Harnessing Collective Structure Knowledge in Data Augmentation for Graph Neural Networks
Rongrong Ma, Guansong Pang, Ling Chen
TL;DR
CoS-GNN addresses the expressiveness gap of classical GNNs by explicitly incorporating a diverse, collective set of node- and graph-level structural features through a dual-graph message passing framework. The approach jointly learns from original node attributes and augmented structural features, then fuses this information at the graph level to produce richer representations. Empirically, CoS-GNN yields substantial improvements on graph classification, anomaly detection, and out-of-distribution generalization across 12 datasets, and demonstrates robustness to structural noise while remaining compatible with other GNN backbones and pooling methods. This work advances graph representation learning by enabling scalable, domain-adaptive structural knowledge integration, with practical impact on various graph-centric tasks.
Abstract
Graph neural networks (GNNs) have achieved state-of-the-art performance in graph representation learning. Message passing neural networks, which learn representations through recursively aggregating information from each node and its neighbors, are among the most commonly-used GNNs. However, a wealth of structural information of individual nodes and full graphs is often ignored in such process, which restricts the expressive power of GNNs. Various graph data augmentation methods that enable the message passing with richer structure knowledge have been introduced as one main way to tackle this issue, but they are often focused on individual structure features and difficult to scale up with more structure features. In this work we propose a novel approach, namely collective structure knowledge-augmented graph neural network (CoS-GNN), in which a new message passing method is introduced to allow GNNs to harness a diverse set of node- and graph-level structure features, together with original node features/attributes, in augmented graphs. In doing so, our approach largely improves the structural knowledge modeling of GNNs in both node and graph levels, resulting in substantially improved graph representations. This is justified by extensive empirical results where CoS-GNN outperforms state-of-the-art models in various graph-level learning tasks, including graph classification, anomaly detection, and out-of-distribution generalization.