Biomedical Engineering Reference
In-Depth Information
to determine cell connections; and the work of
Dellaert and Beer (Dellaert & Beer, 1996), who
propose a model that incorporates the idea of bio-
logical operons to control the model expression,
where the function assumes the mathematical
meaning of a Boolean function.
All these models can be regarded as special
cellular automats. In cellular automats, a starting
cell set in a certain state will turn into a differ-
ent set of cells in different states when the same
transition function is applied to all the cells dur-
ing a determined lapse of time in order to control
the message concurrence among them. The best
known example of cellular automats is Conway's
“Game of Life” (Conway, 1971), where this be-
haviour can be observed perfectly. Whereas the
classical conception specifies the behaviour rules,
the evolutionary models establish the rules by
searching for a specific behaviour. This difference
comes from the mathematical origin of the cellular
automats, whereas the here presented models are
based on biology and embryology.
These models should not be confused with
other concepts that might seem similar, such as
Gene Expression Programming (GEP) (Ferreira,
2006). Although GEP codifies the solution in a
string, similarly as how it is done in the present
work, the solution program is developed in a
tree shape, as in classical genetic programming
(Koza et. al., 1999) which has little or nothing in
common with the presented models.
that is needed for the development of the system.
The genes are activated or transcribed thanks to
the protein shaped-information that exists in the
cytoplasm, and consist of two main parts: the
sequence, which identifies the protein that will
be generated if the gene is transcribed, and the
promoter, which identifies the proteins that are
needed for gene transcription.
Another remarkable aspect of biological genes
is the difference between constitutive genes and
regulating genes. The latter are transcribed only
when the proteins identified in the promoter part
are present. The constitutive genes are always
transcribed, unless inhibited by the presence of
the proteins identified in the promoter part, acting
then as gene oppressors.
The present work has tried to partially model
this structure with the aim of fitting some of
its abilities into a computational model; in this
way, the system would have a structure similar
that is similar to the above and will be detailed
in section 4.
PROPOSED MODEL
Various model variants were developed on the
basis of biological concepts. The present work uses
for its tests an implemented version of our model
used in (Fernandez-Blanco, Dorado, Rabuñal,
Gestal & Pedreira, 2007). The proposed artificial
cellular system is based on the interaction of ar-
tificial cells by means of messages that are called
proteins. These cells can divide themselves, die,
or generate proteins that will act as messages for
themselves as well as for neighbour cells.
The system is supposed to express a global
behaviour towards the generation of structures
in 2D. Such behaviour would emerge from the
information encoded in a set of variables of the
cell that, in analogy with the biological cells, will
be named genes.
One promising application, in which we are
working, could be the compact encoding of adap-
bIOLOGICAL INSPIRATION
A biological cellular system can be categorized
as a complex system following the identification
characteristics of a complex system stated by
Nagl (Nagl, Parish, Paton & Warner, 1998). The
cells of a biological system are mainly determined
by the DNA strand, the genes, and the proteins
contained by the cytoplasm. The DNA is the
structure that holds the gene-encoded information
Search WWH ::




Custom Search