[Date Prev][Date Next][Date Index]
CSLI Calendar, Wednesday, 22 January 2003, vol. 18:17
CSLI CALENDAR OF PUBLIC EVENTS
______________________________________________________________________
22 January 2003 Stanford Vol. 18, No. 17
______________________________________________________________________
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 22 JANUARY 2003 TO 31 JANUARY 2003
WEDNESDAY, 22 JANUARY 2003
3:00pm Reuters Foundation Digital Vision Program Seminar
email for location mailto:digitalvision@csli.stanford.edu
Chris Handley
Chief Information Officer, Stanford University
http://reuters.stanford.edu/seminar_speakers.html
3:45pm Psychology Department Colloquium
Jordan Hall 420:041
Functional MRI of visual cortex:
Binocular rivalry, perceptual learning, and letter recognition
Steve Engel
UCLA
http://www-psych.stanford.edu/news.html
4:15pm EE380: Computer Systems Lab Colloquium
Gates B03 (NEC Auditorium)
Solving High Technology Crime
Gregory S. Crabb
United States Postal Inspector
San Francisco Electronic Crimes Task Force
http://ee380.stanford.edu/contents.html
Abstract below
THURSDAY, 23 JANUARY 2003
11:00am Music 319: CCRMA Hearing Seminar
CCRMA Library, The Knoll
A Cognitive Account of Two Musical Revolutions
David Huron
Ohio State
http://mambo.ucsc.edu/psl/ccrmas/ccrmas.html
Abstract below
12:15pm Stanford Networking Seminar
Gates 104
Clarifying the Fundamentals of HTTP
Jeff Mogul
HP Labs
http://netseminar.stanford.edu/
Abstract below
4:00pm Personality Seminar
Jordan Hall 420:100
Title to be announced
Larissa Tiedens
Stanford GSB
http://www-psych.stanford.edu/news.html#person_lab
4:00pm PARC Forum
George Pake Auditorium at PARC
20,000 Bytes Under the Sea
Emory Kristof
Photographer-in-Residence, National Geographic Society
http://www.parc.com/forum/
4:00pm UC Berkeley BISC Seminar
320 Soda Hall, Berkeley
An Approach to the Mathematical Theory of Perception-Based Information
Victor Korotkikh
Central Queensland University & BISC Program
http://buffy.eecs.berkeley.edu/Seminars/
Abstract below
4:00pm UC Berkeley CIS Seminar
Soda Hall 310 (UC Berkeley)
Generalized Principal Component Analysis (GPCA): an analytic
approach to segmentation of static and dynamics scenes
Rene Vidal
UC Berkeley
http://www.cs.berkeley.edu/~eyal/cis-seminar
Abstract below
4:15pm CSLI Seminar on Computational Learning and Adaptation
Cordura Hall, room 100
Boosting with Averaged Weight Vectors
Nikunj Oza
Computational Science Division, NASA Ames Research Center
http://cll.stanford.edu/scla.html
Abstract below
4:15pm SSP10: Symbolic Systems Forum
Bldg. 380:380C (Math Corner)
Instructive Signals in the Brain
Jennifer Raymond
Neurobiology, Stanford University
http://sbrc.stanford.edu/faculty/sbrc_fac_list/raymond.html
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
Abstract below
4:15pm Neurobiology of Disease Seminar Series
Munzer Auditorium, Beckman B060
Insights into the Neuropathology of Schizophrenia
Edward G. Jones
UC Davis
http://www-med.stanford.edu/sbrc/calendar/
FRIDAY, 24 JANUARY 2003
12 noon EE Special Seminar
Packard EE 202
Suprathreshold Visual Psychophysics and Applications to Image
Compression
Sheila S. Hemami
Electrical & Computer Engineering, Cornell University
http://foulard.ece.cornell.edu
Abstract below
12:30pm CS547: Human-Computer Interaction Seminar
Gates B03
Issues in Personalizing Shared Ubiquitous Devices
David Hilbert and Jonathan Trevor
FX Palo Alto Laboratory
http://www-pcd.stanford.edu/cs547/
Abstract below
2:15pm NLP Reading Group
Math Corner, 380:383P
Computational Morphology
Lauri Karttunen
PARC
http://www-nlp.stanford.edu/nlp/nlpgroup.html
Abstract below
3:15pm Friday Cognitive Seminar
Jordan Hall 420:100
Perception of Action
Elizabetta Zibetti
http://www-psych.stanford.edu/news.html#frisem
3:30pm Semantics and Pragmatics Workshop
Margaret Jacks Hall 460:126
Thoughts on semantic universals and semantic variation
Lisa Matthewson
British Columbia
http://www-linguistics.stanford.edu/semgroup/
Abstract below
4:15pm CS545: Database Seminar
Gates B12
Incremental Validation of XML Databases
Yannis Papakonstantinou
U.C. San Diego
http://www-db.stanford.edu/dbseminar/
Abstract below
MONDAY, 27 JANUARY 2003
3:30pm Social Lab
Jordan Hall 420:050
Social Factors In the Development of Expert Performance:
Why Professional Hockey Players are Five Times More Likely to
be Born in January than December and Other Curious Social Phenomena.
