Enriching the random subspace method with margin theory – a solution for the high-dimensional classification task

The random subspace method (RSM) has proved its excellence in numbers of pattern recognition tasks. However, the standard RSM is limited owing to the randomness in its feature selection procedure that is likely to lead to feature subset having poor class separability. In this paper, a proposal for a...

Full description

Saved in:
Bibliographic Details
Main Authors: Hongyan Xu, Tao Lin, Yingtao Xie, Zhi Chen
Format: Article
Language:English
Published: Taylor & Francis Group 2018-10-01
Series:Connection Science
Subjects:
Online Access:http://dx.doi.org/10.1080/09540091.2018.1512556
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850242539581341696
author Hongyan Xu
Tao Lin
Yingtao Xie
Zhi Chen
author_facet Hongyan Xu
Tao Lin
Yingtao Xie
Zhi Chen
author_sort Hongyan Xu
collection DOAJ
description The random subspace method (RSM) has proved its excellence in numbers of pattern recognition tasks. However, the standard RSM is limited owing to the randomness in its feature selection procedure that is likely to lead to feature subset having poor class separability. In this paper, a proposal for a margin-based criterion has been presented for the evaluation of the true significance of the features, together with the true classification ability of base classifiers, so that both the training phase and integration phase of standard RSM could be enhanced. In the training phase, the random feature selection procedure is enhanced using a weighted random feature selection procedure, in order to improve the classification ability of the base classifier. In the integration phase, the simple majority voting strategy is enhanced using a weighted majority voting strategy for the purpose of assigning the base classifiers with poor classification ability to the lower voting weights. Experimental results on 30 benchmark datasets, together with 6 high-dimensional datasets prove that the recommended approach is capable of better providing classification ability to the usual classification task, in addition to the high-dimensional classification task.
format Article
id doaj-art-b2a67a9d1d2249c2920792a4167324d4
institution OA Journals
issn 0954-0091
1360-0494
language English
publishDate 2018-10-01
publisher Taylor & Francis Group
record_format Article
series Connection Science
spelling doaj-art-b2a67a9d1d2249c2920792a4167324d42025-08-20T02:00:16ZengTaylor & Francis GroupConnection Science0954-00911360-04942018-10-0130440942410.1080/09540091.2018.15125561512556Enriching the random subspace method with margin theory – a solution for the high-dimensional classification taskHongyan Xu0Tao Lin1Yingtao Xie2Zhi Chen3College of Computer Science, Sichuan UniversityCollege of Computer Science, Sichuan UniversityCollege of Computer Science, Sichuan UniversityCollege of Computer Science, Sichuan UniversityThe random subspace method (RSM) has proved its excellence in numbers of pattern recognition tasks. However, the standard RSM is limited owing to the randomness in its feature selection procedure that is likely to lead to feature subset having poor class separability. In this paper, a proposal for a margin-based criterion has been presented for the evaluation of the true significance of the features, together with the true classification ability of base classifiers, so that both the training phase and integration phase of standard RSM could be enhanced. In the training phase, the random feature selection procedure is enhanced using a weighted random feature selection procedure, in order to improve the classification ability of the base classifier. In the integration phase, the simple majority voting strategy is enhanced using a weighted majority voting strategy for the purpose of assigning the base classifiers with poor classification ability to the lower voting weights. Experimental results on 30 benchmark datasets, together with 6 high-dimensional datasets prove that the recommended approach is capable of better providing classification ability to the usual classification task, in addition to the high-dimensional classification task.http://dx.doi.org/10.1080/09540091.2018.1512556random subspace methodwrrsmmargin theoryweighted majority votinghigh-dimensional problem
spellingShingle Hongyan Xu
Tao Lin
Yingtao Xie
Zhi Chen
Enriching the random subspace method with margin theory – a solution for the high-dimensional classification task
Connection Science
random subspace method
wrrsm
margin theory
weighted majority voting
high-dimensional problem
title Enriching the random subspace method with margin theory – a solution for the high-dimensional classification task
title_full Enriching the random subspace method with margin theory – a solution for the high-dimensional classification task
title_fullStr Enriching the random subspace method with margin theory – a solution for the high-dimensional classification task
title_full_unstemmed Enriching the random subspace method with margin theory – a solution for the high-dimensional classification task
title_short Enriching the random subspace method with margin theory – a solution for the high-dimensional classification task
title_sort enriching the random subspace method with margin theory a solution for the high dimensional classification task
topic random subspace method
wrrsm
margin theory
weighted majority voting
high-dimensional problem
url http://dx.doi.org/10.1080/09540091.2018.1512556
work_keys_str_mv AT hongyanxu enrichingtherandomsubspacemethodwithmargintheoryasolutionforthehighdimensionalclassificationtask
AT taolin enrichingtherandomsubspacemethodwithmargintheoryasolutionforthehighdimensionalclassificationtask
AT yingtaoxie enrichingtherandomsubspacemethodwithmargintheoryasolutionforthehighdimensionalclassificationtask
AT zhichen enrichingtherandomsubspacemethodwithmargintheoryasolutionforthehighdimensionalclassificationtask