Quasi-reflection learning arithmetic optimization algorithm firefly search for feature selection.

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• The rapid expansion of digital data necessitates efficient methods for information extraction.
• Feature selection is a vital preprocessing step in machine learning to manage large datasets and improve model performance.

Purpose of the Study:

• To introduce a novel quasi-reflection learning arithmetic optimization algorithm, termed Firefly Search.
• To enhance population diversity and exploitation capabilities for improved feature selection.

Main Methods:

• Developed a wrapper-based feature selection method integrating quasi-reflection learning and Firefly algorithm metaheuristics.
• Tested the algorithm on benchmark functions and standard datasets from UCI and ASU repositories.
• Applied the method to a Corona disease dataset for classification.

Main Results:

• The proposed Firefly Search algorithm demonstrated significant improvements over existing methods.
• Experimental results confirmed the statistical significance of the enhancements in feature selection.
• Effective identification of optimal feature subsets was achieved for classification tasks.

Conclusions:

• The novel Firefly Search algorithm offers a robust and efficient solution for feature selection in machine learning.
• The method shows promise for applications in diverse fields, including medical data analysis.
• This research contributes to advancing data preprocessing techniques for large-scale datasets.