A hybrid gene selection approach for microarray data classification using cellular learning automata and ant colony...

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Microarray data analysis requires effective gene selection to reduce dimensionality and improve classification accuracy.
• Existing methods often face challenges in balancing feature subset size with predictive performance.

Purpose of the Study:

• To propose a hybrid gene selection approach for microarray data.
• To enhance classification accuracy by identifying the smallest effective feature subset.
• To evaluate the proposed method on multiple benchmark datasets.

Main Methods:

• A two-stage approach: primary filtering using Fisher criterion followed by a wrapper method.
• Wrapper method employs cellular learning automata (CLA) optimized with ant colony optimization (ACO).
• Feature subset evaluation using Receiver Operating Characteristic (ROC) curves and classifiers like K-nearest neighbor, support vector machine, and naïve Bayes.

Main Results:

• The proposed approach successfully reduces the search space and computational complexity.
• It identifies the smallest gene subsets while achieving high classification accuracy.
• Consistent performance observed across four different microarray datasets.

Conclusions:

• The hybrid gene selection strategy is effective for microarray data analysis.
• The combination of filtering and wrapper methods with CLA and ACO optimizes feature selection.
• This approach offers a promising solution for efficient and accurate gene selection in bioinformatics.