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.
出处
《计算机应用与软件》
CSCD
2011年第2期287-289,共3页
Computer Applications and Software
关键词
支持向量机
遗传算法
检测精度
入侵检测
Support vector machine Genetic algorithm Detection accuracy Intrusion detection
作者简介
姜春茂，副教授，主研领域：嵌入式系统，操作系统安全。
