26 novembre 2026 à 14h
Abstract: The goal of eXplainable AI (XAI), as stated by DARPA in 2019, is « to provide users with explanations that enable them to understand the system’s overall strengths and weaknesses, convey an understanding of how it will behave in future or different situations, and perhaps permit users to correct the system’s mistakes. » In this talk, I will show how concepts and theories developed in knowledge representation (KR) can be leveraged to reach this goal in some contexts, while providing formal guarantees on the results obtained. In particular, I will focus on the computational interpretability of machine learning (ML) models and on the rectification of tree-based models.
Bio: Pierre Marquis is a Professor at Artois University and IUF Senior Research Fellow. His research activities are mainly concerned with knowledge representation and automated reasoning in a broad sense. For the past few years, his work has been more focused on eXplainable AI.
Amphi IRCICA, 50 avenue Halley, Haute Borne, Villeneuve d'Ascq
français
English