NTLK More on handwriting recognition

From: Mike O'Brien (obrien@leonardo.net)
Date: Mon Sep 04 2000 - 19:06:44 CDT

John Mark Davis sez:

> Could you clarify this a little more?
>> Rosetta was designed to be trainable, as the
>> cursive recognizer is, but they ran out of implementation time and trained
>> it up once with fixed tables.

        As folks are aware, there's an option under the cursive recognizer
to have it "learn" your handwriting. This amounts to changing weights in
a table of possible letter-forms according to which recognitions are
reinforced by being left alone, and which are deprecated by being corrected
by the user after a mis-recognition.

        It was the intention of the implementers of the printed recognition
system, Rosetta, to include this feature in Rosetta as well, but they ran out
of time, and used fixed letter-form weights instead. I don't want to try to
explain it in any more detail than this because I'll likely get it wrong.
The full story is contained in the paper I cited:

Combining Neural Networks and Context-Driven Search for On-Line,
     Printed Handwriting Recognition in the Newton, Yaeger, L., Webb,
     B., Lyon, R., AI Magazine, Volume 19, No. 1, p. 73-89, AAAI (1998)

Mike O'Brien

NewtonTalk brought to you by:

EVOTE.COM -- the ESPN of politics on the Internet! All the players, all the news, and the hottest analysis and features (plus 'toons!) anywhere.... visit http://www.evote.com today!

Need Subscribe/Unsubscribe info?

Visit the NewtonTalk section at http://www.planetnewton.com

This archive was generated by hypermail 2b29 : Tue Sep 12 2000 - 00:00:07 CDT