Supervision : Christophe MARSALA

Co-supervision : LESOT Marie-Jeanne

Defining Differential Explanations: Understanding the Dynamic of Changes in Machine Learning Models

The field of this Ph.D. research is AI/ML interpretability (also called explainable AI or XAI), which is the study of explaining Machine Learning (ML) models. This cross-disciplinary field encompasses computer science, mathematics, and human-computer interface. The goal of the research is to investigate the concept of differential explainability, which refers to the explanation of the differences or evolution between successive versions of an ML model. These successive versions may be necessary for improving the model's performance or adapting the model to a changing context. Differential explainability is important in these situations because it allows for the identification of potential flaws and the explanation of differences between successive versions of a model.