Uncovering the Structural Fairness in Graph Contrastive Learning
This addresses structural unfairness in graph learning for low-degree nodes in real-world graphs with long-tailed degree distributions, offering an incremental improvement over existing GCL methods.
The paper investigates structural fairness in graph contrastive learning (GCL), finding that GCL representations are already fairer to degree bias than graph convolutional networks (GCN), with theoretical analysis linking this to community structure properties. It proposes GRADE, a novel graph augmentation method that applies different strategies to low- and high-degree nodes, validated through extensive experiments on various benchmarks.
Recent studies show that graph convolutional network (GCN) often performs worse for low-degree nodes, exhibiting the so-called structural unfairness for graphs with long-tailed degree distributions prevalent in the real world. Graph contrastive learning (GCL), which marries the power of GCN and contrastive learning, has emerged as a promising self-supervised approach for learning node representations. How does GCL behave in terms of structural fairness? Surprisingly, we find that representations obtained by GCL methods are already fairer to degree bias than those learned by GCN. We theoretically show that this fairness stems from intra-community concentration and inter-community scatter properties of GCL, resulting in a much clear community structure to drive low-degree nodes away from the community boundary. Based on our theoretical analysis, we further devise a novel graph augmentation method, called GRAph contrastive learning for DEgree bias (GRADE), which applies different strategies to low- and high-degree nodes. Extensive experiments on various benchmarks and evaluation protocols validate the effectiveness of the proposed method.