Neighborhood-enhanced unsupervised graph anomaly detection

doi:10.11772/j.issn.1001-9081.2025020216

Journal of Computer Applications ›› 2026, Vol. 46 ›› Issue (2): 458-466.DOI: 10.11772/j.issn.1001-9081.2025020216

• Cyber security • Previous Articles

Neighborhood-enhanced unsupervised graph anomaly detection

Limei DONG¹^,², Yanzi LI¹^,², Jiayin LI¹^,², Li XU¹^,²()

^1.College of Computer and Cyber Security，Fujian Normal University，Fuzhou Fujian 350117，China
^2.Fujian Provincial Key Laboratory of Network Security and Cryptology，Fuzhou Fujian 350117，China

Received:2025-03-06 Revised:2025-04-27 Accepted:2025-05-07 Online:2025-05-16 Published:2026-02-10
Contact: Li XU
About author:DONG Limei， born in 2001， M. S. candidate. Her research interests include anomaly detection， network and information security.
LI Yanzi， born in 1998， Ph. D. candidate. Her research interests include privacy protection， network and information security.
LI Jiayin， born in 1990， Ph.D.， lecturer. His research interests include artificial intelligence security， internet of vehicles security， privacy protection， machine learning.
XU Li， born in 1970， Ph. D.， professor. His research interests include network and information security， big data and informatization， mobile social networks， intelligent information processing. Email:xuli@fjnu.edu.cn
Supported by:
National Natural Science Foundation of China(62471139);Central Government Guided Local Science and Technology Development Special Program(2023L3007);Fujian Provincial Natural Science Foundation(2023J05128);Fujian Provincial Science and Technology Project(2022G02003)

基于邻域增强的无监督图异常检测

董莉梅¹^,², 李雁姿¹^,², 李家印¹^,², 许力¹^,²()

^1.福建师范大学计算机与网络空间安全学院，福州 350117
^2.福建省网络安全与密码技术重点实验室，福州 350117

通讯作者: 许力
作者简介:董莉梅（2001—），女，四川达州人，硕士研究生，主要研究方向：异常检测、网络与信息安全
李雁姿（1998—），女，山东聊城人，博士研究生，主要研究方向：隐私保护、网络与信息安全
李家印（1990—），男，山东梁山人，讲师，博士，主要研究方向：人工智能安全、车联网安全、隐私保护、机器学习
许力（1970—），男，福建福州人，教授，博士，CCF会员，主要研究方向：网络与信息安全、大数据与信息化、移动社会网络、智能信息处理。 Email:xuli@fjnu.edu.cn
基金资助:
国家自然科学基金资助项目(62471139);中央引导地方科技发展专项(2023L3007);福建省自然科学基金资助项目(2023J05128);福建省科技项目(2022G02003)

Abstract

Abstract:

To address the issue of unreliable neighborhood information due to the presence of anomalies， an efficient unsupervised graph anomaly detection method was proposed. In this method， a neighborhood enhancement strategy was utilized to construct multi-type neighborhood sets for central nodes， thereby capturing high-quality node representations and obtaining accurate self-neighborhood similarity. Firstly， an information extraction module based on dynamic neighborhood enhancement was optimized to select the optimal neighborhood strategy adaptively， thereby overcoming the feature homogeneity limitation of traditional fixed neighborhood selection methods during information extraction. Secondly， in order to reduce the interference of redundant information of the nodes during node feature fusion， an anonymous message passing scheme was proposed. In this scheme， features of the nodes were isolated， and neighborhood information was focused on solely， thereby enhancing the quality of message aggregation. Finally， an adaptive weighted anomaly scoring module was designed， using the distance between nodes as an evaluation scale to obtain node anomaly scores， thereby refining the anomaly detection results. Experimental results on five datasets demonstrate that the proposed method outperforms mainstream method CoLA （Anomaly detection on attributed networks via Contrastive self-supervised Learning） in detecting anomalies in complex graph structure， achieving an at least 8.0% AUPRC （Area Under the Precision-Recall Curve） improvement in identifying anomalous samples.

