Information Technology Reference
In-Depth Information
Fig. 1.
Main steps of omni-aiNet
loop where the stop criterion is the number of generations (also defined by the
user). Within this loop, the main steps of the algorithm are executed: Cloning,
Hypermutation, Selection and Gene Duplication. The suppression of individuals
and insertion of new randomly generated ones are made from
N
gs
to
N
gs
gen-
erations (
N
gs
is defined by the user). The value of
N
gs
should be greater than
one to give enough time for the algorithm to explore the vicinity of each solution
before the suppression of similar individuals.
The following Subsections will present a detailed description of the main steps
presented in Figure 1.
4.1
Cloning and Hypermutation
The first step of each generation of the algorithm is the
cloning
phase. In this
phase, for all individuals in the antibody population,
N
c
identical copies (clones)
are generated. The parameter
N
c
must be defined by the user. Then, this pop-
ulation of clones suffer a process of genetic variability known as
hypermutation
.
The hypermutation mechanism applies to each generated clone, a random
variation with rates inversely proportional to its anity to the antigen (also
known as its
fitness
). In this algorithm, it was adopted the
polynomial mutation
mechanism, where a new individual is given by
c
=
c
+
γ
Δ
max
,with
c
the parent individual,
c
the new clone, and
Δ
max
the maximum shift that the
individual can suffer in direction
γ
without violating the domain of the variables.
The value of
γ
is given by:
γ
=
(2
u
)
×
1
η
+1
−
1
,
if
u<
0
.
5
(2)
1
η
+1
,
if
u
1
−
[2(1
−
u
)]
≥
0
.
5
where
u
is a random number with uniform distribution in [0, 1].
