Research map / Multi-agent systems and games / Cooperative multi-agent RL
Scalable Multi-Agent Reinforcement Learning through Intelligent Information Aggregation
Siddharth Nayak, Kenneth Choi, Wenqi Ding, Sydney Dolan, Karthik Gopalakrishnan, Hamsa Balakrishnan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Aggregate local agent and entity information with an attention-based graph network | MAPPO for multi-agent navigation · Target, 3, 7, and 10 agents | Reward: 173.13 -> 205.24; 327.39 -> 399.01; 366.81 -> 429.14 | Improved |
| Use the local-information graph aggregation architecture instead of RMAPPO with global information | MAPPO for multi-agent navigation · Target, 3, 7, and 10 agents | Success rate: 96 -> 100; 88 -> 100; 79 -> 100 | Improved |
| Add the graph information aggregation module to the centralized critic rather than concatenating agent states (ablation) | InforMARL · Target, 3-agent training | Training performance | No clear effect |
| Add the graph information aggregation module to the centralized critic rather than concatenating agent states (ablation) | InforMARL · Target, variable numbers of entities | Critic parameter count | Improved |
| Increase sensing radius from small values (ρ=0.1 or ρ=0.2) to larger values (ρ=0.5, ρ=1, ρ=2, or ρ=5) (ablation) | InforMARL · Target task | Reward | Improved |
| Increase sensing radius from ρ=0.5 to ρ=5 (ablation) | InforMARL · Target task | Reward | No clear effect |
| Use one-hot encoding instead of an embedding layer for entity types (ablation) | InforMARL · Multi-agent navigation | Performance | No clear effect |
Closest papers by meaning
- Multi-Agent Reinforcement Learning in Stochastic Networked Systems (NeurIPS 2021 · similarity 0.89)
- Multi-Agent Common Knowledge Reinforcement Learning (NeurIPS 2019 · similarity 0.88)
- Bayesian Ego-graph inference for Networked Multi-Agent Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- E-MAPP: Efficient Multi-Agent Reinforcement Learning with Parallel Program Guidance (NeurIPS 2022 · similarity 0.88)
- Scalable Multi-Agent Reinforcement Learning for Networked Systems with Average Reward (NeurIPS 2020 · similarity 0.88)
- Multi-Agent Meta-Reinforcement Learning: Sharper Convergence Rates with Task Similarity (NeurIPS 2023 · 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.
