04:00
2026-08-26
arxiv.org
artificial-intelligence
A Formal Methodological Framework for Auditing Robustness and Fidelity in Explainable AI: From Application to Trust Certification
A new arXiv paper (2608.23817v1) proposes an auditing protocol that measures robustness and fidelity of post-hoc explainers like SHAP and LIME, combining them into a single Trust Score. Testing on a M…