<s>
The	O
Fly	B-Algorithm
Algorithm	I-Algorithm
is	O
a	O
type	O
of	O
cooperative	O
coevolution	O
based	O
on	O
the	O
Parisian	O
approach	O
.	O
</s>
<s>
The	O
Fly	B-Algorithm
Algorithm	I-Algorithm
has	O
first	O
been	O
developed	O
in	O
1999	O
in	O
the	O
scope	O
of	O
the	O
application	O
of	O
Evolutionary	B-Algorithm
algorithms	I-Algorithm
to	O
computer	B-Algorithm
stereo	I-Algorithm
vision	I-Algorithm
.	O
</s>
<s>
The	O
first	O
application	O
field	O
of	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
has	O
been	O
stereovision	O
.	O
</s>
<s>
While	O
classical	O
`image	O
priority	O
 '	O
approaches	O
use	O
matching	O
features	O
from	O
the	O
stereo	O
images	O
in	O
order	O
to	O
build	O
a	O
3-D	O
model	O
,	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
directly	O
explores	O
the	O
3-D	O
space	O
and	O
uses	O
image	O
data	O
to	O
evaluate	O
the	O
validity	O
of	O
3-D	O
hypotheses	O
.	O
</s>
<s>
The	O
use	O
of	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
is	O
not	O
strictly	O
restricted	O
to	O
stereo	O
images	O
,	O
as	O
other	O
sensors	O
may	O
be	O
added	O
(	O
e.g.	O
</s>
<s>
Another	O
application	O
field	O
of	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
is	O
reconstruction	O
for	O
emission	O
Tomography	O
in	O
nuclear	O
medicine	O
.	O
</s>
<s>
It	O
is	O
based	O
on	O
the	O
leave-one-out	B-Application
cross-validation	I-Application
principle	O
.	O
</s>
<s>
More	O
recently	O
it	O
has	O
been	O
used	O
in	O
digital	O
art	O
to	O
generate	O
mosaic-like	O
images	O
or	O
spray	O
paint	O
.	O
</s>
<s>
This	O
is	O
implemented	O
using	O
an	O
evolutionary	B-Algorithm
algorithm	I-Algorithm
that	O
includes	O
all	O
the	O
common	O
genetic	O
operators	O
(	O
e.g.	O
</s>
<s>
Cooperative	O
coevolution	O
is	O
a	O
broad	O
class	O
of	O
evolutionary	B-Algorithm
algorithms	I-Algorithm
where	O
a	O
complex	O
problem	O
is	O
solved	O
by	O
decomposing	O
it	O
into	O
subcomponents	O
that	O
are	O
solved	O
independently	O
.	O
</s>
<s>
Similar	O
internal	O
evolutionary	O
engines	O
are	O
considered	O
in	O
classical	O
evolutionary	B-Algorithm
algorithm	I-Algorithm
,	O
cooperative	O
coevolutionary	O
algorithm	O
and	O
Parisian	O
evolution	O
.	O
</s>
<s>
However	O
,	O
Parisian	O
evolutionary	B-Algorithm
algorithms	I-Algorithm
solve	O
a	O
whole	O
problem	O
as	O
a	O
big	O
component	O
.	O
</s>
<s>
Cooperative	O
coevolution	O
and	O
particle	B-Algorithm
swarm	I-Algorithm
optimisation	I-Algorithm
(	O
PSO	B-Algorithm
)	O
share	O
many	O
similarities	O
.	O
</s>
<s>
PSO	B-Algorithm
is	O
inspired	O
by	O
the	O
social	O
behaviour	O
of	O
bird	O
flocking	O
or	O
fish	O
schooling	O
.	O
</s>
<s>
In	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
,	O
the	O
flies	O
aim	O
at	O
building	O
spatial	O
representations	O
of	O
a	O
scene	O
from	O
actual	O
sensor	O
data	O
;	O
flies	O
do	O
not	O
communicate	O
or	O
explicitly	O
cooperate	O
,	O
and	O
do	O
not	O
use	O
