Interpretable Failure Analysis in Multi-Agent Reinforcement Learning Systems

Risal Shahriar Shefin, Debashis Gupta, Thai Le +1

Multi-Agent Reinforcement Learning (MARL) is increasingly deployed in safety-critical domains, yet methods for interpretable failure detection and attribution remain underdeveloped…