OntheGeneralizationBenefitofNoiseinStochasticGradientDescentSamuelL.Smith1ErichElsen1SohamDe1Abstractbatches,orbecauseauthorsoftencomparedifferentbatchsizesunderaconstantepochbudget(suchthatsmallba...
OntheConvergenceofNesterov’sAcceleratedGradientMethodinStochasticSettingsMahmoudAssran123MichaelRabbat23AbstractHowever,thetheoreticalunderstandingofacceleratedmeth-odsremainslimitedwhenusedwithst...
OntheExpressivityofNeuralNetworksforDeepReinforcementLearningKefanDong1YupingLuo2TianheYu3ChelseaFinn3TengyuMa3Abstractwiththeestimateddynamics(Nagabandietal.,2018;Chuaetal.,2018;Wang&Ba,2019).Weco...
OntheConsistencyofTop-kSurrogateLossesForestYang12SanmiKoyejo13Abstractasampleand/orwhenasamplemaycorrespondtomultiplelabels,e.g.,whenanimageofaparkcontainingapondmayThetop-kerrorisoftenemployedtoe...
Onthe(In)tractabilityofComputingNormalizingConstantsfortheProductofDeterminantalPointProcessesNaotoOhsaka1TatsuyaMatsuoka1Abstractdet(AS,S).Considerasubsetselectiontask:givennitems(e.g.,images(Kule...
Onp-normRobustnessofEnsembleDecisionStumpsandTreesYihanWang1HuanZhang2HonggeChen3DuaneBoning3Cho-JuiHsieh2Abstractetal.,2017;Ilyasetal.,2018;Brendeletal.,2018;Chengetal.,2019a;2020),variousalgorith...
OnEfficientConstructionsofCheckpointsYuChen1ZhenmingLiu1BinRen1XinJin2AbstractProducingcheckpointsfrequentlyenablesfailedtrainingprocesstorestartwithminimumwastedtime,andservesEfficientconstruction...
NestedSubspaceArrangementforRepresentationofRelationalDataNozomiHata1ShizuoKaji2AkihiroYoshida1KatsukiFujisawa2Abstracttion,multiplication,differentiation,metric,andtopologysothatvariousoperations,...
MyFairBandit:DistributedLearningofMax-MinFairnesswithMulti-playerBanditsIlaiBistritz1TavorZ.Baharav1AmirLeshem2NicholasBambos1Abstracttheenvironment.Isthereanalternativethatliesinthegapbetweenthetw...
FACT:ADiagnosticforGroupFairnessTrade-offsJoonSikKim12JiahaoChen3AmeetTalwalkar14Abstractfairness,whichmeasureshowagroupofindividualswithcertainprotectedattributesaretreateddifferentlyfromotherGrou...
MeasuringNon-ExpertComprehensionofMachineLearningFairnessMetricsDebjaniSaha1CandiceSchumann1DuncanC.McElfresh1JohnP.Dickerson1MichelleL.Mazurek1MichaelCarlTschantz2Abstractoffairnesscanhavesignific...
Low-lossconnectionofweightvectors:distribution-basedapproachesIvanAnokhin1DmitryYarotsky1AbstractRecentresearchprovidessomefurtherevidenceinfavorofthe“connectedsublevelset”scenario.Aparticulareas...
LinearConvergenceofRandomizedPrimal-DualCoordinateMethodforLarge-scaleLinearConstrainedConvexProgrammingDaoliZhu1LeiZhao2Abstractadditivewithrespecttothefollowingspacedecomposition,Linearconstraine...
LinearLowerBoundsandConditioningofDifferentiableGamesAdamIbrahim1WaïssAzizian2GauthierGidel1IoannisMitliagkas1Abstractmind(Meschederetal.,2017),andtomakemattersworse,havebeenoftentunedsuboptimally...
LEEP:ANewMeasuretoEvaluateTransferabilityofLearnedRepresentationsCuongV.Nguyen1TalHassner2MatthiasSeeger1CedricArchambeau1Abstractchoosegoodsourcemodelsforagiventargettask(Achilleetal.,2019;Baoetal...
LearningtheValuationsofak-demandAgentHanruiZhang1VincentConitzer1AbstractInothercontexts,suchastheallocationofgoods(orbads,e.g.,tasks),agentsareoftenabletomakepayments(orWestudyproblemswherealearne...
LearningTask-AgnosticEmbeddingofMultipleBlack-BoxExpertsforMulti-TaskModelFusionTrongNghiaHoang1ChiThanhLam2BryanKianHsiangLow2PatrickJaillet3Abstractwouldthereforehavedifferentdistributionsorstati...
Learningthepiece-wiseconstantgraphstructureofavaryingIsingmodelBatisteLeBars1PierreHumbert1ArgyrisKalogeratos1NicolasVayatis1Abstractedgebetweentwonodesinthisgraphindicatesthecondi-tionaldependency...
LearningMixturesofGraphsfromEpidemicCascadesJessicaHoffmann1SoumyaBasu1SurbhiGoel1ConstantineCaramanis1AbstractRodriguezetal.,2013;Chengetal.,2014;Zhaoetal.,2015;Liuetal.,2019),detectingthem(Arias-...