[Date Prev][Date Next][Date Index]
CSLI Calendar, Wednesday, 20 November 2002, vol. 18:11
CSLI CALENDAR OF PUBLIC EVENTS
______________________________________________________________________
20 November 2002 Stanford Vol. 18, No. 11
______________________________________________________________________
A weekly publication of the
Center for the Study of Language and Information (CSLI)
Stanford University, Ventura Hall, Stanford, CA 94305-4115
http://www-csli.stanford.edu/
____________
ACTIVITIES FROM 20 NOVEMBER 2002 TO 29 NOVEMBER 2002
WEDNESDAY, 20 NOVEMBER 2002
3:00pm Reuters Foundation Digital Vision Program Seminar
Cordura 100
Collision Course:
Intellectual Property Rights and Traditional Knowledge
Douglas I. Kalish
Biotechnology Consultant and Educator, Palo Alto
http://reuters.stanford.edu/seminar_speakers.html
Information below
3:45pm Psychology Department Colloquium
Jordan Hall 420:041
Exploring the Architecture of Emotion
Robert W. Levenson
University of California
http://www-psych.stanford.edu/news.html
4:00pm Wireless Communication Seminar
Packard 101
Resource Allocation in Multihop Wireless Networks
Rene L. Cruz
Dept. of Electrical & Computer Engineering, UC San Diego
http://www.stanford.edu/group/wcs/schedule.html
4:15pm EE380: Computer Systems Lab Colloquium
Gates B03 (NEC Auditorium)
MEMS on Glass: Technology with applications for fingerprint sensors
S. K. Ganapathi
President and CEO, Fidelica Microsystems
http://ee380.stanford.edu/contents.html
Abstract below
THURSDAY, 21 NOVEMBER 2002
12:15pm Stanford Networking Seminar
Gates 104
Optimizing the Migration of Virtual Computers
Costa Sapuntzakis
Stanford University
http://netseminar.stanford.edu/
Abstract below
4:00pm PARC Forum
George Pake Auditorium at PARC
Our nano-future: What's hype, what's not
Jennifer Fonstad
Managing Director, Draper Fisher Jurvetson
http://www.parc.com/forum/
4:00pm UC Berkeley CIS Seminar
Soda Hall 310 (UC Berkeley)
Reorienting attitude --- rotating the right way.
Berthold Horn
MIT
http://www.cs.berkeley.edu/~eyal/cis-seminar
Abstract below
4:15pm CSLI Seminar on Computational Learning and Adaptation
Cordura Hall, room 100
Maintenance of Diagnostic Case-Based Reasoning Systems
Ioannis Iglezakis
DaimlerChrysler
http://cll.stanford.edu/scla.html
Abstract below
4:15pm SSP: Symbolic Systems Forum
Bldg. 380:380C (Math Corner)
Conversational Interfaces: a Domain-Independent Architecture
for Task-Oriented Dialogues
Alex Gruenstein
Symbolic Systems Program
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
Abstract below
4:15pm Fundamental Themes in Neuroscience Seminar
Munzer Auditorium, Beckman B060
Emotional Synapses
Joseph LeDoux
NYU
http://www-med.stanford.edu/sbrc/calendar/
4:15pm US-ASIA TECHNOLOGY MANAGEMENT CENTER PUBLIC LECTURE SERIES
Gates B01
Seamless Mobility for the Broadband Wireless Industry"
Goli Ameri
Founder & President, eTinium
http://asia.stanford.edu/events/fall02/index.html
Abstract below
5:30pm Stanford Phonology Workshop
Margaret Jacks Hall 460:126
The Armenian Consonant Shift Controversy
Luc Baronian
Linguistics, Stanford
http://www-linguistics.stanford.edu/pinterest/
Abstract below
5:30pm Stanford University Math Organization Talk
Bldg. 370:370
The Geometrization of 3-Dimensional Topology
Steve Kerckhoff
Math, Stanford
http://www.stanford.edu/group/sumo/
FRIDAY, 22 NOVEMBER 2002
10:00am Stanford Humanities Center Seminar
Stanford Humanities Center
Images of Subjectivity
Peter Galison
Harvard University
http://shc.stanford.edu/
11:00am SRI AI Seminar Series
EJ228, SRI International
