基于GA与BP神经网络的网络入侵检测组合模型研究

针对入侵检测系统产生的高维数据的处理问题,提出基于GA与BP神经网络的入侵检测组合模型进行特征选择。为了优化入侵检测分类算法,利用遗传算法适合复杂系统优化的特点,去除入侵检测数据多维特征属性中的冗余部分,将入侵检测数据特征属性对应到染色体,BP神经网络的分类准确率作为种群个体的适应度值,通过遗传算法的全局搜索能力,找出对分类算法最有影响的特征属性组合,从而达到降维的目的。采用KDD99数据集进行分析,实验表明,经过组合算法特征选择的数据能在分类正确率、运算时间、运算稳定性等方面取得更优的效果。

A Research on Network Intrusion Detection Model Based on Genetic Algorithm and Back Propagation Neural Network

A hybrid algorithm was proposed which combined Genetic Algorithm（GA）and Back Propagation Neural Network（BPNN）to select feature characteristics from multi-dimensions data collected by Intrusion Detection System（IDS）.To optimize the classifying algorithm of IDS,GA which fits to complex system optimizing is used to remove the redundancy characteristics from intrusion detection feature.Chromosome denotes the feature characteristics of intrusion detection data,and the classifying accurate ratio of BPNN was used as the fitness value of individuals.The most important feature characteristics effecting classifying algorithm were found by GA which has the global searching ability.As a result,dimension of intrusion detection data was reduced.The KDD99 dataset were analyzed in the experiment.The experimental result showed that the feature data selected by the hybrid algorithm had better effect in several ways which include the classification-accuracy,run time of algorithm,stability of algorithm.