<ul><li>Graph Neural Networks (GNNs) struggle with heterophilic data, where connected nodes may have dissimilar labels, due to homophily assumption and local message passing.</li><li>A new approach proposes creating alternative graph structures by linking nodes with similar structural attributes to improve label homophily.</li><li>Theoretical proof suggests GNN performance improvement by utilizing graphs with fewer false positive edges and considering multiple graph views.</li><li>Structure-Guided GNN (SG-GNN) is introduced as an architecture that processes original and newly created structural graphs to achieve state-of-the-art or highly competitive performance on datasets with heterophilic characteristics.</li></ul>

Adapting to Heterophilic Graph Data with Structure-Guided Neighbor Discovery

Discover more