NetNews Usenet Archive 1992 #16/NN_1992_16.iso/spool/comp/ai/neuraln/2846

Internet Message Format | 1992-07-20 | 22.3 KB

Path: sparky!uunet!cis.ohio-state.edu!ucbvax!CATTELL.PSYCH.UPENN.EDU!neuron-request
From: [e-mail] ("Neuron-Digest Moderator")
Newsgroups: comp.ai.neural-nets
Subject: Neuron Digest V9 #37 (discussion/misc)
Message-ID: <[e-mail]>
Date: 21 Jul 92 05:16:53 GMT
Sender: [e-mail]
Reply-To: "Neuron-Request" <[e-mail]>
Distribution: world
Organization: University of Pennsylvania
Lines: 573

Neuron Digest   Tuesday, 21 Jul 1992
                Volume 9 : Issue 37

Today's Topics:
               Administrivia - ND will go on holiday soon
                   Supporting Nomination Solicitation
                   Practical Books on Neural Networks
                            Re: Sensor fusion
                           What is Khoros (A)
  Combining stochastic and symbolic inference with analog computation..
                             Hopfield Paper
                     Refs on NN in Telecommunication
Request for references on Hybrid systems/applications on Medical Diagnosis.
                     request (papers and references)
                postdoctoral research position available
                  Help! Neuron net texts and software.


Send submissions, questions, address maintenance, and requests for old
issues to "[e-mail]". The ftp archives are
available from cattell.psych.upenn.edu (128.91.2.173). Back issues
requested by mail will eventually be sent, but may take a while.

----------------------------------------------------------------------

Subject: Administrivia - ND will go on holiday soon
From:    "Neuron-Digest Moderator, Peter Marvit" <[e-mail]>
Date:    Tue, 21 Jul 92 00:41:26 -0500

Dear readers,

At the end of this week, your moderator will be going on a three week
holiday. As a result, Neuron Digest will suspend publication until
sometime during the week of 18 August.  I will make another note later
this week as a final reminder.  During the hiatus, of course, email will
accumulate but will not be processed.

As always, thanks for your efforts and submissions.

        -Peter Marvit
         Moderator

: Courtesy of Psychology Department, University of Pennsylvania
: [e-mail]


------------------------------

Subject: Supporting Nomination Solicitation
From:    [e-mail] (ISSNNet)
Organization: International Student Society for Neural Networks
Date:    10 Jul 92 21:12:47 +0000

[[ Editor's Note: Readers are reminded of my comments about a similar
posting in Vol 9 #34. Obviously, this publication is not an endorsement
of the candidate, but I do support this valuable student-run organization
and hope many more people will get involved. -PM ]]

My name is Weilan Wu, the position I am running for is the President or
Vice President of ISSNNET.

I am a graduate student in the Department of Computer Science, Washington
University.

Now I am in the doctoral program, and I am interested in the areas of
neural networks for two years. The work I did included constructing a
feedforward network to recognize the handwritten characters on maps or
mechanical drawings, the other research I have been doing is using the
neural net approach to extract high-level features in spoken speech. Some
of our work will be presented on the Cognitive Science Conference in
Bloomington in August.

Thanks for the supporting.

Weilan

[Please send supporting nominations to: Weilan ([e-mail]) or
 [e-mail].  AJW]

------------------------------

Subject: Practical Books on Neural Networks
From:    [e-mail]
Date:    Sat, 11 Jul 92 21:58:34 -0500

                                        From:     MAJOR PETER G. RAETH
                                        Dept:     AAWP-1
                                        Tel No:   DSN-787, 513-257-5366

Per your request in the recent Neuron Digest, enclosed is a list of books
of which I am familiar that give practical guidance on implementing
neural networks.  This list was composed by myself and Mott Given of
Defense Logistics Agency.

If your search has resulted in other books, we would appreciate receiving
your list.

Thanks.

Best.

Pete.


Practical Neural Network Books

Each of the books listed below gives code (some on disk) and offers
guidance on practical application of various neural network models.
These are the books known to this reviewer.

