Copy Link
Add to Bookmark
Report
AIList Digest Volume 5 Issue 064
AIList Digest Thursday, 5 Mar 1987 Volume 5 : Issue 64
Today's Topics:
Seminars - A Stochastic Genetic Search Method (CMU) &
Hypothesis Formation (Rutgers) &
Creative Analogies in Scientific Progress (UPenn) &
On Visual Formalisms (CMU),
Conference - Workshop on Coupling Symbolic and Numeric Computing &
AAAI Workshop on Planning for Autonomous Mobile Robots &
Computing and Society in Seattle, Preceding AAAI &
HICSS-21 Call For Papers
----------------------------------------------------------------------
Date: Sun 1 Mar 87 17:49:12-EST
From: Dave Ackley <David.Ackley@C.CS.CMU.EDU>
Subject: Seminar - A Stochastic Genetic Search Method (CMU)
David H. Ackley
Carnegie Mellon Computer Science doctoral dissertation defense
Tuesday, February 24, 1987 at 1pm
Wean Hall 5409
"Stochastic iterated genetic hillclimbing"
Abstract
In the "black box function optimization" problem, a search strategy is
required to find an extremal point of a function without knowing the
structure of the function or the range of possible function values.
Solving such problems efficiently requires two abilities. On the one
hand, a strategy must be capable of "learning while searching": It must
gather global information about the space and concentrate the search in
the most promising regions. On the other hand, a strategy must be
capable of "sustained exploration": If a search of the most promising
region does not uncover a satisfactory point, the strategy must redirect
its efforts into other regions of the space.
This dissertation describes a connectionist learning machine that
produces a search strategy called "stochastic iterated genetic
hillclimbing" (SIGH). Viewed over a short period of time, SIGH displays
a coarse-to-fine searching strategy, like simulated annealing and
genetic algorithms. However, in SIGH the convergence process is
reversible. The connectionist implementation makes it possible to
"diverge" the search after it has converged, and to recover
coarse-grained information about the space that was suppressed during
convergence. The successful optimization of a complex function by SIGH
usually involves a series of such converge/diverge cycles.
SIGH can be viewed as a generalization of a genetic algorithm and a
stochastic hillclimbing algorithm, in which genetic search discovers
starting points for subsequent hillclimbing, and hillclimbing biases the
population for subsequent genetic search. Several search
strategies---including SIGH, hillclimbers, genetic algorithms, and
simulated annealing---are tested on a set of illustrative functions and
on a series of graph partitioning problems. SIGH is competitive with
genetic algorithms and simulated annealing in most cases, and markedly
superior in a function where the uphill directions usually lead \away/
from the global maximum. In that case, SIGH's ability to pass
information from one coarse-to-fine search to the next is crucial.
Combinations of genetic and hillclimbing techniques can offer dramatic
performance improvements over either technique alone.
------------------------------
Date: Mon, 2 Mar 87 13:04 EST
From: FAWCETT@RED.RUTGERS.EDU
Subject: Seminar - Hypothesis Formation (Rutgers)
On Thursday, March 19th at 10:30 AM, Prof. Lindley Darden from the
University of Maryland will speak on her work on hypothesis formation. The
room will be announced shortly. An abstract and a summary of her interests
follow.
"Hypothesis Formation Using Part-Whole Interrelations"
Lindley Darden
This paper discusses an implementation, called SUTTON, of
strategies for rediscovering the chromosome theory of heredity.
Walter Sutton formulated the theory in the early 20th century, by
postulating interrelations between the fields of cytology and
genetics. Knowledge from these fields during that period is
represented in a frame-based system, and rules for using
knowledge from one field to guide hypothesis formation in
the other are implemented in LISP. In particular, the discovery
that the gene is part of the chromosome is simulated, and general
rules for part-whole reasoning are investigated, including rules
for inheritance and propagation of causal relations in part-whole
hierarchies.
Keywords: Hypothesis formation, scientific discovery, learning,
identity relation, part-whole relation, causality.
Lindley Darden is an Associate Professor in the Departments of
Philosophy and History and a member of the graduate faculty in
the Committee on the History and Philosophy of Science at the
University of Maryland, College Park. She is currently serving
in the second year of a half-time research appointment in the
University of Maryland Institute for Advanced Computer Studies.
This work was done in collaboration with Roy Rada of the National
Library of Medicine. Her research interests include reasoning in
scientific discovery (including analogical reasoning and
formation of abstract theory types) and knowledge representation
techniques for biological knowledge. Her address is Department
of Philosophy, University of Maryland, College Park, Maryland
20742 and darden@mimsy.umd.edu.
(In addition, Prof. Darden gave an invited talk at last summer's AAAI
entitled "Viewing History of Science as Compiled Hindsight".)
