Riemannian Score-Based Generative Modelling - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2023

Riemannian Score-Based Generative Modelling

Valentin de Bortoli
  • Fonction : Auteur
  • PersonId : 1084097
  • IdRef : 249460300
Emile Mathieu
  • Fonction : Auteur
Michael Hutchinson
  • Fonction : Auteur
James Thornton
  • Fonction : Auteur
Yee Whye Teh
  • Fonction : Auteur
Arnaud Doucet
  • Fonction : Auteur

Résumé

Score-based generative models (SGMs) are a powerful class of generative models that exhibit remarkable empirical performance. Score-based generative modelling (SGM) consists of a ``noising'' stage, whereby a diffusion is used to gradually add Gaussian noise to data, and a generative model, which entails a ``denoising'' process defined by approximating the time-reversal of the diffusion. Existing SGMs assume that data is supported on a Euclidean space, i.e. a manifold with flat geometry. In many domains such as robotics, geoscience or protein modelling, data is often naturally described by distributions living on Riemannian manifolds and current SGM techniques are not appropriate. We introduce here Riemannian Score-based Generative Models (RSGMs), a class of generative models extending SGMs to Riemannian manifolds. We demonstrate our approach on a variety of manifolds, and in particular with earth and climate science spherical data.

Dates et versions

hal-03945480 , version 1 (18-01-2023)

Identifiants

Citer

Valentin de Bortoli, Emile Mathieu, Michael Hutchinson, James Thornton, Yee Whye Teh, et al.. Riemannian Score-Based Generative Modelling. 2023. ⟨hal-03945480⟩
12 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More