GFairHint: Improving Individual Fairness for Graph Neural Networks via Fairness Hint

TL;DR

GFairHint is a novel approach that enhances individual fairness in Graph Neural Networks by learning fairness representations via an auxiliary task, achieving superior fairness with minimal utility loss and computational cost.

Contribution

The paper introduces GFairHint, a method that promotes individual fairness in GNNs, balancing fairness, performance, and efficiency across various models and similarity measures.

Findings

Achieves superior individual fairness in GNNs across multiple datasets.

Maintains comparable utility with significantly less computational cost.

Generalizes across different GNN models and similarity measures.

Abstract

Given the growing concerns about fairness in machine learning and the impressive performance of Graph Neural Networks (GNNs) on graph data learning, algorithmic fairness in GNNs has attracted significant attention. While many existing studies improve fairness at the group level, only a few works promote individual fairness, which renders similar outcomes for similar individuals. A desirable framework that promotes individual fairness should (1) balance between fairness and performance, (2) accommodate two commonly-used individual similarity measures (externally annotated and computed from input features), (3) generalize across various GNN models, and (4) be computationally efficient. Unfortunately, none of the prior work achieves all the desirables. In this work, we propose a novel method, GFairHint, which promotes individual fairness in GNNs and achieves all aforementioned desirables. GFairHint learns fairness representations through an auxiliary link prediction task, and then concatenates the representations with the learned node embeddings in original GNNs as a "fairness hint". Through extensive experimental investigations on five real-world graph datasets under three prevalent GNN models covering both individual similarity measures above, GFairHint achieves the best fairness results in almost all combinations of datasets with various backbone models, while generating comparable utility results, with much less computational cost compared to the previous state-of-the-art (SoTA) method.