The detection of anomalies in graph data is a critical task across various domains, such as fraud detection in social and commercial networks. Traditional Graph Neural Network (GNN) models often struggle with dataset biases, including label bias and keyword bias, which impair their generalization abilities. This paper introduces a novel Debiasing Frequency Adaptive GNN (DFA-GNN) that addresses these challenges by enhancing model accuracy and reducing dataset biases. Unlike previous approaches, DFA-GNN adapts to the complexity of node relationships and the frequency of interaction signals, making it particularly effective for node-based anomaly detection. By decomposing the input graph into several relation graphs and employing a frequency adjusting filter, DFA-GNN selectively processes signals of varying frequencies, catering to both homophily and heterophily conditions. Additionally, our counterfactual inference mechanism mitigates unintended dataset biases, further enhancing the model's prediction accuracy. Our extensive experiments on fraud detection datasets such as Yelp have shown that DFA-GNN has excellent performance in identifying subtle and complex anomalies, outperforming existing models in accuracy and debiasing ability.

Hongjia Pi. Debiasing Frequency Adaptive Graph Neural Network-based Fraud Detector. Academic Journal of Computing & Information Science (2024), Vol. 7, Issue 4: 17-23. https://doi.org/10.25236/AJCIS.2024.070403.

[1] K. Tang, Y. Niu, J. Huang, J. Shi, and H. Zhang. Unbiased Scene Graph Generation from Biased Training. In the Conference on Computer Vision and Pattern Recognition (CVPR), 2020.pages 3716–3725.

[2] M. Jiang, P. Cui, and C. Faloutsos. Suspicious behavior detection: Current trends and future directions. IEEE Intelligent Systems (2016).

[3] C. Hokamp and Q. Liu. Lexically Constrained Decoding for Sequence Generation Using Grid Beam Search. In the Annual Meeting of the Association for Computational Linguistics (ACL), 2017. pages 1535–1546.

[4] Z. Liu, Y. Dou, P. S. Yu, Y. Deng, and H. Peng. Alleviating the Inconsistency Problem of Applying Graph Neural Network to Fraud Detection. SIGIR. 2020. 

[5] J. Zhao, T. Wang, M. Yatskar, V. Ordonez, and K. Chang. Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraints. In the Conference on Empirical Methods in Natural Language Processing (EMNLP), 2017. pages 2979–2989.

[6] Dou, Y., Liu, Z., Sun, L., Deng, Y., Peng, H., and Yu, P. S. Enhancing graph neural network-based fraud detectors against camouflaged fraudsters. In CIKM, pp. 315–324, 2020.

[7] J. Tang, J. Li, Z. Gao, J. Li. Rethinking Graph Neural Networks for Anomaly Detection. Proceedings of the 39th International Conference on Machine Learning, PMLR 162:21076-21089, 2022.

[8] L. Wu, H. Lin, B. Hu, C. Tan, Z. Gao, Z. Liu, and S. Z. Li, “Beyond homophily and homogeneity assumption: Relation-based frequency adaptive graph neural networks,” IEEE Transactions on Neural Networks and Learning Systems, 2023, doi: 10.1109/TNNLS.2022.3230417.

[9] Qian, C., Feng, F., Wen, L., Ma, C., & Xie, P. Counterfactual inference for text classification debiasing. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 5434–5445, 2021.

[10] Y. Zhang, Y. Fan, Y. Ye, L. Zhao, and C. Shi. Key Player Identification in Underground Forums over Attributed Heterogeneous Information Network Embedding Framework. In CIKM. 2019.

[11] W. Hamilton, Z. Ying, and J. Leskovec. Inductive representation learning on large graphs. In NeurIPS. 2017.

[12] T. N. Kipf and M. Welling, “Variational graph auto-encoders,” in Proc. NIPS Workshop Bayesian Deep Learn., 2016, pp. 1–3.

[13] J. Ma, P. Cui, K. Kuang, X. Wang, and W. Zhu, “Disentangled graph convolutional networks,” in Proc. Int. Conf. Mach. Learn., vol. 97, 2019, pp. 4212–4221.

[14] F. R. Chung and F. C. Graham, Spectral Graph Theory. Providence, RI, USA: Amer. Math. Soc., 1997.

[15] G. Zhang, J. Wu, J. Yang, A. Beheshti, S. Xue, C. Zhou, and Q. Z. Sheng, “Fraudre: Fraud detection dual-resistant to graph inconsistency and imbalance,” in Proc. IEEE Int. Conf. Data Mining, 2021, pp. 66–77.

[16] T. N. Kipf and M. Welling, “Semi-supervised classification with graph convolutional networks,” in Proc. Int. Conf. Learn. Represent. (ICLR), 2017, pp. 1–14.

[17] P. Velickovic, G. Cucurull, A. Casanova, A. Romero, P. Liò, and Y. Bengio, “Graph attention networks,” in Proc. Int. Conf. Learn. Represent. (ICLR), 2018, pp. 1–12.