Copy Link
Add to Bookmark
Report
Neuron Digest Volume 06 Number 57
Neuron Digest Monday, 1 Oct 1990 Volume 6 : Issue 57
Today's Topics:
Re: Marr's VISION out of date
length restrictions on announcements
A general NN. Constraint Satisfaction Problem.
A learning algorithm
A BOOK ON CONTROL THEORY AND NEURAL NETWORKS
Minds & Machines (articles available by anonymous ftp)
Technical Report
Call for Papers.
Course change news
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Subject: Re: Marr's VISION out of date
From: michael k finegan <Mike.Finegan@UC.EDU>
Date: Wed, 26 Sep 90 10:45:15 -0400
Steve - I am working in image understanding, not ANN; but applications of
BCS to image analysis/understanding would be most interesting. I have
gone through a few of the BCS papers, but didn't want to implement the
PDE or DE solvers, etc. Can you give me an idea of how computationally
expensive (your) BCS is to use ? Also, since I am working on a system
with several knowledge sources (blackboard, a la VISIONS), can your BCS
be a 'C' module ? Thanks for any info., examples, or references.
- Michael Finegan
mfinegan@uceng.UC.EDU
------------------------------
Subject: length restrictions on announcements
From: ross@zeno.mmwb.ucsf.edu
Date: Thu, 27 Sep 90 13:35:51 -0700
A modest proposal: all conference and journal announcements and requests
for papers be restricted to a 1 page summary including if relevant time,
location, price and correspondent email address(es), with the whole
document available by anonymous ftp. I'm tired of seeing lists of names,
bureacratic statements of purpose (all similar) and order forms. I could
live with lists of titles and am glad to see abstracts on individual
papers especially when the papers are available on the network.
More ambitious: how about a group of bibliographies anonymously ftpable
by subject to be updated from the stream of stuff coming in, so that only
the new ones would appear in mail postings unless accompanied by
comments. Comments could be included in the bibliography too.
Bill Ross
[[ Editor's Note: I sympathize with Bill's lament. It should be noted
that nearly 1/5 of Digest subscribers are on BITNET and probably cannot
use ftp. As for the general suggestion of cutting down lengths of
announcements, I can only hope submitters will consider the request. As
moderator (using my copious spare time), I'm not ready to either organize
the announcement archives nor edit incoming submissions. As regular
readers know, I try to put announcements at the end of Digests and place
discussions and comments at the beginning. At one point, I tried
segregating Digests, but timely information got lost in the shuffle and
it was a royal pain for me. I don't have a good solution for the
trade-off of providing appropriate infomation anf over-doing it. Given
that electrons are relatively cheap, and (as a reader) I don't have to
expend any extra effort to get the recent information which the Digest
provides, the current situation will probabably remain. Of course,
opposing viewpoints cheerfully considered (and published, if desired). -PM ]]
------------------------------
Subject: A general NN. Constraint Satisfaction Problem.
From: qian@icopen.ICO.OLIVETTI.COM (DA QUN QIAN)
Date: Fri, 28 Sep 90 14:07:37 +0100
1. At present I am studying a new type of neural networks. In the neural
network, every arc can simultaneously several values (assume M values)
from input node to output node that the arc connects. The state of the
output node is determined by all its input values N*M if assume there are
N arcs pointing to it, and the output node also owns M values determined
by the N*M values. Who can offer me some references on this type of
neural networks?
2. I am also studying how to apply neural networks to solve constraint
satisfaction problem. I need some examples of description of engineering
problems (such as chemical engineering problem) as constraint
satisfaction problem. Who can offer me resome referfences in which this
kind of examples are reported.
Thanks in advance
Qian Da Qun
Olivetti Artificial Intelligence Center
Olivetti Nuova 3 ICO Piano
Via Jervis 77, 10015 Ivrea(TO)
Italy
Email: qian@icopen.ICO.OLIVETTI.COM
qian%uucp.icopen%it.olivetti.ico.iconet
------------------------------
Subject: A learning algorithm
From: qian@icopen.ICO.OLIVETTI.COM (DA QUN QIAN)
Date: Sun, 30 Sep 90 12:33:31 +0100
I correct my question just sent. The question I am proposing is as follows:
The neural net I am studying has the form:
s(j,t')=f(W(i,j,t), S(i,t), T(i))
s(j,t') is the state of node x(j) at time t',S(i,t) is the state vector
of nodes x(i)s at time t, W(i,j,t) is the weight vector on arcs starting
from x(i)s to x(j) at time t, T(i) is information propagation time vector
along arcs starting from x(i)s to x(j).
