This summer school is part of the Bernoulli Center programme at EPFL, to be held from 1 to 5 September 2025.
The 2025 Summer School on the Mathematical Foundations of Data Science is designed for MSc and PhD students interested in the theoretical underpinnings of data science. As the second installment of a rotating international series (following the 2024 edition in Australia and preceding the 2026 edition in Singapore), the school offers a unique opportunity to learn from leading researchers at the interface of mathematics and machine learning. The confirmed speakers are Francis Bach, Afonso S. Bandeira, Stéphane Mallat, Andrea Montanari and Gabriel Peyré, who will cover topics ranging from structured prediction and random matrix theory to image generation, overfitting, and optimal transport. Participants can expect a rich program combining foundational insights with cutting-edge developments.
Recordings :
Monday
Stéphane Mallat – Image generation from score diffusion to renormalisation group in physics
Afonso S. Bandeira – Random matrices and tensor PCA 1
Afonso S. Bandeira – Random matrices and tensor PCA 2
Tuesday
Stéphane Mallat – Image generation from score diffusion to renormalisation group in physics
Andrea Montanari – Feature learning and overfitting
Wednesday
Andrea Montanari – Feature learning and overfitting
Gabriel Peyré – Diffusion flows and optimal transport in machine learning
Thursday
Gabriel Peyré – Diffusion flows and optimal transport in machine learning
Francis Bach – Structured prediction: supervised learning beyond binary classification
Friday
Francis Bach (INRIA) – Structured prediction: supervised learning beyond binary classification
