Re: I need a critique: using crm114 to train on (very) limited word set - or are there better ways?

Bill Yerazunis escreveu:

 > (speaking of which, Fidelis:  What's the _currrent_ BEST
> coefficients and formula for the best-behaving EDDC so far?)

I'm using that on page 8 of the EDDC slides with two changes:

- Swap "1" and "K3" in the denominator of the second term (a mistake in
the slides);
- Multiply the whole formula by a new factor starting at 0.8 and growing
up to 1 as the number of learnings increase. This reduces initial
instability:

  cfx = 0.8 + (spam_learnings + ham_learnings)/20;
  if (cfx > 1) cfx = 1.0;

The constants are: K1 = 0.25, K2 = 12, K3 = 8;

> 
> Now, CRM114 does not directly output these values; OSBF calculates
> EDDC internally, uses them, and then discards them.  
> 

Yes, you can only see them in trace mode (-T).

-- 
Fidelis


-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/