VIRGINIA MOUNTAIN SECTION
NEWSLETTER
IEEE Region 3, Council 9, Section 65 March 1997
********************************************************
CONTENTS
1. March Meeting
Tour of Cybermotion, Inc.
2. Reservations for March Meeting
3. February Meeting Report
Student Paper Contest, Summary & Winners
4. PACE Patter
Networking Most Effective Job Search Technique
5. Computer/Control/IES Chapter
March Meeting: " Metal Industries Automation Systems -- a
Decade of evolution "
6. Computer/Control/IES Chapter
February Meeting Report: "A Practical Guide to
Neural Networks"
7. IEEE Virginia Council
Descriptive summary by Dan Jackson
8. VMS Info Sources
9. Chapter Chairmen - Notice
10. For Your Information
11. Newsletter Submissions
******************************************************** [1]
March Meeting
Tour of Cybermotion, Inc.
The VMS IEEE will be touring facilities of Cybermotion, Inc. in
Salem. Cybermotion, Inc. has been involved in the design and
manufacture of mobile robot systems, particularly security robots,
since 1984. Our meeting will start with videos about Cybermotion
and their products, will proceed to a tour of the manufacturing
facilities, and conclude with a question and answer session with
John Holland, the founder of the company.
The meeting will start at 7:00 P.M. at Cybermotion, Inc. Note,
there will be no dinner scheduled. Space is limited, so
reservations are imperative.
******************************************************** [2]
Reservations for the March Meeting
Date: Thursday, March 20, 1997
Time: 7:00 P.M.
Place: Cybermotion, Inc.
115 Sheraton Drive, Salem
off Electric Road (Rt. 419) near Quality Inn,
I-81 exit 141
Cost: none
PLEASE MAKE RESERVATIONS BY MONDAY,
MARCH 17, 5:00 P.M.
Roanoke: David Livingston 857-6261
Blacksburg: Anbo Wang 231-4355
Lexington: Dick Skutt 464-7236
Radford and
Christiansburg: Usha Varshney 731-0655
******************************************************** [3]
February MEETING REPORT
The 1997 VMS IEEE Student Papers Contest
The 1997 VMS IEEE Student Papers Contest was held at the
Holiday Inn, Salem on February 20. The contestants were Mr.
David Hepper, a sophomore in the Engineering Transfer program
at Virginia Western Community College; Mr. James Jeter and Mr.
Taylor Edwards, seniors in the Electrical Engineering program at
Virginia Military Institute; and Mr. Paolo Dadone, a masters
candidate from the Electrical Engineering graduate program at
Virginia Polytechnic Institute and State University.
Mr. Hepper presented the first paper entitled "Nanotechnologies:
the Quantum Dot." In the presentation, Mr. Hepper discussed the
limitations involved in shrinking current integrated circuit
technologies to the sub-micron level. These limitations included
strong field effects, removal of heat and quantum effects such as
tunneling. Mr. Hepper then discussed the use of the quantum dot
as a replacement for transistors in switching applications. This
concept involves the control of single electrons in the
representation of logic states. Mr. Hepper concluded by listing
some of the obstacles facing the use of these new technologies.
Mr. Jeter and Mr. Edwards presented the second paper: "Fourier
and Wavelet Analysis of a Speech Pattern." The presenters
started by giving a brief introduction into how speech is produced
and how the Fourier transform has been used to analyze speech
signals. They mentioned that one of the problems with using the
Fourier transform to represent speech is the loss of temporal
information. Mr. Jeter and Mr. Edwards went on to discuss the
use of wavelets and wavelet packets. Using wavelets requires
fewer coefficients than the Fourier transform representation and
temporal information is preserved. They concluded by suggesting
applications of wavelet techniques in areas such as voice
identification security and the identification of speech pathologies.
