基于模拟退火支持向量机的入侵检测系统

为了提高入侵检测系统在小样本集条件下的检测效率,将支持向量机用于网络入侵检测.支持向量机的参数决定了检测效率,然而难以选择合适的参数值,因此提出利用模拟退火算法来优化这些参数,并设计出基于参数优化的支持向量机用于入侵检测.通过对样本数据集中的样本进行实验性检测,并与原始支持向量机入侵检测系统进行比较,结果表明模拟退火支持向量机入侵检测系统检测率高、误报率低,并且缩短了训练时间和检测时间.

Intrusion detection system based on support vector machines with simulated annealing algorithms

In order to improve the detection efficiency of IDS (intrusion detection system) under the small sample conditions, SVM (support vector machine) is employed to IDS. The parameters of SVM are the Key factors of detection efficiency and difficult to choose appropriate parameter values. Therefore, SA (simulated annealing) algorithms are used in the proposed SVM model to optimize the parameter selection and SVM with optimized parameters for intrusion detection is designed. Through applied to treat the sample data and comparison of detection ability between the above detection method and the IDS based on original SVM, the results show that the intrusion detection system based on SVM with SA is efficient, lower false rate, and shorten the training time and detection time.