[Date Prev][Date Next][Date Index]
CSLI Calendar, 16 February 2000, vol. 15:19
C S L I C A L E N D A R O F P U B L I C E V E N T S
______________________________________________________________________
16 February 2000 Stanford Vol. 15, No.19
______________________________________________________________________
A weekly publication of the
Center for the Study of Language and Information (CSLI)
Stanford University, Ventura Hall, Stanford, CA 94305-4115
____________
ACTIVITIES FROM 16 FEBRUARY TO 25 FEBRUARY 2000
WEDNESDAY, 16 FEBRUARY
3:45pm Psychology Colloquium
Building 420:041
How Does Culture Influence What We Feel?
Jeanne Tsai
University of Minn.
http://matia.stanford.edu/html/talks.html#colloq
4:00pm Geometric Analysis Seminar
Building 380:381T
Scattering Theory for Nonlinear
Klein-Gordon Equation with Sobolev Critical Power
Kenji Nakanisi
Kobe University
http://math.stanford.edu/html/seminars.html
4:15pm Broad Area Colloquium For
AI-Geometry-Graphics-Robotics-Vision
TCseq 201, Lecture Hall B
Natural Language Understanding:
Word Spaces, Meaning, Structure,
Psychological Reality, Grammaticality
Chris Manning
Stanford University
http://campus-calendar.stanford.edu/CS/
Abstract below
4:15pm EE380: Computer Systems Laboratory Colloquium
NEC Auditorium, Gates B03
Lessons from Giant-Scale Services
Eric A. Brewer
UC Berkeley & Inktomi
http://www.stanford.edu/class/ee380/schedule.html
Abstract below
4:15pm ME297: Design Theory and Methodology Seminar
Building 560
Contrasting Design Environments:
Product-Oriented vs. Process-Oriented
Greg Twiss
Senior Manager, Product Design Services
http://design.stanford.edu/Courses/me297/#2/16
Abstract below
7:00pm Art, Technology and Culture Colloquium
160 Kroeber Hall (UC Berkeley)
Signal or Noise? The Network Museum
Steve Dietz
Director of New Media, Walker Art Center, MN
http://ieor.berkeley.edu/~goldberg/lecs/
THURSDAY, 17 FEBRUARY
12:15pm CSLI Coglunch
Cordura 100
Natural Selection and Wason's `Selection' Task:
Is Evolutionary Psychology an Empirical Activity?
Keith Stenning
Director, Human Communication Research Centre
University of Edinburgh
http://www-csli.stanford.edu/events/Coglunch/
Abstract below
12:15pm Graphics Lunch
CIS Auditorium, X-101,Paul Allen Center for
Integrated Circuits
Acquiring the Reflectance Field of a Human Face
Paul Debevec
UC Berkeley
http://www-graphics.stanford.edu/glunches/
Abstract below
12:45pm Stanford Networking Seminar
Gates 104
Dynamic Network Measurement Using Active Networks
Geoffrey H. Kuenning
Harvey Mudd College
http://netseminar.stanford.edu/
Abstract below
2:00pm International Computer Science Institute Talk
ICSI, Main Lecture Hall
A Statistical Model for Word Discovery in
Fluent Child-Directed Speech
Anand Venkataraman
Massey University
http://www.icsi.berkeley.edu/talks/Venkataraman.html
Abstract below
2:30pm Mathematics Department Special Lecture
Building 380:383N
S-duality, Eisenstein Series and Hall
Polynomials for Kac-Moody Groups
Mikhail Kapranov
http://math.stanford.edu/html/seminars.html
4:00pm Xerox PARC Forum
George Pake Auditorium at Xerox PARC
Light matters: The Use of Photons in
Computing Environments
Rick Lytel
Sun Microsystems
http://www.parc.xerox.com/ops/projects/forum/
Abstract below
4:15pm Mathematics Colloquium
Rigorous Results on Limiting Regimes of the
Water-Wave Dynamics
Catherine Sulem
University of Toronto
http://math.stanford.edu/html/seminars.html
4:30pm Personality Lab Seminar
Building 420:100
Modeling Short-Term Intraindividual Variability:
Some Findings and Implications
John Nesselroade
University of Virgina
http://matia.stanford.edu/html/talks.html#person_lab
7:30pm Phonology Workshop
Building 460:126
Maximal Word Size and Faithfulness in Modern Hebrew
Adam Ussishkin
Department of Linguistics, UC Santa Cruz
http://www-linguistics.stanford.edu/Linguistics/
8:00pm CCRMA Winter Concert
Auditorium of the Iris & B. Gerald Center for
Visual Arts at Stanford University
http://www-ccrma.stanford.edu/CCRMA/Events.html
FRIDAY, 18 FEBRUARY
11:00am CCRMA Hearing Seminar
CCRMA Ballroom (Main floor of the Knoll at Stanford)
Fitting Hearing Aids on Adults:
Some Research and Clinical Considerations
Gus Haas
http://www-ccrma.stanford.edu/
12:00pm Award-Winning Teachers on Teaching
Terman 217
Workshop on Public Speaking Skills:
Improving Your Vocal Authority in the Classroom
Barbi Scott
Speech Therapist
http://www-ctl.stanford.edu/events.html
12:30pm CS547: Human-Computer Interaction Seminar
Gates B01
Human Factors in Advanced Air
Transportation Technologies
Richard Mogford
NASA Ames Research Center
http://www-pcd.stanford.edu/cs547/
Abstract below
2:30pm Informal Geometry and Topology Seminar
Building 380:383N
Deformations of Minimal Lagrangian Submanifolds
Adrian Butscher
Stanford University
http://math.stanford.edu/html/seminars.html
3:00pm Applied Math Seminar
Building 380:380C
Polarization Effects on Seismic Waves
Miguel Moscoso
Stanford University
http://math.stanford.edu/html/seminars.html
3:15pm Friday Cognitive Seminar
Building 420:100
Analogy in Learning and
Development: Interactions between Relational
Language and Relational Thought
Dedre Gentner
http://matia.stanford.edu/html/talks.html#frisem
3:15pm Philosophy Colloquium
Building 90:92Q
Can Science Know When You're Conscious?
Epistemological Foundations of Consciousness Research
Alvin Goldman
Philosophy, University of Arizona
http://www-csli.stanford.edu/philosophy/
4:15pm Statistics Seminar
Sequoia Hall 200
A Barrier Option of American Type
Hui Wang
Columbia University
http://www-stat.stanford.edu/seminars/seminars.html
SATURDAY, 19 FEBRUARY
7:00pm Art, Technology and Culture Colloquium
160 Kroeber Hall (UC Berkeley)
CRASH Symposium: Critical and Historic
Issues in Net Art
Various Speakers
http://ieor.berkeley.edu/~goldberg/lecs/
TUESDAY, 22 FEBRUARY
2:30pm Statistics Seminar
Sequoia Hall 200
Analysis of Spectral Data via
Wavelets and Functional Linear Model
Tianwen Cai
Purdue University
http://www-stat.stanford.edu/seminars/seminars.html
3:15pm Stanford Learning Lab
Press Warehouse, Press Staff Training Room
C.R.E.A.T.E. (Creating Research Examples
Across the Teaching Enterprise)
Rick Reis
Associate Director,
Global Learning Partnership Program, SLL
http://sll.stanford.edu/speakers/win00.html
4:15pm Computer Musings
Gates B01
Dancing Links
Don Knuth
Stanford University
http://www-cs-faculty.stanford.edu/~knuth/musings.html
4:15pm Statistics Seminar
Sequoia Hall 200
The East (= One-Dimensional Asymmetric
Constrained Ising) Model: Towards Rigorous Results
David Aldous
Department of Statistics, UC Berkeley
http://www-stat.stanford.edu/seminars/seminars.html
THURSDAY, 24 FEBRUARY
12:00pm Award-Winning Teachers on Teaching
Econ 134
Back to the Basics: Presenting
the Foundations of One's Discipline
Peter Sells
Linguistics, Stanford University
http://www-ctl.stanford.edu/events.html
12:15pm CSLI Coglunch
Cordura 100
Predator Avoidance by Wild Bonnet Macaques:
The Roles of Classical Conditioning and Innate
Cognitive Processes.
