Explosion gravitation field algorithm with dust sampling for unconstrained optimization

Xuemei Hu, Lan Huang, Yan Wang (Corresponding Author), Wei Pang (Corresponding Author)

Research output: Contribution to journalArticle

Abstract

Gravitation Field Algorithm (GFA) is a novel optimization algorithm derived from the Solar Nebular Disk Model (SNDM) in astronomy and inspired by the formation process of planets. Although it has achieved good performance when solving many unconstrained optimization problems, which demonstrated its promising application potential in many real-world problems, GFA still has much room for improvement, especially when it comes to the accuracy and efficiency of the algorithm.

In this research, an improved GFA algorithm called Explosion Gravitation Field Algorithm (EGFA) is proposed for unconstrained optimization problems, with the introduction of two strategies: Dust Sampling (DS) and Explosion Operation. The task of DS is to locate the space that contains the optimal solution(s) by initializing the dust population randomly in the search space; while the Explosion Operator is to improve the accuracy of solutions and decrease the probability of the algorithm falling into local optima by generating the new population around the center dust to replace the original population.

A comparison of experimental results on six classical unconstrained benchmark problems with different dimensions demonstrates that the proposed EGFA outperforms the original GFA and several classical metaheuristic optimization algorithms, such as Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), in terms of accuracy and efficiency in lower dimensions. Additionally, the comparison of results on three real datasets indicate that EGFA performs better than the original GFA and k-means for solving clustering problems.
Original languageEnglish
Article number105500
JournalApplied Soft Computing
Volume81
Early online date17 May 2019
DOIs
Publication statusPublished - 31 Aug 2019

Fingerprint

Explosions
Dust
Gravitation
Sampling
Astronomy
Planets
Particle swarm optimization (PSO)
Genetic algorithms

Keywords

  • Explosion gravitation field algorithm
  • Unconstrained optimization
  • Dust Sampling
  • Explosion operation
  • ADJUSTED RAND

Cite this

Explosion gravitation field algorithm with dust sampling for unconstrained optimization. / Hu, Xuemei; Huang, Lan; Wang, Yan (Corresponding Author); Pang, Wei (Corresponding Author).

In: Applied Soft Computing, Vol. 81, 105500, 31.08.2019.

Research output: Contribution to journalArticle

@article{cab3794d53754e0bb12773547e1b79a7,
title = "Explosion gravitation field algorithm with dust sampling for unconstrained optimization",
abstract = "Gravitation Field Algorithm (GFA) is a novel optimization algorithm derived from the Solar Nebular Disk Model (SNDM) in astronomy and inspired by the formation process of planets. Although it has achieved good performance when solving many unconstrained optimization problems, which demonstrated its promising application potential in many real-world problems, GFA still has much room for improvement, especially when it comes to the accuracy and efficiency of the algorithm.In this research, an improved GFA algorithm called Explosion Gravitation Field Algorithm (EGFA) is proposed for unconstrained optimization problems, with the introduction of two strategies: Dust Sampling (DS) and Explosion Operation. The task of DS is to locate the space that contains the optimal solution(s) by initializing the dust population randomly in the search space; while the Explosion Operator is to improve the accuracy of solutions and decrease the probability of the algorithm falling into local optima by generating the new population around the center dust to replace the original population.A comparison of experimental results on six classical unconstrained benchmark problems with different dimensions demonstrates that the proposed EGFA outperforms the original GFA and several classical metaheuristic optimization algorithms, such as Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), in terms of accuracy and efficiency in lower dimensions. Additionally, the comparison of results on three real datasets indicate that EGFA performs better than the original GFA and k-means for solving clustering problems.",
keywords = "Explosion gravitation field algorithm, Unconstrained optimization, Dust Sampling, Explosion operation, ADJUSTED RAND",
author = "Xuemei Hu and Lan Huang and Yan Wang and Wei Pang",
note = "This research was funded by the National Natural Science Foundation of China (Nos. 61572227, 61772227, 61702214), the Development Project of Jilin Province of China (Nos 20170101006JC, 20180414012GH, 20170203002GX, 20190201293JC), Zhuhai Premier-Discipline Enhancement Scheme, China (Grant 2015YXXK02) and Guangdong Premier Key-Discipline Enhancement Scheme, China (Grant 2016GDYSZDXK036). This work was also supported by Jilin Provincial Key Laboratory of Big Date Intelligent Computing, China (No. 20180622002JC).",
year = "2019",
month = "8",
day = "31",
doi = "10.1016/j.asoc.2019.105500",
language = "English",
volume = "81",
journal = "Applied Soft Computing",
issn = "1568-4946",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - Explosion gravitation field algorithm with dust sampling for unconstrained optimization

