Interpretable Failure Analysis In Multi-agent Reinforcement Learning Systems

Abstract

Multi-Agent Reinforcement Learning (MARL) is increasingly deployed in safety-critical domains, yet methods for interpretable failure detection and attribution remain underdeveloped. We introduce a two-stage gradient-based framework that provides interpretable diagnostics for three critical failure analysis tasks: (1) detecting the true initial failure source (Patient-0); (2) validating why non-attacked agents may be flagged first due to domino effects; and (3) tracing how failures propagate through learned coordination pathways. Stage 1 performs interpretable per-agent failure detection via Taylor-remainder analysis of policy-gradient costs, declaring an initial Patient-0 candidate at the first threshold crossing. Stage 2 provides validation through geometric analysis of critic derivatives-first-order sensitivity and directional second-order curvature aggregated over causal windows to construct interpretable contagion graphs. This approach explains "downstream-first" detection anomalie

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keyshefin2026interpretable

Related papers

• Vulnerable Agent Identification In Large-scale Multi-agent Reinforcement Learning (2025)0.00
• Tackling Uncertainties In Multi-agent Reinforcement Learning Through Integration Of Agent Termination Dynamics (2025)2.26
• Multi-agent Reinforcement Learning In Stochastic Networked Systems (2020)0.00
• Fault Tolerant Multi-agent Learning With Adversarial Budget Constraints (2025)0.00
• Understanding Individual Decision-making In Multi-agent Reinforcement Learning: A Dynamical Systems Approach (2025)0.00
• A Principle Of Targeted Intervention For Multi-agent Reinforcement Learning (2025)0.00
• Collaborative Adaptation For Recovery From Unforeseen Malfunctions In Discrete And Continuous MARL Domains (2024)3.58
• Safe Multi-agent Reinforcement Learning With Convergence To Generalized Nash Equilibrium (2024)0.00