State Entropy Maximization with Random Encoders
                 for Efficient Exploration

      Younggyo Seo * 1 Lili Chen * 2 Jinwoo Shin 1 Honglak Lee 3 4 Pieter Abbeel 2 Kimin Lee 2

            Abstract                 proaches encourage agents to visit diverse states, but leave
   Recent exploration methods have proven to be a       unanswered the fundamental question of how to quantify
   recipe for improving sample-efficiency in deep re-     effective exploration in a principled way.
   ...