New Transformer Framework Bridges MARL to SARL for Enhanced Coordination

Researchers have developed the Consensus Multi-Agent Transformer (CMAT), a novel framework designed to bridge cooperative multi-agent reinforcement learning (MARL) with hierarchical single-agent reinforcement learning (SARL). CMAT processes large joint observation spaces using a Transformer encoder and addresses extensive joint action spaces through a hierarchical decision-making mechanism. This mechanism autoregressively generates a high-level consensus vector, enabling agents to reach agreement on strategies in latent space and simultaneously generate order-independent actions. Experiments on StarCraft II, Multi-Agent MuJoCo, and Google Research Football demonstrate CMAT's superior performance compared to existing centralized and sequential MARL methods. AI

IMPACT Introduces a new method for improving coordination and training stability in multi-agent reinforcement learning systems.

RANK_REASON Academic paper detailing a new framework and experimental results. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• CMAT
• Consensus Multi-Agent Transformer
• Google Research Football
• Multi-Agent MuJoCo
• MARL
• StarCraft II
• Transformer
• Zijian Zhao

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →