摘要
提出了一类基于标签传递的半监督模糊聚类模型,得到了其隶属度和聚类中心的、具有简洁形式的迭代求解公式.设计了一种算法将已知的类别标签传递给未标签数据,这些类别标签可以合理、有效地作用于整个数据集,从而增加了标签数据的作用.在人工数据集、乳腺癌数据集以及黄瓜数据集上的实验验证了该聚类方法的有效性.
A new type of semi-supervised fuzzy clustering algorithm based on label propagation is proposed, and the iterative solutions of membership degrees and cluster centers are given in con- cise forms. By propagating labels to unlabeled data, the importance of labeled data is strengthed, and the size of labeled data is enlarged in the clustering process. The effetiveness of the proposed al- gorithm is verified thought experiments on an artificial dataset, bread cancer datasets and a cucumber dataset.
作者
邓朝阳
杨昔阳
李志伟
黄东兰
DENG Chao-yang YANG Xi-yang LI Zhi-wei HUANG Dong-lan(Department of Basic Commonality Curriculum, Quanzhou Medical College, Quanzhou 362000, China Key Laboratory of Intelligent Computing and Information Processing, Quanzhou Normal University, Quanzhou 362000, China)
出处
《福建师范大学学报（自然科学版）》
CAS
CSCD
北大核心
2016年第6期25-33,共9页
Journal of Fujian Normal University：Natural Science Edition
基金
福建省教育厅科技项目(JK2013037)
福建省中青年教师教育项目(JA14262)
关键词
半监督算法
标签传递
模糊聚类
semi-supervised algorithm
label propagation
fuzzy clustering
