• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学

• 论文 • 上一篇    下一篇

基于属性加权的RCM算法

张朋,戴月明,吴定会   

  1. (江南大学物联网工程学院,江苏 无锡 214122)
  • 收稿日期:2015-11-24 修回日期:2016-04-18 出版日期:2018-03-25 发布日期:2018-03-25
  • 基金资助:

    国家863计划(2013AA040405)

A RCM algorithm based on feature weights

ZHANG Peng,DAI Yueming,WU Dinghui   

  1. (School of Internet of Things Engineering,Jiangnan University,Wuxi 214122,China)
  • Received:2015-11-24 Revised:2016-04-18 Online:2018-03-25 Published:2018-03-25

摘要:

传统的粗糙集均值算法RCM的聚类准则是建立在参与聚类的属性同等重要的假设下,而在自然场景下的聚类问题中,不同的属性对聚类结果的影响是不同的。针对该问题,提出了将聚类属性进行加权处理的WRCM算法。具体地,为了筛选出对聚类结果产生关键影响的具有辨别力的聚类属性,算法通过引入权重矩阵将不同的属性赋予不同的属性权重。实验结果表明,本算法可以达到属性选择的效果,从而提高了最终的聚类精确度。

关键词: 粗糙集均值算法, 聚类, 关键属性, 权重矩阵

Abstract:

The clustering criterion of the conventional Rough CMeans (RCM) algorithm is based on the hypothesis that the attributes involved in clustering are equally important. However, in the natural scenario clustering problems, the different attributes have different effects on the clustering results. To address this issue, we propose a weighted RCM by weighting clustering attributes. Specifically, in order to filter out discerning clustering attributes that have a crucial impact on the clustering results, the algorithm assigns different attributes to different attribute weights by introducing a weight matrix. The experimental results show that the proposed method is able to extract the attributes, which improves the clustering accuracy.
 

Key words: Rough C-Means algorithm, clustering, key features, weight matrix