Description
Date depot: 27 août 2021
Titre: Integrating deep learning and physics for dynamical systems with applications to earth science
Directeur de thèse:
Patrick GALLINARI (ISIR (EDITE))
Encadrant :
Nicolas BASKIOTIS (ISIR (EDITE))
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Intelligence artificielle
Resumé: The thesis will explore developments in physico-statistical systems aimed at learning complex dynamics characterizing physical processes in earth science. Two main topics are investigated. The first one concerns combining physics and deep learning with two objectives: complement available physical background and emulate complex dynamics with reduced complexity models. The second one deals with learning the dynamics with scale free methods or developing methods able to handle the dynamics at multiple scales. Applications will be in the domains of earth science and climate modeling.
Résumé dans une autre langue: The thesis will explore developments in physico-statistical systems aimed at learning complex dynamics characterizing physical processes in earth science. Two main topics are investigated. The first one concerns combining physics and deep learning with two objectives: complement available physical background and emulate complex dynamics with reduced complexity models. The second one deals with learning the dynamics with scale free methods or developing methods able to handle the dynamics at multiple scales. Applications will be in the domains of earth science and climate modeling.
Doctorant.e: Wang Thomas Xiaoyuan