Information Technology Reference
In-Depth Information
100
90
80
70
60
50
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
100
Chromosome length
Figure 4.1. Variation of success rate with chromosome length. For this analysis,
chromosomes composed of one gene were used, G = 50, and P = 30. The success
rate was evaluated over 100 identical runs.
also the most parsimonious solutions to the problem at hand. For instance, the
following chromosomes with 13 nodes each:
0123456789012
*++/*/aaaaaaa-[1]
*++*//aaaaaaa-[2]
both code for perfect parsimonious solutions to the target function (4.1).
Note that gene expression programming can evolve solutions efficiently
using large values of h , i.e., is capable of dealing with highly redundant in-
formation. As shown in Figure 4.1, for each problem, there is a chromosome
length that allows the most efficient evolution. And, at least for simple func-
tions, this ideal chromosome length can be easily found. Note also that the
most compact genomes are not the most efficient. This suggests that a cer-
tain redundancy is fundamental to the efficient evolution of good programs
(see a discussion of The Role of Neutrality in Evolution in chapter 12).
As shown in Table 4.1, for this analysis, a population size P of 30 indi-
viduals and an evolutionary time G of 50 generations were used. Unless
 
Search WWH ::




Custom Search