Richard Coss
Department of Psychology, UC Davis
http://www-csli.stanford.edu/events/Coglunch/
Abstract below
12:45pm Stanford Networking Seminar
Gates 104
Power-laws, Fractals, and the Internet Topology
Michalis Faloutsos
UC Riverside
http://netseminar.stanford.edu/
Abstract below
2:00pm International Computer Science Institute Talk
ICSI, Main Lecture Hall
Discrete Curve Evolution: Shape Simplification in
Maps
Thomas Barkowsky
University of Hamburg - Department for Informatics
http://www.icsi.berkeley.edu/talks/
4:15pm Seminar on Computational Learning and Adaptation
Ventura 17
Lazy Bayesian Rules: A Technique for Making
Highly Accurate Predictions
Zijian Zheng
Data Mining Applications
Blue Martini Software, Inc.
http://www-csli.stanford.edu/cll/scla.html
Abstract below
FRIDAY, 25 FEBRUARY
12:30pm CS547: Human-Computer Interaction Seminar
Gates B01
Effects of Humor in Task-Oriented
Human-Computer Interaction
and Computer-Mediated Communication
John Morkes
Trilogy
http://www-pcd.stanford.edu/cs547/
Abstract below
3:30pm Semantics Workshop
Building 460:126
Information Withheld:
The Unspoken Properties of Specific Indefinites
Roger Schwarzschild
Rutgers University
http://www-linguistics.stanford.edu/Linguistics/
____________
BROAD AREA COLLOQUIUM FOR
AI-GEOMETRY-GRAPHICS-ROBOTICS-VISION
on Wednesday, 16 February 2000, 4:15pm
TCseq 201, Lecture Hall B
http://campus-calendar.stanford.edu/CS/
Natural Language Understanding:
Word Spaces, Meaning, Structure,
Psychological Reality, Grammaticality
Chris Manning
Stanford University
In recent years, statistical techniques based on collecting data from
large language corpora have transformed natural language processing --
just as many other areas of AI -- and statistical systems are yielding
better and still improving levels of task performance. In this talk I
will pick out and examine some of what can be done so successfully
with statistical models, from the treatment of words to sentence
parsing, but equally try to connect this work to issues of meaning,
representation, psycholinguistic plausibility, grammaticality, and
progress in achieving artificial intelligence.
About the Speaker:
Christopher Manning is an Assistant Professor of Computer Science and
Linguistics at Stanford University. His research interests include
probabilistic models of language and statistical natural language
processing, constraint-based theories of grammar (HPSG and LFG),
computational lexicography, and syntactic typology. He received his
Ph.D. in linguistics from Stanford University in 1995. From
1994-1996, he was on the faculty of the Computational Linguistics
Program at Carnegie Mellon University, and from 1996-1999 he was "back
home" at the University of Sydney, before returning to Stanford at the
start of this academic year. His most recent book is Foundations of
Statistical Natural Language Processing (MIT Press, 1999, with Hinrich
Schuetze).
