%0 Journal Article
%T 面向云平台安全监控多维数据的离群节点自识别可视化技术<br>Multidimensional data visualization in cloud platform security monitoring
%A 吴頔
%A 王丽娜
%A 余荣威
%A 章鑫
%A 徐来&lt
%A br&gt
%A WU Di
%A WANG Li-na
%A YU Rong-wei
%A ZHANG Xin
%A XU Lai
%J 山东大学学报（理学版）
%D 2017
%R 10.6040/j.issn.1671-9352.2.2016.216
%X 摘要： 通过总结目前云平台安全监控的数据可视化技术,结合具体的多维监控数据探讨可视化技术的应用方法,从时间、节点号、性能指标类型三个维度出发,提出了基于维度压缩与维度切面的性能数据集可视化方法,并在此基础上,应用动态时间规划和卷积神经网络实现离群节点自识别,丰富扩展了警报系统的语义。经实验验证方法可行,能够更直观地展现有效信息,提高云管理员的决策效率。<br>Abstract: Discussed application method of visualization technology by summarized visualization technology of cloud platform security monitoring and combined with the specific multidimensional monitoring data. Started from the three dimensions of time, node number and performance index type, a visualization method of performance data set based on dimension compression and dimension slice is proposed. Based on this, node self-identification of outliers achieved by used the dynamic time planning and convolution neural network, which extends and riches semantics of the alarm system. It was proved that the method was feasible and could show the effective information more intuitively and improve the decision efficiency of the cloud administrator
%K 云安全
%K 云平台安全监控
%K 可视化技术
%K 离群节点自识别
%K &lt
%K br&gt
%K visualization
%K outlier node self-identification
%K cloud security monitoring
%K cloud security
%U http://lxbwk.njournal.sdu.edu.cn/CN/10.6040/j.issn.1671-9352.2.2016.216