On Reinforcement Learning with Adversarial Corruption
              and Its Application to Block MDP

             Tianhao Wu * 1 2 Yunchang Yang * 3 Simon S. Du 4 Liwei Wang 3 5

             Abstract                 is vulnerable to corrupted data stemming from malicious
                                 entities (Huang et al., 2017; Ma et al., 2019), non-malicious
   We study reinforcement learning (RL) in episodic       yet non-stationary behavior, or simply errors in the system.
   tabular M ...