Date: 2nd July 2019
Speaker: Patrick Gallinari, Professor at Sorbonne University & Senior Research Scientist at Criteo AI Lab, Paris, France.
Physics and Machine Learning offer two different perspectives for the modeling of complex natural phenomena. The former relies on prior knowledge of the underlying processes, often accumulated over decades by generation of scientists. The latter proposes an agnostic approach based on the availability of large amounts of data and of generic learning mechanisms. The convergence of these two approaches is one of the great scientific challenges for the next decades. In this presentation I will introduce some preliminary work exploring this direction focusing on the modeling of spatio-temporal processes, typically problems from fluid dynamics. I will present two use cases, one is an example of how to incorporate prior physical knowledge in a deep learning system, and the second one shows how to learn dynamical systems obeying general evolution equations using a deep learning approach.
Patrick Gallinari is a professor at Sorbonne University and researcher at Criteo AI Lab – Paris. His research focuses on statistical learning and deep learning with applications in different fields such as semantic data processing or complex data analysis. Together with colleagues from Sorbonne University, he started to explore in 2017 the development of physico-statistical systems combining the model based approaches of physics and the data processing approaches of statistical learning. He has been a pioneer in the development of Neural Networks in the 90ies. He leads a team whose central theme is Statistical Learning and Deep Learning (https://mlia.lip6.fr). He was director of the Paris 6 computer lab (LIP6) for 9 years (2005 to 2013).