TowardsPracticalMeanBoundsforSmallSamplesMyPhan1PhilipS.Thomas1ErikLearned-Miller1Abstractrefertotwoseparatesettings,theone-endedsupportsetting,inwhichthedistributionisknowntofallintheintervalHisto...
NetworkInferenceandInfluenceMaximizationfromSamplesWeiChen1XiaomingSun23JialinZhang23ZhijieZhang23Abstractnodescalledseedstakessomepieceofinformation(news,ideasoropinions,etc.),andtheinformationspr...
OptimizationfromStructuredSamplesforCoverageFunctionsWeiChen1XiaomingSun23JialinZhang23ZhijieZhang23Abstracteters.Thecomputationaltaskistooptimizeanobjectivefunctiongivenparametersofthemodel.Onesuc...
TransferofSamplesinPolicySearchviaMultipleImportanceSamplingAndreaTirinzoni1MattiaSalvini1MarcelloRestelli1Abstractagentissupposedtoreuseknowledgeacquiredfromasetofsourcetaskstoacceleratethelearnin...
StatisticsandSamplesinDistributionalReinforcementLearningMarkRowland1RobertDadashi2SaurabhKumar2Re´miMunos1MarcG.Bellemare2WillDabney1AbstractthatDRLalgorithmscanbeviewedascombiningastatisti-cales...
SELFIE:RefurbishingUncleanSamplesforRobustDeepLearningHwanjunSong1MinseokKim1Jae-GilLee1Abstracttransitionmatrix(Zhangetal.,2017;Goldberger&Ben-Reuven,2017;Patrinietal.,2017;Changetal.,2017).AsOwin...
WhenSamplesAreStrategicallySelectedHanruiZhang1YuCheng1VincentConitzer1Abstract1988;Friedman,1993)),thereisstilltheconcernthattheagentsendsonlyabiasedcollectionofSamples.IfwedoInstandardclassificat...
NotAllSamplesAreCreatedEqual:DeepLearningwithImportanceSamplingAngelosKatharopoulos12Franc¸oisFleuret12Abstractmodel.Tothisend,weproposeanovelimportancesamplingschemethatacceleratesthetrainingofan...
ImportanceWeightedTransferofSamplesinReinforcementLearningAndreaTirinzoni1AndreaSessa1MatteoPirotta2MarcelloRestelli1Abstracttions,parameters,policies,etc.)andinthecriteriausedtoestablishwhethersuc...
FindingInfluentialTrainingSamplesforGradientBoostedDecisionTreesBorisSharchilev12YuryUstinovsky3PavelSerdyukov2MaartendeRijke1Abstract(2)boostingdeveloper’strustinthemodel’sperformanceinscenarios...
Nystro¨mMethodwithKernelK-means++SamplesasLandmarksDinoOglic12ThomasGa¨rtner2AbstractoreigendecompositionwhichscaleasOn3.ToovercomethiscomputationalshortcomingandscalekernelmethodsWeinvestigate,t...