2.95103956e-02 -6.83873892e-01
2.44965658e-01 6.57749653e-01 1.25686181e+00 5.71221530e-01
4.30481195e-01 1.26135468e+00 -5.46851695e-01 -4.80748087e-01
-3.81675035e-01 -5.43963969e-01 4.59420651e-01 -6.74741268e-01
4.73287225e-01 -4.80746597e-01 -2.42757216e-01 6.11077726e-01
-5.27563214e-01 -2.79655516e-01 7.56783426e-01 9.01556253e-01
-2.15369210e-01 1.34551913e-01 7.28065670e-01 -1.75616771e-01
1.32010078e+00 -5.24949849e-01 -6.07907362e-02 2.65471965e-01
-1.51162946e+00 -6.79452658e-01 5.92337437e-02 6.10521398e-02
-9.40010548e-01 -3.24018598e-01 1.57321900e-01 4.26013693e-02
3.38700339e-02 8.56696188e-01 1.17992736e-01 -4.08544421e-01
5.73218822e-01 -4.42537338e-01 5.85506499e-01 3.40507179e-01
3.12098891e-01 9.17353749e-01 -1.80871800e-01 -8.36521685e-01
2.31979877e-01 2.58829873e-02 -7.10306823e-01 -5.46973884e-01
-3.71707231e-01 -1.04625143e-01 -4.41301912e-01 -5.50251722e-01
6.61868751e-01 -1.08400249e+00 -1.13216484e+00 1.14092998e-01
-9.05594528e-01 -3.04058701e-01 -2.53110290e-01 1.06427228e+00
-6.17412031e-01 3.88408929e-01 -7.82546103e-01 -1.38524485e+00
-3.69120926e-01 -1.41722226e+00 4.09243673e-01 4.66090590e-02
-9.97199893e-01 -8.40351522e-01 2.67648160e-01 5.45320451e-01
4.96301681e-01 7.95856938e-02 4.24109489e-01 -5.26534438e-01
-4.17677052e-02 -4.39139307e-01 2.43367210e-01 5.79606533e-01
2.49869004e-01 2.25816309e-01 -4.92911339e-01 -3.95637512e-01
8.98950577e-01 -8.86660218e-01 -3.71327639e-01 9.28138010e-03
-8.91947225e-02 -2.08457753e-01 2.78932229e-02 -3.69022697e-01
1.13346052e+00 2.12801874e-01 3.70288402e-01 5.86358607e-01
-3.74161289e-03 7.42937863e-01 1.42108381e+00 7.63661146e-01
1.07545722e+00 3.02076906e-01 3.38923931e-01 1.45168528e-01
8.27708304e-01 -2.39647135e-01 6.62803471e-01 1.41272199e+00
-8.54166225e-02 4.19031650e-01 -8.52447867e-01 2.60220587e-01
-1.80717909e+00 -5.38258076e-01 -2.67409861e-01 2.17254543e+00
4.25126672e-01 9.95773450e-02 1.90148056e-01 1.81326970e-01
3.82604659e-01 -2.50418901e-01 6.39681369e-02 -7.09652185e-01
2.79570878e-01 -2.61415523e-02 1.97944231e-02 4.42315608e-01
-3.61326039e-01 7.98084259e-01 7.41623211e+00 -3.40220839e-01
-6.73127353e-01 3.17490846e-01 3.99321288e-01 3.49848688e-01
-6.67533636e-01 4.03978437e-01 1.33339480e-01 -7.90431947e-02
1.14590323e+00 -3.81639302e-01 2.35314503e-01 1.04646993e+00
3.98946494e-01 -4.50561076e-01 -1.02989161e+00 7.14905679e-01
-5.88731989e-02 -3.79286796e-01 -7.75876522e-01 -1.05255805e-01
3.80177408e-01 -4.23012406e-01 -6.84484141e-03 2.58626431e-01
4.72865492e-01 -7.25394130e-01 1.18459034e+00 4.22202110e-01
3.26103777e-01 -5.79694748e-01 1.21261621e+00 6.90997243e-02
-7.50292718e-01 -2.92472601e-01 -1.39789924e-01 -1.05490601e+00
7.21215010e-02 5.29453218e-01 -5.77726364e-01 6.67168736e-01
8.65452647e-01 6.83007360e-01 -2.93667406e-01 4.60268527e-01
-3.20432395e-01 4.09065723e-01 1.85039103e-01 -3.77013773e-01
-7.51743987e-02 -5.43512225e-01 -1.13195688e-01 9.55340326e-01
3.49617332e-01 -3.29700783e-02 -3.20220329e-02 7.56029665e-01
6.02413058e-01 2.92152613e-01 -1.06384099e+00 -3.83183479e-01
9.17292178e-01 1.36725605e+00 -5.02249420e-01 -1.36889890e-01
-8.43145132e-01 1.08625007e+00 1.31287977e-01 2.31874377e-01
-8.78621161e-01 5.68951219e-02 9.09928918e-01 -2.74611473e-01
-2.78096288e-01 -6.16513312e-01 -8.17521691e-01 -7.25094557e-01
-1.83843430e-02 -1.25544274e+00 -1.31484866e-01 -8.48302543e-01
-1.12885869e+00 4.84323382e-01 -2.81251281e-01 -1.07425225e+00
-2.47282796e-02 -7.34592751e-02 -8.71407147e-03 6.01023614e-01
-1.01259196e+00 -1.37806761e+00 -5.10788560e-01 3.55986148e-01
-3.38643283e-01 2.53492534e-01 1.26237977e+00 -6.89136907e-02
-4.43655580e-01 4.11793590e-01 -3.87328535e-01 -6.63567364e-01
1.08764613e+00 -8.10249031e-01 1.70054838e-01 3.42220843e-01
-3.55439395e-01 1.22546864e+00 6.40782475e-01 -1.11645997e+00
-1.89838612e+00 -8.44096988e-02 1.14847481e+00 -7.77733505e-01
1.07794178e+00 -9.01156589e-02 -6.36285782e-01 7.20881104e-01
7.22550511e-01 -6.08402491e-01 1.26505446e+00 4.06028986e-01
-3.29245657e-01 -1.77338853e-01 -1.44508612e+00 5.04049063e-01
7.60772228e-01 -9.52264905e-01 -5.09159684e-01 -2.44012281e-01
8.86627972e-01 8.45178887e-02 -1.74340940e+00 1.90486088e-01
1.16242325e+00 -1.72331011e+00 3.92627537e-01 6.08479157e-02
2.13173106e-01 -2.37298310e-02 -2.86262840e-01 -1.18309891e+00
-7.86992610e-01 -8.41699898e-01 3.83150429e-01 1.11639476e+00
2.20423400e-01 -1.17892635e+00 1.03497073e-01 1.26207304e+00
-4.57019985e-01 -7.79779732e-01 -5.52183628e-01 -2.54343003e-01
-2.89508194e-01 -5.03342569e-01 6.16400421e-01 1.38300931e+00
1.09884667e+00 3.59311700e-01 -2.07834318e-01 -1.02718979e-01
2.18207151e-01 -5.71647465e-01 1.01577389e+00 -1.45674932e+00
-4.14897464e-02 -2.30706751e-01 -3.95681739e-01 -4.46349949e-01
9.88219082e-02 -1.68181047e-01 -2.28222668e-01 -1.49049306e+00
1.33554921e-01 -4.37845796e-01 -2.48350114e-01 2.84033149e-01
7.50205591e-02 -4.64302748e-01 2.67333329e-01 2.93087780e-01
-5.27885139e-01 5.71630716e-01 1.07568574e+00 5.07407069e-01
-1.29998595e-01 -2.03022659e-01 -1.40378797e+00 5.81587613e-01
8.96913946e-01 -2.04076380e-01 -5.43979287e-01 -1.82250701e-02
5.48534453e-01 4.45888668e-01 2.45687947e-01 -1.05009997e+00
2.25768507e-01 -1.64984599e-01 9.38026756e-02 6.53144345e-02
2.38987908e-01 -1.59970832e+00 7.98040450e-01 7.77591705e-01
1.37379661e-01 8.04331183e-01 4.17889021e-02 5.88475950e-02
-2.18254507e-01 -2.85400152e-01 2.83369958e-01 6.28345460e-02
-1.96435198e-01 -5.84744275e-01 -6.52962685e-01 -7.91642487e-01
1.16424537e+00 -4.89149213e-01 -4.72585522e-02 -7.88322151e-01
-1.91785350e-01 -6.77188337e-02 1.31928790e+00 5.12973249e-01
7.34460294e-01 -1.17231619e+00 -1.64121211e-01 -5.58145298e-03
1.69087708e-01 -8.95067453e-01 9.13861440e-04 1.21179891e+00
-7.88733438e-02 4.17853057e-01 -5.80311060e-01 -1.65553808e-01
-8.48572552e-01 5.67196906e-01 1.57136321e-02 2.20607758e-01
-4.44067344e-02 3.50429744e-01 -3.15789431e-01 -2.19231009e-01
8.93243104e-02 -5.06770313e-01 2.59762965e-02 -1.93510458e-01
3.79308574e-02 4.39492375e-01 -1.72190994e-01 -6.77449942e-01
-3.72751623e-01 1.29367515e-01 3.83157313e-01 -3.73301178e-01
8.95202458e-01 -4.65304315e-01 -8.30980301e-01 1.12527645e+00
4.96092677e-01 2.26880416e-01 -7.68951297e-01 6.89556479e-01
1.76813498e-01 -5.20764410e-01 -2.49526888e-01 -9.73464191e-01
-3.39807093e-01 1.83465973e-01 5.42532861e-01 4.33733881e-01
9.00813341e-01 -3.36809933e-01 2.84308016e-01 1.69782028e-01
1.06276691e+00 -1.24736989e+00 2.01690942e-01 3.15523207e-01
1.00767028e+00 -1.02517629e+00 2.27669124e-02 -4.39888388e-01
-1.16385734e+00 7.98092782e-01 8.11216056e-01 2.03346416e-01
9.70506012e-01 3.86184633e-01 2.64544189e-01 -3.46228838e-01
-5.89139342e-01 2.86916792e-01 -4.52042639e-01 7.27967262e-01
1.08912969e+00 5.63741446e-01 -9.78178680e-01 1.33645451e+00
-4.20384794e-01 3.39582801e-01 6.45742536e-01 1.24666452e+00
-3.65511626e-01 -1.21202481e+00 -4.63327676e-01 1.67929634e-01
-1.90045103e-01 2.61828154e-01 -7.59093821e-01 9.89127219e-01
-2.29756519e-01 1.77395904e+00 -5.48123956e-01 -1.22136331e+00
4.93974119e-01 -9.59793851e-02 4.38680053e-01 -1.55987829e-01
-1.43749845e+00 -2.02600941e-01 4.12309676e-01 -6.56569302e-01
-4.78146285e-01 -6.40626252e-01 -8.87559533e-01 -5.33933461e-01
-6.30744815e-01 7.87032902e-01 9.17121708e-01 6.53827429e-01
4.92694408e-01 2.01233298e-01 7.09289432e-01 -3.72985393e-01
-6.80796206e-01 -1.11219764e+00 -7.30782211e-01 1.61915168e-01
7.22445920e-02 -6.68591559e-01 -3.00549626e-01 -5.75495780e-01] | [9.062834739685059, 6.284859657287598] |
1869b4ed-4261-4688-acff-e02e87a63268 | raft-a-real-world-few-shot-text | 2109.14076 | null | https://arxiv.org/abs/2109.14076v3 | https://arxiv.org/pdf/2109.14076v3.pdf | RAFT: A Real-World Few-Shot Text Classification Benchmark | Large pre-trained language models have shown promise for few-shot learning, completing text-based tasks given only a few task-specific examples. Will models soon solve classification tasks that have so far been reserved for human research assistants? Existing benchmarks are not designed to measure progress in applied settings, and so don't directly answer this question. The RAFT benchmark (Real-world Annotated Few-shot Tasks) focuses on naturally occurring tasks and uses an evaluation setup that mirrors deployment. Baseline evaluations on RAFT reveal areas current techniques struggle with: reasoning over long texts and tasks with many classes. Human baselines show that some classification tasks are difficult for non-expert humans, reflecting that real-world value sometimes depends on domain expertise. Yet even non-expert human baseline F1 scores exceed GPT-3 by an average of 0.11. The RAFT datasets and leaderboard will track which model improvements translate into real-world benefits at https://raft.elicit.org . | ['Andreas Stuhlmüller', 'Michael Noetel', 'Alexis Carlier', 'Paul Sedille', 'Carolyn Ashurst', 'Emmie Hine', 'C. Jess Riedel', 'Pegah Maham', 'Abhishek Thakur', 'Lewis Tunstall', 'Eli Lifland', 'Neel Alex'] | 2021-09-28 | null | null | null | null | ['few-shot-text-classification'] | ['natural-language-processing'] | [ 2.44615808e-01 2.34890953e-01 -2.36276105e-01 -3.60058695e-01
-1.00309706e+00 -5.07466555e-01 8.32999766e-01 2.66987700e-02
-8.25961411e-01 7.90506780e-01 3.58039349e-01 -4.17351186e-01
-2.02315822e-02 -4.57491070e-01 -2.82971710e-01 -1.60640135e-01
9.03541222e-02 1.07267082e+00 4.72226202e-01 -6.94771588e-01
2.19409510e-01 -3.18609446e-01 -1.48208511e+00 7.18043685e-01
6.59172833e-01 3.03619206e-01 2.06078961e-01 9.66289937e-01
-3.27298582e-01 1.09209037e+00 -7.62120485e-01 -4.80330706e-01
2.35983551e-01 -1.10631518e-01 -1.12692881e+00 -4.11260277e-01
5.47372818e-01 -3.68111521e-01 -9.41990688e-02 7.09764302e-01
7.73731530e-01 5.79210758e-01 6.29145741e-01 -1.33747375e+00
-8.01452041e-01 1.02251911e+00 -3.28880340e-01 4.64260131e-01
5.37855804e-01 6.93618238e-01 1.27009761e+00 -8.82602751e-01
8.54425967e-01 1.19236732e+00 8.74894619e-01 9.74531054e-01
-1.24487448e+00 -5.74974835e-01 5.67917898e-02 3.98947269e-01
-8.25656891e-01 -6.20905995e-01 -1.17625678e-02 -5.44098258e-01
1.74112928e+00 3.09249073e-01 2.02756375e-01 1.66924202e+00
1.08832456e-01 8.98713529e-01 9.20677364e-01 -4.20810461e-01
2.48783812e-01 1.23033360e-01 7.11726069e-01 3.56828481e-01
2.96201289e-01 -2.86336005e-01 -5.58487117e-01 -2.20445722e-01
1.08247161e-01 1.44852072e-01 -2.87235618e-01 -1.61694154e-01
-1.61222756e+00 7.19027102e-01 -4.23970297e-02 5.90015769e-01
-2.71533251e-01 1.94765292e-02 7.70153761e-01 4.90006417e-01
5.32625437e-01 1.25296128e+00 -7.19845116e-01 -7.78628826e-01
-9.33121383e-01 5.14662862e-01 1.15111423e+00 1.20349479e+00
5.12081861e-01 -2.34105602e-01 -7.15045631e-01 9.98799205e-01
-5.27026236e-01 2.62718439e-01 7.53408253e-01 -1.13102877e+00
5.14005005e-01 3.47392499e-01 4.83176529e-01 -2.20827252e-01
-5.04103959e-01 -1.94051519e-01 -3.07023615e-01 1.87088683e-01
8.12608302e-01 -4.74309772e-01 -8.70831192e-01 1.41604745e+00
-2.28107318e-01 3.62242460e-02 2.38540292e-01 7.04139113e-01
8.15826118e-01 7.00331867e-01 4.55841184e-01 3.19484575e-03
1.34456825e+00 -1.23079908e+00 -6.81180894e-01 -9.27090526e-01
1.06137431e+00 -6.69433892e-01 1.83255291e+00 4.16715443e-01
-1.17888033e+00 -4.26328123e-01 -8.73430610e-01 -2.57456928e-01
-5.27008712e-01 -2.90386498e-01 6.20705962e-01 4.76921201e-01
-8.87106001e-01 6.73489690e-01 -6.20516896e-01 -8.16260934e-01
4.52959687e-01 -9.94246751e-02 -1.95667386e-01 -3.55426162e-01
-1.41210997e+00 1.44469619e+00 3.39781910e-01 -5.37953317e-01
-1.05354452e+00 -1.13456595e+00 -6.12809062e-01 2.72388786e-01
7.42047608e-01 -6.10436320e-01 2.14247298e+00 -3.41820717e-01
-9.95622754e-01 1.06251252e+00 -5.07229753e-02 -6.91142142e-01
8.06612909e-01 -3.72071683e-01 -1.92158073e-01 -2.27684289e-01
3.82587940e-01 6.69670820e-01 4.49206382e-01 -7.19075084e-01
-8.79925013e-01 8.06852151e-03 4.89026040e-01 1.18558213e-01
-4.97243106e-01 2.71055043e-01 5.12948483e-02 -8.84597152e-02
-7.04978824e-01 -6.13391042e-01 -3.08731943e-01 -1.23155927e-02
1.71999544e-01 -7.09573150e-01 6.19951069e-01 -5.54385483e-01
1.00103080e+00 -1.70469820e+00 -2.95352757e-01 -7.80915797e-01
1.80899993e-01 6.89534962e-01 -4.11200613e-01 6.60509944e-01
1.30232394e-01 2.85652369e-01 -8.28175321e-02 -1.36603042e-01
2.51685351e-01 8.34079757e-02 -4.54207778e-01 -8.89676586e-02
2.29789823e-01 1.12756979e+00 -1.37018657e+00 -4.13695067e-01
1.28642678e-01 1.13804284e-02 -3.24602991e-01 8.51133242e-02
-5.38342834e-01 -2.22809598e-01 -2.16505066e-01 4.16816950e-01
2.18986094e-01 -3.86202157e-01 -1.77741393e-01 4.45396125e-01
-1.23260624e-03 2.42357075e-01 -6.61774576e-01 2.06157446e+00
-5.01003146e-01 7.94560194e-01 -2.94106275e-01 -8.83868515e-01
6.86969042e-01 3.75869066e-01 7.62132108e-02 -6.67858958e-01
8.70686257e-04 1.08494960e-01 4.08422142e-01 -8.13900769e-01
5.77785075e-01 -7.35908449e-02 -5.36831445e-04 7.32966721e-01
3.54674846e-01 -2.04857022e-01 6.90406442e-01 4.29863065e-01
1.71332514e+00 -4.62322906e-02 3.65454793e-01 -3.72638494e-01
-9.43137184e-02 6.38382316e-01 4.12135571e-01 1.21341538e+00
-5.82498252e-01 6.20255589e-01 5.48450232e-01 -7.64357448e-01
-1.35380733e+00 -7.00841665e-01 2.74158679e-02 1.91310716e+00
-2.04887465e-01 -5.25945663e-01 -8.00732374e-01 -7.45763659e-01
-1.76533997e-01 1.45949757e+00 -6.04733765e-01 -2.18165860e-01
-1.85014561e-01 -6.02006733e-01 5.38236380e-01 7.11602569e-01
2.93297648e-01 -1.18465948e+00 -9.75068450e-01 3.38286221e-01
-2.47107744e-01 -1.27476752e+00 -4.06774193e-01 3.06105077e-01
-5.74131012e-01 -1.09474444e+00 -1.05067432e+00 -6.16337478e-01
1.84176579e-01 6.46124482e-01 1.51475179e+00 2.45502014e-02
-6.14270568e-01 4.54169363e-01 -5.03371298e-01 -8.87858748e-01
-3.35940391e-01 3.11085254e-01 -1.06286015e-02 -8.51246119e-01
1.00554430e+00 -3.87505233e-01 -3.57663125e-01 4.74209547e-01
-2.87257850e-01 9.07670557e-02 5.33842683e-01 1.09324503e+00
-3.52344602e-01 -4.02660310e-01 5.97718954e-01 -1.34379864e+00
1.10241628e+00 -4.59188581e-01 -6.95987642e-02 6.93728328e-01
-7.20814586e-01 -1.65648013e-01 4.78786916e-01 -7.75536239e-01
-1.33513916e+00 -4.26893920e-01 2.86455631e-01 -3.09230208e-01
-2.42044777e-01 2.74051428e-01 4.41373289e-01 5.85343480e-01
1.66525769e+00 8.00984874e-02 -2.76305348e-01 -2.19962612e-01
2.67062932e-01 7.38883078e-01 2.82670438e-01 -9.01625991e-01
6.36009872e-01 5.53629212e-02 -7.38286197e-01 -8.12186182e-01
-1.20004070e+00 -6.60424232e-01 -4.26459134e-01 5.91651462e-02
8.01806390e-01 -1.08729374e+00 -5.21717668e-01 6.63098395e-02
-1.33725822e+00 -1.08552587e+00 -5.26742578e-01 3.05489838e-01
-4.73043263e-01 -7.91108832e-02 -6.60682321e-01 -9.68804896e-01
-7.27326334e-01 -8.47602069e-01 1.00093067e+00 1.94911256e-01
-9.04531777e-01 -1.00153768e+00 1.50899246e-01 5.72638690e-01
7.21807659e-01 -2.59865522e-01 8.26174736e-01 -1.20170856e+00
-7.13098273e-02 -3.56640637e-01 -2.12206319e-01 2.08112270e-01
-4.42873061e-01 -1.96577534e-01 -1.26352203e+00 -2.44944185e-01
-2.16481760e-01 -9.33518350e-01 7.39324868e-01 1.25239894e-01
7.64679551e-01 4.35418673e-02 -3.89957219e-01 -9.88644734e-03
9.34582412e-01 1.63728967e-01 4.07917380e-01 4.05149996e-01
2.83021778e-01 6.30120099e-01 8.61056507e-01 3.31678748e-01
1.13899708e-01 2.38506883e-01 -2.14540169e-01 4.06745255e-01
-1.03003278e-01 -2.12824956e-01 3.75769466e-01 4.96309906e-01
-2.89220750e-01 -2.65081286e-01 -1.73808765e+00 7.70922244e-01
-2.02368402e+00 -1.17448688e+00 -5.22981510e-02 1.81955552e+00
8.85386467e-01 7.43762553e-01 -1.16763726e-01 -2.36076400e-01
6.68893516e-01 1.82210699e-01 -7.40007102e-01 -5.06398201e-01
2.52785683e-01 3.10046345e-01 1.33773059e-01 3.12024325e-01
-7.55525351e-01 1.12880099e+00 6.99990416e+00 8.65840673e-01
-6.96583867e-01 4.40948218e-01 4.80319619e-01 -4.13223416e-01
-1.73558816e-01 1.90529823e-01 -8.90103459e-01 1.72143355e-01
1.18479002e+00 -9.20896947e-01 3.79569978e-01 1.24350142e+00
-1.22163348e-01 -4.93815616e-02 -1.32550347e+00 7.60348976e-01
1.22151010e-01 -1.11241996e+00 -2.00464413e-01 -3.53909135e-02
8.46571207e-01 4.02226716e-01 -1.73673749e-01 1.43330312e+00
8.74595761e-01 -1.11367631e+00 4.21412468e-01 5.43325186e-01
7.05550969e-01 -3.54428917e-01 7.24383891e-01 9.80009496e-01
-4.76161659e-01 -2.59453595e-01 -8.39457035e-01 -6.61017120e-01
-3.70411202e-02 2.05515966e-01 -1.37461209e+00 -1.74983576e-01
6.57358944e-01 5.91909826e-01 -7.44446874e-01 9.80561733e-01
-3.20577502e-01 6.01101398e-01 1.12093344e-01 -4.86532778e-01
3.88044149e-01 4.21197772e-01 4.35113132e-01 1.38141572e+00
1.14578106e-01 2.66547590e-01 2.05529511e-01 7.79473007e-01
-1.94251865e-01 -1.62651646e-03 -8.36070180e-01 -3.63425881e-01
6.28880918e-01 1.45406425e+00 -5.20747066e-01 -6.82204962e-01
-5.97974777e-01 7.00603127e-01 5.12942195e-01 3.35770875e-01
-5.65733671e-01 -5.65187871e-01 6.92221999e-01 2.90206105e-01
-2.64918506e-01 -2.94439234e-02 -3.30642521e-01 -1.34418178e+00
-1.15446553e-01 -1.11552358e+00 3.14277470e-01 -1.07504153e+00
-1.50968635e+00 4.79912668e-01 8.11823383e-02 -9.47944522e-01
-4.96813953e-01 -8.32255781e-01 -1.05694282e+00 6.99984431e-01
-1.17956674e+00 -9.47499454e-01 -4.71958905e-01 3.09211403e-01
1.27377546e+00 -4.21569228e-01 1.23664725e+00 -1.07698224e-01
-4.30613697e-01 4.55515653e-01 -1.17599301e-01 2.01654732e-01
1.18882573e+00 -1.42111695e+00 1.11971045e+00 6.21354580e-01
6.53209090e-02 6.72013640e-01 1.17114460e+00 -5.45413256e-01
-1.19348586e+00 -7.42437243e-01 9.19970810e-01 -1.19928193e+00
9.65295136e-01 -4.99619395e-01 -1.08029842e+00 9.94394124e-01
4.99385059e-01 2.48447079e-02 5.74684799e-01 7.60265887e-01
-7.02822685e-01 2.35479131e-01 -8.77172768e-01 7.04347551e-01
1.40801668e+00 -6.87105477e-01 -1.22221923e+00 7.81732798e-01
8.78627956e-01 -2.39907220e-01 -6.04284286e-01 6.79868907e-02
4.59326982e-01 -6.58350229e-01 8.73950720e-01 -1.17809129e+00
7.65936553e-01 3.99540454e-01 1.49577022e-01 -1.47180247e+00
-3.21997195e-01 -6.73484206e-01 5.36570102e-02 9.79345083e-01
6.95981026e-01 -3.29845905e-01 7.19910800e-01 1.26206517e+00
-1.57401010e-01 -4.58563358e-01 -4.47812825e-01 -1.00677729e+00
3.20324332e-01 -4.42766547e-01 2.16486275e-01 1.31046605e+00
5.83445549e-01 1.05309272e+00 -2.95009702e-01 -6.23980999e-01
6.25191391e-01 -2.84344137e-01 1.00571060e+00 -1.48614633e+00
-2.51331508e-01 -5.82893789e-01 -1.29688933e-01 -4.23846245e-01
2.20487565e-01 -8.56379211e-01 2.83390641e-01 -1.71977890e+00
5.34742534e-01 -1.55620158e-01 -1.20011307e-01 8.38441312e-01
-4.37092364e-01 -2.49592081e-01 2.43326426e-01 8.75434652e-02
-1.11458898e+00 1.24603480e-01 1.13450813e+00 -4.31576073e-01
4.27631428e-03 -1.44622475e-01 -8.72698367e-01 6.40087962e-01
9.73269045e-01 -3.03802192e-01 -6.63815498e-01 -7.06787288e-01
1.84274137e-01 -1.42761111e-01 8.31051245e-02 -1.36145961e+00
5.67724586e-01 -3.57966453e-01 2.01951206e-01 2.64946856e-02
3.10482532e-01 -3.41754884e-01 -3.46151233e-01 4.28426892e-01
-8.46516788e-01 -1.64959624e-01 2.18904138e-01 3.26336294e-01
1.75253242e-01 -6.73604488e-01 6.22923732e-01 -5.67275107e-01
-1.06023133e+00 1.83195233e-01 -3.22574914e-01 7.37088621e-01
1.22220302e+00 -4.95388061e-02 -9.37358975e-01 -4.31807876e-01
-6.50168717e-01 6.41996801e-01 2.13316947e-01 6.67676032e-01
2.22806752e-01 -8.36521983e-01 -1.05369079e+00 -5.15315771e-01
6.76528871e-01 -7.61347637e-02 3.13123971e-01 6.54218435e-01
-3.36290032e-01 4.31444108e-01 -3.69747519e-01 -3.31165671e-01
-1.20180142e+00 6.68450654e-01 1.17677145e-01 -5.17810345e-01
-8.20232153e-01 1.00830197e+00 -2.82472279e-02 -4.89110321e-01
3.83918613e-01 -1.65720016e-01 -6.61627427e-02 2.22808123e-01
1.10848653e+00 4.52264726e-01 1.82229485e-02 2.01837838e-01
-2.49901451e-02 -5.99204004e-02 -6.13601267e-01 -1.99408859e-01
1.49376488e+00 3.21493715e-01 4.21112210e-01 8.31470370e-01
8.23512435e-01 -7.24628031e-01 -1.22942388e+00 -5.35477161e-01
3.88753474e-01 -3.60688686e-01 -1.61682427e-01 -1.15861130e+00
-1.15494385e-01 1.48961794e+00 2.92972475e-01 1.73037216e-01
4.56467807e-01 -8.94084498e-02 7.98363805e-01 1.15183759e+00
6.17071748e-01 -1.44277239e+00 2.75068849e-01 9.02286232e-01
8.29624772e-01 -1.54314303e+00 -9.56196785e-02 3.90477479e-02
-1.00858259e+00 9.25911665e-01 1.03494227e+00 1.20293744e-01
1.69670746e-01 1.52426034e-01 1.05209313e-02 -1.05490252e-01
-1.60488355e+00 -2.59377122e-01 -9.90950912e-02 7.53939033e-01
7.15878308e-01 3.58286798e-02 -2.21804619e-01 6.98709667e-01
-2.00204879e-01 2.53407449e-01 6.02052510e-01 1.06477225e+00
-9.18183744e-01 -7.83566654e-01 -1.40438944e-01 5.95138848e-01
-2.02853113e-01 -2.37044111e-01 -4.26325858e-01 7.73923278e-01
-2.16179267e-01 9.04376745e-01 -6.27015159e-02 -2.59991556e-01
6.30781770e-01 8.14465702e-01 3.71725023e-01 -1.20351827e+00
-6.04126751e-01 -3.95643562e-01 4.42436486e-01 -4.30277646e-01
1.25052065e-01 -5.59063256e-01 -1.10873616e+00 -5.29558539e-01
-2.84508094e-02 2.48147532e-01 4.02970493e-01 7.98294723e-01
1.19628318e-01 5.02852798e-01 -2.47309580e-01 -7.82350481e-01
-1.06035829e+00 -1.51691651e+00 -2.78609812e-01 4.50996578e-01
4.58168145e-03 -4.22282875e-01 -2.97410846e-01 -1.05236530e-01] | [11.126886367797852, 8.325722694396973] |
19023167-cf76-4a29-98fe-d964af2c1157 | sfmlearner-learning-monocular-depth-ego | 1812.08370 | null | http://arxiv.org/abs/1812.08370v1 | http://arxiv.org/pdf/1812.08370v1.pdf | SfMLearner++: Learning Monocular Depth & Ego-Motion using Meaningful Geometric Constraints | Most geometric approaches to monocular Visual Odometry (VO) provide robust
pose estimates, but sparse or semi-dense depth estimates. Off late, deep
methods have shown good performance in generating dense depths and VO from
monocular images by optimizing the photometric consistency between images.
Despite being intuitive, a naive photometric loss does not ensure proper pixel
correspondences between two views, which is the key factor for accurate depth
and relative pose estimations. It is a well known fact that simply minimizing
such an error is prone to failures.
We propose a method using Epipolar constraints to make the learning more
geometrically sound. We use the Essential matrix, obtained using Nister's Five
Point Algorithm, for enforcing meaningful geometric constraints on the loss,
rather than using it as labels for training. Our method, although simplistic
but more geometrically meaningful, using lesser number of parameters, gives a
comparable performance to state-of-the-art methods which use complex losses and
large networks showing the effectiveness of using epipolar constraints. Such a
geometrically constrained learning method performs successfully even in cases
where simply minimizing the photometric error would fail. | ['Vignesh Prasad', 'Brojeshwar Bhowmick'] | 2018-12-20 | null | null | null | null | ['monocular-visual-odometry'] | ['robots'] | [-2.15658881e-02 2.26376176e-01 1.42959245e-02 -5.70749998e-01
-6.45305634e-01 -4.54712063e-01 7.65320361e-01 -2.28551418e-01
-5.72439790e-01 8.64294410e-01 1.52617171e-01 -2.37105805e-02
2.64064260e-02 -6.37857556e-01 -1.00636697e+00 -5.69909930e-01
2.47524709e-01 7.68797040e-01 2.42805481e-01 7.33388215e-02
4.79855984e-01 7.66196668e-01 -1.75299299e+00 -3.19942057e-01
5.61316013e-01 9.52515602e-01 2.22315833e-01 5.25013328e-01
9.59356651e-02 7.69340575e-01 -3.65813404e-01 -4.49094743e-01
5.81062496e-01 -2.68860519e-01 -7.55631328e-01 1.30814791e-01
1.11506081e+00 -6.13712132e-01 -3.07849050e-01 9.55648363e-01
4.44946855e-01 9.00333747e-02 6.97866380e-01 -1.05784595e+00
-3.33666682e-01 -6.45493343e-02 -4.36534524e-01 -2.14854017e-01
6.29817665e-01 1.75329596e-02 1.26954329e+00 -9.09695625e-01
9.00640547e-01 1.33692753e+00 8.90474617e-01 4.87111777e-01
-1.39982569e+00 -4.54369783e-01 2.71080304e-02 8.56602043e-02
-1.36355257e+00 -5.33917785e-01 7.21520483e-01 -5.22027671e-01
1.05523860e+00 -3.57181169e-02 6.97269738e-01 7.39164233e-01
1.53468356e-01 5.96495748e-01 9.02344346e-01 -5.01499951e-01
1.35661826e-01 2.52625763e-01 -3.40426087e-01 7.65252829e-01
3.50667655e-01 2.91496664e-01 -5.64901888e-01 -7.10514709e-02
1.04909313e+00 4.18024696e-03 -4.88775253e-01 -1.23256421e+00
-9.28721070e-01 8.80709708e-01 6.80974782e-01 -5.88435344e-02
-1.05410852e-01 4.96735811e-01 3.14971316e-03 3.12122315e-01
3.47635448e-01 5.28344512e-01 -2.95563519e-01 -5.92798106e-02
-9.37970817e-01 3.39161962e-01 7.69184113e-01 9.40533221e-01
1.16760707e+00 -1.23266332e-01 5.88276923e-01 6.92575216e-01
4.12048966e-01 2.10612819e-01 1.73404276e-01 -1.58018017e+00
3.89792770e-01 4.70084250e-01 1.83309481e-01 -1.12052011e+00
-5.26471138e-01 -1.42380431e-01 -5.19354403e-01 7.94820607e-01
7.67364681e-01 4.95109670e-02 -1.00471330e+00 1.88617635e+00
2.16085568e-01 -6.97616264e-02 -2.03882456e-01 1.14262104e+00
4.41988587e-01 2.20885530e-01 -5.28924763e-01 1.69407934e-01
7.55463362e-01 -4.57798302e-01 -2.92854577e-01 -5.30238450e-01
5.55081606e-01 -8.14919889e-01 9.20053244e-01 6.35864615e-01
-1.20215607e+00 -2.27041274e-01 -1.38798857e+00 -4.78043318e-01
-7.15609193e-02 -1.39063433e-01 6.90444887e-01 6.30770206e-01
-1.25223505e+00 7.91762769e-01 -7.93439507e-01 -3.96002740e-01
2.12698922e-01 5.60639143e-01 -6.52515233e-01 -1.32599190e-01
-8.39544773e-01 1.23070884e+00 3.82299758e-02 -2.27421001e-01
-7.11000800e-01 -5.86718202e-01 -1.35430133e+00 -1.83302760e-01
2.24815071e-01 -9.68878269e-01 1.04662251e+00 -7.45239556e-01
-1.55061269e+00 1.21175957e+00 -1.57951131e-01 -5.05057335e-01
8.62384319e-01 -4.46803838e-01 5.34639895e-01 2.35511586e-01
-1.24138094e-01 1.25423956e+00 7.40041614e-01 -1.41757309e+00
-3.29918265e-01 -4.17432219e-01 4.05911326e-01 6.30544841e-01
6.77519590e-02 -5.06351829e-01 -4.29256529e-01 -1.05743796e-01
7.65819907e-01 -9.44090247e-01 -1.87722191e-01 6.97814941e-01
-3.72854203e-01 1.83368996e-01 3.61716956e-01 -5.15600443e-01
4.41167384e-01 -1.93316114e+00 1.61993891e-01 2.58457243e-01
2.27919772e-01 -1.37033865e-01 1.90892965e-01 1.33600295e-01
1.06004722e-01 -2.02738300e-01 -2.34454766e-01 -7.87175834e-01
3.46744135e-02 3.30183834e-01 -1.11584943e-02 8.07594240e-01
7.66299963e-02 6.30071342e-01 -7.86739707e-01 -2.30082393e-01
6.46484256e-01 6.67781889e-01 -9.29533601e-01 4.86690961e-02
-1.66957438e-01 4.13373500e-01 1.84803456e-01 3.17463994e-01
5.92194557e-01 -1.97076470e-01 -4.09016535e-02 -1.61531538e-01
2.94919442e-02 6.44963503e-01 -1.55841064e+00 1.97467053e+00
-5.49372017e-01 9.29799974e-01 6.14817031e-02 -6.91667855e-01
9.59794641e-01 1.01265982e-01 4.37388122e-01 -6.37189209e-01
1.90286398e-01 4.29501772e-01 -3.51837218e-01 -9.10361931e-02
5.09452105e-01 -5.13642967e-01 3.43711168e-01 9.87431705e-02
2.76759826e-02 -9.42173064e-01 -4.16328982e-02 1.00246489e-01
7.81049371e-01 6.46267414e-01 2.58920878e-01 -1.47652745e-01
4.58693266e-01 -1.20572127e-01 5.88871181e-01 4.32312220e-01
4.61811153e-03 1.25683117e+00 3.83702576e-01 -4.89084840e-01
-1.46244013e+00 -1.10512948e+00 -3.08226764e-01 6.07409365e-02
3.57685447e-01 -9.44073424e-02 -4.66157913e-01 -2.13365018e-01
2.57598609e-01 4.05934572e-01 -3.16441357e-01 5.00338897e-02
-4.25442845e-01 -4.87700939e-01 3.27404499e-01 4.05949563e-01
4.94328052e-01 -6.16078436e-01 -6.17512167e-01 3.02016735e-02
-9.23950225e-02 -1.20610046e+00 -2.59175330e-01 2.50585765e-01
-1.00162601e+00 -1.03880596e+00 -9.32338536e-01 -6.13897145e-01
6.87406778e-01 3.02501261e-01 1.14412391e+00 -6.08396456e-02
-2.19743416e-01 2.25111514e-01 2.65038431e-01 -1.37815028e-01
-1.17862172e-01 -1.11515671e-01 7.28118271e-02 -3.73990923e-01
3.83108854e-01 -8.40336740e-01 -6.60768807e-01 3.62018377e-01
-6.28211021e-01 3.18251573e-03 3.56830031e-01 8.49326968e-01
5.62346995e-01 -5.06977081e-01 2.67969473e-04 -7.22999454e-01
-6.34230068e-03 4.02871594e-02 -9.13560748e-01 -3.62791240e-01
-6.81434929e-01 2.79117435e-01 4.20504630e-01 8.76000300e-02
-7.58338213e-01 3.99731129e-01 -2.28164703e-01 -5.02053618e-01
-1.30290493e-01 1.11665025e-01 -1.03285059e-01 -5.31780183e-01
1.04437184e+00 -9.07119736e-02 1.60388827e-01 -4.15931493e-01
3.18414778e-01 2.80117273e-01 4.63685840e-01 -3.24824393e-01
9.58751321e-01 8.98344278e-01 4.48445886e-01 -8.50049376e-01
-8.98588419e-01 -6.87929273e-01 -8.88023674e-01 9.20870528e-02
8.23088706e-01 -1.18633962e+00 -6.87416911e-01 3.60781997e-01
-1.16010356e+00 -2.21176252e-01 -1.65797085e-01 7.36914217e-01
-9.28291857e-01 6.89563930e-01 -5.81331074e-01 -7.21102238e-01
7.51464665e-02 -1.04278433e+00 1.23500896e+00 -1.79157570e-01
-2.72826999e-01 -1.13259029e+00 1.81114092e-01 3.10462803e-01
3.86276692e-02 3.12265575e-01 5.55009365e-01 1.05584800e-01
-1.06716311e+00 -1.67909920e-01 -3.57815027e-01 5.21196246e-01
3.73563655e-02 -1.96525872e-01 -1.23616517e+00 -2.79999316e-01
1.07906513e-01 -5.53383708e-01 7.54021943e-01 4.76341933e-01
6.61451399e-01 4.09826227e-02 -5.39858080e-02 9.92030084e-01
1.81476235e+00 -2.78501902e-02 8.98267031e-01 5.21008134e-01
1.06156325e+00 6.77514255e-01 3.69971931e-01 2.34483957e-01
6.38869286e-01 8.42893720e-01 7.17306495e-01 -6.89865947e-02
-9.79735777e-02 -3.12842280e-01 2.40579426e-01 4.37633038e-01
9.01874080e-02 6.26297891e-02 -8.52095664e-01 5.98803163e-01
-1.53322232e+00 -8.42221498e-01 -6.87471405e-02 2.57097387e+00
8.30158651e-01 4.17797536e-01 -4.35428284e-02 1.68901861e-01
1.85408503e-01 1.29088908e-01 -5.52702129e-01 -3.18295807e-01
-1.95719391e-01 3.42786945e-02 8.59351575e-01 1.05075538e+00
-8.88398409e-01 7.64309824e-01 6.24615908e+00 2.02555344e-01
-1.02943838e+00 -1.36743218e-01 3.38456303e-01 -2.57970750e-01
-4.61623758e-01 1.06589265e-01 -9.27795470e-01 1.52665332e-01
4.57505524e-01 1.95866227e-01 4.24981683e-01 7.99777091e-01
1.96408197e-01 -5.37892640e-01 -1.37143254e+00 1.44886923e+00
4.14836049e-01 -1.26556587e+00 -9.23848245e-03 1.67529643e-01
9.18232322e-01 2.35207811e-01 -1.92105442e-01 -3.32134038e-01
4.47798073e-01 -9.79798198e-01 8.60944808e-01 5.10666072e-01
5.92348874e-01 -6.96835756e-01 6.18003070e-01 4.78806049e-01
-7.90045321e-01 2.64198869e-01 -6.77119851e-01 -3.68184447e-01
2.25188404e-01 8.09312224e-01 -7.27505744e-01 3.86237741e-01
5.75382769e-01 8.31670702e-01 -3.14775020e-01 1.26025748e+00
-3.58788431e-01 -2.73571402e-01 -7.94192374e-01 3.00264210e-01
2.63355196e-01 -2.53145248e-01 5.17206907e-01 6.83963478e-01
2.33763576e-01 -4.05600250e-01 -1.07860588e-01 8.53634477e-01
1.90691953e-03 -2.63315309e-02 -1.07439840e+00 5.94697416e-01
3.24150503e-01 8.44719231e-01 -3.33729833e-01 -6.82433099e-02
-5.05359709e-01 9.94155467e-01 4.46159959e-01 1.10614359e-01
-3.48106652e-01 -1.30835146e-01 8.30427885e-01 3.04788291e-01
2.12727979e-01 -5.59329450e-01 -6.32705808e-01 -1.25826299e+00
2.99754947e-01 -5.95902681e-01 1.48155345e-02 -9.46193218e-01
-1.11686051e+00 2.85740912e-01 -6.20452575e-02 -1.40506768e+00
-5.31018853e-01 -9.33820009e-01 -2.09703907e-01 9.65847611e-01
-1.83670044e+00 -7.61556149e-01 -5.00159681e-01 3.94041508e-01
2.09469616e-01 2.39937872e-01 4.97767895e-01 4.56982970e-01
1.03078172e-01 4.40197825e-01 -2.42041945e-01 -2.19469801e-01
8.14222276e-01 -1.54075217e+00 4.01334733e-01 7.57262766e-01
2.59890795e-01 6.11417174e-01 9.19046462e-01 -3.68647546e-01
-1.15761089e+00 -5.81271946e-01 1.04641378e+00 -6.99843764e-01
3.22732776e-01 -4.15047497e-01 -7.34870017e-01 7.88588047e-01
-1.74794912e-01 -2.07382083e-01 3.50574553e-02 -7.82451257e-02
-2.63510257e-01 -6.60799146e-02 -1.26311064e+00 6.40595496e-01
1.23922932e+00 -7.24645317e-01 -4.04376984e-01 4.14182961e-01
5.21461666e-01 -6.63577616e-01 -5.24791121e-01 2.50531524e-01
5.85731804e-01 -1.40283000e+00 1.18325865e+00 7.75557458e-02
3.57001007e-01 -3.93414319e-01 -2.63831049e-01 -1.21653688e+00
1.05196401e-01 -5.95447659e-01 2.27624342e-01 7.71478117e-01
3.00140560e-01 -6.39617741e-01 1.30026495e+00 6.00958109e-01
-1.01067558e-01 -5.66899538e-01 -9.40728188e-01 -9.35295999e-01
1.07829161e-01 -4.78335738e-01 1.57742143e-01 9.07026649e-01
-2.40989655e-01 2.03565940e-01 -4.90030378e-01 1.26460254e-01
8.82529616e-01 -2.43489608e-01 1.17956877e+00 -1.43511593e+00
-3.34651589e-01 -3.37686509e-01 -9.75325167e-01 -1.49941576e+00
-2.45546643e-02 -6.36345744e-01 2.79446751e-01 -1.52901971e+00
-8.50354731e-02 -4.89398718e-01 3.36302429e-01 1.19466729e-01
2.35436261e-01 5.78837812e-01 1.06035091e-01 2.18691185e-01
-7.39737302e-02 4.99209255e-01 1.34590733e+00 2.37777010e-01
-1.92611307e-01 -1.38099626e-01 -2.38379151e-01 1.09078455e+00
4.95673090e-01 -3.33004177e-01 -6.02965057e-01 -6.07510090e-01
6.28055871e-01 -1.64818794e-01 5.94358623e-01 -1.25313759e+00
3.44811291e-01 1.08328499e-01 3.47097665e-01 -6.63067281e-01
9.33807671e-01 -8.89113009e-01 2.24452659e-01 3.65937054e-01
1.99293215e-02 -1.71190277e-02 -1.23001270e-01 3.62868696e-01
-2.49533877e-01 -3.77040505e-01 1.00496030e+00 -4.30615485e-01
-8.51472378e-01 2.95102984e-01 1.74481750e-01 7.76941702e-02
6.27544761e-01 -5.99848449e-01 -8.45293850e-02 -8.55230808e-01
-7.40008473e-01 2.20797323e-02 1.19356465e+00 1.23107590e-01
7.09361315e-01 -1.19831431e+00 -3.30220759e-01 2.93764979e-01
2.49175113e-02 3.21735799e-01 -2.31469914e-01 7.17595994e-01
-1.24264646e+00 4.38516319e-01 -1.80323958e-01 -9.75001812e-01
-1.09996152e+00 2.51787931e-01 8.02392840e-01 -4.67005000e-03
-8.24511886e-01 9.64384615e-01 3.71747166e-01 -7.14890838e-01
4.85946506e-01 -3.16934764e-01 1.21262744e-01 -1.14819787e-01
1.76521927e-01 2.71067321e-01 1.94684118e-01 -5.06262302e-01
-3.44874322e-01 9.76638913e-01 1.47250071e-01 -4.61910605e-01
1.28248775e+00 -4.74840254e-01 2.40660354e-01 4.45978403e-01
1.33635187e+00 1.34109929e-02 -1.77725744e+00 -2.82391399e-01
-7.22650588e-02 -7.90244758e-01 1.19922884e-01 -4.14391369e-01
-8.81303430e-01 1.19395006e+00 4.25760031e-01 -1.78930491e-01
6.03474021e-01 -1.51031420e-01 7.01251507e-01 5.49944162e-01
7.56110847e-01 -9.22209442e-01 1.14718162e-01 6.40317321e-01
7.16308057e-01 -1.40936267e+00 3.95068854e-01 -6.18372917e-01
-2.33930886e-01 8.95628750e-01 5.94756722e-01 -4.58177000e-01
5.07307589e-01 3.44932713e-02 3.46576542e-01 -2.27736190e-01
-3.00733477e-01 -2.12058783e-01 2.87654817e-01 6.21390104e-01
2.85284400e-01 -4.14344966e-01 -1.42201944e-03 -7.21748471e-01
-5.84297717e-01 -1.77713111e-01 6.05280399e-01 8.39243650e-01
-4.87178117e-01 -1.10004747e+00 -3.18330258e-01 1.23383448e-01
-3.50853980e-01 -5.51625118e-02 -3.94292563e-01 8.93287838e-01
-6.06082268e-02 5.94229519e-01 2.33929172e-01 -9.22228619e-02
3.07067662e-01 -5.53854629e-02 9.02928829e-01 -7.51158834e-01
-8.61499384e-02 -1.42134920e-01 1.64885849e-01 -7.88486362e-01
-4.83653247e-01 -7.82509863e-01 -1.18365061e+00 -4.92676288e-01
-1.22611664e-01 -4.15237159e-01 8.39435041e-01 1.13495386e+00
-4.54695150e-02 -1.64700255e-01 4.51958179e-01 -1.24562430e+00
-5.28136671e-01 -5.48647642e-01 -4.92331326e-01 7.01581478e-01
6.39514446e-01 -8.97804677e-01 -7.30655491e-01 -1.38306499e-01] | [8.572245597839355, -2.5295369625091553] |
f0ecbf7f-cc31-4aca-8325-14ba51dd82d9 | ckconv-continuous-kernel-convolution-for | 2102.02611 | null | https://arxiv.org/abs/2102.02611v3 | https://arxiv.org/pdf/2102.02611v3.pdf | CKConv: Continuous Kernel Convolution For Sequential Data | Conventional neural architectures for sequential data present important limitations. Recurrent networks suffer from exploding and vanishing gradients, small effective memory horizons, and must be trained sequentially. Convolutional networks are unable to handle sequences of unknown size and their memory horizon must be defined a priori. In this work, we show that all these problems can be solved by formulating convolutional kernels in CNNs as continuous functions. The resulting Continuous Kernel Convolution (CKConv) allows us to model arbitrarily long sequences in a parallel manner, within a single operation, and without relying on any form of recurrence. We show that Continuous Kernel Convolutional Networks (CKCNNs) obtain state-of-the-art results in multiple datasets, e.g., permuted MNIST, and, thanks to their continuous nature, are able to handle non-uniformly sampled datasets and irregularly-sampled data natively. CKCNNs match or perform better than neural ODEs designed for these purposes in a faster and simpler manner. | ['Mark Hoogendoorn', 'Jakub M. Tomczak', 'Erik J. Bekkers', 'Anna Kuzina', 'David W. Romero'] | 2021-02-04 | ckconv-continuous-kernel-convolution-for-1 | https://openreview.net/forum?id=8FhxBtXSl0 | https://openreview.net/pdf?id=8FhxBtXSl0 | iclr-2022-4 | ['sequential-image-classification'] | ['computer-vision'] | [ 8.70859176e-02 -2.20958829e-01 -2.00583830e-01 -2.96904962e-03
-1.06589265e-01 -6.94897056e-01 4.72626299e-01 -1.56481415e-01
-9.18557882e-01 6.77025557e-01 -4.58237946e-01 -5.51009059e-01
-3.10019076e-01 -6.13173664e-01 -8.43145669e-01 -6.22913718e-01
-4.81044441e-01 2.49702692e-01 4.26890880e-01 -3.39278928e-03
1.11581370e-01 8.97757769e-01 -1.52294207e+00 2.65000731e-01
5.57823360e-01 1.10125434e+00 3.31395566e-01 1.09153616e+00
4.66710106e-02 1.17718446e+00 -1.47620842e-01 -6.28112704e-02
2.53767967e-01 -3.66766185e-01 -7.93810844e-01 -2.59402305e-01
3.84229779e-01 -2.84592152e-01 -5.55539608e-01 7.98923612e-01
2.71534801e-01 2.41443455e-01 5.21696329e-01 -8.76026452e-01
-4.40680802e-01 6.10652030e-01 7.72810206e-02 3.71877134e-01
-4.47863400e-01 9.45933312e-02 6.34802818e-01 -8.75497043e-01
4.45571631e-01 7.44835079e-01 1.24217975e+00 6.66445374e-01
-1.47702181e+00 -4.05939668e-01 1.00082763e-01 4.03884649e-02
-1.18023121e+00 -4.54509199e-01 3.36073577e-01 -2.10991487e-01
1.45927811e+00 3.37527543e-01 7.92935371e-01 1.14368737e+00
2.46185645e-01 7.49159992e-01 7.37811804e-01 -2.53931373e-01
4.20233220e-01 -1.22979254e-01 2.25087181e-01 8.36563587e-01
1.51278526e-01 2.38128617e-01 -3.01551551e-01 -1.63313255e-01
1.21348512e+00 2.86422610e-01 -3.84552926e-01 -2.60492444e-01
-1.49801862e+00 9.09368277e-01 3.85370016e-01 4.15078104e-01
-2.64661670e-01 7.11747229e-01 7.24238336e-01 6.26695335e-01
2.74072707e-01 7.41048217e-01 -7.57579148e-01 -5.63799858e-01
-1.04213333e+00 3.26667935e-01 1.06790102e+00 8.19046855e-01
6.77155018e-01 4.48788911e-01 1.34980172e-01 5.97849607e-01
-4.59320068e-01 2.85018831e-01 8.83053303e-01 -8.32815111e-01
1.79929081e-02 1.80579081e-01 -9.78343468e-03 -6.60041749e-01
-8.24762046e-01 -3.78673106e-01 -1.31856000e+00 3.19241017e-01
4.66041386e-01 -3.11310202e-01 -1.13746583e+00 1.58055556e+00
-1.88521191e-01 6.36250615e-01 2.54954994e-01 5.08965492e-01
4.07652020e-01 7.01172054e-01 -2.61213809e-01 -4.41757292e-02
1.04546010e+00 -1.08632791e+00 -4.61212695e-01 -1.64809167e-01
9.34502304e-01 -5.44749320e-01 1.01398623e+00 5.39130867e-01
-1.08944798e+00 -4.08556491e-01 -1.06605315e+00 1.17695611e-02
-5.45728207e-01 4.54950333e-01 1.07788312e+00 5.49623728e-01
-1.42213368e+00 1.17801666e+00 -1.27707291e+00 -1.80021171e-02
3.51896793e-01 5.49365342e-01 -1.79120794e-01 4.33972150e-01
-1.11820579e+00 8.97849441e-01 7.46680379e-01 4.13244337e-01
-8.52644503e-01 -9.61683691e-01 -7.55902588e-01 1.98679194e-01
4.94842418e-02 -4.70907152e-01 1.49692845e+00 -1.12301934e+00
-1.74247599e+00 5.21695137e-01 6.76629245e-02 -1.11625218e+00
6.00692570e-01 -2.19042867e-01 -1.89691767e-01 6.53314069e-02
-4.73161072e-01 4.10217583e-01 9.18186486e-01 -4.20154780e-01
-4.01887000e-01 1.06966011e-01 9.46735367e-02 -2.60686845e-01
-5.51941454e-01 -1.36821449e-01 -9.26725343e-02 -7.17997134e-01
-1.01845533e-01 -1.17589784e+00 -5.89166343e-01 1.26166046e-01
-4.64145690e-02 1.60515010e-02 9.40631211e-01 -2.59955019e-01
1.23982477e+00 -2.05027461e+00 3.06327403e-01 1.55225722e-02
3.40561301e-01 6.46838486e-01 -8.20484459e-02 3.64507765e-01
-2.12435126e-01 1.57873914e-01 -2.27394834e-01 -2.64321655e-01
-9.01191607e-02 4.84038681e-01 -4.63199973e-01 4.64517027e-01
4.28984642e-01 1.11779952e+00 -9.64302957e-01 -2.08046794e-01
2.55021565e-02 5.97676873e-01 -4.44877893e-01 1.17406465e-01
-3.45199585e-01 -2.45906919e-01 -9.47684944e-02 2.20659152e-01
2.03723565e-01 -6.46277010e-01 6.07972518e-02 -1.54119695e-03
-4.18307781e-01 -7.73405731e-02 -1.02807045e+00 1.52721965e+00
-7.33725011e-01 1.15884829e+00 -8.46988857e-02 -1.21815217e+00
6.85757458e-01 3.71411771e-01 1.75504074e-01 -3.60076338e-01
3.85609567e-02 5.29940963e-01 -7.09841698e-02 -1.75791085e-01
5.08897007e-01 1.01889357e-01 2.54565239e-01 5.89135885e-01
1.91457212e-01 2.02365935e-01 3.92962158e-01 -8.06867033e-02
1.33797133e+00 -2.72322148e-01 2.06455197e-02 -3.51986170e-01
1.79618105e-01 -6.44039959e-02 2.30996877e-01 1.08856857e+00
2.40261003e-01 5.33046365e-01 7.66361833e-01 -1.03134251e+00
-1.56751382e+00 -9.33266282e-01 -1.89530969e-01 8.84623289e-01
-3.39249969e-01 -1.36908442e-01 -6.18212759e-01 -2.23106652e-01
-7.08568245e-02 3.91162395e-01 -9.67926323e-01 -3.04694623e-01
-9.32578743e-01 -8.67050886e-01 1.18185210e+00 8.14889491e-01
5.90265214e-01 -1.31275332e+00 -1.27957356e+00 5.47104239e-01
4.28409100e-01 -1.02705479e+00 -3.27428907e-01 8.98959935e-01
-1.11515915e+00 -8.98174644e-01 -8.58899713e-01 -9.23321247e-01
4.27845627e-01 -4.48136926e-02 1.01328397e+00 2.48929057e-02
-5.55513740e-01 6.29137503e-03 -1.56169683e-01 -1.14812613e-01
-3.38026315e-01 5.85749567e-01 1.10796176e-01 -1.55477151e-01
-6.00309409e-02 -8.13786864e-01 -4.28879172e-01 1.09329462e-01
-1.08565247e+00 1.05542868e-01 7.59172618e-01 1.30875766e+00
2.45047674e-01 -8.16424862e-02 4.78405863e-01 -8.27406049e-01
6.91976011e-01 -2.68229127e-01 -9.77722466e-01 3.56980562e-01
-6.93706691e-01 5.29140055e-01 1.13027465e+00 -1.17672074e+00
-5.78509271e-01 1.70870841e-01 1.05977068e-02 -7.16298699e-01
-1.42424237e-02 5.42218029e-01 6.75864458e-01 -4.28571492e-01
8.87225151e-01 5.19608021e-01 1.81907520e-01 -3.33982438e-01
3.35037470e-01 2.26399258e-01 4.92290288e-01 -5.39921999e-01
4.43268389e-01 4.20364916e-01 4.83300947e-02 -8.92350078e-01
-4.93170172e-01 -1.69327825e-01 -6.88404202e-01 4.00056057e-02
6.44691229e-01 -6.02042556e-01 -1.00614941e+00 8.73821616e-01
-1.32753670e+00 -8.88483167e-01 -4.50002611e-01 4.11755323e-01
-7.32241869e-01 2.34469660e-02 -1.18665135e+00 -7.86250532e-01
-4.83265549e-01 -8.22276652e-01 5.72216511e-01 1.63237795e-01
-1.20070450e-01 -1.19999504e+00 -5.30527756e-02 -8.11134458e-01
9.36083436e-01 1.58352211e-01 8.88087690e-01 -4.70755637e-01
-6.17055297e-01 -2.48840243e-01 -4.19168621e-01 5.46665907e-01
-4.13930379e-02 1.53834790e-01 -8.29965055e-01 -2.69499421e-01
-9.92923602e-02 -5.58978677e-01 1.09171569e+00 2.72777468e-01
1.22343481e+00 -4.23243344e-01 -1.16386198e-01 1.04518235e+00
1.41726542e+00 9.77475345e-02 6.41845405e-01 3.95860612e-01
4.36916143e-01 -1.87759399e-02 -1.44104540e-01 4.39377755e-01
-3.28687876e-01 3.43724608e-01 2.93379009e-01 -6.51008934e-02
1.13676004e-01 1.44294545e-01 2.75804937e-01 8.02961886e-01
-2.50380337e-01 6.55467957e-02 -9.35368121e-01 5.69611490e-01
-2.16619062e+00 -1.12308419e+00 -8.35326165e-02 1.94305682e+00
1.01974928e+00 2.35041901e-01 -1.05702832e-01 -4.75327335e-02
2.44126186e-01 1.62794322e-01 -7.87618279e-01 -5.40261567e-01
-2.77173877e-01 5.36804259e-01 9.43617642e-01 2.00946480e-01
-1.24340725e+00 9.69983160e-01 7.44449472e+00 8.22504044e-01
-1.58329952e+00 -4.33501601e-02 5.79152405e-01 -2.49911860e-01
3.24849412e-02 -2.06332177e-01 -8.10185313e-01 2.46517763e-01
1.43333662e+00 -7.77525529e-02 7.67669976e-01 1.05462480e+00
-1.89192608e-01 1.70926034e-01 -1.08971488e+00 1.11335850e+00
-2.80617952e-01 -2.00054789e+00 -3.32571119e-01 -1.36026233e-01
