A multi-objective optimization algorithm for feature selection problems

Feature selection (FS) is a critical step in data mining, and machine learning algorithms play a crucial role in algorithms performance. It reduces the processing time and accuracy of the categories. In this paper, three different solutions are proposed to FS. In the first solution, the Harris Hawks...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Engineering with computers Ročník 38; číslo Suppl 3; s. 1845 - 1863
Hlavní autoři: Abdollahzadeh, Benyamin, Gharehchopogh, Farhad Soleimanian
Médium: Journal Article
Jazyk:angličtina
Vydáno: London Springer London 01.08.2022
Springer Nature B.V
Témata:
Algorithms
CAE) and Design
Calculus of Variations and Optimal Control; Optimization
Classical Mechanics
Computer Science
Computer-Aided Engineering (CAD
Control
Control methods
Data mining
Datasets
Feature selection
Machine learning
Math. Applications in Chemistry
Mathematical and Computational Engineering
Multiple objective analysis
Optimization algorithms
Original Article
Pareto optimization
Standard data
Statistical tests
Systems Theory
Multiobjective
Feature selection
Bonferroni–Holm
Fruitfly optimization algorithm
Family-wise error rate
Harris hawks optimization
ISSN:0177-0667, 1435-5663
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Feature selection (FS) is a critical step in data mining, and machine learning algorithms play a crucial role in algorithms performance. It reduces the processing time and accuracy of the categories. In this paper, three different solutions are proposed to FS. In the first solution, the Harris Hawks Optimization (HHO) algorithm has been multiplied, and in the second solution, the Fruitfly Optimization Algorithm (FOA) has been multiplied, and in the third solution, these two solutions are hydride and are named MOHHOFOA. The results were tested with MOPSO, NSGA-II, BGWOPSOFS and B-MOABC algorithms for FS on 15 standard data sets with mean, best, worst, standard deviation (STD) criteria. The Wilcoxon statistical test was also used with a significance level of 5% and the Bonferroni–Holm method to control the family-wise error rate. The results are shown in the Pareto front charts, indicating that the proposed solutions' performance on the data set is promising.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0177-0667
1435-5663
DOI:10.1007/s00366-021-01369-9