计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (20): 144-146.DOI: 10.3778/j.issn.1002-8331.2009.20.043

• 数据库、信息处理 • 上一篇    下一篇

FCM算法的改进及仿真实验研究

吕晓燕1,2,罗立民2,李祥生1   

  1. 1.山西医科大学 计算中心,太原 030001
    2.东南大学 计算机科学与工程系,南京 210096
  • 收稿日期:2009-05-11 修回日期:2009-06-19 出版日期:2009-07-11 发布日期:2009-07-11

Impovement on FCM algorithm and simulation research

LV Xiao-yan 1,2,LUO Li-min2,LI Xiang-sheng1   

  1. 1.Computing Center,Shanxi Medical University,Taiyuan 030001,China
    2.Department of Computer Science and Engineering,Southeast University,Nanjing 210096,China
  • Received:2009-05-11 Revised:2009-06-19 Online:2009-07-11 Published:2009-07-11

摘要:

针对FCM原型算法的不足,提出一种新的改进方法,并进行仿真实验研究。利用主成分分析方法对原始数据集的指标进行筛选,应用Relief算法对入选指标计算权重。在此基础上,对FCM算法进行了改进。应用模糊划分系数Fc(R)和平均模糊熵Hc(R)这两个指标对算法的性能进行了评价。仿真实验结果表明,改进后的FCM算法对样本集数据的分类符合率达到了91.5%,其模糊划分系数Fc(R)和平均模糊熵Hc(R)分别为0.924和-0.062。改进后的FCM算法分类性能优于FCM原型算法,在应用中可以取得更好的效果。

关键词: 模糊C均值算法, 主成分分析, Relief算法, 模糊划分系数平均模糊熵

Abstract: To improve the shortcomings of original FCM algorithm,an improved algorithm is proposed.This paper uses principal component analysis to select the features from the original data set,and uses Relief algorithms to calculate the weights of the choosed feathures.Fuzzy division factor Fc(R) and the average fuzzy entropy Hc(R) are used to evaluate the performance of improved FCM algorithm.The improved FCM algorithm is applied for the classification of sample set of data,it can get the accuracy of 91.5%,fuzzy partition coefficient (Fc(R)) and the average fuzzy entropy (Hc(R)) are 0.924 and -0.062.The improved FCM algorithm is more efficient than the original FCM algorithm,the application can be made more effective.

Key words: Fuzzy C-Means(FCM) algorithm, principal component analysis, Relief algorithm, fuzzy partition coefficient, average fuzzy entropy