Double Successive Over-Relaxation Q-Learning With an Extension to Deep Reinforcement Learning | IEEE Journals & Magazine | IEEE Xplore