7.67726660e-01 3.04717332e-01 2.95351088e-01 3.18837225e-01
4.50957477e-01 -1.45109189e+00 7.24765301e-01 6.23971879e-01
7.41241634e-01 -9.43209112e-01 5.92345536e-01 4.01267648e-01
-1.12184834e+00 -1.09280661e-01 -6.96640849e-01 -3.20951641e-01
-7.23418221e-02 5.98467052e-01 -6.80186927e-01 1.70215324e-01
7.00090528e-01 6.41765118e-01 -4.18880403e-01 9.90278125e-01
2.42020115e-01 5.71091235e-01 -6.02578163e-01 -5.74722886e-01
7.79577553e-01 5.19469418e-02 -1.18595637e-01 1.47086918e+00
3.13569039e-01 -2.97311068e-01 -1.15292802e-01 8.94405305e-01
7.34828413e-02 -2.51785338e-01 -5.96785009e-01 -3.42233539e-01
8.67527425e-02 9.93686974e-01 -8.90428483e-01 -3.17264229e-01
-4.05175447e-01 1.38382959e+00 9.39627409e-01 6.59841776e-01
-9.39915776e-01 -7.73702383e-01 8.87864470e-01 -2.20567495e-01
6.49612725e-01 -5.95142782e-01 -2.03943625e-01 -1.35486042e+00
6.05047233e-02 -6.24987781e-01 4.39291634e-02 -4.09325838e-01
-1.00193071e+00 9.10173595e-01 -2.45811194e-01 -9.39346790e-01
-5.66608965e-01 -1.07452452e+00 -4.19552147e-01 7.15772688e-01
-1.41905713e+00 -7.87680686e-01 3.47064100e-02 5.38737237e-01
2.92085320e-01 -2.99741942e-02 1.17743099e+00 2.33842120e-01
-4.00875121e-01 6.16035521e-01 4.36714441e-01 3.97522986e-01
1.03171356e-01 -1.14629722e+00 7.90096402e-01 7.26507485e-01
1.81181118e-01 6.90571725e-01 4.75778580e-01 -2.67380118e-01
-1.47126710e+00 -1.20669854e+00 7.14754105e-01 -4.43123952e-02
9.62318540e-01 -4.25090551e-01 -9.98852432e-01 8.62760425e-01
-1.46901295e-01 5.33158362e-01 2.57302642e-01 3.03777419e-02
-6.51869357e-01 1.80791065e-01 -6.42775536e-01 7.17050135e-01
1.15600955e+00 -6.81039870e-01 -1.85414702e-01 3.62228870e-01
8.54183197e-01 -6.95727646e-01 -7.64378726e-01 3.50504845e-01
7.32066572e-01 -9.38229322e-01 1.02872932e+00 -1.02328706e+00
3.27980012e-01 2.06988938e-02 2.68718630e-01 -1.09881270e+00
-2.12114707e-01 -7.54448771e-01 -4.28721637e-01 3.43762517e-01
6.13383651e-01 -9.93210077e-01 9.50010180e-01 3.51402014e-01
-2.31276184e-01 -1.02310359e+00 -9.94620025e-01 -1.29180467e+00
1.41774863e-01 -4.97062653e-01 3.71848553e-01 9.26787972e-01
-1.03297822e-01 -4.37960541e-03 -5.88606477e-01 7.26167113e-02
3.69148254e-01 1.05317131e-01 3.07359040e-01 -1.12820256e+00
-4.00737047e-01 -8.43530357e-01 -5.67793131e-01 -1.29121411e+00
2.13990957e-01 -6.50354922e-01 1.91824492e-02 -8.46751571e-01
-1.68998018e-01 -6.20440185e-01 -1.40941605e-01 5.64338624e-01
1.88932657e-01 1.56342447e-01 -1.28170490e-01 1.71668097e-01
-3.77422512e-01 3.22877795e-01 8.36161196e-01 1.24362171e-01
-1.80152237e-01 -6.43287823e-02 4.40689027e-02 7.28803158e-01
9.49047506e-01 -2.94328809e-01 -4.57432657e-01 -4.03683126e-01
5.20377636e-01 5.19577507e-03 6.38225555e-01 -1.24330664e+00
6.31969929e-01 5.39453374e-03 4.89469647e-01 -3.90357167e-01
3.13791275e-01 -6.32089078e-01 2.76682556e-01 6.09083295e-01
-4.75807220e-01 4.69139844e-01 4.98595178e-01 4.43649530e-01
-1.34464353e-01 -5.20190358e-01 7.37753570e-01 -3.43669564e-01
-7.10680485e-01 4.50847447e-01 -6.00527465e-01 2.00607795e-02
8.53565454e-01 -1.33026749e-01 -2.00047493e-01 -1.10352844e-01
-8.24913859e-01 -7.96809793e-02 3.93706352e-01 1.38263196e-01
6.95410728e-01 -1.30524981e+00 -3.79069269e-01 3.89618993e-01
-1.41868085e-01 1.16182789e-01 2.15080187e-01 6.91791832e-01
-9.68690693e-01 6.17179394e-01 -9.94843021e-02 -4.78453338e-01
-8.99649918e-01 5.49029946e-01 7.15066671e-01 -5.68866730e-01
-8.46119165e-01 9.07135308e-01 -2.02011377e-01 -5.12009263e-01
4.02568907e-01 -8.77080619e-01 3.09074134e-01 -9.04966667e-02
8.01209807e-01 1.22674748e-01 1.54983819e-01 4.70515564e-02
-5.52866943e-02 3.66872400e-01 -2.01112345e-01 6.50532963e-03
1.37460315e+00 5.52380502e-01 -1.82947382e-01 6.55006588e-01
1.47914505e+00 -7.17487931e-01 -1.48886561e+00 -2.16766343e-01
7.76799321e-02 1.88321322e-02 -2.91665178e-02 -3.31754088e-01
-9.58190322e-01 8.31179619e-01 4.02060330e-01 3.15715462e-01
9.15347874e-01 -2.24734247e-01 8.62739086e-01 9.44360852e-01
1.59179643e-01 -8.62248480e-01 1.84631497e-01 1.10408485e+00
5.81334174e-01 -7.79528677e-01 -3.60994488e-01 -5.28560095e-02
-1.35690004e-01 1.84482908e+00 2.66837656e-01 -3.91199827e-01
8.31490397e-01 7.82928467e-01 -7.66365901e-02 -4.37269807e-02
-1.14735639e+00 -7.65695274e-02 -3.53526953e-03 4.10735846e-01
2.91422099e-01 -3.99156772e-02 1.36343881e-01 1.27732277e-01
-1.96703434e-01 1.96486965e-01 4.95365053e-01 1.13694119e+00
-4.36039507e-01 -7.43473411e-01 2.00778041e-02 5.09851933e-01
-3.13790947e-01 -4.94402975e-01 1.80066109e-01 8.55004191e-01
-3.96732777e-01 1.72112867e-01 3.16065997e-01 -4.57679391e-01
1.24560475e-01 1.31226793e-01 4.78126109e-01 -2.17889935e-01
-6.45244598e-01 -3.98360342e-01 -7.66698504e-04 -6.19704664e-01
-1.01507232e-01 -4.46824133e-01 -1.28980279e+00 -6.11189187e-01
-2.85618156e-01 -1.50403343e-02 6.03917956e-01 7.82667935e-01
3.63133967e-01 5.87919235e-01 2.99743295e-01 -9.86110568e-01
-9.06169474e-01 -7.90733516e-01 -5.45936227e-01 1.56911444e-02
7.30292439e-01 -1.46822631e-01 -5.15208542e-01 -1.61276553e-02] | [7.721799373626709, 3.4045538902282715] |
55a7d349-19de-40b5-9226-78b1f8ec15df | gait-cycle-reconstruction-and-human | 2206.13395 | null | https://arxiv.org/abs/2206.13395v1 | https://arxiv.org/pdf/2206.13395v1.pdf | Gait Cycle Reconstruction and Human Identification from Occluded Sequences | Gait-based person identification from videos captured at surveillance sites using Computer Vision-based techniques is quite challenging since these walking sequences are usually corrupted with occlusion, and a complete cycle of gait is not always available. In this work, we propose an effective neural network-based model to reconstruct the occluded frames in an input sequence before carrying out gait recognition. Specifically, we employ LSTM networks to predict an embedding for each occluded frame both from the forward and the backward directions, and next fuse the predictions from the two LSTMs by employing a network of residual blocks and convolutional layers. While the LSTMs are trained to minimize the mean-squared loss, the fusion network is trained to optimize the pixel-wise cross-entropy loss between the ground-truth and the reconstructed samples. Evaluation of our approach has been done using synthetically occluded sequences generated from the OU-ISIR LP and CASIA-B data and real-occluded sequences present in the TUM-IITKGP data. The effectiveness of the proposed reconstruction model has been verified through the Dice score and gait-based recognition accuracy using some popular gait recognition methods. Comparative study with existing occlusion handling methods in gait recognition highlights the superiority of our proposed occlusion reconstruction approach over the others. | ['Pratik Chattopadhyay', 'Jinesh Jain', 'Manav Mukesh Jain', 'Abhishek Paul'] | 2022-06-20 | null | null | null | null | ['gait-recognition', 'person-identification', 'occlusion-handling'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 1.73466146e-01 -4.04892206e-01 1.18100472e-01 -2.58917302e-01
-4.91845697e-01 2.39250407e-01 2.82639533e-01 -5.64172626e-01
-5.33933580e-01 9.75773096e-01 2.63896435e-01 2.40665823e-01
2.30436951e-01 -7.74092317e-01 -5.78225553e-01 -9.08502340e-01
-4.16217595e-01 1.77277341e-01 -1.44765377e-01 1.41030192e-01
-7.82862157e-02 2.84362316e-01 -1.67372549e+00 1.76693186e-01
6.40527248e-01 1.26464748e+00 -4.32929955e-03 8.31931114e-01
3.93193930e-01 8.56934428e-01 -6.70513511e-01 -6.18432105e-01
3.39318365e-01 -2.22901091e-01 -3.59010965e-01 3.68235260e-01
5.39371312e-01 -7.47455120e-01 -9.73034978e-01 7.09660530e-01
6.72475338e-01 2.78922617e-01 6.76204801e-01 -1.09653902e+00
-4.48660672e-01 -2.77283937e-01 -6.18786931e-01 2.75285274e-01
5.76544225e-01 5.05824864e-01 4.87278223e-01 -5.41535020e-01
7.17104733e-01 1.27303028e+00 1.13185799e+00 3.78747851e-01
-9.61847365e-01 -3.46782118e-01 -2.78086156e-01 8.10861468e-01
-1.24306214e+00 -4.89083499e-01 7.91001022e-01 -5.03590107e-01
1.15753400e+00 -5.02713136e-02 8.32119226e-01 1.44314587e+00
5.39022624e-01 7.06132889e-01 6.82735085e-01 -4.28663462e-01
-2.95295417e-02 -3.96395266e-01 4.81676199e-02 9.37379658e-01
5.03669560e-01 3.80126148e-01 -4.05951321e-01 3.50613333e-02
6.58135235e-01 4.58404981e-02 -4.20204937e-01 -2.54551023e-01
-1.05976140e+00 5.75091362e-01 2.84395188e-01 2.38102794e-01
-8.03391933e-01 3.08342367e-01 8.09227705e-01 2.37795755e-01
5.17039299e-01 -5.22403836e-01 -6.53515011e-02 -6.83290605e-03
-1.40707731e+00 2.85232693e-01 6.58104837e-01 3.47432911e-01
4.05204684e-01 4.69432741e-01 -2.73294568e-01 8.48381817e-01
6.88254058e-01 6.27018273e-01 6.40184641e-01 -8.62968624e-01
7.27171302e-01 1.98402762e-01 3.26007962e-01 -1.54351282e+00
-1.31859779e-01 -3.43243629e-01 -9.85408783e-01 1.74222931e-01
4.85139340e-01 -2.27938488e-01 -9.44132447e-01 1.60382688e+00
1.08392149e-01 3.97537470e-01 4.61615734e-02 1.04568374e+00
7.12028265e-01 5.67224681e-01 3.05721343e-01 -2.37604320e-01
1.23824334e+00 -8.14095974e-01 -9.70529377e-01 -2.17311725e-01
4.98451144e-01 -5.45264065e-01 4.47726071e-01 2.09081292e-01
-9.21553671e-01 -8.76402199e-01 -1.32627749e+00 9.89486799e-02
-1.69219121e-01 6.66732728e-01 -1.14492755e-02 8.32270265e-01
-1.06860769e+00 6.90407395e-01 -9.64953065e-01 -5.40144503e-01
4.95335042e-01 4.60005105e-01 -6.91809237e-01 -2.52242923e-01
-1.32847047e+00 9.50792551e-01 4.44050968e-01 7.12601960e-01
-6.79213226e-01 5.31901699e-03 -1.17086244e+00 1.03890486e-02
-3.28683943e-01 -8.92057300e-01 6.53591812e-01 -8.00558925e-01
-1.13145101e+00 7.53014445e-01 -2.91636676e-01 -9.28272665e-01
1.00112212e+00 -1.52824059e-01 -3.67403001e-01 1.58247918e-01
1.47065312e-01 6.45993590e-01 1.01116264e+00 -8.43758464e-01
-3.52805823e-01 -3.92677873e-01 -5.03844738e-01 2.45823264e-01
-1.82659239e-01 -2.43111074e-01 -2.59935677e-01 -8.63373458e-01
-1.33613823e-02 -6.91002190e-01 1.85269624e-01 1.48827195e-01
-2.91732728e-01 2.08381101e-01 1.29705477e+00 -1.78383088e+00
1.09222448e+00 -1.80127978e+00 -4.03605308e-03 4.03785706e-03
-8.88804123e-02 5.26088715e-01 -1.38925672e-01 1.20058350e-01
-4.73598428e-02 -4.42263365e-01 -3.60775322e-01 -8.80945385e-01
5.03417477e-03 4.16456819e-01 -7.93273523e-02 7.71423459e-01
3.58291268e-02 6.85005486e-01 -5.26199222e-01 -6.59784913e-01
7.52898216e-01 9.72462773e-01 -2.15274617e-01 2.89078325e-01
2.11102098e-01 3.82884830e-01 -3.22893523e-02 7.89122403e-01
8.49317908e-01 2.74630308e-01 1.14839129e-01 -3.88198167e-01
2.86549479e-01 -1.85775250e-01 -1.09065711e+00 1.60043192e+00
-2.71411210e-01 1.01680136e+00 -2.62400150e-01 -1.41721070e+00
9.95402813e-01 7.10780859e-01 6.19488060e-01 -7.01006591e-01
2.72200316e-01 1.40376657e-01 -1.96510628e-01 -1.04788649e+00
3.81316394e-01 1.37125656e-01 2.72548616e-01 1.78326800e-01
1.71956137e-01 7.38568425e-01 4.19476807e-01 -5.22001326e-01
1.01154995e+00 6.02129281e-01 1.08769365e-01 2.78324693e-01
8.46852839e-01 -2.16647923e-01 7.62259841e-01 4.83475149e-01
-7.85647690e-01 6.75847054e-01 1.02214552e-01 -9.59941506e-01
-1.29067028e+00 -9.68753040e-01 1.34025812e-01 3.51790935e-01
-2.39576966e-01 1.25559881e-01 -9.77565885e-01 -4.98401344e-01
-1.98587298e-01 2.90058047e-01 -6.80756211e-01 -7.72841871e-02
-9.68685508e-01 -9.02525902e-01 8.46998334e-01 5.14771283e-01
1.31310678e+00 -1.21144485e+00 -8.45892370e-01 5.04956126e-01
-9.39931989e-01 -1.32943928e+00 -5.06658375e-01 -4.54342395e-01
-8.51509631e-01 -1.12681377e+00 -1.07826948e+00 -9.17034864e-01
4.40133870e-01 4.26452234e-02 8.38198423e-01 9.12540779e-02
-6.39236927e-01 1.48368999e-01 -7.47042894e-02 2.37865597e-01
-7.76420981e-02 -3.05518836e-01 1.87903076e-01 4.29801762e-01
5.73006988e-01 -5.36588490e-01 -7.20414817e-01 2.70125538e-01
-4.95509565e-01 -8.48857164e-02 2.40900025e-01 1.16718507e+00
2.89989501e-01 8.70409757e-02 3.99805516e-01 4.12085392e-02
4.23116118e-01 -1.54997215e-01 -2.83869177e-01 3.37377548e-01
2.00665370e-01 -1.10538960e-01 4.97059733e-01 -3.32150310e-01
-1.08990324e+00 2.09268034e-01 -2.49515668e-01 -7.54858971e-01
-2.66060382e-01 2.55225509e-01 -1.97540045e-01 -8.62520337e-02
3.24702442e-01 5.23831725e-01 8.09350386e-02 -2.81416357e-01
-7.55314827e-02 6.26601279e-01 8.15603614e-01 -3.47334385e-01
4.43053514e-01 4.41703707e-01 -1.46719217e-01 -1.10441375e+00
-9.20351446e-02 -1.04784310e-01 -7.57850349e-01 -6.71957016e-01
1.15584779e+00 -8.34223568e-01 -7.83409655e-01 1.05216873e+00
-1.37817645e+00 -1.45394430e-02 -7.99906924e-02 6.19064033e-01
-7.55928397e-01 9.27299023e-01 -8.89414907e-01 -9.66709495e-01
-6.88172519e-01 -1.38199055e+00 9.12986875e-01 1.43128082e-01
-1.90903068e-01 -1.07888782e+00 8.19985867e-02 5.50702035e-01
3.33956867e-01 7.46812582e-01 5.47534108e-01 -2.04440504e-01
-2.90756464e-01 -6.64301097e-01 -3.20730716e-01 4.11176920e-01
2.68566519e-01 -1.19109573e-02 -9.84591126e-01 -4.49217826e-01
2.54036248e-01 -1.31540567e-01 1.07127154e+00 9.73040521e-01
5.71449339e-01 -4.88184810e-01 -1.74989209e-01 4.92407799e-01
1.39141798e+00 3.17232609e-01 1.08630025e+00 4.11468685e-01
7.08623290e-01 6.71764672e-01 3.60462576e-01 2.82982230e-01
2.79526830e-01 8.55963051e-01 6.16446882e-02 7.67952949e-02
-4.05699521e-01 -9.59654525e-02 6.17288828e-01 7.81171203e-01
-3.71342421e-01 -3.37119043e-01 -5.68410277e-01 4.29925561e-01
-2.02880073e+00 -1.41044843e+00 8.68190601e-02 2.30407786e+00
1.71118155e-01 -5.33005111e-02 2.10313976e-01 6.06836081e-01
1.04244399e+00 3.88573080e-01 -2.94296682e-01 -2.87141562e-01
-2.22229034e-01 -8.73417407e-02 4.31383908e-01 4.54426020e-01
-1.65088809e+00 5.67531705e-01 6.00659084e+00 6.73982739e-01
-1.16202819e+00 9.52777639e-02 7.53802061e-01 8.02260712e-02
6.92300320e-01 -5.32506049e-01 -2.84000337e-01 7.43671656e-01
8.86635721e-01 3.30971241e-01 9.29609090e-02 5.57006717e-01
7.08326459e-01 -9.29710791e-02 -7.68464148e-01 1.20261073e+00
2.37245545e-01 -1.03040123e+00 -5.05823046e-02 1.93114057e-01
3.78860354e-01 -2.70195693e-01 1.25984311e-01 -2.50090826e-02
-8.79933238e-02 -9.32977378e-01 5.51306546e-01 8.36677969e-01
7.31466413e-01 -5.67324042e-01 1.07290828e+00 -5.95367886e-02
-1.55467319e+00 -5.21514378e-02 -4.19741660e-01 -2.70104762e-02
4.48802292e-01 3.80283326e-01 -5.85820019e-01 6.66561544e-01
6.61360919e-01 1.00939095e+00 -3.39849323e-01 1.34142482e+00
7.79352933e-02 4.09659207e-01 -1.79842249e-01 3.33031476e-01
1.28686324e-01 -4.28372860e-01 7.33326852e-01 1.44947195e+00
5.82474589e-01 -2.92627513e-01 3.13130021e-02 6.55567825e-01
1.32944539e-01 -1.88496098e-01 -8.55893314e-01 2.22629219e-01
6.81141391e-03 7.54341483e-01 -5.30365944e-01 -6.06625259e-01
-3.71478587e-01 1.12268233e+00 -3.02443113e-02 4.87428665e-01
-1.02450693e+00 -2.32029438e-01 6.60627604e-01 -9.40539837e-02
3.51829588e-01 -3.10642958e-01 -2.85452694e-01 -1.35279131e+00
3.50566059e-01 -7.09132969e-01 3.99584144e-01 -6.57844126e-01
-1.18077123e+00 6.96982443e-01 -1.16721191e-01 -1.56516957e+00
-5.93475103e-01 -5.17613351e-01 -6.84908807e-01 8.66888165e-01
-9.58075225e-01 -1.28901184e+00 -3.86140883e-01 4.02941704e-01
7.98874438e-01 -3.97001922e-01 6.87653840e-01 9.34772253e-01
-1.01223111e+00 6.12130523e-01 5.20674214e-02 4.68854785e-01
4.54436570e-01 -7.59442508e-01 6.91053331e-01 1.15667975e+00
-2.51699120e-01 1.66356102e-01 9.54172254e-01 -8.60181510e-01
-1.00564659e+00 -1.15293479e+00 9.76012588e-01 2.01271579e-01
1.43053636e-01 1.82983428e-01 -9.00524855e-01 5.86919725e-01
4.06608842e-02 -4.68884744e-02 2.83112884e-01 -7.76836455e-01
-9.44090709e-02 -1.79709420e-02 -1.52403975e+00 3.67234796e-01
8.44012141e-01 -2.97882527e-01 -6.75111532e-01 1.10312402e-01
5.10329604e-02 -4.26957868e-02 -7.75681734e-01 4.38775152e-01
9.89094675e-01 -1.04783547e+00 1.34187078e+00 -2.56505340e-01
2.63724238e-01 -2.53146410e-01 -4.25579756e-01 -1.01037288e+00
-4.93692458e-02 7.30047673e-02 -2.76177973e-01 9.25350487e-01
-1.98910639e-01 -5.70622087e-01 1.19210041e+00 5.53202510e-01
1.01240419e-01 -3.16865802e-01 -1.26978099e+00 -7.25505292e-01
-4.34354246e-01 -2.61110991e-01 1.20339893e-01 6.15667284e-01
-4.07195807e-01 -2.92088240e-01 -1.13034487e+00 -7.17791077e-03
1.16032672e+00 -5.50112367e-01 5.43993413e-01 -9.89747703e-01
-1.69683158e-01 -1.42843977e-01 -1.01333940e+00 -6.17344379e-01
1.86100021e-01 -3.49104017e-01 1.28229093e-02 -1.73094070e+00
-3.88509706e-02 1.28022075e-01 -3.20279822e-02 2.82955885e-01
3.06832455e-02 4.69638616e-01 1.73672751e-01 2.90147007e-01
-2.05028400e-01 7.01163888e-01 8.95782709e-01 -5.39927661e-01
-9.58514065e-02 -2.60682795e-02 3.94174546e-01 6.64440811e-01
6.02892518e-01 -1.94682941e-01 -1.55521080e-01 -3.72061312e-01
-8.79228890e-01 3.41330349e-01 6.53438926e-01 -1.68887520e+00
1.21191561e-01 4.42677081e-01 9.54504669e-01 -6.90004051e-01
8.73330355e-01 -7.00148880e-01 4.59375352e-01 1.13418651e+00
4.25433405e-02 2.14837953e-01 9.24415290e-02 5.07645130e-01
-2.70121306e-01 -3.67757194e-02 7.67857194e-01 -6.45817593e-02
-7.15853751e-01 2.58807391e-01 -6.07312322e-01 -4.22177672e-01
8.21048677e-01 -7.91179061e-01 -1.44423917e-01 -4.18273687e-01
-9.51802671e-01 -2.12863594e-01 2.35723019e-01 2.95269758e-01
1.01026642e+00 -1.55599988e+00 -8.15296471e-01 4.63825703e-01
-3.77588153e-01 -6.11266851e-01 6.58423603e-01 7.32186198e-01
-1.07477367e+00 5.32047093e-01 -8.26137960e-01 -6.02469862e-01
-1.45930374e+00 3.60480696e-01 6.68293595e-01 -5.55613160e-01
-7.52456248e-01 3.74850094e-01 -1.03232421e-01 -1.45485356e-01
4.22777236e-01 -6.96165711e-02 -3.27308387e-01 -2.08067130e-02
5.53520441e-01 7.74773300e-01 5.82885146e-02 -1.19863582e+00
-2.71155983e-01 6.61772132e-01 1.68434784e-01 -2.06864998e-01
1.04651356e+00 -2.25099936e-01 1.61553174e-01 -5.96184656e-02
1.40856612e+00 -1.00030780e+00 -1.51909852e+00 -6.32541329e-02
-7.99639598e-02 -6.82285666e-01 -1.96903184e-01 -3.39633793e-01
-1.31803131e+00 1.06064427e+00 1.33320546e+00 -3.00753802e-01
1.12026298e+00 -1.07691216e+00 1.25117183e+00 2.74096310e-01
3.83878469e-01 -9.74754691e-01 1.27540631e-02 1.63631305e-01
7.04295576e-01 -1.13464689e+00 -2.27099136e-01 -7.05257878e-02
-3.68889451e-01 1.11522746e+00 3.92031431e-01 -3.48804533e-01
3.24556649e-01 -7.34008327e-02 -8.19548965e-02 2.89089620e-01
-4.72689658e-01 1.48010291e-02 1.60644099e-01 1.00684726e+00
2.82859445e-01 3.51749808e-02 -4.20180529e-01 2.33371761e-02
7.06880540e-02 3.87394249e-01 3.09455156e-01 7.63769865e-01
-3.04407805e-01 -8.71610165e-01 -8.29742968e-01 4.93509918e-01
-3.88840169e-01 3.59867513e-01 1.59051493e-01 6.02437556e-01
2.63351232e-01 9.16243613e-01 -9.64665134e-03 -5.25628328e-01
1.35825157e-01 2.39295840e-01 4.84428555e-01 2.67774701e-01
-2.44162261e-01 -6.26070574e-02 2.63973117e-01 -3.75073105e-01
-5.92189252e-01 -8.21110427e-01 -5.37195385e-01 -5.81755102e-01
9.80894715e-02 -2.25678831e-01 4.98272300e-01 1.07003939e+00
-2.00755917e-03 4.57576483e-01 3.39268297e-01 -1.39644814e+00
-2.37844706e-01 -1.01259780e+00 -3.77987295e-01 4.51608121e-01
4.02498662e-01 -9.62680340e-01 -1.65322602e-01 4.02867734e-01] | [14.268715858459473, 1.4436912536621094] |
2237f093-f876-4d12-af43-af34dfd6442a | diverse-trajectory-forecasting-with | 1907.04967 | null | https://arxiv.org/abs/1907.04967v2 | https://arxiv.org/pdf/1907.04967v2.pdf | Diverse Trajectory Forecasting with Determinantal Point Processes | The ability to forecast a set of likely yet diverse possible future behaviors of an agent (e.g., future trajectories of a pedestrian) is essential for safety-critical perception systems (e.g., autonomous vehicles). In particular, a set of possible future behaviors generated by the system must be diverse to account for all possible outcomes in order to take necessary safety precautions. It is not sufficient to maintain a set of the most likely future outcomes because the set may only contain perturbations of a single outcome. While generative models such as variational autoencoders (VAEs) have been shown to be a powerful tool for learning a distribution over future trajectories, randomly drawn samples from the learned implicit likelihood model may not be diverse -- the likelihood model is derived from the training data distribution and the samples will concentrate around the major mode that has most data. In this work, we propose to learn a diversity sampling function (DSF) that generates a diverse and likely set of future trajectories. The DSF maps forecasting context features to a set of latent codes which can be decoded by a generative model (e.g., VAE) into a set of diverse trajectory samples. Concretely, the process of identifying the diverse set of samples is posed as a parameter estimation of the DSF. To learn the parameters of the DSF, the diversity of the trajectory samples is evaluated by a diversity loss based on a determinantal point process (DPP). Gradient descent is performed over the DSF parameters, which in turn move the latent codes of the sample set to find an optimal diverse and likely set of trajectories. Our method is a novel application of DPPs to optimize a set of items (trajectories) in continuous space. We demonstrate the diversity of the trajectories produced by our approach on both low-dimensional 2D trajectory data and high-dimensional human motion data. | ['Ye Yuan', 'Kris Kitani'] | 2019-07-11 | null | https://openreview.net/forum?id=ryxnY3NYPS | https://openreview.net/pdf?id=ryxnY3NYPS | iclr-2020-1 | ['human-pose-forecasting'] | ['computer-vision'] | [-7.30462372e-02 -1.96758490e-02 -1.55044183e-01 -4.71696466e-01
-6.69857264e-01 -5.39520919e-01 8.66912127e-01 -2.24092737e-01
-1.11516081e-01 7.43544281e-01 3.60692799e-01 -2.24373773e-01
-2.17377558e-01 -9.92199361e-01 -9.91789222e-01 -9.83657539e-01
-1.03778057e-01 7.07422137e-01 1.25186741e-01 -2.26104304e-01
8.76336470e-02 7.67701387e-01 -1.90355992e+00 2.02984378e-01
1.16130888e+00 6.30888760e-01 5.48420429e-01 7.04152048e-01
1.92144830e-02 4.11542475e-01 -5.37848771e-01 -2.25885704e-01
1.91795185e-01 -6.13973141e-01 -2.09898323e-01 2.63596028e-01
-1.70661528e-02 -3.73578608e-01 -2.00870171e-01 1.01656723e+00
-1.19645996e-02 4.74996597e-01 1.22314334e+00 -1.54603267e+00
-4.01266396e-01 1.67780966e-01 7.94764310e-02 -4.29215804e-02
1.03333108e-01 3.56911123e-01 8.29762101e-01 -9.64160383e-01
6.61567211e-01 1.24402344e+00 2.13791877e-01 7.32631743e-01
-1.28667343e+00 -5.26584327e-01 3.10736299e-01 2.31369525e-01
-1.30739188e+00 -3.54602098e-01 8.02435875e-01 -6.68081164e-01
5.99688649e-01 2.14131951e-01 9.86883640e-01 1.36481500e+00
5.10612369e-01 8.18467438e-01 5.00595570e-01 6.03422485e-02
8.04296732e-01 1.90664083e-01 -1.65902093e-01 5.39721012e-01
5.76649271e-02 2.81354457e-01 -2.47885317e-01 -3.68027180e-01
4.11060393e-01 1.55222297e-01 -1.45797417e-01 -5.55541754e-01
-8.70021343e-01 1.01513457e+00 1.85742632e-01 -1.08470909e-01
-5.63914001e-01 9.04978812e-02 -4.60629538e-02 -2.87392586e-02
2.01538131e-01 1.75127447e-01 -9.34764221e-02 -1.15485616e-01
-6.22321784e-01 8.12412381e-01 7.51262844e-01 8.44580173e-01
9.41455007e-01 5.58366962e-02 -3.04977655e-01 5.60045183e-01
3.92766476e-01 7.46009350e-01 1.20432600e-01 -1.03675508e+00
4.79766428e-01 4.38105792e-01 4.77593750e-01 -9.08290863e-01
-9.65016261e-02 -3.15942138e-01 -7.68072605e-01 4.06468540e-01
5.36631703e-01 -3.91037583e-01 -9.45457280e-01 2.10895085e+00
4.71726149e-01 2.67554313e-01 2.04609800e-03 9.34700131e-01
2.52676439e-02 1.27588117e+00 -2.34591365e-02 -3.31802815e-01
5.88371158e-01 -4.38771158e-01 -4.09858316e-01 -1.06239296e-01
3.79619390e-01 -3.01134616e-01 8.68712902e-01 2.34076336e-01
-8.63136292e-01 -5.88394105e-01 -8.70248377e-01 4.30842966e-01
-3.74236442e-02 1.40423072e-03 1.07592456e-01 3.85339767e-01
-8.11388850e-01 4.74765092e-01 -9.16373849e-01 -2.22351193e-01
2.71667540e-01 1.76457927e-01 7.39826635e-02 3.78310420e-02
-9.53955889e-01 8.06177318e-01 4.00546849e-01 -2.98451632e-02
-1.51883006e+00 -5.84616840e-01 -7.47733176e-01 5.85643202e-02
2.71551222e-01 -7.42366672e-01 8.48761559e-01 -8.23896825e-01
-1.32529294e+00 1.25589505e-01 -4.31195557e-01 -5.46834230e-01
6.09486639e-01 6.94335962e-04 -3.82477850e-01 -9.24299732e-02
8.67180675e-02 7.88600564e-01 1.17547309e+00 -1.32888138e+00
-9.18318748e-01 -7.13386238e-02 -8.91539305e-02 1.78005144e-01
-1.89200848e-01 -4.63500321e-01 -1.69381961e-01 -3.25851679e-01
-2.04731286e-01 -1.20463371e+00 -2.48983577e-01 3.23727764e-02
-4.28207546e-01 -3.06031972e-01 6.72971249e-01 -5.24868667e-01
1.21148860e+00 -2.01607609e+00 5.15146017e-01 3.35754037e-01
5.74802458e-02 3.27835791e-02 -7.44616017e-02 5.42443335e-01
3.55137378e-01 -1.11650400e-01 -4.05649424e-01 -3.41734022e-01
-5.56462910e-04 4.65688974e-01 -7.24783182e-01 4.76639986e-01
3.50359619e-01 5.26239932e-01 -9.54093516e-01 -5.01336381e-02
3.17231506e-01 4.78655934e-01 -6.82495356e-01 4.01988924e-01
-6.60287857e-01 7.08865702e-01 -8.34705234e-01 9.67839882e-02
5.11705339e-01 6.05008332e-03 -1.60860434e-01 2.34238520e-01
-2.68585831e-01 -1.39504939e-01 -9.35984671e-01 1.13371944e+00
-3.24799240e-01 3.61940295e-01 -4.53921229e-01 -8.47315729e-01
1.14458013e+00 5.51482886e-02 5.51558673e-01 -2.42229447e-01
-2.12723687e-02 2.16971561e-01 7.17544928e-02 -6.03048742e-01
5.15746176e-01 -1.12805992e-01 -1.88720420e-01 2.91187584e-01
-2.60321766e-01 1.77243322e-01 1.18296430e-01 7.16019273e-02
7.92037904e-01 9.56297815e-02 3.32732089e-02 -4.54861857e-02
4.41271484e-01 8.52263570e-02 7.76492178e-01 7.46764481e-01
-1.69434696e-01 6.51582956e-01 5.72945654e-01 -5.08360624e-01
-1.61187208e+00 -1.35732222e+00 -1.65877610e-01 4.28384930e-01
1.47138387e-01 -1.06291585e-01 -7.52353847e-01 -6.05990469e-01
-6.46525696e-02 1.25327718e+00 -5.40513515e-01 -4.73118156e-01
-5.92283309e-01 -5.06738663e-01 7.22687393e-02 1.75359055e-01
2.52831578e-01 -9.66747463e-01 -9.18781757e-01 3.37335378e-01
-1.28554851e-01 -7.85031557e-01 -5.49169064e-01 -2.83639252e-01
-5.06534874e-01 -9.19965327e-01 -5.93372703e-01 -4.39142913e-01
7.52717614e-01 -6.12174384e-02 8.40720057e-01 -4.02395964e-01
1.20835289e-01 2.91080892e-01 -2.85530359e-01 -2.81742871e-01
-8.06462824e-01 -2.31476948e-01 2.81675100e-01 5.54674804e-01
1.59781069e-01 -4.70251799e-01 -5.85202038e-01 2.37020969e-01
-6.86944902e-01 2.03062028e-01 2.62543589e-01 8.74352634e-01
8.44224691e-01 2.36318618e-01 6.93968236e-01 -3.71294677e-01
5.67951798e-01 -1.10185921e+00 -7.16952145e-01 2.57522553e-01
-2.74368376e-01 2.47616693e-01 1.08077025e+00 -7.83023059e-01
-1.09632325e+00 7.19555989e-02 -2.84240454e-01 -7.14737475e-01
-4.22417313e-01 3.53711426e-01 -2.45632440e-01 5.18227935e-01
5.90722978e-01 4.50792819e-01 1.03035234e-01 -1.24910869e-01
4.21414107e-01 4.64891076e-01 2.72813708e-01 -6.96583748e-01
6.29052579e-01 4.79122877e-01 1.62021264e-01 -9.81868505e-01
-5.20668864e-01 -6.19929843e-02 -3.21886748e-01 -7.21805334e-01
9.63230491e-01 -5.82480609e-01 -6.26598418e-01 2.99792320e-01
-1.05524385e+00 -2.75664508e-01 -4.25317466e-01 5.36380351e-01
-8.70565891e-01 -1.00652635e-01 1.64734423e-01 -1.25484324e+00
2.97592729e-01 -1.35672116e+00 9.74305570e-01 3.83646965e-01
-3.95079702e-01 -8.67056072e-01 2.04516068e-01 -1.86725423e-01
2.37320736e-02 5.31436026e-01 1.01694119e+00 -4.97383803e-01
-9.28896427e-01 -1.27063081e-01 5.50111353e-01 2.23876014e-01
3.13278958e-02 2.80870229e-01 -5.91810048e-01 -3.51551682e-01
-1.21648759e-02 -6.19132072e-02 7.49152839e-01 8.71490717e-01
1.06320560e+00 -5.25390446e-01 -5.21415830e-01 7.01775968e-01
1.14151275e+00 7.78884113e-01 6.02756083e-01 5.05621023e-02
5.17381191e-01 7.17017233e-01 6.11721098e-01 5.27196825e-01
5.00540912e-01 7.46825933e-01 5.15836000e-01 6.09323621e-01
1.21579073e-01 -6.46228015e-01 6.28475070e-01 3.49622399e-01
1.39468938e-01 -5.01588345e-01 -7.44531929e-01 7.21276104e-01
-2.01087332e+00 -1.27015328e+00 9.33459997e-02 2.38716602e+00
3.40944976e-01 1.09224908e-01 5.37063777e-01 -1.98879287e-01
6.72349393e-01 -9.54878982e-03 -1.15294015e+00 -3.45255792e-01
2.24095047e-01 -3.79336834e-01 5.09968679e-03 6.12925053e-01
-8.85960639e-01 5.63077033e-01 5.46210146e+00 7.43372262e-01
-9.87534940e-01 -2.13630974e-01 6.34398639e-01 -1.83889076e-01
-8.09279323e-01 3.25520453e-03 -1.05784583e+00 9.75305438e-01
1.04681265e+00 -3.99438888e-01 4.30670828e-01 7.67868519e-01
4.42025930e-01 3.60767134e-02 -1.09772408e+00 7.62464523e-01
-3.08698028e-01 -1.42862332e+00 2.46531591e-01 2.47172728e-01
9.26779270e-01 -9.28577483e-02 4.49225515e-01 2.71642834e-01
3.41037750e-01 -9.30227041e-01 1.16892517e+00 1.08111155e+00
4.62875336e-01 -1.11641645e+00 1.71731040e-01 9.95610893e-01
-9.47463691e-01 -4.37035352e-01 -3.96628171e-01 1.70819014e-01
4.43223566e-01 5.62849760e-01 -8.59850407e-01 7.14839548e-02
3.74123603e-01 6.90610051e-01 -4.19695489e-02 9.56758142e-01
4.52063838e-03 6.17252469e-01 -3.81593347e-01 -3.98891270e-01
3.09488267e-01 -5.74493349e-01 1.15088344e+00 8.53490889e-01
7.95879304e-01 7.64372721e-02 2.21383870e-01 1.32522035e+00
3.15546274e-01 -4.22179282e-01 -9.13295269e-01 1.05360243e-02
6.88982904e-01 7.34032214e-01 -4.46460783e-01 -1.23083681e-01
-3.81375663e-02 6.21864021e-01 3.54459107e-01 7.36291170e-01
-8.85123789e-01 1.20023966e-01 1.19804657e+00 5.91988489e-02
4.72059608e-01 -3.99775684e-01 -1.86363608e-01 -1.08495927e+00
9.64642763e-02 -6.17142618e-01 1.37002259e-01 -4.88419384e-01
-1.29407001e+00 6.64953053e-01 3.22475940e-01 -1.52573287e+00
-8.85859728e-01 -1.76444843e-01 -7.63757110e-01 1.00846553e+00
-1.01385736e+00 -7.49889731e-01 2.01971736e-02 5.11760592e-01
7.38381326e-01 -4.28455234e-01 6.05292737e-01 -2.09102780e-01
-4.98020500e-01 3.39746177e-01 3.45882893e-01 -3.44774574e-01
-1.09709091e-02 -1.06446993e+00 4.97926921e-01 9.54644680e-01
2.57329401e-02 5.07580101e-01 1.11341441e+00 -8.72627258e-01
-1.18143737e+00 -1.31318021e+00 6.81395113e-01 -5.58854342e-01
4.32857215e-01 -3.55020881e-01 -9.16282237e-01 6.20460153e-01
-3.84210885e-01 -2.75921881e-01 2.67888010e-01 -3.66240889e-01
1.80692583e-01 -7.76879489e-02 -1.10740864e+00 9.82962489e-01
9.74571884e-01 -1.26693934e-01 -2.52501905e-01 4.55292575e-02
5.09869039e-01 -2.70945489e-01 -4.39119428e-01 3.00511181e-01
6.38288677e-01 -1.03129864e+00 1.00096810e+00 -7.28038073e-01
4.71779794e-01 -4.75598723e-01 -2.11840793e-01 -1.64748824e+00
-3.81362945e-01 -4.61229175e-01 -4.63290662e-01 1.00338578e+00
3.19705963e-01 -6.68685079e-01 7.15758622e-01 6.67334855e-01
-2.77098626e-01 -1.08381248e+00 -1.03547132e+00 -7.45931804e-01
1.51656061e-01 -6.20586276e-01 8.11524570e-01 2.94548184e-01
-5.37420988e-01 -3.82918231e-02 -4.55750257e-01 5.18171251e-01
9.12549794e-01 1.59255326e-01 8.25524569e-01 -8.38895440e-01
-2.83111036e-01 -3.26438963e-01 -2.78542638e-01 -1.27394938e+00
3.30924928e-01 -8.19209814e-01 1.88817665e-01 -1.48683333e+00
7.64066502e-02 -7.24740565e-01 2.34029412e-01 -1.40676036e-01
-9.18398201e-02 -6.24681652e-01 3.97295624e-01 3.46393555e-01
-2.17686459e-01 1.06163299e+00 1.36494315e+00 -1.31127805e-01
-4.77951974e-01 4.46129888e-01 -4.09098864e-01 7.11510062e-01
7.20973492e-01 -5.12382627e-01 -9.02866125e-01 -9.57364514e-02
2.08612800e-01 3.77898693e-01 4.90320891e-01 -8.92674625e-01
1.84591874e-01 -6.65513039e-01 4.32629079e-01 -8.53601336e-01
5.62355697e-01 -6.78061128e-01 6.09601140e-01 4.05856133e-01
-4.73026842e-01 -1.95955694e-01 -2.95570374e-01 9.14813817e-01
6.09328821e-02 -1.83523133e-01 5.93118131e-01 2.95960028e-02
-6.45950139e-01 5.23329437e-01 -6.41641140e-01 -8.99500623e-02
1.34737170e+00 -3.22356105e-01 -2.82331742e-02 -6.37149811e-01
-7.48131156e-01 5.35013735e-01 4.11041915e-01 5.12089670e-01
1.01670969e+00 -1.53415334e+00 -7.39254892e-01 5.15086174e-01
4.68805097e-02 1.50223628e-01 2.18638301e-01 3.21325392e-01
-1.05644420e-01 1.46675259e-01 -2.06053033e-01 -7.89369702e-01
-6.12774014e-01 4.72951531e-01 5.30783951e-01 3.82664837e-02
-8.29976499e-01 5.72944641e-01 3.17096800e-01 -1.98576868e-01
1.02513051e-02 -3.84497315e-01 -3.75217170e-01 -9.99644399e-02
5.89327812e-01 5.11534989e-01 -4.52839762e-01 -8.62206936e-01
-1.65445581e-01 4.64660496e-01 1.36344731e-01 -3.48190665e-01
1.30024076e+00 -1.71754435e-01 3.28232467e-01 5.59883475e-01
1.32926607e+00 -2.48714760e-01 -2.10661340e+00 3.03576618e-01
-2.51735598e-01 -5.24553180e-01 -3.89792025e-01 -4.34079021e-01
-6.74271166e-01 6.42614722e-01 3.73402178e-01 3.36743504e-01
1.00163782e+00 1.34405643e-01 8.72507155e-01 1.48218274e-01
6.76167250e-01 -8.55834663e-01 -1.43094994e-02 4.75997061e-01
1.02861845e+00 -8.52444410e-01 -5.87888002e-01 6.26741648e-02
-9.29600656e-01 1.16160381e+00 3.38322371e-01 -2.64688879e-01
8.09342265e-01 7.78437629e-02 -2.98916787e-01 -2.98399478e-02
-9.99074459e-01 -2.38639981e-01 4.91698891e-01 7.30155945e-01
-3.92489731e-01 4.05815423e-01 -1.54165134e-01 3.72428715e-01
-3.59444261e-01 -1.56461984e-01 4.74004388e-01 4.13011998e-01
-5.91080487e-01 -8.76221418e-01 -3.71766031e-01 4.34454352e-01
2.93363303e-01 6.69042289e-01 4.97132428e-02 3.58553469e-01
4.07607168e-01 9.38383460e-01 3.00827891e-01 -5.84528625e-01
2.65379816e-01 6.75860569e-02 1.80762038e-01 -4.10569370e-01
6.61660954e-02 -2.19383970e-01 -8.04689005e-02 -4.10632193e-01
-1.05273053e-01 -1.19923377e+00 -1.31685209e+00 -3.28119844e-01
1.89625263e-01 -2.81051453e-02 5.30667901e-01 1.04144895e+00
2.99233347e-01 4.06686127e-01 9.05465484e-01 -8.08560848e-01
-6.29731715e-01 -3.86002451e-01 -3.75451326e-01 5.43012440e-01
6.41541004e-01 -7.55044460e-01 -3.70576829e-01 7.46434554e-02] | [6.430457592010498, 0.7949286699295044] |
b89a91e7-b370-4cec-a1b6-49b1ee84e86f | enabling-efficiency-precision-trade-offs-for | 2106.00730 | null | https://arxiv.org/abs/2106.00730v2 | https://arxiv.org/pdf/2106.00730v2.pdf | Enabling Efficiency-Precision Trade-offs for Label Trees in Extreme Classification | Extreme multi-label classification (XMC) aims to learn a model that can tag data points with a subset of relevant labels from an extremely large label set. Real world e-commerce applications like personalized recommendations and product advertising can be formulated as XMC problems, where the objective is to predict for a user a small subset of items from a catalog of several million products. For such applications, a common approach is to organize these labels into a tree, enabling training and inference times that are logarithmic in the number of labels. While training a model once a label tree is available is well studied, designing the structure of the tree is a difficult task that is not yet well understood, and can dramatically impact both model latency and statistical performance. Existing approaches to tree construction fall at an extreme point, either optimizing exclusively for statistical performance, or for latency. We propose an efficient information theory inspired algorithm to construct intermediary operating points that trade off between the benefits of both. Our algorithm enables interpolation between these objectives, which was not previously possible. We corroborate our theoretical analysis with numerical results, showing that on the Wiki-500K benchmark dataset our method can reduce a proxy for expected latency by up to 28% while maintaining the same accuracy as Parabel. On several datasets derived from e-commerce customer logs, our modified label tree is able to improve this expected latency metric by up to 20% while maintaining the same accuracy. Finally, we discuss challenges in realizing these latency improvements in deployed models. | ['Inderjit S. Dhillon', 'Sujay Sanghavi', 'Kedarnath Kolluri', 'Daniel L. Jiang', 'Tavor Z. Baharav'] | 2021-06-01 | null | null | null | null | ['extreme-multi-label-classification'] | ['methodology'] | [ 2.39122733e-01 -3.43165509e-02 -3.47684860e-01 -6.19430304e-01
-1.15988839e+00 -9.02872801e-01 3.02444547e-01 6.27809882e-01
-3.23065937e-01 3.21029723e-01 -2.78241396e-01 -6.38882339e-01
-4.62453455e-01 -6.02010429e-01 -6.58630073e-01 -5.80018699e-01
-3.88724536e-01 9.86256540e-01 9.96778533e-02 2.41510794e-01
2.37541988e-01 2.22757891e-01 -1.59494495e+00 4.55299526e-01
5.66491067e-01 1.49148273e+00 -1.52449235e-01 6.61160648e-01
-2.43716404e-01 5.95446765e-01 -9.78432074e-02 -5.03186345e-01
5.29484689e-01 2.81191319e-02 -9.58127439e-01 5.97751467e-03
3.64068836e-01 -4.55055274e-02 5.44114888e-01 8.68722796e-01
1.46014035e-01 -1.92183465e-01 6.36568010e-01 -1.25555325e+00
-5.83951101e-02 7.90181637e-01 -7.33847260e-01 -2.31987864e-01
7.26689547e-02 -3.84646565e-01 1.43848491e+00 -4.84154105e-01
4.09382671e-01 8.47802818e-01 6.62061572e-01 1.83815822e-01
-1.60178113e+00 -6.35213256e-01 1.72023281e-01 -5.81234396e-02
-1.44435859e+00 -3.60974759e-01 3.79603148e-01 -3.80405873e-01
6.99031591e-01 4.28287894e-01 2.31564209e-01 4.70193952e-01
9.82871577e-02 5.54237545e-01 1.15631604e+00 -5.86064279e-01
5.91757655e-01 5.30725121e-01 5.95946670e-01 5.11920989e-01
2.89880395e-01 -5.69324121e-02 -5.32792687e-01 -5.88917136e-01
1.98871735e-03 4.81433533e-02 1.75526559e-01 -5.53045809e-01
-8.49285245e-01 9.59915459e-01 2.20320329e-01 -7.92327989e-03
-2.81244576e-01 3.79275084e-01 4.70151663e-01 5.37845969e-01
6.29802525e-01 4.39439237e-01 -8.98618042e-01 -1.24907970e-01
-8.59963715e-01 2.07745552e-01 1.22045791e+00 1.05323517e+00
6.94691420e-01 -7.20117629e-01 1.17452286e-01 6.85584128e-01
3.70747060e-01 3.19096148e-01 1.37284890e-01 -9.78956163e-01
3.93364042e-01 4.76714164e-01 2.17722028e-01 -6.15715146e-01
-5.71812272e-01 -6.60485983e-01 -4.01926517e-01 1.76623896e-01
8.10766399e-01 -8.30600560e-02 -5.64606488e-01 1.61454248e+00
5.00689685e-01 -7.86672086e-02 -2.17960224e-01 6.01072729e-01
-9.41653103e-02 6.00065708e-01 2.23229527e-01 -2.83959150e-01
1.78118610e+00 -7.75720417e-01 -2.01883852e-01 -1.97810069e-01
1.25195813e+00 -8.64803135e-01 9.23148453e-01 6.77470863e-01
-8.75987530e-01 -1.92138538e-01 -1.00864351e+00 9.82710943e-02
-2.77590334e-01 -3.86531383e-01 1.06415009e+00 1.03292668e+00
-9.02701974e-01 7.06607699e-01 -6.84543192e-01 -2.19181567e-01
2.11024493e-01 5.96251547e-01 -2.78965682e-02 -4.71678376e-02
-7.01098263e-01 5.68950295e-01 2.87503839e-01 -3.01029176e-01
-4.40237582e-01 -7.45087028e-01 -2.13818729e-01 3.02597821e-01
6.76506937e-01 -3.79133642e-01 1.51692748e+00 -8.35218072e-01
-1.18993235e+00 6.49066508e-01 -4.10414999e-03 -7.38739908e-01
3.11165631e-01 -1.33361265e-01 -3.48022103e-01 -2.24335074e-01
-1.01754129e-01 5.69678366e-01 5.23959160e-01 -1.26905715e+00
-1.17540264e+00 -6.25556707e-01 2.63033926e-01 7.18350783e-02
-5.14692426e-01 -2.10571419e-02 -5.04002333e-01 -2.32957035e-01
1.70474693e-01 -1.37480605e+00 -3.84330660e-01 -1.60343632e-01
-1.83131218e-01 -3.12903613e-01 4.79787230e-01 -2.71841794e-01
1.12189174e+00 -2.00352740e+00 -4.24511909e-01 6.04032993e-01
4.12531435e-01 -1.95737734e-01 -1.80825442e-02 3.96880955e-01
1.67917013e-01 2.18793780e-01 2.66646594e-01 -4.36318308e-01
1.57689035e-01 -5.20695150e-02 -2.38197580e-01 4.85595196e-01
-4.31908935e-01 7.58609354e-01 -6.32565081e-01 -4.86973405e-01
-1.55824527e-01 1.56833395e-01 -7.92505383e-01 -4.90332842e-02
-5.96888423e-01 -1.68054868e-02 -3.03452045e-01 6.09584928e-01
5.04027426e-01 -7.37134278e-01 6.40163183e-01 -8.63739697e-04
1.87792122e-01 3.25219065e-01 -1.40846813e+00 1.33584917e+00
-8.12692821e-01 4.07496579e-02 8.27832054e-03 -8.44850600e-01
7.02418208e-01 1.52635619e-01 8.45577836e-01 -6.85418844e-01
2.08023101e-01 5.03574729e-01 -1.11993521e-01 9.41286534e-02
3.37465018e-01 -1.12333052e-01 -2.07409382e-01 9.21585798e-01
-3.21655899e-01 4.90719587e-01 8.14077258e-02 6.62342012e-02
1.10176980e+00 -2.64620125e-01 2.38187656e-01 -4.18285817e-01
3.43046710e-02 7.13575408e-02 3.58544737e-01 8.21715891e-01
1.23902112e-01 9.22374278e-02 5.98238528e-01 -6.25699878e-01
-8.79180729e-01 -8.40924025e-01 -3.78928006e-01 1.48129094e+00
5.94570860e-02 -7.54510343e-01 -7.34532952e-01 -9.58702564e-01
2.87040532e-01 7.09330499e-01 -4.61672425e-01 1.41826123e-01
-3.29587430e-01 -6.22566998e-01 -2.18150914e-02 2.33215690e-01
-7.58759119e-03 -4.52339739e-01 -5.84447384e-01 4.42858040e-01
1.40570462e-01 -1.16765380e+00 -5.34998298e-01 6.65683806e-01
-9.01051998e-01 -8.82993162e-01 -1.57018200e-01 -7.37261534e-01
5.11420667e-01 3.27930361e-01 1.23809123e+00 -4.77769412e-03
-5.68496659e-02 9.10359845e-02 -4.13562596e-01 -3.30300361e-01
-4.94664133e-01 4.93461311e-01 6.42048614e-03 2.16398761e-01
5.77295423e-01 -7.72442937e-01 -5.93039513e-01 4.62511003e-01
-6.83914721e-01 1.28404409e-01 5.27898908e-01 6.26718998e-01
6.36070430e-01 3.40565592e-01 5.77901423e-01 -1.54381657e+00
1.78985998e-01 -6.61495745e-01 -1.00906527e+00 2.49268904e-01
-1.43879294e+00 3.40487748e-01 7.60448456e-01 -5.15628815e-01
-6.26627386e-01 5.12465537e-01 -4.70161550e-02 1.34555593e-01
4.78072278e-02 4.00042415e-01 3.28117013e-02 -1.31428689e-01
4.04917806e-01 -1.24705680e-01 -1.25559606e-02 -7.46554255e-01
6.59751475e-01 9.00640011e-01 2.13816702e-01 -7.42425263e-01
3.87694806e-01 3.47658217e-01 1.67337954e-01 -2.68478394e-01
-1.11989295e+00 -7.86329269e-01 -4.65174317e-01 -1.39689110e-02
3.35698545e-01 -6.90153003e-01 -1.29614770e+00 -9.44417566e-02
-6.82018876e-01 -3.09875786e-01 -2.46803001e-01 2.98805356e-01
-6.38369799e-01 7.50867650e-02 -6.48410618e-01 -7.31390119e-01
-5.22044361e-01 -9.70144272e-01 1.13083553e+00 -2.66113788e-01
-2.02001899e-01 -8.34623098e-01 -2.07282677e-01 3.75771046e-01
3.66669774e-01 -4.82189879e-02 1.38386154e+00 -9.59312201e-01
-8.38729203e-01 -4.41253364e-01 -2.38549560e-01 2.03939676e-02
-1.01605564e-01 -5.74431241e-01 -8.82892251e-01 -5.77833235e-01
1.90008059e-02 -2.73546934e-01 6.20017290e-01 1.53676629e-01
1.17893648e+00 -3.72205943e-01 -5.12478769e-01 4.47851598e-01
1.70722234e+00 2.56527305e-01 9.21305194e-02 2.31920972e-01
2.95177430e-01 5.84766030e-01 7.73276508e-01 6.12426937e-01
3.70176673e-01 1.15229046e+00 3.55247557e-01 1.61360309e-01
2.01665476e-01 -1.62071630e-01 4.45218571e-02 7.50967681e-01
3.51291835e-01 -1.92151546e-01 -8.56862545e-01 3.45069200e-01
-1.80387032e+00 -4.65073049e-01 -5.02250902e-02 2.63684940e+00
9.24739122e-01 6.65389478e-01 2.52928853e-01 2.62422800e-01
3.78240526e-01 -3.16697091e-01 -6.39364660e-01 -3.86169434e-01
4.98168647e-01 5.67125492e-02 1.02891445e+00 4.76757735e-01
-1.08586121e+00 6.53123200e-01 5.82639647e+00 8.81482720e-01
-1.03096390e+00 3.38786989e-01 9.01034951e-01 -4.55274940e-01
-2.04861194e-01 4.49918598e-01 -1.26221108e+00 6.13411725e-01
1.59535313e+00 -2.02832043e-01 6.20640934e-01 1.12574220e+00
-1.48207217e-01 -1.28433526e-01 -1.52163684e+00 1.05965841e+00
-3.18811089e-01 -1.23140705e+00 -4.08492893e-01 8.16244423e-01
6.39533162e-01 -1.88697688e-02 1.88224122e-01 4.24101770e-01
6.15677416e-01 -6.99513376e-01 7.20160663e-01 8.56346339e-02
7.71674335e-01 -7.62928784e-01 5.76461792e-01 5.80749512e-01
-1.15136206e+00 -3.40388626e-01 -1.74404919e-01 3.66673581e-02
2.18033865e-01 9.16735947e-01 -9.66391861e-01 1.50150076e-01
4.77137476e-01 1.02613755e-01 -6.50130391e-01 1.02884734e+00
5.31299949e-01 8.43417764e-01 -7.38586366e-01 -2.01492876e-01
2.83502579e-01 -8.67873728e-02 -1.06401071e-01 9.46956158e-01
2.46141300e-01 -1.71639875e-01 3.16765398e-01 2.59022564e-01
-1.22663587e-01 5.73995590e-01 -2.79102951e-01 2.10498318e-01
7.15379655e-01 1.39613974e+00 -1.03423393e+00 -2.61713743e-01
-5.72914839e-01 8.03513229e-01 5.19718170e-01 -1.05980575e-01
-9.65564370e-01 -5.34372367e-02 5.36194980e-01 3.00848633e-01
2.87157238e-01 -1.03597753e-01 -5.67430913e-01 -6.52307212e-01
