Welcome to the December 30, 2013 edition of ACM TechNews, providing timely information for IT professionals three times a week.
Please Note: In observance of the New Year's Day holiday, ACM TechNews will not be published on Wednesday, Jan 1. Publication will resume Friday, Jan. 3.
Updated versions of the ACM TechNews mobile apps are available for Android phones and tablets (click here) and for iPhones (click here) and iPads (click here).
HEADLINES AT A GLANCE
Brainlike Computers, Learning From Experience
The New York Times (12/29/13) John Markoff
A new type of computer chip to be released next year uses a computing approach modeled after the human nervous system to give it the ability to automate programming-heavy tasks as well as to circumvent and tolerate glitches. With a focus on how neurons respond to stimuli and link with other neurons to interpret information, the technology enables computers to digest data while executing a task, and adjust what they do based on the changing signals. The neuromorphic processors are composed of electronic elements that can be connected by wires that emulate synapses. Rather than being programmed, the links between circuits are weighted to correlations in data the processor already has learned, and those weights are then modified as data enters the chip; this triggers a change in values and a spike, producing a signal that travels to other elements, thus altering the neural network. The constantly changing algorithms inspired by biological design enable continuous adaptation and the ability to work around errors. "We're moving from engineering computing systems to something that has many of the characteristics of biological computing," says the California Institute for Telecommunications and Information Technology's Larry Smarr. Driving this new model is the growing body of scientific knowledge about the brain, but that knowledge is still limited, notes Stanford University's Kwabena Boahen.
Federal Judge Upholds NSA's Vast Telephone Data Collection
Los Angeles Times (12/27/13) Timothy M. Phelps
U.S. District Court Judge William H. Pauley on Friday handed down his ruling in a legal challenge brought by the American Civil Liberties Union (ACLU) against the U.S. National Security Agency's telephone metadata collection program, saying the program is both constitutional and a vital part of the effort to protect the nation from terrorist attacks. Pauley's ruling was based on a 1979 U.S. Supreme Court precedent, which holds that telephone metadata is not protected by the Constitution's prohibition on unreasonable searches and seizures because telephone users voluntarily give that data to telecommunications providers and other third parties. In addition, Pauley said the amount of data being collected was not a factor. "The collection of breathtaking amounts of information unprotected by the 4th Amendment does not transform that sweep into a Fourth Amendment" violation, he said. The judge also said the telephone metadata collection program was a necessary response to the Sept. 11 attacks, and the program may have even been able to prevent the attacks had it been in place in 2001. The ACLU is expected to appeal the ruling, setting up the possibility the U.S. Supreme Court will have to reconcile two conflicting lower court opinions about the constitutionality of the metadata collection program.
Report Details Work of Elite NSA Hackers
Associated Press (12/30/13) Raphael Satter
German magazine Der Spiegel reported on Sunday that the U.S. National Security Agency's (NSA) hacking unit intercepts computer deliveries, monitors Microsoft's internal report system, and takes advantage of hardware vulnerabilities. The unit, known as Tailored Access Operations (TAO), has been characterized as a team of elite hackers whose mission, according to Der Spiegel's citing of international NSA documents, was "getting the ungettable." By calling on the U.S. Central Intelligence Agency or the Federal Bureau of Investigation, the Der Spiegel report claims that TAO can intercept electronic devices ordered by targets and outfit them with espionage software before sending the device on to the target. The unit reportedly is able to spy on Microsoft's crash reports, which the German magazine says gives the agency's spies information on how to break into machines running Windows. According to the report, TAO maintains a catalog of high tech gadgets to use on tough cases, and frequently takes advantage of the architectural weaknesses of the Internet to send malware to certain computers. The report also says TAO takes advantage of weaknesses in hardware and software produced by the world's top information technology companies, including U.S. firms.
Laser Demonstration Reveals Bright Future for Space Communication
NASA News (12/23/13) Dewayne Washington
The U.S. National Aeronautics and Space Administration (NASA) has completed its Lunar Laser Communication Demonstration (LLCD) mission, showing significant potential for increasing broadband capabilities in space using laser communications. The LLCD tested laser communication capabilities from a distance of almost 250,000 miles. The mission achieved record-breaking data download speeds to the moon of 622 Mbps and uploads of 20 Mbps, and demonstrated that it could operate as well as any NASA radio system. "Throughout our testing we did not see anything that would prevent the operational use of this technology in the immediate future," says NASA's Don Cornwell. LLCD provided error-free communications during broad daylight and when the moon was low on the horizon, proving that wind and atmospheric turbulence did not significantly impact the system. Cornwell says LLCD was able to download the Lunar Atmosphere and Dust Environment Explorer's (LADEE) entire stored science and spacecraft data of 1 GB in less than five minutes, while downloading the same data using LADEE's onboard radio system would take several days. As a next step, NASA will conduct its Laser Communications Relay Demonstration to confirm continuous laser relay communication capabilities at more than 1 billion bits per second between two Earth stations using a satellite in geosynchronous orbit.
