An Efficient Multi-Agent Q-learning Method Based on Observing the Adversary Agent State Change | IEEE Conference Publication | IEEE Xplore