Keywords : Machine Learning, Neural Networks, Deep Learning, Natural Language Processing, Physics Based Deep Learning.
I am a Professor of Computer Science at Sorbonne University (formerly Pierre et Marie Curie University) in Paris. My research focuses on statistical learning, with applications in areas such as semantic data analysis and complex data modeling. I discovered the field of machine learning in the mid-1980s, when I began working on neural networks—an emerging research topic at the time. I was among the pioneers of this domain in France and Europe and actively contributed to the development of the European research community in neural networks for nearly a decade.
As interest in neural networks temporarily declined, I shifted my research focus to other machine learning topics. During this period, I explored various application domains, including information retrieval, natural language processing, social data analysis, and user modeling. From a methodological perspective, my work concentrated on machine learning for structured data such as sequences, tables and graphs. When neural networks and deep learning regained prominence in the mid-2000s, thanks to its background, the MLIA team was among the first in France to fully reorient its research activities toward this renewed paradigm.
In 2017, I initiated a new research direction in scientific AI, with applications in domains such as Earth sciences and climate modeling. In 2020, I was awarded a national AI Chair on this topic. Since then, scientific AI—and more specifically data-driven modeling of spatio-temporal dynamics in physics—has become my primary research focus (see AI4Science)
I have led the Machine Learning and Information Access (MLIA) team at Sorbonne University for several years. I also served as Director of the Computer Science Laboratory at Sorbonne University (LIP6) from 2005 to 2013, following six years as Vice-Director. In addition, I was Vice-Chair of the Scientific Committee of the Faculty of Engineering at Sorbonne University from 2005 to 2021.