<ul><li>Recent legislative regulations have emphasized the need for accountable and transparent artificial intelligence systems.</li><li>However, the lack of standardized criteria to validate explanation methodologies is hindering the development of trustworthy systems.</li><li>This study focuses on the robustness of explanations in Explainable Artificial Intelligence (XAI) to ensure trust in both the system and the provided explanation.</li><li>The authors propose a novel approach to analyze the robustness of neural network explanations and present an ensemble method to aggregate various explanations.</li></ul>

When Can You Trust Your Explanations? A Robustness Analysis on Feature Importances

Discover more