[Date Prev][Date Next][Date Index]
CSLI Calendar, Wednesday, 28 April 2004, vol. 19:33
CSLI CALENDAR OF PUBLIC EVENTS
______________________________________________________________________
28 April 2004 Stanford Vol. 19, No. 33
______________________________________________________________________
A weekly publication of the
Center for the Study of Language and Information (CSLI)
Stanford University, Cordura Hall, Stanford, CA 94305-4115
http://www-csli.stanford.edu/
____________
ACTIVITIES FROM 28 April 2004 TO 7 MAY 2004
WEDNESDAY, 28 APRIL 2004
4:15pm CSLI Seminar on Computational Learning and Adaptation
Cordura Hall, room 100
"Constructing Features from High-Dimensional Data"
Geoffrey J. Gordon
Center for Automated Learning and Discovery, Carnegie Mellon
http://cll.stanford.edu/scla.html
Abstract below
4:15pm EE380: Computer Systems Lab Colloquium
Gates B03 (NEC Auditorium)
"Flexible Number Representations for Computing with FPGAs"
Oskar Mencer
Imperial College and Maxeler Technologies
http://ee380.stanford.edu/contents.html
Abstract below
4:30pm Psychology Talk
Jordan Hall 420:245
"The basal ganglia and processing beat-based rhythm in
musicians and non-musicians"
Jessica Grahn
MRC Cognition and Brain Sciences Unit, UK
Abstract below
THURSDAY, 29 APRIL 2004
12 noon Psychology Developmental Brownbags
Bldg. 420:050
"On-line processing of grammatically correct and anomalous
sentences in toddlers
Renate Zangl
Psychology, Stanford
http://www-psych.stanford.edu/news.html#dev_brownbag
12 noon RNI/Stanford Seminar Series on Theoretical Neuroscience
BioX/Clark Center, Room S360
"Inferential Processes and the Architecture of Visual Cortex"
Malcolm Young
University of Newcastle
http://www.ncl.ac.uk/biol/staff/profile/m.p.young
http://www.rni.org/seminar2.html
Abstract below
3:00pm Reuters Foundation Digital Vision Program Seminar
Reuters lounge, Cordura Hall
"Business Incubation - A Platform for Entrepreneur and ICT
Growth in Developing Countries"
Barbara L. Harley
http://cgi.stanford.edu/group/reuters/cgi-bin/calendar/index.cgi
4:00pm Personality Lab
Jordan Hall 420:419
"The body stripped down: An existential account of ambivalence
toward the physical body"
Jamie Goldenberg
UC Davis
http://www-psych.stanford.edu/news.html#person_lab
4:00pm SRI AI Seminar Series
EJ228, SRI International
"Towards an Interoperability Framework for Collaborative Tools"
Eugene Eric Kim
Blue Oxen Associates
http://www.ai.sri.com/seminars/
Abstract below
4:00pm UC Berkeley CIS Seminar
Soda Hall 310 (UC Berkeley)
"G Source Separation"
Hagai Attias
http://www.eecs.berkeley.edu/~ywteh/cis-seminar
Abstract below
4:15pm SSP10: Symbolic Systems Forum
Bldg. 380:380C (Math Corner)
"How Real Is the Future?"
John Perry
Philosophy, Stanford University
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
Abstract below
4:15pm Information Systems Seminar
Packard 101
"Dynamic Power Control in a Wireless Channel Subject to a
Quality of Service Constraint"
Baris Ata
Kellogg School of Management
http://isl.stanford.edu/groups/seminar/
Abstract below
5:15pm Archaeology Research Workshop
Bldg 60:61H
"Genes vs agents:
a discussion of the widening theoretical gap in archaeology"
Kristian Kristiansen
Gothenburg University and the Stanford Archaeology Center
http://archaeology.stanford.edu/workshop.html
FRIDAY, 30 APRIL 2004
11:00am CS545: Database Seminar
Gates 104
"Active XML: A Data-Centric Perspective on Web Services"
Omar Benjelloun
INRIA-Futurs
http://www-db.stanford.edu/dbseminar/
Abstract below
(special talk)
12 noon Ethics@Noon
Bldg. 100:101K
"Ethical Writing: Borrowing, Stealing, and Owning"
Hilton Obenzinger
Undergraduate Research Programs and Department of English
http://www.stanford.edu/dept/EIS/lectures_ethics.html
2:00pm NLP Reading Group
Ventura 17
"Speed and Accuracy in Shallow and Deep Stochastic Parsing"
Ron Kaplan, Stefan Riezler, Tracy Holloway King, John T. Maxwell III,
Alexander Vassermann, and Richard Crouch (PARC)
http://www.parc.com/istl/members/riezler/PAPERS/NAACL04.pdf
http://www-nlp.stanford.edu/nlp/nlpgroup.html
Abstract below
3:15pm The Mathematical Representation of Nature
Location to be announced
"Beauty doth of itself persuade:
Quantization, mathematical beauty, and scientific understanding"
Michael Dickson
Indiana University
http://www.stanford.edu/dept/HPST/philosophyworkshop.html
3:15pm Friday Cognitive Seminar
Jordan Hall 420:050
Title to be announced
Ching Kao
http://www-psych.stanford.edu/news.html#frisem
3:30pm Linguistics Department Colloquium
Margaret Jacks Hall 460:126
"Narrativity and Differentiation:
Reading "between the lines" in the Cold War"
Susan Gal
University of Chicago
http://www-linguistics.stanford.edu/colloq/
Abstract below
SATURDAY, 1 MAY 2004
10:00am The Mathematical Representation of Nature
