TAMER: Training an Agent Manually via Evaluative Reinforcement | IEEE Conference Publication | IEEE Xplore