Clustering Structure Sensitive Dissimilarity-Based Sparse Subset Selection.

Saved in:

Bibliographic Details
Title:	Clustering Structure Sensitive Dissimilarity-Based Sparse Subset Selection.
Authors:	Xiaobin Zhi¹ xbzhi@xupt.edu.cn, Bingzhe Li² 1229083133@qq.com, Yuqing Qiu² 1833652606@qq.com, Yigang Qi² 1603600973@qq.com, Yongfei Li² lyf82182010@163.com
Source:	IAENG International Journal of Computer Science. Nov2025, Vol. 52 Issue 11, p4353-4365. 13p.
Subjects:	Clustering algorithms, Subset selection, Mathematical regularization, Data management, Quantitative research, Constrained optimization, Heterogeneity, Algorithms
Abstract:	By cleverly utilizing the theory of sparse representation, sparse subset selection methods aim to identify the most informative subset from a dataset to represent the original data, thereby eliminating data redundancy. The recently proposed dissimilarity-based sparse subset selection (DS3) algorithm demonstrates both theoretical elegance and practical effectiveness. However, the ability of the DS3 algorithm in detecting clustering structures is limited. Specifically, when the dataset exhibits explicit clustering structures, it is essential to carefully adjust the sparsity regularization parameter to select one representative for each cluster. Moreover, when the DS3 algorithm selects multiple representatives for each cluster, the selected representatives often lack sufficient diversity. To address these issues, we propose a clustering structure sensitive DS3 (CSS-DS3) algorithm, which is implemented by adding a low-rank penalty to the member matrix in DS3. When the dataset exhibits a clear clustering structure, the CSS-DS3 algorithm can effectively identify the clustering structure and selects one or more correct and diverse representatives for each cluster, even within a wide range of sparsity regularization parameter values. The proposed algorithm model is solved by the alternating direction method of multipliers (ADMM). Experimental results on both synthetic and real-world datasets demonstrate that the proposed method outperforms state-of-the-art approaches. [ABSTRACT FROM AUTHOR]
	Copyright of IAENG International Journal of Computer Science is the property of International Association of Engineers (IAENG) and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	By cleverly utilizing the theory of sparse representation, sparse subset selection methods aim to identify the most informative subset from a dataset to represent the original data, thereby eliminating data redundancy. The recently proposed dissimilarity-based sparse subset selection (DS3) algorithm demonstrates both theoretical elegance and practical effectiveness. However, the ability of the DS3 algorithm in detecting clustering structures is limited. Specifically, when the dataset exhibits explicit clustering structures, it is essential to carefully adjust the sparsity regularization parameter to select one representative for each cluster. Moreover, when the DS3 algorithm selects multiple representatives for each cluster, the selected representatives often lack sufficient diversity. To address these issues, we propose a clustering structure sensitive DS3 (CSS-DS3) algorithm, which is implemented by adding a low-rank penalty to the member matrix in DS3. When the dataset exhibits a clear clustering structure, the CSS-DS3 algorithm can effectively identify the clustering structure and selects one or more correct and diverse representatives for each cluster, even within a wide range of sparsity regularization parameter values. The proposed algorithm model is solved by the alternating direction method of multipliers (ADMM). Experimental results on both synthetic and real-world datasets demonstrate that the proposed method outperforms state-of-the-art approaches. [ABSTRACT FROM AUTHOR]
ISSN:	1819656X