Florent Krzakala
Florent Krzakala is a researcher working across statistical physics, machine learning, information theory and signal processing. He earned a statistical physics doctorate at Orsay in 2002. With Lenka Zdeborová, he coauthored a review on inference thresholds and algorithms, covering network communities and compressed sensing. His research group moved from Paris to EPFL, where the IdePHICS laboratory was created in September 2020. His teaching and research connect mathematical models of learning with optimization, inference on graphs and neural network theory.
Timeline
-
1999–2004
Training in statistical physics
Krzakala earned a physics master’s degree at Orsay in 1999, a doctorate in statistical physics in 2002 and a postdoctoral period at Rome’s La Sapienza ending in 2004.
-
2015–2016
Recover information from measurements
Krzakala and Lenka Zdeborová coauthored a review on inference thresholds and algorithms. It studies when noisy measurements contain enough information for recovery and how statistical physics can guide efficient algorithms, with examples in graph communities and sparse signal estimation.
-
2020
Create IdePHICS at EPFL
The IdePHICS laboratory began in September 2020, after Krzakala’s work in Paris. Krzakala leads the laboratory, linking physics, learning and information.
-
2023–2024
Explain how networks learn
He gave talks in 2023 on two-layer neural network learning and simple functions. His teaching also covers inference and learning and lecture notes on replica, cavity and message-passing methods, connecting physical theory with machine learning practice.
Sources and credits
Photo: florentkrzakala.com; creator not stated. License: This work is licensed under CC BY NC ND 4.0. Source: florentkrzakala.com. Cropped to a square and resized. Ask us to remove or replace this photo.