François-Xavier Briol

François-Xavier Briol is a statistician and machine learning researcher whose work spans Bayesian quadrature, robust inference and simulation-based inference. His 2015 Frank-Wolfe Bayesian Quadrature method gave probabilistic integration exponential convergence, and later work used maximum mean discrepancy for robust inference. He completed a joint Warwick–Oxford PhD, held postdoctoral roles at Imperial and Cambridge, joined UCL in 2019, and led a Data-Centric Engineering group at the Alan Turing Institute from 2020 to 2023. As of October 2026, he is a Professor of Statistics and Machine Learning at UCL and an ELLIS Unit Director and Scholar; he also shared the 2026 Mardia Prize for workshops linking statistics with fundamental physics.

Now at: Department of Statistical Science, University College London and European Laboratory for Learning and Intelligent Systems

Timeline

2015

Frank-Wolfe Bayesian Quadrature

He coauthors Frank-Wolfe Bayesian Quadrature in 2015 with Chris Oates, Mark Girolami and Michael A. Osborne; the probabilistic integrator gives exponential convergence to the true integral value.
By 2019

A joint PhD and UCL lecturer

By 2019, he completes a Statistics and Machine Learning PhD through a joint CDT at Warwick and Oxford, after bachelor's and master's study in MMORSE at Warwick. A year of postdoctoral research at Imperial College London and Cambridge precedes his 2019 appointment as a lecturer at UCL Statistical Science.
2019

MMD for generative models

He coauthors a 2019 method for statistical inference on generative models with intractable likelihoods using maximum mean discrepancy. The authors show that kernel choice can trade statistical efficiency and robustness.
2020–2023

Data-Centric Engineering at Turing

From 2020 to 2023, he serves as a Group Leader in Data-Centric Engineering at The Alan Turing Institute.
2022

Robust Bayesian inference

In 2022, with Charita Dellaporta, Jeremias Knoblauch and Theodoros Damoulas, he coauthors an MMD posterior bootstrap for robust Bayesian inference on simulator-based models with intractable likelihoods. The paper receives the AISTATS Best Paper Award.
2025

Neural inference across simulators

In 2025, he coauthors Multilevel neural simulation-based inference, which uses multilevel Monte Carlo to combine simulators with different costs and levels of fidelity.
2026

The Mardia Prize

In 2026, he and Purvasha Chakravarti receive the Royal Statistical Society's Mardia Prize for a workshop series connecting statisticians and physicists around fundamental physics.
As of October 2026

Professor and ELLIS director

He is a Professor of Statistics and Machine Learning at UCL Statistical Science and co-leads its Fundamentals of Statistical Machine Learning research group. He also serves as a Unit Director and Scholar in ELLIS Unit London.