<s>
In	O
computer	B-General_Concept
science	I-General_Concept
and	O
operations	O
research	O
,	O
a	O
genetic	B-Algorithm
algorithm	I-Algorithm
(	O
GA	O
)	O
is	O
a	O
metaheuristic	B-Algorithm
inspired	O
by	O
the	O
process	O
of	O
natural	B-Application
selection	I-Application
that	O
belongs	O
to	O
the	O
larger	O
class	O
of	O
evolutionary	B-Algorithm
algorithms	I-Algorithm
(	O
EA	O
)	O
.	O
</s>
<s>
Genetic	B-Algorithm
algorithms	I-Algorithm
are	O
commonly	O
used	O
to	O
generate	O
high-quality	O
solutions	O
to	O
optimization	O
and	O
search	O
problems	O
by	O
relying	O
on	O
biologically	O
inspired	O
operators	O
such	O
as	O
mutation	O
,	O
crossover	O
and	O
selection	O
.	O
</s>
<s>
Some	O
examples	O
of	O
GA	O
applications	O
include	O
optimizing	O
decision	B-Algorithm
trees	I-Algorithm
for	O
better	O
performance	O
,	O
solving	O
sudoku	O
puzzles	O
,	O
hyperparameter	B-General_Concept
optimization	I-General_Concept
,	O
etc	O
.	O
</s>
<s>
In	O
a	O
genetic	B-Algorithm
algorithm	I-Algorithm
,	O
a	O
population	O
of	O
candidate	O
solutions	O
(	O
called	O
individuals	O
,	O
creatures	O
,	O
organisms	O
,	O
or	O
phenotypes	O
)	O
to	O
an	O
optimization	O
problem	O
is	O
evolved	O
toward	O
better	O
solutions	O
.	O
</s>
<s>
The	O
evolution	O
usually	O
starts	O
from	O
a	O
population	O
of	O
randomly	O
generated	O
individuals	O
,	O
and	O
is	O
an	O
iterative	B-Algorithm
process	I-Algorithm
,	O
with	O
the	O
population	O
in	O
each	O
iteration	B-Algorithm
called	O
a	O
generation	O
.	O
</s>
<s>
The	O
new	O
generation	O
of	O
candidate	O
solutions	O
is	O
then	O
used	O
in	O
the	O
next	O
iteration	B-Algorithm
of	O
the	O
algorithm	O
.	O
</s>
<s>
A	O
typical	O
genetic	B-Algorithm
algorithm	I-Algorithm
requires	O
:	O
</s>
<s>
a	O
genetic	B-Algorithm
representation	I-Algorithm
of	O
the	O
solution	O
domain	O
,	O
</s>
<s>
A	O
standard	O
representation	O
of	O
each	O
candidate	O
solution	O
is	O
as	O
an	O
array	B-Data_Structure
of	I-Data_Structure
bits	I-Data_Structure
(	O
also	O
called	O
bit	O
set	O
or	O
bit	B-Data_Structure
string	I-Data_Structure
)	O
.	O
</s>
<s>
The	O
main	O
property	O
that	O
makes	O
these	O
genetic	B-Algorithm
representations	I-Algorithm
convenient	O
is	O
that	O
their	O
parts	O
are	O
easily	O
aligned	O
due	O
to	O
their	O
fixed	O
size	O
,	O
which	O
facilitates	O
simple	O
crossover	O
operations	O
.	O
</s>
<s>
Tree-like	O
representations	O
are	O
explored	O
in	O
genetic	B-Algorithm
programming	I-Algorithm
and	O
graph-form	O
representations	O
are	O
explored	O
in	O
evolutionary	B-Algorithm
programming	I-Algorithm
;	O
a	O
mix	O
of	O
both	O
linear	O
chromosomes	O
and	O
trees	O
is	O
explored	O
in	O
gene	B-Algorithm
expression	I-Algorithm
programming	I-Algorithm
.	O
</s>
<s>
Once	O
the	O
genetic	B-Algorithm
representation	I-Algorithm
and	O
the	O
fitness	O
function	O
are	O
defined	O
,	O
a	O
GA	O
proceeds	O
to	O
initialize	O
a	O
population	O
of	O
solutions	O
and	O
then	O
to	O
improve	O
it	O
through	O
repetitive	O
application	O
of	O
the	O
mutation	O
,	O
crossover	O
,	O
inversion	O
and	O
selection	O
operators	O
.	O
</s>
<s>
The	O
fitness	O
function	O
is	O
defined	O
over	O
the	O
genetic	B-Algorithm
representation	I-Algorithm
and	O
measures	O
the	O
quality	O
of	O
the	O
represented	O
solution	O
.	O
</s>
<s>
For	O
instance	O
,	O
in	O
the	O
knapsack	B-Algorithm
problem	I-Algorithm
one	O
wants	O
to	O
maximize	O
the	O
total	O
value	O
of	O
objects	O
that	O
can	O
be	O
put	O
in	O
a	O
knapsack	O
of	O
some	O
fixed	O
capacity	O
.	O
</s>
<s>
A	O
representation	O
of	O
a	O
solution	O
might	O
be	O
an	O
array	B-Data_Structure
of	I-Data_Structure
bits	I-Data_Structure
,	O
where	O
each	O
bit	O
represents	O
a	O
different	O
object	O
