基于ADASYN、Lasso回归和集成学习的比特币异常交易检测方法

doi:10.3969/j.issn.1671-1122.2026.03.011

摘要/Abstract

摘要：

针对比特币异常交易检测中存在的类别不均衡、特征冗余和单一模型性能较差等问题，文章提出一种基于ADASYN、Lasso回归和集成学习的比特币异常交易检测方法。该方法采用ADASYN算法对不均衡数据集进行过采样，采用Lasso回归进行特征选择，并采用融合多个基分类器的堆叠集成策略实现异常交易识别。在Elliptic++数据集上的实验结果表明，该方法的F1值达到0.7915，与基线RF模型相比提升了14.9%。消融实验结果显示，ADASYN与Lasso回归的协同机制使RF模型性能提升了14.6%。在仅使用20%训练样本的小样本实验中，该方法的F1值为0.6433，性能仅下降18.7%。

关键词: 比特币, 异常交易检测, ADASYN, Lasso回归, 集成学习

Abstract:

Class imbalance, feature redundancy, and insufficient single-model performance pose major challenges in bitcoin transaction anomaly detection. This paper proposed a anomaly detection method for bitcoin transactions based on adaptive synthetic sampling (ADASYN), Lasso regression, and ensemble learning. The method employed the ADASYN algorithm to oversample an imbalanced dataset, used Lasso regression for feature selection, and adopted a stacking ensemble strategy integrating multiple base classifiers for anomalous transaction identification. Experimental results on the Elliptic++ dataset show that the proposed method achieves the F1-score of 0.7915, a 14.9% improvement over the baseline random forest (RF) model. Ablation experiments show that the synergistic effect of ADASYN and Lasso regression contributes to a 14.6% performance improvement for the RF model. In a small-sample experiment with only 20% training samples, the method achieves the F1-score of 0.6433, with a performance degradation of merely 18.7%.

Key words: bitcoin, anomalous transaction detection, ADASYN, Lasso regression, ensemble learning

中图分类号:

TP309

陈潮, 王诺萱, 周胜利. 基于ADASYN、Lasso回归和集成学习的比特币异常交易检测方法[J]. 信息网络安全, 2026, 26(3): 452-461.

CHEN Chao, WANG Nuoxuan, ZHOU Shengli. Anomaly Detection Method for Bitcoin Transactions Based on ADASYN, Lasso Regression, and Ensemble Learning[J]. Netinfo Security, 2026, 26(3): 452-461.

