Guillaume Garrigos
I am Maître de Conférences at the Université Paris 7 Diderot, member of the Laboratoire de Probabilités, Statistique et Modélisation (Team Statistique, données, algorithmes).
Formerly, I used to be a postdoc researcher at the École Normale Supérieure (DMA), member of the NORIA (Numerical Optimal tRansport for ImAging) research group. Before that, I had a two-years postdoc position at the Laboratory for Computational and Statistical Learning, a joint lab between the IIT and MIT. Even before that I did a PhD in Optimization between the Université de Montpellier and the Universidad Santa Maria in Valparaiso, Chile.
My research essentially focuses on dynamical systems associated to optimization problems. In particular, I am interested in algorithms and regularization methods for nonsmooth (eventually nonconvex) problems arising in signal/image processing and machine learning. I am particularly interested in the interplay between the modelization of a learning/inverse problem as an optimization problem, and its resolution. My papers are here.
News
-
Nov282019
Tutorial
Together with Lénaïc Chizat I gave a tutorial on Optimization for Nonconvex Optimization within the SMILE seminar. You can find the (non-animated) slides of my talk here.
-
Nov192019
Machine Learning And Inverse Problems Conference
Together with Luca Calatroni I am organizing the workshop Regularisation for Inverse Problems and Machine Learning. See the website of the event here for more information.
-
Mar062019
Accepted Paper
Our paper Thresholding gradient methods in Hilbert spaces: support identification and linear convergence, written with L. Rosasco and S. Villa, has been accepted in ESAIM: Control, Optimisation and Calculus of Variations. Preprint available on arXiv:1712.00357.
-
Feb212019
Science Communication
I did a presentation to a group of middle-schoolers about the recent success of AI, its limits and the societal challenges which awaits us. Slides available here (in French).
-
Feb182019
Seminar
I gave a talk about Model Consistency for Learning with Mirror-Stratifiable Regularizers at the Séminaire d’Analyse non linéaire et Optimisation of Université d’Avignon, and the Séminaire Parisien de Statistique.
-
Dec242018
Accepted Paper
Our paper Model Consistency for Learning with Mirror-Stratifiable Regularizers, written with J. Fadili, J. Malick and G. Peyré , has been accepted at AISTATS. Preprint on arXiv:1803.08381.
-
Sep012018
New Permanent Position
I joined the University Paris Diderot where I integrated the Laboratoire de Probabilités, Statistique et Modélisation as a Maître de Conférence.
-
Jul062018
Keep Calm And Sketch All You Can
I organized, together with Robert Gower, a session at ISMP titled "A sketchy session: Dimensionality reduction tools for learning", featuring Nicolas Keriven, Arthur Mensch and Alessandro Rudi.