Details of the talk:
- Date: July 11, 2024
- Time: 11:00 a.m. - 12:00 p.m.
- Location: Ground-floor lecture hall, Tübingen AI Center (Maria-von-Linden-Str. 6, 72076 Tübingen)
Talk title: Learning diffusion priors from observations by expectation maximization
Abstract: Diffusion models recently proved to be remarkable priors for Bayesian inverse problems. However, training these models typically requires access to large amounts of clean data, which could prove difficult in some settings. In this work, we present a novel method based on the expectation-maximization algorithm for training diffusion models from incomplete and noisy observations only. Unlike previous works, our method leads to proper diffusion models, which is crucial for downstream tasks. As part of our method, we propose and motivate a new posterior sampling scheme for unconditional diffusion models. We present empirical evidence supporting the effectiveness of our method.
Bio: François Rozet is a PhD student in deep learning under the supervision of Prof. Gilles Louppe at the University of Liège in Belgium. His research mainly consists in developing and applying deep learning methods to Bayesian inference problems in large-scale dynamical systems (oceans, atmospheres, ...). His work lies at the intersection of many subjects, most notably density estimation, generative modeling and inverse problems.