摘要
提出了一种将核Fisher鉴别分析特征抽取与多分类支持向量机算法结合的网络入侵检测技术,扩展了二分类支持向量机,利用经过核Fisher鉴别分析特征抽取后的训练数据构造优化的决策树,从而实现支持向量机的多分类。实验结果表明该算法能够提高检测正确率,同时降低训练时间,取得了良好的效果。
This paper presents a nuclear Fisher discriminant analysis feature extraction and multi-classification support vector machine algorithm combining network intrusion detection technology, the expansion of the two classification support vector machine, to use the nuclear Fisher discriminant analysis feature extraction of training data structure after the optimization the decision tree, support vector machine in order to achieve the multi-classification. Experimental results show that the algorithm be able to improve detection accuracy and reduce training time, and achieved good results.
出处
《科技创业月刊》
2009年第12期178-179,共2页
Journal of Entrepreneurship in Science & Technology