K. Anders Ericsson
Florida State University
http://www-psych.stanford.edu/news.html#social_lab
4:00pm Berkeley Linguistics Department Colloquium
182 Dwinelle (Berkeley)
The search for historical African American English:
Evidence from Mississippi-in-Africa
John Victor Singler
New York University
http://linguistics.berkeley.edu/lingdept/Current/events.html
4:15pm Broad Area Colloquium in AI,
Geometry, Graphics, Robotics, and Vision
TCSeq 201
Shape Recipes
William Freeman
MIT
http://robotics.stanford.edu/ba-colloquium/
Abstract below
7:00pm Presidential Lecture Series
Law School, Room 190
Scholarship and Imagination: The Study of Late Antiquity
Peter R. L. Brown
Rollins Professor of History
Princeton University
http://prelectur.stanford.edu/lecturers/brown/
TUESDAY, 28 JANUARY 2003
5:30pm Syntax Workshop
Margaret Jacks 460:126
Subsumption and Equality: German Partial Fronting in LFG
Annie Zaenen and Ron Kaplan
Palo Alto Research Center
http://www-linguistics.stanford.edu/sssg/
Abstract below
WEDNESDAY, 29 JANUARY 2003
3:00pm Reuters Foundation Digital Vision Program Seminar
email for location mailto:digitalvision@csli.stanford.edu
Larry Lessig
Law, Stanford University
http://reuters.stanford.edu/seminar_speakers.html
Information below
3:45pm Psychology Department Colloquium
Jordan Hall 420:041
Title to be announced
Stanka Fitneva,
Cornell
(Developmental Job Candidate)
http://www-psych.stanford.edu/news.html
4:15pm EE380: Computer Systems Lab Colloquium
Gates B03 (NEC Auditorium)
Chandler: An (Inter)personal Information Manager
Mitch Kapor
Open Source Applications Foundation
http://ee380.stanford.edu/contents.html
Abstract below
7:00pm SCIL Futures of Learning Lecture Series
Wallenberg Hall (Bldg. 160)
Beyond Accountability: Assessment That Supports Adventuresome
Teaching Can assessment fortify innovative teaching and learning?
John D. Bransford
Psychology and Education, Vanderbilt University
http://scil.stanford.edu/
Abstract below
THURSDAY, 30 JANUARY 2003
12:15pm Stanford Networking Seminar
Gates 104
Title to be announced
Sundar Iyer
Stanford University
http://netseminar.stanford.edu/
4:00pm Personality Seminar
Jordan Hall 420:100
Title to be announced
Kelly McGonigal
Stanford
http://www-psych.stanford.edu/news.html#person_lab
4:00pm PARC Forum
George Pake Auditorium at PARC
Kavita Ramdas
President, The Global Fund for Women
http://www.globalfundforwomen.org/
http://www.parc.com/forum/
4:15pm CSLI Seminar on Computational Learning and Adaptation
Cordura Hall, room 100
A stability based method for discovering structure in
clustered data
Asa Ben-Hur
Biochemistry, Stanford University
http://cll.stanford.edu/scla.html
Abstract below
4:15pm SSP10: Symbolic Systems Forum
Bldg. 380:380C (Math Corner)
Why don't we have a theory of higher level brain function and
what can we do about it
Jeff Hawkins
Redwood Neuroscience Institute
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
Abstract below
4:15pm Neurobiology of Disease Seminar Series
Munzer Auditorium, Beckman B060
Nogo Receptor and Axonal Regeneration
Stephen Strittmatter
Yale University
http://www-med.stanford.edu/sbrc/calendar/
FRIDAY, 31 JANUARY 2003
11:00am Berkeley Institute of Cognitive and Brain Seminar
Tolman 5101 (Berkeley)
Gary Marcus
Psychology, New York University
http://socrates.berkeley.edu:4247/seminar.html
12:30pm CS547: Human-Computer Interaction Seminar
Gates B03
Using Words to Search a Thousand Images:
Hierarchical Faceted Metadata in Search Interfaces
Marti Hearst
UC Berkeley SIMS
http://www.sims.berkeley.edu/~hearst/
http://www-pcd.stanford.edu/cs547/
3:15pm Friday Cognitive Seminar
Jordan Hall 420:100
Everything You've Ever Wanted to Know About Discounting of
Perceptual Fluency in Heuristic Judgment but Were Afraid to Ask
Danny Oppenheimer
http://www-psych.stanford.edu/news.html#frisem
3:30pm Semantics and Pragmatics Workshop
Margaret Jacks Hall 460:126
The Semantics of Incorporation
Donka Farkas, University of California, Santa Cruz
(work done with Henriette de Swart)
http://www-linguistics.stanford.edu/semgroup/
4:15pm CS545: Database Seminar
Gates B12
Data Routing Rather than Databases: The Meaning of the Next Wave
of the Web Revolution to Data Management
Adam Bosworth
BEA
http://www-db.stanford.edu/dbseminar/
____________
Stanford Blood Center status: Critical shortage of O- and a
shortage of everything else. For an appointment:
http://bloodcenter.stanford.edu/ or call 650-723-7831. It only takes
an hour of your time.
