Young Researcher to Receive ACM Award for Internet
Performance Measurement Techniques
AScribe Newswire (02/21/08)
University of California, Berkeley associate professor of computer science
Vern Paxson will receive ACM's 2007 Grace Murray Hopper Award for
outstanding young computer professional of the year for his research on
measuring Internet behavior. The award comes with a $35,000 prize funded
by Google. Paxson developed techniques that are used by both the research
community and Internet operators to assess new communications concepts,
improve network performance, and prevent network intrusion. Paxson's
Internet measurement research has brought the scientific process to
measuring the Internet's behavior and the conditions it operates under,
elevating the practice of Internet management to a higher level. Paxson's
work has also allowed the research community to evaluate new ideas and
technologies and identify problems and priorities that are needed for
increased efficiency. Paxson was named an ACM Fellow in 2006 and his 1996
research paper, "End-to-end routing behavior in the Internet," won the
first "Test of Time" award from ACM's Special Interest Group on Data
Communication (SIGCOMM).
Click Here to View Full Article
to the top
A Method for Stealing Critical Data
New York Times (02/22/08) P. C1; Markoff, John
Princeton University researchers have devised a unique and simple
methodology for stealing encrypted information stored on computer hard
disks. The technique involves literally freezing the computer's dynamic
random access memory chips, which have been demonstrated to retain data for
seconds or even minutes after power has been cut off. The data, like the
chips, is frozen in place by chilled air, allowing the keys to be read by
the researchers using pattern-recognition software. Princeton computer
scientist Edward W. Felten wrote in a Web posting that the application of
liquid nitrogen to the chips allows them to retain their data for hours
without power. He noted that this breakthrough "is pretty serious to the
extent people are relying on file protection." The experiment offers clear
proof that Trusted Computing hardware does not apparently halt potential
intrusions, the researchers said. They said they employed special
utilities in the Windows, Macintosh, and Linux operating systems to
compromise encrypted data, and reported that they started probing the
utilities for vulnerabilities last fall after noting a reference to the
persistence of data in memory in a 2005 technical paper authored by
Stanford computer scientists. "This is just another example of how things
aren't quite what they seem when people tell you things are secure," says
SRI International researcher Peter Neumann.
Click Here to View Full Article
to the top
Berkeley Researcher Describes Parallel Path
EE Times (02/21/08) Merritt, Rick
University of California at Berkeley computer science professor David A.
Patterson detailed plans for defining a new parallel programming model for
mainstream computing at an annual gathering on Feb. 21. Unless researchers
successfully define efficient parallel methods, "programming will be so
difficult that people will not get any benefit from new hardware,"
Patterson warned. His plan, which will be followed by a new Parallel
Computing Lab that Patterson will helm at Berkeley, will proceed in five
steps. The lab has already carried out the selection of researchers in
areas that include image recognition, voice recognition, and personal
health to define compelling parallel applications, while the next phase
involves the identification of common patterns through the study of those
applications. An independent group will then devise frameworks and a
composing language to help programmers invent and coordinate parallel
programming modules, and other groups will characterize new operating
system and hardware frameworks to best dovetail with a parallel model along
with tools to help programmers optimize applications. Sources said
Berkeley was chosen by Intel and Microsoft to work on the parallel
programming model challenge through a five-year, $10 million grant, while
Patterson said around 12 faculty members and 40 graduate students will be
lab participants. Separate panel discussions at the Berkeley gathering
focused on the move toward ubiquitous computing, with professor and Inktomi
founder Eric Brewer pointing out that "the IT industry we know is focused
on one-sixth of the population, but there is a tremendous untapped
opportunity in the sheer volume of the people who could be involved."
Click Here to View Full Article
to the top
U.S. Science Funding Hits a Political Wall
HPC Wire (02/22/08) Vol. 17, No. 8, Feldman, Michael
Federal funding for science education and research at the Department of
Energy (DOE) Office of Science, the National Science Foundation (NSF), and
the National Institute of Standards and Technology (NIST) was supposed to
double over the next seven years under the president's American
Competitiveness Initiative, which had bipartisan support on Capitol Hill.