AU - Hu, Xuemei

AU - Huang, Lan

AU - Wang, Yan

AU - Pang, Wei

N1 - This research was funded by the National Natural Science Foundation of China (Nos. 61572227, 61772227, 61702214), the Development Project of Jilin Province of China (Nos 20170101006JC, 20180414012GH, 20170203002GX, 20190201293JC), Zhuhai Premier-Discipline Enhancement Scheme, China (Grant 2015YXXK02) and Guangdong Premier Key-Discipline Enhancement Scheme, China (Grant 2016GDYSZDXK036). This work was also supported by Jilin Provincial Key Laboratory of Big Date Intelligent Computing, China (No. 20180622002JC).

PY - 2019/8/31

Y1 - 2019/8/31

N2 - Gravitation Field Algorithm (GFA) is a novel optimization algorithm derived from the Solar Nebular Disk Model (SNDM) in astronomy and inspired by the formation process of planets. Although it has achieved good performance when solving many unconstrained optimization problems, which demonstrated its promising application potential in many real-world problems, GFA still has much room for improvement, especially when it comes to the accuracy and efficiency of the algorithm.In this research, an improved GFA algorithm called Explosion Gravitation Field Algorithm (EGFA) is proposed for unconstrained optimization problems, with the introduction of two strategies: Dust Sampling (DS) and Explosion Operation. The task of DS is to locate the space that contains the optimal solution(s) by initializing the dust population randomly in the search space; while the Explosion Operator is to improve the accuracy of solutions and decrease the probability of the algorithm falling into local optima by generating the new population around the center dust to replace the original population.A comparison of experimental results on six classical unconstrained benchmark problems with different dimensions demonstrates that the proposed EGFA outperforms the original GFA and several classical metaheuristic optimization algorithms, such as Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), in terms of accuracy and efficiency in lower dimensions. Additionally, the comparison of results on three real datasets indicate that EGFA performs better than the original GFA and k-means for solving clustering problems.

AB - Gravitation Field Algorithm (GFA) is a novel optimization algorithm derived from the Solar Nebular Disk Model (SNDM) in astronomy and inspired by the formation process of planets. Although it has achieved good performance when solving many unconstrained optimization problems, which demonstrated its promising application potential in many real-world problems, GFA still has much room for improvement, especially when it comes to the accuracy and efficiency of the algorithm.In this research, an improved GFA algorithm called Explosion Gravitation Field Algorithm (EGFA) is proposed for unconstrained optimization problems, with the introduction of two strategies: Dust Sampling (DS) and Explosion Operation. The task of DS is to locate the space that contains the optimal solution(s) by initializing the dust population randomly in the search space; while the Explosion Operator is to improve the accuracy of solutions and decrease the probability of the algorithm falling into local optima by generating the new population around the center dust to replace the original population.A comparison of experimental results on six classical unconstrained benchmark problems with different dimensions demonstrates that the proposed EGFA outperforms the original GFA and several classical metaheuristic optimization algorithms, such as Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), in terms of accuracy and efficiency in lower dimensions. Additionally, the comparison of results on three real datasets indicate that EGFA performs better than the original GFA and k-means for solving clustering problems.

KW - Explosion gravitation field algorithm

KW - Unconstrained optimization

KW - Dust Sampling

KW - Explosion operation

KW - ADJUSTED RAND

UR - http://www.mendeley.com/research/explosion-gravitation-field-algorithm-dust-sampling-unconstrained-optimization

U2 - 10.1016/j.asoc.2019.105500

DO - 10.1016/j.asoc.2019.105500

M3 - Article

VL - 81

JO - Applied Soft Computing

JF - Applied Soft Computing

SN - 1568-4946

M1 - 105500

ER -