改进的模糊C均值的增量聚类算法  

Improved Fuzzy C-Means incremental clustering algorithm

作　　者：吴佳[1] 罗可[1] 

机构地区：[1]长沙理工大学计算机与通信工程学院,长沙410076

出　　处：《计算机工程与应用》2011年第23期141-142,207,共3页Computer Engineering and Applications

基　　金：国家自然科学基金No.10926189;No.10871031;湖南省科技计划项目(No.2008FJ3015)~~

摘　　要：针对FCM算法的缺点,提出了一种基于改进的FCM的增量式聚类方法。该算法首先对模糊C均值算法进行加权,并将权系数归一化,然后将改进的算法与增量式聚类算法结合。改进的方法既提高了FCM算法的性能,避免了FCM算法的缺陷,并能够实现增量式聚类,避免了大量的重复计算,并且不受孤立点的影响。实验表明该算法的有效性。A new incremental clustering algorithm is proposed which is based on improved FCM for the shortcomings of FCM algorithm.The algorithm weights FCM which the weights will be normalized,then combines the improved algorithm and the incremental clustering algorithm.Improved method not only improves the performance and avoids the shortcomings of the FCM algorithm,but also be able to achieve incremental clustering,it avoids a lot of double counting and unaffected of outlier.Experiments show that the new algorithm is effective.

关 键 词：聚类分析 模糊C均值算法 增量式聚类 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]