I’m pleased to announce that I’ve successfully defended my Habilitation à Diriger des Recherches (HDR), an important step in the French university system. Unlike a PhD which demonstrates the ability to conduct research, the HDR certifies one’s capability to supervise doctoral research and direct research programs. I’ve taken part in co-advising six doctoral theses and now, with the HDR, can direct research.
The HDR process involved submitting a manuscript that synthesizes my research contributions and vision, followed by a public defense before a committee of established researchers. My HDR manuscript is here and a short abstract from the defense follows. I’m grateful to my committee members for their insightful questions and feedback, and to all my colleagues, students, and collaborators who worked with me along the way.
Methods in Exploration and Interpretability for Automating Discovery
Can algorithms help drive scientific discovery? My research explores this question through two complementary approaches: machine learning algorithms that systematically explore novel and creative solutions, and optimization methods that ensure these discoveries remain interpretable to human researchers. Through case studies in climate science, I demonstrate how these techniques can accelerate our understanding of complex phenomena like shoreline evolution and El Niño prediction. Finally, I propose a framework for how artificial intelligence can augment, rather than replace, human scientific reasoning.
Jury:
- Claire Monteleoni, Rapporteuse
- Stéphane Doncieux, Rapporteur
- Bing Xue, Rapporteuse
- Marc Schoenauer, Examinateur
- Sylvain Cussat-Blanc, Référent
- Frédéric Dehais, Président