Florent Krzakala
Full Professor
Leads the Information, Learning and Physics lab, applying statistical physics to the theory of machine learning and high-dimensional inference.
Florent Krzakala's lab at EPFL works on learning theory & optimization and efficient training & inference. 15 papers since 2023, 4 at top venues, mostly AISTATS and ICML. 6 PhD students. 33 alumni, most often to NVIDIA and Google DeepMind. Applications close .
At a glance
- papers, 2023–2026
- 4/ year
- at T1 venues
- 27%
- PhD students
- 6
- MS students
- 0
- alumni listed
- 33
Papers by yearfrom the lab homepage, 2023–2026
- 0 papers2026 · T1 0 · collecting
- 0 papers2025 · T1 0
- 3 papers2024 · T1 0
- 12 papers2023 · T1 4
T1 venuesOtherVenues: AISTATS, ICML, JMLR, UAI
Research focuspaper titles that use each topic's words
- Efficient training & inference2 papers
- Learning theory & optimization1 paper
- physics1 paper
- statistical physics1 paper
- applying0 papers
Paper list
From the lab homepage.
20243 papers
- Physical Review EGaussian universality of perceptrons with random labels (opens in a new tab)
- Journal Of Physics A-Mathematical And TheoreticalOn the atypical solutions of the symmetric binary perceptron (opens in a new tab)
- Journal Of Statistical Mechanics-Theory And ExperimentStatistical mechanics of the maximum-average submatrix problem (opens in a new tab)
202312 papers
- CLEO/Europe-EQECArtificial Neural Network Training on an Optical Processor via Direct Feedback Alignment (opens in a new tab)
- ICML (T1)Bayes-optimal Learning of Deep Random Networks of Extensive-width
- PLoS Comput BiolBayesian reconstruction of memories stored in neural networks from their connectivity (opens in a new tab)
- ITWCompressed sensing with ℓ 0-norm: statistical physics analysis & algorithms for signal recovery (opens in a new tab)
- IOPscienceError scaling laws for kernel classification under source and capacity conditions (opens in a new tab)
- UAI (T1)Expectation consistency for calibration of neural networks
- Journal Of Statistical Mechanics-Theory And ExperimentFluctuations, bias, variance and ensemble of learners: exact asymptotics for convex losses in high-dimension (opens in a new tab)
- Journal Of Statistical Mechanics-Theory And ExperimentMulti-layer state evolution under random convolutional design (opens in a new tab)
- AISTATS (T1)On double-descent in uncertainty quantification in overparametrized models
- Journal Of Statistical Mechanics-Theory And ExperimentPhase diagram of stochastic gradient descent in high-dimensional two-layer neural networks (opens in a new tab)
- Machine Learning-Science And TechnologyTheoretical characterization of uncertainty in high-dimensional linear classification (opens in a new tab)
- JMLR (T1)Tree-AMP: Compositional Inference with Tree Approximate Message Passing
Where alumni went33 listed, 28% to industry
Industry 7Academia 18
Most often to
- NVIDIA
- Google DeepMind
Also to
- Bocconi University
- EPFL
- Harvard University
Apply to EPFLFall 2027
EDIC Computer and Communication Sciences PhD
- Round
- Round 1
- Opens
- —
- GRE
- optional
- English
- TOEFL 100 / IELTS 7.0
- Fee
- none
Confirmed on the program page.
Similar labsnearest by research summary
- SGNUS
Warut Suksompong
- Trustworthy & safe ML
- USStanford
Carlos Guestrin
- Trustworthy & safe ML
- USUT Austin
Leqi Liu
- Large language models
- CAToronto
Nisarg Shah
- Learning theory & optimization
- CHEPFL
Emmanuel Abbé
- Learning theory & optimization
- USGeorgia Tech
Jacob Abernethy
- Learning theory & optimization
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
EPFLcloses