Benjamin Guedj

Benjamin Guedj is a Professor of Machine Learning and Foundational Artificial Intelligence at University College London (UCL) and a Research Director at Inria (France). He leads the Inria London Programme, fostering Franco-British scientific collaboration. His research focuses on PAC-Bayesian theory, generalisation in machine learning, and frugal AI. He holds a PhD from Sorbonne Université and an Habilitation à diriger des recherches. He has supervised eight PhD students and contributed over 70 research articles. Awards include the NeurIPS Best Reviewer Award and PEDR excellence bonus. He co-founded the ELLIS London Unit and chairs roles in French-British academic initiatives.

Education: PhD in Mathematical Statistics (Sorbonne Université, 2013), MSc in Pure/Applied Mathematics (Sorbonne, 2010–11), postdoctoral research at Sorbonne and DTU Compute. Research interests span statistical learning, PAC-Bayes, computational statistics, and deep learning.

Grants: Led ANR projects BEAGLE/APRIORI (€180k each), Inria's 6PAC project, and EU-funded PERF-AI (€600k). Awards: Multiple best reviewer recognitions, Young Leader of Franco-British Council.

Advising: Current PhD students include Chloé Hashimoto-Cullen and Reuben Adams; alumni include Théophile Cantelobre and Felix Biggs. His group explores PAC-Bayes, generalisation, and AI ethics.

Labs/Teams: Co-director of ELLIS London, Inria Modal team, and UCL Centre for AI. Serves as associate editor for JMLR/TMLR and chairs ICML/PAC-Bayes workshops.