Association for Computing Machinery
Welcome to the July 22, 2015 edition of ACM TechNews, providing timely information for IT professionals three times a week.

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).


Security Researchers Find a Way to Hack Cars
The New York Times (07/21/15) Nicole Perlroth

After two years of labor, security researchers Charlie Miller and Chris Valasek plan to demonstrate a method to hack into and control hundreds of thousands of vehicles at the annual Black Hat and Def Con hacking conferences in August. They used the Internet to monitor cars by their location, determine their rates of speed, turn their blinkers and lights on and off, and manipulate their windshield wipers, radios, navigation, and, in some instances, control brakes and steering. An earlier technique in which Miller and Valasek controlled certain vehicles' steering and speed by plugging into a diagnostic port was of little use to automakers. The researchers tinkered with a Jeep equipped with a car stereo head unit that linked to the Internet via a hardware chip that provides a wireless and a cellular network connection. A defect in the chip enabled Miller and Valasek to scan the Internet for affected vehicles, break into the car stereo head unit, and run their own code. They followed this with a successful hack into another chip in the same head unit that controlled the vehicle's electronics, and they found any car with the same head unit was hackable. Their research will likely be an initial discovery into vulnerabilities and attacks targeting the Internet of Things.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

RoboCup World Championship: University of New South Wales Engineers Defend World Title in China
ABC Online (Australia) (07/22/15) Andrew Griffits

A team of engineers and students from the University of New South Wales (UNSW) has successfully defended the school's title at the RoboCup World Championship, a long-running contest that pits teams of robots against each other in a soccer match. UNSW won the competition last year and successfully defended its title again at this year's competition in Hefei, China, beating a team from the University of Bremen and the German Research Centre for Artificial Intelligence. Sean Harris, a Ph.D. student at UNSW's School of Computer Science and Engineering who led the school's RoboCup team, says this year's game was much more challenging than last year's. Harris says UNSW's main advantage over their German opponents was speed. The teams are not able to control their robots during the match, relying instead on algorithms that are continually honed over the years. Harris says the UNSW team has been writing and improving its code base since the school first entered RoboCup in 1999. The improvements are in part to account for rule changes introduced every year. For example, next year the competitors will use a regular child's soccer ball, rather than the miniature version used this year.

New Computer Program First to Recognize Sketches More Accurately Than a Human
Queen Mary, University of London (07/20/15)

Researchers at the Queen Mary University of London (QMUL) say they have developed technology that is better at recognizing hand-drawn sketches than humans. Called Sketch-a-Net, the researchers say the deep neural network technology is particularly effective because it takes unique information into consideration, such as the order in which strokes are drawn. They say their program is capable of correctly identifying the subject of sketches at a 74.9-percent rate, compared to 73.1 percent for humans. Free-hand sketches are difficult to recognize because they are abstract, varied, and consist of black and white lines. The researchers say the technology could potentially provide a foundation for new ways to interact with computers. The team says it is more natural for people to retrieve specific images by drawing them with a finger on a touchscreen than doing keyword searches. Sketch recognition could "have a huge impact for areas such as police forensics, touchscreen use, and image retrieval, and ultimately will help us get to the bottom of visual understanding," says QMUL lecturer Timothy Hospedales.

Mother Robots Build Children Robots to Experiment With Artificial Evolution
IEEE Spectrum (07/21/15) Evan Ackerman

ETH Zurich researchers sought to bypass some of the limitations of evolutionary robotics by training a "mother robot" to autonomously assemble children robots out of component parts to see how well they move. The researchers note this removes the problem of having what works well in simulation not performing as well as expected in the real world. Once built, the child robots' movements are observed and evaluated, and then the machines are disassembled and their constituent elements returned to the assembly line to be rebuilt into new robots. Successful software "elite" designs are carried over to the next generation without alteration, and they also are mutated or crossbred to produce the rest of the successive generation. Five experiments were run, with 10 robot generations built, evolved, and improved in each instance. The elite designs did not always perform as well when they were retested, even though their designs were identical. Although the researchers attribute this inconsistency to significant variances in the behaviors of some of the robots, they observe "evolutionary pressure tends to select more consistent ones over generations, and usually repeatable genomes remain over generations." Each experiment yielded a fitness increase of more than 40 percent over 10 generations, and the researchers envision this method to be complementary to simulations.