____________
ANNOUNCEMENT
FOURTH ANNUAL STANFORD SEMANTICS FEST
Friday, March 14, 2003
Cordura 100
Stanford University
The Construction of Meaning Workshop is pleased to sponsor the Fourth
Annual Stanford Semantics Fest. The Semantics Fest is once again
intended to serve as a forum for promoting discussion and
collaboration among all those in the Stanford community interested in
the semantics and pragmatics of natural language, as well as their
interface with other modules of grammar. We encourage contributions
by all who share these interests.
Papers (most likely 20 minute talks plus 10 minutes of discussion) are
invited on any topic touching on semantics or pragmatics in natural
language.
Submission of Abstracts:
Abstracts should be submitted by Friday, February 4, 2003. All
abstracts should be in plain text and no more than 500 words long (not
including references). E-mail submissions are preferred. E-mail
abstracts to Judith Tonhauser, juditht@stanford.edu.
Deadlines and Important Dates:
Deadline for abstracts: Monday, Feb 3, 2003
Notification of acceptance: Friday, Feb 14, 2003
Deadline for revised abstracts: Friday, Feb 28, 2003
Semantics Fest: Friday, March 14, 2003
Organizing Committee:
David Beaver, Ivan Garcia, Beth Levin, Judith Tonhauser
This event is sponsored by the Stanford Humanities Center and funded
by a grant from the Mellon Foundation.
____________
EE380: COMPUTER SYSTEMS LAB COLLOQUIUM
on Wednesday, 22 January 2003, 4:15pm to 5:30pm
NEC Auditorium (B03), Gates Computer Science Building
http://ee380.stanford.edu/contents.html
Solving High Technology Crime
Academic Partnership in Crime Fighting
Gregory S. Crabb
United States Postal Inspector
San Francisco Electronic Crimes Task Force
Other participants include:
Robert Rodriguez, Assistant Special Agent in Charge, United States
Secret Service
Richard Perlotto, Cisco Systems
Chris Lalone, Network Security, eBay
Mike Miravalle, CEO, Dolphin Technologies
Fred Demma, Dolphon Technologyies
The San Francisco Electronic Crimes Task Force seeks to engage the
academic community to help us address the technology crimes affecting
our community, our corporate partners and law enforcement. The crimes
affecting our corporate partners include computer hacking,
intellectual property crimes (criminal trademark and copyright
infringement) and identity theft. These crimes are costing the high
technology community billions of dollars and stunting the acceptance
and growth of these technologies to support our economy. Antiquated
investigative methods and poor individual accountability for Internet
communications are some of the greatest challenges facing law
enforcement. The solution to some of these challenges may lie within
the academic community.
The talk will focus on several brief case studies relating our
greatest challenges in fighting high technology crime. Each case study
will be presented by a law enforcement agent and/or corporate partner
of the task force.
About the speaker: The San Francisco Electronic Crimes Task Force is a
group of Federal, state, local investigators and corporate partners,
lead by the U.S. Secret Service, focused on attacking high technology
crime affecting Bay Area companies, locally and globally. The task
force is part of the Secret Service's nation-wide network of
electronic crimes task forces, see http://www.ectaskforce.org .
____________
MUSIC 319: CCRMA HEARING SEMINAR
on Thursday, 23 January 2003, 11:00am
CCRMA Library, The Knoll
http://mambo.ucsc.edu/psl/ccrmas/ccrmas.html
A Cognitive Account of Two Musical Revolutions
David Huron
Ohio State University (and Stanford CCARH :-)
Two major historical revolutions in Western music are analyzed from a
cognitive perspective. One revolution is the displacement of the
Medieval system of modes by the Renaissance major/minor system. An
empirical study of Gregorian chant shows that, by the 12th century,
the eight-fold system of modes was already poised to collapse into two
general schemas -- the precursors of "major" and "minor." This
phenomenon is akin to phonetic "merger" in linguistics.
A second musical revolution is associated with the advent of
"Modernism." Three empirical studies are presented that examine the
compositional strategies of Wagner, Schoenberg and Stravinsky. In
each case, these composers created music consistent with a "reverse
psychology" in which listener expectations were systematically
transgressed: Wagner with respect to cadential expectations,
Schoenberg with respect to tonal expectations, and Stravinsky with
respect to metric expectations. Inexperienced listeners find these
transgressions unsettling, but experienced listeners internalize this
strategy and come to expect the unexpected.
____________
STANFORD NETWORKING SEMINAR
on Thursday, 23 January 2003, 12:40pm (lunch 12:15pm)
Gates 104
http://netseminar.stanford.edu/
Clarifying the Fundamentals of HTTP
Jeff Mogul
HP Labs
The simplicity of HTTP was a major factor in the success of the Web.
However, as both the protocol and its uses have evolved, HTTP has
grown complex. This complexity results in numerous problems, including
confused implementors, interoperability failures, difficulty in
extending the protocol, and a long specification without much
documented rationale.
Many of the problems with HTTP can be traced to unfortunate choices
about fundamental definitions and models. I will analyze the current
(HTTP/1.1) protocol design, showing how it fails in certain cases, and
how to improve these fundamentals. Some problems with HTTP can be
fixed simply by adopting new models and terminology, allowing us to
think more clearly about implementations and extensions. Other
problems require explicit (but compatible) protocol changes.