Rumelhart, David and James McClelland, Explorations in Parallel
Distributed Processing, Cambridge, MA: MIT Press, 1988

Korn, Granino A, Neural Network Experiments on Personal Computers
and Workstations, Cambridge, MA: MIT Press, 1991

Caudill, Maureen and Charles Butler, Understanding Neural
Networks, Vol I and II, Cambridge, MA: MIT Press, 1991

Eberhart, Russell C. and Roy W. Dobbins, Neural Network PC Tools,
Cambridge, MA: Academic Press, 1990

McCord, Nelson M. and W. T. Illingworth, A Practical Guide to
Neural Nets, Reading, MA: Addison-Wesely, 1990

A diskette offered by AI Expert magazine.  This collection of
programs goes along with several articles published over time.

Staff Writers, Neural Teacher, Salt Lake City, UT: Softlabs Corp,
1989

Staff Writers, NeuralWorks Explorer, Pittsburg, PA: NeuralWare,
1990

Aleksander, Igor, An Introduction to Neural Computing, London,
England: Chapman and Hall, 1990

Orchard, G. A. and W. A. Phillips, Neural Computation, East
Sussex, England: LEA Ltd, 1990

Blum, Adam, Neural Networks Programming in C++, New York, NY:
John-Wiley, 1992

Muller, B. and J. Reinhardt, Neural Networks: An Introduction,
New York, NY: Springer-Verlag, 1990

Freeman, James, A. and David M. Skapura, Neural Networks:
Algorithms, Applications, and Programming Techniques, New York,
NY: Addison-Wesley, 1991



------------------------------

Subject: Re: Sensor fusion
From:    Guanghua Zhang <[e-mail]>
Organization: Dept of Computer Science, Heriot-Watt University, Scotland
Date:    Mon, 13 Jul 92 11:49:27 +0000

>Subject: Request for info on sensor fusion
>From:    [e-mail]
>Date:    Fri, 26 Jun 92 11:17:13 -0500
>
>[[ Editor's Note: I have *no* knowledge on this subject.  Any help from
>you readers. -PM ]]
>
>Recently I have been doing some work on sensor fusion for the Air Force.
>Will someone kind enough to provide me with pointers to literature on
>visible and IR sensor fusion. Any assistance is very much appreciated.
>
>Simon Y. Foo, Ph.D.
>Dept. of Electrical Engineering
>FAMU/FSU
>Tallahassee, FL 32306
>email: [e-mail]

Sensor fusion is a big topic, and there are a lot of approaches.
The folloing may be useful to you.

A. Mitiche, T. C. Henderson and R. Laganiere,
"Decision netwroks for multisensor integration in computer vision",
in SPIE: sensor fusion and scene interpretation, pp291-299, 1988

A. Mitiche and J. Aggwrwal,
"Multiple sensor integration/fusion through image processing: a review",
Optical Engineering, vol. 25, n0.3, pp380-386, 1986

P. Ajjimarangsee and T. L. Huntsberger,
"Neural network models for fusion of visible and infrared sensor outputs"
in SPIE: sensor fusion and scene interpretation, pp153-160, 1988

R. C Luo and M. G. Kay,
"Multisensor integration and fusion in intelligent systems",
in IEEE Trans. Sys. Man. and Cyb., Vol.19, No.5, pp901-931, 1989

N. Nandhakumar and J. K. Aggarwal,
Integrated analysis of thermal and visual images for scene interpretation"
in IEEE Trans. Patt. Anal. Mach. Intel., vol.10, no.4, pp469-481, 1988

C. Chu, N. Nandhakumar and J. K. Aggwarl,
Image segmentation using laser radar data"
Pattern recognition, vol.23, no.6, pp569-581, 1990
- --
  Guanghua Zhang               JANET:  [e-mail]
  Dept. of Computer Science,    UUCP:  <EUROPE>!ukc!hwcs!guanghua
  Heriot-Watt Univ., Edinburgh    or:  [e-mail]
  Tel: [phone] ext 4177   Fax:  [phone]


------------------------------

Subject: What is Khoros (A)
From:    Steve Pendergast <[e-mail]>
Date:    Mon, 13 Jul 92 08:59:59 -0800

[[ Editor's note: Vol 9 #35 had a description of how Khoros can be used
for Neural Net modeling. I commented on my ignorance of Khoros itself.
Thanks to Steven Penderast for the following description. -PM ]]

My last submission refered to Khoros, which is unfamiliar to some people.
The following describes this program and its availability:


Khoros

Another popular package is Khoros, a powerful and extensive image
processing and visualization software development environment from the
University of New Mexico. Khoros components include a visual programming
language, code generators for extending the visual language and adding
new application packages to the system, an interactive user interface
editor, an interactive image display package, an extensive library of
image processing, numerical analysis and signal processing routines, and
2D/3D plotting packages.

