Copy Link
Add to Bookmark
Report
AIList Digest Volume 6 Issue 048
AIList Digest Tuesday, 8 Mar 1988 Volume 6 : Issue 48
Today's Topics:
Queries - Approaches to AI & Prototypical Knowledge &
CL for an IBM running VM/CMS HPO 4.2,
AI Tools - CommonLoops & Image Formats & Student Versions of OPS5 &
Constraint Languages,
Theory - Uncertainty and Fuzzy Logic vs Probability
----------------------------------------------------------------------
Date: Sun, 6 Mar 88 16:35:30 EST
From: "Timothy J. Horton" <tjhorton%ai.toronto.edu@RELAY.CS.NET>
Subject: Re: Approaches to AI
jbn@GLACIER.STANFORD.EDU (John B. Nagle) writes (very roughly):
> McCarthy has recently described two paths to artificial intelligence.
>But his two, while the most active, are not the only ones in which substantial
>work is underway. A more general taxonomy might be outlined as follows:
> 1. "Good, old fashioned AI". ... to model the world using formalisms
> related to mathematical logic.
> 2. Neural networks. ... development of massively parallel
> self-organizing systems.
> 3. Engineered artificial life. (bottom-up approach) ... construction
> of robots that function in the real world, using whatever technology
> seems appropriate.
> 4. Study and replication of the detailed structure of biological
> intelligence, without necessarily understanding how it works.
Could anyone fill out this tree a little more?
For instance, what about Woods' work on abstract procedures (not to be
confused with proceduralism)? He wants something more general than logic
-- not throwing it out, but not accepting it as sufficient or appropriate
for the whole job.
What about anything else? Surely there are "mathematical" theoreticians
that hope for something more than logic, that ought to be included here?
The latest issue of "Computational Intelligence" had more than a score
of responses to Drew McDermott's critique of pure logicism, and one heck
of a lot of camps got staked out in the process.
------------------------------
Date: 7 Mar 88 05:05:31 GMT
From: daniel@aragorn.cm.deakin.OZ (Daniel Lui)
Reply-to: daniel@aragorn.OZ (Daniel Lui)
Subject: Request for related research work on Prototypical Knowledge
Can anybody tell me what research work has been done on Prototypical
Knowledge? So far, I found only two related publications:
Janice S. Aikins, "Prototypical Knowledge for Expert Systems",
Artificial Intelligence, Vol. 20, 1983.
Henrik Nordin, "Using Prototypes for Knowledge-Based Consultation and Teaching",
SPIE Vol. 635 Applications of Artificial Intelligence III, 1985.
Thanks in advance.
daniel@aragorn.oz
Daniel Lui >> CSNET:daniel@aragorn.oz <<
Division of Computing >> UUCP: seismo!munnari!aragorn.oz!daniel <<
Deakin University >> decvax!mulga!aragorn.oz!daniel <<
Geelong, Victoria >> ARPA: munnari!aragorn.oz!daniel@seismo.arpa <<
Australia 3217 >> decvax!mulga!aragorn.oz!daniel@Berkeley <<
------------------------------
Date: 23 Feb 88 14:27:20 GMT
From: msu.bitnet!13501jsk@psuvm.bitnet (John Kern)
Subject: Wanted CL for an IBM running VM/CMS HPO 4.2
I am currently working with LISP/VM on an IBM 3090 and I am not satisfied
with it. I would appreicate any information on Common LISP available for
this machine.
Sincerely,
John
------------------------------
Date: Sat, 05 Mar 1988 12:51 EST
From: sidney%acorn@oak.lcs.mit.edu
Subject: commonloops
I thought that this reply to Wei-Han Chu's request for information
would be of more general interest:
Portable CommonLoops (PCL) is available for free from Xerox PARC. Even
though Gold Hill includes a copy with our GCLisp 3.0, that is for the
convenience of our customers who would like a copy, and we do not make
any attempt to support it. PCL is evolving rapidly towards the
emerging CLOS standard, with new releases appearing frequently. It is
currently available for at least 9 Lisp implementations that I know
of. The most current source and documentation is available via
anonymous ftp from parcvax.xerox.com. The file /pub/pcl/get-pcl.text
contains more information. Requests for information about the
CommonLoops mailing list can be sent to commonloops-request@xerox.com.
