一种新的数据流模糊聚类方法.pdfVIP

第37 卷第7 期 电 子 与 信 息 学 报 Vol.37 No.7 2015 年7 月 Journal of Electronics Information Technology Jul. 2015 一种新的数据流模糊聚类方法 孙力娟*①② 陈小东① 韩 崇① 郭 剑①② ①(南京邮电大学计算机学院 南京 210003) ②(南京邮电大学江苏省无线传感网高技术研究重点实验室 南京 210003) 摘 要：针对数据流上的聚类任务受到时间、空间限制等问题，该文提出一种基于权值衰减的数据流模糊微簇聚类 算法(WDSMC)。该算法使用改进的带权值的模糊C 均值算法进行处理，并采用微簇结构和权值时间衰减结构提高 聚类质量。实验表明，相对于现有的数据流加权模糊 C 均值聚类(SWFCM)算法和 StreamKM++算法而言， WDSMC 算法具有更好的聚类精度。 关键词：数据挖掘；数据流；模糊C 均值聚类；权值衰减；微簇聚类 中图分类号： TP301 文献标识码：A 文章编号：1009-5896(2015)07-1620-06 DOI: 10.11999/JEIT141415 New Fuzzy-Clustering Algorithm for Data Stream ①② ① ① ①② Sun Li-juan Chen Xiao-dong Han Chong Guo Jian ① (College of Computer, Nanjing University of Posts and Telecommunications, Nanjing 210003, China) ② (Jiangsu High Technology Research Key Laboratory for Wireless Sensor Networks, Nanjing University of Posts and Telecommunications, Nanjing 210003, China) Abstract: There is a great challenge in the data stream clustering due to a limitation of time and space. In order to solve this problem, a new fuzzy-clustering algorithm, called Weight Decay Streaming Micro Clustering (WDSMC), is presented in this paper. The algorithm uses a reformed weighted Fuzzy C-Means (FCM) algorithm, and improves the quality of clustering by the structures of micro-clusters and weight-decay. Experimental results show that this algorithm has better accuracy than Stream Weight Fuzzy C-Means (SWFCM) and StreamKM++ algorithm. Key words: Data mining; Data stream; Fuzzy C-Means (FCM); Weight decay; Micro-clustering 1 引言 弃；(5