Reinforcement learning from human reward: Discounting in episodic tasks | IEEE Conference Publication | IEEE Xplore