First Hitting Diffusion Models for Generating Manifold, Graph and Categorical Data
Abstract
We propose a family of First Hitting Diffusion Models (FHDM), deep generative models that generate data with a diffusion process that terminates at a random first hitting time. This yields an extension of the standard fixed-time diffusion models that terminate at a pre-specified deterministic time. Although standard diffusion models are designed for continuous unconstrained data, FHDM is naturally designed to learn distributions on continuous as well as a range of discrete and structure domains.
Cite
Text
Ye et al. "First Hitting Diffusion Models for Generating Manifold, Graph and Categorical Data." NeurIPS 2022 Workshops: SBM, 2022.Markdown
[Ye et al. "First Hitting Diffusion Models for Generating Manifold, Graph and Categorical Data." NeurIPS 2022 Workshops: SBM, 2022.](https://mlanthology.org/neuripsw/2022/ye2022neuripsw-first/)BibTeX
@inproceedings{ye2022neuripsw-first,
title = {{First Hitting Diffusion Models for Generating Manifold, Graph and Categorical Data}},
author = {Ye, Mao and Wu, Lemeng and Liu, Qiang},
booktitle = {NeurIPS 2022 Workshops: SBM},
year = {2022},
url = {https://mlanthology.org/neuripsw/2022/ye2022neuripsw-first/}
}