基于人工蜂群算法的手写数字聚类研究.docVIP

Optoelectronics光电子, 2011, 1, 33-38 /10.12677/oe.2011.12007 Published Online December 2011 (/journal/oe) The Research of Handwritten Digit Cluster Based on Artificial Bees Colony Algorithm* Guangbiao Wang , Shuying Yang , Fan Feng , Bokai Wang , Zijuan Jia , Guang Zhu 1 1 1 1 1 2 1 School of Computer and Communications Engineering, Tian Jin University of Technology, Tianjin Radio Film and TV Administration of Shan Dong Province, Shandong Email: guangbiaowang@126.com 2 Received: Sep. 23rd, 2011; revised: Oct. 11th, 2011; accepted: Nov. 3rd, 2011. Abstract: Handwritten digit cluster is an important study of pattern recognition, because of the application of traditional evolutionary algorithm for clustering analysis of handwritten digit has much more problems, such as slow convergence, easily fall into local optimization and so on. To overcome those problems, we present a novel approach to solve the problem of digital clustering by using artificial bees colony algorithm, and propose 3 kinds of operators for bees’ location of updating, and establish a dynamic update of 3 operators in formulas. Finally we elaborate the concrete steps of handwriting digital cluster by using this approach. We do the simulation experiments with some typical handwritten digital instances. Experiments show that our approach can make a good implementation of handwriting digital cluster, overcome the phenomenon of pre- mature convergence, and accelerate the convergence rate in a way. Keywords: Artificial Bees Colony Algorithm; Handwritten Digit Cluster; Combinatorial Optimization 基于人工蜂群算法的手写数字聚类研究* 王光彪，杨淑莹，冯帆，王博凯，贾紫娟，朱光 2 1 1 1 1 1 1 天津理工大学，计算机与通信工程学院，天津 2 山东省广播电影电视局，山东 Email: guangbiaowang@126.com 收稿日期：2011年 9月 23日；修回日期：2011年 10月 11日；录用日期：2011年 11月