Copy Link
Add to Bookmark
Report
AIList Digest Volume 5 Issue 220
AIList Digest Tuesday, 29 Sep 1987 Volume 5 : Issue 220
Today's Topics:
Seminars - Class Hierarchies with Contradictions (AT&T) &
Knowledge-Based Software Development Tools (AT&T) &
The ENGINEOUS Project at GE (NASA Ames) &
Heuristic Functions for Task Scheduling (SMU) &
Autonomous Construction Robots (Lockheed),
Conference - 1988 Workshop on Uncertainty in AI
----------------------------------------------------------------------
Date: Mon 21 Sep 1987 09:13:21
From: dlm%allegra.att.com@RELAY.CS.NET
Subject: Seminar - Class Hierarchies with Contradictions (AT&T)
Time: Thursday, September 10, 1987 1:00pm.
Place: AT&T Bell Laboratories Murray Hill 3D-473
Speaker: Alex Borgida
Affiliation: Rutgers University
Title: Of Quakers and Republicans: A Syntax, Semantics,
and Type Theory for Class Hierarchies with Contradictions
Abstract:
Disparate fields such as Artificial Intelligence, Databases and
Programming Languages have discovered the joys of object-oriented
programming. One of the principal features of this paradigm is the
presence of classes of objects organized in subclass hierarchies,
which provide a form of polymorphism and the notion of inheritance.
The arguments in favour of these mechanisms are concerned with the
ease of developing and modifying programs, but we show that in several
circumstances the same kinds of arguments can be used to undermine the
usual strict interpretation of specialization: namely that a subclass
must be in every way a subtype of its superclass(es). We therefore
propose a syntax that allows the definition of subclasses appearing to
contradict their superclasses, albeit in an explicit and controlled
way. After demonstrating the proper semantics for this construct, we
examine the difficulties of writing correct programs when statements
made about the objects in some class may be contradicted for elements
belonging to a subclass. To solve these difficulties, we propose a
type theory which admits "exceptional subclasses", and consider the
problem of reasoning with these types.
Sponsor: Ron Brachman
------------------------------
Date: Mon 21 Sep 1987 09:13:21
From: dlm%allegra.att.com@RELAY.CS.NET
Subject: Seminar - Knowledge-Based Software Development Tools (AT&T)
Time: Monday Sept 14, 1987 1:30 P.M.
Place: Murray Hill 3D-473 (Lab 1125 conf. room)
Speaker: Douglas Smith
Affiliation: Kestrel Institute
Title: Knowledge-Based Software Development Tools
Abstract: Current research on knowledge-based software development
tools at Kestrel Institute is briefly surveyed. We then focus on
systems for automatically performing algorithm design, deductive
inference, finite differencing, and data structure selection. A
detailed case study shows how these systems could cooperate in
supporting the transformation of a formal specification of a
scheduling problem into efficient, executable code.
Sponsor: Van Kelly
------------------------------
Date: Fri, 25 Sep 87 13:33:41 PDT
From: JARED%PLU@ames-io.ARPA
Subject: Seminar - The ENGINEOUS Project at GE (NASA Ames)
NASA, Ames Research Center
Intelligent Systems Forum
Siu Shing Tong
General Electric
The ENGINEOUS Project at General Electric
Abstract:
The ENGINEOUS project was established in 1985 to address the problem of
designing complex hardware that utilizes a massive number of parameters
and analysis tools. For example, to design an aircraft engine,
approximately 10,000 Fortran programs may be iteratively applied from
conceptual design to final production. The number of relevant parameters
defining a typical engine is estimated to be 300,000. The human
intervention currently required to iterate these programs and parameters,
particularly between programs and disciplines, contributes significantly
to the 7 to 10 years lead time for the development of a new engine.
An experimental system to aid engine designers has been developed and is
being tested. ENGINEOUS makes use of artificial intelligent techniques
(i.e., object oriented programming, knowledge based systems, rapid
prototyping, etc.) to address problems too complex to be effectively
handled by conventional programming techniques. This presentation will
discuss the current status, initial user's experience, and the current
development effort to map ENGINEOUS into a heterogeneous,
distributed/parallel processing environment.
