Title :
Recombination strategy adaptation via evolution of gene linkage
Author :
Smith, Jim ; Fogarty, T.C.
Author_Institution :
Fac. of Comput. Studies & Math., Univ. of the West of England, Bristol, UK
Abstract :
Analyses recombination strategies resulting from evolving gene linkage on a variety of fitness landscapes with known properties. The evolution of “blocks” of linked genes within a gene-pool makes it possible for the recombination strategy to vary in both the amount of genetic material exchanged during the formation of a new individual and in the number of parents contributing genetic material. The strategies evolved are examined in the light of the known properties of the landscapes. The results obtained explain the previously recorded good performance of the algorithm as a self-adaptive mechanism for function optimisation, as the amount and type of recombination is able to adapt to suit the landscape being searched, unlike conventional “fixed” operators
Keywords :
adaptive systems; genetic algorithms; algorithm performance; contributing parents; fitness landscape searching; function optimisation; gene linkage evolution; gene-pool; genetic material; genetic material exchange; individual formation; linked gene blocks; recombination strategy adaptation; self-adaptive mechanism; Algorithm design and analysis; Couplings; Encoding; Genetic algorithms; Genetic mutations; Mathematics; Optimization methods; Robustness; Testing; Time measurement;
Conference_Titel :
Evolutionary Computation, 1996., Proceedings of IEEE International Conference on
Conference_Location :
Nagoya
Print_ISBN :
0-7803-2902-3
DOI :
10.1109/ICEC.1996.542708