TAM: Topology-Aware Margin Loss for Class-Imbalanced Node Classification
This work addresses class imbalance in graph node classification, which is a domain-specific problem for graph learning, and is incremental as it builds on existing methods by incorporating local topology.
The paper tackles the problem of learning unbiased node representations in class-imbalanced graph data by addressing false positives for major nodes caused by ignoring local topology. The proposed Topology-Aware Margin (TAM) method consistently outperforms baselines on various benchmark datasets with GNN architectures.
Learning unbiased node representations under class-imbalanced graph data is challenging due to interactions between adjacent nodes. Existing studies have in common that they compensate the minor class nodes `as a group' according to their overall quantity (ignoring node connections in graph), which inevitably increase the false positive cases for major nodes. We hypothesize that the increase in these false positive cases is highly affected by the label distribution around each node and confirm it experimentally. In addition, in order to handle this issue, we propose Topology-Aware Margin (TAM) to reflect local topology on the learning objective. Our method compares the connectivity pattern of each node with the class-averaged counter-part and adaptively adjusts the margin accordingly based on that. Our method consistently exhibits superiority over the baselines on various node classification benchmark datasets with representative GNN architectures.