From Denoising Diffusions to Denoising Markov Models
From Denoising Diffusions to Denoising Markov Models
复制标题
DOI：
10.48550/arxiv.2211.03595
复制
发表时间：
2022-11
期刊：
影响因子：
--
通讯作者：
Joe Benton;Yuyang Shi;Valentin De Bortoli;George Deligiannidis;A. Doucet
中科院分区：
文献类型：
--
作者：
Joe Benton;Yuyang Shi;Valentin De Bortoli;George Deligiannidis;A. Doucet
去噪扩散是最先进的生成模型，表现出显着的经验性能。它们的工作原理是将数据分布扩散到高斯分布中，然后学习如何逆转这种噪声过程以获得合成数据点。去噪扩散依赖于使用分数匹配的噪声数据密度的对数导数的近似。当只能从先验和似然中采样时，这种模型也可以用于执行近似后验模拟。我们提出了一个统一的框架，将这种方法推广到广泛的一类空间，并导致原来的扩展分数匹配。我们说明了各种应用程序的模型。
Denoising diffusions are state-of-the-art generative models exhibiting remarkable empirical performance. They work by diffusing the data distribution into a Gaussian distribution and then learning to reverse this noising process to obtain synthetic datapoints. The denoising diffusion relies on approximations of the logarithmic derivatives of the noised data densities using score matching. Such models can also be used to perform approximate posterior simulation when one can only sample from the prior and likelihood. We propose a unifying framework generalising this approach to a wide class of spaces and leading to an original extension of score matching. We illustrate the resulting models on various applications.