FACMAC: Factored Multi-Agent Centralised Policy Gradients

Bei Peng, Tabish Rashid, Christian A. Schroeder de Witt, Pierre-Alexandre Kamienny, Philip H.S. Torr, J.W. Böhmer, Shimon Whiteson

Research output: Chapter in Book/Conference proceedings/Edited volumeConference contributionScientificpeer-review

Abstract

We propose FACtored Multi-Agent Centralised policy gradients (FACMAC), a new method for cooperative multi-agent reinforcement learning in both discrete and continuous action spaces. Like MADDPG, a popular multi-agent actor-critic method, our approach uses deep deterministic policy gradients to learn policies. However, FACMAC learns a centralised but factored critic, which combines per-agent utilities into the joint action-value function via a non-linear monotonic function, as in QMIX, a popular multi-agent Q-learning algorithm. However, unlike QMIX, there are no inherent constraints on factoring the critic. We thus also employ a nonmonotonic factorisation and empirically demonstrate that its increased representational capacity allows it to solve some tasks that cannot be solved with monolithic, or monotonically factored critics. In addition, FACMAC uses a centralised policy gradient estimator that optimises over the entire joint action space, rather than optimising over each agent’s action space separately as in MADDPG. This allows for more coordinated policy changes and fully reaps the benefits of a centralised critic. We evaluate FACMAC on variants of the multi-agent particle environments, a novel multi-agent MuJoCo benchmark, and a challenging set of StarCraft II micromanagement tasks. Empirical results demonstrate FACMAC’s superior performance over MADDPG and other baselines on all three domains.
Original languageEnglish
Title of host publicationAdvances in Neural Information Processing Systems 34 (NeurIPS 2021)
Subtitle of host publicationNeurIPS Proceedings
EditorsM. Ranzato, A. Beygelzimer, Y. Dauphin, P.S. Liang , J. Wortman Vaughan
Number of pages14
Publication statusPublished - 2021
Event35th Conference on Neural Information Processing Systems (Virtual) -
Duration: 6 Dec 202114 Dec 2021
Conference number: 35th
https://nips.cc/Conferences/2021

Conference

Conference35th Conference on Neural Information Processing Systems (Virtual)
Abbreviated titleNeurIPS 2021
Period6/12/2114/12/21
Internet address

Fingerprint

Dive into the research topics of 'FACMAC: Factored Multi-Agent Centralised Policy Gradients'. Together they form a unique fingerprint.

Cite this