Novel Bees Algorithm: Stochastic self-adaptive neighborhood

摘要

Several algorithms inspired in recent years by the swarm behavior of honeybees have been developed for a variety of practical applications. The Bees Algorithm (BA) is one of these swarm-based algorithms that imitate the intelligent behaviors of honeybees. The present paper proposes a Novel Bees Algorithm (NBA) that uses a stochastic self-adaptive neighborhood (ssngh) search to improve the original BA. The ssngh automatically and dynamically reflects swarm convergence conditions and frees its settings. Additionally, this paper tests two additional designs for bee relocation as well as the effect on algorithm performance of using fewer recruited bees. Experimental results are compared using 23 benchmark functions. Results demonstrate that the proposed NBA not only frees the parameter settings of the neighborhood ranges of BA but also significantly improves upon the convergence performance of the original BA. Additionally, experimental results indicate that the NBA outperforms the artificial bee colony (ABC) algorithm on 12 benchmark functions, while the ABC outperforms the NBA on only 8 benchmark functions.