An Effective Multi-Label Feature Selection Model Towards Eliminating Noisy Features

Feature selection has devoted a consistently great amount of effort to dimension reduction for various machine learning tasks. Existing feature selection models focus on selecting the most discriminative features for learning targets. However, this strategy is weak in handling two kinds of features,...

Full description

Bibliographic Details
Main Authors: Jun Wang, Yuanyuan Xu, Hengpeng Xu, Zhe Sun, Zhenglu Yang, Jinmao Wei
Format: Article
Language:English
Published: MDPI AG 2020-11-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/10/22/8093
id doaj-5d251b1ab71248ecb1006c27b811dc38
record_format Article
spelling doaj-5d251b1ab71248ecb1006c27b811dc382020-11-25T04:01:05ZengMDPI AGApplied Sciences2076-34172020-11-01108093809310.3390/app10228093An Effective Multi-Label Feature Selection Model Towards Eliminating Noisy FeaturesJun Wang0Yuanyuan Xu1Hengpeng Xu2Zhe Sun3Zhenglu Yang4Jinmao Wei5College of Mathematics and Statistics Science, Ludong University, Yantai 264025, ChinaCollege of Computer Science, Nankai University, Tianjin 300071, ChinaTianjin Key Laboratory of Wireless Mobile Communications and Power Transmission, College of Electronic and Communication Engineering, Tianjin Normal University, Tianjin 300387, ChinaRIKEN National Science Institute, Wako, Saitama 351-0198, JapanCollege of Computer Science, Nankai University, Tianjin 300071, ChinaCollege of Computer Science, Nankai University, Tianjin 300071, ChinaFeature selection has devoted a consistently great amount of effort to dimension reduction for various machine learning tasks. Existing feature selection models focus on selecting the most discriminative features for learning targets. However, this strategy is weak in handling two kinds of features, that is, the irrelevant and redundant ones, which are collectively referred to as noisy features. These features may hamper the construction of optimal low-dimensional subspaces and compromise the learning performance of downstream tasks. In this study, we propose a novel multi-label feature selection approach by embedding label correlations (dubbed ELC) to address these issues. Particularly, we extract label correlations for reliable label space structures and employ them to steer feature selection. In this way, label and feature spaces can be expected to be consistent and noisy features can be effectively eliminated. An extensive experimental evaluation on public benchmarks validated the superiority of ELC.https://www.mdpi.com/2076-3417/10/22/8093feature selectionnoise eliminationspace consistencylabel correlations
collection DOAJ
language English
format Article
sources DOAJ
author Jun Wang
Yuanyuan Xu
Hengpeng Xu
Zhe Sun
Zhenglu Yang
Jinmao Wei
spellingShingle Jun Wang
Yuanyuan Xu
Hengpeng Xu
Zhe Sun
Zhenglu Yang
Jinmao Wei
An Effective Multi-Label Feature Selection Model Towards Eliminating Noisy Features
Applied Sciences
feature selection
noise elimination
space consistency
label correlations
author_facet Jun Wang
Yuanyuan Xu
Hengpeng Xu
Zhe Sun
Zhenglu Yang
Jinmao Wei
author_sort Jun Wang
title An Effective Multi-Label Feature Selection Model Towards Eliminating Noisy Features
title_short An Effective Multi-Label Feature Selection Model Towards Eliminating Noisy Features
title_full An Effective Multi-Label Feature Selection Model Towards Eliminating Noisy Features
title_fullStr An Effective Multi-Label Feature Selection Model Towards Eliminating Noisy Features
title_full_unstemmed An Effective Multi-Label Feature Selection Model Towards Eliminating Noisy Features
title_sort effective multi-label feature selection model towards eliminating noisy features
publisher MDPI AG
series Applied Sciences
issn 2076-3417
publishDate 2020-11-01
description Feature selection has devoted a consistently great amount of effort to dimension reduction for various machine learning tasks. Existing feature selection models focus on selecting the most discriminative features for learning targets. However, this strategy is weak in handling two kinds of features, that is, the irrelevant and redundant ones, which are collectively referred to as noisy features. These features may hamper the construction of optimal low-dimensional subspaces and compromise the learning performance of downstream tasks. In this study, we propose a novel multi-label feature selection approach by embedding label correlations (dubbed ELC) to address these issues. Particularly, we extract label correlations for reliable label space structures and employ them to steer feature selection. In this way, label and feature spaces can be expected to be consistent and noisy features can be effectively eliminated. An extensive experimental evaluation on public benchmarks validated the superiority of ELC.
topic feature selection
noise elimination
space consistency
label correlations
url https://www.mdpi.com/2076-3417/10/22/8093
work_keys_str_mv AT junwang aneffectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT yuanyuanxu aneffectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT hengpengxu aneffectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT zhesun aneffectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT zhengluyang aneffectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT jinmaowei aneffectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT junwang effectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT yuanyuanxu effectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT hengpengxu effectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT zhesun effectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT zhengluyang effectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
AT jinmaowei effectivemultilabelfeatureselectionmodeltowardseliminatingnoisyfeatures
_version_ 1724447707814166528