Biomedical Engineering Reference
In-Depth Information
Figure 6. Example of a type change
developed according to the concept of a correc-
tion template. From the tissue that is developed
by the DNA that is being evaluated, the centroid
is calculated. This point would be the center of
the solution template, which is merely a matrix
of Boolean values representing the figure that is
aimed at. The template could be (and usually is)
smaller than the development environment of
the tissue, which means that every cell that may
not be covered by the template will contribute to
the tissue error with 1.0. The remaining tissue,
covered by the template, will execute the NEXOR
Boolean operation in order to obtain the number
of differences between the template and the tis-
sue. Each difference contributes with a value of
1.0 to the tissue error.
Figure 7 illustrates the use of this method.
We can observe that the error of this tissue with
regard to the template is 2, since we generated
a cell that is not contemplated by the template,
whereas another cell that is present in the template
is really missing.
TESTS
The performance of the model was studied by
applying it to simple 2D spatial problems. We
selected this type of problems because they al-
low us to easily verify the results and simplify
the evaluation functions.
This section describes some of the tests per-
formed with the proposed model, using a Java
implementation on a Pentium Core Duo with 1GB
RAM. The same parameters were used for all the
tests in the GA and the cellular system model.
The GA parameters are the following:
1000 generations
500 population individuals
600 individual length sections (maximum)
90% crossovers and 10% mutations
50 pre-evaluation tissue development
cycles
GA operators: roulette selection, parent
replacement, and one point crossover
Figure 7. Tissue + Template. Example of Template use
Search WWH ::




Custom Search