Generalization in Deep Learning.pdf

Withadirectanalysisofneuralnetworks,thispaperpresentsamathematicallytight generalizationtheorytopartiallyaddressanopenproblemregardingthegeneralization ofdeeplearning.Unlikepreviousbound-basedtheory,ourmaintheoryisquantitatively astightaspossibleforeverydata