Enabling Multi-Agent Transfer Reinforcement Learning via Scenario Independent Representation | IEEE Conference Publication | IEEE Xplore