江汉大学学报(自然科学版) ›› 2014, Vol. 42 ›› Issue (5): 36-39.

• • 上一篇    下一篇

基于万有引力的簇间距离表示法

韩 海   

  1. 江汉大学 数学与计算机科学学院, 湖北 武汉 430056
  • 收稿日期:2014-12-29 修回日期:2014-12-29 出版日期:2014-10-25 发布日期:2014-11-11
  • 作者简介:韩 海 (1968—) , 男, 副教授, 研究方向: 图形图像处理及模式识别。

Description of Distance Between Clusters Based on Gravitation

HAN Hai   

  1. School of Mathematics and Computer Science, Jianghan University, Wuhan 430056, Hubei, China
  • Received:2014-12-29 Revised:2014-12-29 Online:2014-10-25 Published:2014-11-11

摘要: 分析了几种现有聚类算法中簇间距离表示法的优缺点, 并在此基础上提出了一种基于万有引力模型的簇间距离计算方法。该方法模仿物理学中力的合成原理,是对把各质点间引力数值直接相加的重要改进。理论分析及数据计算的结果都表明, 该方法比引力直接相加有更广的适应性。

关键词: 聚类, 簇间距离, 引力, 积分

Abstract: :Analyses the advantages and disadvantages of several existing methods for description of distance between clusters,based on it,presents a calculation method for distance between clusters based on gravitation model. This method is an important improvement for direct addition of gravitation value between each particle, which simulates the synthetic principle of force in physics. Theoretical analysis and computing results show the presented method is more applicable than direct addition of gravitation.

Key words: clustering, distance between clusters, gravitation, integral

中图分类号: