STMARL: A Spatio-Temporal Multi-Agent Reinforcement Learning Approach for Cooperative Traffic Light Control | IEEE Journals & Magazine | IEEE Xplore