Feature Selection Technique for Microarray Data Using Multi-objective Jaya Algorithm Based on Chaos Theory

Microarray datasets are high-dimensional data that contain the gene expression profiling information used for cancer classification. This causes a curse of dimensionality and poses multiple challenges in gene classification. Moreover, the microarray datasets are imbalanced as they have a very small sample size, which deteriorates the performance of machine learning algorithms. In order to address these problems, dimensionality reduction technique is required. Meta-heuristic optimization algorithm-based feature selection is the efficient dimensionality reduction technique. This paper introduces a feature selection technique for microarray data classification using multi-objective Jaya algorithm based on chaos theory. The two objectives: minimization of selected gene subset and maximization of classification accuracy are considered. Chaos theory is used to improve the convergence of Jaya algorithm. Five benchmark microarray datasets were used to evaluate the performance of the proposed feature selection technique. Furthermore, a comparative study is carried out where the performance of the proposed technique is compared with four other competitors.