智能计算在网络入侵检测中的应用研究  

Research on Network Intrusion Detection Algorithm Based on Intelligent Computing

作　　者：王延中[1] 

机构地区：[1]河南质量工程职业学院,河南平顶山467001

出　　处：《科技通报》2013年第2期184-185,188,共3页Bulletin of Science and Technology

基　　金：河南省教育厅课题(2010B5200015)

摘　　要：研究了基于遗传算法和差分进化计算的入侵检测系统中的特征选择技术,差分进化计算在变异过程中并未考虑到适应度大的个体,同时存在过早收敛问题,而遗传算法需要很多的迭代次数才能收敛。针对以上缺点,结合模拟退火算法对差分进化的变异过程进行改进,同时设计合理的适应度函数,使得该算法收敛于最优特征子集。经过Lincoln实验室入侵检测系统评估数据集合MIT’1998测试,改进算法与差分进化算法和遗传算法相比,具有良好的收敛性能,并且收敛特性稳定。Feature selection technique is studied based on differential evolution algorithm and genetic algorithm,differential evolution algorithm has not taken into account of the individual having bigger fitness value,and has the premature problem.Genetic algorithm need more iterative times to converge.In view of the above shortcomings combined with the simulated,an improved algorithm is worked out.Rational fitness function id designed,and the algorithm converges to the optimal feature subset.The improved algorithm is tested using Lincoln laboratory intrusion detection system evaluation data set MIT’1998.The improved algorithm has good convergence performance and stable convergence property compared with differential evolution algorithm and genetic algorithm.

关 键 词：差分进化 模拟退火 变异 适应度函数 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]