GEP Book

  Home
  News
  Author
  Q&A
  Tutorials
  Downloads
  GEP Biblio
  Contacts

  Visit Gepsoft

 

C. FERREIRA In N. Nedjah, L. de M. Mourelle, A. Abraham, eds., Genetic Systems Programming: Theory and Experiences, Studies in Computational Intelligence, Vol. 13, pp. 21-56, Springer-Verlag, 2006.

Automatically Defined Functions in Gene Expression Programming

Results without ADFs
 
The importance of Automatically Defined Functions and the advantages they bring to Evolutionary Computation can only be understood if one analyzes their behavior and how the algorithm copes with their integration. Is evolution still smooth? Are there gains in performance? Is the system still simple or excessively complicated? How does it compare to simpler systems without ADFs? How does one integrate random numerical constants in ADFs? Are these ADFs still manageable or excessively complex? Are there problems that can only be solved with ADFs? These are some of the questions that we will try to address in this work. And for that we are going to start this study by analyzing two simpler systems: the simpler of the two is the unigenic system of GEP that evolves a single parse tree and therefore bares some resemblance to the GP system without ADFs; the other one is the multigenic system of GEP that evolves multiple parse trees linked together by a predefined linking function.

 
Home | Contents | Previous | Next