,	O
and	O
the	O
value	O
of	O
the	O
bit	O
(	O
0	O
or	O
1	O
)	O
represents	O
whether	O
or	O
not	O
the	O
object	O
is	O
in	O
the	O
knapsack	O
.	O
</s>
<s>
In	O
some	O
problems	O
,	O
it	O
is	O
hard	O
or	O
even	O
impossible	O
to	O
define	O
the	O
fitness	O
expression	O
;	O
in	O
these	O
cases	O
,	O
a	O
simulation	B-Application
may	O
be	O
used	O
to	O
determine	O
the	O
fitness	O
function	O
value	O
of	O
a	O
phenotype	O
(	O
e.g.	O
</s>
<s>
computational	O
fluid	O
dynamics	O
is	O
used	O
to	O
determine	O
the	O
air	O
resistance	O
of	O
a	O
vehicle	O
whose	O
shape	O
is	O
encoded	O
as	O
the	O
phenotype	O
)	O
,	O
or	O
even	O
interactive	O
genetic	B-Algorithm
algorithms	I-Algorithm
are	O
used	O
.	O
</s>
<s>
Although	O
crossover	O
and	O
mutation	O
are	O
known	O
as	O
the	O
main	O
genetic	O
operators	O
,	O
it	O
is	O
possible	O
to	O
use	O
other	O
operators	O
such	O
as	O
regrouping	O
,	O
colonization-extinction	O
,	O
or	O
migration	O
in	O
genetic	B-Algorithm
algorithms	I-Algorithm
.	O
</s>
<s>
A	O
very	O
small	O
mutation	O
rate	O
may	O
lead	O
to	O
genetic	B-Application
drift	I-Application
(	O
which	O
is	O
non-ergodic	O
in	O
nature	O
)	O
.	O
</s>
<s>
A	O
recombination	O
rate	O
that	O
is	O
too	O
high	O
may	O
lead	O
to	O
premature	O
convergence	O
of	O
the	O
genetic	B-Algorithm
algorithm	I-Algorithm
.	O
</s>
<s>
In	O
addition	O
to	O
the	O
main	O
operators	O
above	O
,	O
other	O
heuristics	B-Algorithm
may	O
be	O
employed	O
to	O
make	O
the	O
calculation	O
faster	O
or	O
more	O
robust	O
.	O
</s>
<s>
The	O
speciation	O
heuristic	B-Algorithm
penalizes	O
crossover	O
between	O
candidate	O
solutions	O
that	O
are	O
too	O
similar	O
;	O
this	O
encourages	O
population	O
diversity	O
and	O
helps	O
prevent	O
premature	O
convergence	O
to	O
a	O
less	O
optimal	O
solution	O
.	O
</s>
<s>
Genetic	B-Algorithm
algorithms	I-Algorithm
are	O
simple	O
to	O
implement	O
,	O
but	O
their	O
behavior	O
is	O
difficult	O
to	O
understand	O
.	O
</s>
<s>
A	O
description	O
of	O
a	O
heuristic	B-Algorithm
that	O
performs	O
adaptation	O
by	O
identifying	O
and	O
recombining	O
"	O
building	O
blocks	O
"	O
,	O
i.e.	O
</s>
<s>
low	O
order	O
,	O
low	O
defining-length	O
schemata	B-Algorithm
with	O
above	O
average	O
fitness	O
.	O
</s>
<s>
A	O
hypothesis	O
that	O
a	O
genetic	B-Algorithm
algorithm	I-Algorithm
performs	O
adaptation	O
by	O
implicitly	O
and	O
efficiently	O
implementing	O
this	O
heuristic	B-Algorithm
.	O
</s>
<s>
Goldberg	O
describes	O
the	O
heuristic	B-Algorithm
as	O
follows	O
:	O
</s>
<s>
"	O
Short	O
,	O
low	O
order	O
,	O
and	O
highly	O
fit	O
schemata	B-Algorithm
are	O
sampled	O
,	O
recombined	O
[	O
crossed	O
over ]	O
,	O
and	O
resampled	O
to	O
form	O
strings	O
of	O
potentially	O
higher	O
fitness	O
.	O
</s>
<s>
In	O
a	O
way	O
,	O
by	O
working	O
with	O
these	O
particular	O
schemata	B-Algorithm
[	O
the	O
building	O
blocks ]	O
,	O
we	O
have	O
reduced	O
the	O
complexity	O
of	O
our	O
problem	O
;	O
instead	O
of	O
building	O
high-performance	O
strings	O
by	O
trying	O
every	O
conceivable	O
combination	O
,	O
we	O
construct	O
better	O
and	O
better	O
strings	O
from	O
the	O
best	O
partial	O
solutions	O
of	O
past	O
samplings	O
.	O
</s>
<s>
"	O
Because	O
highly	O
fit	O
schemata	B-Algorithm
of	O
low	O
defining	O
length	O
and	O
low	O
order	O
play	O
such	O
an	O
important	O
role	O
in	O
the	O
action	O
of	O
genetic	B-Algorithm
algorithms	I-Algorithm
,	O
we	O
have	O
already	O
given	O
them	O
a	O
special	O
name	O
:	O
building	O
blocks	O
.	O
</s>
<s>
Just	O
as	O
a	O
child	O
creates	O
magnificent	O
fortresses	O
through	O
the	O
arrangement	O
of	O
simple	O
blocks	O
of	O
wood	O
,	O
so	O
does	O
a	O
genetic	B-Algorithm
algorithm	I-Algorithm
seek	O
near	O
optimal	O