1.87558681e-01 -8.76412928e-01 5.12249410e-01 4.67705959e-03
-1.02787030e+00 4.29029286e-01 -1.58652946e-01 -1.11041510e+00
-3.55849206e-01 -5.51543891e-01 2.51702130e-01 6.17435634e-01
-1.20982182e+00 -1.06691301e+00 8.24283734e-02 9.84082930e-03
2.88261801e-01 1.36410221e-01 9.08965528e-01 6.08962357e-01
-2.85131216e-01 8.87405932e-01 6.16168141e-01 -4.93444651e-01
5.63638449e-01 -1.42469060e+00 5.11317134e-01 1.97202116e-01
6.73540950e-01 5.12541831e-01 5.50281346e-01 -3.36126328e-01
-1.45476413e+00 -1.01425827e+00 1.18115222e+00 -5.03059924e-01
5.40780842e-01 -6.87389791e-01 -5.82098126e-01 6.76904798e-01
-2.95370877e-01 6.44139200e-02 7.99191236e-01 8.38850200e-01
-6.89334869e-01 -5.65487266e-01 -9.03206110e-01 4.26085770e-01
9.83534217e-01 -3.57220143e-01 2.60468543e-01 7.19246387e-01
7.56843805e-01 -8.52817520e-02 -1.25492811e+00 1.97667375e-01
8.48636806e-01 -7.26051569e-01 7.36397386e-01 -5.68233430e-01
9.59660858e-02 -5.74671924e-02 -3.24655503e-01 -9.82247174e-01
-3.54454458e-01 -6.52534604e-01 -2.48695686e-01 1.34837818e+00
6.61763310e-01 -5.69462180e-01 1.18660426e+00 7.95621753e-01
4.05717194e-01 -9.64462578e-01 -6.66420281e-01 -9.42458272e-01
-2.52198968e-02 -7.64793932e-01 7.83960342e-01 5.93120098e-01
8.96682031e-03 6.68269753e-01 -3.35223049e-01 1.09670706e-01
8.31852734e-01 5.73582888e-01 6.32062674e-01 -1.52286816e+00
-5.53690255e-01 -3.25429261e-01 -6.75947443e-02 -1.28921962e+00
-5.59845306e-02 -1.18623245e+00 -1.70724690e-01 -1.12209105e+00
3.91477048e-01 -1.28059113e+00 -4.22288775e-01 5.64034343e-01
1.43633500e-01 2.03096613e-01 2.25519300e-01 4.18821901e-01
-9.36493099e-01 -1.18178159e-01 5.22529542e-01 7.99819604e-02
-1.82659328e-01 4.62599635e-01 -9.64608431e-01 4.97253507e-01
6.62860274e-01 -7.97929585e-01 -4.66855466e-01 -1.31384850e-01
5.55381119e-01 2.60873228e-01 -1.07904829e-01 -8.90182316e-01
4.70252186e-01 1.76122561e-02 -6.92219064e-02 -5.13202906e-01
2.29832336e-01 -1.22404325e+00 3.95422041e-01 3.90908360e-01
-7.49333262e-01 -1.69030219e-01 -2.23289445e-01 6.65771365e-01
2.65512288e-01 -3.33172292e-01 7.37222910e-01 7.28026032e-02
-4.11152333e-01 1.83893725e-01 1.04910895e-01 -1.22431368e-01
1.18154216e+00 1.71350107e-01 -2.29825571e-01 -4.25900102e-01
-5.66494405e-01 3.13903600e-01 2.93950230e-01 3.59920740e-01
-8.81991684e-02 -1.25395739e+00 -5.21978974e-01 4.69677225e-02
2.96889454e-01 -3.78485620e-01 -7.50468001e-02 5.87991238e-01
-1.85747549e-01 7.98890889e-01 3.50769013e-01 -5.71932614e-01
-1.32853544e+00 9.12838340e-01 -3.94508876e-02 -6.61908925e-01
-3.88993859e-01 5.92890084e-01 3.96990217e-02 -2.45632455e-01
4.47853237e-01 -1.40352115e-01 1.35773525e-01 6.57324642e-02
6.00248218e-01 3.10866863e-01 5.42794645e-01 -2.66378373e-01
-2.59896785e-01 4.09237713e-01 -5.45546651e-01 8.97261500e-02
1.04120636e+00 -2.34155878e-01 1.29764959e-01 2.85315543e-01
1.58295798e+00 -5.81069058e-03 -1.04608762e+00 -6.48843527e-01
6.86357975e-01 -2.82418102e-01 8.35735127e-02 -9.39621687e-01
-1.00157833e+00 4.48169023e-01 8.02786827e-01 6.21498764e-01
1.10054672e+00 2.37101763e-01 1.08612180e+00 3.81004244e-01
9.45404589e-01 -9.49708045e-01 -1.37164772e-01 1.52490273e-01
2.15030923e-01 -1.17495191e+00 -5.62341474e-02 -6.63630307e-01
-4.36417997e-01 8.34439337e-01 2.35691711e-01 2.52904445e-01
9.92985547e-01 5.83445072e-01 -3.64778936e-02 -2.89777786e-01
-1.25190067e+00 -9.89261344e-02 1.04555497e-02 4.69633602e-02
3.13223988e-01 3.92805576e-01 -3.66516769e-01 6.00183785e-01
-1.51722863e-01 -1.90257262e-02 3.82076427e-02 8.44382584e-01
-5.43537140e-01 -1.44531679e+00 -1.26136497e-01 8.64950418e-01
-7.52069592e-01 -1.75279841e-01 1.78675398e-01 3.95158917e-01
-5.15903421e-02 1.09752023e+00 9.18845609e-02 -4.80749905e-01
2.23967478e-01 2.75672048e-01 1.25222787e-01 -6.51579201e-01
-4.28344220e-01 2.02970788e-01 2.91090012e-01 -5.61045110e-01
5.61927371e-02 -6.56904280e-01 -1.07740223e+00 -4.88827378e-01
-6.16883337e-01 3.71620983e-01 1.04837012e+00 7.53137290e-01
5.19236743e-01 2.24363920e-03 9.91638005e-01 -3.68932396e-01
-1.14153743e+00 -6.04489505e-01 -9.36990857e-01 3.87971908e-01
1.61859649e-03 -5.25294006e-01 -1.81013793e-01 9.62799117e-02] | [9.427563667297363, 4.440122127532959] |
79a776dc-13df-475f-9299-c7d25a768b9a | bias-in-conversational-search-the-double | 2010.10409 | null | https://arxiv.org/abs/2010.10409v1 | https://arxiv.org/pdf/2010.10409v1.pdf | Bias in Conversational Search: The Double-Edged Sword of the Personalized Knowledge Graph | Conversational AI systems are being used in personal devices, providing users with highly personalized content. Personalized knowledge graphs (PKGs) are one of the recently proposed methods to store users' information in a structured form and tailor answers to their liking. Personalization, however, is prone to amplifying bias and contributing to the echo-chamber phenomenon. In this paper, we discuss different types of biases in conversational search systems, with the emphasis on the biases that are related to PKGs. We review existing definitions of bias in the literature: people bias, algorithm bias, and a combination of the two, and further propose different strategies for tackling these biases for conversational search systems. Finally, we discuss methods for measuring bias and evaluating user satisfaction. | ['Arjen P. de Vries', 'Faegheh Hasibi', 'Emma J. Gerritse'] | 2020-10-20 | null | null | null | null | ['conversational-search'] | ['natural-language-processing'] | [-2.40674242e-01 4.39992517e-01 -5.33566475e-01 -5.69553316e-01
-1.50476784e-01 -5.50374568e-01 4.55177456e-01 2.54147917e-01
-3.04918826e-01 8.16483140e-01 7.41659701e-01 -7.60840029e-02
-3.49534929e-01 -6.98355854e-01 5.54793924e-02 -3.08335185e-01
2.29435846e-01 7.61102736e-01 -8.30499455e-02 -6.56584382e-01
9.48650479e-01 8.42022896e-02 -1.59647250e+00 5.35922050e-01
1.30846691e+00 7.94611037e-01 3.56852785e-02 7.46066391e-01
-5.93242943e-01 6.48501456e-01 -8.31909955e-01 -1.08697331e+00
-2.18590200e-01 -3.80277783e-01 -1.11474669e+00 -3.66636992e-01
3.82157713e-01 -2.41165429e-01 -1.14162318e-01 9.30054188e-01
9.92220283e-01 2.39313811e-01 4.47550446e-01 -1.53295076e+00
-1.06887925e+00 8.21118951e-01 1.97248295e-01 4.23726737e-01
1.04132998e+00 -6.69617876e-02 1.13144064e+00 -6.15726769e-01
4.72736180e-01 1.55572331e+00 7.80341744e-01 9.02810812e-01
-1.27674901e+00 -4.88161832e-01 2.92239755e-01 3.39140832e-01
-1.00956118e+00 -5.08409023e-01 8.72502565e-01 -5.90169251e-01
8.22980881e-01 9.50373411e-01 8.76106799e-01 1.38412917e+00
-1.17356619e-02 8.70870471e-01 1.38136947e+00 -3.85360360e-01
4.52459753e-01 1.09085965e+00 8.36669743e-01 3.64927709e-01
5.37540913e-01 -2.77181640e-02 -1.15726531e+00 -1.05043960e+00
2.02458605e-01 -1.43154666e-01 -4.12908822e-01 -4.01802748e-01
-6.86607838e-01 8.80309045e-01 3.05429995e-01 2.94955015e-01
-5.88608205e-01 -1.39181554e-01 1.51071534e-01 2.87106156e-01
7.22853959e-01 9.52697277e-01 -1.73543051e-01 -3.75201136e-01
-2.40333855e-01 5.49819529e-01 1.49383342e+00 1.20630229e+00
7.00652361e-01 -6.19573534e-01 -9.32680309e-01 1.23969543e+00
3.65408719e-01 6.34898245e-01 7.97748327e-01 -8.58669698e-01
-1.25022143e-01 6.47215307e-01 5.38392186e-01 -1.37430668e+00
-3.32944274e-01 -2.78082073e-01 -2.40953982e-01 -5.25289297e-01
1.79360330e-01 -8.69990289e-02 -3.34764719e-01 1.40610230e+00
1.43347740e-01 -7.88234293e-01 -9.32603106e-02 8.88545215e-01
1.24200726e+00 1.67165175e-01 2.60445356e-01 -2.07921192e-01
1.57282400e+00 -9.52166617e-01 -1.26761270e+00 -3.93929929e-01
3.54321748e-01 -5.90951979e-01 1.54039752e+00 1.39161557e-01
-1.06646025e+00 -4.24910896e-02 -6.01342618e-01 1.45613384e-02
-6.23488903e-01 -6.08656466e-01 7.92899549e-01 1.38844585e+00
-1.08125925e+00 5.93456388e-01 2.22507283e-01 -7.78280914e-01
2.29962513e-01 1.64593965e-01 6.32794857e-01 1.83441445e-01
-1.56709290e+00 1.24323225e+00 -3.64400774e-01 -3.41821969e-01
-2.13358432e-01 -5.82033634e-01 -1.99126899e-01 9.60864127e-02
9.65125859e-02 -9.66585219e-01 1.67398095e+00 -8.06396663e-01
-1.72177422e+00 9.86533582e-01 -4.89147514e-01 7.00739026e-02
2.95976937e-01 -3.86436373e-01 -6.31015897e-01 -4.18266803e-01
8.35912675e-02 2.47609809e-01 6.73246384e-01 -1.43362355e+00
-5.51697791e-01 -4.70919997e-01 1.89860240e-01 6.50732219e-01
-5.74454308e-01 2.55798660e-02 -2.29758471e-01 -1.53956652e-01
-3.35673958e-01 -8.99379969e-01 -3.34236711e-01 -5.14547884e-01
-1.92044094e-01 -8.15757990e-01 5.63041091e-01 -5.61814845e-01
1.80346155e+00 -1.68615782e+00 -6.99445009e-02 4.51665640e-01
3.80542636e-01 2.76900500e-01 2.67798126e-01 7.77526081e-01
6.65131390e-01 3.78819764e-01 3.69942278e-01 -1.07612759e-01
3.69494647e-01 1.25034451e-01 6.97059706e-02 -1.57641128e-01
-7.34469175e-01 1.06235135e+00 -1.38133419e+00 -5.41250587e-01
-1.80977821e-01 1.32160902e-01 -5.88493109e-01 3.98180634e-01
-1.07682511e-01 2.70028021e-02 -3.87193292e-01 6.80030286e-01
3.50543141e-01 -1.16654292e-01 3.54510039e-01 1.06296815e-01
6.04262687e-02 6.78979218e-01 -6.38016820e-01 1.18472302e+00
-4.18300211e-01 4.97203857e-01 3.99960242e-02 -4.71184291e-02
9.48923349e-01 2.57264763e-01 1.36441812e-01 -8.67115080e-01
9.16092098e-02 1.47326589e-01 -3.35706860e-01 -8.23357403e-01
9.75501835e-01 -3.65964114e-03 -1.36270434e-01 7.35694766e-01
-3.50028247e-01 -2.81631231e-01 -4.27451096e-02 3.92568469e-01
7.73755312e-01 -7.70529211e-01 5.14724851e-01 -6.48135960e-01
3.56490105e-01 -1.19532757e-01 1.57303855e-01 1.29264545e+00
-5.38554847e-01 -1.75149649e-01 2.81190723e-01 -3.83274764e-01
-6.02937758e-01 -4.63068038e-01 1.38694286e-01 1.56072080e+00
3.57107937e-01 -6.40389681e-01 -7.78347433e-01 -8.45763981e-01
4.29422468e-01 1.20143795e+00 -4.25600350e-01 -3.12147409e-01
8.97211507e-02 -4.82493132e-01 3.57678950e-01 2.40345567e-01
5.08508027e-01 -1.26430857e+00 -4.21167403e-01 -7.44399279e-02
-8.70568752e-01 -5.80150366e-01 -7.82972574e-01 -3.83496583e-01
-8.96780491e-01 -6.76308036e-01 -6.28789365e-01 -4.61882442e-01
4.29179817e-01 5.75936675e-01 1.60165739e+00 3.72750521e-01
1.92319229e-01 1.13237631e+00 -4.30689454e-01 -7.10592568e-01
-1.59203008e-01 -6.28074110e-02 1.39135629e-01 -1.79239899e-01
1.08759606e+00 -3.94316763e-01 -1.05132961e+00 6.82166100e-01
-4.49708372e-01 -3.18681151e-01 2.77505875e-01 7.02344298e-01
-3.92846256e-01 -4.38304991e-01 6.66356027e-01 -1.28477144e+00
1.89227688e+00 -6.56651378e-01 1.25061229e-01 2.48699218e-01
-1.57998419e+00 -1.48895040e-01 -8.90462771e-02 -2.64977098e-01
-1.41192389e+00 -6.77600622e-01 1.19987510e-01 4.97364581e-01
1.40627548e-01 3.31716061e-01 1.19953714e-02 -3.65686297e-01
1.32908714e+00 -2.60958761e-01 -8.91965348e-03 -2.99628407e-01
6.26132965e-01 1.23836577e+00 2.54245438e-02 -4.85641986e-01
1.13452012e-02 2.87654489e-01 -8.45518649e-01 -8.43124270e-01
-6.84782922e-01 -8.09081435e-01 5.10016270e-02 -7.54617214e-01
3.38816851e-01 -2.11093381e-01 -1.13111424e+00 1.47348449e-01
-1.03408921e+00 -1.05056018e-01 -3.29481542e-01 1.21852241e-01
-4.16316658e-01 3.87612283e-01 -5.48732102e-01 -1.39558935e+00
-7.51665890e-01 -5.91050148e-01 4.34697270e-01 6.16053820e-01
-1.12983012e+00 -8.91856670e-01 3.71077359e-01 7.84429729e-01
1.08977473e+00 -5.30838966e-01 7.90617406e-01 -8.43838334e-01
-5.05635627e-02 -3.64629984e-01 3.30247767e-02 -3.22863483e-03
2.30965740e-03 -5.73671877e-01 -1.10290802e+00 -1.22772127e-01
1.88892812e-01 -5.52181415e-02 2.80247867e-01 1.17055222e-01
9.32553768e-01 -8.48694265e-01 -7.36063719e-01 -2.98618823e-01
9.15704489e-01 4.01913375e-01 6.03987515e-01 2.62091517e-01
1.61931887e-01 8.62734616e-01 5.12028515e-01 5.87586641e-01
6.94576025e-01 7.46608853e-01 -3.15344497e-03 3.90923023e-01
6.90124407e-02 -5.08159637e-01 1.51944920e-01 6.54221356e-01
-2.74885148e-01 -4.97226983e-01 -5.55831015e-01 5.40544808e-01
-2.00351739e+00 -8.88967693e-01 -3.51303183e-02 2.15944767e+00
7.92521894e-01 -3.99222583e-01 2.55022079e-01 -1.31065115e-01
9.32557166e-01 1.04996778e-01 -5.11822701e-01 -9.71142113e-01
7.55845681e-02 -2.25877851e-01 4.22946095e-01 8.83576810e-01
-2.67680705e-01 8.36853623e-01 8.19257927e+00 1.40048608e-01
-4.93384868e-01 9.84228328e-02 4.33928639e-01 2.06343015e-03
-8.66069138e-01 4.88623008e-02 -5.65197885e-01 6.24177396e-01
8.47349107e-01 -5.94771266e-01 6.31484568e-01 1.16885114e+00
7.27495039e-03 -3.81868064e-01 -1.10452950e+00 1.16692924e+00
3.41771930e-01 -9.73111808e-01 9.62492302e-02 -1.74435690e-01
7.67024338e-01 -5.75067639e-01 5.33951959e-03 4.14410800e-01
5.63741267e-01 -6.17720127e-01 1.70809999e-01 8.09525788e-01
9.08487737e-02 -3.62663478e-01 6.97169363e-01 2.48283491e-01
-1.48255184e-01 -1.26131177e-01 -2.51503497e-01 -3.59572828e-01
2.92756766e-01 8.29023778e-01 -1.03724182e+00 3.04513029e-03
1.22083437e+00 1.36860609e-01 -5.05301058e-01 9.53330219e-01
1.30814701e-01 3.32450658e-01 3.94468680e-02 -1.09316838e+00
-3.35714489e-01 -2.26632833e-01 8.71434629e-01 1.39699328e+00
5.98886274e-02 4.19367880e-01 -4.19721216e-01 8.24758410e-01
6.16255812e-02 2.69407809e-01 -7.60129333e-01 -1.89032167e-01
9.02452171e-01 1.17123210e+00 -2.61587381e-01 -5.94698668e-01
-7.98957124e-02 1.25396824e+00 2.40828723e-01 3.19963098e-01
-7.22640231e-02 -2.68029243e-01 8.20601583e-01 6.20883852e-02
-4.15182114e-01 1.72797903e-01 -6.91043556e-01 -9.28155065e-01
-1.19467579e-01 -1.25382233e+00 6.01292133e-01 -6.54779315e-01
-1.71286345e+00 3.30147505e-01 -1.16460226e-01 -4.91784185e-01
-1.50983676e-01 -1.55668244e-01 -6.16124988e-01 9.28307652e-01
-9.43777025e-01 -3.44588399e-01 -8.88956606e-01 4.55174357e-01
3.02789986e-01 1.23166724e-03 1.02600348e+00 2.68982381e-01
-1.99338451e-01 6.46689892e-01 -1.85401961e-01 -6.96303189e-01
1.14705527e+00 -1.32228017e+00 2.78126776e-01 -1.34499088e-01
-3.81680310e-01 1.05289841e+00 1.10235476e+00 -8.23026359e-01
-1.30410349e+00 -1.27553657e-01 1.46978307e+00 -1.02439773e+00
1.49359286e-01 -3.33859861e-01 -7.08844543e-01 5.00670373e-01
6.06531382e-01 -1.06123590e+00 1.28657627e+00 1.01597703e+00
-1.40965402e-01 -3.24636213e-02 -1.41386878e+00 8.03231061e-01
1.30789137e+00 -6.74094200e-01 -6.48022532e-01 5.25340736e-01
4.45641011e-01 -2.78846592e-01 -6.77272499e-01 6.37394935e-02
8.83213997e-01 -1.41005433e+00 7.64704466e-01 -6.60691381e-01
1.41287008e-02 2.62864262e-01 1.95887327e-01 -1.88529360e+00
-9.41612780e-01 -9.52553570e-01 -1.00395724e-01 1.31389296e+00
5.10146499e-01 -1.09590030e+00 9.24599051e-01 1.54188490e+00
6.39791563e-02 -5.60865939e-01 -3.02141875e-01 -4.32017207e-01
-3.54708344e-01 -1.33811459e-01 8.49523604e-01 1.15165186e+00
1.20520222e+00 7.28131711e-01 -5.97542763e-01 -2.74237424e-01
4.51149493e-01 -1.68857321e-01 8.15559149e-01 -1.31415296e+00
5.93195595e-02 -8.19975853e-01 1.45252123e-01 -1.18317914e+00
-3.36645335e-01 -6.30435050e-01 7.36554563e-02 -1.55767179e+00
3.74115348e-01 -5.77768683e-01 -4.70487922e-02 -2.94941068e-01
-5.45240879e-01 -2.01105237e-01 -1.29935145e-01 3.21737051e-01
-6.80174112e-01 3.84879380e-01 1.08995891e+00 1.34819850e-01
-4.89160985e-01 2.82693744e-01 -1.48297620e+00 5.10137498e-01
9.49740171e-01 -1.41762376e-01 -4.68310535e-01 2.37203408e-02
8.96844983e-01 -2.96097130e-01 1.45071015e-01 -4.58279520e-01
6.12673759e-01 -1.72946930e-01 -8.07155669e-02 -5.72970092e-01
3.02221060e-01 -7.50389576e-01 3.17469873e-02 3.36824894e-01
-6.76175058e-01 1.08214125e-01 -6.49489239e-02 6.86430871e-01
1.13347955e-01 -4.00434136e-01 2.47605667e-01 -2.75053114e-01
-4.17821258e-01 -1.24372542e-01 -8.48429680e-01 1.55112438e-03
5.14964044e-01 -9.88393575e-02 -6.27593219e-01 -1.12684143e+00
-5.32489002e-01 6.37497067e-01 9.13305804e-02 7.37554312e-01
3.95728856e-01 -1.48887265e+00 -2.43417755e-01 -2.00845182e-01
4.05781627e-01 -7.12486923e-01 3.54112744e-01 6.15728438e-01
-7.06580505e-02 8.20447564e-01 -7.08575025e-02 -2.00346813e-01
-1.35871291e+00 6.41031563e-01 2.38532692e-01 9.93089452e-02
8.41341168e-02 9.47235525e-01 -2.31445655e-01 -5.88445246e-01
5.27841806e-01 1.62952393e-01 -5.94324350e-01 2.26624385e-01
4.88948941e-01 1.05174351e+00 -1.73179489e-02 -4.94614318e-02
-4.28931236e-01 -1.17957192e-02 -1.17794372e-01 -2.64380813e-01
4.40485030e-01 -6.32052958e-01 -4.13217396e-01 6.56552792e-01
5.84351659e-01 2.15090975e-01 -1.00022607e-01 -6.74811482e-01
2.66662657e-01 -8.11326742e-01 -1.67071462e-01 -1.19698155e+00
-4.43959951e-01 2.44536504e-01 5.09067178e-01 8.38081717e-01
5.53420126e-01 1.43835142e-01 7.32283890e-01 5.09873152e-01
2.97354758e-01 -1.79177356e+00 2.10463077e-01 3.47143769e-01
1.13605571e+00 -1.22112036e+00 -5.59190363e-02 -6.12933636e-01
-1.13496614e+00 6.38528764e-01 7.69065320e-01 4.12689656e-01
8.70076478e-01 -4.70518678e-01 3.90694261e-01 -7.08703339e-01
-7.88913190e-01 -6.42854124e-02 5.34693182e-01 7.98274100e-01
8.70454073e-01 3.19272697e-01 -1.21400797e+00 8.81802499e-01
-5.92149317e-01 -4.18398976e-02 2.07504168e-01 9.55952287e-01
-5.49168229e-01 -9.87229109e-01 -3.34567547e-01 7.18735337e-01
-9.90984067e-02 -1.66435719e-01 -1.42446184e+00 4.14670259e-02
-3.97094369e-01 1.58437932e+00 -2.78061062e-01 -7.97205746e-01
6.97336137e-01 5.00396311e-01 1.22436874e-01 -3.43527138e-01
-1.03934395e+00 -6.63196266e-01 9.16213989e-01 -8.31504047e-01
-3.53820294e-01 -5.72056532e-01 -3.94462734e-01 -7.27206826e-01
-5.71262121e-01 7.49259651e-01 6.61679924e-01 5.57320178e-01
7.23579943e-01 -4.81698699e-02 4.64177042e-01 -5.42716682e-01
-3.74658614e-01 -1.26156855e+00 -3.77199948e-01 6.41740084e-01
-8.26516822e-02 -5.68335116e-01 -5.60189784e-01 -5.99182129e-01] | [12.257837295532227, 7.762327671051025] |
fe86d86d-097c-49c8-a1de-07569d7c7b73 | xlid-lexica-cross-lingual-linked-data-lexica | null | null | https://aclanthology.org/L14-1232 | https://aclanthology.org/L14-1232.pdf | xLiD-Lexica: Cross-lingual Linked Data Lexica | In this paper, we introduce our cross-lingual linked data lexica, called xLiD-Lexica, which are constructed by exploiting the multilingual Wikipedia and linked data resources from Linked Open Data (LOD). We provide the cross-lingual groundings of linked data resources from LOD as RDF data, which can be easily integrated into the LOD data sources. In addition, we build a SPARQL endpoint over our xLiD-Lexica to allow users to easily access them using SPARQL query language. Multilingual and cross-lingual information access can be facilitated by the availability of such lexica, e.g., allowing for an easy mapping of natural language expressions in different languages to linked data resources from LOD. Many tasks in natural language processing, such as natural language generation, cross-lingual entity linking, text annotation and question answering, can benefit from our xLiD-Lexica. | ['Michael F{\\"a}rber', 'Achim Rettinger', 'Lei Zhang'] | 2014-05-01 | null | null | null | lrec-2014-5 | ['cross-lingual-entity-linking', 'text-annotation'] | ['natural-language-processing', 'natural-language-processing'] | [-7.95237124e-01 5.39777339e-01 -3.60682517e-01 -4.31477606e-01
-9.94790077e-01 -9.73701000e-01 6.90760374e-01 9.78222132e-01
-3.86951357e-01 1.12631214e+00 6.78473532e-01 -5.35755828e-02
-3.64712119e-01 -1.40040183e+00 -7.53051758e-01 3.41009736e-01
4.68009012e-03 6.86507761e-01 6.16378129e-01 -6.59732938e-01
-4.08343375e-01 -2.49546790e-03 -1.89182520e+00 5.97464323e-01
1.46720850e+00 9.25721705e-01 1.20133497e-01 -8.98263231e-02
-1.48192501e+00 7.00670421e-01 -2.65185148e-01 -9.69652176e-01
-1.76583558e-01 -1.04850315e-01 -1.09901750e+00 -6.77500129e-01
5.11125088e-01 4.94667411e-01 3.27959448e-01 1.29412496e+00
2.99365014e-01 -2.04818785e-01 -1.58284917e-01 -1.35023904e+00
-6.96945250e-01 1.19080317e+00 -2.03334764e-02 -3.87108356e-01
9.79337037e-01 -5.61104596e-01 1.17487991e+00 -8.33742857e-01
1.49811053e+00 1.40270948e+00 5.05565107e-01 2.15236649e-01
-6.28310204e-01 -3.47930163e-01 -2.49498233e-01 1.06417939e-01
-1.50503290e+00 -3.08425039e-01 1.76294878e-01 -3.91715884e-01
8.43394935e-01 4.71379310e-01 4.63755339e-01 4.00211841e-01
-2.20957413e-01 2.79623389e-01 8.86685729e-01 -8.36577475e-01
-1.25907153e-01 3.44522893e-01 1.33564889e-01 9.08221602e-01
8.63691747e-01 -5.82386017e-01 -8.67845058e-01 -2.50575572e-01
1.08576551e-01 -5.19770503e-01 -1.48129657e-01 -4.86594796e-01
-1.23645890e+00 5.66371918e-01 3.78098398e-01 4.70068276e-01
-3.57026875e-01 -3.58940154e-01 7.66229749e-01 8.59708861e-02
4.27237272e-01 4.20313746e-01 -6.51063442e-01 4.77800310e-01
-1.58207908e-01 4.65523720e-01 1.02920341e+00 1.66909051e+00
1.05538607e+00 -5.55080235e-01 1.14994541e-01 9.52050149e-01
6.29403412e-01 5.97625256e-01 5.18580496e-01 -9.68589306e-01
8.61706197e-01 1.07684553e+00 5.27447999e-01 -7.40820646e-01
-4.21455085e-01 3.25569183e-01 -3.70667055e-02 -1.25251666e-01
3.01111281e-01 -4.49958816e-02 -3.15521866e-01 1.53954518e+00
8.89332354e-01 -7.98977077e-01 7.35445023e-01 6.06217682e-01
1.50358891e+00 4.18662280e-01 5.29699206e-01 1.97517276e-02
2.00064206e+00 -4.33353245e-01 -1.30034339e+00 1.77101836e-01
1.24823201e+00 -6.88113272e-01 1.14933658e+00 -3.44561070e-01
-8.96911263e-01 -3.21150303e-01 -8.56671870e-01 -6.35642231e-01
-1.44931746e+00 -5.33946097e-01 6.30230725e-01 1.98091805e-01
-7.25466371e-01 1.05562717e-01 -4.81078386e-01 -9.42922235e-01
3.60036306e-02 -2.51476973e-01 -7.63414264e-01 -1.30587980e-01
-1.99447989e+00 7.92096257e-01 1.16685987e+00 -4.81251687e-01
9.07980744e-03 -9.04678524e-01 -1.42084730e+00 -4.35346901e-01
7.66572595e-01 -6.13629997e-01 7.88335621e-01 -3.89182925e-01
-5.75191736e-01 1.31869256e+00 -2.84352377e-02 -4.52444017e-01
-3.75534147e-02 -8.26135203e-02 -1.30274200e+00 -1.36059567e-01
1.05290806e+00 6.59265935e-01 -1.87518924e-01 -9.21826720e-01
-1.10359454e+00 -6.24062359e-01 4.04606760e-01 9.42969024e-02
-3.36444080e-01 5.65373361e-01 -7.22590208e-01 -4.94326442e-01
-2.03123271e-01 -4.01418328e-01 1.25359342e-01 8.96825194e-02
-3.79856050e-01 -4.88579899e-01 5.71821392e-01 -7.84799278e-01
1.34862471e+00 -1.70728779e+00 -2.15113893e-01 7.89176747e-02
5.87223060e-02 -8.20945725e-02 1.83579028e-01 8.45341444e-01
1.98017761e-01 5.93674541e-01 2.57386379e-02 5.69717348e-01
5.14821410e-01 5.75529873e-01 -1.78419024e-01 -3.46861094e-01
-1.51077524e-01 9.08585846e-01 -1.15109050e+00 -1.04403329e+00
-2.94471830e-01 2.21879572e-01 -3.70838463e-01 -8.81174058e-02
-8.12948525e-01 2.83224642e-01 -8.69495034e-01 6.07621610e-01
3.27117264e-01 -1.36661917e-01 7.12348342e-01 -6.81356788e-01
-6.22047603e-01 6.44018471e-01 -1.12133610e+00 2.36696339e+00
-8.82432580e-01 1.30981117e-01 2.03960091e-02 -4.69007380e-02
9.63489354e-01 6.71955824e-01 6.25252903e-01 -9.87564147e-01
-4.45025951e-01 7.54414499e-01 -7.92716622e-01 -8.80392134e-01
9.88270700e-01 2.51531601e-01 -7.40521967e-01 4.87221390e-01
1.62222013e-01 2.20690459e-01 1.02637386e+00 4.13972139e-01
3.46177012e-01 6.07237041e-01 6.25428736e-01 -5.64500570e-01
7.14989960e-01 5.42032361e-01 5.80158472e-01 1.17578536e-01
5.91996312e-01 -4.99647439e-01 2.78534353e-01 -4.94023174e-01
-1.08744812e+00 -1.02513218e+00 -7.72305608e-01 1.03800559e+00
1.88491847e-02 -1.05495989e+00 -5.89272261e-01 -6.28497303e-01
1.93581253e-01 5.24794161e-01 -2.11079508e-01 7.26002038e-01
-4.40179646e-01 -4.85212594e-01 6.84984207e-01 2.41244629e-01
3.64839137e-01 -9.87002075e-01 -3.68248701e-01 1.28870741e-01
-5.25771737e-01 -1.61508822e+00 -1.81752369e-01 -2.15791881e-01
-1.78920522e-01 -1.17682922e+00 2.87168473e-03 -8.35903525e-01
4.91150141e-01 -3.71631682e-01 1.52502692e+00 -3.87190729e-02
-1.38669789e-01 2.14303240e-01 -5.43036461e-01 -3.97515953e-01
-4.67233330e-01 1.71861425e-01 5.18283853e-03 -3.32553089e-01
6.09914601e-01 -1.07313752e-01 6.38933480e-02 1.51173562e-01
-1.43628228e+00 3.87805887e-02 -4.57175106e-01 1.25715053e-02
1.12013280e+00 2.73879264e-02 8.22217703e-01 -1.42290831e+00
4.51448828e-01 -8.04416478e-01 -1.07902026e+00 7.56210566e-01
-4.00299519e-01 5.82041323e-01 4.21248972e-01 5.18037617e-01
-1.28969967e+00 -1.85349628e-01 -1.96480349e-01 3.25939447e-01
-1.97359696e-01 1.37051463e+00 -9.59169447e-01 2.07434162e-01
4.34610814e-01 -7.08566606e-01 -6.43238127e-01 -9.33301330e-01
1.15557075e+00 6.73668146e-01 6.75585389e-01 -9.64919686e-01
5.63035309e-01 3.84561360e-01 -2.93414056e-01 -3.90535623e-01
-1.14943123e+00 -2.12807611e-01 -8.95704627e-01 -1.64145343e-02
1.27119040e+00 -1.24003482e+00 -3.60719800e-01 -8.50479454e-02
-1.05916035e+00 1.25729755e-01 -5.84794819e-01 2.02643991e-01
-2.22071812e-01 1.35092307e-02 -2.06024051e-01 -1.79950148e-02
-2.91313499e-01 -4.65991914e-01 8.29079628e-01 1.28945649e-01
-3.77499223e-01 -1.33325553e+00 2.49016523e-01 3.97831947e-01
3.97429615e-03 7.39618421e-01 1.41003489e+00 -7.89146960e-01
-2.97101587e-01 -7.69355670e-02 -1.53748646e-01 -4.87201929e-01
2.37428367e-01 5.72879054e-02 -3.67178380e-01 1.36603758e-01
-9.70873117e-01 -4.58926976e-01 -1.98856741e-01 -4.79348600e-01
4.82136250e-01 -6.01143599e-01 -4.28661853e-01 4.77735072e-01
1.80108857e+00 -6.58958852e-02 5.02055109e-01 9.65364218e-01
7.44623244e-01 1.24805689e+00 6.92689955e-01 2.13184744e-01
1.20798385e+00 1.01534677e+00 2.60645241e-01 6.25849888e-02
-2.03228757e-01 -6.94794357e-01 -9.16564539e-02 1.31588972e+00
3.65228117e-01 -1.45755231e-01 -1.16932857e+00 6.53078139e-01
-1.82896864e+00 -7.60250926e-01 -7.12306440e-01 2.18582344e+00
1.16503274e+00 -4.60602701e-01 -6.26214519e-02 -5.43311954e-01
7.93282032e-01 -1.02483649e-02 -1.56132832e-01 -1.16215795e-01
-5.92134714e-01 1.74421877e-01 4.83096033e-01 7.42721796e-01
-7.95772314e-01 1.19056773e+00 5.42707872e+00 5.76905608e-01
-5.88980854e-01 5.91292202e-01 -4.96968687e-01 5.86277284e-02
-9.96874928e-01 4.47872430e-01 -1.39008129e+00 4.37360078e-01
1.08383060e+00 -9.92746115e-01 1.03695534e-01 6.58145666e-01
-6.57988489e-02 2.13187382e-01 -7.01771975e-01 4.72683549e-01
-2.27630720e-01 -1.71952283e+00 2.90854543e-01 1.64678879e-02
7.57833302e-01 2.37495005e-01 -7.08250463e-01 2.41748959e-01
1.02491021e+00 -4.29544121e-01 8.75729561e-01 7.26001740e-01
1.21742690e+00 -6.74686670e-01 5.48909307e-01 -9.17672887e-02
-1.66874754e+00 4.58996415e-01 -3.14075917e-01 8.34133804e-01
5.18817008e-01 8.94549906e-01 -1.22066237e-01 1.31699121e+00
1.01395679e+00 6.31695032e-01 -5.25093615e-01 6.70197666e-01
-3.46997142e-01 -3.65841895e-01 -1.82323232e-01 1.96202517e-01
-1.74470708e-01 -2.34264791e-01 2.73373872e-01 1.09945786e+00
6.59834266e-01 -3.11461359e-01 4.51553345e-01 5.92068732e-01
-6.24903738e-01 9.03675854e-01 -8.42609644e-01 -1.05836034e-01
9.21617091e-01 1.28954470e+00 -2.27079108e-01 -5.61552405e-01
-8.19157302e-01 4.12966549e-01 5.27860403e-01 1.69284731e-01
-5.99793911e-01 -1.01191092e+00 6.73102319e-01 4.35115665e-01
2.86636688e-02 -6.34415969e-02 4.39382941e-01 -1.51838934e+00
3.16766173e-01 -9.28641498e-01 9.57991481e-01 -1.11830950e+00
-1.26398122e+00 8.49587083e-01 2.64858872e-01 -1.21198225e+00
-8.22364748e-01 -3.30216974e-01 2.95154840e-01 8.71726811e-01
-1.61255646e+00 -1.36600828e+00 -1.90503404e-01 8.27369809e-01
-2.53064454e-01 1.44969448e-01 1.21465302e+00 9.10655856e-01
-2.55710244e-01 1.12738786e-02 -1.82739068e-02 3.60205829e-01
8.86185706e-01 -1.14230132e+00 4.46702421e-01 6.97152615e-01
3.33823174e-01 8.48702013e-01 3.54943901e-01 -1.09029543e+00
-1.27834487e+00 -1.43395388e+00 1.72162890e+00 -4.00578052e-01
1.29694664e+00 -3.23750138e-01 -1.27679133e+00 9.56811965e-01
5.44425011e-01 3.38625431e-01 1.04953909e+00 2.30075657e-01
-7.16994107e-01 -2.77919888e-01 -9.85310137e-01 3.73172581e-01
1.12901425e+00 -8.66398752e-01 -5.52762568e-01 6.47883177e-01
8.99838746e-01 -5.54841816e-01 -1.84359813e+00 1.57409236e-01
3.04224610e-01 -4.21371281e-01 6.94887757e-01 -9.20301139e-01
-6.98033348e-02 -7.40228117e-01 -5.10118127e-01 -1.08190203e+00
2.92032123e-01 -3.00698757e-01 4.56738807e-02 1.90455902e+00
7.24098742e-01 -6.43095553e-01 -1.06108617e-02 6.40728533e-01
-1.85197517e-01 2.07807217e-02 -8.97225559e-01 -9.31080997e-01
-1.65105741e-02 -5.90689659e-01 1.51357484e+00 1.43322659e+00
5.54457068e-01 3.15549314e-01 -6.15204871e-02 3.70803416e-01
5.62955558e-01 3.11899126e-01 5.28250098e-01 -1.74121571e+00
2.27683693e-01 1.20591529e-01 -1.29454076e-01 -3.25489074e-01
3.10745478e-01 -1.49425411e+00 -3.61517757e-01 -1.99782455e+00
-3.75439495e-01 -9.25700605e-01 1.84175640e-01 7.11104035e-01
1.60058051e-01 9.96159315e-02 5.86770549e-02 4.03046817e-01
-8.86839151e-01 2.59094149e-01 1.04427087e+00 1.12553820e-01
1.52920946e-01 -1.03114855e+00 -7.43326724e-01 7.88210809e-01
4.00517613e-01 -8.85790050e-01 -1.01787105e-01 -9.86899078e-01
1.22661352e+00 2.29791820e-01 -2.16349233e-02 -7.87262619e-01
3.47522140e-01 -1.38906673e-01 -2.84859419e-01 -3.82803082e-01
-8.07488710e-02 -8.04660261e-01 6.47894204e-01 -1.02567330e-01
-3.71293843e-01 2.08429113e-01 1.53077498e-01 6.74195737e-02
-5.51490605e-01 -5.63615322e-01 2.29012981e-01 -3.93392205e-01
-9.91764069e-01 1.81960464e-01 -4.75601107e-02 7.91550398e-01
8.81385744e-01 4.78981733e-01 -8.78251910e-01 1.71701550e-01
-8.38234067e-01 5.65537632e-01 7.99738348e-01 7.48647809e-01
-2.62854576e-01 -1.88237607e+00 -5.61225414e-01 -2.25002263e-02
8.01763415e-01 -7.62824118e-02 -5.92910089e-02 3.58239114e-01
-6.57408655e-01 8.46164167e-01 -3.74932945e-01 5.34882164e-03
-7.30876505e-01 6.28586829e-01 1.93260685e-01 -4.64859903e-01
-4.18782651e-01 -1.20273054e-01 -4.93733495e-01 -1.13187253e+00
-1.79125652e-01 4.56760749e-02 -5.25758505e-01 4.58818376e-01
7.17126489e-01 4.03078608e-02 3.33576024e-01 -1.08789456e+00
-6.20714366e-01 6.59408092e-01 4.62707102e-01 -2.84858733e-01
1.04081738e+00 -4.46266055e-01 -7.81004846e-01 5.25917768e-01
9.68117118e-01 8.26947331e-01 -2.03222245e-01 -4.06650722e-01
7.95050442e-01 -2.73840457e-01 -2.14487210e-01 -6.87596738e-01
-8.57218564e-01 2.00315386e-01 1.53477609e-01 5.31310439e-01
6.19600892e-01 5.72478235e-01 8.14750791e-01 3.85075957e-01
9.46995556e-01 -9.85307872e-01 -8.47044826e-01 4.80545878e-01
8.28856826e-01 -9.45346475e-01 -1.49530485e-01 -5.66430569e-01
-4.79658008e-01 9.92932200e-01 4.27643329e-01 7.55663812e-01
5.60987055e-01 3.43950182e-01 6.65365517e-01 -8.23688269e-01
-7.06840217e-01 -7.63588727e-01 3.36060017e-01 8.76458168e-01
8.84775221e-01 -1.34335682e-01 -7.13556051e-01 5.98341763e-01
-2.61352897e-01 1.68799371e-01 1.97969347e-01 1.00098681e+00
-4.06924039e-01 -1.78804994e+00 -3.47357213e-01 6.15585186e-02
-6.37719095e-01 -3.76796037e-01 -5.73820025e-02 8.82449090e-01
6.95122123e-01 5.99329591e-01 4.32125360e-01 2.28111684e-01
5.35039067e-01 3.28463107e-01 1.62094668e-01 -8.08803737e-01
-4.52810764e-01 -4.72123384e-01 7.25048602e-01 -5.05229056e-01
-7.94824243e-01 -5.07690370e-01 -1.75504887e+00 -7.40832984e-02
9.45433751e-02 7.93358982e-01 8.62121820e-01 7.09604561e-01
6.40160501e-01 2.37845391e-01 -1.76197112e-01 1.62348181e-01
5.32142460e-01 -3.61953855e-01 -4.81680214e-01 4.75064039e-01
-4.64591026e-01 -3.46588463e-01 3.54927778e-01 4.44436401e-01] | [9.405694961547852, 8.482860565185547] |
4f36b777-03c4-4eff-b98b-08eaf7234980 | towards-a-query-optimal-and-time-efficient | 2106.10374 | null | https://arxiv.org/abs/2106.10374v1 | https://arxiv.org/pdf/2106.10374v1.pdf | Towards a Query-Optimal and Time-Efficient Algorithm for Clustering with a Faulty Oracle | Motivated by applications in crowdsourced entity resolution in database, signed edge prediction in social networks and correlation clustering, Mazumdar and Saha [NIPS 2017] proposed an elegant theoretical model for studying clustering with a faulty oracle. In this model, given a set of $n$ items which belong to $k$ unknown groups (or clusters), our goal is to recover the clusters by asking pairwise queries to an oracle. This oracle can answer the query that ``do items $u$ and $v$ belong to the same cluster?''. However, the answer to each pairwise query errs with probability $\varepsilon$, for some $\varepsilon\in(0,\frac12)$. Mazumdar and Saha provided two algorithms under this model: one algorithm is query-optimal while time-inefficient (i.e., running in quasi-polynomial time), the other is time efficient (i.e., in polynomial time) while query-suboptimal. Larsen, Mitzenmacher and Tsourakakis [WWW 2020] then gave a new time-efficient algorithm for the special case of $2$ clusters, which is query-optimal if the bias $\delta:=1-2\varepsilon$ of the model is large. It was left as an open question whether one can obtain a query-optimal, time-efficient algorithm for the general case of $k$ clusters and other regimes of $\delta$. In this paper, we make progress on the above question and provide a time-efficient algorithm with nearly-optimal query complexity (up to a factor of $O(\log^2 n)$) for all constant $k$ and any $\delta$ in the regime when information-theoretic recovery is possible. Our algorithm is built on a connection to the stochastic block model. | ['Jiapeng Zhang', 'Pan Peng'] | 2021-06-18 | null | null | null | null | ['stochastic-block-model', 'entity-resolution'] | ['graphs', 'natural-language-processing'] | [-1.38089523e-01 3.09204072e-01 -1.28776327e-01 -8.44319072e-03
-1.21352839e+00 -8.68330002e-01 -2.42367581e-01 5.45496941e-01
-5.74573874e-01 7.22452700e-01 -4.92471606e-01 -3.70059073e-01
-6.86657846e-01 -1.26689327e+00 -1.01529038e+00 -8.48079383e-01
-5.87592423e-01 1.02140081e+00 4.21148509e-01 -1.42373621e-01
1.94718093e-01 2.03459665e-01 -1.43103385e+00 -2.34276816e-01
5.78246295e-01 1.02232659e+00 -1.51875094e-01 9.48545635e-01
7.20654102e-03 5.50131738e-01 -3.81037951e-01 -8.31383407e-01
6.70119584e-01 -5.65381467e-01 -1.14305735e+00 -9.95105803e-02
-1.64914683e-01 2.32013911e-01 -2.84798890e-01 1.39747918e+00
3.69368762e-01 -8.14088900e-03 3.28991890e-01 -1.43040061e+00
-4.53102052e-01 9.55375910e-01 -6.78066254e-01 1.37476176e-01
5.03673911e-01 -1.91615462e-01 1.13061070e+00 -5.19895554e-01
7.09610879e-01 7.72851169e-01 6.13858223e-01 3.18510950e-01
-1.21724045e+00 -8.73891830e-01 -2.37490371e-01 1.96860313e-01
-1.93292904e+00 -1.32200971e-01 3.33525270e-01 -1.77507341e-01
4.99530762e-01 6.05171382e-01 3.02899718e-01 -2.23169550e-02
-4.94479299e-01 5.41055381e-01 1.01117027e+00 -5.38047731e-01
5.32545686e-01 1.94464877e-01 6.08731667e-03 6.72011971e-01
7.10815907e-01 -1.69965744e-01 -6.00901306e-01 -5.88205040e-01
3.45323652e-01 -2.45915353e-02 -2.03626022e-01 -2.49732479e-01
-8.21695447e-01 8.13622415e-01 3.45147878e-01 3.80146533e-01
-1.48196518e-01 4.75602090e-01 3.40875946e-02 6.74172163e-01
3.17446858e-01 1.26287878e-01 -5.34645796e-01 3.36303711e-02
-8.39617074e-01 3.88302296e-01 1.22380292e+00 1.30661523e+00
1.08595347e+00 -5.21788538e-01 1.39890507e-01 1.77870631e-01
1.44319907e-01 8.94147038e-01 -1.02810368e-01 -1.31876707e+00
6.17513776e-01 5.92090666e-01 5.88178575e-01 -1.00267756e+00
-2.20372647e-01 -1.20586166e-02 -8.40540290e-01 -1.07649900e-01
8.78980458e-01 -1.24489874e-01 -2.33478829e-01 1.91668689e+00
4.83443171e-01 -8.91881511e-02 -1.02275042e-02 8.40251565e-01
2.07965821e-01 3.37476909e-01 -2.99215525e-01 -5.92111886e-01
1.52842307e+00 -3.45862657e-01 -4.13635492e-01 1.47635520e-01
8.70709181e-01 -5.58541536e-01 7.12612391e-01 4.04159576e-01
-1.39823973e+00 5.55294119e-02 -7.28104055e-01 1.92706510e-01
-3.33140373e-01 -4.54658926e-01 6.10238969e-01 9.41631198e-01
-1.28691459e+00 3.40740770e-01 -5.79036236e-01 -3.37321877e-01
3.62369031e-01 7.10000277e-01 -3.48557949e-01 -3.82147759e-01
-1.03512681e+00 1.79978952e-01 -4.12958767e-03 -2.50848651e-01
-4.86710489e-01 -2.20772982e-01 -4.96816427e-01 -7.73950815e-02
8.29088748e-01 -5.33466458e-01 9.30274904e-01 -7.53554344e-01
-6.64815605e-01 1.05420744e+00 -6.33576512e-01 -6.68266952e-01
5.77410936e-01 5.94698608e-01 -3.69108245e-02 4.07968342e-01
3.77995878e-01 4.62330058e-02 1.87751010e-01 -1.16132724e+00
-8.21942508e-01 -8.78445804e-01 2.68196762e-01 9.19349268e-02
3.12480610e-02 1.98582053e-01 -5.00330567e-01 -4.11014520e-02
4.66814190e-01 -1.14243972e+00 -6.01678312e-01 -8.78308713e-02
-3.02398413e-01 -4.18871880e-01 2.00702175e-01 -2.82057554e-01
1.03701341e+00 -1.99863505e+00 -2.13455319e-01 7.20565140e-01
4.38608855e-01 -1.16189159e-01 1.88660026e-01 6.82624161e-01
2.76579082e-01 6.61122978e-01 -4.56834167e-01 -3.20558459e-01
1.45479009e-01 1.48565978e-01 1.06923878e-01 7.87744403e-01
-4.48123932e-01 6.24434590e-01 -9.23754752e-01 -4.84929711e-01
-4.42831367e-01 -9.35008228e-02 -5.36517441e-01 -2.30459739e-02
-2.22305849e-01 1.07623905e-01 -3.37940663e-01 7.77853727e-01
9.22022700e-01 -6.03738189e-01 4.45766211e-01 4.15622473e-01
6.37575984e-02 -9.60740224e-02 -1.83126235e+00 1.18417346e+00
1.50570229e-01 2.69436389e-01 6.30004287e-01 -1.12264359e+00
5.42088330e-01 4.51639950e-01 6.82647526e-01 -4.43732232e-01
3.61617096e-02 6.34703815e-01 -3.66897494e-01 -4.63236868e-01
5.15357614e-01 -4.10537869e-01 -4.49121416e-01 8.37073445e-01
-4.91438597e-01 2.37924486e-01 1.55984297e-01 4.55463171e-01
1.73607779e+00 -7.21474111e-01 -3.73352952e-02 -2.45993003e-01
2.96485841e-01 8.54527131e-02 4.93386090e-01 1.21990848e+00
-3.35378766e-01 5.39422989e-01 7.61405170e-01 -1.05030477e-01
-8.42464983e-01 -8.28801394e-01 2.24195734e-01 1.03175020e+00
5.81922054e-01 -4.12288606e-01 -8.59628499e-01 -5.45570493e-01
4.68870401e-02 1.00971505e-01 -6.04780257e-01 2.63151050e-01
-3.69963735e-01 -6.60156429e-01 6.33532405e-01 2.98895240e-01
5.80482662e-01 -7.51426041e-01 -3.10525835e-01 1.43963009e-01
-4.24887776e-01 -1.00695539e+00 -4.45101529e-01 2.21076861e-01
-6.93986118e-01 -1.45045221e+00 -3.54584664e-01 -6.36333823e-01
8.01812887e-01 5.67884684e-01 1.02516651e+00 3.72894287e-01
-1.44102871e-02 4.71415997e-01 -3.70536327e-01 -4.05261070e-01
-5.66369519e-02 7.76317939e-02 -1.06619345e-02 -1.40792266e-01
6.37024879e-01 -3.63352597e-01 -8.91798019e-01 3.94680023e-01
-1.01760125e+00 -8.18599761e-01 1.35148212e-01 3.89817566e-01
8.88221741e-01 4.48937237e-01 4.50675011e-01 -1.21284759e+00
2.87119299e-01 -8.40714693e-01 -9.69568729e-01 2.70383120e-01
-8.45730603e-01 -8.35059658e-02 7.52905011e-01 -4.96188886e-02
-1.34741247e-01 2.10829005e-01 4.87914169e-03 -2.70150065e-01
7.72826821e-02 4.45763797e-01 -1.67857349e-01 -1.10282369e-01
6.01916790e-01 1.83430195e-01 -1.13456286e-01 -1.80893958e-01
2.59213179e-01 6.07190073e-01 4.27641928e-01 -5.09993851e-01
8.31017554e-01 9.80823278e-01 2.27794170e-01 -4.92413878e-01
-3.18614542e-01 -7.60373175e-01 -1.29125491e-01 2.07305133e-01
4.32919830e-01 -8.74702871e-01 -1.41581881e+00 1.11343637e-01
-8.53987277e-01 -2.68771410e-01 -5.43813467e-01 1.22193985e-01
-5.51662982e-01 6.03223860e-01 -4.20917779e-01 -1.39689136e+00
-8.67483243e-02 -8.15502346e-01 6.75057232e-01 8.07731003e-02
8.08112323e-02 -7.08381772e-01 -6.20495453e-02 5.42754650e-01
3.00954878e-01 1.21463902e-01 6.46350026e-01 -9.83047068e-01
-9.29487407e-01 -6.32596612e-01 -2.81938136e-01 -2.64341623e-01
-4.37058121e-01 -4.02179778e-01 -6.56458080e-01 -3.59593928e-01
-7.59775983e-04 -1.22061642e-02 4.54578370e-01 2.50725269e-01
1.08123255e+00 -8.11738253e-01 -4.32264328e-01 2.27647066e-01
1.74141479e+00 8.94633457e-02 6.26920640e-01 4.20395136e-02
2.45613545e-01 5.33542812e-01 4.89593267e-01 6.04641795e-01
8.39440048e-01 3.40637207e-01 3.63673717e-01 2.77751803e-01
5.81683457e-01 -2.04676181e-01 5.13194576e-02 5.40209949e-01
-1.30779043e-01 -4.11718667e-01 -6.74326062e-01 1.04394114e+00
-1.91057241e+00 -8.61651421e-01 -3.38765800e-01 2.55910110e+00
8.00100327e-01 -5.40961437e-02 4.64517504e-01 2.57218629e-01
9.71706510e-01 -3.40962142e-01 -4.99888271e-01 -1.98103979e-01
-3.17949653e-01 4.90230322e-01 1.17362392e+00 5.37060201e-01
-6.62422895e-01 7.23909438e-01 5.11222553e+00 8.15882742e-01
-4.87131655e-01 5.08626163e-01 5.86720169e-01 -1.08646020e-01
-5.24420381e-01 4.66504514e-01 -5.62698185e-01 6.80933356e-01
1.33962369e+00 -4.06516612e-01 7.40006328e-01 7.23704278e-01
-1.10487558e-01 -6.11716330e-01 -9.51977849e-01 1.07959449e+00
-1.20856963e-01 -1.17720830e+00 -7.75332451e-01 4.55231577e-01
8.16611111e-01 -2.97934581e-02 -2.35192731e-01 4.85160500e-02
8.34134698e-01 -8.50269198e-01 5.75564861e-01 1.24672323e-01
7.85384297e-01 -9.61477578e-01 9.35657382e-01 8.20952773e-01
-1.33772230e+00 -2.04033703e-01 -3.83755535e-01 -1.64680043e-03
2.17975932e-03 6.05858922e-01 -5.45930386e-01 7.18187749e-01
1.06242859e+00 -1.81881413e-01 -1.47153750e-01 9.76272464e-01
3.34002823e-01 5.25719523e-01 -8.64944160e-01 -1.84558094e-01
-6.24754243e-02 -2.20931143e-01 3.83201063e-01 9.82774913e-01
5.24243355e-01 8.09116304e-01 1.37462646e-01 3.63490820e-01
-5.76421320e-01 2.08781332e-01 -7.79087305e-01 1.21086404e-01
1.11097825e+00 8.83714736e-01 -1.04675674e+00 -2.96132118e-01
-5.48267029e-02 8.03013504e-01 2.51837015e-01 1.76699504e-01
-5.28605044e-01 -5.77842653e-01 2.44582042e-01 5.93943000e-01
6.11306548e-01 -1.70132607e-01 -3.94549817e-01 -8.68756771e-01
3.34418416e-01 -5.77732086e-01 8.66500020e-01 -2.15873703e-01
-1.25491869e+00 2.86032170e-01 -2.25651935e-01 -9.42248702e-01
4.62092347e-02 -2.50131398e-01 -1.97526082e-01 6.60772085e-01
-1.15035415e+00 -6.01029098e-01 7.89571106e-02 9.34089184e-01
-3.14293861e-01 3.59777182e-01 7.76207209e-01 3.08593035e-01
-6.28581792e-02 9.05507922e-01 4.09900397e-01 4.36859615e-02
4.72817183e-01 -1.34852648e+00 -3.29683274e-01 8.54475498e-01
2.91869626e-04 5.53825915e-01 7.98788726e-01 -4.28979665e-01
-1.68116856e+00 -9.67270195e-01 1.36626697e+00 -4.95387048e-01
3.29375684e-01 -4.38051462e-01 -6.64957404e-01 5.12285292e-01
-4.17786241e-02 3.63269895e-01 8.92151952e-01 -1.46974787e-01
-3.26142639e-01 -4.20345485e-01 -1.70476282e+00 1.65811375e-01
1.13612211e+00 -7.42398441e-01 2.17787579e-01 5.41584074e-01
6.59311295e-01 -2.40775302e-01 -9.31332707e-01 1.14659145e-01
2.86293298e-01 -1.08112884e+00 6.23032510e-01 -3.10588062e-01
-1.77428737e-01 -5.08942187e-01 -5.05019069e-01 -4.84739065e-01
2.20792070e-02 -1.02309787e+00 -8.47123265e-02 8.96873713e-01
5.83204091e-01 -7.30519950e-01 1.03846478e+00 8.52168024e-01
5.40894151e-01 -6.66878104e-01 -1.44578779e+00 -6.71946049e-01
1.86750695e-01 -6.15567267e-01 6.77056730e-01 9.00056541e-01
1.97479263e-01 -8.80453959e-02 5.13728615e-03 4.28782523e-01
7.19462752e-01 1.99409276e-01 8.19508135e-01 -1.20517325e+00
-4.10682201e-01 -1.01898074e-01 -1.91057906e-01 -9.34676290e-01
-1.27378613e-01 -7.70571172e-01 2.48290345e-01 -1.26824093e+00
3.47865880e-01 -1.04721248e+00 -6.35201111e-02 2.46001497e-01
1.23534359e-01 3.15800935e-01 1.76252931e-01 4.90134031e-01
-1.26138997e+00 -1.75356895e-01 6.17348611e-01 5.78956343e-02
-1.92559585e-02 3.59879047e-01 -1.02003777e+00 3.45244944e-01
5.38486719e-01 -8.63470733e-01 -1.71643957e-01 -2.03428045e-01
9.42013323e-01 7.17764556e-01 2.66425818e-01 -6.74261212e-01
8.53373408e-01 1.76411867e-02 -3.75338882e-01 -6.24169111e-01
6.11635335e-02 -9.31839824e-01 4.60411936e-01 6.11602783e-01
-3.32625449e-01 2.66491443e-01 -4.43363667e-01 1.13351429e+00
7.03773573e-02 -3.56511414e-01 6.29182875e-01 -3.89864564e-01
4.21114825e-02 5.17379165e-01 -3.51479411e-01 3.11614186e-01
1.19038594e+00 -2.95322508e-01 -3.15702498e-01 -7.02778935e-01
-7.19710350e-01 4.42129076e-01 7.34842360e-01 -3.86891603e-01
2.01581419e-01 -1.00153208e+00 -6.63483083e-01 -1.33781105e-01
1.14953242e-01 4.54691529e-01 2.60514975e-01 1.04926479e+00
-4.07168597e-01 2.38549113e-01 6.88920379e-01 -5.20028114e-01
-1.01391792e+00 6.95658863e-01 2.21817404e-01 -4.91247892e-01
1.91220924e-01 1.01254749e+00 -4.81521606e-01 -3.64654005e-01
