| 基于属性相关性分析的子空间搜索算法 |
| |
| 引用本文： | 夏英,李克非.基于属性相关性分析的子空间搜索算法[J].重庆邮电大学学报(自然科学版),2009,21(4):544-548. |
| |
| 作者姓名： | 夏英 李克非 |
| |
| 作者单位： | 1. 西南交通大学,信息科学与技术学院,成都,610031;重庆邮电大学,计算机学院,重庆,400065
2. 重庆邮电大学,计算机学院,重庆,400065 |
| |
| 基金项目： | 国家自然科学基金，重庆市教育委员会科学技术研究项目 |
| |
| 摘 要： | 在高维子空间聚类算法中,子空间识别质量有助于改善算法性能.提出了基于属性相关性分析的子空间搜索算法,利用相对熵去除冗余属性,根据联合基尼值判断两两非冗余属性之间的相关性,形成属性关系图,并在此属性关系图上搜索相似兴趣子空间.实验结果表明算法参数容易设置,有较好的子空间识别能力.
|
| 关 键 词： | 相对熵 联合基尼值 属性关系图 相似兴趣子空间 |
| 收稿时间： | 6/9/2009 12:00:00 AM |
Subspace search algorithm based on attribute relativity analysis |
| |
| XIA Ying,LI Ke-fei.Subspace search algorithm based on attribute relativity analysis[J].Journal of Chongqing University of Posts and Telecommunications,2009,21(4):544-548. |
| |
| Authors: | XIA Ying LI Ke-fei |
| |
| Institution: | School of Information Science and Technology, Southwest Jiaotong University, Chengdu 600031, P. R. China |
| |
| Abstract: | In the high dimensional subspace clustering algorithm, the quality of identifying the subspace is useful for improving performance. The subspace search algorithm was proposed based on the attribute relativity analysis. The algorithm filters out redundant attributes by computing the relative entropy, then estimates the relativity of the non redundant attributes with united gini coefficient, and forms the attribute relational graph. At last, the similar interesting subspaces are searched in the attribute relational graph. Experimental results show that it is easy to set the input parameters and this algorithm has good ability for subspace identification. |
| |
| Keywords: | relative entropy gini coefficient attribute relational graph similar interesting subspace |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《重庆邮电大学学报(自然科学版)》浏览原始摘要信息 |
| 点击此处可从《重庆邮电大学学报(自然科学版)》下载免费的PDF全文 |
|
