融合特征值与优化划分的改进FCM聚类算法  

Improved FCM clustering algorithm fusing eigenvalues and optimal partition

作　　者：武玉坤[1] 

机构地区：[1]浙江邮电职业技术学院管理与信息学院,浙江绍兴312016

出　　处：《计算机工程与设计》2017年第4期1076-1080,共5页Computer Engineering and Design

基　　金：国家自然科学基金青年科学基金项目(61402053);全国教育信息技术规划课题基金项目(126240629)

摘　　要：通过研究传统FCM和改进FCM算法,针对现有算法中存在收敛速度不够、聚类精度不高等问题,提出一种基于FCM算法的融合特征值和优化划分FCM的改进聚类算法。融合优化划分与特征值,获取特征值向量迭代计算公式,实现自适应更新。实验结果表明,该方法能够有效计算并获取样本特征值向量,改进算法聚类分析效果优于FCM、W-FCM、OP-FCM算法,所提改进算法具有实际应用价值。Through researching the traditional FCM and improved FCM algorithm, aiming at the problems including low conver- gence speed and clustering accuracy in the existing algorithms, a clustering method based on FCM algorithm and fusing feature weighted and improved partition fuzzy C-means was proposed. Fusing feature weighted and improved partition, the feature value vector was ohtained to iterate formula and adaptively updating was then achieved. Experimental results show that the clustering method can effectively calculate and get the sample feature value vector. The improved algorithm has better clustering effects than FCM, weighted FCM and optimal partition FCM algorithm, and the algorithm has application value.

关 键 词：模糊聚类算法 特征值 优化划分 聚类精度 迭代计算公式 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]