Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!pipex!sunic!umdac!cs.umu.se!jan
From: jan@cs.umu.se (Jan T}ngring)
Subject: HNeT = Hype net?
Message-ID: <D2nvJ2.z2@cs.umu.se>
Sender: news@cs.umu.se (News Administrator)
Organization: Dept. of Computing Science, Umea Univ., 901 87 Umea, Sweden
Date: Thu, 19 Jan 1995 16:37:47 GMT
Lines: 62


The quotes below are from:

	John G Sutherland, "The holographic neural method", in "Fuzzy,
	holographic, and parallel intelligence: the sixth-generation
	breaktrough", by Branco Soucek and IRIS group, John Wiley &
	sons, 1992.

What's this then? "Holographic neural technology", "new concept of
information", "new paradigm of intelligence", sounds like really
something! What does it do? Implement functions defined by pairs of
(x,f(x))? No, no, too simple,

	The basis of stimulus-response association is to associate
	one set of analog measurements (stimulus field) to another
	group of measurements, in a manner that the presence of the
	first field invokes the issuance of the second.
	-- ch 2.1, p7.

	analog input states to the holographic based control element
	are transformed in a deterministic manner through the cell in
	accordance with the stimulus-response state mappings
	previously learned.
	-- ch 2.1, p9

Is it any good? Pretty good indeed!

	there exists no limitation in terms to the numbers of analog
	stimulus-response patterns and to the accuracy to which the
	associated mappings may be formed within individual neuron
	cells given a finite dimensionality of the input vector. 
	-- ch 2.1, p7.

What, no errors? Well, kind of, but "error" is such a harsh word 
isn't it?

	The net response vector, therefore, exhibits a small amount
	of error, or fuzzines in recall,
	-- ch 2.2, p19

Its not really an "error", you know, well of course it's an 
error, but,

	The response recall "error" is in fact deterministic in its
	expression of the response contribution from the set of
	nonsimilar stimulus loci,
	-- ch 2.2, p21

Discontinuity boundary errors for example, are very natural. 
When they occur, you know you are on the right track.

	Response values close to the discontinuity boundary and
	exhibiting sufficient recall error may, therefore, induce a
	spastic behavior, whereby response values effectively oscillate
	between the min and max boundary limits of the external real
	number domain. In some sense, this behaviour may be considered
	analogous to the neurological induction of muscle spasms, as
	experienced following excessive physical stress or activity.
	-- ch 2.5, p70

/Jan Tangring

