Biomedical Engineering Reference
In-Depth Information
caching algorithms, compression algorithms
[51]
, random-number generators, and algo-
rithms for automatic parallelization of code
[52]
, to name a few, have been studied. The
spectrum of applications in computer science
spans from the generation of proofs for predi-
cate calculus to the evolution of machine code
for accelerating function evaluation. The gen-
eral tendency is to try to automate the design
process for algorithms of different kinds.
Recently the process of debugging code, i.e., the
correction of errors, has been added to the list
of applications
[53]
. Computer science itself has
many applications, and it is natural that those
areas also benefit indirectly by improving
methods in computer science. For instance, in
the area of computer vision, GP has been used,
among others, for
astronomy and astrophysics; see, for instance,
Refs.
68
and
69
.
Modeling is, however, but one of the applica-
tions of GP in science. Pattern recognition is
another key application used in molecular biol-
ogy and other branches of biology and medicine
as well as in science in general
[70, 71]
. Here, GP
has delivered results that are competitive if not
better than human-generated results
[72, 73]
, a
special area of applications we return to in the
next section. Classification and data mining are
other applications in which GP is in prominent
use
[74, 75]
.
In engineering, GP and other evolutionary
algorithms are used as standalone tools
[76]
or
sometimes in competition or cooperation with
other heuristic methods such as neural networks
or fuzzy systems. The general goal is, again, to
model processes such as material properties
[77]
or production plants or to classify results of pro-
duction. In recent years, design in engineering
has regained some prominence
[78]
. Control of
manmade apparatus is another area in which
GP has been used successfully, with process con-
trol and robot control (e.g., Ref.
79
) the primary
applications.
In business and finance, GP has been used to
predict financial data, notably bancruptcy of
companies
[80, 81]
. The entire area of computa-
tional finance is ripe with applications for GP
(and other evolutionary techniques); see Refs.
82
and 83
. For an early bibliography of business
applications of GP and GAs, the reader is referred
to Ref.
84
. Since, generally speaking, modeling
and prediction are core applications in economic
contexts, GP is an important nonlinear modeling
technique to consider; see, e.g., Refs.
85
and
86
.
In art and entertainment, GP is used to
evolve realistic animation scenes and appeal-
ing visual graphics (see
[87]
for an early exam-
ple). Computer games are another active area
of research and application for GP (see, for
instance,
[88]
). Board games have been studied
with GP-developed strategies, too
[89, 90]
. GP
also has been used in visual art and music
[91]
.
• objectdetection(forexample,Refs.
54
and
55
),
• ilterevolution(forexample,Refs.
56
and
57
),
• edgedetection(forexample,Ref.
58
),
• interestpointdetection(forexample,Ref.
59
),
and
• texturesegmentation(forexample,Ref.
60
).
In addition, the area of software engineering is
a field very fruitful for applications of GP
[61]
.
Query optimization for database applications is
a widespread application of evolutionary com-
putation techniques (see their use in PostgreSQL
and H2,
Refs. 62 and 63
).
Typical applications for GP in science are
those to modeling and pattern recognition.
Modeling certain processes in physics and
chemistry with the unconventional help of evo-
lutionary creativity supports research and
understanding of the systems under study
[64,
65]
. For instance, parameters of models in soil
science can be readily estimated by GP
[66]
. Pre-
dictions based on models generated with GP
have widespread applications. An example
from climate science is
[67]
, where seawater
level is forecast by a GP modeling technique
using past time series. Many modeling applica-
tions for EC methods in general exist in
