<s>
ALOPEX	B-Algorithm
(	O
an	O
acronym	O
from	O
"	O
ALgorithms	O
Of	O
Pattern	O
EXtraction	O
"	O
)	O
is	O
a	O
correlation	O
based	O
machine	O
learning	O
algorithm	O
first	O
proposed	O
by	O
Tzanakou	O
and	O
Harth	O
in	O
1974	O
.	O
</s>
<s>
In	O
machine	O
learning	O
,	O
the	O
goal	O
is	O
to	O
train	O
a	O
system	O
to	O
minimize	O
a	O
cost	O
function	O
or	O
(	O
referring	O
to	O
ALOPEX	B-Algorithm
)	O
a	O
response	O
function	O
.	O
</s>
<s>
Many	O
training	O
algorithms	O
,	O
such	O
as	O
backpropagation	B-Algorithm
,	O
have	O
an	O
inherent	O
susceptibility	O
to	O
getting	O
"	O
stuck	O
"	O
in	O
local	O
minima	O
or	O
maxima	O
of	O
the	O
response	O
function	O
.	O
</s>
<s>
ALOPEX	B-Algorithm
uses	O
a	O
cross-correlation	O
of	O
differences	O
and	O
a	O
stochastic	O
process	O
to	O
overcome	O
this	O
in	O
an	O
attempt	O
to	O
reach	O
the	O
absolute	O
minimum	O
(	O
or	O
maximum	O
)	O
of	O
the	O
response	O
function	O
.	O
</s>
<s>
ALOPEX	B-Algorithm
,	O
in	O
its	O
simplest	O
form	O
is	O
defined	O
by	O
an	O
updating	O
equation	O
:	O
</s>
<s>
Essentially	O
,	O
ALOPEX	B-Algorithm
changes	O
each	O
system	O
variable	O
based	O
on	O
a	O
product	O
of	O
:	O
the	O
previous	O
change	O
in	O
the	O
variable	O
,	O
the	O
resulting	O
change	O
in	O
the	O
cost	O
function	O
,	O
and	O
the	O
learning	O
rate	O
parameter	O
.	O
</s>
