A comprehensive learning based swarm optimization approach for feature selection in gene expression data.

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Gene expression data analysis presents challenges due to high dimensionality and complexity.
• Feature selection is a critical preprocessing step to identify relevant genes.
• Existing methods may struggle with the intricate nature of genomic datasets.

Purpose of the Study:

• To introduce a novel feature selection approach, Comprehensive Learning-Based Swarm Optimization (CLBSO), for gene expression data.
• To enhance the efficiency and accuracy of identifying relevant genes in high-dimensional datasets.
• To improve classification performance in bioinformatics.

Main Methods:

• CLBSO combines ant colony optimization for local search and grasshopper optimization for global exploration.
• The algorithm utilizes pheromone trails for swarm guidance and long jumps to avoid local optima.
• Evaluated on multiple public gene expression datasets against state-of-the-art methods.

Main Results:

• CLBSO demonstrated an average accuracy improvement of 15% over original high-dimensional data.
• Outperformed existing feature selection methods by up to 10%, achieving 97.2% accuracy on the Pancreatic cancer dataset.
• Showed faster convergence and consistent gene subset selection, indicating robustness.

Conclusions:

• CLBSO is a robust and effective method for feature selection in complex gene expression data.
• The approach significantly enhances classification accuracy in bioinformatics.
• CLBSO offers a valuable tool for researchers analyzing genomic datasets.