图/表 11

图1

表1

表2

表3

表4

表5

表6

表7

表8

图2

图3

参考文献 32

[1]	TRZCIONKA M. The Bitcoin-Democratic Money in a Neoliberal Economy[J]. Ad Americam, 2019, 19: 155-173. doi: 10.12797/AdAmericam.19.2018.19.11 URL
[2]	GAJDEK S. Bitcoin as an Example of Cryptocurrency-Current State and Perspectives[J]. Annals of Marketing Management & Economics, 2018, 4(2): 47-56.
[3]	SABRY F, LABDA W, ERBAD A, et al. Cryptocurrencies and Artificial Intelligence: Challenges and Opportunities[J]. IEEE Access, 2020, 8: 175840-175858. doi: 10.1109/Access.6287639 URL
[4]	PHAM T, LEE S. Anomaly Detection in Bitcoin Network Using Unsupervised Learning Methods[EB/OL]. (2017-02-25)[2025-04-28]. http://arxiv.org/abs/1611.03941.
[5]	SONG K, ALI DHRAIEF M, XU Muhua, et al. Identifying Money Laundering Subgraphs on the Blockchain[C]// ACM. The 5th ACM International Conference on AI in Finance. New York: ACM, 2024: 195-203.
[6]	ADLOORI H, DASANAPU V, MERGU A C. Graph Network Models to Detect Illicit Transactions in Block Chain[EB/OL]. (2024-09-23)[2025-05-28]. http://arxiv.org/abs/2410.07150.
[7]	HUMRANAN P, SUPRATID S. A Study on GCN Using Focal Loss on Class-Imbalanced Bitcoin Transaction for Anti-Money Laundering Detection[C]// IEEE. 2023 International Electrical Engineering Congress(iEECON). New York: IEEE, 2023: 101-104.
[8]	OSTERRIEDER J, CHAN S, CHU J, et al. Enhancing Security in Blockchain Networks: Anomalies, Frauds, and Advanced Detection Techniques[EB/OL]. (2024-02-17)[2025-05-10]. http://arxiv.org/abs/2402.11231.
[9]	PETTERSSON R E, ANGELIS J. Combating Money Laundering with Machine Learning-Applicability of Supervised-Learning Algorithms at Cryptocurrency Exchanges[J]. Journal of Money Laundering Control, 2022, 25(4): 766-778. doi: 10.1108/JMLC-09-2021-0106 URL
[10]	ZHU Huijuan, CHEN Jinfu, LI Zhiyuan, et al. Blockchain Anomaly Transaction Detection Method Based on Adaptive Multi-Feature Fusion[J]. Journal on Communications, 2021, 42(5): 41-50. doi: 10.11959/j.issn.1000-436x.2021030
	朱会娟, 陈锦富, 李致远, 等. 基于多特征自适应融合的区块链异常交易检测方法[J]. 通信学报, 2021, 42(5):41-50. doi: 10.11959/j.issn.1000-436x.2021030
[11]	MONAMO P, MARIVATE V, TWALA B. Unsupervised Learning for Robust Bitcoin Fraud Detection[C]// IEEE. 2016 Information Security for South Africa(ISSA). New York: IEEE, 2016: 129-134.
[12]	SAYADI S, BEN R S, CHOUKAIR Z. Anomaly Detection Model over Blockchain Electronic Transactions[C]// IEEE. 2019 15th International Wireless Communications & Mobile Computing Conference(IWCMC). New York: IEEE, 2019: 895-900.
[13]	HASAN M, RAHMAN M S, JANICKE H, et al. Detecting Anomalies in Blockchain Transactions Using Machine Learning Classifiers and Explainability Analysis[EB/OL]. (2024-01-07)[2025-04-28]. https://doi.org/10.1016/j.bcra.2024.100207.
[14]	FELDMAN E V, RUCHAY A N, MATVEEVA V K, et al. Bitcoin Abnormal Transaction Detection Based on Machine Learning[C]// Springer. Recent Trends in Analysis of Images, Social Networks and Texts. Heidelberg: Springer, 2021: 205-215.
[15]	ZHANG Xiaoqi, LI Guangsong, WANG Yongjuan. GAN-Based Abnormal Transaction Detection in Bitcoin[C]// IEEE. 2022 IEEE 7th International Conference on Smart Cloud(SmartCloud). New York: IEEE, 2022: 157-162.
[16]	JIANG Wenlong, YIN Pei, ZHU Wangwei. Cryptocurrency Transaction Fraud Detection Based on Imbalanced Classification with Interpretable Analysis[C]// Springer. New Challenges and Opportunities for Digital-Enabled Intelligent Future. Heidelberg: Springer, 2024: 386-398.
[17]	WANG Chundong, YANG Weijie. Semi-Supervised Blockchain Anomaly Transaction Detection Based on Deep AutoEncoder and Multi-Layer Perceptron[C]// IEEE. 2024 4th International Conference on Digital Society and Intelligent Systems(DSInS). New York: IEEE, 2024: 115-119.
[18]	ALOTIBI J, ALMUTANNI B, ALSUBAIT T, et al. Money Laundering Detection Using Machine Learning and Deep Learning[J]. International Journal of Advanced Computer Science and Applications, 2022, 13(10): 732-738.
[19]	AIRLANGGA G. Deep Learning for Anomaly Detection and Fraud Analysis in Blockchain Transactions of the Open Metaverse[J]. Jurnal Informatika Ekonomi Bisnis, 2024, 6(2): 324-329.
[20]	UMER Q, LI Jianwei, ASHRAF M R, et al. Ensemble Deep Learning-Based Prediction of Fraudulent Cryptocurrency Transactions[J]. IEEE Access, 2023, 11: 95213-95224. doi: 10.1109/ACCESS.2023.3310576 URL
[21]	YUE Yuyu, ZHANG Jixin, ZHANG Mingwu, et al. An Abnormal Account Identification Method by Topology Feature Analysis for Blockchain-Based Transaction Network[EB/OL]. (2024-04-09)[2025-04-28]. https://doi.org/10.3390/electronics13081416.
[22]	LIAO Qian, GU Yijun. Anomaly Transaction Detection in Bitcoin Network Based on LSCP Algorithm[J]. Computer Engineering and Applications, 2022, 58(15): 117-123. doi: 10.3778/j.issn.1002-8331.2105-0028
	廖茜, 顾益军. 基于LSCP算法的比特币网络异常交易检测[J]. 计算机工程与应用, 2022, 58(15):117-123. doi: 10.3778/j.issn.1002-8331.2105-0028
[23]	NERURKAR P. Illegal Activity Detection on Bitcoin Transaction Using Deep Learning[J]. Soft Computing, 2023, 27(9): 5503-5520. doi: 10.1007/s00500-022-07779-1
[24]	POCHER N, ZICHICHI M, MERIZZI F, et al. Detecting Anomalous Cryptocurrency Transactions: An AML/CFT Application of Machine Learning-Based Forensics[EB/OL]. (2023-03-18)[2025-04-28]. https://doi.org/10.48550/arXiv.2206.04803.
[25]	JIAO Feipeng, YUE Ruiqing. BN-GCN: Detecting Abnormal Bitcoin Transactions Using Graph Neural Networks[C]// IEEE. 2024 IEEE 6th International Conference on Power, Intelligent Computing and Systems(ICPICS). New York: IEEE, 2024: 1412-1416.
[26]	ZHENG Yang. GRU-GAT Model for Blockchain Bitcoin Abnormal Transaction Detection[C]// IEEE. 2022 IEEE Conference on Telecommunications, Optics and Computer Science(TOCS). New York: IEEE, 2022: 666-674.
[27]	ZHAO Kewei, DONG Guixin, BIAN Dong. Detection of Illegal Transactions of Cryptocurrency Based on Mutual Information[EB/OL]. (2023-02-27)[2025-04-28]. https://doi.org/10.3390/electronics12071542.
[28]	YAO Chenhao, FU Zhangjie, LIU Tianyi, et al. Attention-Driven Topology Adaptive Graph Convolutional Networks for Bitcoin Abnormal Transaction Detection[C]// IEEE. 2024 Twelfth International Conference on Advanced Cloud and Big Data(CBD). New York: IEEE, 2024: 25-30.
[29]	YU Lian, JING Qi, LI Ruomiao, et al. ParGCN: Abnormal Transaction Detection Based on Graph Neural Networks[C]// IEEE. 2022 IEEE 22nd International Conference on Software Quality, Reliability and Security(QRS). New York: IEEE, 2022: 797-808.
[30]	DING Zhihao, SHI Jieming, LI Qing, et al. Effective Illicit Account Detection on Large Cryptocurrency MultiGraphs[C]// ACM. The 33rd ACM International Conference on Information and Knowledge Management. New York: ACM, 2024: 457-466.
[31]	WEBER M, DOMENICONI G, CHEN Jie, et al. Anti-Money Laundering in Bitcoin: Experimenting with Graph Convolutional Networks for Financial Forensics[EB/OL]. (2019-07-31)[2025-05-20]. http://arxiv.org/abs/1908.02591.
[32]	ELMOUGY Y, LIU Ling. Demystifying Fraudulent Transactions and Illicit Nodes in the Bitcoin Network for Financial Forensics[C]// ACM. The 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. New York: ACM, 2023: 3979-3990.

