基于k近邻密度峰值聚类混合算法的网络入侵检测

Network Intrusion Detection Based on K Nearest Neighbor Density Peak Clustering Hybrid Algorithm

由于传统的入侵检测系统无法识别新型网络入侵问题,在k近邻(KNN)算法和密度峰值聚类(DPC)算法的基础上,提出了一种基于k近邻的密度峰值聚类混合学习算法(DPNN),将DPC用于训练,KNN用于分类,结合KDD-CUP 99数据集作为入侵检测中的标准数据集,并利用DPNN在入侵检测中找到更准确和高效的分类器。实验结果表明,DPNN优于支持向量机(SVM)、k近邻(KNN)等多种机器学习方法,它能够有效地检测入侵攻击并具有良好的性能。

Because of the traditional intrusion detection system can not identify the new network intrusion problem, based on the k nearest neighbor algorithm(KNN) and the density peak clustering(DPC), a density peak clustering hybrid learning model(DPNN) based on the nearest neighbor of k is proposed. DPC is used for training, KNN is used for classification, and KDD-CUP 99 data sets are used as intrusion detection. The standard dataset is detected, and DPNN is used to find more accurate and efficient classifier in intrusion detection. The experimental results show that DPNN is better than support vector machine(SVM), k nearest neighbor(KNN) and other machine learning methods. It can effectively detect intrusion attacks and have good performance.