Arnaud Doucet

Colloquium Arnaud Doucet

Senior Staff Research Scientist, Google DeepMind
Visiting Professor of Statistics, University of Oxford

Thursday, October 15, 2026

4:00 - 5:00 pm (refreshments at 3:30 pm)

Broad Institute Auditorium (415 Main St.) and virtually at broad.io/ewsc

On Some Recent Developments in Diffusion Models

Abstract: Diffusion models have become ubiquitous for generative modelling across various modalities, including speech, audio, and video. In this talk, I will discuss recent methodological developments, focusing on two distinct application domains. First, I will discuss unsupervised speech source separation. While clean training data are unavailable in this setting, I will show how generative modelling ideas can be leveraged to address this challenge. Second, I will discuss very recent work on extending diffusion techniques to discrete data generation.

Biography: Arnaud Doucet obtained his PhD from University Paris-XI in 1997. He has held faculty positions at Cambridge University, the University of British Columbia, the Institute of Statistical Mathematics, and Oxford University, where he was a professor in the Department of Statistics. He was an Institute Mathematical Statistics (IMS) Medallion Lecturer in 2016, was elected an IMS Fellow in 2017, and was awarded the Guy Medal in Silver in 2020 by the Royal Statistical Society. Since 2023, he has been a Senior Staff Research Scientist at Google DeepMind where he leads a team working on generative modelling.