Copy Link
Add to Bookmark
Report
Neuron Digest Volume 11 Number 09
Neuron Digest Saturday, 6 Feb 1993 Volume 11 : Issue 9
Today's Topics:
neural net simulators
NSF funding Opportunites
Job Announcement -- Program Officer, J.S. McDonnell Foundation
Summer institute in computatioal neuroscience - Woods Hole
Call for Volunteers
"neural-net based software for digitization of maps"
Response to "Degeneracy of weights in networks?"
speaker adaptation and normalization
Re: NNs and economics
vision position posting
Send submissions, questions, address maintenance, and requests for old
issues to "neuron-request@cattell.psych.upenn.edu". The ftp archives are
available from cattell.psych.upenn.edu (130.91.68.31). Back issues
requested by mail will eventually be sent, but may take a while.
----------------------------------------------------------------------
Subject: neural net simulators
From: anderson@CSHL.ORG (John Anderson)
Date: 24 Jan 93 20:50:21 +0000
I tried several times to reply directly to the poster, but bounced every
time, so I'm posting to the newsgroup, since others may be interested as
well.
tbrannon@csee.lehigh.edu wrote:
> Could you fill me in on GENESIS? Ie, obtaining, features?
GENESIS is a biological neural net simulator developed by Jim Bower's
group at Caltech. It uses X Windows and has a nice graphical user
interface called XODUS. You can model individual neurons or networks of
them. There are a number of neural net simulators available. Several of
them are discussed and compared in an article in the November 1992 issue
of Trends in Neurosciences:
E. De Schutter (1992) "A consumer guide to neuronal modeling
software" TINS 15:462-464.
There is a table in that article giving addresses for the people
responsible for the packages it discusses. The GENESIS contact is
C. Ploegaert
Division of Biology
216-76
California Institute of Technology
Pasadena, CA 91125
FAX 818-795-2088
genesis@cns.caltech.edu
By the way, that whole issue of TINS is about modeling the nervous
system, so you might want to check it out.
John
- -------
John E. Anderson
W. M. Keck Structural Biology Laboratory
Cold Spring Harbor Laboratory
------------------------------
Subject: NSF funding Opportunites
From: "Paul J. Werbos" <pwerbos@nsf.gov>
Date: Tue, 26 Jan 93 09:45:38 -0500
The Neuroengineering Program at NSF will be broadening the scope of what
it funds a little this year.
In particular, there is circa 100K new money earmarked for "HPCC"
activities. In the past, neuroengineering has not supported
hardware-related research. However, this year it will certainly be able
to support optical neural net work, and probably be able to cofund VLSI
work as well. IN the latter case, however, we will still probably
require coreview and cofunding from the MIPS Division in Computer Science
(for which one should consult Mike Foster, (202)357-7853). Also, there is
circa 100K earmarked for work related to manufacturing; how this
"relation" will be verified is unclear as yet, but use of test problems
of interest to manufacturing (or even letters indicating potential
interest from manufacturers) would certainly help increase the
probability of funding.
Beyond this, the basic funding levels and priorities are essentially
unchanged, as described in the INNS Government Funding book or in the
Handbook of Intelligent Control (D.White and D.Sofge, eds).
It would certainly be legal to copy this email over to other BBS's; in
fact, I hope someone does that kind of thing, to get the word out to the
entire community.
If optically-related proposals do come in, of course, I would hope that
they would also make some contribution to the existing program
priorities, and/or involve the kinds of work which are too fundamental
(e.g. Bell's Theorem effects)s ) to be funded by the existing programs
elsewhere.
As in the past, the deadline for proposals is formally the end of
February, but I will try (without promising anything) to schedule later
proposals for panel review as well.
Best wishes.
Paul J. Werbos
(202)-357-9618
------------------------------
Subject: Job Announcement -- Program Officer, J.S. McDonnell Foundation
From: "John T. Bruer" <P06819JB%WUVMC.bitnet@WUVMD.Wustl.Edu>
Date: Tue, 26 Jan 93 15:49:00 -0600
PROGRAM OFFICER
The James S. McDonnell Foundation, a major private foundation
having special interests in education and the behavioral and biological
sciences, is seeking an energetic, resourceful professional to fill the
position of Program Officer.