performance	O
through	O
the	O
juxtaposition	O
of	O
short	O
,	O
low-order	O
,	O
high-performance	O
schemata	B-Algorithm
,	O
or	O
building	O
blocks.	O
"	O
</s>
<s>
There	O
are	O
limitations	O
of	O
the	O
use	O
of	O
a	O
genetic	B-Algorithm
algorithm	I-Algorithm
compared	O
to	O
alternative	O
optimization	O
algorithms	O
:	O
</s>
<s>
Repeated	O
fitness	O
function	O
evaluation	O
for	O
complex	O
problems	O
is	O
often	O
the	O
most	O
prohibitive	O
and	O
limiting	O
segment	O
of	O
artificial	O
evolutionary	B-Algorithm
algorithms	I-Algorithm
.	O
</s>
<s>
In	O
real	O
world	O
problems	O
such	O
as	O
structural	O
optimization	O
problems	O
,	O
a	O
single	O
function	O
evaluation	O
may	O
require	O
several	O
hours	O
to	O
several	O
days	O
of	O
complete	O
simulation	B-Application
.	O
</s>
<s>
In	O
this	O
case	O
,	O
it	O
may	O
be	O
necessary	O
to	O
forgo	O
an	O
exact	O
evaluation	O
and	O
use	O
an	O
approximated	B-Algorithm
fitness	I-Algorithm
that	O
is	O
computationally	O
efficient	O
.	O
</s>
<s>
It	O
is	O
apparent	O
that	O
amalgamation	O
of	O
approximate	B-Algorithm
models	I-Algorithm
may	O
be	O
one	O
of	O
the	O
most	O
promising	O
approaches	O
to	O
convincingly	O
use	O
GA	O
to	O
solve	O
complex	O
real	O
life	O
problems	O
.	O
</s>
<s>
Genetic	B-Algorithm
algorithms	I-Algorithm
do	O
not	O
scale	O
well	O
with	O
complexity	O
.	O
</s>
<s>
Hence	O
we	O
typically	O
see	O
evolutionary	B-Algorithm
algorithms	I-Algorithm
encoding	O
designs	O
for	O
fan	O
blades	O
instead	O
of	O
engines	O
,	O
building	O
shapes	O
instead	O
of	O
detailed	O
construction	O
plans	O
,	O
and	O
airfoils	O
instead	O
of	O
whole	O
aircraft	O
designs	O
.	O
</s>
<s>
In	O
many	O
problems	O
,	O
GAs	O
have	O
a	O
tendency	O
to	O
converge	O
towards	O
local	B-Algorithm
optima	O
or	O
even	O
arbitrary	O
points	O
rather	O
than	O
the	O
global	O
optimum	O
of	O
the	O
problem	O
.	O
</s>
<s>
The	O
likelihood	O
of	O
this	O
occurring	O
depends	O
on	O
the	O
shape	O
of	O
the	O
fitness	O
landscape	O
:	O
certain	O
problems	O
may	O
provide	O
an	O
easy	O
ascent	O
towards	O
a	O
global	O
optimum	O
,	O
others	O
may	O
make	O
it	O
easier	O
for	O
the	O
function	O
to	O
find	O
the	O
local	B-Algorithm
optima	O
.	O
</s>
<s>
Diversity	O
is	O
important	O
in	O
genetic	B-Algorithm
algorithms	I-Algorithm
(	O
and	O
genetic	B-Algorithm
programming	I-Algorithm
)	O
because	O
crossing	O
over	O
a	O
homogeneous	O
population	O
does	O
not	O
yield	O
new	O
solutions	O
.	O
</s>
<s>
In	O
evolution	B-Algorithm
strategies	I-Algorithm
and	O
evolutionary	B-Algorithm
programming	I-Algorithm
,	O
diversity	O
is	O
not	O
essential	O
because	O
of	O
a	O
greater	O
reliance	O
on	O
mutation	O
.	O
</s>
<s>
Again	O
,	O
evolution	B-Algorithm
strategies	I-Algorithm
and	O
evolutionary	B-Algorithm
programming	I-Algorithm
can	O
be	O
implemented	O
with	O
a	O
so-called	O
"	O
comma	O
strategy	O
"	O
in	O
which	O
parents	O
are	O
not	O
maintained	O
and	O
new	O
parents	O
are	O
selected	O
only	O
from	O
offspring	O
.	O
</s>
<s>
For	O
specific	O
optimization	O
problems	O
and	O
problem	O
instances	O
,	O
other	O
optimization	O
algorithms	O
may	O
be	O
more	O
efficient	O
than	O
genetic	B-Algorithm
algorithms	I-Algorithm
in	O
terms	O
of	O
speed	O
of	O
convergence	O
.	O
</s>
<s>
Alternative	O
and	O
complementary	O
algorithms	O
include	O
evolution	B-Algorithm
strategies	I-Algorithm
,	O
evolutionary	B-Algorithm
programming	I-Algorithm
,	O
simulated	B-Algorithm
annealing	I-Algorithm
,	O
Gaussian	O
adaptation	O
,	O
hill	B-Algorithm
climbing	I-Algorithm
,	O
and	O
swarm	B-Architecture
intelligence	I-Architecture
(	O
e.g.	O
</s>
<s>
:	O
ant	B-Algorithm
colony	I-Algorithm
optimization	I-Algorithm
,	O
particle	B-Algorithm
