基于遗传算法优化SVM的嵌入式网络系统异常入侵检测  

ABNORMAL INTRUSION DETECTION FOR EMBEDDED NETWORK SYSTEM BASED ON GENETIC ALGORITHM OPTIMISED SVM

作　　者：姜春茂[1,2] 张国印[1] 李志聪[2] 

机构地区：[1]哈尔滨工程大学计算机学院,黑龙江哈尔滨150001 [2]哈尔滨师范大学计算机科学技术与信息工程学院,黑龙江哈尔滨150001

出　　处：《计算机应用与软件》2011年第2期287-289,共3页Computer Applications and Software

摘　　要：由于传统嵌入式网络系统入侵检测方法难以获得较高的检测精度,提出基于遗传算法优化的支持向量机(GA-SVM)的网络入侵检测技术。支持向量机分类器能够较好地解决少样本、高维、非线性分类问题。然而,支持向量机训练参数的选择对其分类精度有着很大影响,遗传算法能够同时优化支持向量机的训练参数,采用遗传算法进行支持向量机的训练参数同步优化。实验结果表明,这种遗传算法优化的支持向量机分类入侵检测模型有着很高的检测精度。As conventional intrusion detection method for embedded network system can＇t gain high detection accuracy,the network intrusion detection technology based on genetic algorithm optimised support vector machine（GA-SVM） is proposed in the paper.Support vector machine（SVM） classifier can well resolve non-linear classification problem with small training sample and high dimensions.However,the selection of training parameters of SVM has a great influence on classification accuracy.Genetic algorithm can consume less time to perform the optimisation of SVM parameters simultaneously.Thus,in the paper GA is used to synchronously optimise the SVM training parameters.Experimental results demonstrate that the GA-SVM classification intrusion detection model has very high detection accuracy.

关 键 词：支持向量机 遗传算法 检测精度 入侵检测 

分 类 号：TP393.11[自动化与计算机技术—计算机应用技术]