AutomaticReparameterisationofProbabilisticProgramsMariaI.Gorinova1DaveMoore2MatthewD.Hoffman2Abstractparameterisation(CP).Ifweinsteadworkwithanauxil-iary,standardnormalvariablez˜∼N(0,1),andobtain...
ApproximationGuaranteesofLocalSearchAlgorithmsviaLocalizabilityofSetFunctionsKaitofujii1Abstractproblemoffindingan(approximately)optimalsetfromallfeasiblesets.VariousmachinelearningtaskshavebeenThi...
ApproximationCapabilitiesofNeuralODEsandInvertibleResidualNetworksHanZhang1XiGao1JacobUnterman1TomArodz1Abstractsequence.Then,wecanrepresentthesequencethroughxt+1−xt=fΘ(xt,t),whereΘconsistsoftra...
AndersonAccelerationofProximalGradientMethodsVienV.Mai1MikaelJohansson1Abstractrameters;slightlyover-orunder-estimatingthestrongcon-vexityconstantcanhaveasevereeffectontheoverallper-Andersonacceler...
TheImpactofNeuralNetworkOverparameterizationonGradientConfusionandStochasticGradientDescentKarthikA.Sankararaman12SohamDe3ZhengXu2W.RonnyHuang2TomGoldstein2AbstractClassicalstochasticoptimizationth...
AggregationofMultipleKnockoffsTuan-BinhNguyen12Je´roˆme-AlexisChevalier2BertrandThirion2SylvainArlot1Abstractornot;ii)itrequiresagoodgenerativemodelforfeatures,butposesfewconditionsforthevalidity...
AdversarialRobustnessAgainsttheUnionofMultiplePerturbationModelsPratyushMaini1EricWong2J.ZicoKolter34Abstracttheexistenceofdatapointswhichcanbeadversariallyper-turbedtobemisclassified,butare“close...
AdversarialFiltersofDatasetBiasesRonanLeBras1SwabhaSwayamdipta1ChandraBhagavatula1RowanZellers12MatthewE.Peters1AshishSabharwal1YejinChoi12Abstractwild”(Eykholtetal.,2018;Jia&Liang,2017).Thisphe-n...
AUnifiedTheoryofDecentralizedSGDwithChangingTopologyandLocalUpdatesAnastasiaKoloskova1NicolasLoizou2SadraBoreiri1MartinJaggi1SebastianU.Stich1Abstractetal.,2016;2017;Kairouzetal.,2019)hasemerged,bu...
ASimpleFrameworkforContrastiveLearningofVisualRepresentationsTingChen1SimonKornblith1MohammadNorouzi1GeoffreyHinton1AbstractFigure1.ImageNetTop-1accuracyoflinearclassifierstrainedonrepresentationsl...
AMean-fieldAnalysisofDeepResNetandBeyond:TowardsProvableOptimizationViaOverparameterizationFromDepthYipingLu1ChaoMa2YulongLu3JianfengLu3LexingYing4Abstract1.IntroductionTrainingdeepneuralnetworkswi...
AFinite-TimeAnalysisofQ-LearningwithNeuralNetworkFunctionApproximationPanXu1QuanquanGu1AbstractwhichtriggersalineofresearchondeepreinforcementlearningsuchasDoubleDeepQ-Learning(VanHasseltQ-learning...
VariationalAnnealingofGANs:ALangevinPerspectiveChenyangTao1ShuyangDai1LiqunChen1KeBai1JunyaChen12ChangLiu13RuiyiZhang1GeorgiyBobashev4LawrenceCarin1Abstractapre-specifiedsimpledistributionq(z)throu...
UnderstandingtheOriginsofBiasinWordEmbeddingsMarc-EtienneBrunet12ColleenAlkalay-Houlihan1AshtonAnderson12RichardZemel12Abstract2013a)andGloVe(Penningtonetal.,2014)acquirestereo-typicalhumanbiasesfr...
UniformConvergenceRateoftheKernelDensityEstimatorAdaptivetoIntrinsicVolumeDimensionJisuKim1JaehyeokShin2AlessandroRinaldo2LarryWasserman2Abstractciallyinrecentyears,hasalsobecomeakeystepinmanygeome...
UnderstandingtheImpactofEntropyonPolicyOptimizationZafaraliAhmed12NicolasLeRoux13MohammadNorouzi3DaleSchuurmans34Abstractlis,2000;Greensmithetal.,2004;Schulmanetal.,2015b;Tuckeretal.,2018).Entropyr...
UnderstandingGeometryofEncoder-DecoderCNNsJongChulYe12WoonKyoungSung2Abstractpower,generalizationcapability,andoptimizationland-scapeofDNNshavebecomeanintellectualchallengeforEncoder-decodernetwork...
UnderstandingImpactsofHigh-OrderLossApproximationsandFeaturesinDeepLearningInterpretationSahilSingla1EricWallace1ShiFeng1SoheilFeizi1Abstracthindmodelpredictions?Acommoninterpretationapproachistoid...
TransferofSamplesinPolicySearchviaMultipleImportanceSamplingAndreaTirinzoni1MattiaSalvini1MarcelloRestelli1Abstractagentissupposedtoreuseknowledgeacquiredfromasetofsourcetaskstoacceleratethelearnin...
TrainableDecodingofSetsofSequencesforNeuralSequenceModelsAshwinKalyan1PeterAnderson1StefanLee1DhruvBatra12Abstractinteractionspresentintheimage(Kalyanetal.,2018).Beingabletoproducemultiplerelevanto...