swarm	I-Algorithm
optimization	I-Algorithm
)	O
and	O
methods	O
based	O
on	O
integer	B-Algorithm
linear	I-Algorithm
programming	I-Algorithm
.	O
</s>
<s>
The	O
suitability	O
of	O
genetic	B-Algorithm
algorithms	I-Algorithm
is	O
dependent	O
on	O
the	O
amount	O
of	O
knowledge	O
of	O
the	O
problem	O
;	O
well	O
known	O
problems	O
often	O
have	O
better	O
,	O
more	O
specialized	O
approaches	O
.	O
</s>
<s>
The	O
simplest	O
algorithm	O
represents	O
each	O
chromosome	O
as	O
a	O
bit	B-Data_Structure
string	I-Data_Structure
.	O
</s>
<s>
Typically	O
,	O
numeric	B-Algorithm
parameters	O
can	O
be	O
represented	O
by	O
integers	O
,	O
though	O
it	O
is	O
possible	O
to	O
use	O
floating	B-Algorithm
point	I-Algorithm
representations	I-Algorithm
.	O
</s>
<s>
The	O
floating	B-Algorithm
point	I-Algorithm
representation	I-Algorithm
is	O
natural	O
to	O
evolution	B-Algorithm
strategies	I-Algorithm
and	O
evolutionary	B-Algorithm
programming	I-Algorithm
.	O
</s>
<s>
The	O
notion	O
of	O
real-valued	O
genetic	B-Algorithm
algorithms	I-Algorithm
has	O
been	O
offered	O
but	O
is	O
really	O
a	O
misnomer	O
because	O
it	O
does	O
not	O
really	O
represent	O
the	O
building	O
block	O
theory	O
that	O
was	O
proposed	O
by	O
John	O
Henry	O
Holland	O
in	O
the	O
1970s	O
.	O
</s>
<s>
Other	O
variants	O
treat	O
the	O
chromosome	O
as	O
a	O
list	O
of	O
numbers	O
which	O
are	O
indexes	O
into	O
an	O
instruction	O
table	O
,	O
nodes	O
in	O
a	O
linked	B-Data_Structure
list	I-Data_Structure
,	O
hashes	B-Application
,	O
objects	O
,	O
or	O
any	O
other	O
imaginable	O
data	B-General_Concept
structure	I-General_Concept
.	O
</s>
<s>
Other	O
approaches	O
involve	O
using	O
arrays	O
of	O
real-valued	O
numbers	O
instead	O
of	O
bit	B-Data_Structure
strings	I-Data_Structure
to	O
represent	O
chromosomes	O
.	O
</s>
<s>
Results	O
from	O
the	O
theory	O
of	O
schemata	B-Algorithm
suggest	O
that	O
in	O
general	O
the	O
smaller	O
the	O
alphabet	O
,	O
the	O
better	O
the	O
performance	O
,	O
but	O
it	O
was	O
initially	O
surprising	O
to	O
researchers	O
that	O
good	O
results	O
were	O
obtained	O
from	O
using	O
real-valued	O
chromosomes	O
.	O
</s>
<s>
This	O
was	O
explained	O
as	O
the	O
set	O
of	O
real	O
values	O
in	O
a	O
finite	O
population	O
of	O
chromosomes	O
as	O
forming	O
a	O
virtual	O
alphabet	O
(	O
when	O
selection	O
and	O
recombination	O
are	O
dominant	O
)	O
with	O
a	O
much	O
lower	O
cardinality	O
than	O
would	O
be	O
expected	O
from	O
a	O
floating	B-Algorithm
point	I-Algorithm
representation	I-Algorithm
.	O
</s>
<s>
An	O
expansion	O
of	O
the	O
Genetic	B-Algorithm
Algorithm	I-Algorithm
accessible	O
problem	O
domain	O
can	O
be	O
obtained	O
through	O
more	O
complex	O
encoding	O
of	O
the	O
solution	O
pools	O
by	O
concatenating	O
several	O
types	O
of	O
heterogenously	O
encoded	O
genes	O
into	O
one	O
chromosome	O
.	O
</s>
<s>
This	O
particular	O
form	O
of	O
encoding	O
requires	O
a	O
specialized	O
crossover	O
mechanism	O
that	O
recombines	O
the	O
chromosome	O
by	O
section	O
,	O
and	O
it	O
is	O
a	O
useful	O
tool	O
for	O
the	O
modelling	O
and	O
simulation	B-Application
of	O
complex	O
adaptive	O
systems	O
,	O
especially	O
evolution	O
processes	O
.	O
</s>
<s>
Parallel	B-Operating_System
implementations	O
of	O
genetic	B-Algorithm
algorithms	I-Algorithm
come	O
in	O
two	O
flavors	O
.	O
</s>
<s>
Coarse-grained	O
parallel	B-Operating_System
genetic	B-Algorithm
algorithms	I-Algorithm
assume	O
a	O
population	O
on	O
each	O
of	O
the	O
computer	O
nodes	O
and	O
migration	O
of	O
individuals	O
among	O
the	O
nodes	O
.	O
</s>
<s>
Fine-grained	O
parallel	B-Operating_System
genetic	B-Algorithm
algorithms	I-Algorithm
assume	O
an	O
individual	O
on	O
each	O
processor	O
node	O
which	O
acts	O
