Learning to Play Pac-Xon With Q-Learning and Two Double Q-Learning Variants

doi 10.1109/ssci.2018.8628782