Key words: anomaly detection, unsupervised learning, contrastive learning, neighborhood selection, graph data

摘要：

针对异常的存在导致节点邻域信息不可靠的问题，提出一种高效的无监督图异常检测方法。该方法借助邻域增强策略构建多类型的中心节点的邻域集合，捕捉高质量的节点表示，并获取高准确度的自邻相似度。首先，通过优化一个基于动态邻域增强的信息提取模块，自适应地选择最优邻域策略，从而克服传统固定邻域选择方法在信息提取过程中特征单一的局限性；其次，为了降低节点特征融合时自身冗余信息的干扰，提出一种匿名消息传递方案，该方案能够隔离节点自身特征，只专注于邻域信息，从而提高消息聚合的质量；最后，通过设计一种自适应的加权异常评分模块，以节点之间距离作为评估尺度来获取节点的异常度，从而细化异常检测结果。在5个数据集上的实验结果表明，所提方法在应对复杂图结构的异常检测方面的表现优于现有主流方法CoLA（Anomaly detection on attributed networks via Contrastive self-supervised Learning），其中对异常样本的识别能力指标——AUPRC（Area Under the Precision-Recall Curve）至少提升了8.0%。

关键词: 异常检测, 无监督学习, 对比学习, 邻域选择, 图数据

CLC Number:

TP393

Limei DONG, Yanzi LI, Jiayin LI, Li XU. Neighborhood-enhanced unsupervised graph anomaly detection[J]. Journal of Computer Applications, 2026, 46(2): 458-466.

董莉梅, 李雁姿, 李家印, 许力. 基于邻域增强的无监督图异常检测[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 458-466.