with	O
neighboring	O
individuals	O
for	O
selection	O
and	O
reproduction	O
.	O
</s>
<s>
Other	O
variants	O
,	O
like	O
genetic	B-Algorithm
algorithms	I-Algorithm
for	O
online	B-Algorithm
optimization	I-Algorithm
problems	O
,	O
introduce	O
time-dependence	O
or	O
noise	O
in	O
the	O
fitness	O
function	O
.	O
</s>
<s>
Genetic	B-Algorithm
algorithms	I-Algorithm
with	O
adaptive	O
parameters	O
(	O
adaptive	O
genetic	B-Algorithm
algorithms	I-Algorithm
,	O
AGAs	O
)	O
is	O
another	O
significant	O
and	O
promising	O
variant	O
of	O
genetic	B-Algorithm
algorithms	I-Algorithm
.	O
</s>
<s>
The	O
probabilities	O
of	O
crossover	O
(	O
pc	O
)	O
and	O
mutation	O
(	O
pm	O
)	O
greatly	O
determine	O
the	O
degree	O
of	O
solution	O
accuracy	O
and	O
the	O
convergence	O
speed	O
that	O
genetic	B-Algorithm
algorithms	I-Algorithm
can	O
obtain	O
.	O
</s>
<s>
In	O
AGA	O
(	O
adaptive	O
genetic	B-Algorithm
algorithm	I-Algorithm
)	O
,	O
the	O
adjustment	O
of	O
pc	O
and	O
pm	O
depends	O
on	O
the	O
fitness	O
values	O
of	O
the	O
solutions	O
.	O
</s>
<s>
In	O
CAGA	O
(	O
clustering-based	O
adaptive	O
genetic	B-Algorithm
algorithm	I-Algorithm
)	O
,	O
through	O
the	O
use	O
of	O
clustering	B-Algorithm
analysis	O
to	O
judge	O
the	O
optimization	O
states	O
of	O
the	O
population	O
,	O
the	O
adjustment	O
of	O
pc	O
and	O
pm	O
depends	O
on	O
these	O
optimization	O
states	O
.	O
</s>
<s>
Examples	O
are	O
dominance	O
&	O
co-dominance	O
principles	O
and	O
LIGA	O
(	O
levelized	O
interpolative	O
genetic	B-Algorithm
algorithm	I-Algorithm
)	O
,	O
which	O
combines	O
a	O
flexible	O
GA	O
with	O
modified	O
A*	O
search	O
to	O
tackle	O
search	O
space	O
anisotropicity	O
.	O
</s>
<s>
Other	O
techniques	O
(	O
such	O
as	O
simple	B-Algorithm
hill	I-Algorithm
climbing	I-Algorithm
)	O
are	O
quite	O
efficient	O
at	O
finding	O
absolute	O
optimum	O
in	O
a	O
limited	O
region	O
.	O
</s>
<s>
Alternating	O
GA	O
and	O
hill	B-Algorithm
climbing	I-Algorithm
can	O
improve	O
the	O
efficiency	O
of	O
GA	O
while	O
overcoming	O
the	O
lack	O
of	O
robustness	O
of	O
hill	B-Algorithm
climbing	I-Algorithm
.	O
</s>
<s>
Problems	O
which	O
appear	O
to	O
be	O
particularly	O
appropriate	O
for	O
solution	O
by	O
genetic	B-Algorithm
algorithms	I-Algorithm
include	O
timetabling	O
and	O
scheduling	O
problems	O
,	O
and	O
many	O
scheduling	O
software	O
packages	O
are	O
based	O
on	O
GAs	O
.	O
</s>
<s>
Genetic	B-Algorithm
algorithms	I-Algorithm
are	O
often	O
applied	O
as	O
an	O
approach	O
to	O
solve	O
global	O
optimization	O
problems	O
.	O
</s>
<s>
As	O
a	O
general	O
rule	O
of	O
thumb	O
genetic	B-Algorithm
algorithms	I-Algorithm
might	O
be	O
useful	O
in	O
problem	O
domains	O
that	O
have	O
a	O
complex	O
fitness	O
landscape	O
as	O
mixing	O
,	O
i.e.	O
,	O
mutation	O
in	O
combination	O
with	O
crossover	O
,	O
is	O
designed	O
to	O
move	O
the	O
population	O
away	O
from	O
local	B-Algorithm
optima	O
that	O
a	O
traditional	O
hill	B-Algorithm
climbing	I-Algorithm
algorithm	I-Algorithm
might	O
get	O
stuck	O
in	O
.	O
</s>
<s>
Mutation	O
alone	O
can	O
provide	O
ergodicity	O
of	O
the	O
overall	O
genetic	B-Algorithm
algorithm	I-Algorithm
process	O
(	O
seen	O
as	O
a	O
Markov	O
chain	O
)	O
.	O
</s>
<s>
In	O
his	O
Algorithm	O
Design	O
Manual	O
,	O
Skiena	O
advises	O
against	O
genetic	B-Algorithm
algorithms	I-Algorithm
for	O
any	O
task	O
:	O
</s>
<s>
In	O
1950	O
,	O
Alan	O
Turing	O
proposed	O
a	O
"	O
learning	O
machine	O
"	O
which	O
would	O
parallel	B-Operating_System
the	O
principles	O
of	O
evolution	O
.	O
</s>
<s>
Computer	B-Application
simulation	I-Application
of	O
evolution	O
started	O
as	O
early	O
as	O
in	O
1954	O
with	O
the	O
work	O
of	O
Nils	O
Aall	O
Barricelli	O
,	O
who	O
was	O
using	O
