Fast Reinforcement Learning by Mirror Images | IEEE Conference Publication | IEEE Xplore