The successful candidate will coordinate the foundation's
existing research programs in education and cognitive neuroscience;
assist in identifying and formulating new program areas; evaluate grant
requests and monitor the progress of ongoing grants; interact regularly
with grantees and the foundation's program advisory boards; plan and
coordinate national meetings; assist in preparing and presenting material
to the Board of Directors; and represent the foundation locally and
nationally where appropriate.
Nominees and applicants should demonstrate superior oral and
written communication skills, and have a proven record of strong
administrative abilities. Candidates for the position must hold a
graduate degree in the biological, behavioral, or social sciences and
have had at least five years teaching, research, or administrative
experience. Prior grantmaking experience is not required but will be
considered favorably.
Salary commensurate with experience plus fringe benefits.
The deadline for receipt of application is April 16. Qualified
candidates should send a letter explaining their interest in the
position, resume, salary requirements, and three letters of reference to:
John T. Bruer, Ph.D.
President
James S. McDonnell Foundation
1034 S. Brentwood Blvd., Suite 1610
St. Louis, MO 63117
The James S. McDonnell Foundation is an Equal Opportunity/ Affirmative
Action Employer.
------------------------------
Subject: Summer institute in computatioal neuroscience - Woods Hole
From: cnsag@physics.att.com
Date: Tue, 26 Jan 93 17:28:17 -0500
Marine Biological Laboratory
Woods Hole, Massachusetts
METHODS IN COMPUTATIONAL NEUROSCIENCE
August 3-31, 1993
This is an intensive four week course that addresses issues relevant to
computational neuroscience: the study of how the biophysical and
biochemical properties of neurons and synapses, together with the
architecture of neural circuits, produce animal behavior. The aim of the
course is threefold. First, to teach students how to formulate questions
about computational aspects of a nervous system. Second, to provide the
analytical and numerical simulation tools necessary to answer those
questions. Third, to allow students to meet and interact with established
investigators in this field. The program of the course consists of two
daily lectures and laboratory projects.
The lectures will provide a broad view of computational neuroscience,
while at the same time discussing in detail the interplay between models
and experiments for specific systems. One series of lectures focuses on
the dynamics of individual neurons and synapses, for example, the passive
electrical properties of cells, the basis of excitability, second
messenger and chemical dynamics, and mechanisms and models for
plasticity. A second series of lectures addresses the use of exact models
of single cells, versus reduced neural models, in analysis of networks.
Topics include the dynamics of small circuits, such as those that control
rhythmic motor patterns, as well as the dynamics of large networks. A
third series of lectures addresses development of the nervous system. A
final series considers the coding and processing of external stimuli
within nervous systems. This includes a study of detectability and noise
discrimination and estimation, the transmission and reduction of
information, the phenomenology of receptive fields and cortical maps, and
mechanisms and models for cooperative phenomena in cortex. In addition to
lectures on the above topics, there are reviews on relevant areas of
applied mathematics.
The laboratory section of the course provides each student with a unique
opportunity to advance her or his analytical and numerical simulation
skills in modeling specific aspects of nervous systems. It consists of a
number of tutorial projects, together with an individualized modeling
project. Each student is supplied with a UNIX graphic-color workstation
and has the opportunity to learn about and use state-of-the-art software
designed for the analysis of both single-cell dynamics and large network
properties. The packages include GENESIS and NEURON as well as DSTOOL and
PHASEPLANE. In addition, the general mathematical analysis package
MATHEMATICA, and the interactive graphics package IDL, are available.
The course is designed for advanced graduate students, postdoctoral
fellows, and faculty members who have trained in a variety of
disciplines, including neurobiology, physics, electrical engineering,
computer science, and psychology. A background that includes familiarity
with neurophysiology, basic mathematical skills, i.e., calculus, linear
algebra and ordinary differential equations, and computer programming
skills, is highly desirable. Admission to the course is limited to 23
students.
Directors: David Kleinfeld and David W. Tank, Biological Computation
Research Department, AT&T Bell Laboratories, Murray Hill, NJ 07974.
