Nowadays, due to the growing dimensionality of optimisation problems, numerous studies are dedicated to reduction of metaheuristics computational requirements. Reducing size of the population during optimisation process is one of the promising research trends in the field of Evolutionary Algorithms. The purpose of this paper is to clarify the subject in form of a survey of population size reduction methods already proposed and to present preliminary results of a new method based on the clustering technique. Introduced method was implemented in the framework of Differential Evolution algorithm and verified on a set of real-parameter benchmark functions.

Nowadays, due to the growing dimensionality of optimisation problems, numerous studies are dedicated to reduction of metaheuristics computational requirements. Reducing size of the population during optimisation process is one of the promising research trends in the field of Evolutionary Algorithms. The purpose of this paper is to clarify the subject in form of a survey of population size reduction methods already proposed and to present preliminary results of a new method based on the clustering technique. Introduced method was implemented in the framework of Differential Evolution algorithm and verified on a set of real-parameter benchmark functions.