[Prev][Next][Index][Thread]
CSLI Calendar, 1 April 1998, vol. 13:27
C S L I C A L E N D A R O F P U B L I C E V E N T S
______________________________________________________________________
1 April 1998 Stanford Vol. 13, No. 27
______________________________________________________________________
A weekly publication of the
Center for the Study of Language and Information (CSLI)
Stanford University, Ventura Hall, Stanford, CA 94305-4115
____________
ACTIVITIES DURING 1 APRIL TO 10 APRIL 1998
WEDNESDAY, 1 APRIL
4:15pm EE380: Computer Systems Laboratory Colloquium
Gates B03 (NEC Auditorium)
What Do You Want to Wear Today?
Philippe Kahn
Starfish Software
Abstract below
4:30pm Engineering Economic Systems Colloquium
Room 380:380Y (Math Corner)
Stochastic Games
Professor Tes Raghavan
University of Illinois at Chicago
Abstract below
5:30pm Philosophy Department Colloquium
Bldg. 200:002 (History Corner)
Pictorial Representations
Richard Wollheim
UC Berkeley
THURSDAY, 2 APRIL
3:00pm Computer Science Talk
Gates 104
Statistical Models for Parsing and Information Extraction
Michael Collins
University of Pennsylvania
Abstract below
4:00pm Xerox PARC Forum
George Pake Auditorium, Xerox PARC
Wages, Skills and Technology in the United States and
Canada
Paul Romer
Stanford University
http://www.parc.xerox.com/ops/projects/forum/
4:15pm Frontiers of Neuroscience Seminars
Munzer Auditorium
Cell Fate Determination in the Vertebrate CNS
Dr. Connie Cepko
Harvard Medical School
Host: Dr. Susan McConnell
http://www.stanford.edu/dept/nbio/Spring98.html
4:15pm AI-Vision-Robotics Division Colloquium
Gates 104
Geometric Algorithms in AdeptRAPID
John Craig
Adept Technology
Abstract below
6:30pm Phonology Workshop
Margaret Jacks Hall, 460:146
The Richness of Paradigms, and the Insufficiency of
Underlying Representations in Accounting for Them
Bruce Hayes
Department of Linguistics, UCLA
Abstract below
FRIDAY, 3 APRIL
3:15pm Philosophy Department Colloquium
Bldg. 90:92Q (Philosophy)
The Science of Pragmatics and the Pragmatics of Science:
Scientific Philosophy and Unified Science in Carnap and
Morris
Alan Richardson
University of British Columbia
3:15pm Computer Science Database Seminar
Gates B-12
Internet and Database Related Research at Stanford
Members of the Infolab
Abstract below
3:30pm Linguistics Department Colloquium
Margaret Jacks Hall 460:146
Narrative and Institutional Memory
Charlotte Linde
Institute for Research and Learning,
and Stanford University
Abstract below
SATURDAY, 4 APRIL
all day Second Annual Stanford Undergraduate Philosophy
Conference
Bldg. 60:61H
Information below
SUNDAY, 5 APRIL
10:30am Founders' Day Celebration
honoring the founding of Stanford University
http://www.stanford.edu/~actionj/Founders_Day.html
MONDAY, 6 APRIL
4:30pm Stanford Digital Libraries Seminar
Gates B08
Kevin McKenna
Stanford, New York Times Online
Abstract below
3:30pm Psychology Social Lab
Jordan Hall 420:100
Social Cognitive Theory Of Personality
Albert Bandura
http://matia.stanford.edu/html/talks.html#social_lab
TUESDAY, 7 APRIL
4:15pm Logic Seminar
Room 380:381T
Recent results in metapredicative proof theory, Part II
Thomas Strahm
University of Bern, visiting Stanford
Abstract below
WEDNESDAY, 8 APRIL
4:15pm EE380: Computer Systems Laboratory Colloquium
Gates B03 (NEC Auditorium)
The Internet and Copyright Law
Mark Radcliffe
Gray Cary Ware and Freidenrich
Abstract below
4:30pm Berkeley Phonology Laboratory Colloquium
46 Dwinelle Hall
The Role of Dynamic Cues in Speech Perception, Spoken
Word Recognition, and Phonological Universals
Natasha Warner
University of California, Berkeley
THURSDAY, 9 APRIL
12 noon CSLI Talk
Cordura Hall, Room 100
Advanced Human Interface and New Applications
Herman D'Hooge
Intel, Hilsboro, Oregon
Abstract below
4:00pm Xerox PARC Forum
George Pake Auditorium, Xerox PARC
Plugged In: A Community Technology Access Program
Bart Decrem
4:15pm Philosophy Department Colloquium
room to be announced
The Metaphysics of Separation
Alan Silverman
Ohio State University
co-sponsored with Classics
4:15pm AI-Vision-Robotics Division Colloquium
Gates 104
Computational Mechanisms of Visual Grouping
Jitendra Malik
U.C.Berkeley
Abstract below
4:15pm Mathematics Colloquium
Room 380:380Y
Model-Theoretic Methods in Diophantine-Geometric Problems
Anand Pillay
University of Illinois, visiting MSRI
Abstract below
4:15pm Frontiers of Neuroscience Seminars
Munzer Auditorium
Nerve Growth and Synaptic Plasticity
Dr. Mu-Ming Poo
University of California, San Diego
Host: Dr. Howard Schulman
