基于邻域粗糙集的网络入侵分类诊断组合模型研究

网络入侵诊断直接影响网络正常运行和安全.针对入侵类型复杂,现有分类诊断模型精度有限的问题,提出一种基于邻域粗糙集的网络入侵分类诊断优化模型.首先,运用邻域粗糙集对网络入侵数据进行条件属性的约简,确定关键属性,然后将其作为训练输入构建相关向量机分类诊断模型,并同时运用遗传算法进行超参数优化,提高模型诊断精度和速度.通过KDDCup99数据集对优化模型性能进行检验,结果表明,组合预测方法精确度高于支持向量机、相关向量机和BP神经网络.组合模型诊断精度高、速度快,具有优异的综合性能.

Acombined Model Study of Taxonomic Diagnosis of Network Intrusion Based on Neighborhood Rough Set

Network intrusion diagnosis has a direct impact on the routine operation and security of network.To deal with the problems of complicated intrusion categories and limited accuracy of taxonomic diagnosis of existing models,we propose an optimized taxonomic diagnosis model of network intrusionbased on neighborhood rough set (NRS).First,we use NRS to make a conditional attribute reduction of network intrusion data and determine the key attributes.Then,it acts as the training inputs to build the taxonomic diagnosis model with relevance vector machine (RVM).Meanwhile,genetic algorithm is adopted to make a hyperparameter optimization and enhance the accuracy and speed of model diagnosis.Performance test of the optimized model is carried out with KDDCup99 data sets.Results have revealed that,the combined prediction method enjoys higher accuracy than support vector machine (SVM),RVM and BP neural network.The combined model has a higher accuracy and fast speed,which enjoys excellent comprehensive performance.