TechRxiv
STC.pdf (1020.89 kB)

Star Topology Convolution for Graph Representation Learning

Download (1020.89 kB)
preprint
posted on 02.08.2021, 21:55 by Chong WuChong Wu, Zhenan Feng, Jiangbin Zheng, Houwang Zhang, Jiawang Cao, Hong YAN

We present a novel graph convolutional method called star topology convolution (STC). This method makes graph convolution more similar to conventional convolutional neural networks (CNNs) in Euclidean feature spaces. STC learns subgraphs which have a star topology rather than learning a fixed graph like most spectral methods. Due to the properties of a star topology, STC is graph-scale free (without a fixed graph size constraint). It has fewer parameters in its convolutional filter and is inductive, so it is more flexible and can be applied to large and evolving graphs. The convolutional filter is learnable and localized, similar to CNNs in Euclidean feature spaces, and maintains a good weight sharing property. To test the method, STC was compared with state-of-the-art graph convolutional methods in a supervised learning setting on six node properties prediction benchmark datasets: Cora, Citeseer, Pubmed, PPI, Ogbn-Arxiv, and Ogbn-MAG. The experimental results showed that STC achieved state-of-the-art performance on all these datasets and maintained good robustness. In an essential protein identification task, STC outperformed state-of-the-art essential protein identification methods.

Funding

Hong Kong Research Grants Council (Project 11200818)

City University of Hong Kong (Project 9610460)

Hong Kong Innovation and Technology Commission

History

Email Address of Submitting Author

chongwu2-c@my.cityu.edu.hk

ORCID of Submitting Author

0000-0003-3405-742X

Submitting Author's Institution

Department of Electrical Engineering, City University of Hong Kong

Submitting Author's Country

China