From stability of Langevin diffusion to convergence of proximal MCMC for non-log-concave sampling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Renaud, Marien, De Bortoli, Valentin, Leclaire, Arthur, Papadakis, Nicolas
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915469636665344
author Renaud, Marien
De Bortoli, Valentin
Leclaire, Arthur
Papadakis, Nicolas
author_facet Renaud, Marien
De Bortoli, Valentin
Leclaire, Arthur
Papadakis, Nicolas
contents We consider the problem of sampling distributions stemming from non-convex potentials with Unadjusted Langevin Algorithm (ULA). We prove the stability of the discrete-time ULA to drift approximations under the assumption that the potential is strongly convex at infinity. In many context, e.g. imaging inverse problems, potentials are non-convex and non-smooth. Proximal Stochastic Gradient Langevin Algorithm (PSGLA) is a popular algorithm to handle such potentials. It combines the forward-backward optimization algorithm with a ULA step. Our main stability result combined with properties of the Moreau envelope allows us to derive the first proof of convergence of the PSGLA for non-convex potentials. We empirically validate our methodology on synthetic data and in the context of imaging inverse problems. In particular, we observe that PSGLA exhibits faster convergence rates than Stochastic Gradient Langevin Algorithm for posterior sampling while preserving its restoration properties.
format Preprint
id arxiv_https___arxiv_org_abs_2505_14177
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle From stability of Langevin diffusion to convergence of proximal MCMC for non-log-concave sampling
Renaud, Marien
De Bortoli, Valentin
Leclaire, Arthur
Papadakis, Nicolas
Machine Learning
Computer Vision and Pattern Recognition
We consider the problem of sampling distributions stemming from non-convex potentials with Unadjusted Langevin Algorithm (ULA). We prove the stability of the discrete-time ULA to drift approximations under the assumption that the potential is strongly convex at infinity. In many context, e.g. imaging inverse problems, potentials are non-convex and non-smooth. Proximal Stochastic Gradient Langevin Algorithm (PSGLA) is a popular algorithm to handle such potentials. It combines the forward-backward optimization algorithm with a ULA step. Our main stability result combined with properties of the Moreau envelope allows us to derive the first proof of convergence of the PSGLA for non-convex potentials. We empirically validate our methodology on synthetic data and in the context of imaging inverse problems. In particular, we observe that PSGLA exhibits faster convergence rates than Stochastic Gradient Langevin Algorithm for posterior sampling while preserving its restoration properties.
title From stability of Langevin diffusion to convergence of proximal MCMC for non-log-concave sampling
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.14177