Mr. Dadone gave the final presentation entitled "Adaline: Pattern
Classification." Mr. Dadone began by explaining pattern
classification using a maximum likelihood Gaussian classifier
(MLGC) to separate slightly overlapping data sets into two
distinct classes. He noted the linear structure in the log likelihood
and suggested using an adaptive linear element (Adaline) to
accomplish a similar result. Mr. Dadone explained the operation
of the Adaline and went on to discuss the results of an experiment
comparing the MLGC to the Adaline for a two-class problem
consisting of data generated from Gaussian distributions. The
conclusion was that the two techniques were only marginally
different but that the Adaline had the advantage in that it assumed
no a priori model.
All the presentations were well done, resulting in a very close
contest. It was the judges decision that Mr. Hepper was awarded a
first prize of $100, and Mr. Jeter and Mr. Edwards were awarded
a second prize of $50 in the undergraduate division. Mr. Dadone
was awarded a first prize of $100 in the graduate division. The
VMS offers congratulations to the contestants and also offers its
gratitude to Dr. Russell Churchill and Dr. Usha Varshney for
providing their services as judges.
...Dave Livingston
******************************************************** [4]
PACE Patter
Networking Most Effective Job Search Technique
In the November and January issues of this Newsletter we
described the PACE Conference and issues from the Career Policy
Council. This month we discuss the Members Services Council.
This council consists of committees which develop programs and
services that support individual members, sections and section
leadership. Included are Awards and Recognition,
Communications, Employment Assistance, Alliance of IEEE
Consultants Networks, Surveys, and State Government Activities.
Papers in several of these areas were presented at the annual
PACE Conference last September.
A survey of unemployed IEEE members and those who had
recently found jobs revealed a significant difference in the
perceived importance of "networking" as a successful job search
technique. Networking is the exchanging of information among
individuals to build relationships and help each other. The
reemployed members found networking by far the most effective
job search technique when compared with resumes, ads,
headhunters, consultants, and job fairs. Depending on the source
of reemployment the difference was from 50 percent more
effective to twice as effective. Those still unemployed considered
resumes equally as effective. Maybe that was why they were still
looking for a job. All found consultants the least effective
technique for getting a job.
The Communications Committee, also, was concerned with job
search. Resumes are a means of communication in which an
individual is trying to convince the prospective employer or client
of the value which he or she can bring to the organization. This is
hard enough in person, but to do so on paper is even more so. The
workshop concentrated on characteristics of good resumes, job-
search planning, and skills needed for effective interviewing.
Communications covers, also, external communications to
publicize section activities to the general public through electronic
and print media. The session on "Building a Section Media
Network" covered building the network, types of messages,
methods for delivering the message, and maintaining the network.
Surveys are an important means of determining the needs of our
members, but they are useful only if done correctly. The paper and
session on surveys was intended to be a guide to surveys and
survey techniques as they pertain to IEEE. Smaller surveys done
by sections may lack the statistical validity of the total Institute
surveys but nevertheless provide some useful information. A
session on how to create a new award was part of the program of
the conference. This reviewed the IEEE requirements and
presented a sample of a recent new award as a guide.
If you want more information in any of the areas discussed contact
your section PACE chair, Daniel W. Jackson, 540-774-0484,
d.jackson@ieee.org.
...Dan Jackson
******************************************************** [5]
Computer/Control/IES Chapter
March Meeting
" Metal Industries Automation Systems -- a
Decade of evolution "
R. Srinivasan Metals Automation GE Industrial
Systems
Tuesday, March 11, 1997
5:30 - 7:00 PM Cafeteria A
GE Industrial Systems
1501 Roanoke Boulevard Salem, VA
Metal industry automation systems are large, real-time systems
with stringent performance and uptime requirements.
This talk will present an overview of Level 2 control system
requirements for rolling mills and other metals processes. The
evolution of GE designs to meet these requirements will be
examined. Successes and failures will be discussed.
Variations among mills have been tricky enough to make
standardization very challenging. A synopsis of the quest for the
"perfect software architecture" over the past decade will be
outlined.
R. Srinivasan ("Srini")graduated in Mechanical Engineering from
Indian Institute of Technology, in Madras India. After a brief
period in the IBM mainframe world, most of his career has been
centered around DEC Systems, spanning the RSX, RSTS, RT11,
VMS, ELN Operating Systems, and PDP 11, VAX11 and now the
Alpha/AXP architectures.
