Swarm Intelligence

networks and genetic algorithms, play a relatively minor role in GP im-

plementations. The selection of the function and terminal sets sig-

nificantly depends on the problem domain, however, so this selection

could be thought of as preprocessing.

Formulating the approach to solving a problem with a GP implemen-

tation can be difficult. Discovering what other people have done in simi-

lar circumstances is often helpful. Chapter 26 of Koza’s 1992 book pre-

sents tables to help guide a user in selection of terminal sets, function

sets, population size, and so on. Genetic programming has, as have the

other evolutionary algorithms we’ve discussed, developed ...