InstanceSpecificApproximationsforSubmodularMaximizationEricBalkanski1SharonQian2YaronSinger2Abstract1.IntroductionFormanyoptimizationproblemsinmachinelearn-Acentralchallengeinmachinelearningisthatm...
UnderstandingImpactsofHigh-OrderLossApproximationsandFeaturesinDeepLearningInterpretationSahilSingla1EricWallace1ShiFeng1SoheilFeizi1Abstracthindmodelpredictions?Acommoninterpretationapproachistoid...
OvercomingMean-FieldApproximationsinRecurrentGaussianProcessModelsAlessandroDavideIalongo12MarkvanderWilk3JamesHensman3CarlEdwardRasmussen1Abstractthedistributionoftheoutcome(vonNeumannetal.,1944)....
Linear-ComplexityData-ParallelEarthMover’sDistanceApproximationsKubilayAtasu1ThomasMittelholzer2AbstractInthetextretrievaldomain,anadaptationofEMD,calledWordMoversDistance(WMD),hasemergedasastate-...
VariationalBoosting:IterativelyRefiningPosteriorApproximationsAndrewC.Miller1NicholasJ.Foti2RyanP.Adams13AbstractchainMonteCarlo(MCMC)methodsthatapproximateatargetdistributionwithsamplesdrawnfromaM...
NeuralTaylorApproximations:ConvergenceandExplorationinRectifierNetworksDavidBalduzzi1BrianMcWilliams2TonyButler-Yeoman1AbstractFig.1:ShatteredgradientsinaPL-function.Modernconvolutionalnetworks,inc...
Multi-fidelityBayesianOptimisationwithContinuousApproximationsKirthevasanKandasamy1GautamDasarathy2JeffSchneider1Barnaba´sPo´czos1AbstractTypically,insuchapplications,eachfunctionevaluationisexpe...