Self-Supervised Representation Learning

The Hong Kong University of Science and Technology (Guangzhou)

Data Science and Analytics Thrust

PhD Qualifying Examination

By Mr. Xiasi WANG

Abstract

The effectiveness of deep supervised learning heavily depends on the availability of large amounts of labeled data, which can be expensive to obtain. As a subset of unsupervised learning, self-supervised learning (SSL) has recently garnered significant attention due to its remarkable ability to learn useful features from large collections of unlabeled data. This survey provides a comprehensive review of various SSL methods with a specific focus on computer vision (CV). The reviewed SSL methods are categorized into three types based on their respective pretext tasks. In addition, the multi-view learning framework within the SSL is reviewed and discussed. Based on this framework, a novel method is proposed for learning minimal sufficient representation. Finally, future research directions of SSL are discussed.

PQE Committee

Chairperson: Dr Xin WANG(AI)

Committee Members:

Prof. Yuan YAO (Prime Supervisor)
Prof. Lianwen ZHANG (Co-Supervisor)
Dr. Wenjia WANG (DSA)