Stéphane Mallat will open the Data Science Summer School 2019
This Collège de France professor is a key figure in the data science landscape and for good reason, he was appointed Chair of Data Sciences at the Collège de France in 2017, not to mention the many prizes and awards of his peers.
Since 2008, Stéphane Mallat has been studying the mathematical properties of learning algorithms and deep neural networks for data including a large number of variables. More particularly, his research concerns both the recognition of images or sounds and the prediction of physical properties or the analysis of texts. We will have the opportunity to share with him this opening moment with his lecture: Mathematical mysteries of deep neural networks
Abstract: Deep neural networks obtain impressive results for image, sound and language recognition or to adress complex problems in physics. They are partly responsible of the renewal of artificial intelligence. Yet, we do not understand why they can work so well and why they fail sometimes, which raises many problems of robustness and explainability.
Recognizing or classifying data amounts to approximate phenomena which depend on a very large number of variables. The combinatorial explosion of possibilities makes it potentially impossible to solve. One can learn from data only if the problem is highly structured. Deep neural networks appear to take advantage of these unknown structures. Understanding this "architecture of complexity" involves many branches of mathematics and is related to open questions in physics. I will discuss some approaches and show applications.