Privacy-preserving Q-Learning with Functional Noise
        in Continuous Spaces

         Baoxiang Wang                   Nidhi Hegde
      Borealis AI, Edmonton, Canada          Borealis AI, Edmonton, Canada
       bxiangwang@gmail.com             nidhi.hegde@borealisai.com

                       Abstract
     We consider differentially private algorithms for reinforcement learning in contin-
     uous spaces, such that neighboring reward functions are indistinguishable. This
     protects th ...