Copy Link
Add to Bookmark
Report
Machine Learning List Vol. 5 No. 08
Machine Learning List: Vol. 5 No. 8
Tuesday, April 13, 1993
Contents:
ML93 Workshop on Speedup Learning
Evolutionary Programming
AAAI-93/IAAI-93 Conference Registration Brochure
The Machine Learning List is moderated. Contributions should be relevant to
the scientific study of machine learning. Mail contributions to ml@ics.uci.edu.
Mail requests to be added or deleted to ml-request@ics.uci.edu. Back issues
may be FTP'd from ics.uci.edu in pub/ml-list/V<X>/<N> or N.Z where X and N are
the volume and number of the issue; ID: anonymous PASSWORD: <your mail address>
----------------------------------------------------------------------
From: Prasad Tadepalli <tadepall@chert.CS.ORST.EDU>
Date: Fri, 9 Apr 93 09:32:09 PDT
Subject: ML93 Workshop on Speedup Learning
CALL FOR PARTICIPATION
1-day Workshop on Knowledge Compilation and Speedup Learning
To be held at ML-93
June 30 1993 Amherst, Massachusetts
Description of Workshop
This workshop is a sequel to two successful workshops: the knowledge
compilation workshop which was organized by Jim Bennett, Tom
Dietterich, and Jack Mostow in 1986 and the knowledge compilation and
speedup learning workshop held at Aberdeen, Scotland in July 1992
organized by us.
Knowledge compilation is the problem of converting a declarative
specification or a domain theory to an efficient executable program.
Speedup learning is the problem of improving, or speeding up, a slow
problem solver from experience. These two tasks are closely related
since a declarative specification or a domain theory can be viewed as
a slow problem solver. Several approaches to compilation and speedup
have been explored in the literature: explanation-based learning,
empirical learning, and partial evaluation, among others.
There have been a number of results in speedup learning over the past
few years. While there were some positive results in explanation-based
learning (EBL) (Mitchell,DeJong,Minton,Shavlik), the utility problem,
or the exorbitant computational cost of using the learned knowledge,
proved to be a significant obstacle to further progress (Minton,
Rosenbloom and Tambe). At the 1992 workshop, a new formulation of the
compilation and speedup learning problem in a uniform framework
emerged: the inductive derivation of properties of a problem
distribution and its use in dynamic optimization of a theory. This
relates work in the speedup learning and knowledge compilation area to
statistical learning techniques (how to acquire those properties of a
problem distribution that have efficiency implications), PAC learning
(how to evaluate candidate improvements in a guaranteed manner),
concept learning (how to learn distribution properties from examples),
partial evaluation, logic programming and deductive databases (how to
optimize logical theories using knowledge of problem distribution).
This workshop aims to bring together researchers in machine learning
as well as related communities in theory optimization and refinement
with the goal of improving our understanding of knowledge compilation
and speedup learning. One of our major goals is to consolidate results
and to collect a set of challenge real-world applications for the
community for the next few years. We especially welcome position
papers in this area to critique existing work and to define the agenda
for the next few years.
Specifically, we are interested in the following questions:
-- applications of speedup learning and compilation in real-world tasks.
-- what method classes have been developed for compilation/speedup?
-- how does learning protocol affect the methods?
-- what information about the problem solver and problem distribution
do we need for compiling knowledge or to speed up a problem solver?
-- how does choice of vocabulary influence compilation/speedup learning?
-- what are the conditions under which successful speedup learning can occur?
-- how can we minimize the number of examples needed for speedup learning
algorithms?
-- how can we learn distribution characteristics that are relevant to
speedup?
-- 10 challenge problems for the next round of progress in speedup learning.
This is not an exhaustive list but a sample of questions that we are
interested in. We specifically want to emphasize informal discussions
and exchange of ideas, rather than polished research results. In this
spirit, we welcome position papers on the research directions as well
as rational reconstructions of previous work. Especially encouraged
are papers that propose realistic applications of knowledge
compilation and speedup learning in planning, scheduling,
design/manufacturing and other real-world tasks.
If you wish to present a paper at the workshop, please email an
extended abstract (of a maximum of 1000 words) to
devika@cs.cornell.edu by April 30, 1993. The file should be in ascii
or postscript only. Electronic abstracts are preferred, but if email
is impossible, then please send 4 copies of the abstract to:
Devika Subramanian
Department of Computer Science
5133 Upson Hall
Cornell University
Ithaca, NY 14853
If you do not wish to present a paper but want to attend the workshop,
please send a one page summary of your relevant research and
publications to the same address by the above date.
Important Dates:
April 30 - Abstracts and research descriptions due
May 14 - Acceptance notification
June 1 - Final version of papers due
Organizing Committee:
Doug Fisher, Vanderbilt University
Devika Subramanian, Cornell University
Prasad Tadepalli, Oregon State University.
Program Committee:
Gerry DeJong, University of Illinois at Urbana Champaign.
Doug Fisher, Vanderbilt University
Nick Flann, Utah State University
Steve Minton, NASA Ames Research Center
Tom Mitchell, Carnegie-Mellon University
Paul Rosenbloom, University of Southern California
Devika Subramanian, Cornell University
Prasad Tadepalli, Oregon State University
Frank van Harmelen, University of Amsterdam
------------------------------
Date: Mon, 12 Apr 93 11:59:41 -0400
From: Peter J Angeline <pja@cis.ohio-state.EDU>
Subject: Call for Papers (EP94)
The Third Annual Conference on Evolutionary Programming
CALL FOR PAPERS
February 24-25, 1994
San Diego, California
Evolutionary programming is a stochastic optimization technique that can be
used to address various optimization problems. Papers regarding the theory and
application of evolutionary programming to complex problem solving and
solicited. Topics include, but are not limited to:
automatic control neural network training and design
system identification adaptive representation
forecasting robotics
combinatorial optimization pattern recognition
and the relationship between evolutionary programming and other optimization
methods. On or before June 30, 1993, prospective authors should submit a
100-250 word abstract and three page extended summary of the proposed paper to
the Technical Program Chairman:
Lawrence J. Fogel
ORINCON Corporation
9363 Towne Centre Dr.
San Diego, CA 92121
Authors will be notified of the program committee's decision on or before
September 30, 1993. Completed papers will be due January 15, 1994. Paper
format, page requirements and registration information will be detailed upon
acceptance.
General Chairman: Anthony V. Sebald, UC San Diego
Technical Chairman: Lawrence J. Fogel, ORINCON Corporation
Program Committee:
Peter Angeline, The Ohio State Univ. Gary B. Fogel, UC Los Angeles
Wirt Atmar, AICS Research Inc. Roman Galar, Tech. Univ. Wroclaw
Thomas Back, Univ. Dortmund Douglas Hoskins, The Boeing Company
George Burgin, Titam Systems/Linkabit Gerald Joyce, Scripps Clin./Res. Found.
Michael Conrad, Wayne State Univ. John McDonnell, NCCOSC
David B. Fogel, ORINCON COrporation Stuart Rubin, NCCOSC
Hans-Paul Schwefel, Univ. Dortmund
Finance Chairman:
Bill Porto, ORINCON Corporation
Publicity Co-Chairs:
Ward Page, NCCOSC
Patrick Simpson, ORINCON Corporation
Sponsored by the
Evolutionary Programming Society
In Cooporation with the
IEEE Neural Networks Council
------------------------------
Date: Fri, 9 Apr 93 07:39:06 PDT
From: Rick Skalsky <skalsky@aaai.ORG>
Subject: AAAI-93/IAAI-93 Conference Registration Brochure
AAAI93
ELEVENTH NATIONAL CONFERENCE ON
ARTIFICIAL INTELLIGENCE
IAAI-93
FIFTH INNOVATIVE APPLICATIONS OF
ARTIFICIAL INTELLIGENCE CONFERENCE
July 11-15, 1993
Washington Convention Center, Washington, DC
REGISTRATION BROCHURE
Please join us at AAAI-93
Each year the National Conference on Artificial Intelligence (NCAI)
is the primary large scale forum where the highest quality new research
in artificial intelligence (AI) is presented and discussed. Quality is
maintained by a highly competitive review and selection process in which
fewer than one of every four submitted papers is accepted.
Papers were solicited for this year's conference that describe
significant contributions to all aspects of AI, including the principles
underlying cognition, perception, and action in humans and machines; the
design, application, and evaluation of AI algorithms and intelligent
systems; and the analysis of tasks and domains in which intelligent
systems perform. In recognition of the wide range of methodologies and
research activities legitimately associated with AI, the conference
program includes papers describing both experimental and theoretical
results from all stages of AI research. This year we particularly
encouraged submission of papers that present promising new research
directions by describing innovative concepts, techniques, perspectives,
or observations that are not yet supported by mature results. To be
accepted to the conference, such submissions were required to include
substantial analysis of the ideas, the technology needed to realize them,
and their potential impact.
Because of the essential interdisciplinary nature of AI and the need
to maintain effective communication across sub-specialties, authors were
encouraged to position and motivate their work in the larger context of
the general AI community. While papers concerned with applications of AI
were invited, most such papers can be found in the program of the
Innovative Applications of AI Conference, which is collocated with AAAI-
93.
Session chairs will assure that time remains for questions after every
paper presentation. We invite you to take advantage of the opportunity to
ask incisive questions. The resulting dialog can vitalize a session and be a
catalyst for new insights.
In addition to the refereed papers, the program includes a set
of invited presentations by leaders of the AI research community and
representatives from US government agencies. These presentations
include Herb Simon's keynote address; surveys of major AI research
areas, such as a talk by Ray Reiter on nonmonotonic reasoning research;
symposia on topics of general interest to the AI community, such as a
symposia chaired by Paul Cohen on methods for evaluating AI research;
and a talk by Steve Cross on the Advanced Research Projects Agency's
strategic plan for AI.
The conference this year will include two special events focused on
the use of AI techniques to achieve effective behavior in the real world.
In particular, we will again be holding a mobile robot competition patterned
after the highly successful competition at last year's conference. In
addition, we are introducing a robot building contest in which
participants will design, build, and program small mobile robots on-site
and in real-time.
This multi-faceted conference program is designed to provide attendees
with many opportunities for stimulating and enlightening experiences.
Come join us!
--Richard Fikes & Wendy Lehnert
Cochairs, AAAI-93 Program Committee
AAAI-93 PROGRAM
July 11-15, 1993
The AAAI-93 program holds wide appeal for the varying interests of the
members of the AI community. Highlights include:
Three days of technical paper presentations by top scientists
in the field
A series of invited speakers and panels, including the opening
keynote address by Herbert Simon
Twenty four-hour tutorials that explore evolving techniques
taught by experienced scientists and practitioners in AI
(separate registration fee)
AAAI-93 / IAAI-93 Joint Exhibition, featuring exhibits and
demonstrations
AAAI Robot Exhibition and Competition, combining a live
competition of mobile robots from research labs around the
world with video presentations from several US robot
manufacturers
A series of sixteen small workshops with selected focus.
(Attendance is limited and determined prior to conference.)
AAAI-93 Keynote Address
Artificial Intelligence: An Experimental Science
by Herbert A. Simon, Carnegie Mellon University
A review of the journal Artificial Intelligence shows a rather
steady drift, in recent years, from articles describing and evaluating
specific computer programs that exhibit intelligence, to formal, theoreti-
cal articles that prove theorems about intelligence. This talk will discuss
why a large part of our understanding of intelligence--artificial as well
as natural--will continue to depend upon experimentation, and why much
theory in AI will be relatively qualitative and informal.
Herbert A. Simon's research has ranged from computer science to
psychology, administration, and economics. The thread of continuity
through all his work has been his interest in human decision-making and
problem-solving processes, and the implications of these processes for
social institutions. In the past 25 years, he has made extensive use of
the computer as a tool for both simulating human thinking and augmenting
it with AI.
Born in 1916 in Milwaukee, Wisconsin, Simon was educated in
political science at the University of Chicago (BA, 1936, Ph.D., 1943).
He has held research and faculty positions at the University of California
(Berkeley), Illinois Institute of Technology, and since 1949, Carnegie
Mellon University, where he is Richard King Mellon University
Professor of Computer Science and Psychology.
Simon's writings include Administrative Behavior, Human Problem
Solving, jointly with Allen Newell, The Sciences of the Artificial
Scientific Discovery, with Pat Langley, Gary Bradshaw, and Jan Zytkow,
and Models of My Life (autobiography).
Special Invited Talk
Tiger in a Cage: The Applications of Knowledge-based Systems
by Edward Feigenbaum, Stanford University
Some pioneers of AI dreamed of the super-intelligent computer, whose
problem solving performance would rival or exceed human performance.
Their dream has been partially realized--for narrow areas of human
endeavor--in the programs called expert systems, whose behavior is
often at world-class levels of competence. Their dream was partially
transformed by programs that give intelligent help to humans with
problems (rather than perform super-intelligently). These are called
knowledge systems.
