FOAM - First-Order Accelerated Methods for Machine Learning

Project Summary

FOAM is a bilateral research group led by Alexandre D'aspremont from the SIERRA team of Inria Paris in France and, in Chile, by Cristóbal Guzmán, a researcher at Pontificia Universidad Católica de Chile.

The Franco-Chilean team investigates First-Order Accelerated Methods for machine learning, including optimal convergence rates in statistical learning, convex stochastic optimization, variational inequalities, and fixed-point problems under performance estimation.

The results of this project will have applications in multiple areas, including astronomical data analysis, seriation problems, and training of generative adversarial models (GANs).

In France

• Alexandre d'Aspremont, researcher, FOAM coordinator, École Normale Supérieure, CNRS, project-team SIERRA, Inria Paris Centre, Inria

• Adrien Taylor, researcher, project-team SIERRA, Inria Paris Centre, Inria

• Alessandro Rudi, researcher, project-team SIERRA, Inria Paris Centre, Inria

• Mathieu Barré, PhD student, project-team SIERRA, Inria Paris Centre, Inria

• Manon Romain, PhD student, project-team SIERRA, Inria Paris Centre, Inria

• Clément Lezane, PhD student, project-team SIERRA, Inria Paris Centre, Inria

• Max Kramkimel, undergraduate student, École Polytechnique

In Chile

• Cristóbal Guzmán, researcher, FOAM coordinator, Pontificia Universidad Católica de Chile

• Carlos Sing-Long, researcher, Pontificia Universidad Católica de Chile

• Juan Pablo Flores, MSc student, Pontificia Universidad Católica de Chile

• Roberto Cominetti, researcher, Universidad Adolfo Ibáñez

• Juan Pablo Contreras, postdoctoral researcher, Universidad Católica del Norte

• Luis Martí, researcher, Inria Chile

• Nayat Sánchez-Pi, researcher, Inria Chile

Former members

• Rodolfo Palma, researcher, Inria Chile

• Santiago Armstrong, MSc student, Pontificia Universidad Católica de Chile

• Radu-Alexandru Dragomir, PhD student, equipo-proyecto SIERRA, Inria

• Patricio Ulloa, MSc student, Pontificia Universidad Católica de Chile

• Gregoire Mialon, PhD student, project-team SIERRA, Inria Paris Centre, Inria

• INRIA ASSOCIATE TEAM

• FOAM: First-Order Accelerated Methods for Machine Learning