Victor Dheur

prof_pic.jpg

Bâtiment De Vinci

Avenue Victor

Maistriau 15

7000 Mons, Belgium

About me

I am a PhD student in the Big Data and Machine Learning Lab at the University of Mons (UMONS), advised by Souhaib Ben Taieb.

My research focuses on principled and efficient uncertainty quantification to improve the reliability of machine learning models. Primarily, my work involves statistical frameworks such as model calibration, conformal prediction, and strictly proper scoring rules. These methods are compatible with a range of machine learning models, including generative models such as normalizing flows and diffusion models. I am also interested in connections to risk control and Bayesian methods.

Applications of interest include problems involving tabular data, temporal data and computer vision.

Selected publications

  1. A unified comparative study with generalized conformity scores for multi-output conformal regression
    Victor DheurMatteo Fontana, Yorick Estievenart, Naomi Desobry, and Souhaib Ben Taieb
    In The 42th International Conference on Machine Learning, 2025
  2. Rectifying conformity scores for better conditional coverage
    Vincent Plassier*, Alexander Fishkov*, Victor Dheur*, Mohsen Guizani, Souhaib Ben TaiebMaxim Panov, and Eric Moulines
    In The 42th International Conference on Machine Learning, 2025
    (* denotes equal contribution.)
  3. Probabilistic Calibration by Design for Neural Network Regression
    Victor Dheur, and Souhaib Ben Taieb
    In The 27th International Conference on Artificial Intelligence and Statistics, 2024