Pantheon SEMPARIS Le serveur des séminaires parisiens Paris

Status Confirmed
Seminar Series MATH-IHES
Subjects hep-th
Date Monday 25 November 2019
Time 10:30
Institute IHES
Seminar Room Centre de conférences Marilyn et James Simons
Speaker's Last Name Mallat
Speaker's First Name Stéphane
Speaker's Email Address
Speaker's Institution Collège de France
Title Multiscale Models for Image Classification and Physics with Deep Networks
Abstract Approximating high-dimensional functionals with low-dimensional models is a central issue of machine learning, image processing, physics and mathematics. Deep convolutional networks are able to approximate such functionals over a wide range of applications. This talk shows that these computational architectures take advantage of scale separation, symmetries and sparse representations. We introduce simplified architectures which can be anlalyzed mathematically. Scale separations is performed with wavelets and scale interactions are captured through phase coherence. We show applications to image classificaiton and generation as well as regression of quantum molecular energies and modelization of turbulence flows.
arXiv Preprint Number
Comments Nokia-IHES Workshop

To Generate a poster for this seminar : [ Postscript | PDF ]

[ Annonces ]    [ Abonnements ]    [ Archive ]    [ Aide ]    [ ]
[ English version ]