2.89139509e-01 -1.83608420e-02 5.37991285e-01 -1.05975546e-01
4.04548973e-01 6.10784411e-01 1.28418341e-01 -4.98897582e-01
-4.43892449e-01 3.28449875e-01 2.97909498e-01 -5.27630031e-01
1.08168185e+00 -5.51976204e-01 -5.12346208e-01 2.82385111e-01
1.22357619e+00 3.27365994e-01 -6.31741047e-01 -5.80878437e-01
2.32760534e-01 -3.27293068e-01 -7.12374270e-01 -3.67249787e-01
-1.11096692e+00 4.20780569e-01 6.19643271e-01 1.00107682e+00
1.09803283e+00 4.25495267e-01 7.42694378e-01 4.80482966e-01
1.07822788e+00 -1.10947907e+00 -3.22310120e-01 1.71065584e-01
1.47903055e-01 -1.18111002e+00 -1.85171023e-01 -3.67010057e-01
-3.54849279e-01 7.02283025e-01 3.23157817e-01 -1.86644047e-01
8.51290405e-01 1.96615785e-01 -3.40853870e-01 -3.88833046e-01
-7.35626757e-01 -3.69171381e-01 -4.78737026e-01 4.24542129e-01
-1.52021237e-02 3.39669853e-01 -6.57287121e-01 1.04079580e+00
-2.15899512e-01 3.20573635e-02 4.60608870e-01 1.02133977e+00
-5.56047618e-01 -1.10653222e+00 -5.26291728e-01 5.01311481e-01
-8.29970598e-01 2.81138197e-02 -3.40814352e-01 6.68686986e-01
4.48547751e-01 1.32697177e+00 -3.13856713e-02 -3.19122344e-01
1.87154263e-01 -5.69162592e-02 2.28492424e-01 -4.50286329e-01
-6.87668145e-01 -1.14402743e-02 -1.33391306e-01 -4.34437692e-01
-5.93364954e-01 -6.79824233e-01 -1.60102427e+00 -7.76948750e-01
-5.64136267e-01 7.87903070e-01 5.52738070e-01 9.63039696e-01
4.54990774e-01 -5.55954397e-01 9.83663082e-01 6.06501848e-02
-4.27858084e-01 -6.50123775e-01 -9.70534086e-01 1.42621323e-01
-5.50620072e-03 -3.79365012e-02 -6.67282522e-01 -1.42791122e-03] | [6.783349514007568, 5.000453948974609] |
554d94fb-d3d5-41af-b366-8046a1c64bcd | gait-recognition-in-the-wild-with-dense-3d | 2204.02569 | null | https://arxiv.org/abs/2204.02569v1 | https://arxiv.org/pdf/2204.02569v1.pdf | Gait Recognition in the Wild with Dense 3D Representations and A Benchmark | Existing studies for gait recognition are dominated by 2D representations like the silhouette or skeleton of the human body in constrained scenes. However, humans live and walk in the unconstrained 3D space, so projecting the 3D human body onto the 2D plane will discard a lot of crucial information like the viewpoint, shape, and dynamics for gait recognition. Therefore, this paper aims to explore dense 3D representations for gait recognition in the wild, which is a practical yet neglected problem. In particular, we propose a novel framework to explore the 3D Skinned Multi-Person Linear (SMPL) model of the human body for gait recognition, named SMPLGait. Our framework has two elaborately-designed branches of which one extracts appearance features from silhouettes, the other learns knowledge of 3D viewpoints and shapes from the 3D SMPL model. In addition, due to the lack of suitable datasets, we build the first large-scale 3D representation-based gait recognition dataset, named Gait3D. It contains 4,000 subjects and over 25,000 sequences extracted from 39 cameras in an unconstrained indoor scene. More importantly, it provides 3D SMPL models recovered from video frames which can provide dense 3D information of body shape, viewpoint, and dynamics. Based on Gait3D, we comprehensively compare our method with existing gait recognition approaches, which reflects the superior performance of our framework and the potential of 3D representations for gait recognition in the wild. The code and dataset are available at https://gait3d.github.io. | ['Tao Mei', 'Chenggang Yan', 'Lingxiao He', 'Wu Liu', 'Xinchen Liu', 'Jinkai Zheng'] | 2022-04-06 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Zheng_Gait_Recognition_in_the_Wild_With_Dense_3D_Representations_and_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Zheng_Gait_Recognition_in_the_Wild_With_Dense_3D_Representations_and_CVPR_2022_paper.pdf | cvpr-2022-1 | ['gait-recognition-in-the-wild'] | ['computer-vision'] | [-2.72907972e-01 -4.51790154e-01 -1.35416791e-01 -2.12938013e-03
-5.06605208e-02 -2.08918348e-01 2.62733281e-01 -6.37799323e-01
-1.17399618e-01 2.66733021e-01 4.41047132e-01 2.85805196e-01
2.64254689e-01 -5.54148734e-01 -3.52553904e-01 -8.59350622e-01
-2.83339292e-01 5.46208918e-01 1.08500943e-01 -1.98989153e-01
-2.12688297e-01 5.56673050e-01 -1.51930475e+00 -2.94495881e-01
3.61140102e-01 8.35597873e-01 -1.86866730e-01 6.02990150e-01
3.63384336e-01 2.02122554e-01 -3.90115827e-01 -3.04050833e-01
2.94101596e-01 -2.94052303e-01 -4.16219205e-01 6.38005257e-01
4.21203792e-01 -5.16152620e-01 -9.09261644e-01 6.21357381e-01
6.24582529e-01 1.05981775e-01 7.43180633e-01 -1.33175004e+00
-4.40840960e-01 -4.41289932e-01 -7.03742027e-01 6.95049539e-02
8.15153897e-01 3.88641179e-01 6.97466075e-01 -8.17065120e-01
5.91457903e-01 1.39490724e+00 8.20490181e-01 6.84415936e-01
-8.53347719e-01 -3.51123244e-01 9.72106084e-02 2.49196276e-01
-1.39387023e+00 -3.66563290e-01 8.64359438e-01 -5.31139672e-01
8.46935689e-01 1.09617561e-01 1.24704146e+00 1.42397952e+00
1.94744751e-01 1.03291976e+00 7.27597117e-01 -2.37500250e-01
-1.46422219e-02 -6.75607443e-01 1.03729345e-01 9.16706324e-01
6.14461780e-01 1.21814452e-01 -4.81199175e-01 -1.15063250e-01
1.28418887e+00 2.56995410e-01 -2.83873141e-01 -7.83669233e-01
-1.36976433e+00 3.04299295e-01 2.81063735e-01 5.84082259e-03
-2.17526242e-01 1.74667403e-01 3.81599963e-01 1.41461909e-01
2.57880330e-01 -4.77661133e-01 -5.74169457e-02 -3.58587325e-01
-6.59421921e-01 4.30033535e-01 6.87780857e-01 8.87780666e-01
4.29160029e-01 2.98471659e-01 2.25230828e-01 7.87906349e-01
6.40594602e-01 1.03768921e+00 5.20687878e-01 -8.00880432e-01
5.90235353e-01 7.44350493e-01 -1.62477233e-02 -1.39783704e+00
-5.05799234e-01 -5.33674192e-03 -1.07097006e+00 -1.37618944e-01
4.71450090e-01 2.11366177e-01 -8.20071340e-01 1.38404489e+00
4.76079375e-01 1.28700167e-01 -3.04522723e-01 1.33967960e+00
8.74942243e-01 2.88247138e-01 -2.70816237e-01 1.00985609e-01
1.42705381e+00 -8.42317104e-01 -3.95754665e-01 -3.39905173e-01
4.69116628e-01 -4.10953343e-01 7.87797987e-01 1.43227845e-01
-8.57924342e-01 -6.05985224e-01 -8.84721935e-01 -1.11094723e-02
-1.03707192e-02 3.21708381e-01 4.50090885e-01 7.69475579e-01
-7.54321218e-01 4.46093321e-01 -1.23369539e+00 -7.98968196e-01
2.73977548e-01 2.55325228e-01 -7.06441164e-01 -3.80974650e-01
-9.21374500e-01 7.80288815e-01 -4.05605808e-02 3.57433259e-01
-4.87060368e-01 -4.33842763e-02 -1.28384566e+00 -3.73039305e-01
3.31354231e-01 -1.19200838e+00 7.14043617e-01 -1.06607229e-01
-1.33986914e+00 1.32692683e+00 -1.54511884e-01 -4.07790467e-02
8.07068229e-01 -4.55811322e-01 -2.55104721e-01 2.40608856e-01
6.13819659e-02 5.21850698e-02 1.04646015e+00 -8.49683583e-01
1.42362058e-01 -8.30372989e-01 -2.44216442e-01 1.88056588e-01
-9.28670983e-04 -2.26499021e-01 -8.74718368e-01 -8.32658648e-01
4.12521541e-01 -1.05604291e+00 -1.51558995e-01 2.77424335e-01
-3.95877659e-01 3.51719446e-02 7.49181151e-01 -9.19140637e-01
9.49361265e-01 -2.20619631e+00 4.10574496e-01 6.51737442e-03
1.82124645e-01 2.82579392e-01 1.08777760e-02 3.50376129e-01
1.07127532e-01 -1.11943334e-01 -1.22535728e-01 -6.26524329e-01
2.20586166e-01 5.53614199e-01 9.01821163e-03 9.13001776e-01
1.69758350e-01 8.76508296e-01 -8.28109503e-01 -6.60768807e-01
4.79223847e-01 6.12163365e-01 -6.38024449e-01 2.38360420e-01
3.91930044e-01 6.88494563e-01 -7.71186888e-01 1.22534740e+00
7.53549397e-01 -1.67609423e-01 7.95329884e-02 -3.44586611e-01
2.31656298e-01 -1.47831485e-01 -1.34914017e+00 1.75190282e+00
1.12900749e-01 3.50420289e-02 -9.63874981e-02 -1.14140952e+00
1.14033151e+00 3.03671151e-01 1.01306999e+00 -4.88519609e-01
2.15325862e-01 2.41151869e-01 -4.15048689e-01 -5.27203798e-01
1.63569570e-01 5.81190102e-02 -3.30273509e-01 2.70814568e-01
1.23984523e-01 8.19238126e-02 1.18220925e-01 -1.87111452e-01
1.05475485e+00 6.26053274e-01 6.06404543e-01 3.25383246e-02
5.51089585e-01 -3.38394433e-01 7.92458415e-01 3.61895412e-01
-6.88215435e-01 9.62759852e-01 2.47811675e-01 -7.93582916e-01
-9.66395259e-01 -1.44084752e+00 1.66601270e-01 2.74014622e-01
2.27328315e-01 -5.41134715e-01 -4.27972972e-01 -4.08977538e-01
2.15573296e-01 -3.46585035e-01 -4.19349372e-01 -3.15827250e-01
-9.92957830e-01 -7.06611812e-01 5.64172447e-01 7.14329660e-01
8.55746269e-01 -7.09366858e-01 -7.51252770e-01 2.88880318e-02
-5.20647287e-01 -1.31610763e+00 -8.22035193e-01 -4.51040894e-01
-1.04491568e+00 -1.38707387e+00 -1.06062865e+00 -6.23573005e-01
6.63547754e-01 3.28209132e-01 9.90643084e-01 1.95215404e-01
-5.90118766e-01 7.20551133e-01 -3.78708690e-01 2.73697883e-01
2.27937698e-01 -3.38452548e-01 6.83731675e-01 1.24003805e-01
3.62922966e-01 -8.50912273e-01 -6.55777335e-01 6.87574327e-01
-2.39292756e-01 -5.80228912e-03 3.59692365e-01 7.38264441e-01
6.30723596e-01 -1.73302025e-01 -5.15533499e-02 -1.06862441e-01
-3.26125063e-02 -3.12519670e-01 -1.13454692e-01 -1.08789159e-02
1.74512371e-01 -2.91489780e-01 3.54593784e-01 -4.26691502e-01
-5.65475225e-01 2.72230536e-01 -1.74595177e-01 -8.46120298e-01
-3.26083153e-01 2.32100263e-01 -3.95640492e-01 8.60089287e-02
2.60272980e-01 5.17511010e-01 3.37003499e-01 -8.43228519e-01
2.08047964e-02 2.92867303e-01 6.50757670e-01 -8.44799459e-01
1.02126586e+00 6.83455825e-01 2.89774746e-01 -1.07738805e+00
-4.87756610e-01 -6.19538307e-01 -1.30106866e+00 -5.11246383e-01
6.34823978e-01 -9.83979523e-01 -8.54487598e-01 9.34334397e-01
-7.86322355e-01 -8.73133391e-02 -2.04245344e-01 5.83319426e-01
-1.06316185e+00 9.60907638e-01 -6.57861412e-01 -8.67551208e-01
-3.86375874e-01 -8.58556092e-01 1.39647388e+00 -1.37933008e-02
-3.33179861e-01 -8.47821772e-01 1.65189087e-01 5.75231373e-01
-1.11461461e-01 7.56600499e-01 5.47357440e-01 -1.01049952e-01
-3.60167891e-01 -4.05623376e-01 9.69976634e-02 2.18808785e-01
3.31876785e-01 -4.46184464e-02 -5.90255857e-01 -4.54342127e-01
-9.66122150e-02 -2.69688576e-01 7.98707128e-01 4.82315391e-01
6.12522066e-01 -1.46956161e-01 -2.97165245e-01 8.15090656e-01
9.61406946e-01 -2.17419446e-01 5.80298007e-01 3.07335496e-01
9.89209950e-01 4.57144797e-01 5.92120767e-01 6.91602647e-01
6.32069170e-01 1.01550949e+00 2.70354837e-01 2.18191162e-01
-3.44314367e-01 -4.80148882e-01 7.69671559e-01 1.23506343e+00
-7.34657228e-01 1.44405887e-01 -9.70947087e-01 3.85314643e-01
-1.85369360e+00 -9.57972348e-01 -3.02744713e-02 2.29505396e+00
1.87273294e-01 -5.89784011e-02 7.12810695e-01 3.65876883e-01
6.99124038e-01 3.22334737e-01 -7.60281920e-01 2.36034915e-01
-8.67764428e-02 -6.12991415e-02 4.96504717e-02 -1.60740931e-02
-1.24459755e+00 6.17154300e-01 5.74428654e+00 3.61076385e-01
-7.69138932e-01 -2.72164166e-01 -2.68119648e-02 -6.92638084e-02
4.12431210e-01 -3.39840263e-01 -8.64861488e-01 5.15226662e-01
5.18797755e-01 -1.65937424e-01 2.99666412e-02 8.65373671e-01
3.44584495e-01 2.41854146e-01 -1.06395817e+00 1.47549355e+00
2.05956087e-01 -8.14048171e-01 2.56650388e-01 4.62777793e-01
2.66426504e-01 -3.03822368e-01 -1.01027988e-01 2.55320817e-01
-1.10396765e-01 -7.64707267e-01 5.75326562e-01 5.69081962e-01
8.08993459e-01 -4.56430137e-01 6.24089062e-01 4.74698931e-01
-1.80084443e+00 1.44141406e-01 -3.74059707e-01 -1.66981429e-01
3.73817801e-01 4.18083936e-01 -1.86429113e-01 8.17858458e-01
8.62955809e-01 1.57099724e+00 -4.56224203e-01 1.18049240e+00
-1.20477632e-01 5.09397864e-01 -4.58495140e-01 3.50689769e-01
-1.65591925e-01 -4.36041713e-01 6.79416299e-01 9.83799458e-01
5.62716424e-01 1.64381951e-01 5.27560234e-01 7.37026393e-01
2.43380174e-01 -9.92087424e-02 -8.34662855e-01 -1.31015152e-01
5.42799979e-02 8.02305818e-01 -5.53681731e-01 -2.04259437e-02
-4.20431107e-01 1.05643177e+00 -9.87726748e-02 2.33678937e-01
-6.75159693e-01 1.49757311e-01 1.06617224e+00 2.37931833e-01
3.22517276e-01 -7.74778724e-01 9.33243427e-03 -1.88009405e+00
4.08695042e-01 -7.75477409e-01 5.54465771e-01 -5.93485832e-01
-1.52669549e+00 2.19536573e-01 9.22924578e-02 -1.67237806e+00
-3.21315914e-01 -9.04774070e-01 -5.38884580e-01 6.08177185e-01
-9.92096603e-01 -1.30674469e+00 -4.57582951e-01 9.51599300e-01
4.66378480e-01 -3.37637402e-02 8.29392970e-01 3.69417518e-01
-6.80012286e-01 4.11944985e-01 -2.07003653e-01 6.23686194e-01
5.20363748e-01 -8.43963802e-01 9.17279184e-01 5.65502644e-01
-4.68370179e-03 5.57576120e-01 3.82020533e-01 -7.91795909e-01
-2.01231766e+00 -8.64418924e-01 7.38020778e-01 -7.14512229e-01
4.50445443e-01 -3.96367818e-01 -7.40342379e-01 6.48095429e-01
-7.36163080e-01 2.74778366e-01 6.72698200e-01 -2.59804614e-02
-3.55838150e-01 9.91110802e-02 -9.99843478e-01 6.71022415e-01
1.63625920e+00 -2.00447366e-01 -8.23743463e-01 1.07213706e-01
1.19648330e-01 -5.20229876e-01 -1.03099060e+00 4.07755584e-01
8.90608370e-01 -7.44297206e-01 1.57020164e+00 -4.23925608e-01
1.70104712e-01 -3.52377683e-01 -4.29474056e-01 -1.07150018e+00
-2.46614084e-01 -2.43110269e-01 -7.33822465e-01 7.41543591e-01
-7.25638449e-01 -5.45802236e-01 1.07682514e+00 3.27193350e-01
-7.46296486e-03 -6.25145793e-01 -1.20844328e+00 -1.09138262e+00
-7.66193541e-03 -3.31734747e-01 4.08823937e-01 6.27630949e-01
-7.15585798e-02 -6.79226220e-02 -8.65123391e-01 1.39174774e-01
1.11394906e+00 2.88213938e-01 1.12618327e+00 -1.34162295e+00
-1.42393842e-01 -6.46036640e-02 -1.31915891e+00 -1.45324075e+00
3.78401279e-02 -6.49880826e-01 -3.03823590e-01 -1.41456878e+00
2.54892945e-01 2.83162296e-02 -2.14244761e-02 4.83149379e-01
4.14887181e-04 2.91921586e-01 3.74072790e-01 5.53437829e-01
-4.19903308e-01 9.22475398e-01 1.52011132e+00 -2.24179044e-01
1.35294497e-01 1.92906156e-01 -9.95815843e-02 8.87266517e-01
6.20462120e-01 -4.33462001e-02 -1.89895004e-01 -3.17760497e-01
-4.00817573e-01 1.63877368e-01 7.74534404e-01 -1.21019650e+00
-1.53481180e-03 -1.62213251e-01 6.09617889e-01 -7.76152790e-01
7.19609380e-01 -6.91845775e-01 2.07109883e-01 5.92299402e-01
4.60195214e-01 -1.89859346e-02 -8.55465010e-02 6.76097691e-01
-1.11288652e-01 2.56038398e-01 4.83327925e-01 -4.93042290e-01
-9.21568096e-01 6.95880413e-01 -3.63171816e-01 2.14876667e-01
8.03626835e-01 -8.52954090e-01 -2.84710415e-02 -2.10560963e-01
-1.01145542e+00 2.14696705e-01 7.90334642e-01 5.64455926e-01
9.54431236e-01 -1.69879627e+00 -5.99814773e-01 6.77609801e-01
1.24867164e-01 -7.11752921e-02 5.38579643e-01 8.47824812e-01
-4.77467507e-01 3.47204357e-01 -6.17140651e-01 -9.42965031e-01
-1.32682121e+00 2.89440870e-01 4.96949047e-01 -2.58715510e-01
-1.21159029e+00 2.62296617e-01 1.14550263e-01 -6.50639117e-01
4.96260077e-02 -2.64091998e-01 -8.22955072e-02 -1.77642524e-01
4.79448646e-01 5.10554731e-01 -1.87431306e-01 -1.26674998e+00
-6.82923973e-01 1.26140320e+00 2.55704820e-01 3.61705482e-01
1.29870868e+00 -2.78247297e-01 2.15400666e-01 5.12352228e-01
1.13905621e+00 -4.32547212e-01 -1.41511559e+00 -2.74909437e-01
-2.57011920e-01 -8.96079123e-01 -5.41813970e-01 -1.52932525e-01
-1.15920377e+00 9.15202081e-01 4.57182199e-01 -4.10297364e-01
9.50202107e-01 -1.00402735e-01 1.02198088e+00 3.30342025e-01
9.59443569e-01 -8.35210025e-01 4.11783069e-01 6.35759890e-01
8.82650614e-01 -1.04070675e+00 1.51673451e-01 -6.35587573e-01
-5.00202119e-01 1.16710341e+00 5.69472075e-01 -3.66338491e-01
6.30139410e-01 -8.79796445e-02 -1.22044086e-01 -1.73552632e-01
-3.43666792e-01 -1.27334401e-01 4.08300698e-01 1.10121942e+00
1.95797831e-01 1.76154658e-01 1.30789652e-01 6.09377146e-01
-2.41029054e-01 1.72297791e-01 2.81463116e-01 1.15169990e+00
-1.97658807e-01 -9.47900772e-01 -6.18950725e-01 1.71279356e-01
5.00260219e-02 5.68722069e-01 -2.36432806e-01 7.69847393e-01
-4.37428392e-02 6.55977130e-01 -6.83012679e-02 -6.05670691e-01
7.84669638e-01 1.03549480e-01 7.37723589e-01 -5.04282773e-01
1.00804225e-01 -3.33051346e-02 -7.52910301e-02 -8.55095923e-01
-4.77434486e-01 -9.68405962e-01 -1.01704872e+00 -6.10876322e-01
3.45466465e-01 -2.56377935e-01 1.56859428e-01 8.33616436e-01
2.06410944e-01 5.56471832e-02 3.88875127e-01 -1.42941332e+00
-6.43967927e-01 -5.47276914e-01 -1.02890861e+00 6.38385713e-01
4.27934140e-01 -1.27427840e+00 -2.72410989e-01 3.20391864e-01] | [14.283770561218262, 1.4142816066741943] |
5f7086b1-a5ba-4c67-ad8a-f45eeadb2344 | describing-unseen-videos-via-multi | 2008.07935 | null | https://arxiv.org/abs/2008.07935v2 | https://arxiv.org/pdf/2008.07935v2.pdf | Describing Unseen Videos via Multi-Modal Cooperative Dialog Agents | With the arising concerns for the AI systems provided with direct access to abundant sensitive information, researchers seek to develop more reliable AI with implicit information sources. To this end, in this paper, we introduce a new task called video description via two multi-modal cooperative dialog agents, whose ultimate goal is for one conversational agent to describe an unseen video based on the dialog and two static frames. Specifically, one of the intelligent agents - Q-BOT - is given two static frames from the beginning and the end of the video, as well as a finite number of opportunities to ask relevant natural language questions before describing the unseen video. A-BOT, the other agent who has already seen the entire video, assists Q-BOT to accomplish the goal by providing answers to those questions. We propose a QA-Cooperative Network with a dynamic dialog history update learning mechanism to transfer knowledge from A-BOT to Q-BOT, thus helping Q-BOT to better describe the video. Extensive experiments demonstrate that Q-BOT can effectively learn to describe an unseen video by the proposed model and the cooperative learning method, achieving the promising performance where Q-BOT is given the full ground truth history dialog. | ['Yu Wu', 'Yi Yang', 'Ye Zhu', 'Yan Yan'] | 2020-08-18 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/4257_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123680154.pdf | eccv-2020-8 | ['video-description'] | ['computer-vision'] | [-1.47044092e-01 4.20609444e-01 -2.37265140e-01 -3.10323536e-01
-8.21653366e-01 -6.35742188e-01 4.21910375e-01 -3.58413130e-01
-4.60450351e-01 8.70440781e-01 2.65619308e-01 3.71123821e-01
1.54857934e-01 -3.91885936e-01 -1.72396719e-01 -7.02980697e-01
-6.29624575e-02 7.02731490e-01 6.62140131e-01 -5.17154813e-01
1.67820007e-01 -7.56505430e-02 -1.12777555e+00 3.48848879e-01
4.26382899e-01 8.76972973e-01 7.58721054e-01 7.72286952e-01
-1.08409546e-01 1.65258014e+00 -6.53523743e-01 -4.82598871e-01
9.68996361e-02 -6.04254901e-01 -1.17111969e+00 5.06043136e-01
-3.44348745e-03 -1.24381506e+00 -7.95726895e-01 1.14915252e+00
9.74881947e-02 5.29395878e-01 4.89136130e-01 -1.67302680e+00
-6.63883269e-01 5.57080567e-01 -1.51365981e-01 3.48062485e-01
7.72285700e-01 8.35645318e-01 9.64212775e-01 -4.53795046e-01
9.97111022e-01 1.70946300e+00 1.55337721e-01 1.18151987e+00
-5.71156919e-01 -3.70251149e-01 2.41214827e-01 6.78607285e-01
-1.16736293e+00 -4.47360992e-01 9.99073803e-01 -3.40808749e-01
5.66195130e-01 -6.17002957e-02 4.79712814e-01 1.14526629e+00
1.75021753e-01 8.84987772e-01 7.00592458e-01 -7.39303753e-02
2.33999506e-01 4.16710168e-01 1.55065745e-01 8.83918643e-01
-5.00604928e-01 -8.84931609e-02 -7.72341013e-01 -2.46043503e-01
6.96999490e-01 1.09119348e-01 -3.25027257e-01 -1.16088182e-01
-1.27883875e+00 9.55344141e-01 2.21943274e-01 3.54179233e-01
-5.30710816e-01 1.42846908e-02 3.13501418e-01 5.91774642e-01
1.35716528e-01 3.76259148e-01 -1.81316659e-01 -3.07896793e-01
-1.39627218e-01 3.16456854e-01 1.11932099e+00 1.30806684e+00
9.97167826e-01 6.47311360e-02 -1.60297230e-01 3.51799607e-01
3.86934400e-01 3.37738693e-01 4.05785710e-01 -1.76447833e+00
4.78812516e-01 6.93651497e-01 5.79117477e-01 -9.42198694e-01
-4.23251837e-02 4.79931146e-01 -4.83628571e-01 4.55051996e-02
4.50993866e-01 -5.91215253e-01 -3.54855210e-01 1.83261538e+00
5.12523472e-01 2.22404711e-02 6.19522631e-01 1.19183099e+00
9.07259524e-01 1.19531679e+00 1.50130406e-01 -7.89307177e-01
1.42850053e+00 -1.26246643e+00 -1.11788225e+00 -2.54495084e-01
3.30348015e-01 -4.56343055e-01 8.11455548e-01 2.20851898e-01
-1.14975321e+00 -6.60700500e-01 -7.59633005e-01 -8.36876631e-02
5.50447740e-02 -7.68427029e-02 1.80948645e-01 -3.21984813e-02
-1.11296546e+00 7.05704391e-02 -5.06435692e-01 -6.21366918e-01
-5.50342239e-02 3.70247215e-01 -3.13862771e-01 -3.71344090e-01
-1.42702746e+00 7.69723833e-01 6.02762222e-01 -4.66225073e-02
-1.41486847e+00 1.59702934e-02 -6.88520074e-01 6.28770441e-02
1.08476090e+00 -6.36864722e-01 1.86851048e+00 -1.27439034e+00
-1.72899139e+00 3.83766621e-01 -1.28281578e-01 -3.21478426e-01
3.47196668e-01 -1.18927479e-01 -1.64428294e-01 8.19196880e-01
2.20365211e-01 9.54377890e-01 8.49354923e-01 -1.43430078e+00
-1.00919878e+00 -3.85347813e-01 8.05309057e-01 7.13861763e-01
-3.44274789e-01 1.62183233e-02 -6.71861589e-01 -1.65658027e-01
-1.82537153e-01 -9.42691028e-01 -4.01628576e-02 1.14914151e-02
-1.15539268e-01 -6.45003140e-01 1.30313921e+00 -6.74998879e-01
7.93761075e-01 -1.98701131e+00 4.48893607e-01 -4.97063994e-01
4.60375577e-01 2.31623441e-01 -2.11531609e-01 5.82109690e-01
6.26292288e-01 -3.01711053e-01 1.08561523e-01 -2.68758088e-01
-2.67944247e-01 4.48387444e-01 -1.50182307e-01 1.54221222e-01
1.23571463e-01 7.65995026e-01 -1.26392400e+00 -9.23246622e-01
2.77851056e-02 1.90884694e-02 -3.79244179e-01 1.07807684e+00
-6.95622742e-01 6.54550314e-01 -9.52211320e-01 3.92825961e-01
1.97816879e-01 -4.69056249e-01 3.06904227e-01 -1.50468096e-01
6.13282658e-02 -2.47983292e-01 -7.22499907e-01 1.67338157e+00
9.69086960e-02 5.14207482e-01 6.03960693e-01 -8.87624741e-01
7.82535255e-01 7.93655396e-01 5.33626795e-01 -4.01233494e-01
2.38861978e-01 -1.63530365e-01 3.31148915e-02 -1.05228579e+00
4.47130829e-01 -8.15226510e-02 -2.22453728e-01 6.05894327e-01
3.17778498e-01 -6.81305230e-02 7.25250170e-02 7.34049976e-01
9.75132108e-01 -3.05580925e-02 2.86686271e-01 3.44196647e-01
8.67875397e-01 4.82282698e-01 6.09107018e-01 8.01571429e-01
-6.81664288e-01 -1.09243356e-01 4.84918624e-01 -4.49688137e-01
-7.77713418e-01 -6.48567379e-01 9.09716845e-01 1.36777723e+00
8.28143001e-01 -2.01476187e-01 -7.25424767e-01 -8.31618667e-01
-4.81154710e-01 5.33512533e-01 -3.55310053e-01 -2.63176411e-01
-6.40986741e-01 2.59159386e-01 -3.80107015e-02 9.29163322e-02
1.14997220e+00 -1.42124426e+00 -7.34098494e-01 2.73910373e-01
-7.82356501e-01 -1.28164375e+00 -8.65689576e-01 -2.65347868e-01
-4.90335256e-01 -9.13181305e-01 -6.16342366e-01 -1.00612772e+00
3.57414544e-01 6.00874066e-01 8.30307186e-01 2.32534423e-01
2.89399952e-01 8.71422172e-01 -6.44830525e-01 -3.39481793e-02
-8.61737311e-01 -1.65005580e-01 3.02702617e-02 1.56106666e-01
4.18270499e-01 -1.83931500e-01 -2.64646143e-01 6.60359025e-01
-8.55696380e-01 1.98264554e-01 4.51774716e-01 7.60905445e-01
5.27136512e-02 -4.68759518e-03 7.03786612e-01 -6.70890450e-01
5.67856491e-01 -8.20677996e-01 -2.99526066e-01 2.79120713e-01
-1.27005592e-01 1.39357504e-02 8.73508096e-01 -8.09534848e-01
-1.47319591e+00 1.52821064e-01 2.76336014e-01 -6.52160704e-01
-2.23275378e-01 3.65953505e-01 -3.05741131e-01 2.17496708e-01
5.14026582e-01 3.07369858e-01 2.25426123e-01 3.08184102e-02
3.25975239e-01 8.86275709e-01 9.64506209e-01 -3.68018478e-01
7.99548209e-01 3.73417020e-01 -4.80889052e-01 -8.60581100e-01
-9.48847651e-01 -8.30723166e-01 -6.59583211e-01 -8.58927548e-01
1.24341226e+00 -8.55331182e-01 -1.21552408e+00 5.03302872e-01
-1.77750182e+00 -3.09545010e-01 9.59968939e-03 3.42498392e-01
-8.98457885e-01 5.64480722e-01 -5.50312579e-01 -1.03765571e+00
-4.91227321e-02 -1.07541656e+00 7.00119138e-01 5.17040968e-01
6.71443045e-02 -8.53019714e-01 -4.24483605e-02 7.67684698e-01
1.57075197e-01 -1.92880869e-01 6.25947356e-01 -8.16891372e-01
-1.04460073e+00 1.14216171e-01 -1.43034667e-01 3.02905440e-02
3.14045340e-01 -5.06569922e-01 -7.55045772e-01 -3.38620692e-01
4.44665641e-01 -8.49609256e-01 2.42858067e-01 1.08067337e-02
5.34126103e-01 -7.67553985e-01 -2.75389344e-01 -8.39357451e-02
9.58833039e-01 9.43798065e-01 2.21411869e-01 -4.71721403e-02
4.77265179e-01 9.97194409e-01 1.13751268e+00 4.84610826e-01
7.16614187e-01 7.44615614e-01 6.92733824e-01 3.47929865e-01
3.06442678e-02 -2.34067917e-01 6.20113075e-01 8.76605093e-01
4.81860191e-02 -6.29051447e-01 -5.42761505e-01 5.56708097e-01
-2.42076588e+00 -1.25915754e+00 4.36182499e-01 1.59081066e+00
8.21863770e-01 -1.07898481e-01 3.57111156e-01 -4.77864325e-01
1.01666892e+00 3.71864676e-01 -1.11338484e+00 1.75295368e-01
4.46263291e-02 -9.94869232e-01 -1.16330877e-01 5.89970291e-01
-8.38473558e-01 1.28604460e+00 5.49907351e+00 5.21524549e-01
-7.70847023e-01 2.94961631e-01 5.21989167e-01 1.97419658e-01
8.11519623e-02 1.47093162e-01 -7.57591963e-01 4.11047101e-01
1.04924190e+00 -4.07772958e-01 7.44509280e-01 8.49805951e-01
4.38195109e-01 -3.78390253e-01 -1.18003547e+00 9.70728934e-01
3.08225572e-01 -1.34310043e+00 2.23294839e-01 -2.48957142e-01
4.95371044e-01 -4.83023763e-01 -1.77419022e-01 5.35096467e-01
5.87437689e-01 -5.97692490e-01 5.84740460e-01 6.18096888e-01
6.67233467e-01 -4.71568227e-01 8.06920648e-01 1.03600490e+00
-1.23556340e+00 -5.42925715e-01 -4.01266098e-01 4.29173326e-03
3.65894169e-01 -5.02050757e-01 -1.03846240e+00 3.87300938e-01
5.43728411e-01 6.50729597e-01 -2.46632975e-02 4.56853241e-01
7.59572685e-02 2.97855377e-01 1.94951534e-01 -3.07716936e-01
5.21632135e-01 -2.47213691e-01 8.10690701e-01 6.24498665e-01
-5.70016429e-02 1.06937432e+00 7.68539071e-01 7.55115330e-01
-1.38261855e-01 -1.49929240e-01 -6.16619170e-01 -1.47052422e-01
6.36636078e-01 1.18783236e+00 -3.13672900e-01 -6.00165308e-01
-6.04205966e-01 1.01287377e+00 3.43451858e-01 3.98071796e-01
-6.96478248e-01 -2.23484650e-01 1.92555204e-01 -4.67985645e-02
-5.83099201e-02 -1.65235177e-01 7.78354764e-01 -1.04302168e+00
-3.27781022e-01 -1.10111356e+00 6.85856819e-01 -1.44434130e+00
-1.15379226e+00 8.06269884e-01 3.20789963e-01 -1.12831676e+00
-9.07404661e-01 -1.98753491e-01 -5.15428364e-01 4.57329005e-01
-1.36099803e+00 -1.15574157e+00 -6.06908262e-01 1.09617531e+00
1.43562520e+00 -6.16736948e-01 4.67562169e-01 -1.55766904e-01
-2.34419346e-01 -9.15267039e-03 -2.89653391e-01 2.91161925e-01
8.26558113e-01 -1.01874721e+00 -2.63346374e-01 5.09085059e-01
-1.90316156e-01 1.40636832e-01 9.44058239e-01 -8.12119544e-01
-1.57384276e+00 -8.59546006e-01 4.86117631e-01 -2.66887724e-01
7.04531848e-01 6.28453568e-02 -1.03138292e+00 8.72568667e-01
5.51826477e-01 -4.50792223e-01 1.23418726e-01 -8.36604953e-01
1.24096647e-01 5.74302897e-02 -1.22190166e+00 5.23738146e-01
7.87846506e-01 -5.46246171e-01 -1.08460081e+00 6.04816258e-01
1.15945470e+00 -2.96136409e-01 -4.88442510e-01 -1.37598485e-01
1.87804326e-01 -8.59378934e-01 5.82072020e-01 -7.23016143e-01
3.89290005e-01 -1.22160666e-01 -1.77829683e-01 -1.18306541e+00
-1.37424752e-01 -1.16958761e+00 -1.67309061e-01 9.90487814e-01
-5.00300750e-02 -2.02743500e-01 9.39624190e-01 7.51454949e-01
-4.02186699e-02 -1.88501582e-01 -8.57544482e-01 -4.76676881e-01
-3.96341592e-01 1.77817449e-01 2.82826990e-01 6.80604041e-01
3.35364401e-01 6.93410218e-01 -9.68980014e-01 2.16826975e-01
3.99190873e-01 -4.28779498e-02 1.08223104e+00 -1.20767796e+00
-1.99149638e-01 3.09874088e-01 1.14936624e-02 -1.71054435e+00
4.71846908e-01 -1.37249008e-01 4.52511996e-01 -1.34113061e+00
5.16469121e-01 1.94800124e-01 3.41226250e-01 3.10642809e-01
-2.13106081e-01 -2.29441479e-01 4.11855370e-01 8.77745926e-01
-1.30555999e+00 6.06884301e-01 1.65896344e+00 -3.10931593e-01
-3.99103731e-01 6.60109371e-02 -3.86233926e-01 7.24876285e-01
4.22009379e-01 -4.06904042e-01 -8.37880969e-01 -2.40282863e-01
-3.37905973e-01 1.13947856e+00 2.68059403e-01 -7.59607971e-01
8.00929129e-01 -6.29874408e-01 -2.84279406e-01 -5.94006240e-01
8.90355527e-01 -1.06070948e+00 8.53051022e-02 5.28427958e-01
-6.29417598e-01 2.13481769e-01 -2.25421727e-01 9.15170133e-01
-3.76508892e-01 -5.44269681e-01 6.49801373e-01 -6.06962919e-01
-1.24013710e+00 4.18970585e-01 -7.14464307e-01 2.29913387e-02
1.35365474e+00 -2.14182138e-01 -7.39216805e-01 -1.34425163e+00
-8.42089891e-01 8.01399827e-01 2.42859110e-01 3.45327407e-01
7.55061328e-01 -1.07312536e+00 -5.57253420e-01 -3.38076442e-01
2.09384672e-02 -1.51541442e-01 5.23268402e-01 3.03763509e-01
-2.56766617e-01 4.28481281e-01 -5.21615446e-01 -5.26666641e-01
-1.36534762e+00 9.12905514e-01 3.11673939e-01 -2.25742400e-01
-3.84593517e-01 4.23798293e-01 4.22635704e-01 -3.30532156e-02
4.61768717e-01 2.72825599e-01 -6.32563472e-01 2.54060812e-02
5.32785475e-01 2.67321289e-01 -7.12366939e-01 -9.97972488e-01
3.70974727e-02 2.04622045e-01 -1.71405539e-01 -4.75741386e-01
1.05017817e+00 -8.10157895e-01 7.78361261e-02 6.25100255e-01
1.02865207e+00 -5.43487549e-01 -1.70121539e+00 -6.75182343e-01
-2.57180840e-01 -4.99446720e-01 -2.08562613e-01 -6.03541493e-01
-8.74178588e-01 5.29234946e-01 4.07099605e-01 5.38251221e-01
8.81018281e-01 3.59669298e-01 9.84843075e-01 1.03527987e+00
6.23778224e-01 -9.71130013e-01 1.15236402e+00 4.29927021e-01
8.79581988e-01 -1.54011691e+00 -2.84318805e-01 -3.12365532e-01
-1.33343828e+00 1.12853754e+00 1.09536517e+00 2.25929871e-01
2.21256673e-01 -2.41443649e-01 2.83643603e-01 -3.78865570e-01
-1.30540013e+00 -9.88651887e-02 -5.34549169e-02 8.17639649e-01
-4.41202730e-01 -4.44599777e-01 2.20452398e-01 4.26516086e-01
1.08387135e-01 -8.80463123e-02 7.95878351e-01 8.55210423e-01
-8.98246109e-01 -6.91984892e-01 -3.87772053e-01 8.02858323e-02
2.14462485e-02 5.71971655e-01 -7.08747208e-01 8.97244692e-01
-1.70259997e-01 1.47555494e+00 1.26532346e-01 -3.50287110e-01
1.09750651e-01 6.71863705e-02 -1.63409919e-01 -7.31003344e-01
-5.12016952e-01 1.40511066e-01 8.35421011e-02 -4.86484408e-01
-8.57815742e-01 -6.09644473e-01 -1.18342316e+00 -3.35480899e-01
-3.46531183e-01 5.60987532e-01 6.07321821e-02 1.05681109e+00
3.43730673e-02 2.43548229e-02 9.54317629e-01 -7.01958358e-01
-7.14948893e-01 -1.13017476e+00 -5.21096408e-01 5.29034257e-01
5.24643064e-01 -6.78054571e-01 -5.57750881e-01 5.99609256e-01] | [10.840975761413574, 1.126764178276062] |
b6953196-3888-4148-8ae3-7c859c532000 | semantic-aware-graph-matching-mechanism-for | 2304.11275 | null | https://arxiv.org/abs/2304.11275v1 | https://arxiv.org/pdf/2304.11275v1.pdf | Semantic-Aware Graph Matching Mechanism for Multi-Label Image Recognition | Multi-label image recognition aims to predict a set of labels that present in an image. The key to deal with such problem is to mine the associations between image contents and labels, and further obtain the correct assignments between images and their labels. In this paper, we treat each image as a bag of instances, and formulate the task of multi-label image recognition as an instance-label matching selection problem. To model such problem, we propose an innovative Semantic-aware Graph Matching framework for Multi-Label image recognition (ML-SGM), in which Graph Matching mechanism is introduced owing to its good performance of excavating the instance and label relationship. The framework explicitly establishes category correlations and instance-label correspondences by modeling the relation among content-aware (instance) and semantic-aware (label) category representations, to facilitate multi-label image understanding and reduce the dependency of large amounts of training samples for each category. Specifically, we first construct an instance spatial graph and a label semantic graph respectively and then incorporate them into a constructed assignment graph by connecting each instance to all labels. Subsequently, the graph network block is adopted to aggregate and update all nodes and edges state on the assignment graph to form structured representations for each instance and label. Our network finally derives a prediction score for each instance-label correspondence and optimizes such correspondence with a weighted cross-entropy loss. Empirical results conducted on generic multi-label image recognition demonstrate the superiority of our proposed method. Moreover, the proposed method also shows advantages in multi-label recognition with partial labels and multi-label few-shot learning, as well as outperforms current state-of-the-art methods with a clear margin. | ['Yang Wang', 'Songhe Feng', 'Yanan Wu'] | 2023-04-21 | null | null | null | null | ['graph-matching'] | ['graphs'] | [ 7.91288912e-01 -1.05580300e-01 -5.30062795e-01 -6.65314257e-01
-8.01717162e-01 -2.23870009e-01 3.86139423e-01 3.58995378e-01
5.71334325e-02 2.19778493e-01 -1.09220974e-01 2.73759156e-01
-4.42106843e-01 -7.39406407e-01 -3.85344416e-01 -7.77718365e-01
3.86404455e-01 4.05821204e-01 -8.74568336e-03 4.10081923e-01
3.13038349e-01 2.16603756e-01 -1.86586654e+00 3.01582187e-01
5.92880368e-01 1.27455795e+00 4.33463931e-01 1.77371904e-01
-3.68971854e-01 1.04480612e+00 -1.26514554e-01 -3.43328863e-01
-8.54549482e-02 -5.19613564e-01 -1.02863026e+00 7.94488430e-01
5.53874075e-01 1.98228166e-01 -9.55049396e-02 1.26603806e+00
2.04421654e-01 2.99410433e-01 9.10896242e-01 -1.47261620e+00
-8.30296397e-01 2.67154694e-01 -8.54413748e-01 -1.17849968e-01
9.44033638e-02 -1.78465813e-01 1.25580490e+00 -8.80334854e-01
5.81159770e-01 1.19852757e+00 5.32262266e-01 4.81062442e-01
-1.09483063e+00 -7.40723789e-01 3.56394261e-01 4.10788834e-01
-1.47332513e+00 -2.84566104e-01 8.19360912e-01 -4.02790576e-01
5.82801461e-01 1.48238674e-01 2.95441031e-01 7.27981269e-01
-1.30492583e-01 8.57367039e-01 1.16683471e+00 -5.52743554e-01
-6.75169751e-02 1.28691763e-01 5.08354425e-01 1.03772569e+00
-7.41558820e-02 -2.41287395e-01 -4.57442045e-01 -1.34649813e-01
4.20621127e-01 5.28731585e-01 -9.52502489e-02 -5.20758808e-01
-1.05849862e+00 7.90120125e-01 4.13473338e-01 4.12916094e-01
-3.58847082e-01 2.28326991e-01 4.63543624e-01 7.53329396e-02
5.00294864e-01 2.25869343e-02 -2.31579199e-01 4.56244171e-01
-8.02180350e-01 -3.74209106e-01 4.25183207e-01 1.24111938e+00
1.26199937e+00 -3.64450246e-01 -2.24721014e-01 1.37316072e+00
6.13913178e-01 2.77969807e-01 5.82332253e-01 -6.78742468e-01
3.16925496e-01 9.94491518e-01 -4.74903345e-01 -1.36775672e+00
-4.88285810e-01 -3.80196214e-01 -9.15084779e-01 -2.40444347e-01
-4.32771668e-02 4.28472906e-01 -9.25995529e-01 1.63621902e+00
5.07849872e-01 9.30688858e-01 5.04231965e-03 6.91698611e-01
1.08620131e+00 5.49498439e-01 6.16938591e-01 -5.87508023e-01
1.54474986e+00 -1.20979536e+00 -5.84942400e-01 -4.47466582e-01
8.77583325e-01 -4.84941363e-01 7.98553646e-01 -1.42797455e-01
-4.86559361e-01 -6.65840447e-01 -8.19446087e-01 1.59129605e-01
-3.56121153e-01 1.02152146e-01 5.64605892e-01 3.31333458e-01
-6.41542912e-01 6.13147676e-01 -1.10373616e-01 -4.22677189e-01
5.03914237e-01 3.64374459e-01 -4.93610978e-01 -6.36890173e-01
-1.09334111e+00 7.36848950e-01 7.84891427e-01 -7.43311420e-02
-8.10665667e-01 -5.77326715e-01 -1.17314124e+00 -2.94003519e-03
5.85979700e-01 -5.27926683e-01 7.19897270e-01 -1.17112851e+00
-7.64838219e-01 1.30387223e+00 -2.35378161e-01 1.24125622e-01
-2.87418455e-01 5.59345365e-01 -7.85308540e-01 2.42664263e-01
3.20241749e-01 5.95695317e-01 7.89529800e-01 -1.66050601e+00
-9.95510936e-01 -5.25091171e-01 4.07006070e-02 3.76679480e-01
-4.13386375e-01 9.02970508e-03 -5.13306499e-01 -4.29478198e-01
5.21762073e-01 -8.10608983e-01 -1.43809235e-02 -4.21807915e-01
-3.71988416e-01 -3.91835511e-01 7.32033134e-01 -5.44020653e-01
1.17474580e+00 -2.02377415e+00 1.35465875e-01 3.00734282e-01
3.05311769e-01 -8.47907737e-02 -4.43165988e-01 1.24908209e-01
-1.74633920e-01 3.52545306e-02 -2.50769466e-01 -5.18925250e-01
-9.62266922e-02 3.38793576e-01 1.14391588e-01 5.94155908e-01
-5.74367195e-02 9.88898695e-01 -9.68543231e-01 -9.52105284e-01
3.35349202e-01 2.44514212e-01 -7.67375380e-02 2.17924431e-01
-1.41893774e-01 2.79867053e-01 -5.59567511e-01 1.09375107e+00
5.90918124e-01 -8.51214051e-01 4.77968812e-01 -6.33918166e-01
3.81001502e-01 -5.71956277e-01 -1.37396836e+00 1.71096516e+00
-6.63724244e-01 -2.36730259e-02 -2.59103954e-01 -1.36025739e+00
1.02131331e+00 2.12085143e-01 8.24933708e-01 -7.12615132e-01
2.24478468e-01 1.11846447e-01 -7.78590620e-01 -6.42605186e-01
2.44732529e-01 -3.38079005e-01 -1.63687721e-01 5.81455588e-01
4.19848621e-01 2.49379486e-01 8.35674778e-02 1.32427409e-01
7.53539562e-01 -3.96164991e-02 6.44949019e-01 -1.89562757e-02
8.09078395e-01 -2.57073939e-01 7.25850821e-01 4.87905324e-01
-1.76268503e-01 3.41387451e-01 3.55665460e-02 -3.53533357e-01
-6.25376225e-01 -5.99018216e-01 1.23971403e-02 1.35506499e+00
7.70636439e-01 -2.50964552e-01 -3.92568111e-01 -1.16282773e+00
8.71266574e-02 5.38965344e-01 -6.32551253e-01 -1.89105526e-01
-2.21034721e-01 -7.62433410e-01 1.62031129e-01 1.81498542e-01
5.08237123e-01 -1.21758556e+00 1.19434595e-02 1.00847147e-01
-3.27699900e-01 -9.26575899e-01 -6.09002471e-01 1.80866316e-01
-6.37624741e-01 -1.46583819e+00 -2.79121697e-01 -1.37031329e+00
8.91706049e-01 6.13006949e-01 9.20610964e-01 3.80396456e-01
-4.53540266e-01 5.70324183e-01 -4.31195855e-01 2.71445006e-01
-2.33750716e-01 -2.36732841e-01 -2.44805455e-01 6.82789385e-01
4.40669537e-01 -4.88319159e-01 -3.95340472e-01 5.56076169e-01
-9.79014397e-01 1.83877498e-01 5.75308681e-01 1.04499590e+00
1.12809658e+00 1.82489395e-01 7.85901070e-01 -1.22927475e+00
2.85521835e-01 -7.30509698e-01 -3.38726193e-01 9.04554069e-01
-1.02811253e+00 -8.75135437e-02 4.40958589e-01 -3.51628691e-01
-9.52546358e-01 4.86881226e-01 3.98209155e-01 -5.99796832e-01
-2.90511072e-01 5.48852444e-01 -3.82468849e-01 -4.05007392e-01
1.18091501e-01 4.04837459e-01 -8.39303285e-02 -3.07537884e-01
7.08490193e-01 7.49885380e-01 2.89616704e-01 -3.77429873e-01
3.25629324e-01 2.84061491e-01 3.06147426e-01 -4.77968901e-01
-1.34641242e+00 -1.10402918e+00 -8.54246378e-01 -5.58092058e-01
1.05182672e+00 -9.41011786e-01 -8.23626995e-01 4.22460675e-01
-9.09947991e-01 2.13928178e-01 6.27247021e-02 2.61999935e-01
-5.93656242e-01 6.17325068e-01 -5.94935477e-01 -6.52949929e-01
-1.69740587e-01 -1.20123088e+00 1.18488789e+00 4.36378598e-01
1.23960100e-01 -1.35912120e+00 -7.90952221e-02 7.30528593e-01
-5.94234988e-02 2.25460678e-01 1.12333381e+00 -7.68781424e-01
-5.38436055e-01 -1.00910805e-01 -7.40782499e-01 2.42735535e-01
2.35305429e-01 -4.75215614e-01 -9.77073193e-01 -3.84015709e-01
-6.87768385e-02 -5.70743382e-01 9.15293574e-01 2.28277683e-01
1.19114053e+00 -2.54239857e-01 -7.33108461e-01 4.22390699e-01
1.92677772e+00 1.86769217e-01 1.22386172e-01 1.30067259e-01
1.16512811e+00 7.72710562e-01 9.04429972e-01 3.40319127e-01
5.32517493e-01 7.04534471e-01 5.93980134e-01 -1.21227326e-02
-2.22928137e-01 -2.19698206e-01 -1.75140217e-01 1.08791149e+00
4.18799460e-01 -2.88285583e-01 -6.99807882e-01 4.14349556e-01
-2.12129807e+00 -9.66340303e-01 -4.71295305e-02 1.88899469e+00
6.00286782e-01 -4.34157997e-01 -1.50857657e-01 -2.31765062e-01
1.32950675e+00 3.98079395e-01 -6.73619032e-01 2.00018212e-02
-3.65405008e-02 -1.22920975e-01 4.53090012e-01 3.34815770e-01
-1.40712488e+00 8.86297703e-01 5.43065882e+00 1.22967398e+00
-6.79998755e-01 4.19489950e-01 7.98891246e-01 2.85574168e-01
-2.12275803e-01 1.11163467e-01 -8.92485559e-01 3.56480896e-01
7.24556506e-01 -1.71125811e-02 4.84763086e-01 9.77244198e-01
-2.61530876e-01 8.15367401e-02 -1.05160129e+00 1.37662888e+00
6.39974713e-01 -1.26936340e+00 2.89413124e-01 9.52094942e-02
7.61045516e-01 -4.37628180e-01 -2.17109114e-01 2.64251769e-01
2.21437052e-01 -1.00454426e+00 4.33612257e-01 6.81648791e-01
9.30864275e-01 -7.07279325e-01 6.16846085e-01 2.90969908e-01
-1.82217050e+00 -2.77912050e-01 -4.47577804e-01 3.71567607e-01
8.80276412e-02 5.25958002e-01 -4.36515868e-01 8.76078784e-01
1.62643977e-02 1.24937606e+00 -6.83581829e-01 8.36959600e-01
-9.55712632e-04 1.54758349e-01 2.22680435e-01 1.11859336e-01
3.16927463e-01 -3.94691676e-01 -8.58945996e-02 1.04133761e+00
2.06576124e-01 1.64220348e-01 8.19549859e-01 5.74838638e-01
-2.14586496e-01 5.46677709e-01 -5.02725184e-01 1.24461092e-01
6.92970932e-01 1.52927518e+00 -9.52067018e-01 -4.30829227e-01
-7.74551690e-01 1.09713960e+00 6.80052519e-01 2.29145795e-01
-8.33341777e-01 -1.56069279e-01 3.82897168e-01 -4.54923272e-01
4.93712239e-02 4.41351414e-01 -2.71604747e-01 -9.31471109e-01
-1.41959235e-01 -5.49608588e-01 9.64160502e-01 -7.55546749e-01
-1.42936373e+00 3.18565071e-01 -2.69789368e-01 -1.25965571e+00
1.81102052e-01 -2.37745866e-01 -2.59216130e-01 6.91857457e-01
-1.63130867e+00 -1.79269290e+00 -5.67870200e-01 6.67497754e-01
6.22702181e-01 -1.55480176e-01 1.08941126e+00 5.47742426e-01
-6.22079968e-01 6.11555994e-01 9.52996537e-02 -2.29112729e-02
5.99418938e-01 -9.09761071e-01 -2.93416083e-01 5.90041578e-01
4.37587172e-01 2.29962066e-01 8.14406574e-02 -7.33049750e-01
-1.08145738e+00 -1.56882215e+00 9.55105364e-01 -1.88513473e-01
5.18051386e-01 1.18966080e-01 -8.47824752e-01 6.39868438e-01
-1.94021925e-01 3.18525732e-01 8.96731794e-01 7.91505352e-02
-5.67897320e-01 -2.10647807e-01 -1.14908326e+00 2.44508937e-01
1.27934813e+00 -7.80283570e-01 -4.36918437e-01 7.26817310e-01
7.90543914e-01 -5.22111543e-03 -1.12010336e+00 4.51803356e-01
3.28020811e-01 -6.80869818e-01 1.00547564e+00 -4.27394718e-01
4.28752065e-01 -4.08182979e-01 -3.79711866e-01 -1.34933865e+00
-7.42635727e-01 2.92629421e-01 6.89212605e-02 1.53797412e+00
3.51735681e-01 -4.17696327e-01 6.56706393e-01 4.81754363e-01
-6.25459328e-02 -7.66065061e-01 -7.91999638e-01 -5.98597169e-01
-4.93254602e-01 -2.35326618e-01 5.93861759e-01 1.40373349e+00
-3.20467502e-02 4.25983429e-01 -6.32907629e-01 3.29401761e-01
9.71622109e-01 5.91188610e-01 1.21890076e-01 -1.30416381e+00
-1.61108807e-01 -4.01174575e-01 -6.23324513e-01 -6.69437170e-01
7.49220312e-01 -1.54295933e+00 1.50376256e-03 -1.70949912e+00
7.86702335e-01 -8.37234735e-01 -8.66864562e-01 7.25845039e-01
-2.94218957e-01 3.80713195e-01 2.05024764e-01 5.16668260e-01
-1.20911229e+00 3.47142994e-01 1.15783024e+00 -5.63821077e-01
2.46116951e-01 -2.66729951e-01 -6.89173758e-01 4.68582392e-01
4.28882122e-01 -6.10451162e-01 -6.15440965e-01 -8.12952891e-02
-1.95253745e-01 2.70820498e-01 2.73162216e-01 -8.63357663e-01
4.05457407e-01 -4.81105417e-01 1.10023223e-01 -2.59409398e-01
2.84577727e-01 -1.04071224e+00 4.86453205e-01 2.56183803e-01
-6.31940782e-01 -3.91808271e-01 -3.22949857e-01 9.76537526e-01
-3.54335636e-01 -5.18209457e-01 9.87266243e-01 -3.15832466e-01
-1.51166499e+00 7.48089373e-01 2.55241960e-01 8.36647227e-02
1.52399790e+00 -2.51415104e-01 -3.33235174e-01 1.77949250e-01
-9.42336559e-01 4.03762907e-01 3.80462199e-01 5.64283848e-01
7.69801736e-01 -1.82943606e+00 -2.73791790e-01 1.91411108e-01
8.10480535e-01 -3.68699342e-01 7.83602655e-01 5.57865798e-01
5.27581386e-02 1.59450784e-01 -5.54678915e-03 -5.33570588e-01
-1.64439428e+00 8.90490174e-01 2.39804402e-01 -3.56361836e-01
-4.74387020e-01 8.29840541e-01 3.28709126e-01 -4.82474983e-01
1.95965081e-01 3.60845715e-01 -5.88901758e-01 5.30710816e-01
4.84661877e-01 2.13678703e-01 -1.52331278e-01 -1.15656233e+00
-4.81036842e-01 1.11474228e+00 -1.04831066e-02 6.26596391e-01
1.10153639e+00 -4.59023416e-01 -4.22522157e-01 5.94570220e-01
1.70444262e+00 -7.18155861e-01 -8.08941722e-01 -6.37517750e-01
1.41870692e-01 -5.37024915e-01 2.53379308e-02 -6.46848023e-01
-1.34762740e+00 4.84817564e-01 5.89387476e-01 -7.58689567e-02
1.07712281e+00 1.53727442e-01 7.23467946e-01 -3.20683680e-02
4.73793417e-01 -1.11487043e+00 3.65891337e-01 1.94378793e-01
3.50323111e-01 -1.55354333e+00 -1.37024954e-01 -6.47189438e-01
-7.00038791e-01 1.03520930e+00 7.03856528e-01 2.56720483e-01
8.55437219e-01 -3.43855858e-01 -5.06206118e-02 -6.05325580e-01
-4.37893450e-01 -4.61424083e-01 4.43992645e-01 4.88395989e-01
6.30176067e-02 3.22451502e-01 -2.00019404e-01 3.83516490e-01
7.22017825e-01 -1.17790014e-01 2.61852201e-02 5.72813511e-01
-7.99529552e-01 -1.03970158e+00 -5.56292310e-02 7.06118166e-01
-3.13214585e-02 -9.13455486e-02 1.09754331e-01 3.25771451e-01