Because knowledge is of such central importance to late twentieth
century firms and economies, these two types of knowledge-based
computer systems offer great economic and competitive leverage. The
systems offer remarkable cost savings; some dramatically "hot selling"
products; great return-on-investment; speedup of professional work by
factors of ten to several hundred; improved quality of human decision
making (often reducing errors to zero); and the preservation and
"publishing" of knowledge assets of a firm. These benefits will be made
vivid by descriptions of knowledge-based systems of prominence in
1993.
These stories of successful applications, repeated a thousand fold
around the world, show that knowledge-based technology is a tiger. Rarely
does a technology arise that offers such a wide range of important benefits
of this magnitude. Yet as the technology moved through the phase of early
adoption to general industry adoption, the response has been cautious,
slow, and "linear" (rather than exponential). The tiger is in a cage, and
we do not yet understand what the bars of the cage are made of. Are there
fundamental flaws in the technology that are somehow not evident in "best
practice" systems? Is there a specific set of technology transfer problems
that arise with knowledge systems but not with other kinds of systems? It
is important to the economy to free this competitive tiger, but to do so we
must understand its cage.
Edward Feigenbaum is Professor of Computer Science at Stanford
University and co-Scientific Director of Stanford's Heuristic
Programming Project. A Past President of AAAI, Feigenbaum serves on
the DARPA Advisory Committee for Information Science and Technology.
Dr. Feigenbaum was elected to the National Academy of Engineering, the
Productivity Hall of Fame of the Republic of Singapore, and the American
College of Medical Informatics.
Professor Feigenbaum received his BS and Ph.D. from Carnegie Mellon
University. His writings include The Handbook of Artificial Intelligence,
coedited with Avron Barr and Paul Cohen, Computers and Thought;
Applications of Artificial Intelligence in Organic Chemistry: The DENDRAL
Program; The Fifth Generation, with Pamela McCorduck; and The Rise of
the Expert Company, with Pamela McCorduck and Penny Nii.
Please join us for IAAI-93
The Fifth Annual Conference on Innovative Applications of Artificial
Intelligence will showcase the most impressive deployed AI applications of
the past year. These applications are winners of a worldwide competition
for the best uses of AI technology to solve real-world problems. Winning
applications need to be fully deployed and achieve significant business
benefit. The organizations honored this year will include many of the most
prestigious names in the business world (AT&T, Boeing, Compaq, Ford,
IBM, Nynex) and in the government (US Air Force, Department of the
Energy, NASA). The IAAI conferences continue to demonstrate and
showcase the importance of using AI technology within critical business
functions.
Applications will be presented in talks that are accompanied by
audiovisual presentations and live demonstrations. Meet-the-author
discussions at the end of each session encourage close interaction between
presenters and other conference participants. IAAI also includes the AI
On-Line panels focusing on issues of particular interest to the business
and government communities. For the first time, IAAI-93 will also
include invited talks on the emergence of AI as a critical technology in
helping organizations cope with change and competition.
IAAI-93 sessions have been scheduled to allow participants to attend
Herb Simon's address as well as to engage in some of the other AAAI
activities, including tutorials, workshops, and the exhibition. Please join
us for a stimulating and rewarding conference!
--Philip Klahr Elizabeth Byrnes
IAAI-93 Chair IAAI-93 Cochair
IAAI-93 Program Paper Presentations
Sixteen deployed applications will be featured this year at IAAI, on topics
such as space science experimentation, material deficiency analysis, eddy
current evaluation, intelligent information repositories, outside plant
engineering, operations research, jetliner design, and producting
scheduling.
AI-on-Line
Now entering its fourth year, AI-on-Line is a series of issue-oriented
panels and talks. Short presentations by users of deployed applications
are followed by intensive, interactive discussions. Topics for 1993
include:
- Emerging Technologies
- Transitioning AI Technology into Government Information Systems
- KBS Technology for Industrial Use
- How to Market AI
AAAI-93 / IAAI-93 Joint Exhibition
The exhibit program will offer exhibits and demonstrations by the leading
suppliers of AI hardware and software as well as AI consultants and
publishers displaying the latest in AI books and periodicals. Graphic
presentations by selected exhibitors will be featured in the Applications
Pavilion. At the time of publication, 1993 exhibitors include AAAI Press,
Ablex Publishing Corporation, Addison-Wesley Publishing Company, AI
Expert Magazine, A. K. Peters Publishing, Andersen Consulting,
Cambridge University Press, Cognitive Systems Inc., Covia Technologies,
Elsevier Science Publishing Company, Exsys Inc., Franz Inc., Gensym
Corporation, Harlequin Limited, ILOG, Inc., International Association of
Knowledge Engineers, John Wiley & Sons, Inc., Kluwer Academic
Publishers, Lawrence Erlbaum Associates, Inc., Lucid, Inc., Morgan
Kaufmann Publishers, Naval Research Lab, Neuron Data, OXKO
Corporation, PC AI Magazine, Pergamon Press Inc., Production Systems
Technologies, Southwest Research Institute, Springer-Verlag, Inc. ,
Talarian Corporation, The MIT Press, Venue, Ward Systems Group, Inc.
Robot Exhibition and Competition
Inaugurated last year, this exciting event will feature a competition
among mobile robot entries from research laboratories and universities
around the world. The competition will involve three tasks that are
typical of an office environment, and that push the limits of current
mobile robotics research.
- Escape from the office! Head-to-head slalom event, out of an office
and across a finish line.
- Find and deliver. Robots search an office complex trying to find an
object and bring it back.
- Block pushing. The added challenge of manipulating the environment
to produce a prespecified pattern.
At the time of print, approximately twenty participants on the
forefront of mobile robotics research are readying their robots for the
contest. We expect an exciting event, with many teams returning from
last year, including the 1992 champion University of Michigan team, who
will defend their overall title against an array of hungry challengers.
In addition to the competition, the exhibition will include displays and
videos of the latest experimental robotics research. Highlights will
include the results of a Legolike robot building workshop conducted just
before the exhibition, and recent research on multi-robot coordination.
Robot Building Event
Lynn Andrea Stein and David Miller
Do you suffer from robot envy? Do you bore colleagues and sponsors with
dreary simulations when what you want is robots putting on a show? Are
you sick and tired of people asking if you've implemented your system on
a real robot?
Well here's your chance to change all that. Register now for the First
Annual AAAI Robot Building and Talent Event Spectacular!
In this event, participants will design, build, and program small
mobile robots on-site and in real-time. The event will begin with the
Mobile Robots I tutorial on Monday morning and culminate in a mini-
robot-contest on Thursday. At the conference, each team will receive a kit
containing sensors, motors, a simple microprocessor control board, and
LEGO(TM). Over the course of the week, teams will assemble these kits
into working examples of modern robotic technology.
Constructing a capable robot is not a trivial task. In designing this
event, we have taken several steps to ensure that there will be a high rate
of success among the participants:
- All participants will be required to attend the Mobile Robots I
tutorial. This will ensure that all participants will have a grounding in
the critical technical areas.
- Specific materials describing the equipment and some possible
techniques to be used in building the robots will be distributed to all
participants in advance of the AAAI conference. Participants should arrive
at AAAI with preliminary designs for their robots.
- The robot kits will have preassembled and tested electronics and
connectors. These steps will greatly reduce the time required for system
debugging and wiring--the two most time consuming, but least productive
parts of robot building.
- The robot structures will be assembled from LEGO-Technics(TM),
which facilitate rapid prototyping of mechanical systems.
- All participants will be pre-registered and will be sent literature
and documentation of the kit and programming environment well in
advance of the event.
- The event will be collocated with the robot competition. This will
allow participants to draw on the expertise of established roboticists.
- The event will provide teaching assistants who have extensive
experience with the laboratory kits.
Participants may pre-register in teams (of three or four people) or as
individuals; individuals will be assigned to teams by the event organizers.
Participants must also be pre-registered for the Mobile Robots I:
Instantiating Intelligent Agents Tutorial. Basic familiarity with computer
programming is assumed. However, no prior experience with mobile
robotics, hardware, or mechanical design is required or expected.
Teams may purchase their robot kits to take home.
Space is limited. To avoid disappointment, we recommend that you
register early!
1993 AAAI Tutorials
The AAAI tutorial program for 1993 features twenty four-hour tutorials
that explore evolving techniques. Each tutorial is taught by experienced
scientists and practitioners in AI. A separate registration fee applies to
each tutorial.
- AI in Customer Service and Support, Including Help Desks (SA1)
Avron Barr and Anil Rewari
- Automating the Design of Effective Graphics (MA1)
Steven Feiner, Jock Mackinlay, and Joe Marks
- Building Expert Systems in the Real World (SA2)
Tod Hayes Loofbourrow and Ed Mahler
- Business Process Re-engineering: Using AI to Change the Organization
(SP5)
Neal M. Goldsmith and Robert A. Friedenberg
- Case-Based Reasoning: Theory and Practice (MA3)
Kevin Ashley and Evangelos Simoudis
- Computational Challenges from Molecular Biology (MA4)
Peter Karp and Russ B. Altman
- Design and Implementation of an Intelligent Multimedia Tutor (SP2)
Beverly Park Woolf and Tom Murray
- Distributed Artificial Intelligence Tools (SP4)
Edmund H. Durfee and Katia P. Sycara
- Experimental Methods in AI (MP4)
Paul Cohen and Bruce Porter
- Genetic Algorithms and Genetics-Based Machine Learning (MP3)
David E. Goldberg and John R. Koza
- Intelligent Technologies in Transportation (SA5)
Lynden Tennison and Scott Smits
- Manufacturing Applications of Integrated Knowledge-Based Systems
(SP1)
Thomas S. Kaczmarek
- Mobile Robots I: Instantiating Intelligent Agents (MA2)
David P. Miller and Marc G. Slack
- Mobile Robots II: Architectures for Reaction and Deliberation (MP2)
R. James Firby and Reid G. Simmons
- Multistrategy Learning (SP3)
R. S. Michalski and G. Tecuci
- Principles of Probabilistic Diagnosis (SA4)
Max Henrion and Eric J. Horvitz
- Probabilistic Causal Modeling (MP5)
Judea Pearl
- Qualitative Reasoning for Design and Diagnosis Applications (MP1)
Robert Milne and Louise Trave-Massuyes
- Statistical Models in Natural Language Processing (MA5)
Eugene Charniak
- Symbolic and Neural Network Approaches to Machine Learning (SA3)
Haym Hirsh and Jude Shavlik
SA1 (Sunday, 9 am - 1 pm, July 11)
AI in Customer Service and Support, Including Help Desks
Avron Barr, Inference Corporation, and Anil Rewari, Digital Equipment
Corporation
This tutorial will survey the use of AI technology in customer service and
support--areas that are poised to be leading areas for revenue growth for
many companies in the 1990's. It is exciting to note that in addition to the
conventional rule-based approaches, many of the AI systems currently
fielded in customer service and support are using more complex and
powerful AI techniques. In this tutorial, we first explore AI techniques
being used in developing intelligent applications and tools, such as case-
based reasoning, semantic networks, model-based reasoning, neural nets,
fuzzy logic, natural language processing, and distributed AI. We also
describe real applications in service organizations that use such
techniques.
Second, we will focus on service and support areas where AI techniques
are being effectively used, such as knowledge-based troubleshooting
systems, intelligent information management systems, force planning and
dispatch systems, maintenance planning applications, analysis of field
feedback data, and automatic letter generation applications. Help desks and
call centers will be discussed in greater detail. We then describe and
compare some of the popular shells that are available to build service and
support applications. Finally, we look at certain current areas of AI
research such as knowledge sharing, multi-functional knowledge bases,
machine learning, and distributed AI, and argue that customer service and
support activities are good testbeds for research and applications using
these techniques.
Prerequisite Knowledge: Some familiarity with AI.
Avron Barr is Director of Marketing for Inference Corporation, a
leading supplier of advanced application development tools. Barr coedited
The Handbook of Artificial Intelligence, and was a cofounder of
Teknowledge.
Anil Rewari is a Principal Software Engineer at Digital Equipment
Corporation's AI Technology Center where he has worked on diagnostic and
advisory systems for customer service and support using a variety of
advanced AI techniques. Rewari holds a Master's degree in Computer
Science (AI) from the University of Massachusetts in Amherst.
SA2 (Sunday, 9 am - 1 pm, July 11)
Building Expert Systems in the Real World
Tod Hayes Loofbourrow, Foundation Technologies, Inc.; and Ed Mahler,
DuPont
Building Expert Systems in the Real World will give you an understanding
of how those companies that have been most successful in applying
knowledge-based systems technology have organized, performed, and
managed their activities. The tutorial will give participants a look behind
the technology at the organizational steps taken by corporate and
divisional managers, project managers, knowledge engineers, functional
specialists, and data processing professionals to successfully build
integrated knowledge-based systems, and successfully manage knowledge-
based systems projects.
You will gain an understanding of the key factors that have led
organizations to success in developing integrated knowledge-based
systems programs, and an understanding of the strategic choices facing
individuals and organizations charged with building knowledge-based
systems. You will also learn the concrete steps you can take to improve
your ability to successfully develop knowledge-based systems. The
tutorial will stress diverse corporate and government examples, and will
make use of numerous case studies.
