基于聚类的LS-SVM的入侵检测方法研究

本文针对最小二乘法支持向量机在入侵检测中的训练效率低下的缺点,将聚类方法应用其中。该方法主要用来对数据集进行剪枝,有效地减少距离分类面较远的数据集合数量,而使用靠近聚类中心的数据集合作为有效的样本集合,减少样本的训练时间,提高训练效率。实验表明,使用聚类方法提高了最小二乘法支持向量机的训练效率,而且对入侵检测有很好的效果。

Research of The Intrusion Detection Method Based On Clustering LS-SVM

For the shortcomings of training inefficient of Intrusion Detection based on least squares support vector machine, we use clustering. This method can realize data pruning and effectively reduce the amount of data set by using the cluster center set close to the distinguishing boundary, which may reduce the samples of the training time and improve the training efficiency. The experimental result shows that using clustering can improve the efficiency of LS-SVM, and it is effective for intrusion detection