Using a non-Gaussian prior
If the prior distributions on
the factors are not
Gaussian, some
orientations will be better
than others
It is better to generate
the data from factor
values that have high
probability under the
one big value and one
small value is more
likely than two medium
values that have the
same sum of squares.
If the prior for each hidden
activity is
the iso-probability contours
are straight lines at 45