Biomedical Engineering Reference
In-Depth Information
1987). A trained ANN can be considered as an
“expert” in the category of information it has
received to analyse. The two main advantages
of the ANN are:
copying the evolutionary behaviour of species:
from an initial random population of solutions,
this population is evolved mainly by means of
selection, mutation and crossover genetic opera-
tors, taken from natural evolution. By applying
this set of operations, the population goes through
an iterative process in which it reaches different
states, each one is called generation. As a result of
this process, the population is expected to reach a
generation in which it contains a good solution to
the problem. In GAs the solutions of the problem
are codified as a string of bits or real numbers.
While there are many different types of selec-
tion operation, the most common type is roulette
whell selection. In roulette wheel selection, indi-
viduals are given a probability of being selected
that is directly proportionate to their fitness. Two
individuals are then chosen randomly based on
these probabilities and produce offspring. There
are many different kinds of crossover opera-
tion, but the most common type is a single point
crossover. In single point crossover, you choose
a point at which you swap one part of gens from
a parent to the other.
Genetic Algorithms are very effective way of
quickly finding a reasonable solution to a complex
problem. They do an excellent job of searching
trough a large and complex search spaces. Genetic
Algorithms are most effective in a great space
for which little is known and we didn't any more
about the problem that we want to solve. You may
know exactly what you want a solution to do but
have no idea how you want it to go about doing
it. This is where genetic algorithms thrive. They
produce solutions that solve the problem in ways
you may never have even considered.
•
Adaptive learning:
An ability to learn
how to do tasks based on the data given for
training.
•
Fault tolerance:
Partial destruction of the
network leads to the corresponding degrada-
tion of performance.However, some network
capabilities may be retained even with major
network damage.
The ANN's have shown to be a powerful tool
in many different applications. If significant vari-
ables are known, but not their exact relationships,
an ANN is able to perform a kind of function fit-
ting by using multiple parameters on the existing
information and predict the possible relationships
for the coming future.
In recent years, the need to develop more pow-
erful systems which are capable of solving time
problems. Recurrent ANN is the best option in the
ANN field (Williams et al., 1989) for this kind of
problems. This kind of network is more complex
than traditional ones, so that the problems of net-
work development are more acute. Their structure
has a much higher number of connections, which
complicates both the training process (Hee et al.,
1997) and the architecture adjustment.
Genetic Algorithm
A GA (Holland, 1975) is a searching method
developed by John Holland and based in the
emulation of the natural evolution. This method
is contained in the Evolutionary Computation
techniques.
A GA (Goldberg, 1989) is a search technique
inspired in the world of biology. More specifi-
cally, the Evolution theory by Charles Darwin
(Darwin, 1859) is taken as a basis for its working.
GAs are used to solve optimization problems by
biological Neuron Dynamics
The activation output of a neuron cell is explained
from the electrical characteristics of the cell's
membrane (Taylor et al., 2003). This membrane is
more permeable to the potassium ions (K+) than to
the sodium ions. The potassium chemical gradient
Search WWH ::
Custom Search