However, fiscal 2008 appropriations ended up being 70 percent short for
NIST, 77 percent short for NSF, and 92 percent short for the DOE because of
partisan maneuvering. President Bush's decision to set a cap on 2008
appropriations spurred the Democrat-controlled Congress to challenge him
with an appropriations proposal that was $23 billion above the cap, and
Bush responded with a veto threat that prompted the Democrats to lower
their proposal to $11 billion over the cap; the president and the GOP
refused to accept this and forced the Democrats to meet the original budget
cap. The result has been a DOE-implemented funding freeze on university
research, significant reductions in graduate research fellowships and basic
research projects for NSF, and a cutback on scientists and engineers at
NIST. Experts generally lament the decline of funding for basic science
research, warning that it breeds complacency among the science community
and general public, and threatens America's global technology leadership by
giving other countries with heavy research and tech work force investments
a competitive edge. However, funding for the DOE Office of Science's
Advanced Scientific Computing Research program, responsible for developing
and managing the high-end computing resources of DOE's R&D initiatives, got
a substantial boost compared to 2007.
Click Here to View Full Article
to the top
Securing Cyberspace Among Top Technological Challenges of
21st Century, Panel Says
Network World (02/19/08)
Securing cyberspace was recently named as one of the top 14 technical
challenges for this century by a National Academy of Engineering panel.
The panel said that electronic computing and communication contain some of
the most complex challenges engineering has ever faced, including ensuring
the confidentiality and integrity of transmitted information, deterring
identity theft, and preventing electronic terrorist attacks that could
disable transportation, communication, and power grids. The panel noted
that serious breaches of security in financial and military computer
systems have already taken place, and that identity theft is a growing
problem, but research and development in security systems has not
progressed much beyond a strategy amounting to fixing problems and cobbling
together security patches after vulnerabilities are discovered. Other
great technological challenges the panel identified include advancing
health care informatics, improving virtual reality, engineering better
medicine, and preventing nuclear terror. The National Science Foundation
agrees with the panel and cited cybersecurity as an area it wants the
United States to invest more resources in.
Click Here to View Full Article
to the top
Can R&D Still Drive the Next Big Thing?
ThomasNet (02/19/08) Butcher, David R.
Plunkett Research estimates that global spending on research and
development topped $1 trillion in 2006, while the latest Global R&D report
from Battelle and R&D Magazine projects that worldwide R&D spending in 2008
will exceed 2007 figures by nearly 8 percent, for a total of $1.21
trillion. R&D has primarily been carried out and funded in North America,
Europe, and Asia by the 30 developed member nations of the Organization for
Economic Cooperation and Development. A Science and Engineering Indicators
2008 report from the National Science Board estimates that R&D spending has
climbed rapidly in selected Asian and Latin American economies and
elsewhere for almost 10 years, while the most rapid and sustained R&D
expansion from 1995 to 2005 has been that of China. University R&D
projects are continuing to be fed by government research dollars, while a
greater portion of U.S. R&D funding is being channeled into company-owned
or outsourced labs abroad. "The outsourcing and offshoring of research and
engineering projects are skyrocketing" on the strength of low costs,
immense talent pools, and the speed and quality of overseas innovation,
especially in India and China, Plunkett says. Japan and the United States
currently lead the world in terms of R&D as a percentage of gross domestic
project, but both countries' combined R&D portion fell from 56 percent of
the global total to 48 percent between 1995 and 2005, according to the NSB
report. Despite a record high of approximately $340 billion spent on R&D
by the United States in 2006, the NSB report observes a multiyear decline
in federal support for basic and applied research, while ACM concluded in
December 2007 that Congress is sanctioning R&D spending increases that are
not in proportion to the rate of inflation, and including allocations for
construction projects that are extraneous to its basic research funding
responsibilities.
