一种基于AC-RBF神经网络的网络安全态势预测方法

为了准确地把握网络安全发展态势,提出了一种基于自适应聚类径向基函数(adaptive clustering radical basis function,AC-RBF)神经网络的网络安全态势预测(network security situation prediction, NSSP)方法?该方法对网络安全态势样本自适应聚类,获得了神经网络隐层节点数,采用梯度下降法训练神经网络,寻找网络安全态势样本之间的非线性映射关系,利用该关系对未来时刻网络安全态势进行了预测? 仿真实验表明,相对于 K-均值 RBF 神经网络及支持向量机(support vector machine,SVM)预测模型,该方法在神经网络规模较小的情况下,不仅能够反映网络安全态势的总体趋势,而且还提高了预测精度,能够提供给网络安全管理员一个直观的网络安全态势图 ?

A method of network security situation prediction based on AC-RBF neural network

To grasp the trend of network security,a method of network security situation prediction (NSSP)based on adaptive clustering radical basis function(AC-RBF)neural network is proposed.The neural network hidden nodes are obtained by clustering the network security situation samples adaptively.We can train the neural network by gradient descent and find out the nonlinear relations among situation samples, to predict the future security situation.Experiment results show that,compared with the k-means RBF neural network and Support Vector Machine(SVM)prediction model,the proposed method can not only reflect the general trend of network security situation,but also can improve the prediction accuracy in the case of small-scale neural network.Finally,the proposed method can provide the network administrators with an intuitive network security situation map.