Bowen Island, British Columbia
Bowen Island, British Columbia
I am a probabilist working on the mathematics of machine learning, with tools from random matrix theory, high-dimensional probability, and statistical physics. I am especially interested in nonlinear random matrices, such as those arising in neural networks, and in how algorithms recover hidden structure from noisy high-dimensional data, as well as in spin glasses.
Below is a list of my publications and preprints in reverse chronological order.
Preprints:
Symmetry-Aware Feature Learning: A Polynomial Separation for Multi-Index Models
Joint with Jivan Waber, Yatin Dandi, and Florent Krzakala, 2026
[arXiv]
Spectral phase transitions in Gaussian multi-index models
Joint with Florent Krzakala and Pierre Mergny, 2026
[arXiv]
Publications:
Global law of conjugate kernel random matrices with heavy-tailed weights
Joint with Alice Guionnet
Electronic Journal of Probability, 31(8), 1--58, 2026
[Journal version] [arXiv]
Permutation recovery of spikes in noisy high-dimensional tensor estimation
Joint with Gérard Ben Arous and Cédric Gerbelot
Stochastic Analysis and Applications 2025, 425--476, Springer, Cham, 2026
[Journal version] [arXiv]
Stochastic gradient descent in high dimensions for multi-spiked tensor PCA
Joint with Gérard Ben Arous and Cédric Gerbelot
Communications on Pure and Applied Mathematics, 79(10), 2291--2369, 2026
[Journal version] [arXiv]
Langevin dynamics for high-dimensional optimization: the case of multi-spiked tensor PCA
Joint with Gérard Ben Arous and Cédric Gerbelot
Probability Theory and Related Fields, 196(1-2), 643--749, 2026
[Journal version] [arXiv]
Topological complexity of spiked random polynomials and finite-rank spherical integrals
Annals of Applied Probability, 36(5), 4535--4592, 2026
[Journal version] [arXiv]
Analysis of one-hidden-layer Neural Networks via the Resolvent Method
Joint with Dominik Schröder
NeurIPS, 34, 2021
[NeurIPS] [arXiv]