Improved k-means algorithm based on optimizing initial cluster centers

作     者：Wu, Di Zhang, Yongjing Yang, Fengqin Cheng, Siying Sun, Hongguang Sun, Tieli 

作者机构：School of Computer Science and Information Technology Northeast Normal University No. 2555 Jingyue Avenue Changchun 130117 China Key Laboratory of Intelligent Information Processing of Jilin Universities Northeast Normal University No. 2555 Jingyue Avenue Changchun 130117 China 

出 版 物：《ICIC Express Letters》 (ICIC Express Lett.)

年 卷 期：2013年第7卷第3 B期

页      面：991-996页

核心收录：

主　　题：Clustering algorithms 

摘      要：The k-means algorithm is known to be a fast clustering algorithm. However, it is sensitive to initial cluster centers and the number K must be pre-set. To solve the problem, the k-means algorithm based on Max-Min distance algorithm has been introduced, but it cannot handle noise data. In this paper, we present the improved Unweighted Pair Group Method with Arithmetic Mean (UPGMA) algorithm with Max-Min distance method (IU-M) k-means algorithm, data sampling and the improved UPGMA algorithm are used before the Max-Min distance algorithm to get favorable initial cluster centers, and the number K can be determined intelligently. Comparative experiments are done on three benchmark datasets. The results demonstrate that the IU-M k-means algorithm is efficient and effective. © 2013 ICIC International.