Hybrid Whale Optimization Algorithm with simulated annealing for feature selection

•Four hybrid feature selection methods for classification task are proposed.•Our hybrid method combines Whale Optimization Algorithm with simulated annealing.•Eighteen UCI datasets were used in the experiments.•Our approaches result a higher accuracy by using less number of features. Hybrid metaheur...

Full description

Saved in:
Bibliographic Details
Published in:Neurocomputing (Amsterdam) Vol. 260; pp. 302 - 312
Main Authors: Mafarja, Majdi M., Mirjalili, Seyedali
Format: Journal Article
Language:English
Published: Elsevier B.V 18.10.2017
Subjects:
Classification
Feature selection
Hybrid optimization
Optimization
Simulated annealing
Whale Optimization Algorithm
WOA
Whale Optimization Algorithm
Feature selection
Hybrid optimization
Simulated annealing
Classification
WOA
Optimization
ISSN:0925-2312, 1872-8286
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Four hybrid feature selection methods for classification task are proposed.•Our hybrid method combines Whale Optimization Algorithm with simulated annealing.•Eighteen UCI datasets were used in the experiments.•Our approaches result a higher accuracy by using less number of features. Hybrid metaheuristics are of the most interesting recent trends in optimization and memetic algorithms. In this paper, two hybridization models are used to design different feature selection techniques based on Whale Optimization Algorithm (WOA). In the first model, Simulated Annealing (SA) algorithm is embedded in WOA algorithm, while it is used to improve the best solution found after each iteration of WOA algorithm in the second model. The goal of using SA here is to enhance the exploitation by searching the most promising regions located by WOA algorithm. The performance of the proposed approaches is evaluated on 18 standard benchmark datasets from UCI repository and compared with three well-known wrapper feature selection methods in the literature. The experimental results confirm the efficiency of the proposed approaches in improving the classification accuracy compared to other wrapper-based algorithms, which insures the ability of WOA algorithm in searching the feature space and selecting the most informative attributes for classification tasks.
ISSN:0925-2312
1872-8286
DOI:10.1016/j.neucom.2017.04.053