The main goal of the Khoros software project is to create and provide an
integrated software development environment for information processing
and data visualization. The Khoros software system is now being used as
a foundation to improve productivity and promote software reuse in a
wide variety of application domains.  A powerful feature of the Khoros
system is the high-level, abstract visual language that can be employed
to significantly boost the productivity of the researcher.

Central to the Khoros system is the need for a consistent yet flexible
user interface development system that provides cohesiveness to the vast
number of programs that make up the Khoros system.  Automated tools
assist in maintenance as well as development of programs. The software
structure that embodies this system provides for extensibility and
portability, and allows for easy tailoring to target specific application
domains and processing environments.

The Khoros system integrates multiple user interface modes, code
generators, instructional aids, data visualization, distributed computing
and information processing.  The result is a comprehensive tool for
computational research and development.  The Khoros infrastructure
consists of six major subsystems:

(1)  A high-level visual language that may be extended to fit the
application domain.

(2)  A user interface development system consisting of a User Interface
Specification (UIS) and code generators that use the UIS to generate
code for all programs created under the Khoros system.

(3)  An interoperable data exchange format (VIFF) supported by a
comprehensive automatic storage conversion library.

(4)  Application specific data display and processing libraries; image
processing, digital signal processing, numerical analysis, data and file
conversion, graphics display, and image display.

(5)  A set of interactive X Windows-based programs address such areas as
image display, colormap manipulation, animation, plotting, warping of
image data, and surface visualization.

(6)  A set of meta-system calls provide a foundation for distributed
computing and efficient data transport. Khoros provides user selectable
compute locations via the visual language, computer supported
cooperative work via a distributed X mechanism, and remote access to
data via a distributed data transport server.

The software structure that embodies this system provides for
extensibility and portability, and allows for easy tailoring to target
specific application domains and pro- cessing environments. Khoros is a
successful demonstration of how research programming, application
programming, infor- mation processing, data visualization, instruction,
documen- tation, and maintenance can be integrated to build a state-
of-the-art software environment.

Feature Descriptions

THE VISUAL LANGUAGE

The visual language of Khoros, cantata, is a graphically expressed,
data-flow oriented language.  The user builds a cantata application
program ("workspace") by connecting pro- cessing nodes ("glyphs") to
form a data flow graph.  Glyphs are selected from the many library
routines available with Khoros; the user may also create new library
routines using the Khoros user interface development system, and make
them available to cantata.  The visual language is considered to be
large-grain because each glyph represents an entire program.  "Control"
glyphs (providing if/else conditions, loops, etc) and a expression
parser extend the functionality of the underlying data flow methodology.

Visual "procedures", each representing a segment of the net- work of
glyphs in a workspace, can be used to add structure to the visual
language and help to manage the complexity often associated with visual
programming.  A dynamic execu- tion scheduler allows the user to
interactively execute the entire flow graph across a heterogeneous
computer network. The execution can be set to either a demand driven or
data driven model depending on the application and desired level of
interactivity.

Host Configuration Requirements

To run Khoros, you must have a UNIX platform running the X11R3, X11R4,
X11R5, or OpenWindows servers. The Khoros system consists of ~450,000
lines of C code;  you will need a maximum of 120 Megabytes of storage
for source, documentation and binaries.  To compile Khoros, the system
requires X11R4 (patch levels 1-18) or X11R5 client software from MIT and
an additional 50 Megabytes of storage.

The Khoros user community has ported Khoros to various other platforms
such as the Convex, the Sequent, Data General Aviion, MacII, and 386/486
machines. At this time, these platforms are not directly supported by
the Khoros Group.