Disclaimer: While I work for Gold Hill, this message is my own
personal reply to a request for information and is not the official
word from either Gold Hill Computers or anybody at Xerox PARC.
Sidney Markowitz <sidney%acorn@oak.lcs.mit.edu>
------------------------------
Date: 2 Mar 88 19:45:06 GMT
From: hao!noao!stsci!sims@ames.arc.nasa.gov (Jim Sims)
Subject: Re: File formats for image data ?
I tried to reply but our mailer is brain-dead...
Most astronomers use FITS, which allows multiple data types, multiple groups
and types of data in the same file, and extension to the format. Any
observatory or astronomy program at a university can get you the full scoop.
--
Jim Sims Space Telescope Science Institute Baltimore, MD 21218
UUCP: {arizona,decvax,hao,ihnp4}!noao!stsci!sims
SPAN: {SCIVAX,KEPLER}::SIMS
ARPA: sims@stsci.edu
------------------------------
Date: 3 Mar 88 19:24:19 GMT
From: ptsfa!pacbell!att-ih!alberta!ahmed@AMES.ARC.NASA.GOV (Ahmed
Mohammed)
Subject: Student versions of OPS5
Computer Thought Corp. has begun shipping students versions
of the expert system lang. These are operational on IBM PC/XT/AT
on one 5.25'' floppy drive
The system should accomplish for rule-based systems what
Borland's Turbo Prolog has already accomplished for logic
programming: mass distribution and increased public understanding.
The package costs $255 for Graduate student version,
$90 for undergraduate version.
------------------------------
Date: 07 Mar 88 15:08:30 EST (Mon)
From: sas@bfly-vax.bbn.com
Subject: AIList V6 #45 - Constraint Languages
I just got back from vacation and I cannot remember if I sent this out
before I left or not, but:
I was the lead engineer on the TK!Solver product back in the early
80's so I did a bit of constraint language research. Although
TK!Solver was (and still is) a software product, the only publications
are the manual itself and a number of software reviews. TK!Solver let
you enter equations and then use a Newton-Raphson solver to find
solutions. We tried to generalize the then popular MBA calculators
which let you manipulate PV, FV, T and/or i and given three out of
four compute the fourth. Inflation and interest rates were much
higher then and MBA's worried about IRR instead of market share.
TK!Solver was based on some work done by Milos Konopasek at U of
Manchester, Georgia Tech and NCSU. While I might be able to find some
copies of his papers, you could try looking them up under the name QAS
or the Question Answering System. He used it to help teach textile
engineering. TK!Solver made a number of improvements.
If you want to follow this vein you might try finding Bob Light's
MIT Mechanical Engineering thesis done in the mid 80's. He combined
back solution techniques with more traditional CAD rendering
techniques.
I am not sure if anyone mentioned the Sutherland's MIT doctoral thesis
which is kind of the grand daddy of constraint systems done in the
early 60's. The stressed trestle example in Borning's 80-81 Thinglab
paper (PARC, I think) originally appeared in this one.
I'd also recommend Guy Steele's MIT thesis on a discreet state
constraint system and Gosling's CMU thesis which uses a constraint
system to compile closed form algebraic solutions to constraint
problems.
Depending on you interests in constraint systems you could look at
Negroponte's Architecture Machines, an MIT press book which discusses
the kinds of constraint systems actual designers would be interested
in using. There were a number of architects working with constraint
systems which have developed, at least partially, into modern CAD
systems. You might try looking up some papers by Tim Johnson, Yona
Friedman, or Masanori Nagashima if you are interested in this sort of
thing. These systems tried to be useful during the early stages of
design, rather than during the final drafting. Most of these
languages were visual rather than textual, but a solution was found by
adding and manipulating constraints.
