repetitive training

[Boris 'pi' Piwinger]

David Relson wrote:

>> 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.
> 
> 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?"

But this is not what the training was made for. The messages
used in training are carefully chosen depending on those
parameters. By this you can just generate an appropriate
cutoff. Changing this must fail then.

As we could see from Greg's experiment is that we basically
had the same number of messages to train with in each round,
so there was no hope of closing off. A clear sign that there
was no improvement.

pi