Policy Continuation with Hindsight Inverse Dynamics

         Hao Sun1 , Zhizhong Li1 , Xiaotong Liu2 , Dahua Lin1 , Bolei Zhou1
           1
            The Chinese University of Hong Kong, 2 Peking University

                        Abstract
      Solving goal-oriented tasks is an important but challenging problem in reinforce-
      ment learning (RL). For such tasks, the rewards are often sparse, making it difficult
      to learn a policy effectively. To tackle this difficulty, we propose  ...