一种面向数据流贝叶斯分类的显露模式挖掘方法  

Mining Emerging Patterns for Bayesian Classifier of Data Stream

作　　者：李志杰 刘基旺 廖旭红 江华 Li ZhiJie;Liu Jiwang;Liao Xuhong;Jiang Hua(School of Information Science and Engineering,Hunan Institute of Science and Technology,Yueyang 414006)

机构地区：[1]湖南理工学院信息科学与工程学院,岳阳414006

出　　处：《现代计算机》2021年第32期38-41,64,共5页Modern Computer

基　　金：湖南省自然科学基金面上项目(2019JJ40111)。

摘　　要：显露模式考虑模式在目标类与对立类数据集合中的支持度,首先必须是频繁模式,IncMine是著名的数据流频繁闭合模式挖掘算法。然而,现有的频繁模式挖掘算法面向事务数据流,项集没有类标约束,不能进一步挖掘显露模式用作数据流分类。本文提出一种基于IncMine的显露模式挖掘方法(emerging patterns based on incMine,EPBIM)。EPBIM改进IncMine算法,挖掘带类值约束的频繁闭合项集,并获取显露模式用于贝叶斯分类。在MOA平台上运行多个真实与模拟数据流,实验结果验证了该方法的有效性。Emerging patterns firstly must be frequent patterns, which consider the support of patterns in the target class and opposite class data sets. IncMine is a famous frequent closed pattern mining algorithm. However, the existing frequent pattern mining algorithms are transaction oriented. Frequent itemsets do not have class labels and can not further mine emerging patterns for data stream classification. This paper proposes mining emerging patterns based on IncMine(EPBIM). EPBIM mine frequent closed itemsets with class value constraints by improved IncMine, and obtain emerging patterns for Bayesian classification. Several real and simulated data streams are run on the MOA platform, and the experimental results verify the effectiveness of the method.

关 键 词：数据流分类 贝叶斯 IncMine算法 频繁闭合模式 显露模式 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] TP181[自动化与计算机技术—计算机科学与技术]