Accelerating Safe Reinforcement Learning
              with Constraint-mismatched Baseline Policies

          Tsung-Yen Yang 1 Justinian Rosca 2 Karthik Narasimhan 1 Peter J. Ramadge 1

             Abstract                 or other costs. For instance, when you drive an unfamiliar
                                  vehicle, you do so cautiously to ensure safety, while adapt-
   We consider the problem of reinforcement learn-
                                  ing your driving technique to the ve ...