Prerequisite Knowledge: No prerequisites are required or assumed,
although familiarity with knowledge-based systems is helpful. This
strategic and tactical focused tutorial is targeted at functional specialists
in all business disciplines and their supervision; those charged with
building and managing knowledge-based system projects; individuals
interested in shaping organizational behavior and facilitating business
process redesign; information systems professionals and their managers;
and knowledge engineers.
Tod Hayes Loofbourrow is President and CEO of Foundation
Technologies, Inc., a knowledge technology consulting firm and a teacher
of AI courses at Harvard University. He also writes the "Managing
Knowledge" column in Expert Systems magazine and is a columnist for the
Software Engineering Journal.
Ed Mahler is known for his highly successful, no-nonsense, business-
oriented approach to intelligent systems applications while Program
Manager for AI at DuPont. Ed is CEO of E. G. Mahler and Associates, Inc., a
knowledge management consulting group located in Wilmington, Delaware.
Ed received his B.S. and Ph.D. degrees in chemical engineering from the
University of Texas.
SA3 (Sunday, 9 am - 1 pm, July 11)
Symbolic and Neural Network Approaches to Machine Learning
Haym Hirsh, Rutgers University; and Jude Shavlik, University of
Wisconsin
Machine learning dates back to the beginnings of AI, but has seen its most
vibrant growth in the last ten years. Building programs that can learn has
seen success along two complementary fronts: symbolic and neural-
network approaches to inductive learning. After an initial survey of
machine learning, this tutorial will focus on symbolic and neural network
approaches to inductive learning from examples. The problem is defined
as follows: given descriptions of a set of examples each labeled as
belonging to a particular class, determine a procedure for correctly
assigning new examples to these classes. We provide an overview of both
symbolic and neural-network approaches to this problem in a single,
unified light that highlights their commonalties and relative strengths and
weaknesses. Quinlan's ID3 and Rumelhart's Backpropagation algorithms
will be described and illustrated with simple examples. The approaches
taken and the results obtained in applying these algorithms to real-world
tasks will be covered. The tutorial will also describe some recent
comparisons between ID3 and Backpropagation using real-world data sets.
Finally, we discuss the strengths and weaknesses of this form of machine
learning, and current research problems in the area.
Prerequisite Knowledge: This intermediate-level tutorial is addressed
at computer scientists with introductory textbook experience in AI.
Dr. Haym Hirsh is Assistant Professor of Computer Science at Rutgers
University. He received his Ph.D. degree in Computer Science from
Stanford University. He is the author of Incremental Version-Space
Merging: A General Framework for Concept Learning. His current
research interests include applications of machine learning in both
molecular biology and ship design, as well as computational issues for
inductive learning.
Dr. Jude Shavlik is Assistant Professor of Computer Science at the
University of Wisconsin. He received his Computer Science Ph.D. from
the University of Illinois. He is the author of Extending Explanation-
Based Learning by Generalizing the Structure of Explanations, and
coeditor of Readings in Machine Learning. His current research interests
include comparing and combining symbolic and neural network
approaches to machine learning, as well as the application of machine
learning techniques to problems in the Human Genome Project.
SA4 (Sunday, 9 am - 1 pm, July 11)
Principles of Probabilistic Diagnosis
Max Henrion, IDSR; and Eric J. Horvitz, Rockwell International
Probability and utility theory provide a set of general principles for
reasoning and decision making under uncertainty. Now, development of
practical knowledge representations and efficient software tools based on
these principles are leading to their application to knowledge-based
systems for diagnosis and decision support. This tutorial provides you
with an overview of probability-based reasoning methodologies.
In this tutorial, you will gain an intuitive feel for key principles of
probabilistic reasoning and decision analysis; and a critical appreciation
of what kinds of problem are good candidates for application of these
methods, and the critical issues for developing successful applications.
We will focus on the use of Bayesian belief networks and influence
diagrams, as effective representations for knowledge engineering, and for
probabilistic reasoning. We will discuss qualitative and quantitative
approaches, as well as how to obtain numerical probabilities from expert
judgment or data. We will review several successful probability-based
reasoning systems, fielded in real-world applications in machine
trouble-shooting and in medicine. We will also provide an overview of
current software tools. Finally, we will review ongoing research in
uncertainty in AI, describing the challenges and research opportunities
ahead in the construction and use of decision-theoretic reasoning systems.
Prerequisite Knowledge: Previous introductory college level exposure
to probability will be helpful --but is not essential.
Dr. Max Henrion is the Director of the Institute of Decision Systems
Research (IDSR), in Palo Alto, CA, President of Lumina Decision
Systems, and a Consulting Associate Professor at Stanford University. He
received his Ph.D. from Carnegie Mellon University, and was founding
president of the Association for Uncertainty in AI. He is a coauthor of
Uncertainty, and editor of two volumes of Uncertainty and AI .
Dr. Eric Horvitz is a principal investigator at the Palo Alto Laboratory
of the Rockwell International Science Center and a research affiliate at the
Medical Computer Science Group of Stanford's Knowledge Systems
Laboratory. He received his Ph.D. from Stanford University. His research
interests include the utility-directed control of inference in large
knowledge bases, flexible procedures for determining ideal actions under
varying resources, and techniques for exploiting the relationships
between deliberative and compiled reasoning.
SA5 (Sunday, 9 am - 1 pm, July 11)
Intelligent Technologies in Transportation
Lynden Tennison, Union Pacific Railroad; and Scott Smits, American
Airlines
This tutorial will cover the business areas within transportation
companies that offer significant opportunities for intelligent technologies.
The focus of the tutorial will be identifying these areas and providing a
matrixed approach to match technologies to these business areas. The
session will address technologies such as: rule based languages and
systems, operations research techniques, heuristic search techniques,
inexact reasoning, neural networks, object technologies, intelligent text
retrieval and imaging.
Prerequisite Knowledge: The recommended audience includes
transportation company automation system planners, developers and
managers. Industry consultants or product providers would also benefit
from this session.
Lynden Tennison is the Director of Distributed Computing at Union
Pacific Railroad. He joined Union Pacific from American Airlines where
he managed the knowledge systems organization. He had past assignments
with Southwestern Bell Telephone, AT&T, and LTV Aerospace.
SP1 (Sunday, 2 pm - 6 pm, July 11)
Manufacturing Applications of Integrated Knowledge-Based Systems
Thomas S. Kaczmarek, General Motors
The manufacturing organizations that survive today's intensely
competitive environment will be those that understand the importance of
knowledge. The world class manufacturers of tomorrow are taking action
today to insure that their workers are applying corporate knowledge to all
phases of their business. Integrated knowledge based systems are being
used to support corporate objectives for agility and quality. Integrated
knowledge based systems strive to take full advantage of all personal,
organizational and information technology capabilities in a balanced
approach.
Prerequisite Knowledge: This tutorial will present a survey of
opportunities and success stories for integrated knowledge based systems
in the manufacturing enterprise. The territory covered will range from
deriving customer requirements through design, engineering, tooling,
scheduling, operations, supplier interaction, logistics, sales, marketing,
customer support , and customer feedback. The tutorial will also present
lessons learned to help you take advantage of the opportunities. The
lessons learned will help you to prepare your organization to exploit
knowledge based systems and provide you with a focused approach for
introducing integrated knowledge based systems that will lead you from
simple to more complex applications.
Thomas S. Kaczmarek, Ph.D. did his graduate work at the University of
Pennsylvania in AI. He was founding manager of the Detroit based
knowledge engineering facility of Inference Corporation. While with
Inference, he was instrumental in initiating and managing many
applications developed for Ford Motor Company. Three of these
applications have been presented at the Innovative Applications of
Artificial Intelligence. Dr. Kaczmarek is now at GM where he serves as a
senior staff member responsible for knowledge based architecture and the
integration of knowledge based technology with CAD, CAM, CAE, and CIM
systems.
SP2 (Sunday, 2 pm - 6 pm, July 11)
Design and Implementation of an Intelligent Multimedia Tutor
Beverly Park Woolf, University of Massachusetts; and Tom Murray, Asea
Brown Boveri Inc.
Recent success in achieving efficient and reliable intelligent tutors
demonstrates that these systems can be twice as effective as classroom
teaching and can teach in 1/3 the time. This tutorial offers you techniques
for building flexible and customizable intelligent tutors and for modifying
existing training systems towards increased flexibility and intelligence.
We focus on building tutoring systems and developing interface tools for
increasing the knowledge of tutors.
We will demonstrate working intelligent tutors, including systems
that tutor about complex industrial boilers, concepts of high-school
statics and electricity, and advanced cases of cardiac failure. We will
describe noteworthy systems in the military, industry and education
along with an overview of the capabilities and limitations of current
intelligent tutors.
A modular approach will be discussed for making traditional
computer-based instructional systems more intelligent by selectively
adding simple yet powerful features. High-level design specifications will
be proposed along with a structure for supporting the organization of
domain and pedagogical knowledge. Practical tools and shells will be
demonstrated to enable teachers and trainers to transfer their knowledge
to a tutor. For each tutor component we will discuss content covered;
alternative representations; control structures for perusing the content
knowledge; and possible knowledge acquisition tools.
Prerequisite Knowledge: This tutorial is intended for those who will
participate in developing tutoring, training and advisory systems, as well
as those who would like a basic understanding of the technology. Minimal
familiarity with AI and some computing experience is helpful, but not
prerequisite.
Beverly Park Woolf is a research scientist at the University of
Massachusetts. She has a Ph.D. in Computer Science and an Ed. D. in
Education, both from the University of Massachusetts and has more than
15 years experience in educational computer science research,
production of intelligent tutoring systems and development of multimedia
systems.
Tom Murray has a Ph.D. in Education. He is affiliated with the Advanced
Computation Systems Group of Asea Brown Boveri Inc. His interests
include knowledge acquisition of tutoring (including building tools) and
domain knowledge.
SP3 (Sunday, 2 pm - 6 pm, July 11)
Multistrategy Learning
R. S. Michalski and G. Tecuci, George Mason University
Multistrategy learning is concerned with developing systems that
integrate two or more inferential strategies or computational
mechanisms. One of the central and the fastest growing new directions in
machine learning, multistrategy learning systems integrate empirical
induction with explanation-based learning, deduction with abduction and
analogy, quantitative and qualitative discovery, symbolic and neural net
learning, symbolic and genetic algorithm-based learning. Because
integrated methods are complementary, multistrategy learning systems
sometimes apply to a wider range of practical problems than the more
traditional monostrategy systems.
In this tutorial we present an overview of methods, systems and
applications of multistrategy learning. First, we will review basic issues
in machine learning, analyze learning strategies, and compare existing
methodologies. Advantages, limitations and appropriate application areas
for monostrategy learning methods will be summarized. Next, we present
theoretical foundations, methods and representative systems for
multistrategy learning, followed by applications of multistrategy
systems, in particular, to areas such as automated knowledge acquisition,
knowledge discovery in databases, planning, robotics, engineering design,
technical and medical decision making, and computer vision. Finally, we
will discuss major current and prospective research directions in
multistrategy learning. The materials for the tutorial will include an
extensive bibliography of this field.
Prerequisite Knowledge: This tutorial is designed for those who are
interested in building advanced learning systems and applying them to
various practical domains. Basic introductory-course AI knowledge is
highly desirable.
Ryszard S. Michalski is PRC Chaired Professor of Computer Science
and Systems Engineering, and Director of the Center for Artificial
Intelligence at George Mason University. He cofounded the field of machine
learning, co-edited Machine Learning , and co-founded Machine Learning
journal.
G. Tecuci is Associate Professor of Computer Science at George Mason
University. He has published over fifty papers in machine learning,
knowledge acquisition, AI, advanced robotics, graph theory and compilers.
SP4 (Sunday, 2 pm - 6 pm, July 11)
Distributed Artificial Intelligence Tools
Edmund H. Durfee, University of Michigan; and Katia P. Sycara, Carnegie
Mellon University
This tutorial will thoroughly survey problems, techniques and
applications in contemporary DAI, in preparation for building DAI
systems or as background for doing research to advance the state of DAI
practice. Particular emphasis will be given to DAI tools and to
methodologies for building and evaluating DAI systems. Real-world DAI
applications, such as heterogeneous intelligent information systems and
coordinated intelligent vehicles, are only beginning to emerge. We hope,
through this tutorial, to speed this process by introducing DAI algorithms
and paradigms to attendees in practical, rather than only theoretical,
terms.
Prerequisite Knowledge: The tutorial presumes knowledge of AI at the
level of an introductory course, and familiarity with such general
concepts as object-oriented systems, planning, heuristic search,
knowledge-based systems, reasoning under uncertainty, and so on.
Edmund H. Durfee received his Ph.D. degree in computer and
information science from the University of Massachusetts, Amherst. He is
currently an Assistant Professor in the Dept. of Electrical Engineering
and Computer Science at the University of Michigan, where his interests
are in distributed AI, planning, blackboard systems, and real-time
problem solving. He is the author of Coordination of Distributed Problem
Solvers, and is also a 1991 recipient of a Presidential Young Investigator
award from the National Science Foundation.
