<s>
Joint	B-Algorithm
Approximation	I-Algorithm
Diagonalization	I-Algorithm
of	I-Algorithm
Eigen-matrices	I-Algorithm
(	O
JADE	B-Algorithm
)	O
is	O
an	O
algorithm	O
for	O
independent	B-Algorithm
component	I-Algorithm
analysis	I-Algorithm
that	O
separates	O
observed	O
mixed	O
signals	O
into	O
latent	O
source	O
signals	O
by	O
exploiting	O
fourth	O
order	O
moments	O
.	O
</s>
<s>
The	O
motivation	O
for	O
this	O
measure	O
is	O
that	O
Gaussian	O
distributions	O
possess	O
zero	O
excess	O
kurtosis	B-Error_Name
,	O
and	O
with	O
non-Gaussianity	O
being	O
a	O
canonical	O
assumption	O
of	O
ICA	O
,	O
JADE	B-Algorithm
seeks	O
an	O
orthogonal	O
rotation	O
of	O
the	O
observed	O
mixed	O
vectors	O
to	O
estimate	O
source	O
vectors	O
which	O
possess	O
high	O
values	O
of	O
excess	O
kurtosis	B-Error_Name
.	O
</s>