4.73801434e-01 1.17358887e+00 -1.41862765e-01 -6.94783449e-01
2.19371915e-01 3.49453330e-01 2.60463446e-01 -8.83842885e-01
-3.64106447e-01 -2.03279093e-01 3.91888805e-02 -5.59340119e-01
-8.22868705e-01 -5.06606877e-01 -1.21533132e+00 -3.58659104e-02
-6.08973980e-01 1.99906640e-02 5.56135595e-01 1.16171181e+00
3.47003937e-01 6.74610853e-01 8.70500505e-01 -5.48420966e-01
-2.98769176e-01 -7.51643777e-01 -1.15532196e+00 1.09176683e+00
-1.29483312e-01 -8.77626359e-01 -4.76595849e-01 1.01187274e-01] | [9.746611595153809, 4.0315680503845215] |
72775b62-6f66-4119-8e7e-4178c0554173 | channel-attention-networks-for-robust-mr | 2012.01241 | null | https://arxiv.org/abs/2012.01241v1 | https://arxiv.org/pdf/2012.01241v1.pdf | Channel Attention Networks for Robust MR Fingerprinting Matching | Magnetic Resonance Fingerprinting (MRF) enables simultaneous mapping of multiple tissue parameters such as T1 and T2 relaxation times. The working principle of MRF relies on varying acquisition parameters pseudo-randomly, so that each tissue generates its unique signal evolution during scanning. Even though MRF provides faster scanning, it has disadvantages such as erroneous and slow generation of the corresponding parametric maps, which needs to be improved. Moreover, there is a need for explainable architectures for understanding the guiding signals to generate accurate parametric maps. In this paper, we addressed both of these shortcomings by proposing a novel neural network architecture consisting of a channel-wise attention module and a fully convolutional network. The proposed approach, evaluated over 3 simulated MRF signals, reduces error in the reconstruction of tissue parameters by 8.88% for T1 and 75.44% for T2 with respect to state-of-the-art methods. Another contribution of this study is a new channel selection method: attention-based channel selection. Furthermore, the effect of patch size and temporal frames of MRF signal on channel reduction are analyzed by employing a channel-wise attention. | ['Ilkay Oksuz', 'Devrim Unay', 'Andrew P. King', 'Claudia Prieto', 'Gastao Cruz', 'Eda Ozgu Ersoy', 'Ebru Navruz', 'Refik Soyak'] | 2020-12-02 | null | null | null | null | ['magnetic-resonance-fingerprinting'] | ['medical'] | [ 6.20591700e-01 3.54579203e-02 1.09058537e-01 -3.47250611e-01
-7.75503099e-01 -1.76475689e-01 2.43979871e-01 6.55206665e-02
-4.90830630e-01 7.29604602e-01 -9.00072791e-03 -3.18202078e-02
-3.28167498e-01 -4.82199311e-01 -8.15101147e-01 -8.99355054e-01
-2.47222662e-01 1.46307185e-01 3.21145117e-01 1.05642535e-01
5.02447724e-01 6.03801668e-01 -1.31193864e+00 2.80102074e-01
9.25265372e-01 1.06430018e+00 7.85294771e-01 4.81788099e-01
-8.38084817e-02 5.18898129e-01 -4.85002846e-01 1.23486549e-01
7.31677040e-02 -5.73572576e-01 -7.90067971e-01 -2.72601634e-01
3.94208103e-01 -2.69074351e-01 -2.80629247e-01 1.08551669e+00
9.54715014e-01 7.52199590e-02 5.67876697e-01 -5.76123416e-01
-6.14761710e-01 1.03577542e+00 -5.07003546e-01 6.25575125e-01
-3.19103226e-02 3.41240950e-02 3.73493999e-01 -5.52808404e-01
7.06427753e-01 6.43973649e-01 7.52270103e-01 5.81109881e-01
-1.46120656e+00 -5.94921231e-01 -1.66431844e-01 3.37603629e-01
-1.29232073e+00 -3.31799150e-01 7.82886922e-01 -5.84398985e-01
9.60263014e-01 1.93999574e-01 6.38261378e-01 7.97847033e-01
7.17747509e-01 4.00298744e-01 1.38482058e+00 -5.24314880e-01
7.49008963e-03 1.30653009e-01 1.89689919e-02 3.45096856e-01
1.67490672e-02 3.73594277e-02 -3.34840417e-01 3.79767902e-02
1.18373215e+00 -3.64680052e-01 -5.75877070e-01 -1.84239611e-01
-1.54911625e+00 6.45378053e-01 5.42888701e-01 7.60504067e-01
-5.87597609e-01 4.11429584e-01 3.02607208e-01 -2.08940846e-03
1.59828857e-01 7.94630468e-01 -2.15227231e-01 9.03605968e-02
-9.98358727e-01 4.68702763e-02 2.12848768e-01 4.89937961e-01
3.83012742e-01 5.57470098e-02 -5.86572349e-01 7.12041616e-01
3.08039784e-01 5.25717258e-01 8.61847997e-01 -1.00612926e+00
3.19388765e-03 -2.34476700e-02 -5.95838763e-02 -1.03665018e+00
-9.02515054e-01 -6.04844093e-01 -6.70427024e-01 -6.69808164e-02
4.69432384e-01 -4.81731668e-02 -9.91161227e-01 1.81734455e+00
1.55456111e-01 1.41456634e-01 -4.91124392e-01 1.37741172e+00
7.09043682e-01 4.22636211e-01 5.10311276e-02 -3.09312046e-01
1.19976783e+00 -7.67419994e-01 -9.89495993e-01 1.57780200e-01
1.73397377e-01 -6.15697145e-01 9.24510956e-01 2.23974675e-01
-1.11722803e+00 -6.17170632e-01 -1.00271606e+00 2.86772221e-01
-5.28693460e-02 3.71976383e-02 6.93800628e-01 8.32971871e-01
-1.08062589e+00 7.31269300e-01 -7.98577845e-01 -8.36274922e-02
3.12484920e-01 6.28683448e-01 -2.50161856e-01 1.76854044e-01
-1.49773610e+00 9.74696517e-01 3.35420877e-01 3.44186068e-01
-8.52569401e-01 -1.15432334e+00 -4.14816737e-01 -1.01829380e-01
-8.02950934e-02 -6.33685589e-01 1.08996224e+00 -9.36701477e-01
-1.66421950e+00 5.80355525e-01 8.40537697e-02 -5.81230402e-01
5.64377010e-01 3.25503498e-02 -4.60732132e-01 4.96440560e-01
1.23176523e-01 8.13151658e-01 8.52691531e-01 -1.03268778e+00
-6.20402172e-02 -4.86380368e-01 -1.87001407e-01 3.66615579e-02
1.62565336e-02 -1.63720220e-01 -3.78858477e-01 -6.20588124e-01
2.82152236e-01 -8.95724773e-01 -4.10315990e-01 -1.34242579e-01
-3.09368759e-01 5.30651987e-01 4.26467299e-01 -8.91580880e-01
1.22671211e+00 -1.90250897e+00 5.02551869e-02 3.40201944e-01
2.94150770e-01 -1.28483428e-02 -9.67786089e-02 -4.66300361e-02
-3.30135733e-01 2.60848664e-02 -4.03302103e-01 2.72354037e-01
-4.60482121e-01 -3.58361393e-01 1.11752093e-01 6.65623844e-01
-2.10185461e-02 8.83846879e-01 -8.48911941e-01 -3.51787388e-01
3.62406850e-01 8.37154508e-01 -3.99760336e-01 -1.01334721e-01
1.47344694e-01 1.04784703e+00 -3.69583666e-01 3.67235988e-01
7.33207047e-01 -5.06526753e-02 2.18451455e-01 -6.81715429e-01
-4.12941843e-01 -3.62567455e-02 -8.37580144e-01 1.94433248e+00
-4.72424924e-01 7.19129741e-01 4.76761907e-02 -1.05521929e+00
9.62728977e-01 4.08598036e-01 9.52407837e-01 -1.10990727e+00
3.47943395e-01 4.54698503e-01 3.98988456e-01 -7.65993178e-01
2.64824599e-01 -2.73753345e-01 2.32578307e-01 4.67473656e-01
6.20400570e-02 1.11263888e-02 -4.94964011e-02 -1.33126155e-01
8.26565325e-01 -7.45375548e-03 -2.24457055e-01 -6.06089652e-01
6.28088295e-01 -3.11988801e-01 2.41021290e-01 9.84534800e-01
-3.80375773e-01 7.38412380e-01 2.93041140e-01 -5.96042812e-01
-1.03217518e+00 -6.73675358e-01 -6.49703622e-01 5.14004707e-01
2.44257912e-01 1.72515839e-01 -9.18438852e-01 -3.71330649e-01
9.42408387e-03 5.66322923e-01 -1.04512560e+00 -9.60109606e-02
-6.57565892e-01 -1.11651242e+00 4.47501063e-01 3.94691795e-01
4.51417685e-01 -1.19588506e+00 -1.01778364e+00 5.88606536e-01
-3.01557869e-01 -9.68933284e-01 -4.16414022e-01 2.91819066e-01
-1.13159633e+00 -7.75235474e-01 -1.05765545e+00 -4.07550365e-01
5.41734099e-01 6.21541776e-02 7.99832284e-01 -2.67819762e-01
-5.33025861e-01 1.52547896e-01 -2.84231693e-01 -1.89155117e-01
-3.30670267e-01 3.91643167e-01 -2.88582861e-01 2.38914758e-01
-1.52994350e-01 -4.71265405e-01 -9.08913553e-01 2.24352807e-01
-7.94457138e-01 6.49198890e-02 8.89647782e-01 9.81624365e-01
9.53402340e-01 -2.37238869e-01 7.19132781e-01 -1.04438579e+00
5.61758459e-01 -3.87693703e-01 -3.59094977e-01 2.00215057e-01
-6.41656816e-01 2.32032716e-01 5.22256792e-01 -4.64374483e-01
-9.35239315e-01 1.92081153e-01 -9.64636728e-02 -3.76714200e-01
-9.10655409e-02 3.08068216e-01 2.96593606e-01 -4.97804016e-01
6.87517524e-01 4.99337822e-01 3.02452117e-01 -2.90053308e-01
1.41340494e-02 4.56334591e-01 6.08024240e-01 -3.22129548e-01
1.23465404e-01 4.50352997e-01 1.70859825e-02 -8.23437274e-01
-2.96267807e-01 -1.15408286e-01 -7.27963626e-01 -6.13335907e-01
9.29441094e-01 -4.53740478e-01 -4.96044129e-01 5.85768163e-01
-1.10112762e+00 -3.42714220e-01 -1.07776307e-01 7.59809971e-01
-5.48632383e-01 2.90293247e-01 -4.79774207e-01 -4.98129100e-01
-6.14317119e-01 -1.62774539e+00 8.05922508e-01 1.30657658e-01
-8.02630633e-02 -8.09073091e-01 -2.67440099e-02 3.71935904e-01
1.09563017e+00 3.66777927e-01 8.74529719e-01 -1.96434423e-01
-4.52382594e-01 -2.68529598e-02 -1.59408316e-01 5.57135837e-03
1.85336433e-02 -3.69025052e-01 -1.08524966e+00 -2.14913592e-01
-2.84875534e-03 1.47374153e-01 7.39022911e-01 1.23470616e+00
1.47386181e+00 2.48393685e-01 -1.85593233e-01 8.27817142e-01
1.35516763e+00 5.01527846e-01 8.79011631e-01 4.08027261e-01
5.99125504e-01 5.80736995e-01 2.78827906e-01 3.88180584e-01
3.31717990e-02 9.39051032e-01 2.95397967e-01 -1.49503335e-01
-4.89623696e-01 1.49701059e-01 -7.18314871e-02 6.60448551e-01
-2.36378178e-01 2.78250992e-01 -8.75506103e-01 4.70179498e-01
-1.35390031e+00 -7.96547651e-01 -3.05460066e-01 2.30074239e+00
6.49186671e-01 -2.23375902e-01 -1.19840261e-03 3.17028388e-02
9.73652601e-01 -6.56792969e-02 -7.57939279e-01 -2.35200047e-01
7.78089538e-02 2.57015973e-01 8.58472764e-01 5.66634059e-01
-1.05722189e+00 4.42109585e-01 6.69520473e+00 6.19710028e-01
-1.82563484e+00 3.08056682e-01 6.73299432e-01 -5.28911538e-02
-4.95644897e-01 -2.99898177e-01 -4.02539104e-01 6.91887915e-01
1.08170593e+00 9.81120765e-02 5.63936353e-01 2.87621826e-01
4.42053139e-01 -1.20715074e-01 -5.10760069e-01 1.07539654e+00
1.16284892e-01 -1.31472337e+00 -2.31163889e-01 -2.42110670e-01
5.36633432e-01 9.95667204e-02 7.15975538e-02 -1.96435615e-01
-7.73802638e-01 -1.16645062e+00 8.55438769e-01 7.37327218e-01
1.13355517e+00 -5.65150261e-01 8.24227452e-01 -1.99761286e-01
-8.43719542e-01 6.89085424e-02 -2.14238182e-01 4.29514259e-01
2.09512115e-01 8.43122542e-01 -8.28770757e-01 4.85390306e-01
7.50393093e-01 3.76711160e-01 -4.79546368e-01 1.25128305e+00
6.32769987e-02 4.20771182e-01 -1.60405695e-01 2.58581769e-02
-2.01768950e-02 1.35068133e-01 5.36301851e-01 1.33917582e+00
3.58958870e-01 -7.35290125e-02 -2.77153790e-01 1.04322422e+00
3.20680439e-01 2.40652487e-01 -2.81609148e-01 2.77741924e-02
5.39937079e-01 1.30520761e+00 -1.00615728e+00 -6.57585934e-02
-1.98794499e-01 8.33910465e-01 -1.83773473e-01 2.58241028e-01
-9.59446073e-01 -5.29201448e-01 -3.45184319e-02 5.81874371e-01
1.26658157e-01 4.15197238e-02 -2.92109400e-01 -7.99090445e-01
-1.54314816e-01 -5.64567029e-01 -7.17430115e-02 -6.60493433e-01
-8.95572841e-01 9.02614772e-01 -8.86604562e-02 -8.92021298e-01
-5.12998700e-02 -3.17872107e-01 -2.82482028e-01 1.01316524e+00
-1.54717207e+00 -8.65091264e-01 -4.37494904e-01 4.73062754e-01
1.75111681e-01 6.32013455e-02 6.61564410e-01 6.92019343e-01
-3.69159788e-01 7.30635285e-01 2.42021661e-02 -2.84590870e-01
7.62534857e-01 -1.16581774e+00 1.22189388e-01 5.25145888e-01
-2.65589207e-01 6.07429385e-01 6.36654437e-01 -5.11377811e-01
-1.33758008e+00 -8.58696461e-01 5.20071447e-01 1.31286114e-01
3.06480169e-01 -6.95868805e-02 -9.13188994e-01 1.89608961e-01
-8.54666978e-02 2.31551394e-01 4.50262576e-01 -4.23366219e-01
8.77635404e-02 -3.04910868e-01 -1.35010600e+00 1.16128117e-01
4.66234446e-01 -4.24054921e-01 -6.62615970e-02 3.92850004e-02
3.95940602e-01 -5.82463861e-01 -1.16396880e+00 2.42273003e-01
1.02864182e+00 -1.05753243e+00 8.45589459e-01 2.51235180e-02
2.19278708e-01 -3.70130330e-01 6.88389838e-02 -1.29578626e+00
-6.00681543e-01 -4.58945364e-01 7.46839270e-02 5.92203259e-01
4.93206352e-01 -7.98837006e-01 7.33782589e-01 4.90766615e-01
-4.38921541e-01 -7.71575451e-01 -1.08652604e+00 -4.46668178e-01
6.70689270e-02 -3.52474779e-01 5.06580234e-01 9.32375133e-01
-3.37215096e-01 4.86688279e-02 -3.64246845e-01 7.19377026e-02
6.11806154e-01 -6.05309233e-02 8.03651959e-02 -9.17369127e-01
-4.51688200e-01 -4.60414052e-01 -4.38382506e-01 -6.54967368e-01
-1.80098549e-01 -9.97818470e-01 1.26998425e-01 -1.18775165e+00
2.82155812e-01 -6.06234252e-01 -6.08718336e-01 2.11303607e-01
1.51016545e-02 3.37196410e-01 1.97969258e-01 2.11859509e-01
-1.18487209e-01 2.44698152e-01 1.81824386e+00 -7.93787017e-02
-1.86700717e-01 -2.98002124e-01 -5.47401249e-01 1.13377638e-01
8.18137228e-01 -2.94488996e-01 -3.89075488e-01 -5.49423277e-01
-1.25447288e-01 4.92309302e-01 2.85020918e-01 -1.18118870e+00
1.26110598e-01 3.92704278e-01 6.37318909e-01 -4.13944155e-01
-2.26940792e-02 -7.23305941e-01 5.21544278e-01 7.71700799e-01
-5.27601838e-01 -2.57392898e-02 3.36440444e-01 2.58645296e-01
-1.94603533e-01 -3.36700737e-01 1.09299076e+00 -2.37834424e-01
-6.44579649e-01 1.56759158e-01 -3.99442732e-01 -2.79525936e-01
9.95887220e-01 -3.51097405e-01 6.55611977e-02 -1.56470552e-01
-7.70631015e-01 -1.86194330e-01 6.49783686e-02 1.74880385e-01
5.76291919e-01 -1.15874410e+00 -7.31014013e-01 4.39480156e-01
-2.36647591e-01 -4.22554135e-01 9.84142661e-01 1.46707571e+00
-6.92970157e-01 5.85450709e-01 -6.17504478e-01 -6.89592719e-01
-7.20718145e-01 2.92427391e-01 8.89463305e-01 -1.45881519e-01
-5.86994827e-01 8.54726851e-01 2.79903207e-02 -3.90740484e-01
-5.13268262e-02 -3.84518981e-01 -3.18608999e-01 2.19043214e-02
5.86674213e-01 2.73544192e-01 3.97155702e-01 -5.91218114e-01
-4.55688179e-01 7.93612838e-01 -1.82279721e-01 -1.40645623e-01
1.18450630e+00 -1.94060922e-01 -3.02947871e-02 2.80594617e-01
1.05376172e+00 -1.78529575e-01 -1.23576045e+00 2.30578780e-01
-3.53891164e-01 -4.43464994e-01 6.74674034e-01 -1.28862345e+00
-1.51770163e+00 9.48678613e-01 1.19927871e+00 1.28197940e-02
1.18482852e+00 -2.17578948e-01 7.67228663e-01 -3.28247964e-01
3.97647023e-01 -8.76987457e-01 -2.85054147e-01 2.07765013e-01
9.78608251e-01 -1.02616966e+00 -2.53107697e-01 -3.07216018e-01
-5.01611650e-01 1.27417767e+00 3.99815083e-01 9.31465253e-02
4.14092779e-01 3.66343319e-01 1.58214629e-01 -2.83011526e-01
-2.85844594e-01 3.03228945e-01 3.25426072e-01 6.82383597e-01
8.84610534e-01 6.70245960e-02 -5.85936189e-01 5.84907651e-01
9.45941955e-02 2.32131228e-01 3.33394915e-01 5.74457645e-01
-2.25050822e-01 -8.68734956e-01 -5.26392937e-01 5.75234771e-01
-8.27006578e-01 -5.04969880e-02 1.97032213e-01 6.35921657e-01
2.23260179e-01 3.46449524e-01 -5.42544294e-03 -3.97570699e-01
2.53858447e-01 -1.65700674e-01 8.47783625e-01 -2.29318798e-01
-7.72212505e-01 2.60193884e-01 -2.03593239e-01 -6.76806688e-01
-5.24516821e-01 -7.13246644e-01 -1.31316459e+00 6.91989586e-02
-4.13886458e-01 3.33064832e-02 1.00828648e+00 6.65267825e-01
3.94381464e-01 9.59491611e-01 5.29628098e-01 -1.02459538e+00
-2.75857031e-01 -1.01745355e+00 -7.82706738e-01 3.54099661e-01
1.99096799e-01 -7.93771684e-01 -6.82233796e-02 -1.02727830e-01] | [13.609705924987793, -2.404474973678589] |
3b840650-dd71-463e-affc-5be981d39f25 | soft-prompt-tuning-for-large-language-models | 2306.04735 | null | https://arxiv.org/abs/2306.04735v1 | https://arxiv.org/pdf/2306.04735v1.pdf | Soft-prompt Tuning for Large Language Models to Evaluate Bias | Prompting large language models has gained immense popularity in recent years due to the advantage of producing good results even without the need for labelled data. However, this requires prompt tuning to get optimal prompts that lead to better model performances. In this paper, we explore the use of soft-prompt tuning on sentiment classification task to quantify the biases of large language models (LLMs) such as Open Pre-trained Transformers (OPT) and Galactica language model. Since these models are trained on real-world data that could be prone to bias toward certain groups of populations, it is important to identify these underlying issues. Using soft-prompts to evaluate bias gives us the extra advantage of avoiding the human-bias injection that can be caused by manually designed prompts. We check the model biases on different sensitive attributes using the group fairness (bias) and find interesting bias patterns. Since LLMs have been used in the industry in various applications, it is crucial to identify the biases before deploying these models in practice. We open-source our pipeline and encourage industry researchers to adapt our work to their use cases. | ['Faiza Khan Khattak', 'Laleh Seyyed-Kalantari', 'Deval Pandya', 'Sevil Zanjani Miyandoab', 'David Emerson', 'Jacob-Junqi Tian'] | 2023-06-07 | null | null | null | null | ['sentiment-analysis'] | ['natural-language-processing'] | [-5.76954335e-03 1.97432950e-01 -2.07727224e-01 -8.90542328e-01
-5.56562483e-01 -6.58808291e-01 6.67005241e-01 4.80286598e-01
-6.00098073e-01 4.70747322e-01 3.48976851e-01 -4.81157333e-01
1.01898260e-01 -5.73911965e-01 -5.73878169e-01 -5.52743018e-01
1.35236591e-01 3.57173979e-01 1.80292979e-01 -2.11586520e-01
5.61835885e-01 1.25234976e-01 -1.33878696e+00 4.61649567e-01
9.21399772e-01 8.03102493e-01 -9.21020582e-02 3.34462374e-01
-2.64755338e-01 9.67136919e-01 -8.09312820e-01 -8.38800848e-01
4.00449574e-01 -1.35161117e-01 -6.48303151e-01 -4.25943524e-01
5.78072429e-01 -1.17437243e-01 6.21694505e-01 1.14066482e+00
7.56635606e-01 -2.03571528e-01 6.15220189e-01 -1.51326072e+00
-4.42947060e-01 7.77960896e-01 -6.72842801e-01 2.93878406e-01
7.03511983e-02 2.82904625e-01 1.14190793e+00 -4.15697873e-01
4.29691821e-01 1.56509233e+00 6.55747712e-01 4.40963209e-01
-1.27204776e+00 -1.04258454e+00 2.68697500e-01 -9.95404199e-02
-9.91597772e-01 -4.82945025e-01 6.07404590e-01 -7.12067485e-01
7.76338041e-01 2.33349428e-01 6.16856627e-02 1.40422797e+00
2.53742486e-01 6.02321088e-01 1.56698811e+00 -4.58375722e-01
4.42046314e-01 8.20296407e-01 5.26406646e-01 3.14188629e-01
3.73475701e-01 2.93355465e-01 -5.23430467e-01 -5.46759069e-01
2.79657304e-01 -1.83104441e-01 7.62792081e-02 -1.56808257e-01
-1.01320326e+00 1.17208183e+00 5.06686687e-01 1.00568414e-01
-4.66101021e-01 -8.67801383e-02 3.85435998e-01 2.52544940e-01
7.24447787e-01 9.86455917e-01 -8.29236865e-01 -1.36371478e-01
-9.13738132e-01 4.52673674e-01 7.22790062e-01 4.40710872e-01
6.38583660e-01 -3.88718545e-01 -2.74138004e-01 1.12998092e+00
3.54012072e-01 3.76610190e-01 5.03448606e-01 -8.26873124e-01
3.46120596e-01 8.58626544e-01 2.55708396e-01 -8.84757638e-01
-5.31626761e-01 -4.79492128e-01 -5.32394052e-01 1.02830566e-01
6.91831887e-01 -3.35894048e-01 -5.09826243e-01 1.77876604e+00
3.58642071e-01 -3.10447842e-01 -7.06820562e-02 8.81123424e-01
5.99579275e-01 4.07459766e-01 6.07037604e-01 1.21188335e-01
1.50318217e+00 -5.73246837e-01 -5.09959877e-01 -4.76247847e-01
9.44217622e-01 -9.45196807e-01 1.68688595e+00 2.31695563e-01
-6.59850657e-01 -2.42203981e-01 -7.29383945e-01 2.89584190e-01
-3.91514778e-01 -2.67706454e-01 7.63881266e-01 9.67945099e-01
-7.30037689e-01 4.53607827e-01 -3.75849098e-01 -5.92647672e-01
3.98474157e-01 4.22785342e-01 4.78347726e-02 1.51500136e-01
-1.41737556e+00 9.23430562e-01 -4.26805858e-03 -2.77125567e-01
-4.87283647e-01 -9.98261452e-01 -5.62787831e-01 -8.67287964e-02
3.67916316e-01 -3.84397238e-01 1.48562908e+00 -1.46956325e+00
-1.07679915e+00 9.15137589e-01 -1.40251592e-01 -2.28366166e-01
6.60879672e-01 -2.15163633e-01 -2.16073349e-01 -5.56912303e-01
1.66104451e-01 7.61766016e-01 6.92498684e-01 -1.01568234e+00
-6.46910012e-01 -2.60746270e-01 2.64592022e-01 -1.46353081e-01
-6.80731356e-01 7.54741788e-01 1.88323289e-01 -5.55271506e-01
-5.51112950e-01 -1.08987582e+00 -4.95342195e-01 -4.95217860e-01
-2.09284946e-01 -3.51510942e-01 3.87605190e-01 -5.03073037e-01
1.18022943e+00 -1.89967668e+00 -5.08301377e-01 2.38731623e-01
-2.09795274e-02 2.46815056e-01 -1.82849616e-01 2.88227588e-01
-6.63986756e-03 4.70639080e-01 1.41344339e-01 -3.52936164e-02
2.18326617e-02 -7.05562308e-02 -3.72163564e-01 1.86951607e-01
3.05619448e-01 7.19633520e-01 -7.71984220e-01 -5.43275356e-01
-9.06887576e-02 2.09245920e-01 -8.70970309e-01 3.23779076e-01
-2.26139307e-01 2.25257307e-01 -2.11061105e-01 5.06616414e-01
5.89772761e-01 -3.69354039e-01 -3.99352275e-02 1.73966791e-02
-1.16755245e-02 6.56673610e-01 -1.02327859e+00 7.72695959e-01
-5.22844732e-01 3.64097118e-01 5.29250018e-02 -5.19764125e-01
8.41248155e-01 4.87765968e-02 1.35733590e-01 -9.22495902e-01
2.81230658e-01 3.64863783e-01 3.67596149e-01 -4.18576092e-01
5.87169111e-01 -3.70086193e-01 -1.80374607e-01 7.35223413e-01
-2.83971816e-01 3.32701835e-03 -3.96872088e-02 2.53043622e-01
8.27503026e-01 -4.42208230e-01 2.24999875e-01 -6.27155304e-01
2.02984720e-01 -1.55354425e-01 5.86893737e-01 6.41099930e-01
-2.93286949e-01 3.76409441e-01 7.74017930e-01 -4.85617161e-01
-7.40265846e-01 -5.22666574e-01 -1.57034546e-01 1.54206002e+00
-2.79041708e-01 -3.54836464e-01 -6.75358891e-01 -1.01690328e+00
1.13268733e-01 9.05401707e-01 -7.16058433e-01 -1.00004420e-01
-1.94336489e-01 -1.15181828e+00 3.35443825e-01 4.76725519e-01
1.68733329e-01 -1.07149673e+00 -7.57693410e-01 -9.98213887e-02
-2.26338074e-01 -1.01633489e+00 -3.56797844e-01 3.16243798e-01
-6.04817331e-01 -9.09858525e-01 -6.83424652e-01 -3.02814960e-01
8.11778724e-01 -2.47500706e-02 1.43605101e+00 9.19362437e-03
1.14903808e-01 1.23662136e-01 -3.10854375e-01 -1.08303320e+00
-6.72628582e-01 4.14464414e-01 1.05588451e-01 -3.78243835e-03
6.89164221e-01 -3.25219259e-02 -4.37920392e-01 7.03435242e-01
-8.03473771e-01 -2.93345144e-03 5.97358406e-01 5.89028478e-01
-4.08246443e-02 -1.87928095e-01 8.54557455e-01 -1.39386344e+00
1.08338368e+00 -5.28510749e-01 -8.41748238e-01 1.66559562e-01
-8.96456838e-01 2.36828119e-01 4.64196473e-01 -6.02895141e-01
-8.88941765e-01 -4.17912900e-01 -9.92731377e-02 1.77696720e-01
-1.39367580e-01 4.24392909e-01 -9.80094522e-02 7.10718483e-02
1.10179842e+00 -7.38952994e-01 -2.06179410e-01 -4.12744671e-01
1.21231982e-02 1.04422247e+00 -2.43314818e-01 -7.36976802e-01
4.84542876e-01 1.55935436e-01 -3.89871806e-01 -4.61104929e-01
-1.15718842e+00 -4.04398173e-01 -1.77203774e-01 -1.05470188e-01
5.36795318e-01 -9.48263168e-01 -6.36423528e-01 3.13253194e-01
-9.77812707e-01 -7.09123075e-01 1.15502225e-02 1.19566776e-01
3.38235758e-02 -2.81934477e-02 -3.16481650e-01 -9.20148313e-01
-4.40133482e-01 -1.13609052e+00 9.15001988e-01 3.49589080e-01
-9.19340074e-01 -1.01188338e+00 9.10210460e-02 3.99915457e-01
6.85915530e-01 -1.01716585e-01 9.74442005e-01 -1.02954602e+00
-1.51878268e-01 -1.78606242e-01 -2.29097024e-01 4.17705268e-01
6.80589527e-02 1.73796177e-01 -1.29504144e+00 -3.09121042e-01
-2.26819992e-01 -5.99001884e-01 5.50160229e-01 3.51321459e-01
1.04052055e+00 -3.35868329e-01 -1.62862465e-01 1.82738036e-01
1.08338904e+00 -1.38671756e-01 2.24673197e-01 4.96522903e-01
5.35424411e-01 1.18023336e+00 7.51688421e-01 2.82795817e-01
6.15287364e-01 5.27345717e-01 1.80929944e-01 -2.94380695e-01
3.53858620e-01 -2.37223461e-01 5.57502925e-01 3.41360271e-01
2.06563979e-01 1.03010773e-03 -1.14448917e+00 5.32153010e-01
-1.66161370e+00 -6.10894203e-01 -2.13055164e-01 2.38421845e+00
8.95122707e-01 5.44819236e-01 3.69437248e-01 -6.90193921e-02
4.94917452e-01 1.77410334e-01 -3.78531665e-01 -8.92755449e-01
-2.89881229e-03 -4.00428995e-02 6.93062007e-01 4.48207349e-01
-8.47188056e-01 8.01862717e-01 6.42911482e+00 5.39204597e-01
-1.46201253e+00 4.12118696e-02 1.15203452e+00 -3.28357428e-01
-5.37007391e-01 1.45099789e-01 -1.05240440e+00 6.05772674e-01
1.27293074e+00 3.72996144e-02 2.85784900e-01 8.60067070e-01
4.47771370e-01 -3.55165541e-01 -1.09398401e+00 7.80348837e-01
-2.17097655e-01 -8.10842216e-01 -2.53873855e-01 1.56146795e-01
8.75361741e-01 1.92917213e-01 6.12191819e-02 5.99134922e-01
7.67224193e-01 -9.98601973e-01 7.58964002e-01 -2.97112893e-02
4.06542510e-01 -6.06215179e-01 1.01054120e+00 5.14641047e-01
-2.41696626e-01 -2.66063005e-01 -4.77439940e-01 -4.11396980e-01
-1.54866368e-01 1.11190546e+00 -1.21405160e+00 -1.44833699e-01
8.39335144e-01 2.45324269e-01 -9.66534317e-01 7.79297948e-01
-2.27871150e-01 1.11644578e+00 -3.71957809e-01 -4.29183722e-01
6.71704561e-02 1.40162900e-01 7.17187896e-02 1.17411017e+00
-1.67493541e-02 -3.16362590e-01 -1.33274635e-02 7.87525773e-01
-1.25052020e-01 4.79119837e-01 -4.55904335e-01 -2.34571725e-01
3.58514369e-01 1.57444024e+00 -7.18020856e-01 -1.89902812e-01
-4.86226082e-01 4.38735992e-01 2.28952542e-01 1.99421197e-01
-6.37220025e-01 -5.73126189e-02 7.16507971e-01 4.82698232e-01
-3.07449132e-01 2.86516458e-01 -6.18158996e-01 -9.19221520e-01
-1.48114249e-01 -1.37671185e+00 4.81505245e-01 -5.26787639e-01
-1.53332460e+00 3.67804796e-01 -3.98903824e-02 -7.64595985e-01
-2.72411734e-01 -4.91692573e-01 -6.17799938e-01 1.02391875e+00
-1.63030601e+00 -9.89160955e-01 -2.17776090e-01 1.80580661e-01
2.69729704e-01 -5.10207098e-03 6.66769147e-01 3.49949986e-01
-5.24170518e-01 8.45680714e-01 -4.59494680e-01 -6.99468702e-02
1.50537312e+00 -1.22537327e+00 5.03264666e-01 4.81478661e-01
-8.16107318e-02 9.10449624e-01 1.03606451e+00 -5.88664949e-01
-5.25787950e-01 -9.52340126e-01 1.30017304e+00 -9.79101479e-01
5.93160570e-01 -6.27011001e-01 -8.69952381e-01 5.79619110e-01
2.88586676e-01 -2.29088575e-01 1.04765475e+00 7.28410840e-01
-4.56391633e-01 -2.91520566e-01 -1.06096113e+00 6.26367569e-01
6.36701763e-01 -3.28523695e-01 -3.37877274e-01 2.31382042e-01
5.32512069e-01 -1.07182994e-01 -5.68469703e-01 3.67688209e-01
5.02125859e-01 -1.01968133e+00 5.16240180e-01 -7.64025807e-01
5.46801090e-01 1.95474643e-02 -9.16105509e-03 -1.58752060e+00
-2.02877492e-01 -4.25077587e-01 5.96596777e-01 1.67751563e+00
8.13768506e-01 -1.00924575e+00 5.71985364e-01 1.25399959e+00
5.17104745e-01 -6.85089290e-01 -2.52712339e-01 -4.16369975e-01
2.51734346e-01 -4.78560597e-01 8.33642483e-01 1.23748970e+00
-1.30992770e-01 4.86957043e-01 -2.73365527e-01 1.39185429e-01
3.88105184e-01 -8.18114653e-02 1.00149703e+00 -1.38645339e+00
-1.52125701e-01 -4.16291237e-01 2.35588271e-02 -4.27286863e-01
1.86423063e-01 -7.39746451e-01 -8.54171440e-02 -1.14922869e+00
4.56887811e-01 -8.70114267e-01 -5.35320878e-01 6.83970332e-01
-6.13912463e-01 5.94457015e-02 1.72525167e-01 1.63924992e-02
-6.13242805e-01 1.99582070e-01 7.84913003e-01 1.15587421e-01
-1.44484386e-01 1.44127205e-01 -1.42114735e+00 8.33723366e-01
8.60493481e-01 -8.30260813e-01 -3.01212996e-01 -2.23840818e-01
8.66658390e-01 -6.96030259e-01 3.13139498e-01 -6.21646583e-01
-1.72200069e-01 -2.90567577e-01 1.87582165e-01 -1.92778081e-01
-2.23938674e-01 -8.09965432e-01 -2.09650889e-01 2.66928494e-01
-6.74436748e-01 2.43486539e-01 2.39491254e-01 3.35780941e-02
5.17265056e-04 -2.56104022e-01 8.26259315e-01 -1.80032521e-01
-3.58057767e-01 -1.21172488e-01 -2.43634000e-01 2.82367170e-01
8.10482144e-01 3.19270790e-01 -5.11839986e-01 -4.20695573e-01
9.44818631e-02 4.02478546e-01 7.54093468e-01 7.75073469e-01
-1.36089310e-01 -9.38368976e-01 -6.12574875e-01 1.81592360e-01
4.22278225e-01 -3.26646060e-01 -8.41322765e-02 9.00518775e-01
-1.44440919e-01 2.86190450e-01 -9.27476734e-02 -6.38760567e-01
-1.30781865e+00 4.81504619e-01 1.90747023e-01 -4.57227439e-01
5.53665347e-02 8.97505462e-01 3.44679981e-01 -8.01403165e-01
1.22548975e-01 -5.12035072e-01 -1.72527418e-01 2.78866589e-01
5.56071639e-01 3.22806299e-01 2.14110851e-01 -3.27510625e-01
-5.37817776e-01 2.79491663e-01 -2.53527343e-01 -7.71788731e-02
1.09879041e+00 1.19993545e-01 -1.04273073e-01 5.68979442e-01
9.99699771e-01 4.36439455e-01 -1.05595410e+00 -1.26411527e-01
3.85893643e-01 -4.98741627e-01 9.11065415e-02 -1.26115716e+00
-8.54839206e-01 9.45474505e-01 6.46687210e-01 3.84694338e-01
7.86219716e-01 -1.09521665e-01 4.81384248e-01 -3.01286541e-02
4.13620949e-01 -1.19868934e+00 -8.46580509e-03 3.39365900e-01
6.62923455e-01 -1.77728403e+00 -1.08112134e-01 -5.73715605e-02
-1.01452744e+00 5.27247727e-01 6.87923908e-01 2.05322891e-01
5.02247870e-01 3.38567942e-01 9.19278681e-01 -2.32019454e-01
-9.30832207e-01 1.84712559e-01 1.45241752e-01 3.39219302e-01
8.90592694e-01 2.45887846e-01 -4.08463806e-01 8.54028702e-01
-4.95959610e-01 -3.46707627e-02 3.90237868e-01 6.26359820e-01
-1.11313738e-01 -1.35344732e+00 -6.72807813e-01 7.31618345e-01
-8.58730435e-01 -1.30419701e-01 -7.75984466e-01 5.20960093e-01
5.41417934e-02 1.12758470e+00 -1.07613362e-01 -2.95122206e-01
4.70442384e-01 2.23045349e-01 2.54481938e-02 -6.59566402e-01
-1.14912868e+00 -2.05817714e-01 3.66334021e-01 -5.41774452e-01
-3.03053379e-01 -7.66561091e-01 -9.72547352e-01 -3.31322253e-01
-3.99735153e-01 1.44841194e-01 7.74618804e-01 7.73781657e-01
4.20941025e-01 3.57605398e-01 4.73249704e-01 -4.52040941e-01
-8.01406085e-01 -1.17333007e+00 -2.33962849e-01 7.42733419e-01
1.01746477e-01 -5.74054539e-01 -4.43507880e-01 -2.80632734e-01] | [9.797323226928711, 7.808390140533447] |
41680c12-9b57-4037-b164-07f916611468 | motion-inductive-self-supervised-object | 2210.00221 | null | https://arxiv.org/abs/2210.00221v1 | https://arxiv.org/pdf/2210.00221v1.pdf | Motion-inductive Self-supervised Object Discovery in Videos | In this paper, we consider the task of unsupervised object discovery in videos. Previous works have shown promising results via processing optical flows to segment objects. However, taking flow as input brings about two drawbacks. First, flow cannot capture sufficient cues when objects remain static or partially occluded. Second, it is challenging to establish temporal coherency from flow-only input, due to the missing texture information. To tackle these limitations, we propose a model for directly processing consecutive RGB frames, and infer the optical flow between any pair of frames using a layered representation, with the opacity channels being treated as the segmentation. Additionally, to enforce object permanence, we apply temporal consistency loss on the inferred masks from randomly-paired frames, which refer to the motions at different paces, and encourage the model to segment the objects even if they may not move at the current time point. Experimentally, we demonstrate superior performance over previous state-of-the-art methods on three public video segmentation datasets (DAVIS2016, SegTrackv2, and FBMS-59), while being computationally efficient by avoiding the overhead of computing optical flow as input. | ['Qi Tian', 'Hongkai Xiong', 'Xiaopeng Zhang', 'Rui Qian', 'Yabo Chen', 'Weidi Xie', 'Shuangrui Ding'] | 2022-10-01 | null | null | null | null | ['object-discovery', 'unsupervised-object-segmentation', 'object-discovery-in-videos'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 3.17113936e-01 -1.98023617e-01 -7.81049579e-02 -1.48190230e-01
-1.86806381e-01 -5.57439864e-01 3.24730456e-01 -1.33919239e-01
-5.05608201e-01 7.62821555e-01 -2.71416381e-02 3.78608704e-02
6.73624650e-02 -6.55948102e-01 -7.44087875e-01 -6.45518363e-01
-2.46890366e-01 1.06322847e-01 9.30454195e-01 3.09401125e-01
2.66153961e-01 4.72011030e-01 -1.79508889e+00 4.23691869e-01
9.50571001e-01 1.09725022e+00 2.61535585e-01 8.50968361e-01
-7.42849857e-02 1.20100534e+00 -2.71575987e-01 -2.42868990e-01
3.18948805e-01 -4.33437020e-01 -1.19752550e+00 4.48299587e-01
8.80475163e-01 -7.75915444e-01 -5.89034915e-01 9.70444858e-01
-1.86791569e-01 5.35845995e-01 4.29001391e-01 -1.35300159e+00
-3.83015931e-01 1.20486259e-01 -6.40058935e-01 5.39135516e-01
4.55525249e-01 4.73173589e-01 8.07242990e-01 -6.48736238e-01
9.25298870e-01 1.20516610e+00 1.89580321e-01 6.53028071e-01
-1.31719959e+00 -4.16806936e-01 6.40933812e-01 4.61432397e-01
-1.13245833e+00 -6.04585826e-01 7.05952108e-01 -7.06666470e-01
6.46714568e-01 3.59539777e-01 8.44996750e-01 9.41597641e-01
-1.86186135e-01 1.07061994e+00 9.63870049e-01 -1.18066752e-02
2.83429265e-01 -2.20535979e-01 1.27016515e-01 7.72212744e-01
1.48470044e-01 -4.69143838e-02 -7.68712044e-01 6.66545406e-02
9.43087935e-01 1.15575027e-02 -6.98813498e-01 -2.20072746e-01
-1.44930387e+00 3.27594876e-01 4.45464760e-01 9.13800672e-02
-4.27735686e-01 2.27123141e-01 9.80839729e-02 -2.59433091e-02
4.02334422e-01 9.04678553e-02 -4.31763977e-01 -2.36998573e-01
-1.19878173e+00 1.44184485e-01 5.56596816e-01 8.89279008e-01
1.05195510e+00 -1.67054534e-01 -2.76402920e-01 1.60800800e-01
3.41459990e-01 1.80000275e-01 9.01296884e-02 -1.51313436e+00
5.02592444e-01 3.41395706e-01 3.71457756e-01 -1.07601249e+00
-1.57164320e-01 1.84908733e-01 -5.99731922e-01 1.78782180e-01
9.12478387e-01 1.17779166e-01 -1.09337056e+00 1.42671633e+00
5.38009882e-01 9.27530706e-01 -4.19859216e-02 1.38140440e+00
7.81076610e-01 6.89440131e-01 1.37913138e-01 -4.13246512e-01
1.09845269e+00 -1.30436623e+00 -7.68705547e-01 -2.86127657e-01
4.16060627e-01 -7.41071999e-01 8.13746274e-01 5.23460150e-01
-1.22194517e+00 -6.65923834e-01 -7.53826022e-01 -1.58955887e-01
1.64631195e-02 -1.94848880e-01 8.37274492e-01 3.15287203e-01
-9.25464809e-01 6.98540688e-01 -1.36927760e+00 -1.29599556e-01
5.59726536e-01 1.46015480e-01 -2.96367973e-01 -2.11251780e-01
-8.93060863e-01 2.00257123e-01 4.92197126e-01 3.57666403e-01
-9.05684710e-01 -7.00750113e-01 -7.25995421e-01 -1.10128634e-01
6.05339289e-01 -6.61279202e-01 8.65992963e-01 -1.13505673e+00
-1.45745623e+00 5.03909826e-01 -5.28763294e-01 -2.90540397e-01
9.52213228e-01 -3.08653682e-01 -7.21258223e-02 5.57133377e-01
6.54042512e-02 9.83059287e-01 7.66465724e-01 -1.32985461e+00
-8.66639793e-01 -9.70705077e-02 4.62105244e-01 1.10356182e-01
-2.40764081e-01 -6.42928854e-02 -1.14257383e+00 -3.98597717e-01
3.69569242e-01 -1.02063322e+00 -2.49912620e-01 3.39167982e-01
-4.94447768e-01 2.99548730e-03 1.14452124e+00 -5.82071126e-01
1.14405203e+00 -2.20855236e+00 3.35943937e-01 -1.20438576e-01
2.01767802e-01 2.30587244e-01 1.19184582e-02 -2.78060675e-01
3.69996458e-01 9.65529308e-02 -5.77099562e-01 -4.92919147e-01
-4.95719820e-01 5.36642790e-01 -2.82553583e-01 4.99428898e-01
3.82691771e-01 8.70529830e-01 -1.25468040e+00 -8.23318541e-01
4.82901931e-01 4.85585451e-01 -7.49346972e-01 2.60595620e-01
-5.49959004e-01 9.88765240e-01 -5.71077824e-01 6.11134171e-01
8.50179195e-01 -4.14557606e-01 -7.87885562e-02 -2.67810553e-01
-1.93009794e-01 2.27878764e-01 -1.43740940e+00 1.87799311e+00
-2.09112670e-02 6.97081566e-01 6.70982450e-02 -7.44666100e-01
1.57652736e-01 2.11947158e-01 8.99557173e-01 -4.78896528e-01
1.70656070e-02 -1.59099344e-02 -1.21865690e-01 -9.02528107e-01
5.52688658e-01 4.94319111e-01 4.50257689e-01 4.93708402e-01
-6.05164580e-02 4.62414697e-03 7.53384888e-01 2.17706129e-01
9.68626857e-01 4.64013815e-01 -4.84674811e-01 9.85957533e-02
5.45719385e-01 -1.03711866e-01 8.62082183e-01 7.16368675e-01
-5.60232282e-01 1.04224622e+00 5.83843410e-01 -4.60205942e-01
-4.64045644e-01 -1.11300385e+00 -9.37877670e-02 8.04963827e-01
7.36310184e-01 -3.51849347e-01 -6.09196246e-01 -8.41747940e-01
-1.55922472e-01 2.02097982e-01 -6.71868503e-01 2.97649473e-01
-8.32057059e-01 -6.26868784e-01 1.36366203e-01 5.03496349e-01
6.10099912e-01 -1.03411758e+00 -1.02996540e+00 3.14342499e-01
-7.88627803e-01 -1.70956898e+00 -4.90692735e-01 -2.61676282e-01
-9.77076054e-01 -1.30258489e+00 -5.83943307e-01 -5.70422232e-01
6.61859989e-01 5.21525562e-01 1.05680180e+00 3.87161642e-01
-3.67827713e-01 2.39921287e-01 -1.92602426e-01 3.10529411e-01
-7.33859278e-03 -1.46472082e-01 -2.91693896e-01 4.80143458e-01
-1.01233535e-01 -3.03919971e-01 -9.86021459e-01 4.07862782e-01
-1.10775912e+00 3.13502878e-01 -3.37651148e-02 4.36091483e-01
7.13636279e-01 -1.76720202e-01 -1.37177080e-01 -7.62941301e-01
-2.06823349e-01 -2.75236845e-01 -4.80220586e-01 2.40971863e-01
-2.31246315e-02 -1.39263406e-01 2.72399157e-01 -4.33992088e-01
-1.15629888e+00 3.02204460e-01 3.68783176e-01 -7.62606561e-01
-2.55719632e-01 1.22592673e-01 6.46442249e-02 4.90605608e-02
2.47698128e-01 1.11703172e-01 -1.77503124e-01 -1.99314892e-01
3.68620664e-01 2.39410758e-01 6.65976524e-01 -5.57259023e-01
5.76357543e-01 1.11898923e+00 -9.93519351e-02 -8.68811309e-01
-8.53475928e-01 -6.50556743e-01 -9.05291617e-01 -5.62762678e-01
1.25623930e+00 -7.63017058e-01 -6.59108222e-01 5.53565860e-01
-1.39414465e+00 -4.86788124e-01 -5.13341650e-02 6.50447905e-01
-5.45858800e-01 6.11181974e-01 -7.79624820e-01 -7.73540318e-01
1.16423890e-01 -1.37009203e+00 9.53921556e-01 4.88959551e-01
-1.10837026e-02 -8.58813882e-01 -2.76570439e-01 4.68713045e-01
9.59975943e-02 3.96432906e-01 3.76916081e-01 2.20193744e-01
-1.32328498e+00 2.96215534e-01 -3.50780964e-01 2.04210714e-01
2.77686507e-01 5.28179705e-01 -1.07923353e+00 -1.16778806e-01
-2.38769159e-01 -2.39363983e-01 1.11179435e+00 4.77299839e-01
1.29152715e+00 -1.66993111e-01 -3.46895188e-01 8.57370853e-01
1.14148808e+00 1.28232777e-01 6.81551039e-01 2.54460454e-01
1.06343842e+00 7.88024366e-01 6.15828395e-01 2.85547823e-01
3.52355480e-01 5.12284994e-01 5.69361031e-01 -9.34998021e-02
-3.62204760e-01 3.25488634e-02 2.55584657e-01 5.36383629e-01
-4.18904215e-01 -4.32330430e-01 -6.26711249e-01 4.98277217e-01
-2.16093540e+00 -9.25023973e-01 -4.51771528e-01 2.18145394e+00
6.22164667e-01 2.36537889e-01 1.80540551e-02 -5.35673276e-02
5.36999047e-01 3.79074275e-01 -5.46494067e-01 2.66631335e-01
-1.76128224e-01 -6.67231977e-02 3.88550520e-01 6.69873238e-01
-1.34666145e+00 1.08277535e+00 5.70917368e+00 1.54381603e-01
-1.24795687e+00 8.06492642e-02 7.71029294e-01 -4.28787410e-01
-1.64868474e-01 2.67068446e-01 -4.43052649e-01 6.04842126e-01
3.88259023e-01 3.75264496e-01 4.68172967e-01 2.03243554e-01
4.08060312e-01 -5.93980908e-01 -1.11367559e+00 8.29532981e-01
-3.80450301e-02 -1.20350385e+00 -7.49571249e-02 -1.73225373e-01
1.03352380e+00 2.49477606e-02 -1.08464271e-01 -2.47804239e-01
-1.03034660e-01 -6.61095738e-01 1.01070833e+00 7.18636096e-01
3.70494246e-01 -3.05618048e-01 2.33200386e-01 1.13520354e-01
-1.28260136e+00 1.57953158e-01 -1.49402946e-01 -1.49767622e-01
5.27143836e-01 6.48050070e-01 -2.32302979e-01 5.53787351e-01
1.04163933e+00 1.04670179e+00 -5.67971468e-01 1.17721677e+00
-2.83112347e-01 5.98194003e-01 -5.23293853e-01 4.71185386e-01
3.46275419e-01 -4.24959958e-01 4.70520735e-01 1.03383005e+00
-2.50820164e-02 9.48801860e-02 4.68489319e-01 1.01491642e+00
1.86386749e-01 -2.81534642e-01 -1.10850088e-01 6.45904392e-02
1.46429881e-01 1.13980532e+00 -1.30129921e+00 -5.34040332e-01
-5.73958158e-01 1.29239750e+00 1.28916904e-01 8.37953329e-01
-9.04845655e-01 -5.68339750e-02 8.57091546e-01 1.98520571e-01
3.23628068e-01 -4.64150399e-01 -7.14410171e-02 -1.51106286e+00
3.38824302e-01 -2.76173949e-01 3.28457177e-01 -7.28475988e-01
-8.96164536e-01 4.70894396e-01 -2.16661748e-02 -1.29147100e+00
-7.53344670e-02 -3.88645470e-01 -4.09005612e-01 5.77804565e-01
-1.73512888e+00 -8.15798819e-01 -5.38298547e-01 5.99720061e-01
6.14143491e-01 6.04642391e-01 1.81024253e-01 4.74587500e-01
-6.64863050e-01 8.82178918e-02 -2.25050911e-01 2.86607146e-01
6.47646606e-01 -1.01619446e+00 1.81272268e-01 1.22925663e+00
2.85769612e-01 3.40608001e-01 4.93663311e-01 -6.64954007e-01
-1.25694144e+00 -1.09024990e+00 6.03022277e-01 -4.55358297e-01
5.60916007e-01 -3.78777176e-01 -1.23318124e+00 5.49301803e-01
-1.00061148e-02 8.22363555e-01 1.37140989e-01 -4.60985690e-01
-8.90151262e-02 1.31904274e-01 -7.16032207e-01 6.00684345e-01
1.27150393e+00 -4.22906041e-01 -1.82271227e-01 2.48593450e-01
7.43007720e-01 -5.97395718e-01 -6.26353264e-01 3.64477187e-01
5.00360191e-01 -1.28409863e+00 1.03016806e+00 -5.15109599e-01
5.70803404e-01 -7.95958579e-01 1.54358894e-01 -7.57745683e-01
5.86185195e-02 -7.56198108e-01 -4.61123824e-01 1.20872760e+00
5.78248799e-02 -3.66828620e-01 9.77212250e-01 8.71783674e-01
5.24592623e-02 -5.18365502e-01 -9.34008956e-01 -5.89530528e-01
-4.57001656e-01 -6.37563229e-01 2.66694695e-01 9.05409098e-01
-4.21606362e-01 -2.00908601e-01 -3.14806432e-01 3.45112711e-01
6.28184676e-01 2.14164168e-01 7.21619368e-01 -9.92179215e-01
-4.04462844e-01 -3.95679414e-01 -4.47201431e-01 -1.44392920e+00
1.94168940e-01 -5.41771591e-01 3.43182653e-01 -1.43264818e+00
1.08115360e-01 -4.63945419e-01 -3.54069948e-01 3.28919798e-01
-5.18370450e-01 5.13327241e-01 4.59885180e-01 4.68900651e-01
-8.66146564e-01 3.55806231e-01 1.70912075e+00 -2.84777969e-01
-3.75889212e-01 -1.66162297e-01 -9.99875590e-02 8.86929333e-01
3.14836293e-01 -3.45069617e-01 -4.71961766e-01 -6.98136032e-01
-1.45184025e-01 2.25039080e-01 6.28679633e-01 -1.01974678e+00
1.60344481e-01 -4.65050876e-01 3.65757644e-01 -5.79453945e-01
3.57141614e-01 -7.34405637e-01 1.75037190e-01 4.96456802e-01
-1.94285125e-01 -1.19146086e-01 4.69878279e-02 8.14761758e-01
-2.57689893e-01 -2.06636831e-01 6.13841116e-01 -9.81711373e-02
-9.79005396e-01 6.91252053e-01 -4.45488632e-01 1.83939382e-01
1.06936514e+00 -3.40454638e-01 -2.09474400e-01 -6.37757778e-03
-6.97965503e-01 3.49872649e-01 6.23690426e-01 6.31336570e-01
6.20591819e-01 -9.21453774e-01 -3.76393765e-01 3.32202703e-01
1.73535403e-02 5.24606884e-01 4.36666310e-01 1.05503356e+00
-7.86051095e-01 7.16657639e-02 -6.74645454e-02 -1.08976948e+00
-9.96000767e-01 5.67451477e-01 2.61973858e-01 1.31384686e-01
-8.37615967e-01 9.04831707e-01 5.05295038e-01 2.94166297e-01
3.04961264e-01 -6.64507389e-01 -2.01003224e-01 1.35468334e-01
5.96159399e-01 4.35324788e-01 -7.44810179e-02 -7.58590758e-01
-4.38351542e-01 7.27244318e-01 -3.28260809e-02 -1.81728914e-01
8.90938222e-01 -3.83648366e-01 -1.05431445e-01 4.74604368e-01
1.16344202e+00 -2.44958892e-01 -2.07612658e+00 -2.37553284e-01
-9.26877633e-02 -1.01695776e+00 -1.24588329e-02 -2.74679363e-01
-1.57269025e+00 8.63403976e-01 3.57828051e-01 1.66795269e-01
1.01863134e+00 -5.07334322e-02 9.24764752e-01 3.88917103e-02
2.97470570e-01 -9.33054090e-01 1.01573214e-01 4.78326738e-01
2.46958017e-01 -1.24109685e+00 -1.91529095e-02 -8.93681586e-01
-4.85030591e-01 1.18944967e+00 8.20051908e-01 8.66413787e-02
5.07346451e-01 1.39262378e-01 5.31942211e-02 1.32128060e-01
-6.94273889e-01 -3.91531587e-01 3.01762819e-01 4.27567661e-01
3.26470703e-01 -2.78549969e-01 -8.57954547e-02 -2.47984454e-01
2.81216204e-01 1.28611580e-01 6.45268202e-01 1.05068600e+00
-1.63310036e-01 -6.88389421e-01 -3.86913389e-01 2.89078504e-01
-4.50564563e-01 1.17132008e-01 -1.57945901e-01 5.92402935e-01
2.22024351e-01 1.06141889e+00 4.78119195e-01 -6.13768473e-02
1.25242174e-01 -2.95203030e-01 6.26330435e-01 -3.63916993e-01
-1.66714862e-01 2.02269673e-01 -2.13908702e-01 -9.73121643e-01
-1.05937350e+00 -8.05615485e-01 -1.53696907e+00 -3.71665396e-02
-2.08574161e-01 -2.40896717e-02 2.27270365e-01 1.05868018e+00
1.80194959e-01 6.15759850e-01 4.83478844e-01 -1.22538459e+00
2.45780021e-01 -4.97732908e-01 -3.21313083e-01 7.79749632e-01
6.43654525e-01 -7.66786933e-01 -4.44558531e-01 6.99676394e-01] | [9.142990112304688, -0.3020530343055725] |
b0deb7ec-6a37-41b4-80a9-cffc00154406 | microscopic-nuclei-classification | 1811.03447 | null | http://arxiv.org/abs/1811.03447v1 | http://arxiv.org/pdf/1811.03447v1.pdf | Microscopic Nuclei Classification, Segmentation and Detection with improved Deep Convolutional Neural Network (DCNN) Approaches | Due to cellular heterogeneity, cell nuclei classification, segmentation, and
detection from pathological images are challenging tasks. In the last few
years, Deep Convolutional Neural Networks (DCNN) approaches have been shown
state-of-the-art (SOTA) performance on histopathological imaging in different
studies. In this work, we have proposed different advanced DCNN models and
evaluated for nuclei classification, segmentation, and detection. First, the
Densely Connected Recurrent Convolutional Network (DCRN) model is used for
nuclei classification. Second, Recurrent Residual U-Net (R2U-Net) is applied
for nuclei segmentation. Third, the R2U-Net regression model which is named
UD-Net is used for nuclei detection from pathological images. The experiments
are conducted with different datasets including Routine Colon Cancer(RCC)
classification and detection dataset, and Nuclei Segmentation Challenge 2018
dataset. The experimental results show that the proposed DCNN models provide
superior performance compared to the existing approaches for nuclei
classification, segmentation, and detection tasks. The results are evaluated
with different performance metrics including precision, recall, Dice
Coefficient (DC), Means Squared Errors (MSE), F1-score, and overall accuracy.