Katia P. Sycara is a Research Scientist in the School of Computer
Science at Carnegie Mellon University. She is also the Director of the
Laboratory for Enterprise Integration, where she is conducting research
in investigating and integrating decision making across the manufacturing
product life cycle. She received her Ph.D. in Computer Science from
Georgia Institute of Technology, and is area editor for Group Decision and
Negotiation journal.
SP5 (Sunday, 2 pm - 6 pm, July 11)
Business Process Re-engineering: Using AI to Change the Organization
Neal M. Goldsmith, Business Technology; and Robert A. Friedenberg,
Inference Corporation
There are three central issues in planning for and managing advanced
information systems in business: re-designing business processes (also
known as business process re-engineering or BPR) through the
application of advanced information systems; aligning systems strategy
with corporate priorities; and increasing return on investments in
emerging information systems through effective technology transfer and
change management. If those three things are done well, the battle has
been won; if they aren't, it doesn't matter how well anything else is done,
we might as well save our money and go home. This tutorial will assist AI
professionals in responding to these three crucial challenges.
This tutorial will begin with an overview of basic concepts of strategic
alignment, BPR and change management. Following the overview session,
case examples will be provided of the techniques used, in some cases
literally to turn companies around through re-engineering with
knowledge-based tools. After the case examples, attendees will be invited
to describe their own experiences and to begin to apply the concepts and
cases directly.
Prerequisite Knowledge: This is an introductory tutorial. No formal
business background is required.
Neal M. Goldsmith, Ph.D. is publisher of Business Technology, a
periodical for CIOs that focuses on tools and techniques in business
process re-engineering, strategic alignment, ROI and change management.
He is also President of Tribeca Research, a technology management
consulting firm.
Robert A. Friedenberg, Ph.D. is Vice President of Inference
Corporation's Business Process Re-Engineering Consulting Group. Prior
to joining Inference Corporation, he was responsible for Coopers and
Lybrand's Financial Services and Retail Industry Decision Support
practices, and a faculty member of the University of Washington.
MA1 (Monday, 9 am - 1 pm, July 12)
Automating the Design of Effective Graphics
Steven Feiner, Columbia University; Jock Mackinlay, Xerox PARC; and
Joe Marks, DEC CRL
The notion of a linguistically articulate computer system--one that can
compose natural-language utterances to communicate given information-
-is the ultimate goal of research in natural-language generation. A
complementary notion is that of a graphically articulate computer
system, one that can design effective graphics automatically to
communicate given information. In this tutorial, we will survey recent
research on graphically articulate systems, emphasizing the potential
roles for such systems in the next generation of intelligent user
interfaces.
In this tutorial, we will present a comprehensive survey of previous
and current research on the automated design of effective graphics,
organized by the graphic being designed, the type of information being
presented, and the target domain; a description of themes and algorithmic
techniques that have arisen in the development of graphically articulate
systems; detailed case studies of projects that concern business graphics,
2D and 3D graphics for technical documentation, and automated
cartography; and possible future applications of this technology.
Prerequisite Knowledge: The tutorial describes current research in an
area that is in the intersection of computer graphics, AI, and computer
human interaction. The course is intended for intermediate and
experienced researchers and developers who are interested in
methodologies for intelligent user interfaces. A general background in
graphical user interfaces and information presentation will be assumed.
Some familiarity with computer-graphics and AI techniques will be
useful, though all relevant concepts and terms will be explained.
Steven Feiner is an Associate Professor of Computer Science at
Columbia University. His research focuses on interactive 3D graphics,
applications of AI to graphics, user interfaces, hypermedia, and virtual
worlds. He is co-author of Computer Graphics: Principles and Practice.
Jock Mackinlay is a member of the User Interface Research group at
Xerox PARC. He has worked on automating the design of various aspects of
user interfaces, including graphical presentations and input devices, and
on incorporating 3D animation into user interfaces.
Joe Marks is a member of the research staff at DEC CRL. His current
research interests include automated cartography, animation, and
evolutionary computation.
MA2 (Monday, 9 am - 1 pm, July 12)
Mobile Robots I: Instantiating Intelligent Agents
David P. Miller, MIT AI Lab, and JPL; and Marc G. Slack, MITRE Corp.
Much of the current AI research deals with the actions of embedded agents.
Simulations of an agent's environment are often inadequate for effective
system evaluation. This tutorial will give you critical information to
start embedding your systems in physical agents--mobile robots that can
interact with real environments. This tutorial will concentrate on tools
and techniques for allowing a physical agent to interact with the real
world, including, techniques for mobility and manipulation, sensing,
perception, reactive control, and navigation will be covered. Issues of
integration of these capabilities with high-level reasoning techniques
will also be introduced, providing a connection to Mobile Robots II. We
hope to sensitize you to the relevant issues of dealing with mechanical
agents and provide you with the tools to create and evaluate your work in
the context of these agents. We will demonstrate and discuss the current
state of the art in mobile robotics.
Prerequisite Knowledge: This tutorial assumes that the participants
have a basic understanding of common AI techniques and programming
languages. No previous experience with robotics is required. We
recommend you follow this tutorial with Mobile Robots II for a more in
depth look at the architectural principles involved in dealing with
multiple goals, resource constraints, and other aspects of higher-level
reasoning.
David Miller received his Ph.D. in Computer Science from Yale
University. He joined NASA's Jet Propulsion Laboratory in 1988 where
he ran the Robotic Intelligence Research Group for several years. He is
currently on sabbatical at the MIT Artificial Intelligence Laboratory.
Marc Slack received his Ph.D. in Computer Science from Virginia
Polytechnic Institute. He is currently a MITRE corporation lead scientist
in the Autonomous Systems Laboratory. His current efforts include the
development of a robot independent control architecture and support to
NASA Johnson's Autonomous systems research program.
MA3 (Monday, 9 am - 1 pm, July 12)
Case-Based Reasoning: Theory and Practice
Kevin Ashley, University of Pittsburgh, and Evangelos Simoudis,
Lockheed
The objective of the tutorial is to present the mechanisms and techniques
underlying Case-Based Reasoning (CBR) with special emphasis on CBR's
potential as an alternative to current expert systems techniques for the
development of applications. We will explore both theoretical and
practical issues, identify advantages of CBR vis a vis other reasoning
methods, and characterize application domains where CBR seems to be
most promising.
Prerequisite Knowledge: This tutorial is intended for professionals
interested in current CBR research issues, managers interested in
alternative methodologies for expert system development, and knowledge
engineers considering developing CBR applications. There are no special
prerequisites for the tutorial although some experience in designing
expert systems and familiarity with basic AI concepts and approaches to
representing knowledge and controlling inference will be helpful.
Dr. Kevin Ashley is an Assistant Professor of Law at the University of
Pittsburgh School of Law and a Research Scientist at the Learning
Research and Development Center. He also is an Adjunct Assistant
Professor of Computer Science. His research interests include case-based
and analogical reasoning, argumentation and explanation and developing
computer systems to assist attorneys in the teaching and practice of law.
He received a J. D. from Harvard Law School and a Ph.D. in computer
science from the University of Massachusetts .
Evangelos Simoudis is a research scientist in the AI Center of Lockheed
Corporation, where he is directing research on knowledge discovery from
multimedia databases, and an adjunct assistant professor at the Computer
Engineering Department of Santa Clara University. Simoudis received his
Ph.D. in Computer Science from Brandeis University in 1991. He has
consulted and taught courses in the United States and abroad in the areas of
case-based reasoning, knowledge-based design, and distributed AI.
MA4 (Monday, 9 am - 1 pm, July 12)
Computational Challenges from Molecular Biology
Peter Karp, SRI International; and Russ B. Altman, Stanford Program in
Medical Informatics
Computational problems in molecular biology provide a rich set of
challenges for AI researchers. These problems have the potential to
motivate the development of more powerful AI techniques, and to shift the
focus of AI researchers from toy problems to real problems with large
potential payoffs. This tutorial will introduce computational problems in
molecular biology to computer scientists, with an emphasis on challenges
to AI. We will provide a brief road map to computational biology in
general, and then focus on those problems that are particularly
important. Attendees will be exposed to a smorgasbord of problems and
provided with a clear problem definition, review of approaches that have
been tried, summary of progress-to-date , and distillation of chief
lessons that remain within that problem area.
We will begin by surveying computational biology in general,
including the fundamental biological notions that will be used throughout
the tutorial. We will then provide a breadth-first summary of those
problems that may be amenable to solution by AI techniques. Finally, we
will discuss four particular problems in more depth, examining issues in
technical detail, in order to stress the types of problem that arise when
dealing with biological problem and show some of the major successes in
the field. We will also discuss the research culture of this area.
Prerequisite Knowledge: Tutorial attendees should have a firm
understanding of basic issues in computer science. Registrants will
receive advance readings on the critical biological concepts.
Russ B. Altman, MD, Ph.D. is an Assistant Professor of Medicine at
Stanford University. He received his Ph.D. and MD from Stanford
University. His research interests currently focus on new methods for
the analysis and prediction of protein structure, especially with respect
to probabilistic algorithms and evaluation of uncertainty. He also is
interested in the use of abstract representations of protein and nucleic
acids for the purposes of more efficient computation.
Peter D. Karp, Ph.D. is a Computer Scientist in the AI Center at SRI
International. He received his Ph.D. degree in Computer Science from
Stanford University. His current research interests focus on building
large biological knowledge bases to support taskssuch as design,
simulation, and machine learning.
MA5 (Monday, 9 am - 1 pm, July 12)
Statistical Models in Natural Language Processing
Eugene Charniak, Brown University
Most traditional programs for understanding natural language are limited
in domain and fragile in use. One response to this has been a shift to
statistical techniques. These techniques work by collecting statistics on
large bodies of text and then use the statistics to perform simple tasks,
typically without much restriction on domain, and without "bombing. "
What makes these methods particularly powerful is that while sometimes
the text must be annotated with the correct answers, there is a powerful
training algorithm that often allows the models to be trained from raw
test. Because they work by training on a large body of text, statistical
techniques trade data availability and processing power for "smarts."
One problem that confronts people considering statistical techniques is
the seeming impenetrability of the mathematics that underlie them. We
will try to overcome this problem first by stressing the intuitive aspects
whenever possible and second by showing that the mathematics is not
really all that difficult. Tutorial topics will include: everything you need
to know about probability theory, statistical models of language and how
they relate to speech recognition, entropy and cross-entropy as measures
of language models, trigrams as a simple language model, sparse-data
problems in trigram models, word-tagging models, finding the best tags
using the Viterbi algorithm, how statistical models can be improved using
forward-backward training, context-free grammars and chart-parsing,
probabilistic context-free grammars and probabilistic chart-parsing,
finding the most likely parse, grammar learning, statistical models of
prepositional phrase attachment, techniques that can classify words by
meaning, finding word senses, statistical models of word-sense
disambiguation, applications of statistical models, hidden Markov models,
deriving the trigram model from basic assumptions, deriving word-
tagging models, finding the probability of a string according to HMM,
training an HMM, and training a context-free grammar.
Eugene Charniak is Professor of Computer Science and Cognitive
Science at Brown University and the Chairman of the Department of
Computer Science. He received his Ph.D. from MIT and is the author of
Computational Semantics with Yorick Wilks, Artificial Intelligence
Programming with Chris Riesbeck, Drew McDermott, and James Meehan,
and Introduction to Artificial Intelligence with Drew McDermott.
MP1 (Monday, 2 pm - 6 pm, July 12)
Qualitative Reasoning for Design & Diagnosis Applications
Robert Milne, Intelligent Applications Ltd.; and Louise Trave-Massuyes,
LAAS
Qualitative reasoning is an increasingly significant, but relatively new AI
technology now mature enough for serious applications. For applications
where the limitations of rule-based approaches cause serious problems,
qualitative reasoning techniques provide considerable efficiencies for the
development and long term support of an application.
In this tutorial we concentrate on how to develop successful
applications using qualitative reasoning. Case studies and examples of
application oriented work is used to integrate an introduction to the
fundamentals of qualitative reasoning with guidance on how to make
applications successful. We explore why most prototypes never become
successful applications, and describe strategies to increase the chances of
success.
Tutorial topics include qualitative representations, qualitative models,
causal representations, qualitative propagation, qualitative simulation,
selecting the appropriate paradigm, developing diagnostic, design and
other systems, and strategies for building successful applications. We
will also summarize current application oriented work using qualitative
reasoning, the techniques used by each and the motivations for using
qualitative reasoning.
Prerequisite Knowledge: This tutorial will appeal to those involved in
engineering development who would like to know how they can use
qualitative reasoning for practical applications. A background in basic AI
techniques is helpful, but not essential.
Dr. Robert Milne is the Managing Director of Intelligent Applications
Ltd., Edinburgh, Scotland He has a B. Sc. in Electrical Engineering and
Computer Science with special emphasis on AI from MIT and a Ph.D. in AI
from Edinburgh University.