http://www.stanford.edu/dept/nbio/Spring98.html
7:30pm Phonology Workshop
Margaret Jacks Hall, 460:146
The Optimization of Portuguese Verb Representations:
Vowel Harmony Revisited
James Giangola
General magic
Abstract below
FRIDAY, 10 APRIL
3:15pm Computer Science Database Seminar
Gates B-12
SGML Documents and their Manipulation in Object and
Object-Relational Databases
Eric Neuhold
ICSI, Berkeley and GMD IPSI, Germany
http://www-db.stanford.edu/dbseminar/dbseminar.html
____________
EE380: COMPUTER SYSTEMS COLLOQUIUM
on Wednesday, 1 April 1998, 4:15pm to 5:30pm
NEC Auditorium (B03), Gates Computer Science Building
http://www.stanford.edu/class/ee380/
What Do You Want to Wear Today?
Philippe Kahn
CEO and Chairman
Starfish Software
mailto:philippe@starfish.com
What if you could wear your information? What if you could move
through your whole day without having to think about bringing your
information, your handheld or laptop computer with you? What if you
could have it all, anywhere, anytime?
You wake up and start your day, get in the shower, and ask yourself:
"Where do I want to go today?" It isn't just advertising hype. It's
probably the right thing to do. Then you pour yourself a cup of
coffee, 100% Java or not.
Then you ask yourself the all-important question: "What do I want to
wear today?"
You're not really thinking about that laptop or that handheld computer
that you'll carry with you all day. You're thinking about your shoes,
your belt, your shirt or blouse, your pants. Once you've decided what
to wear, for the rest of the day, you'll rarely think about what
you're wearing again. You simply wear your clothes.
You're dressed for success. You can focus on getting things done. Now,
it's time to think about the digital tools that you will need
throughout your day. What will you be taking with you? Is it the
7.4-pound power laptop, the 4-pound sub notebook, or the 5-ounce
pen-based handheld? Chances are that whichever choice you make, you'll
have to lug it with you all day, it will feel heavier and heavier with
every hour that passes, pack it in and out of your briefcase, and not
lose it! This is portable technology. It sure isn't wearable, but it's
a step in the right direction.
In prior columns, I discussed (the concept of what Starfish terms)
Connected Information Devices. They are small devices such as pagers,
smart phones or REX cards that can synchronize important information
such as address books and calendars and give you access to this
important information anywhere, anytime.
But what does wearable mean? A wearable device is one that you put on
in the morning and it's always with you. No need to think about it
further. Just like a pager or cellular phone clipped onto your belt..
Wearable Connected Information Devices are small, sturdy and can
synchronize in a timely manner with information that is important to
you. You also never want to worry about the state of your batteries
during the day.
Like a small pager or a StarTac(TM) cellular phone, you want to be
able to operate them with one hand, while your other hand is free to
do something else. You really shouldn't try to operate any of these
devices while driving!
Is this a completely new concept? Yes and No. Remember Maxwell Smart
from the television series "Get Smart?" Remember his shoephone? We all
have seen numerous James Bond movies. Then there is Dick Tracy and his
famous watch. So the concept of Wearable Connected Information Devices
is not completely new.
What is new is that wearable technology is now available for you and I
to purchase at very reasonable prices and allows synchronization and
connectivity to our information. Like a StarTac phone or a REX card.
That technology is generally not extrapolated from conventional
computer-based technology. That is because of form factor and battery
life requirements. In fact, the kind of technology that is really
useful here is more attune to watch class technology than to
conventional PC technology.
Let's take a look back at the history of personal computers. With the
first PC, the Micral, released in 1973, there was a radical shift in
form factors. Before the Micral, we had Mainframe computers. The
Altair, Apple 2 and the IBM PC followed the Micral later. It is
interesting to note that the form factors of desktop computers haven't
really changed much since 1973. There is more computing power and
storage capacity. However, most of that computing power and storage
capacity is actually used to provide for a more intuitive user
interface as well as multimedia capability.
