Copy Link
Add to Bookmark
Report
AIList Digest Volume 6 Issue 089
AIList Digest Thursday, 5 May 1988 Volume 6 : Issue 89
Today's Topics:
Seminars - Butterfly Lisp (NASA) &
Joshua Database System (SRI) &
Concept Acquisition in Noisy Environments (CMU) &
Computational Models in AI (CMU) &
Expert Database Systems (SRI),
Conferences - 22nd Carnegie Symposium on Cognition (CMU) &
WESTEX-88 Conference on Expert Systems
----------------------------------------------------------------------
Date: Fri, 29 Apr 88 11:14:49 PDT
From: CHIN%PLU@ames-io.ARPA
Subject: Seminar - Butterfly Lisp (NASA)
National Aeronautics and Space Administration
Ames Research Center
SEMINAR ANNOUNCEMENT
SPEAKER: Seth Steinberg
TOPIC: Butterfly Lisp
The BBN Butterfly is a shared memory multiprocessor which can be
configured with up to 256 processors. Over the last several years we
have developed a Lisp system which takes advantage of this hardware,
by providing for the parallel execution of lightweight tasks in a
shared Lisp world. Tasks are created using the future mechanism which
automatically provides for data directed task synchronization. Other
tasking and synchronizing techniques may be used as well.
In an attempt to understand how parallel programs execute, we installed
a tracing system to record the fine details of tasking and
synchronization. This data is then used to produce a picture showing
both the task creation tree and the task synchronization tree. These
pictures have shed light on the behavior of a number of parallel
programs.
Butterfly Lisp can be described as locally serial, since, whenever
possible, sections of code which contain no parallel constructs will
execute as they would on a serial processor. Preserving this
property while combining such features as tasking, unwind protection
and special variables has required us to make a number of changes in
the way these features are implemented.
The system is now being ported to run on the BBN GP-1000 processor
under the Mach operating system and a new compiler is being tested
which takes advantage of Common Lisp type declarations.
BIOGRAPHY:
Seth Steinberg has been working with computers since the late '60's
and has an M.S. in Computer Science from MIT. He spent seven years
working for Nicholas Negroponte doing systems programming, language
development and graphics at the Architecture Machine Group. (Now the
nucleus of the Arts and Media Technology Center) In 1979, he joined
Software Arts where he designed and developed TK!Solver, a constraint
relaxation system for personal computers. More recently he has been
working on a parallel Lisp implementation for the BBN Butterfly
multiprocessor.
================================================================
DATE: Monday, TIME: 2:00 - 3:00 pm BLDG. 244 Room 103
May 2, 1988 --------------
POINT OF CONTACT: Marlene Chin PHONE NUMBER: (415) 694-6525
NET ADDRESS: chin%plu@ames-io.arpa
***************************************************************************
VISITORS ARE WELCOME: Register and obtain vehicle pass at Ames Visitor
Reception Building (N-253) or the Security Station near Gate 18. Do not
use the Navy Main Gate.
Non-citizens (except Permanent Residents) must have prior approval from the
Director's Office one week in advance. Submit requests to the point of
contact indicated above. Non-citizens must register at the Visitor
Reception Building. Permanent Residents are required to show Alien
Registration Card at the time of registration.
***************************************************************************
------------------------------
Date: Fri 29 Apr 88 15:18:06-PDT
From: NIELSEN@KL.SRI.COM (Norm R. Nielsen)
Subject: Seminar - Joshua Database System (SRI)
Information Industries Divisional Seminar
Joshua: A System That Provides
Syntactically Uniform Access to Heterogeneously
Implemented Data Bases
Steve Anthony
Symbolics
May 11 at 10 AM, BS-208
Joshua is a system developed at Symbolics that provides
syntactically uniform access to heterogeneously implemented
knowledge bases. Its power comes from the observation that
there is a "protocol of inference" consisting of a small set
of abstract actions, each of which can be implemented in many
ways. The object-oriented programming facilities of Flavors
have been used to control the choice of implementation. Each
statement in the language represents an instance of a class
identified with that statement's predicate. Steps of the
protocol are implemented by methods inherited from the
classes. Since inheritance of protocol methods is a compile-
time operation, very fine-grained control can be achieved
with little run-time cost.
Joshua offers two major advantages. First, a Joshua
programmer can easily change his or her program to use more
efficient data structures without changing the rule set or
other knowledge-level structures. Second, it is easy to
build interfaces which incorporate existing tools into
Joshua, without having to modify those tools.
