Information Technology Reference
In-Depth Information
In nature, at least in periods of stasis, the rates of mutation are tightly
controlled and usually very small, and adaptation and evolution occur very
smoothly. In GEP, populations are always evolving at high creative rates in
what Eldredge and Gould called creative periods in their theory of punctu-
ated equilibrium (Eldredge and Gould 1972). So, in artificial evolutionary
systems, evolution can be much faster and kept in constant turmoil, allowing
the discovery of very good solutions in record time.
Let's now take a closer look at the structure of the best individual of gen-
eration 1 (chromosome 6). Its expression is shown in Figure 3.6. As you can
see, this program is slightly better than the best of the previous generation,
solving correctly a total of seven fitness cases. As for its structure, we can
easily guess that it is a descendant of the best of the previous generation
(chromosome 4), as they only differ at position 2 in the second gene:
01234560123456
AbObcbcOA
a
caac-[0,4] = 6
AbObcbcOA
A
caac-[1,6] = 7
As you can see, this new individual suffered just one point-mutation during
reproduction, changing the “a” at position 2 in gene 2 into “A”. This resulted
in a bulkier sub-ET
2
encoding a different Boolean expression (compare Fig-
ures 3.4 and 3.6), conferring to this new individual a slightly better perform-
ance than its mother.
Figure 3.7 shows the next generation (generation 2) created in this evolu-
tionary process. And as you can see, the best individual of this generation
(chromosome 1) has maximum fitness and therefore codes for a perfect
a.
0123456
AbObcbcOAAcaac
0123456
b.
O
A
O
b
O
A
A
c
c
a
a
c
b
Figure 3.6.
Expression of the best individual of generation 1 (chromosome 6).
a)
The chromosome composed of two genes.
b)
The program encoded in the
chromosome (the linking function is shown in gray). This Boolean function
solves correctly seven out of eight fitness cases (see Table 3.1).