The first truly transportable computer appeared in 1981, it was the
Osborne 1. It was soon to be followed by the Compaq. These machines
were basically oscilloscope cases retrofitted to house a complete
personal computer. They needed AC to function. A few years later,
around 1984, the first Clamshell device appeared. It was the Data
General One. Most laptops today are derived from that original design
with much improvement. They weigh 3.5 to 7.5 pounds.
To achieve more portability several companies designed handheld
computers. Over the years, the Tandy 100, the HP 100 LX, the Psion and
many others have had some measure of success. Pen-based devices
appeared and disappeared and will probably start reappearing again. In
that category, the Newton in particular broke new ground for ease of
use. But synchronization was essentially absent, a retail price that
was much too high and a form factor that was too big. In 1995, the
PalmPilot was introduced and became the most successful pen-based
handheld device and demonstrated that there is a real market for small
portable devices that easily synchronize with personal computers.
In October 1997, REX was introduced by Rolodex/Franklin after more
than three years of development by Starfish. The form factor shift is
radical: REX is ultra thin, the size of a credit card and weighs only
1.4 Ounces. It has 6 months of battery life, and synchronizes directly
through the PC Card slot. REX has sold as many units in its first
three months as PalmPilot sold in its first six months.
REX is the first truly Wearable Connected Information Device. You wear
it just like you carry your credit cards. Pull it out when you need
it. Synchronize it when you get close to your PC. The conceptual model
is familiar, simple and efficient, while the device itself is
completely non-obtrusive. You can put it in your shirt pocket in the
morning and forget about it all day long.
What is in store for Wearable Connected Information Devices? Of course
there is the entire wireless world, the world of specialized medical
wearable devices and many other developments that we will soon be
discussing. Unfortunately, because writing this column is only a
secondary occupation and that my primary occupation is to work with
our device development teams and our partners on next generation
devices, I obviously can't divulge all of our plans now. But rest
assured that in the next few years you'll see many innovative Wearable
Connected Information Devices.
I would like to make two predictions:
Prediction #1: Within the next 5 years most of us will make use of
several Wearable Connected Information Devices that all synchronize
with the same information and work synergistically.
Prediction #2: By the year 2002, the unit sales of Wearable Connected
Information Devices will surpass the unit sales of personal computers.
Biography: Philippe Kahn founded Starfish in 1994 with Sonia Lee to
develop the TrueSync technology platform for the Connected Information
Device industry. In addition to his responsibilities at Starfish,
Philippe champions freedom of speech, open standards, innovation and
spirited competition. BYTE Magazine names Philippe as one of the Top
20 Most Important People in the history of the computer industry.
A French-born mathematician, Philippe's graduate and post-graduate
work was in pure mathematics. In the 1970s, he worked in Zurich,
Switzerland under Niklaus Wirth on Pascal, the computer programming
language. Contrary to conventional belief, The Computer Museum
( http://www.tcm.org/ ) now recognizes the Micral as the first
personal computer available outside a build-it-yourself kit. This
Intel-based machine was developed in France by Andre Truong. Philippe
was one of the first key programmers for this system which preceded
the Altair by more than one year.
After moving to the United States in 1982, Philippe founded Borland
International. Under his direction, and without venture capital
funding, Borland grew from a start-up to one of the leading suppliers
of professional software development tools with several thousand
employees worldwide and $500 Million in revenues. While at Borland, he
spearheaded the object-computing revolution and the move to
component-based software, changing software development methodologies
forever.
Anticipating the emergence of wireless and wireline devices as well as
the fact that the Internet would cause the personal computer to
quickly evolve from a document creation and data computation tool to a
wide area communications tool, Philippe left Borland in early 1994 to
create Starfish with Sonia Lee.
After several years developing the TrueSync technology platform,
Starfish is now the leading supplier of wearable technology components
for the Connected Information Device Industry. In August 1997,
Starfish introduced Rex, the first wearable Connected Information
Device. Key partners and customers of Starfish include companies such
as Motorola, Rolodex Electronics, General Magic and others.
Besides working with the teams at Starfish, Philippe spends his time
with his four children, Laura, Estelle, Samuel and Sophie, with whom
he loves to practice his favorite hobbies: jazz, martial arts,
Siberian huskies, dirt biking, snowboarding and skiing.
