Improving Your Graph Neural Networks: A High-Frequency Booster
This addresses performance issues in GNNs for graph-structured data, particularly in heterophilic graphs, but appears incremental as it builds on existing signal processing insights.
The paper tackles the problems of over-smoothing and heterophily in graph neural networks (GNNs) for semi-supervised node classification by incorporating high-frequency information, resulting in up to a 3.6% improvement over popular baselines.
Graph neural networks (GNNs) hold the promise of learning efficient representations of graph-structured data, and one of its most important applications is semi-supervised node classification. However, in this application, GNN frameworks tend to fail due to the following issues: over-smoothing and heterophily. The most popular GNNs are known to be focused on the message-passing framework, and recent research shows that these GNNs are often bounded by low-pass filters from a signal processing perspective. We thus incorporate high-frequency information into GNNs to alleviate this genetic problem. In this paper, we argue that the complement of the original graph incorporates a high-pass filter and propose Complement Laplacian Regularization (CLAR) for an efficient enhancement of high-frequency components. The experimental results demonstrate that CLAR helps GNNs tackle over-smoothing, improving the expressiveness of heterophilic graphs, which adds up to 3.6% improvement over popular baselines and ensures topological robustness.