Lorenzo Croissant

ENSAE, Office 3106,

5 Av. Le Chatelier,

91120, Palaiseau, France

Research and Teaching Assistant (A.T.E.R.) at CREST in ENSAE. Member of the FairPlay Inria team (joint with CREST and Criteo AI Lab). Mathematics Ph.D. from Université Paris Dauphine and Criteo AI Lab under the supervision of Bruno Bouchard and Marc Abeille.

I am an applied mathematician with interests in Analysis and Probability, and in particular in questions of decision theory under uncertainty. I have worked one Control Theory, Optimal Transport, and Optim Reinforcement Learning, Bandits, Control Theory, Reinforcement Learning, Auction Theory, and Optimal Stopping.

Welcome to my Home page! You can find an organised overview of my work in projects, or directly visit the list of publications. Events collects things I organise, and on blog I may post small musings. My CV is available as a page or as a direct download.

news

Jun 30, 2025	Travelling to COLT and ICML this month! At COLT I will be helping Clair Vernade with the WiML event, while at ICML, I will be presenting my work on learning-augmented algorithms as a poster.
May 3, 2025	My first work in algorithms with predictions (a.k.a. learning-augmented algorithms) has been accepted at ICML 2025!
Mar 1, 2025	The second part of my work in the diffusion limit project is published in SPA.

selected publications

Diffusive limit approximation of pure jump optimal ergodic control problems

Marc Abeille, Bruno Bouchard, and Lorenzo Croissant

Stochastic Processes and their Applications Mar 2025

arXiv Bib HTML PDF Blog Project

@article{ABC25,
  title = {Diffusive limit approximation of pure jump optimal ergodic control problems},
  volume = {181},
  copyright = {All rights reserved},
  issn = {03044149},
  doi = {10.1016/j.spa.2024.104536},
  language = {en},
  urldate = {2024-12-05},
  journal = {Stochastic Processes and their Applications},
  author = {Abeille, Marc and Bouchard, Bruno and Croissant, Lorenzo},
  month = mar,
  year = {2025},
  pages = {104536},
}

Pareto-Optimality, Smoothness, and Stochasticity in Learning-Augmented One-Max-Search

Ziyad Benomar, Lorenzo Croissant, Vianney Perchet, and 1 more author

Feb 2025

Abs arXiv Bib HTML PDF Blog Project

One-max search is a classic problem in online decision-making, in which a trader acts on a sequence of revealed prices and accepts one of them irrevocably to maximise its profit. The problem has been studied both in probabilistic and in worst-case settings, notably through competitive analysis, and more recently in learning-augmented settings in which the trader has access to a prediction on the sequence. However, existing approaches either lack smoothness, or do not achieve optimal worst-case guarantees: they do not attain the best possible trade-off between the consistency and the robustness of the algorithm. We close this gap by presenting the first algorithm that simultaneously achieves both of these important objectives. Furthermore, we show how to leverage the obtained smoothness to provide an analysis of one-max search in stochastic learning-augmented settings which capture randomness in both the observed prices and the prediction.
@misc{Benomar25, title = {Pareto-{Optimality}, {Smoothness}, and {Stochasticity} in {Learning}-{Augmented} {One}-{Max}-{Search}}, copyright = {All rights reserved}, urldate = {2025-02-11}, publisher = {arXiv}, author = {Benomar, Ziyad and Croissant, Lorenzo and Perchet, Vianney and Angelopoulos, Spyros}, month = feb, year = {2025}, keywords = {Computer Science - Artificial Intelligence, Computer Science - Data Structures and Algorithms}, }
Bandit Optimal Transport

Lorenzo Croissant

Feb 2025

Abs arXiv Bib HTML PDF Blog Project

Despite the impressive progress in statistical Optimal Transport (OT) in recent years, there has been little interest in the study of the }emph{sequential learning} of OT. Surprisingly so, as this problem is both practically motivated and a challenging extension of existing settings such as linear bandits. This article considers (for the first time) the stochastic bandit problem of learning to solve generic Kantorovich and entropic OT problems from repeated interactions when the marginals are known but the cost is unknown. We provide \tildeO(\sqrtT) regret algorithms for both problems by extending linear bandits on Hilbert spaces. These results provide a reduction to infinite-dimensional linear bandits. To deal with the dimension, we provide a method to exploit the intrinsic regularity of the cost to learn, yielding corresponding regret bounds which interpolate between \tildeO(\sqrtT) and \tildeO(T).
@misc{Croissant_BOT_25, title = {Bandit {Optimal} {Transport}}, author = {Croissant, Lorenzo}, month = feb, year = {2025}, keywords = {Bandit Algorithms, Optimal transport}, }