Abstract: Feature selection is a core problem for network intrusion detection, in order to improve the detection rate of network intrusion, a network intrusion detection model（AFSA-SVM） is proposed based on Artificial Fish Swarm Algorithm and Support Vector Machine. The feature subset is coded as the position of adult fish, and the detection rate of 5 cross validation for SVM training model is taken as evaluation criteria of the feature subset, and then the fish feeding, clustering and rear-end behavior are imitated to find the optimal feature subset. The intrusion detection model is built based on the optimal feature subset. The simulation experiment is carried out on the KDD CUP 99 data. The results show that, compared with the Particle Swarm Optimization algorithm, Genetic Algorithm and all features, the proposed algorithm has improved detection efficiency and the detection rate of the network intrusion, so it is an efficient intrusion detection model.

Key words: feature selection, Artificial Fish Swarm Algorithm（AFSA）, Support Vector Machine（SVM）, intrusion detection

摘要： 特征选择是网络入侵检测研究中的核心问题，为了提高网络入侵检测率，提出一种人工鱼群算法（AFSA）和支持向量机（SVM）相融合的网络入侵检测模型（AFSA-SVM）。将网络特征子集编码成人工鱼的位置，以5折交叉验证SVM训练模型检测率作为特征子集优劣的评价标准，通过模拟鱼群的觅食、聚群及追尾行为找到最优特征子集，SVM根据最优特征子集进行网络入侵检测，并采用KDD CUP 99数据集进行仿真测试。仿真结果表明，相对于粒子群优化算法、遗传算法和原始特征法，AFSA-SVM提高了入侵检测效率和检测率，是一种有效的网络入侵检测模型。

关键词: 特征选择, 人工鱼群算法, 支持向量机, 网络入侵检测