An attraction basin estimating genetic algorithm for multimodal optimization
Zhuoran Xu, Mikko Polojärvi, Masahito Yamamoto, Masashi Furukawa · 2013
Radii-based niching evolutionary algorithms are criticized for the difficulty of the proper choice of the radius parameter. Detect-multimodal method enables the identification of niches without an explicit user-defined radius parameter. Although robust, the detect-multimodal method based algorithms are computationally expensive. We propose a novel algorithm called Attraction Basin Estimating Genetic Algorithm (ABE), which estimates the radius parameter based on the detect-multimodal method and uses the estimated radius to identify niches. Our experiments demonstrate that ABE has a similar ability to solve the multimodal optimization problem as Topological Species Conservation algorithm which is based on the detect-multimodal method, but much more efficiently.