分布式数据流上的连续异常检测

近年来,数据流异常检测在决策支持和监测等领域有着广泛的应用前景,并成为数据管理与挖掘的研究热点.针对该问题提出了相应的异常定义及检测算法,理论分析表明:与现有异常检测算法相比较,提出的算法具有良好的性能和效率,更适合于数据流应用.

Continuous Outlier Detection over Distributed Data Streams

In recent years,outlier detection over data stream has been becoming a hot research topic in data management and mining field,due to its potential applications in decision-making and monitoring applications.Stream data often have these characteristics: distributed,high speed,huge amount and only "one-scan",while people more care about obtaining the outliers quickly and progressively.In this paper,firstly,we propose a novel outlier definition and corresponding algorithm.Furthermore,theoretical analysis shows that the proposed method is both efficient and effective compared with existing outlier detection algorithms,and more suitable for data stream.