A solvable generative model with a linear, one-step denoiser

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
1. Verfasser: Halder, Indranil
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913978151600128
author Halder, Indranil
author_facet Halder, Indranil
contents We develop an analytically tractable single-step diffusion model based on a linear denoiser and present an explicit formula for the Kullback-Leibler divergence between the generated and sampling distribution, taken to be isotropic Gaussian, showing the effect of finite diffusion time and noise scale. Our study further reveals that the monotonic fall phase of Kullback-Leibler divergence begins when the training dataset size reaches the dimension of the data points. Finally, for large-scale practical diffusion models, we explain why a higher number of diffusion steps enhances production quality based on the theoretical arguments presented before.
format Preprint
id arxiv_https___arxiv_org_abs_2411_17807
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A solvable generative model with a linear, one-step denoiser
Halder, Indranil
Machine Learning
Computer Vision and Pattern Recognition
We develop an analytically tractable single-step diffusion model based on a linear denoiser and present an explicit formula for the Kullback-Leibler divergence between the generated and sampling distribution, taken to be isotropic Gaussian, showing the effect of finite diffusion time and noise scale. Our study further reveals that the monotonic fall phase of Kullback-Leibler divergence begins when the training dataset size reaches the dimension of the data points. Finally, for large-scale practical diffusion models, we explain why a higher number of diffusion steps enhances production quality based on the theoretical arguments presented before.
title A solvable generative model with a linear, one-step denoiser
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2411.17807