We have achieved around 3.4% and 4.5% better F-1 score for nuclei
classification and detection tasks compared to recently published DCNN based
method. In addition, R2U-Net shows around 92.15% testing accuracy in term of
DC. These improved methods will help for pathological practices for better
quantitative analysis of nuclei in Whole Slide Images(WSI) which ultimately
will help for better understanding of different types of cancer in clinical
workflow. | ['Md Zahangir Alom', 'Tarek M. Taha', 'Vijayan K. Asari', 'Chris Yakopcic'] | 2018-11-08 | null | null | null | null | ['nuclei-classification'] | ['medical'] | [ 2.14578927e-01 -1.19298421e-01 -2.89704204e-01 5.16240448e-02
-9.91598547e-01 -5.12891889e-01 3.15242559e-01 4.21931475e-01
-6.63242519e-01 9.30691898e-01 9.35752094e-02 -4.65957046e-01
1.21015497e-01 -7.54044950e-01 -6.54308274e-02 -1.37760758e+00
-5.70408031e-02 3.36215347e-01 3.06456506e-01 1.49109541e-02
2.91373044e-01 8.32237005e-01 -1.00637925e+00 1.92565724e-01
6.37141347e-01 8.88744712e-01 1.74831189e-02 1.05037248e+00
7.96072707e-02 8.17422152e-01 -2.77109951e-01 8.03687721e-02
-2.62224257e-01 -3.12672257e-01 -8.19081604e-01 -4.16233331e-01
1.68233970e-03 -2.01588318e-01 -3.09253067e-01 1.09898102e+00
9.11225319e-01 -1.87188864e-01 9.17036533e-01 -6.53989851e-01
-3.21518898e-01 5.81438422e-01 -6.93104863e-01 6.08142674e-01
-2.31005222e-01 3.13404292e-01 6.18148983e-01 -4.20571268e-01
9.76284027e-01 7.22196281e-01 9.84742880e-01 6.88223004e-01
-8.81170690e-01 -6.05079770e-01 -7.38809586e-01 -1.51104614e-01
-1.52313256e+00 -3.05579510e-02 5.19107245e-02 -3.25573176e-01
8.75136256e-01 3.57334465e-01 6.35445893e-01 5.95530152e-01
5.80895722e-01 9.28006172e-01 1.03379619e+00 -1.58593506e-01
3.24049592e-02 1.98655147e-02 3.92183959e-01 6.01672530e-01
3.42221886e-01 -2.48154804e-01 2.75411576e-01 1.88762471e-01
7.18937993e-01 1.60307541e-01 -3.65635067e-01 3.43860745e-01
-1.29049432e+00 7.15870321e-01 6.47538841e-01 8.75427783e-01
-1.42960995e-01 2.25141898e-01 1.00962162e+00 -1.47729054e-01
4.75339115e-01 8.64814147e-02 -3.14752817e-01 -9.82235819e-02
-7.83854663e-01 1.75961658e-01 4.37003464e-01 4.85995620e-01
6.18430264e-02 -2.16521516e-01 -7.54037619e-01 8.06785703e-01
2.57612020e-01 4.89052087e-01 1.09095562e+00 -4.87868607e-01
-1.73786998e-01 7.66215384e-01 -2.06264317e-01 -7.94016540e-01
-1.07686603e+00 -7.61124372e-01 -1.43778563e+00 -2.05178112e-01
4.46371198e-01 -1.54028097e-02 -1.18507004e+00 1.36214066e+00
3.58990252e-01 1.61083564e-01 3.45447361e-01 7.93295741e-01
1.29849434e+00 5.73670983e-01 2.19602078e-01 -2.77077943e-01
1.44238985e+00 -7.96700597e-01 -7.25629687e-01 5.14440298e-01
1.51276433e+00 -7.60813534e-01 5.33650517e-01 2.84432676e-02
-8.14753175e-01 -2.44256392e-01 -8.45270574e-01 -1.73663914e-01
-5.99206984e-01 5.74755013e-01 6.67324662e-01 7.07964599e-01
-1.20585513e+00 3.57246071e-01 -1.04633856e+00 -8.75853777e-01
9.09676492e-01 7.44551241e-01 -3.79089028e-01 -8.19372311e-02
-8.49335372e-01 4.90053773e-01 2.69384295e-01 1.77980021e-01
-9.08311665e-01 -8.50242198e-01 -5.06391227e-01 -1.06915623e-01
-1.51166439e-01 -6.14019215e-01 1.23559213e+00 -1.07530043e-01
-1.12434340e+00 1.13105190e+00 8.70078728e-02 -4.67551708e-01
2.68668026e-01 5.95029175e-01 -4.02747281e-02 1.58324704e-01
5.51019572e-02 9.08697546e-01 -3.37610275e-01 -6.94449306e-01
-7.14382529e-01 -4.66501623e-01 -4.66470629e-01 1.73422381e-01
-8.47160295e-02 -2.71269649e-01 -4.26003546e-01 -4.25783008e-01
1.49256587e-01 -1.02925253e+00 -3.32097441e-01 -1.36188775e-01
-6.55031443e-01 -4.19655204e-01 1.06241608e+00 -6.33730173e-01
1.04978526e+00 -2.12222505e+00 -1.65695772e-01 2.33849734e-01
3.86187136e-01 5.77619255e-01 9.29640308e-02 -1.70540452e-01
1.44770458e-01 5.41881561e-01 7.57207572e-02 -2.85578877e-01
-3.11294764e-01 -1.12584770e-01 6.27833307e-01 8.38621557e-01
-2.41857418e-03 1.13876605e+00 -7.69358814e-01 -6.22851372e-01
1.00215793e-01 6.97922349e-01 -2.65897483e-01 9.75416787e-03
1.66742966e-01 3.46974164e-01 -2.77889073e-01 1.24003673e+00
6.69791341e-01 -4.28696752e-01 2.38550812e-01 -4.08101827e-01
1.77029938e-01 -3.11559856e-01 -6.27626896e-01 1.16182780e+00
3.85861956e-02 7.98829436e-01 -2.94673648e-02 -6.76892936e-01
7.65915275e-01 4.21549827e-01 4.47831273e-01 -7.52972364e-01
8.02755177e-01 4.44079459e-01 2.76760817e-01 -5.56042790e-01
2.28667304e-01 -1.89839095e-01 2.47996837e-01 1.52411476e-01
1.43547915e-02 2.01134488e-01 5.62985599e-01 1.09242134e-01
1.32396233e+00 -5.00025511e-01 4.57949758e-01 -3.42898399e-01
8.54352772e-01 1.06205028e-02 5.54719687e-01 4.79459256e-01
-7.32396066e-01 7.61394858e-01 8.86236668e-01 -3.52618009e-01
-9.34319019e-01 -5.82332432e-01 -2.70958215e-01 5.06378710e-01
-1.55605212e-01 1.61959693e-01 -6.66269064e-01 -7.09208965e-01
-1.65910974e-01 -2.06048433e-02 -1.04917717e+00 1.34142101e-01
-4.39421415e-01 -1.29197550e+00 1.17224681e+00 7.69900262e-01
6.19095564e-01 -1.10618818e+00 -3.00678015e-01 3.42219360e-02
1.51563426e-02 -9.16713476e-01 -3.03356886e-01 2.20577061e-01
-1.02020240e+00 -1.40227735e+00 -1.09279418e+00 -9.72577989e-01
8.86247337e-01 1.24995753e-01 7.33777463e-01 4.31621224e-01
-7.59650350e-01 -1.42812774e-01 -2.53053337e-01 -2.65162647e-01
-4.12081450e-01 2.83380508e-01 -3.15267086e-01 -3.96766692e-01
5.54311335e-01 -2.09418714e-01 -9.21780944e-01 6.14033043e-02
-1.01172733e+00 -2.38238811e-01 9.70333397e-01 1.04187918e+00
9.88134980e-01 -2.15068068e-02 5.22153974e-01 -1.31194329e+00
5.25387824e-01 -5.23767889e-01 -4.03747469e-01 8.14696178e-02
-4.07207310e-01 -5.38603246e-01 6.02069497e-01 -1.72637239e-01
-6.83057547e-01 -7.81732425e-02 -3.27550918e-01 -2.10715160e-01
-1.50024712e-01 6.78378880e-01 4.50200975e-01 -5.13980351e-02
6.54306412e-01 1.28924146e-01 3.05066556e-01 1.56535611e-01
-2.68842369e-01 8.10596108e-01 4.86856073e-01 1.37023017e-01
1.36963725e-01 5.73462307e-01 3.66867036e-01 -5.16770720e-01
-6.89054549e-01 -9.48355138e-01 -3.86546254e-01 -1.59382522e-01
1.02920341e+00 -8.04790437e-01 -1.03134966e+00 8.08012724e-01
-6.79229856e-01 -3.42923373e-01 9.48224366e-02 5.52392960e-01
-1.16575338e-01 1.00268632e-01 -1.24730444e+00 -6.94297135e-01
-1.09214079e+00 -1.25076985e+00 9.76872861e-01 8.28365147e-01
-1.00040242e-01 -1.17988145e+00 2.54872203e-01 3.96173358e-01
6.57601118e-01 5.33234000e-01 1.02924883e+00 -9.19544458e-01
-1.92729503e-01 -5.17257273e-01 -5.66890180e-01 1.23910829e-01
1.36411786e-01 4.40700173e-01 -9.08843696e-01 -4.38958734e-01
-5.93006670e-01 -3.49033207e-01 1.03085518e+00 9.89364743e-01
1.23561227e+00 8.16761330e-03 -9.49163079e-01 7.29234993e-01
1.67820597e+00 4.55331177e-01 8.90969157e-01 2.67218769e-01
5.42258203e-01 8.26621726e-02 5.49376547e-01 2.19018359e-04
6.05140850e-02 -2.93312948e-02 3.31176162e-01 -5.50617039e-01
-4.20673490e-01 2.67505258e-01 -1.72034517e-01 7.20405102e-01
1.82603635e-02 -4.54154909e-01 -1.22571743e+00 7.92029500e-01
-1.69320071e+00 -8.38231027e-01 -3.15946311e-01 1.56990826e+00
6.91157520e-01 1.04777858e-01 -1.09325517e-02 3.01375568e-01
8.01110923e-01 -3.13546717e-01 -5.82836568e-01 -2.03741685e-01
-6.18275926e-02 2.39492148e-01 5.95234394e-01 8.02756622e-02
-1.24249494e+00 6.01800263e-01 5.92067575e+00 1.13231397e+00
-1.36865556e+00 1.32415459e-01 1.47801375e+00 1.91919357e-01
1.75498515e-01 -5.45590937e-01 -9.41671550e-01 2.70706952e-01
7.60821462e-01 1.56608656e-01 -2.24188060e-01 5.03002524e-01
2.34850481e-01 -3.63530397e-01 -6.70905471e-01 7.78640687e-01
-1.14909895e-01 -1.63952065e+00 -1.13437720e-01 3.43575120e-01
8.96453023e-01 3.27794075e-01 1.88627452e-01 4.96089667e-01
8.02830085e-02 -1.30810404e+00 -3.90749007e-01 6.61836386e-01
1.08476961e+00 -1.06537151e+00 1.96305680e+00 1.87427074e-01
-1.00113678e+00 1.49161741e-01 -4.85869795e-01 4.68542039e-01
-3.97905111e-01 7.31243730e-01 -1.30908668e+00 3.55908066e-01
5.06445169e-01 7.31463373e-01 -5.52650571e-01 1.17192721e+00
5.09160638e-01 7.94251978e-01 -2.48715565e-01 -7.03284085e-01
3.97732258e-01 4.61114570e-04 4.93179038e-02 1.52721226e+00
4.41816568e-01 2.92494476e-01 -2.02531159e-01 6.98632538e-01
-3.41795117e-01 2.07680330e-01 -1.76987201e-01 -3.05298269e-01
2.89742827e-01 1.99893963e+00 -1.65236008e+00 -2.51003146e-01
2.46713925e-02 2.18492195e-01 3.33235823e-02 4.99719940e-02
-7.85299957e-01 -6.64755225e-01 1.75477818e-01 5.53384684e-02
9.73789692e-02 4.40640986e-01 -2.60585606e-01 -4.68066841e-01
-6.94657326e-01 -5.52431226e-01 4.90968108e-01 -4.98168290e-01
-1.23141789e+00 4.66477424e-01 -4.10058677e-01 -1.08493829e+00
2.98395455e-01 -7.09870219e-01 -7.00662732e-01 5.23358762e-01
-1.49887359e+00 -1.08997333e+00 -4.64504361e-01 1.14350967e-01
3.12357932e-01 -1.42383128e-01 8.29685271e-01 2.67011493e-01
-8.56440485e-01 6.99554980e-01 3.62519324e-01 4.84352231e-01
5.05228341e-01 -1.28962398e+00 -2.19219267e-01 3.30062598e-01
-8.04383814e-01 5.68604171e-01 2.11493358e-01 -5.77825189e-01
-1.10595465e+00 -1.45659626e+00 5.49725831e-01 5.63225448e-02
3.45584244e-01 8.60819593e-02 -6.46516383e-01 3.99908185e-01
4.50074002e-02 5.06450832e-01 1.26033700e+00 -3.07034999e-01
2.02964157e-01 -4.69641201e-02 -1.55459738e+00 4.92244512e-01
2.78587550e-01 -1.73584729e-01 4.63850409e-01 4.67819780e-01
4.70240533e-01 -6.92234635e-01 -1.12507403e+00 7.43476689e-01
6.62816584e-01 -8.85637760e-01 5.77674627e-01 -1.72766790e-01
2.42950529e-01 -5.08294523e-01 -2.76230555e-02 -8.49503636e-01
-4.62820858e-01 1.10732242e-01 2.42202207e-01 9.83379304e-01
5.52749753e-01 -3.48047107e-01 1.10865557e+00 -8.68041217e-02
-2.24949241e-01 -1.53336465e+00 -8.51771832e-01 -1.46767020e-01
1.42158434e-01 1.10434756e-01 2.13476911e-01 8.89870763e-01
-6.17348477e-02 1.96495101e-01 2.51415849e-01 -1.64404973e-01
3.34395260e-01 -4.34872419e-01 4.40903306e-01 -9.65921044e-01
2.24775091e-01 -7.59222269e-01 -8.77907157e-01 -4.16977525e-01
-1.74368486e-01 -1.02943563e+00 -1.83606505e-01 -1.65385163e+00
8.07008684e-01 -3.25554460e-01 -5.92183948e-01 4.90822375e-01
-1.56749859e-01 8.46965969e-01 -2.49719039e-01 1.94228142e-01
-6.29989207e-01 3.54476795e-02 1.47675097e+00 -4.29847598e-01
-1.22988485e-01 -4.58707884e-02 -6.33170664e-01 4.32623506e-01
1.06966650e+00 -4.00621027e-01 -1.58195630e-01 3.56699109e-01
9.19248387e-02 1.57268643e-02 2.25227065e-02 -1.22020686e+00
3.97146851e-01 8.22597146e-02 9.16132808e-01 -1.15867400e+00
-1.48640677e-01 -2.21377626e-01 2.36428287e-02 9.28361237e-01
-2.90986687e-01 -2.52410054e-01 1.95276693e-01 3.51567179e-01
-2.04695582e-01 -1.37523338e-01 1.02447569e+00 -2.57106096e-01
-8.54526758e-02 3.72858077e-01 -6.37259424e-01 -4.14346486e-01
1.25791550e+00 -5.11872053e-01 -7.56726265e-01 7.74496645e-02
-7.47172892e-01 4.49973911e-01 1.92827657e-01 -3.37893218e-01
5.27210414e-01 -1.28626764e+00 -7.25005150e-01 -1.42283008e-01
1.47010237e-01 3.38731349e-01 6.86659992e-01 1.56248331e+00
-1.20262003e+00 7.27922499e-01 -1.07722089e-01 -9.66364145e-01
-1.50122643e+00 7.45761171e-02 6.88548565e-01 -9.36103404e-01
-2.40485653e-01 1.20322478e+00 5.97493500e-02 -7.18011200e-01
2.36245334e-01 -7.45721519e-01 -7.17763901e-01 -6.18656874e-02
3.88308734e-01 5.55613697e-01 2.73822129e-01 -5.32436848e-01
-2.78532207e-01 4.72177982e-01 -6.48183405e-01 5.84926188e-01
1.11550784e+00 2.12908342e-01 -4.06175315e-01 3.77731532e-01
1.34857702e+00 -3.19426715e-01 -6.85931504e-01 1.42835245e-01
-2.63937593e-01 2.24920973e-01 2.11904421e-01 -1.06082880e+00
-1.33446705e+00 7.34408319e-01 1.11110389e+00 9.81075540e-02
1.11784565e+00 -2.25987628e-01 9.24525380e-01 1.61528319e-01
-1.93676442e-01 -8.85197341e-01 -1.15987159e-01 4.57492948e-01
1.26067460e-01 -1.06927192e+00 1.47279218e-01 -2.70204633e-01
-3.28681976e-01 1.32558668e+00 6.38385415e-01 2.05201586e-03
6.07067347e-01 4.62859273e-01 2.66754121e-01 -4.02300477e-01
-8.34217310e-01 -1.66598082e-01 -1.39725894e-01 4.25039887e-01
8.70455384e-01 2.47785971e-01 -4.13576365e-01 5.82931042e-01
3.92360315e-02 1.26211435e-01 6.07890606e-01 8.53740752e-01
-4.84438807e-01 -5.89608312e-01 -8.03372487e-02 8.99937987e-01
-9.23860788e-01 9.80088040e-02 -3.88650775e-01 1.08485198e+00
-1.62947434e-03 4.99983490e-01 4.13944572e-02 -2.73659885e-01
-7.98409954e-02 -3.28052849e-01 1.96599111e-01 -4.64438945e-01
-7.79233634e-01 3.02596897e-01 -2.62984753e-01 -7.48358965e-02
-5.64558208e-01 -4.36511934e-01 -1.68242610e+00 -4.51176018e-01
-6.28046095e-01 1.54000089e-01 5.03578842e-01 7.79012859e-01
5.30130789e-02 9.54158306e-01 3.61708164e-01 -4.49944407e-01
-1.08174875e-01 -1.38874269e+00 -8.53764236e-01 -6.85303286e-02
2.89214879e-01 -2.82064021e-01 -2.93722123e-01 -1.79130733e-01] | [15.09542465209961, -3.036947011947632] |
e0c42457-92bd-4963-913f-65e86169c45a | icitris-causal-representation-learning-for | 2206.06169 | null | https://arxiv.org/abs/2206.06169v2 | https://arxiv.org/pdf/2206.06169v2.pdf | Causal Representation Learning for Instantaneous and Temporal Effects in Interactive Systems | Causal representation learning is the task of identifying the underlying causal variables and their relations from high-dimensional observations, such as images. Recent work has shown that one can reconstruct the causal variables from temporal sequences of observations under the assumption that there are no instantaneous causal relations between them. In practical applications, however, our measurement or frame rate might be slower than many of the causal effects. This effectively creates "instantaneous" effects and invalidates previous identifiability results. To address this issue, we propose iCITRIS, a causal representation learning method that allows for instantaneous effects in intervened temporal sequences when intervention targets can be observed, e.g., as actions of an agent. iCITRIS identifies the potentially multidimensional causal variables from temporal observations, while simultaneously using a differentiable causal discovery method to learn their causal graph. In experiments on three datasets of interactive systems, iCITRIS accurately identifies the causal variables and their causal graph. | ['Efstratios Gavves', 'Taco Cohen', 'Yuki M. Asano', 'Sindy Löwe', 'Sara Magliacane', 'Phillip Lippe'] | 2022-06-13 | null | null | null | null | ['temporal-sequences'] | ['reasoning'] | [ 5.22402406e-01 1.40861114e-02 -6.69244826e-01 -1.41331494e-01
-2.01788396e-01 -6.68710351e-01 7.89352834e-01 -1.27953384e-02
2.00586826e-01 8.79588485e-01 6.10267222e-01 -5.84248066e-01
-7.07378805e-01 -7.27583230e-01 -9.71290529e-01 -6.40918255e-01
-9.24663544e-01 3.80549490e-01 -1.48365080e-01 3.15345675e-01
1.85582057e-01 3.44351441e-01 -1.39244986e+00 2.75134593e-01
3.77505809e-01 3.21946472e-01 9.20478106e-02 1.09659469e+00
3.31362220e-03 1.33214414e+00 -6.02471471e-01 3.13108236e-01
1.29105691e-02 -4.87881809e-01 -6.48163736e-01 -2.03666985e-01
3.73810232e-01 -7.22763956e-01 -7.04993248e-01 5.39030135e-01
-2.10099909e-02 -1.77693814e-01 6.93189740e-01 -1.69115806e+00
-7.16217458e-01 1.01668441e+00 -8.94827068e-01 4.80117142e-01
6.57771230e-01 8.75876993e-02 1.23009217e+00 -3.54556650e-01
5.75325012e-01 1.79742002e+00 2.33627617e-01 2.23820299e-01
-1.70161664e+00 -8.58714461e-01 5.92394054e-01 3.83735448e-01
-9.45072055e-01 -3.33210558e-01 6.58344448e-01 -8.19252968e-01
5.59567809e-01 3.36761773e-01 3.19447905e-01 1.50645387e+00
5.22022843e-01 7.16070056e-01 7.87649095e-01 -2.55596220e-01
1.75799295e-01 -7.02854693e-01 6.83128089e-02 6.30794942e-01
9.01259556e-02 6.25538349e-01 -7.87762403e-01 -6.29186690e-01
1.32126987e+00 2.35823870e-01 -3.33728909e-01 -2.76194364e-01
-1.54518104e+00 8.78177643e-01 1.71891600e-01 9.38922688e-02
-6.19914949e-01 8.01827490e-01 3.27814162e-01 5.22807717e-01
2.51394808e-01 3.61018956e-01 -6.21127665e-01 -4.50000539e-02
-2.38465697e-01 2.18599930e-01 6.30726695e-01 7.82418191e-01
3.27374697e-01 -3.70801799e-02 -1.91133678e-01 1.61707640e-01
1.74211651e-01 5.24451733e-01 -1.47063900e-02 -9.73825574e-01
2.38011241e-01 5.62446415e-01 4.63048637e-01 -1.21079457e+00
-4.01463896e-01 2.68704474e-01 -8.20327520e-01 -2.73736995e-02
5.35119414e-01 -3.47799867e-01 -9.72863674e-01 1.90627885e+00
3.26099753e-01 1.04705715e+00 -2.27261990e-01 1.02841997e+00
3.59897763e-01 7.34898329e-01 3.18564087e-01 -7.24668086e-01
9.39532340e-01 7.76270330e-02 -1.07101619e+00 7.50339404e-02
2.68405557e-01 -5.89892983e-01 1.06521440e+00 1.59504399e-01
-6.88542843e-01 -2.13320032e-01 -6.13395572e-01 5.11145711e-01
1.06314115e-01 -1.28463894e-01 1.18892694e+00 9.80711952e-02
-6.96162879e-01 4.42963928e-01 -9.87758517e-01 -9.31450650e-02
9.51873064e-02 4.07122672e-01 -1.44640222e-01 -8.77192020e-02
-1.43063366e+00 3.33342195e-01 -4.48776409e-02 -1.49347961e-01
-1.60917306e+00 -1.20702243e+00 -4.43633378e-01 6.42237812e-02
6.72984123e-01 -6.34420335e-01 1.25440586e+00 -9.80304658e-01
-1.08069599e+00 -6.38992190e-02 -3.63234431e-01 -2.82356381e-01
3.47005427e-01 -4.62025255e-01 -5.06031096e-01 3.86266559e-02
1.79574788e-01 -4.31128079e-03 1.08811343e+00 -1.27583075e+00
-7.10375071e-01 -3.31125200e-01 3.82945657e-01 2.89710630e-02
-2.00328361e-02 -1.02418385e-01 -3.00086975e-01 -3.25604916e-01
-5.41842170e-02 -9.18402672e-01 -3.41127694e-01 -2.41032392e-01
-5.29560924e-01 -2.56084442e-01 1.20538986e+00 -4.26216930e-01
1.23546875e+00 -2.04565024e+00 3.63492787e-01 8.96477923e-02
5.17982841e-01 -4.90404189e-01 -1.58376843e-01 6.03397071e-01
-6.80204868e-01 4.33494031e-01 1.86897025e-01 2.45673910e-01
-3.92578095e-01 4.09213871e-01 -6.34332716e-01 8.24308634e-01
1.89087003e-01 6.19433880e-01 -1.29571092e+00 -2.68316418e-01
3.92856330e-01 3.60419095e-01 -2.22528711e-01 4.69847292e-01
-5.89019775e-01 7.52025008e-01 -7.96286225e-01 1.19233780e-01
1.56496212e-01 -4.14647073e-01 6.59365296e-01 4.72340807e-02
-2.26807147e-01 2.87908196e-01 -1.28112364e+00 1.16505778e+00
-4.64809477e-01 7.59012163e-01 -2.04890788e-01 -9.66223240e-01
3.70462418e-01 7.31023073e-01 1.09244251e+00 -7.47606277e-01
-1.57547563e-01 -4.12250042e-01 8.17355793e-03 -7.39426017e-01
-6.54225647e-02 -4.90048043e-02 -1.63192794e-01 6.05670214e-01
-3.51501465e-01 6.22419417e-01 4.19938518e-03 5.75179160e-01
1.71104717e+00 -2.00560898e-01 2.63514578e-01 -9.22767743e-02
-9.26934555e-02 1.47018302e-02 6.05006158e-01 8.51846159e-01
1.50930420e-01 1.83938742e-01 1.19170845e+00 -5.47993898e-01
-9.41439450e-01 -1.32648158e+00 9.96304303e-02 9.97650504e-01
1.53335422e-01 -3.64651024e-01 -1.61141485e-01 -7.80165732e-01
2.97809511e-01 7.09216833e-01 -1.02203906e+00 -1.09767787e-01
-6.42647922e-01 -5.84026217e-01 1.62930861e-01 5.00158608e-01
-2.66087800e-01 -7.45621502e-01 -6.70963168e-01 2.57495582e-01
-2.67615229e-01 -7.42099702e-01 -4.69743490e-01 7.12832361e-02
-6.98915541e-01 -1.53514874e+00 -4.53955214e-03 8.17443207e-02
6.72047496e-01 3.14057678e-01 1.17077816e+00 -1.05512977e-01
-4.05295461e-01 6.09520257e-01 -8.32638070e-02 -1.76174730e-01
-3.50450516e-01 -5.26801527e-01 2.56858200e-01 2.95121484e-02
-3.44164558e-02 -6.14480615e-01 -5.08676887e-01 2.63193458e-01
-7.07457006e-01 5.59129044e-02 4.39002603e-01 8.67504179e-01
4.66512203e-01 4.64446843e-01 7.30443597e-01 -9.82950330e-01
6.23824239e-01 -9.56287265e-01 -8.50010633e-01 1.32254109e-01
-5.49719930e-01 1.56792775e-01 6.55733705e-01 -9.13379133e-01
-1.31431782e+00 2.33537927e-01 8.29570770e-01 -8.36020708e-01
-1.34264588e-01 7.47953832e-01 -1.26450941e-01 7.20922768e-01
7.30208039e-01 -3.84582341e-01 -3.59508663e-01 -1.72606990e-01
5.58525383e-01 5.39096892e-02 4.20004755e-01 -6.32518768e-01
5.52148283e-01 4.37355995e-01 1.92241505e-01 -6.21598482e-01
-5.82137644e-01 -4.11816537e-01 -5.05024731e-01 -3.71823668e-01
6.62653923e-01 -8.46585989e-01 -1.27764428e+00 -1.68864071e-01
-1.23577309e+00 -4.45911855e-01 -8.64757448e-02 6.47461414e-01
-5.29818952e-01 -2.55041033e-01 -4.87091571e-01 -9.77251589e-01
4.68490213e-01 -7.07898438e-01 9.99447167e-01 -3.91597971e-02
-7.47790873e-01 -1.11326480e+00 1.37796029e-01 -1.99548736e-01
-2.37893313e-01 4.46417183e-01 1.14369011e+00 7.73959309e-02
-9.32488620e-01 -5.16899228e-02 -1.69957787e-01 -5.41723788e-01
5.36807954e-01 4.56422985e-01 -5.77490330e-01 -1.04937486e-01
-4.58239377e-01 2.07663670e-01 5.11199772e-01 9.83772755e-01
1.19404137e+00 -7.70203829e-01 -7.57864892e-01 4.43235599e-02
1.20322824e+00 6.47858918e-01 4.25929755e-01 -3.25330347e-01
9.55706656e-01 6.15506053e-01 6.66399419e-01 7.00089097e-01
2.05938414e-01 5.49573660e-01 6.07591331e-01 -2.82070398e-01
3.82330529e-02 -5.82415998e-01 3.23719680e-01 6.37593716e-02
2.47207917e-02 -5.11381626e-01 -9.63939250e-01 6.75276041e-01
-2.30553269e+00 -1.22469831e+00 -7.15439022e-01 2.26537228e+00
7.84097791e-01 -1.48034349e-01 1.23295277e-01 -2.41912216e-01
8.34137201e-01 6.44886540e-03 -1.00323594e+00 -9.23565105e-02
1.47970349e-01 -2.49730632e-01 8.65967453e-01 5.24879932e-01
-1.09699774e+00 5.71630001e-01 7.35068464e+00 -1.90184936e-02
-9.93297458e-01 -3.34760845e-02 6.51034176e-01 -3.94164383e-01
-5.58731019e-01 2.33280346e-01 -3.40801626e-01 3.74479741e-01
1.20916414e+00 -5.51065981e-01 4.19037521e-01 3.67748916e-01
1.06584716e+00 -2.26319339e-02 -1.54034221e+00 6.30746007e-01
-5.78038871e-01 -1.46914494e+00 -1.16655836e-02 1.48742273e-01
7.29648948e-01 -2.65900522e-01 2.59164330e-02 -1.68205008e-01
1.25833201e+00 -1.30629814e+00 4.05329436e-01 6.50660217e-01
8.37598264e-01 -7.07362592e-01 1.74394980e-01 1.51523754e-01
-1.04315400e+00 -4.56472218e-01 7.08166230e-03 -5.37393868e-01
3.02723080e-01 8.73584926e-01 -1.08471668e+00 3.28261673e-01
6.55739129e-01 1.01502812e+00 1.13381118e-01 5.47422767e-01
-4.52092826e-01 1.13968706e+00 -1.30791396e-01 2.48639971e-01
-1.84939981e-01 1.57250147e-02 5.53294539e-01 8.22999358e-01
1.40505448e-01 4.71065909e-01 2.02842072e-01 8.66916180e-01
9.39801857e-02 -4.14348334e-01 -9.07744765e-01 -3.91194493e-01
6.53236628e-01 7.04353690e-01 -4.71759140e-01 -2.32723027e-01
-3.71744722e-01 6.28604054e-01 9.87071078e-03 7.40101576e-01
-1.03313303e+00 4.76256192e-01 1.07676685e+00 2.43390441e-01
-1.02815256e-01 -3.96624446e-01 -3.43559414e-01 -9.95366156e-01
-3.25970262e-01 -8.12029541e-01 6.13857388e-01 -5.79121470e-01
-1.21110463e+00 -2.58663744e-01 2.84117281e-01 -8.94761860e-01
-4.18048710e-01 3.88039127e-02 -5.95414519e-01 8.68165910e-01
-9.24149454e-01 -7.91424453e-01 1.02573805e-01 8.43512118e-01
5.51811397e-01 1.87916026e-01 7.51697004e-01 1.42204389e-01
-7.82132924e-01 -1.16645165e-01 -1.76121399e-01 -1.43567473e-01
6.87704682e-01 -1.37274313e+00 2.15208381e-01 7.41014302e-01
-8.06239527e-03 6.95661306e-01 1.04275095e+00 -1.13657546e+00
-1.88084972e+00 -1.00844336e+00 5.63743770e-01 -4.07827944e-01
1.15702116e+00 -3.18379521e-01 -8.30494702e-01 1.04232752e+00
5.71490414e-02 -1.52746022e-01 3.60594153e-01 6.52099431e-01
-4.20939267e-01 1.55028865e-01 -5.82836270e-01 8.38979244e-01
1.26271653e+00 -4.26444799e-01 -3.01799178e-01 4.21901435e-01
1.03223896e+00 -1.07813723e-01 -7.75281906e-01 2.23593414e-01
5.50997853e-01 -5.14795840e-01 1.16929257e+00 -9.27071929e-01
8.22125375e-01 -4.46872741e-01 1.35624096e-01 -1.41992366e+00
-4.75244880e-01 -6.74963355e-01 -4.64433253e-01 1.13576758e+00
2.65677035e-01 -3.44680190e-01 5.97850621e-01 7.28888988e-01
4.62700188e-01 -5.65564446e-02 -8.19876611e-01 -5.86549461e-01
-3.20582747e-01 -3.88328761e-01 5.83641648e-01 1.41137528e+00
6.91527594e-03 7.36985564e-01 -8.19188595e-01 8.01134467e-01
7.92946637e-01 2.94896305e-01 6.44260526e-01 -1.27942359e+00
-4.28730458e-01 -1.39399081e-01 -1.62254915e-01 -7.53665030e-01
3.60167801e-01 -1.47415459e-01 3.54764126e-02 -1.32058430e+00
2.88633406e-01 -7.08250046e-01 -2.12490812e-01 6.15946174e-01
-3.34040314e-01 -7.54435718e-01 -2.14635238e-01 3.05120915e-01
-1.03116810e-01 3.34810346e-01 1.28246617e+00 -2.47332081e-01
-3.66598159e-01 -4.38667685e-02 -4.78808790e-01 5.38718462e-01
6.08625472e-01 -5.41638792e-01 -1.08717442e+00 -2.16254398e-01
2.48097777e-01 9.74202752e-01 7.21562982e-01 -1.46298617e-01
7.41892532e-02 -9.95982587e-01 3.51174355e-01 -3.57484370e-01
-9.09474387e-04 -8.08554828e-01 6.57285273e-01 4.44356024e-01
-7.50738919e-01 1.29134014e-01 1.00094467e-01 1.14229465e+00
4.76386175e-02 4.72016722e-01 2.57480532e-01 8.76186863e-02
-7.95952141e-01 2.95909494e-01 -5.95261037e-01 -4.05955076e-01
1.12227356e+00 4.13601846e-01 -5.29607415e-01 -6.94405437e-01
-7.20288396e-01 3.78273040e-01 8.02882761e-03 7.41418660e-01
7.19734907e-01 -1.20698583e+00 -6.83314562e-01 -1.09523669e-01
-1.66184664e-01 -2.20458940e-01 3.10036570e-01 7.27229774e-01
3.60571951e-01 3.04258049e-01 1.73347324e-01 -6.24053478e-01
-1.45765126e+00 9.43242252e-01 1.89526692e-01 -1.60706550e-01
-7.12899446e-01 4.64807600e-01 9.05172527e-01 7.02531189e-02
4.30251323e-02 -6.15190454e-02 -2.56551027e-01 6.83115944e-02
6.71642959e-01 3.94080490e-01 -6.08891368e-01 -4.79507223e-02
-2.12669522e-01 -1.47093646e-02 -1.09954609e-03 -1.84859484e-01
1.36899507e+00 -2.53575832e-01 -4.31705788e-02 8.64682555e-01
1.06094313e+00 -2.46929958e-01 -1.69213104e+00 -2.08474696e-02
1.28313944e-01 -7.69490421e-01 2.00236961e-01 -8.36720645e-01
-1.08939528e+00 4.83406782e-01 5.58279812e-01 5.07990956e-01
1.12183273e+00 2.62967348e-01 1.73055306e-01 -2.19643012e-01
2.70036489e-01 -4.64987487e-01 2.06379563e-01 2.04766825e-01
1.05060518e+00 -1.04841459e+00 -2.70823594e-02 -6.49714589e-01
-2.47816369e-01 8.96101236e-01 3.92114013e-01 -3.31942551e-02
5.42153299e-01 6.53109193e-01 -2.18977258e-01 -5.17965317e-01
-1.35400629e+00 1.22286573e-01 1.81917138e-02 5.33454180e-01
6.31259382e-01 7.43790448e-01 -7.98520073e-02 -1.62206665e-02
2.26133689e-01 -7.00981449e-03 7.38536239e-01 6.72994077e-01
2.70664245e-01 -9.39071059e-01 -6.99625671e-01 5.56896329e-01
-1.71321481e-01 -2.43472066e-02 -4.30391908e-01 7.63672233e-01
-2.53920466e-01 1.19467652e+00 3.45528245e-01 -3.48621190e-01
5.17862260e-01 -3.50931674e-01 2.43866012e-01 -4.04252529e-01
1.05641186e-01 7.74392933e-02 1.88112065e-01 -1.02906728e+00
-3.67497355e-01 -1.18472826e+00 -1.45106232e+00 -5.90526223e-01
-5.93256764e-02 -2.23142028e-01 3.27625364e-01 8.89675617e-01
3.89056772e-01 1.10783911e+00 8.42246532e-01 -2.89288908e-01
-5.54494075e-02 -5.55579007e-01 -3.45922381e-01 4.90378201e-01
7.80318439e-01 -1.00554562e+00 -3.58935326e-01 6.04068696e-01] | [7.823699474334717, 5.247799873352051] |
a42e0d9a-8758-427a-9d89-ce5a5206c0ae | can-chatgpt-and-bard-generate-aligned | 2304.05372 | null | https://arxiv.org/abs/2304.05372v1 | https://arxiv.org/pdf/2304.05372v1.pdf | Can ChatGPT and Bard Generate Aligned Assessment Items? A Reliability Analysis against Human Performance | ChatGPT and Bard are AI chatbots based on Large Language Models (LLM) that are slated to promise different applications in diverse areas. In education, these AI technologies have been tested for applications in assessment and teaching. In assessment, AI has long been used in automated essay scoring and automated item generation. One psychometric property that these tools must have to assist or replace humans in assessment is high reliability in terms of agreement between AI scores and human raters. In this paper, we measure the reliability of OpenAI ChatGP and Google Bard LLMs tools against experienced and trained humans in perceiving and rating the complexity of writing prompts. Intraclass correlation (ICC) as a performance metric showed that the inter-reliability of both the OpenAI ChatGPT and the Google Bard were low against the gold standard of human ratings. | ['Abdolvahab Khademi'] | 2023-04-09 | null | null | null | null | ['automated-essay-scoring'] | ['natural-language-processing'] | [-5.17058611e-01 2.24683881e-01 2.17840467e-02 -1.85670093e-01
-7.93958783e-01 -7.33791947e-01 4.08665925e-01 2.71401346e-01
-4.59023327e-01 6.45623863e-01 -4.07815538e-02 -4.72234309e-01
-5.63555419e-01 -5.46293557e-01 4.82815914e-02 -1.27838835e-01
4.37088102e-01 8.35902393e-01 3.30497891e-01 -2.76549906e-01
7.34200478e-01 1.20117627e-01 -1.33699918e+00 4.08718258e-01
1.35942531e+00 6.20973647e-01 2.26620361e-01 9.78244841e-01
-6.05396092e-01 1.48486376e+00 -1.42469573e+00 -9.23411727e-01
-1.49537489e-01 -5.52264750e-01 -9.58374739e-01 -4.61498648e-01
6.19123280e-01 -3.53900105e-01 3.22846740e-01 7.58684695e-01
5.54505169e-01 9.50921923e-02 5.80253541e-01 -1.16324723e+00
-1.10878122e+00 7.34884918e-01 -1.44980401e-01 8.86534899e-02
1.10469067e+00 1.55052319e-01 1.03663146e+00 -7.11769879e-01
4.14933085e-01 1.14433408e+00 8.08882713e-01 5.04737318e-01
-8.90942872e-01 -6.84919834e-01 -5.91466367e-01 4.72343802e-01
-9.45202231e-01 1.37185991e-01 3.07194889e-01 -9.24731851e-01
9.44339454e-01 1.55071512e-01 1.09370434e+00 7.79709399e-01
5.13895631e-01 5.44269383e-01 1.51023662e+00 -6.76313639e-01
4.70265299e-01 7.43454218e-01 2.03048646e-01 6.40399754e-01
2.15532839e-01 -5.14530599e-01 -9.41508710e-01 -1.91415340e-01
8.52322340e-01 -3.46917957e-01 3.04123461e-01 2.23934814e-01
-9.17921841e-01 9.97053623e-01 1.92741960e-01 5.64757705e-01
-3.14086318e-01 -1.21498749e-01 2.58355439e-01 8.65031958e-01
5.33134520e-01 9.99019265e-01 -1.39507860e-01 -1.16127455e+00
-4.74840283e-01 1.53783187e-01 1.13193917e+00 7.85820067e-01
8.86023492e-02 3.59390154e-02 -2.69634932e-01 1.27712202e+00
3.34554404e-01 4.45345402e-01 1.06305265e+00 -1.00751066e+00
3.57674778e-01 1.23037028e+00 -4.40733768e-02 -7.25799799e-01
-1.62997395e-01 -2.98876520e-02 -1.79138914e-01 4.19801444e-01
6.43703640e-01 -1.21612139e-01 -3.87431383e-01 1.04195642e+00
9.38244089e-02 -3.68420959e-01 -1.64257050e-01 8.73041332e-01
9.95629609e-01 5.46439886e-01 3.96732390e-01 1.18816100e-01
1.32155490e+00 -9.07496035e-01 -8.47384036e-01 -2.21737280e-01
1.03015363e+00 -9.33014989e-01 1.77564609e+00 7.63940692e-01
-1.45329559e+00 -6.44636035e-01 -8.66793573e-01 -1.03648029e-01
-2.24887118e-01 -1.32541120e-01 5.09853721e-01 1.08827293e+00
-1.23492968e+00 3.94875556e-01 -2.02756211e-01 -3.42616916e-01
1.71190113e-01 2.77114421e-01 -3.91424686e-01 3.39707620e-02
-1.04063368e+00 1.58411849e+00 -1.98803633e-01 -3.45418692e-01
-5.41805089e-01 -6.34805858e-01 -3.73606950e-01 2.66549528e-01
-2.17697412e-01 -1.85711861e-01 1.85250890e+00 -7.15497434e-01
-2.31192183e+00 9.83047307e-01 5.66496551e-01 -3.03444088e-01
4.82776672e-01 -4.42312956e-01 1.20787032e-01 4.57887761e-02
2.04181820e-01 1.61411643e-01 2.19178885e-01 -2.15247661e-01
-4.49078083e-01 -2.81124741e-01 -5.65473747e-04 4.47327465e-01
-7.12376654e-01 4.24989641e-01 4.05442685e-01 -2.19649628e-01
-1.18441112e-01 -6.64932847e-01 1.22856960e-01 -1.70373097e-01
3.46644163e-01 -1.00657022e+00 5.05205512e-01 -6.33198440e-01
1.29042447e+00 -1.67030180e+00 -2.39409342e-01 2.36794606e-01
3.30639422e-01 7.16333568e-01 -9.63338763e-02 6.63008213e-01
5.92054307e-01 2.24204749e-01 4.61354792e-01 1.32225722e-01
2.43933529e-01 -3.82528566e-02 2.55730003e-01 8.38889703e-02
-2.20553383e-01 8.96790504e-01 -1.07271361e+00 -8.39978814e-01
2.77741373e-01 5.27313538e-02 -2.80349195e-01 7.73977280e-01
1.60764921e-02 1.24817163e-01 -2.80606687e-01 2.86314279e-01
-8.28085318e-02 -2.04946622e-01 -1.31486490e-01 1.05735457e+00
-1.15595073e-01 8.63951743e-01 -7.25542903e-01 1.47994041e+00
-6.12618744e-01 7.96293139e-01 -1.55330166e-01 -3.39367539e-01
1.54700065e+00 8.47403288e-01 1.46785796e-01 -8.50858808e-01
2.74854843e-02 5.10800481e-01 6.13729239e-01 -6.54428303e-01
2.96707094e-01 4.26510572e-02 7.31119066e-02 1.22098720e+00
5.17755508e-01 -9.02065933e-01 1.80141300e-01 3.19479167e-01
1.22197068e+00 -8.51050913e-02 6.49230599e-01 -2.52019048e-01
4.32513893e-01 2.87872497e-02 -8.95929411e-02 8.85412514e-01
-2.30199829e-01 2.06983481e-02 5.19196808e-01 -4.87456650e-01
-9.04297054e-01 -8.52701783e-01 1.41515195e-01 1.40273774e+00
-6.13017380e-01 -3.32742184e-01 -1.09018934e+00 -5.68569720e-01
-3.42668682e-01 8.72002900e-01 -1.24929242e-01 -1.01153322e-01
-3.58277746e-02 1.01064548e-01 6.93085313e-01 2.56961852e-01
1.35779470e-01 -1.50284815e+00 -7.53795922e-01 5.28282821e-01
6.01337589e-02 -9.10632014e-01 -1.75636962e-01 -1.96846008e-01
-6.84135199e-01 -7.02148020e-01 -6.22658670e-01 -8.62646043e-01
1.57318130e-01 1.62248552e-01 1.30497074e+00 4.87148255e-01
-1.74865842e-01 5.97647190e-01 -6.04997754e-01 -9.78139818e-01
-8.15310955e-01 3.38709980e-01 5.13749421e-02 -8.53931427e-01
8.23139846e-01 -5.95626950e-01 -1.05392203e-01 5.12262881e-01
-3.60611528e-01 1.54040039e-01 7.91564167e-01 5.13593972e-01
-3.94396037e-01 -8.17511439e-01 9.01457071e-01 -7.03142047e-01
1.56793201e+00 -2.70402908e-01 -4.62802768e-01 3.02129835e-01
-8.38202477e-01 -4.53621626e-01 3.46080273e-01 -6.18517637e-01
-8.39092970e-01 -7.31470466e-01 -2.04050809e-01 6.33481797e-03
-2.63267428e-01 6.28341019e-01 5.63907444e-01 -4.98193204e-01
9.11344647e-01 -2.29179025e-01 1.71739042e-01 -2.60897040e-01
-8.31723809e-02 1.21933472e+00 1.09706879e-01 -8.13058197e-01
3.42113972e-01 -6.87907338e-01 -3.65583926e-01 -7.52553701e-01
-8.91941309e-01 -4.18528378e-01 -5.56890845e-01 -8.22045386e-01
6.97481573e-01 -6.79002404e-01 -1.30361795e+00 3.33895981e-01
-1.20964992e+00 -6.43348634e-01 -3.14735413e-01 7.52510488e-01
-5.54733038e-01 -2.05761325e-02 -1.11402655e+00 -1.10078597e+00
-7.26156712e-01 -9.74495947e-01 5.11537611e-01 7.30403781e-01
-9.08588409e-01 -1.06316268e+00 4.03200448e-01 8.99689794e-01
5.35893857e-01 -4.96614397e-01 9.08455491e-01 -1.23142755e+00
4.77057062e-02 -7.55349219e-01 1.41716287e-01 3.43755424e-01
-1.45825043e-01 3.75127681e-02 -8.16207647e-01 3.02687466e-01
8.07383657e-02 -8.74325812e-01 -2.66794533e-01 9.87536460e-03
6.88728094e-01 -5.27050078e-01 5.60111523e-01 -2.66948909e-01
8.95295978e-01 4.81792301e-01 4.56468999e-01 5.74739397e-01
4.32332665e-01 7.00694621e-01 6.06715918e-01 1.28601313e-01
1.35961547e-01 5.14459372e-01 -2.55559891e-01 4.68256325e-01
2.17381388e-01 -3.42761129e-01 7.24491060e-01 1.58802295e+00
-3.65439057e-01 -1.12564914e-01 -1.40879500e+00 3.61272037e-01
-1.67764294e+00 -8.49589765e-01 -3.43202472e-01 2.07144737e+00
9.70402181e-01 2.32882723e-01 2.68843740e-01 1.62013993e-01
2.73320198e-01 -3.25248390e-01 -7.62323011e-03 -1.48348749e+00
4.47757691e-01 6.14654481e-01 -1.73904866e-01 5.10749042e-01
-2.33925954e-02 7.91275442e-01 6.72835827e+00 7.65069008e-01
-5.96611977e-01 3.19444358e-01 4.21877831e-01 1.08534157e-01
3.81190516e-02 -3.52346241e-01 -4.59563076e-01 4.30666238e-01
1.48742747e+00 -4.79663640e-01 2.95609802e-01 1.15523589e+00
9.08200666e-02 -2.11517528e-01 -9.55593228e-01 5.14739335e-01
2.20219176e-02 -1.04288030e+00 -1.66362435e-01 1.54190548e-02
8.98903012e-01 -3.20116132e-01 -1.71099976e-01 4.58772421e-01
7.59282470e-01 -1.09615183e+00 5.60100377e-01 4.13380146e-01
5.78296661e-01 -3.25268775e-01 1.22822762e+00 6.21257842e-01
-4.89069760e-01 -6.60215393e-02 -4.91861075e-01 -9.10343766e-01
-3.26207727e-01 -5.55895120e-02 -1.55056334e+00 -2.10557312e-01
5.08192182e-01 1.45453632e-01 -6.26983464e-01 1.16536486e+00
-7.71750748e-01 1.04547799e+00 -5.19750901e-02 -9.74223673e-01
2.92703629e-01 -5.82013190e-01 1.44755706e-01 9.43559408e-01
4.02472258e-01 1.37929216e-01 8.77474546e-02 7.09070563e-01
1.44142896e-01 4.81634110e-01 -4.92718339e-01 -3.19335163e-01
8.76683474e-01 1.28375971e+00 -4.31472152e-01 -3.06834519e-01
-4.27522659e-01 5.07108748e-01 4.74180192e-01 -2.32064217e-01
-4.22778547e-01 -3.76206934e-01 5.08968711e-01 2.27056742e-01
-6.12520993e-01 -2.15810671e-01 -8.82149935e-01 -4.44039583e-01
-2.25654647e-01 -1.08121800e+00 2.09386796e-01 -1.24386311e+00
-1.40259659e+00 4.08608943e-01 -3.51745516e-01 -1.21206820e+00
-3.94288838e-01 -6.32618368e-01 -1.01041174e+00 8.64015460e-01
-4.34312999e-01 -8.35424364e-01 -5.27560651e-01 3.40829492e-01
7.20745206e-01 -3.60015243e-01 1.09221482e+00 3.58937010e-02
-1.24226667e-01 5.92664719e-01 -1.97718531e-01 1.83141217e-01
8.28261197e-01 -1.42207909e+00 1.18603677e-01 1.55505300e-01
3.16399753e-01 4.43324894e-01 4.91887271e-01 -5.32505989e-01
-1.10803378e+00 -1.00456968e-01 9.86586034e-01 -7.25489020e-01
8.68448436e-01 -1.82003289e-01 -1.00059092e+00 5.27339935e-01
4.11089689e-01 -8.45111668e-01 1.08031964e+00 2.65118033e-01
-2.28375763e-01 7.63093904e-02 -1.06228948e+00 3.26941520e-01
4.94367927e-01 -6.79702580e-01 -9.26845729e-01 5.06693900e-01
4.06655282e-01 -3.67953300e-01 -1.29233348e+00 -3.84757638e-01
8.94731104e-01 -1.01604855e+00 5.71682811e-01 -6.74826622e-01
8.04389000e-01 4.71008599e-01 6.07386589e-01 -1.39525521e+00
-2.14560047e-01 -5.70217133e-01 9.58586484e-02 1.27647567e+00
2.66210705e-01 -5.18119395e-01 7.83601642e-01 1.08132625e+00
-3.13734561e-01 -7.04294920e-01 -1.00477982e+00 -6.29288912e-01
2.59237289e-01 -2.60040641e-01 2.76312470e-01 9.39998925e-01
1.00043130e+00 6.17336035e-01 -1.25569582e-01 -6.01050436e-01
1.86177954e-01 -3.45462590e-01 9.64757740e-01 -1.77694511e+00
1.88749284e-01 -7.17106521e-01 -7.24914789e-01 -3.51962090e-01
-2.51206577e-01 -5.29061437e-01 4.50215973e-02 -1.78977132e+00
5.12044691e-03 -2.68463284e-01 7.70994276e-02 3.49723667e-01
-3.29796225e-01 1.15806207e-01 3.18342537e-01 1.58274233e-01
-7.63329923e-01 1.73431709e-01 1.17350411e+00 2.46270984e-01
-3.47711056e-01 8.64888206e-02 -4.85705256e-01 8.70244205e-01
9.44549203e-01 -7.26892114e-01 -2.35015541e-01 -4.22583699e-01
5.98976314e-01 2.00482056e-01 -1.56878624e-02 -1.34114540e+00
4.13773298e-01 -3.67436916e-01 1.20050959e-01 2.86004603e-01
1.43630192e-01 -6.98141575e-01 -3.17351103e-01 3.06376219e-01
-6.69994891e-01 4.67101574e-01 -6.81740195e-02 -1.62390500e-01
-2.22778633e-01 -7.64229119e-01 5.04342258e-01 -3.11708272e-01
-2.81056166e-01 -2.59860337e-01 -9.66202676e-01 2.61042323e-02
8.39890122e-01 -4.23398137e-01 -5.31482697e-01 -1.01385677e+00
-6.46375120e-02 9.28110331e-02 2.22482786e-01 4.39300179e-01
4.96046752e-01 -1.02646184e+00 -6.22347951e-01 -9.33715180e-02
-7.06506297e-02 -2.56018877e-01 -9.26267058e-02 8.77094924e-01
-8.96673322e-01 6.32410645e-01 -6.96793795e-01 -3.79914820e-01
-1.47965348e+00 -2.06497088e-01 5.79248145e-02 -4.25645947e-01
-1.35422170e-01 9.98358905e-01 -4.25733149e-01 -4.25933927e-01
1.92061931e-01 -1.76398918e-01 -3.91295910e-01 -1.20190464e-01
7.83146322e-01 7.20054030e-01 1.69837505e-01 -1.35530069e-01
2.47429594e-01 2.57887065e-01 -1.85556908e-03 -4.08618033e-01
1.32736015e+00 2.89290160e-01 -1.64187402e-01 1.08586657e+00
3.52703840e-01 1.94563985e-01 -3.68246794e-01 5.84547296e-02
4.82094824e-01 -5.07271111e-01 -4.97018307e-01 -1.16769123e+00
-2.80000437e-02 1.19769502e+00 4.50195581e-01 6.54981077e-01
3.37489009e-01 -1.62377268e-01 3.58498245e-01 6.55052781e-01
4.51572657e-01 -1.38116407e+00 4.81827050e-01 7.20418572e-01
8.33322346e-01 -1.01867843e+00 -1.82693265e-02 1.94274448e-02
-1.07165074e+00 1.28549612e+00 1.26797569e+00 -1.29013300e-01
1.30952165e-01 4.75325957e-02 4.52150017e-01 -2.94137239e-01
-9.34536278e-01 2.74191976e-01 5.56160927e-01 4.37655777e-01
1.27533555e+00 2.21314117e-01 -8.48683774e-01 8.62179816e-01
-8.65553498e-01 2.30550289e-01 8.54725718e-01 8.08292806e-01
-9.04477596e-01 -8.62654507e-01 -6.68608308e-01 6.45140648e-01
-3.57241511e-01 9.49803740e-02 -1.00957859e+00 7.03665912e-01
-3.02283257e-01 1.15893161e+00 -4.42435741e-02 -5.28987110e-01
2.55024105e-01 6.73739672e-01 2.99851596e-01 -9.85762477e-01
-1.40512347e+00 -6.47698283e-01 1.79779872e-01 -2.16163233e-01
-1.19048595e-01 -2.40877628e-01 -7.39152968e-01 -5.74123025e-01
-6.47121966e-01 8.92302573e-01 9.28670824e-01 1.10455537e+00
8.02132189e-02 1.22528598e-01 -2.69535892e-02 -3.78840625e-01
-7.75842488e-01 -1.82170153e+00 -5.39135456e-01 2.07612082e-01
-2.96521723e-01 -4.07475919e-01 -1.31719232e-01 -4.78343576e-01] | [11.288352012634277, 9.270907402038574] |
21f94e70-4eae-4e3f-88cc-5ffeeb3eb6a3 | controllable-deep-melody-generation-via | 2109.00663 | null | https://arxiv.org/abs/2109.00663v1 | https://arxiv.org/pdf/2109.00663v1.pdf | Controllable deep melody generation via hierarchical music structure representation | Recent advances in deep learning have expanded possibilities to generate music, but generating a customizable full piece of music with consistent long-term structure remains a challenge. This paper introduces MusicFrameworks, a hierarchical music structure representation and a multi-step generative process to create a full-length melody guided by long-term repetitive structure, chord, melodic contour, and rhythm constraints. We first organize the full melody with section and phrase-level structure. To generate melody in each phrase, we generate rhythm and basic melody using two separate transformer-based networks, and then generate the melody conditioned on the basic melody, rhythm and chords in an auto-regressive manner. By factoring music generation into sub-problems, our approach allows simpler models and requires less data. To customize or add variety, one can alter chords, basic melody, and rhythm structure in the music frameworks, letting our networks generate the melody accordingly. Additionally, we introduce new features to encode musical positional information, rhythm patterns, and melodic contours based on musical domain knowledge. A listening test reveals that melodies generated by our method are rated as good as or better than human-composed music in the POP909 dataset about half the time. | ['Roger B. Dannenberg', 'Celso Gomes', 'Zeyu Jin', 'Shuqi Dai'] | 2021-09-02 | null | null | null | null | ['music-generation', 'music-generation'] | ['audio', 'music'] | [ 1.99235559e-01 -8.56954083e-02 1.80344909e-01 6.82090446e-02
