Information Technology Reference
In-Depth Information
Note that two of the best individuals of this generation are the clone of the
best individual of the initial population (chromosome 0) and a minor variant
of this individual (chromosome 3) encoding a slightly different DT that solves
a slightly different set of sample cases; the other one (chromosome 16) is
totally unrelated to the best of the previous generation and encodes a com-
pletely different decision tree that also happens to have fitness 10.
In the next generation a new individual was created, chromosome 7, slightly
better than the best individuals of the previous generations. Its expression is
shown in Figure 9.5. The complete population is shown below (the best indi-
vidual is shown in bold):
Generation N: 2
0123456789012345
ObOTOabbababbaab-[ 0] = 10
ObOTOabbababbaab-[ 1] = 10
HTWbaaababbbbabb-[ 2] = 6
ObHHWbbaabaabaab-[ 3] = 7
OHbWTabaababbaab-[ 4] = 8
OTWTOabaababaaaa-[ 5] = 8
OTOWbabbbbbbaabb-[ 6] = 5
HOWbaabababbbaba-[ 7] = 11
OTWTOabaabaabbaa-[ 8] = 10
OTWTOabbaabbbbba-[ 9] = 7
OTHTWabbbbabbaab-[10] = 5
Wabbabbaabababab-[11] = 5
WWbWWbbbabbbbabb-[12] = 5
HWaHbaabaabaabba-[13] = 9
WWbWWbbbabbbbabb-[14] = 5
WTaaHbabbbaaaaba-[15] = 8
OTWTOabaabaabbaa-[16] = 10
TOabHbaabbababaa-[17] = 5
TOObWababbababaa-[18] = 6
OTWTOabaabaabbaa-[19] = 10
It is worth pointing out that the structure of the best DT of this generation
is totally unrelated to the best-of-generation of the previous populations (com-
pare Figures 9.4 and 9.5). As you can see, it encodes a very compact decision
tree with just eight nodes that solves a total of 11 instances.
In the next generation a new individual was created, chromosome 11, that
slightly outperforms the best decision tree of the previous generation. Its
expression is shown in Figure 9.6. The complete population is shown below
(the best of generation is shown in bold):
Search WWH ::




Custom Search