Stanford Researchers: It Is Trivially Easy to Match Metadata to Real People
The Atlantic (12/24/13) Rebecca J. Rosen
Stanford researchers have demonstrated that the specific identity of a caller can be determined using only limited metadata. In an experiment with volunteers who agreed to use the Android app MetaPhone, researchers Jonathan Mayer and Patrick Mutchler pinpointed phone numbers' owners within hours. "We randomly sampled 5,000 numbers from our crowdsourced MetaPhone dataset and queried the Yelp, Google Places, and Facebook directories," the researchers note. "With little marginal effort and just those three sources--all free and public--we matched 1,356 (27.1 percent) of the numbers." By adding human analysis, the researchers could associate an individual or a business with 60 of 100 random numbers in under an hour, and that figure rose to 73 when the three online directories were added. Using an inexpensive, consumer-oriented data aggregator, Google Search, and the three initial sources, the researchers linked a name to 91 of the 100 numbers. The researchers say their work is significant in light of recent efforts by government officials to defend U.S. National Security Agency (NSA) phone data collection by stressing the metadata does not include specific names. "If a few academic researchers can get this far this quickly, it's difficult to believe the NSA would have any trouble identifying the overwhelming majority of American phone numbers," the researchers say.
A Breakthrough for Speeding Satellite Feeds
Technology Review (12/23/13) David Talbot
Researchers at the Hamilton Institute, the Massachusetts Institute of Technology (MIT), and MIT's Lincoln Laboratory say they have developed a data-coding technology based on transmission control protocol (TCP) they say could more than double bandwidth on satellite Internet connections, improving service to developing countries and fixing problems with live video feeds. The gains could be as much as 20-fold in conditions where data losses are exceptionally large, according to the researchers. When regular TCP is used on wireless networks, some bandwidth gets wasted on back-and-forth traffic to recover the inevitably dropped portions of a signal. The new TCP model sends mathematical functions describing multiple packets so that a receiving device can solve for missing ones without having to regather them. "You transfer more packets than what you normally would, but you don't have to retransmit," says Inmarsat's Ammar Khan. The new coding technology has been tested in lab simulations, and in the coming year it is expected to be tested in an environment that emulates the long travel times and typical data losses found on satellites. After those rounds of testing, if the system shows promise, it will be tested on real satellites and considered for commercialization, according to Khan.
Smartphone as Mentor: How Tech Could Change Behavior
University of Michigan News Service (12/20/13) Nicole Casal Moore
This fall, University of Michigan professor Jasprit Singh taught a class called Imagine, Innovate, Act!, in which students from the engineering, art, music, and health fields designed mobile apps to help users set and meet wellness milestones. "The goal of this course was to bring harmony between what we know and what we do," Singh says. The students created apps that focused on delivering messages to users at a set time or place. For example, an app called Balance was designed to offer easy and routine access to short exercise videos that could improve coordination and prevent falls. Another app, WeeAddition, guides women through pregnancy, while Joggle is a collaborative art, poetry, and music app that encourages creativity. Another app, College Granny, aims to help students balance studying and socializing to develop healthy habits in both parts of their lives. "The app would be able to have a constant, almost living presence in the user's life, and could thus help them form their decisions even more than if it were just a browsing app," says master's student Diana Sussman. Singh thinks the course's concept could continue to grow with other programs because of the do-it-yourself nature of the class.
Motion-Captured Laughs Make Animations More Amusing
New Scientist (12/20/13) Paul Marks
University College London (UCL) researchers are working on the European Commission's Ilhaire project, which aims to make chatbot avatars laugh more realistically. The work is based on the knowledge that in everyday conversation people tend to laugh about once every minute, and psychologists say keeping that up is important for making people feel engaged with someone they are talking to, according to UCL's Harry Griffin. As part of the research, nine volunteers dressed up in motion-capture suits and watched videos and played games designed to induce laughter. Thirty-two observers then classified the laughter, recreated in stick figure animations, in categories such as hilarious, social, awkward, or fake. "Even a simple stick figure can create a very powerful impression of someone laughing, as we are very familiar with this type of motion," Griffin says. "And from the observers' perceptions of these animations we can infer what movements are important for conveying different kinds of laughter." The researchers used the stick figures to develop more realistic three-dimensional animated avatars.
