Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FnRGNN: Distribution-aware Fairness in Graph Neural Network

Soyoung Park, Sungsu Lim

TL;DR

FnRGNN tackles fairness in node-level regression with graph neural networks by introducing a distribution-aware in-processing framework. It interlocks three interventions—structure-level edge reweighting, representation-level MMD alignment, and prediction-level distribution matching via Sinkhorn divergence—into a unified training objective. Empirical results on four real-world datasets show reduced group disparities (MG, VG, WD) while preserving predictive accuracy, outperforming baselines across structure, representation, and prediction levels. The approach advances fairness in continuous graph predictions and offers practical potential for high-stakes applications, with code available for reproducibility.

Abstract

Graph Neural Networks (GNNs) excel at learning from structured data, yet fairness in regression tasks remains underexplored. Existing approaches mainly target classification and representation-level debiasing, which cannot fully address the continuous nature of node-level regression. We propose FnRGNN, a fairness-aware in-processing framework for GNN-based node regression that applies interventions at three levels: (i) structure-level edge reweighting, (ii) representation-level alignment via MMD, and (iii) prediction-level normalization through Sinkhorn-based distribution matching. This multi-level strategy ensures robust fairness under complex graph topologies. Experiments on four real-world datasets demonstrate that FnRGNN reduces group disparities without sacrificing performance. Code is available at https://github.com/sybeam27/FnRGNN.

FnRGNN: Distribution-aware Fairness in Graph Neural Network

TL;DR

FnRGNN tackles fairness in node-level regression with graph neural networks by introducing a distribution-aware in-processing framework. It interlocks three interventions—structure-level edge reweighting, representation-level MMD alignment, and prediction-level distribution matching via Sinkhorn divergence—into a unified training objective. Empirical results on four real-world datasets show reduced group disparities (MG, VG, WD) while preserving predictive accuracy, outperforming baselines across structure, representation, and prediction levels. The approach advances fairness in continuous graph predictions and offers practical potential for high-stakes applications, with code available for reproducibility.

Abstract

Graph Neural Networks (GNNs) excel at learning from structured data, yet fairness in regression tasks remains underexplored. Existing approaches mainly target classification and representation-level debiasing, which cannot fully address the continuous nature of node-level regression. We propose FnRGNN, a fairness-aware in-processing framework for GNN-based node regression that applies interventions at three levels: (i) structure-level edge reweighting, (ii) representation-level alignment via MMD, and (iii) prediction-level normalization through Sinkhorn-based distribution matching. This multi-level strategy ensures robust fairness under complex graph topologies. Experiments on four real-world datasets demonstrate that FnRGNN reduces group disparities without sacrificing performance. Code is available at https://github.com/sybeam27/FnRGNN.
Paper Structure (18 sections, 6 equations, 1 figure, 3 tables, 1 algorithm)

This paper contains 18 sections, 6 equations, 1 figure, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Preliminaries
  4. Graph Neural Networks
  5. Fairness in Node Regression
  6. Methodology
  7. Edge Reweighting for Fair Structure
  8. Group Alignment for Fair Representation
  9. Distribution Matching for Fair Prediction
  10. Training Procedure
  11. Experiments
  12. Experiment Settings
  13. Datasets
  14. Baselines and Implementation
  15. Evaluation Metrics
  16. ...and 3 more sections

Figures (1)

  • Figure 1: Overview of FnRGNN, which integrates fairness interventions at three levels: (i) structure, (ii) representation, and (iii) prediction. Each component addresses group-level disparities in node-level regression, as introduced in Sec. \ref{['sec:methodology']}.