-7.16019869e-01 -1.14370298e+00 3.91551554e-01 -2.76165962e-01
-6.27669543e-02 6.43921554e-01 5.25581777e-01 -4.02466916e-02
-4.67869669e-01 -8.99858117e-01 -4.53737587e-01 -5.69234729e-01
8.04034695e-02 3.84849191e-01 -1.61480457e-02 -7.00466633e-01
3.17600340e-01 3.06042522e-01 -1.53921902e+00 7.17149019e-01
6.13569438e-01 8.09056818e-01 1.70055002e-01 9.52149153e-01
3.94831499e-04 6.45134747e-01 -1.01478827e+00 -3.10236067e-01
5.94824851e-01 -1.03540611e+00 -4.85660464e-01 -1.18394069e-01
1.69407725e-01 4.72592562e-02 9.05642062e-02 6.04973972e-01
7.27817416e-01 2.67684549e-01 6.08028352e-01 -7.95073390e-01
-5.81316829e-01 1.49477673e+00 -2.65687287e-01 -3.50305617e-01
3.83216560e-01 1.80934489e-01 1.35219729e+00 -7.19596267e-01
4.55775052e-01 9.18712318e-01 1.01503015e+00 5.57203889e-01
-1.55644155e+00 -8.83649230e-01 -2.67907470e-01 -1.12652950e-01
-1.40409219e+00 -1.15350768e-01 1.14946866e+00 -4.14803714e-01
7.19740748e-01 4.19920802e-01 1.20854557e+00 1.03615272e+00
2.20932901e-01 4.69567895e-01 6.25539362e-01 -5.10303557e-01
1.49968192e-01 -4.36377913e-01 -4.41003472e-01 1.86751693e-01
-3.82153928e-01 2.76983440e-01 -5.40728331e-01 -2.59998720e-02
1.10544205e+00 -4.41167504e-01 -1.36231985e-02 -1.34512380e-01
-1.42012882e+00 7.27739930e-01 3.86780351e-01 5.07423878e-01
-4.51409340e-01 5.03704250e-02 3.64447713e-01 3.22582781e-01
-2.05316424e-01 1.22607017e+00 -2.75246263e-01 -1.91041440e-01
-1.41282952e+00 9.09939528e-01 8.46417665e-01 5.38883626e-01
4.93160933e-01 4.59239870e-01 -5.50764680e-01 1.14051604e+00
-2.19288096e-01 3.43416095e-01 8.20875287e-01 -1.20946515e+00
1.30879745e-01 2.69997418e-01 -8.81531313e-02 -8.48223329e-01
-3.43904793e-01 -7.90503860e-01 -1.05981863e+00 4.39323515e-01
4.02340680e-01 -1.45719737e-01 -8.26308787e-01 1.94024873e+00
-2.49052078e-01 1.12151742e-01 -8.82172659e-02 7.52871633e-01
6.10551596e-01 8.35286915e-01 -2.58284450e-01 -2.24683046e-01
1.17809725e+00 -8.00451398e-01 -5.54351628e-01 1.86918452e-01
-1.24820307e-01 -9.94132459e-01 1.39708734e+00 7.49487340e-01
-1.58347344e+00 -1.01968825e+00 -1.26573157e+00 -4.63420972e-02
6.73241839e-02 1.37136370e-01 4.18789357e-01 4.73070681e-01
-8.75037313e-01 9.19492245e-01 -3.58063281e-01 3.81865323e-01
-5.85755482e-02 2.61584014e-01 1.91937666e-02 7.68410444e-01
-1.27996862e+00 4.58830148e-01 7.84729242e-01 -9.63901803e-02
-7.78261006e-01 -9.11976218e-01 -7.89631426e-01 3.06382596e-01
1.08702509e-02 -9.67071116e-01 1.50170994e+00 -8.77342880e-01
-1.92643631e+00 5.85456371e-01 3.78655642e-01 -5.61235428e-01
2.89356977e-01 1.21362835e-01 -3.64505321e-01 -2.70514905e-01
7.42011610e-03 8.07374120e-01 7.95411348e-01 -1.02852499e+00
-5.15426815e-01 1.25451237e-01 -3.61816101e-02 3.76970232e-01
-2.29802594e-01 -2.85682887e-01 -4.96575117e-01 -1.41994524e+00
1.12084016e-01 -1.07900715e+00 -2.31439248e-01 -6.40470803e-01
-6.15867555e-01 -5.92430979e-02 1.10248446e-01 -4.42786932e-01
1.81871736e+00 -2.24720597e+00 5.31175494e-01 3.74751151e-01
-1.12360291e-01 -8.66790861e-03 -4.19768065e-01 3.89166683e-01
-3.03433895e-01 2.77476516e-02 -2.84560412e-01 -9.01696384e-02
2.79774308e-01 -7.03013465e-02 -6.68242991e-01 -4.67061132e-01
-1.15441764e-03 9.72507298e-01 -7.30019271e-01 -9.28118080e-02
-1.95389846e-03 5.04247725e-01 -9.88490641e-01 9.18469355e-02
-5.11712432e-01 4.77233469e-01 8.82376581e-02 3.21764737e-01
1.58225209e-01 1.00243889e-01 2.27113411e-01 -4.29156035e-01
-2.73452282e-01 6.34435296e-01 -1.31820703e+00 1.95810711e+00
-5.56761146e-01 3.62985432e-01 -3.79492998e-01 -4.62381929e-01
1.37261331e+00 4.74002987e-01 4.50754941e-01 -6.13851249e-01
8.99808202e-03 3.13445449e-01 3.55234712e-01 6.90970197e-03
7.81478047e-01 -6.11372054e-01 -5.81098676e-01 7.17543483e-01
-3.45240645e-02 -6.12795591e-01 4.52419460e-01 -3.16815555e-01
8.51853669e-01 4.75985140e-01 4.39796537e-01 1.06384829e-01
2.41291910e-01 -5.40233552e-02 3.73755395e-01 6.82808459e-01
5.61352730e-01 9.91056442e-01 3.62294316e-01 -6.06582522e-01
-1.35665393e+00 -1.37161326e+00 2.12264329e-01 1.40563464e+00
-4.17747974e-01 -7.60051310e-01 -6.66638494e-01 6.27983660e-02
-2.64184296e-01 6.91465735e-01 -5.40477514e-01 -9.21115130e-02
-1.04896557e+00 -6.11286938e-01 9.21425045e-01 4.81097430e-01
9.59744528e-02 -1.82802904e+00 -4.87016886e-01 6.33065879e-01
-3.93539548e-01 -3.63031596e-01 -8.00391436e-01 3.48655403e-01
-5.76878667e-01 -6.98166192e-01 -7.86933661e-01 -1.04833579e+00
-8.24268013e-02 -4.14848477e-01 1.32317197e+00 -2.43925408e-01
-3.20286930e-01 -4.33293879e-01 -1.85821235e-01 -3.35462391e-01
-6.12279713e-01 3.99082541e-01 -2.02645753e-02 -2.90785879e-01
-3.59990597e-01 -1.09063363e+00 -5.26979685e-01 2.56860703e-01
-9.89422083e-01 4.05490190e-01 5.10544658e-01 8.12058568e-01
9.38904822e-01 2.55439639e-01 7.75767267e-01 -5.52721083e-01
1.04853570e+00 -7.54346922e-02 -2.72633463e-01 -8.11652839e-02
-3.01011294e-01 3.98711041e-02 8.02735627e-01 -7.88038015e-01
-7.98154593e-01 2.29956746e-01 -3.52908343e-01 -4.08990562e-01
1.06396094e-01 6.46591961e-01 -2.78488826e-02 5.43054521e-01
1.19270945e+00 3.89037997e-01 -2.96778321e-01 -4.60439861e-01
7.08837330e-01 2.69550532e-01 1.14645886e+00 -8.39030445e-01
9.50658262e-01 -3.51422327e-03 -1.66914433e-01 -3.16048622e-01
-9.20942664e-01 7.32344836e-02 -7.60593176e-01 -7.02485964e-02
7.24072099e-01 -6.95709825e-01 -6.56489909e-01 3.12398463e-01
-1.04382181e+00 -5.02670169e-01 -9.47865307e-01 2.25515455e-01
-9.08151805e-01 -3.85138462e-03 -7.39075601e-01 -4.91330236e-01
-6.23990297e-01 -8.65370274e-01 8.66769612e-01 -3.16324015e-03
-8.84663761e-01 -4.46121842e-01 4.88711268e-01 -4.42431681e-02
3.87964040e-01 5.32722533e-01 9.72875655e-01 -2.24271312e-01
-4.68129188e-01 5.05041778e-02 3.32079560e-01 2.31876120e-01
4.18897659e-01 1.72984879e-02 -6.39557958e-01 -8.23660567e-02
-2.06104726e-01 -3.00011307e-01 7.35384881e-01 3.36291045e-01
1.09242499e+00 -4.54635412e-01 4.01377082e-01 1.02226079e+00
1.00515938e+00 5.54661453e-01 6.78244293e-01 3.16452146e-01
5.84950864e-01 4.50801402e-01 2.46477023e-01 6.21503055e-01
9.18591395e-02 9.27494586e-01 -2.05302313e-02 1.58795729e-01
-4.14950490e-01 -7.31556475e-01 4.41700786e-01 1.11253130e+00
-4.00511533e-01 -1.36723399e-01 -6.65794492e-01 5.36646187e-01
-1.59715927e+00 -1.58731818e+00 2.23859116e-01 1.92411911e+00
1.26194823e+00 2.55844355e-01 6.31616056e-01 7.04117179e-01
6.34028196e-01 6.10414408e-02 -5.03284812e-01 -3.83984804e-01
-3.37516576e-01 9.09309864e-01 -1.61087960e-01 2.64808327e-01
-7.43743062e-01 1.09675956e+00 7.09190655e+00 9.59497571e-01
-1.34112394e+00 -4.74717975e-01 3.22128832e-01 -4.33174044e-01
-7.06653476e-01 -1.95575997e-01 -5.39548993e-01 3.70277524e-01
6.21739924e-01 -3.59985530e-01 1.07996500e+00 5.63504577e-01
2.22105935e-01 7.64077187e-01 -1.07451105e+00 1.02097857e+00
7.52641559e-02 -1.77458298e+00 4.87425029e-01 -2.08963260e-01
8.75860393e-01 -4.20881242e-01 3.49013895e-01 6.34986997e-01
3.32161069e-01 -1.40526497e+00 1.17125094e+00 5.23458779e-01
1.07045805e+00 -9.40916538e-01 1.54136181e-01 2.17855856e-01
-1.53478479e+00 -2.07588494e-01 -1.91079993e-02 -3.08755338e-01
2.52403319e-01 1.17248610e-01 -8.23525608e-01 4.60789949e-01
5.43904305e-01 4.60408479e-01 -3.86544138e-01 1.06128359e+00
-2.99180686e-01 7.13009357e-01 -2.24969730e-01 2.49380782e-01
5.49933463e-02 -1.88273057e-01 4.84998822e-01 1.16795421e+00
7.86090612e-01 2.14191563e-02 1.88502848e-01 1.16464436e+00
4.13051359e-02 2.04869017e-01 -2.12364227e-01 -1.03922121e-01
6.45310819e-01 1.21128309e+00 -6.39128804e-01 -1.40396893e-01
3.13011348e-01 9.06102002e-01 2.77169123e-02 2.09090367e-01
-7.51669109e-01 -4.02801007e-01 2.27680832e-01 2.53755599e-01
3.13133091e-01 -1.43793762e-01 -5.57980061e-01 -8.22942019e-01
-1.44315928e-01 -1.29314053e+00 3.91071469e-01 -1.10410094e+00
-1.27966928e+00 1.02457225e+00 -2.67522246e-01 -1.49569213e+00
-1.08902824e+00 -1.56406119e-01 -8.95457745e-01 9.66928899e-01
-7.04761565e-01 -1.15842545e+00 -1.02539964e-01 6.01601422e-01
6.40196145e-01 -4.55171615e-01 1.19130325e+00 1.37309551e-01
6.92964867e-02 6.87718213e-01 -3.86327028e-01 2.71583170e-01
7.34808385e-01 -1.24466658e+00 6.81843102e-01 2.69896269e-01
7.95235395e-01 6.48528457e-01 6.41244829e-01 -5.10176599e-01
-6.47581697e-01 -9.70641553e-01 7.97158003e-01 -3.17112535e-01
6.02938116e-01 -3.94696385e-01 -6.81048989e-01 4.24873084e-01
2.82215476e-01 -6.13339901e-01 1.00696039e+00 2.23628089e-01
-4.90406901e-01 -1.45078197e-01 -5.78406334e-01 9.99420226e-01
1.00574005e+00 -5.21562815e-01 -9.71824586e-01 -1.87211201e-01
8.26620102e-01 -5.57361662e-01 -7.77265787e-01 5.50179303e-01
1.15037823e+00 -1.15815628e+00 1.04567432e+00 -2.63239771e-01
5.30721188e-01 -7.20551193e-01 -4.22925949e-02 -1.56928885e+00
-8.93192053e-01 -1.14159703e+00 1.13146059e-01 1.02399230e+00
6.85196161e-01 2.92046070e-01 8.19606483e-01 -3.59014958e-01
-3.61344993e-01 -4.99928385e-01 -5.49578488e-01 -6.52952433e-01
9.29740518e-02 -6.48061097e-01 1.10264194e+00 9.13871646e-01
4.39388910e-03 6.42897189e-01 -6.40209734e-01 -3.18624884e-01
2.34052628e-01 6.13653958e-01 8.78084242e-01 -1.30316544e+00
-1.02227771e+00 -9.34285045e-01 1.83278918e-02 -7.58661509e-01
-1.67986080e-01 -1.16547489e+00 -1.61516406e-02 -1.24318421e+00
3.55489291e-02 -4.52261448e-01 -7.41195917e-01 6.59417093e-01
6.86469972e-02 8.31823707e-01 6.77301645e-01 2.68785834e-01
-1.30255625e-01 4.99079257e-01 1.43413222e+00 -2.26462960e-01
-8.34008813e-01 2.24366635e-01 -8.54074299e-01 7.57239521e-01
8.82351279e-01 -2.56410301e-01 -5.32060385e-01 -1.46620780e-01
6.42836809e-01 4.76910561e-01 1.11785248e-01 -1.35045612e+00
1.68929517e-01 -1.18100852e-01 6.25005543e-01 -7.52251565e-01
4.12616998e-01 -2.67457932e-01 7.39723742e-01 3.82101238e-01
-6.97374880e-01 3.03087443e-01 3.10065448e-01 -1.66442301e-02
-3.54332387e-01 -1.28917411e-01 6.56290472e-01 -1.76114544e-01
-1.30615145e-01 1.22068375e-01 -2.75626808e-01 -5.46092987e-02
4.44950700e-01 -4.30698305e-01 2.39174500e-01 -4.57589656e-01
-1.28230488e+00 -4.66054887e-01 2.21720085e-01 7.63168454e-01
3.20441961e-01 -1.79896402e+00 -9.90168571e-01 5.28495610e-01
1.54856583e-02 -6.54176474e-02 2.45586470e-01 2.39723865e-02
-3.94708574e-01 8.62379149e-02 -6.43674612e-01 -3.92880738e-01
-9.44830298e-01 3.40879470e-01 2.57191509e-01 -4.43150938e-01
-4.86557364e-01 6.71175361e-01 2.45310038e-01 -5.18422782e-01
1.02791749e-01 -4.36151117e-01 -3.74242514e-01 2.28653759e-01
4.20989007e-01 6.04921728e-02 -1.91924930e-01 -3.95600796e-01
1.74951077e-01 8.02065313e-01 3.92223954e-01 -6.64476097e-01
1.27091122e+00 2.48265147e-01 -8.85638744e-02 7.18772590e-01
4.57248658e-01 6.51991367e-01 -1.16464221e+00 1.36107653e-01
-9.70396772e-02 1.03078678e-01 -4.33681458e-01 -1.13853252e+00
-7.74695516e-01 3.65935177e-01 6.96045831e-02 2.91729778e-01
1.32403421e+00 -2.83581465e-01 8.78026783e-01 3.00070167e-01
3.40206593e-01 -8.27307642e-01 5.35044432e-01 8.59399974e-01
1.35902941e+00 -2.54595071e-01 -2.94467121e-01 2.10041273e-02
-6.91686809e-01 1.08224440e+00 3.98508817e-01 -4.33431774e-01
4.65876967e-01 4.29540128e-01 1.52680159e-01 2.18941003e-01
-7.17172682e-01 -2.27583442e-02 6.91422999e-01 3.05686057e-01
7.78631449e-01 1.25893295e-01 -2.72200227e-01 1.13391662e+00
-1.53463304e+00 -6.28266484e-02 3.46809238e-01 1.25147104e-01
-4.04796153e-01 -1.43599355e+00 -4.36977029e-01 5.14631644e-02
-3.75006407e-01 -4.16939020e-01 -4.39853460e-01 5.55291474e-01
7.13367760e-01 6.15553021e-01 2.96514452e-01 -7.17068315e-01
4.01140034e-01 1.30002588e-01 6.16804302e-01 -7.68626332e-01
-9.39302862e-01 6.27759576e-01 -5.90875298e-02 -2.72670478e-01
-4.27668616e-02 -3.32789123e-01 -1.34198868e+00 -5.17070591e-02
5.68296500e-02 3.15569311e-01 4.30196553e-01 3.97924870e-01
2.15684161e-01 9.84386146e-01 6.47103548e-01 -1.18365705e+00
-3.92933816e-01 -1.06801987e+00 -5.15875518e-01 4.73405361e-01
4.28134613e-02 -2.87104659e-02 -9.02854949e-02 3.30714941e-01] | [16.056365966796875, 5.5233893394470215] |
bd358b7a-52c1-44e9-aa3f-027d29e905f7 | uncertainty-aware-multi-view-representation | 2201.05776 | null | https://arxiv.org/abs/2201.05776v1 | https://arxiv.org/pdf/2201.05776v1.pdf | Uncertainty-Aware Multi-View Representation Learning | Learning from different data views by exploring the underlying complementary information among them can endow the representation with stronger expressive ability. However, high-dimensional features tend to contain noise, and furthermore, the quality of data usually varies for different samples (even for different views), i.e., one view may be informative for one sample but not the case for another. Therefore, it is quite challenging to integrate multi-view noisy data under unsupervised setting. Traditional multi-view methods either simply treat each view with equal importance or tune the weights of different views to fixed values, which are insufficient to capture the dynamic noise in multi-view data. In this work, we devise a novel unsupervised multi-view learning approach, termed as Dynamic Uncertainty-Aware Networks (DUA-Nets). Guided by the uncertainty of data estimated from the generation perspective, intrinsic information from multiple views is integrated to obtain noise-free representations. Under the help of uncertainty, DUA-Nets weigh each view of individual sample according to data quality so that the high-quality samples (or views) can be fully exploited while the effects from the noisy samples (or views) will be alleviated. Our model achieves superior performance in extensive experiments and shows the robustness to noisy data. | ['QinGhua Hu', 'Changqing Zhang', 'Zongbo Han', 'Yu Geng'] | 2022-01-15 | null | null | null | null | ['multi-view-learning'] | ['computer-vision'] | [-1.22011460e-01 -3.64505649e-02 -2.29269385e-01 -5.90801120e-01
-8.89023781e-01 -5.31625271e-01 3.75451833e-01 -1.06288485e-01
1.28685206e-01 4.68551695e-01 6.60101533e-01 5.99821270e-01
-3.78786534e-01 -9.19885397e-01 -4.69521582e-01 -1.02826369e+00
4.09668833e-01 3.69362056e-01 -8.80712196e-02 6.33352995e-02
-1.87839806e-01 1.66454613e-02 -1.43129373e+00 3.82277906e-01
8.30080986e-01 1.00488925e+00 3.01215142e-01 2.09379867e-01
-2.27813795e-01 3.60471725e-01 -5.18118858e-01 -2.82603860e-01
2.41524950e-01 -2.48649776e-01 -5.17220646e-02 6.20243251e-01
1.55328795e-01 -2.93216765e-01 -2.64924288e-01 1.51399696e+00
5.04626870e-01 1.03497148e-01 4.94080931e-01 -1.12797368e+00
-7.32553244e-01 8.18852127e-01 -6.74992442e-01 -6.57796711e-02
1.55544952e-01 6.11910298e-02 1.12025261e+00 -9.86907482e-01
5.91699362e-01 1.59513259e+00 1.57946602e-01 4.84857887e-01
-1.43311560e+00 -5.10041595e-01 9.11164224e-01 1.00921548e-03
-1.07492626e+00 -6.09933674e-01 1.27916169e+00 -2.59208679e-01
-7.27124587e-02 -3.95409130e-02 6.53819501e-01 1.54900801e+00
6.39144480e-02 1.14892006e+00 1.18526280e+00 2.49987260e-01
3.39066386e-01 2.41588235e-01 2.72191875e-02 2.82881230e-01
4.75398332e-01 -8.07359908e-03 -6.66811407e-01 -2.05507830e-01
3.68896902e-01 4.97480780e-01 -4.62292284e-01 -6.77937627e-01
-1.20114541e+00 6.96733356e-01 2.48964310e-01 1.25630081e-01
-4.10030097e-01 -2.80931741e-01 5.46393752e-01 2.80878991e-01
4.61279571e-01 -4.51986603e-02 -5.04518509e-01 9.06349942e-02
-5.23949385e-01 2.68667527e-02 4.72031742e-01 9.87249315e-01
8.38456333e-01 1.03850767e-01 1.94526967e-02 9.11887944e-01
5.62178493e-01 5.89736819e-01 4.14641887e-01 -8.35220039e-01
7.87782729e-01 7.41170704e-01 7.86860362e-02 -1.25929284e+00
-2.55604029e-01 -5.36213398e-01 -1.38428342e+00 2.47060686e-01
2.81057984e-01 -2.53397405e-01 -7.28109837e-01 1.95439947e+00
3.48831058e-01 -1.17673486e-01 3.22772443e-01 8.00029099e-01
8.74188006e-01 6.83756769e-01 -3.51141304e-01 -5.94899416e-01
1.17449594e+00 -5.86673379e-01 -9.35233414e-01 -2.64707655e-01
-3.79097275e-03 -5.08069336e-01 8.09310555e-01 8.17883909e-01
-8.29545617e-01 -6.25280261e-01 -1.19566882e+00 4.85981286e-01
-2.60748193e-02 4.34686020e-02 1.41099751e-01 4.34265256e-01
-2.66719401e-01 4.91159886e-01 -7.70902872e-01 2.61788875e-01
5.91819286e-01 9.99074243e-03 -4.56399769e-01 -5.56081116e-01
-1.01667380e+00 2.69965738e-01 3.12537134e-01 2.92003989e-01
-1.08255804e+00 -4.22918260e-01 -7.51666188e-01 -1.29195796e-02
1.07797408e+00 -5.84652424e-01 6.68411314e-01 -8.20265293e-01
-1.11466825e+00 1.03695348e-01 -2.64277130e-01 1.19170263e-01
6.19490504e-01 -3.00555468e-01 -6.29588783e-01 -4.06703167e-02
2.10600495e-01 2.14350626e-01 1.13257027e+00 -1.93539131e+00
-4.52133179e-01 -6.53924525e-01 1.93215936e-01 3.39911699e-01
-2.07119018e-01 -6.22165859e-01 -4.96381372e-01 -5.82382143e-01
7.11079180e-01 -6.47757292e-01 -1.04948603e-01 1.11973025e-02
-5.05651414e-01 9.67854038e-02 9.26239848e-01 -2.84674764e-01
1.06179643e+00 -2.15943241e+00 4.98442829e-01 1.63015723e-01
4.54869151e-01 -6.85273260e-02 -7.01347813e-02 3.36962670e-01
6.91491663e-02 1.95228696e-01 -2.09053963e-01 -2.04324484e-01
-2.05991745e-01 6.68679833e-01 -1.88789696e-01 3.44767481e-01
1.06900334e-01 5.25499284e-01 -1.14201629e+00 -2.97768682e-01
2.04561770e-01 3.31397176e-01 -2.78888762e-01 2.31352612e-01
-1.55110762e-01 6.43346131e-01 -8.61809850e-01 6.85120463e-01
9.79176104e-01 -3.33434403e-01 1.08925126e-01 -6.52378738e-01
3.01198691e-01 -2.25776464e-01 -1.89815235e+00 1.65275049e+00
-5.39135873e-01 -5.10124080e-02 2.59019256e-01 -8.66071165e-01
8.29173982e-01 4.59924549e-01 4.28186685e-01 -3.27506840e-01
-1.90483332e-01 -4.42204252e-02 -2.90500261e-02 -5.83724976e-01
1.41412020e-01 -3.39997530e-01 -4.09557857e-02 5.94921768e-01
1.77864656e-01 4.22982611e-02 -1.64736565e-02 1.76477194e-01
5.32518685e-01 7.14864135e-02 4.08310175e-01 7.60941878e-02
5.45627952e-01 -8.08928907e-01 1.15492404e+00 6.51993513e-01
-2.28685483e-01 8.65023136e-01 7.13402331e-01 -3.78304452e-01
-9.16485667e-01 -1.05123699e+00 -3.08725517e-02 6.90848827e-01
2.47384578e-01 -4.31833595e-01 -3.44670594e-01 -1.02602839e+00
-1.48015827e-01 4.40487504e-01 -6.92520976e-01 -2.60174781e-01
-1.10724673e-01 -8.06382298e-01 -6.18383624e-02 3.89931440e-01
4.39315289e-01 -6.00520551e-01 -1.68895826e-01 3.37867111e-01
-2.93386668e-01 -9.73464072e-01 -4.39583808e-01 1.73362866e-01
-9.36253726e-01 -1.05973077e+00 -4.10130590e-01 -5.75017594e-02
7.41993606e-01 5.33171892e-01 9.73049104e-01 -3.46776187e-01
3.53064209e-01 4.02942151e-01 -4.80866313e-01 -2.55886525e-01
-4.07885730e-01 -2.88336366e-01 2.85940677e-01 4.75885004e-01
1.79368138e-01 -8.85283351e-01 -3.88868958e-01 3.79544050e-01
-1.26307166e+00 -5.02473153e-02 5.27038455e-01 1.21419632e+00
9.46932912e-01 4.73226070e-01 7.73568690e-01 -1.00948381e+00
4.73107487e-01 -6.69199526e-01 -4.78573143e-01 4.11134005e-01
-5.65010905e-01 2.46575743e-01 8.79403710e-01 -4.26348656e-01
-1.27448499e+00 -1.24474704e-01 2.02625960e-01 -8.25596392e-01
-1.59269035e-01 6.24097764e-01 -9.81062770e-01 4.40053374e-01
3.92968982e-01 2.34839797e-01 3.06161940e-02 -4.46609259e-01
4.62712914e-01 3.80531520e-01 1.17836788e-01 -6.99696243e-01
7.58183479e-01 6.95453644e-01 1.84676284e-03 -4.81211782e-01
-1.24924362e+00 -2.09550887e-01 -5.84067583e-01 -2.59305686e-01
4.57222730e-01 -1.22023416e+00 -1.94885492e-01 4.01576817e-01
-6.82626963e-01 6.98822856e-01 -3.98366868e-01 3.81524891e-01
-1.34695962e-01 4.79000688e-01 -1.25989109e-01 -1.06277001e+00
9.05336887e-02 -1.51521087e+00 1.03580475e+00 4.15076792e-01
2.85080492e-01 -9.05573964e-01 -2.34458163e-01 3.29230249e-01
-1.26108853e-03 1.91531956e-01 8.14972103e-01 -6.81114674e-01
-6.39541507e-01 -2.56171167e-01 -2.55103949e-02 6.81143284e-01
5.96127927e-01 4.15156633e-02 -1.19706380e+00 -2.51900584e-01
6.30622387e-01 -4.84093159e-01 9.52547789e-01 3.85840684e-01
1.23700094e+00 -3.34007353e-01 -4.32782024e-02 2.31779590e-01
1.43285561e+00 -1.01188952e-02 2.26655602e-01 -1.48600534e-01
8.62622917e-01 7.55142570e-01 5.47612965e-01 8.42581511e-01
2.75750577e-01 3.62973839e-01 7.50278175e-01 5.19005120e-01
3.70715618e-01 -3.56244951e-01 3.91489118e-01 1.14334643e+00
9.76806059e-02 -4.15351272e-01 -3.86745006e-01 4.02100861e-01
-1.98009741e+00 -1.03635454e+00 1.37077928e-01 2.28923202e+00
4.53310758e-01 3.46578807e-01 -1.00327000e-01 -1.92260053e-02
7.26091683e-01 6.70555234e-01 -9.03519034e-01 3.70536685e-01
-4.50394332e-01 -6.64547384e-01 -4.22240160e-02 2.41963759e-01
-1.00849891e+00 1.58598870e-01 5.04724932e+00 8.12907517e-01
-6.26763940e-01 3.86743546e-02 8.37864816e-01 -3.78183633e-01
-9.92181540e-01 2.88087595e-02 -5.95257223e-01 6.50176406e-01
4.62404847e-01 -1.69686154e-01 4.33184773e-01 8.19923401e-01
3.25964838e-01 -2.23106727e-01 -1.06867242e+00 9.39912319e-01
7.75155425e-02 -1.02765107e+00 3.55151772e-01 1.75626338e-01
9.43420589e-01 -7.30187073e-02 1.07123218e-01 1.71419173e-01
2.52224505e-01 -5.83339036e-01 5.59185445e-01 8.96260679e-01
2.48471633e-01 -1.04020655e+00 8.36070240e-01 7.36399591e-01
-1.18879831e+00 -2.73367614e-01 -5.91500819e-01 3.29027504e-01
2.13763043e-01 1.20170283e+00 3.80402394e-02 1.06634188e+00
5.63190401e-01 1.08406270e+00 -3.55136096e-01 3.93974632e-01
-1.00358658e-01 3.68166268e-01 -2.34444186e-01 4.84266251e-01
1.15892611e-01 -6.28460824e-01 8.73727918e-01 4.00982440e-01
3.44090790e-01 6.15310818e-02 4.99280721e-01 8.96988750e-01
-1.83593661e-01 -1.48915052e-01 -7.33391821e-01 -7.86243752e-03
4.83610600e-01 1.29745221e+00 -3.23976308e-01 -3.95505279e-01
-7.41763413e-01 6.25278533e-01 3.25270772e-01 3.76322031e-01
-2.41900131e-01 2.78234750e-01 7.12706089e-01 -3.04029554e-01
3.06923449e-01 1.03135772e-01 -2.66151369e-01 -1.72841465e+00
2.79145867e-01 -1.03009272e+00 5.14093101e-01 -7.16588795e-01
-2.00689411e+00 3.78498763e-01 -4.24562879e-02 -1.91591740e+00
-8.80552381e-02 -1.90667957e-01 -4.42409068e-01 8.44494641e-01
-1.22335863e+00 -9.28459942e-01 -1.43848866e-01 3.83530408e-01
6.47593439e-01 -2.20763221e-01 6.16863906e-01 1.14938132e-02
-6.06920302e-01 3.00938874e-01 4.67376024e-01 -1.56864405e-01
7.09170640e-01 -1.29424810e+00 -1.52110204e-01 7.73868620e-01
2.48531833e-01 6.16588116e-01 5.61220288e-01 -6.48665071e-01
-1.52915311e+00 -1.00356507e+00 1.13086581e-01 -2.40222991e-01
5.37088096e-01 -2.52690196e-01 -1.25144815e+00 4.05421942e-01
3.79829332e-02 4.14380759e-01 8.92253101e-01 3.37403804e-01
-5.86152911e-01 -4.37871307e-01 -1.17642224e+00 4.48856294e-01
7.42615700e-01 -5.62219381e-01 -4.84888434e-01 7.37213567e-02
7.37270057e-01 -2.68107533e-01 -1.00041616e+00 4.31734413e-01
4.78591174e-01 -1.41185904e+00 7.50333607e-01 -5.39780140e-01
5.55143118e-01 -4.83726263e-01 -5.34058571e-01 -1.78380919e+00
-3.26957524e-01 -1.36798933e-01 -4.24094617e-01 1.54369009e+00
2.01451644e-01 -4.83896703e-01 5.31377733e-01 5.96554577e-01
1.07329033e-01 -8.66292894e-01 -9.87096131e-01 -5.05666196e-01
-2.65325218e-01 -5.94725788e-01 7.23858893e-01 7.72253752e-01
-3.62512469e-01 4.03438270e-01 -8.57652068e-01 4.73570675e-01
8.54823291e-01 3.94642293e-01 5.26602745e-01 -1.31580222e+00
-4.78624582e-01 -1.61163002e-01 -2.96640664e-01 -9.38488543e-01
-5.74300475e-02 -5.51987886e-01 -3.81179377e-02 -1.35447764e+00
4.98027116e-01 -2.21501425e-01 -5.15966773e-01 -1.33035961e-03
-5.64830244e-01 -4.12719667e-01 2.64484584e-01 3.56963724e-01
-5.99167347e-01 9.31395888e-01 1.54650700e+00 -2.35365421e-01
-1.49367517e-02 3.57820362e-01 -9.97018218e-01 9.66448843e-01
3.37745726e-01 -4.35303867e-01 -8.03411186e-01 -3.55100125e-01
4.14575487e-01 3.54214132e-01 1.26982361e-01 -8.77629340e-01
7.60212392e-02 -2.15458438e-01 7.31174111e-01 -6.90402329e-01
3.50813895e-01 -1.10241520e+00 1.89893216e-01 -8.37778896e-02
-2.47902170e-01 -2.71974981e-01 -3.13554704e-01 1.45138037e+00
-4.80022728e-01 -2.67896742e-01 6.16167784e-01 -5.80405116e-01
-3.26956838e-01 5.96399844e-01 -1.30490154e-01 2.02099457e-01
6.10990644e-01 -1.07209466e-01 -2.00329483e-01 -7.24092245e-01
-1.05195665e+00 5.32000661e-01 4.37959373e-01 4.25907671e-01
6.34676397e-01 -1.65297627e+00 -5.98688185e-01 3.60427499e-01
5.94034314e-01 4.55258310e-01 9.07941103e-01 4.80536699e-01
7.07769871e-01 -2.62706757e-01 1.90193467e-02 -7.61313438e-01
-9.33676958e-01 8.64057779e-01 1.09446734e-01 -3.66971970e-01
-5.21293044e-01 5.57822824e-01 5.13990521e-01 -5.49390912e-01
4.26722653e-02 -1.61735624e-01 -4.91238624e-01 6.54437959e-01
6.26051366e-01 3.59193146e-01 -2.26618424e-01 -7.05375910e-01
1.70752164e-02 5.18532038e-01 -1.73057511e-01 -1.41030457e-02
1.52998376e+00 -6.60366476e-01 4.53967042e-02 1.21439159e+00
1.09298766e+00 1.21306211e-01 -1.59718168e+00 -7.46338964e-01
-4.18729454e-01 -6.63370490e-01 5.58083318e-02 -6.63572550e-01
-1.42573667e+00 8.30095828e-01 4.26703602e-01 2.87949204e-01
1.20259738e+00 -4.48253118e-02 4.42531139e-01 3.78954947e-01
4.36282963e-01 -1.16621327e+00 2.04674095e-01 1.98670462e-01
9.01704371e-01 -1.56797433e+00 2.08411619e-01 -3.41670096e-01
-1.01399231e+00 1.12695372e+00 6.86057210e-01 4.45634015e-02
8.40952098e-01 9.04336795e-02 -4.69541252e-02 -3.14615190e-01
-9.86943781e-01 1.26606822e-01 2.97015309e-01 5.15992403e-01
-1.74044341e-01 1.86143294e-01 3.27034712e-01 8.92608523e-01
4.56479281e-01 -3.16862226e-01 5.82886934e-01 7.40039825e-01
-3.13898772e-01 -1.02700341e+00 -4.34427708e-01 7.90459275e-01
-2.10069910e-01 2.94097036e-01 -1.08313784e-01 4.34485078e-01
3.69096607e-01 9.16896045e-01 -3.76896530e-01 -4.88354653e-01
3.79237324e-01 1.33770719e-01 1.44642398e-01 -5.99864662e-01
-1.22037992e-01 6.39506638e-01 -6.68628663e-02 -5.22643387e-01
-7.30439544e-01 -9.58383739e-01 -7.96866059e-01 7.35000893e-02
-3.89742106e-01 -1.22869670e-01 1.94062859e-01 1.08010232e+00
3.43917578e-01 6.73074007e-01 1.02661300e+00 -7.14355350e-01
-7.73561597e-01 -8.17954063e-01 -9.75185156e-01 3.88891786e-01
4.35201496e-01 -8.54299784e-01 -6.47481561e-01 -2.15724066e-01] | [8.485212326049805, 4.563709735870361] |
6517e182-474e-498a-b8ea-c7dcb0f93a91 | transformers-generalize-deepsets-and-can-be-1 | null | null | https://openreview.net/forum?id=scn3RYn1DYx | https://openreview.net/pdf?id=scn3RYn1DYx | Transformers Generalize DeepSets and Can be Extended to Graphs & Hypergraphs | We present a generalization of Transformers to any-order permutation invariant data (sets, graphs, and hypergraphs). We begin by observing that Transformers generalize DeepSets, or first-order (set-input) permutation invariant MLPs. Then, based on recently characterized higher-order invariant MLPs, we extend the concept of self-attention to higher orders and propose higher-order Transformers for order-$k$ data ($k=2$ for graphs and $k>2$ for hypergraphs). Unfortunately, higher-order Transformers turn out to have prohibitive complexity $\mathcal{O}(n^{2k})$ to the number of input nodes $n$. To address this problem, we present sparse higher-order Transformers that have quadratic complexity to the number of input hyperedges, and further adopt the kernel attention approach to reduce the complexity to linear. In particular, we show that the sparse second-order Transformers with kernel attention are theoretically more expressive than message passing operations while having an asymptotically identical complexity. Our models achieve significant performance improvement over invariant MLPs and message-passing graph neural networks in large-scale graph regression and set-to-(hyper)graph prediction tasks. Our implementation is available at https://github.com/jw9730/hot. | ['Seunghoon Hong', 'Saeyoon Oh', 'Jinwoo Kim'] | 2021-05-21 | null | null | null | neurips-2021-12 | ['graph-regression'] | ['graphs'] | [ 3.76431555e-01 4.71439391e-01 -1.15219578e-01 -2.17053697e-01
-7.21651912e-01 -5.49194455e-01 2.68426239e-01 2.90425509e-01
-3.20195466e-01 4.98824209e-01 -7.27246106e-02 -7.88844407e-01
-5.64635515e-01 -1.25426054e+00 -1.39241183e+00 -7.05325425e-01
-8.53538454e-01 5.89622498e-01 9.00460333e-02 -5.27542308e-02
1.16317764e-01 5.19869328e-01 -1.52065969e+00 2.35753790e-01
5.21615863e-01 9.61093605e-01 -1.53324276e-01 1.07768726e+00
3.53165232e-02 7.70918608e-01 -1.85135990e-01 -6.14507079e-01
3.97945374e-01 -1.96878035e-02 -1.28504181e+00 -2.43581325e-01
8.45010042e-01 -5.31157888e-02 -8.70343804e-01 1.27617717e+00
2.87852526e-01 2.86839813e-01 4.34986681e-01 -1.44975412e+00
-1.16603124e+00 1.27070737e+00 -2.82909453e-01 4.58202988e-01
-5.57272993e-02 1.13775522e-01 1.52923000e+00 -4.13457274e-01
4.69075926e-02 1.30185223e+00 7.43973374e-01 1.69296116e-01
-1.40424383e+00 -7.01725423e-01 3.80129904e-01 3.20339978e-01
-1.31386626e+00 -2.47881263e-01 6.77415490e-01 -2.56528258e-01
1.55232346e+00 5.94667971e-01 6.57831013e-01 6.45752847e-01
1.85104325e-01 5.44244289e-01 7.43598580e-01 -2.79116839e-01
-1.27583161e-01 -3.11133623e-01 6.06633186e-01 1.34001756e+00
4.63947624e-01 -1.38517410e-01 -1.34999067e-01 -1.37159675e-01
8.34664941e-01 -1.59579143e-02 -2.97861189e-01 -2.21583202e-01
-1.06558383e+00 8.47646952e-01 8.44897568e-01 2.85716861e-01
-5.92382662e-02 9.10517931e-01 3.71930450e-01 6.84627891e-01
1.80126041e-01 5.40409029e-01 -7.18350768e-01 7.12295324e-02
-1.46179363e-01 -2.77047694e-01 9.24967170e-01 1.09828448e+00
1.01933384e+00 2.66259730e-01 -1.42034784e-01 6.99588776e-01
2.01128628e-02 5.88696420e-01 2.63008595e-01 -5.63849688e-01
4.62713301e-01 8.12256396e-01 -4.98569995e-01 -1.07789719e+00
-6.38373196e-01 -7.00050354e-01 -1.22941637e+00 -2.89194196e-01
2.39353225e-01 3.41975093e-01 -1.10656011e+00 1.93536019e+00
-1.90883249e-01 1.03925586e-01 -2.52904266e-01 2.12399125e-01
7.81177044e-01 7.15057969e-01 -9.00010020e-02 3.21508870e-02
1.49114299e+00 -8.57640505e-01 -8.58341008e-02 -2.43083805e-01
8.21393251e-01 -6.10007532e-02 1.49467468e+00 9.24186036e-02
-1.20218205e+00 -2.93073922e-01 -8.52224529e-01 -1.75003573e-01
-6.73783422e-01 -2.74507552e-01 8.85839999e-01 6.44424498e-01
-1.64234710e+00 7.77439415e-01 -7.55605578e-01 -2.90608883e-01
4.34748203e-01 9.05796587e-01 -3.49593818e-01 2.90204845e-02
-1.20884764e+00 5.32693863e-01 5.14735758e-01 -4.09174128e-04
-6.79326177e-01 -9.65066135e-01 -1.19833016e+00 5.94250858e-01
4.48865354e-01 -6.05699658e-01 1.03870249e+00 -6.32843614e-01
-1.21344543e+00 7.70421505e-01 -1.10360943e-01 -5.56234658e-01
-2.30868235e-01 2.02968553e-01 -1.86856896e-01 7.03854635e-02
-1.39132529e-01 3.65028918e-01 6.36496603e-01 -7.47508526e-01
-4.12490398e-01 -2.23703921e-01 5.82179308e-01 4.80537638e-02
-5.73075891e-01 -2.04386234e-01 -5.00185907e-01 -3.33384842e-01
2.89267749e-02 -1.08958113e+00 -1.62918404e-01 -4.64099914e-01
-6.34942472e-01 -3.87175918e-01 2.47638687e-01 -5.28316498e-01
1.08081686e+00 -2.10718513e+00 1.60862103e-01 4.51560676e-01
6.89312994e-01 1.55964240e-01 -3.84037346e-01 4.01104033e-01
-3.97724241e-01 3.82096320e-01 -3.96423519e-01 7.22981915e-02
3.91295761e-01 3.52825522e-01 1.86556168e-02 4.37827229e-01
1.76260844e-01 1.31134915e+00 -7.95685232e-01 -1.99652210e-01
-3.71306241e-02 2.67253309e-01 -8.28284800e-01 -1.75962314e-01
-9.47775990e-02 -3.59930158e-01 -1.26824096e-01 5.23044109e-01
4.17367876e-01 -9.81444836e-01 2.27693498e-01 -3.94917876e-01
4.60701883e-01 5.23242831e-01 -9.91168976e-01 1.14689624e+00
-4.41685498e-01 5.99878788e-01 -8.61383975e-02 -1.34051812e+00
3.16110820e-01 5.40784709e-02 3.85720581e-01 -6.19664729e-01
1.04130983e-01 5.65949008e-02 2.52823561e-01 -1.66382268e-01
3.64204049e-01 -1.45119190e-01 -3.16667736e-01 3.39733452e-01
2.70990491e-01 2.22101003e-01 4.09000903e-01 3.13564092e-01
1.71223032e+00 -5.60906172e-01 1.48491874e-01 -3.23056579e-01
2.92345852e-01 -4.20105278e-01 2.32725143e-01 1.06328738e+00
2.99275760e-03 -6.10304549e-02 1.15909135e+00 -4.81269747e-01
-8.99998128e-01 -1.12605166e+00 2.45549619e-01 1.51720166e+00
-2.29723021e-01 -3.24416161e-01 -5.26134372e-01 -6.08044326e-01
2.52805978e-01 4.53874379e-01 -7.82061577e-01 -3.54691297e-01
-6.50509536e-01 -9.33131516e-01 8.16738069e-01 8.15361679e-01
2.96557754e-01 -1.12234807e+00 -6.73008338e-02 3.82214859e-02
1.15775637e-01 -1.04552114e+00 -8.21149826e-01 7.46570110e-01
-7.71592975e-01 -1.16000032e+00 -3.22821140e-01 -1.15268075e+00
1.03002679e+00 -8.58158097e-02 1.16233623e+00 1.36161998e-01
-2.32556567e-01 6.01153493e-01 4.42102440e-02 -1.53207600e-01
-2.27180794e-01 3.71637970e-01 5.48279844e-02 -1.12847738e-01
1.30657986e-01 -8.54443729e-01 -2.78889567e-01 -5.34576923e-02
-1.00400817e+00 -2.87574053e-01 8.69880974e-01 8.65817785e-01
5.94577968e-01 4.00580794e-01 1.59986347e-01 -1.18916559e+00
5.66676617e-01 -1.83726430e-01 -8.20996165e-01 2.92029917e-01
-3.83059770e-01 5.32355666e-01 1.10012829e+00 -5.35497010e-01
-1.66818559e-01 -8.51088613e-02 -6.61490336e-02 -5.73982298e-01
1.53659776e-01 7.92271495e-01 -1.33938536e-01 -4.63734448e-01
3.99064511e-01 5.41596472e-01 -1.51521385e-01 -1.49551392e-01
4.96264607e-01 5.47126532e-02 4.74193841e-01 -5.24431586e-01
1.02753687e+00 1.98824719e-01 3.01107913e-01 -9.49358761e-01
-4.82366174e-01 -6.45384341e-02 -3.54963124e-01 3.40364605e-01
3.98009241e-01 -5.05992889e-01 -1.33469820e+00 4.73983198e-01
-7.84001708e-01 -6.83402658e-01 -2.79383272e-01 2.12436467e-01
-5.08348763e-01 4.36316788e-01 -1.12714982e+00 -5.66593170e-01
-5.37212789e-01 -1.06821418e+00 6.80444062e-01 -2.34689757e-01
1.57926485e-01 -1.00871742e+00 -2.31793225e-01 -8.84167999e-02
4.18035358e-01 -2.15509444e-01 1.78377330e+00 -9.62671995e-01
-8.92054915e-01 -1.82537943e-01 -5.51040590e-01 3.05284739e-01
1.08925225e-02 -4.49908108e-01 -5.90850592e-01 -4.52800542e-01
-4.83585626e-01 -1.25910297e-01 9.67479706e-01 4.51666385e-01
1.52096927e+00 -9.80274498e-01 -3.83400500e-01 8.98272574e-01
1.37898839e+00 -6.71450943e-02 4.89093661e-01 -1.44062653e-01
1.07467031e+00 8.45643505e-02 -4.52925354e-01 2.98258420e-02
5.71024418e-01 1.79737985e-01 5.06655753e-01 -8.55900347e-02
3.78598347e-02 -3.46199065e-01 3.28629494e-01 1.15057421e+00
1.28362954e-01 -3.59007061e-01 -7.79385805e-01 6.64729118e-01
-1.51645708e+00 -7.24045336e-01 -6.91301003e-02 2.22755218e+00
5.69260359e-01 3.58425409e-01 1.15821220e-01 1.99395567e-01
6.84067726e-01 1.08050086e-01 -4.62675184e-01 -7.38796532e-01
-2.27413792e-02 6.26870155e-01 1.17446589e+00 6.73736215e-01
-1.14045835e+00 9.49197233e-01 5.83026934e+00 8.98373425e-01
-9.81383383e-01 -1.07919216e-01 5.22021651e-01 -4.80207503e-02
-6.71003997e-01 -7.20172152e-02 -4.37472522e-01 3.48101646e-01
1.17311406e+00 -1.01215420e-02 1.07927084e+00 6.91679120e-01
-5.84084392e-01 4.49203491e-01 -1.47035611e+00 9.31007385e-01
-2.46292707e-02 -1.31772947e+00 1.63432345e-01 2.15493947e-01
1.98681667e-01 3.88909698e-01 1.95975170e-01 6.98541045e-01
7.54496872e-01 -1.21863842e+00 2.49064952e-01 7.45536238e-02
7.74111807e-01 -6.28082871e-01 4.34875101e-01 1.37607278e-02
-1.51935577e+00 -3.21627796e-01 -4.78400916e-01 -1.13534167e-01
-1.28198281e-01 4.55157667e-01 -8.06083918e-01 4.95722324e-01
8.83123040e-01 5.22456169e-01 -5.31188846e-01 7.12392211e-01
2.66410597e-02 5.41671932e-01 -7.39828110e-01 -3.30870122e-01
2.80436814e-01 1.04029022e-01 3.65270674e-01 1.07589102e+00
2.40491614e-01 2.83801734e-01 8.16443488e-02 5.55998206e-01
-6.00330651e-01 -1.02143556e-01 -7.67889619e-01 -4.39742893e-01
3.75590026e-01 1.08630776e+00 -8.15881729e-01 -2.76582688e-01
-3.19537342e-01 8.23187709e-01 9.13401306e-01 3.13434303e-01
-9.59795296e-01 -8.22435558e-01 6.25428081e-01 6.44951612e-02
6.97625816e-01 -2.86150962e-01 1.01536408e-01 -1.01074338e+00
3.12059317e-02 -7.08828568e-01 8.08325648e-01 -4.26704615e-01
-1.38705003e+00 5.90601623e-01 -8.51018950e-02 -5.53218126e-01
1.70075253e-01 -1.06141365e+00 -3.63049507e-01 5.22668779e-01
-1.41526997e+00 -1.13492763e+00 1.78578749e-01 5.82122862e-01
-1.38135731e-01 1.01221472e-01 9.22309160e-01 4.71413910e-01
-5.99916458e-01 1.13242340e+00 -3.36824954e-02 4.15628314e-01
4.45007980e-02 -1.46207404e+00 7.38763094e-01 7.85869181e-01
3.27526510e-01 7.39904463e-01 2.68787384e-01 -2.78724372e-01
-1.97828424e+00 -1.19658840e+00 8.77066910e-01 -3.76008064e-01
9.75227118e-01 -5.87181687e-01 -8.64106476e-01 1.32963455e+00
7.89270271e-03 3.13728541e-01 5.71989477e-01 5.40761411e-01
-6.50636137e-01 -3.30322564e-01 -8.78437340e-01 7.64742255e-01
1.53060317e+00 -7.56611049e-01 -1.26946911e-01 3.72383654e-01
1.06087923e+00 -3.69981200e-01 -1.08885944e+00 4.49799359e-01
3.55697632e-01 -4.89534110e-01 9.80273366e-01 -8.58209610e-01
4.80196811e-02 1.27227694e-01 -2.88034678e-01 -1.17588270e+00
-9.45554256e-01 -7.74973035e-01 -2.74696767e-01 6.04135752e-01
7.74869323e-01 -1.23904479e+00 9.54977632e-01 2.46626049e-01
-3.18185806e-01 -8.00766706e-01 -1.00123131e+00 -8.51155102e-01
8.14496428e-02 -4.76218522e-01 6.62352502e-01 1.02034402e+00
2.60159969e-01 3.89367521e-01 -1.48325697e-01 6.58360064e-01
4.72844958e-01 1.68254584e-01 5.32495856e-01 -1.13281167e+00
-6.82469726e-01 -7.73909330e-01 -6.46247268e-01 -1.15776753e+00
2.92956859e-01 -1.54455721e+00 -1.12772904e-01 -1.38468552e+00
1.96435466e-01 -5.33231258e-01 -6.91087425e-01 1.15228367e+00
-5.03318571e-02 2.07898527e-01 -1.63915530e-02 -3.89737576e-01
-6.90830171e-01 3.18168610e-01 9.85279500e-01 -3.96637231e-01
-3.29711027e-02 -1.10718936e-01 -1.06152987e+00 4.85481679e-01
8.59115899e-01 -3.99930865e-01 -3.95860732e-01 -6.51090920e-01
5.24284959e-01 -9.45717469e-02 4.29715455e-01 -7.97158897e-01
2.69376576e-01 4.08879668e-02 -2.33079735e-02 -2.92475313e-01
3.37638319e-01 -5.65613151e-01 -6.37915507e-02 7.68502116e-01
-4.09386843e-01 4.21335369e-01 4.18930531e-01 6.32253349e-01
2.83758909e-01 6.42777234e-02 6.67571604e-01 -2.52409399e-01
-3.53778213e-01 7.78628409e-01 -3.19395572e-01 2.56947786e-01
6.25909626e-01 -1.75307289e-01 -6.18508279e-01 -4.61154073e-01
-6.82390809e-01 2.64564812e-01 4.93979901e-02 2.06919894e-01
4.54592735e-01 -1.03398001e+00 -5.13236582e-01 4.35012490e-01
5.72673157e-02 2.29548514e-02 3.05355549e-01 9.53494966e-01
-4.67118144e-01 6.04833007e-01 -5.53893670e-02 -3.98698270e-01
-1.11185634e+00 8.19042802e-01 3.05785686e-01 -5.45726776e-01
-5.08981526e-01 1.20252800e+00 3.36868912e-01 -7.89058089e-01
4.19797480e-01 -9.04491603e-01 3.48698556e-01 -3.57857049e-01
5.50968461e-02 3.38377535e-01 2.75788486e-01 -2.51623303e-01
-4.85186338e-01 4.31324095e-01 -1.38695091e-01 4.16367203e-01
1.42705452e+00 4.07724112e-01 -5.16288996e-01 1.62927702e-01
1.59059596e+00 -9.89845693e-02 -7.25702941e-01 -4.14296359e-01
-1.27773523e-01 7.73332547e-03 -3.51509571e-01 -3.63208562e-01
-1.27386725e+00 6.63165092e-01 2.01035678e-01 7.51848280e-01
1.14022708e+00 2.89806902e-01 6.12025321e-01 9.48505819e-01
2.81112015e-01 -4.77739632e-01 -1.96630761e-01 7.41946399e-01
6.40195131e-01 -8.21714044e-01 -2.00748205e-01 -5.43311715e-01
-3.49578448e-03 7.05532134e-01 5.35828412e-01 -2.44216040e-01
9.49689984e-01 2.64879882e-01 -6.19128704e-01 -3.22672039e-01
-8.29630435e-01 -1.94607452e-01 2.81510264e-01 6.14453316e-01
1.40216514e-01 2.00576141e-01 5.86914318e-03 5.57793736e-01
-4.46267873e-01 -2.94761866e-01 4.28932846e-01 6.89760089e-01
-3.82524490e-01 -9.59926009e-01 9.06894729e-02 1.17774618e+00
-4.40375656e-01 -6.66167676e-01 -1.30602136e-01 8.31050575e-01
-3.85065079e-01 4.98269588e-01 2.93001354e-01 -4.95629132e-01
3.08503538e-01 1.66631624e-01 7.17406511e-01 -5.82617164e-01
-4.99860168e-01 -5.25288463e-01 1.72371536e-01 -5.20539701e-01
1.39865175e-01 -1.74546957e-01 -1.12590492e+00 -7.01676607e-01
-8.36651996e-02 3.08396332e-02 1.01204984e-01 6.00608945e-01
6.42587364e-01 7.11760879e-01 3.79073083e-01 -5.37579715e-01
-5.71976900e-01 -8.31201017e-01 -7.10325122e-01 1.50320992e-01
4.21887577e-01 -2.56386936e-01 -5.03192306e-01 -1.90483049e-01] | [6.950056552886963, 6.284908294677734] |
