Ok, if you say so. Do you think it would be worth considering adding PCA related options to SSVD stuff?
On Tue, Sep 6, 2011 at 1:51 PM, Ted Dunning <[email protected]> wrote: > On Tue, Sep 6, 2011 at 12:42 PM, Dmitriy Lyubimov <[email protected]> wrote: > >> >> yes, there are tricks but that would change SVD contract and amend the >> method, which is why I asked "without changing SVD contract", i.e. >> without changing what A, U, Sigma and V mean in SVD step. >> > > In a conventional program, changing the data types isn't really changing the > contract. > > But I would guess that the MAHOUT-792 sequential code would run in roughly > finite time on the data from the original poster. A multi-threaded version > would have an even better chance. >
