Moment Matching Denoising Gibbs Sampling
Abstract
Energy-Based Models (EBMs) offer a versatile framework for modelling complex data distributions. However, training and sampling from EBMs continue to pose significant challenges. The widely-used Denoising Score Matching (DSM) method for scalable EBM training suffers from inconsistency issues, causing the energy model to learn a noisy data distribution. In this work, we propose an efficient sampling framework: (pseudo)-Gibbs sampling with moment matching, which enables effective sampling from the underlying clean model when given a noisy model that has been well-trained via DSM. We explore the benefits of our approach compared to related methods and demonstrate how to scale the method to high-dimensional datasets.
Cite
Text
Zhang et al. "Moment Matching Denoising Gibbs Sampling." Neural Information Processing Systems, 2023.Markdown
[Zhang et al. "Moment Matching Denoising Gibbs Sampling." Neural Information Processing Systems, 2023.](https://mlanthology.org/neurips/2023/zhang2023neurips-moment/)BibTeX
@inproceedings{zhang2023neurips-moment,
title = {{Moment Matching Denoising Gibbs Sampling}},
author = {Zhang, Mingtian and Hawkins-Hooker, Alex and Paige, Brooks and Barber, David},
booktitle = {Neural Information Processing Systems},
year = {2023},
url = {https://mlanthology.org/neurips/2023/zhang2023neurips-moment/}
}