Stereo using multiple views
Motilal Agrawal
http://www.ai.sri.com/seminars/
Abstract below
12 noon Ethics@Noon
Bldg. 100:101k
A Computer Scientist Looks at Ethics
Nils Nilsson
Computer Science, Stanford
http://www.stanford.edu/dept/EIS/noon.htm
12 noon Logical Methods in the Humanities
Bldg. 380:383N
Proof-theoretic semantics for classical mathematics
William W. Tait
Philosophy, University of Chicago
http://shc.stanford.edu/shc/2002-2003/02-03workshops.html
Abstract below
12:30pm CS547: Human-Computer Interaction Seminar
Gates B01
The Pebbles Project: Using Hand-Held Computers With Other Devices
Brad A. Myers
Human Computer Interaction Institute, Carnegie Mellon University
http://www-pcd.stanford.edu/cs547/
Abstract below
2:15pm NLP Reading Group
Margaret Jacks Hall, 460:301
A Graph Theoretic Model for Unsupervised Lexical Acquisition
Dominic Widdows
CSLI
http://www-nlp.stanford.edu/nlp/nlpgroup.html
Abstract below
3:15pm Philosophy Department Colloquium
Bldg. 90:92Q
Dialectic and Logic: The Truth of Axioms
William Tait
University of Chicago
http://www-philosophy.stanford.edu/ce.html
3:15pm ME394: Design Forum
Terman 556
Prof. Fritz Prinz on his research
http://me.stanford.edu/faculty/facultydir/prinz.html
http://www.stanford.edu/class/me394/
3:30pm Linguistics Department Colloquium
Margaret Jacks Hall 460:126
Blocking effects in voice: A bidirectional OT account
Judith Aissen
UC Santa Cruz
http://www-linguistics.stanford.edu/colloq/
Abstract below
MONDAY, 25 NOVEMBER 2002
3:00pm Reuters Foundation Digital Vision Program Seminar
Cordura Hall, Room 100
Technology and Terrorism
Abraham D. Sofaer
Senior Fellow, Hoover Institution and Professor of Law, Stanford
http://reuters.stanford.edu/seminar_speakers.html
Information below
4:00pm Berkeley Linguistics Department Colloquium
182 Dwinelle (Berkeley)
Directionality and output-driven phonology: the case of Hakha Lai
Larry M. Hyman & Kenneth VanBik
http://linguistics.berkeley.edu/lingdept/Current/events.html
Abstract below
4:15pm Broad Area Colloquium in AI,
Geometry, Graphics, Robotics, and Vision
TCSeq 200
Title to be announced
Marshall Bern
PARC
http://robotics.stanford.edu/ba-colloquium/
TUESDAY, 26 NOVEMBER 2002
4:15pm Logic Seminar
Bldg. 380:381T (math corner)
Classification Theory
Rami Grossberg
Carnegie-Mellon University
http://www-logic.stanford.edu/seminars.html
Abstract below
4:15pm SNRC Industry Seminar
Packard 101
Electronics and Information Technology -
Core Competencies for a Car Company
Klause Schaaf
VW USA
http://snrc.stanford.edu/events/industry-seminar/
Abstract below
THURSDAY, 28 NOVEMBER 2002
University Holiday - Eat well
FRIDAY, 29 NOVEMBER 2002
University Holiday - Sleep well
____________
Stanford Blood Bank status: Critical shortage of O-
and a shortage of A+, B+, and A-. For an appointment:
http://bloodcenter.stanford.edu/ or call 650-723-7831.
It only takes an hour of your time.
____________
REUTERS FOUNDATION DIGITAL VISION PROGRAM SEMINAR
on Wednesday, 20 November 2002, 3:00pm - 5:00pm
Cordura Hall, Room 100
http://reuters.stanford.edu/seminar_speakers.html
Collision Course:
Intellectual Property Rights and Traditional Knowledge
Douglas I. Kalish
Biotechnology Consultant and Educator, Palo Alto
Doug Kalish is an educator, consultant, and serial entrepreneur who
has founded or been an early executive in three companies. With over
30 years of IT, KM, and management experience, Doug consults in
technology forecasting, strategic technology planning, electronic
business, knowledge management, and biotechnology.
