Florent Krzakala

Full Professor

École polytechnique fédérale de Lausanne

About me

Florent Krzakala is a full professor at École polytechnique fédérale de Lausanne in Switzerland. His research interests include Statistical Physics, Machine Learning, Statistics, Signal Processing, Computer Science and Computational Optics. He leads the IdePHIcs “Information, Learning and Physics” laboratory in the Physics and Engineering departments in EPFL. He is also the founder and scientific advisor of the startup Lighton.

mylabs

I am curently looking for Ph.D students and postdocs. Send me a mail!

Curriculum Vitae

Download Complete CV

Full Professor, Physics and EE, EPFL, Switzerland, Since 2020
Professor UPMC and Researcher at Ecole Normale Superieure, Paris, 2013-2020
Member of the Institut Universitaire de France, 2015-2020
Holder of a Prairie Institute AI Chair, 2019-2020
Member and Fellow of the ELLIS society, Since 2019
Holder of the chair CFM-ENS on datascience, 2016-2020
Visiting Professor @ Duke University, Maths Dept., 2018
Visiting Scientist @ Simons Institute in Berkeley, 2016
Visiting Scientist @ Los Alamos National Labs, 2008
Maitre de Conference (Associate Professor) in ESPCI Paristech, 2004 - 2013

Interests

Statistical Physics
Machine learning
Statistics
Computer Science
Random Optimization
Signal Processing
Information theory
Inference on graphs
Computational optics

Education

Postdoc, 2004
Roma, La Sapienza
PhD in Statistical Physics, 2002
Orsay, Paris XI
MSc in Physics, 1999
Orsay, Paris XI

Some recent talks online

Click here for the youtube channel, or browse with the arrows

Youtube Channel

Generalization in Machine Learning: Insights from Simple Models

Invited talk at the Workshop on the Theory of Overparameterized Machine Learning (TOPML) 2021

Watch talk

How hard is it to learn from data generated by a probabilistic model?

EPFL Lausanne

Watch talk

Statistics and Computations in high dimensions

The Alan Turing Institute

Watch talk

Statistical Physics and Machine Learning: Old ideas for new problems

KITP Conference: At the Crossroad of Physics and Machine Learning

Watch talk

Learning with (realistic?) synthetic data

NeurIPS2019 workshop: Science meets Engineering of Deep Learning

Watch talk

Selected review of 30 years of physics research on artificial neural networks

KITP Conference: The Rough High-Dimensional Landscape Problem

Watch talk

Rigourous proof of the Replica Formula for Low-Rank Factorization

A lecture at the 2018 Beg Rohu school on Deep Learning

Watch talk

On Statistical Physics of Inference

Disordered serendipity: a glassy path to discovery; A workshop in honour of Giorgio Parisi’s 70th birthday

Watch talk

How Does Gradient Descent Find a Good Minima, Instead of Spurious Ones?

Talk given at the Flatiron institute in Newyork, April 30, 2019

Watch talk

Teaching

Current or recent classes

Computation Science

Lecture given in the international master Physics of Complex Systems on computational science

Fundamentals of inference and learning EE-411

EPFL set of lectures on the theory of statistics, inference, and machine learning.

Introduction to Machine Learning

An introductory pratical course by Florent Krzakala and Antoine Baker, Ecole Doctorale EDPIF 2019

ML & AI pour la physique et les physiciens

Cours Master 1, Université Paris Sorbonne 2019-2010

Mecanique quantique et théorie des groupes

Cours donné à l’EPFL à partir de 2022

Numerical Physics and Machine Learning

Lecture Master 2, Ecole Normale Superieure 2020-2021, ICFP

Statistical Physics For Optimization and Learning

A set of Lectures given at Duke in 2018 by Lenka Zdeborova and Florent Krzakala

Statistical Physics For Optimization and Learning

A set of Lectures given at EPFL in 2021 by Lenka Zdeborova and Florent Krzakala

The SPHINX group

Meet the team working in the Sphinx Group. Our interdisplinary research group embodies physicists, mathematicians, and computer scientists working together to solve problems ranging from statistics to signal processing and machine learning. Feel free to check the group github repo.

The SPHINX team acknowledge funding from: