An artificial bee colony algorithm with adaptive heterogeneous competition for global optimization problems

Artificial bee colony (ABC) algorithm is an efficient bio-inspired optimizer proposed recently. Though it has gained great popularity, ABC suffers from its slow convergence and poor generalization on various problem landscapes. To address the issues, an augmented ABC with adaptive heterogeneous competition (ABC-AHC) is proposed in this study. In ABC-AHC, two bee swarms with each conducting heterogeneous but complementary capabilities are implemented to improve the search capabilities on various problem spaces. To dynamically adjust the search behaviors, an adaptive mechanism is developed to trigger the competition and migration between the bee swarms. Comparative studies are conducted for parameter tuning and the heterogeneous searching (HST). Existing algorithms including ABC variants and non-ABC variants are adopted to validate the performance of ABC-AHC. Numerical comparisons are conducted on 30D and 100D benchmark functions, CEC 2014 test function, random function and the real-world problems. Numerical results demonstrate that the proposed strategies significantly enhance ABC's search capability and convergence speed on the various benchmark functions.