neuron-request@HPLABS.HP.COM ("Neuron-Digest Moderator Peter Marvit") (09/12/89)
Neuron Digest Monday, 11 Sep 1989
Volume 5 : Issue 38
Today's Topics:
Alternatives to backprop
eigensystems
reply to eigenvalue question
Higher order invariant pattern recognition
Looking for the Rochester Neural Net simulator...
Re: Looking for the Rochester Neural Net simulator...
Re: Looking for the Rochester Neural Net simulator...
Re: Looking for the Rochester Neural Net simulator...
Mactivation 3.2 - ftp
Re: Mactivation - ftp
Mac Version of PDP Software
Neuron Digest V5 #37
one-shot learning
When does a hopfield net converge ?
Re: When does a (hopfield) net converge?
EMCSR 1990, Second Call for Papers
ICNC Conference Announcement March 1990 in Germany
NIPS update
Send submissions, questions, address maintenance and requests for old issues to
"neuron-request@hplabs.hp.com" or "{any backbone,uunet}!hplabs!neuron-request"
Use "ftp" to get old issues from hplpm.hpl.hp.com (15.255.176.205).
------------------------------------------------------------
Subject: Alternatives to backprop
From: vaidyana@cb.ecn.purdue.edu (Ramaswamy Vaidyanathan)
Organization: Purdue University Engineering Computer Network
Date: 07 Sep 89 06:59:13 +0000
Hi!
I am interested in comparing the performance of other supervised
learning algorithms, vis-a-vis backpropagation. I am using backprop
for what is essentially a pattern recognition problem. The network
should be able to generalize well. The inputs are continuous, but a
a one hidden layer net seems sufficient.
Can some knowledgable netter suggest suitable alternatives, with
appropriate references. Has anyone done similar work before?
Please email your responses to
vaidyana@lips.ecn.purdue.edu
or
vaidyana@cb.ecn.purdue.edu
I will summarise the responses and post for the benefit of all.
Thanks in advance.
Vaidya.
------------------------------
Subject: eigensystems
From: chambers@watdcsu.waterloo.edu (Mike Chambers - Independent Studies)
Organization: U of Waterloo, Ontario
Date: 06 Sep 89 15:54:43 +0000
Could anyone supply references to the use of neural nets for solving
eigenvalue/vector problems ?
thanks
------------------------------
Subject: reply to eigenvalue question
From: gary@desi.ucsd.edu (Gary Cottrell)
Organization: Computer Science and Engineering, UC San Diego
Date: 10 Sep 89 22:56:14 +0000
My mail is not getting through, so I'll post this:
To: chambers
Subject: eigenvalues
I have found that a linear encoder network (e.g., 64x16x64, use a low
learning rate of about .01 with back prop) will converge on a solution that
spans the principal subspace, even if it doesn't line up on the principal
components. See also Sanger's article in NIPS 88 (the book is 89), and
Baldi's article in the same volume, as well as Linsker's for an alternative
(ed Touretzy, Morgan Kaufmann).
gary cottrell 619-534-6640 Secretary: 619-534-5288
Computer Science and Engineering C-014
UCSD, La Jolla, Ca. 92093
gary@cs.ucsd.edu (ARPA)
{ucbvax,decvax,akgua,dcdwest}!sdcsvax!gary (USENET)
gcottrell@ucsd.edu (BITNET)
------------------------------
Subject: Higher order invariant pattern recognition
From: plonski@primrose.Aero.Org (Mike Plonski)
Organization: The Aerospace Corporation
Date: 07 Sep 89 21:56:18 +0000
Has anyone implemented any of Gile's work on invariant pattern
recognition using higher order nets. I was able to get similiar
results to Gile's
(3rd ref.) for the small 2D patterns (3 by 5 input field) that he
used, but I found that it was also possible to converge to spurious states.
I also found that when I increase the input field to 16x16 pixels, I
would always converge to a spurious state. I don't think that there is
a coding problem,since I get similiar results for the smaller input
field. Has anyone
else experimented with this and would like to share their comments.
