arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.05906cs.LGcs.AIcs.MA
收藏

网络聚合马尔可夫博弈中的风险敏感多智能体强化学习

Risk-Sensitive Multi-Agent Reinforcement Learning in Network Aggregative Markov Games

University of Tehran(德黑兰大学)

Hafez Ghaemi, Hamed Kebriaei, Alireza Ramezani Moghaddam, Majid Nili Ahamdabadi

AI总结:

针对网络聚合马尔可夫博弈,提出分布式嵌套CPT-AC算法,将累积前景理论风险引入多智能体强化学习,证明收敛于主观马尔可夫完美纳什均衡,实验显示损失厌恶影响社交隔离倾向。

摘要:

经典的多智能体强化学习(MARL)假设智能体具有风险中性和完全客观性。然而,在智能体需要考虑或建模人类经济或社会偏好的场景中,必须将风险概念纳入强化学习优化问题。这在涉及其他人类或非人类智能体(可能具有各自的风险敏感策略)的多智能体强化学习中尤为重要。在本工作中,我们考虑了结合累积前景理论(CPT)的风险敏感且非合作的多智能体强化学习,CPT是一种非凸风险度量,也是相干风险度量的推广。CPT能够解释人类的损失厌恶以及他们高估/低估小/大概率的倾向。我们提出了一种基于分布式采样的演员-评论家(AC)算法,该算法结合了CPT风险,用于网络聚合马尔可夫博弈(NAMGs),我们称之为分布式嵌套CPT-AC。在一组假设下,我们证明了该算法在NAMGs中收敛到主观的马尔可夫完美纳什均衡。实验结果表明,我们的算法获得的主观CPT策略可能与风险中性策略不同,并且具有更高损失厌恶的智能体在NAMGs中更倾向于社交隔离。

英文摘要:

Classical multi-agent reinforcement learning (MARL) assumes risk neutrality and complete objectivity for agents. However, in settings where agents need to consider or model human economic or social preferences, a notion of risk must be incorporated into the RL optimization problem. This will be of greater importance in MARL where other human or non-human agents are involved, possibly with their own risk-sensitive policies. In this work, we consider risk-sensitive and non-cooperative MARL with cumulative prospect theory (CPT), a non-convex risk measure and a generalization of coherent measures of risk. CPT is capable of explaining loss aversion in humans and their tendency to overestimate/underestimate small/large probabilities. We propose a distributed sampling-based actor-critic (AC) algorithm with CPT risk for network aggregative Markov games (NAMGs), which we call Distributed Nested CPT-AC. Under a set of assumptions, we prove the convergence of the algorithm to a subjective notion of Markov perfect Nash equilibrium in NAMGs. The experimental results show that subjective CPT policies obtained by our algorithm can be different from the risk-neutral ones, and agents with a higher loss aversion are more inclined to socially isolate themselves in an NAMG.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