repetitive training

[David Relson]

On Tue, 09 Mar 2004 13:10:50 +0100
Boris 'pi' Piwinger wrote:

> Greg Louis wrote:

...[snip]...

> > The most important disagreement we have is over the value of
> > reoptimizing parameters after training; I claim it's moderately
> > useful to do so, pi seems to feel it's too far from what happens in
> > production
> 
> Even worse, I feel it will break the training so far which
> is made to fit with the given parameters.
> 
> So what I would like to see in you latest experiment are the
> actual parameters used for each round. And possibly for each
> round the error rate before and after the tuning.

pi,

Try thinking of it this way:

The wordlist contains a history of old ham and spam and we're going to
tune with a bunch of new messages.  The question to answer is "What are
the best parameter values given this history and these new messages?"

David