Aleš Zamuda (Author), Janez Brest (Author), Borko Bošković (Author), Viljem Žumer (Author)

Abstract

V članku predstavljamo študijo samoprilagodljivih krmilnih parametrov algoritma diferencialne evolucije za večkriterijsko optimizacijo, ki ga krmili samoprilagoditveni mehanizem, predstavljen v evolucijskih strategijah. Samoprilagajanje parametrov omogoča danemu evolucijskemu algoritmu učinkovitejše iskanje, saj se algoritem lahko prilagodi optimizacijskemu problemu, ki ga rešuje. Z eksperimentom prikažemo dejanske vrednosti in spreminjanje samoprilagodljivih krmilnih parametrov na znanih testnih funkcijah.

Keywords

evolucijsko računanje;diferencialna evolucija;večkriterijska optimizacija;samoprilagoditev;algoritmi;

Data

Language: Slovenian
Year of publishing:
Typology: 1.01 - Original Scientific Article
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: Elektrotehniška zveza Slovenije
UDC: 004.89.021
COBISS: 12933654 Link will open in a new window
ISSN: 0013-5852
Views: 766
Downloads: 29
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: A study in self-adaptation of control parameters in the DEMOwSA algorithm
Secondary abstract: In this paper we present an experimental analysis showing that the self-adaptation of control parameters plays an important role in the multiobjective optimization process (refer to Figure I for notion of multiobjective optimality). Experimental results of a self-adaptive differential evolution algorithm are evaluated on the set of benchmark functions provided for the CEC 2007 Special session on Performance Assessment & Competition on Multi-objective Optimization Algorithms, as seen in Tables 2-7. Self-adaptation is proven to statistically outperform fixed parameters, using t-test on the empirical results in these tables. The values of control parameters are encoded in each individual (see Figure 2) and changed during the optimization process. They depend on the nature of the problem being solved, as can be seen in Table I and Figures 3 and 4 which show how using self-adaptation good control parameters are obtained to improve the search results.
Secondary keywords: evolutionary computation;differential evolution;multi-objective optimization;self-adaptation;
URN: URN:NBN:SI
Type (COBISS): Not categorized
Pages: str. 223-228
Volume: ǂLetn. ǂ75
Issue: ǂšt. ǂ4
Chronology: 2008
ID: 1735580