Usually learning algorithms are used to modify the values of W(i,j,t). Now
I intend to study the learning algorithm by which not only the values of
W(i,j,t) will be modified, but also the time T(i), i.e., information
propagation time, will be modified.
Who can offer me some references on this type of learning algorithms.
Thanks in advance.
Qian Da Qun
Artificial Inteligence Center
Olivetti Nuova ICO 3 Piano
Via Jervis 77, 10015 Ivrea(TO)
Italy
Email: qian@icopen.ico.olivetti.com
------------------------------
Subject: A BOOK ON CONTROL THEORY AND NEURAL NETWORKS
From: qian@icopen.ICO.OLIVETTI.COM (DA QUN QIAN)
Date: Mon, 01 Oct 90 13:24:35 +0100
Last month someone sent an email to this newsgroup in which he/she
recommended a book on control theory and neural networks to us. I lost
the email. Who can send me this email ?
Thanks in advance.
Qian Da Qun
Email: qian@icopen.ICO.OLIVETTI.COM
------------------------------
Subject: Minds & Machines (articles available by anonymous ftp)
From: harnad@phoenix.Princeton.EDU (Stevan Harnad)
Organization: Princeton University, Princeton, New Jersey
Date: 20 Sep 90 15:22:02 +0000
The following article is retrievable by anonymous ftp as (compressed)
file otherminds.Z from directory /pub/harnad on princeton.edu (retrieve
it in "binary" mode)
Other bodies, other minds:
A machine incarnation of an old philosophical problem
[To appear in: Minds and Machines 1: 1991]
Stevan Harnad
Department of Psychology
Princeton University
Princeton NJ 08544
ABSTRACT: Explaining the mind by building machines with minds runs into
the other-minds problem: How can we tell whether any body other than our
own has a mind when the only way to know is by BEING the other body? In
practice we all use some form of Turing Test: If it can DO everything a
body with a mind can do such that we can't tell them apart, we have no
basis for doubting it has a mind. But what is "everything" a body with a
mind can do? Turing's original "pen-pal" version (the TT) only tested
linguistic capacity, but Searle has shown that a mindless
symbol-manipulator could pass the TT undetected. The Total Turing Test
(TTT) calls for all of our linguistic AND robotic capacities; immune to
Searle's argument, it suggests how to ground a symbol manipulating system
in the capacity to pick out the objects its symbols refer to. No Turing
Test, however, can guarantee that a body has a mind. Worse, nothing in
the explanation of its successful performance requires a model to have a
mind at all. Minds are hence very different from the unobservables of
physics (e.g., quarks, superstrings); and Turing Testing, though
essential for machine-modeling the mind, can really only yield an
explanation of the body.
KEYWORDS: artificial intelligence; causality; cognition; computation;
explanation; mind/body problem; other-minds problem; robotics; Searle;
symbol grounding; Turing Test.
Other papers available from the same directory:
symbol.Z (The Symbol Grounding Problem, Physica D 1990)
searle.Z (Minds, Machines and Searle, J. Th. Exp. AI 1989)
categorization.Z (Category Induction and Representation, UP 1987)
Stevan Harnad Department of Psychology Princeton University
harnad@clarity.princeton.edu / harnad@pucc.bitnet / srh@flash.bellcore.com
harnad@learning.siemens.com / harnad@elbereth.rutgers.edu / (609)-921-7771
------------------------------
Subject: Technical Report
From: ANDERSON%BROWNCOG.BITNET@MITVMA.MIT.EDU
Date: Fri, 21 Sep 90 15:00:00 -0400
A technical report is available:
"Why, having so many neurons, do we have so few thoughts?"
Technical Report 90-1
Brown University Department of Cognitive and Linguistic Sciences
James A. Anderson
Department of Cognitive and Linguistic Sciences
Box 1978
Brown University
Providence, RI 02912
This is a chapter to appear in: Relating Theory and Data Edited by W.E.