About the speaker: Jeffrey C. Mogul received an S.B. from the
Massachusetts Institute of Technology in 1979, an M.S. from Stanford
University in 1980, and his PhD from the Stanford University Computer
Science Department in 1986. Dr. Mogul has been an active participant
in the Internet community, and is the author or co-author of several
Internet Standards; he contributed extensively to the HTTP/1.1
specification. From 1986 to 2002, he was a researcher at Digital's
(and Compaq's) Western Research Laboratory, and is now at HP Labs,
working on network and operating systems issues for large-scale
computer systems, and on improving performance of the Internet and the
World Wide Web. Jeff is a Fellow of the ACM, and a member of Sigma Xi
and CPSR. He was Program Committee Chair for the Winter 1994 USENIX
Technical Conference, the IEEE TCOS Sixth Workshop on Hot Topics in
Operating Systems, and the Second Workshop on Industrial Experiences
with Systems Software.
____________
BERKELEY BISC SEMINAR
on Thursday, 23 January 2003, 4:00pm-5:00pm
Tolman 5101 (Berkeley)
http://buffy.eecs.berkeley.edu/Seminars/
An Approach to the Mathematical Theory of Perception-Based Information
Victor Korotkikh
Central Queensland University & BISC Program
Fundamental ideas by Lotfi Zadeh about perception-based information
open a way to explore new avenues and raise deep questions. One of the
most important is the mathematical theory of perception-based
information.
The laws of human information processing may be still far away from
our knowledge and experience. It seems that we would be in a better
position if we could attempt to understand them within an irreducible
theory, i.e., a theory not allowing a deeper explanatory base. In this
case there would be no further reductions possible. In other words,
nobody could ask to explain such a theory and its basic notions in
terms of a more profound theory. Clearly, these very strong
requirements challenge our belief that we could find an irreducible
theory to start with. For this purpose even basic notions originated
from space-time cannot be viewed irreducible because it is
increasingly regarded as an approximation to reality.
In the paper we approach the mathematical theory of perception-based
information by using a theory that captures a new type of processes.
These processes come to life in imaginative way, i.e., they are the
formations of integer relations. The formations have a natural order:
starting with integers, integer relations as elements of a level form
more complex integer relations as elements of the next one. The theory
of integer relations and their formations is irreducible. The
existence of integer relations and their formations, based on a
universal principal, are completely determined by arithmetic. In the
theory integers appear with a new meaning: they are the ultimate
building blocks from which the integer relations are formed.
The integer relations are true propositions. The formations of the
integer relations set up a structure of these propositions. In this
structure each proposition is given by how it is formed or inferenced
and is a building block in the formation of other propositions. The
structure has natural ability to integrate information. For example,
two pieces of information represented in the structure as propositions
may turn to be components of a process forming new propositions.
There is an isomorphism between integer relations and their formations
from one side and two-dimensional geometrical patterns and their
formations from another. Consequently, an integer relation or
proposition of the structure can be viewed as a geometrical pattern.
This geometrical pattern results from a universal transformation
applied to geometrical patterns of propositions from which the
proposition is inferenced. The area of the geometrical pattern can
measure the proposition.
The isomorphism opens a way to view propositions and their inference
in geometrical terms. This allows us to consider geometrical
transformations that make propositions and their inference imprecise
but satisfy certain properties. In this context a perception may be
defined as a class of geometrical patterns with the geometrical
pattern of a corresponding proposition at a center of the class. The
geometrical transformations may be seen as operations for processing
of the perceptions and suggest an approach to model human information
processing.
An application in combinatorial optimization is considered.
____________
UC BERKELEY CIS SEMINAR
on Thursday, 23 January 2003, 4:00pm
Soda Hall 310 (UC Berkeley)
http://www.cs.berkeley.edu/~eyal/cis-seminar
Generalized Principal Component Analysis (GPCA):
An analytic approach to segmentation of static and dynamics scenes
Rene Vidal
UC Berkeley
Segmentation is usually though of as a "chicken-and-egg" problem. In
order to estimate a mixture of models one needs to first segment the
data and in order to segment the data one needs to know the model for
each class. Therefore, segmentation is usually solved in two-stages
(1) data clustering and (2) model fitting, or else iteratively using
the Expectation Maximization (EM) algorithm.
In this talk, we will show that for a wide class of segmentation
problems (mixtures of subspaces, eigenvector segmentation, mixtures of
affine models, mixtures of fundamental matrices), the
"chicken-and-egg" dilemma can be completely solved using
algebraic-geometric techniques. In fact it is possible to use all the
data simultaneously to recover all the models without previously
segmenting the data. In the absence of noise this can be done in
polynomial time using linear techniques. Furthermore the solution is
closed form if and only if the number of groups is less than or equal
to 4. In the presence of zero-mean Gaussian noise, the algebraic
solution leads to an optimal objective function that depends on the
model parameters and not on the segmentation of the data.
We present applications on intensity and motion (3D and affine) based
image segmentation, and compare with K-means and EM.
