Date: 06/08/2021, 2-3 pm
Deep Graph Contrastive Learning
Self-supervised learning (SSL) has been extensively studied to alleviate the label sparsity problem in deep models. Recent self-supervised learning techniques are converging around the central theme of Contrastive Learning (CL), which aims to maximize the consistency of representations under multiple views of the input data. However, the development of contrastive learning for graph-structured data is still in its infancy. In this presentation, I will present recent advances in graph contrastive learning. In particular, I will present a general graph CL paradigm, and some recent works on graph CL. In addition, I will share some ideas and outline future directions regarding this topic.
Song Xiangyu received his Bachelor’s degree from Beijing Jiaotong University in China in 2013. Currently, he is a PhD student in the School of Information Technology at Deakin University, Australia. His research interests include data mining and analytics, learning analytics and deep learning.