C-HDESHO: Cancer Classification Framework using Single Objective Meta—heuristic and Machine learning Approaches

Abstract

Microarray gene expression data holds the potential for diagnosis and prognosis of various genetic diseases. It is also used extensively in designing cancer classification techniques. But the enormity of genomic features and the lesser number of samples data make cancer classification a tedious task. This paper presents a novel hybrid metaheuristic optimization algorithm which is based on Differential Evolution (DE) and recently developed Spotted Hyena Optimizer (SHO) named as Hybrid Differential Evolutions and Spotted Hyena Optimizer (HDESHO) for cancer classification. The main contribution of this algorithm is to improve the mutation strategy of differential evolution using the spotted hyena optimizer algorithm. After the initial gene selection different machine learning algorithms were employed for performing cancer classification. The results state that the proposed approach outperforms as compared to the method discussed in the literature.