____________
CSLI SEMINAR ON COMPUTATIONAL LEARNING AND ADAPTATION
on Thursday, 23 January 2003, 4:15pm to 5:30pm
Cordura 100
http://cll.stanford.edu/scla.html
Boosting with Averaged Weight Vectors
Nikunj C. Oza
Computational Sciences Division, NASA Ames Research Center
AdaBoost is a well-known ensemble learning algorithm that constructs
its constituent or base models in sequence. A key step in AdaBoost is
constructing a distribution over the training examples to create each
base model. This distribution, represented as a vector, is constructed
to be orthogonal to the vector of mistakes made by the previous base
model in the sequence. The idea is to make the next base model's
errors uncorrelated with those of the previous model. Some researchers
have pointed out the intuition that it is probably better to construct
a distribution that is orthogonal to the mistake vectors of all the
previous base models, but that this is not always possible. We present
an algorithm that attempts to come as close as possible to this goal
in an efficient manner. We present results demonstrating significant
improvement over AdaBoost and the Totally Corrective boosting
algorithm, which also attempts to satisfy this goal.
____________
SSP10: SYMBOLIC SYSTEMS FORUM
on Thursday, 23 January 2003, 4:15pm
Bldg. 380:380C (Math Corner)
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
Instructive Signals in the Brain
Jennifer Raymond
Neurobiology, Stanford University
http://sbrc.stanford.edu/faculty/sbrc_fac_list/raymond.html
Learning involves a change or set of changes in the brain. What are
the neural events that trigger such changes? Which neurons carry the
instructive signals that guide the cellular events underlying
learning? We are using a simple motor learning task as an
experimental system for exploring these questions.
____________
EE SPECIAL SEMINAR
on Friday, 24 January 2003, 12 noon
Packard EE 202
Suprathreshold Visual Psychophysics and
Applications to Image Compression
Sheila S. Hemami
School of Electrical & Computer Engineering, Cornell University
http://foulard.ece.cornell.edu
Image compression has reached the peak of efficiency in terms of
treating images as traditional "signals," employing efficient
transformations, correlation-based models, and context-based coding.
Human visual system characteristics have been successfully applied to
high-rate signal-based compression, where stimuli such as
compression-induced distortions are below the visibility threshold;
these distortions and this compression regime are called subthreshold.
Operation of such signal-based compression algorithms in the
suprathreshold regime (i.e., low-rate compression), in which
compression-induced distortions are clearly visible, has to date
operated based on visual system rules-of-thumb and has produced
moderate success. However, a unifying approach to produce
higher-quality low-rate images is lacking. In this talk I will
describe our work toward a robust, unified approach to psychovisually
motivated low-rate image compression. Experimentation and results
have produced a robust distortion allocation strategy based on RMS
contrast which is applicable to wavelet-based compression and can be
used successfully over the entire range of compression ratios.
About the speaker: Sheila S. Hemami received her PhD in EE at Stanford
University in 1994. She is currently an Associate Professor of
Electrical Engineering and Director of the Visual Communications Lab
at Cornell University.
____________
CS547: HUMAN-COMPUTER INTERACTION SEMINAR
on Friday, 24 January 2003, 12:30-2:00pm
Gates B03
http://www-pcd.stanford.edu/cs547/
Issues in Personalizing Shared Ubiquitous Devices
David Hilbert and Jonathan Trevor
FX Palo Alto Laboratory
Everywhere we go, we are surrounded by shared devices: TVs, stereos,
and appliances in the home; copiers, fax machines, and projectors in
the office; phones and vending machines in public. Because these
devices don't know who we are, they provide the same interface and
functionality to everyone. This lack of personalization in the real
world is reminiscent of the World Wide Web in its infancy--no matter
who you were, you saw the same Web pages as everyone else. Today,
personalization has made the Web more friendly, efficient, and
profitable. Our research seeks to personalize shared real world
devices to reap some of the same benefits. This talk covers three
topics: (1) the design of the PIPs system for personalizing multi-user
document devices such as projectors, shared displays, and
multi-function copiers; (2) the experimental method we employed, which
is applicable to other HCI, CSCW, and ubiquitous systems research; and
(3) how data collected over the past two years has helped us refine
and evolve our research and prototypes.
About the speakers: David Hilbert is a research scientist at FX Palo
Alto Laboratory. His research interests are in the design and
evaluation of practical HCI, CSCW, and ubiquitous computing
applications. He received a PhD in information and computer science
from the University of California, Irvine. He is a member of IEEE,
ACM, and the Phi Beta Kappa Society.
Jonathan Trevor is a senior research scientist at FX Palo Alto
Laboratory, where he works in ubiquitous systems and
computer-supported cooperative work. His current research interests
are in the development of readily accessible groupware and
human-computer interaction applications across a range of technologies
and platforms. He received a PhD in computer science from the
University of Lancaster.
____________
NLP READING GROUP
on Friday, 24 January 2003, 2:15pm
Math Corner, 380:383P
http://www-nlp.stanford.edu/nlp/nlpgroup.html
Computational Morphology
Lauri Karttunen
Palo Alto Research Center
Computational Morphology is concerned with the analysis and generation
of words. The analysis of an inflected surface form such as "leaves"
maps it to one or more lexical representations (= lemmas) such as
"leaf"-Noun Plural, "leave"-Noun Plural, and "leave"-Verb Sg3. The
generation from a lexical form such as "hang"-Verb Past produces one
or more surface forms such as "hanged" and "hung".