Thanks.
%A C. L. Giles
%A T. Maxwell
%T Learning, invariance, and generalization in high order neural networks
%J Applied Optics
%V 26
%N 23
%D 1 |DEC| 1987
%P 4972-4978
%A C. L. Giles
%A R. D. Griffin
%A T. Maxwell
%T Encoding Geometric Invariances in Higher Order Neural Networks
%J Neural Information Processing Systems
%I AIP Conf. Proc.
%E D. Z. Anderson
%D 1988
%P 301-309
%A T. Maxwell
%A C. L. Giles
%A Y. C. Lee
%A H. H. Chen
%T Transformation Invariance Using Higher Order Correlations in
Neural Net Architectures
%J Proc. IEEE Intl. Conf. on Systems, Man, and Cybernetics
%C Atlanta, GA
%D 14-17 |OCT| 1986
%P 627-632
%K HOLU, group theory, tolerance, temporal invariance
. . .__. The opinions expressed herin are soley
|\./| !__! Michael Plonski those of the author and do not represent
| | | "plonski@aero.org" those of The Aerospace Corporation.
_______________________________________________________________________________
We have been having some problems with our net connection lately so, if mail
bounces back, then try "@brl.mil:plonski@aero.org"
------------------------------
Subject: Looking for the Rochester Neural Net simulator...
From: jtn@potomac.ads.com (John T. Nelson)
Organization: Advanced Decision Systems, Arlington VA
Date: 07 Sep 89 13:55:46 +0000
Does anyone know where I can find source code to the Rochester Neural
Network simulator? We're interested in experiementing with neural networks
and GA simulators.
Your opinions and any pointers are appreciated.
------------------------------
Subject: Re: Looking for the Rochester Neural Net simulator...
From: plonski@primrose.Aero.Org (Mike Plonski)
Date: 08 Sep 89 20:42:27 +0000
I am including the message I was sent when I subscribed to the
simulator-users mailing group. The only significant change
that I know of is that there is now an X-windows version of the
graphical interface available from the same source. I don't
remember the specific files to get, but it was fairly obvious
once you ftp over. I have gotten one of the X-windows versions
to work, but not the other. The forwarded message contains all the
necessary information to obtain RCS.
. . .__. The opinions expressed herin are soley
|\./| !__! Michael Plonski those of the author and do not represent
| | | "plonski@aero.org" those of The Aerospace Corporation.
_______________________________________________________________________________
Date: Fri, 7 Apr 89 12:32:09 EDT
From: bukys@cs.rochester.edu
Subject: welcome to the simulator-users mailing list
Welcome to the <simulator-users@cs.rochester.edu> mailing list. On
this list you will find the other users of the Rochester Connectionist
Simulator.
Please forgive my tardiness in adding you to the mailing list.
If it's any consolation, there has been little traffic on the list.
If you have any ADMINISTRATIVE requests, such as mailing list additions
or deletions, please send them to
<simulator-request@cs.rochester.edu>
Please send BUG REPORTS to
<simulator-bugs@cs.rochester.edu>
We are interested in fixing bugs, but can't make any promises!
Please make your bug reports as specific as possible.
===============================================================================
The Rochester Connectionist Simulator, version 4.1, is available by
anonymous FTP from CS.Rochester.Edu, in the directory public/rcs.
Version 4.1 was released on April 5, 1988, and can be found in the file
"rcs_v4.1.tar.Z". It is a compressed tar file, size 837215 bytes.
When uncompressed, it should be size 3092480 bytes, and running the
Unix "sum" program on it should yield "62514 3020".
This is too big to mail electronically, so please don't ask.
REMEMBER TO USE THE "TYPE TENEX" or "TYPE BINARY" COMMAND IN FTP WHEN
RETRIEVING COMPRESSED TAR FILES!!!
Official bug reports also reside in this directory:
rcs_v4.1.patch.01 fixes a SERIOUS problem in the logging feature.
rcs_v4.1.patch.02 fixes some minor syntactic problems in the doc.
rcs_v4.1.patch.03 a missing "return" makes commands noops on SUN4
The patches can be applied by hand if necessary, but you will make your
life easier if you obtain the widely-available "patch" program, and redirect
each patch file into "patch -p".
