Title: An improved firefly algorithm for numerical optimisation
Authors: Xiangqin Xiang
Addresses: Laboratory of Network and Intelligent Information Management, Hefei University, Hefei 230601, China
Abstract: Firefly algorithm (FA) is a recently proposed meta-heuristic optimisation technique, which has shown good performance on many optimisation problems. In the original FA, each firefly is attracted by any other brighter firefly (better fitness value). By the attraction, fireflies maybe moved to better positions. However, the attraction does not guarantee whether a firefly is moved to a better position. Sometimes, the attraction may move a firefly to a worse position. Therefore, the search of firefly is oscillated during the evolution. In this paper, we present an improved firefly algorithm (IFA), which employs a greedy selection method to guarantee that a firefly is not moved to worse positions. To verify the performance of IFA, a set of well-known benchmark functions are used in the experiments. Experimental results show that the IFA achieves better results than the original FA.
Keywords: firefly algorithm; swarm intelligence; numerical optimisation; global optimisation; greedy selection.
DOI: 10.1504/IJCSM.2015.069466
International Journal of Computing Science and Mathematics, 2015 Vol.6 No.2, pp.201 - 210
Received: 17 Sep 2014
Accepted: 12 Nov 2014
Published online: 17 May 2015 *