Cryptography Expert Wins ACM Award for Advances in
Protecting Privacy of Information Retrieval
AScribe Newswire (05/22/08)
ACM has named Microsoft Research Silicon Valley Lab researcher Sergey
Yekhanin the winner of the 2007 Doctoral Dissertation Award. Yekhanin has
developed a new way to keep a query private when the user is accessing a
public database. With new families of private information retrieval
schemes and a special kind of error-correcting codes known as locally
decodable codes, the research supports the kind of anonymity that will
improve the security and use of cyber-infrastructure. The research has
also helped further protection for data storage, secure multi-party
computation, and computational complexity. MIT nominated Yekhanin, whose
dissertation is titled "Locally Decodable Codes and Private Information
Retrieval Schemes." He will receive the Doctoral Dissertation Award and
its $20,000 prize at the annual ACM Awards Banquet on June 21, 2008, in San
Francisco, Calif. Benny Applebaum, a post doctoral candidate at Princeton
University, Yan Liu, a research staff member at IBM Research, and Vincent
Conitzer, assistant professor of computer science and economics at Duke
University, received honorable mention and will each receive a $10,000
prize.
Click Here to View Full Article
to the top
Computer Programs Decide Humans' Fates, Set Social
Policy, Panelists Say
Wired News (05/22/08) Singel, Ryan
The growing intelligence of computer applications acting as agents of
users raises privacy and legal issues, said Brooklyn College of the City of
New York professor Samir Chopra, who spoke at a panel during ACM's
Computers, Freedom and Privacy conference in New Haven, Conn. Other
panelists emphasized that computer systems are already acting as agents and
making decisions such as if someone is a known terrorist, a likely threat
at the border, or a deadbeat parent late on child support. University of
Maryland law professor Danielle Citron noted that government agencies are
now using software to make the decisions, instead of just for decision
support. Citron cited Colorado's public benefits computer system that
determines whether people are eligible for Medicaid and food stamps, which
denied food stamps to people with prior drug convictions, a direct
violation of state and federal law. "Humans trust computers," Citron said.
"And the programmers didn't check for legal compliance; they checked for
bugs." Such errors are likely to worsen as more decisions are delegated to
computer systems, he said.
Click Here to View Full Article
to the top
E-Voting Banned by Dutch Government
InterGovWorld.com (05/21/08) Udo de Haes, Andreas
The Netherlands has banned the use of electronic voting machines in future
elections due to concerns that the technology was too vulnerable to
eavesdropping. "Developing new equipment furthermore requires a large
investment, both financially and in terms of organization," according to
the Ministry of Internal Affairs. "The administration judges that this
offers insufficient added value over voting by paper and pencil." The
Dutch government also banned voting printers, which were criticized by a
group of experts led by Bart Jacobs, a professor at Radboud University in
Nijmegen, over similar security concerns. The Netherlands will make use of
electronic vote counting, and will conduct tests to improve its
effectiveness. The local activist group "Wij vertrouwen stemcomputers
niet" (We don't trust voting computers), led by computer hacker Rop
Gonggrijp, declared the decision a victory for those who want verifiable
election results.
Click Here to View Full Article
to the top
'U2 3D' Film Director Selected as SIGGRAPH 2008 Featured
Speaker
Business Wire (05/20/08)
The Irish artist who co-directed and produced a film on the rock band U2
will be a featured speaker at ACM's SIGGRAPH 2008. Catherine Owens will
present "Giving Technology Emotion: From the Artist's Mind to 'U2 3D,'"
which will address her experiences involving the first digital 3D,
multi-camera, real-time production. "Catherine's ground-breaking work with
'U2 3D' is an excellent example of how the computer graphics industry
continues to evolve and push the boundaries for the next generation," says
Jacquelyn Martino, SIGGRAPH 2008 Conference Chair from the IBM T.J. Watson
Research Center. "Even the most complex technologies and films begin with
an idea and emotion." Owens also directed the band's "Original of the
Species" music video, which makes use of CG motion capture technology and
is considered a precursor for her work on 'U2 3D.' Supportive feedback led
SIGGRAPH to implement a multiple featured speaker format last year.
