Next: Robustness
Up: Behavior under the ICA
Previous: Consistency
Asymptotic variance is
one criterion for choosing the
function G to be used in the contrast function. Comparison of, say,
the traces of the asymptotic covariance matrices of two estimators
enables direct comparison of the mean-square error of the estimators.
In [18], evaluation of asymptotic variances was addressed
using a related
family of contrast functions. In fact, it can be
seen that the results in [18] are valid even in this case,
and thus we have the following theorem:
Theorem 2
The trace of the asymptotic (co)variance
of
is minimized when
G is of the form
|
(10) |
where
fi(.) is the density function of
si, and
k1,
k2,
k3 are
arbitrary constants.
For simplicity, one can
choose
.
Thus the optimal contrast function is the same as
the one obtained by the maximum likelihood approach
[34], or the infomax approach [3]. Almost
identical results have also
been obtained in [5] for another algorithm.
The theorem above treats, however, the one-unit case instead of the
multi-unit case treated by the other authors.
Next: Robustness
Up: Behavior under the ICA
Previous: Consistency
Aapo Hyvarinen
1999-04-23