Saved in:
Bibliographic Details
Main Authors: Vargas, Francisco, Padhy, Shreyas, Blessing, Denis, Nüsken, Nikolas
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2307.01050
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918012367405056
author Vargas, Francisco
Padhy, Shreyas
Blessing, Denis
Nüsken, Nikolas
author_facet Vargas, Francisco
Padhy, Shreyas
Blessing, Denis
Nüsken, Nikolas
contents Connecting optimal transport and variational inference, we present a principled and systematic framework for sampling and generative modelling centred around divergences on path space. Our work culminates in the development of the \emph{Controlled Monte Carlo Diffusion} sampler (CMCD) for Bayesian computation, a score-based annealing technique that crucially adapts both forward and backward dynamics in a diffusion model. On the way, we clarify the relationship between the EM-algorithm and iterative proportional fitting (IPF) for Schr{ö}dinger bridges, deriving as well a regularised objective that bypasses the iterative bottleneck of standard IPF-updates. Finally, we show that CMCD has a strong foundation in the Jarzinsky and Crooks identities from statistical physics, and that it convincingly outperforms competing approaches across a wide array of experiments.
format Preprint
id arxiv_https___arxiv_org_abs_2307_01050
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Transport meets Variational Inference: Controlled Monte Carlo Diffusions
Vargas, Francisco
Padhy, Shreyas
Blessing, Denis
Nüsken, Nikolas
Machine Learning
Connecting optimal transport and variational inference, we present a principled and systematic framework for sampling and generative modelling centred around divergences on path space. Our work culminates in the development of the \emph{Controlled Monte Carlo Diffusion} sampler (CMCD) for Bayesian computation, a score-based annealing technique that crucially adapts both forward and backward dynamics in a diffusion model. On the way, we clarify the relationship between the EM-algorithm and iterative proportional fitting (IPF) for Schr{ö}dinger bridges, deriving as well a regularised objective that bypasses the iterative bottleneck of standard IPF-updates. Finally, we show that CMCD has a strong foundation in the Jarzinsky and Crooks identities from statistical physics, and that it convincingly outperforms competing approaches across a wide array of experiments.
title Transport meets Variational Inference: Controlled Monte Carlo Diffusions
topic Machine Learning
url https://arxiv.org/abs/2307.01050