第一个人 (2015)
最早是GAN。但是它似乎只能学习部分类别。例如用动物图片训练，最终只能生成狗。
Sohl-Dickstein当时是斯坦福的博士后。他喜好nonequilibrium thermodynamics。
Crucially, each step is reversible — with small enough steps, you can go from a simple distribution back to a complex one.
前向过程产生一个分布可以从中采样。
“The sequence of transformations very slowly turns your data distribution into just a big noise ball,” said Sohl-Dickstein. This “forward process” leaves you with a distribution you can sample from with ease.
花了好几个月：
Sohl-Dickstein recalls the first outputs of his diffusion model. “You’d squint and be like, ‘I think that colored blob looks like a truck,’” he said. “I’d spent so many months of my life staring at different patterns of pixels and trying to see structure that I was like, ‘This is way more structured than I’d ever gotten before.’ I was very excited.”
结果：图像效果不好。生成速度慢。
第二个人 宋颺 (2019)
In 2019, he and his adviser published a novel method for building generative models that didn’t estimate the probability distribution of the data (the high-dimensional surface). Instead, it estimated the gradient of the distribution (think of it as the slope of the high-dimensional surface).
也就是估计
第三个人 Jonathan Ho (2020)
把上述2人的工作结合起来。发明了DDPM。[1]
