CHQ: a multi-agent reinforcement learning scheme for partially observable Markov decision processes | IEEE Conference Publication | IEEE Xplore