What Is a 'Computer' Anymore?
The Atlantic (07/20/15) Adrienne Lafrance

Leading computer scientists and technologists say the definition of "computer" is shifting as the underlying technology converges and accelerates. "We're making an architectural change, not just a technology change," says Lawrence Livermore National Laboratory's James Brase. "The new kinds of capabilities--it won't be a linear scale--this will be a major leap." The change in architecture refers to projects to assemble a computer with human brain-like action and learning capabilities, with a concentration on enabling pattern recognition and far more processing power. Argonne National Laboratory director Peter Littlewood argues there will be no single computing model, with changes forthcoming in the area of data integration as well as architecture, as a result of modeling next-generation systems on the brain's networked mechanisms. "We're now on the verge of being able to map the brain at a scale where you can see a synapse," Littlewood notes. "And if you map the brain down to the scale of synapse, and you take all of that data, that's about a zettabyte of data." IBM Cognitive Computing Group founder Dharmendra Modha, recipient of the ACM Gordon Bell Prize in 2009, foresees computers becoming increasingly adaptable to humans, instead of vice-versa, with the emergence of brain-inspired computing and its integration into modern computing infrastructures.

Siting Wind Farms More Quickly, Cheaply
MIT News (07/17/15) Larry Hardesty

Massachusetts Institute of Technology (MIT) researchers say they have developed a statistical technique that yields better wind-speed predictions than existing techniques. The researchers say their breakthrough could save power companies time and money, especially in the evaluation of sites for offshore wind farms, where maintaining measurement stations is particularly costly. The first novelty of the technique is that it can factor in data from up to 15 or more weather stations, in some cases. In addition, it is not restricted to Gaussian probability distributions, also known as bell curves. The new model also can use different types of distributions to characterize data from different sites, and it can combine them in multiple ways. Another aspect of the new model is its ability to use nonparametric distributions, and to find nonlinear correlations between data sets. The researchers applied the new model to data collected from an anemometer on top of the MIT Museum. The researchers used three months worth of wind data to predict wind speeds over the next two years three times as accurately as existing models could with eight months of data. "This methodology has strong practical value, and I am convinced that it will be applicable to many other real-life problems," says Nanyang Technological University professor Justin Dauwels.

Seales' Research Team Reveals Biblical Text From Damaged Scroll
University of Kentucky News (07/20/15) Whitney Harder

Researchers at the University of Kentucky (UK), in collaboration with colleagues in Israel, have used software developed at the university to "virtually unwrap" and examine the contents of a damaged scroll that is at least 1,500 years old. The scroll was discovered in 1970 as part of an archeological excavation of the synagogue in Ein Gedi, Israel, but it had been damaged by fire at some point and could not be unwrapped. A UK team led by professor Brent Seales used software developed at the university to analyze micro-computed tomography scans of the scroll performed in Israel. The software first enhances the details of the scan so it can tell which areas have ink on them and which do not, then uses the presence of writing to generate a three-dimensional (3D) surface within the volume of the burnt scroll. The surface is then rendered as a high-quality 3D surface with the writing as part of its texture, producing a flattened, and in this case legible, image of the scroll as it would look when unrolled. When they virtually unrolled the Ein Gedi scroll, the researchers found it contained passages from the Book of Leviticus.

Neuroscience-Based Algorithms Make for Better Networks
CMU News (07/16/15) Byron Spice

Researchers at Carnegie Mellon University (CMU) and the Salk Institute for Biological Studies say they have, for the first time, determined the rate at which the developing brain eliminates unneeded connections between neurons during early childhood, which they say could be used to improve the robustness and efficiency of distributed computational networks. The researchers created an algorithm for designing computational networks based on the brain-pruning approach observed in a mouse model's somatosensory cortex over time. The researchers used simulations and theoretical analysis to determine the neuroscience-based algorithm produced networks that are much more efficient and robust than the current engineering methods. In the brain pruning-based networks, the flow of information was more direct and provided multiple paths for information to reach the same endpoint, which minimizes the risk of network failure. "It turns out that this neuroscience-based approach could offer something new for computer scientists and engineers to think about as they build networks," says CMU professor Alison Barth. The researchers applied the algorithm to flight data from the U.S. Department of Transportation, and found the algorithm created the most efficient and robust routes to enable passengers to reach their destinations. "There's a lot that the brain can teach us about computing, and a lot that computer science can do to help us understand how neural networks function," Barth says.

White House Wants Agencies to Prioritize Emerging Tech in Next Year's Budget (07/17/15) Mohana Ravindranath

The White House recently published a memorandum that directs U.S. agencies to prioritize emerging technology and big data in the fiscal year 2017 budget. Agencies should "prioritize investments in enabling technologies that benefit multiple sectors of the economy, such as nanotechnology, robotics, the Materials Genome Initiative, and cyber-physical systems and their application to smart cities," the memo says. General topics include "advanced manufacturing and industries of the future" and "information technology and high-performance computing." The White House sees big data analytics as potentially helping improve national security. The document also says agencies should focus on programs that "foster innovation," such as Grand Challenges with prize rewards, and collaboration with the Maker Movement. Small Business Innovation Research programs and Small Business Technology Transfer awards should bolster agency technology priorities. Moreover, the memo says budget requests to the Office of Management and Budget should prioritize investments that could help improve access to federal data.

