Explorative Binary Gray Wolf Optimizer with Quadratic Interpolation for Feature Selection.

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• High dimensionality in large datasets hinders data mining efficiency.
• Feature selection is crucial for dimensionality reduction and improved classification accuracy.

Purpose of the Study:

• To propose a novel binary Gray Wolf Optimization algorithm for feature selection in classification tasks.
• To enhance the exploration and exploitation capabilities for optimal feature subset selection.

Main Methods:

• Utilizing historical optimal positions to guide search agent exploration.
• Implementing a quadratic interpolation technique to balance population diversity and local exploitation.
• Incorporating chaotic perturbations to prevent premature convergence and promote global search.
• Employing a novel transfer function for effective binary space optimization.

Main Results:

• The proposed algorithm demonstrated superior performance in feature selection.
• Experimental results showed significant improvements in classification accuracy across 32 datasets.
• The method outperformed other advanced feature selection algorithms.

Conclusions:

• The novel binary Gray Wolf Optimization algorithm effectively addresses high-dimensional data challenges.
• The integrated techniques enhance search capability, diversity, and convergence accuracy.
• The algorithm provides an effective solution for optimal feature subset selection in machine learning.