Breaking the Entanglement of Homophily and Heterophily in Semi-supervised Node Classification

arXiv:2312.04111v2 Announce Type: replace
Abstract: Recently, graph neural networks (GNNs) have shown prominent performance in semi-supervised node classification by leveraging knowledge from the graph database. However, most existing GNNs follow the homophily assumption, where connected nodes are more likely to exhibit similar feature distributions and the same labels, and such an assumption has proven to be vulnerable in a growing number of practical applications. As a supplement, heterophily reflects dissimilarity in connected nodes, which has gained significant attention in graph …

abstract arxiv breaking classification cs.ai cs.lg cs.si database entanglement feature gnns graph graph database graph neural networks however knowledge labels networks neural networks node nodes performance semi-supervised type