GP Rule Problem

Home  News  Author  Q&A  Tutorials  Downloads  GEP Biblio  Contacts  Visit Gepsoft

  GP rule
 (CA rule for the density-classification problem discovered by GP)

  These kind of problems are similar to the Boolean multiplexer problems, with the difference that the solution is not known beforehand: the only thing known is a truth table with 128 transition states. Thus, these problems clearly show that GEP can be used efficiently in electronic circuit design. The rule table [1] has 7 arguments which are the 3 neighbour cells to the left (c, b, a) of a central cell, the 3 neighbour cells to the right (1, 2, 3), and the central cell (u).

 

Table 1
Parameters for the GP-rule problem
 
Function set A,O,I,X,D,R,M
Terminal set c,b,a,u,1,2,3
Number of fitness cases 128
Number of runs 100
Number of generations 100
Population size 100
Success rate 27%

Notes: A- AND; O- OR; I- if (x,y,z) function: if x is 1, the second argument is evaluated, otherwise the third argument is evaluated; X- XOR; D- NAND; R- NOR; M- majority (x,y,z) function.

Download the executable and find yourself a Boolean expression for the GP-rule and see that the GP-rule, like the GKL-rule, is a function of only 5 arguments (c,a,u,1,3).

Bibliography:
1. Koza, J. R., Bennett III, F. H., Andre, D., and Keane, M. A. (1999). Genetic Programming III: Darwinian Invention and Problem Solving. San Francisco: Morgan Kaufmann Publishers.

Back

***

Last update: 23/July/2013
  		© Candida Ferreira
All rights reserved.