Generalized Data Distribution Iteration

Jiajun Fan, Changnan Xiao

Abstract: To obtain higher sample efficiency and superior final performance simultaneously has been one of the major challenges for deep reinforcement learning (DRL). Previous work could handle one of these ...