A Novel Crossover Operator Based on Grey Wolf Optimizer Applied to Feature Selection Problem

Alzheimer’s disease (AD) poses challenges in feature selection due to the limited number of features relative to the large number of samples. To address this issue, we propose a novel crossover operator called the Grey Wolf Optimizer Genetic Algorithm (GWOGA) to enhance classification accuracy with a reduced number of features in AD research. GWOGA combines the Grey Wolf Optimizer and a genetic algorithm, addressing limitations and improving feature selection performance. We compared GWOGA with five state-of-the-art feature selection methods and seven crossover operators on integrated and real AD datasets. Results demonstrate that GWOGA outperforms all other methods in terms of accuracy and dimensionality, achieving similar classification accuracy with a significantly smaller feature set. The fitness score analysis indicates that GWOGA identifies a subset of features effectively representing the entire dataset, despite not always having the highest accuracy among state-of-the-art methods. GWOGA shows promise for future AD research, providing a balance between accuracy and computation time and exhibiting potential for identifying novel AD biomarkers.