Welcome to the October 18, 2010 edition of ACM TechNews, providing timely information for IT professionals three times a week.
HEADLINES AT A GLANCE
Internet Security Plan Under Review Would Alert Users to Hacker Takeover
Associated Press (10/18/10) Lolita C. Baldor
An Australian program allowing Internet service providers (ISPs) to alert customers if their computers are commandeered by hackers and restrict online access if they do not correct the problem is being considered by the U.S. government. Certain sections of the plan have kindled the interest of experts and U.S. officials, but any government attempt to monitor or regulate the Internet could spark public opposition. White House cybercoordinator Howard Schmidt says the United States is studying a number of voluntary ways to help small businesses and the public better shield themselves online, and possibilities include provisions in the Australian effort that enable customers to receive alerts from their ISPs if their computer is hijacked by hackers via a botnet. However, officials are not advocating an option in the program that permits ISPs to block or limit Internet access by customers who fail to fix their infected computers, arguing that this would be technically problematic and face heavy resistance. Center for Strategic and International Studies fellow James Lewis says that ISPs playing a part in defending online customers from cyberattack is an inevitability, but Harris Corp.'s Dale Meyerrose cautions that voluntary programs will be insufficient. "We need to have things that have more teeth in them, like standards," Meyerrose says.
The Second Coming of TSUBAME
HPC Wire (10/14/10) Michael Feldman
Tokyo Tech's TSUBAME 2.0 will officially be named the fastest Japanese supercomputer next month, with a raw performance of 2.4 petaflops that surpasses the total floating point operations per second capability of all other government and academic supercomputers currently in Japan. TSUBAME is part of a new generation of general-purpose graphics processing unit-powered systems penetrating large research institutions across the globe. First-generation TSUBAME systems were built with Sun Microsystems equipment, but Tokyo Tech turned to Hewlett-Packard to co-design the second-generation TSUBAME along with NEC. TSUBAME 2.0 is one-quarter the size and will consume about one-quarter the power of Oak Ridge Laboratory's Jaguar supercomputer while having roughly the same peak performance. TSUBAME project leader and professor Satoshi Matsuoka says the supercomputer will be particularly useful for real-world applications in climate and weather projections, biomolecular simulation, tsunami modeling, CFD codes, and other scientific codes.
From Handwritten CAPTCHAs to "Smart Rooms," Tech Solutions Start With Pattern Recognition
University of Buffalo NewsCenter (10/14/10) Ellen Goldbaum
University of Buffalo (UB) researchers have developed a method for using handwritten CAPTCHAs to identify humans online. "Here at UB's Center for Unified Biometrics, we're the only ones who have proposed and thoroughly studied handwritten CAPTCHAs," says UB computer scientist Venu Govindaraju. "Our perspective is that humans are good at reading handwriting, machines are not." Their research is based on pattern recognition. UB researchers also are using pattern-recognition techniques to develop smart-room biometric technologies that are applicable in larger arenas, such as shopping centers, airports, and other transportation centers. Biometric technologies under investigation include hand gestures and facial, voice, and gait recognition. "This, too, is all pattern recognition, but instead of letters, here, we're trying to standardize gestures," Govindaraju says. "It's like developing an alphabet of gestures so machines can be programmed to do gesture recognition." The researchers say smart-room technologies could be used to monitor people living in assisted-living facilities or in offices and other environments for security purposes.
Artificial Intelligence Has a Feel for Laboratory Science
The Engineer (United Kingdom) (10/14/10)
The best student paper award at the recent Discovery Science international conference went to two Southampton University doctoral candidates who developed an artificial intelligence system for performing scientific experiments in a lab. Chris Lovell developed the artificial experimenter software, and Gareth Jones developed the platform for conducting experiments manually. The software examines the available data, builds hypotheses, and chooses the experiments to perform, all without human interaction. The artificial experimenter is designed to learn from a small number of experiments, and to question whether the data is correct. Jones also is building an automated lab-on-chip platform that will be integrated with the artificial intelligence software to allow full autonomous microscale experimentation. "The artificial experimenter will provide a tool for scientists, which will not only allow them to reduce experimentation costs, but will also allow them to redirect their time from monotonous characterization experiments, to analyzing the results, building theories, and determining uses for those results," the researchers say in their paper.