Terrace Room, 4th floor, Margaret Jacks Hall
"A case of mathematical uncooperativeness:
Classical field theories and the classical ideal of theories"
Mathias Frisch
University of Maryland
http://www.stanford.edu/dept/HPST/philosophyworkshop.html
2:00pm The Mathematical Representation of Nature
Terrace Room, 4th floor, Margaret Jacks Hall
"Hume and mechanics' complexities"
Mark Wilson
University of Pittsburgh
http://www.stanford.edu/dept/HPST/philosophyworkshop.html
MONDAY, 3 MAY 2004
12 noon UC Berkeley Developmental Psychology Colloquium
3105 Tolman (Berkeley)
"The Child Learns to Think for Speaking:
Puzzles of Crosslinguistic Diversity in Form-Meaning Mappings"
Dan Slobin
Psychology, UC Berkeley
http://psychology.berkeley.edu/admin/colloquia.html
4:15pm CS528: Broad Area Colloquium in AI,
Geometry, Graphics, Robotics, and Vision
TCSeq 200
"Differential topology and combinatorial algorithms, and where
this unlikely marriage works"
Herbert Edelsbrunner
Duke
http://graphics.stanford.edu/courses/cs528/
Abstract below
TUESDAY, 4 MAY 2004
2:45pm CS548: Internet and Distributed Systems Seminar
Skilling Auditorium
"High Availability Techniques and Case Studies from USBank"
Greg Messer
Manager, Enterprise Performance Engineering,
http://cs548.stanford.edu/schedule.shtml
Abstract below
3:00pm Reuters Foundation Digital Vision Program Seminar
Reuters lounge, Cordura Hall
"Democracy: Coming Soon to a Computer Near You"
Todd Davies
http://cgi.stanford.edu/group/reuters/cgi-bin/calendar/index.cgi
4:15pm SNRC Industry Seminar
Gates B03
"Mars Rover Information Portal"
Ronald Mak
NASA Ames Research Center
http://snrc.stanford.edu/events/industry-seminar/
Abstract below
WEDNESDAY, 5 MAY 2004
12 noon UC Berkeley Anthropology Talks
Seminar Room, 2251 College, ARF (Berkeley)
"Learning and stone tool production in Paleolithic France"
Kathy Sterling
http://anthropology.berkeley.edu/news.html
3:45pm Psychology Department Colloquium
Jordan Hall 420:041
Title to be announced
Ken Kendler
Virginia Commonwealth University
http://www-psych.stanford.edu/news.html
4:15pm EE380: Computer Systems Lab Colloquium
Gates B03 (NEC Auditorium)
"Scalable Visual Comparison of Biological Trees and Sequences"
Tamara Munzner
University of British Columbia
http://ee380.stanford.edu/contents.html
Abstract below
THURSDAY, 6 MAY 2004
12:15pm CSLI CogLunch
Cordura Hall, Room 100
Title to be announced
Paul Kay
http://www-csli.stanford.edu/events/Coglunch/
3:00pm Reuters Foundation Digital Vision Program Seminar
Reuters lounge, Cordura Hall
"Designing a Sea Change"
Lauralee Alben
http://cgi.stanford.edu/group/reuters/cgi-bin/calendar/index.cgi
4:00pm Personality Lab
Jordan Hall 420:419
"'I just can't stop thinking about it...': Information-processing
biases, cognitive inhibition, and rumination in depression"
Jutta Joormann,
Psychology, Stanford
http://www-psych.stanford.edu/news.html#person_lab
4:00pm SRI AI Seminar Series
EJ228, SRI International
Title to be announced
Elin Ronby Pedersen
http://www.ai.sri.com/seminars/
4:15pm SSP10: Symbolic Systems Forum
Bldg. 380:380C (Math Corner)
"Engineering Deep Grammars"
Tracy King
Palo Alto Research Center
Consulting Associate Professor, Symbolic Systems Program
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
Abstract below
4:15pm Information Systems Seminar
Packard 101
"Time Synchronization and Distributed Modulation in Large
Sensor Networks (or 'How to Make a Large Random Field by
Stitching Together Very Many Small Ones')"
Sergio D. Servetto
Cornell University
http://isl.stanford.edu/groups/seminar/
Abstract below
5:30pm Stanford Phonology Workshop
Margaret Jacks Hall 460:126
"A French Dialect Continuum Across North America"
Luc Baronian
Stanford University
http://www-linguistics.stanford.edu/pinterest/
Abstract below
FRIDAY, 7 MAY 2004
12:30pm CS547: Human-Computer Interaction Seminar
Gates B01
"Understanding the Requirements for Developing and Designing
Free/Open Source Software"
Walt Scacchi
UC Irvine
http://www-pcd.stanford.edu/cs547/
Abstract below
2:00pm NLP Reading Group
Ventura 17
"Unsupervised multilingual sentence boundary disambiguation"
Jan Strunk
Linguistics, Stanford
http://www-nlp.stanford.edu/nlp/nlpgroup.html
Abstract below
3:15pm Philosophy Department Colloquium
Bldg. 90:92Q
Title to be announced
Dorothea Frede
University of Hamburg/Stanford CASBS
http://www-philosophy.stanford.edu/ce.html
3:15pm Friday Cognitive Seminar
Jordan Hall 420:050
Title to be announced
Michelle Gumbrecht
http://www-psych.stanford.edu/news.html#frisem
3:30pm Semantics and Pragmatics Workshop
Margaret Jacks Hall 460:126
Title to be announced
Mark Baker
Rutgers
http://www-linguistics.stanford.edu/semgroup/
____________
Stanford Blood Center status: Shortage of A+, O-, and O+. For an
appointment: http://bloodcenter.stanford.edu/ or call 650-723-7831.
