Loading…

Multimodal Bare-Bone Niching Differential Evolution in Feature Selection

Feature selection (FS) is to select relevant feature subsets from a given feature set to reduce the dimensions of data. In the classification task, feature selection can not only reduce the calculation cost but also improve the classification accuracy. In this paper, a multimodal bare-bone niching d...

Full description

Saved in:
Bibliographic Details
Main Authors: Hu, Xiao-Min, Guo, Zi-Wen
Format: Conference Proceeding
Language:English
Subjects:
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Feature selection (FS) is to select relevant feature subsets from a given feature set to reduce the dimensions of data. In the classification task, feature selection can not only reduce the calculation cost but also improve the classification accuracy. In this paper, a multimodal bare-bone niching differential evolution algorithm (MBNDE) is proposed for FS in classification problems. MBNDE finds multiple feature combinations that can achieve the highest classification rate. Three different strategies have been applied to MBNDE in the niching process, and the corresponding algorithms are termed MBNDE-CC (crowding and cluster), MBNDE-CS (crowding and speciation), as well as MBNDE (index-based). A modified 3-NN classifier is designed to evaluate the classification rate of different feature combinations. The results of the proposed method and some existing multimodal and unimodal algorithms for FS are compared. The proposed algorithm shows promising performance.
ISSN:2577-1655
DOI:10.1109/SMC52423.2021.9658633