____________
COMPUTER SYSTEMS LABORATORY COLLOQUIUM
on Wednesday, 20 November 2002, 4:15pm
NEC Auditorium, Gates Computer Science Building B03
http://ee380.stanford.edu
MEMS on Glass: Technology with applications for fingerprint sensors
S. K. Ganapathi
President and CEO, Fidelica Microsystems
Fidelica Microsystems develops MEMS products in which low-cost at high
volume is the primary competitive advantage over products already in
the marketplace. The emphasis on cost has led Fidelica to develop its
first product, a low-power, compact, fingerprint sensor, for
production on glass substrates in factories that produce flat panel
displays (FPDs). An ASIC is flip-chip bonded to the substrate, so
that the substrate need not contain any active electronic elements.
The Fidelica approach has been to develop new designs that use
well-established methods and facilities for high-volume production.
By building MEMS devices on glass, Fidelica is able to exploit several
advantages over more traditional MEMS on silicon.
First, because the substrates are much larger, and cost less per unit
area to process, further miniaturization of device geometries is not
the only path to improved cost-competitiveness. This difference is
especially valuable in cases such as the fingerprint sensor, where
application, rather than process resolution, determines the device
size.
Second, because the resolution requirements are not extreme, the MEMS
devices can be produced in facilities that may not have the latest
photolithography tools, further reducing the costs, and providing
volume production flows to fabrication facilities that might already
be seeing falling demand.
Third, because the substrate need not include active electronics, the
choice of materials in the device is broadened, so that the material
for a layer might be better suited for the function of that layer.
Fourth, the optical transparency of the substrate opens the
possibility for several future enhancements of the fingerprint
sensor, in particular.
About the speaker: Dr. Ganapathi (K. G.) has a Ph. D. in Materials
Science and Engineering from Ohio State University. He started his
career as a post-doctoral research fellow at the Center for Magnetic
Recording Research at the University of California, San Diego. He
then served in various engineering management capacities with Applied
Magnetics Corporation and Read-Rite Corporation, before joining the
founding team of DAS Devices, Inc., a startup in thin film recording
heads. DAS was sold to Applied Magnetics Corporation in February
1999. Following the merger, Dr. Ganapathi served as Vice President
and General Manager of the San Jose division of Applied Magnetics,
with responsibility for all R, engineering, customer qualification,
and prototype operations. Dr. Ganapathi founded Fidelica Microsystems
in June 1999.
____________
STANFORD NETWORKING SEMINAR
on Thursday, 21 November 2002, 12:45pm (lunch 12:15pm)
Gates 104
http://netseminar.stanford.edu/
Optimizing the Migration of Virtual Computers
Costa Sapuntzakis
Stanford University
Tired of carrying a laptop? With the help of VMware, we can take the
state of your PC and move it around a network to a machine close to
you.
Tired of administering your own system? Over the network, we'll send
you a pre-configured PC with a mix of applications and OS that have
been extensively tested for compatibility. Oh, and that PC has the
latest security patches too.
PCs hold a huge amount of data; they have gigabytes of local disk and
hundreds of megabytes in RAM. How do we move so much data? In this
talk, I'll show you four ways to reduce how much travels over the
network. With these techniques and a DSL link, your PC state can move
between home and work while you commute.
About the speaker: Costa has been working towards his PhD at Stanford
University since 1998. Before that, he spent five years at MIT,
getting both a Bachelors and Masters in Computer Science. His recent
work includes the iSCSI block storage transport and the Collective
computing utility. His interests include programming languages,
operating systems, networks, and storage.
____________
UC BERKELEY CIS SEMINAR
on Thursday, 21 November 2002, 4:00pm - 5:30pm
Soda Hall 310, UC Berkeley
http://www.cs.berkeley.edu/~eyal/cis-seminar
Reorienting attitude -- rotating the right way
Berthold Horn
MIT
In machine vision, graphics, robot kinematics, protein folding, motion
planning, and so on, one needs to deal with rotations on a regular
basis. While there are few arguments over how to represent
translation, eight different notations for rotations are commonly
used. Why is rotation so much harder to deal with than translation?
It should be easy to compose rotations, interpolate between rotations,
sample rotations in a regular fashion (or randomly), integrate over a
set of rotations, compute an average of a function over a set of
attitudes, and speak of the "space of rotations". The notation should
be neither highly redundant, nor subject to complicated constraints or
singularities.
Euler angles (but which of the 24 "standard" definitions?),
orthonormal matrices, Gibb's vectors, axis and angle, and so on are
not satisfactory when viewed this way.
