Comparative Study of Artificial Bee Colony Algorithm and Real Coded Genetic Algorithm for Analysing Their Performances and Development of a New Algorithmic Framework

This paper compares performance of the artificial bee colony algorithm (ABC) and the real coded genetic algorithm (RCGA) on a suite of 9 standard benchmark problems. The problem suite comprises a diverse set of unimodal, multimodal and rotated multimodal numerical optimization functions and the comparison criteria include (i) solution quality, (ii) convergence speed, (iii) robustness, and (iv) scalability to test efficacy of the algorithms. To the best knowledge of the authors, such a comprehensive comparative study of the two algorithms is not available in the literature. An empirical study shows that the RCGA has advantages over the ABC in terms of all the criteria for the unimodal and the rotated multimodal functions. On other hand, the ABC outperforms the RCGA in terms of solution quality for the multimodal functions. Therefore, based on the insights gained out of this comparative study, the authors propose an algorithm ABC-GA with new algorithmic framework that comprises advantages of both the ABC and the GA. An empirical study of the proposed algorithm ABC-GA shows its promising performance as the obtained results are superior to both the comparative algorithms for all the problems in all the criteria.