Research map / Multi-agent systems and games / Cooperative multi-agent RL
Dual Self-Awareness Value Decomposition Framework without Individual Global Max for Cooperative MARL
Zhiwei Xu, Bin Zhang, Dapeng Li, Guangchong Zhou, Zeren Zhang, Guoliang Fan
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 |
|---|---|---|---|
| increase number of sampled joint actions M from 1 (ablation) | DAVE · SMAC (three scenarios) | median test win rate | Improved |
| increase M beyond 10 (to 50, 100) (ablation) | DAVE · SMAC (three scenarios) | median test win rate | No clear effect |
| add anti-ego exploration (lambda_init > 0) in easy scenarios (ablation) | DAVE · SMAC easy scenarios | training speed | Hurt |
| use high lambda_init in hard scenarios (ablation) | DAVE · SMAC hard scenarios | median test win rate | Improved |
| add a monotonic mixing network to DAVE (ablation) | DAVE · Matrix games | convergence to optimum | Hurt |
| add a monotonic mixing network to DAVE (ablation) | DAVE · SMAC | median test win rate | Hurt |
| add DAVE dual self-awareness framework to QMIX | QMIX · SMACv2 | median win rate | Improved |
| remove absolute functions from QMIX (IGM-Free QMIX) (ablation) | QMIX · SMACv2 | median win rate | Hurt |
| add DAVE framework to QPLEX | QPLEX · SMACv2 | median win rate | Improved |
| add DAVE framework to QMIX | QMIX · Multi-Agent MuJoCo | return | Improved |
| add action information to QMIX mixing network (ablation) | QMIX · Matrix games | convergence to optimum | No clear effect |
| replace IGM-based policy-gradient critics with MADDPG/MAPPO (no IGM) | MADDPG · Matrix Game I k=0 | proportion converging to optimal 8: 20% | Hurt |
Closest papers by meaning
- Retaining Suboptimal Actions to Follow Shifting Optima in Multi-Agent Reinforcement Learning (ICLR 2026 · similarity 0.86)
- Value Function Decomposition for Iterative Design of Reinforcement Learning Agents (NeurIPS 2022 · similarity 0.86)
- Deconfounded Value Decomposition for Multi-Agent Reinforcement Learning (ICML 2022 · similarity 0.86)
- Rethinking Individual Global Max in Cooperative Multi-Agent Reinforcement Learning (NeurIPS 2022 · similarity 0.86)
- Towards Understanding Cooperative Multi-Agent Q-Learning with Value Factorization (NeurIPS 2021 · similarity 0.86)
- Optimistic Value Instructors for Cooperative Multi-Agent Reinforcement Learning (AAAI 2024 · similarity 0.85)
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.
