计算机应用与软件2017，Vol.34Issue(12)：256-259,4.DOI:10.3969/j.issn.1000-386x.2017.12.049
基于相对熵的数据流概念漂移检测算法
A DATA FLOW CONCEPTUAL DRIFT DETECTION ALGORITHM BASED ON RELATIVE ENTROPY
摘要
Abstract
Aiming at the problem of concept drift in data stream,this paper proposed a conceptual drift detection algorithm based on relative entropy based on decision tree as a classifier.The proposed algorithm combined the accuracy and relative entropy of the classifier as a criterion for judging whether the data block was drilled or not.The method was verified by 5 data sets.The algorithm obtained the optimal result on the four data sets,and the suboptimal result was obtained on the other data set.The experimental results showed that this method not only detected the occurrence of concept drift effectively,but also improved the accuracy of the classifier.关键词
数据流/概念漂移/相对熵/决策树Key words
Data stream/Concept drift/Relative entropy/Decision tree分类
信息技术与安全科学引用本文复制引用
杨帆,张永..基于相对熵的数据流概念漂移检测算法[J].计算机应用与软件,2017,34(12):256-259,4.基金项目
国家自然科学基金面上项目(61373127). （61373127）
