Software That Learns From Users
Technology Review (11/30/07) Naone, Erica
University of Washington computer science professor Pedro Domingos is
developing CALO, a massive, four-year-old artificial intelligence project
to help computers understand human intentions. The DARPA-funded project
involves researchers from 25 universities and corporations focusing on many
areas of artificial intelligence, including machine learning,
natural-language processing, and Semantic Web technologies. CALO, which
stands for "cognitive assistant that learns and organizes," tries to help
users by managing information about key people and projects, understanding
and organizing information from meetings, and learning and automating
routine tasks. For example, CALO can learn about projects and who is
involved in those projects, so emails from those people can be given
priority and categorized based on subject matter. CALO can also be used to
make transcripts of meetings through voice recognition, or perform routine
tasks such as purchasing books online, searching for a hotel that meets
specific criteria, scheduling meetings, and coordinating people's
schedules. The ultimate goal is to build an artificial intelligence that
can serve as a personal assistant that can learn about a user's needs and
preferences and adapt to them without having to be reprogrammed. "It's an
amazingly large thing, and it's insanely ambitious," Domingos says. "But
if CALO succeeds, it'll be quite a revolution."
Click Here to View Full Article
to the top
20 Percent of Election Printouts Were Unreadable
Plain Dealer (Cleveland) (11/28/07) Guillen, Joe
Recently discovered problems with the paper records produced by electronic
voting machines in Cuyahoga Country, Ohio, could make a recount after next
year's presidential election a disaster. More than 20 percent of the paper
printouts from touch-screen voting machines were found to be unreadable.
The recount was necessary because the vote counting software crashed twice
on election night and the margin of victory was one-half of one percent or
less. Election workers found the unreadable ballots while conducting a
recount of two races, which involved only 17 of the county's 1,436
precincts. Board of Elections director Jane Platten says recounting the
ballots for the entire county in the 2008 presidential election could take
more than a week. Cuyahoga County uses Premier Elections Solutions
(formerly Diebold) touch-screen voting machines that store votes on a
memory card inside each machine. During the election a paper record of
each vote is printed on a long reel of paper that is stored inside the
machine. The paper record is used during recounts, but can be damaged or
unreadable, usually because of a paper jam while printing. Premier
Elections Solutions' Chris Riggall says the company will investigate the
situation.
Click Here to View Full Article
to the top
General Motors, Virginia Tech Scientists Collaborate to
Advance Neuroinformatics
Virginia Tech News (11/28/07) Daniilidi, Christina
Technological advancements in sensing technology makes it possible to take
more accurate measurements of brain activity, something computer scientists
and neuroscientists say could lead to the discovery of the complex neuronal
networks in the brain that allow for simple, automatic movements such as
reaching for a glass of water. Virginia Tech and General Motors Research
are opening the Laboratory for Neuroinformatics for the purpose of creating
algorithms that process the massive amounts of data neuroscientists collect
from the brain. The lab will be co-directed by Virginia Tech computer
science professor Naren Ramakrishnan and General Motors research scientist
K.P. Unnikrishnan. "Neuroscientists are making the transition from
studying neurons to studying networks--the sequences of firings and spikes
of activity across big groups of neurons," Ramakrishnan says. "What we are
trying to do is analyze all this data and discover something about the
network--the connections and relationships." Unnikrishnan says the many
possible applications of neuroscience-related research include analyzing
data from cars and maintaining vehicle health. But even greater
applications are possible, Unnikrishnan says. "Creation of brain-machine
interfaces is the next frontier," Unnikrishnan says. "Giving senses to
people who have lost them--vision, touch, hearing, and motor--would be a
contribution to humanity."
Click Here to View Full Article
to the top
New Grant Program Designed for 'Transformative' Computing
Research
Chronicle of Higher Education (11/30/07) Vol. 54, No. 14, P. A23;
Carnevale, Dan
The Cyber-Enabled Discovery and Innovation program will award $26 million
in grants next year to support research into a wider range of uses for
high-powered computing. Recipients will be called on to apply
computational thinking to real-world problems involving engineering and
computer science, as well as for biology, economics, and other social
sciences. Sirin Tekinay, head of the National Science Foundation's new
program, says advanced computational thinking is about the process of
sorting out data, deriving knowledge, gaining an understanding of
complexity, and then developing new sociotechnical systems. All areas of
science and engineering stand to benefit from this type of computing, which
has helped clear the way for the decoding of the human genome, the
production of complex real-time, satellite-aided maps, and the development
of the Internet. For the program, innovation is defined as research that
has the potential to produce transformative outcomes. Research
institutions can apply for more than one grant, but a researcher cannot be
named in more than two proposals during a competition cycle.
