Break the Wall Between Homophily and Heterophily for Graph Representation Learning

Xiao Liu, Lijun Zhang, Hui Guan

Keywords: Graph Neural Networks, Graph Homophily, Graph Heterophily

TL;DR: This work proposes a new GNN model called OGNN (Omnipotent Graph Neural Network) that extracts different aspects of graph representations to generalize well on the whole spectrum of homophily.

Abstract: Homophily and heterophily are intrinsic properties of graphs that describe whether two linked nodes share similar properties. Although many Graph Neural Network (GNN) models have been proposed, it remains unclear how to design a model so that it can generalize well to the whole spectrum of homophily. This work addresses the challenge by identifying three graph features, including the ego node feature, the aggregated node feature, and the graph structure feature, that are essential for graph representation learning. It further proposes a new GNN model called OGNN (Omnipotent Graph Neural Network) that extracts all three graph features and adaptively fuses them to achieve generalizability across the whole spectrum of homophily. Extensive experiments on both synthetic and real datasets demonstrate the superiority (average rank 1.56) of our OGNN compared with state-of-the-art methods. Our code will be available at https://*.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: Deep Learning and representational learning