基于遗传优化神经网络的网络入侵特征检测  

Network intrusion detection based on neural network optimized by GA

作　　者：陈鸿星[1] 

机构地区：[1]江西师范大学数学与信息科学学院,南昌330022

出　　处：《计算机工程与应用》2014年第14期78-81,共4页Computer Engineering and Applications

摘　　要：为了提高网络入侵检测正确率,提出一种遗传优化神经网络的网络入侵特征选择和检测算法。该方法先将网络状态特征和RBF神经网络参数作为遗传算法的个体,把检测正确率作为适应度函数;然后利用遗传算法的选择、交叉和变异等操作对网络状态特征和RBF神经网络参数进行优化,最后利用KDD 1999数据集对算法性能进行测试。测试结果表明:遗传优化神经网络能够快速获得最优网络状态特征和分类器参数,同时提高了网络入侵检测正确率。To improve network intrusion detection rate, a network intrusion detection algorithm based on RBF neural net-work optimized by Genetic Algorithm（GA）is proposed in this paper. Firstly, the feature and RBFNN parameters are taken as genetic algorithm individuals while the network intrusion detection rate as the evaluation function, secondly, the opti-mal features and parameters are selected by selection, crossover and mutation, lastly, the algorithm performance is tested by KDD 1999 data. The results from performance analysis show that the proposed algorithm can select the optimal fea-tures and parameters quickly, and improves the network intrusion detection rate.

关 键 词：网络入侵 特征选择 遗传算法 径向基函数(RBF)神经网络 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]