Convergence of a Human-in-the-Loop Policy-Gradient Algorithm With Eligibility Trace Under Reward, Policy, and Advantage Feedback

Autor:	Shah, Ishaan, Halpern, David, Asadi, Kavosh, Littman, Michael L.
Rok vydání:	2021
Předmět:	Computer Science - Machine Learning Computer Science - Artificial Intelligence Computer Science - Data Structures and Algorithms Computer Science - Human-Computer Interaction
Druh dokumentu:	Working Paper
Popis:	Fluid human-agent communication is essential for the future of human-in-the-loop reinforcement learning. An agent must respond appropriately to feedback from its human trainer even before they have significant experience working together. Therefore, it is important that learning agents respond well to various feedback schemes human trainers are likely to provide. This work analyzes the COnvergent Actor-Critic by Humans (COACH) algorithm under three different types of feedback-policy feedback, reward feedback, and advantage feedback. For these three feedback types, we find that COACH can behave sub-optimally. We propose a variant of COACH, episodic COACH (E-COACH), which we prove converges for all three types. We compare our COACH variant with two other reinforcement-learning algorithms: Q-learning and TAMER. Comment: Accepted into ICML 2021 workshops Human-AI Collaboration in Sequential Decision-Making and Human in the Loop Learning
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2109.07054 Zobrazit plný text záznamu View this record from Arxiv