------------------------------
Date: Tue, 3 Mar 87 19:11 EST
From: Tim Finin <Tim@cis.upenn.edu>
Subject: Seminar - Creative Analogies in Scientific Progress (UPenn)
SPECIAL JOINT COLLOQUIUM
Computer Science, Psychology and Physics
University of Pennsylvania
THE ROLE OF CREATIVE ANALOGIES IN SCIENTIFIC PROGRESS: COMPUTER MODELING
Professor Douglas R. Hofstadter, University of Michigan
2:30 p.m. Wednesday, March 4, 1987
Tea served at 2:00 in the Faculty Lounge (2E17)
David Rittenhouse Lab - Auditorium A1
The Copycat project is a computer model of analogical thought processes,
particularly ones in which a creative or daring leap is made of the sort that
when done in science often postulates new theoretical constructs or objects
(genes, particles, etc.). Examples of such analogies in science will be
presented and the copycat model will be discussed.
------------------------------
Date: 3 Mar 87 10:23:53 EST
From: Theona.Stefanis@g.cs.cmu.edu
Subject: Seminar - On Visual Formalisms (CMU)
PS SEMINAR
MONDAY, 9 March
WeH 5409
3:30
On Visual Formalisms
David Harel
Weizmann Inst., Rehovot, Israel
(At CMU for the year)
A general mathematical object of diagrammatic nature, the higraph, is
presented. Higraphs borrow and extend ideas from Venn-diagrams, graphs and
hypergraphs. They constitute a visual formalism for describing various
kinds of complex entities, particularly those that involve many sets of
objects having intricate structural (i.e., set-theoretic) interrelationships
as well as additional relations af dynamic, causal or other nature.
Higraphs appear to have many applications, as well as a rich theory that
awaits further research. We shall exhibit a number of applications in
database theory (entity-relationship diagrams), artificial intelligence
(semantic and associative nets) and concurrent reactive systems (statecharts).
Statecharts constitute a natural extension of conventional state-transition
diagrams in ways that make them appropriate for describing large real-world
systems, and they will be described in the talk in some detail.
__________________________________________
------------------------------
Date: 4 Mar 87 02:29:21 GMT
From: ssc-vax!bcsaic!tedk@BEAVER.CS.WASHINGTON.EDU (Ted Kitzmiller)
Subject: Conference - Workshop on Coupling Symbolic and Numeric
Computing
Due to electronic mail system problems and other manifestations of Murphy's
law, the deadline for paper submittals to the workshop on coupling symbolic
and numeric computing (see AAAI magazine Winter issue) has been extended
until late March.
If you had previously sent me an electronic mail message about the workshop
and have not received a response, please resend your message. It appears
that in many instances in which I responded to queries about the workshop
via the network, the responses were not successfully delivered. Unfortunately,
in these instances no evidence of a problem was indicated.
Please contact me at the e-mail address, telephone, or mail address below
(if you have not done so within the last week) if you are interested.
Please include both your phone number and US mail address along with
an explicit e-mail incantation to your site.
Ted Kitzmiller
Boeing Advanced Technology Center
US Mail: MS 7L-64 / PO Box 24346 / Seattle / Washington / 98124-0346
Parcel Post: MS 7L-64 / 2760 160th Avenue SE / Bellevue / Washington / 98008
Phone: (206) 865-3227 E-mail: tedk@boeing.com
------------------------------
Date: Thu, 26-FEB-1987 15:45 EST
From: MILLER%VTCS1.BITNET@wiscvm.wisc.edu
Subject: Conference - AAAI Workshop on PLANNING FOR AUTONOMOUS MOBILE
ROBOTS
Call for Participation and abstracts:
Workshop on PLANNING FOR AUTONOMOUS MOBILE ROBOTS
July 16, 1987, The University of Washington,
Seattle, WA
Sponsored by AAAI
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Most mobile robot projects have concentrated on robots with specific
missions (e.g., complete errands one and two, follow this road for three
miles). Yet a truly autonomous robot would have its mission described
at a much higher level. Its programming would have to derive specific
tasks to be accomplished based on unpredictable (and perhaps not even
previously classifiable) conditions in its environment. This opens new
issues for the type of planning system necessary to guiding autonomous
robots.
The purpose of this workshop would be to discuss the planning and
knowledge requirements of an autonomous exploratory robot such as a Mars
Rover. How would such a robot decide on a course? What kind of risk
assessment is necessary before deciding to make a dangerous observation?
What types of knowledge are necessary for recognizing something as being
interesting, or dangerous? What role will physical knowledge play in
safe navigation? Is either incremental or opportunistic planning
necessary for dealing with a dynamic world? What kind of demands would
the planning system place on the sensory system?
Among the topics of interest are:
*Spatial Representation *Map Building
*Planning Under Uncertainty *Risk Analysis
*Planning in Dynamic Domains *Physical Reasoning
*Spatial and Temporal Reasoning *Sensor Coordination
*Experience-Based Planning *Route Planning
Those interested in participating in the workshop should submit a short
abstract (no more than two pages) of your work you would wish to
present. Mail two copies of your abstract (hard copy only) before April
15, 1987, to either of the workshop organizers. Invitations for
workshop participation will be sent out by May 15, 1987.
