Valentine Huré
I am a third year PhD student working with Amélie Lambert and Zacharie Alès. Our research focuses on the construction of optimal classification trees using mathematical programming. Initially, we focused on improving state-of-the-art mathematical programs. By combining this work with a new algorithm for learning classification tree parameters, we have significantly reduced computing time while maintaining or improving learning performances. We are now in the process of developing a method to scale the construction of optimal classification trees. Approaches based on the direct resolution of mathematical programs are unable to build high-performance trees beyond a few thousand data sets. Our method consists in clustering data to solve smaller mathematical programs.
2024
Articles de revue
- New optimization models for optimal classification trees. In Computers and Operations Research, 164: 106515, 2024. doi www
Non publié
- Clustering data for the Optimal Classication Tree Problem. , working paper or preprint. www
2022
Articles de conférence
- Modélisations d'arbres de décision optimaux. In 23ème congrès annuel de la société Franc caise de Recherche Opérationnelle et d'Aide `a la Décision (ROADEF 22), Lyon, France, 2022. www
- New optimization models for optimal classification trees. In 32nd European Conference on Operational Research (EURO 2022), Espoo, Finland, 2022. www