ActiveFeatureAcquisitionwithGenerativeSurrogateModelsYangLi1JunierB.Oliva1Abstractparadigmissuccessfulinamultitudeofdomains,itisincon-gruouswiththeexpectationsofmanyreal-worldintelligentManyreal-wo...
OntheConsistencyofTop-kSurrogateLossesForestYang12SanmiKoyejo13Abstractasampleand/orwhenasamplemaycorrespondtomultiplelabels,e.g.,whenanimageofaparkcontainingapondmayThetop-kerrorisoftenemployedtoe...
SurrogateLossesforOnlineLearningofStepsizesinStochasticNon-ConvexOptimizationZhenxunZhuang1AshokCutkosky2FrancescoOrabona13Abstractstepsizeηt>0.Inordertoachieveafastconvergence,thestepsizesmustbec...
LeveragingLow-RankRelationsBetweenSurrogateTasksinStructuredPredictionGiuliaLuise1DimitrisStamos1MassimilianoPontil12CarloCiliberto13AbstractAmongthemostwell-establishedstrategiesforstructuredpredi...