<s>
Neural	B-Algorithm
network	I-Algorithm
software	I-Algorithm
is	O
used	O
to	O
simulate	O
,	O
research	O
,	O
develop	O
,	O
and	O
apply	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
,	O
software	O
concepts	O
adapted	O
from	O
biological	B-General_Concept
neural	I-General_Concept
networks	I-General_Concept
,	O
and	O
in	O
some	O
cases	O
,	O
a	O
wider	O
array	O
of	O
adaptive	O
systems	O
such	O
as	O
artificial	B-Application
intelligence	I-Application
and	O
machine	O
learning	O
.	O
</s>
<s>
Neural	B-Architecture
network	I-Architecture
simulators	O
are	O
software	O
applications	O
that	O
are	O
used	O
to	O
simulate	O
the	O
behavior	O
of	O
artificial	O
or	O
biological	B-General_Concept
neural	I-General_Concept
networks	I-General_Concept
.	O
</s>
<s>
They	O
focus	O
on	O
one	O
or	O
a	O
limited	O
number	O
of	O
specific	O
types	O
of	O
neural	B-Architecture
networks	I-Architecture
.	O
</s>
<s>
They	O
are	O
typically	O
stand-alone	O
and	O
not	O
intended	O
to	O
produce	O
general	O
neural	B-Architecture
networks	I-Architecture
that	O
can	O
be	O
integrated	O
in	O
other	O
software	B-Application
.	I-Application
</s>
<s>
Simulators	O
usually	O
have	O
some	O
form	O
of	O
built-in	O
visualization	B-Application
to	O
monitor	O
the	O
training	O
process	O
.	O
</s>
<s>
Some	O
simulators	O
also	O
visualize	O
the	O
physical	O
structure	O
of	O
the	O
neural	B-Architecture
network	I-Architecture
.	O
</s>
<s>
Historically	O
,	O
the	O
most	O
common	O
type	O
of	O
neural	B-Algorithm
network	I-Algorithm
software	I-Algorithm
was	O
intended	O
for	O
researching	O
neural	B-Architecture
network	I-Architecture
structures	O
and	O
algorithms	O
.	O
</s>
<s>
The	O
primary	O
purpose	O
of	O
this	O
type	O
of	O
software	O
is	O
,	O
through	O
simulation	O
,	O
to	O
gain	O
a	O
better	O
understanding	O
of	O
the	O
behavior	O
and	O
the	O
properties	O
of	O
neural	B-Architecture
networks	I-Architecture
.	O
</s>
<s>
Today	O
in	O
the	O
study	O
of	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
,	O
simulators	O
have	O
largely	O
been	O
replaced	O
by	O
more	O
general	O
component	O
based	O
development	O
environments	O
as	O
research	O
platforms	O
.	O
</s>
<s>
Commonly	O
used	O
artificial	B-Architecture
neural	I-Architecture
network	I-Architecture
simulators	O
include	O
the	O
Stuttgart	B-Algorithm
Neural	I-Algorithm
Network	I-Algorithm
Simulator	I-Algorithm
(	O
SNNS	B-Algorithm
)	O
,	O
Emergent	B-Algorithm
and	O
Neural	B-Algorithm
Lab	I-Algorithm
.	O
</s>
<s>
In	O
the	O
study	O
of	O
biological	B-General_Concept
neural	I-General_Concept
networks	I-General_Concept
however	O
,	O
simulation	O
software	O
is	O
still	O
the	O
only	O
available	O
approach	O
.	O
</s>
<s>
Commonly	O
used	O
biological	O
network	O
simulators	O
include	O
Neuron	O
,	O
GENESIS	B-Application
,	O
NEST	B-Application
and	O
Brian	B-Application
.	O
</s>
<s>
Unlike	O
the	O
research	O
simulators	O
,	O
data	O
analysis	O
simulators	O
are	O
intended	O
for	O
practical	O
applications	O
of	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
.	O
</s>
<s>
Their	O
primary	O
focus	O
is	O
on	O
data	B-Application
mining	I-Application
and	O
forecasting	O
.	O
</s>
<s>
Data	O
analysis	O
simulators	O
usually	O
have	O
some	O
form	O
of	O
preprocessing	B-General_Concept
capabilities	O
.	O
</s>
<s>
Unlike	O
the	O
more	O
general	O
development	O
environments	O
