Improved adaptive coding learning for artificial bee colony algorithms

Recently, the artificial bee colony (ABC) algorithm has become increasingly popular in the field of evolutionary computing and manystate- of-the-art ABC variants (ABCs) have been developed. It has found that ABCs are optimal for separable problems, but suffer drastic performance losses for non-separable problems. Driven by this phenomenon, improved adaptive encoding learning (IAEL) has been integrated into ABCs (IAEL+ABCs) to enhance their performance for non-separable problems. In IAEL+ABCs, the cumulative population distribution information is utilized to establish an Eigen coordinate system that can effectively increase the improvement interval of variables, and thus make the population converge quickly in the early stage of evolution. In addition, a multivariable perturbation strategy serves as a supplementary method for reducing the risk of ABCs falling into local optima in complex multimodal non-separable problems. For comparison purposes, all experiments were conducted on CEC2014 competition benchmark suite. The experimental results show that the proposed IAEL+ABCs perform better than their corresponding ABCs and previously developed AEL+ABCs.