Margaux Zaffran

Since November 2024, I am a postdoctoral researcher in the Statistics Department of UC Berkeley, working with Ryan Tibshirani.

In June 2024 I completed my PhD in statistics, under the supervision of Aymeric Dieuleveut from CMAP, Ecole Polytechnique, of Julie Josse, from Inria, and of Olivier Féron (director of the FiME laboratory) and Yannig Goude from EDF R&D.

I am interested in predictive uncertainty quantification, especially with distribution-free finite-sample guarantees in a post-hoc setting. During my PhD, I had a particular emphasis in the context of time series or missing features.

News

• Very happy to announce that I succesfully defended my PhD last Tuesday! A huge thanks to the members of the jury. Slides available, as well as the manuscript.
• Really enthusiastic that our tutorial’s proposition, Distribution-Free Predictive Uncertainty Quantification: Strengths and Limits of Conformal Prediction, with Aymeric Dieuleveut, was accepted at UAI and ICML! Pick your favorite European destination for the summer and see you there :D
• Super excited to announce that I have been named one of the Fondation L’Oréal-UNESCO 2023 French Young Talents, along with 34 amazing French women scientists at the doctoral or post-doctoral level!

Upcoming events

December 1-5, 2025. I will teach a week-long course on conformal prediction at the 2025 ECAS-SFdS School Towards Reliable Machine Learning, on the French riviera. Don’t hesitate to register, and enjoy December’s sun!

Community

I am the current president of the Young Statisticians Group of the French Statistical Society (SFdS).
We have organized the 11th Young Statisticians and Probabilists day. Thanks to all the speakers for the amazing talks! You can find the slides and their contact here (very soon).

I served as a conference reviewer for AISTATS 2023 (10% top reviewer), for NeurIPS 2023 and as a journal reviewer for the Journal of the Royal Statistical Society, Series B, for the International Journal of Forecasting, for the Machine Learning journal and for the Journal of Business and Economics Statistics.

News archive

• Just released an introductive tutorial (slides) on Conformal Prediction! Going from the standard split mean-regression case towards beyond exchangeability, passing by conditional coverage, classification, and no (other) splitting strategies.
• Super happy that the work with Aymeric Dieuleveut, Julie Josse, and Yaniv Romano, Conformal Prediction with Missing Values, was accepted to ICML 2023!
• I was very honored to receive the Séphora Berrebi Scholarship for Women in Advanced Mathematics. Many thanks to the Séphora Berrebi Association.
• From August to October 2022, I was visiting Yaniv Romano and his group at the Technion. Thanks for hosting me! I also thank the FMJH and LMH for funding this stay abroad.
• Our paper, Adaptive Conformal Predictions for Time Series, has been accepted to ICML 2022!
• My Master’s thesis was awarded the prize of the best end-of-studies project by ENSTA Alumni. Thanks a lot to them, and to Philippe Naveau who supervised me!
• I presented Adaptive Conformal Predictions for Time Series at the International Seminar on Distribution-Free Statistics (ISDFS). Thanks a lot Anastasios for the kind invitation!