[mlpack] regression using sparse matrices/vectors

Ryan Curtin gth671b at mail.gatech.edu
Tue Dec 10 12:09:13 EST 2013


On Mon, Dec 09, 2013 at 08:21:48PM -0500, Darryl Reeves wrote:
> Hello,
> 
> I am interested in using mlpack in a project that I am working on. I know
> that there is some support for sparse matrices in the project but can you
> let me know if this support is available for mlpack's regression methods?

Hello Darryl,

There are a couple ways sparse matrices could be used in the context of
regression.  If your input data features are sparse, then this is pretty
simple (although some minor code modification needs to be done to
templatize the regression algorithms so you can use a sparse matrix).
But the parameters of the model will not be sparse.

On the other hand, if you are looking for a sparse parameter vector, you
could use LARS with L1 regularization.

Does this help?

Thanks,

Ryan

-- 
Ryan Curtin       | "Weeee!"
ryan at igglybob.com |   - Bobby


More information about the mlpack mailing list