Availability & Support

Khoros is available via anonymous ftp at no charge (from
pprg.eece.unm.edu, 129.24.24.10, in the /pub/khoros directory), or a
tape and printed documentation can be ordered for $250.00. Support is
provided via both a user's group mailing-list: [e-mail]
(send subscription requests to khoros- [e-mail]); and
a USENET Newsgroup:  comp.soft- sys.khoros. For more information, send
e-mail to [e-mail].  All communication to the
Khoros group should be via e-mail.

CURRENT STATUS OF KHOROS (Patch Level 5)

The Khoros system has been applied to a diverse spectrum of areas
including medical research, ecological studies, remote sensing, virtual
reality, telecommunications, and optics. The Khoros project continues to
evolve as new developments occur in the continued research and
development of software environments for information processing, data
visualization and software development.  Khoros is being used as a
teach- ing tool at several Universities.  Release 1.0 of Khoros was
offered free of charge via public ftp access on May 10, 1991; updates
have since been provided in patch form. Release 2.0 of Khoros is
expected to be out in late 1992.

==============================
Hope this helps

Stephen L Pendergast, Senior Scientist/Engineer, Hughes Aircraft Company
Ground Systems Group, PO Box 3310 Bldg 604/MS B152, Fullerton, CA 92634
Voice: [phone]  Personal - No Company/Official Responsibility
Email: [e-mail]           Fax: [phone]


------------------------------

Subject: Combining stochastic and symbolic inference with analog computation..
From:    Badri Roysam <[e-mail]>
Date:    Sat, 11 Jul 92 10:07:52 -0500


I am writing this note to bring our recent paper, which was published in
the signal processing literature, to the attention of neural network
researchers:

Badrinath Roysam and Michael I. Miller, "Combining Stochastic and
Syntactic Processing with Analog Computation Methods", Digital Signal
Processing, Academic Press, Volume 2, pp. 48-64, April 1992.

Some key words:

generalized bayesian hypothesis testing
regular grammars
stochastic diffusions
positron emission tomography
3-D confocal microscopy
noise removal

Badri Roysam
Assistant Professor
Department of Electrical, Computer and Systems Engineering
Rensselaer Polytechnic Institute, Troy, NY 12180
Office:518-276-8067, Home:518-283-0417, Fax:[phone]
Lab: 518-276-8207


------------------------------

Subject: Hopfield Paper
From:    [e-mail]
Date:    Wed, 15 Jul 92 14:18:00 -0800

[[ Editor's Note: From this note, I can only assume a request to Hopfield
himself (at Caltech?) produced nothing.  Readers, any volunteers? -PM ]]

I am looking for a clean reprint of Hopfield's original paper that
appeared in the Proc. Natl. Acad. Sci. I am trying to include this in a
collection of reprints I am editing and not able to locate a CLEAN copy.
If someone has a clean reprint and willing to lend it to me, I will
guarantee that it will be returned back to you safely after the printer
takes a photocopy of it.

Here is the complete reference.

Hopfield, J. J. Neural networks and physical systems with emergent
collective computational properties. Proc. Natl. Acad. Sciences, Vol. 79,
pp 2554 to 2558, April 1982.

A xerox copy from a bound volume is not acceptable because of the poor
quality at the edges. Can someone help?

Thanks

V. Rao Vemuri
Editor-in-Chief
Computer Society Press of IEEE

Dept. of Applied Science
University of California at Davis
Livermore, CA 94550
[phone]


------------------------------

Subject: Refs on NN in Telecommunication
From:    "P. de Bruyne" <[e-mail]>
Date:    Thu, 16 Jul 92 08:26:00 +0100

I would like to add the following reference on NN in telecommunications
to your list:

SPREAD-SPECTRUM DIGITAL SIGNAL SYNCHRONIZATION USING NEURAL NETWORKS by
de Bruyne, P., Kjelsen, O., and Sacroug, O., Laboratory of Communication
Technology ETHZ Switzerland. To be pubished in " Proceedings of the IEEE
International Conference on Security Technology" 1992 (October,
Atlanta,GA) 13 pages, 11 Refs.


