Gene Selection Using Hybrid Multi-Objective Cuckoo Search Algorithm With Evolutionary Operators for Cancer Microarray Data

Microarray data play a huge role in recognizing a proper cancer diagnosis and classification. In most microarray data set consist of thousands of genes, but the majority number of genes are irrelevant to the diseases. An efficient algorithm for gene selection becomes important to deal with large microarray data. The main challenge is to analyze and select the relevant genes with maximum classification accuracy. Various algorithms were proposed for gene classification in previous studies, however, limited success was succeeded due to the selection of many genes in the high-dimensional microarray data. This study proposed and developed a hybrid multi-objective cuckoo search with evolutionary operators for gene selection. Evolutionary operators that are used in this article were double mutation and single crossover operators. The motivation behind this research is to improve the dimensions' values and explorative search abilities. Multi-objective cuckoo search with evolutionary operators employed the selection of informative genes among the high-dimensional cancer microarray data. Experiments were conducted on seven publicly available and high-dimensional cancer microarray data sets. These microarray data sets consist of approximately 2000 to 15000 genes. The results from the experiments concluded that the developed algorithm, multi-objective cuckoo search with evolutionary operators outperforms cuckoo search and multi-objective cuckoo search algorithms with a smaller number of selected significant genes.