home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!pipex!warwick!uknet!edcastle!hwcs!scottd
- From: scottd@cs.hw.ac.uk (Scott Dunn)
- Newsgroups: comp.ai.neural-nets
- Subject: RCS - units with multiple sites?
- Message-ID: <1992Nov22.134423.21010@cs.hw.ac.uk>
- Date: 22 Nov 92 13:44:23 GMT
- Sender: news@cs.hw.ac.uk (News Administrator)
- Organization: Dept of Computing & Electrical Engineering, Heriot-Watt University, Scotland
- Lines: 14
-
- I have been looking at the Rochester Connectionist Simulator as an
- example of how to design a flexible and open NN simulator.
-
- The one big problem I have is with with sites. Precisely what purpose
- do they serve. I know what the manual says - "allows differential treatment
- of inputs" but there are no examples in the manual with multiple sites.
-
- IMHO, allowing for lack of experience, they complicate the error determination
- without adding much in the way of flexibilty - I am obviously missing something,
- probably fairly obvious, so, could someone please explain further.
-
- Cheers,
-
- Scott.
-