TrainingCNNswithSelectiveAllocationofChannelsJongheonJeong1JinwooShin123AbstractChenetal.,2018)andanytime/adaptivenetworks(Figurnovetal.,2017;Bolukbasietal.,2017;Huangetal.,2018a).Recentprogressind...
TowardsaUnifiedAnalysisofRandomFourierFeaturesZhuLi1Jean-FrançoisTon1DinoOglic2DinoSejdinovic1Abstractimplicitcomputationofaninnerproductbetweenrichfea-turerepresentationsofdatathroughthekerneleva...
TowardsaDeepandUnifiedUnderstandingofDeepNeuralModelsinNLP1ChaoyuGuan2XitingWang2QuanshiZhang1RunjinChen1DiHe3XingXie2Abstract(a)Gradient-basedmethod(b)OursWedefineaunifiedinformation-basedmeasureF...
TowardsUnderstandingtheImportanceofNoiseinTrainingNeuralNetworksMoZhou1TianyiLiu2YanLi2DachaoLin1EnluZhou2TuoZhao2AbstractSimplefirstorderalgorithmssuchasStochasticGradientDescent(SGD)anditsvariant...
TopologicalDataAnalysisofDecisionBoundarieswithApplicationtoModelSelectionKarthikeyanNatesanRamamurthy1KushR.Varshney1KrishnanMody12Abstractofneuralnetworkdecisionboundaries.Persistenthomologyinvol...
TightKernelQueryComplexityofKernelRidgeRegressionandKernelk-meansClusteringManuelFerna´ndez1DavidP.Woodruff1TaisukeYasuda2Abstractbetweentwodatapointswiththeirinnerproductafterap-plyingakernelmap,...
TheImplicitFairnessCriterionofUnconstrainedLearningLydiaT.Liu1MaxSimchowitz1MoritzHardt1Abstractormoresensitiveattributes,aswellasarelatedcriterioncalledsufficiency(e.g.,Barocasetal.,2018),atthecos...
TheInformation-TheoreticValueofUnlabeledDatainSemi-SupervisedLearningAlexanderGolovnev1Da´vidPa´l2Bala´zsSzo¨re´nyi2Abstractofalgorithmsindexedbythe(uncountablymany)distri-butionsoverthedomain...
TheEffectofNetworkWidthonStochasticGradientDescentandGeneralization:anEmpiricalStudyDanielS.Park12JaschaSohl-Dickstein1QuocV.Le1SamuelL.Smith3AbstractWilsonetal.,2017;Sagunetal.,2017;Mandtetal.,201...
TheAdvantagesofMultipleClassesforReducingOverfittingfromTestSetReuseVitalyFeldman12RoyFrostig1MoritzHardt34Abstracttransferrathergracefullytoanewlycollectedtestsetcol-lectedfromthesamesourceaccordi...
SurrogateLossesforOnlineLearningofStepsizesinStochasticNon-ConvexOptimizationZhenxunZhuang1AshokCutkosky2FrancescoOrabona13Abstractstepsizeηt>0.Inordertoachieveafastconvergence,thestepsizesmustbec...
StatisticalFoundationsofVirtualDemocracyAnsonKahng1MinKyungLee1RiteshNoothigattu1ArielD.Procaccia1AlexandrosPsomas1Abstractpledilemmas;second,learnmodelsoftheirpreferences,whichgeneralizetoany(prev...
SpectralClusteringofSignedGraphsviaMatrixPowerMeansPedroMercado12FrancescoTudisco3MatthiasHein2Abstractexample,canexpresspositiveinteractions,likefriendshipandtrust,andnegativeones,likeenmityanddis...
SimilarityofNeuralNetworkRepresentationsRevisitedSimonKornblith1MohammadNorouzi1HonglakLee1GeoffreyHinton1AbstractThispaperinvestigatestheproblemofmeasuringsimilari-tiesbetweendeepneuralnetworkrepr...
SensitivityAnalysisofLinearStructuralCausalModelsCarlosCinelli1DanielKumor2BryantChen3JudeaPearl1EliasBareinboim2Abstractdertoobtaincausalclaims.Theseassumptionsareusuallyencodedastheabsenceofcerta...
ScalableTrainingofInferenceNetworksforGaussian-ProcessModelsJiaxinShi1MohammadEmtiyazKhan2JunZhu1AbstractFigure1.InferencenetworksforGPsareanalternativewaytopredictoutputf∗giventestinputsx∗.Infer...
RobustEstimationofTreeStructuredGaussianGraphicalModelsAshishKatiyar1JessicaHoffmann1ConstantineCaramanis1Abstractmatrix,Σ,wehaveaccessonlytoM=Σ+D,thesumofthecovariancematrixandadiagonalmatrix.In...
RefinedComplexityofPCAwithOutliersFedorFomin1PetrGolovach1FahadPanolan1KirillSimonov1Abstractlow-rankapproximationofdatamatrixMbysolvingPrincipalcomponentanalysis(PCA)isoneoftheminimizeM−L2mostfun...
RegretCircuits:ComposabilityofRegretMinimizersGabrieleFarina1ChristianKroer2TuomasSandholm1345Abstractvariants,alongwithotherscalabilitytechniquessuchasreal-timeendgamesolving(Ganzfried&Sandholm,20...