RE: [MLUG] AI - any suggestions?

["Brent Deterding" <EMAIL:PROTECTED>]

Very true - more than a few week project. For our NN firewall our engine is a
modified version of Neuralware - seems to work OK.

-- Brent

-----Original Message-----
From: EMAIL:PROTECTED [mailto:EMAIL:PROTECTED]On
Behalf Of Kmicic
Sent: Wednesday, October 31, 2001 9:35 PM
To: EMAIL:PROTECTED
Subject: Re: [MLUG] AI - any suggestions?


<snip>

as for neural nets, in Mark's place i wouldn't try to get too ambitious.
backprop net is a dumb classifier, but it can take lots of trial and
error to find the right parameters for it to learn. for that reason i'd
stick to pure classification problems (this is CECS 375, right?). there
are benchmark datasets on the net, just grab one of those and see
whether your program can match the standards. there are lots of issues
involved even in such simple training (not to mention tracking bugs in
your backprop implementation) to keep you busy for the rest of the
semester.

/paul
--
To unsubscribe, go to http://mlug.missouri.edu/members/edit.php

Archives are available at http://mlug.missouri.edu/list-archives/

--
To unsubscribe, go to http://mlug.missouri.edu/members/edit.php

Archives are available at http://mlug.missouri.edu/list-archives/