Camp Gives Middle School Girls Hands-On Experience in Engineering
UC Berkeley NewsCenter (07/16/15) Sarah Yang

The University of California, Berkeley's Girls in Engineering summer camps are part of a pilot program launched last year that is designed to narrow the gender gap in science, technology, engineering, and math (STEM) fields. Instructors are professors, postdoctoral researchers, and graduate and undergraduate students, covering topics ranging from nanotechnology to data science. Several studies have shown many girls start to lose interest in STEM fields during middle school. "Our goal is to keep the girls from losing interest, to keep the momentum going," says Berkeley professor Claire Tomlin. The Berkeley camps also emphasize the need for soft skills, such as communications and presentation proficiencies. At the beginning of each session, girls are grouped into teams of five and asked to identify a problem and discuss ways to solve it. On the last day of camp, each team gives a presentation in front of camp staff and family members. "We do show them an academic perspective, but they also need to see the industry side of engineering, which is why we arranged field trips to local tech companies," Tomlin says. Camp organizers hope the effort will lead to greater retention of women in the STEM pipeline.

After 85-Year Search, Massless Particle with Promise for Next-Generation Electronics Discovered
Princeton University (07/16/15) Morgan Kelly

Princeton University researchers have confirmed the existence of Weyl fermions, which were first theorized in 1929. Weyl fermions could allow for a nearly free and efficient flow of electricity in electronics, and thus greater power, in next-generation computers. Unlike electrons, Weyl fermions are massless and possess a high degree of mobility, and the particle's spin is both in the same direction as its motion, known as being right-handed, and in the opposite direction from which it moves, known as left-handed. The researchers also found Weyl fermions can be used to create massless electrons that move very quickly with no backscattering. "These are very fast electrons that behave like unidirectional light beams and can be used for new types of quantum computing," says Princeton professor M. Zahid Hasan. The researchers had previously theorized Weyl fermions could exist in a tantalum arsenide crystal. Such crystals were loaded into a scanning tunneling spectromicroscope cooled to near absolute zero and suspended from the ceiling to prevent even atom-sized vibrations. The spectromicroscope determined the crystal matched the theoretical specifications for hosting a Weyl fermion. "This work really shows why research is so fascinating, because it involved both rational, logical thinking, and also sparks and inspiration," says Princeton researcher Su-Yang Xu.

Web 2.0 (and Beyond): Developing the Next Generation of Connectivity
Government Technology (07/17/15) Colin Wood

The U.S. National Science Foundation's Global Environment for Network Innovations (GENI), a network of more than 50 sites in more than 30 countries, was established in 2007 to test next-generation networking concepts without being limited by the Internet. A recent partnership between GENI and non-profit US Ignite hosted an event demonstrating how a low-latency fiber network could enable unprecedented functionality for public safety officials, such as a simulated vehicle crash that activated an automatic notification to a police station, while a drone shot video at the crash site for information-gathering purposes. Software-defined networks (SDN) partly drive GENI's innovation. "SDN is the notion that you separate the data and the control into separate channels, so that way when you want to have a connection to a healthcare provider, the SDN checks [that the user has adequate clearance to do this] and then it creates what's called a flow, keeping it separate from other traffic," notes US Ignite's Glenn Ricart. More security is one benefit of the segregation offered by SDN, which also ameliorates bandwidth constraints and lowers latency via cross-network data distribution. GENI products such as the OpenFlow open SDN standard are migrating to the regular Internet, while GENI project director Mark Berman says the network also aims to deliver detailed data flow control and intelligent software that accommodates fast-moving mobile devices.

CMU Professor at Forefront of Machine Learning Research
The Pittsburgh Post-Gazette (07/16/15) Mark Roth

Carnegie Mellon University (CMU) professor Tom Mitchell is a pioneer in the field of machine learning, which is having a sizable impact on numerous aspects of people's everyday lives, ranging from credit card transaction validation to email to personalized ads and news posts on social media. Such programs no longer operate on the basis of simple rules, but continuously learn from real-life experiences and revise their operations over time. Mitchell notes in an interview many commercial programs being used today are instances of supervised machine learning, in which the programs train themselves on a database with known results, and then apply that knowledge to support a model of what is happening with new data. Among the advantages of machine-learning programs over groups of people seeking the same patterns is the lack of any biases in their work, according to Siftscience software engineer Doug Beeferman. Mitchell and CMU psychology professor Marcel Just have used unsupervised machine-learning software to seek out patterns in data it has never encountered before, in order to predict words or phrases people are thinking of by studying their brain activation patterns in a functional magnetic resonance imaging machine. Mitchell projects future efforts will focus on "never-ending learning programs" that might run for years at a time.

Abstract News © Copyright 2015 INFORMATION, INC.
Powered by 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