摘要： 基于孤立点挖掘，探讨了异常检测的可行性，将孤立点挖掘方法应用到入侵检测中，采用编码映射方法对符号型数据进行处理，利用主成分分析对编码映射后扩展的属性进行降维。仿真实验验证了该方法的可行性。

关键词: 入侵检测, 孤立点, 相似系数, 编码映射, 主成分分析

Abstract: This paper discusses feasibility of anomaly detection based on outlier mining, presents an anomaly detection method by using an outlier detection method based on similar coefficient sum. It transforms the character feature into numerical value by code mapping, and uses Principal Components Analysis(PCA) to reduce dimension. Implementation method is proposed, feasibility of the method is proved.

Key words: intrusion detection, outlier, similar coefficient, code mapping, Principal Components Analysis(PCA)

中图分类号: 

• TP391