Steve Anthony will discuss the capabilities and design of
Joshua, followed by some demonstrations. We will be meeting
in the Intelligent System Laboratory's computer room so that
the demonstrations can be run live on the Symbolics 3670.
------------------------------
Date: Sun, 01 May 88 00:49:10 EDT
From: Anurag.Acharya@CENTRO.SOAR.CS.CMU.EDU
Subject: Seminar - Concept Acquisition in Noisy Environments (CMU)
Dates: 9-May-88
Time: 11:30 am
Place: WeH 7220
Type: Machine Learning Seminar
Duration: 1 hr.
Who: Francesco Bergadano (CS Dept., U. di Torino, Italy)
Automated Concept Acquisition in Noisy Environments
F. Bergadano, A. Giordana, L. Saitta
Computer Science Dpt
Universita di Torino, Torino, Italy
A learning method for concept acquisition from examples
in noisy environments is presented. The learned knowledge
is expressed in the form of production rules, organized into
separate clusters and linked together in a graph structure;
A continuous-valued semantics is associated
to the description language and each rule is affected
by a certainty factor.
The learning process is guided by a top-down control
strategy, through successive specialization steps. Search
is strongly focalized by task-oriented heuristics and
by the available domain knowledge. The methodology
been tested on a problem in the field of speech recognition,
and the obtained results are presented and discussed.
Mr. Bergadano is also interested in the following topics:
inductive learning, explanation based learning, integration of inductive
and deductive approaches to Machine Learning, and applications of learning
systems to Pattern Recognition, and would like to meet people working
in these areas.
------------------------------
Date: Sun, 01 May 88 00:54:53 EDT
From: Anurag.Acharya@CENTRO.SOAR.CS.CMU.EDU
Subject: Seminar - Computational Models in AI (CMU)
THEORY/AI SEMINAR
Jiawei Hong, Courant Institute
Friday, May 6
2:00 p.m.
4605 WEH
Connectionist and Other Computational Models in AI
This talk consists of three problems:
1. The well known connectionist models defined by an nxn real weight matrix
(notice that an arbitrary real number may have infinite information in it),
can be simulated by a non-uniform circuit of O(n^3 log n) Boolean gates
(thus the total information is finite) with time slowdown O(log n). Therefore
the connectionist model do not have more computational power than other
parallel computation hardwares.
2. In the future, computers may consist of digital elements as well as analogue
elements. Which kind of analogue element does help? We prove that a kind of
analogue element does help only if
(1). the analogue element has very very high precision: exponentially many
significant bits, or
(2). the analogue element can very efficiently compute a problem which is
NOT in NC. Both are unlikely true.
3. Can human brains be simulated by computers in the future? Under some rea-
sonable assumptions, we proved that this is possible. (there is no intrinsic
difficulty, like NP-hardness.) We are optimistic.
------------------------------
Date: Mon, 2 May 88 09:52:23 PDT
From: seminars@csl.sri.com (contact lunt@csl.sri.com)
Subject: Seminar - Expert Database Systems (SRI)
SRI COMPUTER SCIENCE LAB SEMINAR ANNOUNCEMENT:
A FLARE FOR EXPERT DATABASE SYSTEMS IN STARBURST
Shel Finkelstein
IBM Almaden Research Center
Monday, May 9 at 4:00 pm
SRI International, Conference Room B, Building A
Expert systems technology is being integrated into standard data
processing environments. These environments have requirements not
typically emphasized by expert systems, including storage management,
shared access/update of data and knowledge, efficient access to data and
knowledge, persistent data and authorization. Database systems meet
these requirements. On the other hand, database systems do not have the
representation and search capabilities required for certain
applications. Expert systems have these capabilities.
The goal of the Starburst project at IBM Almaden Research Center is to
do exploratory systems research towards building a portable, extensible,
distributed relational database management system. One extension that
is a new direction (called "Flare") in Starburst is for expert database
systems. In this talk we discuss the relationship between expert
systems and database systems and describe some areas for research in
expert database systems in Starburst.
NOTE FOR VISITORS TO SRI:
Please arrive at least 10 minutes early in order to sign in and
be shown to the conference room.
SRI is located at 333 Ravenswood Avenue in Menlo Park. Visitors
may park in the visitors lot in front of Building A (red brick
building at 333 Ravenswood Ave) or in the conference parking area
at the corner of Ravenswood and Middlefield. The seminar room is in
Building A. Visitors should sign in at the reception desk in the
Building A lobby.