____________
ENGINEERING ECONOMIC SYSTEMS COLLOQUIUM
on Wednesday, 1 April 1998, 4:30pm
Math Corner 380:380Y
http://www.stanford.edu/class/eesor406/
Stochastic Games
Professor Tes Raghavan
University of Illinois at Chicago
This is a general talk on stochastic games with finite state and
action spaces that covers the existence of stationary optimal
strategies for discounted and structured undiscounted stochastic
games, and algorithms for some of them. An undiscounted stochastic
games has a value; however, even for the simplest such games, it is
necessary to use past history and compromise with non-stationary
epsilon-optimal behavior strategies. For non-zero-sum games,
equilibria fail to exist in stationary strategies, even for the
simplest of stochastic games like those with perfect information. The
existence of an equilibrium payoff for non-zero-sum n-person
stochastic games is still open. The talk will be accessible to a
general audience of M.S. and Ph.D. degree students with some knowledge
of zero-sum matrix games.
____________
COMPUTER SCIENCE TALK
on Thursday, 2 April 1998, 3:00pm
Gates 104
Statistical Models for Parsing and Information Extraction
Michael Collins
University of Pennsylvania
The vast amount of information now available in electronic form has
led to increasing demand for applications that process natural
language. (Example applications include machine translation,
summarization and information extraction). Accurate methods for
parsing unrestricted text will almost certainly be a key component in
these applications.
Unfortunately, the traditional approach to syntactic analysis --
writing a grammar by hand -- has encountered two major problems.
First, ambiguity: even moderate-length sentences often receive
thousands of analyses, with no indication of which is correct.
Second, coverage: constructing an exhaustive grammar of English has
proved to be extremely difficult owing to the huge number of rules
needed.
In this talk I will describe my work on machine learning methods for
parsing. A statistical model is trained from a corpus of sentences
that have been annotated for syntactic structure. Competing analyses
for a test data sentence can then be ranked by their probability under
the model; moreover the most probable analysis can be efficiently
found. I will show how careful design of the model can lead to
linguistically motivated parameters, and crucially to parameters that
condition heavily on lexical information. The resulting models
recover constituents in Wall Street Journal text with 88% accuracy,
the best published results on this task.
As an example application using the output of the parser, I will
describe more recent work on statistical models for information
extraction, specifically on the task addressed in the Darpa-sponsored
6th Message Understanding Conference (extraction of information about
management succession events from newspaper text).
____________
AI-VISION-ROBOTICS DIVISION COLLOQUIUM
on Thursday, 2 April 1998, 4:15pm until 5:30pm
Gates 104
http://www-formal.stanford.edu/aicolloq/
Geometric Algorithms in AdeptRAPID
John J. Craig
Director of R&D
SILMA Division of Adept Technology
AdeptRAPID is a software package used to design and simulate
robot-based flexible automation cells. Its goal is to simplify the
conception and design of automation installations used for
applications such as mechanical assembly, material handling, or
packaging. Ease of use is of paramount importance so that this
technology can truly impact the way industrial automation is
accomplished. The need for ease of use drives the need for the
simulation system to behave like the actual physical world. The more
the simulator acts like the real world, the simpler the user interface
paradigm for the user, since the physical world is the one we are all
familiar with. At the same time, trade-offs for computational speed
and other factors have driven a design in which a particular "slice"
of reality is simulated, while many details are not. AdeptRAPID is
well-suited as a host for a variety of geometric algorithms. The need
to model various portions of the real world, as well as the need to
unburden the user by making various computations of a geometric nature
drive the need for such algorithms. AdeptRAPID provides the
environment in which some advanced algorithms can be brought to bear
on real problems occurring in industry. This paper presents the
current state of development of the AdeptRAPID simulator. Attention is
paid to the various geometric algorithms already in place as well as
the need for more and better algorithms.
____________
STANFORD PHONOLOGY WORKSHOP
on Thursday, 2 April 1998, 6:30pm
Margaret Jacks Hall 460:146
http://www-linguistics.stanford.edu/Linguistics/pinterest/
*Note change from the usual time*
The Richness of Paradigms, and the Insufficiency of Underlying
Representations in Accounting for Them
Bruce Hayes
UCLA
The thesis of this talk is that the informational content of paradigms
is richer, in a specific way, than has been commonly realized.
Consider the famous "Wug test" of Berko (1958): the subject is given a
fictional word of the language in some inflected form, and is asked to
provide another inflected form for the same stem. This test, which
replicates real life experience, is in traditional theories supposed
to invoke the following logical sequence:
(1) a) Undo phonology in the form one is given.
b) Peel off any affixes from the form that results, to
achieve the underlying form of the stem.
c) Add the appropriate affixes for the target inflected form.
d) Apply phonology in the forward direction, to obtain the
desired surface form.
