<s>
Waffles	B-Algorithm
is	O
a	O
collection	O
of	O
command-line	O
tools	O
for	O
performing	O
machine	O
learning	O
operations	O
developed	O
at	O
Brigham	O
Young	O
University	O
.	O
</s>
<s>
These	O
tools	O
are	O
written	O
in	O
C++	B-Language
,	O
and	O
are	O
available	O
under	O
the	O
GNU	B-Application
Lesser	I-Application
General	I-Application
Public	I-Application
License	I-Application
.	O
</s>
<s>
The	O
Waffles	B-Algorithm
machine	O
learning	O
toolkit	O
contains	O
command-line	O
tools	O
for	O
performing	O
various	O
operations	O
related	O
to	O
machine	O
learning	O
,	O
data	B-Application
mining	I-Application
,	O
and	O
predictive	B-General_Concept
modeling	I-General_Concept
.	O
</s>
<s>
The	O
primary	O
focus	O
of	O
Waffles	B-Algorithm
is	O
to	O
provide	O
tools	O
that	O
are	O
simple	O
to	O
use	O
in	O
scripted	O
experiments	O
or	O
processes	O
.	O
</s>
<s>
For	O
example	O
,	O
the	O
supervised	O
learning	O
algorithms	O
included	O
in	O
Waffles	B-Algorithm
are	O
all	O
designed	O
to	O
support	O
multi-dimensional	O
labels	O
,	O
classification	B-General_Concept
and	O
regression	O
,	O
automatically	O
impute	O
missing	O
values	O
,	O
and	O
automatically	O
apply	O
necessary	O
filters	O
to	O
transform	O
the	O
data	O
to	O
a	O
type	O
that	O
the	O
algorithm	O
can	O
support	O
,	O
such	O
that	O
arbitrary	O
learning	O
algorithms	O
can	O
be	O
used	O
with	O
arbitrary	O
data	O
sets	O
.	O
</s>
<s>
The	O
algorithms	O
provided	O
in	O
Waffles	B-Algorithm
also	O
have	O
the	O
ability	O
to	O
automatically	O
tune	O
their	O
own	O
parameters	O
(	O
with	O
the	O
cost	O
of	O
additional	O
computational	O
overhead	O
)	O
.	O
</s>
<s>
Because	O
Waffles	B-Algorithm
is	O
designed	O
for	O
script-ability	O
,	O
it	O
deliberately	O
avoids	O
presenting	O
its	O
tools	O
in	O
a	O
graphical	O
environment	O
.	O
</s>
<s>
All	O
of	O
the	O
Waffles	B-Algorithm
tools	O
are	O
implemented	O
as	O
thin	O
wrappers	O
around	O
functionality	O
in	O
a	O
C++	B-Language
class	O
library	O
.	O
</s>
<s>
Waffles	B-Algorithm
was	O
first	O
released	O
as	O
an	O
open	O
source	O
project	O
in	O
2005	O
.	O
</s>
<s>
Waffles	B-Algorithm
is	O
not	O
an	O
acronym	O
—	O
the	O
toolkit	O
was	O
named	O
after	O
the	O
food	O
for	O
historical	O
reasons	O
.	O
</s>
<s>
Some	O
of	O
the	O
advantages	O
of	O
Waffles	B-Algorithm
in	O
contrast	O
with	O
other	O
popular	O
open	O
source	O
machine	O
learning	O
toolkits	O
include	O
:	O
</s>
<s>
Waffles	B-Algorithm
automatically	O
takes	O
care	O
of	O
many	O
issues	O
related	O
to	O
data	O
format	O
in	O
order	O
to	O
simplify	O
its	O
tools	O
.	O
</s>
<s>
Because	O
it	O
is	O
implemented	O
in	O
C++	B-Language
,	O
many	O
of	O
its	O
algorithms	O
are	O
particularly	O
fast	O
.	O
</s>
<s>
The	O
functionality	O
included	O
in	O
Waffles	B-Algorithm
is	O
very	O
broad	O
,	O
including	O
algorithms	O
for	O
dimensionality	B-Algorithm
reduction	I-Algorithm
,	O
collaborative	B-Algorithm
filtering	I-Algorithm
,	O
visualization	O
,	O
clustering	O
,	O
supervised	O
learning	O
,	O
optimization	O
,	O
linear	O
algebra	O
,	O
data	O
transformation	O
,	O
image	O
and	O
signal	O
processing	O
,	O
policy	O
learning	O
,	O
and	O
sparse	O
matrix	O
operations	O
.	O
</s>
<s>
Although	O
Waffles	B-Algorithm
provides	O
significant	O
breadth	O
,	O
it	O
lacks	O
the	O
depth	O
of	O
many	O
toolkits	O
that	O
focus	O
on	O
a	O
particular	O
area	O
of	O
machine	O
learning	O
.	O
</s>
<s>
The	O
Weka	B-Language
(	O
machine	O
learning	O
)	O
toolkit	O
,	O
for	O
example	O
,	O
provides	O
many	O
more	O
classification	B-General_Concept
algorithms	O
than	O
Waffles	B-Algorithm
provides	O
.	O
</s>
<s>
Waffles	B-Algorithm
only	O
has	O
a	O
limited	O
graphical	O
interface	O
.	O
</s>