The two challenges in modeling natural language morphology are:
Morphotactics
Words are typically composed of smaller units: stems and affixes that
must be combined in a certain order. Most languages build words by
concatenation but some languages also exhibit non-concatenative
processes such as interdigitation (Arabic) and reduplication (Malay).
Morphological Alternations
The shape of the components that make up a word often depends on
their environment. For example, the comparative degree of English
Adjectives is sometimes expressed by -"er", sometimes by -"r", and
the stem may also vary, as in "bigger".
Computational linguists generally take it for granted that the
relation between the surface forms of a language and their
corresponding lemmas can be described as a *regular relation*. If the
relation is regular, it can be defined using the metalanguage of
regular expressions; and, with a suitable compiler, the regular
expression source code can be compiled into a finite-state transducer
that implements the relation computationally. In the resulting lexical
transducer, each path (= sequence of states and arcs) from the initial
state to a final state represents a mapping between a surface form and
its lemma.
Comprehensive lexical transducers have been created for a great number
of languages including most European languages, Turkish, Arabic,
Korean, and Japanese among others. They are commercially available
through companies such as Inxight. Lexical transducers are generally
compact in size (a few hundred kilobytes) and they can be applied for
analysis and generation at very high speeds (thousands of words per
second).
The success of finite-state morphology has so far had no impact within
linguistics as an academic discipline. Practical issues that arise in
the context of real-life applications such as completeness of
coverage, physical size, and speed of applications are irrelevant from
an academic morphologist's point of view. The main purpose of a
morphologist writing to an audience of fellow linguists is to be
convincing that his theory of word formation provides a more
insightful and elegant account of this aspect of the human linguistic
endowment than the competing theories and formalisms.
In this talk I will demonstrate that the linguistic approach known as
*realizational morphology* (Zwicky, Anderson, Stump), as presented by
Stump, is in fact a finite-state model. Stump's recent book,
"Inflectional Morphology" (2001), lays out a rich set of principles
and notational conventions designed to capture important linguistic
generalizations. In spite of the apparent complexity of Stump's
formalism, the system as a whole is no more powerful than a collection
of regular relations. Consequently, a Stump-style description of the
morphology of a particular language such as Lingala or Bulgarian can
be compiled into a finite-state transducer that maps the underlying
lexical representations directly into the corresponding surface forms
or forms, and vice versa, yielding a single lexical transducer. It
turns out that this theory is a finite-state model that does not know
its name. It takes no advantage of the nice formal and computational
properties it in fact has.
____________
STANFORD SEMANTICS AND PRAGMATICS WORKSHOP
on Friday, 24 January 2003, 3:30pm
Margaret Jacks Hall 460:126
http://www-linguistics.stanford.edu/semgroup/
Thoughts on semantic universals and semantic variation
Lisa Matthewson
British Columbia
This talk tackles the following two-part question: What is universal
in semantics, and how can semantics vary cross-linguistically? Based
partly on two specific case studies (determiners and tense in Lillooet
Salish), and partly on conceptual reasoning, I'll put forward the
three claims in (1).
(1) i. In both syntax and semantics, our null hypothesis should be
that all languages are identical.
ii. Semantic variation is not necessarily tied to syntactic
variation; the former exists even in the absence of the latter.
iii. A theory that admits semantic parameters is more restrictive
than one which doesn't.
The basic idea is that semantic theory should not abdicate to
syntactic theory all responsibility for setting limits on variation.
Once we admit that semantics can vary in ways which do not necessarily
derive from differing syntactic structures, we can begin to articulate
constraints on semantic variation; we will then end up with a more
restricted theory of both semantics and syntax. Finally, the case
studies will also lead to the substantive claim given in (2).
(2) The inventory of functional projections is invariant
cross-linguistically; the lexical semantics of functional heads
varies cross-linguistically.
____________
CS545: DATABASE SEMINAR
on Friday, 24 January 2003, 4:15pm - 5:15pm
Gates B12
http://www-db.stanford.edu/dbseminar/
Incremental Validation of XML Databases
Yannis Papakonstantinou
U.C. San Diego
We discuss algorithmic and systems issues on the incremental schema
validation of XML databases with respect to DTDs, XML Schemas and
XQuery's type system under updates consisting of element tag
renamings, insertions and deletions. For DTDs we provide a worst-case
O(m log n) incremental validation algorithm using an auxiliary
structure of size O(n), where n is the size of the document and m is
the number of updates. Note that a brute force solution requires O(n)
steps: following an update, we run the updated lists of elements
through automatons that correspond to regular expressions in the
DTD. For XML Schemas and XQuery types, the problem is harder, since an
update to a single node may have global repercussions for the typing
of the tree. We provide a worst-case O(m log^2 n) algorithm that is
based on the use of tree automata and an O(n) auxiliary data
structure. Next we provide experimental results showing that for real
DTDs the worst case is relatively rare. Furthermore, we provide an
alternative auxiliary structure consisting of only a schema-dependent
set of counters for each list of elements in the database. This
auxiliary structure enables O(1) incremental validation in most cases,
using an algorithm that exploits a form of "locality".
