Improving Fairness in Graph Neural Networks via Counterfactual Debiasing

TL;DR

This paper introduces Fair-ICD, a novel counterfactual data augmentation method for GNNs that improves fairness by creating diverse neighborhoods and using adversarial training, outperforming existing bias mitigation techniques.

Contribution

The paper proposes a new counterfactual data augmentation approach combined with adversarial training to mitigate bias in GNNs, addressing limitations of existing filtering methods.

Findings

Fair-ICD improves fairness metrics significantly.

The method maintains high predictive accuracy.

Experiments on standard datasets validate effectiveness.

Abstract

Graph Neural Networks (GNNs) have been successful in modeling graph-structured data. However, similar to other machine learning models, GNNs can exhibit bias in predictions based on attributes like race and gender. Moreover, bias in GNNs can be exacerbated by the graph structure and message-passing mechanisms. Recent cutting-edge methods propose mitigating bias by filtering out sensitive information from input or representations, like edge dropping or feature masking. Yet, we argue that such strategies may unintentionally eliminate non-sensitive features, leading to a compromised balance between predictive accuracy and fairness. To tackle this challenge, we present a novel approach utilizing counterfactual data augmentation for bias mitigation. This method involves creating diverse neighborhoods using counterfactuals before message passing, facilitating unbiased node representations learning from the augmented graph. Subsequently, an adversarial discriminator is employed to diminish bias in predictions by conventional GNN classifiers. Our proposed technique, Fair-ICD, ensures the fairness of GNNs under moderate conditions. Experiments on standard datasets using three GNN backbones demonstrate that Fair-ICD notably enhances fairness metrics while preserving high predictive performance.