Copy Link
Add to Bookmark
Report
AIList Digest Volume 4 Issue 095
AIList Digest Monday, 21 Apr 1986 Volume 4 : Issue 95
Today's Topics:
Seminars - Learning Robots, Approximate Theories (Rutgers) &
A Localized Model of Concurrency (SRI) &
Analogical Representations in Naive Physics (Edinburgh) &
Learning Apprentice Systems (SU) &
A Theory of Diagnosis (SU) &
A Formal Logic for Planning (UPenn) &
Editorial Comprehension in Op-Ed (UTexas) &
Run-Length Code for Geographical Information (SMU),
Conference - Symbolics User Group National Symposium &
Workshop on Engineering Design &
ACM SIGMOD & Design Automation & Computers and Math
----------------------------------------------------------------------
Date: 15 Apr 86 11:03:24 EST
From: PRASAD@RED.RUTGERS.EDU
Subject: Seminar - Learning Robots, Approximate Theories (Rutgers)
MACHINE LEARNING COLLOQUIUM
Learning Robots as Users and Refiners of Approximate Theories
Tom Mitchell
Rutgers University
11 AM, April 29, 1986
#423, Hill Center
This talk will describe some recent (and fairly tentative) research
toward building a learning robot. The robot is viewed as having an
approximate theory of its world, which it uses to guide problem
solving, and which is in turn refined as the robot gains experience.
The initial theory may contain fairly abstract assertions such as
"executing motor commands causes changes in the configuration of parts
of oneself", "coming into physical contact with a rigid object often
causes changes in its position", and "changes in the configuation of
physical objects correlate with changes in the visual appearance of
the object". This abstract theory is used by the robot to construct
PLAUSIBLE plans for achieving its goals. When these plans are
executed, the world provides feedback--training data which is useful
for refining the theory. This training data is generalized by a
combined explanation based/empirical method (the approximate theory is
used to contruct plausible explanations which are verified and refined
empirically). Tom Fawcett and I have recently begun implementing parts
of this system, but many open research issues remain.
------------------------------
Date: Wed 16 Apr 86 11:59:17-PST
From: LANSKY@SRI-AI.ARPA
Subject: Seminar - A Localized Model of Concurrency (SRI)
A LOCALIZED MODEL OF CONCURRENCY
Fernando Pereira (PEREIRA@SRI-AI)
SRI International, AI Center
11:00 AM, MONDAY, April 21
SRI International, Building E, Room EJ228 (new conference room)
In this talk I will give an informal overview of a structural theory
of concurrency that I have been developing with Luis Monteiro. The
main goal of our theory is to model the way in which local
interactions between components of a system lead to global behavior.
The theory, which is based on the mathematical concept of sheaf,
allows us to model precisely the idea of processes interacting
through common behavior at shared locations. In contrast to
behavioral models, ours keeps track of the individual contributions
of subsystems to overall system behavior, allowing a finer-grained
analysis of subsystem interactions.
>From event signatures that specify relations of independence and exclusivity
between events, we construct spaces of locations where activity may occur.
Behaviors are then modeled as elements of sheaves of monoids over those
spaces and processes as certain sets of behaviors. The construction of the
model, and in particular its avoidance of interleaving, gives it very
convenient mathematical properties --- sheaves of behavior monoids are to
event signatures what free monoids are to alphabets. The theory also allows
us to identify on purely structural grounds event signatures with a
potential for deadlock.
Time permitting, I will engage in rambling speculation as to possible
applications of the theory.
VISITORS: Please arrive 5 minutes before the seminar (10:55), as
you must now be escorted from the reception desk.
------------------------------
Date: Mon, 14 Apr 86 11:07:53 GMT
From: Gideon Sahar <gideon%edai.edinburgh.ac.uk@cs.ucl.ac.uk>
Subject: Seminar - Analogical Representations in Naive Physics (Edinburgh)
EDINBURGH AI SEMINARS
Date: Wednesday 16th April 1986
Time: 2.00 p.m.
Place: Department of Artificial Intelligence
Seminar Room - F10
80 South Bridge
EDINBURGH EH1 1HN.
Professor Bernard Meltzer, Joint Research Centre, Ispra Establishment, Italy w
will give a seminar entitled - ``Analogical Representations in
Modelling Naive Physics".
Ideas and experimental results will be presented on the use of
analogical representations of knowledge, in Sloman's sense, that is,
ones which bear a structural similarity to what is represented. This
was done for the qualitative modelling of the everyday behaviour of
objects and substances like strings, liquids and gases, represented by
pixel sets built up from message-passing between adjacent base
elements. These messages embody a very small number of local
constraints derived from naive observation such as material continuity
and non-copenetrability.
