Understanding GNNs and Homophily in Dynamic Node Classification
This work addresses the problem of improving GNNs for dynamic graphs, which is incremental as it extends existing homophily analysis from static to dynamic settings.
The authors tackled the problem of understanding graph neural network (GNN) performance in dynamic node classification by introducing a new measure called dynamic homophily, which correlates with GNN discriminative performance and reveals that popular GNNs are not robust to low dynamic homophily.
Homophily, as a measure, has been critical to increasing our understanding of graph neural networks (GNNs). However, to date this measure has only been analyzed in the context of static graphs. In our work, we explore homophily in dynamic settings. Focusing on graph convolutional networks (GCNs), we demonstrate theoretically that in dynamic settings, current GCN discriminative performance is characterized by the probability that a node's future label is the same as its neighbors' current labels. Based on this insight, we propose dynamic homophily, a new measure of homophily that applies in the dynamic setting. This new measure correlates with GNN discriminative performance and sheds light on how to potentially design more powerful GNNs for dynamic graphs. Leveraging a variety of dynamic node classification datasets, we demonstrate that popular GNNs are not robust to low dynamic homophily. Going forward, our work represents an important step towards understanding homophily and GNN performance in dynamic node classification.