<s>
Compositional	B-Algorithm
pattern-producing	I-Algorithm
networks	I-Algorithm
(	O
CPPNs	B-Algorithm
)	O
are	O
a	O
variation	O
of	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
(	O
ANNs	O
)	O
that	O
have	O
an	O
architecture	O
whose	O
evolution	O
is	O
guided	O
by	O
genetic	O
algorithms	O
.	O
</s>
<s>
While	O
ANNs	O
often	O
contain	O
only	O
sigmoid	B-Algorithm
functions	I-Algorithm
and	O
sometimes	O
Gaussian	O
functions	O
,	O
CPPNs	B-Algorithm
can	O
include	O
both	O
types	O
of	O
functions	O
and	O
many	O
others	O
.	O
</s>
<s>
Linear	O
functions	O
can	O
be	O
employed	O
to	O
produce	O
linear	O
or	O
fractal-like	O
patterns	O
.	O
</s>
<s>
Thus	O
,	O
the	O
architect	O
of	O
a	O
CPPN-based	O
genetic	B-Algorithm
art	I-Algorithm
system	O
can	O
bias	O
the	O
types	O
of	O
patterns	O
it	O
generates	O
by	O
deciding	O
the	O
set	O
of	O
canonical	O
functions	O
to	O
include	O
.	O
</s>
<s>
Furthermore	O
,	O
unlike	O
typical	O
ANNs	O
,	O
CPPNs	B-Algorithm
are	O
applied	O
across	O
the	O
entire	O
space	O
of	O
possible	O
inputs	O
so	O
that	O
they	O
can	O
represent	O
a	O
complete	O
image	O
.	O
</s>
<s>
Since	O
they	O
are	O
compositions	O
of	O
functions	O
,	O
CPPNs	B-Algorithm
in	O
effect	O
encode	O
images	O
at	O
infinite	O
resolution	O
and	O
can	O
be	O
sampled	O
for	O
a	O
particular	O
display	O
at	O
whatever	O
resolution	O
is	O
optimal	O
.	O
</s>
<s>
CPPNs	B-Algorithm
can	O
be	O
evolved	O
through	O
neuroevolution	B-Algorithm
techniques	O
such	O
as	O
neuroevolution	B-Algorithm
of	I-Algorithm
augmenting	I-Algorithm
topologies	I-Algorithm
(	O
called	O
CPPN-NEAT	O
)	O
.	O
</s>
<s>
CPPNs	B-Algorithm
have	O
been	O
shown	O
to	O
be	O
a	O
very	O
powerful	O
encoding	O
when	O
evolving	O
the	O
following	O
:	O
</s>
<s>
Neural	B-Architecture
networks	I-Architecture
,	O
via	O
the	O
HyperNEAT	B-Algorithm
algorithm	O
,	O
</s>
