Reinforcement Learning-based Decentralized Optimal Control for Large-Scale Multi-agent System by Using Neural Networks and Discrete-time Mean Field Games | IEEE Conference Publication | IEEE Xplore