Path: utzoo!attcan!uunet!mcsun!ukc!edcastle!aipna!jre
From: jre@aipna.ed.ac.uk (John Ellis      )
Newsgroups: comp.ai.neural-nets
Subject: A Network for Not Representing Knowledge
Message-ID: <2606@aipna.ed.ac.uk>
Date: 13 Jul 90 13:16:29 GMT
Reply-To: jre@aipna.ed.ac.uk (John Ellis      (MSc 89/90))
Organization: Dept of AI, Edinburgh University, UK.
Lines: 16

Dear all,

I'm looking for information about Kanerva's Sparse Distributed memory. I
have built and SDM simulator on a transputer array, and now that I'm
getting some results out of it, I'd like to be able to compare them with
other peoples'. I'm thinking of just some basic data, like how the
recall fidelity degrades as more (random) patterns are stuffed in, what
the optimum circle radius is for varius pattern widths, and what happens
when you use non-uniform distributions for the patterns and/or address
locations. 
  Also, if anyone has done a transputer based implementation, I'd be
interested to here of any techniques for not going completely mad.

Thanks in advance,

John Ellis