Guardado en:
Detalles Bibliográficos
Autores principales: Yu, Zhendong, Huang, Haiping
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2405.11932
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917883753267200
author Yu, Zhendong
Huang, Haiping
author_facet Yu, Zhendong
Huang, Haiping
contents Generative diffusion models apply the concept of Langevin dynamics in physics to machine leaning, attracting a lot of interests from engineering, statistics and physics, but a complete picture about inherent mechanisms is still lacking. In this paper, we provide a transparent physics analysis of diffusion models, formulating the fluctuation theorem, entropy production, equilibrium measure, and Franz-Parisi potential to understand the dynamic process and intrinsic phase transitions. Our analysis is rooted in a path integral representation of both forward and backward dynamics, and in treating the reverse diffusion generative process as a statistical inference, where the time-dependent state variables serve as quenched disorder akin to that in spin glass theory. Our study thus links stochastic thermodynamics, statistical inference and geometry based analysis together to yield a coherent picture about how the generative diffusion models work.
format Preprint
id arxiv_https___arxiv_org_abs_2405_11932
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Nonequilbrium physics of generative diffusion models
Yu, Zhendong
Huang, Haiping
Statistical Mechanics
Disordered Systems and Neural Networks
Machine Learning
Generative diffusion models apply the concept of Langevin dynamics in physics to machine leaning, attracting a lot of interests from engineering, statistics and physics, but a complete picture about inherent mechanisms is still lacking. In this paper, we provide a transparent physics analysis of diffusion models, formulating the fluctuation theorem, entropy production, equilibrium measure, and Franz-Parisi potential to understand the dynamic process and intrinsic phase transitions. Our analysis is rooted in a path integral representation of both forward and backward dynamics, and in treating the reverse diffusion generative process as a statistical inference, where the time-dependent state variables serve as quenched disorder akin to that in spin glass theory. Our study thus links stochastic thermodynamics, statistical inference and geometry based analysis together to yield a coherent picture about how the generative diffusion models work.
title Nonequilbrium physics of generative diffusion models
topic Statistical Mechanics
Disordered Systems and Neural Networks
Machine Learning
url https://arxiv.org/abs/2405.11932