Q-SAT: Value Factorization with Self-Attention for Deep Multi-Agent Reinforcement Learning | IEEE Conference Publication | IEEE Xplore