Click Here to View Full Article
- Web Link May Require Paid Subscription
to the top
Continued Growth in Science and Engineering Doctorate
Production
CRA Bulletin (11/28/07) Vegso, Jay
The number of doctorates awarded in science and engineering (S&E) fields
has risen for the fourth consecutive year, according to the National
Science Foundation. Last year the United States produced 29,854 doctorate
degrees in S&E fields, an increase of nearly 7 percent from the previous
year. Computer science doctorates led the way with a 28 percent increase
to 1,452 degrees, following a double-digit increase in CS doctorates from
the previous year. CS doctorates are up 79 percent since 2002 and now
represent a considerable share of not only S&E doctorates but all doctorate
degrees. Non-U.S. citizens have been key to the growth in CS doctorate
degree production. In the mid-to-late 1990s permanent or temporary visa
holders received about half of CS doctorates, but last year they accounted
for 61 percent. CS doctorates to U.S. citizens rose 42 percent from 2002
to 2006, but jumped 115 percent for non-U.S. citizens over the same
period.
Click Here to View Full Article
to the top
Software Strikes a Chord for Disabled Students
eSchool News (11/29/07)
Rensselaer Polytechnic Institute's "Adaptive Use Musical Instruments for
the Physically Challenged" program enables students with severe physical
disabilities to make music by just moving their heads. The system uses a
digital video camera to track a student's head movements on a computer
screen and then translates the movements into piano scales or drum beats.
Zane Van Dusen, a RPI undergraduate student in computer science and
electronic media arts and communication, developed the idea of using a
digital video camera to track the user's head. A cursor is digitally
placed on a portion of the student's head, usually the tip of the nose, to
follow the user's movements. As the cursor moves, sounds are created based
on the user's movements. Moving the head completely in one direction will
create a scale climb on the piano or a quick series of drum beats or a drum
roll. The project's ultimate goal is to eventually enable students to
compose their own pieces to help students learn the creative process and
build communication skills. "The client or patient doesn't have to be a
musician to participate," says the American Musical Therapy Association's
Al Bumanis. "The goal is not usually a performance, it's increasing
communication skills, understanding, relearning lost skills."
Click Here to View Full Article
to the top
Robots Dazzle at Japanese Exhibit
Associated Press (11/29/07) Tabuchi, Hirkoko
At the 2007 International Robot Exhibition, Japan's largest robotics
convention, several revolutionary robots were on display, showing why Japan
is a world leader in service and industrial robotics. One robot, called
Simroid for "simulator humanoid," is a human-like robot that dentistry
students can practice procedures on. Simroid has realistic skin, eyes, a
mouth fitted with replica teeth, and sensors where nerve endings would be
to alert the student when he or she is drilling too close to the nerve.
Simroid designers are still ironing out several bugs, including a function
that allows students to inject anesthetic into the robot's gums. Another
robot, called Mr. Cube, uses color sensors and a pair of dexterous hands to
solve a Rubik's Cube puzzle. Although Mr. Cube is significantly slower
than humans at solving the puzzle, the ability to quickly detect and
differentiate between colors is a breakthrough in industrial robotics.
Meanwhile, a panda-shaped robot developed by Waseda University uses a Web
camera and software to scan a person's face for smiles to help relieve
stress by making people laugh. When a hint of a smile is detected the
robot joins in the celebration by giggling and wiggling its arms and legs.
Japan had more than 370,000 robots in use in 2005, about 40 percent of the
global total, or about 32 robots for every 1,000 Japanese manufacturing
employees.
Click Here to View Full Article
to the top
Microsoft Preps Parallel Developer Tool
eWeek (11/29/07) Taft, Darryl K.
Microsoft has released an early preview of ParallelFX (Parallel Extensions
to the .Net Framework), a set of programming tools designed to help
developers approach issues related to coding for parallel environments.
ParallelFX contains new APIs to make programming on the .Net Framework
simpler and to support documentation and samples. Microsoft's S. "Soma"
Somasegar wrote in a blog post that ParallelFX runs on the .Net Framework
3.5 and relies on features available in C# 3.0 and Visual Basic 9.0.
