Improvement of Reinforcement Learning With Supermodularity | IEEE Journals & Magazine | IEEE Xplore