结合粗糙集和禁忌搜索的网络流量特征选择

Feature selection of network traffic using a rough set and tabu search

针对网络流量特征属性的优化选择问题,提出了一种结合粗糙集和禁忌搜索的网络流量特征选择方法(RS-TS).该方法通过粗糙集算法对网络流量特征属性进行约简,将所得到的特征子集作为禁忌搜索的初始解,并利用禁忌搜索得到最优特征子集.实验验证RS-TS方法优于基于GA的特征选择方法和基于IG的特征选择方法,能够有效地去除网络流量的冗余特征属性,提高网络流量分类精度.

A feature selection of network traffic using a rough set and tabu search （RS-TS） was proposed for the purpose of optimization in the feature selection of traffic classification. This approach reduced the traffic feature attribute with a rough set and established the feature subset as the initial value of a tabu search, as well as the optimal feature subset on the basis of a tabu search. The optimal feature subset with a tabu search can be selected on the basis of a feature subset. In contrast with the traditional feature selection methods based on GA and IG, RS-TS was vaXidated optimal by experimental results. It can diminish redundant feature attribution of network traffic effectively and greatly improve the classification accuracy.