Based as they are on fundamental phenomenological properties of the
physical world, these programs turned out to have capacities for
solving other problems than those for which they were designed.
The use of such programs in integrated reasoning and problem-solving
systems, and the relationship of this approach to those of classical
physics and current AI ones in qualitative physics will also be
discussed.
------------------------------
Date: Wed 16 Apr 86 10:10:57-PST
From: Anne Richardson <RICHARDSON@SU-SCORE.ARPA>
Subject: Seminar - Learning Apprentice Systems (SU)
DAY: April 22, 1986
EVENT: CS 520 AI Seminar
PLACE: Terman Auditorium
TIME: 11:00
TITLE: Learning Apprentice Systems
PERSON: Tom Mitchell
FROM: Rutgers University
This talk introduces a class of knowledge-based systems called
Learning Apprentices: systems that provide interactive aid in solving
some problem and acquire new knowledge by observing the actions of
their users. The talk focuses on a particular Learning Apprentice,
called LEAP, which is presently being developed in the domain of
digital circuit design. By analyzing circuit fragments contributed by
its users, LEAP infers rules that allow it to recommend similar
circuits in subsequent cases. We discuss the type of problem solving
architecture, knowledge organization, and learning methods required to
support such learning apprentices in a variety of domains.
------------------------------
Date: 16 Apr 86 1742 PST
From: Vladimir Lifschitz <VAL@SU-AI.ARPA>
Subject: Seminar - A Theory of Diagnosis (SU)
A THEORY OF DIAGNOSIS FROM FIRST PRINCIPLES
Raymond Reiter
Department of Computer Science, University of Toronto
and
The Canadian Institute for Advanced Research
Thursday, April 17, 4pm
MJH 252
Suppose given a description of a system, together with an
observation of the system's behaviour which conflicts with the way
the system is meant to behave. The diagnostic problem is to determine
those components of the system which, when assumed to be functioning
abnormally, will explain the discrepancy between the observed and
correct system behaviour.
We propose a general theory for this problem.The theory requires
only that the system be described in a suitable logic. Moreover, there
are many such suitable logics, e.g., first order, temporal, dynamic,
etc. As a result, the theory accomodates diagnostic reasoning in a wide
variety of practical settings, including digital and analogue circuits,
medicine, and database updates. The theory leads to an algorithm for
computing all diagnoses, and to various results concerning principles
of measurement for discriminating between competing diagnoses. Finally,
the theory reveals close connections between diagnostic reasoning and
non-monotonic reasoning.
------------------------------
Date: Wed, 16 Apr 86 14:22 EST
From: Tim Finin <Tim%upenn.csnet@CSNET-RELAY.ARPA>
Subject: Seminar - A Formal Logic for Planning (UPenn)
COLLOQUIUM
3pm Thursday, April 17, 1986
216 Moore School - University of Pennsylvania
A FORMAL LOGIC THAT SUPPORTS PLANNING WITH EXTERNAL EVENTS
AND CONCURRENT ACTIONS
Richard Pelavin - University of Rochester
A formal logic will be presented that provides a foundation for a theory of
plans in temporally rich domains. These domains include actions that occur
over intervals that may overlap in time. Thus, we can represent plans with
concurrent actions. We also can treat domains with external events, i.e.
actions by other agents and natural forces, that the planner may need to
interact with. These interactions include the prevention of an event, the
assurance of the successful completion of an event, and the performance an
action that is enabled by some external event.
The logic is an extension of a linear time logic (Allen's interval logic) with
a modal operator expressing temporal possibility and a counterfactual-like
modality that can be used to encode what can and cannot be done by the planning
agent. The semantic model consists of a set of possible worlds related by two
accessibility relations in terms of which the modalities are interpreted. The
approach of interpreting a counterfactual-like modality in terms of an
accessibility relation derives from Lewis' and Stalnaker's semantic theories of
conditionals.
------------------------------
Date: Thu, 17 Apr 86 11:09:17 CST
From: Rose M. Herring <roseh@ratliff.CS.UTEXAS.EDU>
Subject: Seminar - Editorial Comprehension in Op-Ed (UTexas)
University of Texas
Computer Sciences Department
COLLOQUIUM
SPEAKER: Sergio Alvarado
University of California at Los Angeles
TITLE: Editorial Comprehension in OpEd Through Argument Units
DATE: Tuesday, April 22, 1986
PLACE: TAY 3.144
TIME: 11 - 12 noon
OpEd (Opinions to/from the Editor) is a computer program
that reads short polito-economic editorial segments and answers
questions about their contents. For OpEd, understanding editori-
als involves: (1) applying a large amount of domain-specific
knowledge; (2) recognizing beliefs and belief relationships; (3)
following reasoning about plans and goals; (4) applying abstract
knowledge of argumentation; (5) mapping text into conceptual
representation; and (6) indexing recognized concepts for later
retrieval during question answering.