Click Here to View Full Article
to the top
Stanford Camera Chip Can See in 3D
CNet (02/21/08) Shankland, Stephen
Stanford University researchers have developed a multi-aperture image
sensor that is capable of judging the distance of subjects in a snapshot.
The sensor can see objects differently than the light detectors used in
ordinary digital cameras. Instead of using the entire sensor to capture a
single representation of the image, the 3-megapixel sensor prototype breaks
the scene into small, slightly overlapping 16x16-pixel patches called
subarrays. Each subarray uses its own lens to view the photo subject.
After a picture is taken, image-processing software analyzes the minor
location differences for the same element in different subarrays to
determine the distance between objects. The result is a photograph
accompanied by a "depth map" that describes each pixel's red, blue, and
green light components and how far away each pixel is. Currently, the
researchers do not have a specific file format for the data, but the depth
information can be attached to a JPEG file as metadata. The multi-aperture
approach could also help reduce noise, or the colored speckles that occur
more often in digital photography at higher ISO sensitivity settings.
Noise is reduced because the multiple subarrays capture the same views,
making it easier to distinguish the true color of an object from off-color
noise. Each subarray can also be calibrated to record a specific color,
which would reduce the "color crosstalk" of current image sensors.
However, the sensor requires about 10 times as much processing power as
conventional on-chip image processing, and resolution is lower than the raw
number on the overall sensor because the same subject matter is captured by
multiple pixels.
Click Here to View Full Article
to the top
A Google Competition, With a Robotic Moon Landing as a
Goal
New York Times (02/22/08) P. C3; Stone, Brad
Google is sponsoring the Google Lunar X Prize, a competition with $30
million prizes for the first two teams to land a robotic rover on the moon
and send back images and other data. Ten teams have announced their
intention to participate in the competition. The teams include
universities, open-source engineers, and aerospace startups. One team is
led by renowned roboticist William L. Whitaker, a Carnegie Mellon
University professor. Other teams include an affiliation of four
universities and two major aerospace companies in Italy, and a group that
is a loose association of engineers coordinating their efforts online.
"This is about developing a new generation of technology that is cheaper,
can be used more often, and will enable a new wave of explorers," says X
Prize Foundation Chairman Peter H. Diamandis. Google will award $20
million to the first team that lands on the moon, sends a packet of data
pack to earth, travels at least 500 meters, and sends a second packet. The
second team to accomplish those goals will be awarded $5 million. Bonuses
are being offered for completing tasks such as visiting a historic landing
site and finding and detecting lunar ice. The prize amounts start to
decline if the goals are not accomplished by 2012. "The idea we can help
spur the return to the moon and maybe even do it more quickly than some of
the national plans is really exciting to me," says Google co-founder Sergey
Brin.
Click Here to View Full Article
to the top
Applying Theory at Microsoft
Technology Review (02/21/08) Naone, Erica
Microsoft Research's Jennifer Chayes was recently named the managing
director for the new Microsoft Research New England lab, which opens in
Cambridge, Mass., in July. Her research has had wide-reaching applications
on the Internet, including search, keyword advertising, recommendation
systems, and social networks. When asked what has changed in the 11 years
she has worked for Microsoft, Chayes recalled a time when a colleague said
that their work had moved so much closer to applications and Chayes
responded by saying that applications had moved closer to their work. By
working in industry instead of academia, Chayes says she has been able to
hear about problems in the field much more quickly and that she was able to
be one of the first people to model such problems. Chayes recently has
been researching how to monetize social networks, focusing on finding a
balance between collecting enough data to make appropriate recommendations
while protecting peoples' private information. Her lab will work with
sociologists, psychologists, and economists in order to gain new insights
in how to approach their work. "With all the data that we've got and the
kinds of things that we want to do, I think it's really time for
mathematicians and computer scientists to start interacting with
sociologists and psychologists," she says. "I'm not an expert in what
people want. I can just model what people want."
