Self-adaptive pseudo-parallel differential evolution algorithm

被引：0

作者：

Xie, Datong ^{[1
,2
]}

Ding, Lixin ^{[1
]}

Du, Xin ^{[3
]}

Hu, Yurong ^{[1
]}

Wang, Shenwen ^{[1
]}

机构：

[1] State Key Lab. of Software Engineering, School of Computer, Wuhan University, Wuhan 430072, China

[2] Department of Information Management Engineering, Fujian Commercial College, Fuzhou 350000, China

[3] Faculty of Software, Fujian Normal University, Fuzhou 350000, China

来源：

Journal of Computational Information Systems | 2012年 / 8卷 / 08期

关键词：

Benchmarking - Optimization - Gaussian distribution - Evolutionary algorithms - Topology;

D O I：

暂无

中图分类号：

学科分类号：

摘要：

Differential evolution (DE) is a simple, fast and effective metaheuristic optimization algorithm. However, it is confronted with the difficulty to choose proper parameters to avoid trapping into local optima or stagnation. A self-adaptive pseudo-parallel differential evolution, SPDE, is proposed in this paper. The main idea of the algorithm consists of two aspects. First, it employs a self-adaptive mechanism which codes the two control parameters of DE into individuals, and makes parameters vary with Gaussian models based on population statistics information. Second, multi-population with ring topology where the best individuals or randomly selected individuals rotate is utilized to diversify the population. It is shown that SPDE is not only better than basic DE, CHC, G-CMA-ES, GODE and SOUPDE but also comparable to SaDE, jDElscop, GaDE on convergence performance by experiments on II high-dimensional benchmark problems. © 2012 Binary Information Press.

引用

页码：3403 / 3411