Dr. Louise Trave-Massuyes is a CNRS Researcher at the French
National Laboratory LAAS (Laboratory for Automation and Systems
Analysis) working on qualitative reasoning techniques for dynamic
systems supervision applications in Toulouse, France. She has a Ph.D.
from INSA (Toulouse, France).
MP2 (Monday, 2 pm - 6 pm, July 12)
Mobile Robots II: Architectures for Reaction & Deliberation
R. James Firby, University of Chicago; and Reid G. Simmons, Carnegie
Mellon University
This tutorial will introduce concepts in architectures for controlling
autonomous mobile robots. We will begin by presenting factors that
influence modern robotic architectures: actuator and sensor control,
complex changing environments, and the desire to perform a wide variety
of tasks. The tutorial will quickly make the connection with the reactive
(behavior-based) concepts detailed in Mobile Robots I and examine them
further through case studies that illustrate specific architecture design
principles. We continue with case studies that describe alternative
deliberative (hierarchical) and hybrid architecture designs. The
strengths and weaknesses of each methodology will be analyzed in terms of
the types of tasks and environments it readily supports. Particular
attention will be paid to the problems of dealing with uncertainty,
complexity, and resource constraints. Finally, we will discuss techniques
to guarantee properties of architectures, such as hard real-time
response, consistency, or absence of resource deadlock.
Prerequisite Knowledge: This tutorial is intended for those developing
or evaluating realistic mobile robotic systems. No previous experience
with robotics is required but a basic understanding of AI planning
concepts is essential. Mobile Robots I is recommended but not required.
R. James Firby is an Assistant Professor of Computer Science at the
University of Chicago. He received his Ph.D. from Yale University. His
main interest is the construction of systems that interact with complex,
changing environments.
Reid G. Simmons is a research scientist in the School of Computer
Science and Robotics Institute at Carnegie Mellon University. He received
his Ph.D. from MIT. His research has focused on developing self-reliant
robots that can autonomously operate over extended periods of time in
unknown, unstructured environments.
MP3 (Monday, 2 pm - 6 pm, July 12)
Genetic Algorithms and Genetics-Based Machine Learning
David E. Goldberg, University of Illinois at Urbana-Champaign; and John
R. Koza, Stanford University
This tutorial will introduce participants to the ideas and applications of
genetic algorithms (GAs)--computer search procedures based on the
mechanics of natural genetics and natural selection--and genetics-based
machine learning (GBML)--machine learning techniques that use genetic
algorithms and their derivatives. GAs and GBML are receiving increased
attention in practical yet difficult search and machine learning problems
across a spectrum of disciplines. We review the mechanics of a simple
genetic algorithm and consider the implicit parallelism that underlies its
power. A parade of current search applications is reviewed as are more
advanced GA techniques such as niching and messy GAs. The two most
prominent techniques of GMBL, classifier systems and genetic
programming, are also surveyed.
Prerequisite Knowledge: Knowledge of genetic algorithms or biological
concepts is not assumed. A general familiarity with computers and
programming is required.
David E. Goldberg is an Associate Professor of General Engineering and
the Beckman Institute at the University of Illinois at Urbana-Champaign.
He received his Ph.D. from the University of Michigan and has written
papers on the application and foundations of genetic algorithms. He is the
author of Genetic Algorithms in Search, Optimization, and Machine
Learning. His recent studies have considered the theory of deception, the
role of noise in GA convergence, and the theory and development of messy
GAs.
John R. Koza is a Consulting Professor of Computer Science at Stanford
University. He received his Ph.D. in Computer Science from the
University of Michigan in the field of machine learning and induction. He
currently is investigating the artificial breeding of computer programs
and has recently completed Genetic Programming, a book that surveys
these efforts. Between 1973 and 1987 he was chief executive officer of
Scientific Games Incorporated in Atlanta, and he is currently a principal
in Third Millennium Venture Capital Limited in California.
MP4 (Monday, 2 pm - 6 pm, July 12)
Experimental Methods in AI
Paul Cohen, University of Massachusetts at Amherst; and Bruce Porter,
University of Texas at Austin
This tutorial will introduce, through case studies, designs for exploratory
and confirming experiments, a few statistical techniques for analyzing
results, and techniques for evaluating what results mean to one's research
and development program. Our objective is to provide methods by which
researchers and practitioners can test hypotheses and substantiate claims
of system performance. We will focus on experiments in knowledge-based
systems, machine learning and planning, although much of our discussion
will apply to broader areas of AI. The case studies illustrate techniques
for measuring and comparing levels of performance, analyzing the effects
of adding knowledge to a system, finding interactions between components
of systems, and isolating causes of poor performance. We will discuss
some tricky problems, including designing representative test sets and
getting representative samples of data; and we will describe some open
problems, for which convincing techniques are not yet available, such as
generalizing results from one system to others.
Prerequisite Knowledge: Some familiarity with knowledge-based
systems and basic machine learning and planning techniques is helpful but
not essential. No knowledge of statistics is assumed.
Dr. Paul Cohen is an Associate Professor at the University of
Massachusetts at Amherst, where he directs the Experimental Knowledge
Systems Lab. His research concerns planning in uncertain, real-time
environments. Cohen has been involved in the development of research
methods for AI and computer science in general for several years. Cohen
received his Ph.D. from Stanford University, and was an editor of The
Handbook of Artificial Intelligence, Volumes III and IV.
Bruce Porter is an Associate Professor of Computer Sciences at the
University of Texas at Austin. His current research is developing large
knowledge bases and methods for generating coherent explanations to
answer questions using them. His earlier research with Ray Bareiss
produced the Protos knowledge acquisition system and a comprehensive
evaluation of its performance. Porter received his Ph.D. from the
University of California at Irvine, and was honored with a Presidential
Young Investigator Award from the National Science Foundation in 1988.
MP5 (Monday, 2 pm - 6 pm, July 12)
Probabilistic Causal Modeling
Judea Pearl, UCLA
Probabilistic causal models provide effective tools for combining
evidence, generating explanations and predicting the effect of actions in
uncertain environments . This tutorial will cover graphical models as a
language for encoding probabilistic causal knowledge; methods of inducing
causes-effect relationships from statistical data; inference methods of
interpreting multiple evidence, given a probabilistic causal model; and
symbolic methods of reasoning with causal rules, observations and
actions, using order-of-magnitude abstractions of probabilities.
Prerequisite Knowledge: The tutorial is designed for statisticians and
data-analysts interested in newly developed graphical techniques of model
selection for prediction and policy analysis, machine learning
researchers exploring methods of uncovering stable relationships and
action-related theories in data, and researchers in automated reasoning
seeking a coherent framework for analyzing abduction, diagnosis, belief
revision, belief updating, counterfactual queries, temporal prediction,
attention management, actions, persistence and causation.
Judea Pearl is a Professor of Computer Science at UCLA where he is
the Director of the Cognitive Systems Laboratory. He received his Ph.D. in
Electrical Engineering from the Polytechnic Institute of Brooklyn,
Brooklyn, NY. His current interests include: knowledge representation,
probabilistic reasoning, constraint processing, nonstandard logics,
distributed computation, and learning.
Pearl is the author of Heuristics and Probabilistic Reasoning In
Intelligent Systems, and the editor of Search and Heuristics, and Readings
in Uncertain Reasoning (with G. Shafer).
IAAI-93 Preliminary Program
(Subject to change)
Monday, July 12
8:30 - 9:00 am
Opening Remarks
Phil Klahr, IAAI Conference Chair
"What Makes an AI Application Innovative?"
9:00 - 9:30 am
Computer Aided Parts Estimation
Adam Cunningham, Ford Motor Company and Robert Smart, Inference
Europe Ltd.
9:30 - 10:00 am
Expanding the Utility of Legacy Systems: MPSA-The Master Production
Scheduler's Assistant. COLES-The Customer Order Loading Expert System
Joseph McManus, AT&T Network Systems and Teresa Garland, Inference
Corporation
10:00 - 10:20 am
Break
10:20 - 10:50 am
The Application Software Factory: A Knowledge Based Approach to
Software Engineering
Stu Burton, Celite Corporation; Kent Swanson and Lisa Leonard, Andersen
Consulting
10:50 am - 11:50 pm
IAAI Invited Talk
How to Market AI. . . . . . . . NOT! OrIIf AI is a Wireless Telegraph, What
Is a Radio?
Joe Carter, Andersen Consulting
There are eight billion people in the world and most of them are under
employed. The AI community has spent the last three decades trying to
emulate those under-employed resources. What businessperson in his or
her right mind would invest in an artificial substitute for a surplus
commodity? AI is a bomb looking for a war that doesn't exist outside the
laboratory.
But amongst all the chaff of AI hype, there is plenty of wheat. In fact,
the early hype may actually understate the value of the underlying AI
technology. The AI community set out to build a wireless telegraph and
ended up producing a radio. Can the AI equivalent of the television be far
behind?
Based on a decade of experience in successfully commercializing the
output of the AI community, Mr. Carter will share his thoughts on
packaging AI technology for prime-time consumption.
Joe Carter has worldwide responsibility for Andersen Consulting's
practice in multimedia, imaging, and knowledge management technologies.
Andersen Consulting's personnel under Mr. Carter's direction provide
specialized training, R&D, Methodology, Marketing, and System
Development services for Andersen's clients and 25,000 regional
consultants around the world.
11:50 - 12:20 pm
Meet the Authors
12:20 - 2:00 pm
Lunch
2:00 - 3:30 pm
AI-on-Line Panel
"What's Next on KBS Technology for Industrial Use"
Organized by Herb Schorr, USC Information Sciences Institute
3:30 - 3:50 pm
Break
3:50 - 4:20 pm
ESDS: Materials Technology Knowledge Bases Supporting Design of Boeing
Jetliners
Mark A. Dahl, The Boeing Company
4:20 - 4:50 pm
Managing Product Quality By Integrating Operations Research And
Artificial Intelligence Technologies
Charles S. Moon, Thomas M. Moore and Suheil M. Nasser, IBM Industrial
Sector Division
4:50 - 5:20 pm
COMPAQ QuickSource: Providing the Consumer with the Power of
Artificial Intelligence
Trung Nguyen and Mary Czerwinski, Compaq Computer Corporation
5:20 - 5:50 pm
Meet the Authors
6:00 - 7:00 pm
IAAI Opening Reception
7:30 - 8:30 pm
Invited Talk: "Tiger in a Cage: The Applications of KB Systems"
Edward Feigenbaum, Stanford University
Tuesday, July 13
8:30 - 10:00 am
AAAI Keynote Address
Herbert Simon, Carnegie Mellon University
10:00 - 10:30 am
Break
10:30 - 11:00 am
Tennessee Offender Management Information System
David Reynolds and Tim Beck, Andersen Consulting
11:00 - 11:30 am
OPERA: A Highly Interactive Expert System for Outside Plant Engineering
Gary Lazarus, Lien Tran and Marty Baade, NYNEX Science & Technology
11:30 am - 12:00 pm
A Knowledge-Based Configurator that Supports Sales, Engineering, and
Manufacturing at AT&T Network Systems
Jon R. Wright, Elia S. Weixelbaum, Gregg T. Vesonder, Karen E. Brown,
Stephen R. Palmer, Jay I. Berman and Harry H. Moore, AT&T Bell
Laboratories
12:00 - 12:30 pm
Meet the Authors
12:30 - 2:00 pm
Lunch
2:00 - 3:30 pm
AI-on-Line Panel
Emerging AI Technologies
Organized by Neena Buck, New Science Associates
3:30 - 3:50 pm
Break
3:50 - 4:20 pm
Digitized Expert PICTures (DEPICT) An Intelligent Information
Repository
George Gallant and Janet Thygesen, IBM Corporation
4:20 - 4:50 pm
Diagnostic Yield Characterization Expert (DYCE) A Diagnostic Knowledge
Based System Shell for Automated Data Analysis
Donald D. Pierson and George J. Gallant, IBM Corporation
4:50 - 5:20 pm
Pitch Expert: A Problem-Solving System for Kraft Mills
A. Kowalski, D. Bouchard, L. H. Allen, Y. Larin and O. Vadas, Pulp and
Paper Research Institute of Canada
5:20 - 5:50 pm
Meet the Authors
6:00 - 7:00 pm
Join in on the AAAI Opening Reception!
Wednesday, July 14
8:30 - 9:00 am
The DRAIR Advisor: A Knowledge-Based System for Material Deficiency
Analysis
Brian L. Robey, Pamela K. Fink, Sanjeev Venkatesan and Carol L. Redfield,
Southwest Research Institute; Jerry W. Ferguson, US Air Force
9:00 - 9:30 am
Dodger, an Expert System for Eddy Current Evaluation
Arthur J. Levy, Jane E. Oppenlander, David M. Brudnoy, James M.
Englund, Kent C. Loomis and Arnold M. Barsky, General Electric Co.
9:30 - 10:00 am
GCESS: A Symptom Driven Diagnostic Shell and Related Applications
Peter Holtzman, Inference Corporation
10:00 - 10:20 am
Break
10:20 - 10:50 am
PI-in-a-Box: A Knowledge-based System for Space Science
Experimentation
Richard Frainier and Nicolas Groleau, RECOM Technologies, Inc.; Lyman
Hazelton, Laurence Young and Peter Szolovits, Massachusetts Institute of
Technology; Silvano Colombano and Irving Statler, NASA Ames Research
Center; Michael Compton, Sterling Software.
