An Improved Binary Walrus Optimizer with Golden Sine Disturbance and Population Regeneration Mechanism to Solve...

Area of Science:

• Machine Learning
• Data Mining
• Computational Intelligence

Background:

• Feature selection (FS) is crucial for dimensionality reduction in high-dimensional data.
• Metaheuristic algorithms offer powerful search capabilities for effective FS.
• Existing methods face challenges in balancing exploration and exploitation and escaping local optima.

Purpose of the Study:

• To propose a novel improved binary walrus optimizer (WO) algorithm, BGEPWO, for enhanced feature selection.
• To improve population diversity, algorithm stability, and convergence speed.
• To increase the algorithm's capacity to escape local optima and expand the search range.

Main Methods:

• Initialization using an iterative chaotic map with infinite collapses (ICMIC) for diversity.
• Introduction of an adaptive operator for stability and exploration-exploitation balance.
• Implementation of a population regeneration mechanism, elite opposition-based learning (EOBL), and a golden sine strategy for optimization.

Main Results:

• BGEPWO demonstrated superior performance across 21 datasets compared to the standard BWO and 10 other algorithms.
• Significant improvements were observed in fitness value, number of selected features, and F1-score.
• The algorithm achieved higher accuracy, better feature reduction, and stronger convergence.

Conclusions:

• BGEPWO effectively addresses feature selection challenges in high-dimensional data.
• The proposed enhancements lead to increased population diversity, balanced exploration-exploitation, and improved local optima evasion.
• BGEPWO offers a robust and efficient solution for machine learning and data mining applications.