基于局部异常因子的无监督学习集成异常检测  

Ensemble anomaly detection with unsupervised learning method based on local outlier factor algorithm

作　　者：周绪川[1] 蔡利平[1] 

机构地区：[1]西南民族大学计算机科学与技术学院,成都610041

出　　处：《西南民族大学学报（自然科学版）》2012年第3期444-448,共5页Journal of Southwest Minzu University(Natural Science Edition)

基　　金：西南民族大学中央高校基本科研业务费专项资金项目(11NZYBS09;12NZYQN27);西南民族大学自然科学基金项目(10NYB003)

摘　　要：研究了基于局部异常因子(LOF)的无监督学习模型共享的集成学习异常检测方法,首先在局部采用LOF无监督学习得到检测模型,然后通过交换局部模型的方式实现集成异常检测.该方法能够从检测数据中自动发现异常样本,无需预先了解数据的分布特征,不对数据进行任何假设,适用范围广.方法通过交换检测模型实现数据有效信息的共享,相比集中式方法,减少了网络传输耗费.实验仿真表明,该方法能取得优于或和集中模型相当的检测性能.Detecting anomalous behavior from terabytes of collected record data has emerged as a crucial component for many systems for Data Mining System. Very often, processing record data collected from various locations or providers cannot be directly aggregated for anomaly analysis due to the proprietary nature of the data. Unsupervised learning can get anomaly samples without any hypothesis and knowing the distribute features of the data set. This paper proposes a novel general framework for anomaly detection from distributed data sources based on local outlier factor （LOF）. In the proposed method, LOF algorithm is first applied to data from individual provider and then their unsupervised learning detection models are combined. The experiments performed on simulated data show that particular anomaly detection algorithms and combining methods are more suitable for the task of distributed anomaly detection than others.

关 键 词：局部异常因子 无监督学习 集成学习 异常检测 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]