摘要
针对网络入侵特征高维性和特征冗余严重等问题,提出一种K近邻算法(KNN)和改进人工鱼群算法选择特征的网络入侵检测模型(AFSA-KNN)。计算特征之间的关联度,采用KNN算法消除原始网络数据中的冗余特征;将得到的特征子集作为AFSA初始解,通过模拟鱼群的觅食、聚群及追尾行为找到最优特征子集;建立网络入侵检测分类器。实验结果表明,AFSA-KNN有效消除了冗余特征,减少分类器输入维数,提高了网络入侵检测正确率和检测速度。
For the serious problems exist in the network intrusion, such as high dimension and redundancy, a network intrusion detection model based on K nearest neighbor algorithm and improved artificial fish swarm algorithm was presented. Firstly, the correlation degree between features was computed, and KNN algorithm was used to eliminate redundant features in the original network data. Then the obtained feature subsets were taken as the initial solution of AFSA, and the simulation of feeding, clus- tering and the following behavior was used to find the best subset of features. Finally, the network intrusion detection classifier was established. The results show that AFSA-KNN can effectively eliminate redundant features and reduce the input dimension of classifier. It can also improve the network intrusion detection accuracy and detection speed.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第8期2675-2679,共5页
Computer Engineering and Design
关键词
特征选择
入侵检测
K近邻
特征关联性
人工鱼群算法
feature selection
intrusion detection
KNN
feature relevance
artificial fish swarm algorithm