Copy Link
Add to Bookmark
Report
AIList Digest Volume 5 Issue 219
AIList Digest Tuesday, 22 Sep 1987 Volume 5 : Issue 219
Today's Topics:
Neural Nets - Shift Invariance & References,
Philosophy - Natural Kinds & Computer Science
----------------------------------------------------------------------
Date: 19 Sep 87 18:18:56 GMT
From: maiden@sdcsvax.ucsd.edu (VLSI Layout Project)
Subject: Re: Neural Net Literature, shifts in attention
Someone sent me mail about a citation for Fukushima's network that
handled "shifts in attention". I lost the address. If that person
receives this information through this channel, I'd appreciate a
e-mail letter.
"A Neural Network Model for Selective Attention in Visual Pattern
Recognition," K. Fukushima, _Biological Cybernetics_ 55: 5-15 (1986).
"A Hierarchical Neural Network Model for Associative Memory,"
K. Fukushima, _Biological Cybernetics_ 50: 105-113 (1984).
"Neocognitron: A Self-organizing Neural Network Model for a Mechanism
of Pattern Recognition Unaffected by Shift in Position,"
K. Fukushima, _Biological Cybernetics_ 36: 193-202 (1980).
The same person mentioned about vision-like systems, so here are some
interesting physiologically grounded network papers:
"A Self-Organizing Neural Network Sharing Features of the Mammalian
Visual System," H. Frohn, H. Geiger, and W. Singer, _Biological
Cybernetics_ 55: 333-343 (1987).
"Associative Recognition and Storage in a Model Network of
Physiological Neurons," J. Buhmann and K. Shulten, _Biological
Cybernetics_ 54: 319-335 (1986).
Concerning selection:
"Neural networks that learn temporal sequences by selection," S. Dehaene,
J. Changeux, and J. Nadal, _Proceeding of the National Academy of
Sciences, USA_ 84: 2727-2731 (1987).
I hope this of help. I apologize for the delay; my bibliography on
neural networks spans an entire file cabinet and is severely disorganized
after the last move.
Edward K. Y. Jung
------------------------------------------------------------------------
UUCP: {seismo|decwrl}!sdcsvax!maiden ARPA: maiden@sdcsvax.ucsd.edu
------------------------------
Date: 18 Sep 87 13:49:36 GMT
From: ihnp4!homxb!homxc!del@ucbvax.Berkeley.EDU (D.LEASURE)
Subject: Neural Net Literature
In article <598@artecon.artecon.UUCP>, donahue@artecon.artecon.UUCP
(Brian D. Donahue) writes:
> Does anyone know of a good introductory article/book to neural networks?
We're using Rumelhart and MCClelland's 2 (I've heard a rumor that a
third volume is out) volume set on the Parallel Distributed Processing
Project in a seminar at Rutgers. I've only 8 chapters of it, but it
covers a lot of ground in neuroscience, cognitive psychology
(though some would disagree that such models are really cog-psy),
and computing. I recommend it. It's only $25 for both volumes in
paperback.
--
David E. Leasure - AT&T Bell Laboratories - (201) 615-5307
------------------------------
Date: Fri, 18 Sep 87 12:00:12 GMT
From: Caroline Knight <cdfk%hplb.csnet@RELAY.CS.NET>
Subject: Is Computer Science Science? Or is it Art?
Randy Martens says:-
"There is, however, Computer Engineering. (and Software Engineering,
and Systems Engineering etc.). Science is the discovery of the new.
Engineering takes what the scientists have found, and finds ways
to do useful things with it."
If this is so, my first question is
Who are the relevant scientists and what have they discovered?
-*-
As an AI researcher I'm always discovering new things - although
possibly not interesting in the same way as Newton's laws of motion or
Einstein's theory of general relativity - they are still potentially
new knowledge. (Most people must be content to play with grains of
sand not pebbles!)
However I would defend an engineer's creativity and ability to
experiment - they too discover new things but with a different aim in
mind and a different form of reporting than the scientist.