Figures/Tables 12

References 34

[1]	樊琳娜，李城龙，吴毅超，等. 物联网设备识别及异常检测研究综述［J］. 软件学报， 2024， 35（1）： 288-308.
	FAN L N， LI C L， WU Y C， et al. Survey on IoT device identification and anomaly detection［J］. Journal of Software， 2024， 35（1）： 288-308.
[2]	孙澈，武玉伟，贾云得. 上下文建模与推理的视频异常事件检测［J］. 计算机学报， 2024， 47（10）： 2368-2386.
	SUN C， WU Y W， JIA Y D. Context modeling and reasoning for video anomaly event detection［J］. Chinese Journal of Computers， 2024， 47（10）： 2368-2386.
[3]	LI A， QIU C， KLOFT M， et al. Deep anomaly detection under labeling budget constraints［C］// Proceedings of the 40th International Conference on Machine Learning. New York： JMLR.org， 2023： 19882-19910.
[4]	LIU Y， DING K， LIU H， et al. GOOD-D： on unsupervised graph out-of-distribution detection［C］// Proceedings of the 16th ACM International Conference on Web Search and Data Mining. New York： ACM， 2023： 339-347.
[5]	方介泼，陶重犇. 应对零日攻击的混合车联网入侵检测系统［J］. 计算机应用， 2024， 44（9）： 2763-2769.
	FANG J P， TAO C B. Hybrid internet of vehicles intrusion detection system for zero-day attacks［J］. Journal of Computer Applications， 2024， 44（9）： 2763-2769.
[6]	AHMED I， HU X B， ACHARYA M P， et al. Neighborhood structure assisted non-negative matrix factorization and its application in unsupervised point-wise anomaly detection［J］. Journal of Machine Learning Research， 2021， 22： 1-32.
[7]	WEI Z， XIE X， ZHANG X. Maritime anomaly detection based on a support vector machine［J］. Soft Computing， 2022， 26（21）： 11553-11566.
[8]	CHOI H， KIM M， LEE G， et al. Unsupervised learning approach for network intrusion detection system using autoencoders［J］. The Journal of Supercomputing， 2019， 75（9）： 5597-5621.
[9]	KIPF T N， WELLING M. Variational graph auto-encoders［EB/OL］. ［2025-04-24］..
[10]	YUAN X， HUANG B， WANG Y， et al. Deep learning-based feature representation and its application for soft sensor modeling with variable-wise weighted SAE［J］. IEEE Transactions on Industrial Informatics， 2018， 14（7）： 3235-3243.
[11]	SCHLEGL T， SEEBÖCK P， WALDSTEIN S M， et al. Unsupervised anomaly detection with generative adversarial networks to guide marker discovery［C］// Proceedings of the 2017 International Conference on Information Processing in Medical Imaging， LNCS 10265. Cham： Springer， 2017： 146-157.
[12]	DEECKE L， VANDERMEULEN R， RUFF L， et al. Image anomaly detection with generative adversarial networks［C］// Proceedings of the 2018 Joint European Conference on Machine Learning and Knowledge Discovery in Databases， LNCS 11051. Cham： Springer， 2019： 3-17.
[13]	SCHLEGL T， SEEBÖCK P， WALDSTEIN S M， et al. f-AnoGAN： Fast unsupervised anomaly detection with generative adversarial networks［J］. Medical Image Analysis， 2019， 54： 30-44.
[14]	DING K， LI J， BHANUSHALI R， et al. Deep anomaly detection on attributed networks［C］// Proceedings of the 2019 SIAM International Conference on Data Mining. Philadelphia， PA： SIAM， 2019： 594-602.
[15]	FAN H， ZHANG F， LI Z. AnomalyDAE： dual autoencoder for anomaly detection on attributed networks［C］// Proceedings of the 2020 IEEE International Conference on Acoustics， Speech and Signal Processing. Piscataway： IEEE， 2020： 5685-5689.
[16]	LIU F， MA X， WU J， et al. DAGAD： data augmentation for graph anomaly detection［C］// Proceedings of the 2022 IEEE International Conference on Data Mining. Piscataway： IEEE， 2022： 259-268.
[17]	LIU Y， LI Z， PAN S， et al. Anomaly detection on attributed networks via contrastive self-supervised learning［J］. IEEE Transactions on Neural Networks and Learning Systems， 2022， 33（6）： 2378-2392.
[18]	JIN M， LIU Y， ZHENG Y， et al. ANEMONE： graph anomaly detection with multi-scale contrastive learning［C］// Proceedings of the 30th ACM International Conference on Information and Knowledge Management. New York： ACM， 2021： 3122-3126.
[19]	ZHANG J， WANG S， CHEN S. Reconstruction enhanced multi-view contrastive learning for anomaly detection on attributed networks［C］// Proceedings of the 31st International Joint Conference on Artificial Intelligence. California： ijcai.org， 2022： 2376-2382.
[20]	ZHENG Y， JIN M， LIU Y， et al. Generative and contrastive self-supervised learning for graph anomaly detection［J］. IEEE Transactions on Knowledge and Data Engineering， 2023， 35（12）： 12220-12233.
[21]	GU X， AKOGLU L， RINALDO A. Statistical analysis of nearest neighbor methods for anomaly detection［C］// Proceedings of the 33rd International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2019： 10923-10933.
[22]	LI J， DANI H， HU X， et al. Radar： residual analysis for anomaly detection in attributed networks［C］// Proceedings of the 26th International Joint Conference on Artificial Intelligence. California： ijcai.org， 2017： 2152-2158.
[23]	PENG Z， LUO M， LI J， et al. ANOMALOUS： a joint modeling approach for anomaly detection on attributed networks［C］// Proceedings of the 27th International Joint Conference on Artificial Intelligence. California： ijcai.org， 2018： 3513-3519.
[24]	DING K， LI J， AGARWAL N， et al. Inductive anomaly detection on attributed networks［C］// Proceedings of the 29th International Joint Conferences on Artificial Intelligence. California： ijcai.org， 2020： 1288-1294.
[25]	BO D， WANG X， SHI C， et al. Beyond low-frequency information in graph convolutional networks［C］// Proceedings of the 35th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2021： 3950-3957.
[26]	MA M， NA S， WANG H. AEGCN： an autoencoder-constrained graph convolutional network［J］. Neurocomputing， 2021， 432： 21-31.
[27]	HE J， XU Q， JIANG Y， et al. ADA-GAD： anomaly-denoised autoencoders for graph anomaly detection［C］// Proceedings of the 38th AAAI International Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2024： 8481-8489.
[28]	CHANG W， LIU K， DING K， et al. Multitask active learning for graph anomaly detection［EB/OL］. ［2025-04-24］..
[29]	DUAN J， WANG S， ZHANG P， et al. Graph anomaly detection via multi-scale contrastive learning networks with augmented view［C］// Proceedings of the 37th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2023： 7459-7467.
[30]	DUAN J， ZHANG P， WANG S， et al. Normality learning-based graph anomaly detection via multi-scale contrastive learning［C］// Proceedings of the 31st ACM International Conference on Multimedia. New York： ACM， 2023： 7502-7511.
[31]	PAN J， LIU Y， ZHENG Y， et al. PREM： a simple yet effective approach for node-level graph anomaly detection［C］// Proceedings of the 2023 IEEE International Conference on Data Mining. Piscataway： IEEE， 2023： 1253-1258.
[32]	BEI Y， ZHOU S， TAN Q， et al. Reinforcement neighborhood selection for unsupervised graph anomaly detection［C］// Proceedings of the 2023 IEEE International Conference on Data Mining. Piscataway： IEEE， 2023： 11-20.
[33]	DUAN J， XIAO B， WANG S， et al. ARISE： graph anomaly detection on attributed networks via substructure awareness［J］. IEEE Transactions on Neural Networks and Learning Systems， 2024， 35（12）： 18172-18185.
[34]	KONG X， LIU J， LI H， et al. Graph anomaly detection via diffusion enhanced multi-view contrastive learning［J］. Knowledge-Based Systems， 2025， 311： No.113093.