,	O
data	O
analysis	O
simulators	O
use	O
a	O
relatively	O
simple	O
static	O
neural	B-Architecture
network	I-Architecture
that	O
can	O
be	O
configured	O
.	O
</s>
<s>
Neural	B-Language
Designer	I-Language
is	O
one	O
example	O
of	O
a	O
data	O
analysis	O
simulator	O
.	O
</s>
<s>
were	O
released	O
in	O
1986-87	O
,	O
they	O
provided	O
some	O
relatively	O
simple	O
software	B-Application
.	I-Application
</s>
<s>
The	O
original	O
PDP	O
software	O
was	O
developed	O
into	O
a	O
more	O
powerful	O
package	O
called	O
PDP++	B-Algorithm
,	O
which	O
in	O
turn	O
has	O
become	O
an	O
even	O
more	O
powerful	O
platform	O
called	O
Emergent	B-Algorithm
.	O
</s>
<s>
With	O
each	O
development	O
,	O
the	B-Application
software	I-Application
has	O
become	O
more	O
powerful	O
,	O
but	O
also	O
more	O
daunting	O
for	O
use	O
by	O
beginners	O
.	O
</s>
<s>
In	O
2012	O
,	O
Wintempla	O
included	O
a	O
namespace	O
called	O
NN	O
with	O
a	O
set	O
of	O
C++	B-Language
classes	O
to	O
implement	O
:	O
feed	O
forward	O
networks	O
,	O
probabilistic	O
neural	B-Architecture
networks	I-Architecture
and	O
Kohonen	O
networks	O
.	O
</s>
<s>
Neural	B-Algorithm
Lab	I-Algorithm
is	O
based	O
on	O
Wintempla	O
classes	O
.	O
</s>
<s>
Neural	B-Algorithm
Lab	I-Algorithm
tutorial	O
and	O
Wintempla	O
tutorial	O
explains	O
some	O
of	O
these	O
classes	O
for	O
neural	B-Architecture
networks	I-Architecture
.	O
</s>
<s>
Development	O
environments	O
for	O
neural	B-Architecture
networks	I-Architecture
differ	O
from	O
the	B-Application
software	I-Application
described	O
above	O
primarily	O
on	O
two	O
accounts	O
–	O
they	O
can	O
be	O
used	O
to	O
develop	O
custom	O
types	O
of	O
neural	B-Architecture
networks	I-Architecture
and	O
they	O
support	O
deployment	O
of	O
the	O
neural	B-Architecture
network	I-Architecture
outside	O
the	O
environment	O
.	O
</s>
<s>
In	O
some	O
cases	O
they	O
have	O
advanced	O
preprocessing	B-General_Concept
,	O
analysis	O
and	O
visualization	B-Application
capabilities	O
.	O
</s>
<s>
A	O
more	O
modern	O
type	O
of	O
development	O
environments	O
that	O
are	O
currently	O
favored	O
in	O
both	O
industrial	O
and	O
scientific	O
use	O
are	O
based	O
on	O
a	O
component	B-Architecture
based	I-Architecture
paradigm	I-Architecture
.	O
</s>
<s>
The	O
neural	B-Architecture
network	I-Architecture
is	O
constructed	O
by	O
connecting	O
adaptive	O
filter	O
components	O
in	O
a	O
pipe	O
filter	O
flow	O
.	O
</s>
<s>
With	O
the	O
advent	O
of	O
component-based	B-Architecture
frameworks	O
such	O
as	O
.NET	O
and	O
Java	B-Language
,	O
component	O
based	O
development	O
environments	O
are	O
capable	O
of	O
deploying	O
the	O
developed	O
neural	B-Architecture
network	I-Architecture
to	O
these	O
frameworks	O
as	O
inheritable	O
components	O
.	O
</s>
<s>
In	O
addition	O
some	O
software	O
can	O
also	O
deploy	O
these	O
components	O
to	O
several	O
platforms	O
,	O
such	O
as	O
embedded	B-Architecture
systems	I-Architecture
.	O
</s>
<s>
Component	O
based	O
development	O
environments	O
include	O
:	O
Peltarion	B-Algorithm
Synapse	I-Algorithm
,	O
NeuroDimension	O
NeuroSolutions	B-Algorithm
,	O
Scientific	B-Application
Software	I-Application
Neuro	O
Laboratory	O
,	O
and	O
the	O
LIONsolver	B-Application
integrated	O
software	B-Application
.	I-Application
</s>
<s>
Free	O
open	B-Application
source	I-Application
component	O
based	O
environments	O
