A Survey on Image Clustering with Deep Learning

PhD Qualifying Examination


Title: "A Survey on Image Clustering with Deep Learning"

by

Mr. Xingzhi ZHOU


Abstract:

Clustering is a fundamental unsupervised machine learning problem that aims to 
group instances without any supervised signal. Clustering can discover 
underlying structures and has practical applications in various fields, such as 
social analysis, bioinformatics, and computer vision. However, traditional 
clustering algorithms struggle in handling complex and high-dimensional data, 
such as images, videos, and graphs. Because these algorithms are unable to 
effectively model semantic distances between instances. In this survey, we 
provide an overview of deep image clustering from the perspective of 
representation learning modules. We focus on how these modules address the 
challenge of modeling semantic distance in images through the use of deep 
neural networks. We explore various deep representation learning techniques for 
image clustering, including autoencoders, maximizing mutual information, 
generative models, and self-supervised learning. Finally, we discuss the 
challenges that arise in deep image clustering and suggest promising directions 
for future research.


Date:                   Tuesday, 18 April 2023

Time:                   10:00am - 12:00noon

Venue:                  Room 4475
                         Lifts 25/26

Committee Members:      Prof. Nevin Zhang (Supervisor)
                         Prof. Raymond Wong (Chairperson)
 			Dr. Hao Chen
 			Dr. Dan Xu


**** ALL are Welcome ****