Click Here to View Full Article
to the top
Black Hat Conference: Experts Develop Cybersecurity
Recommendations for Next President
InformationWeek (02/20/08) Hoover, J. Nicholas
Members of the recently convened Cyber Commission for the 44th President
discussed the goals of the panel during the recent Black Hat security
conference in Washington, D.C. The commission will spend the next nine
months coming up with recommendations for creating a national cybersecurity
policy. "This is one of the central issues for national security and we
want to make sure it doesn't go away," says James Lewis, director of the
technology and public policy program for the Center for Strategic and
International Studies, which supports the panel. The commission is not a
government-mandated commission, but its membership, which includes two
sitting members of Congress and Jerry Dixon, former executive director of
the National Cyber Security Division at the Department of Homeland
Security, could give the panel some clout. The panel plans to focus on
defining a clear command and control structure for federal cybersecurity,
standardizing technology procurement procedures across federal agencies,
and determining the research and development agenda.
Click Here to View Full Article
to the top
Learning About Brains From Computers, and Vice
Versa
MIT News (02/16/08) Chandler, David
MIT researcher Tomaso Poggio recently combined parallel research efforts
that examined how computers and the brain perform various functions. One
project researched using complex computational methods to understanding how
the brain works, while the other focused on improving the abilities of
computers to perform difficult tasks that human brains excel at, such as
understanding complex visual images. Poggio says the turning point came
last year when he and his team used a model vision system designed to be a
theoretical analysis of how certain pathways in the brain work to interpret
a series of photographs. Despite the fact that the model was not developed
for that purpose, it turned out to be just as good, if not better, than the
best existing computer vision systems at recognizing certain types of
complex scenes. "This is the first time a model has been able to reproduce
human behavior on that kind of task," Poggio says. The experiments
involved a task that is easy for people but difficult for computers, such
as showing a picture and asking if there are any animals in the scene. The
picture was shown for only a fraction of a second so that information would
be processed in the visual cortex, one of the brain's most complex areas.
Reaching an understanding of how the visual cortex works could be a
significant step toward understanding the brain. "Computational models are
beginning to provide powerful new insights into the key problems of how the
brain works," Poggio says.
Click Here to View Full Article
to the top
Military Aims to Seal Leaky Networks
Investor's Business Daily (02/20/08) P. A5; Riley, Sheila
To minimize future security leaks, military and academic researchers are
looking for ways to make computer networks more secure, including
techniques for understanding the thought process of network intruders.
These efforts have lead to a rapidly emerging field of research called
"intrusion prediction," which aims to predict the actions of intruders and
minimize the damage if they succeed in infiltrating a network. Researchers
are working to predict the actions of hackers by analyzing past breaches.
The project involves researchers from the Rochester Institute of
Technology, the University at Buffalo, Pennsylvania State University, the
U.S. Air Force, and CUBRC, a nonprofit research and development company.
Technology developed by the researchers can send alert messages and follow
and project actions of multiple attackers simultaneously. Rochester
Institute of Technology professor Shanchieh Jay Yang says intrusion
prediction is intended to be one part of a larger security system. "We
believe this work is quite pioneering," Yang says, but he notes that it is
not a cure-all. "Very sophisticated, experienced, and good hackers will
always find a way to get into your network," he says. "Networks are built
to allow remote access." The Navy is convinced that the strategy is
valuable and will help protect important information. "We are all over
that stuff," says Jim Granger, a technical director for the U.S. Navy's
Cyber Defense Operations Command. "You want to get from the reactive to
the proactive to the predictive."