ca7a050a-bfe9-401a-b687-d475583538ae | optimal-transport-for-offline-imitation | 2303.13971 | null | https://arxiv.org/abs/2303.13971v1 | https://arxiv.org/pdf/2303.13971v1.pdf | Optimal Transport for Offline Imitation Learning | With the advent of large datasets, offline reinforcement learning (RL) is a promising framework for learning good decision-making policies without the need to interact with the real environment. However, offline RL requires the dataset to be reward-annotated, which presents practical challenges when reward engineering is difficult or when obtaining reward annotations is labor-intensive. In this paper, we introduce Optimal Transport Reward labeling (OTR), an algorithm that assigns rewards to offline trajectories, with a few high-quality demonstrations. OTR's key idea is to use optimal transport to compute an optimal alignment between an unlabeled trajectory in the dataset and an expert demonstration to obtain a similarity measure that can be interpreted as a reward, which can then be used by an offline RL algorithm to learn the policy. OTR is easy to implement and computationally efficient. On D4RL benchmarks, we show that OTR with a single demonstration can consistently match the performance of offline RL with ground-truth rewards. | ['Marc Peter Deisenroth', 'Edward Grefenstette', 'samuel cohen', 'Zhengyao Jiang', 'Yicheng Luo'] | 2023-03-24 | null | null | null | null | ['offline-rl', 'd4rl'] | ['playing-games', 'robots'] | [-2.13771060e-01 1.30452320e-01 -5.62798202e-01 -2.37182230e-01
-9.91637111e-01 -9.54721928e-01 4.06759262e-01 3.13856691e-01
-7.26120234e-01 9.55574572e-01 9.27850790e-03 -4.62720543e-01
1.51452199e-02 -5.30739427e-01 -1.01366079e+00 -5.43841839e-01
-3.11330736e-01 5.85955799e-01 1.61826238e-01 2.55451053e-02
1.97344735e-01 6.70958459e-01 -1.38791060e+00 -1.78516895e-01
7.79177129e-01 9.29542422e-01 4.43121016e-01 8.68587732e-01
1.12308942e-01 1.03778422e+00 -4.15054709e-01 -3.94798554e-02
6.13703609e-01 -3.40574056e-01 -9.60266769e-01 1.60929877e-02
-3.57008576e-02 -1.00368023e+00 -3.48690808e-01 9.24985886e-01
1.66748807e-01 4.62846637e-01 5.66526592e-01 -1.55503833e+00
-1.58451825e-01 6.56257927e-01 -1.39930859e-01 -4.19183113e-02
3.72203052e-01 8.19418728e-01 1.11125302e+00 -3.50542128e-01
9.37267900e-01 1.09949040e+00 1.95672601e-01 6.11325562e-01
-1.12379313e+00 -2.67625898e-01 2.52651572e-01 2.03862116e-01
-7.73300827e-01 -4.20195125e-02 4.21656638e-01 -5.02520025e-01
7.95862913e-01 -1.84867159e-01 9.68120635e-01 9.59162056e-01
-1.47592455e-01 1.14723504e+00 1.07041824e+00 -4.62601781e-02
6.70790672e-01 -1.88454464e-01 -2.24175230e-01 8.16299379e-01
-1.87211886e-01 6.46132231e-01 -4.05868918e-01 1.24495439e-01
8.01642299e-01 7.05235600e-02 9.72959399e-02 -6.34342372e-01
-1.55449951e+00 6.07502401e-01 6.60723269e-01 -2.70639271e-01
-5.81289113e-01 7.92784095e-01 3.77878040e-01 5.85352540e-01
-1.03001870e-01 7.37523973e-01 -4.01314646e-01 -7.04346240e-01
-4.34327334e-01 5.50350070e-01 5.78036189e-01 1.01844323e+00
9.04959261e-01 8.97083357e-02 -2.26899564e-01 2.21543774e-01
1.56540662e-01 6.57038689e-01 9.91162434e-02 -1.76349068e+00
6.00759327e-01 4.26886767e-01 8.10158730e-01 -2.40424722e-01
-1.16012104e-01 1.31452635e-01 -1.18215634e-02 7.51244426e-01
8.47595990e-01 -4.13335919e-01 -5.87232471e-01 1.56106317e+00
5.28814137e-01 4.44450170e-01 4.38929111e-01 1.22285903e+00
1.97505504e-02 6.13676548e-01 -1.08429715e-02 4.71647196e-02
5.24366379e-01 -1.19750750e+00 -1.98710665e-01 3.11982520e-02
9.93384480e-01 -3.62281680e-01 1.28945124e+00 2.90513098e-01
-9.20925677e-01 -3.63506466e-01 -8.72441530e-01 2.78262869e-02
-4.55511138e-02 1.14884444e-01 4.81651932e-01 1.59733653e-01
-8.89287531e-01 1.19625497e+00 -1.06056035e+00 -1.19811662e-01
5.10345459e-01 4.62603062e-01 -1.31213501e-01 -1.11815415e-01
-6.15989685e-01 1.01138866e+00 2.30206415e-01 -1.89879164e-01
-1.92473459e+00 -6.41183197e-01 -7.37329543e-01 -1.77632928e-01
7.31535316e-01 -2.67247796e-01 1.87085152e+00 -8.77328396e-01
-1.90182960e+00 2.93281436e-01 1.52081177e-01 -6.66180551e-01
7.73182154e-01 -2.81388313e-01 1.07629083e-01 2.17822775e-01
1.10613897e-01 9.89147961e-01 8.25978100e-01 -1.04594898e+00
-9.22700405e-01 8.92003775e-02 5.19881427e-01 4.80737239e-01
2.07889885e-01 -2.03974530e-01 -9.76181254e-02 2.42034178e-02
-5.03320098e-01 -1.17797041e+00 -5.84132671e-01 2.54780143e-01
-2.65016735e-01 -4.42249089e-01 8.17031920e-01 -4.10050064e-01
5.05876064e-01 -1.95073056e+00 2.40959719e-01 1.85879946e-01
4.43411656e-02 1.18820943e-01 -3.18578392e-01 4.42459822e-01
4.55750853e-01 -2.73961008e-01 -7.91971684e-02 -1.81991935e-01
3.34084958e-01 5.60139060e-01 -4.55742866e-01 5.47487557e-01
1.53748453e-01 1.03528428e+00 -1.63795292e+00 -4.19483274e-01
4.04218435e-01 -2.38635883e-01 -6.60054028e-01 6.94778144e-01
-8.59493732e-01 9.66706872e-01 -7.05354512e-01 6.08990073e-01
-2.35517062e-02 -1.59563795e-01 1.31867215e-01 3.28879267e-01
-3.24554354e-01 3.41018647e-01 -1.02108419e+00 1.81704569e+00
-4.80957448e-01 5.98649979e-01 -6.02744482e-02 -7.95055389e-01
8.20918679e-01 1.50272161e-01 7.50153005e-01 -5.45639873e-01
1.49632201e-01 4.44832504e-01 -1.19548544e-01 -6.33950353e-01
5.77446997e-01 1.72254831e-01 -2.27989569e-01 7.79923022e-01
1.54413171e-02 -3.91201705e-01 2.22743988e-01 2.14111879e-01
1.39278030e+00 7.36536324e-01 -5.53105660e-02 2.20060751e-01
9.53553393e-02 5.12722373e-01 4.68548924e-01 6.84198081e-01
-3.34400892e-01 -7.71524981e-02 4.38833982e-01 -4.77308661e-01
-1.27952790e+00 -1.14280164e+00 6.07851028e-01 7.82326400e-01
3.44895363e-01 -1.58101514e-01 -7.12876379e-01 -1.14872169e+00
1.99527264e-01 6.13820136e-01 -2.21714586e-01 5.98105043e-03
-6.02604628e-01 2.94685125e-01 2.97552884e-01 6.60210967e-01
2.97323704e-01 -1.16324544e+00 -1.16050637e+00 3.28461558e-01
-8.11593309e-02 -1.10776746e+00 -6.95780456e-01 2.40528435e-01
-9.29986954e-01 -1.13311160e+00 -6.48956835e-01 -4.61730570e-01
9.60866630e-01 2.00126261e-01 8.09635580e-01 1.41866505e-01
1.15913607e-03 6.97979808e-01 -2.91403979e-01 -2.89575130e-01
-5.61348557e-01 -9.24173817e-02 1.35991916e-01 -3.82821590e-01
-1.55430958e-01 -2.74548352e-01 -5.57570457e-01 2.70428628e-01
-3.99956465e-01 -9.96446386e-02 4.57719415e-01 7.66335249e-01
8.91901433e-01 -1.79665983e-01 6.47449255e-01 -4.97090459e-01
5.78500807e-01 -2.72955835e-01 -1.29806459e+00 4.69841883e-02
-6.72468007e-01 5.21578789e-01 9.71146822e-01 -5.59286833e-01
-6.14322424e-01 4.81152892e-01 1.35001987e-01 -8.85333240e-01
-8.95357952e-02 2.73894668e-01 1.77148968e-01 -4.34035435e-02
5.91330469e-01 -2.89735459e-02 5.83705902e-02 -2.61287302e-01
7.21694410e-01 4.56041157e-01 5.66524625e-01 -1.09103739e+00
6.88086092e-01 2.33341157e-01 1.57021254e-01 -2.28115588e-01
-9.49917674e-01 -4.39828008e-01 -7.19026387e-01 -6.70588493e-01
7.90511250e-01 -6.87806368e-01 -1.25212932e+00 6.35064468e-02
-8.98624837e-01 -1.34617937e+00 -8.37527692e-01 7.30354488e-01
-1.17750561e+00 9.48327258e-02 -3.14493775e-01 -8.89531434e-01
1.32896170e-01 -1.47349763e+00 9.76173818e-01 2.25986987e-01
-4.56990786e-02 -7.13150859e-01 7.91717321e-02 2.06573412e-01
-2.96508502e-02 1.59766480e-01 6.00816309e-01 -2.97975242e-01
-1.18989646e+00 1.34491054e-02 -1.64458212e-02 3.71587247e-01
-1.00737929e-01 -1.59825105e-02 -6.56289697e-01 -3.62080187e-01
-6.01408899e-01 -9.19405878e-01 2.65034795e-01 1.60097361e-01
1.28402328e+00 -4.34646070e-01 -1.02402598e-01 2.62403369e-01
1.25662112e+00 2.32236266e-01 2.96358585e-01 4.05729294e-01
5.46034873e-01 4.75376815e-01 1.44429326e+00 4.74182397e-01
5.63867629e-01 5.41912556e-01 6.88382685e-01 2.33143926e-01
2.47314405e-02 -8.13949525e-01 7.10900545e-01 5.35931885e-01
-1.08332179e-01 1.70028567e-01 -6.09206915e-01 5.91151953e-01
-2.25898862e+00 -1.00919580e+00 1.55834690e-01 2.28530598e+00
8.33961070e-01 1.33350166e-02 4.61458534e-01 -1.37590289e-01
3.19629073e-01 -1.95762813e-01 -1.12944543e+00 -4.04607922e-01
5.18235564e-01 7.59208994e-03 8.21738660e-01 4.98220444e-01
-7.62435377e-01 1.07544982e+00 6.55447340e+00 4.94140118e-01
-1.07313883e+00 -6.28404692e-02 3.23927850e-01 -1.51366636e-01
-1.57840312e-01 2.58249670e-01 -5.87971210e-01 4.68562901e-01
1.05916405e+00 -2.83251017e-01 1.14290047e+00 1.23585117e+00
6.09582424e-01 -3.45861584e-01 -1.58537686e+00 8.07008505e-01
-6.79865777e-01 -1.37312436e+00 -5.52050173e-01 1.14181422e-01
8.65470052e-01 2.75779128e-01 -2.79489886e-02 6.20190322e-01
1.18631816e+00 -9.13247943e-01 8.85734499e-01 4.36390698e-01
6.84757650e-01 -8.65241170e-01 3.16974133e-01 7.84188986e-01
-1.02282798e+00 -2.92967230e-01 -3.55415702e-01 -9.68993157e-02
1.55499756e-01 5.92746399e-02 -1.22851360e+00 2.50623375e-01
4.22600925e-01 9.23203766e-01 8.65684673e-02 1.04289842e+00
-6.78285778e-01 4.33996737e-01 -3.35827947e-01 -4.43971753e-01
6.82433128e-01 -3.31172407e-01 5.02780616e-01 6.06626332e-01
3.39595497e-01 -1.16219819e-01 7.65331268e-01 8.76994491e-01
-1.72385678e-01 -1.96800932e-01 -7.16867089e-01 -3.22510540e-01
5.00538528e-01 1.25495815e+00 -5.50780773e-01 -3.96699727e-01
4.61154105e-03 9.81429756e-01 6.59491837e-01 2.74519384e-01
-9.21344638e-01 -5.46669103e-02 7.51489162e-01 -1.91816941e-01
1.84825405e-01 -5.55605054e-01 1.73257917e-01 -8.70340765e-01
-9.53230858e-02 -7.60211706e-01 1.08608797e-01 -9.51092601e-01
-8.10889602e-01 8.46638083e-02 -6.45926744e-02 -1.63691521e+00
-5.74579418e-01 -5.04589438e-01 -3.85764599e-01 5.13681471e-01
-1.83435655e+00 -7.20682442e-01 -2.69316494e-01 5.91623843e-01
5.11931121e-01 -4.95744646e-02 6.13461137e-01 1.06609099e-01
-3.22151929e-01 2.61656731e-01 7.58016258e-02 2.94995718e-02
4.55824792e-01 -1.68610561e+00 2.45138496e-01 5.86613894e-01
1.76068902e-01 8.65382627e-02 6.38674974e-01 -6.17948890e-01
-1.73327768e+00 -1.30006993e+00 2.59032190e-01 -3.38282436e-01
9.12409127e-01 8.03052932e-02 -4.14437354e-01 7.79478431e-01
-4.73991819e-02 2.90603429e-01 1.41680732e-01 -3.36607605e-01
-1.57284334e-01 -7.27262050e-02 -1.14962780e+00 7.69907296e-01
9.43009853e-01 -4.19632167e-01 -2.79416829e-01 4.28945333e-01
9.30250823e-01 -8.28303933e-01 -9.39218938e-01 -1.11945361e-01
3.91528368e-01 -5.86606801e-01 7.24074602e-01 -7.44761765e-01
3.59003872e-01 -5.62630355e-01 1.42670958e-03 -1.60218251e+00
3.32119972e-01 -9.65403736e-01 -2.47188747e-01 6.35889053e-01
4.61351693e-01 -4.42016006e-01 8.55028331e-01 6.25530660e-01
-3.15582365e-01 -8.88958454e-01 -7.32306242e-01 -1.27965879e+00
-1.43159881e-01 -4.98820275e-01 8.86231899e-01 6.32383585e-01
1.07792154e-01 9.04072542e-03 -4.00918037e-01 1.67989194e-01
7.19324112e-01 3.01534355e-01 1.22805941e+00 -8.29437733e-01
-4.46495205e-01 -1.66682646e-01 -9.41282585e-02 -1.54238224e+00
5.02159536e-01 -1.01757276e+00 5.93700171e-01 -1.65782905e+00
-2.51278639e-01 -9.85267103e-01 -1.39216796e-01 6.12600327e-01
2.59672672e-01 -4.85154837e-01 3.41965467e-01 2.80047506e-01
-1.00221634e+00 7.13156700e-01 1.80440366e+00 -5.27359024e-02
-3.33339125e-01 2.59422332e-01 -1.06767975e-01 5.89919388e-01
9.40083921e-01 -7.23474562e-01 -6.85959876e-01 -3.91280919e-01
2.69641224e-02 4.79579240e-01 4.84414369e-01 -9.93616998e-01
2.45674804e-01 -7.01969326e-01 -4.37546745e-02 -6.43027127e-01
4.50307608e-01 -1.01796842e+00 -1.58190802e-01 6.25578582e-01
-7.30521023e-01 1.97613522e-01 -2.10463375e-01 7.78473139e-01
1.45309329e-01 -5.46333730e-01 5.90728819e-01 -1.47864848e-01
-8.03309500e-01 5.44650495e-01 -4.20539737e-01 1.65890187e-01
1.28840768e+00 8.20084661e-02 -2.33060107e-01 -4.25031602e-01
-5.45633793e-01 7.89530575e-01 6.95093215e-01 1.93177760e-01
6.99215293e-01 -1.37965775e+00 -2.32341394e-01 -3.25817525e-01
-4.62921783e-02 2.34027565e-01 -1.91769257e-01 6.11371994e-01
-3.45371038e-01 1.42231196e-01 -1.74949899e-01 -5.65273225e-01
-9.22758818e-01 5.80777586e-01 3.60523254e-01 -3.89068455e-01
-8.95770133e-01 3.68139267e-01 -5.19550979e-01 -6.05706155e-01
4.96752083e-01 -7.23985970e-01 1.22800417e-01 -4.30185169e-01
2.90059805e-01 5.20982742e-01 -3.97212714e-01 -1.59050912e-01
1.23466283e-01 1.69453666e-01 3.58528852e-01 -5.37114859e-01
1.28297901e+00 1.15393065e-01 4.11290646e-01 5.33077776e-01
1.02693903e+00 -3.71334553e-01 -2.23788571e+00 -7.23702461e-02
2.84436226e-01 -6.51250124e-01 -6.24015406e-02 -7.37255752e-01
-9.21858549e-01 7.52988219e-01 3.11098963e-01 -9.56089571e-02
5.46513855e-01 -1.16200186e-01 1.10945857e+00 9.11934733e-01
8.88640344e-01 -1.54647589e+00 7.07924426e-01 4.56722856e-01
6.10399902e-01 -1.31647944e+00 -2.06023037e-01 1.79394647e-01
-1.01353443e+00 1.00115132e+00 6.69034839e-01 -3.63264531e-01
2.58039504e-01 2.15155110e-01 -5.13215400e-02 2.20784284e-02
-7.96730578e-01 -4.66808468e-01 -5.23694120e-02 7.80931294e-01
-4.10726607e-01 2.19450757e-01 3.34132671e-01 2.37759277e-02
-1.36515334e-01 3.93481284e-01 6.37401879e-01 1.07751703e+00
-6.66802287e-01 -1.35028780e+00 -8.35739542e-03 3.45545679e-01
-1.36386827e-02 4.98094529e-01 -1.23109266e-01 5.48123479e-01
-1.28013343e-01 8.42687130e-01 -1.01736531e-01 -3.08702528e-01
2.98219919e-01 -2.63312042e-01 7.28504658e-01 -6.98666394e-01
-5.21495879e-01 -1.97305188e-01 1.58830494e-01 -9.85590994e-01
-3.97696078e-01 -8.14157486e-01 -1.95568800e+00 -3.23241204e-01
6.41017919e-03 1.47365779e-01 7.76893318e-01 1.02938759e+00
2.08234861e-01 3.46522480e-01 9.63077962e-01 -8.85628283e-01
-1.10756910e+00 -3.58667135e-01 -3.45412880e-01 2.93089569e-01
6.10557854e-01 -8.34574282e-01 -1.47049233e-01 1.41523769e-02] | [4.237392902374268, 1.6066488027572632] |
5ed66b8f-21c7-44da-9b43-880cf02e5d51 | diffusion-hyperfeatures-searching-through | 2305.14334 | null | https://arxiv.org/abs/2305.14334v1 | https://arxiv.org/pdf/2305.14334v1.pdf | Diffusion Hyperfeatures: Searching Through Time and Space for Semantic Correspondence | Diffusion models have been shown to be capable of generating high-quality images, suggesting that they could contain meaningful internal representations. Unfortunately, the feature maps that encode a diffusion model's internal information are spread not only over layers of the network, but also over diffusion timesteps, making it challenging to extract useful descriptors. We propose Diffusion Hyperfeatures, a framework for consolidating multi-scale and multi-timestep feature maps into per-pixel feature descriptors that can be used for downstream tasks. These descriptors can be extracted for both synthetic and real images using the generation and inversion processes. We evaluate the utility of our Diffusion Hyperfeatures on the task of semantic keypoint correspondence: our method achieves superior performance on the SPair-71k real image benchmark. We also demonstrate that our method is flexible and transferable: our feature aggregation network trained on the inversion features of real image pairs can be used on the generation features of synthetic image pairs with unseen objects and compositions. Our code is available at \url{https://diffusion-hyperfeatures.github.io}. | ['Trevor Darrell', 'Aleksander Holynski', 'Dong Huk Park', 'Lisa Dunlap', 'Grace Luo'] | 2023-05-23 | null | null | null | null | ['semantic-correspondence'] | ['computer-vision'] | [ 1.18474394e-01 8.42362922e-03 5.38971461e-02 -4.00485456e-01
-1.01080585e+00 -5.71515560e-01 1.15909493e+00 -3.70571278e-02
-2.66595274e-01 4.51456070e-01 4.51125264e-01 2.18035847e-01
-2.22048119e-01 -1.03328443e+00 -5.82845509e-01 -8.74056220e-01
-1.13250479e-01 4.07808095e-01 3.44307125e-01 -2.89543808e-01
2.65682101e-01 6.60639763e-01 -1.67747664e+00 6.66963041e-01
4.13481295e-01 1.30298150e+00 2.82563210e-01 7.27973282e-01
-4.43539256e-03 9.25905764e-01 -6.52051508e-01 -1.03641100e-01
4.78699833e-01 -4.32733178e-01 -9.46707904e-01 -1.89253837e-02
6.02960169e-01 -5.49320877e-01 -7.37097859e-01 8.98993254e-01
6.42174006e-01 7.03877211e-02 1.04652286e+00 -1.31191814e+00
-1.13843453e+00 3.57473135e-01 -5.12627661e-01 5.98812580e-01
2.48365253e-01 4.95853007e-01 1.08699906e+00 -9.62209642e-01
1.19001722e+00 1.20261478e+00 5.16274989e-01 5.45205832e-01
-1.19055629e+00 -7.16593862e-01 -3.83943826e-01 1.58603176e-01
-1.13940823e+00 -6.17790461e-01 5.51811934e-01 -6.35055721e-01
1.17676985e+00 -8.44812952e-03 7.28226483e-01 1.24829865e+00
2.47537747e-01 9.55260277e-01 1.06069922e+00 3.18253636e-02
3.96713242e-02 -8.19810927e-02 -9.54235047e-02 8.97628427e-01
1.50619103e-02 4.19711620e-01 -7.60556698e-01 -2.32464224e-01
1.10334909e+00 -2.64366437e-02 -2.02874243e-01 -2.14698538e-01
-1.66228485e+00 1.06835127e+00 8.92938912e-01 2.32977107e-01
-5.40453196e-01 5.46188891e-01 1.34264961e-01 6.07697964e-01
6.72726631e-01 6.15900159e-01 -2.68807054e-01 -1.99541956e-01
-5.42188406e-01 4.17280108e-01 5.08932292e-01 9.00195062e-01
9.79418635e-01 -2.30353266e-01 -3.54150116e-01 8.12343419e-01
-6.45000711e-02 6.02852285e-01 6.91391170e-01 -1.35104251e+00
2.62814671e-01 4.83514071e-01 1.75784808e-02 -9.52223480e-01
-2.97570437e-01 -1.11968651e-01 -9.22201097e-01 1.67263597e-01
2.95834601e-01 -8.59210268e-03 -1.02737796e+00 1.61646307e+00
3.85073483e-01 9.73668993e-02 1.15251146e-01 7.07952499e-01
9.71768975e-01 7.32612729e-01 -2.03761712e-01 3.40546668e-01
1.14966488e+00 -1.09300065e+00 -2.42763251e-01 2.69013755e-02
7.91028738e-01 -8.60374629e-01 8.42148602e-01 -1.18313241e-03
-1.15831470e+00 -4.48845357e-01 -7.41015375e-01 -2.44330436e-01
-5.58960557e-01 -2.44484365e-01 7.21755505e-01 -2.66664252e-02
-1.41124678e+00 1.02510309e+00 -6.88480794e-01 -3.49964559e-01
8.09449732e-01 3.11056703e-01 -5.36057293e-01 -2.37612307e-01
-1.10598648e+00 6.83889568e-01 1.71560794e-02 -1.96096957e-01
-1.13737941e+00 -8.44683647e-01 -7.14725256e-01 -1.37281805e-01
-3.04583490e-01 -9.79080260e-01 1.16621137e+00 -8.10552359e-01
-1.08396935e+00 7.10743606e-01 4.96913157e-02 -3.55471134e-01
5.72538018e-01 2.41408736e-01 -1.29258931e-01 5.83511949e-01
5.24123192e-01 1.21385992e+00 1.20795166e+00 -8.26968908e-01
-7.32610166e-01 -3.92001033e-01 -5.38036190e-02 1.62421405e-01
-4.59315181e-01 -9.87651646e-02 -2.88359165e-01 -8.06355357e-01
1.64495986e-02 -9.96769667e-01 -2.39526212e-01 3.98513526e-01
-2.96058029e-01 -2.55713880e-01 9.19289708e-01 -3.43596399e-01
6.36917055e-01 -2.06679988e+00 1.56234547e-01 1.88734949e-01
4.44769561e-01 -5.88178970e-02 -6.10087872e-01 3.49649340e-01
1.43529728e-01 -2.30901614e-02 -3.24687183e-01 -1.38401046e-01
-1.75046727e-01 4.87740268e-04 -2.39457175e-01 4.34832305e-01
4.49513435e-01 1.37347162e+00 -1.00865853e+00 -2.90433258e-01
5.82037270e-02 7.57724643e-01 -3.81185561e-01 1.74658015e-01
-1.75832018e-01 4.60503101e-01 -6.52005613e-01 4.81718838e-01
3.42505962e-01 -7.67165244e-01 -4.34295923e-01 -1.99956745e-01
3.15938145e-01 4.02562916e-02 -8.13672006e-01 1.82540965e+00
-4.22990680e-01 8.64644527e-01 -2.82190293e-01 -6.01466179e-01
7.85506368e-01 3.72043960e-02 6.19931400e-01 -8.65170419e-01
-1.12245847e-02 2.87503570e-01 -2.02448651e-01 -3.67999583e-01
5.00880778e-01 1.10897236e-01 3.46999131e-02 1.07294703e+00
2.65122116e-01 -5.53486049e-01 2.34005913e-01 4.64957327e-01
1.52222991e+00 -1.18366621e-01 -2.22652867e-01 -3.08404833e-01
8.26994777e-02 1.15309201e-01 2.29216535e-02 8.13736141e-01
6.22873604e-02 9.15080667e-01 2.45321363e-01 -5.45642197e-01
-1.29171443e+00 -1.19124806e+00 -1.98881671e-01 7.84285605e-01
1.98066741e-01 -4.56831217e-01 -5.53524673e-01 -7.40982056e-01
3.06070179e-01 1.30167842e-01 -9.89648402e-01 -2.56142735e-01
-1.78696111e-01 -8.78459275e-01 6.19041324e-01 5.97698271e-01
5.04137158e-01 -1.14125323e+00 -4.91108656e-01 2.52835602e-01
-1.75257951e-01 -9.61114585e-01 -3.86847615e-01 -1.50772640e-02
-7.01811910e-01 -1.04026973e+00 -1.00636804e+00 -6.21039510e-01
7.65881598e-01 5.66522598e-01 1.07073402e+00 1.81674361e-01
-6.66398525e-01 6.24087632e-01 -3.20229977e-01 -6.91774637e-02
-5.53170443e-01 6.66824058e-02 -2.05386877e-01 -2.46457048e-02
1.82874933e-01 -5.10593832e-01 -1.03092706e+00 3.27266723e-01
-1.07317102e+00 5.94110973e-02 4.59126592e-01 9.84114110e-01
6.09643102e-01 -7.66906366e-02 5.25066078e-01 -8.91808867e-01
8.78546894e-01 -5.96488774e-01 -3.43736380e-01 4.08150367e-02
-5.48673689e-01 3.32784712e-01 2.35257030e-01 -3.56707096e-01
-9.25931811e-01 -1.67132951e-02 -1.77424967e-01 -3.94519269e-01
3.41620371e-02 8.27969089e-02 5.19174218e-01 -4.30049717e-01
9.38097119e-01 2.42571890e-01 2.14820132e-01 -2.88254619e-01
7.44270623e-01 5.91884434e-01 2.24836752e-01 -5.90449512e-01
7.81018734e-01 9.56593752e-01 2.07830016e-02 -9.57122684e-01
-7.27296412e-01 -4.19743210e-01 -5.48831344e-01 1.40095232e-02
7.83405840e-01 -1.03564119e+00 -2.58308053e-01 7.96809316e-01
-1.09032679e+00 -5.82850158e-01 -7.24694073e-01 1.56702310e-01
-9.05812263e-01 1.80312783e-01 -1.01651800e+00 -5.15225017e-03
-4.23149019e-01 -1.28160632e+00 1.47396839e+00 5.46453036e-02
-2.34584227e-01 -1.12503862e+00 1.87413469e-02 1.95560262e-01
6.40959740e-01 1.82653725e-01 7.97226846e-01 -3.29311490e-01
-9.30899143e-01 -1.28493100e-01 -5.88434458e-01 3.32125813e-01
1.61719188e-01 -2.66191121e-02 -1.13648820e+00 -3.25569689e-01
-2.76129961e-01 -7.67131567e-01 1.37123394e+00 3.69917005e-01
1.08237672e+00 -2.02436611e-01 -4.35583562e-01 6.96537375e-01
1.16880620e+00 -4.21942919e-01 5.61278522e-01 2.37759307e-01
6.27710879e-01 6.90049767e-01 2.81116784e-01 3.23198885e-01
4.56405461e-01 5.28051376e-01 1.30766422e-01 -2.53265977e-01
-7.15832233e-01 -1.86419189e-01 1.55029282e-01 6.80386662e-01
2.78598983e-02 -1.69684514e-01 -9.15035069e-01 7.05214441e-01
-1.69734359e+00 -1.04372096e+00 1.92974117e-02 1.66164601e+00
7.61116862e-01 -9.46217850e-02 -7.72506297e-02 -2.68311232e-01
4.50206339e-01 3.73664469e-01 -6.22401297e-01 -2.32879678e-03
-4.27872241e-01 8.26339722e-02 4.34528291e-01 4.19966072e-01
-9.42880034e-01 9.26864088e-01 6.65702963e+00 9.04403865e-01
-1.11334515e+00 3.40156078e-01 8.55622709e-01 -1.74609423e-01
-5.87755740e-01 -2.10377783e-01 -6.59082234e-01 3.36405873e-01
8.54067683e-01 -3.14697802e-01 3.84578198e-01 6.25140846e-01
-1.85111836e-01 -2.47868132e-02 -1.13056552e+00 9.97918069e-01
-2.07474716e-02 -1.91781688e+00 3.96692723e-01 2.65832812e-01
1.22103870e+00 7.38530755e-01 4.79903758e-01 -1.33221760e-01
7.10383594e-01 -9.81044590e-01 5.37760258e-01 6.11880720e-01
9.37122047e-01 -3.27099264e-01 3.32447857e-01 9.05288905e-02
-9.38947141e-01 -7.41970912e-02 -5.81310987e-01 2.48880878e-01
3.30141559e-02 6.31480932e-01 -8.35245371e-01 1.95031568e-01
7.58881271e-01 1.17027581e+00 -7.81694412e-01 8.78638685e-01
-1.18561529e-01 7.14704394e-02 -4.13412780e-01 1.23956740e-01
4.46237713e-01 2.34408360e-02 2.94161469e-01 1.12271392e+00
5.79292834e-01 -1.48032280e-02 -2.00745910e-01 1.08076453e+00
-2.17963457e-01 -7.69493058e-02 -1.13397837e+00 -2.76112586e-01
2.18397096e-01 1.32716513e+00 -7.50131130e-01 -5.52692235e-01
-3.44516218e-01 1.19928360e+00 5.90807438e-01 4.58827257e-01
-4.49378729e-01 -3.06005627e-01 9.00142908e-01 1.55902669e-01
3.61670673e-01 -1.91932261e-01 2.18446568e-01 -1.25266171e+00
-2.55136102e-01 -6.33939207e-01 2.61008769e-01 -1.04123402e+00
-1.73553300e+00 9.19956386e-01 -1.09308667e-01 -1.03511345e+00
-4.99894083e-01 -6.07083976e-01 -5.05247056e-01 8.12411070e-01
-1.59002030e+00 -1.17113328e+00 -5.40735006e-01 7.12330222e-01
5.32414615e-01 -1.71672702e-01 7.02607691e-01 8.85463357e-02
-6.92496225e-02 1.61394611e-01 3.82118881e-01 7.91314319e-02
6.16853774e-01 -1.09630251e+00 9.04890418e-01 3.55925500e-01
4.18932319e-01 4.29967791e-01 2.55982995e-01 -5.24757564e-01
-1.23413110e+00 -1.27008140e+00 5.86493611e-01 -8.00870895e-01
8.71474743e-01 -2.71565497e-01 -8.09826434e-01 5.55967748e-01
1.28379747e-01 3.69172573e-01 3.62983614e-01 -3.29575807e-01
-5.11238098e-01 4.74572107e-02 -9.74062502e-01 4.38968539e-01
1.38668430e+00 -8.18269253e-01 -2.29010060e-01 7.14834571e-01
7.36501575e-01 -2.22428903e-01 -1.19171906e+00 1.58539832e-01
4.60544258e-01 -1.07387829e+00 1.21667647e+00 -4.43198860e-01
7.48656332e-01 9.35157388e-02 -1.77552961e-02 -1.50228179e+00
-5.18282235e-01 -5.44792950e-01 -7.51064122e-02 9.06780958e-01
6.53449476e-01 -7.36386001e-01 5.31379402e-01 4.36111450e-01
3.13227803e-01 -8.28691483e-01 -8.04622352e-01 -8.11739802e-01
2.54295945e-01 -2.35766098e-01 8.48613381e-01 8.66949677e-01
-2.82395631e-01 3.80642116e-01 2.41793692e-02 -2.97681183e-01
7.03850985e-01 2.65584558e-01 6.47961318e-01 -1.18898451e+00
-2.02803344e-01 -5.77226579e-01 -5.37225127e-01 -1.07227063e+00
2.40809679e-01 -1.23455477e+00 -2.18574509e-01 -1.67976248e+00
1.68428019e-01 -8.39903295e-01 -2.85904318e-01 3.19979757e-01
1.50182277e-01 7.18358219e-01 5.31141758e-02 6.35671914e-01
-4.42644894e-01 6.66944563e-01 1.75714445e+00 -4.90049273e-01
1.24771021e-01 -4.36730057e-01 -4.82613534e-01 4.20674384e-01
8.00478935e-01 -5.36561966e-01 -6.01322055e-01 -8.04654956e-01
2.33099192e-01 -9.09682289e-02 5.86491406e-01 -1.00374329e+00
3.03255856e-01 -5.23773767e-03 5.84877670e-01 -2.49043077e-01
5.64407468e-01 -4.16836619e-01 9.37249884e-02 2.44873583e-01
-4.25971508e-01 2.47529104e-01 -7.22063705e-02 7.38127708e-01
-2.40743443e-01 -1.11036628e-01 7.26026058e-01 -4.19577420e-01
-7.96269715e-01 7.60556340e-01 -6.56100810e-02 1.86097085e-01
1.03434098e+00 -1.49184331e-01 -7.07263947e-01 -5.36363006e-01
-5.03106236e-01 -6.58932254e-02 7.90953338e-01 6.49734318e-01
8.35498095e-01 -1.53134513e+00 -7.07075715e-01 4.99013036e-01
4.06026721e-01 -9.83564835e-03 2.51207590e-01 6.97427869e-01
-5.55468202e-01 1.66150212e-01 -3.15340579e-01 -7.02063322e-01
-7.93024659e-01 4.55658704e-01 3.34815800e-01 7.91950300e-02
-9.37994361e-01 1.12961626e+00 5.77400625e-01 -3.07474941e-01
-1.84198424e-01 -2.61664122e-01 1.66435108e-01 1.37086675e-01
6.77083790e-01 1.43699259e-01 -8.15183297e-02 -8.48112285e-01
1.34707363e-02 8.29954863e-01 -2.61001348e-01 -2.89289206e-01
1.57265568e+00 -1.48985475e-01 -1.33149162e-01 9.90067720e-02
1.73263490e+00 -3.78195763e-01 -1.41045105e+00 -4.41539913e-01
-2.06023857e-01 -5.64532101e-01 6.90438375e-02 -5.36217809e-01
-1.20773935e+00 9.79271591e-01 5.32437086e-01 2.27123410e-01
8.18018138e-01 3.56155843e-01 1.08067751e+00 4.24914986e-01
2.93098748e-01 -9.30815458e-01 5.52733481e-01 2.77484298e-01
1.11009598e+00 -1.29491043e+00 -2.66524553e-01 -1.41550794e-01
-5.79675674e-01 1.07268631e+00 3.38939637e-01 -2.53184199e-01
8.41926515e-01 1.69235453e-01 2.03198567e-01 -6.66228831e-01
-1.07216632e+00 -1.15698792e-01 2.59335965e-01 7.66989231e-01
1.02792986e-01 -1.63122475e-01 2.70918339e-01 -1.25466734e-01
-1.69362456e-01 8.38104403e-04 5.36596894e-01 8.25145364e-01
-3.94342452e-01 -9.69107926e-01 9.96955931e-02 7.89504647e-01
1.50887901e-02 -2.25164399e-01 -4.14256752e-01 4.82772976e-01
-7.52543435e-02 6.33941293e-01 2.61661083e-01 -3.81142795e-01
2.40124837e-02 -3.04534853e-01 6.30489647e-01 -5.70385575e-01
-4.45069730e-01 -3.31893533e-01 3.52043211e-02 -9.02783751e-01
-4.46643561e-01 -5.53410947e-01 -1.12978530e+00 -3.95309448e-01
-5.92887811e-02 -2.29328386e-02 6.00332856e-01 5.58300793e-01
7.30154216e-01 2.42412314e-01 6.66074157e-01 -9.41092849e-01
-6.01208627e-01 -8.82774174e-01 -5.69224119e-01 8.78148198e-01
4.86615717e-01 -7.08595335e-01 -3.98461848e-01 6.15111813e-02] | [11.026674270629883, 0.06942462176084518] |
08ba64a0-a623-46a8-ae1a-e8ee1e278ed9 | improving-3d-aware-image-synthesis-with-a | 2209.15637 | null | https://arxiv.org/abs/2209.15637v1 | https://arxiv.org/pdf/2209.15637v1.pdf | Improving 3D-aware Image Synthesis with A Geometry-aware Discriminator | 3D-aware image synthesis aims at learning a generative model that can render photo-realistic 2D images while capturing decent underlying 3D shapes. A popular solution is to adopt the generative adversarial network (GAN) and replace the generator with a 3D renderer, where volume rendering with neural radiance field (NeRF) is commonly used. Despite the advancement of synthesis quality, existing methods fail to obtain moderate 3D shapes. We argue that, considering the two-player game in the formulation of GANs, only making the generator 3D-aware is not enough. In other words, displacing the generative mechanism only offers the capability, but not the guarantee, of producing 3D-aware images, because the supervision of the generator primarily comes from the discriminator. To address this issue, we propose GeoD through learning a geometry-aware discriminator to improve 3D-aware GANs. Concretely, besides differentiating real and fake samples from the 2D image space, the discriminator is additionally asked to derive the geometry information from the inputs, which is then applied as the guidance of the generator. Such a simple yet effective design facilitates learning substantially more accurate 3D shapes. Extensive experiments on various generator architectures and training datasets verify the superiority of GeoD over state-of-the-art alternatives. Moreover, our approach is registered as a general framework such that a more capable discriminator (i.e., with a third task of novel view synthesis beyond domain classification and geometry extraction) can further assist the generator with a better multi-view consistency. | ['Dit-yan Yeung', 'Qifeng Chen', 'Deli Zhao', 'Yujun Shen', 'Yinghao Xu', 'Zifan Shi'] | 2022-09-30 | null | null | null | null | ['3d-aware-image-synthesis'] | ['computer-vision'] | [ 2.73471713e-01 4.27729011e-01 1.12689771e-01 -1.12632588e-01
-7.54023433e-01 -8.01470399e-01 7.50361741e-01 -6.15312338e-01
1.11897565e-01 6.57179117e-01 -1.24515638e-01 -2.20309243e-01
3.15861762e-01 -1.23002875e+00 -9.34271276e-01 -9.65368986e-01
4.99440581e-01 4.94633198e-01 -2.77750909e-01 -3.86504263e-01
8.05265233e-02 7.52792120e-01 -1.36844850e+00 7.02939406e-02
1.21183634e+00 1.20358038e+00 3.90352458e-02 4.76184607e-01
3.17583187e-03 5.17258763e-01 -7.59015620e-01 -7.50174642e-01
7.54984915e-01 -7.90508568e-01 -3.66855413e-01 3.07938963e-01
4.40053254e-01 -8.02470148e-01 -1.74163491e-01 9.50541139e-01
5.53710163e-01 -1.33504078e-01 8.71017039e-01 -1.25486374e+00
-8.26985240e-01 -4.12240904e-03 -5.23856699e-01 -5.02054453e-01
3.69929165e-01 4.97967184e-01 7.29733825e-01 -6.94634080e-01
5.45779347e-01 1.23711169e+00 4.11486596e-01 7.39675403e-01
-1.19250453e+00 -6.53573453e-01 4.78031561e-02 -4.36442316e-01
-1.24972534e+00 -1.38119295e-01 1.35258341e+00 -3.76596302e-01
2.71222293e-01 4.56674904e-01 7.65943468e-01 1.49822676e+00
9.57840830e-02 6.79360867e-01 1.27120042e+00 -1.74924538e-01
2.51823813e-01 2.60976195e-01 -7.79163241e-01 6.12185299e-01
1.45579726e-01 4.05793399e-01 -1.47915423e-01 -4.31931838e-02
1.18080282e+00 -1.34424880e-01 -5.01135111e-01 -6.15626812e-01
-8.70033562e-01 8.22530329e-01 5.36811590e-01 -2.03106701e-02
-5.48323929e-01 -5.25283441e-02 5.30873351e-02 1.64533705e-01
5.83089054e-01 6.38097465e-01 3.34243407e-04 4.38880473e-02
-8.59766483e-01 4.07242835e-01 6.73492074e-01 9.74716067e-01
7.56843328e-01 6.03215694e-01 -2.81624615e-01 3.98639500e-01
1.52859747e-01 6.17364109e-01 1.55872792e-01 -9.44059253e-01
4.61183071e-01 8.07773054e-01 1.61481917e-01 -1.06996715e+00
2.17307016e-01 -5.96021056e-01 -1.13132703e+00 8.35614145e-01
5.08799493e-01 -5.44643868e-03 -8.90233636e-01 1.71315253e+00
5.06949365e-01 8.03598985e-02 9.12733078e-02 1.16172814e+00
5.61189294e-01 5.41823447e-01 -2.70537436e-01 2.39940267e-03
1.03428626e+00 -6.61624849e-01 -3.51695031e-01 -1.35505751e-01
2.89398581e-01 -6.75373256e-01 1.14321482e+00 4.67668772e-01
-1.25161183e+00 -6.08726621e-01 -1.17060530e+00 -6.49338737e-02
-2.52457380e-01 1.44799769e-01 4.68201727e-01 8.37632239e-01
-8.28131139e-01 4.22644258e-01 -4.67924267e-01 1.71878695e-01
6.34262562e-01 -5.97051792e-02 -2.93577820e-01 -6.16313256e-02
-1.15402818e+00 7.71753669e-01 1.54999122e-01 8.48967433e-02
-1.08525038e+00 -7.27392137e-01 -8.52361679e-01 -9.08634365e-02
2.90397793e-01 -1.06209898e+00 8.11118126e-01 -1.33704698e+00
-1.75609970e+00 9.37633574e-01 3.09148669e-01 -1.99770525e-01
1.23530281e+00 1.41859174e-01 -5.70322946e-02 2.96114683e-01
-6.30985796e-02 5.71343243e-01 1.37691700e+00 -1.86297286e+00
-2.36060381e-01 -3.36069733e-01 5.16975999e-01 2.95059800e-01
-3.55308168e-02 -7.42539704e-01 -1.89015806e-01 -8.13311040e-01
4.80549447e-02 -8.74793351e-01 2.43580118e-02 2.69175649e-01
-6.27321601e-01 1.13840684e-01 7.04708755e-01 -8.35538208e-01
4.83199358e-01 -2.11589384e+00 2.01311216e-01 1.87399194e-01
3.18432420e-01 2.95649022e-01 -1.27742469e-01 2.44933173e-01
-1.52297691e-01 2.53677309e-01 -2.31352463e-01 -5.52210987e-01
1.30027503e-01 1.30499795e-01 -5.10403931e-01 4.53148425e-01
5.90804815e-01 1.19836080e+00 -9.18141544e-01 -1.34532258e-01
4.34192151e-01 6.72006905e-01 -5.83635271e-01 4.51740563e-01
-3.09087843e-01 1.02175915e+00 -6.44362390e-01 4.95429635e-01
1.06882405e+00 -9.77377146e-02 -5.54038957e-03 -3.07986140e-01
1.43857732e-01 1.09107397e-01 -1.03414965e+00 1.73472095e+00
-6.51790321e-01 1.69114500e-01 3.32862847e-02 -1.01264572e+00
1.22798419e+00 2.86974698e-01 2.91867197e-01 -8.45872641e-01
1.50540411e-01 2.89513320e-01 -2.69906551e-01 -2.46445820e-01
3.72422040e-01 -3.23783219e-01 -5.18504903e-02 3.17138731e-01
-2.10573226e-01 -6.50865436e-01 -5.46866894e-01 8.82733762e-02
6.34717286e-01 5.41067839e-01 1.46981999e-01 7.40206838e-02
5.49662828e-01 -2.71796882e-01 4.15506929e-01 4.97663438e-01
1.85492069e-01 9.82902050e-01 6.03653073e-01 -3.77542585e-01
-1.34843397e+00 -1.18699610e+00 2.01722577e-01 2.93931007e-01
1.70341805e-01 5.32051064e-02 -9.29442525e-01 -9.62682188e-01
-9.73441601e-02 9.18539524e-01 -7.89068162e-01 -3.50661635e-01
-6.43858194e-01 -2.32709661e-01 5.80287814e-01 3.58724654e-01
7.37397492e-01 -7.07405388e-01 -5.80390692e-01 -1.15293689e-01
-1.22331940e-01 -1.03482759e+00 -3.37403864e-01 -1.36923179e-01
-7.13913560e-01 -9.66280878e-01 -9.17469203e-01 -4.29085553e-01
8.42442930e-01 1.62650675e-01 1.11405396e+00 1.32155597e-01
1.21793456e-01 1.21063150e-01 -3.19457680e-01 -2.19895482e-01
-9.22973812e-01 -1.44935936e-01 -2.29068235e-01 1.67176977e-01
-3.20501626e-01 -9.51313436e-01 -8.18936825e-01 2.32757047e-01
-1.00517774e+00 5.16676664e-01 6.38565779e-01 8.55320632e-01
4.81251121e-01 -5.15329826e-04 5.95406532e-01 -8.89183998e-01
2.37519145e-01 -1.68171719e-01 -7.08428741e-01 1.29709423e-01
-5.36563218e-01 -7.54144937e-02 1.20979047e+00 -4.84147847e-01
-1.15612483e+00 -8.12614113e-02 -3.69890362e-01 -7.63649464e-01
-2.69081056e-01 -1.62776306e-01 -7.85493493e-01 -2.88796216e-01
6.50817871e-01 6.41980946e-01 1.06947623e-01 -2.59279072e-01
5.63279867e-01 3.60730052e-01 3.69232595e-01 -6.33443236e-01
1.30757391e+00 6.88733697e-01 2.11586475e-01 -4.88904625e-01
-4.63698685e-01 2.86123663e-01 -4.31812257e-01 -2.42585659e-01
7.99321890e-01 -9.49849546e-01 -6.48079693e-01 6.73660755e-01
-1.24485290e+00 -2.52005875e-01 -5.28593898e-01 9.69139412e-02
-8.00318301e-01 2.62830615e-01 -3.04013640e-01 -9.39075053e-01
-2.45522380e-01 -1.30512369e+00 1.38589990e+00 8.51793364e-02
1.56231999e-01 -8.99095774e-01 -2.86609799e-01 5.55705488e-01
2.79534191e-01 8.96599174e-01 1.09442735e+00 -2.33426630e-01
-9.41819549e-01 -2.28921965e-01 -2.21770763e-01 7.53664732e-01
3.71114820e-01 -2.12039620e-01 -1.29654813e+00 -2.63243437e-01
4.16589350e-01 -2.91931838e-01 4.52505738e-01 7.54684508e-02
1.24437392e+00 -5.52188575e-01 9.50425342e-02 9.61799860e-01
1.47110724e+00 2.24023178e-01 9.76374447e-01 -7.73917735e-02
1.09403658e+00 5.99383771e-01 3.35738093e-01 3.62932265e-01
4.52211201e-01 7.19254792e-01 9.28481698e-01 -2.79814869e-01
-3.95135850e-01 -7.10234344e-01 1.92831099e-01 4.32885528e-01
-3.07805747e-01 -5.19808412e-01 -4.84738618e-01 4.44776155e-02
-1.35405910e+00 -8.13377619e-01 9.58713219e-02 2.20735550e+00
8.08943152e-01 -7.18123047e-03 5.23493849e-02 1.74408898e-01
5.25734723e-01 2.51068503e-01 -7.49846160e-01 -2.97269046e-01
-2.70177394e-01 1.52736202e-01 2.13893071e-01 2.84980327e-01
-6.87922537e-01 7.34227657e-01 4.74646854e+00 1.11960685e+00
-1.42590940e+00 -9.76676643e-02 9.68798816e-01 1.35826468e-01
-8.71727943e-01 -9.91162509e-02 -3.23753297e-01 6.39102161e-01
2.13179439e-01 1.26353577e-01 6.13465965e-01 8.17577481e-01
2.65820026e-01 1.66649848e-01 -1.08857906e+00 1.04425359e+00
1.84858620e-01 -1.33547747e+00 4.56062764e-01 4.56458777e-01
8.76232564e-01 -7.36532569e-01 2.09932432e-01 1.45998016e-01
2.38694269e-02 -1.11734915e+00 1.08762264e+00 6.40835285e-01
1.15417504e+00 -8.98891985e-01 4.30937678e-01 6.50125861e-01
-8.23527277e-01 2.84781605e-01 -1.25461683e-01 7.50032589e-02
2.38988176e-01 6.44532740e-01 -6.68930590e-01 1.14982617e+00
3.96525443e-01 4.32198614e-01 -3.96263838e-01 4.82044280e-01
-5.43518066e-01 2.19564259e-01 -9.47762355e-02 3.56310815e-01
1.43797114e-01 -5.21210551e-01 7.65776277e-01 5.56682885e-01
5.03097892e-01 8.24259818e-02 9.14966047e-04 1.40352130e+00
-1.82163119e-01 -1.39439911e-01 -9.20340121e-01 1.27180457e-01
3.51294935e-01 1.19380307e+00 -5.36612391e-01 -1.11458935e-01
-1.76423043e-02 1.28370559e+00 2.40806133e-01 4.53857005e-01
-1.03380597e+00 -4.59290156e-03 5.93854547e-01 3.32656741e-01
3.42156351e-01 -9.24581662e-02 -5.40388048e-01 -1.12445104e+00
2.79537469e-01 -1.15771317e+00 -1.49389759e-01 -9.40503180e-01
-1.49761498e+00 7.19388306e-01 -2.52185285e-01 -1.56709254e+00
-5.56696415e-01 -5.00276864e-01 -5.78848422e-01 1.14409459e+00
-1.63721681e+00 -1.62805963e+00 -4.90615845e-01 5.94466805e-01
2.67830521e-01 1.59957185e-02 7.27748036e-01 1.35327086e-01
-1.86623439e-01 8.13477695e-01 -3.31359297e-01 1.38987422e-01
4.91903454e-01 -1.23577762e+00 4.24798131e-01 8.57781589e-01
-1.96012575e-02 3.27907354e-01 5.05173147e-01 -5.78434646e-01
-1.58518791e+00 -1.21662438e+00 1.91588283e-01 -5.40505886e-01
1.38462946e-01 -5.49224496e-01 -8.40627611e-01 2.54251659e-01
7.34110847e-02 -4.87280451e-02 2.22693518e-01 -6.72080278e-01
-3.65770966e-01 -7.35801756e-02 -1.47211933e+00 6.36485279e-01
1.20684576e+00 -5.75375199e-01 -2.23222673e-01 -7.24195167e-02
6.81148410e-01 -7.02575028e-01 -7.08962381e-01 4.08156008e-01
4.22847599e-01 -1.25985658e+00 1.15494561e+00 -2.74464428e-01
7.53736615e-01 -3.89291406e-01 -1.49203137e-01 -1.53244328e+00
9.06999782e-02 -6.90993786e-01 -2.16507047e-01 1.42178667e+00
-8.02756399e-02 -8.05118382e-01 8.85581851e-01 4.52089220e-01
-1.72132492e-01 -8.69066596e-01 -9.25740600e-01 -9.13932323e-01
4.25727189e-01 -3.78811210e-01 1.01103699e+00 9.69561458e-01
-8.62388134e-01 9.95152965e-02 -6.17280483e-01 1.82841882e-01
6.32856488e-01 4.39523548e-01 1.04875672e+00 -9.32275832e-01
-4.69576091e-01 -4.16625887e-01 -3.73178750e-01 -1.28095269e+00
1.71471640e-01 -7.82314777e-01 -7.53174201e-02 -1.24466968e+00
-2.88737416e-01 -7.10187376e-01 3.39231819e-01 1.81923911e-01
-1.08974926e-01 3.69000614e-01 3.49226028e-01 7.96897337e-02
9.19979811e-02 8.80556285e-01 1.97192729e+00 -1.77362606e-01
3.66919599e-02 1.51051521e-01 -9.09254134e-01 5.73303103e-01
7.09740877e-01 -2.38624662e-01 -6.64784610e-01 -3.93982977e-01
3.02893192e-01 1.14746787e-01 1.00780654e+00 -6.72728360e-01
-1.86860234e-01 -5.55783324e-02 5.91546595e-01 -3.65170509e-01
5.50362587e-01 -9.22457218e-01 3.16816002e-01 1.82015702e-01
6.66832030e-02 -2.61669755e-01 -6.41352385e-02 4.42538857e-01
-6.40286431e-02 -5.13221473e-02 8.05489957e-01 -2.30984390e-01
-2.75042564e-01 4.84489322e-01 2.65249819e-01 1.73691675e-01
8.94085944e-01 -4.81644481e-01 -2.97925502e-01 -6.25817060e-01
-2.74295777e-01 -2.02794179e-01 1.05588114e+00 3.00582469e-01
6.16890073e-01 -1.57420003e+00 -5.90086818e-01 5.45686603e-01
5.98699413e-02 5.14572382e-01 4.28688645e-01 4.56416637e-01
-3.96059513e-01 -3.46007347e-02 -1.71924397e-01 -5.84454954e-01
-5.25038242e-01 5.92326522e-01 4.91540045e-01 -2.30718493e-01
-6.03874862e-01 5.84911585e-01 6.93314433e-01 -4.55094665e-01
-1.91164706e-02 -1.99408606e-01 2.12149620e-01 -9.79504734e-02
2.87333995e-01 9.31616127e-02 8.64193812e-02 -4.51060623e-01
-1.45858735e-01 4.96300191e-01 4.36077386e-01 5.58837913e-02
1.19457126e+00 -5.52667491e-02 1.35773107e-01 1.34121165e-01
1.12780976e+00 1.93621591e-01 -1.79788864e+00 2.36536816e-01
-7.75205851e-01 -6.84257329e-01 -7.11285844e-02 -8.20688009e-01
-1.39753771e+00 9.07772601e-01 3.80559027e-01 2.66662419e-01
1.31230879e+00 -2.10564926e-01 8.70389819e-01 -2.16736317e-01
5.81017196e-01 -5.53154528e-01 2.29888514e-01 3.55544835e-02
1.12416029e+00 -1.21994328e+00 -2.14385509e-01 -5.13904274e-01
-6.55813754e-01 9.78214443e-01 6.63378239e-01 -3.63449246e-01
2.20796525e-01 1.20987348e-01 4.19102311e-02 -1.45432875e-01
-1.62958324e-01 1.07388496e-01 4.85959530e-01 8.00411284e-01
2.65592318e-02 7.61357397e-02 1.58391595e-01 5.84180474e-01
-4.44800675e-01 -2.53055096e-01 3.55471283e-01 4.62563634e-01
2.62249887e-01 -1.13897347e+00 -2.70958781e-01 6.16104342e-02
-6.20287098e-02 -1.55066401e-02 -3.53414625e-01 8.97730052e-01
4.79273766e-01 6.92188859e-01 -1.12260021e-01 -3.81236434e-01
3.89612645e-01 -1.30886838e-01 5.87930024e-01 -4.02637631e-01
-5.38286090e-01 -6.40941188e-02 -2.09731057e-01 -6.79588377e-01
-2.11865172e-01 -2.51959443e-01 -8.16225529e-01 -4.54754382e-01
-7.77123645e-02 -1.37157828e-01 5.94208181e-01 7.56061256e-01
3.49188983e-01 5.19370079e-01 1.11152554e+00 -1.00905478e+00
-7.77527213e-01 -5.39545715e-01 -5.38835704e-01 5.80419064e-01
3.65813345e-01 -6.48771942e-01 -5.84894478e-01 -1.19569667e-01] | [9.297171592712402, -3.24824857711792] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.