Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
97 tokens/sec
GPT-4o
53 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
5 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Alleviating Structural Distribution Shift in Graph Anomaly Detection (2401.14155v1)

Published 25 Jan 2024 in cs.LG and cs.AI

Abstract: Graph anomaly detection (GAD) is a challenging binary classification problem due to its different structural distribution between anomalies and normal nodes -- abnormal nodes are a minority, therefore holding high heterophily and low homophily compared to normal nodes. Furthermore, due to various time factors and the annotation preferences of human experts, the heterophily and homophily can change across training and testing data, which is called structural distribution shift (SDS) in this paper. The mainstream methods are built on graph neural networks (GNNs), benefiting the classification of normals from aggregating homophilous neighbors, yet ignoring the SDS issue for anomalies and suffering from poor generalization. This work solves the problem from a feature view. We observe that the degree of SDS varies between anomalies and normal nodes. Hence to address the issue, the key lies in resisting high heterophily for anomalies meanwhile benefiting the learning of normals from homophily. We tease out the anomaly features on which we constrain to mitigate the effect of heterophilous neighbors and make them invariant. We term our proposed framework as Graph Decomposition Network (GDN). Extensive experiments are conducted on two benchmark datasets, and the proposed framework achieves a remarkable performance boost in GAD, especially in an SDS environment where anomalies have largely different structural distribution across training and testing environments. Codes are open-sourced in https://github.com/blacksingular/wsdm_GDN.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (51)
  1. Graph based anomaly detection and description: a survey. Data Min. Knowl. Discov. 29, 3 (2015), 626–688.
  2. Outlier Resistant Unsupervised Deep Architectures for Attributed Network Embedding. In WSDM. 25–33.
  3. Can Abnormality be Detected by Graph Neural Networks?. In IJCAI.
  4. F-fade: Frequency factorization for anomaly detection in edge streams. In WSDM. 589–597.
  5. GCCAD: Graph Contrastive Coding for Anomaly Detection. CoRR abs/2108.07516 (2021).
  6. Counterfactual samples synthesizing for robust visual question answering. In CVPR. 10800–10809.
  7. When Does A Spectral Graph Neural Network Fail in Node Classification? CoRR abs/2202.07902 (2022).
  8. Causal understanding of fake news dissemination on social media. In KDD. 148–157.
  9. Gregory C Chow. 1960. Tests of equality between sets of coefficients in two linear regressions. Econometrica: Journal of the Econometric Society (1960), 591–605.
  10. Deep anomaly detection on attributed networks. In SDM. 594–602.
  11. Interactive anomaly detection on attributed networks. In WSDM. 357–365.
  12. Enhancing graph neural network-based fraud detectors against camouflaged fraudsters. In CIKM. 315–324.
  13. Debiased Graph Neural Networks with Agnostic Label Selection Bias. TNNLS (2022).
  14. Rumor Detection with Self-supervised Learning on Texts and Social Graph. arXiv preprint arXiv:2204.08838 (2022).
  15. Inductive Representation Learning on Large Graphs. In NIPS.
  16. Instance-dependent pu learning by bayesian optimal relabeling. arXiv preprint arXiv:1808.02180 (2018).
  17. Invariant causal prediction for nonlinear models. Journal of Causal Inference 6, 2 (2018).
  18. beta-vae: Learning basic visual concepts with a constrained variational framework. In ICLR.
  19. GraphLIME: Local Interpretable Model Explanations for Graph Neural Networks. TKDE (2022).
  20. Hyunjik Kim and Andriy Mnih. 2018. Disentangling by factorising. In ICML. 2649–2658.
  21. Thomas N. Kipf and Max Welling. 2017. Semi-Supervised Classification with Graph Convolutional Networks. In ICLR.
  22. Deep domain generalization via conditional invariant adversarial networks. In ECCV. 624–639.
  23. Semi-supervised embedding in attributed networks with outliers. In SDM. 153–161.
  24. Intention-aware heterogeneous graph attention networks for fraud transactions detection. In KDD. 3280–3288.
  25. Pick and choose: a GNN-based imbalanced learning approach for fraud detection. In WWW. 3168–3177.
  26. Alleviating the inconsistency problem of applying graph neural network to fraud detection. In SIGIR. 1569–1572.
  27. Subgroup generalization and fairness of graph neural networks. In NIPS.
  28. A comprehensive survey on graph anomaly detection with deep learning. TKDE (2021).
  29. Julian John McAuley and Jure Leskovec. 2013. From amateurs to connoisseurs: modeling the evolution of user expertise through online reviews. In WWW. 897–908.
  30. Causal inference by using invariant prediction: identification and confidence intervals. Journal of the Royal Statistical Society. Series B (Statistical Methodology) (2016), 947–1012.
  31. Invariant causal prediction for sequential data. J. Amer. Statist. Assoc. 114, 527 (2019), 1264–1276.
  32. Explainability methods for graph convolutional neural networks. In CVPR. 10772–10781.
  33. Shebuti Rayana and Leman Akoglu. 2015. Collective opinion spam detection: Bridging review networks and metadata. In KDD. 985–994.
  34. Everett M Rogers and Dilip K Bhowmik. 1970. Homophily-heterophily: Relational concepts for communication research. Public opinion quarterly 34, 4 (1970).
  35. Modeling relational data with graph convolutional networks. In ESWC. 593–607.
  36. Grad-cam: Visual explanations from deep networks via gradient-based localization. In ICCV. 618–626.
  37. Disentangled Generative Causal Representation Learning. CoRR abs/2010.02637 (2020).
  38. Stable learning via differentiated variable decorrelation. In KDD. 2185–2193.
  39. Towards Out-Of-Distribution Generalization: A Survey. CoRR abs/2108.13624 (2021).
  40. Rethinking Graph Neural Networks for Anomaly Detection. In ICML. 21076–21089.
  41. Graph Attention Networks. In ICLR.
  42. A semi-supervised graph attentive network for financial fraud detection. In ICDM. 598–607.
  43. Hongwei Wang and Jure Leskovec. 2020. Unifying graph convolutional neural networks and label propagation. arXiv preprint arXiv:2002.06755 (2020).
  44. Fdgars: Fraudster detection via graph convolutional networks in online app review system. In WWW (Companion Volume). 310–316.
  45. Decoupling representation learning and classification for gnn-based anomaly detection. In SIGIR. 1239–1248.
  46. Towards Distribution Shift of Node-Level Prediction on Graphs: An Invariance Perspective. In ICLR.
  47. CausalVAE: disentangled representation learning via neural structural causal models. In CVPR. 9593–9602.
  48. Bianca Zadrozny. 2004. Learning and evaluating classifiers under sample selection bias. In ICML. 114.
  49. FRAUDRE: Fraud Detection Dual-Resistant to Graph Inconsistency and Imbalance. In ICDM. 867–876.
  50. GCN-Based User Representation Learning for Unifying Robust Recommendation and Fraudster Detection. In SIGIR. ACM, 689–698.
  51. Shift-robust gnns: Overcoming the limitations of localized graph training data. In NIPS.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (6)
  1. Yuan Gao (336 papers)
  2. Xiang Wang (279 papers)
  3. Xiangnan He (200 papers)
  4. Zhenguang Liu (55 papers)
  5. Huamin Feng (6 papers)
  6. Yongdong Zhang (119 papers)
Citations (48)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now
X Twitter Logo Streamline Icon: https://streamlinehq.com

Tweets