When to Trust Your Model:
          Model-Based Policy Optimization

       Michael Janner Justin Fu    Marvin Zhang   Sergey Levine
              University of California, Berkeley
       {janner, justinjfu, marvin, svlevine}@eecs.berkeley.edu

                        Abstract
     Designing effective model-based reinforcement learning algorithms is difficult
     because the ease of data generation must be weighed against the bias of model-
     generated data. In this paper, we study the role ...