Building Shortcuts between Distant Nodes with Biaffine Mapping for Graph Convolutional Networks
This addresses a key bottleneck in graph neural networks for researchers and practitioners, offering a shallow yet expressive alternative to deep architectures, though it is incremental as it builds on existing GCN frameworks.
The paper tackles the paradox in graph convolutional networks where shallow architectures limit learning from high-order neighbors and deep ones suffer from over-smoothing, by introducing biaffine mapping to enable shallow GCNs to capture long-distance dependencies with one-hop message passing, resulting in significant outperformance over state-of-the-art GCNs on semi-supervised node classification across nine benchmark datasets.
Multiple recent studies show a paradox in graph convolutional networks (GCNs), that is, shallow architectures limit the capability of learning information from high-order neighbors, while deep architectures suffer from over-smoothing or over-squashing. To enjoy the simplicity of shallow architectures and overcome their limits of neighborhood extension, in this work, we introduce Biaffine technique to improve the expressiveness of graph convolutional networks with a shallow architecture. The core design of our method is to learn direct dependency on long-distance neighbors for nodes, with which only one-hop message passing is capable of capturing rich information for node representation. Besides, we propose a multi-view contrastive learning method to exploit the representations learned from long-distance dependencies. Extensive experiments on nine graph benchmark datasets suggest that the shallow biaffine graph convolutional networks (BAGCN) significantly outperforms state-of-the-art GCNs (with deep or shallow architectures) on semi-supervised node classification. We further verify the effectiveness of biaffine design in node representation learning and the performance consistency on different sizes of training data.