home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.neural-nets
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!rpi!ghost.dsi.unimi.it!lombardo
- From: lombardo@ghost.dsi.unimi.it (franco lombardo)
- Subject: Neurons on [0,1]
- Organization: Computer Science Dep. - Milan University
- Date: Thu, 17 Dec 1992 11:36:22 GMT
- Message-ID: <1992Dec17.113622.12586@ghost.dsi.unimi.it>
- Lines: 20
-
- Hi,
- I'm trying to theach to a net with one hidden layer a
- function from [0,1] to [0,1], i. e. a function taking its values
- in the whole interval [0,1], using back-prop.
- Which kind of function colud I use instead of the classical
- sigmoidal function?
- Colud you tell me some references about this?
-
- Thanks in advance!
-
- Franco Lombardo.
-
- SORRY FOR MY BAD ENGLISH!
-
-
- *******************************************************************************
- Franco Lombardo | Dipartimento di Scienze dell'Informazione,
- lombardo@ghost.dsi.unimi.it | Universita' degli Studi di Milano, Italy.
- *******************************************************************************
-
-