BriefGPT.xyz
Feb, 2021
改进降噪扩散概率模型
Improved Denoising Diffusion Probabilistic Models
HTML
PDF
Alex Nichol, Prafulla Dhariwal
TL;DR
通过对DDPM模型进行简单修改，可以在保持高质量样本的同时达到具有竞争力的对数似然值，并学习反向扩散过程的方差，从而使用数量级更少的正向传递采样。使用精度和召回率比较DDPM和GAN模型的性能，并证明这些模型的样本质量和似然值可以与模型容量和训练计算平稳地提高。
Abstract
Denoising diffusion probabilistic models (
ddpm
) are a class of
generative models
which have recently been shown to produce excellent samples. We show that with a few simple modifications, DDPMs can also achieve c
→