计算机应用研究2024，Vol.41Issue(1)：21-25,31,6.DOI:10.19734/j.issn.1001-3695.2023.05.0207
基于GAT与SVM的区块链异常交易检测
Blockchain anomaly transaction detection based on GAT and SVM
摘要
Abstract
Public chains face numerous problems of malicious transactions and illegal cryptographic activities because of their transparency and openness,which cause anomalous transactions in blockchains and cause serious damage to users'assets and information security.To address the problem of blockchain abnormal transactions,this paper proposed a blockchain abnormal transaction detection method based on the fusion of GAT and SVM,which focused on the features and connections of the local structure of the blockchain transaction graph neighbor nodes-GAS.In GAS,it utilized the random forest to evaluate the importance of transaction data features of nodes,and selected the top 140 important features in descending order.Then,com-bining with the features of neighboring nodes,it used GAT to update the features of the current node.The updated features served as input to SVM for anomaly detection.Experimental results demonstrate that compared to non-integrated methods,GAS shows superior performance in detecting anomalies,with an accuracy rate of 98.11％,precision of 94.01％,and recall rate of 85.48％.关键词
区块链/图注意力神经网络/异常交易检测/支持向量机Key words
blockchain/graph attention network/abnormal transaction detection/SVM分类
信息技术与安全科学引用本文复制引用
谭朋柳,周叶..基于GAT与SVM的区块链异常交易检测[J].计算机应用研究,2024,41(1):21-25,31,6.基金项目
国家自然科学基金资助项目(61961029) （61961029）
江西省科技厅重点研发计划项目(20171ACE50025) （20171ACE50025）
