Graph Self-supervised Learning: A Comprehensive Introduction and its Potential Applications (2022)

Description:
Deep learning on graphs has attracted significant interest recently, where graph neural networks (GNN) has achieved remarkable success in a large rank of graph-structured data domains, including e-commerce, traffic, drug design, friendship recommendation, and so on. However, most deep learning studies on graphs focus on (semi-) supervised learning scenarios, which require sufficient labeled data for effective network training. Their performance can be seriously degraded when labels are extremely limited. To address the shortcomings of (semi-) supervised learning, self-supervised learning (SSL) provides a promising learning paradigm that reduces the dependence on manual labels. Different from SSL on other domains like computer vision and natural language processing, SSL on graphs has an exclusive background, design ideas, and taxonomies. In this course, we will present a timely and comprehensive introduction of the existing approaches which employ SSL techniques for graph data. In doing so, we construct a unified framework that mathematically formalizes the paradigm of graph SSL, and then we will also briefly introduce our several recent works on SSL based GNN with extremely limited labels. Finally, we will discuss the remaining challenges and potential future directions in this research field.

Prerequisites:
The course requires basic knowledge of machine learning, especially, deep learning.

Organizer: Chenjuan Guo & Miao Zhang

Lecturers: Shirui Pan

ECTS: 3.0

Time: 9:00-12:00 on 23, 24, 25, 27 May 2022

Place: Zoom

Number of seats: 60

Deadline: 2 May 2022

Important information concerning PhD courses:
We have over some time experienced problems with no-show for both project and general courses. It has now reached a point where we are forced to take action. Therefore, the Doctoral School has decided to introduce a no-show fee of DKK 3.000 for each course where the student does not show up. Cancellations are accepted no later than 2 weeks before start of the course. Registered illness is of course an acceptable reason for not showing up on those days. Furthermore, all courses open for registration approximately four months before start. This can hopefully also provide new students a chance to register for courses during the year. We look forward to your registrations.