Hamilton's quaternions, on the other hand, provide a compact notation
that allows one to elegantly address these issues. Some problems that
had previously eluded analytic solution have been solved in closed
form using this notation. Examples will be presented in
photogrammetry and manipulator kinematics.
Quaternion notation is related to a number of other notations for
rotations, including Pauli spin matrices, Euler parameters, and
stereographic projection combined with a particular conformal
transformation in the complex plane.
Quaternion algebra lies in a series of algebras between that of
complex numbers and that of the octonions. The multiplication rule
for each of these algebras is based recursively on that of the next
lower level.
However, the further one goes in this series, the more desirable
properties are lost. So, quaternion multiplication is not
commutative, and octonion multiplication is not even associative.
An attempt has also been made to build an analysis for quaternions,
analogous to that for complex numbers -- alas with little success so
far.
____________
CSLI SEMINAR ON COMPUTATIONAL LEARNING AND ADAPTATION
on Thursday, 21 November 2002, 4:15pm to 5:30pm
Cordura 100
http://cll.stanford.edu/scla.html
Maintenance of Diagnostic Case-Based Reasoning Systems
Ioannis Iglezakis
DaimlerChrysler AG
During the last decade, case-based reasoning has evolved from initial
ideas that originated in cognitive science to a well established
intelligent technology suitable for various applications. One
consequence is that the focus of current case-based reasoning research
has moved from basic issues - like representation, acquisition,
retrieval, and indexing - toward the maintenance of large-scale
systems. However most methods for maintenance only consider removing
cases to restoring prediction accuracy. As a consequence, knowledge is
lost and prediction accuracy is viewed as the only requirement useful
for satisfying customers. This talk considers ways to improve cases
besides removing them and applies a methodology for maintaining the
knowledge of diagnostic case-based reasoning systems. The methodology
indirectly measures and monitors customer requirements and restores
them if necessary.
____________
SSP: SYMBOLIC SYSTEMS FORUM
on Thursday, 21 November 2002, 4:15pm
Bldg. 380:380C (Math Corner)
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
Conversational Interfaces: a Domain-Independent Architecture
for Task-Oriented Dialogues
or
Conversations with Robots, Cars, and Toasters
Alex Gruenstein
Symbolic Systems Program
Humans use natural language to get just about everything done in their
everyday lives, yet they are surrounded by electronic devices like
TVs, VCRs, and even powerful computers which can't understand a word
they say. At the same time, there are many groups of people working
on creating even more intelligent, complex devices capable of doing
everything from vacuuming on their own to flying helicopters
autonomously. As electronic devices become more complex and capable
there arises an immediate need for better interfaces, ones which allow
for humans to interact naturally with complex devices and agents.
Because humans use natural language so effortlessly, it emerges as one
clear choice for such an interface. Clearly, however, it would be
undesirable, complex, and expensive to design from scratch a natural
language interface for each new device which is invented. In this
talk, I will describe an architecture which can be quickly and easily
adapted to a wide variety of devices.
Draft available at: http://www.stanford.edu/~alexgru/thesis/
____________
US-ASIA TECHNOLOGY MANAGEMENT CENTER PUBLIC LECTURE SERIES
on Thursday, 21 November 2002, 4:15pm - 5:30pm
Gates Computer Science Bldg, Room B01, Stanford University
http://asia.stanford.edu/events/fall02/index.html
Seamless Mobility for the Broadband Wireless Industry
Goli Ameri
Founder and President, eTinium
There is a powerful trend toward seamless mobility in the wireless
industry where mobile professionals today and eventually all consumers
in the future will want to communicate and be able to do their daily
business anytime, anywhere. As a result, there is real demand for
ubiquitous connectivity between a wide variety of mobile devices and
access technologies, which (at least for now) include wireless WANs
and wireless LANs. Roaming and communications between these
technologies is therefore a must for seamless mobility to occur. To
realize the potential of seamless mobility and ensure continued
profitability, service providers have to focus equally on WLAN
implementations as they do on their cellular WWAN networks. Wi-Fi and
traditional wireless services are adjuncts that can co-exist and
succeed together and provide consumers what they want and when they
want it. Customers will use these technologies for different reasons
and at different times. The 2.5G and 3G technologies such as GPRS,
EDGE, CDMA 1XRTT and CDMA 1xEV-DO will be used for applications
requiring instant gratification and bursty data: e-mail, calendar
access, text messaging and MMS among others. But wireless LANs will
be used in specific locations where users need access to their
corporate files and Intranets. The talk will focus on 3G's business
and technical shortcomings and show how service providers can take
advantage of WLAN deployments to make up for 3G's revenue shortfall.