the	O
computer	O
at	O
the	O
Institute	O
for	O
Advanced	O
Study	O
in	O
Princeton	O
,	O
New	O
Jersey	O
.	O
</s>
<s>
Starting	O
in	O
1957	O
,	O
the	O
Australian	O
quantitative	O
geneticist	O
Alex	O
Fraser	O
published	O
a	O
series	O
of	O
papers	O
on	O
simulation	B-Application
of	O
artificial	O
selection	O
of	O
organisms	O
with	O
multiple	O
loci	O
controlling	O
a	O
measurable	O
trait	O
.	O
</s>
<s>
From	O
these	O
beginnings	O
,	O
computer	B-Application
simulation	I-Application
of	O
evolution	O
by	O
biologists	O
became	O
more	O
common	O
in	O
the	O
early	O
1960s	O
,	O
and	O
the	O
methods	O
were	O
described	O
in	O
books	O
by	O
Fraser	O
and	O
Burnell	O
(	O
1970	O
)	O
and	O
Crosby	O
(	O
1973	O
)	O
.	O
</s>
<s>
Fraser	O
's	O
simulations	O
included	O
all	O
of	O
the	O
essential	O
elements	O
of	O
modern	O
genetic	B-Algorithm
algorithms	I-Algorithm
.	O
</s>
<s>
Bremermann	O
's	O
research	O
also	O
included	O
the	O
elements	O
of	O
modern	O
genetic	B-Algorithm
algorithms	I-Algorithm
.	O
</s>
<s>
Although	O
Barricelli	O
,	O
in	O
work	O
he	O
reported	O
in	O
1963	O
,	O
had	O
simulated	O
the	O
evolution	O
of	O
ability	O
to	O
play	O
a	O
simple	O
game	O
,	O
artificial	B-Algorithm
evolution	I-Algorithm
only	O
became	O
a	O
widely	O
recognized	O
optimization	O
method	O
as	O
a	O
result	O
of	O
the	O
work	O
of	O
Ingo	O
Rechenberg	O
and	O
Hans-Paul	O
Schwefel	O
in	O
the	O
1960s	O
and	O
early	O
1970s	O
Rechenberg	O
's	O
group	O
was	O
able	O
to	O
solve	O
complex	O
engineering	O
problems	O
through	O
evolution	B-Algorithm
strategies	I-Algorithm
.	O
</s>
<s>
Another	O
approach	O
was	O
the	O
evolutionary	B-Algorithm
programming	I-Algorithm
technique	O
of	O
Lawrence	O
J	O
.	O
Fogel	O
,	O
which	O
was	O
proposed	O
for	O
generating	O
artificial	O
intelligence	O
.	O
</s>
<s>
Evolutionary	B-Algorithm
programming	I-Algorithm
originally	O
used	O
finite	O
state	O
machines	O
for	O
predicting	O
environments	O
,	O
and	O
used	O
variation	O
and	O
selection	O
to	O
optimize	O
the	O
predictive	O
logics	O
.	O
</s>
<s>
Genetic	B-Algorithm
algorithms	I-Algorithm
in	O
particular	O
became	O
popular	O
through	O
the	O
work	O
of	O
John	O
Holland	O
in	O
the	O
early	O
1970s	O
,	O
and	O
particularly	O
his	O
book	O
Adaptation	O
in	O
Natural	O
and	O
Artificial	O
Systems	O
(	O
1975	O
)	O
.	O
</s>
<s>
Holland	O
introduced	O
a	O
formalized	O
framework	O
for	O
predicting	O
the	O
quality	O
of	O
the	O
next	O
generation	O
,	O
known	O
as	O
Holland	O
's	O
Schema	B-Algorithm
Theorem	O
.	O
</s>
<s>
Research	O
in	O
GAs	O
remained	O
largely	O
theoretical	O
until	O
the	O
mid-1980s	O
,	O
when	O
The	O
First	O
International	O
Conference	O
on	O
Genetic	B-Algorithm
Algorithms	I-Algorithm
was	O
held	O
in	O
Pittsburgh	O
,	O
Pennsylvania	O
.	O
</s>
<s>
In	O
the	O
late	O
1980s	O
,	O
General	O
Electric	O
started	O
selling	O
the	O
world	O
's	O
first	O
genetic	B-Algorithm
algorithm	I-Algorithm
product	O
,	O
a	O
mainframe-based	O
toolkit	O
designed	O
for	O
industrial	O
processes	O
.	O
</s>
<s>
In	O
1989	O
,	O
Axcelis	O
,	O
Inc	O
.	O
released	O
Evolver	B-General_Concept
,	O
the	O
world	O
's	O
first	O
commercial	O
GA	O
product	O
for	O
desktop	O
computers	O
.	O
</s>
<s>
The	O
New	O
York	O
Times	O
technology	O
writer	O
John	O
Markoff	O
wrote	O
about	O
Evolver	B-General_Concept
in	O
1990	O
,	O
and	O
it	O
remained	O
the	O
only	O
interactive	O
commercial	O
genetic	B-Algorithm
algorithm	I-Algorithm
until	O
1995	O
.	O
</s>
<s>
Evolver	B-General_Concept
was	O
sold	O
to	O
Palisade	O
in	O
1997	O
,	O
translated	O
into	O
several	O
languages	O
,	O
and	O
is	O
currently	O
in	O
its	O
6th	O
version	O
.	O
</s>
<s>
Since	O
the	O
1990s	O
,	O
MATLAB	B-Language
has	O
built	O
in	O
three	O
