Geometric Regularity in Deterministic Sampling Dynamics of Diffusion-based Generative Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chen, Defang, Zhou, Zhenyu, Wang, Can, Lyu, Siwei
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914193412718592
author Chen, Defang
Zhou, Zhenyu
Wang, Can
Lyu, Siwei
author_facet Chen, Defang
Zhou, Zhenyu
Wang, Can
Lyu, Siwei
contents Diffusion-based generative models employ stochastic differential equations (SDEs) and their equivalent probability flow ordinary differential equations (ODEs) to establish a smooth transformation between complex high-dimensional data distributions and tractable prior distributions. In this paper, we reveal a striking geometric regularity in the deterministic sampling dynamics of diffusion generative models: each simulated sampling trajectory along the gradient field lies within an extremely low-dimensional subspace, and all trajectories exhibit an almost identical boomerang shape, regardless of the model architecture, applied conditions, or generated content. We characterize several intriguing properties of these trajectories, particularly under closed-form solutions based on kernel-estimated data modeling. We also demonstrate a practical application of the discovered trajectory regularity by proposing a dynamic programming-based scheme to better align the sampling time schedule with the underlying trajectory structure. This simple strategy requires minimal modification to existing deterministic numerical solvers, incurs negligible computational overhead, and achieves superior image generation performance, especially in regions with only 5 - 10 function evaluations.
format Preprint
id arxiv_https___arxiv_org_abs_2506_10177
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Geometric Regularity in Deterministic Sampling Dynamics of Diffusion-based Generative Models
Chen, Defang
Zhou, Zhenyu
Wang, Can
Lyu, Siwei
Machine Learning
Statistical Mechanics
Computer Vision and Pattern Recognition
Diffusion-based generative models employ stochastic differential equations (SDEs) and their equivalent probability flow ordinary differential equations (ODEs) to establish a smooth transformation between complex high-dimensional data distributions and tractable prior distributions. In this paper, we reveal a striking geometric regularity in the deterministic sampling dynamics of diffusion generative models: each simulated sampling trajectory along the gradient field lies within an extremely low-dimensional subspace, and all trajectories exhibit an almost identical boomerang shape, regardless of the model architecture, applied conditions, or generated content. We characterize several intriguing properties of these trajectories, particularly under closed-form solutions based on kernel-estimated data modeling. We also demonstrate a practical application of the discovered trajectory regularity by proposing a dynamic programming-based scheme to better align the sampling time schedule with the underlying trajectory structure. This simple strategy requires minimal modification to existing deterministic numerical solvers, incurs negligible computational overhead, and achieves superior image generation performance, especially in regions with only 5 - 10 function evaluations.
title Geometric Regularity in Deterministic Sampling Dynamics of Diffusion-based Generative Models
topic Machine Learning
Statistical Mechanics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.10177