Multi-Agent Reinforcement Learning with Submodular Reward

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chen, Wenjing, Qian, Chengyuan, Xing, Shuo, Zhou, Yi, Crawford, Victoria
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917321047539712
author Chen, Wenjing
Qian, Chengyuan
Xing, Shuo
Zhou, Yi
Crawford, Victoria
author_facet Chen, Wenjing
Qian, Chengyuan
Xing, Shuo
Zhou, Yi
Crawford, Victoria
contents In this paper, we study cooperative multi-agent reinforcement learning (MARL) where the joint reward exhibits submodularity, which is a natural property capturing diminishing marginal returns when adding agents to a team. Unlike standard MARL with additive rewards, submodular rewards model realistic scenarios where agent contributions overlap (e.g., multi-drone surveillance, collaborative exploration). We provide the first formal framework for this setting and develop algorithms with provable guarantees on sample efficiency and regret bound. For known dynamics, our greedy policy optimization achieves a $1/2$-approximation with polynomial complexity in the number of agents $K$, overcoming the exponential curse of dimensionality inherent in joint policy optimization. For unknown dynamics, we propose a UCB-based learning algorithm achieving a $1/2$-regret of $O(H^2KS\sqrt{AT})$ over $T$ episodes.
format Preprint
id arxiv_https___arxiv_org_abs_2603_06810
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Multi-Agent Reinforcement Learning with Submodular Reward
Chen, Wenjing
Qian, Chengyuan
Xing, Shuo
Zhou, Yi
Crawford, Victoria
Machine Learning
Data Structures and Algorithms
In this paper, we study cooperative multi-agent reinforcement learning (MARL) where the joint reward exhibits submodularity, which is a natural property capturing diminishing marginal returns when adding agents to a team. Unlike standard MARL with additive rewards, submodular rewards model realistic scenarios where agent contributions overlap (e.g., multi-drone surveillance, collaborative exploration). We provide the first formal framework for this setting and develop algorithms with provable guarantees on sample efficiency and regret bound. For known dynamics, our greedy policy optimization achieves a $1/2$-approximation with polynomial complexity in the number of agents $K$, overcoming the exponential curse of dimensionality inherent in joint policy optimization. For unknown dynamics, we propose a UCB-based learning algorithm achieving a $1/2$-regret of $O(H^2KS\sqrt{AT})$ over $T$ episodes.
title Multi-Agent Reinforcement Learning with Submodular Reward
topic Machine Learning
Data Structures and Algorithms
url https://arxiv.org/abs/2603.06810