In Chapter 3 we described early schema theories for genetic programming These are based on the idea of schemata as components within the program tree. In this chapter we will consider two GP schema theories of a different kind. These schemata have the advantage that not only can they be considered as parts of programs but, also that they are true subsets of the search space. In Section 4.1 we summarise the schema theory proposed by Justinian Rosca [Rosca, 1997a] that is applicable to GP with standard crossover. How ever, the main focus of the chapter is on our schema theory. This is based on a concept of schema which is much closer to the original concept of schema in genetic algorithms and it is applicable to GP (and to different kinds of GAs) when one-point crossover is used This is a generalisation of the corresponding GA operator in which the same crossover point is selected in both parents.
KeywordsCrossover Point Initialisation Method Schema Definition Schema Disruption Standard Crossover
Unable to display preview. Download preview PDF.