ParallelFX also includes imperative data and task parallelism APIs,
including parallel "for" and "foreach" loops, to make the transition from
sequential to parallel programs simpler, as well as declarative data
parallelism in the form of data parallel implementation of LINQ-to-Objects,
which allows users to run LINQ queries on multiple processors. A new MSDN
center dedicated to concurrent programming was also launched with the
ParallelFX release and features a collection of whitepapers, including one
that describes the broader vision for parallel computing at Microsoft. "The
shift to multi- and many-core processors that is currently underway
presents an exciting opportunity for everyone in the software industry,"
Somasegar writes in his blog. "With an expected increase of 10 to 100
times today's compute power, the opportunities to deliver powerful and
immersive new user experiences and business value are just awesome."
Click Here to View Full Article
to the top
Petascale Computers: The Next Supercomputing Wave
IT News Australia (11/29/07) Tay, Liz
Academics are focusing their attention on petascale computers that can
perform 1 quadrillion, or 1 million billion, operations per second, almost
10 times faster than today's fastest supercomputers. Petascale computing
is expected to create solutions to global challenges such as environmental
sustainability, disease prevention, and disaster recovery. "Petascale
Computing: Algorithms and Applications," by Georgia Tech computing
professor David A. Bader, was recently released, becoming the first
published collection on petascale techniques for computational science and
engineering. Bader says the past 50 years has seen a fundamental change in
the scientific method, with computation joining theory and experimentation
as a means for scientific discovery. "Computational science enables us to
investigate phenomena where economics or constraints preclude
experimentation, evaluate complex models and manage massive data volumes,
model processes across interdisciplinary boundaries, and transform business
and engineering practices," Bader says. However, petascale computing will
also create new challenges in designing algorithms and applications.
"Several areas are important for this task: scalable algorithm design for
massive concurrency, computational science and engineering applications,
petascale tools, programming methodologies, performance analyses, and
scientific visualization," Bader says. He expects to see the first peak
petascale systems in 2008, with sustained petascale systems following
shortly behind.
Click Here to View Full Article
to the top
An Open Approach to Smarter Homes
ICT Results (11/29/07)
Today's advanced electronic devices could become the foundation for the
smart homes of the future if they could be designed to work together
intelligently. Home automation systems have become more common and
consumer electronics are increasingly network compatible, but so far no one
has united all of the technology in a home, which could lead to fridges
that can send a message to the television announcing that the door has been
left open or heating systems that turn on or off automatically when someone
enters or leaves the house. "People are finding themselves with all these
networkable devices and are wondering where the applications are that can
use these devices to make life easier and how they could be of more value
together than individually," says Philips researcher Maddy Janse. The
major obstacles preventing such smart homes are a lack of interoperability
between individual devices and the need for context-aware artificial
intelligence to manage the devices. The European Union-funded Amigo
project, coordinated by Janse, is developing a middleware platform that
will allow all networkable devices in a home to communicate as well as
provide artificial intelligence to control the devices. The Amigo system
consists of a base middleware layer, an intelligent user services layer,
and a programming and development framework so developers can create
individual applications and services. Amigo's software is open source to
encourage consumer electronics and telecom firms to develop products and
services for home networks and to ensure interoperability with different
brands.
Click Here to View Full Article
to the top
Getting to Core of the Problem
Chicago Sun-Times (11/28/07) Guy, Sandra
A Ph.D. student at the University of Illinois at Chicago (UIC) has
developed CoreWall, software that scientists at the Antarctic Drilling
program in Antarctica are using to study rock cores more effectively. When
the researchers drill for core samples to determine what the climate of the
Earth was like millions of years ago, they do not have much time to collect
data because the cores shrivel up in about a week. With CoreWall,
developed by Julian Chen, the researchers are able to develop
full-resolution digital images of the core and upload the images back to
the United States overnight using a satellite Internet connection.
CoreWall also features a visualization tool that is capable of enlarging
the high-resolution core photos for closer examination as well as
annotation. "It's something new that commercial Photoshop packages don't
have," says UIC computer science professor Jason Leigh, director of UIC's
Electronic Visualization Laboratory. "Now that we have low-cost computers
and display screens, scientists can look at the cores in their perfect form
when they are first dug out" and preserve the images. The UIC team has
received a grant from the National Science Foundation to add new
capabilities to the software, and has proposed developing "mashup" software
that would allow the scientists to gather and share data quickly over
superfast networks.
