Title:

Application of multiple criteria evolutionary algorithms to vector optimisation, decision support and reference point approaches, Journal of Telecommunications and Information Technology, 2003, nr 3

Creator:

Wierzbicki, Andrzej P. ; Szczepanski, Marcin

Subject and Keywords:

vector optimisation ; evolutionary algorithms ; reference point techniques ; nadirpoint estimation

Description:

Multiple criteria evolutionary algorithms, being essentially parallel in their character, are a natural instrument of finding a representation of entire Pareto set (set of solutions and outcomes non-dominated in criteria space) for vector optimization problems. However, it is well known that Pareto sets for problems with more than two criteria might become complicated and their representation very time-consuming. Thus, the application of such algorithms is essentially limited to bi-criteria problems or to vector optimisation problems with more criteria but of simple structure. Even in such cases, there are problems related to various important aspects of vector optimisation, such as the uniformity of representation of Pareto set, stopping tests or the accuracy of representing Pareto set, that are not fully covered by the broad literature on evolutionary algorithms in vector optimisation. These problems and related computational tests and experience are discussed in the paper. In order to apply evolutionary algorithms for decision support, it would be helpful to use them in an interactive mode. However, evolutionary algorithms are in their essence global and of batch type. Nevertheless, it is possible to introduce interactive aspects to evolutionary algorithms by focusing them on a part of Pareto set. The results of experimental tests of such modifications of evolutionary algorithms for vector optimisation are presented in the paper. Another issue related to vector optimisation problems with more than two criteria is the computational difficulty of estimating nadir points of Pareto set. The paper describes the use of diverse variants of evolutionary algorithms to the estimation of nadir points, together with experimental evidence.

Publisher:

Instytut Łączności - Państwowy Instytut Badawczy, Warszawa

Date:

2003, nr 3

Resource Type:

artykuł

Format:

application/pdf

Resource Identifier:

ISSN 1509-4553, on-line: ISSN 1899-8852

DOI:

10.26636/jtit.2003.3.194

ISSN:

1509-4553

eISSN:

1899-8852

Source:

Journal of Telecommunications and Information Technology

Language:

ang

Rights Management:

Biblioteka Naukowa Instytutu Łączności