数据集	节点数	边数	特征数	异常数
Cora	2 078	5 429	1 433	150
CiteSeer	3 327	4 732	3 703	150
PubMed	19 717	88 648	500	600
ACM	16 484	71 980	8 337	650
Flickr	7 575	239 738	12 407	450

数据集	节点数	边数	特征数	异常数
Cora	2 078	5 429	1 433	150
CiteSeer	3 327	4 732	3 703	150
PubMed	19 717	88 648	500	600
ACM	16 484	71 980	8 337	650
Flickr	7 575	239 738	12 407	450

方法	Cora	CiteSeer	PubMed	ACM	Flickr
DOMINANT^［14］	0.812 8	0.819 7	0.789 6	0.756 1	0.743 6
CoLA^［17］	0.904 6	0.889 4	0.946 8	0.837 5	0.761 6
ANEMONE^［18］	0.920 7	0.928 2	0.953 4	0.849 2	0.767 9
GRADATE^［29］	0.903 2	0.890 4	0.945 0	0.858 0	0.752 2
SL-GAD^［20］	0.917 8	0.922 1	0.962 7	0.814 3	0.796 5
Sub-CR^［19］	0.902 3	0.972 7	0.968 7	0.806 0	0.792 1
PREM^［31］	0.951 0	0.977 9	0.971 9	0.905 6	0.861 3
RAND^［32］	0.923 8	0.954 7	0.952 3	0.844 2	0.762 5
ARISE^［33］	0.943 8	0.931 0	0.971 2	0.874 2	0.703 8
DE-GAD^［34］	0.926 7	0.938 0	0.974 2	0.875 4	0.815 2
本文方法	0.966 1	0.988 5	0.976 4	0.917 8	0.882 0

方法	Cora	CiteSeer	PubMed	ACM	Flickr
DOMINANT^［14］	0.812 8	0.819 7	0.789 6	0.756 1	0.743 6
CoLA^［17］	0.904 6	0.889 4	0.946 8	0.837 5	0.761 6
ANEMONE^［18］	0.920 7	0.928 2	0.953 4	0.849 2	0.767 9
GRADATE^［29］	0.903 2	0.890 4	0.945 0	0.858 0	0.752 2
SL-GAD^［20］	0.917 8	0.922 1	0.962 7	0.814 3	0.796 5
Sub-CR^［19］	0.902 3	0.972 7	0.968 7	0.806 0	0.792 1
PREM^［31］	0.951 0	0.977 9	0.971 9	0.905 6	0.861 3
RAND^［32］	0.923 8	0.954 7	0.952 3	0.844 2	0.762 5
ARISE^［33］	0.943 8	0.931 0	0.971 2	0.874 2	0.703 8
DE-GAD^［34］	0.926 7	0.938 0	0.974 2	0.875 4	0.815 2
本文方法	0.966 1	0.988 5	0.976 4	0.917 8	0.882 0

方法	Cora	CiteSeer	PubMed	ACM	Flickr
DOMINANT^［14］	0.447 4	0.330 8	OOM	0.355 3
CoLA^［17］	0.506 5	0.430 3	0.416 9	0.321 4
ANEMONE^［18］	0.532 0	0.628 1	0.427 4	0.353 9
SL-GAD^［20］	0.602 2	0.526 3	0.635 1	0.388 6
Sub-CR^［19］	0.592 2	0.608 3	0.585 4	0.346 9
RAND^［32］	0.497 9	0.544 0	0.445 4	0.392 0	0.367 0
本文方法	0.669 3	0.773 6	0.686 1	0.561 5	0.473 6

