RiskQ: Risk-sensitive Multi-Agent Reinforcement Learning Value Factorization

arXiv:2311.01753v2 Announce Type: replace-cross
Abstract: Multi-agent systems are characterized by environmental uncertainty, varying policies of agents, and partial observability, which result in significant risks. In the context of Multi-Agent Reinforcement Learning (MARL), learning coordinated and decentralized policies that are sensitive to risk is challenging. To formulate the coordination requirements in risk-sensitive MARL, we introduce the Risk-sensitive Individual-Global-Max (RIGM) principle as a generalization of the Individual-Global-Max (IGM) and Distributional IGM (DIGM) principles. This principle requires that the collection of risk-sensitive action …

agent arxiv cs.ai cs.lg cs.ma factorization multi-agent reinforcement reinforcement learning risk type value