IMPORTANT: Visitors from Communist Bloc countries should make the
necessary arrangements with Liz Luntzel (415) 859-3285 as soon as possible.
------------------------------
Date: Sun, 01 May 88 00:53:38 EDT
From: Anurag.Acharya@CENTRO.SOAR.CS.CMU.EDU
Subject: Conference - 22nd Carnegie Symposium on Cognition (CMU)
The 22nd Carnegie Symposium on Cognition will be on the topic of
Architectures for Intelligence. Talks will be in the Adamson wing,
as usual. Inclosed is a schedule. Contact Kurt VanLehn or Elaine
Benjamin (x4964, VanLehn@psy.cmu.edu) for more information.
--------------------------------------------------------------
Monday, May 16
8:30 Coffee
8:45 Welcome -- David Klahr, Head, Psychology Department
9:00 John Laird (U. Michigan), Allen Newell (CMU), Paul Rosenbloom (ISI
"Towards completing the Soar architecture"
10:00 Coffee
10:15 Michael Genesereth (Stanford) -- "Deliberate agents"
11:15 Barbara Hayes-Roth (Stanford) -- "Making intelligent systems adapt
12:15 Lunch
1:30 Tom Mitchell (CMU) -- "Theo: A framework for constructing
self-improving systems."
2:30 Rodney Brooks (MIT) -- "How to build creatures rather than
isolated cognitive imitators."
3:30 Coffee
3:45 Jamie Carbonell (CMU) -- TBA
4:45 Bill Clancey (Xerox IRL) -- "Intelligent architectures and knowledge
engineering: A commentary"
5:45 Adjourn
-------------------------------------------------------------------------
Tuesday, May 17
8:45 Coffee
9:00 John Anderson (CMU) "The status of cognitive architectures
in a rational analysis"
10:00 Coffee
10:15 Kurt VanLehn (CMU) "Flexibility and robustness in
the execution of cognitive procedures"
11:15 Geoff Hinton (Toronto) "The transition from serial to
parallel processing in a connectionist network"
12:15 Lunch
1:30 Walter Schneider and William Oliver (LRDC) -- "An instructable
connectionist/control architecture: Using rule based instructions
accomplish connectionist learning in a human time scale."
2:30 Jay McClelland (CMU) -- "Nature, nurture and connections:
Explorations in network architecture"
3:30 Coffee
3:45 Zenon Pylyshyn (W. Ontario) -- "Architectures and strong equivalence
A commentary"
4:45 Adjourn
------------------------------
Date: Tue, 3 May 88 15:04:30 PDT
From: Greg Jordan <gjordan@cirm.northrop.com>
Subject: Conference - WESTEX-88 Conference on Expert Systems
The third annual WESTEX conference, sponsored by the Western Committee
of the Computer Society of the IEEE and the IEEE Los Angeles Council,
will be held June 28-30, 1988 at the Anaheim Marriott Hotel, Anaheim,
California.
This year special emphasis will be given to management issues associated
with fielding successful applications. Over the past two years a great
deal of experience has been gained in the basic technology and early
development of expert systems. As a result, many systems have been
successfully prototyped for a broad range of applications. A common
problem facing the field today is transitioning from prototypes to
fielded solutions. WESTEX-88 will focus on the issues and problems
that must be solved in this transition.
The program will feature the internationally prominent speaker, Professor
Edward Feigenbaum, Stanford University, who will present new observations
and predictions regarding expert systems from his decades of leadership
and experience.
On Tuesday, June 28, three tracks of tutorials will be presented by
leading expert system practitioners: "Basic Concepts", "Advanced
Concepts" and "Special Topics".
A two-day program featuring well-known invited speakers and submitted
papers will be presented on Wednesday and Thursday, June 29 and 30.
Technical exhibits featuring expert system products, tools and
environments from commercial artificial intelligence hardware and
software development firms are an important part of the conference. The
exhibits provide potential users the opportunity to see and compare
first hand what is available and to make contact with interested vendors.
For exhibit information contact Martha B. Wolf, Electronics Conventions
Management, 8110 Airport Blvd., Los Angeles, CA 90045, or call
1-800-421-6816(US) or 1-800-262-4208 (CA only).
------------------------------
End of AIList Digest
********************