Faculty: Joseph J. Atick (Rockefeller), William Bialek (NEC), Rodney
Douglas (MRC), Bard Ermentrout (Pittsburgh), William N. Frost (UT
Houston), Apostolos P. Georgopoulos (Minnesota), Charles M. Gray (Salk),
John J. Hopfield (Caltech), Christof Koch (Caltech), Nancy Kopell
(Boston), Stephen M. Kosslyn (Harvard), John E. Lisman (Brandeis),
Rodolfo R. Llinas (NYU), Eve E. Marder (Brandeis), John H. R. Maunsell
(Baylor), David A. McCormick (Yale), Bruce L. McNaughton (Arizona),
Kenneth D. Miller (Caltech), John Rinzel (NIH), David A. Robinson (Johns
Hopkins), Idan Segev (Hebrew U), Terrence J. Sejnowski (Salk), H.
Sebastian Seung (AT&T), Arthur Sherman (NIH), Karen Sigvardt (UC Davis),
Haim Sompolinsky (Hebrew U), Michael Stryker (UCSF), Roger Traub (IBM),
David Van Essen (Washington U).
Laboratory Instructors: David Berkowitz (Yale), David Golomb (NIH),
Michael Hines (Duke), Roderick Jensen (Texas A&M), Terrance Kovacs
(AT&T), Rafael Yuste (AT&T).
For further information and application forms:
Dorianne Chrysler
Admissions Coordinator
Marine Biological Laboratory
Woods Hole, MA 02543, USA
(508) 548-3705, ext. 401
TUITION:
$1,500 (includes room and board)
Partial financial aid is available
APPLICATION DEADLINE:
May 21, 1993
------------------------------
Subject: Call for Volunteers
From: "Andrew J. Worth" <andy@cma.MGH.Harvard.Edu>
Date: Wed, 27 Jan 93 11:23:36 -0500
- ---------------------------------------------------------------------------
** CALL FOR VOLUNTEERS **
1993 IEEE INTERNATIONAL CONFERENCE ON NEURAL NETWORKS
SECOND IEEE INTERNATIONAL CONFERENCE ON FUZZY SYSTEMS
March 28 - April 1, 1993
San Francisco Hilton
San Francisco, California, USA
- ---------------------------------------------------------------------------
Volunteer positions are available for both FUZZ-IEEE'93 and ICNN'93.
If you or anyone you know would like to exchange admittance to the
conference for working as a volunteer, please respond directly to me at
the e-mail address below.
In the past, volunteers have given approximately 20 hours of labor
(spread out over the entire conference) to receive:
- admittance to the conference
- a full set of proceedings
- attendance to a limited number of tutorials (while working)
Volunteer positions include helping at:
- Stuffing Conference Proceedings
- Poster Sessions
- Technical Sessions
- Evening Plenary Sessions
- Social Events
- OPTIONAL duty: Tutorials
If you are interested in volunteering, please respond directly to me
with the following information:
- Electronic Mail Address
- Last Name, First Name
- Address
- Country
- Phone and FAX number
Positions will be filled on a first commit first served basis. There
will be no funding available for volunteer's travel and lodging
expenses.
PLEASE RESPOND TO:
andy@cma.mgh.harvard.edu
Thank you,
Andy.
=---------------------------------------------------------------------=
Andrew J. Worth Center for Morphometric Analysis
Volunteer Coordinator Neuroscience Center
ICNN'93 / FUZZ-IEEE'93 Massachusetts General Hospital-East
(617) 726-5711, FAX:726-5677 Building 149, 13th St., MA 02129 USA
andy@cma.mgh.harvard.edu
=---------------------------------------------------------------------=
------------------------------
Subject: "neural-net based software for digitization of maps"
From: VEMURI@icdc.llnl.gov
Date: Thu, 28 Jan 93 15:14:00 -0800
A colleague from India is asking me if there is any effort on "neural
network based software for the digitization of maps and photos (maps and
photos presumably obtained from remotely-sensed data)" Is there any one
who can help answer this question? Thanks
Rao Vemuri
Dept. of Applied Science
UC Davis, Livermore, CA
vemuri@icdc.llnl.gov
------------------------------
Subject: Response to "Degeneracy of weights in networks?"