It only takes an hour of your time.
____________
CSLI SEMINAR ON COMPUTATIONAL LEARNING AND ADAPTATION
on Wednesday, 28 April 2004, 4:15pm to 5:30pm
Cordura 100
http://cll.stanford.edu/scla.html
"Constructing Features from High-Dimensional Data"
Geoffrey J. Gordon
Center for Automated Learning and Discovery, Carnegie Mellon
An important machine learning problem is feature construction:
learning functions which reduce raw, high-dimensional data to a
lower-dimensional set of features, while still retaining as much
information from the raw data as possible. This sort of dimensionality
reduction can help filter out noise, and it can be an important
pre-processing step before running learning algorithms whose
computational complexity or sample complexity grow quickly with input
dimensionality. In this talk, I will describe our experiments with
feature construction from data recorded using our team of mobile
robots. These algorithms include nonlinear principal components
analysis, which we have used to compress belief states in planning
problems, and predictive state representations, which we have used to
compress time series of laser rangefinder readings. While our
experiments are based on robot sensor data, the algorithms and results
should be relevant to other applications of dimensionality reduction
such as text, link structure, or images.
____________
EE380: COMPUTER SYSTEMS LAB COLLOQUIUM
on Wednesday, 28 April 2004, 4:15pm to 5:30pm
NEC Auditorium (B03), Gates Computer Science Building
http://ee380.stanford.edu/contents.html
"Flexible Number Representations for Computing with FPGAs"
Oskar Mencer
Imperial College and Maxeler Technologies
As programmable hardware matures into a competitive platform for
computation, we are faced with the opportunity to optimize the number
representations possibly for each variable of our programs.
Dynamically, we can even adapt the number representation to input data
and phases of the program.
Adapting the number representation consists of two parts: (1) the
encoding, and (2) the number of bits used to represent the values. We
are here not limited to standard encodings such as integer or IEEE
floating point, but can adapt the encoding to the function that is
being computed. The second step is then to find the precision
requirement which we can accomplish, for example, via automatic
differentiation of the program.
This research is enabled by ASC, A Stream Compiler, which provides a
hardware programming environment for FPGAs. The programmer gets access
to the programmable hardware on the algorithm level, but also on the
architecture, arithmetic and gate levels, all within a single C++
program. As a consequence, ASC combines the performance of hand
designed circuits with the productivity of the software development
process.
About the speaker: Oskar Mencer teaches at Imperial College in London
and, on the side, runs Maxeler Technologies, a startup commercializing
ASC technology. (See http://www.maxeler.com for information about
Maxeler Technologies.)
He received his Ph.D. and M.S. in Electrical Engineering from Stanford
University in 2000 and 1997 respectively, and a B.S. degree in
Computer Engineering from the Technion/Israel in 1994. His research
interests span computer architecture, computer arithmetic, VLSI
microarchitecture, CAD and reconfigurable (custom) computing. More
specifically, he is interested in improving the programmability of
FPGAs and making the compute power of reconfigurable computers
accessible to the computer science community.
____________
PSYCHOLOGY TALK
on Wednesday, 28 April 2004, 4:30pm
Jordan Hall 420:245
"The basal ganglia and processing beat-based rhythm in
musicians and non-musicians"
Jessica Grahn
MRC Cognition and Brain Sciences Unit, UK
Regular rhythmic patterns engage us. We tap our toes and bob our heads
to the beat, clap our hands or drum on the steering wheel. This
behaviour has led to the suggestion that we may use a beat-based
timing mechanism for these types of regular temporal patterns, and an
interval-based timer when temporal patterns have no regular beat. In
beat-based timing, neural machinery synchronizes to a perceived beat,
and durations to be timed are encoded and reproduced in reference to
this beat. The interval timer encodes individual durations, recalling
them at any time. Beat-based timing may offer an accuracy benefit or
working memory load reduction, though behavioural evidence for this
has been conflicting. This talk presents both behavioural and neural
evidence for the existence of beat-based timing. The results indicate
that beat-based timing does exist, and is mediated by the basal
ganglia in both musicians and non-musicians. In addition, musicians
show higher levels of activation in brain areas not specifically
involved in beat-based timing.
About the Speaker: Jessica Grahn is a PhD student working with Matthew
Brett and Robert Carlyon at the MRC Cognitive and Brain Sciences Unit.