SIGGRAPH 2008 is scheduled for Aug. 11-15, at the Los Angeles Convention
Center.
Click Here to View Full Article
to the top
Open-Source File Format Is to Be a Part of Microsoft
Office
New York Times (05/22/08) P. C2; O'Brien, Kevin J.
Microsoft says it plans to give customers the ability to open, edit, and
save documents in Open Document Format within Microsoft Office 2007
software starting next year. The free update will allow users to save text
documents in ODF format and adjust Office 2007 to automatically save
documents in ODF. Starting next year, Microsoft will also allow users to
open and save files in Adobe's Portable Document Format 1.5 and PDF/A
formats. Microsoft's Chris Capossela says the decision comes from
Microsoft's commitment to making its programs more compatible with rival
software. Ivar Jachwitz of Standards Norway, the Norwegian national
standards-settings body, says proof of Microsoft's commitment to ODF and
interoperability will be seen next year when the updated version of Office
is available to users. "We have heard a lot of promises from Microsoft,
but as of yet, we are hoping for results," Jachwitz says. The
International Organization for Standards (ISO) designated ODF as the
world's first global interchangeable document standard in May 2006, but
Microsoft recently won ISO approval for Office Open XML (OOXML), a
competing interchangeable format that the company plans to develop into an
open format.
Click Here to View Full Article
to the top
Google Carves an Android Path Through Open-Source
World
CNet (05/22/08) Shankland, Stephen
Google's Android project marks an attempt to apply open-source programming
to develop a software stack for mobile devices, and Android elements that
will emerge as open-source software include a music and audio decoder from
PacketVideo and speech-recognition software from Nuance. But a major
challenge of the project is bringing in outside developers on a venture
whose code has been gestating within the company in a proprietary manner.
"The community comes at the early inception of a product, not when you
decide you're ready to ship a product," says Red Hat CTO Brian Stevens.
However, Google engineering director Andy Rubin says Android had to begin
as a closed-source project so that it could achieve stability before a
transition to open source. Google will offer a certification test suite
based on the work of project maintainers so that compatibility among
different versions of Android will be sustained. Rubin says it was
Google's desire to avoid the GNU General Public License (GPL) that
motivated its decision to engage in certain projects on a solitary basis.
"The thing that worries me about GPL is this: Suppose Samsung wants to
build a phone that's different in features and functionality than [one
from] LG," Rubin says. "If everything on the phone was GPL, any
applications or user interface enhancements that Samsung did, they would
have to contribute back. At the application layer, GPL doesn't work."
Click Here to View Full Article
to the top
45th Design Automation Conference Offers Six Full Day
Tutorials
Business Wire (05/19/08)
The Design Automation Conference (DAC) will offer six educational,
full-day tutorials designed to give middle management and design engineers
an opportunity to improve their skills in a number of design methods.
ACM's Special Interest Group on Design Automation (SIGDA) sponsors DAC,
which takes place June 8-13, 2008, in Anaheim, Calif. The tutorials will
be held Monday, June 9 and Friday, June 13. "Bridging a Verification Gap:
C++ to RTL for Practical Design" will describe how to use high level
synthesis technology to create the RTL description and formal verification
technology to verify the RTL. And "Robust Analog/Mixed-Signal Design" will
focus on the art of analog design, from the viewpoint of
analog/mixed-signal (AMS) designers, and the design of robust analog
circuits. Other tutorials include "Programming Massively Parallel
Processors: the NVIDIA Experience," "DFM Revisited: A Comprehensive
Analysis of Variability at all Levels of Abstraction," "Low Power
Techniques for SoC Design," and "System Level Design for Embedded Systems."
"We encourage design engineers, CAD developers, managers in design, and
EDA companies to consider attending a full-day tutorial this year," says
Narendra Shenoy, 45th DAC Tutorial Chair. For more information on the
full-day tutorials, and to register, visit
www.dac.com.