10:50 - 11:50 am
Invited Talk
Bob Kahn
11:50 am - 12:20 pm
Meet the Authors
12:20 - 2:00 pm
Lunch Break
2:00 - 3:00 pm
AI-on-Line Panel
Transitioning AI Technology into Government Information Systems
Organized by Ted Senator, Department of the Treasury
AAAI Preliminary Program
Tuesday, July 13
8:30-9:50 am
Keynote Address
Herbert A. Simon, Carnegie Mellon University
9:50 - 10:15 pm
Break
10:15 - 11:55 am
Session 1
Diagnostic Reasoning
10:15 - 10:40 am
Hybrid Case-Based Reasoning for the Diagnosis of Complex Devices
M. P. Feret and J. I. Glasgow, Queen's University
10:40 - 11:05 am
An Epistemology for Clinically Significant Trends
Ira J. Haimowitz, MIT Laboratory for Computer Science and Isaac S.
Kohane, Harvard Medical School
11:05 - 11:30 am
A Framework for Model-Based Repair
Ying Sun and Daniel S. Weld, University of Washington
11:30 - 11:55 am
Multiple Dimensions of Generalization In
Model-Based Troubleshooting
Randall Davis and Paul Resnick, MIT AI Laboratory
10:15 - 11:55 am
Session 2
Intelligent User Interfaces
10:15 - 10:40 am
Generating Explanations of Device Behavior Using Compositional Modeling
and Causal Ordering
Patrice O. Gautier and Thomas R. Gruber, Stanford University
10:40 - 11:05 am
Building Synthesis Models to Support Early Stage Product and Process
Design
R. Bharat Rao and Stephen C-Y. Lu, University of Illinois at Urbana-
Champaign
11:05 - 11:30 am
A Conversational Model of Multimodal Interaction
Adelheit Stein and Ulrich Thiel, German National Center for Computer
Science
11:30 - 11:55 am
Generating Natural Language Descriptions with Examples: Differences
between Introductory and Advanced Texts
Vibhu O. Mittal and Ccile L. Paris, USC/ISI
10:15 - 11:55 am
Session 3
Search--1
10:15 - 10:40 am
Conjunctive Width Heuristics for Maximal Constraint Satisfaction
Richard J. Wallace and Eugene C. Freuder, University of New Hampshire
10:40 - 11:05 am
Decomposition of Domains Based on the Micro-structure of Finite
Constraint-Satisfaction Problems
Philippe Jgou, Universit de Provence
11:05 - 11:30 am
Innovative Design as Systematic Search
Dorothy Neville and Daniel S. Weld, University of Washington
11:30 - 11:55 am
Time-Saving Tips for Problem Solving with Incomplete Information
Michael R. Genesereth and Illah R. Nourbakhsh, Stanford University
11:55 am-1:30 pm
Lunch
1:30 - 3:10 pm
Session 4
Reasoning about Physical Systems--I
1:30 - 1:55 pm
CFRL: A Language for Specifying the Causal Functionality of Engineered
Devices
Marcos Vescovi, Yumi Iwasaki and Richard Fikes, Stanford University; B.
Chandrasekaran, Ohio State University
1:55 - 2:20 pm
Model Simplification in Fluid Mechanics
Kenneth Man-Kam Yip, Yale University
2:20 - 2:45 pm
Numerical Behavior Envelopes for Qualitative Models
Herbert Kay and Benjamin Kuipers, University of Texas at Austin
2:45 - 3:10 pm
A Qualitative Method to Construct Phase Portraits
Wood Wai Lee, Dowell Schlumberger and Benjamin J. Kuipers,
University of Texas
1:30 - 3:10 pm
Session 5
Natural Language Sentence Analysis
1:30 - 1:55 pm
Having Your Cake and Eating It Too: Autonomy and Interaction in a Model of
Sentence Processing
Kurt P. Eiselt and Kavi Mahesh, Georgia Institute of Technology; Jennifer
K. Holbrook, Albion College
1:55 - 2:20 pm
Efficient Heuristic Natural Language Parsing
Christian R. Huyck and Steven L. Lytinen, University of Michigan
2:20 - 2:45 pm
Machine Translation of Spatial Expressions: Defining the Relation between
an Interlingua and a Knowledge Representation System
Bonnie J. Dorr and Clare R. Voss, University of Maryland
2:45 - 3:10 pm
Towards a Reading Coach that Listens: Automated Detection of Oral Reading
Errors
Jack Mostow, Alexander G. Hauptmann, Lin Lawrence Chase and Steven
Roth, Carnegie Mellon University
1:30 - 3:10 pm
Session 6
Search--II
1:30 - 1:55 pm
Depth-First Versus Best-First Search: New Results
Weixiong Zhang and Richard E. Korf, University of California, Los Angeles
1:55 - 2:20 pm
Pruning Duplicate Nodes in Depth-First Search
Larry A. Taylor and Richard E. Korf, University of California, Los Angeles
2:20 - 2:45 pm
Iterative Weakening: Optimal and Near-Optimal Policies for the Selection
of Search Bias
Foster John Provost, University of Pittsburgh
2:45 - 3:10 pm
Generating Effective Admissible Heuristics by Abstraction and
Reconstitution
Armand Prieditis and Bhaskar Janakiraman, University of California,
Davis
3:10 - 3:30
Break
3:30 - 5:10 pm
Session 7
Reasoning about Physical Systems--II
3:30 - 3:55 pm
Ideal Physical Systems
Brian Falkenhainer, Xerox Corporation/Cornell University
3:55 - 4:20 pm
Intelligent Model Selection for Hill Climbing Search in Computer-Aided
Design
Thomas Ellman and John Keane, Rutgers University
4:20 - 4:45 pm
Sensible Scenes: Visual Understanding of Complex Structures through
Causal Analysis
Matthew Brand, Lawrence Birnbaum and Paul Cooper, Northwestern
University
4:45 - 5:10 pm
Understanding Linkages
Howard E. Shrobe, MIT and Symbolics, Inc.
3:30 - 5:10 pm
Session 8
Natural Language Generation
3:30 - 3:55 pm
Corpus Analysis for Revision-Based Generation of Complex Sentences
Jacques Robin and Kathy McKeown, Columbia University
3:55 - 4:20 pm
Generating Argumentative Judgment Determiners
Michael Elhadad, Ben Gurion University of the Negev
4:20 - 4:45 pm
Bidirectional Chart Generation of Natural Language Texts
Masahiko Haruno, Yasuharu Den and Yuji Matsumoto, Kyoto University
4:45 - 5:10 pm
Communicative Acts for Generating Natural Language Arguments
Mark T. Maybury, The MITRE Corporation
3:30 - 5:10 pm
Session 9
Plan Generation
3:30 - 3:55 pm
Granularity in Multi-Method Planning
Soowon Lee and Paul S Rosenbloom, University of Southern California
3:55 - 4:20 pm
An Average Case Analysis of Planning
Tom Bylander, Ohio State University
4:20 - 4:45 pm
Postponing Conflicts in Nonlinear Planning
David E. Smith, Rockwell International and Mark A. Peot, Stanford
University
4:45 - 5:10 pm
Threat-Removal Strategies for Nonlinear Planning
Mark A. Peot, Stanford University and David E. Smith, Rockwell
International
6:00 - 7:00 pm
AAAI Opening Reception
Wednesday, July 14
8:30 - 9:50 am
Presidential Address
Pat Hayes
9:50 - 10:15 am
Break
10:15 - 11:55 am
Session 10
Trainable Natural Language Systems
10:15 - 10:40 am
A Case-Based Approach to Knowledge Acquisition for Domain-Specific
Sentence Analysis
Claire Cardie, University of Massachusetts
10:40 - 11:05 am
Automatically Constructing a Dictionary for Information Extraction Tasks
Ellen Riloff, University of Massachusetts
11:05 - 11:30 am
Learning Semantic Grammars with Constructive Inductive Logic
Programming
John M. Zelle and Raymond J. Mooney, University of Texas
11:30 - 11:55 am
KITSS: A Knowledge-Based Translation System for Test Scenarios
Van E. Kelly and Mark A. Jones, AT&T Bell Laboratories
10:15 - 11:55 am
Session 11
Distributed Problem Solving--I
10:15 - 10:40 am
Solving the Really Hard Problems with Cooperative Search
Tad Hogg and Colin P. Williams, Xerox PARC
10:40 - 11:05 am
A Fast First-Cut Protocol for Agent Coordination
Andrew P. Kosoresow, Stanford University
11:05 - 11:30 am
A One-shot Dynamic Coordination Algorithm for Distributed Sensor
Networks
Keith Decker and Victor Lesser, University of Massachusetts
11:30 - 11:55 am
An Implementation of the Contract Net Protocol Based on Marginal Cost
Calculations
Tuomas Sandholm, University of Massachusetts
10:15 - 11:55 am
Session 12
Qualitative Reasoning
10:15 - 10:40 am
Qualitatively Describing Objects Using Spatial Prepositions
Alicia Abella and John R. Kender, Columbia University
10:40 - 11:05 am
Efficient Reasoning in Qualitative Probabilistic Networks
Marek J. Druzdzel, Carnegie Mellon University and Max Henrion,
Rockwell International Science Center
11:05 - 11:30 am
Numeric Reasoning with Relative Orders of Magnitude
Philippe Dague, Universit Paris Nord
11:30 - 11:55 am
Generating Quasi-symbolic Representation of Three-Dimensional Flow
Toyoaki Nishida, Kyoto University
11:55 am-1:30 pm
Lunch
1:30 - 2:45 pm
Session 13
Discourse Analysis
1:30 - 1:55 pm
An Optimizing Method for Structuring Inferentially Linked Discourse
Ingrid Zukerman and Richard McConachy, Monash University
1:55 - 2:20 pm
A Methodology for Development of Dialogue Managers for Natural Language
Interfaces
Arne Jnsson, Linkping University
2:20 - 2:45 pm
Mutual Beliefs of Multiple Conversants: A Computational Model of
Collaboration in Air Traffic Control
David G. Novick and Karen Ward, Oregon Graduate Institute of Science and
Technology
1:30 - 3:10 pm
Session 14
Distributed Problem Solving--II
1:30 - 1:55 pm
IPUS: An Architecture for Integrated Signal Processing and Signal
Interpretation in Complex Environments
Victor Lesser, Izaskun Gallastegi and Frank Klassner, University of
Massachusetts; Hamid Nawab, Boston University
1:55 - 2:20 pm
Overeager Reciprocal Rationality and Mixed Strategy Equilibria
Edmund H. Durfee and Jaeho Lee, University of Michigan, Piotr J.
Gmytrasiewicz, Hebrew University
2:20 - 2:45 pm
Agents Contracting Tasks in Non-Collaborative Environments
Sarit Kraus, Bar Ilan University Ramat Gan
2:45 - 3:10 pm
Quantitative Modeling of Complex Computational Task Environments
Keith Decker and Victor Lesser, University of Massachusetts
1:30 - 3:10 pm
Session 15
Automated Reasoning--I
1:30 - 1:55 pm
The Breakout Method For Escaping From Local Minima
Paul Morris, IntelliCorp
1:55 - 2:20 pm
Towards an Understanding of Hill-climbing Procedures for SAT
Ian P. Gent and Toby Walsh, University of Edinburgh
2:20 - 2:45 pm
An Empirical Study of Greedy Local Search for Satisfiability Testing
Henry A. Kautz and Bart Selman, AT&T Bell Laboratories
2:45 - 3:10 pm
Experimental Results on the Cross-Over Point in Satisfiability Problems
James M. Crawford and Larry D. Auton, AT&T Bell Laboratories
3:10 - 3:50 pm
Break
3:30 - 4:45 pm
Session 16
Statistically-Based Natural Language Processing
3:30 - 3:55 pm
Equations for Part-of-Speech Tagging
Eugene Charniak, Curtis Hendrickson, Neil Jacobson and Mike Perkowitz,
Brown University
3:55 - 4:20 pm
Using an Annotated Language Corpus as a Virtual Stochastic Grammar
Rens Bod, University of Amsterdam
4:20 - 4:45 pm
Estimating Probability Distribution Over Hypotheses with Variable
Unification
Dekai Wu, Hong Kong University of Science and Technology
3:30 - 5:10 pm
Session 17
Vision Processing
3:30 - 3:55 pm
Learning Object Models from Appearance
Hiroshi Murase and Shree K. Nayar, Columbia University
3:55 - 4:20 pm
Polly: A Vision-Based Artificial Agent
Ian Horswill, MIT AI Laboratory
4:20 - 4:45 pm
On the Qualitative Structure of Temporally Evolving Visual Motion Fields
Richard P. Wildes, SRI David Sarnoff Research Center
4:45 - 5:10 pm
Range Estimation From Focus Using an Active Non-frontal Imaging
Camera
Arun Krishnan and Narendra Ahuja, University of Illinois
3:30 - 5:10 pm
Session 18
Automated Reasoning--II
3:30 - 3:55 pm
On Computing Minimal Models
Rachel Ben-Eliyahu, University of California, Los Angeles and Rina
Dechter, University of California, Irvine
3:55 - 4:20 pm
Reasoning With Characteristic Models
Henry A. Kautz, Michael J. Kearns and Bart Selman, AT&T Bell
Laboratories
4:20 - 4:45 pm
Rough Resolution: A Refinement of Resolution to Remove Large Literals
Heng Chu and David A. Plaisted, University of North Carolina
4:45 - 5:10 pm
On the Adequateness of the Connection Method
Antje Beringer and Steffen Hlldobler, Technische Hochschule Darmstadt
Thursday, July 15
8:30 - 10:10 am
Session 19
Machine Learning
8:30 - 8:55 am
Finding Accurate Frontiers: A Knowledge-Intensive Approach to
Relational Learning
Michael Pazzani and Clifford Brunk, University of California, Irvine
8:55 - 9:20 am
OC1: A Randomized Algorithm for Building Oblique Decision Trees
Sreerama K. Murthy, Simon Kasif and Steven Salzberg, Johns Hopkins
University; Richard Beigel, Yale University
9:20 - 9:45 am
Probabilistic Prediction of Protein Secondary Structure Using Causal
Networks
Arthur L Delcher, Loyola College in Maryland, Simon Kasif, Harry R.
