<s>
Information	B-Algorithm
fuzzy	I-Algorithm
networks	I-Algorithm
(	O
IFN	O
)	O
is	O
a	O
greedy	B-Algorithm
machine	O
learning	O
algorithm	O
for	O
supervised	B-General_Concept
learning	I-General_Concept
.	O
</s>
<s>
The	O
data	B-General_Concept
structure	I-General_Concept
produced	O
by	O
the	O
learning	O
algorithm	O
is	O
also	O
called	O
Info	O
Fuzzy	O
Network	O
.	O
</s>
<s>
IFN	O
construction	O
is	O
quite	O
similar	O
to	O
decision	B-Algorithm
trees	I-Algorithm
 '	I-Algorithm
construction	O
.	O
</s>
<s>
However	O
,	O
IFN	O
constructs	O
a	O
directed	O
graph	O
and	O
not	O
a	O
tree	B-Application
.	O
</s>
<s>
IFN	O
also	O
uses	O
the	O
conditional	O
mutual	O
information	O
metric	O
in	O
order	O
to	O
choose	O
features	O
during	O
the	O
construction	O
stage	O
while	O
decision	B-Algorithm
trees	I-Algorithm
usually	O
use	O
other	O
metrics	O
like	O
entropy	O
or	O
gini	O
.	O
</s>
<s>
The	O
IFN	O
model	O
partially	O
solves	O
the	O
fragmentation	O
problem	O
that	O
occurs	O
in	O
decision	B-Algorithm
trees	I-Algorithm
(	O
the	O
deeper	O
the	O
node	O
the	O
less	O
records	O
it	O
represent	O
.	O
</s>
<s>
Hence	O
,	O
the	O
number	O
of	O
records	O
might	O
be	O
too	O
low	O
for	O
statistical	B-General_Concept
significance	I-General_Concept
indication	O
)	O
since	O
the	O
entire	O
set	O
of	O
records	O
is	O
used	O
in	O
every	O
layer	O
.	O
</s>
<s>
IFN	O
was	O
compared	O
on	O
few	O
common	O
datasets	O
to	O
the	O
c4.5	B-Algorithm
decision	B-Algorithm
tree	I-Algorithm
algorithm	O
.	O
</s>
<s>
The	O
accuracy	O
of	O
the	O
IFN	O
was	O
smaller	O
than	O
the	O
one	O
of	O
the	O
decision	B-Algorithm
tree	I-Algorithm
.	O
</s>
<s>
Input	O
:	O
a	O
list	O
of	O
input	O
variables	O
that	O
can	O
be	O
used	O
,	O
a	O
list	O
of	O
data	O
records	O
(	O
training	O
set	O
)	O
and	O
a	O
minimal	O
statistical	B-General_Concept
significance	I-General_Concept
used	O
to	O
decide	O
whether	O
to	O
split	O
a	O
node	O
or	O
not	O
(	O
default	O
0.1	O
%	O
)	O
.	O
</s>
<s>
Create	O
the	O
root	B-Application
node	I-Application
and	O
the	O
layer	O
of	O
the	O
target	O
variable	O
.	O
</s>
<s>
Loop	O
until	O
we	O
have	O
used	O
up	O
all	O
the	O
attributes	O
or	O
it	O
cannot	O
improve	O
the	O
conditional	O
mutual	O
information	O
any	O
more	O
with	O
any	O
statistical	B-General_Concept
significance	I-General_Concept
.	O
</s>
<s>
Verify	O
that	O
the	O
contribution	O
of	O
the	O
attribute	O
has	O
statistical	B-General_Concept
significance	I-General_Concept
using	O
the	O
likelihood	B-General_Concept
ratio	I-General_Concept
test	I-General_Concept
.	O
</s>
<s>
Split	O
any	O
node	O
in	O
the	O
previous	O
layer	O
if	O
the	O
contribution	O
of	O
the	O
current	O
attribute	O
has	O
statistical	B-General_Concept
significance	I-General_Concept
.	O
</s>
