集成基于EP的分类器用于数据流入侵检测

Integrating EP-based Classifiers for Intrusion Detection On Data Streams

本文提出了一种集成基于EP的分类器用于数据流入侵检测的模型EEPCDS(Ensemble of EP-based Classifiers on Data Stream)。该模型选择滑动窗口中的多个时间段数据来生成多个EP分类器,并且通过加权投票表决对未知样本进行分类,检测入侵行为。EEPCDS能适应数据流环境下的概念漂移,并且能实现较好的目标类召回率和精度的平衡,以及较高的分类准确率。

This paper proposed a new approach,called EEPCDS(Ensemble of EP-based Classifiers on Data Stream),for intrusion detection on Data Stream.EEPCDS constructs fixed number of EP-based Classifiers from different chunks,and integrated these classifiers for intrusion detection.EEPCDS not only has a high accuracy,but also achieves a good balance of recall and precision.