____________
EE380: COMPUTER SYSTEMS LABORATORY COLLOQUIUM
on Wednesday, 16 February 2000, 4:15pm
NEC Auditorium, Gates B03
http://www.stanford.edu/class/ee380/schedule.html
Lessons from Giant-Scale Services
Eric A. Brewer
UC Berkeley & Inktomi
The last five years has seen an incredible rise in the use of
infrastructure services; Yahoo! has over 400M page views per day,
Inktomi handles more than 1B searches per month, and AOL users visit
over 4B pages per day through AOL's caching infrastructure. These are
giant-scale services and they represent the broad movement of
capabilities into the infrastructure. This new class of systems
present extreme challenges for scalability, availability, and online
evolution. In this talk, I attempt to define this space, show how some
of the problems have been solved, and provide some new ways to think
about these problems.
About the speaker:
Dr. Brewer is a professor at UC Berkeley, where he focuses on Internet
infrastructure, security, and mobile computing. In 1996, he co-founded
Inktomi, which provides scalable applications for the Internet,
including search engines, network products, and e-commerce
solutions. He is a Sloan Fellow, an Okawa Fellow, a Global Leader for
Tomorrow (World Economic Forum), and a member for Forbes "E-Gang" of
12 mavericks of Internet.
____________
ME297: DESIGN THEORY AND METHODOLOGY SEMINAR
on Wednesday, 16 February 2000, 4:15pm
Building 560
http://design.stanford.edu/Courses/me297/#2/16
Contrasting Design Environments:
Product-Oriented vs. Process-Oriented
Greg Twiss
Senior Manager, Product Design Services
Design is considered a creative activity, particularly when it
involves the development of new products. In this week's presentation,
Greg Twiss will draw on his experiences in two very different design
environments to compare and contrast the opportunities for creative
design in each. In the first, a design consultancy, he encountered a
great deal of variety in terms of the products designed, regularly
challenging him to apply his technical design skills in new ways. In
the second situation, a large networking equipment manufacturer, he
found a more stable line of product types, but greater opportunity to
design the processes by which these products were developed. Greg will
discuss his work in both environments, including the reasons he has
found both to be professionally and personally rewarding.
Greg Twiss is a Senior Manager of Product Design Services at Cisco
Systems, developing networking devices for the small/medium business
and home environments. His groups are responsible for mechanical
engineering, industrial design, PCB layout, power architecture, and
regulatory compliance functions. Before joining Cisco, Greg was a
project manager at IDEO Product Development in Palo Alto for four
years. His most recent project was leading the mechanical design team
on the Palm V product for 3Com-Palm Computing. He was also involved
with product development for Apple and Dell Computer, Samsung,
Hewlett-Packard, and BMW. Greg also helped launch "IDEO U", a
consulting business based on developing and presenting innovation and
design methodology seminars for IDEO clients. Greg is also an
independent design consultant and has developed several
biomedical-related products. He designed, patented, and licensed an
electronic catheter tracking device, and is co-developing a product
that reduces catheter-related infections. He has also filed over ten
patents in the consumer electronics and networking fields. Greg holds
BS degrees in Mechanical and Aeronautical Engineering from the
University of California at Davis, and an MS degree in Aeronautics and
Astronautics at Stanford University.
_____________
CSLI COGLUNCH
on Thursday, 17 February 2000, 12:15pm to 1:30pm
Cordura 100
http://www-csli.stanford.edu/events/
Natural Selection and Wason's `Selection' Task:
Is Evolutionary Psychology an Empirical Activity?
Keith Stenning
Director, Human Communication Research Centre
University of Edinburgh
I will talk about some recent work with Michiel van Lambalgen on
conditional reasoning in Wason's selection task. The dominant theory
of the most discussed `content effect' in this task is an
`evolutionary psychology explanation' due to Cosmides. I will present
some analysis, along with some tutorial dialogue evidence of a novel
kind, which reveals how little empirical curiosity has been shown
toward both the task itself, and the nature of this particular
evolutionary theory.
More generally, `rational analysis' theories (Oaksford & Chater) of
the task are also shown to be deficient in their attention to meaning.
Our argument is that the psychology of reasoning cannot afford to
ignore theories of semantics of natural language, and could be greatly
empirically enriched if semantic insights were allowed to guide their
investigations.
