Copy Link
Add to Bookmark
Report
AIList Digest Volume 2 Issue 044
AIList Digest Sunday, 8 Apr 1984 Volume 2 : Issue 44
Today's Topics:
Cellular Automata - References,
Image Understanding - Expert System for Radiograph Analysis,
Education - Model AI Curriculum,
AI Funding - Alan Kay Review & Strategic Computing,
Seminars - Language Structures Time Change & Automatic Programming
----------------------------------------------------------------------
Date: 1 Apr 84 13:44:53-PST (Sun)
From: decvax!genrad!wjh12!vaxine!pct @ Ucb-Vax
Subject: Re: Cellular Automata
Article-I.D.: vaxine.221
There is a big review article by Wolfram in Reviews of Modern Physics
v. 55 no. 3 p. 601 (July 1983) with a large list of references
------------------------------
Date: 2 Apr 84 18:41:37-PST (Mon)
From: hplabs!hao!seismo!rochester!ur-laser!bill @ Ucb-Vax
Subject: An expert system for reading chest radiographs
Article-I.D.: ur-laser.136
I have developed an "expert" system that analyzes chest ra-
diographs for tumors. This system was tested on 37 films
that contain nodules. It is capable of finding the nodules
in 92% of the films. In studies of mass screenings of ra-
diographs by radiologists it was found that the radiologists
miss 25-30% of all nodules < 1cm. A Rib Expert determines
whether a candidate nodule (possible tumor) is a rib. A No-
dule Expert, a linear-discriminant-based pattern recognizer,
classifies candidate nodules. All candidate nodules that are
classified as any type of nodule are presented to a radiolo-
gist for further inspection. Radiologists can recognize no-
dules as such once they are pointed out. If you are in-
terested in this work or want leads to other methods of au-
tomated chest film analysis, which are listed in the bi-
bliographies, contact Peggy Meeker ((716)275-7737, {allegra,
seismo}!rochester!peg) at the Computer Science Dept at the
University of Rochester and request the following TRs:
Lampeter, W.A. "Design, tuning, and performance evaluation
of an automated pulmonary nodule detection system." TR-120,
Computer Science Department, University of Rochester, Ro-
chester NY, 1983.
Lampeter, W.A. "Three image experts which help distinguish
tumors from non-tumors," TR-123, Computer Science Depart-
ment, University of Rochester, Rochester NY, 1984.
Other works of possible interest:
Ballard, D. H., J. Sklansky. "Tumor detection in radio-
graphs," Computers in Biomedical Research, 6, 299-321,
1973.
Jagoe, J.R., "Reading chest radiographs for pneumoconiosis
by computer," Brit. J. Ind. Med., 32, 267-272, 1975.
Toriwaki, J. et.al. Pattern recognition of chest x-ray im-
ages. Comp Grap Pat Recog, 2, 252-271, 1973.
Bill Lampeter
Department of Radiology
School of Medicine and Dentistry
University of Rochester
(716) 275-5101 or (716) 275-3194
{seismo, allegra}!rochester!ur-laser!bill
------------------------------
Date: Sat 7 Apr 84 21:30:27-PST
From: Ken Laws <Laws@SRI-AI.ARPA>
Subject: AI Curriculum
The April issue of IEEE Computer discusses computers in education.
The first article is on a model curriculum for Computer Science, and
pages 12-13 describe a sample curriculum for AI. About 20 references
to suggested AI texts and articles are also given.
-- Ken Laws
------------------------------
Date: 6 Apr 84 21:16:53 PST (Friday)
From: Ron Newman <Newman.es@Xerox.ARPA>
Subject: Alan Kay on DARPA research & Mansfield amendment
Excerpted from an interview in the April 1984 issue of ST.Mac magazine
(Softalk's magazine for the Macintosh). All [bracketed phrases] are as
in the original.
Alan: Things haven't been the same in computer science since two things
happened. The awful thing that happened was the Mansfield amendment in
1969. The amendment was a congressional reaction to pressure from the
population about the Vietnam war, mostly uninformed pressure.
What it did was force all military funding to be put under the
scrutiny of Congress and to be diverted only to military-type things.
All of a sudden, everything was different at ARPA [the Advanced Research
Projects Agency].
Q: ARPA became DARPA (the Defense Advanced Research Projects Agency) at
that point?
Alan: ARPA became DARPA. The last good thing to be done had already
been funded, which was the Arpanet [a network of communicating computers
around the world that allows scientists to send messages to each other].
That was finished in 1970. That was the end of ARPA funders [program
directors] being drawn from the ARPA community.
During the golden age at ARPA, the funding was much less than it is now,
but it was wide open.
Q: More creative work was done?
Alan: Yeah. Their whole theory--partially because the managers of ARPA
were scientists themselves--was "we fund people, not projects. If we
can understand what these guys are doing, we should probably be off
doing it ourselves. We'll just dump half this money for three years and
take our lumps."
They took percentages, like you have to in real research. And, God, did
they get some great stuff!
~~~~End of excerpt~~~~~
Alan makes lots of other brash statements in this article too. I'll
leave you with just this one:
"I'd just as soon send all the engineers around here in Silicon Valley
to the Outback of Australia until they have read something like 'The
Federalist Papers' or Adam Smith's 'Wealth of Nations' or *something*,
for God's sake....what they're doing is actually vandalizing an entire
generation of kids by acting as though things like Basic have value."
