Feature selection using max dynamic relevancy and min redundancy

Feature selection algorithms based on three-way interaction information have been widely studied. However, most of these traditional algorithms only consider class-dependent redundancy, which can lead to an underestimation of redundancy. To address this issue, a feature selection algorithm based on...

Full description

Saved in:
Bibliographic Details
Published inPattern analysis and applications : PAA Vol. 26; no. 2; pp. 631 - 643
Main Authors Yin, Kexin, Zhai, Junren, Xie, Aifeng, Zhu, Jianqi
Format Journal Article
LanguageEnglish
Published London Springer London 01.05.2023
Springer Nature B.V
Subjects
Algorithms
Computer Science
Pattern Recognition
Redundancy
Theoretical Advances
Feature selection
Interaction information
Class-independent redundancy
Mutual information
Online AccessGet full text

Cover

More Information
Summary:Feature selection algorithms based on three-way interaction information have been widely studied. However, most of these traditional algorithms only consider class-dependent redundancy, which can lead to an underestimation of redundancy. To address this issue, a feature selection algorithm based on maximum dynamic relevancy minimum redundancy is proposed. The algorithm first proposes a quality coefficient to estimate the feature relevancy. Then we introduce class-independent redundancy to solve the issue of not fully considering redundancy, and propose adaptive coefficients to optimize the algorithm. To ensure the effectiveness of the algorithm, experimental comparisons are carried on 19 benchmark data sets with six algorithms. The results show that the proposed algorithm outperforms other algorithms in terms of performance.
ISSN:1433-7541
1433-755X
DOI:10.1007/s10044-023-01138-y