The DAAO (Data, Machine Learning, and Optimization) group of the GDR RO is pleased to announce the online tutorial of Thibaut Vidal on combinatorial optimization and interpretable machine learning on June 21, 2021 from 2 pm to 4 pm (CEST).
Title:
Combinatorial optimization and interpretable machine learning
Speaker:
Thibaut Vidal
CIRRELT, Département de Mathématiques et Génie Industriel, Ecole Polytechnique de Montréal
Département d'Informatique, Université Pontificale Catholique de Rio de Janeiro
Abstract:
The use of machine learning algorithms in finance, medicine, and many other domains can profoundly impact human lives. Consequently, extensive efforts have been made to improve machine learning pipelines, making them more accurate, robust, and interpretable. In this seminar, we explore the synergy between combinatorial optimization algorithms and the machine learning domain. We focus on tree ensembles (including random forests and gradient boosting), a popular family of models with good empirical performance which is often used as a more transparent replacement to neural networks. We discuss interpretation, explanation and simplification techniques for these models from combinatorial optimization lenses. We illustrate the use of dynamic programming algorithms for simplifying tree ensembles by constructing a single ---minimal-size--- decision tree that faithfully reproduces the decision function of a tree ensemble. Moreover, we show that the search for counterfactual explanations (i.e., small and plausible perturbations of the input that modify the classification outcome) can be cast as a mixed integer program and efficiently solved. We conclude the talk with other research perspectives connected to the application of combinational optimization techniques in interpretable machine learning.
Registration
Registrations are now closed.
Organizers
Jérôme Malick
Axel Parmentier