This talk will also cover the drivers and obstacles that must be
addressed to achieve growth in the wireless LAN market such as
roaming, billing, security, seamless authentication and handovers.
About the speaker: Goli Ameri is the founder and president of eTinium,
a telecom consulting and market research firm specializing in wireless
technologies. She is the author of over 30 wireless studies, the
latest of which is the upcoming study titled "Seamless Mobility: The
Marriage of 3G and Wireless LANs" to be published by the International
Engineering Consortium (IEC). Ms. Ameri is quoted regularly in such
publications as RCR Wireless News, Wireless Week, Internet Week, EE
Times, national Business Journals, the San Jose Mercury News, Multex
Investor, Internet Telephony, and the X-Change magazine amongst
others. She is invited regularly as a speaker to industry conferences
worldwide including CTIA, SuperNet, National Wireless Engineering
Conference, 802.11b and SMI Conferences in Europe. She also writes a
bi-monthly analysis column for Telephony magazine.
Free Admission, contact: viji@stanford.edu
____________
STANFORD PHONOLOGY WORKSHOP
on Thursday, 21 November 2002, 5:30pm
Margaret Jacks Hall 460:126
http://www-linguistics.stanford.edu/pinterest/
The Armenian Consonant Shift Controversy
Luc Baronian
Linguistics, Stanford
Standard Western Armenian (SWA) and Standard Eastern Armenian (SEA)
show inverse voicing of stops and affricates. Taking TH, T, D to
represent respectively voiceless aspirate, plain voiceless and plain
voiced stops and affricates, we get a SEA:SWA set of correspondences
of TH:TH; T:D; D:TH. While SEA shows the same system as Classical
Armenian (CA), SWA shows D in the same positions as
Proto-Indo-European (PIE). The voicing facts of SWA raise the
question of whether all modern dialects of Armenian are descendants of
CA or whether the SWA pronunciation reflects a prior stage of
Armenian, Proto-Armenian (PA), which was neither PIE nor CA. Under
this latter view, SWA's pronunciation would not be descendant of that
of CA, but of that of a dialect contemporary with CA. Under the other
view, the SWA pronunciation facts are the result of a sound shift that
operated after the CA stage, in other words, PA=CA.
Armenologists and tenants of the Glottalic theory for the
reconstruction of PIE tend to favor the PA=CA theory. Garrett (1998)
provides many arguments based on Armenian facts against the Glottalic
theory. However, he fails to address two problems that face the
PA=/=CA hypothesis: a problematic sound change needs to be postulated
from PA to Middle Armenian and the systematicity of voicing
correspondences in borrowings is left unexplained.
On the other hand, the PA=CA hypothesis can stand alone without the
Glottalic theory, but it then still faces two challenges, one of
economy, one of naturalness. The latter is the most problematic, as
it states that T > D regularly in SWA, a very unnatural sound change.
The general problem is that theorists have made little effort to go
beyond the facts that support their position, while Armenologists'
concern with the facts has made them ignore theoretical challenges.
Unsurprisingly then, the solution to the controversy lies in a careful
examination of all the facts, especially the status of borrowings in
dialects, and an explanation backed by a solid theory of sound change.
In this workshop, I will argue that theoretically PA=/=CA is the
preferable solution, but that careful dialectal comparison work is
still needed to confirm the hypothesis.
____________
SRI AI SEMINAR SERIES
on Friday, 22 November 2002, 11:00am
EJ228, SRI International
http://www.ai.sri.com/seminars/
Stereo using multiple views
Motilal Agrawal
The ability to automatically and accurately extract models of real
world using stereo images is an important problem in computer vision.
Applications range from virtual reality, human computer interaction to
automated navigation and surveillance. In this talk I will give a
brief overview of "state of the art" stereo algorithms. I will then
present a window-based binocular stereo algorithm using graph cuts.
Finally, I will present a probabilistic framework for stereo using
multiple views.
Note for Visitors to SRI: Please arrive at least 10 minutes early in
order to sign in and be escorted 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 E, and should call extension
2592 to be escorted to the meeting room.
