The THUG sampler

AI generated image representing samples around a manifold.

Abstract

Sampling from a probability density constrained to a manifold is of importance in numerous applications arising in statistical physics, statistics or machine learning. Sampling from such constrained densities, in particular using an MCMC approach, poses significant challenges and it is only recently that correct solutions have been proposed. The resulting algorithms can however be computationally expensive. We propose a relaxation of the problem where the support constraint is replaced with that of sampling from a small neighbourhood of the manifold. We develop a family of bespoke and efficient algorithms adapted to this problem and demonstrate empirically their computational superiority, which comes at the expense of a modest bias.

Date
Jul 17, 2022 9:00 AM — Jul 22, 2022 5:00 PM
Location
Linz, Austria
Avatar
Mauro Camara Escudero
Research Associate in Statistical Machine Learning

My research interests include approximate manifold sampling and generative models.

Related