From: Yaakov Stein <stein@galaxy.huji.ac.il>
Date: Sat, 30 Jan 93 22:26:04 +0200
In Neuron Digest Volume 11 Issue 6 from Monday 25 Jan 1993
jjb@watson.ibm.com asked
> Before training a homogeneous artificial neural network, the expectation
> value for the trainable weights are all identical. After training,
> these weights will assume an organized pattern representing in some
> way the information encoded in the network by the training. Most
> of the artificial networks I have seen described are structurally
> degenerate: a pattern of three weights 1,2,3 cannot give different
> answers from 2,1,3 or 3,2,1. This degeneracy, a result of the
> homogeneity, is not present in bio-nets: they are not homogeneous.
>
> Am I wrong about this degeneracy? Does it not cause grief in numerical
> training to "optimal" weights?
I'm not quite sure what "homogeneous" means in this query, but if he's
talking about a fully connected feedforward net, then he is certainly not
wrong. The hidden units in a given layer are indistinguishable, and
interchanging two hidden units fed by different weights produces a net
equivalent to the first. This symmetry is difficult to exploit while
training continuous valued weights, but should not cause any grief.
Since the hidden units indistinguishable, and any number can have the
same couplings, they obey what physicists call Bose - Einstein
statistics. The fact that hidden units are Bosons and other symmetries
help us in estimating the capacity of such networks. I discuss these
questions at length in a paper entitled "On the Dichotomization Capacity
of Binary Sets using Feedforward Neural Networks", which will be
available soon.
Yaakov Stein
------------------------------
Subject: speaker adaptation and normalization
From: WEYMAERE@lem.rug.ac.be
Date: Wed, 03 Feb 93 11:33:00 +0100
[[ Editor's Note: As many of you know, one of the major problems in
speeach recognitio has been speaker independence. I hope the summary
will appear in a future issue of this Digest. -PM ]]
Dear Neural Networkers,
I'm looking for references on speaker normalization and speaker
adaptation with neural networks. Is somebody aware of the existence of
such articles? I will send a summary of the reactions to those who are
interested.
Thanks in advance.
Nico Weymaere
------------------------------
Subject: Re: NNs and economics
From: tarroux@biologie.ens.fr (Philippe TARROUX)
Date: Thu, 04 Feb 93 13:31:03 +0100
I don't really know a precise reference on the use of neural nets in
economics, but are you aware of the work done by the Santa Fe Institute
(Santa Fe New Mexico) the aim of which is to promote studies on the
science of complexity. I know they are interested in the application of
adaptive systems, learning procedures, neural networks and genetic
algorithms to complex systems modeling including economics. There were
two working groups there last year on the Theory of money and on
Theoretical computation in economics. The leader of the Santa Fe
Institute research network entitled 'Economy as an evolving, complex
system' is Martin Shubik Yale University. May be you should contact them:
SFI
1660 Old Pecos Trail
Suite A
Santa Fe, NM 87501
505/984-8800
to email try ginger@santafe.edu
Good luck
------------------------------
Subject: vision position posting
From: ira@linus.mitre.org
Date: Fri, 05 Feb 93 10:06:53 -0500
Neural Network Vision Research Position
The MITRE Corporation is looking for a Vision Modeler with an excellent
math background, knowledge of signal processing techniques, considerable
experience modeling biological low-level vision processes and broad
knowledge of current neural network learning algorithm research.
This is an *applied* research position which has as its goal the
application of vision modeling techniques to real tasks such as 2D and 3D
object recognition in synthetic and real world imagery. This position
requires software implementation of models in C language. The position
may also involve management responsibilities.
The position is located in Bedford, Massachusetts. We are looking for
someone with availability within the next two months. Note: US
CITIZENSHIP REQUIRED. Interested applicants should send a resume and
representative publications to:
Ira Smotroff
Lead Scientist
The MITRE Corporation
MS K331
202 Burlington Rd.
Bedford, MA 01730-1420
------------------------------
End of Neuron Digest [Volume 11 Issue 9]
****************************************
