Mathematical Mysteries of Deep Neural Networks
September 7, 2022 at 5 pm CET
Deep neural networks have spectacular applications to classification, regression or generation of images, sounds, languages and most physical data. They are responsible for the renewal of artificial intelligence. Yet, we mostly do not understand the underlying mathematics, allowing them to avoid the curse of dimensionality. By taking examples and inspiration from statistical physics, this talk will emphasize the role of concentration phenomena, multiscale interactions and invariants. Applications will be shown in physics and image classification problems.
Stéphane Mallat was born in France in 1962. He is professor at the Collège de France on the chair of Data Sciences. After his PhD at University of Pennsylvania in 1988, he became Professor of computer science and mathematics at the Courant Institute of NYU in New York for 10 years, then at Ecole Polytechnique and Ecole Normale Supérieure in Paris. He also was the co-founder and CEO of a semiconductor start-up company from 2001 to 2007.
Stéphane Mallat is a member of the French Academy of sciences, of the Academy of Technologies and a foreign member of the US National Academy of Engineering. His research interests include machine learning, signal processing and harmonic analysis. He developed the multi-resolution wavelet theory with applications to image processing and sparse representation in dictionaries with matching pursuits. He now works on mathematical understanding of deep neural networks, and their applications.