FairGNN-Bagging: Balancing model performance and algorithmic fairness in graph neural networks using ensemble learning

• Juan Carlos Elias Obando Valdivia USP
• Marcel Rodrigues de Barros USP
• Artur Jordão Lima Correia USP
• Anna Helena Reali Costa USP

DOI: https://doi.org/10.5753/eniac.2025.14477

Abstract

Graph Neural Networks (GNNs) are powerful tools for learning from graph-structured data, with applications in social networks, bioinformatics, and fraud detection. However, GNNs can inherit and amplify biases present in data, leading to unfair predictions regarding sensitive attributes such as gender or race. To mitigate these issues, fairness-aware frameworks like NIFTY and its variant with Biased Edge Dropout have been proposed. In this work, we introduce FairGNN-Bagging, an ensemble framework that combines NIFTY with Bagging to improve fairness in node classification tasks. Our method uses graph perturbations to generate an augmented graph dataset to train an ensemble of GNNs in parallel. These models are then aggregated via majority voting or fairness-aware selective voting. We evaluate our approach on three real-world datasets, showing that it achieves better fairness metrics while preserving or improving predictive performance.

References

Agarwal, C., Lakkaraju, H., and Zitnik, M. (2021). Towards a unified framework for fair and stable graph representation learning. In Uncertainty in Artificial Intelligence, pages 2114–2124. PMLR.

Breiman, L. (1996). Bagging predictors. Machine learning, 24:123–140.

Bromley, J., Guyon, I., LeCun, Y., Säckinger, E., and Shah, R. (1993). Signature verification using a ”siamese” time delay neural network. In Proceedings of the 6th International Conference on Neural Information Processing Systems, page 737–744.

Chakravarty, A., Sarkar, T., Ghosh, N., Sethuraman, R., and Sheet, D. (2020). Learning decision ensemble using a graph neural network for comorbidity aware chest radiograph screening. In 2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), pages 1234–1237.

Cynthia, Hardt, M., Pitassi, T., Reingold, O., and Zemel, R. (2012). Fairness through awareness. In Proceedings of the 3rd Innovations in Theoretical Computer Science Conference, ITCS ’12, page 214–226, New York, NY, USA. ACM.

Dai, E., Zhao, T., Zhu, H., Xu, J., Guo, Z., Liu, H., Tang, J., and Wang, S. (2024). A comprehensive survey on trustworthy graph neural networks: Privacy, robustness, fairness, and explainability. Machine Intelligence Research, pages 1–51.

Dong, Y., Liu, N., Jalaian, B., and Li, J. (2022). Edits: Modeling and mitigating data bias for graph neural networks. In Proceedings of the ACM web conference 2022, pages 1259–1269.

Dong, Y., Ma, J., Wang, S., Chen, C., and Li, J. (2023). Fairness in graph mining: A survey. IEEE TKDE, 35(10):10583–10602.

Dua, D., Graff, C., et al. (2017). UCI machine learning repository, 2017. 7(1):62.

Franco, D., D’Amato, V. S., Pasa, L., Navarin, N., and Oneto, L. (2024). Fair graph representation learning: Empowering nifty via biased edge dropout and fair attribute preprocessing. Neurocomput., 563(C).

Hamilton, W., Ying, Z., and Leskovec, J. (2017a). Inductive representation learning on large graphs. Advances in neural information processing systems, 30.

Hamilton, W. L., Ying, R., and Leskovec, J. (2017b). Representation learning on graphs: Methods and applications. IEEE Data Eng. Bull., 40(3):52–74.

Jordan, K. L. and Freiburger, T. L. (2015). The effect of race/ethnicity on sentencing: Examining sentence type, jail length, and prison length. Journal of Ethnicity in Criminal Justice, 13(3):179–196.

Kang, J., He, J., Maciejewski, R., and Tong, H. (2020). Inform: Individual fairness on graph mining. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, page 379–389, NY, USA. ACM.

Kipf, T. N. and Welling, M. (2017). Semi-supervised classification with graph convolutional networks. In 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, Conference Track Proceedings. OpenReview.net.

Kosasih, E. E., Cabezas, J., Sumba, X., Bielak, P., Tagowski, K., Idanwekhai, K., Tjandra, B. A., and Jamasb, A. R. (2021). On graph neural network ensembles for large-scale molecular property prediction. arXiv preprint arXiv:2106.15529.

Kusner, M. J., Loftus, J., Russell, C., and Silva, R. (2017). Counterfactual fairness. Advances in neural information processing systems, 30.

Lin, Q., Yu, S., Sun, K., Zhao, W., Alfarraj, O., Tolba, A., and Xia, F. (2022). Robust graph neural networks via ensemble learning. Mathematics, 10(8).

Nagarajan, A., Stevens, J. R., and Raghunathan, A. (2022). Efficient ensembles of graph neural networks. In Proceedings of the 59th ACM/IEEE Design Automation Conference, pages 187–192.

Pearl, J. (2009). Causal inference in statistics: An overview.

Rong, Y., Huang, W., Xu, T., and Huang, J. (2019). Dropedge: Towards deep graph convolutional networks on node classification. In International Conference on Learning Representations.

Spinelli, I., Scardapane, S., Hussain, A., and Uncini, A. (2021). Fairdrop: Biased edge dropout for enhancing fairness in graph representation learning. IEEE Transactions on Artificial Intelligence, 3(3):344–354.

Wang, H. and Leskovec, J. (2020). Unifying graph convolutional neural networks and label propagation. arXiv preprint arXiv:2002.06755.

Wei, W., Qiao, M., and Jadav, D. (2023). GNN-Ensemble: Towards Random Decision Graph Neural Networks . In 2023 IEEE International Conference on Big Data (Big-Data), pages 956–965, Los Alamitos, CA, USA. IEEE Computer Society.

Wu, Z., Pan, S., Chen, F., Long, G., Zhang, C., and Philip, S. Y. (2020). A comprehensive survey on graph neural networks. IEEE transactions on neural networks and learning systems, 32(1):4–24.

Yeh, I.-C. and hui Lien, C. (2009). The comparisons of data mining techniques for the predictive accuracy of probability of default of credit card clients. Expert Syst. Appl., 36:2473–2480.

Zhang, H., Wu, B., Yuan, X., Pan, S., Tong, H., and Pei, J. (2024). Trustworthy graph neural networks: Aspects, methods, and trends. Proceedings of the IEEE, 112(2):97–139.