____________
RNI/STANFORD SEMINAR SERIES IN THEORETICAL NEUROSCIENCE
on Thursday, 29 April 2004, 12 noon - 1pm
BioX/Clark Center, Room S360
http://www.rni.org/seminar2.html
"Inferential Processes and the Architecture of Visual Cortex"
Malcolm Young
University of Newcastle
http://www.ncl.ac.uk/biol/staff/profile/m.p.young
A striking aspect of vision is that observers do not see what their
eyes tell their brains (e.g. Young 1994). This mismatch is clearly
evident in the fact that apparently detailed coloured texture is
seen in extra-foveal vision, where such information is very modest
or absent in the input from the retina, and in other cases, where
information at one location and time influences the likelihood that
information will be present at a different location, and at a later
time. These considerations, and detailed observations of what
neuroanatomy can tell us of the processing architecture in visual
cortex, suggest that the visual system may undertake an inferential
style of computation, rather than analysing bottom-up information
from the eyes, which latter model is in the ascendant in visual
neurophysiology.
We have been exploring the different predictions that the inference
and traditional models make, when the prior probabilities of stimuli
are manipulated, as they are during normal vision. In one paradigm,
we use spatiotemporal sequences of oriented bars to provide the visual
system with "reason to believe" that a bar of particular orientation
will be presented at a particular time. We then vary the congruence,
incongruence and presence of the bar in the RF in relation to these
prior expectations. The results so far are striking: V1 neurons are
often more modulated by these prior and distant events - in many cases
more than 400ms, and 6-10 degrees away from their classical RF - than
they are by events within their RF. Similar results relate to an
experimental paradigm that tests whether V1 neurons really suffer from
the "aperture problem." The interaction between priors and likelihood
functions is well fitted by a Bayesian model, and only poorly fitted
by the traditional view that V1 cells are local filters of one kind or
another. Similarly, Bayesian models fit better than mismatch models,
such as that of Rao and Ballard. These results suggest that V1
neurons are signaling the posterior probability of visual stimuli,
on the basis both of prior knowledge about the world and information
from the eye, effectively imputing feature constellations to the
visual world, rather than simply analysing local contours.
About the Speaker: Professor Malcolm Young is a Provost of Science,
Agriculture and Engineering at the University of Newcastle in England.
His research addresses the question of how the brain works by
investigating how the brain is organized anatomically; what processes
go on inside it; how these processes interact to cause behaviour,
particularly visual behaviour; and what biological constraints must
be met by the brain. His laboratory was instrumental in developing
the neuroinformatics approach - computer-based collation, management,
and analysis of neuroscience data.
____________
SRI AI SEMINAR SERIES
on Thursday, 29 April 2004, 4:00pm - 5:30pm
EJ228, SRI International
http://www.ai.sri.com/seminars/
"Towards an Interoperability Framework for Collaborative Tools"
Eugene Eric Kim
Blue Oxen Associates
Almost 40 years ago, Doug Engelbart demonstrated the first
collaborative software tools at a conference in San Francisco. The
demonstration stunned the audience, who gave Engelbart a standing
ovation. In many ways, today's collaborative tools aren't
significantly better than the software demonstrated that day. One of
the biggest problems is that these tools do not interoperate well with
each other. Kim will discuss the need for a shared conceptual
framework for collaborative tools, which will help reveal new
possibilities for interoperability and improvement.
About the Speaker: Eugene is Executive Director and cofounder of Blue
Oxen Associates, a socially-conscious think tank devoted to improving
collaboration and knowledge management. Previously, he worked as an
independent consultant, specializing in software development, project
management, and strategic planning, and before that, he was Senior
Technical Editor at Dr. Dobb's Journal. He worked closely with Doug
Engelbart on the Open Hyperdocument System, and served on the Core
Planning Committee of Engelbart's Bootstrap Alliance. Eugene has
written for a variety of publications, and is currently writing a book
on the history of free software. Eugene received his A.B. in History
and Science from Harvard University.
____________
UC BERKELEY CIS SEMINAR
on Thursday, 29 April 2004, 4:00pm-5:00pm
Soda Hall 310 (UC Berkeley)
http://www.eecs.berkeley.edu/~ywteh/cis-seminar
"G Source Separation"
Hagai Attias
Some graphical model based algorithms for source separation have two
separate versions, one for clean data (zero noise mixing) and the
other for noisy data (mixing with additive noise). One such algorithm
is independent factor analysis (IFA). One might expect noisy IFA to
reduce to clean IFA as the noise level approaches zero. However, noisy
IFA reduces to PCA in that limit, and performs poorly on low noise
data where clean IFA performs well. I will discuss a simple method,
called the G-trick, for deriving a new noisy IFA algorithm that
performs optimally for any noise level, and reduces to clean IFA in
the zero noise limit. I will also discuss extensions to reverberant
data (convolutive mixing) and to incorporating rich source models.
____________
SSP10: SYMBOLIC SYSTEMS FORUM
on Thursday, 29 April 2004, 4:15pm
Bldg. 380:380C (Math Corner)
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
"How Real Is the Future?"
John Perry
Philosophy, Stanford University
John McTaggart Ellis McTaggart argued that time is not real. He thought
that the concept of time requires that events have, successively, the
properties of being future, being present, and being past; this is the
source of temporal change, the essence of time. These are A-properties.
The 2000 election was once in the future, then it was present, and
now it is in the past. But our concept of time also requires that all
temporal properties derive from the order of events in time; the 2000
election was after the 1996 election and before the 2004 election.
