计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (36): 9-11.
• 博士论坛 • 上一篇 下一篇
胡 健,孙金花
收稿日期:
修回日期:
出版日期:
发布日期:
HU Jian,SUN Jinhua
Received:
Revised:
Online:
Published:
摘要: 为提高聚类集成算法效率,弥补以往聚类集成算法的不足,确保多数聚类成员分簇的均匀无偏差,提出了一种新的基于聚类集成的多目标聚类分析框架,并利用系统能量理论定义了多目标聚类问题的优化目标函数。在此基础上,设计了一种启发式的K-ETMC聚类集成算法,并对Iris、Wine、Soybean三个数据集进行了快速有效的聚类分析,通过与MCLA,HGPA,CSPA三个典型聚类集成算法比较表明:该算法聚类效果较好,能够有效地改善聚类结果。
关键词: 聚类, 集成学习, 多目标优化, 系统能量理论
Abstract: To raise the efficiency of cluster ensemble algorithm,make up the disadvantage of traditional cluster algorithm and assure the uniform no-deviation of clustering algorithm of a lot of cluster members,a new cluster ensemble with multi-objective cluster analysis framework is proposed in this paper.The optimal object function of multi-objective cluster is defined by using system energy theory.On the basis,a heuristic algorithm K-ETMC is designed,which is used to make a fast and effective clustering analysis on Iris,Wine and Soybean data sets.The experiment with MCLA,HGPA and CSPA typical algorithms comparison shows that cluster effect is better,which can effectively improve cluster result.
Key words: cluster, ensemble study, multi-objective optimization, system energy theory
胡 健,孙金花. 基于系统能量理论的多目标优化聚类集成研究[J]. 计算机工程与应用, 2011, 47(36): 9-11.
HU Jian,SUN Jinhua. Multi-objective cluster ensemble with system energy theory[J]. Computer Engineering and Applications, 2011, 47(36): 9-11.
0 / 推荐
导出引用管理器 EndNote|Ris|BibTeX
链接本文: http://cea.ceaj.org/CN/
http://cea.ceaj.org/CN/Y2011/V47/I36/9