一种基于改进支持向量机的入侵检测方法研究

摘要

提出基于粒子群优化(Particle Swarm Optimization,PSO)算法和支持向量机(Support Vector Machines,SVM)的入侵检测方法,为优化SVM性能,使用PSO的全局搜索特性寻找SVM的最优参数C和σ；为避免PSO算法陷入局部最优,引入变异操作,找到最优参数组合后进行基于PSO SVM入侵检测算法的训练和检测,解决了入侵检测系统准确度难题.仿真实验表明该方法的检测率为92.8％,误报率为6.911 9％,漏报率为9.708 7％,对KDDCUP竞赛的最佳结果有一定程度的提高,实验结果验证了该算法的有效性和可行性.%An intrusion detection method based on SVM combined with PSO is proposed. The global search characteristic of PSO is used to search for the best SVM's parameter: C and σ, and mutation operation is introduced in PSO in order to obtain globally optimal solutions. After finding the optimal C and σ, training and testing operation of intrusion detection system based on PSOSVM are performed. It has high real-time and accuracy. The simulation results show that the detection rate is 92.8%, false alarm is 6.9119%and losing alarm is 9.7087%. It verifies the effectiveness and feasibility of the proposed algorithm.