Click Here to View Full Article
- Web Link May Require Paid Subscription
to the top
Canadian Student Maps Brain Power to Image Search
Computerworld Canada (11/28/07) Schick, Shane
University of Ottawa Master's student Kris Woodbeck is mapping how the
human brain interacts with technology to power a search engine for visual
images. The search engine mimics how the brain processes visual
information and capitalizes on the processing capabilities of graphics
processors. "The brain is very parallel. There's lots of things going on
at once," Woodbeck says. "Graphics processors are also very parallel, so
it's a case of almost mapping the brain onto graphics processors, getting
them to process visual information more effectively." Woodbeck believes
his research has potential for use in medical and military applications as
well as facial recognition. Search engine specialist Guy Creese says
vendors are struggling to find the right kind of artificial intelligence to
extract the content of an image to create accurate metadata. "In text,
you've got a lot of metadata compared to images," Creese says. "For
images, it might be when you took it, with what camera, with what exposure,
that's about it ... How do you surface that metadata so it becomes much
more searchable?" Creese says the biggest problem is that indexing image
content is a manually intensive job that most organizations do not have the
manpower to accomplish. Woodbeck says he has been testing his search
engine on academic datasets that include between 60,000 and 100,000
images.
Click Here to View Full Article
to the top
Argonne's Nuclear Energy Research Moves Toward Greater
Reliance on Computer Simulation
Innovations Report (11/29/07) Hardin, Angela
The U.S. Department of Energy's Argonne National Laboratory is
increasingly relying on computer simulation and modeling to carry out
nuclear energy research. "The traditional approach to developing nuclear
energy technologies is to do a bunch of experiments to demonstrate a
process or reaction," notes Argonne's program manager for the Global
Nuclear Energy Partnership Mark Peters. "What Argonne is doing is creating
a set of integrated models that demonstrate and validate new technologies,
using a smaller number of experiments." Argonne's nuclear simulation
project leader Andrew Siegel adds that virtual experimentation can
substantially lower facilities' costs by improving the identification and
targeting of the physical experiments underlying their design. He says
Argonne computational researchers are developing SHARP (Simulation-based
High-efficiency Reactor Prototyping) software components that digitally
emulate physical processes that transpire within a reactor core. The SHARP
toolkit has been devised to exploit the lab's Advanced Leadership Computing
Facility featuring IBM's Blue Gene/P computer, which runs at a sustained
rate of 1 petaflop per second. SHARP could ultimately supplant computer
codes that are used to carry out safety assessments of aging nuclear
reactors, and Siegel says simulation tools such as SHARP could potentially
save millions of dollars in reactor design development and assembly.
Click Here to View Full Article
to the top
Futurologist Predicts Life in 2030
VNUNet (11/26/07) Williams, Ian
Futurologist and author Ray Hammond predicts that by 2030 the Internet
will evolve into a super-intelligent network, our bodies will contain
neurological interfaces, robots will play a major role in our daily lives,
and replacement organs will help extend the average life span to 130.
Hammond's predictions are part of a report, "The World in 2030," which was
produced independently following a year-long study. "If you think this
picture of life in 2030 sounds unrealistic, consider this: how many people
in 1985 would have thought that computers and mobile phones would play such
a central role in our lives today?" Hammond says. He says that no one can
accurately predict the future, but that the report identifies key trends
that are likely to shape the coming decades leading to 2030. "One thing is
certain: the rapid change that we have seen since the 1980s will not slow
down," Hammond says. "It will speed up so much that, in some ways, our
lives in 2030 will be unrecognizable today." People will be wirelessly
tagged for their own protection, with data on location and health
constantly being transmitted so help can be called in case of emergency or
sudden illness. The Internet will develop into a "super combined Web" that
is always on and always connected to every device.
Click Here to View Full Article
to the top
Torvalds on Where Linux Is Headed in 2008
InformationWeek (11/25/07) Babcock, Charles
Linus Torvalds says he believes Linux development is a lot more efficient
than any other commercial development method, not only for the kernel but
also for satellite products surrounding the kernel. Torvalds says that
Linux virtualization efforts particularly benefit from an open source model
because virtualization can mean many different things to different people.
The open source model prevents one person's, or company's, interest from
dominating the project. Torvalds says that current work on upcoming
kernels includes a lot of hardware-related work, both in terms of
peripheral drivers and platform changes. Graphics and wireless networking
and weaknesses in current Linux systems are receiving a lot of attention,
as is virtualization and switching to solid-state drives (SSD) disks.