Goldberg and William H. Hsu, Johns Hopkins University
9:45 - 10:10 am
Learning Non-Linearly Separable Boolean Functions With Linear
Threshold Unit Trees and Madaline-Style Networks
Mehran Sahami, Stanford University
8:30 - 10:10 am
Session 20
Large Scale Knowledge Bases
8:30 - 8:55 am
Automated Index Generation for Constructing Large-Scale Conversational
Hypermedia Systems
Richard Osgood and Ray Bareiss, Northwestern University
8:55 - 9:20 am
Case-Method: A Methodology for Building Large-Scale Case-Based
Systems
Hiroaki Kitano, Hideo Shimazu and Akihiro Shibata, NEC Corporation
9:20 - 9:45 am
Matching 100,000 Learned Rules
Robert B. Doorenbos, Carnegie Mellon University
9:45 - 10:10 am
Massively Parallel Support for Computationally Effective Recognition
Queries
Matthew P. Evett, William A. Andersen and James A. Hendler, University
of Maryland
8:30 - 10:10 am
Session 21
Real-Time Planning and Simulation
8:30 - 8:55 am
Real-Time Self-Explanatory Simulation
Franz G. Amador, Adam Finkelstein and Daniel S. Weld, University of
Washington
8:55 - 9:20 am
Task Interdependencies in Design-to-time Real-time Scheduling
Alan Garvey, Marty Humphrey, Victor Lesser, University of
Massachusetts
9:20 - 9:45 am
Planning With Deadlines in Stochastic Domains
Thomas Dean, Leslie Kaelbling, Jak Kirman and Ann Nicholson, Brown
University
9:45 - 10:10 am
A Comparison of Action-Based Hierarchies and Decision Trees for Real-
Time Performance
David Ash and Barbara Hayes-Roth, Stanford University
8:30 - 10:10 am
Session 22
Representation for Actions and Motion
8:30 - 8:55 am
Towards Knowledge-Level Analysis of Motion Planning
Ronen I. Brafman, Jean-Claude Latombe and Yoav Shoham, Stanford
University
8:55 - 9:20 am
The Semantics of Event Prevention
Charles L. Ortiz, Jr., University of Pennsylvania
9:20 - 9:45 am
EL: A Formal, Yet Natural, Comprehensive Knowledge Representation
Chung Hee Hwang and Lenhart K. Schubert, University of Rochester
9:45 - 10:10 am
The Frame Problem and Knowledge-Producing Actions
Richard B. Scherl and Hector J. Levesque, University of Toronto
10:10-10:30 am
Break
10:30 am-11:45 pm
Session 23
Novel Methods in Machine Learning
10:30 - 10:55 am
Learning Interface Agents
Pattie Maes and Robyn Kozierok, MIT Media-Lab
10:55 - 11:20 am
Learning from an Approximate Theory and Noisy Examples
Somkiat Tangkitvanich and Masamichi Shimura, Tokyo Institute of
Technology
11:20 - 11:45 am
Scientific Model-Building as Search in Matrix Spaces
Ra l E. Valds-Prez, Jan M. Zytkow and Herbert A. Simon, Carnegie
Mellon University
10:30 am-12:10 pm
Session 24
Case-Based Reasoning
10:30 - 10:55 am
Projective Visualization: Acting from Experience
Marc Goodman, Cognitive Systems Inc., and Brandeis University
10:55 - 11:20 am
A Framework and an Analysis of Current Proposals for the Case-Based
Organization and Representation of Procedural Knowledge
Roland Zito-Wolf and Richard Alterman, Brandeis University
11:20 - 11:45 am
Case-Based Diagnostic Analysis in a Blackboard Architecture
Edwina L. Rissland, Jody J. Daniels, Zachary B. Rubinstein and David B.
Skalak, University of Massachusetts
11:45 am-12:10 pm
Representing and Using Procedural Knowledge for Doing Geometry Proofs
Thomas F. McDougal, Kristian J. Hammond, University of Chicago
10:30 am-12:10 pm
Session 25
Rule-Based Reasoning
10:30 - 10:55 am
Exploring the Structure of Rule Based Systems
C. Grossner, A. Preece, P. Gokul Chander, T. Radhakrishnan, C. Y. Suen,
Concordia University
10:55 - 11:20 am
Comprehensibility Improvement of Tabular Knowledge Bases
Atsushi Sugiura and Yoshiyuki Koseki, NEC Corporation; Maximilian
Riesenhuber, Johann Wolfgang Goethe-University
11:20 - 11:45 am
Supporting and Optimizing Full Unification in a Forward Chaining Rule
System
Howard E. Shrobe, MIT and Symbolics, Inc.
11:45 am-12:10 pm
The Paradoxical Success of Fuzzy Logic
Charles Elkan, University of California, San Diego
10:30 am-12:10 pm
Session 26
Representation and Reasoning
10:30 - 10:55 am
All They Know About--Preliminary Report
Gerhard Lakemeyer, University of Bonn
10:55 - 11:20 am
Nonmonotonic Reasoning with Many Agents
Joseph Y. Halpern, IBM Almaden Research Center
11:20 - 11:45 am
Conditional Belief Revision: Model-Based and Syntactic Approaches
Craig Boutilier, University of British Columbia; Moiss Goldszmidt,
Rockwell International
11:45 am-12:10 pm
Abduction As Belief Revision: A Model of Preferred Explanations
Craig Boutilier and Veronica Becher, University of British Columbia
12:10 - 1:30 pm
Lunch
1:30 - 3:10 pm
Session 27
Plan Learning
1:30 - 1:55 pm
Permissive Planning: A Machine Learning Approach to Linking Internal
and External Worlds
Gerald DeJong and Scott Bennett, University of Illinois
1:55 - 2:20 pm
On the Masking Effect
Milind Tambe, Carnegie Mellon University; Paul S. Rosenbloom, USC/ISI
2:20 - 2:45 pm
Relative Utility of EBG based Plan Reuse in Partial Ordering Versus Total
Ordering Planning
Subbarao Kambhampati and Jengchin Chen, Arizona State University
2:45 - 3:10 pm
Learning Plan Transformations from Self-Questions: A Memory-Based
Approach
R. Oehlmann, D. Sleeman, P. Edwards, King's College
1:30 - 3:10 pm
Session 28
Constraint-Based Reasoning--I
1:30 - 1:55 pm
Nondeterministic Lisp as a Substrate for Constraint Logic Programming
Jeffrey Mark Siskind, University of Pennsylvania; David Allen
McAllester, MIT AI Laboratory
1:55 - 2:20 pm
Arc-Consistency and Arc-Consistency Again
Christian Bessire, University of Montpellier II; Marie-Odile Cordier,
University of Rennes I
2:20 - 2:45 pm
On the Consistency of General Constraint-Satisfaction Problems
Philippe Jgou, Universit de Provence
2:45 - 3:10 pm
Extending Deep Structure
Colin P. Williams and Tad Hogg, Xerox PARC
1:30 - 3:10 pm
Session 29
Nonmonotonic Logic--I
1:30 - 1:55 pm
Generating Explicit Orderings for Nonmonotonic Logics
James Cussens, King's College; Anthony Hunter, Imperial College; Ashwin
Srinivasan, Oxford University
1:55 - 2:20 pm
Minimal Belief and Negation as Failure: A Feasible Approach
Antje Beringer and Torsten Schaub, Technische Hochschule Darmstadt
2:20 - 2:45 pm
Algebraic Sematics for Cumulative Inference Operations
Zbigniew Stachniak, University of Toronto
2:45 - 3:10 pm
Subnormal Modal Logics for Knowledge Representation
Grigori Schwarz, Stanford University; Miroslaw Truszczynski,
University of Kentucky
3:10 - 3:30 pm
Break
3:30 - 5:10 pm
Session 30
Complexity in Machine Learning
3:30 - 3:55 pm
Cryptographic Limitations on Learning One-Clause Logic Programs
William W. Cohen, AT&T Bell Laboratories
3:55 - 4:20 pm
Learnability in Inductive Logic Programming: Some Basic Results and
Techniques
Michael Frazier and C. David Page Jr., University of Illinois
4:20 - 4:45 pm
Complexity Analysis of Reinforcement Learning
Sven Koenig and Reid G. Simmons, Carnegie Mellon
4:45 - 5:10 pm
Pac-Learning a Restricted Class of Recursive Logic Programs
William W. Cohen, AT&T Bell Laboratories
3:30 - 5:10 pm
Session 31
Constraint-Based Reasoning--II
3:30 - 3:55 pm
A Constraint Decomposition Method for Spatio-Temporal Configuration
Problems
Toshikazu Tanimoto, Digital Japan
3:55 - 4:20 pm
Coping With Disjunctions in Temporal Constraint Satisfaction Problems
Eddie Schwalb and Rina Dechter, University of California, Irvine
4:20 - 4:45 pm
Integrating Heuristics for Constraint Satisfaction Problems: A Case Study
Steven Minton, NASA Ames Research Center
4:45 - 5:10 pm
Slack-Based Heuristics for Constraint Satisfaction Scheduling
Stephen F. Smith and Cheng-Chung Cheng, Carnegie Mellon University
3:30 - 5:10 pm
Session 32
Nonmonotonic Logic--II
3:30 - 3:55 pm
Propositional Logic of Context Extended Abstract
Sasa Buvac and Ian A. Mason, Stanford University
3:55 - 4:20 pm
A Context-based Framework for Default Logics
Philippe Besnard, IRISA; Torsten Schaub, TH Darmstadt
4:20 - 4:45 pm
Reasoning Precisely with Vague Concepts
Nita Goyal and Yoav Shoham, Stanford University
4:45 - 5:10 pm
Restricted Monotonicity
Vladimir Lifschitz, University of Texas at Austin
AAAI-93 Technical Program Registration
July 13-15, 1992
Your AAAI-93 technical program registration includes admission to
technical sessions, invited talks, the AAAI-93 / IAAI-93 joint exhibition,
AI-on-Line panels, the AAAI-93 opening reception, and the AAAI-93
conference Proceedings.
Onsite Registration will be located at the entrance of Exhibit Hall B on
the upper level of the Washington Convention Center, 900 Ninth Street,
NW., Washington DC.
IAAI-93 Registration
July 13-15, 1992
Your IAAI-93 registration includes admission to IAAI-93, the AI-on-Line
sessions, the AAAI-93 / IAAI-93 joint exhibition, the AAAI-93 opening
address, the IAAI-93 and AAAI-93 opening receptions, and the IAAI
conference Proceedings.
IAAI-93 / AAAI-93
Registration Fees
Early Registration (Postmarked by 21 May)
AAAI Members
Regular $260 Student $100
Nonmembers
Regular $305 Student $165
Early Bird Package Registration (Postmarked by 21 May)
AAAI Members
Regular $730 Student $330
Nonmembers
Regular $870 Student $430
Late Registration (Postmarked by 11 June)
AAAI Members
Regular $310 Student $100
Nonmembers
Regular $360 Student $165
On-Site Registration (Postmarked after 11 June or onsite. Hours below.)
AAAI Members
Regular $380 Student $105
Nonmembers
Regular $420 Student $180
AAAI-93 / IAAI-93 Combined Registration
To attend both the AAAI-93 and the IAAI-93 Conferences, and to receive
the proceedings of both, please add $100.00 to the appropriate
registration fee above.
AAAI-93 / IAAI-93 Early Bird Package Registration
This year, AAAI is offering a special-value Early Bird Package
Registration for attendees registering before May 21th. Included in your
registration fee are the AAAI-93 Conference Registration, the IAAI-93
Conference Registration, and two tutorials (of your choice). Sorry, no
substitutions . Early-bird registrations are non-transferable.