Here, I discuss OpEd's abstract knowledge of argumenta-
tion. In OpEd, knowledge of argument structure is organized by
memory structures called Argument Units (AUs). These structures
package belief support and attack relationships and reasoning
chains. When combined with domain-specific knowledge, AUs can be
used to understand and generate arguments involving plans, goals,
and beliefs. Thus, argument comprehension is viewed in OpEd fun-
damentally as the process of accessing and instantiating these
units.
A description of OpEd's architecture and examples of its
current input/output behavior are also presented.
COFFEE AT 10:30 in TAY 3.128
------------------------------
Date: WED, 10 JAN 84 17:02:23 CDT
From: E1AR0002%SMUVM1.BITNET@WISCVM.WISC.EDU
Subject: Seminar - Run-Length Code for Geographical Information (SMU)
A Spatial Knowledge Structure Based on Run-Length-Code for a Geographical
Information System
Speaker: Erland Jungert
Illinois Institute of Technology
Location: 315SIC, Southern Methodist University, CS
Time: 3PM
Run-Length-Code (RLC) is an example of a simple data structure used
mainly for compacting images. A method where RLC is used as an object
oriented data structure for Geographical Information Systems (GIS) will be
presented. The usage of this object structure as a basis for spatial
reasoning while regarding the RLC-objects as part of a spatial knowledge
structure will be discussed.
------------------------------
Date: Sun, 20 Apr 86 14:45:46 pst
From: grover@aids-unix (Mark Grover)
Subject: Conference - Symbolics User Group National Symposium
Registration materials are now available to the National Symposium of the
Symbolics Lisp Users Group, to be held June 2-6, 1986 at Georgetown University
in Washington, DC. Both tutorials and technical sessions will be held. The
theme of this year's Symposium is "Programming in Style". Many interesting
and exciting guests are expected. Registration materials and housing
information can be obtained via telephone or US mail to:
Symbolics National Symposium
Attn: Annmarie Pittman
655 15th St. NW #300
Washington, DC 20005
(202) 639-4228
------------------------------
Date: Thu, 17 Apr 86 08:40:33 -0500
From: sriram@ATHENA.MIT.EDU
Subject: Conference - Workshop on Engineering Design
AAAI-86 WORKSHOP ON
KNOWLEDGE-BASED EXPERT SYSTEMS FOR ENGINEERING DESIGN
In the 60's, AI researchers explored weak methods applicable to a very
broad class of problems. In the 70's, we created knowledge-intensive,
"strong" methods for solving quite specific types of problems. A major
trend in the 80's is to identify coherent problem classes of
intermediate generality; the proof of coherence is in the further
identification of correspondingly general problem-solving methods. For
instance, "classification problems" have been defined, a general but
still knowledge-based classification problem-solving process and
system architecture have been laid out, and tools exist for
facilitating development of classification systems.
Design problems also appear to constitute a coherent problem class. At
present, however, we are only beginning the enterprise of: defining
this class; formalizing a model of the design problem-solving process
and design system architecture; and creating tools for developing
design systems.
To model (and ultimately facilitate) human designers and their
enormous flexibility in terms of conventional AI "primitives" requires
integrating such diverse functions as refinement techniques,
constraint reasoning, and goal satisfaction, and encoding these
functions in such varied forms as rules, heuristics, and algorithms.
Viewing design tools as "knowledge-based expert systems" provides a
framework for capturing such diversity.
The purpose of this workshop is to provide a forum in which both
engineers and computer scientists can discuss knowledge-based
frameworks for organizing and developing useful engineering design
systems.