While SSDs are currently too expensive to create a major change, Torvalds
expects them to play a bigger role in 2008 as they can make a significant
difference in reducing latency.
Click Here to View Full Article
to the top
Digital Preservation: Alliance Set to Tackle Science's
New Frontier
European Science Foundation (11/22/07)
The creation of a European digital information infrastructure that
maintains accessibility to scientific works is the goal of the Alliance for
Permanent Access, a coalition of major national and international
scientific organizations dedicated to digital preservation that was
launched at the Second International Conference on Permanent Access to the
Records of Science. The alliance includes such groups as the European
Science Foundation, CERN, ESA, the Max Planck Society, and libraries.
Meeting the coming challenges of creating the digital information
infrastructure requires the support of scientific communities, according to
the alliance. The organization will also focus on the development of
funding models and economic analyses to evaluate the cost of sharing and
accessing data and find ways to embed these costs within all funding
mechanisms for science. Stakeholders generally concur that data must be
retained in a manner that ensures open access, interoperability so that
datasets can be compared within and across scientific disciplines, and
repositories must be furnished to fulfill these requirements in a
quality-controlled and sustainable way. The European Union realizes that a
cultural shift is required, and the European Commission has assumed the
role of leveraging stakeholders and devising policy efforts on a strategic
and technical level, with an emphasis on digitization and digital
preservation. Projects the commission is undertaking include the
establishment of economic incentives for preserving data, and a proposal to
develop a study on the socio-economic drivers and ramifications of
longer-term digital preservation is underway. Up next for the alliance is
the creation of a forum on preservation and access, and the development of
a manual of good practices.
Click Here to View Full Article
to the top
Computer Simulations Advance Beyond Hollywood
New Scientist (12/01/07)No. 2632, P. 28; Marshall, Jessica
University of California, San Diego computer scientist Henrik Wann Jensen
says photorealistic computer graphics have advanced to the point where they
can be used in other industries besides gaming and special visual effects
for movies. For example, the software developed for realistic hair
simulation in "King Kong" could be used to virtualize hair product
applications, sparing companies the expense of manufacturing the products
before trying them out. Skin rendering has also made enormous progress,
and an important advance was the recognition of skin's translucency, which
led to the development of software that takes this property into account.
Jensen is now boosting the realism of skin models with a version that
divides sub-surface light into an epidermis layer containing models of two
kinds of melanin and a dermis that contains hemoglobin. "Without this,
there's a uniformity to the skin that may not be quite right," he notes.
"Things start to look a bit like wax." Jensen plans to embed individual
fibers and cells within the skin layers, and he says such models could be
employed by the cosmetics industry to generate more natural-looking
foundation. Jensen adds that a future version of the skin model could be
used to simulate the propagation of light of particular intensities through
the skin of cancer patients, which could be used to ascertain the proper
dosage of laser or radiation therapy.
Click Here to View Full Article
- Web Link May Require Paid Subscription
to the top
Hire Learning
Redmond Developer News (11/07) Richards, Kathleen
A major decrease in U.S. computer science enrollment is leading to a
paucity of enterprise-level graduates, sparking concerns and projections
about the makeup of the future IT workforce. Experts such as Northwest
Cadence's Jeff Levinson note that CS majors' coursework generally fails to
equip students with the real-world experience and business skills that are
an increasingly critical component of IT positions. "When CS graduates
come out of school, 95 percent of the time they haven't seen or heard of
use cases, have never written or read a requirements document, and don't
possess any soft skills or understanding of business consequences," he
laments. Some institutions are working with tech companies such as Google
and IBM to overhaul curriculums, bolster research programs, and draw a
wider range of students, particularly women and minorities. Associate dean
of California State University Fullerton Dorota Huizinga maintains that
there has been very little change in the software engineering curriculum
over the years, and in many undergraduate programs students are exposed to
a few descriptive courses that do not adequately train them for the
discipline, while there is also little concentration on the design of user
interfaces and improving the friendliness of software and services.
Director of Microsoft Research's External Research and Programs Group
Sailesh Chutani believes CS enrollment can be boosted by placing computing
in a socially relevant context or in the context of interesting fields such
as robotics and gaming. Some people are suggesting that students should
receive at the very least a Master of Science in CS or software engineering
in order to qualify as a professional. "Once you've gone to the master's
level, chances are you have more depth and you're more likely to fit right
into what the industry is trying to do," says Chutani.
Click Here to View Full Article
to the top
