On Mon, Sep 10, 2012 at 1:11 PM, Pat Ferrel <[email protected]> wrote:

> Got it.
>
> Maybe it would be enough to do the projection into PCA space and back
> project only a very very few individual vectors (cluster centroids and
> special docs).


Yes.   On-demand fold-out works well.


> As to density we can also do with only the top few terms so we could
> drastically compress the dense vector. Still that would require calculating
> the inverse of V I think.
>

The inverse of V is trivial.  It is orthonormal so the inverse is the
transpose.

Reply via email to