If you are interested in the interaction of generative grammatical
constraints and explicit situational constraints (as are often found
in natural structures) I'll mention the SAR design people based in
Eindhoven, though I am only familiar with the MIT contingent including
Habraken, Gerzso and Govela. (I won't mention the fascinating
politics of this design methodology although there are a number of
good stories).
For more on the interaction of constraint and construction you should
check out the classic On Growth and Form by Thompson (or is it Thomas)
which turns up now and then.
Still jet lagged
after all these years,
Seth
------------------------------
Date: 7 Mar 88 04:31:24 GMT
From: Eric Neufeld <emneufeld%watdragon.waterloo.edu@RELAY.CS.NET>
Reply-to: Eric Neufeld <emneufeld%watdragon.waterloo.edu@RELAY.CS.NET>
Subject: Re: Uncertainty and FUZZY LOGIC VS PROBABILITY
In article <8802291913.AA13911@dalcsug.UUCP> creelman@dalcsug.UUCP
(Paul Creelman) writes:
>
>Subject: Uncertainty and FUZZY LOGIC VS PROBABILITY
>Newsgroups: comp.ai.digest
>Keywords: UNCERTAINTY, PROBABILITY
>
> There appears to be some discussion about uncertainty by Eric Neufeld
> and others. According to Spiegelhalter, the use of probability for
> representing uncertainty in expert systems is the wrong method.
First favour I would like to ask of the net: Something has happened with
our news feed: I have seen nothing of this controversy since my original
posting. Would someone, possibly the moderator, be so kind as to mail me the
controversy? [Done. -- KIL]
To continue the discussion:
> it is inappropriate because uncertainty in knowledge does not match the
> chance mechanism of an observable event. It is unnecessary since no meaning
> must be attached to numbers, but instead the rank order of hypotheses is
> often all that matters in an expert system.
I have heard Ben-Bassat say that even the rank ordering is unimportant in
*applications*. Physicians want to know *possible* diagnoses, relative
strengths are what is important. (My apologies to Dr. Ben-Basset if this is
incorrect.) But so what? That is an opinion. Suppose rank ordering is
important as Spiegelhalter suggests. The use of numbers in probability
theory can be viewed merely as a convention. Nothing precludes the use of
probability as a way of deriving rank orderings. One of my favourite papers
is Koopman's which eliminates the numbers (in the preamble) with the hope of
restoring the primal intuition of probability. The numbers are later added
for consistency with the mathematical theory.
>Furthermore probability is
> somewhat impractical since it requires too many estimates of prior
> probabilities, fails to distinguish ignorance from uncertainty, and
> fails to provide an explanation of conclusions. I must agree. Down with
> probability!
You contradict yourself! Probability tells us that it is not trivial to
distinguish ignorance from uncertainty. Probability tells us that truth is
*independent* of explanation (i.e., given our knowledge of your symptoms,
the probability of disease X is 0.xx (or rank ordering 3) REGARDLESS OF THE
EXPLANATION or ARGUMENT used to get the diagnosis). But that is not what
probability is for. It is used to measure (relative) strength in an
argument.
> Surely what is needed is a simplified version of Shafer's evidence theory
> which deals with all possible subsets of the possible variable values, the
> frame of discernment. A number is associated with each subset which measures
> the certainty that the actual variable value is in that subset. Suppose we
> coarsen the uncertainty measure by reducing the number of subsets specified.
I would say surely not! Professor Kyburg has, more than a year ago, shown
that the theory of Dempster-Shafer is equivalent to an
interval-valued theory of probability, with some added statistical
assumptions. That is not to say that the D-S model has no useful
applications. I will take the liberty of paraphrasing Dr. Kyburg, who
concludes his article by saying that there is nothing wrong with variations
on the theory of probability containing such statistical assumptions, but
these assumptions should be in full view, up-front, for all to criticize
constructively.
>
> Paul Creelman
--
,
w
q
------------------------------
End of AIList Digest
********************
