AdaFair-MARL Framework Enforces Adaptive Fairness in Multi-Agent RL
Researchers propose AdaFair-MARL, a constrained cooperative multi-agent reinforcement learning framework that enforces workload fairness as an explicit constraint. Unlike fixed penalties or heuristic reward-shaping, AdaFair-MARL uses a primal-dual update to maintain balanced agent contributions while optimizing team performance. The approach addresses training instability and conflicting incentives in heterogeneous multi-agent systems. The paper is available on arXiv under identifier 2511.14135.
Key facts
- AdaFair-MARL formulates workload fairness as an explicit constraint.
- Core algorithmic component is a primal-dual update.
- Targets heterogeneous multi-agent systems with shared objectives.
- Overcomes limitations of fixed fairness penalties and heuristic reward-shaping.
- Published on arXiv with ID 2511.14135.
- Announce type: replace-cross.
Entities
Institutions
- arXiv