---
title: "Finite-Sample Convergence in Networked Average Reward MARL: Decentralization Pitfalls and Entropy Remedies"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.04072v1"
permalink: "https://feed.astralor.com/items/itm_8ad5cab6983785e9aa52a5ea"
published: "2026-11-02T23:34:29.000Z"
score: 58
channel: "ai"
tags: ["AI Agent","RAG","Finite-Sample Convergence","Networked Average Reward","MARL","Decentralization Pitfalls","Entropy Remedies","Multi-Agent Reinforcement Learning"]
---

# Finite-Sample Convergence in Networked Average Reward MARL: Decentralization Pitfalls and Entropy Remedies

## 速览

We study the problem of average reward Multi-Agent Reinforcement Learning (MARL) where agents interact within a network. Each agent in the network conducts local updates with information within its $k$-hop neighborhood …

## 关键点

- We study the problem of average reward Multi-Agent Reinforcement Learning (MARL) where agents interact within a network.
- Each agent in the network conducts local updates with information within its $k$-hop neighborhood and collaboratively maximizes the overall reward of the entire network.
- We first provide impossibility results in achieving convergence to the global optimal joint policies in our setting.

## 全文

We study the problem of average reward Multi-Agent Reinforcement Learning (MARL) where agents interact within a network. Each agent in the network conducts local updates with information within its $k$-hop neighborhood and collaboratively maximizes the overall reward of the entire network. We first provide impossibility results in achieving convergence to the global optimal joint policies in our setting. These impossibility results highlight the challenges of decentralized learning with local information in multi-agent systems, namely the issues of multiagency, where each agent independently optimizes its policy, and partial observability, where agents have limited visibility into the global state. Given these challenges which indicate that decentralized policy optimization can generally be certified only up to stationarity, we provide finite-sample convergence guarantees for a decentralized actor-critic algorithm with linear function approximation that converges to an approximate stationary point with small gradient norm. To overcome such impossibility results, we further show that adding fixed-state entropy regularization reshapes the objective landscape which leads to even stronger global convergence guarantees for our proposed actor-critic algorithm.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_8ad5cab6983785e9aa52a5ea