Ensemble Bootstrapping for Q-Learning

                 Oren Peer 1 Chen Tessler 1 Nadav Merlis 1 Ron Meir 1

             Abstract                  focuses on learning the value-function. The value represents
   Q-learning (QL), a common reinforcement learn-         the expected, discounted, reward-to-go that the agent will
   ing algorithm, suffers from over-estimation bias        obtain. In particular, such methods learn the optimal pol-
   due to the maximization term in the optimal          ...