Patrick Gallinari

Professor at Sorbonne Université, Criteo AI Lab in Paris

Patrick Gallinari is a professor at Sorbonne University, affiliated with the ISIR laboratory (Institute of Intelligent Systems and Robotics), and a distinguished researcher at the Criteo AI Lab in Paris. A pioneer in the field of neural networks, his research focuses on statistical and deep learning with applications to semantic and complex data modeling. Since 2018, he has initiated research at the intersection of machine learning and physics, with a particular emphasis on spatio-temporal dynamics, and has contributed to seminal work in this field. He was awarded a national AI Chair (2020–2026) titled “Deep Learning for Physical Processes with Applications to Earth System Science,” and a second chair (2025–2031) from the PostGenAI@Paris cluster titled “Deep Learning for Science: Modeling Fluid Dynamics in Engineering and Climate Physics.”

https://pages.isir.upmc.fr/gallinari/

Title: AI4Science: From Equations to Learning Machines

For centuries, science has progressed through the explicit formulation of physical laws — like equations derived from first principles. The advent of data-driven modeling is now challenging this tradition and redefining the foundations of scientific inquiry.

This talk explores how AI for Science is transforming research across disciplines, moving machine learning from a mere computational tool to a core scientific paradigm. Deep learning, in particular, is enabling powerful empirical models that capture intricate physical dynamics and can generalize across diverse conditions.

The first part of the presentation briefly surveys striking examples from climate and weather forecasting, materials science, biology and drug discovery. The second part highlights key advances from our own group on modeling physical dynamical systems, emphasizing both the opportunities and the challenges of deploying machine learning in this field.