____________
GRAPHICS LUNCH
on Thursday, 17 February 2000, 12:15pm
CIS Auditorium, X-101,Paul Allen Center for Integrated Circuits
http://www-graphics.stanford.edu/glunches/
Acquiring the Reflectance Field of a Human Face
Paul Debevec
UC Berkeley
We present a method to acquire the spatially varying reflectance
properties of a human face and use these measurements to render the
face under arbitrary changes in lighting and viewpoint. We first
acquire a set of images of the face from two viewpoints under a dense
sampling of incident illumination directions. We then construct a
reflectance function image for each observed image pixel from its
values over the space of illumination directions. From the reflectance
functions, we can directly generate images of the face from the
original viewpoints in any form of sampled or computed
illumination. We then fit each pixel's reflectance function to a
parameterized reflectance model adapted to experimental data so that
the pixel's reflectance may be computed from novel viewpoints. We
demonstrate our technique with synthetic renderings of a person's face
under novel illumination and viewpoints.
Joint work with Tim Hawkins, Chris Tchou, H.P. Duiker,
Westley Sarokin, and Mark Sagar
____________
STANFORD NETWORKING SEMINAR
on Thursday, 17 February 2000, 12:45pm
Gates 104
http://netseminar.stanford.edu/
Dynamic Network Measurement Using Active Networks
Geoffrey H. Kuenning
Harvey Mudd College
Internet measurement has always been a difficult task, due to the
necessity for collecting data at multiple locations. Active networks,
and the active network backbone (ABONE) in particular, provide new
capabilities that can support much more powerful and flexible
measurements than have previously been practical. We describe the
design of a measurement system that takes advantage of this new
infrastructure to collect previously unavailable data, and present the
results of preliminary measurements using the system.
About the speaker:
Geoffrey H. Kuenning is an assistant professor of computer science at
Harvey Mudd College in Claremont, California. His research interests
include file systems, operating systems, mobile computing, distributed
systems, active networks, and disconnected operation. Dr. Kuenning has
also worked as a software consultant in the fields of operating
systems, embedded systems, networking, and graphics. He is a member of
ACM, IEEE, Usenix, CPSR, and a number of ACM SIGs.
____________
INTERNATIONAL COMPUTER SCIENCE INSTITUTE TALK
on Thursday, 17 February 2000, 2:00pm to 3:00pm
ICSI, Main Lecture Hall
http://www.icsi.berkeley.edu/talks/Venkataraman.html
A Statistical Model for Word Discovery in
Fluent Child-Directed Speech
Anand Venkataraman
Massey University
A statistical model for segmentation and word discovery in fluent
child directed speech is presented. An incremental unsupervised
learning algorithm to infer word boundaries based on this model is
described. Although the algorithm is presented as an unsupervised
learner, empirical results are presented showing improved performance
with training size that is consistent with predictions from learning
theory.
This talk will be held in the Main Lecture Hall at ICSI, 1947 Center
Street, Sixth Floor, Berkeley, CA 94704-1198 (On Center between Milvia
and Martin Luther King Jr. Way)
____________
XEROX PARC FORUM
on Thursday, 17 February 2000, 4:00pm to 5:00pm
George Pake Auditorium at Xerox PARC
http://www.parc.xerox.com/ops/projects/forum/
Light matters: The Use of Photons in Computing Environments
Rick Lytel
Sun Microsystems
There are of order one hundred undersea cables carrying telephony
around the world. A typical transcontinental optical fiber now carries
of order 100 Gbps of traffic, and this is well below its fundamental
data carrying capacity. There are probably optical fibers in your
neighborhood carrying your phone calls, DSL traffic, and cable
channels. Why are there no optical devices in your computer? Scaling
of computer systems will determine when optics gets implemented. The
talk examines scaling of Silicon-based computers to determine where
and when optics will be utilized. The implications of scaling for more
exotic technologies, such as molecular electronics and quantum
computation, will also be described.