Naturally, this procedure is not always deterministic. If the
phonology under (a) is *neutralizing*, there will be multiple
possibilities. In principle, all of them should be possible Wug test
answers.
The procedure of (1) is quite indirect; a priori, one might have
imagine alternative procedures that map inflected forms directly, one
from another. Current theorizing about Correspondence Constraints
operating on members of a paradigm raises just this possibility.
If languages really do map members of the paradigm directly, one from
another, there arises the possibility that paradigms involve a greater
*richness of predictability relations* than the orthodox scheme in (1)
implies. Empirically, we could detect this if we found that speakers
systematically give *fewer answers* in Wug testing than the procedure
of (1) would lead us to expect. Such an observation would show that
speakers are making use of knowledge that (1) implies they shouldn't
possess.
In this talk, I propose to discuss cases of this sort: briefly in
Korean, more extensively in English and most crucially in Yidiny. For
the latter two cases, I hope to show that the only way of saving the
underlying-representation approach, namely placing constraints on
underlying forms, will not suffice. What emerges is this: underlying
representations deserve our skepticism, not because they are fancy or
abstract, but because they *aren't good enough* to accomplish what we
wanted them to do.
____________
SEMINAR ON PEOPLE, COMPUTER, AND DESIGN
on Friday, 3 April 1998, 12:30-2:00pm
Gates B03 (NEC Classroom)
http://www-pcd.stanford.edu/seminar/
(SITN Channel E2)
Comic Chat: Research and Productization War Stories
David Kurlander
Microsoft Corporation
mailto:djk@microsoft.com
Comics have a rich visual vocabulary and people find them appealing. They
are also an effective form of communication. We have built a system called
Comic Chat, that represents on-line communications in the form of comics.
Comic Chat automates numerous aspects of comics generation, including
balloon construction and layout, the placement and orientation of comic
characters, the default selection of character gestures and expressions, the
placement of panel breaks, and the choice of zoom factor for the virtual
camera. It uses techniques that we learned by working with a professional
comic artist, to create comics depicting textual conversations. Comic Chat
started out as a project at Microsoft Research, and now ships with the
latest version of Windows '95, Internet Explorer 4, The Microsoft Network,
and will be included as part of Windows '98 and Windows NT 5. This talk will
describe the technology behind Comic Chat and our experiences trying to
bring it to market.
Biography: Dr. Kurlander joined Microsoft Research over five years
ago. One of his recent projects, called Comic Chat, was a SIGGRAPH
paper in 1996, and has since become an extremely successful product,
shipping millions of copies. Kurlander graduated from Harvard
University with an A.B. in applied mathematics. Later he received his
Ph.D. in computer science from Columbia University for work in
graphical editing and example-based user interfaces. He has worked at
diverse places including the Pentagon and Xerox PARC.
____________
COMPUTER SCIENCE DATABASE SEMINAR
on Friday, 3 April 1998, 3:15pm - 4:30pm
Gates B-12
http://www-db.stanford.edu/dbseminar/dbseminar.html
Internet and Database Related Research at Stanford
Members of the InfoLab
Interactive Query and Search in Semistructured Databases
Roy Goldman and Jennifer Widom
Project: Lore
Semistructured graph-based databases have been proposed as well-suited
stores for World-Wide Web data. Yet so far, languages for querying
such data are too complex for casual Web users. Further, proposed
query approaches do not take advantage of the interactive nature of
typical Web sessions--users are proficient at iteratively refining
their Web explorations. In this paper we propose a new model for
interactively querying and searching semistructured databases. Users
can begin with a simple keyword search, dynamically browse the
structure of the result, and then submit further refining queries.
Enabling this model exposes new requirements of a semistructured
database management system that are not apparent under traditional
database uses. We demonstrate the importance of efficient keyword
search, structural summaries of query results, and support for inverse
pointers. We also describe some preliminary solutions to these
technical issues.
Google
Sergey Brin
Project: Google
Data Integration over the Web
Ramana Yerneni
Project: TSIMMIS
Mediation architecture is well suited for integrating data over the
web. We explore the possibility of exploring TSIMMIS technology for
this task. Specifically, we shall address the issues surrounding the
construction of wrappers and mediators, the query capabilities of data
sources, and the efficiency of query processing over the web.
Resolving Semantic Heterogeneity
Gio Wiederhold
Project: Scalable Knowledge Composition
After we learn to deal with heterogeneity in hardware systems,
operating systems, communication protocols, database systems and data
schemas, we are still faced with the problem that the meaning of the
words also varies. We will sketch an approach we are developing to
define methods that can resolve such differences for specific
application contexts.
