Les séminaires du CENTAL ont pour but de réunir des enseignants, des étudiants et des chercheurs (du monde académique ou de l'industrie) intéressés par le traitement automatique de langues. Les séminaires sont gratuits et ouverts à tous et ont généralement lieu le mardi de 11h à 12h dans les locaux du Collège Léon Dupriez. Si vous souhaitez être informé par courrier électronique des séminaires que nous organisons et des actualités du CENTAL, nous vous proposons de vous inscrire à la liste de diffusion du CENTAL en indiquant votre adresse électronique dans le formulaire.
Organisation 2021-22
Calendrier 2021-22
12 octobre 2021 — Adrien Bibal
Interpretability and Explainability in Machine Learning
Adrien Bibal, Postdoctoral Researcher in Machine Learning and NLP at UCLouvain
Abstract :
Machine learning models are becoming more and more complex for the sake of performance. However, in many situations, the way in which the model is computed must be somewhat transparent. For instance, in some countries, the reasons for credit denial must legally be provided. Furthermore, in science, it is often not the predictive performance of the model that is sought, but the knowledge that can be extracted from it. Interpretability is a property of models that characterizes the degree to which models are understandable by their users, while explainability is the capacity the model to be explained, if it is not interpretable. In this seminar, we will introduce these two concepts, as well as the issues related to their implementation and evaluation.
Références :
Bibal, A., & Frenay, B. (2016). Interpretability of machine learning models and representations: an introduction. In European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (pp. 77-82).
Lipton, Z. C. (2018). The mythos of model interpretability: in machine learning, the concept of interpretability is both important and slippery. Queue, 16(3), 31-57.
Guidotti, R., Monreale, A., Ruggieri, S., Turini, F., Giannotti, F., & Pedreschi, D. (2018). A survey of methods for explaining black box models. ACM Computing Surveys, 51(5), 1-42.
Bibal, A., Lognoul, M., De Streel, A., & Frénay, B. (2021). Legal requirements on explainability in machine learning. Artificial Intelligence and Law, 29(2), 149-169.
Rudin, C. (2019). Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature Machine Intelligence, 1(5), 206-215.