An improved artificial bee colony and its application

Inspired by the intelligence of the foraging behavior of honeybee swarms, the artificial bee colony (ABC) algorithm has shown competitive performance among evolutionary algorithms. However, despite its strong global search ability, its slow convergence speed remains a weakness. To compensate for this shortcoming and further improve the standard ABC, based on a novel chaotic-based operator and a new neighbor selection strategy, our paper proposes a novel update equation and an improved dimension-selection strategy for employed bees to strike a good balance between global search and local tuning abilities. To evaluate its performance in both theoretical and practical problems, a series of benchmark functions and two real-world applications (Lennard-Jones potential problem and feature selection problem) are used for testing in our paper. Experimental results demonstrate that our proposed strategies help the improved ABC achieve more favorable results than the compared algorithms.