A hybrid P2P and master-slave architecture for intelligent multi-agent reinforcement learning in a distributed computing environment: A case study | IEEE Conference Publication | IEEE Xplore