计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (34): 158-160.

• 数据库、信号与信息处理 • 上一篇    下一篇

FCM迭代增强与划分混合聚类算法

陈 磊1,牛秦洲2   

  1. 1.马鞍山钢铁股份有限公司 第四钢轧总厂,安徽 马鞍山 243000
    2.桂林理工大学 信息科学与工程学院,广西 桂林 541004
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-12-01 发布日期:2011-12-01

Hybrid iterate boosting and space portioning clustering based FCM algorithm

CHEN Lei1,NIU Qinzhou2   

  1. 1.No.4 Steel Making & Rolling Plant of Maanshan Iron & Steel Co.,Ltd.,Maanshan,Anhui 243000,China
    2.Guilin University of Technology,College of Information Science and Engineering,Guilin,Guangxi 541004,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-12-01 Published:2011-12-01

摘要: 提出了一种以迭代增强和空间划分为基础的模糊C均值聚类方法,利用弱学习理论在每次迭代之后将产生的训练集合重新归并,在原有划分集的基础上通过分布质量权重选举方法更新产生最优假设划分集,克服了传统的简单重复训练方法的聚类效果不理想的缺点。通过形状分类实验和聚类量化指标对比,证明了该方法具有分类质量高、形状分割彻底的优点。

关键词: 迭代增强, 聚类, 空间划分, 弱学习

Abstract: A hybrid Iterate boosting and space portioning clustering based FCM algorithm is proposed.Under the foundation of original portion set,the iterated optimize cluster hypothesis updates with the fraction of distribution weight voting,making use of the weak learning exoteric to remerge with the training set after every iterate process.Compared with the traditional simple repeat training clustering method,a disadvantage has been got over.It confirms that the algorithm has high quality of classification and entirely shape portioning effect in the shape classification experiment and quantity index contrast.

Key words: iterate boosting, clustering, space portioning, weak learning