Gong Gao

Expert Behavior Prior Reinforcement Learning

Gong Gao, Weidong Zhao, Xianhui Liu, Ning Jia

Behavior prior reinforcement learning (BPRL) has emerged as a promising paradigm to improve sample efficiency in online reinforcement learning (RL) by leveraging policy priors derived from offline demonstrations. However, most existing BPRL methods rely on static offline datasets…Read more