Information Technology Reference
In-Depth Information
observed for mutation, is very different from the plots obtained both for trans-
position and recombination. Note that only mutation can climb “Mount Ever-
est” and, indeed, systems undergoing mutation can be easily tuned so that
populations stay, albeit precariously, poised at the peak as small variations
on mutation rate have dramatic effects within the finger region.
As Figure 12.1 emphasizes, mutation has a tremendous creative power
and, indeed, this operator alone is more than sufficient to evolve solutions to
virtually all problems. Nonetheless, other genetic operators can be and are
regularly used in GEP both for practical and theoretical reasons. For instance,
RIS transposition is interesting as it challenges our views on the importance
of too drastic modifications in evolution. Moreover, it is known that muta-
tion is not sufficiently innovative to account for all the wonders of nature.
Consider, for instance, the creation of the eukaryotes by symbiosis (Margulis
1970) or the simpler phenomenon of gene duplication (Lynch and Conery
2000) also observed in gene expression programming.
Also interesting are the results obtained for IS and RIS transposition. Par-
ticularly interesting is the fact that these two kinds of simple intrachromosomal
transposition far exceed all kinds of recombination in the first place and that
RIS transposition is slightly more efficient than IS transposition. It is worth
pointing out that with RIS transposition the first position of a gene is always
the target. And this means that, at the phenotype level, the root of sub-ETs is
modified. Indeed, this kind of modification is one of the most disruptive and
is similar to a frameshift mutation occurring at the beginning of a protein
gene. Note also that the transforming power of both kinds of transposition is
slightly smaller than mutation (compare maximum performance obtained
for each operator).
Also worth discussing are the results obtained for the three kinds of re-
combination. Recall that two-point recombination is the most disruptive of
the recombinational operators and, as shown in Figure 12.1, it is also the
most efficient kind of recombination. Not surprisingly, the most conserva-
tive of the recombinational operators - gene recombination - is also the less
efficient. In addition, it is worth noticing that all the recombinational mecha-
nisms analyzed here, even the most conservative, are more disruptive than
the homologous recombination that occurs during sexual reproduction be-
cause, in GEP, the exchanged genes rarely are homologous.
One of the unsolved questions of biology is the role of sex in evolution
(Margulis and Sagan 1986) and, most of the times, biological sex in its over-
whelming diversity is confounded with the homologous recombination that
Search WWH ::




Custom Search