But these order properties, the B-properties, never change, and so
cannot be the source of the changing properties A-properties. So the
whole thing is a unreconcilable mess, and the concept of time makes no
sense, so there is no time.
Contemporary philosophers of time tend to agree that time is real, but
don't agree on what to do about the properties of being future,
present, and past. Some argue that these are indexical or
token-reflexive properties. When I say now "The 2004 election is in
the future," the condition of truth on my remark is that the 2004
election is later than my remark. The relation between the 2004 and
my remark never change. So there is no problem finding the fact that
makes my statement true in the B-series. Others argue that this
analysis leaves the nature of time utterly mysterious, since temporal
change, the essence of time, is left out. We find the A-series within
the B-series at the cost of losing the essence of time. Moreover, it
seems that on this analysis there would be no past, present, or future
without utterances, or at least thoughts, for events to be before,
simultaneous with, or later than.
I'll argue that the indexical analysis of the properties of being
present, past, and future is not right. There are no future events,
for events are not real, do not exist, until they happen. If there is
no event later than an event E, E is present. If there are events
later than it, E is past. Events go from being present to being past
in virtue of new events coming into existence. The B-series facts
change just as the A-series facts do, and are the basis for them.
Time is real, temporal change is unique, and it does not require
utterances or thoughts. Confusion on these matters is due, among
other things, to not distinguishing representations, models, and
reality, and to the overly facile way we talk about propositions being
true at times.
This talk covers work I am doing with Thomas Hofweber.
About the Speaker: John Perry is Henry Waldgrave Stuart Professor of
Philosophy at Stanford University.
____________
INFORMATION SYSTEMS SEMINAR
on Thursday, 29 April 2004, 4:15pm-5:15pm
Packard 101
http://isl.stanford.edu/groups/seminar/
"Dynamic Power Control in a Wireless Channel
Subject to a Quality of Service Constraint"
Baris Ata
Kellogg School of Management
A controller dynamically chooses a state-dependent transmission rate
on a static, point-to-point wireless link by varying transmission
power over time. The transmitter is modeled as a finite-buffer
Markovian queue with adjustable service rates. That is, data packets
arrive to the system according to a Poisson process and packet size is
exponentially distributed. The controller chooses a transmission rate
from a fixed set $A$ of available values, depending on the backlog in
the system. The objective is to minimize long-run average energy
consumption subject to a quality of service constraint, which is
expressed as an upper bound on the packet drop rate. An explicit
formula is developed for the optimal transmission rate as a function
of the packet queue length. In the second half of the talk I will
extend the model and the results to a fading channel.
____________
CS545: DATABASE SEMINAR
on Friday, 30 April 2004, 11:00am - 12 noon
Gates 104
http://www-db.stanford.edu/dbseminar/
(special seminar, note unusual time and place)
"Active XML: A Data-Centric Perspective on Web Services"
Omar Benjelloun
INRIA-Futurs
This talk introduces Active XML (AXML, for short), a declarative
framework that harnesses Web services for data integration, and is put
to work in a peer-to-peer architecture.
An AXML document is an XML document that may contain embedded calls to
Web services. An AXML "peer" is a repository of AXML documents. On
the one hand, it acts as a client, by invoking the calls to Web
services embedded in its documents. On the other hand, a peer acts as
a server, by providing Web services that can be declaratively
specified as queries over the AXML documents it contains. The AXML
approach allows to gracefully combine stored information with data
defined in an intensional manner (as service calls). The fact that
peers can exchange a mixture of materialized and intensional data (via
AXML documents) leads to a very powerful distributed data management
paradigm.
The AXML approach induces a number of technically challenging
problems, both theoretical and practical, which I will describe in the
talk. I will also present a prototype that we developed, and mention
some interesting applications.
____________
NLP READING GROUP
on Friday, 30 April 2004, 2:00pm - 3:30pm
Ventura 17
http://www-nlp.stanford.edu/nlp/nlpgroup.html
"Speed and Accuracy in Shallow and Deep Stochastic Parsing"
Ron Kaplan, Stefan Riezler, Tracy Holloway King, John T. Maxwell III,
Alexander Vassermann, and Richard Crouch (PARC)
http://www.parc.com/istl/members/riezler/PAPERS/NAACL04.pdf
This paper reports some experiments that compare the accuracy and
performance of two stochastic parsing systems. The currently popular
Collin s parser is a shallow parser whose output contains more
detailed semantically-relevant information than other such parsers.
The XLE parser is a deep-parsing system that couples a Lexical
Functional Grammar to a log-linear disambiguation component and
provides much richer representations. We measured the accuracy of both
systems against a gold standard of the PARC 700 dependency bank, and
also measured their processing times. We found the deep-parsing system
to be more accurate than the Collins parser with only a slight
reduction in parsing speed.
____________
LINGUISTICS DEPARTMENT COLLOQUIUM
on Friday, 30 April 2004, 3:30pm
Margaret Jacks Hall, Room 460:126
http://www-linguistics.stanford.edu/colloq/
"Narrativity and Differentiation:
Reading "between the lines" in the Cold War"
Susan Gal
University of Chicago
We know that linguistic differentiation (style) signals the
differentiation of social identities. This insight has been developed
very effectively with reference to non-referential indexes (i.e.
phonological and sometimes syntactic variants). Personal narratives,
by contrast, have usually been analyzed as a means for creating
identity. In the study of narrative strategies, there has been little
attention to processes of social differentiation. My aim is to
examine a Cold War text -- the novelistic representation of a
bureaucratic encounter in Budapest -- in order to show how we can
watch social differentiation emerge through the poetic organization of
conversational interaction in narrative. The study of this text
proves ethnographically revealing too, showing the forms of coercion,
collusion and collaboration that were the hallmarks of state-citizen
relations in the communist period in eastern Europe.