Click Here to View Full Article
to the top
MIT Helps Develop New Image-Recognition Software
MIT News (05/21/08) Chandler, David
MIT researchers led by Antonio Torralba, a professor in MIT's Computer
Science and Artificial Intelligence Laboratory, has found that it takes
only a few pixels of information to identify the subject of an image. The
researchers say their work could lead to advances in the automated
identification of online images and eventually could enable computers to
see like humans do. The researchers were seeking the shortest numerical
representation that can be derived from an image while still providing a
useful representation of its content. Being able to create a small
representation would be an important step toward automatically cataloging
the billions of images on the Internet. Automatic identification would
also provide a way to index pictures downloaded from a digital camera
without having to go through each picture individually. The technology
could eventually enable robots to autonomously understand the data
collected by their cameras, allowing them to determine where they are and
what they are viewing. To discover how little image information is needed
for people to recognize the subject of a picture, the team reduced images
to lower and lower resolutions and determined how many images at each level
people could identify. Torralba says people's familiarity with images
allowed them to recognize them even when coded into numerical
representations containing as little as 256 to 1024 bits of data. He says
using such small amounts of data per image makes it possible to search for
similar pictures, and because the method uses the entire image, it can be
applied to large datasets without human intervention.
Click Here to View Full Article
to the top
Experts Warn of Cyber Terrorism Threat
Associated Press (05/20/08) Zappei, Julia; Vatvani, Chandni
The International Multilateral Partnership Against Cyber Terrorism, which
involves both public and private groups from the around the world, will
create a new center in Malaysia to fight cyberterrorism. The center is
likely to open by the end of 2008 and will provide such services as
emergency response and training. Information technology has changed how
terrorists operate, said Hamadoun Toure, secretary general of the United
Nations' International Telecommunication Union, at a May 20 conference in
Malaysia attended by representatives from more than 30 countries. He said
that cybersecurity needs to be incorporated into "every aspect of keeping
ourselves, our countries, and our world safe." Malaysian Prime Minister
Abdullah Ahmad Badawi said that nations must work together to safeguard
facilities such as nuclear power plants, dams, telecommunication networks,
and energy services from cyberterrorism.
Click Here to View Full Article
to the top
Faster Wireless Networks
Technology Review (05/21/08) Graham-Rowe, Duncan
The U.S. military is underwriting a project to develop a new generation of
mobile ad-hoc networks that facilitate faster and more reliable tactical
communications between personnel and vehicles, says BAE Systems' Greg
Lauer. His company has developed a new wireless-network protocol that
sends a description of the data rather than the data itself, which will be
tested within the next year by the U.S. Defense Advanced Research Projects
Agency (DARPA). It has been demonstrated under simulation that the
efficiency of a network using the protocol was five times higher than that
of a traditional network. The project that fostered the protocol's
development hints at the potential of network coding, which can eliminate
data bottlenecks by having the destination node receive independent packets
from various sources, with each packet containing a fragment of data that
meshes together to retrieve the complete content once all the packets have
been received and processed. MIT professor Muriel Medard says this method
consumes less bandwidth and also makes it unnecessary to keep track of
which node sent what. Under the aegis of the DARPA-funded program, BAE and
MIT developed protocols that could be used to send information to multiple
destinations by tapping network-coding principles. In simulation, the
network utilizing the protocol was shown to reduce bandwidth requirements
to one-fifth of what a conventional network needs without a downgrade in
quality. Medard believes network coding could eventually be harnessed to
uncover evidence of data tampering.
Click Here to View Full Article
to the top
European Research Consortium Turns Focus on Transactional
Memory
HPC Wire (05/21/08)
The VELOX project is a three-year European-led research effort to develop
seamless transactional memory (TM) systems for multi-core computers that
integrate well at all levels of the system stack. The project is backed by
the VELOX consortium, coordinated by the Barcelona Supercomputing Center,
that unites nine research and systems integration organizations. As
multi-core chips become more popular, eventually becoming the architecture
of choice for mainstream computing, they will require programs to be
rewritten in parallel for computers that have multiple processing cores.
One of the fundamental issues in developing parallel programs is finding a
coordinated and orderly way of accessing shared data. The TM programming
paradigm has a strong chance to become the approach of choice for handling
multi-core data. Combining sequences of concurrent operations into atomic
transactions could lead to a significant reduction in the complexity of
programming and verification, making parts of the code appear to be
sequential without needing to program fine-grained locks. The VELOX
project will develop an integrated TM stack that can span a system from the
underlying hardware to the high-end applications.
