DeltaGrad:RapidretrainingofmachinelearningmodelsYinjunWu1EdgarDobriban2SusanB.Davidson1AbstractFigure1.RunningtimeofourDeltaGradalgorithmforretrainingalogisticregressionmodelonRCV1asafunctionofthef...
DeepMolecularProgramming:ANaturalImplementationofBinary-WeightReLUNeuralNetworksMarkoVasic1CameronChalk1SarfrazKhurshid1DavidSoloveichik1AbstractChemicalcontrolmodulescompatiblewiththechemicalenvir...
CurseofDimensionalityonRandomizedSmoothingforCertifiableRobustnessAounonKumar1AlexanderLevine1TomGoldstein1SoheilFeizi1AbstractbasedmethodssuchasFGSM(Goodfellowetal.,2015)andprojectedgradientdescen...
CoresetsforClusteringinGraphsofBoundedTreewidthDanielBaker1VladimirBraverman1LingxiaoHuang2ShaofengH.-C.Jiang3RobertKrauthgamer3XuanWu1Abstractthesecontexts,thefocusisonedge-weightedgraphsG=(V,E)wi...
CoresetsforData-efficientTrainingofMachineLearningModelsBaharanMirzasoleiman1JeffBilmes2JureLeskovec1AbstractTrainingmachinelearningmodelsoftenreducestooptimiz-Incrementalgradient(IG)methods,suchas...
ConvergenceRatesofVariationalInferenceinSparseDeepLearningBadr-EddineChérief-Abdellatif1AbstractModernapproximateinferencemainlyreliesonvariationalinference(VI),withsometimesaflavorofsamplingtech-...
ConvergenceofaStochasticGradientMethodwithMomentumforNon-SmoothNon-ConvexOptimizationVienV.Mai1MikaelJohansson1AbstractThisfunctionclassisveryrichandimportantinoptimiza-tion(Rockafellar,1982;Vial,1...
ControllingOverestimationBiaswithTruncatedMixtureofContinuousDistributionalQuantileCriticsArseniiKuznetsov1PavelShvechikov12AlexanderGrishin13DmitryVetrov13AbstractThrun&Schwartz(1993)elucidatetheo...
ConciseExplanationsofNeuralNetworksusingAdversarialTrainingPrasadChalasani1JiefengChen2AmritaRoyChowdhury2SomeshJha12XiWu3Abstract1.IntroductionWeshownewconnectionsbetweenadversarialDespitetherecen...
ComposableSketchesforFunctionsofFrequencies:BeyondtheWorstCaseEdithCohen12OfirGeri34RasmusPagh156Abstractdatawithsmallcomputationalresources(time,communi-cation,andspace).Suchsketchessupportprocess...
ComplexityofFindingStationaryPointsofNonsmoothNonconvexFunctionsJingzhaoZhang1HongzhouLin1StefanieJegelka1SuvritSra1AliJadbabaie1AbstractTable1.Whentheproblemisnonconvexandnonsmooth,find-inga-stati...
ClosingtheconvergencegapofSGDwithoutreplacementShashankRajput1AnantGupta1DimitrisPapailiopoulos1Abstractthechoiceofasingleorasubsetofsampledfunctionsfi,andαrepresentsthestepsize.With-andwithoutre-...
CLUB:AContrastiveLog-ratioUpperBoundofMutualInformationPengyuCheng1WeituoHao1ShuyangDai1JiachangLiu1ZheGan2LawrenceCarin1Abstract2015),andmachinelearning(Chenetal.,2016;Alemietal.,2016;Hjelmetal.,2...
ChoiceSetOptimizationUnderDiscreteChoiceModelsofGroupDecisionsKiranTomlinson1AustinR.Benson1Abstractatownmightpreferdifferentgovernmentpolicies.Thewaythatpeoplemakechoicesorexhibitpref-Muchofthecom...
BreakingtheCurseofSpaceExplosion:TowardsEffcientNASwithCurriculumSearchYongGuo1YaofoChen12YinZheng3PeilinZhao4JianChen15JunzhouHuang4MingkuiTan1AbstractStandardNASwithNASwithcurriculumsearchfixedse...
BreakingtheCurseofManyAgents:ProvableMeanEmbeddingQ-IterationforMean-FieldReinforcementLearningLingxiaoWang1ZhuoranYang2ZhaoranWang1Abstract1998;Dzˇeroskietal.,2001;Guestrinetal.,2002;Karetal.,201...
BoundingtheFairnessandAccuracyofClassifiersfromPopulationStatisticsSivanSabato12andEladYom-Tov2Abstractfairnesswithrespecttothestateofresidence(assumingaUS-basedpopulation).Accuracyandfairnessareea...
BoostingforControlofDynamicalSystemsNamanAgarwal1NatalyBrukhim21EladHazan21ZhouLu2Abstractofthesystem,andhenceitisoftennota-prioriclearhowtoobtainameaningfulguaranteewhenswitchingbetweenorWestudyth...
BayesianSparsificationofDeepC-valuedNetworksIvanNazarov1EvgenyBurnaev1AbstractthatcausedegeneratetransformationsinR-valuednetworkswithtwicethefeaturedimensions.Theirstudydemon-Withcontinualminiatur...