Training Neural Samplers with Reverse Diffusive KL Divergence

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: He, Jiajun, Chen, Wenlin, Zhang, Mingtian, Barber, David, Hernández-Lobato, José Miguel
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912257392246784
author He, Jiajun
Chen, Wenlin
Zhang, Mingtian
Barber, David
Hernández-Lobato, José Miguel
author_facet He, Jiajun
Chen, Wenlin
Zhang, Mingtian
Barber, David
Hernández-Lobato, José Miguel
contents Training generative models to sample from unnormalized density functions is an important and challenging task in machine learning. Traditional training methods often rely on the reverse Kullback-Leibler (KL) divergence due to its tractability. However, the mode-seeking behavior of reverse KL hinders effective approximation of multi-modal target distributions. To address this, we propose to minimize the reverse KL along diffusion trajectories of both model and target densities. We refer to this objective as the reverse diffusive KL divergence, which allows the model to capture multiple modes. Leveraging this objective, we train neural samplers that can efficiently generate samples from the target distribution in one step. We demonstrate that our method enhances sampling performance across various Boltzmann distributions, including both synthetic multi-modal densities and n-body particle systems.
format Preprint
id arxiv_https___arxiv_org_abs_2410_12456
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Training Neural Samplers with Reverse Diffusive KL Divergence
He, Jiajun
Chen, Wenlin
Zhang, Mingtian
Barber, David
Hernández-Lobato, José Miguel
Machine Learning
Training generative models to sample from unnormalized density functions is an important and challenging task in machine learning. Traditional training methods often rely on the reverse Kullback-Leibler (KL) divergence due to its tractability. However, the mode-seeking behavior of reverse KL hinders effective approximation of multi-modal target distributions. To address this, we propose to minimize the reverse KL along diffusion trajectories of both model and target densities. We refer to this objective as the reverse diffusive KL divergence, which allows the model to capture multiple modes. Leveraging this objective, we train neural samplers that can efficiently generate samples from the target distribution in one step. We demonstrate that our method enhances sampling performance across various Boltzmann distributions, including both synthetic multi-modal densities and n-body particle systems.
title Training Neural Samplers with Reverse Diffusive KL Divergence
topic Machine Learning
url https://arxiv.org/abs/2410.12456