Generalized Information-Theoretic Criterion for Multi-Label Feature Selection
Multi-label feature selection that identifies important features from the original feature set of multi-labeled datasets has been attracting considerable attention owing to its generality compared to conventional single-label feature selection. The unimportant features are filtered by scoring the de...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2019-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8756255/ |
| _version_ | 1818643240844263424 |
|---|---|
| author | Wangduk Seo Dae-Won Kim Jaesung Lee |
| author_facet | Wangduk Seo Dae-Won Kim Jaesung Lee |
| author_sort | Wangduk Seo |
| collection | DOAJ |
| description | Multi-label feature selection that identifies important features from the original feature set of multi-labeled datasets has been attracting considerable attention owing to its generality compared to conventional single-label feature selection. The unimportant features are filtered by scoring the dependency of features to labels. In conventional multi-label feature filter studies, the score function is obtained by approximating a dependency measure such as joint entropy because direct calculation is often impractical due to the presence of multiple labels with limited training patterns. Although the efficacy of approximation can differ depending on the characteristics of the multi-label dataset, conventional methods presume a certain approximation method, leading to a degenerated feature subset if the presumed approximation is inappropriate for the given dataset. In this study, we propose a strategy for selecting an approximation among a series of approximations depending on the number of involved variables and consequently instantiate a score function based on the chosen approximation. The experimental results demonstrate that the proposed strategy and score function outperform conventional multi-label feature selection methods. |
| first_indexed | 2024-12-16T23:55:49Z |
| format | Article |
| id | doaj.art-aca604391c494280b585067bb1bea632 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-16T23:55:49Z |
| publishDate | 2019-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-aca604391c494280b585067bb1bea6322022-12-21T22:11:12ZengIEEEIEEE Access2169-35362019-01-01712285412286310.1109/ACCESS.2019.29274008756255Generalized Information-Theoretic Criterion for Multi-Label Feature SelectionWangduk Seo0https://orcid.org/0000-0003-4806-1614Dae-Won Kim1https://orcid.org/0000-0001-7124-1141Jaesung Lee2https://orcid.org/0000-0002-3757-3510School of Computer Science and Engineering, Chung-Ang University, Seoul, South KoreaSchool of Computer Science and Engineering, Chung-Ang University, Seoul, South KoreaSchool of Computer Science and Engineering, Chung-Ang University, Seoul, South KoreaMulti-label feature selection that identifies important features from the original feature set of multi-labeled datasets has been attracting considerable attention owing to its generality compared to conventional single-label feature selection. The unimportant features are filtered by scoring the dependency of features to labels. In conventional multi-label feature filter studies, the score function is obtained by approximating a dependency measure such as joint entropy because direct calculation is often impractical due to the presence of multiple labels with limited training patterns. Although the efficacy of approximation can differ depending on the characteristics of the multi-label dataset, conventional methods presume a certain approximation method, leading to a degenerated feature subset if the presumed approximation is inappropriate for the given dataset. In this study, we propose a strategy for selecting an approximation among a series of approximations depending on the number of involved variables and consequently instantiate a score function based on the chosen approximation. The experimental results demonstrate that the proposed strategy and score function outperform conventional multi-label feature selection methods.https://ieeexplore.ieee.org/document/8756255/Machine learningmulti-label learningmulti-label feature selectioninformation entropy |
| spellingShingle | Wangduk Seo Dae-Won Kim Jaesung Lee Generalized Information-Theoretic Criterion for Multi-Label Feature Selection IEEE Access Machine learning multi-label learning multi-label feature selection information entropy |
| title | Generalized Information-Theoretic Criterion for Multi-Label Feature Selection |
| title_full | Generalized Information-Theoretic Criterion for Multi-Label Feature Selection |
| title_fullStr | Generalized Information-Theoretic Criterion for Multi-Label Feature Selection |
| title_full_unstemmed | Generalized Information-Theoretic Criterion for Multi-Label Feature Selection |
| title_short | Generalized Information-Theoretic Criterion for Multi-Label Feature Selection |
| title_sort | generalized information theoretic criterion for multi label feature selection |
| topic | Machine learning multi-label learning multi-label feature selection information entropy |
| url | https://ieeexplore.ieee.org/document/8756255/ |
| work_keys_str_mv | AT wangdukseo generalizedinformationtheoreticcriterionformultilabelfeatureselection AT daewonkim generalizedinformationtheoreticcriterionformultilabelfeatureselection AT jaesunglee generalizedinformationtheoreticcriterionformultilabelfeatureselection |