Research map / Multi-agent systems and games / Cooperative multi-agent RL
Multi-agent Dynamic Algorithm Configuration
Ke Xue, Jiacheng Xu, Lei Yuan, Miqing Li, Chao Qian, Zongzhang Zhang, Yang Yu
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 |
|---|---|---|---|
| Use multi-agent VDN to jointly tune four heterogeneous MOEA/D hyperparameter types rather than the single-agent DQN configuration | MOEA/D · MaMo test problems | IGD: DQN is significantly worse than MA-DAC on all the problems | Improved |
| Use multi-agent VDN to jointly tune four heterogeneous MOEA/D hyperparameter types rather than four independent UCB agents | MOEA/D · MaMo train and test problems | IGD: Train: 0/9/0; Test: 0/15/0 ($+/$-/$\approx$ for MA-UCB) | Improved |
| Add adaptive weight tuning (ablation) | MA-DAC (M) · MaMo train problems | IGD: 4.656E-02 (3.80E-04) -> 3.839E-02 (5.35E-04) | Improved |
| Add neighborhood-size tuning (ablation) | MA-DAC (M) · MaMo train problems | IGD: 3.914E-02 (8.43E-04) -> 3.839E-02 (5.35E-04) | Improved |
| Add reproduction-operator-type tuning (ablation) | MA-DAC (M) · MaMo train problems | IGD: 3.935E-02 (6.72E-04) -> 3.839E-02 (5.35E-04) | Improved |
| Add reproduction-operator-parameter tuning (ablation) | MA-DAC (M) · MaMo train problems | IGD: 3.919E-02 (5.91E-04) -> 3.839E-02 (5.35E-04) | Improved |
| Train a separate policy for each objective-count class rather than mix objective counts in training (ablation) | MA-DAC · MaMo test problems, 3/5/7 objectives | average rank: 1.4; 2.6; 2.6 -> 1.8; 2.8; 3.4 (MA-DAC (M) vs MA-DAC (3)) | Mixed |
| Replace the proposed reward with R1 (ablation) | MA-DAC · MaMo training problems | average rank: 2.67 | Hurt |
| Replace the proposed reward with R2 (ablation) | MA-DAC · MaMo training problems | average rank: 3.11 | No clear effect |
| Replace the proposed reward with R3 (ablation) | MA-DAC · MaMo training problems | average rank: 2.11 | No clear effect |
Closest papers by meaning
- Sequential Multi-Agent Dynamic Algorithm Configuration (NeurIPS 2025 · similarity 0.94)
- Graph-Supported Dynamic Algorithm Configuration for Multi-Objective Combinatorial Optimization (ICML 2025 · similarity 0.88)
- Controlling Behavioral Diversity in Multi-Agent Reinforcement Learning (ICML 2024 · similarity 0.87)
- ConfigX: Modular Configuration for Evolutionary Algorithms via Multitask Reinforcement Learning (AAAI 2025 · similarity 0.87)
- HyperMARL: Adaptive Hypernetworks for Multi-Agent RL (NeurIPS 2025 · similarity 0.87)
- Efficient Multi-agent Reinforcement Learning by Planning (ICLR 2024 · similarity 0.86)
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.
