一种半聚类的异常入侵检测算法

针对基于监督学习的入侵检测算法所面临的训练样本不足的问题，提出了一种结合改进k 近邻法的基于半监督聚类的异常入侵检测算法，利用少量的标记数据改善算法的学习能力，并实现了对新攻击类型的检测。实验结果表明，在标记数据极少的情况下，算法的检测结果明显好于非监督学习的算法，接近于监督学习的检测算法。

An anomaly intrusion detection algorithm based on semi-supervised clustering

An anomaly intrusion detection algorithm based on semi-supervised clustering along with k-nearest neighbor was presented. It could solve the problem of the insufficiency of training samples that the intrusion detection algorithms based on supervised learning face. The algorithm exploited minimal labeled data to improve its learning capability, and novelty detection could also be carried out. The experiment results manifest that the detection results of the algorithm precedes the one based on unsupervised learning remarkably, and approaches the one based on supervised learning while the labeled data are few.