摘要
交互终端界面可分为桌面、非桌面、可见及不可见几种类型,因此交互界面的大数据具有多样化和无序化特点,导致AI交互终端安全攻击事件时有发生。为增强AI交互安全性,提出AI交互终端大数据异常入侵风险识别方法。构建AI交互终端数据模型,获得入侵数据参变量在终端内簇首节点中的布局函数,构建异常数据入侵节点的路由拓扑模型。根据能量损耗测量频谱,得到交互终端数据布局全部簇的位置,拟合终端数据信息流二维信号。基于此,利用主成分分析提取异常值,筛选关联密切信息,保证成分互不干扰。根据排列数据簇,划分正常数据与异常入侵数据,利用网络发生器,得到入侵特征和权值,完成AI交互终端大数据异常入侵风险识别。实验结果显示,所提方法的AI交互终端大数据异常识别率可达95%以上,误检率低于10%,确保了AI终端的安全,可有效减小用户损失。
Interactive terminals can be divided into desktop,non-desktop,visible and invisible interfaces.In order to enhance the security of AI interaction,a method of identifying invasion risks of abnormal big data in AI interaction terminals was proposed.Firstly,AI interactive terminal data model was constructed to obtain the layout function of intrusion data parameters of cluster head nodes within the terminal.Secondly,the routing topology model of abnormal data intrusion nodes was built.And then the positions of all clusters in the interactive terminal data layout were obtained by the spectrum measurement of energy loss.Moreover,the two-dimensional signals of terminal data flow were fitted.On this basis,principal component analysis was adopted to extract abnormal values and screen the information with close relations,thus ensuring that components did not interfere with each other.According to the arrangement of data clusters,abnormal intrusion data were separated from normal data.In the meanwhile,the network generator was used to find intrusion characteristics and weights.Finally,the abnormal intrusion risk identification of big data in AI interactive terminals was completed.Experimental results show that the recognition rate of abnormal data in AI interactive terminals can reach more than 95%,and the false detection rate is less than 10%,so this method ensures the safety of AI terminals and effectively reduces user losses.
作者
刘明
张弘
LIU Ming;ZHANG Hong(School of Information Engineering,Zhengzhou University of Industry Technology,Zhengzhou Henan 450000,China;School of Electronic Technology,Information Engineering University,Zhengzhou Henan 450000,China)
出处
《计算机仿真》
北大核心
2022年第11期467-471,共5页
Computer Simulation
关键词
交互终端
大数据异常入侵
入侵风险识别
数据特征提取
对角矩阵
AI interactive terminal
Abnormal intrusion of big data
Intrusion risk identification
Data feature extraction
Diagonal matrix