____________
CS528: BROAD AREA COLLOQUIUM FOR
AI-GEOMETRY-GRAPHICS-VISION-ROBOTICS
on Monday, 3 May 2004, 4:15pm
TCSeq 200
http://graphics.stanford.edu/courses/cs528/
"Differential topology and combinatorial algorithms,
and where this unlikely marriage works"
Herbert Edelsbrunner
Arts and Sciences Professor of Computer Science and Mathematics, Duke
In this talk, I will present some of our recent work in computational
topology. I will introduce concepts from differential topology (Morse
functions, Reeb graphs, Jacobi sets, ...) and algebraic topology
(Betti numbers, persistence, ...) and discuss how they can be computed
for piecewise linear data. There are numerous applications of these
ideas and I will focus primarily on problems in structural biology.
About the Speaker: Herbert Edelsbrunner received his Ph.D. in
Technical Mathematics in 1982 from the University of Technology in
Graz, Austria. He was faculty at the University of Illinois at
Urbana-Champaign from 1985 until he joined Duke University in
1999. His primary research interests are in algorithms, geometry and
topology, and structural biology. He published two textbooks in
computational geometry. In 1996, he co-founded Raindrop Geomagic, a
software company that specializes in geometric modeling and shape
reconstruction from scan data.
____________
CS548: INTERNET AND DISTRIBUTED SYSTEMS SEMINAR
on Tuesday, 4 May 2004, 2:45pm - 4:00pm
Skilling Auditorium
http://cs548.stanford.edu/schedule.shtml
"High Availability Techniques and Case Studies from USBank"
Greg Messer
Manager, Enterprise Performance Engineering, US Bancorp
I will describe the practice of "Systems Performance Engineering and
Management" in business and real-time settings. The reason for having
a "Performance Engineering and Management" discipline in industry is
to ensure that an application's performance requirements are met prior
to being released to production, and continue to do so while in
production. I will discuss the science behind the practice, how it is
implemented within its methods, software and roles, and the skill set
necessary to do it. In addition, two case studies will be presented:
the first from a business application point of view, the second from a
real-time service point of view.
____________
SNRC INDUSTRY SEMINAR
on Tuesday, 4 May 2004, 4:15pm
Gates B03
http://snrc.stanford.edu/events/industry-seminar/
"Mars Rover Information Portal"
Ronald Mak
Computer Scientist - NASA Ames Research Center
http://www.arc.nasa.gov/
The Collaborative Information Portal is mission-critical software
developed at NASA Ames for the current Mars Exploration Rover
mission. CIP is used inside mission control at the Jet Propulsion
Laboratory and by NASA scientists and researchers throughout the
world. It displays current staff and event schedules and clocks that
show times in various Mars and Earth time zones. CIP users can
securely download and view data and image files from the mission file
servers, and they can send and receive broadcast messages. CIP plays
a major role in ensuring that mission personnel work together
effectively and efficiently, especially since the mission runs on Mars
time.
CIP is a three-tier enterprise application based on industry-standard
Java technologies (Swing, J2EE, etc.) and web services. This talk
will present an overview of the Mars mission and discuss CIP's
architecture and key components. We'll take a closer look at the CIP
middleware, which handles the remote service calls from hundreds of
client applications.
About the Speaker: Ronald Mak is a computer scientist employed by the
University of California at Santa Cruz. He works in the University
Affiliated Research Center (UARC) at the NASA Ames Research Center.
Ron was the architect and lead for the CIP middleware, and currently,
he does research and development at Ames and mission support at
JPL. Prior to NASA, he spent over 15 years developing enterprise
software in industry. Ron has written books on numerical computing and
on compiler writing, and he has a B.S. in the Mathematical Sciences
and a M.S. in Computer Science from Stanford University.
____________
EE380: COMPUTER SYSTEMS LAB COLLOQUIUM
on Wednesday, 5 May 2004, 4:15pm to 5:30pm
NEC Auditorium (B03), Gates Computer Science Building
http://ee380.stanford.edu/contents.html
"Scalable Visual Comparison of Biological Trees and Sequences"
Tamara Munzner
University of British Columbia
We present several visualization systems for comparing and browsing
evolutionary trees and genomic sequences. These systems use the
Focus+Context navigational metaphor of allowing users to fluidly
stretch and shrink parts of the view, as if manipulating a rubber
sheet with the borders tacked down. We introduce cognitive scalability
to this approach by guaranteeing the visibility of landmarks at all
times, so that users can stay oriented as they explore complex
datasets. In our systems, landmarks can be regions of difference
between datasets, or the results of a search, or user-chosen regions.
This technique, which we call "accordion drawing", supports smooth
realtime transitions between a big-picture overview and a drilled-down
views that show details in context.
