Information Technology Reference
In-Depth Information
constants, drawn from the integer interval [0, 10] will be used and, as usual,
will be represented by the numerals 0-9, thus giving R = {0, ..., 9}. The
number of hits will be again used to evaluate the fitness of the evolving
decision trees. The complete list of parameters used per run in this experi-
ment is given in Table 9.8.
And as you can see in Table 9.8, GEP decision trees are very good at
learning from this dataset with an average best-of-run fitness of 90.73. Fur-
thermore, the best-of-experiment solution is, as usual, considerably above
average. It was created in generation 6861 of run 85 and encodes a decision
tree with 28 nodes:
MBOFBPcGBIbBcREbbdcbcdcbbcadabdbcbd...
Table 9.8
Performance and settings used in the lymphography problem.
Number of runs
100
Number of generations
20,000
Population size
30
Number of training instances
104
Number of testing instances
44
Number of attributes
18
Attribute set
A-R
Terminal set / Classes
a b c d
Random constants array length
10
Random constants type
Integer
Random constants range
[0, 10]
Maximum arity
8
Head length
15
Gene length
136
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
0.01
Fitness function
Eq. (3.8)
Average best-of-run fitness
90.73