Click Here to View Full Article
to the top
Microsoft Gives Students a Peek at the Future
San Francisco Chronicle (05/23/08) P. C1; Gage, Deborah
Microsoft invited visitors to its Mountain View, Calif., research lab on
May 22nd to demonstrate about a dozen of the company's most innovative
projects. The visitors included high school students, who were given a
private lunch with Microsoft researchers. Microsoft employs about 800
people at Microsoft Research Silicon Valley and four other research labs
around the world, including one in both China and India. A sixth lab is
scheduled to open this summer in Cambridge, Mass. Projects usually
originate with researchers and then find their place in a product. One of
the projects on display fights botnets by automatically detecting computer
servers that send spam, which can be identified because the servers'
Internet protocol addresses keep changing. Another project, called Pinq,
obscures personal data during online searchers, and could be used by
companies looking to keep sensitive financial and health care data safe.
Microsoft's Silicon Valley lab is also working on tools to improve online
markets, determine how online search results should be ranked, and how bugs
can be found in large networks.
Click Here to View Full Article
to the top
The Very Model of a Modern Transistor
ICT Results (05/21/08)
Performance models for two types of power transistors will lead to more
efficient smart electrical circuits, which will make technologies such as
cars and home appliances more reliable and environmentally friendly.
Improved transistors would use energy sources in a more efficient and
economical manner, leading to greener power supplies and electronics. The
double-diffused metal oxide semiconductor (DMOS) and the lateral-insulated
gate bipolar transistor (LIGBT) are both important power transistors in the
electronics market, but a lack of accurate models on how DMOS and LIGBT
behave under different conditions has lead semiconductor manufacturers to
overcompensate in their design. The over-dimensioning of the power circuit
chips has resulted in a waste of the energy they consume and of the
materials used to make them. Researchers working on the European Union
funded Robuspic project say they have developed the models needed to make
DMOS and LIGBT chips less costly and better for the environment. The
researchers say semiconductor and system manufacturers can use the models
to design more efficient power transistors and smart circuits.
Click Here to View Full Article
to the top
Army Aims to Take Guesswork Out of Cyberdefense
Government Computer News (05/20/08) Jackson, William
The Cyber-Threat Analytics (Cyber-TA) project, funded by the Army Research
office, will create a global system to gather and correlate security events
to provide users with early warnings on upcoming attacks, as well as aid in
the configuration on sensors, filters, and other devices intended to detect
and respond to such events. The project's goal is to create software that
can be used to help program security devices. Livio Ricciulli, chief
scientist at MetaFlows, a project participant, compares the Cyber-TA's
tools to Google's page-ranking algorithms, and says the project is applying
similar principles to cybersecurity warfare. Open-source organization
Emerging Threats is also participating in the project, providing
specialized threat signatures to complement signature updates from
Sourcefire for its open-source Snort intrusion detection and prevention
system. Ricciulli says the project wants to create a way of configuring
sensors with a global understanding of what is happening around them.
MetaFlows is updating previous Cyber-TA research by expanding algorithms
for programming network security devices. The project is funded through
the end of 2009, and additional National Science Foundation funding will
last through 2010.
Click Here to View Full Article
to the top
Better Business Decisions With Real-Time Data
University of Wisconsin-Milwaukee (05/21/08)
University of Wisconsin-Milwaukee (UWM) professor Hermant Jain is
developing a cross-disciplinary Real Time Enterprises Research Program that
will attach intelligent cyber devices to physical objects through the
Internet. The program will enable smart appliances to communicate with the
electric company to find the least expensive time to run a cycle, giving
them the ability to shut down or turn on based on programmed decision
rules. Other applications could include tracking attendance at a summer
festival by monitoring ticket bar codes, or improving health care by
continuously monitoring patients in the hospital and at home. Jain says
the decreasing cost of networks and sensor devices has made it practical
for many businesses to deploy real-time information systems. UWM professor
Matthew Petering, who has been working with Jain on programming and other
aspects of software involved in real-time systems, says there has been
"explosive" growth in the use of micro- to nanoscale embedded devices and
sensors in all aspects of manufacturing. "We are working to develop a
service-oriented, event-driven, smart cyber-agent (SES) approach for
real-time management of global manufacturing enterprises that combines the
ideas of both centralized and decentralized control," Petering says.