Click Here to View Full Article
to the top
Wizkid Makes Its Debut at the Museum of Modern Art
Ecole Polytechnique Federale de Lausanne (02/19/08) Luy, Florence
The Museum of Modern Art in New York will offer visitors an opportunity to
interact with WizKid, a machine that is part computer and part robot and is
trained to follow them and their movements. As part of MoMA's Design and
the Elastic Mind exhibit, which is scheduled to run from Feb. 24 to May 12,
2008, visitors will interact with Wizkid, which has a screen on a mobile
neck that moves about like a head and is capable of asking questions
nonverbally by expressing interest, confusion, and pleasure with its body
language. It also remembers visitors when they come back in range and
continues their conversations. "Wizkid gets us AFK--away from
keyboard--and back into the physical world," says Ecole Polytechnique
Federale de Lausanne engineer Frederic Kaplan. "Unlike a personal
computer, it doesn't force the human to accommodate, and it's fundamentally
social and multi-user." Visitors will see themselves on Wizkid's screen
surrounded by a "halo" of interactive elements that can be selected by
waving their hands. Wizkid has a "creature of habit" function that will
learn preferences, and would know that the user wants to listen to some
light jazz when he or she returns home. Kaplan developed Wizkid with
Martino d'Esposito, an industrial designer who teaches at the University of
Art and Design Lausanne.
Click Here to View Full Article
to the top
University of Illinois Develops Free, Easy-to-Use Web
Tool Kit for Archivists
University of Illinois at Urbana-Champaign (02/19/08) Lynn, Andrea
University of Illinois, Urbana-Champaign Library archivists have developed
Archon, a new, freely-available online collections management program.
Archon was designed specifically for archivists with limited access to
technological resources and knowledge, says Scott Schwartz, one of Archon's
developers and the archivist for music and fine arts at UIUC. Schwartz
says the program is adaptable to any institutional setting. "We wanted our
application to be particularly useful to small, one-person repositories
that have been unable to take full advantage of current tools under
development," he says. Assistant university archivist and co-project
director Chris Prom says the software makes an archive's holdings far more
accessible to its users, and notes that it automatically creates its own
searchable Web site. Schwartz says the group's primary goal was to create
a tool that can provide immediate public access to information on the
various collections of historical documents and records found in archives.
"We took a minimalist approach, yet didn't sacrifice the standards of the
profession," Schwartz says. "We recognized what people and researchers
need to access collections of historical documents preserved in archives,
and developed the tools to help put archivists and the public in the
driver's seat to meet these important access needs."
Click Here to View Full Article
to the top
Fast-Learning Computer Translates From Four
Languages
ICT Results (02/18/08)
METIS II is a European-based machine-translation project that has
demonstrated an inexpensive technique for translating documents from Dutch,
German, Greek, or Spanish to English. Machine translation currently works
best for formal texts in specialized areas with unambiguous vocabulary and
limited sentence patterns. The European Union has been supporting research
in this field since the large Eurotra project in the 1980s, which used a
rules-based approach that taught a computer the rules of syntax and applied
them to translate texts from one language to another. However, starting in
the early 1990s, a new concept of statistical translation has gained in
popularity. Statistical translation replaces rules with statistical
methods that are based on a text corpus--a large body of written material,
up to tens of millions of words--that is intended to be representative of a
language. Parallel corpora contain the same material in two or more
languages that the computer uses to compare corpora and learn how words and
expressions in one language translate to another. Parallel corpora are
expensive and rare and exist only in a very few languages. METIS II
researchers are employing statistical machine translation without a
parallel corpora resource by using monolingual corpora for the target
language. Using a single corpus requires using a dictionary for the
vocabulary and a way of understanding syntax. METIS II matches patterns at
the "chunk" level by matching phrases or fragments of a sentence instead of
the entire sentence, which makes the pattern matching more efficient.
Click Here to View Full Article
to the top
New Research Takes Aim at Oral Cancer
University of Minnesota News (02/19/08)
An interdisciplinary team of researchers at the University of Minnesota is
combining medical research and computer science to advance medical
technology and fight oral cancer. The goal is to be able to identify
proteins that lead to oral cancer and to create a technique that can fight
the disease at its earliest stages. The researchers are working on a new
three-step process that can detect three to four times more proteins than
other methods. As part of the project, University of Minnesota computer
science professor John Carlis and doctoral student Getiria Onsongo are
using databases and data-modeling techniques to analyze the medical data
and extract and visualize the desired information. Meanwhile, public
health professor and lead researcher Baolin Wu is using complex statistical
models to determine how many proteins, out of the thousands found in human
saliva, to pinpoint for the study. Wu is also analyzing data on how the
proteins interact with each other that could potentially lead to cancerous
cells. Being able to detect signs early signs of oral cancer is critical
to increasing the survival rate.
