聚类分析的竞争决策算法  

Competitive Decision Algorithm for Clustering Analysis

作　　者：刘勇[1,2] 宁爱兵[1] 许秋艳[3] 马良[1] 

机构地区：[1]上海理工大学管理学院,上海200093 [2]盐城工学院基础教学部,江苏盐城224051 [3]盐城工学院信息工程学院,江苏盐城224051

出　　处：《数学的实践与认识》2008年第21期58-62,共5页Mathematics in Practice and Theory

基　　金：国家自然科学基金项目资助(70871081);上海市重点学科建设项目资助(T0502);上海市高校选拔培养优秀青年教师科研专项基金项目资助(21012)

摘　　要：聚类分析是数据挖掘的重要技术,是一种无监督的学习方式,可根据数据间的相似程度,将数据进行分类.竞争决策算法是一种基于竞争造就优化和决策左右结果的新型优化算法,针对聚类分析的特点,设计了一种竞争决策算法进行求解,经实验测试和验证,并与其它算法的结果进行比较,获得了较好的结果.Clustering analysis is an important technique in data mining. It is an unsupervised learning technique and is a division of data into groups of similar objects. A competitive decision algorithm is a new optimization algorithm based on the characteristics that competition builds optimization and the result of competition hinges on decision. Based on some characteristics of clustering analysis, this paper provides a competitive decision algorithm for clustering analysis. The experiments are tested and the results show that the clustering result of competitive decision algorithm is better than that of some other algorithms.

关 键 词：聚类分析 竞争力函数 决策函数 竞争决策均衡 竞争决策算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]