NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/comp/ai/neuraln/4297

Text File | 1992-11-17 | 1.3 KB | 27 lines

Newsgroups: comp.ai.neural-nets
Path: sparky!uunet!charon.amdahl.com!pacbell.com!sgiblab!spool.mu.edu!agate!doc.ic.ac.uk!syma!stevew
From: [e-mail] (Steve Williams)
Subject: Help wanted for coarse coding scheme.
Message-ID: <[e-mail]>
Organization: University of Sussex
References: <none>
Date: Tue, 17 Nov 1992 14:51:55 GMT
Lines: 16

I am doing some research into the possibilities of a SRN (Simple Recurrent
Network) being useful as a psychological model of the object concept. Details
of this aide I have run into rather a tricky problem, namely, how to transform
a spatial array cointaining the 2d coordinates of a single object, into the
three coordinates specifying the location of the object in a coarse coding
scheme. I have considered generating a look up table in a 3d array which maps
onto the 2d spatial array but this would be cumbersome to code and not as
flexible as I would like - I would have to alter the contents of the look up
table everytime I adjusted the size of the network and thus its receptive
field. If anyone has any ideas then I would be interested to know them. If you
require any furthur information then please e-mail me at my Janet address, as
this explanation of the situation is a little compact!

Thanks in advance,
Steve Williams (e-mail: [e-mail] or
[e-mail])