Date: Thursday, October 1, 1987
Time: 1:30PM
Location: Bldg. 258, rm. 127, the auditorium
Inquires: Alison Andrews, (415) 694-6741, andrews%ear@ames-io.ARPA, or
David Jared, (415) 694-6525, jared%plu@ames-io.ARPA
VISITORS ARE WELCOME: Register and obtain vehicle pass at Ames Visitor
Reception Building (N-253) or the Security Station near Gate 18. Do not
use the Navy Main Gate.
Non-citizens (except Permanent Residents) must have prior approval from the
Director's Office one week in advance. Submit requests to the point of
contact indicated above. Non-citizens must register at the Visitor
Reception Building. Permanent Residents are required to show Alien
Registration Card at the time of registration.
------------------------------
Date: Sat, 26 Sep 1987 22:55 CST
From: Leff (Southern Methodist University)
<E1AR0002%SMUVM1.BITNET@wiscvm.wisc.edu>
Subject: Seminar - Heuristic Functions for Task Scheduling (SMU)
Three Practical Heuristic Functions for Task Scheduling:
Descriptions and Analyses
SPEAKER: Mingfang Wang (mwang%smu@csnet-relay) LOCATION: 315 SIC
Southern Methodist University TIME: 1:30 pm
ABSTRACT
Generally, task scheduling in a multi-processing environment is an
NP-hard problem. Here, three task scheduling algorithms using different
heuristic functions are presented and analyzed. These algorithms fall into
the category of \fIpriority list\fR methods. The algorithms are analyzed
both analytically and through simulations. The trade-off is between the time
complexity of the task scheduling and the optimalily of the schedule. A fast
algorithm can have a time complexity of O(n * log n),
but the task schedule produced by the algorithm is not as good as those with
time complexity of O(n^2).
------------------------------
Date: Mon, 28 Sep 87 13:28 CDT
From: SULLIVAN%lockheed.com@RELAY.CS.NET
Subject: Seminar - Autonomous Construction Robots (Lockheed)
FROM: JOSEPH W. SULLIVAN O/90-06 B/259 (415)354-5213
SUBJECT: AIC COLLOQUIUM
The Lockheed AI Center is pleased to announce a presentation by
Dr. Michael R. Genesereth of the Logic Group at Stanford
University. An abstract of the presentation is provided below.
Proposal for Ten Years of Research on
Autonomous Construction Robots
Michael R. Genesereth, Ph.D.
DATE: 14 October 1987
TIME: 3:30
PLACE: Lockheed Artificial Intelligence Center
Main Conference Room
2710 Sand Hill Rd. (Lockheed Bld. #259)
Menlo Park
One of the boldest promises of Artificial Intelligence is
the creation of an autonomous robot, one that is capable of
functioning appropriately in an arbitrary environment so as to
achieve an arbitrary goal. The environment and goal are
described in advance by the robot's client, in as much or as
little detail as he desires. Given this description, the robot
then acts autonomously, sensing and acting on its environment in
a manner appropriate to the client's goal.
Although there have been efforts in the past to build such
robots, these efforts have not met with great success due to
limitations on various technological fronts. In recent years,
however, there has been significant progress on these fronts;
and, in light of this progress, it appears likely that, with
additional research and a strong effort at integration, it should
be possible within ten years to achieve this goal.
This talk describes one particular research project aimed at
achieving this goal. The project is a collaborative venture of
the Logic Group and the Robot Reasoning Group of Stanford
University and is just getting underway.
In order to ground our research and development, we have chosen
to concentrate on autonomous robots that are experts at the
construction of electromechanical artifacts. Insofar as good
methodology involves verification of proper construction, our
robots will also need to be experts at the testing of artifacts,
the diagnosis of observed failures, and their repair.
We believe this project to be a good one for several reasons.
