BayesG learns dynamic sparse interaction structures via Bayesian variational inference on ego-graphs for decentralized networked MARL, showing better performance than baselines on traffic control with up to 167 agents.
Each agent’s state evolution depends on its immediate neighbors’ actions, not the global joint action of all agents
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.MA 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Bayesian Ego-graph Inference for Networked Multi-Agent Reinforcement Learning
BayesG learns dynamic sparse interaction structures via Bayesian variational inference on ego-graphs for decentralized networked MARL, showing better performance than baselines on traffic control with up to 167 agents.