摘要
针对LS-SVM的网络入侵检测技术存在检测率低和误判率高的缺点,结合果蝇优化算法的快速寻优和全局最优的优点,提出一种FOA优化LS-SVM的网络入侵检测方法。通过FOA优化LS-SVM的惩罚因子C和核函数参数g,实现网络入侵类型的检测。以KDD99 CUP数据集为研究对象,实验结果表明,FOA-LSSVM算法在分类性能和分类准确率上都优于单纯的LS-SVM和BP,FOA-LSSVM算法的网络入侵检测率平均高达96.33%。
For network intrusion detection technology using SVM has the shortcomingsollow and high rateol lalse positives,combined with Iruit flying optimization algorithm rapid optimization and the advantages ol the global optimal, FOA is applied to optimize LS-SVM for network intrusion detection. Through FOA optimizing LS-SVM penalty factor C and the kernel function parameterg ,it realizes network intrusion type detection. Taking KDD 99 CUP dataset as study on object,the experimental results showed that FOA-LSSVM algorithms on classification performance and classification accuracy rate have better than LS-SVM and BP , the average detection rate is up to 96. 33% , FOA- LSSVM algorithm has much higher detection rate than LS- SVM and BP .
出处
《信息技术》
2017年第2期173-176,共4页
Information Technology
关键词
网络入侵
支持向量机
果蝇优化算法
检测率
network intrusion
support vector machine
fruit flying optimization algorithm
detection rate