Alberto Foresti

EURECOM, Campus SophiaTech

450 route des Chappes

06410 Biot, France

I am a final-year PhD student in machine learning at EURECOM in Sophia Antipolis, France, advised by Pietro Michiardi and Giulio Franzese.

My research sits at the intersection of generative modeling and information theory. I am interested in continuous and discrete diffusion models, optimal transport and Schrödinger bridges, and in the probability and statistical physics behind them. Recent work includes:

Information estimation with discrete diffusion: INFO-SEDD, a discrete-diffusion estimator of mutual information and entropy for high-dimensional discrete data (ICLR 2026).
Sampling schedules for discrete diffusion: optimization-free schedules derived from entropy production, dynamical activity and the discrete 1-Wasserstein distance.
Benchmarking mutual information estimators: a copula-based framework and a suite of synthetic, flow-based and image tasks that reveal where each class of estimator fails.

Before my PhD I received an MSc in Data Science and Engineering (110/110 cum laude) and a BSc in Computer Engineering from the Polytechnic of Turin, and was part of the Alta Scuola Politecnica honours program of the Polytechnics of Turin and Milan.

news

Oct 09, 2026	I was recognized as a Top Reviewer at NeurIPS 2026.
Jul 15, 2026	Two new preprints: Towards Diverse and Comprehensive Benchmarks for Mutual Information Estimation and From Atoms to Entropy: Optimal Noise Allocation for Diffusion Training in the Convex Regime.
Feb 06, 2026	New preprint: Improved Sampling Schedules for Discrete Diffusion Models.
Jan 26, 2026	Information Estimation with Discrete Diffusion was accepted at ICLR 2026.
Oct 01, 2025	WS2, on weakly supervised segmentation for waste sorting, appears at the ICCV 2025 Workshops.

selected publications

ICLR

Information Estimation with Discrete Diffusion

Alberto Foresti, Giulio Franzese, and Pietro Michiardi

In International Conference on Learning Representations, 2026

Bib PDF Code

@inproceedings{foresti2026information,
  title = {Information Estimation with Discrete Diffusion},
  author = {Foresti, Alberto and Franzese, Giulio and Michiardi, Pietro},
  booktitle = {International Conference on Learning Representations},
  volume = {2026},
  pages = {21001--21043},
  year = {2026},
}

arXiv

Improved Sampling Schedules for Discrete Diffusion Models

Alberto Foresti, Mustapha Bounoua, Giulio Franzese, and 2 more authors

arXiv preprint arXiv:2602.06849, 2026

arXiv Bib

@article{foresti2026improved,
  title = {Improved Sampling Schedules for Discrete Diffusion Models},
  author = {Foresti, Alberto and Bounoua, Mustapha and Franzese, Giulio and Ambrogioni, Luca and Michiardi, Pietro},
  journal = {arXiv preprint arXiv:2602.06849},
  year = {2026},
}

arXiv

Towards Diverse and Comprehensive Benchmarks for Mutual Information Estimation

Alberto Foresti, Ivan Butakov, Alexander Tolmachev, and 3 more authors

arXiv preprint arXiv:2607.03487, 2026

arXiv Bib Code

@article{foresti2026towards,
  title = {Towards Diverse and Comprehensive Benchmarks for Mutual Information Estimation},
  author = {Foresti, Alberto and Butakov, Ivan and Tolmachev, Alexander and Franzese, Giulio and Frolov, Alexey and Michiardi, Pietro},
  journal = {arXiv preprint arXiv:2607.03487},
  year = {2026},
}