Voice2Series:ReprogrammingAcousticModelsforTimeSeriesClassificationChao-HanHuckYang1Yun-YunTsai2Pin-YuChen3AbstracttargetreprogrammedsourceoutputtargetoutputLearningtoclassifytimeserieswithlimitedd...
VectorQuantizedModelsforPlanningSherjilOzair12YazheLi1AliRazavi1IoannisAntonoglou1AäronvandenOord1OriolVinyals1AbstractDota2(OpenAIetal.,2019)andStarCraftII(Vinyalsetal.,2019),orrobotics(OpenAIeta...
Synthesizer:RethinkingSelf-AttentionforTransformerModelsYiTay1DaraBahri1DonaldMetzler1Da-ChengJuan1ZheZhao1CheZheng1Abstractwidelyattributedtothisself-attentionmechanismsincefullyconnectedtokengrap...
StabilizingEquilibriumModelsbyJacobianRegularizationShaojieBai1VladlenKoltun2J.ZicoKolter1Abstractequilibriumzbytheimplicitfunctiontheorem(Krantz&Parks,2012),irrespectiveofthemethodusedtosolveforDe...
RobustInferenceforHigh-DimensionalLinearModelsviaResidualRandomizationY.SamuelWang1SiKaiLee1PanosToulis1MladenKolar1Abstractsamplesplittingprocedurestoformvalidp-values.Zhang&Zhang(2014),VandeGeere...
RegularizingtowardsCausalInvariance:LinearModelswithProxiesMichaelOberst1NikolajThams2JonasPeters2DavidSontag1AbstractβxAβyWeproposeamethodforlearninglinearmod-XαYelswhosepredictiveperformanceis...
Zoo-Tuning:AdaptiveTransferfromaZooofModelsYangShu1ZhiKou1ZhangjieCao1JianminWang1MingshengLong1AbstractleverageModelspretrainedonlarge-scaledatasets(Rus-sakovskyetal.,2015)andfine-tunethemodelonth...
ProvablyStrictGeneralisationBenefitforEquivariantModelsBrynElesedy1SheheryarZaidi2AbstractasSokolicetal.(2017);Sannai&Imaizumi(2019),coveronlytheworstcaseperformanceofalgorithms.TheseworksItiswidel...
ProvableLipschitzCertificationforGenerativeModelsMattJordan1AlexandrosG.Dimakis1Abstractthroughavarietyofnonlinearoperatorsandrelatethistoa2-dimensionalgeometricproblemthatwesolveop-Wepresentascala...
PriorImage-ConstrainedReconstructionusingStyle-BasedGenerativeModelsVarunA.Kelkar1MarkA.Anastasio1Abstractqualityofimagesgenerated,invertibilityandimagerepre-sentation,andameaningfulcontrolofsemant...
ParallelandFlexibleSamplingfromAutoregressiveModelsviaLangevinDynamicsVivekJayaram1JohnThickstun1AbstractlimittheusefulnessofautoregressiveModelsinpracticalsettings.First,ancestralsamplinghastimeco...
OnEstimationinLatentVariableModelsGuanhuaFang,PingLiCognitiveComputingLabBaiduResearch10900NE8thStBellevueWA98004USA{guanhuafang,liping11}@baidu.comAbstractvariablescanbeaggregatedinamodeltoreprese...
OnEnergy-BasedModelswithOverparametrizedShallowNeuralNetworksCarlesDomingo-Enrich1AlbertoBietti2EricVanden-Eijnden1JoanBruna12Abstracttionaltoexp{−f(x)}.Suchenergy-basedModels(EBMs)originateinstat...
OfflineContextualBanditswithOverparameterizedModelsDavidBrandfonbrener1WilliamF.Whitney1RajeshRanganath1JoanBruna1AbstractIncontrast,thebestperformanceinmodernsupervisedlearningisoftenachievedbymas...
OnVariationalInferenceinBiclusteringModelsGuanhuaFang,PingLiCognitiveComputingLabBaiduResearch10900NE8thStBellevueWA98004USA{guanhuafang,liping11}@baidu.comAbstractstudies(Prelic´etal.,2006;GuandL...
LogME:PracticalAssessmentofPre-trainedModelsforTransferLearningKaichaoYou∗1YongLiu∗1JianminWang1MingshengLong1Abstracttrainedwithlarge-scalesuperviseddata(Dengetal.,2009;Russakovskyetal.,2015)and...
LearningTransferableVisualModelsFromNaturalLanguageSupervisionAlecRadford1JongWookKim1ChrisHallacy1AdityaRamesh1GabrielGoh1SandhiniAgarwal1GirishSastry1AmandaAskell1PamelaMishkin1JackClark1Gretchen...
InferenceforNetworkRegressionModelswithCommunityStructureMengjiePan1TylerH.McCormick2BaileyK.Fosdick3AbstractInthiswork,wefocusonacasewherecontinuousrelationsbetweenpairsofactorsaremodeledasalinear...
ImprovedContrastiveDivergenceTrainingofEnergy-BasedModelsYilunDu1ShuangLi1JoshuaTenenbaum1IgorMordatch2AbstractFigure1:(Left)128x128samplesonunconditionalCelebA-HQ.(Right)128x128samplesonunconditio...
ImprovedDenoisingDiffusionProbabilisticModelsAlexNichol1PrafullaDhariwal1AbstractandVAEs(Kingma&Welling,2013).Thisraisesvariousquestions,suchaswhetherDDPMsarecapableofcapturingDenoisingdiffusionpro...