Research map / Multi-agent systems and games / Cooperative multi-agent RL
Sparse Topology-Aware Pairwise Scoring for Large-Scale Multi-Agent Reinforcement Learning
Zhibo Deng, Feng Liang, Yong Zhang, Xiaoxi Zhang, Xiping Hu
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 |
|---|---|---|---|
| plug SOPS into QMIX | QMIX · IMP Correlated, 50 agents | normalized discounted reward: 18.04 -> 45.59 | Improved |
| plug SOPS into QPLEX | QPLEX · IMP Correlated, 50 agents | normalized discounted reward: 20.10 -> 46.26 | Improved |
| plug SOPS into SHAQ | SHAQ · IMP Correlated, 50 agents | normalized discounted reward: -14.88 -> 32.41 | Improved |
| plug SOPS into SHAQ | SHAQ · IMP Uncorrelated, 50 agents | normalized discounted reward: -7.82 -> 15.46 | Improved |
| plug SOPS into QPLEX | QPLEX · IMP OWF, 50 agents | normalized discounted reward: 60.74 -> 62.93 | Improved |
| plug SOPS into QMIX | QMIX · IMP OWF, 50 agents | normalized discounted reward: 61.38 -> 65.46 | Improved |
| replace static exponential backbone EC-S with learned sparse selection (SOPS) | EC-S · IMP OWF, 100 agents | normalized discounted reward: 63.76 -> 66.40 | Improved |
| replace one-peer exponential schedule EC-O with SOPS | EC-O · IMP Correlated, 100 agents | normalized discounted reward: 18.85 -> 23.55 | Improved |
| SOPS under test-time message dropout p=0.5 versus no dropout | SOPS · Battle-100 win rate | win rate: 0.98 -> 0.74 | Hurt |
| EC-O under test-time message dropout p=0.5 versus no dropout | EC-O · Battle-100 win rate | win rate: 0.93 -> 0.46 | Hurt |
| use linear temperature annealing instead of exponential annealing (ablation) | SOPS · Battle 20 and 42 agents | win rate | Improved |
| use linear annealing instead of fixed Gumbel temperature (0.1, 0.5, 1.0, 2.0) (ablation) | SOPS · Battle 20 and 42 agents | win rate | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- QPLEX — QPLEX: Duplex Dueling Multi-Agent Q-Learning (ICLR 2021) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Exponential Topology-enabled Scalable Communication in Multi-agent Reinforcement Learning (ICLR 2025 · similarity 0.92)
- Bridging Training and Execution via Dynamic Directed Graph-Based Communication in Cooperative Multi-Agent Systems (AAAI 2025 · similarity 0.90)
- Bayesian Ego-graph inference for Networked Multi-Agent Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- Learning Efficient and Robust Multi-Agent Communication via Graph Information Bottleneck (AAAI 2024 · similarity 0.88)
- Graph Diffusion for Robust Multi-Agent Coordination (ICML 2025 · similarity 0.88)
- Multi-Agent Reinforcement Learning with Communication-Constrained Priors (NeurIPS 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.
