摘要
研究一种基于群智能算法优化神经网络的网络安全事件分析模型,使用遗传算法和LMS算法对常规RBF神经网络中的隐含层神经元个数、基函数中心以及各层连接阈值和权值进行优化,得到最优解,从而提高RBF神经网络模型的训练效率和精度,提高基于RBF神经网络的网络安全事件分析效率和准确度。使用KDD CUP99数据集中的网络入侵事件数据对研究的网络安全入侵事件分析模型进行实例研究,测试结果表明,该分析模型相比常规神经网络算法建立的模型具有更高的识别准确率,能够准确识别分析正常事件和四种网络攻击入侵事件。
A network security event analysis model based on swarm intelligence algorithm optimizing neural network is stu- died. The genetic algorithm and LMS algorithm are used to optimize the hidden layer neurons quantity, basis function center, connection threshold and weight of each layer of the conventional RBF neural network, so as to obtain the optimal solution, im- prove the training efficiency and accuracy of the RBF neural network model, and the efficiency and accuracy of the network se- curity event analysis based on RBF neural network. The network intrusion event data in KDD CUP99 dataset is used to perform the instance study for the network security intrusion events analysis model. The test results show that the analysis model has more recognition accuracy than the model established by the conventional neural network algorithm, and can accurately identify and analyze the normal events and four network attack events.
作者
高峰
GAO Feng(Software Engineering Institute, Chongqing University of Arts and Sciences, Chongqing 402160, China)
出处
《现代电子技术》
北大核心
2016年第21期123-126,共4页
Modern Electronics Technique
作者简介
高峰（1982-），男，重庆长寿人，实验师。研究方向为计算机科学与技术、网络工程。