____________
LOGICAL METHODS IN THE HUMANITIES
Stanford Humanities Center Workshop
on Friday, 22 November 2002, 12 noon - 1:15pm
Bldg. 380:383N
http://shc.stanford.edu/shc/2002-2003/02-03workshops.html
Proof-theoretic semantics for classical mathematics
William W. Tait
Philosophy, U. of Chicago
The picture of mathematics as being about constructing objects of
various sorts and proving the constructed objects equal or unequal is
an attractive one, going back at least to Euclid. On this picture,
what counts as a mathematical object is specified once and for all by
effective rules of construction.
In the last century, this picture arose in a richer form with
Brouwer's intuitionism. In his hands (for example, in his proof of the
Bar Theorem), proofs themselves became constructed mathematical
objects, the objects of mathematical study, and with Heyting's
development of intuitionistic logic, this conception of proof became
quite explicit. Today it finds its most elegant expression in the
Curry-Howard theory of types, in which a proposition may be regarded,
at least in principle, as simply a type of object, namely the type of
its proofs. When we speak of 'proof-theoretic semantics' for
mathematics, it is of course this point of view that we have in mind.
Much of my discussion applies equally to constructive mathematics. But
the type-theoretic point of view remains, for many people, restricted
to the domain of constructive mathematics. The term "classical" is
included in the title to indicate that, on the contrary, classical
mathematics can also be understood in this way and does not need to be
founded on an inchoate picture of truth-functional semantics in the
big-model-in-the-sky, a picture that can in any case never be
coherently realized.
About the speaker: W.W. Tait is Professor Emeritus, Department of
Philosophy and the Committee on Conceptual Foundations of Science,
University of Chicago
____________
CS547: HCI SEMINAR ON PEOPLE, COMPUTERS, AND DESIGN
on Friday, 22 November, 2002, 12:30pm - 2:00pm
Gates B01 (HP Classroom) and SITN
http://hci.stanford.edu/seminar
The Pebbles Project: Using Hand-Held Computers With Other Devices
Brad A. Myers
Human Computer Interaction Institute, Carnegie Mellon University
With today's wireless technologies, such as BlueTooth and IEEE 802.11,
connecting handheld computers and conventional computers together are
becoming no longer an occasional event for synchronization. Instead,
the devices are frequently in close, interactive communication. Many
environments, such as offices, meeting rooms and classrooms, already
contain computers, and the smart homes of the future will have
ubiquitous embedded computation. Household and office appliances will
soon have wireless communication abilities. When the user enters one
of these environments carrying a handheld or wearable computer, how
will that computer interact with the environment? The Pebbles project
is exploring the many ways that small handheld Personal Digital
Assistants (PDAs) such as PalmOS devices or Pocket PC / Windows CE
devices can serve as a useful adjunct to the "fixed" computers in
those situations. For meetings, our applications allow the presenter
to use a PDA to have better control of presentations, and allow the
audience to actively participate with their own PDAs. For the office,
other Pebbles applications allow the PDA to be used as an extra input
and output device. For the home, we are exploring the use of the PDA
as a customizable, intelligent "personal universal controller" (PUC)
for appliances. For classrooms, we are investigating how the
students' handhelds can enhance testing and notetaking when they are
connected to the instructor's PC. For the disabled, we are
investigating how PDAs can serve as assistive devices for access to
computers and appliances. This talk will provide an overview of our
Pebbles project, including a live demonstration of our systems
(available for download from our web site) and a discussion of future
plans. For more information, see http://www.pebbles.hcii..cmu.edu
About the speaker: Brad A. Myers is a Senior Research Scientist in
the Human-Computer Interaction Institute in the School of Computer
Science at Carnegie Mellon University. He is the author or editor of
over 230 publications, including three books, and he is on the
editorial board of five journals. He has consulted for over 40
companies on user interface design and implementation. Myers received
a PhD in computer science at the University of Toronto where he
developed the Peridot UIMS. He received the MS and BSc degrees from
the Massachusetts Institute of Technology during which time he was a
research intern at Xerox PARC. From 1980 until 1983, he worked at
PERQ Systems Corporation. His research interests include user
interfaces, handheld computers, programming languages for kids, User
Interface Development Systems, Programming by Example, Visual
Programming, interaction techniques, window management, and
programming environments. He belongs to SIGCHI, ACM, IEEE, IEEE
Computer Society, and Computer Professionals for Social Responsibility.
