面向多维数据的异常点检测模型设计  

Design of outlier detection model for multidimensional data

作　　者：马勇[1] 杨敏[2] 朱琳[1] Ma Yong;Yang Min;Zhu Lin(Network Information Center Inner Mongolia University of Science and Technology Baotou Medical College,Baotou 014040,China;Dean′s Office Inner Mongolia University of Science and Technology Baotou Medical College,Baotou 014040,China)

机构地区：[1]内蒙古科技大学包头医学院网络信息中心,内蒙古包头014040 [2]内蒙古科技大学包头医学院教务处,内蒙古包头014040

出　　处：《网络安全与数据治理》2023年第7期85-90,共6页CYBER SECURITY AND DATA GOVERNANCE

基　　金：包头医学院自然科学类青苗计划(BYJJ-ZRQM202320)。

摘　　要：为了在大数据环境下快速、精准地挖掘异常点,保障网络安全,提出了一种面向多维数据的异常点检测模型设计方案。该方案利用长短期记忆网络(LSTM)存储任意时间段的多维数据,并使用图卷积网络提取完整数据结构,同时加入惩罚参数和均方误差来缩小异常点出现范围。此外,还利用编码器和解码器构建变分自编码器函数模型,使其能够解读正常数据子特征,并通过编码重建损失函数来计算数据异常度量,从而实现异常点检测。经过实验验证,该方法表现出较高的检测正确率和运行效率,具有极高的应用价值。In order to quickly and accurately mine outliers in the big data environment and ensure network security,we propose a design scheme for multidimensional data oriented outlier detection model.In this scheme,the long short memory network(LSTM) is used to store multi-dimensional data in any period of time,and the graph convolution network is used to extract the complete data structure.At the same time,penalty parameters and mean square error are added to narrow the range of outliers.In addition,we also use the encoder and decoder to build a variational self encoder function model,so that it can interpret the normal data sub features,and calculate the data anomaly measurement through the coding reconstruction loss function,so as to achieve outlier detection.After experimental verification,this method exhibits high detection accuracy and operational efficiency,and has high application value.

关 键 词：编码损失函数 变分自编码器 异常点检测 长短期记忆网络 多维数据 

分 类 号：TP995[自动化与计算机技术]