Vincent Plassier

Vincent Plassier

PhD in Machine Learning

Vincent Plassier, PhD candidate

Hello, I just defended my PhD a couple of months ago at Ecole Polytechnique. My PhD was conducted at École Polytechnique (CMAP) on a grant from Huawei Technologies (Lagrange Research Center). My advisors were Eric Moulines and Alain Durmus.

I am currently working on federated Monte Carlo methods with applications to large-scale Bayesian inference. More broadly, I am interested in the following topics:

  • Machine learning applications
  • Distributed/Federated Monte Carlo methods
  • Uncertainty quantification via Conformal Prediction/Bayesian Fusion

Previously, I graduated from Ecole Normale Supérieure Paris-Saclay, where I earned a master’s degree in Applied Mathematics. Additionally, I hold a master’s degree Mathematics, Vision and Learning, from ENS Paris-Saclay (MVA).

News.

  • 23-29/07/2023: I will be presenting my paper at the Hawaii Convention Center.
  • 24-27/04/2023: Presented my paper at AISTAT2023, held at Palacio de Congresos de València, Spain.
  • 24/04/2023: Excited to announce that my submission titled “Conformal Prediction for Federated Uncertainty Quantification Under Label Shift” has been accepted for inclusion in the proceedings of ICML 2023. (ArXiv)
  • 20/01/2023: Delighted to share that our paper titled “Federated Averaging Langevin Dynamics: Toward a unified theory of and new algorithms” has been accepted for AISTATS 2023. (Proceedings)
  • 13-17/03/2023: Participated in the workshop on “Statistics, Learning, Simulation, and Image” held in Hyères.
  • 24-28/10/2022: Attended an international workshop at CIRM on the Computational methods for unifying multiple statistical analyses (Bayesian Fusion).
  • 24-30/07/2022: Participated in the “Math for Machine Learning Summer School” (link) at Mohammed VI University in Ben Guerir, Morocco. I also had the opportunity to deliver a talk during the event.
  • 07-11/03/2022: Engaged in the workshop titled “New Challenges in Statistical Learning” held in Font-Romeu.
  • 01/2022: Our paper titled “QLSD: Quantised Langevin Stochastic Dynamics for Bayesian Federated Learning” has been accepted for AISTATS 2022 (Proceedings).
  • 03/2022: Collaborated with François Portier and Johan Segers on the paper “Risk bounds when learning infinitely many response functions by ordinary linear regression”. The paper has been accepted for publication in the Annales de l’Institut Henri Poincaré (ArXiv).
  • 06/2021: Attended the workshop on “Recent advances in machine learning and uncertainty” at CIRM, Marseille.
  • 05/2021: “DG-LMC: A Turn-key and Scalable Synchronous Distributed MCMC Algorithm via Langevin Monte Carlo within Gibbs” accepted to ICML 2021 (ArXiv).