Robots 'Think' With Their Hands
ICT Results (10/18/10)
The European PACO-PLUS project is developing a new approach to robot cognition based on a theory called object-action complexes (OACs). Robotic cognition theorists believe that if perception and interaction can be developed, intelligence will emerge spontaneously. OACs are types of software and hardware that are designed to enable robots to think about objects in terms of the actions that can be performed on that object. The PACO-PLUS approach imitates the learning processes of young infants, using trial and error and watching other people to learn new information. A key part of the research involves working with humanoid robots. "Humanoid robots are artificial embodiments with complex and rich perceptual and motor capabilities, which make them ... the most suitable experimental platform to study cognition and cognitive information-processing," says Karlsruhe Institute of Technology researcher Tamim Asfour.
Parallel Java Programming System Launched by University
eWeek (10/14/10) Darryl K. Taft
University of Illinois at Urbana-Champaign researchers, led by professor Vikram Adve and Ph.D. student Robert Bocchino, have launched a project to develop a Deterministic Parallel Java (DPJ) implementation. The researchers say the parallel implementation of Java would be the first to guarantee deterministic semantics without run-time checks for general-purpose, object-oriented programs. "The broad goal of our project is to provide deterministic-by-default semantics for an object-oriented, imperative parallel language, using primarily compile-time checking," according to the DPJ website. The new system will help developers rewrite parts of parallel Java applications to simplify debugging, testing, and long-term maintenance. "It is our contention that parallel programming is much easier than concurrent programming; in particular, it is seldom necessary to use nondeterministic code," according to the DPJ site. The researchers say they launched the project because they wanted to develop a language that supports programming styles developers find most familiar and productive, such as mainstream object-oriented programming languages.
Campaign Builds to Construct Babbage Analytical Engine
BBC News (10/14/10) Jonathan Fildes
A campaign is underway to raise funds to reconstruct Charles Babbage's analytical engine, which he first proposed in 1837. The steam-powered analytical engine is regarded as the first design of a general-purpose computer that could be reprogrammed to carry out specific tasks. Computer historian Doron Swade says building the machine could answer "profound historical questions." More than 1,600 people have already pledged money to support the effort, and the campaign wants to get donations from 50,000 people to launch the project. Although no one has ever build a complete analytical engine, Babbage's son and Swade have built parts of it. Babbage created many different designs for the device, and author John Graham-Cumming, who is leading the campaign, wants to recreate a design known as Plan 28. Graham-Cumming says the first steps are digitizing Babbage's papers and deciphering his drawings. "We would then need to build a [three-dimensional] simulation of the engine [on a computer]," he says. "We can then debug it and it would make it available to everyone around the world."
Supplying Less, Revealing More
Technology Review (10/15/10) Erica Naone
IBM researchers are developing Social Lens, software that plugs into a company's in-house social networking software and filters updates made on the corporate network to reveal which posts are most relevant. Social Lens is designed to sift information by topic rather than to adapt based on what the user reads, says IBM researcher Elizabeth Daly. The program also uses social information, but is not tied specifically to a user's social network. A small pilot user study with IBM employees found that people scored posts from Social Lens as most interesting, compared with posts retrieved from within a user's social network and with a simple feed of recent posts, says IBM researcher Michael Muller. IBM researchers also are developing Audrey, a system that attempts to solve the same problem by focusing on personalization. Meanwhile, the Palo Alto Research Center's Augmented Social Cognition team and Microsoft's Fuse Labs also are developing social networking tools for business users. However, Joanne Cantor, outreach director of the Center for Communication Research at the University of Wisconsin-Madison, cautions that many users reject such tools due to concerns that they will filter out important data.
Robot Arm Punches Human to Obey Asimov's Rules
New Scientist (10/13/10) Paul Marks
University of Ljubljana researcher Borut Povse is conducting experiments in which a robot limb repeatedly hits human volunteers on the arm to evaluate human-robot pain thresholds in order to facilitate adherence to Isaac Asimov's first law of robotics, which prohibits robots from injuring people. Povse says accidental collisions between robots and humans are unavoidable, and the experiments are "taking the first steps to defining the limits of the speed and acceleration of robots, and the ideal size and shape of the tools they use, so they can safely interact with humans." The tests will continue using an artificial human arm for the purpose of modeling the physical impact of far more severe collisions. The objective is to limit the speed a robot should move at when it detects humans in close proximity to avoid harming them. Baylor College of Medicine biomechanics specialist Michael Liebschner questions the use of pain as a measure of outcome. "Pain is very subjective," he notes. "Nobody cares if you have a stinging pain when a robot hits you--what you want to prevent is injury, because that's when litigation starts."
