On Stability and Classification Tools for Genetic Algorithms

Issue title: Advances in Artificial Intelligence and Applications

Article type: Research Article

Authors: Kotowski, Stefan | Kosiński, Witold | Michalewicz, Zbigniew | Synak, Piotr | Brocki, Łukasz

Affiliations: Faculty of Computer Science, Polish-Japanese Institute of Information Technology ul. Koszykowa 86, 02-008 Warsaw, Poland. E-mail: [email protected]; [email protected] | School of Computer Science, University of Adelaide Adelaide, SA 5005, Australia. E-mail: [email protected] | Faculty of Computer Science, Polish-Japanese Institute of Information Technology ul. Koszykowa 86, 02-008 Warsaw, Poland. E-mail: [email protected]; [email protected]

Abstract: Convergence of genetic algorithms in the form of asymptotic stability requirements is discussed. Some tools to measure convergence properties of genetic algorithms are introduced. A classification procedure is proposed that is based on the following conjecture: the entropy and the fractal dimension of trajectories of genetic algorithms produced by them are quantities that can characterize the algorithms. The role of these quantities as invariants of the algorithm classes is discussed together with the compression ratio of points of genetic algorithms.

Keywords: Genetic Algorithm, Classification, Convergence

DOI: 10.3233/FI-2009-189

Journal: Fundamenta Informaticae, vol. 96, no. 4, pp. 477-491, 2009