BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Bayesian imaging with deep generative priors - Marcelo Pereyra (He
 riot-Watt University)
DTSTART:20230620T130000Z
DTEND:20230620T140000Z
UID:TALK201550@talks.cam.ac.uk
DESCRIPTION:This talk presents a new Bayesian analysis and computation met
 hodology to perform inference in high-dimensional problems where the prior
  knowledge is available in the form of a dataset of training examples\, wh
 ich we consider to be a sample from the marginal distribution of the unkno
 wn quantity of interest. Following the manifold hypothesis\, which states 
 that high-dimensional physical quantities encountered in the real world of
 ten lie along a low-dimensional latent manifold inside the ambient space\,
  we construct a prior distribution that is supported on a low-dimensional 
 manifold which is encoded by a deep neural network. The manifold and the d
 istribution supported on the manifold can then be learnt from the training
  data by using modern machine learning techniques for generative modelling
 . We study the resulting Bayesian models theoretically and empirically by 
 using a range of challenging imaging inverse problems and where we perform
  analyses related to uncertainty quantification\, hypothesis testing\, and
  model selection in the absence of ground truth.
LOCATION:Seminar Room 2\, Newton Institute
END:VEVENT
END:VCALENDAR