Stochastic interpolants with data-dependent couplings

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Albergo, Michael S., Goldstein, Mark, Boffi, Nicholas M., Ranganath, Rajesh, Vanden-Eijnden, Eric
Format: Preprint
Publié: 2023
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866914954976690176
author Albergo, Michael S.
Goldstein, Mark
Boffi, Nicholas M.
Ranganath, Rajesh
Vanden-Eijnden, Eric
author_facet Albergo, Michael S.
Goldstein, Mark
Boffi, Nicholas M.
Ranganath, Rajesh
Vanden-Eijnden, Eric
contents Generative models inspired by dynamical transport of measure -- such as flows and diffusions -- construct a continuous-time map between two probability densities. Conventionally, one of these is the target density, only accessible through samples, while the other is taken as a simple base density that is data-agnostic. In this work, using the framework of stochastic interpolants, we formalize how to \textit{couple} the base and the target densities, whereby samples from the base are computed conditionally given samples from the target in a way that is different from (but does preclude) incorporating information about class labels or continuous embeddings. This enables us to construct dynamical transport maps that serve as conditional generative models. We show that these transport maps can be learned by solving a simple square loss regression problem analogous to the standard independent setting. We demonstrate the usefulness of constructing dependent couplings in practice through experiments in super-resolution and in-painting.
format Preprint
id arxiv_https___arxiv_org_abs_2310_03725
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Stochastic interpolants with data-dependent couplings
Albergo, Michael S.
Goldstein, Mark
Boffi, Nicholas M.
Ranganath, Rajesh
Vanden-Eijnden, Eric
Machine Learning
Generative models inspired by dynamical transport of measure -- such as flows and diffusions -- construct a continuous-time map between two probability densities. Conventionally, one of these is the target density, only accessible through samples, while the other is taken as a simple base density that is data-agnostic. In this work, using the framework of stochastic interpolants, we formalize how to \textit{couple} the base and the target densities, whereby samples from the base are computed conditionally given samples from the target in a way that is different from (but does preclude) incorporating information about class labels or continuous embeddings. This enables us to construct dynamical transport maps that serve as conditional generative models. We show that these transport maps can be learned by solving a simple square loss regression problem analogous to the standard independent setting. We demonstrate the usefulness of constructing dependent couplings in practice through experiments in super-resolution and in-painting.
title Stochastic interpolants with data-dependent couplings
topic Machine Learning
url https://arxiv.org/abs/2310.03725