TOPICS TO BE DISCUSSED
1. Definition of "a design problem".
2. A general model of the design process.
3. Knowledge representation formalisms for design.
4. Problem-solving strategies required for design.
5. Existing frameworks and for design.
6. Existing architectures for design aids.
7. Capabilities and tools desired by engineers.
8. Automated vs.interactive design aids.
9. Software environments and tools for developing design aids.
ORGANIZERS
Sriram [sriram@athena.mit.edu] and Chris Tong [tong@red.rutgers.edu]
PARTICIPATION
The workshop will take place on Monday, August 11, at the University
of Pennsylvania. Participation in the workshop is by invitation,
limited to 35 participants. Those wishing to be invited should submit
four copies of a 1000-word abstract describing their work in AI and
engineering design to Sriram, 1-253b, Dept. of Civil Engineering,
M. I.T., Cambridge, MA 02139 OR to Chris Tong, Dept. of Computer
Science, Rutgers University, New Brunswick, NJ 08903. The deadline
for application is May 30, 1986. Invitations will be issued by July 1.
------------------------------
Date: WED, 10 JAN 84 17:02:23 CDT
From: E1AR0002%SMUVM1.BITNET@WISCVM.WISC.EDU
Subject: Conferences - ACM SIGMOD & Design Automation & Computers and Math
1986 ACM SIGMOD international Conference on the management of data
May 28-30 1986 Washington DC
Session 6a Logic and Databases 11:00 - 12:30 Thursday May 30
A. Van Gelder "A Message Passing Framework for Logical Query Evaluation"
A. Rosenthal, S. Heller, U. Dayal, F. Manola "Traversal Recursion:
A Practical Approach to Supporting Recursive Applications:
G. Gardarin C. DeMaindreville "Evaluation of Databse Recursive Logic
as Recurrent Function Series"
Session 7 Query Processing Thursday May 30 2:00 - 3:30
J. C. Freytag "Rule Based Transformation of Relational Queries into
Interactive Programs"
Session 9a Rule Based Systems
M. T. Harandi T. Schang S. Cohen "Rule Base Management Using Meta
Knowledge"
T. Imelinksi "Query Processing in Deductive Databases with Incomplete
Information"
Q. Chu "A Rule-Based Object/Task Modelling Approach"
__________________________________________________________________________
Twenty Third ACM/IEEE Design Automation Conference
June 29-July 2, 1986 Las Vegas, Nevada
Session 4 Intelligent Systems Time Monday 10:30 - 12:00
An Expert System Paradigm for Design
Forrest D. Brewer, Daniel D. Gajaski
University of Illinois at Urbana
Session 12 Timing Verification Monday Monday 4:00 - 5:30
Reasoning About Digital Systems Using Temporal Logic
G. Venkatesh
Session 14 Test Generation Techniques Tuesday 8:00 - 10:00
A Heuristic Chip-Level Test Generation Algorithm
Daniel S. Barclay, James R. Armstrong
Virginia Polytechnic Institute
Session 29 Hardware Design Languages Tuesday 3:30 - 5:30
A Design Rule Database System to Support Technology Adaptable Applications
Hilary J. Kahn, J. S. Aude
University of Manchester
Session 34 Expert Systems for Design Automation Wednesday 8:00 - 10:00
A Rule-Based Logic Circuit Synthesis System for CMOS Gate Arrays
Takao Saito, Hiroyuki Sugimoto, Masami Yamazaki, NObuaki Kawato
Fujitsu Labs
FLUTE - A Floorplanning Agent for Full Custom VLSI Design
Hityuki Wantanabe, Bryan Ackland
AT&T Bell Laboratories, Holmdel NJ
Knowledge-Based Optimal Ill Circuit Generator From Conventional Logic
Descriptions
T. Watanabe, T. Masuishi, T. Nishiyama, N. Horie
Hitachi
PEARL: An Expert System for Power Supply Layout
Ed DeJesus
DEC
Session 38 Short Papers: Representing and Manipulating VLSI Design
Wednesday 10:30 - 12:00
Precedent Based Reasoning About VLSI Structures
Richard H. Lasthrop, Robert S. Kirk MIT and Gould AMI respectively
A Frame Based System for Representing Knowledge About VLSI Design
Hassan K. Reghbati, W. Stephen ADolph, Amar Sanmugasundam
Simon Fraser University
Session 39 Timing Verification Wednesday 10:30 - 12:00
A Rule Based Approach to Unifying Functional and Fault Simulation and
Timing Verification
Sujmit Ghosh
AT & T
Session 42 Database II Wednesday 1:30 - 3:30
Rules-Based Object Clustering: A Data Structure for Symbolic VLSI
Synthesis and Analysis
Robert P. Larsen Rockwell International Corporation
__________________________________________________________________________
A Conference on Computers and Mathematics July 30 - August 1
Stanford University
Woodrow Bledsoe
Automated Theorem Proving and Artificial Intelligence
Rudiger Loos
Tarski's Dream
------------------------------
End of AIList Digest
********************
