Asymptotic Learning Curves for Diffusion Models with Random Features Score and Manifold Data

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: George, Anand Jerry, Macris, Nicolas
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911587676192768
author George, Anand Jerry
Macris, Nicolas
author_facet George, Anand Jerry
Macris, Nicolas
contents We study the theoretical behavior of denoising score matching--the learning task associated to diffusion models--when the data distribution is supported on a low-dimensional manifold and the score is parameterized using a random feature neural network. We derive asymptotically exact expressions for the test, train, and score errors in the high-dimensional limit. Our analysis reveals that, for linear manifolds the sample complexity required to learn the score function scales linearly with the intrinsic dimension of the manifold, rather than with the ambient dimension. Perhaps surprisingly, the benefits of low-dimensional structure starts to diminish once we have a non-linear manifold. These results indicate that diffusion models can benefit from structured data; however, the dependence on the specific type of structure is subtle and intricate.
format Preprint
id arxiv_https___arxiv_org_abs_2603_22962
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Asymptotic Learning Curves for Diffusion Models with Random Features Score and Manifold Data
George, Anand Jerry
Macris, Nicolas
Machine Learning
We study the theoretical behavior of denoising score matching--the learning task associated to diffusion models--when the data distribution is supported on a low-dimensional manifold and the score is parameterized using a random feature neural network. We derive asymptotically exact expressions for the test, train, and score errors in the high-dimensional limit. Our analysis reveals that, for linear manifolds the sample complexity required to learn the score function scales linearly with the intrinsic dimension of the manifold, rather than with the ambient dimension. Perhaps surprisingly, the benefits of low-dimensional structure starts to diminish once we have a non-linear manifold. These results indicate that diffusion models can benefit from structured data; however, the dependence on the specific type of structure is subtle and intricate.
title Asymptotic Learning Curves for Diffusion Models with Random Features Score and Manifold Data
topic Machine Learning
url https://arxiv.org/abs/2603.22962