About the speaker: Yannis Papakonstantinou serves on the Faculty of
Computer Science and Engineering at the University of California, San
Diego, since 1996. His research is in the intersection of database and
Internet technologies. Yannis has published over forty five research
articles in scientific conferences and journals, given tutorials at
major conferences, and served on journal editorial boards and program
committees for numerous international conferences and symposiums. He
was the co-Chair of WebDB 2002, is the General Chair of ACM SIGMOD
2003 and the Vice PC Chair for the "XML, Metadata and Semistructured
Data" track of IEEE ICDE 2004. In 1998, Yannis received the NSF CAREER
award for his work on integrating heterogeneous data. In 2000 Yannis
founded Enosys Software, which built the first generally available
distributed XQuery processor, along with software for XML-based
integration of distributed sources. Yannis holds a Diploma of
Electrical Engineering from the National Technical University of
Athens and MS and Ph.D. in Computer Science from Stanford University
(1997).
____________
BROAD AREA COLLOQUIUM FOR
AI-GEOMETRY-GRAPHICS-VISION-ROBOTICS
on Monday, 27 January 2003, 4:15pm
TCSeq 201
http://robotics.stanford.edu/ba-colloquium/
Shape Recipes
William T. Freeman
Artificial Intelligence Laboratory
Massachusetts Institute of Technology
The goal of low-level vision is to estimate an underlying scene, given
an observed image. Real-world scenes (e.g., shapes, or patterns of
reflectance) can be complex, conventionally requiring high dimensional
representations which can be hard to estimate and to store. But in
many situations, scene and image are closely related and it is
possible to learn a functional relationship between them. The scene
information can then be represented in reference to the image, where
the functional specifies how to translate the image into the
associated scene. We illustrate the use of this representation for
encoding shape information. We show that this representation has
appealing properties such as low dimensionality, locality, and slow
variation across space and scale. These properties allow us to improve
initial shape estimates from modalities such as stereo. I'll also show
two 5 minute "shorts": presentations showing results I'm excited about
on the topics of non-parametric belief propagation and separating
shading and reflectance.
About the speaker: William T. Freeman is an Associate Professor of
Electrical Engineering and Computer Science at the Artificial
Intelligence Laboratory at MIT, joining the faculty in September,
2001. From 1992 - 2001 he worked at Mitsubishi Electric Research Labs
(MERL), most recently as Sr. Research Scientist and Associate
Director. He studied computer vision for his PhD in 1992 from the
Massachusetts Institute of Technology, and received a BS in physics
and MS in electrical engineering from Stanford in 1979, and an MS in
applied physics from Cornell in 1981. His current research interests
include machine learning applied to computer vision and graphics, and
Bayesian models of visual perception. Previous research topics include
steerable filters and pyramids, the generic viewpoint assumption,
color constancy, computer vision for computer games, and separating
"style and content". Hobbies include kite aerial photography (see
http://www.ai.mit.edu/people/wtf/ ) and ping-pong.
References:
Shape recipes:
ftp://publications.ai.mit.edu/ai-publications/2002/AIM-2002-016.pdf
ftp://publications.ai.mit.edu/ai-publications/2002/AIM-2002-019.pdf
Non-parametric belief propagation:
ftp://publications.ai.mit.edu/ai-publications/2002/AIM-2002-020.pdf
Separating shading and reflectance
ftp://publications.ai.mit.edu/ai-publications/2002/AIM-2002-015.pdf
____________
SYNTAX WORKSHOP
on Tuesday, 28 January 2003, 5:30pm
Margaret Jacks Hall 460:126
http://www-linguistics.stanford.edu/sssg/
Subsumption and Equality: German Partial Fronting in LFG
Annie Zaenen and Ron Kaplan
Palo Alto Research Center
Equi-controllers or 'raised' elements are in general assumed to occur
in the matrix clause, and to control or bind in one or another way an
element in the embedded clause. Last year we presented an analysis of
some French raising and equi facts in which the equi controller or the
'raised' subject is actually in the embedded clause in the
c-structure. We argued that to model these facts we needed both
equality and subsumption relations. In this talk we will discuss the
interaction between raising and partial fronting in German and show
that the same types of relation are needed here. We will briefly
mention some other cases of 'backward' control or raising, discussed
in recent literature.
____________
REUTERS FOUNDATION DIGITAL VISION PROGRAM SEMINAR
on Wednesday, 29 January 2003, 3:00pm-4:30pm
email for location mailto:digitalvision@csli.stanford.edu
http://reuters.stanford.edu/seminar_speakers.html
Larry Lessig
Law Professor, Stanford University
Professor Lessig is one of the nation's top experts on constitutional
law and cyberspace. He received his B.A. in economics and B.S. in
management from the University of Pennsylvania, his M.A. in philosophy
from Trinity College, Cambridge, and his J.D. from Yale Law School.
After law school he clerked for Judge Richard Posner of the Seventh
Circuit Court of Appeals and Justice Antonin Scalia on the U.S.
Supreme Court.
Before joining the Stanford law faculty, he was a Professor of Law at
the University of Chicago from 1991 to 1997. From 1997 to 2000, he was
at the Harvard Law School, where he was the Jack N. and Lillian R.
Berkman Professor for Entrepreneurial Legal Studies. In 1999-2000,
Professor Lessig was a fellow at the Wissenschaftskolleg zu Berlin.