Faster Websites, More Reliable Data
MIT News (10/14/10) Larry Hardesty
Massachusetts Institute of Technology (MIT) researchers have developed TxCache, a database caching system that eliminates certain types of asymmetric data retrieval while making database caches easier to program. TxCache is designed to solve the problem of making sure that data cached on local servers is as current as the data stored in the master database. The MIT system can handle transactions, sets of computations that are treated as a block, which means that none of the computations will be performed unless all of them can be performed. TxCache makes it easier for programmers to manage caches, says MIT graduate student Dan Ports, who led the system's development along with professor Barbara Liskov, who received the 2008 ACM A.M. Turing Award. Ports says TxCache ensures that programmers can change variables in a line of code just once, and have the cached copies be automatically updated everywhere. The system has to track what data are cached where, and which data depend on each other, Liskov says. The researchers say that during testing, Websites using TxCache were more than five times faster.
Shogi Computer Beats Female Champ Shimizu
Mainichi Daily News (Japan) (10/12/10)
A computer has defeated the top female player in shogi, the Japanese chess game. The Akara 2010 took advantage of a questionable move made by Ichiyo Shimizu during the middle of the match and won the game in 86 moves. The Information Processing Society of Japan approached the Japan Shogi Association to set up the special game between the Akara 2010 and Shimizu at the University of Tokyo. The computer used a system that combined four shogi software programs--Gekisashi, GPS Shogi, YSS, and Bonanza. Akara 2010's strategy was based on what the four shogi software programs considered to be the best move. "It made no eccentric moves, and from partway through it felt like I was playing against a human," Shimizu says. Future University Hakodate professor Hitoshi Matsubara says the computer's victory was the culmination of decades of work. "I started developing shogi software 35 years ago, and for the software to become this strong is enough for me to forget all the hard work," he says.
Computational Power Against Noise
Empa (10/12/10) Martina Peter
The Swiss Federal Office for the Environment is working with Empa's Acoustics and Noise Control Laboratory to develop a computer model to simulate noise levels along the Swiss rail network. The sonRAIL model will provide noise maps and calculate the sound exposure of individual buildings. Federal and local authorities can use sonRAIL to calculate sound levels from existing and planned rail lines and to measure the effectiveness of the countermeasures. The intensity of the train noise is based on factors such as the kind of train, its speed, the presence of cliffs or buildings that can reflect the noise, the construction of the track bed, the local topography, and the weather. Empa researchers, led by Kurt Eggenschwiler, note that a powerful computational system is required to measure all of the different factors. Empa uses the Ipazia computer cluster, which provides high computational performance and allows for parallel calculations on several processors.
Immersed in Their Work, Together
CITRIS Newsletter (10/10) Gordy Slack
Researcher Oliver Kreylos at the University of California Davis' Institute for Data Analysis and Visualization has developed tools for converting huge data sets into three-dimensional (3-D) models that can be projected like holograms into virtual space, supporting an interactive CAVE environment for researchers. Kreylos says the CAVE can now facilitate collaboration between users "across the street or across the continent" through a combination of teleconferencing and immersive imaging due to a partnership with the University of California, Berkeley's Tele-Immersion Lab. Tele-immersion could be especially beneficial to doctors who want to view medical data and collaborate with colleagues in other locations. Part of the project between the Berkeley and Davis facilities is to devise simple and efficient methods to enable one user to work in the CAVE, or a CAVE-like environment in a hospital, and for another user to have a less sophisticated but more affordable setup in a doctor's office. Another element of the project is a real-time video-capturing system, which captures 3-D images of the users and sends them to the collaborators via a broadband link. The system's third component is a 3-D visualizer, a programming tool kit for converting, developing, and processing datasets into 3-D models.
Abstract News © Copyright 2010 INFORMATION, INC.
To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.