RevisitingPeng’sQ(λ)forModernReinforcementLearningTadashiKozuno1YunhaoTang2MarkRowland3Re´miMunos4StevenKapturowski3WillDabney3MichalValko4DavidAbel3Abstract1996;Watkins,1989;Peng&Williams,1994;...
AKernelTheoryofModernDataAugmentationTriDao1AlbertGu1AlexanderJ.Ratner1VirginiaSmith2ChristopherDeSa3ChristopherRe´1Abstractasregularizertomaketheresultingmodelmorerobust,andprovideresourcestodata...
OnCalibrationofModernNeuralNetworksChuanGuo1GeoffPleiss1YuSun1KilianQ.Weinberger1AbstractLeNet(1998)ResNet(2016)1.0CIFAR-100CIFAR-100Confidencecalibration–theproblemofpredict-ingprobabilityestimat...