Other files of interest:
rcs_v4.1.note.01 the doc's missing "urcsmanual" TeX files.
rcs_v4.1.doc.tar.Z the manual, in postscript format, ready to print
===============================================================================
If you are unable to obtain anonymous FTP access to the simulator
distribution, you can still order a copy the old way. Contact
Peg Meeker
Computer Science Department
University of Rochester
Rochester, NY 14627
(USA)
or <connect@cs.rochester.edu>, and she will send you the approriate forms.
We are currently charging $150 for a distribution tape and a manual.
If you have TeX and a PostScript printer, you should be able to produce
your own copy of the 181-page manual. If you want a paper copy of the
manual anyway, send a check for $10 per manual (payable to the
University of Rochester) to Peg Meeker Peet at the above address.
We do not have the facilities for generating invoices, so payment is
required with any order.
===============================================================================
------------------------------
Subject: Re: Looking for the Rochester Neural Net simulator...
From: goddard@aic.dpl.scg.hac.com (Nigel Goddard (213)317-5674)
Organization: Hughes Aircraft Co., El Segundo, CA
Date: 08 Sep 89 23:28:28 +0000
In article <8039@potomac.ads.com> jtn@potomac.ads.com (John T. Nelson) writes:
>
>Does anyone know where I can find source code to the Rochester Neural
>Network simulator? ...
Anonymous ftp from cs.rochester.edu. cd to /public/rcs and retrieve
file README. Or send mail to connect@cs.rochester.edu
Nigel Goddard
------------------------------
Subject: Re: Looking for the Rochester Neural Net simulator...
From: goddard@aic.dpl.scg.hac.com (Nigel Goddard (213)317-5674)
Organization: Hughes Aircraft Co., El Segundo, CA
Date: 10 Sep 89 00:02:29 +0000
In article plonski@primrose.Aero.Org (Mike Plonski) writes:
>
>I am including the message I was sent when I subscribed to the
>simulator-users mailing group. The only significant change
>that I know of is that there is now an X-windows version of the
>graphical interface available from the same source. I don't
>remember the specific files to get, but it was fairly obvious
>once you ftp over. I have gotten one of the X-windows versions
>to work, but not the other. The forwarded message contains all the
>necessary information to obtain RCS.
>
Oh well, the cat is out of the bag. The X interface that you will
find in the ftp directories is an inferior development version. If
you want to play with it, the files are xgi_X11r3.tar.z and
xgi_popups.tar.Z. Don't bother with xgi.tar.Z, it won't work. The
beta release is about to go out to selected sites. I advise waiting
for the full release of 4.2. At that time dynamic code-loading under
SunOS 4.0 will also be fixed (thanks to the GNU project).
Nigel Goddard
------------------------------
Subject: Mactivation 3.2 - ftp
From: mikek@fred.colorado.edu (Mike Kranzdorf)
Organization: University of Colorado, Boulder
Date: Sun, 23 Jul 89 22:17:31 +0000
Mactivation is an introductory-type connectionist/neural network
simulator for all Macintoshes. You can't build much with it, but
it is very interactive and lets you diddle with lots of parameters.
It is free, and includes a 30 page user's manual/connectionism intro.
Instructors may use it in courses if desired. Version 3.2 is now
available via anonymous ftp on boulder.colorado.edu (in pub). It
is a binhex'ed stuffit file. Enjoy,
mike kranzdorf mikek@boulder.colorado.edu
PS. You can also get a copy on disk by sending me:
1) $5 (US) cash or check, OR
2) one 3 1/2 inch disk AND a self addressed *stamped* envelope.
Mike Kranzdorf
5943 Sugarloaf Rd.
Boulder, CO 80302
------------------------------
Subject: Re: Mactivation - ftp
From: mikek@fred.colorado.edu (Mike Kranzdorf)
Organization: University of Colorado, Boulder
Date: Sun, 30 Jul 89 19:00:11 +0000
Some clarifications:
The file is on boulder.colorado.edu in /pub.
