组合核函数相关向量机的网络安全态势预测  

Network security situation forecasting based on combining kernels function relevance vector machine

作　　者：刘付民[1] 王静咏[1] 张治斌[1] 

机构地区：[1]河南理工大学计算机科学与技术学院,河南焦作454000

出　　处：《计算机应用研究》2016年第8期2417-2419,2424,共4页Application Research of Computers

基　　金：河南省软科学研究计划资助项目(102400450034);河南省科技计划基金资助项目(092300410216)

摘　　要：为了提高网络安全态势的预测精度,针对单一核函数的局限性,提出了一种组合核函数相关向量机的网络安全态势预测模型。首先对网络安全态势时间序列进行重新构造,得到相关向量机的学习样本,然后采用多项式和高斯核函数构建组合核函数,并采用组合核函数相关向量机对网络安全态势样本进行学习,建立网络安全态势预测模型,最后对网络安全态势预测性能进行测试。实验结果表明,相对于单一核函数相关向量机以及其他网络安全态势预测模型,组合核函数相关向量机提高了网络安全态势的预测准确性,可以满足网络安全态势预测的实际应用需求。In order to improve the prediction accuracy of network security situation, due to limitations of single kernel func- tion, this paper put forward a network security situation forecasting model based on combining kernels function relevance vector machine. Firstly, it struclured the network security time series to get learning samples of the relevant vector machine, and then constructed the combined kernel function with polynomial and Gauss kernel function, and used relevance vector machine with combined kernel function to establish the forecast model of network security situation. Finally it tested the performance of the model. The results show that, compared with the single kernel relevance vector machine and other network security situation forecasting models, relevance vector machine with combined kernel function has improved the prediction accuracy of network security situation, and carl be used to meet the prediction needs of network security situation.

关 键 词：网络安全 组合核函数 相关向量机 预测模型 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]