For this lecture on the Internet, look under Computer Science 547 in
http://scpd.stanford.edu/scpd/students/courseList.asp
For more information about the project in general use, see
http://interactivity.stanford.edu
____________
NLP READING GROUP
on Friday, 22 November 2002, 2:15pm - 3:30pm
Margaret Jacks Hall, Room 460:301
http://www-nlp.stanford.edu/nlp/nlpgroup.html
A Graph Theoretic Model for Unsupervised Lexical Acquisition
Dominic Widdows
CSLI
Hand-built lexical resources are costly to build and maintain, and
notoriously brittle when applied to new domain or tasks.
Automatically assembling semantic knowledge about the meanings of
individual words and phrases is increasingly important for the
scalability of NLP systems. We will give an overview of some of the
resources, techniques and models used for building and representing
lexical information, especially for nominals, before focusing on one
of the models we have developed at CSLI.
The paper presents an unsupervised method for assembling semantic
knowledge from a part-of-speech tagged corpus using graph algorithms.
The graph model is built by linking pairs of words which participate
in particular syntactic relationships. We focus on the symmetric
relationship between pairs of nouns which occur together in lists. A
trial lexicon is built for ten categories of words (such as "musical
instruments" or "diseases"), using a "prototypical" example as a
"seed". An incremental cluster-building algorithm using this model
achieves 82% accuracy at a lexical acquisition task, evaluated against
WordNet classes. This result is at least state-of-the-art and to our
knowledge, better than any other results at similar tasks. The model
naturally realizes domain and corpus specific ambiguities as distinct
components in the graph surrounding an ambiguous word.
The similarities and differences between our approach and those of
other researchers is much more interesting than just comparing
percentage accuracies. Models using different sizes of corpora,
different levels of linguistic processing (such as parsing to find
mutual dependencies), and different machine learning techniques appear
to produce strikingly different lexicons, and the future almost
certainly lies in successfully combining methods.
For an online demo or copy of this paper, see:
http://infomap.stanford.edu/graphs
http://infomap.stanford.edu/papers/lexical-graphs.pdf
http://infomap.stanford.edu/papers/lexical-graphs.ps
____________
LINGUISTICS DEPARTMENT COLLOQUIUM
on Friday, 22 November 2002, 3:30pm
Margaret Jacks Hall, Room 460:126
Blocking effects in voice: A bidirectional OT account
Judith Aissen
UC Santa Cruz
In recent years, arguments that optimization within OT must proceed in
two directions (production-directed and comprehension directed) have
come from various sources. These include asymmetries in the
linguistic abilities of children, learning algorithms, lexical
blocking effects, freezing effects, superiority effects, and anaphora.
These phenomena tend to be 'soft' in the sense that preferences can be
overridden by context. I consider here the distribution of active and
passive in a number of languages which are characterized by relatively
little surface configurationality and argue that a satisfactory
account requires bidirectional optimization. In such languages,
diathesis displays a highly structured pattern of interpretational
preferences and blocking effects, a pattern which can be characterized
if optimization is bidirectional, but not if it is unidirectional.
Interestingly, these effects are generally resistant to context and
plausibility. Voice thus provides evidence of a clearly syntactic
sort for bidirectional OT.
Food and drinks to follow.
____________
REUTERS FOUNDATION DIGITAL VISION PROGRAM SEMINAR
on Monday, 25 November 2002, 3:00pm-4:30pm
Cordura Hall, Room 100
http://reuters.stanford.edu/seminar_speakers.html
Technology and Terrorism
Abraham D. Sofaer
Senior Fellow, Hoover Institution
Professor of Law, Stanford
Abe Sofaer, who served as legal adviser to the U.S. Department of
State from 1985 to 1990, was appointed the first George P. Shultz
Distinguished Scholar and Senior Fellow at the Hoover Institution in
1994.
Named in honor of former U.S. secretary of state George P. Shultz, the
appointment is awarded to a senior scholar of international prominence
whose broad vision, knowledge, and skill can be brought to bear on the
problems presented by a radically transformed global environment.
Sofaer's work has focused on separation of powers issues in the
American system of government, including the power over war, and on
issues related to international law, terrorism, diplomacy, national
security, the Middle East conflict, and water resources. He teaches a
course on transnational law at the Stanford Law School. During his
distinguished career, Sofaer has been a prosecutor, legal educator,
judge, government official, and attorney in private practice.