There is an anonymous ftp account which accepts any password.
The file is called mactivation.3.2.sit.hqz.Z
It is stuffited (w/ version 1.5.1), binhexed (v 4.0), and compressed
(on unix ala Lempel-Ziv).
I don't know how to get it to your mac, please don't ask me. Ask
someone who knows how at your site.
Yes, it will get to comp.binaries.mac.
--mikek
------------------------------
Subject: Mac Version of PDP Software
From: "James L. McClelland" <jlm+@andrew.cmu.edu>
Date: Fri, 08 Sep 89 10:02:52 -0400
A version of the PDP software for the Macintosh, and the accompanying
handbook, are now available from MIT Press.
To avoid any misunderstanding, I wanted to make clear that this is a
STRAIGHT PORT of the IBM PC version of the PDP software. This means that
the software does not make use of mac point-and-click or graphics
capabilities. The software is compatible with the finder and multifinder,
but otherwise functionally identical to the IBM PC version. The handbook is
unchanged, with the exception of a small section on using the software on
Macs that can be found at the back of the Book.
Users who want to use the software on PC's or Unix systems should still get
the IBM PC version. The Mac Version is works on a wide range of Macs, but
not on any other computers.
Reference:
McClelland, J. L. and Rumelhart, D. E. Explorations in
Parallel Distributed Processing: A Handbook of models, programs
and exercises. [IBM PC Edition/Macintosh Edition.]
Cambridge, MA: MIT Press, [1988/1989].
------------------------------
Subject: Neuron Digest V5 #37
From: phil@goldhill.com
Date: Wed, 06 Sep 89 09:33:56 -0400
Hi. I was wondering if i could get a relatively non-technical explanation of
neural networks and their applications, training requirements, and
evaluation criteria for my wife. She's a market researcher and her boss has
become very excited about neural networks and their application to document
management and image processing and wants her to "write it up".
Information about the costs associated with training and delivering an
application would be very helpful as well.
Thanks very much.
Phil Stanhope : phil@goldhill.com
Gold Hill Computers
Cambridge, MA 02139
------------------------------
Subject: one-shot learning
From: vub.vub.ac.be!arti9!chen@prlb.philips.be (Chung-Chih Chen)
Date: Thu, 07 Sep 89 15:20:50 +0100
Hallo, everybody.
I don't like back-propagation because you have to present the input many
times. I prefer one-shot learning where you present the input only once.
But nothing is perfect. While one-shot learning is fast, its property of
generalization is worse than back-propagation's. I am trying to find more
papers about one-shot learning and to improve its generalization. If we can
build a NN which is fast and generalize very well, that would be great.
Right?
If you know any paper about one-shot learning, please tell me.
Thanks.
Chung-Chih Chen
Artificial Intelligence Lab.
(Building K, 4th floor)
Free University of Brussels
Pleinlaan 2
1050 Brussels, BELGIUM
(email: chen@arti9.vub.ac.be)
------------------------------
Subject: When does a hopfield net converge ?
From: mike@bucasb.BU.EDU (Michael Cohen)
Date: Wed, 06 Sep 89 09:34:05 -0400
See Morris W. Hirsch's forthcomming paper in Neural Networks for a Survey.
Many more systems than Cohen Grossberg ... are now known to be stable. For
example a system with arbitrary positive weights which is non-competitive
and bounded is known to converge with probabibility one.
-mike
Michael Cohen ---- Center for Adaptive Systems
Boston University (617-353-7857)
Email: mike@bucasb.bu.edu
Smail: Michael Cohen
Center for Adaptive System
Department of Mathematics, Boston University
111 Cummington Street
Boston, Mass 02215
------------------------------
Subject: Re: When does a (hopfield) net converge?
