Empirical Investigation of Multiparent Recombination Operators in Evolution Strategies

Evolutionary Computation - Tập 5 Số 3 - Trang 347-365 - 1997
A. E. Eiben1, Thomas Bäck2
1Department of Mathematics and Computer Science Leiden University Niels Bohrweg 1 NL-2333 CA, Leiden The Netherlands [email protected]#TAB#
2Informatik Centrum Dortmund Center for Applied Systems Analysis Joseph-von-Fraunhofer-Strasse 20 D-44227 Dortmund, Germany and Department of Mathematics and Computer Science Leiden University Niels Bohrweg 1 NL-2333 CA, Leiden The Netherlands

Tóm tắt

An extension of evolution strategies to multiparent recombination involving a variable number ϱ of parents to create an offspring individual is proposed. The extension is experimentally evaluated on a test suite of functions differing in their modality and separability and the regular/irregular arrangement of their local optima. Multiparent diagonal crossover and uniform scanning crossover and a multiparent version of intermediary recombination are considered in the experiments. The performance of the algorithm is observed to depend on the particular combination of recombination operator and objective function. In most of the cases a significant increase in performance is observed as the number of parents increases. However, there might also be no significant impact of recombination at all, and for one of the unimodal objective functions, the performance is observed to deteriorate over the course of evolution for certain choices of the recombination operator and the number of parents. Additional experiments with a skewed initialization of the population clarify that intermediary recombination does not cause a search bias toward the origin of the coordinate system in the case of domains of variables that are symmetric around zero.

Từ khóa


Tài liệu tham khảo

Back, T. (1996). Evolutionary algorithms in theory and practice. New York: Oxford University Press.

Back, T., Fogel, D. & Michalewicz, Z. (Eds.) (1997). Handbook of evolutionary computation. New York: Oxford University Press.

Back, T, & Michalewicz, 2. (1997). Test landscapes. In T. Back, D. Fogel, & Z. Michalewicz (Eds.), Handbook of evolutionary computation (pp. B2.7: 14-B2.7:20). New York: Oxford University Press.

Back, T, & Schwefel, H.P. (1993). An overview ofevolutionary algorithms for parameter optimization. Evolutionary Computation, 1 (I), 1-2 3.

Belew, K. & Booker, L. (Eds.) (1991). Proceedings of the Foziith Oiteirzntional Conference on Genetic :llgorithms. San Illateo, Ch: Morgan Kaufmann.

Bersini, H. Dorigo, AI., Langerman, S., Seront, G. & Gambardella, L. (1996). Results of the First International Contest on Evolutionary Optimization. In Proceedings ofthe Third IEEE Conference o n Eyoliitioimiy Compritatioii (pp. 61 1-61 5 ) . Piscataway, NJ: IEEE Press.

Toward LIP, Compiitntron, 3, 1

Bremermann, H., Kogson, hl. & Salaff, S. (1966). Global properties of evolution processes. In H. Pattee, E. Edlsack, L. Fein, & A. Callahan, (Eds.),\ktwnlnzrtomtita nizd use~dsimdations (pp. 3-41). Washington, DC: Spartan Rooks.

Eihen, A. (1997). Multi-parent recornbination. In T. Back, D. Fogel, & Z. Michalewicz (Eds.), Haiidbook of FL'OhtiOliN1? ro7npiitntion (pp. C3.3.7: 1-C3.3.7:9). New York: Oxford University Press.

E:iben, A, Kaue, P.E., 8i Kuttkay, Z. (1 994). Genetic algorithms with multi-parent recombination. In Y. Davidor, R-P. Schwefel, & R. Manner, (Eds.), Proceedingsofibe Third Conjii-ence on PnrallelProblem S0Li.i~.g-fi07/1.littnre [Lecture Notes in Computer Science 8661 (pp. 78-87). Berlin: Springer-Verlag.

Eiben, A, R Schippers, C. (1996). Multi-parent's niche: n-an; crossovers on NK-landscapes. In I I.LI. \'oipt, 1%'. Eheling, I. Kechenberg, & H.P. Schwefel (Eds.), Piaceeding.softhe Feud Conference 011 f'nt-~lkel h n b h i i Sohiizgfifi.otn %twr [Lecture Notes in Computer Science 1 1411 (pp. 3 19-328). 13erlin: Springer-Verlag.

Thông tin
Thông tin xuất bản

Evolutionary Computation

Tập 5 Số 3

347-365

Thông tin tác giả