High-Dimensional Asymptotics and Dataset Selection for Private Transfer Learning
External datasets can improve prediction, but distribution shifts and privacy noise can erase those gains. This work develops...
A theory lab forunderstandinglearning systems.
We use statistical physics, probability, and computer science to understand why learning systems generalise, fail, and develop bias, and how to build safer and more intelligent AI.
Explore our researchOur latest works
The Interplay of Data Structure and Imbalance in the Learning Dynamics of Diffusion Models
Real-world datasets differ across classes in both structure and frequency, but most theory for diffusion models assumes homogeneous...
News
We are recruiting two MSc students for a thesis on generative AI for cell painting experiments, led by Flavio Nicoletti and co-supervised by Télio Cropsal in collaboration with the AI Laboratory for Molecular Engineering (AIME). See our opportunities page or read the full listing and apply.
New preprint! Filip, Edwige, Stefano, and Marco released High-Dimensional Asymptotics and Dataset Selection for Private Transfer Learning, studying when private external datasets improve prediction and how to select them using only summary statistics.
Stefano participated in the Wallenberg Advanced Scientific Forum (WASF) workshop on Symmetries in Neural Networks at Rånäs Castle, Sweden, from September 29 to October 2, 2026.
Great news! Our paper The Interplay of Data Structure and Imbalance in the Learning Dynamics of Diffusion Models has been accepted to NeurIPS 2026. Congratulations to Flavio, Chenxiao, Enrico, Luca, and Stefano!
Meet the group.
We work across machine learning, statistical physics, cognitive science, and high-dimensional probability at Chalmers and the University of Gothenburg, with active links to Wits.
Chenxiao Ma
PhD Student
Flavio Nicoletti
Postdoc Researcher
Loek van Rossem
PhD student
Stefano Sarao Mannelli
Assistant Professor
Our group