____________
LINGUISTICS DEPARTMENT COLLOQUIUM
on Friday, 3 April 1998, 3:30pm
Margaret Jacks Hall, Room 460:146
http://www-linguistics.stanford.edu/colloq/colloq.html
Narrative and Institutional Memory
Charlotte Linde
Institute for Research on Learning
and Stanford University
This study provides a taxonomic structure for the discourses which
constitute institutional memory and the ways in which such memory
structures identity of an institution and of people within an
institution. It proposes the dimensions of written - oral, official -
unofficial, and intentional - secondary, and discusses how these
categories shape the landscape of memory, with particular attention to
informal and formal oral narratives told over a long period of time,
and their relation to materially preserved forms of memory. The social
aspect of memory is remembering: for example, a file cabinet preserves
some of a corporations' written memories, but it does so only if
people use it as a resource for remembering. Similarly a narrative
functions in an institution only if it is regularly told, and has the
potential to be transmitted to new tellers. Therefore, a taxonomy of
occasions for narrative remembering is also presented. Finally, this
study examines who in institutions have or do not have the right to
represent the institution's memory, and the ways in which silences and
erasures can be recognized and studied.
____________
SECOND ANNUAL STANFORD UNDERGRADUATE PHILOSOPHY CONFERENCE
on Saturday, 4 April 1998
Bldg. 60:61H
http://www.stanford.edu/group/dualist/conference.html
Sponsored by The Dualist
and The Undergraduate Philosophy Association
Philip Kitcher, Professor of Philosophy, UCSD
"Scientific Progress and Human Flourishing": The ways in which
the sciences make progress, focusing not only on advances in
knowledge, but also on how knowledge affects the ways in which
people live.
Peter Godfrey-Smith, Professor of Philosophy, Stanford
"On Genes and Codes": How do claims about genetic coding
relate to philosophical analyses of meaning, coding, and
information?
Eric Beerbohm, Senior, Political Science and Ethics in Society, Stanford
"Where the Action _Isn't_: On the Site of Distributive
Justice"
Johannes Kratz, Senior, Philosophy, Stanford
"Hume: Between Moral Subjectivism and Realism"
Kritika Yegnashankaran, Senior, Philosophy, Stanford
"The Philosophy of Biology"
____________
STANFORD DIGITAL LIBRARIES SEMINAR
on Monday, 6 April 1998, 4:30pm
Gates Building, B08
http://diglib.stanford.edu/diglib/seminars/seminars.html
News in the Digital Age
Kevin McKenna
As the Internet and other new media emerge, traditional publishers are
finding themselves with new opportunities and new competitors. What
have they learned so far about this new communications frontier, and
how will it affect their mission and their utility? My presentation
will discuss the ways in which digital technologies are allowing The
New York Times and other news organizations to expand their reach and
role, including the marketing of their archives to consumers through
the Web and the use of digital libraries as a reporting tool.
Biography: Kevin McKenna, a Knight Journalism Fellow for 1997-98 at
Stanford University, has been an editor at The New York Times for 13
years. From 1995 to 1997, he served as editorial director of The
Times's electronic media group and was the founding editor of The New
York Times on the Web. He began his career as a reporter for the
Associated Press and spent six years as an editor at the International
Herald Tribune in Paris. He graduated from the University of Southern
California and has a master's degree in journalism from Columbia
University.
____________
LOGIC SEMINAR
on Tuesday, 7 April 1998, 4:15pm
Math Corner 380:381T
http://www-philosophy.stanford.edu/Logic/
Recent Results in Metapredicative Proof Theory, Part II
Thomas Strahm
University of Bern, visiting Stanford
Metapredicativity is a new general term in proof theory which
describes the analysis and study of formal systems whose
proof-theoretic strength is beyond the Feferman-Schuette ordinal
Gamma-0 but which are nevertheless amenable to predicative methods.
In this talk we give a general survey and introduction to
metapredicativity. In particular, we discuss various examples of
metapredicative systems, including (i) subsystems of second order
arithmetic, (ii) first and second order fixed point theories, (iii)
extensions of Kripke-Platek set theory without foundation, and (iv)
systems of explicit mathematics with universes.
Relevant keywords for our talk are: arithmetical transfinite recursion
and dependent choice; restricted bar induction; transfinite
hierarchies of fixed points; transfinite fixed point recursion; hyper
inaccessibility, Mahloness and Pi-3 reflection without foundation;
universe operators.
