regularized local reconstruction for clustering
文献类型：会议论文
| 作者 | Sun Jun ; Shen Zhiyong ; Su Bai ; Shen Yidong |
| 出版日期 | 2009 |
| 会议名称 | 13th Pacific-Asia Conference on Knowledge and Data Mining |
| 会议日期 | APR 27-30, |
| 会议地点 | Bangkok, THAILAND |
| 关键词 | Clustering Local Reconstruction Entropy Regularization |
| 英文摘要 | Motivated by the local reconstruction approach to discovering low dimensional structure in high dimensional data; we propose a novel clustering algorithm that effectively utilizes local reconstruction information. We obtain the local reconstruction weights by minimizing the reconstruction error between each data point; and the reconstruction from its neighbors. An entropy regularization terns is incorporated into the reconstruction objective function so that the smoothness of the reconstruction weights call be explicitly controlled. The reconstruction weights are then used to obtain the clustering result by employing spectral clustering techniques. Experimental results on a number of datasets demonstrate that our algorithm performs well relative to other approaches, which validate the effectiveness of our approach for clustering. |
| 会议主办者 | Sirindhorn Int Inst Technol, Thammasat Univ, Chulalonkorn Univ, Asian Inst Technol, Natl Elect & Comp Technol Ctr, Thailand Convent & Exhibit Bureau, AF Off Sci Res, Asian Off Aerosp Res & Dev |
| 会议录 | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
 |
| 会议录出版者 | ADVANCES IN KNOWLEDGE DISCOVERY AND DATA MINING, PROCEEDINGS |
| 会议录出版地 | HEIDELBERGER PLATZ 3, D-14197 BERLIN, GERMANY |
| ISSN号 | 0302-9743 |
| ISBN号 | 978-3-642-01306-5 |
| 源URL | [http://124.16.136.157/handle/311060/8360]  |
| 专题 | 软件研究所_计算机科学国家重点实验室 _会议论文 |
| 推荐引用方式 GB/T 7714 | Sun Jun,Shen Zhiyong,Su Bai,et al. regularized local reconstruction for clustering[C]. 见:13th Pacific-Asia Conference on Knowledge and Data Mining. Bangkok, THAILAND. APR 27-30,. |
入库方式： OAI收割
来源：软件研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。