Research map / Multi-agent systems and games / Cooperative multi-agent RL
Coordination Connectivity: Shared Initialization Shapes the Joint-Policy Landscape in MARL
Yixiang Fan, Zhiqiang Pu, Hao Ma, Dongmin Li, Weihao Zhang, Yinxing Dai, Shijie Zheng, Jingjing Huang
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Studies whether linear interpolation between joint policies in cooperative MARL preserves team performance, finding shared initialization shapes the joint-policy landscape (coordination connectivity diagnostic).
Problem
Mode connectivity is known for single neural networks, but it is unclear whether high-performing cooperative multi-agent teams (sets of decentralized policies) are connected by low-loss paths in parameter space.
Approach
Introduces coordination connectivity, a diagnostic that tests whether linear interpolation between the actor parameters of two joint policies preserves coordinated team performance, and shows shared initialization shapes this landscape.
Key results
- Shared initialization shapes the joint-policy landscape in cooperative MARL, per the abstract's central claim (quantitative details not stated in the provided abstract).
Experimental setup
Metrics: team performance along interpolation path
Builds on: mode connectivity
Closest papers by meaning
- Multi-agent In-context Coordination via Decentralized Memory Retrieval (AAAI 2026 · similarity 0.90)
- Revisiting Some Common Practices in Cooperative Multi-Agent Reinforcement Learning (ICML 2022 · similarity 0.90)
- HyperMARL: Adaptive Hypernetworks for Multi-Agent RL (NeurIPS 2025 · similarity 0.89)
- Enhancing Cooperative Multi-Agent Reinforcement Learning with State Modelling and Adversarial Exploration (ICML 2025 · similarity 0.89)
- Cooperative Policy Agreement: Learning Diverse Policy for Offline MARL (AAAI 2025 · similarity 0.88)
- Investigating Relational State Abstraction in Collaborative MARL (AAAI 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