In 1986 Srini came to GE Drive Systems as a contract
programmer. He joined GE full time in 1989. He has been
involved in the evolution of the Metals Level 2 Automation
Systems ever since.
Since 1991 or so, Srini has concentrated on providing Operator
Interface solutions which exposed him to the innards of several
generations of GE Automation Systems.
All IEEE Members and Guests are Welcome. There is no charge.
Come to the main entrance near the flagpole. Meetings start
promptly at 5:30, so come early.
RESERVATIONS are appreciated! Call Dave Geer, 540-387-
7359 by 5 PM Monday, March 10
******************************************************** [6]
February Meeting Report
"A Practical Guide to Neural Networks"
Dr. David L. Livingston
Dr. Livingston began his lecture with an overview of the
biological inspiration for Artificial Neural Networks (ANNs).
Biological neurons consist of dendrites, soma, and axons, which
perform input, processing, and output of neural "signals". The
communication channel between neurons is the synaptic junction
between the axon of one neuron and the synapse of another. The
synaptic junction forms a threshold, across which the ionic signal
must fire. Communication between neurons exhibits frequency
modulation. As the signals increase in strength, the axons fire
more rapidly. A biological neuron can be modeled as a weighted
sum of inputs, plus a bias. The output is the bounded value of the
sum. A smooth function, such as hyperbolic tangent, is usually
chosen as the bound. Networks of these artificial neurons can
exhibit learning by automatically adjusting the input weights.
Artificial neurons can be connected in many ways. One
commonly used is the feedforward network. In this configuration
neurons are arranged in layers. The input layer simply accepts
signals. This layer is connected to one or more hidden layers,
connected in cascade. The final layer is the output layer. Outputs
from the network appear only at this layer. A number of learning
algorithms can be used to adjust the weights. One that has proved
useful is backward propagation. In order to use this method, a
large number of sets of input and output vectors are required. The
output vector represents the correct result for the corresponding
input vector. Beginning at some initial state, the network is
presented with each input vector in turn. The result is compared to
the correct output, and a portion of the error is propagated back
through the network to adjust the weights at each layer. A
gradient descent method, using the partial derivative of the error
with respect to each weight, is used to calculate the new weights
at each iteration. After many iterations, if we are lucky, the
weights will converge to stable values, and the network is
"trained".
Other methods of learning were briefly mentioned, such as
competitive learning, adaptive resonance theory, and annealing
algorithms.
Although artificial neural nets have been used to solve certain
classes of problems, there are many obstacles to general
application. The optimal number of layers, and the number of
nodes in each layer is not known in advance. The gradient decent
learning algorithm may be fooled by local peaks and valleys in the
error surface. The rate of learning, governed by the amount of
error fed back at each iteration, and the convergence criteria, also
affect the accuracy of the final configuration. And then there is
the question of when does learning stop?
Dr. Livingston included historical milestones in the development
of ANNs, and concluded with a summary of future work. About
15 people attended, including engineers from GE, and professors
from Virginia Tech and Washington and Lee.
... Dave Geer
******************************************************** [7]
IEEE VIRGINIA COUNCIL
Very few members of the Virginia Mountain Section are aware of
the existence of the Virginia Council. The IEEE Virginia Council
is Area 1 of Region 3 of the IEEE. Region 3 has nine areas
covering the nine states east of the Mississippi River including
Kentucky, Virginia, and all states south plus Jamaica.
Purpose:
1. Stimulate member participation in the Institute.
2. Improve communications among and for its member sections.
3. Coordinate activities of its member sections.
4. Promote statewide meetings and activities for the benefit of its
members.
5. Enhance status of membership in the IEEE.
6. Provide strong, united representation of its member sections
in the Regional and National activities of the Institute.
Member Sections:
Those sections in Virginia which are in Region 3.
1. Central Virginia Section.
2. Hampton Roads Section.
3. Richmond Section.
4. Virginia Mountain Section.
Officers; Four:
Chair, Vice-Chair, Secretary, Treasurer.