TreeJuxtaposer is a system that allows users to compare large trees of
several hundred thousand nodes. We propose a new methodology for
detailed structural comparison between two trees and provide a new
nearly-linear algorithm for computing the best corresponding node from
one tree to another. SequenceJuxtaposer is a sequence visualization
tool for the exploration and comparison of biomolecular sequences.
SequenceJuxtaposer supports interaction at 20 frames per second when
browsing a single sequence of over 1.2 million base pairs, or large
collections of sequences up to 2 million total base pairs. In both
systems, all runtime rendering algorithms are sublinear in the total
number of tree nodes or base pairs in the dataset, and all
preprocessing is subquadratic in those variables. Our rendering
algorithms accomodate high-resolution displays as well as standard
screen sizes. With these systems, we are able to quickly observe many
features that had previously required significant analysis to
discover.
We also present the TJC and TJC-Q systems, which allow users to browse
extremely large trees using the accordion drawing technique. TJC is a
system that supports browsing trees up to 15 million nodes by
exploiting leading-edge graphics hardware while TJC-Q allows browsing
trees up to 5 million node on commodity platforms. Both of these
systems use a fast new algorithm for drawing and culling and benefit
from a complete redesign of all data structures for more efficient
memory usage and reduced preprocessing time. In addition to browsing
evolutionary trees, these tools are also useful in many other
application domains where there is a demand for large-scale tree
browsing, including network management, software engineering, database
integration, and genealogy.
About the speaker: Tamara Munzner has been an assistant professor in
the University of British Columbia Department of Computer Science
since 2002. Her current research interests are information
visualization, graph drawing, dimensionality reduction, and
interactive computer graphics. She was a research scientist from 2000
to 2002 at the Compaq Systems Research Center in California, and
earned her PhD from Stanford between 1995 and 2000. She was on the
technical staff of The Geometry Center, a mathematical visualization
research group at the University of Minnesota, from 1991 to 1995.
____________
SSP10: SYMBOLIC SYSTEMS FORUM
on Thursday, 6 May 2004, 4:15pm
Bldg. 380:380C (Math Corner)
http://symsys.stanford.edu:8081/ssp-dynamic/servlet/ssp_events
"Engineering Deep Grammars"
Tracy King
Researcher, Palo Alto Research Center
Consulting Associate Professor, Symbolic Systems Program
Parsing systems based on deep grammars mark explicitly a variety of
syntactic and semantic dependencies and should therefore provide
crucial support for meaning-sensitive NLP applications. However,
common wisdom has it that parsing systems based on deep linguistic
grammars are too difficult to produce, lack coverage and robustness,
are massively ambiguous, and have poor run-time performance. In this
talk, I will discuss ways in which deep grammars can be engineered to
overcome these problems, focusing on techniques used in the
implementation of the broad-coverage LFG-based grammars produced by
the ParGram project.
About the Speaker: Tracy Holloway King is a member of the research
staff at the Palo Alto Research Center and a consulting associate
professor with the Symbolic Systems Program.
____________
INFORMATION SYSTEMS SEMINAR
on Thursday, 6 May 2004, 4:15pm-5:15pm
Packard 101
http://isl.stanford.edu/groups/seminar/
"Time Synchronization and Distributed Modulation
in Large Sensor Networks
(or 'How to Make a Large Random Field by
Stitching Together Very Many Small Ones')"
Sergio D. Servetto
Cornell University
Joint work with An-swol Hu
We study limit properties of random processes obtained as a properly
scaled sum of a large number of randomly shifted pulses. Such
processes arise naturally in a sensor networking application, where a
large number of power-constrained radio transmitters coordinate their
access to a Gaussian multiple access channel, to cooperate in
generating a strong information-bearing waveform. In this talk we will
see how, under very mild technical assumptions easily met in practice,
the resulting limit waveform has some useful deterministic properties:
it is continuous, with known roots and polarity. We will also see how
these properties lead to efficient and practical implementations of
two essential primitive sensor networking functions: global time
synchronization, and reachback communication. Time permitting, we will
discuss connections with some old problems in network information
theory (and some new ones too), and we will overview experimental work
on these topics currently under way.
____________
STANFORD PHONOLOGY WORKSHOP
on Thursday, 6 May 2004, 5:30pm
Margaret Jacks Hall 460:126
http://www-linguistics.stanford.edu/pinterest/
"A French Dialect Continuum Across North America"
Luc Baronian
Stanford University
Besides presenting interesting grammatical phenomena, the study of the
French dialects and creoles of the Americas involves complex contact
situations between Africans, (Native) Americans and Europeans, and poses a
significant challenge for language change and classification. Therefore
sociolinguists, historical linguists and even theoretical linguists have
much to learn from their study.
Based on striking resemblances, Hull (1968) first proposed that the French
dialects and creoles of the colonies were all descendent from a maritime
jargon used in the ports of Western France. While linguists who have
concentrated on the genesis of Quebec French (Barbaud, Asselin &
McLaughlin, Morin, Poirier) have all rejected this strong hypothesis at
least for Quebec, a satisfactory global account of the genesis of these
languages and their relation to one another as well as to 17th century
dialects of French is still lacking.
I propose a traditional answer to part of this larger question, based
on recent fieldwork in Louisiana and research at the Library of
Congress on old recordings of former Missouri and Indiana varieties of
French. I show that a former dialect continuum emerges from Canada to
Louisiana, along the St Lawrence and Mississippi rivers.
