Reproduce “Context-Aware Bayesian Network Actor-Critic Methods for Cooperative Multi-Agent Reinforcement Learning” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Dingyang Chen, Qi Zhang

The authors introduce a context-aware multi-agent policy that lets each agent condition its action on selected teammates’ actions through a learned directed acyclic graph. Local policies use each agent’s observation and its graph parents’ actions, while training later encourages an empty graph. In two tested StarCraft II scenarios, this schedule finishes above independently trained policies; this is not a general guarantee.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.