Reinforcement Learning with Goal-Distance Gradient

Kai Jiang, XiaoLong Qin

Reinforcement learning usually uses the feedback rewards of environmental to train agents. But the rewards in the actual environment are sparse, and even some environments will not…