Abstract

Diffusion models have emerged as the new state-of-the-art family of deep generative models. They are designed to generate high-quality images by iteratively refining a random noise input; They outperforms many classic sota models in terms of performance; Several improved versions are also emerging, such as consistency models. The purpose of this report is to let people quickly understand the principle, the relation with other models, and stimulate the research interest. At the same time, several classical papers are interpreted, and the application of the API of TensorFlow implementation of diffusion models is demonstrated. Hopefully diffusion models will help you solve the problem.

Speaker Intro

Congwei Song received the master degree in applied mathematics from the Institute of Science in Zhejiang University of Technology, and the Ph.D. degree in basic mathematics from the Department of Mathematics, Zhejiang University, worked in Zhijiang College of Zhejiang University of Technology as an assistant from 2014 to 2021, from 2021 on, worked in BIMSA as asistant researcher. His research interests include machine learning, as well as wavelet analysis and harmonic analysis.