derivative-free	B-Algorithm
optimization	I-Algorithm
heuristic	B-Algorithm
algorithms	I-Algorithm
(	O
simulated	B-Algorithm
annealing	I-Algorithm
,	O
particle	B-Algorithm
swarm	I-Algorithm
optimization	I-Algorithm
,	O
genetic	B-Algorithm
algorithm	I-Algorithm
)	O
and	O
two	O
direct	O
search	B-Application
algorithms	I-Application
(	O
simplex	O
search	O
,	O
pattern	O
search	O
)	O
.	O
</s>
<s>
Genetic	B-Algorithm
algorithms	I-Algorithm
are	O
a	O
sub-field	O
:	O
</s>
<s>
Evolutionary	B-Algorithm
algorithms	I-Algorithm
is	O
a	O
sub-field	O
of	O
evolutionary	O
computing	O
.	O
</s>
<s>
Evolution	B-Algorithm
strategies	I-Algorithm
(	O
ES	O
,	O
see	O
Rechenberg	O
,	O
1994	O
)	O
evolve	O
individuals	O
by	O
means	O
of	O
mutation	O
and	O
intermediate	O
or	O
discrete	O
recombination	O
.	O
</s>
<s>
De-randomization	O
of	O
self-adaptation	O
has	O
led	O
to	O
the	O
contemporary	O
Covariance	B-Algorithm
Matrix	I-Algorithm
Adaptation	I-Algorithm
Evolution	I-Algorithm
Strategy	I-Algorithm
(	O
CMA-ES	B-Algorithm
)	O
.	O
</s>
<s>
Evolutionary	B-Algorithm
programming	I-Algorithm
(	O
EP	O
)	O
involves	O
populations	O
of	O
solutions	O
with	O
primarily	O
mutation	O
and	O
selection	O
and	O
arbitrary	O
representations	O
.	O
</s>
<s>
Genetic	B-Algorithm
programming	I-Algorithm
(	O
GP	O
)	O
is	O
a	O
related	O
technique	O
popularized	O
by	O
John	O
Koza	O
in	O
which	O
computer	O
programs	O
,	O
rather	O
than	O
function	O
parameters	O
,	O
are	O
optimized	O
.	O
</s>
<s>
Genetic	B-Algorithm
programming	I-Algorithm
often	O
uses	O
tree-based	B-Application
internal	O
data	B-General_Concept
structures	I-General_Concept
to	O
represent	O
the	O
computer	O
programs	O
for	O
adaptation	O
instead	O
of	O
the	O
list	O
structures	O
typical	O
of	O
genetic	B-Algorithm
algorithms	I-Algorithm
.	O
</s>
<s>
There	O
are	O
many	O
variants	O
of	O
Genetic	B-Algorithm
Programming	I-Algorithm
,	O
including	O
Cartesian	B-Algorithm
genetic	I-Algorithm
programming	I-Algorithm
,	O
Gene	B-Algorithm
expression	I-Algorithm
programming	I-Algorithm
,	O
grammatical	B-Algorithm
evolution	I-Algorithm
,	O
Linear	B-Algorithm
genetic	I-Algorithm
programming	I-Algorithm
,	O
Multi	B-Algorithm
expression	I-Algorithm
programming	I-Algorithm
etc	O
.	O
</s>
<s>
Grouping	O
genetic	B-Algorithm
algorithm	I-Algorithm
(	O
GGA	O
)	O
is	O
an	O
evolution	O
of	O
the	O
GA	O
where	O
the	O
focus	O
is	O
shifted	O
from	O
individual	O
items	O
,	O
like	O
in	O
classical	O
GAs	O
,	O
to	O
groups	O
or	O
subset	O
of	O
items	O
.	O
</s>
<s>
clustering	B-Algorithm
or	O
partitioning	O
problems	O
where	O
a	O
set	O
of	O
items	O
must	O
be	O
split	O
into	O
disjoint	O
group	O
of	O
items	O
in	O
an	O
optimal	O
way	O
,	O
would	O
better	O
be	O
achieved	O
by	O
making	O
characteristics	O
of	O
the	O
groups	O
of	O
items	O
equivalent	O
to	O
genes	O
.	O
</s>
<s>
These	O
kind	O
of	O
problems	O
include	O
bin	O
packing	O
,	O
line	O
balancing	O
,	O
clustering	B-Algorithm
with	O
respect	O
to	O
a	O
distance	O
measure	O
,	O
equal	O
piles	O
,	O
etc.	O
,	O
on	O
which	O
classic	O
GAs	O
proved	O
to	O
perform	O
poorly	O
.	O
</s>
<s>
Interactive	O
evolutionary	B-Algorithm
algorithms	I-Algorithm
are	O
evolutionary	B-Algorithm
algorithms	I-Algorithm
that	O
use	O
human	O
evaluation	O
.	O
</s>
<s>
Swarm	B-Architecture
intelligence	I-Architecture
is	O
a	O
sub-field	O
of	O
evolutionary	O
computing	O
.	O
</s>
<s>
Ant	B-Algorithm
colony	I-Algorithm
optimization	I-Algorithm
(	O
ACO	O
)	O
uses	O
many	O
ants	O
(	O
or	O
agents	O
)	O
equipped	O
with	O
a	O
pheromone	O
model	O
to	O
traverse	O
the	O
solution	O
space	O
and	O
find	O
locally	O
productive	O
areas	O
.	O
</s>
<s>
Although	O
considered	O
an	O
Estimation	O
of	O
