About

I am Stefano Sarao Mannelli, a tenure-track Assistant Professor in the Data Science and AI division in the Computer Science department of Chalmers University of Technology and Gothenburg University, and Visiting Lecturer at the University of the Witwatersrand. Prior to my current position, I worked as a postdoc with Andrew Saxe at the University College London and the University of Oxford and obtained a Ph.D. in Physics applied to Machine Learning at the University of Paris-Saclay supervised by Lenka Zdeborova. My research focuses on analysing machine learning problems using a model-based approach, where the complexity of the problem is reduced to obtain a parsimonious solvable model that still captures the phenomenon of interest. In my previous works, I applied several variations of this approach to study problems in learning, such as transfer learning, continual learning, and curriculum learning.

Recent publications

arXiv preprint

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 data. This work develops a high-dimensional analytical framework...

ICML 2026 (Spotlight)

Position: the Stochastic Parrot in the Coal Mine. Model Collapse is a Threat to Low-Resource Communities

Model collapse can degrade generative models when they are trained on outputs from earlier models. This position paper argues that the problem compounds existing concerns around...

ICML 2026

Sharp description of local minima in the loss landscape of high-dimensional two-layer ReLU neural networks

We study the population loss landscape of two-layer ReLU networks in a realisable teacher-student setting with Gaussian covariates. The work shows that local minima admit an...

Proceedings of the Analytical Connectionism Schools 2023--2024, PMLR 320:15-41, 2026

Thinking of Neural Networks Like a Physicist: The Statistical Physics of Machine Learning

This pedagogical paper introduces statistical-physics approaches to machine learning, based on material presented at Analytical Connectionism 2023. It reviews how tools such as the replica method...

Phys. Rev. E 112, 025304 (2025)

Bias-inducing geometries: an exactly solvable data model with fairness implications

Machine learning (ML) may be oblivious to human bias but it is not immune to its perpetuation. Marginalisation and iniquitous group representation are often traceable in...

Experience and activities

Research experience

WASP-AI Assistant Professor Tenure Track Department of Computer Science and Engineering, Chalmers University of Technology and Gothenburg University, Gothenburg, Sweden
Sept. 2024 – Present
Senior Research Fellow Gatsby Computational Neuroscience Unit and Sainsbury Wellcome Centre, University College London, London, UK
Jan. 2023 – Aug. 2024
Supervisor: Andrew Saxe
Research Fellow Gatsby Computational Neuroscience Unit and Sainsbury Wellcome Centre, University College London, London, UK
Jun. 2021 – Dec. 2022
Supervisor: Andrew Saxe
Research Fellow Department of Experimental Psychology, University of Oxford, Oxford, UK
Oct. 2020 – Jun. 2021
Supervisor: Andrew Saxe
Visiting Researcher New York University (NYU), New York, USA
Mar. 2020 – Apr. 2020
Visiting Researcher Kavli Institute for Theoretical Physics (KITP), University of California Santa Barbara, Santa Barbara, USA
Jan. 2019 – Mar. 2019
Visiting Researcher Duke University, Durham, USA
Feb. 2018
Research Intern IPhT, CEA, Saclay, France
Mar. 2017 – Jul. 2017
Supervisor: Lenka Zdeborová
Research Intern CMLA, ENS Cachan, Cachan, France
Jan. 2016 – Apr. 2016
Supervisor: Nicolas Vayatis

Research projects

Bias Generation and Amplification Developing a theoretical framework to identify the key factors contributing to ML misbehaviour against population subgroups.
Bridging Biological and Artificial Neural Networks Investigating the learning differences between biological and artificial neural networks to improve both NN models and our understanding of the brain.
Connecting Dynamics and Landscapes in High Dimensions Reconciling and identifying potential limitations of loss landscape properties in relation to the asymptotic performance of ML systems.

Event organization

KITP Program on Statistical Physics & Neurobiology of Learning in High-Level Cognition Organizer (future event), Santa Barbara, USA
Jun.-Jul. 2027
School on Analytical Connectionism School Director (future event), Gothenburg, Sweden
Aug. 2026
Secured 400,000 SEk
Workshop on the Mathematical Foundations of AI Organizer, Gothenburg, Sweden
Jun. 2026
Secured 165,000 SEk
Workshop on LLMs and Responsible AI Organizer, Gothenburg, Sweden
Mar. 2026
EurIPS Workshop - Unifying Perspectives on Learning Biases Organizer, Copenhagen, Denmark
Dec. 2025
School on Analytical Connectionism Organizer, London, UK
Aug. 2025
Secured £25K funding
Workshop in Advancements in High-Dimensional Methods for Machine Learning Organizer, Gothenburg, Sweden
May 2025
Secured 135,500 SEk
Introduction to Statistical Physics for ML Theory Organizer, Gothenburg, Sweden
Apr. 2025
School on Analytical Connectionism Organizer, New York, USA
Aug. 2024
Secured $152K funding
Workshop on Analytical Approaches for Neural Network Dynamics Organizer, Paris, France
Oct. 2023
Secured €10K funding
School and Workshop on Analytical Connectionism Organizer, London, UK
Aug.–Sep. 2023
Secured £42K funding

Students and postdocs

Jie Huang Master student
Loek Von Rossem PhD student (3rd year)
Co-supervisor: Andrew Saxe
Chenxiao Ma PhD student
Start: Feb. 2025
Flavio Nicoletti Postdoc
Start: Mar. 2025

Grants and awards

Research Visit Grant Helge Ax:son Johnsons stiftelse
Jun. 2026
Amount: 39,000 SEK
Vetenskapsfond Wilhelm och Martina Lundgrens stiftelser
Jan. 2026
Amount: 50,000 SEK
AI Alignment Grant AISI
Jan. 2026
Amount: 969,811.36 GBP
Academic Grant CM Lerici Foundation
Jan. 2025
Amount: 60,000 SEK
Travel Grant Guarantor of Brains
Jul. 2024
Amount: £1,000
Travel Grant G-Research
Feb. 2024
Amount: £1,400
UK–IT Trustworthy AI Exchange Programme The Alan Turing Institute
Nov. 2023
Amount: £4,500
SCGB Conference Awards Simons Foundation
Mar. 2023
Amount: £3,000
Ph.D. Scholarship CEA
Sept. 2017
Mobility Grant "Tesi su proposta" Politecnico di Torino
Feb. 2017 & Feb. 2016
Amount: €2,200 each
Accommodation Scholarship "Borsa Talenti" Fondazione CEUR
Sept. 2011, 2012 & 2013
3rd Prize – Individual Math Competition Alfa Class Fondazione CRT
Sept. 2012
Amount: €500