Rick Lytel is a distinguished engineer at Sun Microsystems. He
received a Ph.D. in theoretical high-energy physics at Stanford
University in 1980, and a B.A. in physics from the University of
Md. Baltimore County in 1975. He has started two photonics companies
and was EVP and General Manager of one of them. He ran an R&D group in
quantum optics at Lockheed, and now at Sun, is applying optics to Sun
server and network products. He has 70 publications and has presented
a few hundred invited and contributed talks and papers at conferences,
universities, and industries.
____________
CS547: HUMAN-COMPUTER INTERACTION SEMINAR
on Friday, 18 February 2000, 12:30pm to 2:00pm
Gates B01
http://www-pcd.stanford.edu/cs547/
Human Factors in Advanced Air Transportation Technologies
Richard Mogford
NASA Ames Research Center
The US National Airspace System (NAS) is composed of a huge network of
surveillance, communication, and data processing equipment that
enables the management all commercial air traffic. Operated by the
Federal Aviation Administration (FAA), the NAS ensures safe and
efficient airline flight operations. A workforce of over 14,000 air
traffic controllers located in air traffic control towers, Terminal
Radar Control facilities, and Air Route Traffic Control Centers
monitors aircraft movements and provides clearances to pilots. The
management and display of information for air traffic controllers and
maintainers raises many human factors issues that are the subject of
research by the FAA, NASA, academia, and industry.
The NAS is currently being modernized to provide controllers with new
equipment, including large-format color radar displays. Work is also
underway to develop decision support automation tools to assist
controllers with their work. The FAA has been conducting intensive
human factors work on these new systems as they near deployment. NASA
and other research establishments have been exploring new concepts and
automation tools that will improve the efficiency and safety of the
NAS.
This seminar will provide an overview of the NAS and how the FAA and
NASA address critical human factors issues that affect system safety
and performance. Examples of important projects will be provided as
well as a discussion of specific human factors techniques and tools.
Richard H. Mogford received his BA in psychology from York University
in 1974 and his MA in psychology from Sonoma State University in
1978. Dr. Mogford obtained his Ph.D. in experimental psychology and
human factors from Carleton University in 1990. He worked for
Transport Canada on the Canadian Automated Air Traffic System and was
employed until 1999 at the FAA William J. Hughes Technical Center in
Atlantic City, NJ. There he was involved in a variety of air traffic
control human factors projects. He is currently working as the Manager
of Human Factors for the Advanced Air Transportation Technology
Project at the NASA Ames Research Center. Dr. Mogford's research
interests focus on human-machine interaction in air traffic control
and flight deck systems.
____________
CSLI COGLUNCH
on Thursday, 24 February 2000, 12:15pm to 1:30pm
Cordura 100
http://www-csli.stanford.edu/events/Coglunch/
Predator Avoidance by Wild Bonnet Macaques:
The Roles of Classical Conditioning and Innate Cognitive Processes
Richard Coss
Department of Psychology, UC Davis
Wild bonnets macaques (Macaca radiata) living in urban and forest
settings react to conspecific alarm calls by running up trees to seek
refuge. The alarm calls of neighboring species, such as langurs and
sambar deer, are also provocative to these monkeys, but only among
troops that hear these alarm calls routinely in the forest. Juvenile
monkeys treat interspecific alarm calls as signifying danger,
indicating that the time frame for learning occurs during infant
development. A Pavlovian process of associative learning is proposed
to explain how bonnet macaques in the forest learn the alarm calls of
neighboring species. Although alarm calls provide information about
the potential presence of a predator, bonnet macaques rely on vision
to detect stealthy predators, such a pythons and leopards, partly
occluded by vegetation. Experience is not a prerequisite for leopard
recognition and juveniles are more excited by snakes than are adults,
suggesting that long-term experience attenuates the reactivity to
snakes. Unlike snakes, leopards are provocative to both juveniles and
adults. Experimental evidence will be presented showing that the
felid configuration and the spotted yellow coat are essential
leopard-recognition cues, possibly because these cues have been
available historically as contextual sources for natural
selection. Similar processes of natural selection might explain the
high reactivity of other primates to leopards. Together, these
examples provide a backdrop for evaluating how early learning and
evolved cognitive systems promote survival in a dangerous world.