SMART CARS: Vehicles are Becoming 'Portals' to the Information Highway
Asahi Shimbun (12/20/13) Hitoki Nakagawa
In the three years since Google announced it was developing an autonomous car for use on public roads, traditional car manufacturers are finally starting their own development programs. "Autonomous cars should connect to the Internet when driving through residential areas" to move to the next level, says Continental engineer Ibro Muharemovic. Autonomous cars need a lot of detailed information, including the location of bumps in the road and the speed at which nearby people are walking. There are so many decisions that may need to be made in various situations, it would be more efficient to connect to the Internet and cross-check radar information, rather than installing high-performance computers that can process huge amounts of data. "Now is the time for us to look at vehicles on the road the same way we look at smartphones, laptops, and tablets, as pieces of a much bigger and richer network," says Ford Motor executive chairman Bill Ford. In addition, controlling various functions via apps, such as Internet radio stations and diagnostics for vehicular devices, is a simple way of making cars more user-friendly, according to General Motors executive director Timothy R. Nixon. The potential for software applications is increasing because most vehicular components are now controlled electronically by microcomputers.
Stanford and Google Team Up to Simulate Key Drug Receptor
Stanford News (12/19/13) Bjorn Carey
Researchers at Stanford University and Google have led a first-of-its-kind atom-level simulation of a cell's G protein-coupled receptor (GPCR) that could help improve drug design and lead to specialized scientific projects on cloud computer systems. The project was the first to be completed using Google Exacycle's cloud computing platform, which lets scientists use Google servers during low network demand times to run big data queries. The team simulated a GPCR called beta 2 adrenergic receptor site transforming between its two base configurations. Approximately 40 percent of all medications act on GPCRs, and knowing exact atom locations is critical for drug development. In the past, maps of atoms and other receptors have been generated using X-ray crystallography, but the technique can only visualize a molecule in its resting state and intermediate forms could be medically significant. "The computational burden of a model that is faithful to atomic details is very high," says Stanford professor Vijay Pande. "A very fast computer processor can compute a billionth of a second of this reaction in one computer day." Using Google's Exacycle cloud computing system, the team leveraged a distributed network of computers to process data in parallel.
Four Research Companies Pursue Artificial Intelligence for Image Intelligence
Military & Aerospace Electronics (12/19/13) John Keller
Researchers at Teledyne Scientific & Imaging, Siemens, Carnegie Mellon University, and HRL Laboratories are collaborating on the Intelligence Advanced Research Projects Agency's Knowledge Representation in Neural Systems (KRNS) program to improve image intelligence. The artificial intelligence research project aims to create and evaluate new theories that explain how conceptual knowledge is represented in the human brain within spatial and changing patterns of neural activity. The companies will work on systems that forecast neural activity patterns associated with certain concepts and that interpret concepts represented within measured patterns of neural activity. All neural activity data gathering will be noninvasive, using methods such as functional magnetic resonance imaging and magnetoencephalography. The KRNS program could assist intelligence analysts with deep exploration of images, video, text, and other data sets. Analysts use conceptual knowledge to resolve ambiguities, make inferences, and draw conclusions about intelligence data. Officials say understanding the brain's representation of conceptual knowledge will aid in the creation of analysis tools that acquire, organize, and use knowledge more efficiently. The project could lead to new techniques for training intelligence analysts and linguists.
Researchers Simulate a Worm's Internal Muscle Sensations With Code
IO9 (12/19/13) George Dvorsky
Researchers have reached a milestone in an effort to create a virtual nematode worm in a computer. The OpenWorm Project has used algorithms alone to artificially recreate internal muscle sensation. The developers are using the open source channel to crack the long-term memory code of the C. elegans nematode worm, which has 302 neurons and about 6,000 synapses. In addition to achieving a building block for worm locomotion, the team used a smooth particle hydrodynamics simulator to create an environment for the virtual worm to swim in. The researchers computed one-third of a second of movement, which took 72 hours to calculate, but they would like to extend the length of time and attach a synthetic 302-neuron brain to it. The goal is to reproduce real-world behavior, validate it against experimental recordings, and let people play and interact with the worm from their browsers via Geppetto.
Abstract News © Copyright 2013 INFORMATION, INC.
To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe
|