One officer from each section. Each office rotates among the
sections. 1997: Chair, Richmond; Vice-Chair, Hampton Roads;
Secretary, Virginia Mountain; Treasurer, Central Virginia. Term
of office: two years.
Executive Committee:
The four officers, the Junior Past Chair, and the Chair of each of
the four member sections. The chairs of the member sections serve
only during their term as a Section Chair.
The Virginia Council is a partner with the IEEE Northern
Virginia Section, part of Region 2, in the Virginia Coordinating
Council. This body provides a cooperating vehicle through which
all IEEE members in Virginia can work together in activities of
benefit to all. One of the primary activities of this latter group is
communication with the state government particularly in
cooperation with other Virginia engineering organizations on
matters of concern to engineers of all disciplines.
The IEEE Virginia Council has been a member of the Virginia
Engineering Awards Committee, (includes ASME, ASCE, VSPE
and Consulting Engineers Council). This group awards the Crozet
Award, to recognize outstanding engineers in non-federal
government service, and the Pletta Award for outstanding
engineering educators.
The Virginia Council has an infrequent newsletter which goes to
the leaders of the sections for information about the council and
the other sections in Virginia.
Meetings of the Virginia Council executive committee are held
quarterly, usually in Richmond, the most central location. In the
past the meetings have been moved around the state. The Virginia
Council has sponsored two statewide technical conferences,
VACon '82 in Hampton, and VACon '84 in Blacksburg at
Virginia Tech.
As an entity of the IEEE, the Virginia Council has bylaws which
guide its operation. If any readers wish a copy let me know at
. I will send you the six pages by US mail.
...Daniel W. Jackson
******************************************************** [8]
VMS Info Sources
Electronic Newsletter
Over 15 percent of our members now subscribe. If you have not
yet signed up, you are strongly encouraged to do so.
VMS Home Page
About VMS, its Chapters, officers, by-laws, and links to IEEE,
Region 3, the Student Chapter, and a few more. Still an infant so,
please take a look and send your suggestions. You can do that
while you are looking at it. The URL is:
http://fiddle.ee.vt.edu/ieeevms
To Subscribe to any or all three, see instructions in the For
Your Information section.
******************************************************** [9]
Chapter Chairmen
Places have been reserved in the Home Page for any and all
kinds of information about your Chapter: History, function,
requirements, activities, schedules, flash announcements,
meetings, speakers ... . Please send your material to the editor or
submit it with the auto-mail feature.
******************************************************** [10]
For Your Information
E-MAIL subscriptions:
Newsletter
Send the one-line message:
SUBSCRIBE IEEEVMS your-name
(e.g.: SUBSCRIBE IEEEVMS Jane Doe)
to: listserv@VTVM1.cc.vt.edu
to receive your copy automatically.
IEEEVMS_Info:
Send the one-line message:
SUBSCRIBE IEEEVMS_INFO your-name
(e.g.:SUBSCRIBE IEEEVMS_info John Doe)
to: listserv@listserv.vt.edu
For News, Bulletins, Archives, Your material.
Visit Your WEB page at:
http://fiddle.ee.vt.edu/ieeevms
Section and Chapter organization, meeting notices, reports
Past issues of the Newsletter, PACE Patter,
Access to the VMS_Info archives
Easy subscriptions to the ListServers
Links to everything in IEEE, as well as the VMS, and
Misc. Sites of technical interest.
(Submit your ideas on-line)
******************************************************** [11]
Anyone may submit material for the Newsletter
Submissions Information:
Deadline: Monday following each meeting
Submit To: Editor: (See last page)
e-mail: jhfslf@swva.net
Fax: 55 2-0261
call first:552-0052
e-mail: awang@vt.edu
Change of Address
US Mail: IEEE Service Center
445 Hoes Lane, PO Box 1331
Piscataway, NJ 08855-1331
e-mail: address.change@ieee.org
Tel: 800-678-IEEE
Fax: 908-981-9667
Web: http://www.ieee.org/i3e_mailer.html
(Changes are forwarded to the Newsletter)