GEP Book

  Home
  News
  Author
  Q&A
  Tutorials
  Downloads
  GEP Biblio
  Contacts

  Visit Gepsoft

 

C. FERREIRA In E. Lutton, J. A. Foster, J. Miller, C. Ryan, and A. G. B. Tettamanzi, eds., Proceedings of the 4th European Conference on Genetic Programming, Lecture Notes in Computer Science, Vol. 2278, pages 51-60, Springer-Verlag, Berlin, Germany, 2002.

Discovery of the Boolean Functions to the Best Density-Classification Rules Using Gene Expression Programming

Coevolution1 Rule
 

The Coevolution1 rule has a performance of 85.1% and was discovered using a coevolutionary approach between GA-evolved rules and ICs [8].

Gene expression programming found the solution to the Coevolution1 rule in four phases. Firstly, chromosomes composed of nine genes of length 22 each were used. The sub-ETs were linked by IF, forming a huge multi-subunit ET. An intermediate solution with fitness 121 was found after 47093 generations of a small initial population of 30 individuals. Then, this intermediate solution was used as seed, and after 11785 generations a better solution with fitness 123 was found. Then, the genes of this intermediate solution were increased from 22 to 34 and populations of 50 individuals were left to evolve for 50000 generations. On generation 37362 an intermediate solution with fitness 127 was discovered. Finally, this program was used as seed to create another population, this time with 50 individuals. After 15189 generations the following perfect solution to the Coevolution1 rule was found:

0123456789012345678901234567890123
MAD21RMDuOa1ab1ucb31bbu1acc31bu1c1
XXM1ROM3Rubaaa3322c2ba13abcu3c1uc1
NOaANMbXIXbb3buc33bc2bba3b21u11uc1
XRRID1ADbDA2cb1bcuu1cbu133u1aaaa3c
AOaOIAcONuDbcca33auauba3332b11b2bb
MXIODuRXOA31122b3cb3a3bau1cbbbb312
MuIADIOXINMa1cua1uu1bu2cab1cuccc2a
NNIMOODR1ODuu31u132u12babc1bu2aub2
IuIIRDOXIMb2cu312ua23a2c3222ba2ab3

where the sub-ETs are linked 3-by-3 with IF, forming three big clusters that are, in their turn, linked also 3-by-3 with IF.

Home | Contents | Previous | Next