On Fri, Jul 11, 2008 at 9:10 PM, Robin Anil <[EMAIL PROTECTED]> wrote:

> On Sat, Jul 12, 2008 at 5:43 AM, Ted Dunning <[EMAIL PROTECTED]>
> wrote:
>
> >  Remember that sparsity can be
> > relative to an offset so that if a particular count is common for a
> class,
> > feature compound, you can carry along that count as a default.
>
>
> I am putting the formulation here. If anyone gets an idea please post
>
> Can you formulate this in terms of matrix operations.

Start with the example x feature matrix.

What happens from there (in terms of matrices).

Cant say i have the plan from what you said. Will have to read up on good
> ordering techniques to see if it can be scaled up.
>
>
See this book:
http://www.amazon.com/Computations-Hopkins-Studies-Mathematical-Sciences/dp/0801854148

-- 
ted

Reply via email to