Click Here to View Full Article
to the top
Japan: Google's Real-Life Lab
BusinessWeek (02/14/08) Hall, Kenji
Japan's widespread use of wireless broadband has made the country a sort
of unofficial testing lab for Google as it tries to refine mobile search
technology. Japan's 100 million cell phone users represent the most
diverse group of mobile subscribers. Google tests in a variety of
locations, but the Japanese are often the most critical because they are
just as likely to use a phone to access the Internet as a PC, and at speeds
that rival fixed-line broadband. Japanese carriers have also offered such
services for years, and many Web sites in Japan are formatted for cell
phones. Google is working with the two top Japanese wireless operators,
which have a combined subscriber base of 82 million. "Our fundamental
strategy is to take ideas from Japan and apply them to other markets," says
Google's Emmanuel Sauquet. Japan's influence is why Gmail users will soon
be able to include "emoji," or small animated cartoons and emoticons, in
their messages. Google relies on user-experience groups to determine what
mobile Web surfers like. Participants are given phones with Internet
access and asked to complete simple tasks, either in Google's lab or on the
streets of Tokyo. Google also conducts what it calls 1 percent tests,
which is when a small portion of users see different layouts, fonts, and
other features. The goal is to determine what changes make the service
easier to use. For example, Google found that letting users choose a
default neighborhood can make searching faster.
Click Here to View Full Article
to the top
So an EECS Prof and an Undergrad Walk Into a Computer
Lab...
University of California, Berkeley (02/13/08) Bergman, Barry
University of California, Berkeley professor Ken Goldberg and student Tavi
Nathanson are developing Jestor version 4.0, a computer program that rates
jokes and gives suggestions based on each user's personal sense of humor.
The program was first launched in 1998 and has been upgraded several times.
Goldberg says Nathanson joined the project in 2006 and has overhauled the
program. Jester sends users suggested jokes by first having the user
respond to a series of eight jokes to establish their personal preferences.
The program uses collaborative filtering to make joke recommendations.
The jokes that other users with similar preferences found humorous are
submitted to the user, who then rates those jokes, helping further define
the user's preferences. "The computer doesn't know anything about the
jokes--every joke is just a black box, the same as it would be for a movie
or a film," Goldberg says. Goldberg and Nathanson are also working on a
program tentatively called "Donation Dashboard" that will recommend
portfolios of nonprofit organizations to potential supporters based on
their interests, desired donation amounts, and other preferences.
Click Here to View Full Article
to the top
Looks Familiar
The Engineer (02/11/08)
York University computer vision lecturer William Smith is trying to make
the process of reconstructing a 3D image of a face from a single 2D image
faster and more accurate. Smith's work combines the advantages of two
face-recognition approaches. The first, more sophisticated approach uses a
morphable statistical model of facial appearance, while the second uses
classical shape-from-shading techniques. The statistical approach uses a
model and adjusts the parameters to try to fit the model to an image.
Smith says the advantages of this technique are that it works well on real
images and can recognize a variety of different faces. One weakness,
however, is that the shape it re-creates is completely determined by what
the system learned earlier, which means that it is not able to reconstruct
atypical features it has not seen before. The shape-from-shading
techniques are intended to compensate for this shortcoming. The techniques
interpret changes in brightness and darkness to detect changes in the
direction of the surface of the face. "The issue of face expression is
quite difficult with morphable model approaches--you would need hundreds of
examples of every expression," Smith says. "So we are looking at using
shading information to recover 3D shapes when they might be in a completely
different expression, then we can feed that back into the model so the
model will improve and learn to recognize these images."
Click Here to View Full Article
to the top