____________
EE380: COMPUTER SYSTEMS COLLOQUIUM
on Wednesday, 8 April 1998, 4:15pm to 5:30pm
NEC Auditorium (B03), Gates Computer Science Building
http://www.stanford.edu/class/ee380/
The Internet and Copyright Law
Mark Radcliffe
Gray Cary Ware and Freidenrich
Copyright law is critical for doing business on the Internet, yet
copyright law is based on a world of printed books and movies which
are fixed in tangible objects. It is also a "national" law which
differs from country to country. How does copyright law need to be
changed to meet the new challenges of the Internet?
____________
CSLI TALK
on Thursday, 9 April 1998, 12 noon
Cordura Hall, Room 100
http://www-csli.stanford.edu/csli/Coglunch/
Advanced Human Interfaces and New Applications Research at Intel
Dave Cobbley and Herman D'Hooge
Intel
Intel, as part of the Intel Architecture Labs, has recently started an
"Advanced Human Interface and New Applications" research lab. This
new applications lab will conduct long-term research focused on the
discovery and exploration of compelling new uses of computing. The
presentation will provide an overview of the Intel Architecture Labs,
the new applications research lab, and current research focus areas.
____________
AI-VISION-ROBOTICS DIVISION COLLOQUIUM
on Thursday, 2 April 1998, 4:15pm until 5:30pm
Gates 104
http://www-formal.stanford.edu/aicolloq/
Computational Mechanisms of Visual Grouping
Jitendra Malik
Computer Science Division
University of California at Berkeley
http://www.cs.berkeley.edu/~malik/
I shall argue that early and intermediate level visual processing be
modeled as a three stage process. The first stage is a measurement
stage carried out with spatiotemporal receptive fields tuned to
orientation, spatial frequency, opponent color, and short-range
motion. The second stage is a grouping stage resulting in the
formation of regions of coherent brightness, color and texture. Call
these 'proto-surfaces'. The third stage results in the formation of
surfaces/objects with attached properties such as lightness, object
motion, occlusion relationships (figure-ground), depth, slant-tilt etc
and is based on the combined operation of Gestalt grouping factors,
shape cues, and can be partially influenced by knowledge of familiar
configurations.
The first stage can and has been modeled by many researchers using
tools of linear system analysis. We offer a novel approach to the
second stage by modeling it as the process of finding a partition of
the image into regions such that there is high similarity within a
region and low similarity across regions. This is made precise as the
'Normalized cut' criterion which can be optimized by solving an
eigenvalue problem. The resulting eigenvectors provide a hierarchical
partitioning of the image into regions ordered according to salience.
Brightness, color, texture, motion similarity, proximity and good
continuation can all be encoded into this framework. We show results
on complex images of natural scenes which demonstrate the significant
superiority of this technique over classical approaches such as those
based on edge detection, MRFs etc. Phenomena such as subjective
contours emerge as side consequences.
Our work on the third stage is preliminary; I shall argue on
computational and psychophysical grounds that modular shape processing
should be abandoned, and that grouping driven by ecological statistics
is as crucial as shape cues driven by ecological optics.
This is joint work with Jianbo Shi, Serge Belongie and Thomas Leung.
____________
MATHEMATICS COLLOQUIUM
on Thursday, 9 April 1998, 4:15 PM
Room 380:380Y (Math Corner)
Model-Theoretic Methods in Diophantine-Geometric Problems
Anand Pillay
University of Illinois, visiting MSRI
I discuss how the model theory of differential fields is able to
provide good bounds for certain finiteness theorems in
algebraic-diophantine geometry. Some straightforward examples concern
the number of generic points on X intersection \Gamma where X is a
plane curve over Q^{alg} and \Gamma a finitely generated subgroup of
C^{*}xC^{*}.
____________
STANFORD PHONOLOGY WORKSHOP
on Thursday, 9 April 1998, 7:30pm
Margaret Jacks Hall 460:146
http://www-linguistics.stanford.edu/Linguistics/pinterest/
The Optimization of Portuguese Verb Representations:
Vowel Harmony Revisited
James Giangola
General Magic
Three relatively recent phonological theories facilitate an
explanatorily superior account of the historically knotty problem of
Vowel Harmony (VH) in the Portuguese verb conjugation.
(1) Optimality Theory (OT, Prince & Smolensky 1993): The grammar
consists of a hierarchy of ranked and violable constraints against
which competing outputs are evaluated in parallel.
(2) Grammar Optimization (Inkelas 1995): "The optimal grammar is the
most transparent, i.e. the one in which alternations are maximally
structure-filling (Kiparsky 1993)". Underlying Representations
(URs) are thus posited to achieve maximum transparency.
(3) Archiphonemic Underspecification (Inkelas 1995): The
underspecification of alternating, predictable structures is
necessary in order to achieve Grammar Optimization.
