Competitive Multi-agent Deep Reinforcement Learning with Counterfactual Thinking | IEEE Conference Publication | IEEE Xplore