Click Here to View Full Article
to the top
Cat Brain Could Provide Bionic Eye Firmware
New Scientist (05/21/08) Axt, Barbara
Software developed by researchers at the Smith-Kettlewell Eye Research
Institute can perceive moving images in much the same way as a cat's brain,
a development that researchers hope will lead to implants that make it
possible for people to see without an optic eye nerve. The researchers
developed the software by recording the responses of 49 individual neurons
in the part of a cat's brain called the lateral geniculate nucleus (LGN),
which receives and processes visual information from the retina, via the
optic nerve, before sending it on to the cerebral cortex. Using a mix of
simple stimuli such as dots and bars, and then shifting to more complex
moving artificial scenes, the researchers determined the basics of the
LGN's response to visual features. The collected data made it possible to
build a software model of the LGN that can approximate how the neurons
would respond to real scenes. The model was tested against scenes recorded
on a camera attached to a cat's head. The model's predictions were 80
percent accurate when viewing artificial scenes, but accuracy fell to 60
percent when viewing the scenes from the camera or scenes from a movie with
independently moving elements. Nevertheless, the researchers say being
able to predict LGN activity from moving images at all is significant.
Click Here to View Full Article
to the top
Schools, Businesses Must Adapt to 'Thumb Generation,'
Study Says
Network World (05/15/08) Fontana, John
Generation Y is placing new technology demands on schools and will also
put the same pressure on their future employers in the next three to five
years, says "Technologies to Teach the Thumb Generation," a new Basex
report. The report says that young people are used to being connected and
having access to a number of computing devices, and the modern classroom
will need to be equipped with high-tech tools to help them process and
retain information. The report cites classroom-capture systems, which
would allow students to access lectures and materials at a later time, and
interactive white boards, which use the pull model (student to teacher) for
the flow of information, as potential tools for the next-generation
learning space. In the workplace, those born from 1981 to 2000 will be
well prepared to use social-networking tools, similar to the way the
current generation was ready to use technologies such as instant messaging.
Such technology will be a key to retention and improved productivity, so
companies will need to educate themselves on security and the proper use of
the tools in the workplace.
Click Here to View Full Article
to the top
Web Searching in a Multlingual World
Communications of the ACM (05/08) Vol. 51, No. 5, P. 32; Chung, Wingyan
The abundance of multilingual Web sites carries significant implications
for Web searching, and Web search portals seek to address this challenge,
writes Santa Clara University's Wingyan Chung. He reviews search engines
in the emerging languages of Spanish, Chinese, and Arabic, and notes for
instance that the Arabic Web accounts for less than 1 percent of total Web
content despite the fact that the language is spoken by over 284 million
people in nearly two dozen countries. Chung says it is a typical feature
for Spanish, Chinese, and Arabic search engines to present results as lists
of textual items, which can restrict comprehension and analysis by users.
He recommends that any prospective search-engine developer should carry out
a domain analysis before building a portal in any specific language,
including a review of existing portals and technologies to guarantee
comprehensive coverage. The mitigation of information overload requires
pre- and post-retrieval analysis, which is supported by modules that
include encoding conversion, summarization, categorization, and
visualization. Chung rated three prototype search portals--the Chinese
CBizPort, the Spanish SBizPort, and the Arabic AMedPort--by comparing them
to benchmark search engines in each of the three languages, and the results
indicate that Web searching in a multilingual world is supported by the
framework. Though information overload is indeed alleviated by
post-retrieval analysis methods, the degree of such improvement varies
across domains. On the basis of his findings, Chung advises system
developers and IT managers to embed browse support and analysis tools
within their online search systems and portals to enhance traditional
textual list displays, while remaining cognizant of the fact that such
tools are still susceptible to error mainly because of vagaries in
natural-language processing and high computational costs.
Click Here to View Full Article
to the top