____________
SEMINAR ON COMPUTATIONAL LEARNING AND ADAPTATION
on Thursday, 24 February 2000, 4:15pm to 5:30pm
Ventura 17
http://www-csli.stanford.edu/cll/scla.html
Lazy Bayesian Rules: A Technique for Making Highly
Accurate Predictions
Zijian Zheng
Data Mining Applications
Blue Martini Software, Inc.
The naive Bayesian classifier provides a simple and effective approach
to classifier learning, but its attribute independence assumption is
often violated in the real world. A number of approaches have sought
to alleviate this problem. A Bayesian tree learning algorithm builds a
decision tree, and generates a local naive Bayesian classifier at each
leaf. The tests leading to a leaf can alleviate attribute
inter-dependencies for the local naive Bayesian classifier. However,
Bayesian tree learning still suffers from the replication,
fragmentation, and small disjunct problems of tree learning. While
inferred Bayesian trees demonstrate low average prediction error
rates, there is reason to believe that error rates will be higher for
those leaves with few training examples. In this talk, I present an
application of lazy learning techniques to Bayesian tree induction and
presents the resulting lazy Bayesian rule learning algorithm, called
LBR. For each test example, it builds a most appropriate rule with a
local naive Bayesian classifier as its consequent. I will show that,
on average, this new algorithm obtains lower error rates significantly
more often than the reverse in comparison to a naive Bayesian
classifier, C4.5, a Bayesian tree learning algorithm, a constructive
Bayesian classifier that eliminates attributes and constructs new
attributes using Cartesian products of existing nominal attributes,
and a lazy decision tree learning algorithm in a wide cross-selection
of natural domains. It also outperforms, although the result is not
statistically significant, a selective naive Bayesian classifier. I
will also demonstrate using experiments with these domains that the
computational requirements of LBR are reasonable.
Furthermore, I analyze the LBR algorithm using the bias and variance
decomposition. I will show that LBR significantly reduces the bias of
naive Bayesian classification at a cost of a slight increase in
variance. Empirical comparison of LBR with boosting decision trees, a
technique being considered as a breakthrough in recent machine
learning research, shows that LBR has, on average, significantly lower
variance and higher bias. As a result of the interaction of these
effects, the average prediction error of LBR over a range of learning
tasks is at a level directly comparable to boosting. Empirical
comparison of LBR with bagging decision trees shows that LBR has lower
average variance and bias, and thus lower average error.
All these results suggest that LBR provides a very competitive
learning technique where error minimization is an important criterion.
____________
STANFORD NETWORKING SEMINAR
on Friday, 25 February 2000, 12:45pm to 1:15pm
Gates 104
http://netseminar.stanford.edu/
Power-laws, Fractals, and the Internet Topology
Michalis Faloutsos
UC Riverside
Despite the apparent randomness of the Internet, we discover some
surprisingly simple power-laws of the Internet topology. These
power-laws hold for several snapshots of the Internet, between
November 1997 and Aug 1999, despite a 50% growth of its size during
that period. We show that our power-laws fit the real data very well
resulting in correlation coefficients of 96% or higher. Our
observations provide a novel perspective of the structure of the
Internet. The power-laws describe concisely skewed distributions of
graph properties such as the node outdegree.
An open question is why such regularities exist in something as ad-hoc
and "random" as the Internet. We attempt to provide an intuitive
exaplantion.
Joint work with Petros Faloutsos, Christos Faloutsos and George
Siganos.
