BIBHU: Biomarker Identification Using Bio-inspired Evolutionary Hybrid Unique Machine Learning Model

The Firefly algorithm (FA) is one of the most efficient optimization algorithms proposed according to the attraction and flashing characteristics of the firefly. FA generated the population randomly and considered them as fireflies. Premature convergence with poor global exploration is the main issue that still exists with the FA while dealing with high-dimensional datasets and complex problems. To avoid this issue, a hybrid machine learning model named SVGA-RCBFA is proposed. In this model, a combination of self-variant genetic algorithm (SVGA) is hybridized with a return cost-based FA (RCBFA) metaheuristic algorithm proposed to achieve the advantage of both algorithms with microarray cancer datasets. In this hybrid approach, individuals in the new generation are generated through SVGA processes and RCBFA mechanisms to prevent local optima, promote sufficient diversity of solutions, and balance exploration and exploitation tendencies. The proposed hybrid model BIBHU is tested with different cancer microarray datasets, the outcome validates the superiority of this model comparison with classical GA-FA. At the end of the study, a fair comparison is carried out with existing models relevant to the same dataset.