David Miller David Atkinson
562 McBryde Mail Stop 510-202
Department of Computer Science Jet Propulsion Laboratory
Virginia Tech Cal Tech
Blacksburg, VA 24061 4800 Oak Grove Drive
Pasadena, CA, 91109
(703) 961-5605 (818) 577-6603
miller%vtcs1@bitnet-relay.arpa atkinson@usc-ecl.arpa
------------------------------
Date: Tue, 03 Mar 87 08:59:31 PST
From: jon@june.cs.washington.edu (Jon Jacky)
Subject: Conference - Computing and Society in Seattle, Preceding AAAI
(This was sent around in early December - due date 4/1 now approaching)
Call for Papers
DIRECTIONS AND IMPLICATIONS OF ADVANCED COMPUTING
Seattle, Washington July 12, 1987
The adoption of current computing technology, and of technologies that
seem likely to emerge in the near future, will have a significant impact
on the military, on financial affairs, on privacy and civil liberty, on
the medical and educational professions, and on commerce and business.
The aim of the symposium is to consider these influences in a social and
political context as well as a technical one. The social implications of
current computing technology, particularly in artificial intelligence, are
such that attempts to separate science and policy are unrealistic. We
therefore solicit papers that directly address the wide range of ethical
and moral questions that lie at the junction of science and policy.
Within this broad context, we request papers that address the following
particular topics. The scope of the topics includes, but is not limited
to, the sub-topics listed.
RESEARCH FUNDING DEFENSE APPLICATIONS
- Sources of Research Funding - Machine Autonomy and the Conduct of War
- Effects of Research Funding - Practical Limits to the Automation of War
- Funding Alternatives - Can An Automated Defense System Make War
Obsolete?
COMPUTING IN A DEMOCRATIC SOCIETY COMPUTERS IN THE PUBLIC INTEREST
- Community Access - Computing Access for Handicapped People
- Computerized Voting - Resource Modeling
- Civil Liberties - Arbitration and Conflict Resolution
- Risks of the New Technology - Educational, Medical and Legal Software
- Computing and the Future of Work
Submissions will be read by members of the program committee, with the
assistance of outside referees. The program committee includes Andrew
Black (U. WA), Alan Borning (U. WA), Jonathan Jacky (U. WA), Nancy Leveson
(UCI), Abbe Mowshowitz (CCNY), Herb Simon (CMU) and Terry Winograd
(Stanford).
Complete papers, not exceeding 6000 words, should include an abstract,
and a heading indicating to which topic it relates. Papers related to
AI and/or in-progress work will be favored. Submissions will be judged
on clarity, insight, significance, and originality. Papers (3 copies)
are due by April 1, 1987. Notices of acceptance or rejection will be
mailed by May 1, 1987. Camera ready copy will be due by June 1, 1987.
Proceedings will be distributed at the Symposium, and will be on sale
during the 1987 AAAI conference.
For further information contact Jonathan Jacky (206-548-4117) or Doug
Schuler (206-783-0145).
Sponsored by Computer Professionals for Social Responsibility
P.O. Box 85481
Seattle, WA 98105
------------------------------
Date: Tue 3 Mar 87 16:30:34-EST
From: Gail E. Kaiser <KAISER@CS.COLUMBIA.EDU>
Subject: Conference - HICSS-21 Call For Papers
CALL FOR PAPERS
21ST ANNUAL
HAWAII INTERNATIONAL CONFERENCE ON SYSTEM SCIENCES
(HICSS-21)
Papers are invited for the session(s) on Use of AI Techniques in Software
Design and Implementation in the software track of the 21st annual Hawaii
International Conference on System Sciences (HICSS-21), to be held in Kona,
Hawaii next January 5-8, 1988.
Topics of interest include, but are not limited to, the following artificial
intelligence areas as they apply to software design and implementation,
particularly for large-scale software systems. Techniques may apply to any or
all phases of the software development process: project management,
requirements, functional specification, design specification, modular
decomposition, coding, integration, testing, maintenance, documentation,
delivery, etc. Example applications are given in parentheses.
- Automatic deduction (detecting inconsistencies among programmers'
assumptions, automatic programming)
- Knowledge representation (semantic nets, frames, etc. for
representing programming information)
- Learning (self-tuning of software tools to specific programs,
generalization of program fragments to support reusability)
- Natural language (matching functionality of program parts with the
corresponding program documentation, explaining program components
and their interactions to new project member)
- Planning (detecting interactions among planned changes)
- Rule-based systems (program transformation, performance tuning)
- Search (retrieval of reusable program fragments)
Six copies of the full paper (maximum 20 double-spaced pages) should be sent
to the session chairman at the address given below. Papers must arrive by July
1, 1987. Authors will be notified of acceptance by September 7, 1987.
Camera-ready copies will be due by October 19, 1987.
Session chairman: Prof. Gail E. Kaiser, Columbia University, Department of
Computer Science, New York, NY 10027. Phone: 212-280-3856. Electronic mail:
kaiser@cs.columbia.edu, ...!columbia!cs!kaiser
Software track chairman: Dr. Bruce D. Shriver, IBM T.J. Watson Research
Center, P.O. Box 704, Yorktown Heights, NY 10598. Phone: 914-789-7626.
Electronic mail: shriver@ibm.com
------------------------------
End of AIList Digest
********************
