基于非欧式距离的模糊核聚类算法在入侵检测的应用

摘要

模糊核C‐均值聚类（KFCM ）的主要思想是在模糊C‐均值聚类（FCM ）中引入核函数，样本点被非线性变换映射到高维特征空间进行聚类，解决了高维数据空间的聚类问题。同经典的 FCM 算法及其派生算法一样，KFCM 算法对噪声或野值数据敏感。论文在KFCM基础上，利用鲁棒统计观点对目标函数进行改进，通过引入非欧式距离度量代替欧氏距离度量，提高其对噪声或野值数据的抗干扰能力。将该算法用于构建入侵检测系统模型并通过模拟仿真实验表明，改进算法有效解决了传统的聚类算法在入侵检测中稳定性差，检测准确率低的问题。%Based on FCM and by introducing the kernel function ,Kernel fuzzy C‐means clustering (KFCM ) make the sample points nonlinear mapped to a high‐dimensional feature space for clustering which can solve the problem of high dimen‐sional data space clustering .Like the classic FCM clustering algorithm and its Derived algorithm ,KFCM clustering algorithm is sensitive to noises or outliers .Based on modified objective function by using the robust statistical view ,a new non‐Euclide‐an distance is introduced to replace the Euclidean distance which can improve anti‐jamming capability of noise or outliers data . The improved algorithm is used to construct the model of intrusion detection system .Our experimental results show the pro‐posed algorithm can solve poor stability and low detection accuracy of the traditional clustering algorithms in intrusion detec‐tion .