I am a maître de conférence (~associate professor) at Université Paris-Saclay in Orsay (south of Paris) in the field of statistics. My research interests are multiple testing and its applications to biological problems, and the theoretical foundations of machine learning and deep learning.

Previously, I worked as a data scientist on patient and asset flow at Intelligent Locations. Before that, I did my PhD in applied mathematics at the Laboratoire de Probabilité, Statistique et Modélisation (LPSM) of Sorbonne Université, under the supervision of Etienne Roquain and Pierre Neuvial. During my PhD I worked on multiple testing procedures using weighting and heterogeneity, as well as post hoc false positive bounds for selective inference. Before my PhD, I studied mathematics and engineering at École polytechnique and Université Paris-Sud.

Here is my Curriculum Vitae (updated 2023-06-06).

Research interests

• Multiple testing and FDR control: theory and application to omics data
• p-value weighting and power optimality
• Discrete multiple testing
• Selective inference, post hoc inference and joint false positive control
• Deep learning, expressivity of neural networks
• Counterfactual inference, meta-learning, state space learning, reinforcement learning

Publications

See Education for my PhD thesis.

Published papers

Post hoc false positive control for structured hypotheses, G. Durand, G. Blanchard, P. Neuvial, E. Roquain, Scandinavian Journal of Statistics 47.4 (2020), pp. 1114–1148. arXiv version: arXiv:1807.01470.

Adaptive p-value weighting with power optimality, G. Durand, Electronic Journal of Statistics 13.2 (2019), pp. 3336–3385.

New FDR bounds for discrete and heterogeneous tests, S. Döhler, E. Roquain, G. Durand, Electronic Journal of Statistics 12.1 (2018), pp. 1867–1900.

Performance of epistasis detection methods in semi-simulated GWAS, C. Chatelain, G. Durand, V. Thuillier, F. Augé, BMC Bioinformatics 19.231 (2018).

Fixation probability in a two-locus intersexual selection model, G. Durand, S. Lessard, Theoretical population biology 109 (2016), pp. 75–87.

Preprints

DiscreteFDR: An R package for controlling the false discovery rate for discrete test statistics, G. Durand, F. Junge, S. Döhler, E. Roquain, arXiv:1904.02054.

Software

Contributions to the R package sansSouci on GitHub. Post hoc bounds on the number of false positives of any selected set of hypotheses.

R package DiscreteFDR on CRAN. Multiple testing procedures controlling the FDR for p-values with discrete support.

Talks

Contrôle post hoc des faux positifs pour des hypothèses structurées, Journée de rentrée du LMO, October 2021, Orsay. Slides.

Contrôle post hoc des faux positifs pour des hypothèses structurées, Séminaire Statistique de l’IRMA, March 2021, Strasbourg (videoconference). Slides.

Contrôle post hoc des faux positifs pour des hypothèses structurées, Séminaire de probabilités et statistiques du LAMA, March 2021, Champs-sur-Marne. Slides.

Improved post hoc bounds for localized signal, Young Researchers’ Meeting in Mathematical Statistics, September 2018, Paris. Slides.

Optimal data-driven weighting procedure with grouped hypotheses and pi_0-adaptation, Workshop: Post-selection Inference and Multiple Testing, February 2018, Toulouse. Poster.

Tests multiples : généralités, problème du weighting optimal, Journée des thésards ESP, October 2017, Toulouse. Slides.

Adaptive data-driven optimal weighting, Statistique Mathématique et Applications, September 2017, Fréjus. Slides.

Step-up procedure with data-driven optimal weights for grouped hypotheses, Multiple Comparison Procedures, June 2017, Riverside. Slides.

BH procedure using data-driven optimal weights for grouped hypotheses, CMStatistics, December 2016, Sevilla. Slides.

An extension of the Benjamini and Hochberg procedure using data-driven optimal weights with grouped hypothesis, Journées MAS, August 2016, Grenoble. Slides.

Teaching

As a maître de conférences at Université Paris-Saclay:

Pré-requis de statistique, lecture course, M1 Mathématiques et IA, 2023+, 6 hours/year. Essentiels de théorie de la mesure.

During my PhD:

Introduction aux probabilités, practical tutorials, 1st year students, Polytech Sorbonne, 2017, 8 hours.

Mesure, intégration, probabilités, practical tutorials, 1st year students, Institut de statistique de Sorbonne Université, 2015-2016, 90 hours/year.

Other teaching experiences:

Work experience

Maître de conférence, Université Paris-Saclay, Laboratoire de Mathématiques d’Orsay, since 2021.

Data Scientist, Intelligent Locations, 2018-2021. Study and simulation of patient and asset flow inside multiple hospital services, software development using Pyton and SQL.

PhD student, Sorbonne Université, 2015-2018. See Education.

Internship student in bioinformatics, Sanofi, under the supervision of Franck Augé, 2015 (6 months).

Research assistant in population genetics, Université de Montréal, under the supervision of Sabin Lessard, 2014 (4 months).

Internship student in development, Lipigas, under the supervision of Camilo Muñoz, 2013 (6 weeks).

Teaching assistant in elementary schools, Académie de Créteil, under the supervision of Axel Jean, 2011-2012 (7 months).

Education

PhD, LPSM, Sorbonne Université, 2015-2018, Paris. Here is my manuscript (with minor corrections with respect to the TEL repository version) and here are my defense slides.

Master 2 Mathématiques pour les sciences du vivant, Université Paris-Sud (now Université Paris-Saclay), 2014-2015, Orsay.

Diplôme de l’X, École polytechnique, 2011-2015, Palaiseau.

Classe Préparatoire aux Grandes Écoles, Lycée Henri IV, 2009-2012, Paris.

Baccalauréat scientifique, Lycée Guillaume Budé, Limeil-Brévannes.