include	O
Encog	B-Language
and	O
Neuroph	B-Language
.	O
</s>
<s>
A	O
disadvantage	O
of	O
component-based	B-Architecture
development	I-Architecture
environments	O
is	O
that	O
they	O
are	O
more	O
complex	O
than	O
simulators	O
.	O
</s>
<s>
The	O
majority	O
implementations	O
of	O
neural	B-Architecture
networks	I-Architecture
available	O
are	O
however	O
custom	O
implementations	O
in	O
various	O
programming	O
languages	O
and	O
on	O
various	O
platforms	O
.	O
</s>
<s>
Basic	O
types	O
of	O
neural	B-Architecture
networks	I-Architecture
are	O
simple	O
to	O
implement	O
directly	O
.	O
</s>
<s>
There	O
are	O
also	O
many	O
programming	B-Library
libraries	I-Library
that	O
contain	O
neural	B-Architecture
network	I-Architecture
functionality	O
and	O
that	O
can	O
be	O
used	O
in	O
custom	O
implementations	O
(	O
such	O
as	O
TensorFlow	B-Language
,	O
Theano	B-Algorithm
,	O
etc.	O
,	O
typically	O
providing	O
bindings	O
to	O
languages	O
such	O
as	O
Python	B-Language
,	O
C++	B-Language
,	O
Java	B-Language
)	O
.	O
</s>
<s>
In	O
order	O
for	O
neural	B-Architecture
network	I-Architecture
models	I-Architecture
to	O
be	O
shared	O
by	O
different	O
applications	O
,	O
a	O
common	O
language	O
is	O
necessary	O
.	O
</s>
<s>
The	O
Predictive	B-Application
Model	I-Application
Markup	I-Application
Language	I-Application
(	O
PMML	B-Application
)	O
has	O
been	O
proposed	O
to	O
address	O
this	O
need	O
.	O
</s>
<s>
PMML	B-Application
is	O
an	O
XML-based	O
language	O
which	O
provides	O
a	O
way	O
for	O
applications	O
to	O
define	O
and	O
share	O
neural	B-Architecture
network	I-Architecture
models	I-Architecture
(	O
and	O
other	O
data	B-Application
mining	I-Application
models	O
)	O
between	O
PMML	B-Application
compliant	O
applications	O
.	O
</s>
<s>
PMML	B-Application
provides	O
applications	O
a	O
vendor-independent	O
method	O
of	O
defining	O
models	O
so	O
that	O
proprietary	O
issues	O
and	O
incompatibilities	O
are	O
no	O
longer	O
a	O
barrier	O
to	O
the	O
exchange	O
of	O
models	O
between	O
applications	O
.	O
</s>
<s>
Previously	O
,	O
this	O
was	O
very	O
difficult	O
,	O
but	O
with	O
PMML	B-Application
,	O
the	O
exchange	O
of	O
models	O
between	O
compliant	O
applications	O
is	O
now	O
straightforward	O
.	O
</s>
<s>
A	O
range	O
of	O
products	O
are	O
being	O
offered	O
to	O
produce	O
and	O
consume	O
PMML	B-Application
.	O
</s>
<s>
This	O
ever-growing	O
list	O
includes	O
the	O
following	O
neural	B-Architecture
network	I-Architecture
products	O
:	O
</s>
<s>
R	O
:	O
produces	O
PMML	B-Application
for	O
neural	B-Architecture
nets	I-Architecture
and	O
other	O
machine	O
learning	O
models	O
via	O
the	O
package	O
pmml	B-Application
.	O
</s>
<s>
SAS	O
Enterprise	O
Miner	O
:	O
produces	O
PMML	B-Application
for	O
several	O
mining	O
models	O
,	O
including	O
neural	B-Architecture
networks	I-Architecture
,	O
linear	O
and	O
logistic	O
regression	O
,	O
decision	O
trees	O
,	O
and	O
other	O
data	B-Application
mining	I-Application
models	O
.	O
</s>
<s>
SPSS	O
:	O
produces	O
PMML	B-Application
for	O
neural	B-Architecture
networks	I-Architecture
as	O
well	O
as	O
many	O
other	O
mining	O
models	O
.	O
</s>
<s>
STATISTICA	O
:	O
produces	O
PMML	B-Application
for	O
neural	B-Architecture
networks	I-Architecture
,	O
data	B-Application
mining	I-Application
models	O
and	O
traditional	O
statistical	O
models	O
.	O
</s>
