PCSVM: A Hybrid Approach using Particle Swarm Optimization and Cuckoo Search for Effective Cancer Diagnosis

Today, cancer is increasingly recognized as a significant contributor to the rising death toll around the globe. Therefore, early identification of cancer raises the degree to which patients may recover, and one of the main concepts of this approach is known as machine learning (ML). For ML the important factor is the presence of a proper dataset and biopsy and microarray datasets are two varieties of datasets available to help in developing ML-based models. The biopsy dataset does not have any genetic information and due to this limitation present behind the biopsy dataset, the researchers are looking at the microarray data. The microarray data are used for the diagnosis and classification of cancer disease which makes the data colossal. But to examine a large number of datasets is a challenging task. To avoid this kind of issue feature selection is one of the best solution and classification algorithms are present in machine learning which selects the relevant features that help in constructing a better model for classification. As a result of this better accuracy is obtained in disease classification which helps in prevention. The primary objective of this research work is to provide an integrated model PCSVM or PSO-CS-SVM based on Particle swarm optimization (PSO), and cuckoo search algorithm (CS) algorithm used as feature selection and optimization algorithm respectively. In addition, the support vector machine (SVM) is being used as the classifier. The state-of-the-art approaches to machine learning, such as Decision Tree, Neural Network, and Random Forest, are used to compare the performance of the proposed PSO-CS-SVM.