Enhancing firefly algorithm with courtship learning

Hu Peng,Wenhua Zhu,Changshou Deng,Zhijian Wu

DOI: https://doi.org/10.1016/j.ins.2020.05.111

IF: 8.1

2021-01-01

Information Sciences

Abstract:The firefly algorithm (FA) is a nature-inspired heuristic optimization algorithm based on the luminescence and attraction behavior of fireflies. Although the FA can effectively solve complex optimization problems, it suffers from premature convergence because of its simple full attraction model. However, in nature, fireflies exhibit luminous behavior to attract mates. The attractiveness between fireflies of opposite sexes depends not only on the light intensity they emit, but also on their individual size, location, and other factors. In the original FA, all fireflies are assumed to be the same, i.e., they have no gender-based difference, which is not true biologically. Therefore, in this paper, we proposed a novel courtship learning (CL) framework to enhance the performance of the FA. In the proposed framework, the population is divided into female and male subpopulations. The female archiving mechanism is adopted to select excellent fireflies, which are assumed to be female fireflies. When the selected male firefly emits light that is less bright than that of the current firefly, a female individual will be selected from the female archive to guide the movement of the selected male firefly. Comprehensive experiments are conducted on the CEC 2013 benchmark set and the proposed CL framework is integrated with other advanced FA variants to verify its effects. Experimental results confirm that the proposed framework significantly enhances the performance of the original FA and advanced FA variants.

computer science, information systems