Erfan Mirzaei

Stanserhorn, Switzerland.

Hi. I am a third-year ELLIS Ph.D. student in Applied Mathematics and Machine Learning at Istituto Italiano di Tecnologia & École Polytechnique & Università di Genova. I am fortunate to be advised by Massimiliano Pontil and Karim Lounici.

My overarching goal is to explain why and how modern machine learning algorithms generalize. To bridge the gap between empirical success and statistical theory, I leverage advanced probability tools, specifically:

Concentration Inequalities
The PAC-Bayesian Framework
Information Theory

Currently, I am investigating the generalization properties of stochastic dynamical systems—such as Gibbs and Langevin Monte Carlo algorithms—particularly within the interpolation regime. I am deeply interested in establishing rigorous, data-dependent theoretical bounds that hold up in complex, realistic settings.

If you are interested in discussing machine learning theory, deep learning theory, or potential collaborations, feel free to reach out via email!

Selected Publications

Generalization of Gibbs and Langevin Monte Carlo Algorithms in the Interpolation Regime

Andreas Maurer, Erfan Mirzaei, and Massimiliano Pontil

arXiv preprint arXiv:2510.06028, 2025

arXiv Bib

@article{maurer2025generalization,
  title = {Generalization of Gibbs and Langevin Monte Carlo Algorithms in the Interpolation Regime},
  author = {Maurer, Andreas and Mirzaei, Erfan and Pontil, Massimiliano},
  journal = {arXiv preprint arXiv:2510.06028},
  year = {2025},
}