Tutorial Program Registration
July 11-12, 1993
The Tutorial Program Registration includes admission to one tutorial, the
AAAI-93 / IAAI-93 Joint Exhibition, the AI-on-Line panels at IAAI-93,
and the tutorial syllabus. Prices quoted are per tutorial. A maximum of
four may be taken due to parallel schedules.
Tutorial Fee Schedule
Early Registration (Postmarked by 21 May)
AAAI Members
Regular $205 Student $75
Nonmembers
Regular $260 Student $95
Late Registration (Postmarked by 11 June)
AAAI Members
Regular $245 Student $100
Nonmembers
Regular $290 Student $120
On-Site Registration (Postmarked after 11 June or onsite. Hours are
listed below.)
AAAI Members
Regular $310 Student $130
Nonmembers
Regular $330 Student $155
Workshop Registration
July 11-12, 1993
Workshop registration is limited to active participants determined by the
organizer prior to the conference. Those individuals attending workshops
only are subject to a $125.00 per workshop registration fee.
Payment & Registration Information
Prepayment of registration fees is required. Checks, international money
orders, bank transfers and traveler's checks must be in US dollars. Amex,
MasterCard, Visa, and government purchase orders are also accepted.
Registrations postmarked after the June 11 deadline will be subject to
on-site registration fees. The deadline for refund requests is June 18,
1993. All refund requests must be made in writing. A $75. 00 processing
fee will be assessed for all refunds. Student registrations must be
accompanied by proof of full-time student status.
Registration forms and inquiries should be directed to:
AAAI-93 / IAAI-93
445 Burgess Drive
Menlo Park, California 94025-3496 USA
415-328-3123; Fax: 415-321-4457
Email ncai@aaai. org.
On-Site Registration will be located at the entrance of Exhibit Hall B
on the upper level of the Washington Convention Center, 900 Ninth
Street, NW, Washington, DC.
Registration hours will be Sunday, July 11 through Tuesday, July 15
from 7:30 am - 6:00 pm. On Wednesday, July 14 and Thursday, July 15,
hours will be 8:00 am - 5:00 pm. All attendees must pick up their
registration packets for admittance to programs.
Child Care Services
Child care services are available from Lipton Corporate Child Care
Center, 655 15th Street., NW, Washington, DC 20005, (202) 416-
6875. Lipton Corporate Child Care Center is located one block from the
White House and approximately one mile from the Washington Convention
Center. Lipton is a fully licensed child care center used by corporate
individuals and conference attendees in the Washington DC area. Rates are
$12 per hour. Reservations must be made in advance, and directly with
Lipton Corporate Child Care Center.
(This information for your convenience, and does not represent an
endorsement of Lipton Corporate Child Care Center by AAAI.)
Housing
AAAI has reserved a block of rooms in Washington DC properties at
reduced conference rates. To qualify for these rates, housing reservations
must be made with the Washington DC Convention & Visitors Bureau
housing office. The deadline for reservations is June 8, 1993.
To make housing reservations, no form is required. You must call the
Housing Bureau :
Metropolitan Washington Area:
1-202-842-2930
US & Canada 1-800-535-3336
International Attendees 1-202-842-2930
Please have the following information available prior to calling for
reservations:
1. Name of convention attending
2. 1st and 2nd choice of hotel
3. Arrival/departure dates
4. Number of rooms required
5. Type of room (single, double, triple, quad)
6. Number of persons in party
7. Arrival rime
8. Credit card name, number and expiration date
9. Names of all occupants of room(s)
10. Address
11. Telephone number
All changes/cancellations prior to June 8, 1993 should be made
directly with the Housing Bureau. After this date, please contact the
HOTEL directly with any changes or cancellations. Room cancellations
must be received by assigned hotel at least 72 hours prior to arrival for
refund.
Confirmation will be sent to you from the Housing Bureau. A deposit is
not required if a credit card number has been given. In order to guarantee
a reservation without a credit card, send deposit amount indicated on the
confirmation form directly to the hotel within 15 days of receipt of the
confirmation.
International Attendees
International attendees may call the metro area number listed above or
fax reservation requests to 202-789-7037. (Please, only international
fax requests will be honored. US and Canadian attendees must use the
metro area or "800" number to make reservations.)
Mailing address for International Attendees ONLY:
AAAI Housing Bureau
1212 New York Avenue, NW, 6th Floor
Washington, DC 20005
Confirmation will be sent to you from the Housing Bureau. A deposit is
not required if a credit card number has been given. In order to guarantee
a reservation without a credit card, send deposit amount indicated on the
confirmation form directly to the hotel within 15 days of receipt of the
confirmation.
All changes/cancellations prior to June 8, 1993 should be made
directly with the Housing Bureau. After this date, please contact the
HOTEL directly with any changes or cancellations. Room cancellations
must be received by assigned hotel at least 72 hours prior to arrival for
refund.
Headquarters Hotel:
Grand Hyatt-Washington
1000 H Street, NW
Washington, DC 20001
Single: $121.00
Double: $141.00
Triple: $165.00
Quad: $190.00
Distance to Center: Across the street
Holiday Inn Crowne Plaza
Metro Center
775 12th Street, NW
Washington, DC 20005
Single: $105.00
Double: $125.00
Triple: n/a
Quad: n/a
Distance to Center: One block
Hotel rooms are priced as singles (1 person, 1 bed), doubles (2 persons,
2 beds), triples (3 persons, 2 beds) or quads (4 persons, 2 beds).
Student Housing
AAAI has reserved a block of dormitory rooms at George Washington
University for student housing during the conference. The University is
located within two blocks of a MetroRail station that can take you directly
across the street from the Washington Convention Center. Accommodations
include linen service.
Rates per person
- Two-three persons per room:
$25 per person per night
- One person per room:
$40 per person per night
Additional nights can be reserved based upon availability. No meals are
included.
Student housing reservations and deposits must be received by the AAAI
office no later than June 11, 1993. Prepayment of housing fees is
required. Checks, international money orders, bank transfers, and
traveler's checks must be in US currency. American Express,
MasterCard, and Visa, are also accepted. Refund requests must be made in
writing. The deadline for refund requests is June 18, 1993. A $50
processing fee will be assessed for all refunds.
Student housing is restricted to full-time graduate or undergraduate
students enrolled in a accredited college or university program. Proof of
full-time status must accompany the student housing form. Housing forms
and inquiries should be directed to:
AAAI-93/IAAI-93
445 Burgess Drive
Menlo Park, California 94025-3496 USA
415-328-3123; FAX: 415-321-4457
email: ncai@aaai.org
Air Transportation and Car Rentals
The American Association for Artificial Intelligence has selected United
Airlines as the official airline carrier and Hertz Rental Car as the official
car rental agency. You can make your airline and car rental reservations
by calling the United Airlines Specialized Meeting Reservations Center
directly at 800-521-4041 or through any travel agent. Be sure to
specify that you are traveling to the AAAI National Conference and identify
our reference # 538AG. By using this reference number, you will also
qualify for conference discounts on your Hertz Rental Car. For on-site
travel needs, The Grand Hyatt Hotel has airline ticket information
available at the concierge's desk located in the hotel lobby.
Car Rental Terms and Conditions:
Applicable charges for taxes, optional refueling service, PAI, PEC and LIS
are extra. Optional LDW may be purchased at $13 or less per day. Rates
are nondiscountable nor usable with any promotion or coupon, with the
exception of Hertz gift certificates with PC#27952. Rentals are subject
to Hertz minimal rental and driving age of twenty-five, driver's license
and credit requirements as well as car availability. Weekend rentals are
available for pick-up between noon Thursday and noon Sunday and must
be returned no later than 11:59 pm Monday. Weekend minimum rentals
are as follows: Thursday pick-ups, three days; Friday pick-ups, two
days; Saturday and Sunday pick-ups, one day. Weekly rentals are
available from five to seven days and must be kept over a Saturday night.
Saturday Night Keep rentals are available for pick-up any day of the week
and must be kept over a Saturday night.
Ground Transportation
The following information provided is the best available at press time.
Please confirm fares when making reservations.
Airport Connections:
Several companies provide service from Washington Dulles Airport and
Washington National Airport to downtown Washington, DC. A sampling of
companies and their one-way rates are shown below. Contact the company
directly for reservations.
Washington Flyer
703/892-6800
Washington Dulles International to downtown Washington, DC
Fare: $15; $24 round trip
Washington National Airport to downtown Washington, DC
Fare: $7; $12 round trip
DC, MD and VA cabs
From Washington National Airport to downtown Washington, DC
Fare: approximately $8 (Local cab listings available at Washington
National Airport)
Parking
There is no parking available at the Washington Convention Center.
However, parking facilities are available across the street from the
Convention Center located at the establishments listed below.
The Grand Hyatt Hotel
1000 H Street, NW
Parking is available for both guests and non-guests of this hotel at
$12.00 a day.
Atlantic Garage, Inc.
Techworld 999 Ninth Street
Parking is available at $9.00 a day.
Bus
The Washington, DC Greyhound/Trailways terminal is located at First & L
Streets, approximately eight blocks from the Convention Center.
301/565-2662.
Rail
Amtrak has more than fifty trains daily that link Washington to
Baltimore, Philadelphia, New York, and other east coast cities. The
Historic Union Station Amtrak depot is located at Massachusetts Avenue
and North Capitol Street, approximately ten blocks from the Convention
Center. For Amtrak reservations or information, call 800-872-7245.
Metrorail
Washington's modern subway system is known as the Metrorail. Visitors
can reach most major attractions in the city on the Metrorail, using the
easy-to-follow directional displays at the stations. The system also
provides rail links to the Maryland and Virginia suburbs. To travel on the
Metrorail, just purchase a farecard at a vending machine in any of the
stations. Trains operate every 10 minutes on average from 5:30 am until
Midnight, Monday through Friday. Weekend hours are 8 am until midnight
on Saturday, and 10 am until midnight on Sunday. For Metrorail or
Metrobus--providing connecting bus service--information, call 202-
637-7000.
Disclaimer
In offering United Airlines, Hertz Rental, Lipton Corporate Child Care,
George Washington University and all other service providers
(hereinafter referred to as "Supplier(s)" for the Innovative Applications
Conference and the National Conference of Artificial Intelligence, AAAI
acts only in the capacity of agent for the Suppliers which are the
providers of the service. Because AAAI has no control over the personnel,
equipment or operations of providers of accommodations or other services
included as part of the AAAI-93 or IAAI-93 program, AAAI assumes no
responsibility for and will not be liable for any personal delay,
inconveniences or other damage suffered by conference attendees which
may arise by reason of (1) any wrongful or negligent acts or omissions
on the part of any Supplier or its employees, (2) any defect in or failure
of any vehicle, equipment or instrumentality owned, operated or
otherwise used by any Supplier, or (3) any wrongful or negligent acts or
omissions on the part of any other party not under the control, direct or
otherwise, of AAAI.
Area Attractions
US Capitol
National Mall (East End). 202-224-3121; 202-225-6827
See the Rotunda, Statuary Hall with two statues from each state, the House
of Representatives, the Senate, and the original Supreme Court Chamber.
Open daily, 9 am through 8 pm. Tours 9 am through 3:45 pm. Free.
The Smithsonian Institution
202-357-1729
This is the world's largest museum complex with fourteen museums,
including the National Air and Space Museum and the National Zoo. All
museums are open daily, 10 am through 5:30 pm. Admission is free to all
Smithsonian Museums. Free walk-in "highlights" tours available to the
public in most museums.
The White House
1600 Pennsylvania Avenue, NW. 202-456-2200 or 202-456-7041
This is the home of every US President since 1800. Five of the mansion's
132 rooms are open to the public: the East Room, the Green Room, the
Blue Room, the Red Room, and the State Dining Room. Open Tuesday-
Saturday, 10 am to noon. No tickets are needed Labor Day to Memorial
Day. Congressional guided tours are available through your
Congressperson or Senator; to arrange a tour, call your Representative or
Senator's local office.
The Library of Congress
10 First Street, SE. 202-707-6400
The Library of Congress offers tours every Monday through Friday, 9 am
to 4 pm. Free.
Washington, DC Visitor Information Center
1455 Pennsylvania Avenue, NW. 202-789-7038
The Visitor Information Center is located one block from the White House.
Visitors can pick up a variety of free literature and maps and receive
advice on what to see and do while in the nation's capital. Operated by the
Washington, DC Convention and Visitors Association. Open Monday through
Saturday, 9 am through 5 pm. Closed Sundays and major holidays.
Tech 2000
800 K St., NW, #60 (located within the Tech World Complex). 202-
842-0500
Tech 2000 is the world's first permanent gallery of interactive
multimedia containing more the 60 "hands-on" exhibits. Open Tuesday -
Sunday, 11 am - 5 pm. Admission: $5 adult; $4 student; $3 children
under 12 and senior citizens.
PLEASE SEND ALL INFORMATION REQUESTS TO NCAI@AAAI.ORG or call
(415) 328-3123 if you need further information, a conference
order form, or if you wish to register with your credit card.
------------------------------
End of ML-LIST (Digest format)
****************************************
