On the disintegration of the stochastic majority vote: From PAC-Bayesian bounds to a self-bounding algorithm
arXiv:2609.16803, 2026
[link]
About
Julien Bastian
PhD student in Computer Science
Since October 2024, I have been a PhD student in Computer Science at the Hubert Curien Laboratory, Université Jean Monnet (France) in Saint-Étienne, where I am a member of the Machine Learning team. My research lies in the field of statistical machine learning, with a particular focus on algorithmic fairness and PAC-Bayesian theory. I am interested in understanding how learning algorithms can be designed with (and from) theoretical guarantees while taking fairness-related constraints into account.
My PhD is supervised by Christine Largeron, Emilie Morvant, and Guillaume Metzler, and is funded by the french project ANR Famous (Fairness-Aware Multimodal Learning Under Structured Assumptions). The goal of my research is to study how PAC-Bayesian tools can contribute to the analysis and design of fair learning algorithms. More broadly, my work aims to connect theoretical machine learning with questions related to the fairness of predictive models.
Research interests
- Fairness and Bias in Machine Learning
- PAC-Bayesian theory
- Statistical machine learning
- Learning theory
- Generalization guarantees
- Supervised learning
- Self-bounding algorithms
Research
Preprints
PAC-Bayesian Generalization Guarantees for Fairness on Stochastic and Deterministic Classifiers
arXiv:2602.11722, 2026
[link]
French Conferences
Détection non supervisée d'anomalies dans les images satellites pour le monitoring des surfaces océaniques par une composition d'ACP robuste et d'un test d'adéquation sur la distance de Wasserstein entre processus ponctuels
Conférence sur l'Apprentissage automatique (CAp), 2024
Seminars
Bias and Fairness in Artificial Intelligence
Pint of Science · Technology and Ethics: Two Faces of AI · Rouen · May 2026
Joint talk with Hind Atbir.
Towards Fair Learning with Multiple Sensitive Attributes under a PAC-Bayesian Perspective
ANR FAMOUS Project Workshop · Université Jean Monnet, Saint-Étienne, France · November 2025
PAC-Bayesian Generalization Guarantees for Fairness
ANR FAMOUS Project Workshop · Aix-Marseille Université, France · March 2025