Deep Unsupervised Learning using Nonequilibrium Thermodynamics

Author: Jascha Sohl-Dickstein
Link: http://proceedings.mlr.press/v37/sohl-dickstein15.pdf
Score: ⭐️⭐️⭐️⭐️⭐️
Status: Done
Type: Academic Journal
备注: 首篇扩散模型论文

A central problem in machine learning involves modeling complex data-sets using highly flexible families of probability distributions in which learning, sampling, inference, and evaluation are still analytically or computationally tractable. Here, we develop an approach that simultaneously achieves both flexibility and tractability. The essential idea, inspired by non-equilibrium statistical physics, is to systematically and slowly destroy structure in a data distribution through an iterative forward diffusion process. We then learn a reverse diffusion process that restores structure in data, yielding a highly flexible and tractable generative