Saved in:
Bibliographic Details
Main Author: Hyvärinen, Aapo
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2410.05837
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914046561746944
author Hyvärinen, Aapo
author_facet Hyvärinen, Aapo
contents The Langevin algorithm is a classic method for sampling from a given pdf in a real space. In its basic version, it only requires knowledge of the gradient of the log-density, also called the score function. However, in deep learning, it is often easier to learn the so-called "noisy-data score function", i.e. the gradient of the log-density of noisy data, more precisely when Gaussian noise is added to the data. Such an estimate is biased and complicates the use of the Langevin method. Here, we propose a noise-corrected version of the Langevin algorithm, where the bias due to noisy data is removed, at least regarding first-order terms. Unlike diffusion models, our algorithm only needs to know the noisy score function for one single noise level. We further propose a simple special case which has an interesting intuitive interpretation of iteratively adding noise the data and then attempting to remove half of that noise.
format Preprint
id arxiv_https___arxiv_org_abs_2410_05837
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A noise-corrected Langevin algorithm and sampling by half-denoising
Hyvärinen, Aapo
Machine Learning
The Langevin algorithm is a classic method for sampling from a given pdf in a real space. In its basic version, it only requires knowledge of the gradient of the log-density, also called the score function. However, in deep learning, it is often easier to learn the so-called "noisy-data score function", i.e. the gradient of the log-density of noisy data, more precisely when Gaussian noise is added to the data. Such an estimate is biased and complicates the use of the Langevin method. Here, we propose a noise-corrected version of the Langevin algorithm, where the bias due to noisy data is removed, at least regarding first-order terms. Unlike diffusion models, our algorithm only needs to know the noisy score function for one single noise level. We further propose a simple special case which has an interesting intuitive interpretation of iteratively adding noise the data and then attempting to remove half of that noise.
title A noise-corrected Langevin algorithm and sampling by half-denoising
topic Machine Learning
url https://arxiv.org/abs/2410.05837