Information Technology Reference
In-Depth Information
represented by the numerals 0-9. The fitness function will consist of the number
of hits and, therefore, will be evaluated by equation (3.8). The population size
and number of generations will be exactly the same used in the GEP-MO
(see Table 4.7) and MCS-MO (see Table 6.8) experiments so that we can
compare the performance of all these different algorithms. The complete list
of parameters used in this experiment is shown in Table 9.6.
And as you can see by comparing Table 9.6 with Tables 4.7 and 6.8, the
performance of the EDT-RNC algorithm is very similar to the performance
obtained in the MCS-MO experiment and both of them are considerably
better than the GEP-MO system (average best-of-run fitness of 147.52 for
the decision trees, 147.20 for the multicellular system with multiple outputs,
and 146.48 for the GEP-MO algorithm).
Table 9.6
Performance and settings used in the iris problem.
Number of runs
100
Number of generations
20,000
Population size
30
Number of fitness cases
150
Attribute set
S T P Q
Terminal set / Classes
a b c
Random constants array length
10
Random constants type
Rational
Random constants range
[0, 10]
Head length
10
Gene length
31
Mutation rate
0.044
Inversion rate
0.1
IS transposition rate
0.1
RIS transposition rate
0.1
One-point recombination rate
0.3
Two-point recombination rate
0.3
Dc-specific mutation rate
0.044
Dc-specific inversion rate
0.1
Dc-specific transposition rate
0.1
Random constants mutation rate
0.1
Fitness function
Eq. (3.8)
Average best-of-run fitness
147.52
