不确定数据基于密度的局部异常点检测

发布时间：2018-01-08 22:08

本文关键词：不确定数据基于密度的局部异常点检测　出处：《计算机学报》2017年10期 　论文类型：期刊论文

【摘要】：不确定数据作为一种新型的数据模型,被广泛应用于金融、基于位置的服务、移动物体监测、传感器网络等许多类型应用领域.近年来出现的面向不确定数据的分析处理技术已成为数据库、数据挖掘等领域的研究热点.许多传统的数据挖掘技术已经被扩展并应用到不确定数据的分析和管理,异常点检测是数据挖掘领域重要的技术,用来发现行为或特征不同于其他对象的数据对象.当数据对象的性质和行为明显区别于它的近邻时,则被为视为异常点.异常点检测在许多方面有着广泛的应用,如网络入侵检测、信用卡诈骗、环境监测等.该文研究不确定数据基于密度的局部异常点检测,每个不确定数据由几个离散的可能实例组成.首先,提出了基于特定不确定数据模型的局部异常点定义.为了能够快速地检测局部异常点,在不展开可能世界的前提下,提出了基础算法UDOL(Uncertain Density-based Local Outlier).然后,又提出在不精确计算概率的情况下,通过估计局部异常点因子的检测算法PUDOL(Pruning on Uncertain Density-based Local Outlier),可以有效地减少计算量.最后,通过大量的实验验证该文提出算法的性能.实验结果证明,该文所提出的算法是解决不确定数据基于密度的局部异常点检测的有效方法.
[Abstract]:As a new data model, uncertain data is widely used in finance, location-based services, mobile object monitoring. Sensor networks and many other types of applications. In recent years, uncertain data oriented analysis and processing technology has become a database. Many traditional data mining techniques have been extended and applied to the analysis and management of uncertain data. Anomaly point detection is an important technology in the field of data mining. Used to discover data objects whose behavior or characteristics are different from those of other objects. When the nature and behavior of a data object are significantly different from its immediate neighbors. Anomaly detection has been widely used in many fields, such as network intrusion detection, credit card fraud, environmental monitoring, etc. Each uncertain data consists of several discrete possible instances. Firstly, the definition of local outliers based on specific uncertain data model is proposed to detect local outliers quickly. Without expanding the possible world, the basic algorithm UDOL(Uncertain Density-based Local Outlier.Then. It is also proposed that the probability is calculated inaccurately. A detection algorithm PUDOL by estimating local outliers. Pruning on Uncertain Density-based Local Outlier. Finally, the performance of the proposed algorithm is verified by a large number of experiments. The algorithm proposed in this paper is an effective method for the detection of local outliers based on density in uncertain data.
【作者单位】：沈阳建筑大学信息与控制工程学院;沈阳航空航天大学计算机学院;
【基金】：国家自然科学基金(611602323) 中国博士后科学基金(2016M591455) 辽宁省博士启动基金(201601209)资助~~
【分类号】：TP311.13
【正文快照】： 引近年来,随着人们对数据采集和处理技术理解的不断深入,不确定数据得到了广泛的重视.在许多现实的应用中,普遍存在着不确定数据[1],例如基于位置的服务、传感器网络、射频识别等领域.由于不确定数据中存在着不确定性,传统的数据分析技术无法有效地处理不确定数据,这就引发

【相似文献】