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

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Ghaemi, Hafez, Kebriaei, Hamed, Moghaddam, Alireza Ramezani, Ahamdabadi, Majid Nili
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917585182785536
author Ghaemi, Hafez
Kebriaei, Hamed
Moghaddam, Alireza Ramezani
Ahamdabadi, Majid Nili
author_facet Ghaemi, Hafez
Kebriaei, Hamed
Moghaddam, Alireza Ramezani
Ahamdabadi, Majid Nili
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2402_05906
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Risk-Sensitive Multi-Agent Reinforcement Learning in Network Aggregative Markov Games
Ghaemi, Hafez
Kebriaei, Hamed
Moghaddam, Alireza Ramezani
Ahamdabadi, Majid Nili
Machine Learning
Artificial Intelligence
Multiagent Systems
I.2.6; I.2.11
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.
title Risk-Sensitive Multi-Agent Reinforcement Learning in Network Aggregative Markov Games
topic Machine Learning
Artificial Intelligence
Multiagent Systems
I.2.6; I.2.11
url https://arxiv.org/abs/2402.05906