Neighborhood-enhanced unsupervised graph anomaly detection

基于邻域增强的无监督图异常检测

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 12

References 34

Related Articles 15

Recommended Articles

Metrics

方法	Cora	CiteSeer	PubMed	ACM	Flickr
N-en	0.735 4	0.779 5	0.874 8	0.722 9	0.697 3
本文方法	0.966 1	0.988 5	0.976 4	0.917 8	0.882 0

方法	Cora	CiteSeer	PubMed	ACM	Flickr
N-ano	0.895 6	0.949 4	0.937 4	0.882 0	0.802 9
本文方法	0.966 1	0.988 5	0.976 4	0.917 8	0.882 0

方法	Cora	CiteSeer	PubMed	ACM	Flickr
N-wed	0.888 1	0.941 8	0.905 4	0.864 0	0.840 5
本文方法	0.966 1	0.988 5	0.976 4	0.917 8	0.882 0

[1]	Hu LUO, Mingshu ZHANG. Rumor detection method based on cross-modal attention mechanism and contrastive learning [J]. Journal of Computer Applications, 2026, 46(2): 361-367.
[2]	Qi ZHONG, Shufen ZHANG, Zhenbo ZHANG, Yinlong JIAN, Zhongrui JING. Detection and defense mechanism for poisoning attacks to federated learning [J]. Journal of Computer Applications, 2026, 46(2): 445-457.
[3]	Wen LI, Kairong LI, Kai YANG. Subgraph-aware contrastive learning with data augmentation [J]. Journal of Computer Applications, 2026, 46(1): 1-9.
[4]	Xingyao YANG, Zheng QI, Jiong YU, Zulian ZHANG, Shuai MA, Hongtao SHEN. Session-based recommendation model based on time-aware and space-enhanced dual channel graph neural network [J]. Journal of Computer Applications, 2026, 46(1): 104-112.
[5]	Ziyang CHENG, Ruizhang HUANG, Jingjing XUE. Deep evolutionary topic clustering model [J]. Journal of Computer Applications, 2026, 46(1): 85-94.
[6]	Chao LIU, Yanhua YU. Knowledge-aware recommendation model combining denoising strategy and multi-view contrastive learning [J]. Journal of Computer Applications, 2025, 45(9): 2827-2837.
[7]	Zhixiong XU, Bo LI, Xiaoyong BIAN, Qiren HU. Adversarial sample embedded attention U-Net for 3D medical image segmentation [J]. Journal of Computer Applications, 2025, 45(9): 3011-3016.
[8]	Zhiyuan WANG, Tao PENG, Jie YANG. Integrating internal and external data for out-of-distribution detection training and testing [J]. Journal of Computer Applications, 2025, 45(8): 2497-2506.
[9]	Jin XIE, Surong CHU, Yan QIANG, Juanjuan ZHAO, Hua ZHANG, Yong GAO. Dual-branch distribution consistency contrastive learning model for hard negative sample identification in chest X-rays [J]. Journal of Computer Applications, 2025, 45(7): 2369-2377.
[10]	Zhenzhou WANG, Fangfang GUO, Jingfang SU, He SU, Jianchao WANG. Robustness optimization method of visual model for intelligent inspection [J]. Journal of Computer Applications, 2025, 45(7): 2361-2368.
[11]	Yulin HE, Xu LI, Yingting HE, Laizhong CUI, Zhexue HUANG. Subspace Gaussian mixture model clustering ensemble algorithm based on maximum mean discrepancy [J]. Journal of Computer Applications, 2025, 45(6): 1712-1723.
[12]	Mingfeng YU, Yongbin QIN, Ruizhang HUANG, Yanping CHEN, Chuan LIN. Multi-label text classification method based on contrastive learning enhanced dual-attention mechanism [J]. Journal of Computer Applications, 2025, 45(6): 1732-1740.
[13]	Wenjing YAN, Ruidong WANG, Min ZUO, Qingchuan ZHANG. Recipe recommendation model based on hierarchical learning of flavor embedding heterogeneous graph [J]. Journal of Computer Applications, 2025, 45(6): 1869-1878.
[14]	Ying HUANG, Shengmei GAO, Guang CHEN, Su LIU. Low-light image enhancement network combining signal-to-noise ratio guided dual-branch structure and histogram equalization [J]. Journal of Computer Applications, 2025, 45(6): 1971-1979.
[15]	Chaoying JIANG, Qian LI, Ning LIU, Lei LIU, Lizhen CUI. Readmission prediction model based on graph contrastive learning [J]. Journal of Computer Applications, 2025, 45(6): 1784-1792.