Journal of Information Security Reserach ›› 2023, Vol. 9 ›› Issue (2): 98-.

    Next Articles

Research on Blockchain Anomaly Transaction Detection Technology  Based on Stacking Ensemble Learning

  

  • Online:2023-02-01 Published:2023-01-24

基于Stacking集成学习的区块链异常交易检测技术研究

王志强;王姿旖;倪安发;   

  1. (北京电子科技学院北京102627)
  • 通讯作者: 王志强 博士,副教授.主要研究方向为网络空间安全和漏洞挖掘. wangzq@besti.edu.cn
  • 作者简介:王志强 博士,副教授.主要研究方向为网络空间安全和漏洞挖掘. wangzq@besti.edu.cn 王姿旖 硕士研究生.主要研究方向为网络空间安全. 1157942740@qq.com 倪安发 硕士研究生.主要研究方向为网络空间安全. anfa_ni@foxmail.com

Abstract: In order to efficiently detect abnormal transactions on the blockchain, this paper proposes a method  based on Stacking integration learning. Firstly, XGBoost, LightGBM, CatBoost and LCE are used as the base classifier, and MLP is used as the metaclassifier, and the MLP_Stacking integrated learning algorithm is designed. Secondly,SUNDO is used for data augmentation to solve the problem of serious imbalance in data sets; Finally, a multimodel joint feature sorting algorithm is designed to generate an optimal subset of features, and the resulting optimal subset of features is used as the input data set of the MLP_Stacking for classification training to achieve model optimization. This paper experiments at the open source dataset provided by Kaggle platform , and the experimental results show that the SUNDO data generation method can effectively improve the performance of each classifier, and  the training effect of the integrated model designed in this paper is obviously better than that of the individual model.

Key words: class imbalance, Stacking, anomaly detection, blockchain, ensemble learning

摘要: 摘要为了高效地进行区块链异常交易检测,提出了一种基于Stacking集成学习的区块链异常交易检测方法.首先,采用XGBoost,LightGBM,CatBoost,LCE作为基分类器,采用MLP作为元分类器,设计了MLP_Stacking集成学习算法;其次,利用SUNDO进行数据扩充,解决数据集中严重类不均衡问题;最后,设计多模型联合特征排序算法,生成最优特征子集,将得到的最优特征子集作为MLP_Stacking输入数据集进行分类训练,通过网格搜索优化参数实现模型优化.实验采用Kaggle平台提供的开源数据集,实验结果显示采用SUNDO数据生成方法能有效提高各分类器性能,在此基础上,设计的集成模型训练效果明显优于单个模型.


关键词: 类不平衡, Stacking, 异常检测, 区块链, 集成学习