____________
BERKELEY LINGUISTICS DEPARTMENT COLLOQUIUM
on Monday, 25 November 2002, 4:00pm
182 Dwinelle (Berkeley)
http://linguistics.berkeley.edu/lingdept/Current/events.html
Directionality and output-driven phonology:
the case of Hakha Lai
Larry M. Hyman & Kenneth VanBik
In this paper we present an analysis of the tone system of Hakha Lai,
a Kuki-Chin language spoken in Chin State, Burma and Mizoram State,
India. After presenting the basic system, we show how all of the tone
sandhi observed on bisyllabic combinations of the three tones in the
language, HL (falling), LH (rising) and L, are driven by a single
output constraint, which we call the No Contour Principle (NoCP).
While an output-driven (OT) account of the bisyllabic alternations is
straightforward (and revealing), the analysis self-destructs once we
consider trisyllabic and longer sequences of tones, where the NoCP is
regularly violated in two ways. The first violation results from the
right-to-left application of one of the tone sandhi rules. We show
that every one of the six principles that Chen (2000, 2002) invokes to
predict directionality (to which we add Howard's 1971 directional
iterative rule theory) would predict left-to-right directionality
(which, in turn, would not violate the NoCP) rather than the observed
right-to-left. The second type of violation of the NoCP arises from
what would be accounted for by rule ordering in a derivational
framework. Various "fix-ups" are considered and rejected. We then
argue for a direct mapping approach similar to Goldsmith (1993) and
Lakoff (1993), but where cross-level input/output constraints
("rules") are ranked, as in OT. In comparison with the (rejected)
output-driven analyses, the direct-mapping analysis captures the Hakha
Lai tonal facts with great simplicity (and in a psycholinguistically
plausible way). We conclude by considering possible objections to such
an approach and some of the implications of our
findings--specifically, the question of whether the violable universal
constraints (or "ideals") of OT ought to be in the grammar at all.
____________
LOGIC SEMINAR
on Tuesday, 26 November 2002, 4:15 p.m
Math Corner 380:381T
Classification Theory
Rami Grossberg
Carnegie-Mellon University
The most important development in the past 40 years in model theory is
in establishing the field of Classification Theory (formerly known as
Stability Theory). This turned out to be an important bridge between
powerful set theoretic methodology, primarily combinatorial set theory
(and to a lesser extent independence results) and classical branches
of mathematics like number theory.
I will attempt to survey some of the deeper developments and concepts
from a modern point of view, starting with Morley and ending with
Shelah. I will keep the prerequisites to a minimum. Less than the
first few weeks of an undergraduate course in mathematical logic will
be enough.
____________
SNRC INDUSTRY SEMINAR
on Tuesday, 26 November 2002, 4:15pm
Packard EE 101
http://snrc.stanford.edu/events/industry-seminar/
Electronics and Information Technology:
Core Competencies for a Car Company
Klaus Schaaf
Research Director, Electronics Research Lab, VW America
In the future about 80% of all new developments for cars will be
influenced, if not dominated by electronics and information
technologies. The Volkswagen group has embraced this trend and has
developed an electronics strategy to make sure that Volkswagen will
stay one of the most competitive car companies.
In my talk I will focus on these challenges, show how the Volkswagen
group already has dealt with them and how the ERL is an important
keystone to make these new technologies available to the Volkswagen
group. I will present some new features which will come to Volkswagen
cars soon and which show how electronics will play an ever more
important part in the future.
About the speaker: Klaus Schaaf is heading the research activities of
the Electronics Research Lab (ERL) of Volkswagen since summer of 2001.
He has been with Volkswagen since 1988. Before he came to Silicon
Valley he was head of Acoustic Research at Volkswagen headquarters in
Wolfsburg, Germany. He holds a Ph.D. from the University of
Goettingen in Applied Physics.
Food and drink will be provided in a small reception from 5:15-6pm.
____________
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
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 (i.e., no abstracts). 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/Archive/calendar/current.shtml
and the archives at
http://www-csli.stanford.edu/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.
and
news://news.stanford.edu/su.events
Information about CSLI's research program is available at
http://www-csli.stanford.edu/
For maps to the Stanford University campus see
http://www.stanford.edu/home/visitors/maps.html
____________