However I believe that in software there is a better analogy with art
and illustration than engineering or science. I have noticed that this
is not welcomed by many people in computing but this might be because
they know so little of the thought processes and planning that go on
behind the development of, say, a still life or an advertising poster.
Like software art is frequently pliable and reworkable; like software
there are many different methods and philosophies (many not employed
explicitly by experts although there are procedures for producing
certain types of work), rules of thumb and conventions; there are
great practioners and many more humble industrious ones; there are
different schools of thought and also ferverent arguments about such
low level things as Acrylics or Oils, sable brushes or manmade fibre
(here ethical issues also creep in), the "rightness" of working from a
photograph, etc. In illustration and advertising the artist might be
given a very wide but constrained brief or a very tightly specified
mock-up to work from. A work of art or an ad are often the results of
a carefully executed plan (although the results are not always quite
was expected).
I have also watched both good artists and good software makers at work
and several similarities struck me: the light sketch with more work
put into some of the trickier areas, experimentation with different
compositions, throwing out or completely removing bits, putting
finisihing touches which change the whole although are little enough
in themselves.
What is useful that can come of this analogy? Here are some
suggetions:-
Training: An artist will frequently learn their own style through
meticulous study of previous greats (whose great software is there for
us to emmulate?).
At first working from nature is important although more freedom and
greater abstraction will come later. An artist must learn to see and
understand - this is something which many software workers could do
with applying.
Aids: An artist has sketch pads for roughs or capture of structure or
examples of detail. The organisation of these is often less than
perfect - in software we have a better chance of providing this
although currently our best attempts such as Lisp machines and
environments like POPLOG are still very much less than perfect too.
Aids for producing mockups - for instance cartoonists use sheets of
shading which can be cut to fit the required area - in software we
need some such things to allow us to prototype with hints at detail
without putting it all in.
Aids for throwing stuff away! How many novices or less than expert
programmers cling to the stuff they've written when it needs throwing
out and redesigning from scratch! This is like the advice given in
school not to use an eraser - of course eventually the artist knows
when it is worth using one but at first it is better to concentrate on
developing the ability to create smoothly and without fiddling.
Well I guess I've gone on long enough - I'd be pleased to reply to
anyone interested in this point of view - thanks for reading this
far!
Caroline Knight cdfk@lb.hp.co.uk
cdfk@hplb.csnet
Tel: (0272) 799910 x4040 Telex: (0270) 449206
Fax: (0272) 790076
HPLabs, Hewlett Packard Ltd, Filton Rd, Stoke Gifford, BRISTOL
BS16 1NY
Everything I write is from me personally and does not represent
Hewlett Packard in any way.
------------------------------
Date: Sat 19 Sep 87 16:03:18-EDT
From: Albert Boulanger <ABOULANGER@G.BBN.COM>
Subject: Generalization & Natural Kinds
To add some beef to much of this natural kinds discussion, I suggest
that those interested in the issue of natural kinds and generalization take
a look at a recent paper by Roger Shepard:
"Toward a Universal Law of Generalization for Psychological Science"
Science, 11 September 1987, 1317-1323
From the abstract:
A psychological space is established for any set of stimuli
by determining metric distances between the stimuli such
that the probability that a response learned to any stimulus
will generalize to any other is an invariant monotonic function
of the distance between them. To a good approximation, this
probability of generalization (i) decays exponentially with this
distance, and (ii) does so in accordance with one of two
metrics, depending on the relation between the dimensions along
which the stimuli vary. These empirical regularities are
mathematically derivable from universal principles of natural
kinds and probabilistic geometry that may, through evolutionary
internalization, tend to govern the behaviors of all sentient
organisms.
Albert Boulanger
BBN Labs
------------------------------
Date: 21 Sep 87 11:31:00 EST
From: cugini@icst-ecf.arpa
Reply-to: <cugini@icst-ecf.arpa>
Subject: Natural kinds
Gilbert Cockton writes:
> I'd like to continue the sociological perspective on this debate.
> Rule number 1 in sociology is forget about "naturalness" - only
> sociobiologists are really into "nature" now, and look at the foul
> images of man that they've tried to pass off as science (e.g. Dworkin).
