A self-adaptive differential evolutionary algorithm based on population reduction with minimum distance. (1st January 2014)

• Record Type: Journal Article
• Authors: Yang, Ming; Guan, Jing; Cai, Zhihua; Li, Changhe
• Is Part Of: International journal of innovative computing and applications. Volume 6:Number 1(2014)
• Page Start: 13
• Page End: 24
• Publication Date: 2014-01-01
• Subjects: differential evolution -- self-adaptation -- population reduction -- difference vector; Evolutionary computation -- Periodicals; Neural networks (Computer science) -- Periodicals; Genetic programming (Computer science) -- Periodicals; Biologically-inspired computing -- Periodicals; Swarm intelligence -- Periodicals; Quantum computers -- Periodicals; 006.3
• Languages: English
• ISSNs: 1751-648X

Abstract

In differential evolution (DE), many adaptive algorithms have been proposed for parameter adaptation. However, they mainly focus on tuning the mutation factor F and crossover probability CR . The adaptation of population size NP has not been widely studied in the scope of DE. Reducing population size could save computational resources and hence accelerate the convergence speed of algorithms. This is beneficial for algorithms to solve the optimisation problems which need expensive evaluations. However, population reduction may weaken the population diversity, hence, it will result in population premature. In this paper, we propose a novel population reduction method for jDE algorithm, called dynNPMinD-jDE. When the population reduction criterion is satisfied, dynNPMinD-jDE selects the best individual and the pair individuals with minimal-step difference vectors to form a new population. To enhance the population diversity, dynNPMinD-jDE adopts the adaptive mechanisms of F and CR . dynNPMinDjDE is tested on a set of 38 scalable benchmark functions. The results show that dynNPMinD-jDE can get better results on most functions, and the convergence speed becomes faster and faster as each population reduction.