A comparative study of the Bees Algorithm as a tool for function optimisation

The Bees Algorithm is a parameter optimisation method that mimics the foraging behaviour of honey bees. This paper presents an experimental study of the performance of the Bees Algorithm. Its strengths and weaknesses are analysed, and the most suitable parameterizations in relation to different optimisation tasks are revealed. The robustness of the optimisation results to reasonable modifications of the fitness landscape is studied for a large set of parameterizations. The Bees Algorithm is tested on 18 custom-made function minimisation benchmarks, and its performance compared to that of two state-of-the-art biologically inspired optimisation methods. Thanks to their two-dimensional nature, the proposed fitness landscapes are easy to visualise. Experimental evidence indicates also that they constitute a varied and challenging set of test cases, useful to reveal the specific abilities and biases of the search algorithms. In addition, the performance of the Bees Algorithm and the other two optimisation methods is tested on four real-world minimisation tasks from the literature. The results obtained on the benchmarks prove the effectiveness and robustness of the bees foraging metaphor, in particular on the most deceptive and high-dimensional fitness landscapes. They also confirm the ability of the Bees Algorithm to solve complex real-world optimisation tasks.