This seems a somewhat abrupt dismissal of natural kinds, which has
lately attracted some support by people such as Saul Kripke, who is
neither a computer scientist, dumb, nor politically unreliable
(although he IS a philosopher, and is thereby suspect, no doubt).
The (philosophically) serious question is to what extent our shared
concepts ("dog", "star", "electron", "chair", "penguin", "integer",
"prime number") are merely arbitrary social conventions, and to what
extent they reflect objective reality (the old nominalist-realist
debate). A sharper re-phrasing of the question might be:
To what extent would *any* recognizably rational being share our
conceptual framework, given exposure to the same physical environment?
(Eg, would Martians have a concept of "star"?).
I believe there have been anthropological studies, for instance,
showing that Indian classifications of animals and plants line
up reasonably well with the conventional Western taxonomy.
If there are natural kinds, their relevance to some AI work seems
obvious.
John Cugini <Cugini@icst-ecf.arpa>
------------------------------
Date: 21 Sep 87 17:40:08 GMT
From: Michael Shafto <aurora!shafto@ames.arpa>
Reply-to: shafto@aurora.UUCP (Michael Shafto)
Subject: Re: Materials Science
I would like an explanation of why Materials Science is
particularly "scientific" compared to other "<foo> Science"
disciplines. In particular, Materials Science doesn't seem
any more "scientific" (or less) than Computer Science.
Mike Shafto
------------------------------
Date: 21 Sep 87 17:52:53 GMT
From: shafto@AMES-AURORA.ARPA (Michael Shafto)
Subject: Re: Is Computer Science Science?
Alfred North Whitehead called mathematics the "science of
abstract forms." If that's too Platonic, then call it
"the science of abstract descriptions." I think if you
adopt the position that Real Science is about Nature, and
that mathematics is not Real Science, then you'll eventually
end up (with no further help from me) saying either
(a) mathematicians don't make discoveries, or (b) they
make discoveries about the properties of formal systems
or systems of abstract descriptions, and that THESE are
not part of Nature. If you follow (a), then you confine
yourself to a limited group of discussants who share your
idiosyncratic notion of 'discovery'; if you follow (b), then
you put the content of mathematics somewhere outside Nature.
Exactly where, I don't know.
Someone (perhaps Lakatos or Feyerabend) said that scientists
know about as much about science as fish know about
hydrology. This is well illustrated whenever scientists
quit DOING science and start talking about it.
Mike Shafto
------------------------------
Date: 21 Sep 87 19:00:13 GMT
From: pioneer!eugene@ames.arpa (Eugene Miya N.)
Subject: Re: Is Computer Science Science?
A couple of more recommended readings which came to me after a short
conversion with Denning:
"Cargo Cult Science" by Richard Feynman last chapter (1974 Comm. Addr.
at Caltech) in his Autobiography which I reread before bed last evening.
"An Empirical Study of FORTRAN Programs" Software -- Practice and
Experience by Don Knuth Feb. 1971, see intro and conclusions.
Knuth's paper in American Math. Monthly on the differences between
Algorithmic and Mathematical Thinking, around 1985.
These along with Simon, etc. mentioned earlier.
Unfortunately, I would say CS exhibits some cargo cult characteristics.
This does not have to be, we can change it.
From the Rock of Ages Home for Retired Hackers:
--eugene miya
NASA Ames Research Center
eugene@ames-aurora.ARPA
"You trust the `reply' command with all those different mailers out there?"
"Send mail, avoid follow-ups. If enough, I'll summarize."
{hplabs,hao,ihnp4,decwrl,allegra,tektronix,menlo70}!ames!aurora!eugene
------------------------------
Date: 21 Sep 87 22:49:56 GMT
From: pioneer!eugene@AMES.ARPA (Eugene Miya N.)
Subject: Re: Is Computer Science Science?
Oh yeah, one more reference thought on the way to lunch:
W. Daniel Hillis The Connection Machine, MIT Press, 1986,
Last Chapter entitled something like "Why Computer Science is No Good"
Says CS lacks scale, symmetry, and locality of effect.
--eugene
------------------------------
End of AIList Digest
********************