The following three verbs illustrate VH, each representing a distinct
conjugation class.
conjugation class [a]-Theme [e]-Theme [i]- Theme
infinitive l[e]v[a]r 'to take' b[e]b[e]r 'to drink' v[e]st[i]r 'to dress'
1sg.pres.ind l[LOW e']vo b[e]bo v[i]sto
infinitive g[o]st[a]r 'to like' m[o]v[e]r 'to move' d[o]rm[i]r 'to sleep'
1sg.pres.ind g[LOW 'o']sto m[o]vo d[u]rmio
Assuming a simple vowel matrix consisting of heights high (i, u), mid
(e, o), and low (LOW e, o), one may generalize that Root Vowels (RVs)
orthographically represented by "e" and "o" assume the height of a
non-surfacing Theme Vowel (TV). Notably, all three TVs (/i/, /e/, /a/)
can trigger VH, while RVs /i/, /a/, /u/ are invariant.
The proposed analysis exploits the full predictability in verbs of
stress placement and the MID [e]/[o] ~ LOW [e]/[o]alternation, which
would require (chez Inkelas 1995) that we posit archiphonemes of /E/
and /O/, underspecified for [low]. This moves paves the way for
Grammar Optimization. In addition, an OT analysis portrays VH as the
resolution of conflicting constraints: on one hand, constraints
prohibit the TV from surfacing in certain contexts; on the other hand,
constraints require that features be parsed. Resolving this tension,
underspecified RVs salvage the stranded height features of unparsed
TVs.
Assuming archiphonemic underspecification, a constraint-based approach
offers considerable explanatory power. For example,
(1) constraint interaction accounts for the 'if-and-ony-if'
relationship between VH and Truncation in a non-stipulative
manner.
(2) Constraint interaction accounts for all three TVs /a/, /e/, and
/i/ as VH triggers, as well as for the invariance of RVs /i/, /a/,
and /u/, since these are fully specified.
(3) This analysis, in contrast to previous approaches, exploits the
predictability of MID [e]/[o] vs. LOW [e]/[o], along with stress
placement, in verbs.
In contrast, previous generative analyses have assumed three fully
specified low RVs, LOW /e/, /a/, and LOW /o/, which are based on
nominal forms, and posit an alpha-notation VH rule which causes only
mid and low RVs (but not /a/), to harmonize in height with the TV. A
rule of Truncation then suppresses the TV; thus, e.g., /vEst-i-o/ -->
vist-i-o --> v[i]sto ('I dress').
Such analyses engender a number of problems.
(1) The mutual dependency of VH and Truncation is handled
stipulatively via a counterbleeding rule order.
(2) Formulations of the VH rule fail to explain the invariance of RVs
/i/, /a/, and /u/.
(3) The underlying distinction between MID /e, o/ and LOW /e, o/
results in an opaque grammar; underlying structure is consistently
undone or replaced by the action of the VH rule.
(4) For many verbs, it is impossible to ascertain a fully specified
underlying RV.
These problems are handled as a matter of course in the proposed
analysis.
The solution is relatively cost-free since (1) the constraints reflect
general principles suggested by the data themselves; (2) the ranking
of constraints is built into the very architecture of OT grammar; and
(3) archiphonemic underspecification has been shown to be
indispensable for Grammar Optimization cross-linguistically. Finally,
this analysis brings to light a case of VH involving both [high] and
[low], a scenario for which there are purportedly "no known cases" in
the literature (Van der Hulst & Van der Weijer 1996).
____________
END MATERIAL
The CSLI Calendar appears weekly on Wednesdays throughout the academic
year. Announcements, abstracts, and other information to appear in
the Calendar should be submitted to the editor, who reserves the right
to decide what does or does not go in the calendar
mailto:incalendar@csli.stanford.edu.
Requests to be added to the mailing list should be sent to
mailto:majordomo@csli.stanford.edu. With the lines in the body of
the text of either
subscribe csli-calendar
for the long form or
subscribe csli-short-calendar
for the short form. Problems with subscribing or unsubscribing should
be sent to owner-csli-calendar@csli.stanford.edu.
The full current issue is at
http://www-csli.stanford.edu/csli/Archive/calendar/current.html
and the archives at
http://www-csli.stanford.edu/csli/Archive/calendar/.
People on most of the CSLI computers can type 'help csli-calendar' to
see the current issue.
The CSLI Calendar is also posted each week to
news://nntp-csli.stanford.edu/csli.bboard.
Information about CSLI's research program is available at
http://www-csli.stanford.edu/csli/.
For maps to the Stanford University campus see
http://www.stanford.edu/home/visitors/maps.html.
____________