------------------------------
Date: 5 Apr 84 17:56:01 PST (Thursday)
From: Ron Newman <Newman.es@Xerox.ARPA>
Subject: Strategic Computing in Electronic News 3/19/84
[personal comment follows at end of article--RN]
"DOD Strategic Computing to get $95M in Funding"
Electronic News, March 19, 1984, page 18
by Lloyd Schwartz
WASHINGTON (FNS)--A virtual doubling of the funds for the Defense
Department's Strategic Computing initiative in fiscal year 1985--from
$50 million to $95 million--represents the first step in providing
"dramatic new computational capabilities to meet future critical defense
needs," Pentagon officials reported to Congress.
They said that, as computer capability evolves, "men and computers
will operate as collaborators in the control of complex weapon systems."
It boiled down to, they added, future "wars by computer," with the side
possessing the superior technology prevailing.
Dr. Robert S. Cooper, director of DOD's Defense Advanced Research
Projects Agency (DARPA), describing the program as well under way,
explained it is using a new idea, employing multiprocessor architecture
to reach for a new generation of computers with as much as 10,000 times
the computing capability of hardware available today.
The computers, endowed with artificial intelligence, will be capable
of solving extraordinarily complex problems involving human beings,
understanding speech and responding in kind, Dr. Cooper indicated to
the House Armed Services Committee. They also will require a whole new
system of prototyping, it was added.
Dr. Cooper testified that while computers are already widely employed
in defense, current computers have inflexible program logic and are
limited in their ability to adapt to unanticipated enemy actions in the
field. The problem, he noted, is exacerbated by the increasing pace and
complexity of modern warfare.
"The Strategic Computing program will confront this challenge by
producing adaptive, intelligent computers specifically aimed at critical
military applications," the DARPA chief continued. "These new machines
will be designed to solve complex problems in reasoning. Special
symbolic processors will employ expert human knowledge contained in
radical new memory systems to aid humans in controlling the operation of
complex military systems.
"The new generation computers will understand connected human speech
conveyed to them in natural English sentences, as well as be able to see
and understand visible images obtained from TV and other sensors."
Dr. Cooper noted DARPA has already demonstrated a limited voice
message system in which a computer recognized and understood human
speech to receive its commands. The computer was able to respond
verbally, using synthesized speech, although it possesses a limited
vocabulary.
Another example of technological advancement, Dr. Cooper noted, was
DARPA's recent success in applying a finely-focused ion beam in the
maskless fabrication of integrated circuits. He said this work is
continuing and "could result in a major breakthrough in ultimately
achieving a large-scale maskless fabrication capability."
Summing up, the DARPA chief declared "In the future, supercomputers
with reasoning ability and natural language interfaces with military
commanders will be able to participate in military assessment and may be
able to simulate and predict the consequences of various proposed
courses of military action. This will allow the commander and his staff
to focus on the larger strategic issues, rather than have to manage the
enormous information flow that will characterize the battles of the
future."
Dr. Cooper added that the balance of military power in the future
"could well depend on successful application of 'superintelligent
computers' to the control of highly-effective advanced weapons."
~~~~~End of Electronic News article~~~~~
Comments:
1. In the past, defenders of DARPA funded computer research have
asserted that the military and civilian industry have the same goals, so
that what's good for the Pentagon is good for the commercial market too.
But now we have a program whose goal, in the Pentagon's own words, is to
produce "adaptive, intelligent computers ***specifically aimed at
critical military applications***."
[Sorry if I'm injecting any personal bias here, but this seems to be a
non sequitur. Past military research (e.g., image understanding) was
also targeted at critical military applications; that didn't prevent
it from also being useful or even critical to civilian industry. The
strategic computing effort need not be different. All that has changed
is the military's boldness in expressing its own importance, about which
it may or may not be right. -- KIL]
2. Everyone knows how backward Soviet computer science and industry
are, so who is he talking about when he refers to " 'wars by computer,"
with the side possessing the superior technology prevailing" ? Once
again, the U.S. leads the way into a new round of the arms race.
/Ron
------------------------------
Date: Fri 6 Apr 84 11:57:07-PST
From: PENTLAND@SRI-AI.ARPA
Subject: Issues in Language, Perception and Cognition
[Forwarded from the CSLI bboard by Laws@SRI-AI.]
**** Due to a scheduling conflict, there has been a room change, to 050 ****
WHO: Len Talmy, Cognitive Science Program and German Dept., UC Berkeley
WHAT: How Language Structures its Concepts
WHEN: Monday April 9 12:00 noon
WHERE: Room 380-50
------------------------------
Date: Thu 5 Apr 84 17:02:34-PST
From: Richard Treitel <TREITEL@SUMEX-AIM.ARPA>
Subject: Automatic Deduction talk
[Forward from the Stanford bboard by Laws@SRI-AI.]
Monday, April 9th in MJH 301 at 2:30.
THE ORIGIN OF BINARY-SEARCH ALGORITHMS
Richard Waldinger
Artificial Intelligence Center
SRI International
Many of the most efficient numerical algorithms employ a binary search, in
which the number we are looking for belongs to an interval that is divided in
half at each iteration. We consider how such algorithms might be derived from
their specifications.
We follow a deductive approach, in which programming is regarded as a kind
of theorem proving. By systematic application of this approach, several
integer and real-number algorithms for such functions as the square root and
quotient have been derived. Some of these derivations have been carried out on
an interactive program-synthesis system. The programs we obtained are
different from what we originally expected.
------------------------------
End of AIList Digest
********************
