|
Abstract:
|
Successful use of EAs depends on the choice of the representation - that is, the genotype and the genotype-phenotype mapping - and on the used search operators. The question whether a certain representation leads to higher EA performance can only be answered when the operators applied are considered. Indirect representations use standard data structures (e.g. strings, vectors) and standard off- the-shelf search operators. To evaluate the solution, the genotype is mapped to the phenotype space. Key concepts to analyse the impact of representation-operator combinations on EA performance are (1) locality, (2) redundancy, and (3) bias. Locality is a result of the interplay between the search operator and the genotype-phenotype mapping. Representations are redundant if the number of phenotypes exceeds the number of possible genotypes. Furthermore, redundant representations can lead to biased encodings if some phenotypes are on average represented by a larger number of genotypes. Finally, a bias need not be the result of the representation but can also be caused by the search operator. The tutorial gives a brief overview about existing guidelines for representation design, illustrates different aspects of representations, gives a brief overview of theoretical models describing the different aspects, and illustrates their relevance with practical examples.
|