First of all, the robots produced are likely to be applicable to
many military and industrial applications, e.g. small-scale
manufacturing, space-station assembly, planetary exploration,
engineering behind enemy lines, and operations at radioactive and
toxic chemical sites. Secondly, we believe the project will be
beneficial for research in both Artificial Intelligence and
Robotics by forcing the integration of results from disciplines
that have over the years grown apart. Finally, we believe that
the project, given its university setting, will have educational
benefit by once again holding up for students the exciting goal
of creating autonomous robots.
------------------------------
Date: Thu, 24 Sep 87 06:58:41 PDT
From: Ross Shachter <SHACHTER@SUMEX-AIM.STANFORD.EDU>
Subject: Conference - 1988 Workshop on Uncertainty in AI
CALL FOR PARTICIPATION
Fourth Workshop on Uncertainty in Artificial Intelligence
Sponsored by AAAI
St. Paul, Minnesota, August 19-21, 1988
(preceding the AAAI Conference)
This is the fourth annual AAAI Workshop on Uncertainty in AI. The
first three workshops have been successful and productive, involving
many of the top researchers in the field. The first two workshop
proceedings have been published in the North-Holland Intelligence and
Pattern Recognition series, and the third proceedings is in press.
The general subject is automated or interactive reasoning under
uncertainty.
This year's emphasis is on the control of uncertain reasoning
processes and the issues of knowledge engineering in uncertain
domains. The most effective way to make points, compare approaches,
and clarify issues in these research areas is through demonstration in
applications, so these are especially encouraged, although more
theoretical research papers are also welcome. Many of the ideas
discussed at earlier workshops have been incorporated into prototype
and production software. This year we would especially like to
encourage the demonstration of some of these systems.
The key to the success of past workshops has been the ability to
interact with leading researchers in all aspects of the field. There
will be ample opportunity for informal discussions as well as panel
discussion to focus and debate the issues. In order to maintain this
interaction, all accepted papers will appear in the proceedings and be
presented in poster sessions. This format worked well at the 1987
workshop, and participants requested that it be done again.
Papers are invited on the following topics:
* Applications: results, implementation problems and experiences,
analyses of the experiences of end users
* Knowledge engineering under uncertainty: problem structuring,
corrections for bias, consensus among experts, man-machine interface
and human-in-the-loop systems
* Control of uncertain reasoning processes
* Different uncertainty calculi: theoretical and empirical
comparisons, transformations between representations, criteria for
decision making, axiomatic frameworks
* Revision of beliefs in an uncertain environment
* Robotics: uncertainty in perception and control
* Planning: generation of feasible plans under uncertainty
* Development of standard test cases
* Other uncertainty in AI issues
Papers will be carefully reviewed. Space is limited, so prospective
attendees are urged to submit a paper with the intention of active
participation in the workshop. Preference will be given to papers
that have demonstrated their approach in real applications.
Nonetheless, the underlying methodology should be supported by solid
theory to encourage discussion on a scientific basis. Again, all
accepted papers will be included in the proceedings and presented in
poster sessions.
Four copies of a paper should be sent to the program chairman by March
31, 1988. (No extended abstracts will be accepted.) Acceptances will
be sent by May 25, 1987. Final (camera ready) papers incorporating
the reviewers' comments must be received by July 15, 1988. There is
an eight page limit on the camera-ready copy. (A few extra pages are
available for a nominal fee.) Copies of the proceedings will be
available at the workshop.
General Chair: Program Chair:
Tod Levitt Ross Shachter
Advanced Decision Systems Center for Health Policy
201 San Antonio Circle 125 Old Chemistry Building
Suite 286 Duke University
Mountain View, CA 94040 Durham, NC 27706
(415) 941-3912 (919) 684-4424, 684-3023, 942-5852
levitt@ads.arpa shachter@sumex-aim.stanford.edu
Program Committee: P. Bonissone, P. Cheeseman, L. Kanal, J. Lemmer, T.
Levitt, R. Patil, J. Pearl, E. Ruspini, R. Shachter, G. Shafer
------------------------------
End of AIList Digest
********************
