PhD Dissertation Proposal by WEN Zhihao | Generalizing Graph Neural Network across Graphs and Time

Graph-structured data widely exist in diverse real-world scenarios, such as social networks, e-commerce graphs, citation graphs, and biological networks. Analysis of these graphs can uncover valuable insights about their respective application domains. Meanwhile, graph representation learning provides an effective way to the graph analytics problem.

However, most previous works focused on learning node representation from a single fixed graph, while many real-world scenarios require representations to be quickly generated for unseen nodes, new edges, or entirely new graphs. This inductive ability is essential for high-throughtput machine learning systems, operating on evolving graphs and constantly meeting unseen nodes (e.g., posts on Wikipedia, users and items on Amazon). This inductive graph representation problem is quite difficult, compared to the transductive setting, for that generalizing to unseen nodes requires new subgraphs containing the new nodes to be aligned to the neural network trained already. And We further explore the inductive graph representation learning from two more detailed perspectives: (1) Generalizing GNN across graphs, in which we tackle with the problem of semi-supervised node classification across graphs; (2) Generalizing GNN across time, in which we mainly solve the problem of temporal link prediction.

WEN Zhihao is a Ph.D. candidate in the School of Computing and Information Systems, Singapore Management University, supervised by Assistant Professor FANG Yuan. He received his Bachelor's Dual-Degree in Electronic Engineering & E-commerce from University of Electronic Science and Technology of China. His research mainly focuses on graph neural network generalization.