2018
Salcedo-Sanz, Sancho; Aybar-Ruíz, A.; Camacho-Gómez, C.; Pereira, E.
Efficient fractal-based mutation in evolutionary algorithms from iterated function systems Journal Article
In: Communications in Nonlinear Science and Numerical Simulation, vol. 56, pp. 434-446, 2018, ISSN: 1007-5704.
Abstract | Links | BibTeX | Tags: Evolutionary programming, Fractals, Iterated function systems, Mutation procedures design
@article{SALCEDOSANZ2018434,
title = {Efficient fractal-based mutation in evolutionary algorithms from iterated function systems},
author = {Sancho Salcedo-Sanz and A. Aybar-Ruíz and C. Camacho-Gómez and E. Pereira},
url = {https://www.sciencedirect.com/science/article/pii/S1007570417302915},
doi = {https://doi.org/10.1016/j.cnsns.2017.08.010},
issn = {1007-5704},
year = {2018},
date = {2018-01-01},
urldate = {2018-01-01},
journal = {Communications in Nonlinear Science and Numerical Simulation},
volume = {56},
pages = {434-446},
abstract = {In this paper we present a new mutation procedure for Evolutionary Programming (EP) approaches, based on Iterated Function Systems (IFSs). The new mutation procedure proposed consists of considering a set of IFS which are able to generate fractal structures in a two-dimensional phase space, and use them to modify a current individual of the EP algorithm, instead of using random numbers from different probability density functions. We test this new proposal in a set of benchmark functions for continuous optimization problems. In this case, we compare the proposed mutation against classical Evolutionary Programming approaches, with mutations based on Gaussian, Cauchy and chaotic maps. We also include a discussion on the IFS-based mutation in a real application of Tuned Mass Dumper (TMD) location and optimization for vibration cancellation in buildings. In both practical cases, the proposed EP with the IFS-based mutation obtained extremely competitive results compared to alternative classical mutation operators.},
keywords = {Evolutionary programming, Fractals, Iterated function systems, Mutation procedures design},
pubstate = {published},
tppubtype = {article}
}
In this paper we present a new mutation procedure for Evolutionary Programming (EP) approaches, based on Iterated Function Systems (IFSs). The new mutation procedure proposed consists of considering a set of IFS which are able to generate fractal structures in a two-dimensional phase space, and use them to modify a current individual of the EP algorithm, instead of using random numbers from different probability density functions. We test this new proposal in a set of benchmark functions for continuous optimization problems. In this case, we compare the proposed mutation against classical Evolutionary Programming approaches, with mutations based on Gaussian, Cauchy and chaotic maps. We also include a discussion on the IFS-based mutation in a real application of Tuned Mass Dumper (TMD) location and optimization for vibration cancellation in buildings. In both practical cases, the proposed EP with the IFS-based mutation obtained extremely competitive results compared to alternative classical mutation operators.