Grouping and Reflection of the Artificial Bee Colony Algorithm for High-Dimensional Numerical Optimization Problems

The artificial bee colony (ABC) algorithm, inspired by the cooperative foraging behaviors observed in bees, is a prominent example of a swarm intelligence algorithm that offers significant advantages in optimization problems. However, the efficacy of the ABC algorithm is limited in high-dimensional scenarios or when handling multimodal functions, which contain many local optima because of the random nature of the one-dimensional search process for improving the position in the employed and onlooker bee phases. As a result, ABC has a limited ability to obtain the optimum result (slow convergence rate). To address this limitation, this research introduces Grouping and Reflection of the Artificial Bee Colony (GRABC), a distinctive adaptation of the traditional ABC algorithm meticulously tailored to meet the specific demands of high-dimensional numerical optimization problems by balancing exploration and exploitation processes. GRABC strategically incorporates vector reflection and inertial weighting to formulate equations vleft and vright, which enhance both the employed and onlooker bee phases, substantially improving the convergence speed and improving the exploitation process. Moreover, the integration of grouping bees facilitates the exploration of food sources by promoting diversification and improving the exploration process. Additionally, an equation is derived to accurately compute the new positions of scout bees (exploration process), accounting for the possibility of becoming stuck in local optima and considering the proper limit values. The effectiveness of GRABC is thoroughly evaluated using 32 numerical benchmark functions, mostly including CEC 2017, which encompasses 100 dimensions. The empirical findings compellingly demonstrate that the GRABC algorithm outperforms alternative methodologies in terms of solution quality and convergence characteristics, as substantiated by comprehensive assessments that include metrics such as the worst, best, and average results and standard deviations.