The results are similar to Andrew Garrett's account of the origin of
Greek, where he proposes that closely related Indo-European dialects
gradually moved closer together by undergoing the same phonological
changes. In the case of the North American French dialects, the
evidence suggests that distinct dialects underwent regional changes
that gradually brought them closer to neighboring dialects, while
distancing them from dialects from which they were originally closer.
____________
CS547: HUMAN-COMPUTER INTERACTION SEMINAR
on Friday, 7 May 2004, 12:30-2:00pm
Gates B01
http://www-pcd.stanford.edu/cs547/
"Understanding the Requirements for
Developing and Designing Free/Open Source Software"
Walt Scacchi
Institute for Software Research, UC Irvine
In this talk, I present findings from an ongoing empirical study of
social processes, technical system configurations, organizational
contexts, and interrelationships that give rise to free/open source
software. The focus is directed at understanding what the requirements
for designing F/OSS may be. Multiple kinds of software development
informalisms are found to play a critical role in the elicitation,
analysis, specification, validation, and management of the
requirements and design of F/OSS systems, artifacts, processes,
project portals, and communities. Subsequently, understanding the
roles these informalisms take in the development process for F/OSS is
the focus of this presentation.
Research publications and related information at:
http://www.isr.uci.edu/research-open-source.html
About the Speaker: Walt Scacchi is senior research computer scientist
and research faculty member at the Institute for Software Research,
and director of research for the Laboratory for Game Culture and
Technology, at the University of California Irvine. He received his
Ph.D. in Information and Computer Science at UC Irvine in 1981. He
joined ISR in 1999 after serving on the faculty at the University of
Southern California for 18 years. From 1981 to 1991 he founded and
directed the USC System Factory, and from 1993 to 1998 he directed the
USC ATRIUM Laboratory. His interests include open source software
development, software process engineering, software acquisition and
electronic commerce, and organizational studies of system
development. He is an active researcher with more than 100 research
papers, and consults widely to clients in industry and government
agencies.
____________
NLP READING GROUP
on Friday, 7 May 2004, 2:00pm - 3:30pm
Ventura 17
http://www-nlp.stanford.edu/nlp/nlpgroup.html
"Unsupervised multilingual sentence boundary disambiguation"
Jan Strunk
Linguistics, Stanford
The detection of sentence boundaries constitutes a preliminary, yet
important step in electronic text processing. A majority of linguistic
analyses and methods in computational linguistics posit the sentence
as a crucial unit. What is more, sentence boundary detection itself
can be seen as a typical instance of one of the major problems in
computational linguistics: ambiguity resolution. In many languages of
the world, the marker for the sentence boundary, the period is not
used unequivocally. In particular, it is also used to mark
abbreviations. To make this ambiguity even more problematic the period
can as well be used to mark an abbreviation at the end of a sentence.
Although being a comparatively minor task in language processing,
sentence boundary detection has received some attention in the past
decade, as can be witnessed by approaches such as Palmer/Hearst
(1997), Grefenstette (1999), Schmid (2000) and Mikheev (2002). It is a
common property of these approaches that the developed methods have
been applied to a small range of languages only, usually covering one
to three languages. In such a setting, it remains unclear how well the
suggested methods operate if more languages or different types of
corpora are considered.
I will present an approach to sentence boundary detection that I
developed together with Tibor Kiss at the University of Bochum which
is supposed to be language and corpus independent and determines
sentence boundaries with high accuracy. We assume that the
determination of abbreviations is an easier task than the detection of
sentence boundaries, and base our approach on the detection of
abbreviations as a first step. This approach does not make use of
additional annotations, POS tagging, or precompiled lists to support
sentence boundary detection. Also, it does not use orthographical
information as primary evidence and is thus also suited to process
single-case text. It focuses on robustness and flexibility in that it
can be applied to a variety of languages without any adjustment. The
basic algorithm has been determined experimentally on the basis of
large unannotated corpora from German and English. We have applied the
results to further corpora from German and English as well as to
corpora from nine other languages (Brazilian Portuguese, Dutch,
Estonian, French, Italian, Norwegian, Spanish, Swedish, and Turkish).
In the evaluations reported, the mean accuracy of sentence boundary
detection is 98.47 %, the mean accuracy for abbreviation detection is
99.05%.
I will give an introduction into our general approach to abbreviation
detection and the additional heuristics we use to detect abbreviations
at the end of sentences and some more special cases. I will also
report on the results we have obtained when we tested our algorithm on
eleven different languages.
Grefenstette, Gregory. (1999). Tokenization.. In van Halteren, H.
(Ed.) Syntactic Wordclass Tagging. Dordrecht: Kluwer Academic
Publishers, pp. 117-133.
Mikheev, Andrei (2002). Periods, Capitalized Words, etc. Computation
Linguistics 28(3), pages 289-318.
Palmer, David D. and Matti A. Hearst (1997). Adaptive multilingual
sentence boundary disambiguation. Computational Linguistics 23(3),
pages 241-267.
Schmid, Helmut (2000). Unsupervised Learning of Period Disambiguation
for Tokenisation. Internal Report, IMS, University of Stuttgart.
http://www.stanford.edu/~jstrunk/pubeng.html
____________
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
____________