|
This chapter is wholly dedicated to the amazing family of techniques known as
genetic algorithms. As a matter of fact, not so amazing, since similar principles in nature have
produced complex entities such as humans. This chapter starts by describing the basic properties
and motivation underlying the genetic algorithm approach. After outlining the approach, the genetic
operators are introduced and the whole approach is exemplified with respect to a simple example.
The next section covers the important concept of schemata
, introduced by Holland as part of his
approach to develop a theoretical basis for genetic algorithms, and the
schemata theorem. These
concepts are discussed in terms of the bandit problem. The last section in this chapter discuss how
to control the cost of calculating the fitness function for large populations in terms of sampling
the individuals. The concept is illustrated for a simple situation. This section concludes by discussing
the interesting issue of co-evolution of parasites,
which explores the idea that the competition between two species tend to improve the fitness of both.
|
|
|