基于遗传算法支持向量机的网络入侵预测  

Support Vector Machines Based on Genetic Algorithm for Network Intrusion Prediction

作　　者：谢志强[1] 

机构地区：[1]广东外语艺术职业学院,广东广州510507

出　　处：《计算机仿真》2010年第8期110-113,共4页Computer Simulation

摘　　要：在预测网络安全问题的研究中,针对网络入侵检测优化问题,为了改变传统入侵检测算法存在训练精度高,预测精度相当低的过拟合难题,提出一种基于遗传算法的支持向量机。支持向量机首先利用遗传算法搜索最优的支持向量机参数,然后用得到的最优参数来训练,利用训练得到的最优算法模型对测试集进行建模预测。并利用支持向量机对KDD 1999 CUP数据集进行了仿真。实验结果表明,方法在降低训练时间的同时有着很好的检测率,优于经典的神经网络算法,方法提高了预测效率。In the study of network security predicting,aiming at the optimization of network intrusion detection,a genetic algorithm based on support vector machine is proposed,in order to sovle the problems that the traditional intrusion detection algorithm has high accuracy and over-fitting with relatively low prediction accuracy.First,genetic algorithm is used to search optimal parameters of support vector machines.Then the optimal parameters are trained.The optimal algorithm is used in predicting model on the test set.The support vector machine is used to simulate the KDD 1999 CUP data set.The results show that the method has very good detection rate while the training time is reduced,better than the classical neural network algorithm.

关 键 词：遗传算法 支持向量机 网络入侵 检测 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]