A Maximum Divergence Approach to Optimal Policy in Deep Reinforcement Learning | IEEE Journals & Magazine | IEEE Xplore