一种基于机器学习算法的网络安全评估方法

A network security evaluation method based on machine learning algorithm

互联网给人们生活和工作带来便利的同时,也增加了网络安全风险发生的概率。为减少这一危害的出现,提出一种基于机器学习的网络安全预测评估方法,该方法是利用支持向量机的非线性拟合能力和果蝇优化算法的全局优化能力,在时间序列的基础上建立网络安全态势评估模型。所提出的预测模型的决定系数为0.906997774231,最佳核函数参数为9.55886673069,输入维数为12,惩罚因子为26.6112511992。相较于RBF、PSO-SVM两种预测模型,FOA-SVM网络安全预测模型的准确率、AUC值、F1值分别为81.2%、0.83、0.83,能够更好地完成网络安全评估。

Internet not only brings convenience to people’s life and work,but also improves the probability of network security incidents.In order to reduce this harm,this study proposes a network security prediction and evaluation method based on machine learning.The method uses the nonlinear fitting ability of support vector machine and the global optimization ability of drosophila optimization algorithm to establish the network security situation assessment model on the basis of time series.The decision coefficient of the model is 0.906997774231,the optimal kernel function parameter is9.55886673069,the input dimension is 12,and the penalty factor is 26.6112511992.Compared with RBF and PSO-SVM,the accuracy rate,AUC value and F1 value of FOA-SVM network security prediction model are 81.2%,0.83 and 0.83 respectively,which can better complete the network security assessment.