基于聚类的未标识数据的入侵检测

Intrusion Detection Based on Clustering and Unlabeled Data

自动化入侵检测是入侵检测的重要研究方向。传统的入侵检测由于依赖标识数据进行训练,不能做到自动更新规则库和检测新的入侵。提出一种自动检测入侵的方法———基于聚类(C lustering)的未标识数据的检测。它不依赖分类标识数据进行训练,能检测到未知的入侵而保持着很低的误报率。

Automatical Intrusion Detection System is becoming more and more important in the area of Intrusion Detection System（IDS）. Traditional IDS＇s which rely on labeled datas to train , can＇t update the rules and detect intrusions automatically. This paper presents a frame work for automatically detecting intrusions：intrusion detection based on clustering and unlabeled data. It doesn＇t rely on labeled datas to train and can detect the new intrusions keeping low false positive rate.