His "Law of Cyberspace" class, taught while he was a visiting
professor at Yale in 1995, was one of the first of its kind offered at
a law school. Professor Lessig teaches and writes in the areas of
constitutional law, contracts, comparative constitutional law, and
cyberspace, with a particular emphasis on such fundamentals as the
First Amendment and free speech, and copyright law. Professor Lessig
has consulted extensively with policy makers about the regulation of
cyberspace. He has also been active in a number of high-profile
Internet-related lawsuits, including Napster, the Microsoft antitrust
case, and the merger of AT&T and MediaOne. At the time he served as
Special Master to Judge Thomas Penfield Jackson in the Microsoft
antitrust case, Time magazine called Lawrence Lessig a "leading
thinker on how to adapt ancient principles to the new digital age."
His book, Code, and Other Laws of Cyberspace, is a thoughtful
exploration of intellectual property rights, free speech, and privacy
on the Web. Many of his published articles are in the field of
Internet regulation, exploring the nexus of regulation and cyberspace.
He has given dozens of lectures on the same topic. Formerly a regular
columnist for the Industry Standard, he has also contributed essays to
The Wall Street Journal, the Washington Post, the Los Angeles Times,
and The Boston Globe. In 2001, Random House published his latest book,
The Future of Ideas. Lawrence Lessig gave his oral argument before the
Supreme Court in the Eldred v. Ashcroft case in October of 2002.
____________
EE380: COMPUTER SYSTEMS LAB COLLOQUIUM
on Wednesday, 29 January 2003, 4:15pm to 5:30pm
NEC Auditorium (B03), Gates Computer Science Building
http://ee380.stanford.edu/contents.html
Chandler: An (Inter)personal Information Manager
Mitch Kapor
Chair, Open Source Applications Foundation
http://www.osafoundation.org/
Mitchell Kapor will speak about the Chandler project, an open source
Personal Information Manager now under development. In addition to
innovations in the handling of email, calendar, and contact
management, Chandler will emphasize sharing and collaboration and is
targeted to information-centric users.
About the speaker: Mitchell Kapor is the founder and Chair and of the
Open Source Applications Foundation. In 1982 he founded Lotus
Development Corporation and designed Lotus 1-2-3, the "killer app"
which made the PC ubiquitous in business. He co-founded of the
Electronic Frontier Foundation in 1990, an organization working in the
public interest to protect privacy and free expression on the
Internet.
____________
SCIL FUTURES OF LEARNING LECTURE SERIES
on Wednesday, 29 January 2003, 7:00pm
Wallenberg Hall (Bldg. 160)
http://scil.stanford.edu/
Beyond Accountability:
Assessment That Supports Adventuresome Teaching
Can assessment fortify innovative teaching and learning?
John D. Bransford
Vanderbilt University
Accountability is becoming increasingly important in education--it
forces educators to think carefully about goals and how to measure
them. But it can be detrimental if assessment inadvertently inhibits
educational experiences that enable students to lead successful
lives. Dr. Bransford discusses technology-enabled assessment
environments that have the potential to support development skills,
knowledge, and attitudes that people need for successful lifelong
learning.
About the speaker: John D. Bransford is Centennial Professor of
Psychology and Education and co-director of the Learning Technology
Center at Vanderbilt University. Author of seven books and hundreds of
articles and presentations, Dr. Bransford is an internationally
renowned scholar in cognition and technology.
____________
CSLI SEMINAR ON COMPUTATIONAL LEARNING AND ADAPTATION
on Thursday, 30 January 2003, 4:15pm to 5:30pm
Cordura 100
http://cll.stanford.edu/scla.html
A stability based method for discovering structure in clustered data
Asa Ben-Hur
Biochemistry, Stanford
http://tx.technion.ac.il/~asa/
Most clustering algorithms provide a clustering of a dataset
regardless of whether the data actually has cluster structure or not.
To address this issue, we present a method for assessing the presence
of structure in clustered data. The method is based on the idea that a
"good" clustering should be stable under perturbations of the data. We
characterize stability using the a similarity measure between a
reference clustering and clusterings obtained from sub-samples of the
data. High similarities indicate a stable clustering pattern. We argue
that stability is a desirable feature of a clustering solution that
implies the existence of cluster structure. The proposed method can be
used with any clustering algorithm; it provides a means of rationally
defining an optimum number of clusters, choosing various aspects of
the clustering algorithm, including feature selection. We show results
on several datasets using a hierarchical clustering algorithm, and
demonstrate with the method that using a few leading principal
components enhances cluster structure.
____________
SSP10: SYMBOLIC SYSTEMS FORUM
on Thursday, 30 January 2003, 4:15pm
Bldg. 380:380C (Math Corner)
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
Why don't we have a theory of higher level brain function and
what can we do about it
Jeff Hawkins
Redwood Neuroscience Institute
In this talk I will give an overview of the difficulties we face in
developing a theory of cognition and provide a perspective on why we
have made so little progress to date. I will discuss the problem from
both the biological and computational perspective suggesting specific
mistakes we have made, such as misunderstanding the role of feedback and
ignoring the role of time. There is reason to hope we can make major
progress in this field in the near term. I will lay out the research
agenda that can make this happen and what a theory of cognition will
look like.
____________
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
____________