Inproceedings,

Controlling Code Growth in Genetic Programming

.
Advances in Soft Computing, page 166--171. De Montfort University, Leicester, UK, Physica-Verlag, (2000)

Abstract

It has been known since the early days of Genetic Programming that the evolutionary process tends to stagnate after a certain number of generations. Furthermore, during the evolutionary process, there is an inexorable and sometimes exponential increase in the average size of programs in a population. This has been called bloat. The major cause of this bloat appears to be caused by introns - sections of non-functional code. In this paper introns are categorised and different approaches to the control of code growth are described.

Tags

Users

  • @brazovayeye

Comments and Reviews