One Parameter Differential Evolution(OPDE)for Numerical Benchmark Problems

摘要：

　　Differential Evolution (DE) can be simplified in the sense that the number of existing parameter is decreased from two parameters to only one parameter.We eliminate the scaling factor, F, and replace this by a uniform random number within 0, 1.As such, it is easy to tune the crossover rate, CR, through parameter sensitivity analysis.In this analysis, the algorithm is run for 50 trials from 0.1 to 1.0 with a step increment of 0.1 on 23 benchmark problems.Results show that using the optimal CR, there is room for improvement in some of the benchmark problems.With the advantage and simplicity of a single parameter, it is significantly easier to tune this parameter and thus take the full advantage of the algorithm.The proposed algorithm here has a significant benefit when applied to real-world problems as it saves time in obtaining the best parameter setting for optimal performance.

关键词： Benchmark problems one parameter differential evolution parameter sensitivity analysis

作者: Y.Kang T.O.Ting Xin-She Yang Shi Cheng

作者单位: Department of Electrical and Electronic Engineering, Xian Jiaotong-Liverpool University, Suzhou, Ji School of Science and Technology, Middlesex University Hendon Campus, London, UK Department of Electrical and Electronic Engineering, Xian Jiaotong-Liverpool University, Suzhou, Ji

会议类型: 国际会议

会议名称: 4th international Conference,ICSI2013(第4届群体智能国际会议)

会议地点: 哈尔滨

会议语种:英文

页码: 431-438

在线出版日期: 2013-06-12（万方平台首次上网日期，不代表论文的发表时间）