Hockley and S. Lewandowsky, Hillsdale, NJ: Erlbaum (LEA)
Abstract
Experimental cognitive psychology often involves recording two quite
distinct kinds of data. The first is whether the computation itself is
done correctly or incorrectly and the second records how long it took to
get an answer. Neural network computations are often loosely described
as being `brain-like.' This suggests that it might be possible to model
experimental reaction time data simply by seeing how long it takes for
the network to generate the answer and error data by looking at the
computed results in the same system. Simple feedforward nets usually do
not give direct computation time data. However, network models realizing
dynamical systems can give `reaction times' directly by noting the time
required for the network computation to be completed. In some cases
genuine random processes are necessary to generate differing reaction
times, but in other cases deterministic, noise free systems can also give
distributions of reaction times.
This report can be obtained by sending an email message to:
LI700008@brownvm.BITNET
or
anderson@browncog.BITNET
and asking for Cognitive Science Technical Report 90-1 on reaction times,
or by sending a note by regular mail to the address above.
------------------------------
Subject: Call for Papers.
From: /PN=JAMES.L.RASH/O=GSFCMAIL/PRMD=GSFC/ADMD=TELEMAIL/C=US/@sprint.com
Date: 01 Oct 90 21:24:00 +0000
Call for Papers
1991 Goddard Conference on
Space Applications of Artificial Intelligence
May 14 & 15, 1991
NASA Goddard Space Flight Center
Greenbelt, Maryland
The Sixth Annual Goddard Conference on Space Applications of
Artificial Intelligence will focus on AI research and applications
relevant to space systems, space operations, and space science.
Topics will include, but are not limited to:
o knowledge-based spacecraft command & control
o expert system management & methodologies
o distributed knowledge-based systems
o intelligent database management
o fault-tolerant rule-based systems
o simulation-based reasoning
o fault isolation & diagnosis
o knowledge acquisition
o robotics & telerobotics
o planning & scheduling
o neural networks
o image analysis
Original, unpublished papers are now being solicited for the conference.
Abstracts should be no longer than one page. Five copies of the abstract
should be submitted by November 1, 1990 along with the authorUs name,
affiliation, address and telephone number. Notification of tentative
acceptance will be given by November 15, 1990. Papers should be no
longer than 15 pages and must be submitted in camera-ready form for final
acceptance by February 1, 1991.
Accepted papers will be presented formally or as poster presentations,
which may include demonstrations. All accepted papers will be published
in the conference proceedings as an official NASA document, and select
papers will appear in a special issue of the international journal
Telematics and Informatics. There will be a conference award for Best
Paper.
No commercial presentations will be accepted.
Send abstracts to:
Jonathan Hartley
NASA/GSFC
Code 522
Greenbelt, MD 20771
For further info call:
(301) 286-3150
------------------------------
Subject: Course change news
From: elsberry@arrisun3.arl.utexas.edu (Wes Elsberry)
Date: Mon, 01 Oct 90 20:22:16 -0500
Applied Neural Networks Computing course changes
The upcoming UCLA short course on Applied Neural Networks Computing
(December 3-6) has had two topics added to the course curriculum:
Biomedical Engineering Applications and Macroeconomic Applications. The
Biomedical Engineering Applications portion will include a look at an
advanced Adaptive Resonance Theory architecture network model used for
identification of the HIV virus. The Macroeconomic Applications portion
will look at non-geometric and non-parametric models of macroeconomic
systems.
The course is offered by the University of California at Los Angeles
Extension, with Dr. Harold Szu of the Office of Naval Research teaching
the course. The catalog list of topics includes the following: Nonconvex
Optimization; Problem Solving by Fixed Point Learning Systems; Solving
Image Processing and Automated Pattern Recognition Problems; Designing to
Solve Particular Problems; Strategy of Neural Nets for Human Visual
System; Dynamic Reconfigurable Nets; and Application of Neural Networks
According to Underlying Principles.
As it says in the catalog, "For technical information about the course,
call Harold Szu at (202) 767-1493. For registration information, call
the Short Course program Office at (213) 825-3344."
------------------------------
End of Neuron Digest [Volume 6 Issue 57]
****************************************