------------------------------

Subject: Request for references on Hybrid systems/applications on Medical Diagnosis.
From:    "Heitor S. Lopes" <[e-mail]>
Organization: Universidade Federal de Santa Catarina/BRASIL
Date:    Thu, 16 Jul 92 10:47:22 -0300

      I would appreciate to receive references of papers/technical
reports on hybrid systems, specially when applied to medical diagnosis.
E-mail's of the authors would be also interesting. I will post a summary
to the net of the responses received. Thanks a lot.


------------------------------

Subject: request (papers and references)
From:    Jose Leomar <[e-mail]>
Organization: Universidade Federal de Santa Catarina/BRASIL
Date:    Thu, 16 Jul 92 11:29:20 -0300

       Dear Marvit,

I'de like to receive papers and references about pattern recognition and
pattern classification. We have dificulties to obtain  good  papers  and
books about neural networks in Brazil. I will go to use the paradigm  in
my thesis. My thesis involve the classification of chromossomes and pos-
terior analisys. I'd like too, to know wich the better net to use   this
case.
Please, sorry eventual English errors.

       Sincerly yours,
                           Jose Leomar Todesco
Acknowledge-To: <EPS3JLT@BRUFSC>


------------------------------

Subject: postdoctoral research position available
From:    Tony_Prescott <[e-mail]>
Date:    Wed, 15 Jul 92 12:45:10 +0000


                    POSTDOCTORAL RESEARCH POSITION AVAILABLE

                  Artificial Intelligence Vision Research Unit
                          Sheffield University, England.


A postdoctoral research position has become available for two years (in
the first instance) from the 1st of October 1992 to work on neural
network control architectures for an autonomous mobile robot.

The goal of the research is to develop small, fast moving, autonomous
mobile vehicles capable of real-time navigation in an unstructured indoor
environment.  The research aims to develop navigation methods that can
exploit the low resolution optic flow information provided by a novel
real-time visual sensor. The project is in collaboration with Oxford
University Department of Engineering who will be building the
transputer-based robot platform and sensor systems.

The research at AIVRU will be concerned with the development of neural
network control strategies for obstacle avoidance and map-building, this
will build on work already conducted in simulation using reinforcement
learning methods.  The postdoctoral reseacher will be concerned with
implementation and integration of control systems on the robot platform
and with the development of new algorithms in simulation and on the
vehicle.

AIVRU is one of the largest and best-funded AI research groups in the UK.
Current projects include 3D Object and scene modelling, the development
of a fast transputer based vision engine, and 3D vision for vehicle
trajectory control.  There is ongoing neural network research in
reinforcement learning, object recognition and control algorithms for a
stereo camera rig.  The group has industrial collaborations with IBM, GEC
and Toyota.  AIVRU also conducts research in psychophysics and being
based in the Department of Psychology at Sheffield has strong links with
the Neuroscience and Cognition research groups.

Sheffield is the fifth largest city in Britain, it is situated in the
heart of England on the edge of the Peak District one of the country's
finest national parks.

For further information please write to:

Professor John Mayhew,
Artificial Intelligence Vision Research Unit,
University of Sheffield,
Sheffield S10 2TN.

tel: [phone] ext 6554
fax: [phone]

or email to:

[e-mail]  (Tony Prescott)





------------------------------

Subject: Help! Neuron net texts and software.
From:    [e-mail] (Alan Macdonald)
Date:    Sun, 19 Jul 92 12:23:19 -0600

[[ Editor's Note: I hope the posting earlier in this issue by Peter Raeth
will be of some assistance.  It seems time again for another "beginner
book/article list." Volunteers...again? -PM ]]

I am planning a seminar in neural nets for Fall semester. It will meet
about 14 times over the fall semester. The students are computer science
majors with knowledge of calculus and linear algebra. I (the prof) am
reading various articles about the subject, but have not found an
appropriate text and software (preferably integrated). I am late in
getting started on this because I just returned from a year abroad. I
would appreciate any suggestions you all might have.


                                Thank you,

                                Alan Macdonald



------------------------------

End of Neuron Digest [Volume 9 Issue 37]
****************************************