An Enhancement of Bees Algorithm Using Probabilistic Model

The behavior of bee swarms foraging in nature inspires the bees algorithm (BA). Even though the bees algorithm has already undergone several enhancements over the past decade, there is still room for enhancement to increase the convergence speed and prevent the algorithm from getting trapped in the local optimum. Therefore, this paper proposes an enhancement inspired by estimation distribution algorithms (EDAs) by applying the statistical method known as probabilistic modeling. The proposed enhancement aims to gradually improve the performance of global search with new additions. In this paper, three types of distribution are proposed: the bees algorithm with normal distribution (BAND), the bees algorithm with triangular distribution (BATD), and finally the bees algorithm with multivariate normal distribution (BAMVD). Subsequently, the enhanced BA was tested with continuous benchmark functions and compared with the standard BA version. The result showed that the enhanced BA searches faster and more precisely than the standard BA in most functions.