摘要
当前网络流量日趋复杂,给网络管理带来许多困难.为了准确地识别出网络中的各种流量,本文以支持向量机为分类器,以流的统计学特征为分类依据,提出一种组合式特征选择算法,该算法首先快速去除和分类不相关的特征,针对余下的特征,再利用遗传算法引导特征的选择和支持向量机模型参数的寻优,最终获得了最优的特征集和最佳的支持向量机分类模型.经过实验验证,基于该算法的网络流量识别方法在识别P2P流量时能以更少的特征获得更高的分类准确率.
Nowadays network traffic is increasingly complicated which brings many difficulties to network management. In order to accurately identify network traffic, this paper uses support vector machine as classifier and flow statistic features as discriminators, and proposes a hybrid feature selection algorithm. Firstly this algorithm quickly eliminates those flow features irrelevant to class, then facing to the rest features it makes use of genetic algorithm to select flow features and optimize the parameters of support vector ma- chine model, finally it outputs the best flow feature set and the optimized support vector machine model. The experimental results in- dicate that the improved accuracy and the fewer features are achieved by applying this algorithm to P2P traffic identification.
出处
《小型微型计算机系统》
CSCD
北大核心
2012年第2期325-329,共5页
Journal of Chinese Computer Systems
基金
国家"八六三"高技术研究发展计划基金项目(2009AA01A346)资助
关键词
P2P流量识别
流统计学特征
支持向量机
遗传算法
特征选择
P2P traffic identification
flow statistic feature
support vector machine
genetic algorithm
feature selection
