<s>
In	O
statistics	O
,	O
EM	O
(	O
expectation	B-Algorithm
maximization	I-Algorithm
)	O
algorithm	O
handles	O
latent	O
variables	O
,	O
while	O
GMM	O
is	O
the	O
Gaussian	O
mixture	O
model	O
.	O
</s>
<s>
is	O
a	O
random	B-General_Concept
vector	I-General_Concept
such	O
as	O
,	O
and	O
from	O
medical	O
studies	O
it	O
is	O
known	O
that	O
are	O
normally	O
distributed	O
in	O
each	O
group	O
,	O
i.e.	O
</s>
<s>
not	O
observed	O
)	O
,	O
with	O
unlabeled	O
scenario	O
,	O
the	O
Expectation	B-Algorithm
Maximization	I-Algorithm
Algorithm	I-Algorithm
is	O
needed	O
to	O
estimate	O
as	O
well	O
as	O
other	O
parameters	O
.	O
</s>
<s>
With	O
the	O
EM	B-Algorithm
algorithm	I-Algorithm
,	O
some	O
underlying	O
pattern	O
in	O
the	O
data	O
can	O
be	O
found	O
,	O
along	O
with	O
the	O
estimation	O
of	O
the	O
parameters	O
.	O
</s>
<s>
The	O
wide	O
application	O
of	O
this	O
circumstance	O
in	O
machine	O
learning	O
is	O
what	O
makes	O
EM	B-Algorithm
algorithm	I-Algorithm
so	O
important	O
.	O
</s>
<s>
The	O
EM	B-Algorithm
algorithm	I-Algorithm
consists	O
of	O
two	O
steps	O
:	O
the	O
E-step	O
and	O
the	O
M-step	O
.	O
</s>
