Performance evaluation of artificial bee colony optimization and new selection schemes

Abstract

The artificial bee colony optimization (ABC) is a population-based algorithm for function optimization that is inspired by the foraging behavior of bees. The population consists of two types of artificial bees: employed bees (EBs) which scout for new, good solutions and onlooker bees (OBs) that search in the neighborhood of solutions found by the EBs. In this paper we study in detail the influence of ABC's parameters on its optimization behavior. It is also investigated whether the use of OBs is always advantageous. Moreover, we propose two new variants of ABC which use new methods for the position update of the artificial bees. Extensive empirical tests were performed to compare the new variants with the standard ABC and several other metaheuristics on a set of benchmark functions. Our findings show that the ideal parameter values depend on the hardness of the optimization goal and that the standard values suggested in the literature should be applied with care. Moreover, it is shown that in some situations it is advantageous to use OBs but in others it is not. In addition, a potential problem of the ABC is identified, namely that it performs worse on many functions when the optimum is not located at the center of the search space.Finally it is shown that the new ABC variants improve the algorithm's performance and achieve very good performance in comparison to other metaheuristics under standard as well as hard optimization goals.

Population-based heuristic algorithms for continuous and mixed discrete-continuous optimization problems

“…In particular, it was found to be relatively poor performing on composite and non-separable function as well as having a slow convergence rate towards high quality solutions [Akay and Karaboga, 2012]. Therefore, in the following years, a number of modifications of the original ABC algorithm were introduced trying to improve performance [Alataş, 2010, Aydın et al, 2012, Banharnsakun et al, 2011, Diwold et al, 2011a, Gao and Liu, 2011, Kang et al, 2011, Zhu and Kwong, 2010. Unfortunately, so far there is no comprehensive comparative evaluation of the performance of ABC variants on a significantly large benchmark set available.…”

“…where x gbest,j is the j-th element of the global-best solution, ψ i,j is a random number drawn uniformly at random in [0, C] where C is a constant that is set to one by Diwold et al [2011a] or used as a parameter to be set by Zhu and Kwong [2010]. This modification is inspired by the usage of the global-best solution to influence particles in particle swarm optimization; it is a rather straightforward modification that, as we will see later, is important to obtain a significantly improved performance.…”

“…Diwold et al [2011a] have also proposed a second modification to the ABC algorithm, which uses the global-best solution as in the GbABC variant (with C = 1) but which additionally changes the selection of the neighboring solution x k in Equation E.6. Let d(x i , x k ) be the Euclidean distance between two solutions x i and x k .…”

“…Thus, the closer a solution x k is to x i , the higher is the probability of choosing it. The intuition of this choice is that it is more probable to fine a better solution by searching between two good solutions that are probably also close to each other in the solutions space [Diwold et al, 2011a].…”

“…Already in earlier papers, the impact of parameter settings on the performance of ABC was studied , Diwold et al, 2011a, Karaboga and Basturk, 2008. Conclusions were, for example, that a too low value of the limit parameter leads to poor performance, that ABC's performance is relatively robust given the limit value as determined through parameter lf is large enough Karaboga, 2009, Diwold et al, 2011a], and that the population size does not need to be fine-tuned in order to obtain satisfactorily good results .…”