From: hirsch@math.berkeley.edu (M. W. Hirsch)
Organization: Math Department, University of California at Berkeley
Date: Wed, 06 Sep 89 15:09:16 -0700
The question of when a neural net converges is a special case of the very
general question of when a dynamical system converges. For continuous time
nets some sufficient conditions are discussed in my article in the
forthcoming issue of Neural Networks (September 1989). (An earlier version
appeared in Volume 2 of the Proceedings of the June 1987 IEEE conference on
neural nets in San Diego.) Hopfield nets are mathematically a special case
of the nets considered by Michael Cohen and Stephen Grossberg, in "Absolute
Stability of Global Apttern Formation and Parallel Memory Storage by
Competitive Neural Networks", IEEE Transactions SMC-13, 1983, pp. 815-826;
discussed in Grossberg's article in vol. 1, no. 1 (1988) of Neural Networks.
--M. W. Hirsch
------------------------------
Subject: EMCSR 1990, Second Call for Papers
From: mcvax!ai-vie!georg@uunet.UU.NET (Georg Dorffner)
Date: Fri, 01 Sep 89 13:40:37 -0100
Second Call for Papers
EMCSR 90
TENTH EUROPEAN MEETING
ON
CYBERNETICS AND SYSTEMS RESEARCH
April 17-20, 1990
University of Vienna, Austria
Session M:
Parallel Distributed Processing
in Humans and Machines
Chairs:
D.Touretzky (Carnegie Mellon, Pittsburgh, PA)
G.Dorffner (Vienna, Austria)
Other Sessions at the meeting will be:
A: General Systems Methodology
B: Fuzzy Sets, Approximate Reasoning and Knowledge-based Systems
C: Designing and Systems
D: Humanity, Architecture and Conceptualization
E: Cybernetics in Biology and Medicine
F: Cybernetics in Socio-Economic Systems
G: Workshop: Managing Change: Institutional Transition in the
Private and Public Sector
H: Innovation Systems in Management and Public Policy
I: Systems Engineering and Artificial Intelligence for Peace
Research
J: Communication and Computers
K: Software Development for Systems Theory
L: Artificial Intelligence
N: Impacts of Artificial Intelligence
The conference is organized by the Austrian Society for
Cybernetic Studies (chair: Robert Trappl).
SUBMISSION OF PAPERS:
For symposium M, all contributions in the fields of PDP,
connectionism, and neural networks are welcome.
Acceptance of contributors will be determined on the basis of
Draft Final Papers. These papers must not exceed 7 single-spaced
A4 pages (maximum 50 lines, final size will be 8.5 x 6 inches),
in English. They have to contain the final text to be
submitted, however, graphs and pictures need not be of
reproducible quality.
The Draft Final Paper must carry the title, author(s) name(s),
and affiliation in this order. Please specify the symposium in
which you would like to present the paper (one of the letters
above). Each scientist shall submit only 1 paper.
Please send t h r e e copies of the Draft Final Paper to:
EMCSR 90 - Conference Secretariat
Austrian Society for Cybernetic Studies
Schottengasse 3
A-1010 Vienna, Austria
Deadline for submission: Oct 15, 1989
Authors will be notified about acceptance no later than Nov 20,
1989. They will then be provided with the detailed instructions
for the preperation of the Final Paper.
Proceedings containing all accepted papers will be printed.