distribution	O
algorithm	O
,	O
Particle	B-Algorithm
swarm	I-Algorithm
optimization	I-Algorithm
(	O
PSO	O
)	O
is	O
a	O
computational	O
method	O
for	O
multi-parameter	O
optimization	O
which	O
also	O
uses	O
population-based	O
approach	O
.	O
</s>
<s>
Like	O
genetic	B-Algorithm
algorithms	I-Algorithm
,	O
the	O
PSO	O
method	O
depends	O
on	O
information	O
sharing	O
among	O
population	O
members	O
.	O
</s>
<s>
Evolutionary	O
computation	O
is	O
a	O
sub-field	O
of	O
the	O
metaheuristic	B-Algorithm
methods	O
.	O
</s>
<s>
Memetic	B-Algorithm
algorithm	I-Algorithm
(	O
MA	O
)	O
,	O
often	O
called	O
hybrid	B-Algorithm
genetic	I-Algorithm
algorithm	I-Algorithm
among	O
others	O
,	O
is	O
a	O
population-based	O
method	O
in	O
which	O
solutions	O
are	O
also	O
subject	O
to	O
local	B-Algorithm
improvement	O
phases	O
.	O
</s>
<s>
The	O
idea	O
of	O
memetic	B-Algorithm
algorithms	I-Algorithm
comes	O
from	O
memes	O
,	O
which	O
unlike	O
genes	O
,	O
can	O
adapt	O
themselves	O
.	O
</s>
<s>
In	O
some	O
problem	O
areas	O
they	O
are	O
shown	O
to	O
be	O
more	O
efficient	O
than	O
traditional	O
evolutionary	B-Algorithm
algorithms	I-Algorithm
.	O
</s>
<s>
Cultural	B-Algorithm
algorithm	I-Algorithm
(	O
CA	O
)	O
consists	O
of	O
the	O
population	O
component	O
almost	O
identical	O
to	O
that	O
of	O
the	O
genetic	B-Algorithm
algorithm	I-Algorithm
and	O
,	O
in	O
addition	O
,	O
a	O
knowledge	O
component	O
called	O
the	O
belief	O
space	O
.	O
</s>
<s>
Differential	B-Algorithm
evolution	I-Algorithm
(	O
DE	O
)	O
inspired	O
by	O
migration	O
of	O
superorganisms	O
.	O
</s>
<s>
Therefore	O
it	O
has	O
a	O
certain	O
"	O
ambition	O
"	O
to	O
avoid	O
local	B-Algorithm
peaks	O
in	O
the	O
fitness	O
landscape	O
.	O
</s>
<s>
Metaheuristic	B-Algorithm
methods	O
broadly	O
fall	O
within	O
stochastic	B-Algorithm
optimisation	I-Algorithm
methods	O
.	O
</s>
<s>
Simulated	B-Algorithm
annealing	I-Algorithm
(	O
SA	O
)	O
is	O
a	O
related	O
global	O
optimization	O
technique	O
that	O
traverses	O
the	O
search	O
space	O
by	O
testing	O
random	O
mutations	O
on	O
an	O
individual	O
solution	O
.	O
</s>
<s>
Tabu	B-Algorithm
search	I-Algorithm
(	O
TS	O
)	O
is	O
similar	O
to	O
simulated	B-Algorithm
annealing	I-Algorithm
in	O
that	O
both	O
traverse	O
the	O
solution	O
space	O
by	O
testing	O
mutations	O
of	O
an	O
individual	O
solution	O
.	O
</s>
<s>
While	O
simulated	B-Algorithm
annealing	I-Algorithm
generates	O
only	O
one	O
mutated	O
solution	O
,	O
tabu	B-Algorithm
search	I-Algorithm
generates	O
many	O
mutated	O
solutions	O
and	O
moves	O
to	O
the	O
solution	O
with	O
the	O
lowest	O
energy	O
of	O
those	O
generated	O
.	O
</s>
<s>
Extremal	B-Algorithm
optimization	I-Algorithm
(	O
EO	O
)	O
Unlike	O
GAs	O
,	O
which	O
work	O
with	O
a	O
population	O
of	O
candidate	O
solutions	O
,	O
EO	O
evolves	O
a	O
single	O
solution	O
and	O
makes	O
local	B-Algorithm
modifications	O
to	O
the	O
worst	O
components	O
.	O
</s>
<s>
The	O
parameters	O
are	O
updated	O
via	O
cross-entropy	O
minimization	O
,	O
so	O
as	O
to	O
generate	O
better	O
samples	O
in	O
the	O
next	O
iteration	B-Algorithm
.	O
</s>
<s>
Reactive	O
search	O
optimization	O
(	O
RSO	O
)	O
advocates	O
the	O
integration	O
of	O
sub-symbolic	O
machine	O
learning	O
techniques	O
into	O
search	O
heuristics	B-Algorithm
for	O
solving	O
complex	O
optimization	O
problems	O
.	O
</s>
<s>
Methodologies	O
of	O
interest	O
for	O
Reactive	O
Search	O
include	O
machine	O
learning	O
and	O
statistics	O
,	O
in	O
particular	O
reinforcement	O
learning	O
,	O
active	B-General_Concept
or	I-General_Concept
query	I-General_Concept
learning	I-General_Concept
,	O
neural	B-Architecture
networks	I-Architecture
,	O
and	O
metaheuristics	B-Algorithm
.	O
</s>