数据集	异常交易		正常交易		交易总数/条
数据集	数量/条	比例	数量/条	比例	交易总数/条
训练集	2770	11.66%	20989	88.34%	23759
验证集	692	11.65%	5248	88.35%	5940
测试集	1083	6.63%	15263	93.37%	16346
合计	4545	9.87%	41500	90.13%	46045

模型	超参数	数值	搜索范围
LR	最大迭代次数	1000次	[500,2000]
	正则化参数	1.0	[0.01,10]
	随机种子	42	固定
XGB	最大树深度	5	[3,10]
	学习率	0.08	[0.01,0.5]
	树数量	400棵	[50,500]
	L2正则化参数	0.8	[0,10]
	最小子节点权重	2	[1,10]
	随机种子	42	固定
MLP	隐藏层结构	[100,100]	{[50,50],[100,100],[200,200]}
	激活函数	ReLU	固定
	学习率η	0.001	[0.0001,0.01]
	最大迭代次数	1000次	[500,2000]
	批量大小	200	[100,500]
	随机种子	42	固定
RF	树数量	300棵	[50,500]
	最大深度	15	[10,30]
	每次分裂特征数	40个	$\left\{ {{\log }_{2}}50,50 \right\}$
	最小分裂样本数	3个	[2,10]
	随机种子	42	固定

处理方法	精确率	召回率	F1值
无处理	0.6605	0.7202	0.6890
ROS	0.6997	0.7165	0.7080
SMOTE	0.9132	0.6316	0.7467
ADASYN	0.9898	0.6260	0.7670

特征名称	Lasso系数
Aggregate_feature_55	0.0527
Aggregate_feature_8	0.0387
Local_feature_87	0.0366
Local_feature_86	0.0245
Aggregate_feature_49	0.0234
Aggregate_feature_19	0.0198
Local_feature_48	0.0162
Local_feature_27	0.0152
Local_feature_63	0.0147
Local_feature_20	0.0135

模型	特征集	精确率	召回率	F1值
LR	完整	0.2962	0.7091	0.4178
LR	筛选	0.3615	0.6436	0.4630
XGB	完整	0.9928	0.6399	0.7782
XGB	筛选	0.9916	0.6519	0.7866
RF	完整	0.9898	0.6260	0.7670
RF	筛选	0.9412	0.6796	0.7893
MLP	完整	0.7734	0.6334	0.6964
MLP	筛选	0.7527	0.5762	0.6527