Data Augmentation with Diffusion Model

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering

Final Year Thesis Oral Defense

Title: "Data Augmentation with Diffusion Model"

by

TSE Wai Chung

Abstract:

The recent development of diffusion models fostered breakthroughs in many 
machine learning fields. Some new research have shown potential in applying 
diffusion model to effectively augment data. In this research, we propose a 
novel approach, applying a new embedding replacement style transfer method for 
vision data augmentation with diffusion models. Experimentally our method 
enhances the downstream classifier's domain generalization ability and computes 
faster than some current methods.


Date            : 25 April 2024 (Thursday)

Time            : 13:45 - 14:25

Venue           : Room 5501 (near lifts 25/26), HKUST

Advisor         : Prof. ZHANG Nevin Lianwen

2nd Reader      : Dr. CHEN Long