Ovidiu Gheorghies, Henri Luchian, Adriana Gheorghies

The aim of this paper is to show that exploiting knowledgeextracted from the optimization process is important for thesuccess of an evolutionary solver. In the context of NK fitnesslandscapes, we identify two causes for the difficulty ofan optimization problem: the intrinsic combinatorial difficultyand the random-search hybridization. We apply theseconcepts for the royal road fitness landscape. Experimentalresults indicate that Integrated-Adaptive Genetic Algorithms(IAGA) are particularly suited for tackling randomsearchhybridization. A learn-as-you-go system aimed at afine-grained adaptation of operators behavior increases thesolving power and convergence speed of IAGA. We concludethat the royal road problem is actually being “royal”for the traditional GA, but for a class of adaptive geneticalgorithms.

Full Document (PDF)


author = " Ovidiu Gheorghie{c s} and Henri Luchian and Adriana Gheorghie{c s}",
title = " Walking the Royal Road with Integrated-Adaptive Genetic Algorithms",
institution = "``Al.I.Cuza'' University of Ia{c s}i, 
                 Faculty of Computer Science",
year = "2005",
number = "TR 05-04",
note = "URL:http://www.infoiasi.ro/~tr/tr.pl.cgi"