Computer Engineering and Applications ›› 2011, Vol. 47 ›› Issue (24): 136-138.

• 图形、图像、模式识别 • Previous Articles     Next Articles

Semi-supervised mercer-kernel based fuzzy clustering algorithm with pairwise constraints and attribute weighted

HE Yangcheng,WANG Shitong,JIANG Nan   

  1. School of Digital Media,Jiangnan University,Wuxi,Jiangsu 214122,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-08-21 Published:2011-08-21

成对约束的属性加权半监督模糊核聚类算法

贺杨成,王士同,江 南   

  1. 江南大学 数字媒体学院,江苏 无锡 214122

Abstract: Clustering with constraints is an active area in machine learning and data mining.A semi-supervised mercer-kernel based fuzzy clustering algorithm with pairwise constraints and attributes weighted is proposed which incorporates both semi-
supervised learning technique and the kernel method into the traditional fuzzy clustering algorithm.The proposed algorithm performs clustering in high feature space mapped by a mercer kernels and considers the imbalance between the attributes fully.This algorithm experimentally outperforms a similar semi-supervised fuzzy clustering approach,i.e Pairwise Constrained Competitive Agglomeration(PCCA).

Key words: semi-supervised clustering, pairwise constraints, kernel, fuzzy clustering

摘要: 在机器学习和数据挖掘中,带约束的半监督聚类是一个活跃的研究领域。为了利用约束条件获得表现更优异的聚类效果,提出了一种成对约束的属性加权半监督聚类算法,该方法充分考虑了属性间的不平衡性,在传统模糊聚类算法中融合半监督学习机制并通过Mercer核把原始的观察空间映射到高维特征空间。实验结果表明,该算法优于相似的成对约束的竞争群算法(PCCA)。

关键词: 半监督聚类, 成对约束, 核, 模糊聚类