Instruction for reinforcement learning agent based on sub-rewards and forgetting | IEEE Conference Publication | IEEE Xplore