About the speaker:
Michalis Faloutsos is an assistant professor in the Department of
Computer Science at the University of California, Riverside. He
received his PhD and Master's from the University of Toronto and his
5-year BA from the National Technical University of Athens (NTUA).
____________
CS547: HUMAN-COMPUTER INTERACTION SEMINAR
on Friday, 25 February 2000, 12:30pm to 2:00pm
Gates B01
http://www-pcd.stanford.edu/cs547/
Effects of Humor in Task-Oriented Human-Computer Interaction
and Computer-Mediated Communication
John Morkes
Trilogy
The use of humor is becoming more prevalent in people's interactions
with computers. (Recent examples are Microsoft's Office Assistant and
the application "electronic Laugh Out Loud,"). However, little
published research exists on whether humor is a positive or a negative
in task-oriented human-computer interaction (HCI). The prevailing
notion is that humor distracts users, wastes their time, and may cause
them to take their work less seriously. Four experiments examined the
effects of successful and unsuccessful humor in task situations
involving HCI and computer-mediated communication (CMC). The studies
used the same between-subjects design and essentially the same
experimental method. Data from the studies were compared in a direct
test of the Social Responses to Communication Technologies (SRCT)
claim that people respond to humans and computers in identical ways.
Experiments 1 and 2 each had a 2-condition (successful humor or
control) between-subjects design. In the first experiment,
participants worked on a task, ostensibly with another person in a
different room, via a networked computer (CMC). All participants
received preprogrammed comments, differing only in whether they
contained humor. Humor participants rated the "other person" as more
likable and reported greater cooperation with and similarity to the
"other person." They also made more jokes and responded more
sociably. Task time and the amount of effort participants put into the
task were unaffected by humor. In the second experiment, participants
were told they were interacting with a computer in another room
(HCI). The results from Experiment 2 were generally consistent with
those from Experiment 1. However, HCI participants were less sociable,
demonstrated less mirth, felt less similar to their interaction
partner, and spent less time on the task. The results suggest both
that humor may enhance likability of an interface without distracting
users and that SRCT theory should be revised.
Experiments 3 and 4 replicated and extended the first two experiments,
with additional measures and conditions that used jokes seen as
unfunny. For each of these experiments, a 3-condition (successful
humor, unsuccessful humor or control) between-subjects design was
used. Results from the CMC study, Experiment 3, were mostly consistent
with those from Experiment 1, and they showed few differences between
measures for the unsuccessful-humor and control groups. New measures
showed no effects for negative affect or recall memory, but the
unsuccessful-humor group responded more unsociably than the
others. The Experiment 4 results were generally consistent with those
from Experiment 3 and SRCT theory. But comparisons of the data from
the HCI and CMC experiments showed that compared to HCI participants,
CMC participants felt more similar to their interaction partner, were
more sociable, demonstrated more mirth and more negative affect, and
liked their partner less, again suggesting a revision to the SRCT
model.
The talk will include a discussion of the study's implications for
interaction design and SRCT theory, examples of humor in user
interfaces, and guidelines for the use of humor in HCI. This talk is
based on a paper (by John Morkes, Hadyn K. Kernal and Clifford Nass)
accepted for publication in the journal Human-Computer Interaction.
John Morkes is the director of the Human-Computer Interaction Group at
Trilogy Software, a provider of business-to-consumer and
business-to-business e-commerce software. He has a Ph.D. in
Communication Theory and Research from Stanford.
____________
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. Problems with subscribing or
unsubscribing should be sent to owner-csli-calendar@csli.stanford.edu.
The full current issue is at
http://www-csli.stanford.edu/csli/Archive/calendar/current.html
and the archives at
http://www-csli.stanford.edu/csli/Archive/calendar/.
People on most of the CSLI computers can type 'help csli-calendar' to
see the current issue.
The CSLI Calendar is also posted each week to
news://nntp-csli.stanford.edu/csli.bboard.
and
news://nntp.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.
____________