一种核模糊聚类的改进及其在育肥猪出栏中的应用  

AN IMPROVEMENT ON KERNEL FUZZY CLUSTERING ALGORITHM AND ITS APPLICATION IN FINISHING PIG SLAUGHTER

作　　者：王越[1] 黄靖华[1] 张凯文[1] 

机构地区：[1]重庆理工大学计算机科学与工程学院

出　　处：《计算机应用与软件》2013年第1期120-125,共6页Computer Applications and Software

基　　金：重庆市科委攻关项目(2010GGB097)

摘　　要：核模糊聚类算法不适用于含孤立点与噪声点的数据,并且对初始化中心敏感。针对此种情况,结合减法聚类,对样本加权,放宽隶属度归一化条件,提出基于减法聚类的加权核模糊聚类。通过IRIS和WINE数据集证实改进算法比传统的核聚类算法具有更高的健壮性与抗噪性,并将改进后的算法运用在育肥猪出栏中,验证了算法的实用性与可行性。Kernel fuzzy clustering algorithm does not fit the data with outliers and noise points, and is sensitive to initialisation centre. Ac- cording to the above, combining the subtractive clustering, weighting on the sample and relaxing the membership norrnalisation conditions, a weighted kernel fuzzy clustering algorithm based on subtractive clustering is proposed. It is proved by IRIS and WINE data set that the im- proved algorithm has better robustness and noise restraint than the traditional kernel clustering algorithm. The improved algorithm has been applied to finishing pig slaughter. Its practicality and the feasibility are validated.

关 键 词：核模糊聚类 减法聚类 加权 育肥猪出栏 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]