Enhanced chimp hierarchy optimization algorithm with adaptive lens imaging for feature selection in data...

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• Feature selection is crucial in machine learning for data preprocessing.
• The Chimp Optimization Algorithm (CHoA) offers fast convergence but suffers from weak exploration and local optima issues in feature selection.
• Ineffective feature selection leads to suboptimal performance in machine learning models.

Purpose of the Study:

• To propose an Enhanced Chimp Hierarchy Optimization Algorithm (ALI-CHoASH) to address the limitations of CHoA in feature selection.
• To improve the exploration and exploitation capabilities of the Chimp Optimization Algorithm.
• To achieve optimal feature subset selection for classification problems.

Main Methods:

• Designed a chimp social hierarchy with a social class factor to model individual relationships.
• Introduced novel attacking prey and autonomous search strategies to enhance optimization.
• Implemented an adaptive lens imaging back-learning strategy to prevent local optima and improve diversity.
• Validated the algorithm on high-dimensional datasets.

Main Results:

• ALI-CHoASH demonstrated improved exploration and exploitation capabilities compared to standard CHoA.
• The proposed algorithm achieved superior performance in classification accuracy and reduced feature subset size.
• ALI-CHoASH showed competitive or superior results against eight state-of-the-art feature selection methods.
• The algorithm effectively avoided falling into local optima, particularly in later iterations.

Conclusions:

• The ALI-CHoASH algorithm offers a significant advancement in feature selection for machine learning.
• The enhanced social hierarchy and adaptive strategies effectively improve optimization performance.
• ALI-CHoASH provides a robust and efficient solution for selecting optimal feature subsets, enhancing classification accuracy and reducing computational load.