For further information write to the above address, call +43 222
535 32 810, or send email to: sec@ai-vie.uucp
Questions concerning symposium M (Parallel Distributed
Processing) can be directed to Georg Dorffner (same address as
secretariat), email: georg@ai-vie.uucp
------------------------------
Subject: ICNC Conference Announcement March 1990 in Germany
From: <ECKMILLE%DD0RUD81.BITNET@CUNYVM.CUNY.EDU>,
ECKMILLE@DD0RUD81.BITNET (Rolf Eckmiller, Duesseldorf, FRG)
Date: Thu, 07 Sep 89 09:14:00 +0700
CALL FOR PAPERS 8/89
International Conference on:
PARALLEL PROCESSING IN NEURAL SYSTEMS AND COMPUTERS (ICNC)
- 10th Cybernetics Congress of the DGK -
19. - 21. March, 1990, Heinrich-Heine-Universitaet Duesseldorf (FRG)
Organizing Committee: R. Eckmiller (chair), Duesseldorf (FRG)
G. Hartmann, Paderborn (FRG)
G. Hauske, Muenchen (FRG)
C. v.d. Malsburg Los Angeles (USA)
W. v. Seelen Bochum (FRG)
Conference Language: English
Topics: 1) New Concepts in Neuroscience and Computational Neuroscience
2) Massively Parallel Computers (e.g. SUPRENUM, Transputer Systems)
3) Structure and Function of Biological Neural Systems
4) Self Organization versus Programming in Parallel Computers
5) Optical Computers and Molecular Computers
6) Parallel Processing in Artificial Intelligence
Activities: *) Invited Lectures by American and European Scientists on the
listed Topics
*) Oral Presentations (15 + 5 min.)
*) Posters Presentations
*) Exhibition of Books, Neural Systems, and Computers
Invited Speakers include:
J.R Barker/Glasgow (UK) G. Carpenter/Boston (USA)
J. Feldman/Berkeley (USA) A. Cremers/Dortmund (FRG)
H. Haarer/Bayreuth (FRG) K. Fukushima/Osaka (Japan)
T. Kohonen/Espoo (Finland) H. Haken/Stuttgart (FRG)
D. Psaltis/Pasadena (USA) W. Reichardt/Tuebingen (FRG)
U. Trottenberg/Bonn (FRG)
Deadline for Submission of Papers:>> 15 November, 1989 << !!!!!!!
(4 camera-ready pages per contribution)
Publication: Conference Proceedings will be available at the conference
as hard cover book (Elsevier Science Publ.)
Registration Fees: Before 1 October, 1989 = 150,- DM including Proceedings
After 1 October, 1989 = 200,- DM " "
Students = 100,- DM " "
Students = 50,- DM without Proceedings
ICNC-Conference Secretariat:
Dr. R. Eckmiller
Heinrich-Heine-Universitaet Duesseldorf
Div. Biocybernetics
Universitaetsstrasse 1 Tel.: (211) 311-5204
D-4000 Duesseldorf (FRG) e-mail: ECKMILLE@DD0RUD81.BITNET
Please complete and mail the request of 2nd Announcement to the ICNC-Conference
Secretariat.
cut here - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
To ICNC-Conference Secretariat
Dr. R. Eckmiller
Heinrich-Heine-Universitaet Duesseldorf
Div. Biocybernetics
Universitaetsstrasse 1
D-4000 Duesseldorf (FRG)
Request for 2nd Announcement of ICNC
19. - 21. March, 1990, Duesseldorf (FRG)
Last Name:
___________________________________________________________________
First Name:
__________________________________________________________________
Affiliation:
_________________________________________________________________
Street:
______________________________________________________________________
City:
________________________________________________________________________
Country:
_____________________________________________________________________
Tel.: e-mail:
______________________________ _________________________________
I intend to participate at the ICNC ( ) as student (yes / no)
I intend to submit a contribution on the topic No. ( )
as oral presentation ( )
as poster presentation ( )
as oral paper or poster ( )
Please, send the 2nd Announcement of ICNC to my above address.
Date: Sincerely,
_________________________ ___________________________________
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
------------------------------
Subject: NIPS update
From: jose@tractatus.bellcore.com (Stephen J Hanson)
Date: Thu, 07 Sep 89 10:06:29 -0400
**** NIPS89 Update ****
We've just finished putting the program for the conference together and have
a preliminary program for the workshops. A mailing to authors will go out
this week, with registration information. Those who requested this
information but are not authors will hear from us starting in another week.
If you received a postcard from us acknowledging receipt of your paper, you
are on our authors' mailing list. If you haven't requested the registration
packet, you can do so by writing to
Kathie Hibbard
NIPS89 Local Committee
University of Colorado Eng'g Center
Campus Box 425
Boulder, CO 80309-0425
------------------------------
End of Neurons Digest
*********************