摘要
提出了建立在概率典型性和聚类排斥基础上的一个新型无噪声模糊聚类方法RTCM,给出了它的迭代算法过程,并验证了它的收敛性.首先引述了一般的聚类方法,它们主要分为两种:噪声聚类,如模糊c均值(FCM)、可能模糊c均值(FPCM);无噪声聚类,如NC、PCM等,然后给出了RTCM算法模型和过程,并验证了它的局部收敛性.该算法解决噪声环境下的数据聚类问题,避免了重叠聚类.对比试验表明,该算法改善了噪声环境下FCM,NC、PCM、FPCM的聚类中心质量,有效地解决了PCM在近邻聚类数据中的聚类重叠问题.
This paper presents a new noise-resistant fuzzy clustering approach based on probabilistic typicalities and cluster repulsion, denoted as RTCM, proposes a new iteration algorithm, furthermore, validates its converge. Firstly, some existed fuzzy clustering methods are introduced, which commonly include two types： noise clustering, such as FCM, FPCM, and noise-resistant fuzzy clustering, such as NC, PCM, and then the objective function and iteration algorithm are proposed, the proof is given for its converge properties. Experimental results show that our algorithm improves the quality of clustering centers of data sets including noise comparing to FCM, NC, PCM, FPCM, solves the coincident clusters in PCM.
出处
《小型微型计算机系统》
CSCD
北大核心
2005年第9期1536-1539,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金(60272019和60321002)资助.
关键词
模糊聚类
概率典型性
聚类排斥
无噪声模糊聚类
fuzzy clustering
probabilistic typicalities
cluster repulsion
noise-resistant fuzzy clustering
