Graphic process units‐based chicken swarm optimization algorithm for function optimization problems

Abstract

This article focuses on how to design an efficient GPU-based chicken swarm optimization (CSO) algorithm (GCSO), so as to improve diversity and speed up convergence by running a large number of populations in parallel. GCSO mainly improves the sequential CSO in three aspects: (i) GCSO modifies the location updating equation of the rooster and proposes a parallel iterative strategy to transform the sequential iteration process into a parallel iterative process, thereby achieving fine-grained parallelism and improving the convergence speed. (ii) A multirange search strategy is proposed to build different neighborhoods for each flock on the graphic process units (GPU), so that each flock searched in their respective neighborhoods, thus increasing the density and diversity of the search, and making it not easy to fall into a local optimum. (iii) A new column storage structure is designed to meet the requirement of coalescent access on GPU. Twelve benchmark functions are selected to compare GCSO algorithm with some sequential intelligence optimization algorithms and the GPU-based particle swarm algorithm. The results show that the GCSO is able to obtain a speedup up to 163.09× compared with the CSO and achieve better optimization results in terms of both optimization accuracy and convergence speed than some intelligence optimization algorithms.

An Improved Chicken Swarm Optimization Algorithm Based on Adaptive Mutation Learning Strategy

“…And it was applied to the human-computer interaction. Lin et al [34] improved location update formula of rooster and transformed the sequential iteration process into a parallel iterative process, thereby improving the convergence speed. And it applied to optimize image processing unit.…”

An advanced hybrid algorithm for nonlinear function optimization with real world applications

“…) and rand (0, 1) < p u t i,j ; otherwise (9) where f(⋅): fitness function values and p: random value in (0, 1].…”

“…1 Presently, a bunch of optimization methods known as meta-heuristics algorithms (MAs) has been introduced to solve COPs, to overcome drawbacks of traditional optimization methods. According to mechanical differences the MAs can be categorized into four groups as follows: swarm intelligence algorithms (SIAs): inspired from behavior of social insects or animals like particle swarm optimization (PSO), 2 artificial bee colony (ABC), 3,4 animal migration optimization (AMO), 5 whale optimization algorithm (WOA), 6,7 social spider optimization (SSO), 8 chicken swarm optimization (CSO), 9 wind driven dragonfly algorithm (WDDA), 10 firefly algorithm (FA), 11 and so forth; evolutionary algorithms (EAs)-inspired from biology such as differential evolution (DE), 12 genetic algorithm (GA), 13 and so forth; physics based algorithms (PBAs): inspired by the rules governing a natural phenomenon like harmony search (HS), 14 gravitational search algorithm (GSA), 15 and so forth and human behavior based algorithms (HBAs): inspired from the human being like teaching-learning-based optimization (TLBO), 16 gaining sharing knowledge based Algorithm (GSK), 17 and so forth.…”

PECSO: An Improved Chicken Swarm Optimization Algorithm with Performance-Enhanced Strategy and Its Application

“…Yu et al [ 21 ] proposed a hybrid localization scheme for mine monitoring using a CSO algorithm and wheel graph, which minimized the inter-cluster complexity and improved the localization accuracy. Lin et al [ 22 ] designed a CSO algorithm (GCSO) based on a high-efficiency graphics processing unit, which increased the population diversity and accelerated the convergence speed through parallel operations.…”