Hybrid evolutionary programming using adaptive Lévy mutation and modified Nelder–Mead method

Pang, J, He, J ORCID: https://orcid.org/0000-0002-5616-4691 and Dong, H, 2018. Hybrid evolutionary programming using adaptive Lévy mutation and modified Nelder–Mead method. Soft Computing. ISSN 1432-7643

Preview

Text 11928_He.pdf - Post-print Download (373kB) | Preview

Abstract

Evolutionary programming has been widely applied to solve global optimization problems. Its performance is related to both mutation operators and fitness landscapes. In order to make evolutionary programming more efficient, its mutation operator should adapt to fitness landscapes. The paper presents novel hybrid evolutionary programming with adaptive Lévy mutation, in which the shape parameter of Lévy probability distribution adapts to the roughness of local fitness landscapes. Furthermore, a modified Nelder-Mead method is added to evolutionary programming for enhancing its exploitation ability. The proposed algorithm is tested on 39 selected benchmark functions and also benchmark functions in CEC2005 and CEC2017. The experimental results demonstrate that the overall performance of the proposed algorithm is better than other algorithms in terms of the solution accuracy.

Item Type:	Journal article
Publication Title:	Soft Computing
Creators:	Pang, J., He, J. and Dong, H.
Publisher:	Springer
Date:	2 August 2018
ISSN:	1432-7643
Identifiers:	Number Type 10.1007/s00500-018-3422-4 DOI 3422 Publisher Item Identifier
Divisions:	Schools > School of Science and Technology
Record created by:	Jill Tomkinson
Date Added:	14 Sep 2018 10:10
Last Modified:	02 Aug 2019 03:00
URI:	https://irep.ntu.ac.uk/id/eprint/34490

Actions (login required)

Edit View

Statistics

Views

Downloads