home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cis.ohio-state.edu!zaphod.mps.ohio-state.edu!caen!hellgate.utah.edu!dog.ee.lbl.gov!nosc!ryptyde!jim
- From: jim@netlink.cts.com (Jim Bowery)
- Newsgroups: comp.ai.neural-nets
- Subject: Re: Quality of training data
- Message-ID: <4eHBoB1w164w@netlink.cts.com>
- Date: 22 Jul 92 14:14:38 GMT
- References: <1992Jul20.104344.28201@iitb.fhg.de>
- Organization: NetLink Online Communications, San Diego CA
- Lines: 16
-
- The selection of training data and the order in which it is presented
- (assuming one is not using a batch mode) has been studied in education
- theory by Zigfried Engleman and applied in classroom learning programs
- quite successfully. The theory is called "conceptual learning" and it
- involves queueing the stimulus response pairs so as to time their
- reintroduction to match the time when the student is most likely to
- begin to "forget" that particular stimulous-response pair. I think
- Marty Siegel at the University of Illinois (at least he used to be
- there) might have some references, although a person who might be
- more conversant in the mathematics of the theory as applied to
- stochastic approximations like neural nets would be Lou Dibello.
-
- --
- INTERNET: jim@netlink.cts.com (Jim Bowery)
- UUCP: ...!ryptyde!netlink!jim
- NetLink Online Communications * Public Access in San Diego, CA (619) 453-1115
-