Antithetic Noise in Diffusion Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jia, Jing, Liu, Sifan, Song, Bowen, Yuan, Wei, Shen, Liyue, Wang, Guanyang
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914292493713408
author Jia, Jing
Liu, Sifan
Song, Bowen
Yuan, Wei
Shen, Liyue
Wang, Guanyang
author_facet Jia, Jing
Liu, Sifan
Song, Bowen
Yuan, Wei
Shen, Liyue
Wang, Guanyang
contents We systematically study antithetic initial noise in diffusion models, discovering that pairing each noise sample with its negation consistently produces strong negative correlation. This universal phenomenon holds across datasets, model architectures, conditional and unconditional sampling, and even other generative models such as VAEs and Normalizing Flows. To explain it, we combine experiments and theory and propose a \textit{symmetry conjecture} that the learned score function is approximately affine antisymmetric (odd symmetry up to a constant shift), supported by empirical evidence. This negative correlation leads to substantially more reliable uncertainty quantification with up to $90\%$ narrower confidence intervals. We demonstrate these gains on tasks including estimating pixel-wise statistics and evaluating diffusion inverse solvers. We also provide extensions with randomized quasi-Monte Carlo noise designs for uncertainty quantification, and explore additional applications of the antithetic noise design to improve image editing and generation diversity. Our framework is training-free, model-agnostic, and adds no runtime overhead. Code is available at https://github.com/jjia131/Antithetic-Noise-in-Diffusion-Models-page.
format Preprint
id arxiv_https___arxiv_org_abs_2506_06185
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Antithetic Noise in Diffusion Models
Jia, Jing
Liu, Sifan
Song, Bowen
Yuan, Wei
Shen, Liyue
Wang, Guanyang
Machine Learning
Numerical Analysis
Computation
We systematically study antithetic initial noise in diffusion models, discovering that pairing each noise sample with its negation consistently produces strong negative correlation. This universal phenomenon holds across datasets, model architectures, conditional and unconditional sampling, and even other generative models such as VAEs and Normalizing Flows. To explain it, we combine experiments and theory and propose a \textit{symmetry conjecture} that the learned score function is approximately affine antisymmetric (odd symmetry up to a constant shift), supported by empirical evidence. This negative correlation leads to substantially more reliable uncertainty quantification with up to $90\%$ narrower confidence intervals. We demonstrate these gains on tasks including estimating pixel-wise statistics and evaluating diffusion inverse solvers. We also provide extensions with randomized quasi-Monte Carlo noise designs for uncertainty quantification, and explore additional applications of the antithetic noise design to improve image editing and generation diversity. Our framework is training-free, model-agnostic, and adds no runtime overhead. Code is available at https://github.com/jjia131/Antithetic-Noise-in-Diffusion-Models-page.
title Antithetic Noise in Diffusion Models
topic Machine Learning
Numerical Analysis
Computation
url https://arxiv.org/abs/2506.06185