基于网格搜索的 SVM 在入侵检测中的应用

摘要

随着网络的快速普及和发展，网络安全问题日益突出，如何保障网络安全已经成为一个国际化问题。在众多方法中，入侵检测技术是解决这一问题的有效手段。文中将支持向量机方法运用在入侵检测中。首先，介绍了基于 SVM 的入侵检测技术研究现状；然后，将网格搜索算法应用在 SVM 参数寻优中；最后，通过实验，将 PSO 算法、GA 算法、网格搜索算法对 SVM 参数优化的结果进行比较。实验结果表明，使用网格搜索法对 SVM 参数进行优化，具有最好的泛化精度，并且在此基础上，对数据集进行归一化处理，将大幅度减少构建分类器的迭代次数，从而减少预测时间。因此，可以认为基于网格搜索的支持向量机能够很好地实现入侵检测。%With the rapid popularization and development of network,network security problems are becoming increasingly prominent. How to guarantee the security of the network has become an international problem. Among the many methods,intrusion detection technol-ogy is an effective means to solve this problem. In this paper,Support Vector Machine (SVM) method will be used in intrusion detec-tion. First of all,the current situation of the intrusion detection technology is introduced based on SVM. Secondly,the grid search algo-rithm is used into the optimization of the SVM’s parameters. At last,bring the result of the SVM’s parameters that based on PSO algo-rithm,GA algorithm and grid search algorithm into comparison. The results of the experiment show that using the grid search method for optimization of SVM’s parameters has the best generalized accuracy,and on this basis,the normalization of dataset will greatly reduce the number of the classifier’s iterations,so as to reduce the forecast time. Therefore,it is considered that SVM based on grid search can real-ize the intrusion detection excellently.