any	O
behavioural	O
model	O
.	O
</s>
<s>
However	O
,	O
the	O
solution	O
of	O
the	O
optimisation	O
problem	O
in	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
is	O
the	O
population	O
(	O
or	O
a	O
subset	O
of	O
the	O
population	O
)	O
:	O
The	O
flies	O
implicitly	O
collaborate	O
to	O
build	O
the	O
solution	O
.	O
</s>
<s>
In	O
PSO	B-Algorithm
the	O
solution	O
is	O
a	O
single	O
particle	O
,	O
the	O
one	O
with	O
the	O
best	O
fitness	O
.	O
</s>
<s>
Another	O
main	O
difference	O
between	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
and	O
with	O
PSO	B-Algorithm
is	O
that	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
is	O
not	O
based	O
on	O
any	O
behavioural	O
model	O
but	O
only	O
builds	O
a	O
geometrical	O
representation	O
.	O
</s>
<s>
Tomography	O
reconstruction	O
is	O
an	O
inverse	O
problem	O
that	O
is	O
often	O
ill-posed	B-Algorithm
due	O
to	O
missing	O
data	O
and/or	O
noise	O
.	O
</s>
<s>
The	O
input	O
data	O
of	O
a	O
reconstruction	O
algorithm	O
may	O
be	O
given	O
as	O
the	O
Radon	B-Algorithm
transform	I-Algorithm
or	O
sinogram	O
of	O
the	O
data	O
to	O
reconstruct	O
.	O
</s>
<s>
In	O
this	O
case	O
the	O
reconstruction	O
corresponds	O
to	O
the	O
inversion	O
of	O
the	O
Radon	B-Algorithm
transform	I-Algorithm
:	O
</s>
<s>
The	O
Fly	B-Algorithm
Algorithm	I-Algorithm
is	O
an	O
example	O
of	O
iterative	B-Algorithm
reconstruction	I-Algorithm
.	O
</s>
<s>
Iterative	O
methods	O
in	O
tomographic	B-Algorithm
reconstruction	I-Algorithm
are	O
relatively	O
easy	O
to	O
model	O
:	O
</s>
<s>
The	O
pseudocode	B-Language
below	O
is	O
a	O
step-by-step	O
description	O
of	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
for	O
tomographic	B-Algorithm
reconstruction	I-Algorithm
.	O
</s>
<s>
A	O
JavaScript	B-Language
implementation	O
can	O
be	O
found	O
on	O
.	O
</s>
<s>
In	O
this	O
example	O
,	O
an	O
input	O
image	O
is	O
to	O
be	O
approximated	O
by	O
a	O
set	O
of	O
tiles	O
(	O
for	O
example	O
as	O
in	O
an	O
ancient	O
mosaic	B-Application
)	O
.	O
</s>
<s>
Using	O
a	O
classical	O
evolutionary	B-Algorithm
algorithm	I-Algorithm
where	O
the	O
answer	O
of	O
the	O
optimisation	O
problem	O
is	O
the	O
best	O
individual	O
,	O
the	O
genome	O
of	O
an	O
individual	O
would	O
be	O
made	O
up	O
of	O
45,000	O
genes	O
.	O
</s>
<s>
Using	O
the	O
Fly	B-Algorithm
Algorithm	I-Algorithm
,	O
every	O
individual	O
mimics	O
a	O
tile	O
and	O
can	O
be	O
individually	O
evaluated	O
using	O
its	O
local	O
fitness	O
to	O
assess	O
its	O
contribution	O
to	O
the	O
population	O
's	O
performance	O
(	O
the	O
global	O
fitness	O
)	O
.	O
</s>
<s>
In	O
her	O
work	O
,	O
Z	O
.	O
Ali	O
Aboodd	O
uses	O
OpenGL	B-Application
to	O
generate	O
different	O
effects	O
(	O
e.g.	O
</s>
<s>
mosaics	B-Application
,	O
or	O
spray	O
paint	O
)	O
.	O
</s>
<s>
For	O
speeding	O
up	O
the	O
evaluation	O
of	O
the	O
fitness	O
functions	O
,	O
OpenCL	B-Application
is	O
used	O
too	O
.	O
</s>
