Association for Computing Machinery
Welcome to the November 16, 2016 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).


U.S., China Vie for Supercomputing Supremacy
HPC Wire (11/14/16) Tiffany Trader

The 48th edition of the TOP500 supercomputing list shows a clear rivalry between China and the U.S. for supremacy, with each country boasting 171 systems on the list. The previous edition had China dethroning the U.S. in terms of system share and performance, but now the U.S. has a narrow lead in total performance of 33.9 percent, versus China's 33.3 percent. Chinese machines--the TaihuLight and Tianhe-2--occupy first and second place on the current TOP500. The 93-petaflops TaihuLight has a more than fivefold FLOPS (FLoating-point Operations Per Second) lead over that of the 17.6-petaflops Titan, the fastest U.S. supercomputer. The U.S. is planning to revitalize its supercomputing sector for 2018-2019 with the CORAL systems coming online, but China will not remain idle. China says it is not only deploying the Wuxi supercomputer, but also planning to upgrade one or two more major systems into the 100-petaflops realm and constructing three prototype systems ramping up toward their 2020 exascale goal. The collective performance of all TOP500 supercomputers currently amounts to 672 petaflops, a 60-percent increase from a year ago. A year from now, total aggregate performance should exceed 1,000 petaflops, or 1 exaflops, if the growth rate remains above 50 percent.

Machine-Learning Algorithm Can Show Whether State Secrets Are Properly Classified
Technology Review (11/14/16)

Machine-learning algorithms can be used to determine how information from intelligence agencies should be classified, according to researchers from Brazilian think tank Fundacao Getulio Vargas and Columbia University. Using a variety of machine-learning approaches, the researchers studied more than 1 million declassified U.S. State Department cables from the 1970s. Each cable was a text message exchanged between the State Department and a diplomatic mission, with cables labeled as secret, confidential, limited official use, or unclassified. Messages also contained a date, a sender and receiver, and a subject. The researchers determined how cable information correlated with the classification label and used an algorithm to predict whether a cable was classified. The relative frequency of certain words in the message text was the best indicator of classification. Sender and recipient data also could be used to identify sensitive information, but this method was more likely to produce false positives. When the algorithm combined the different metadata in its analysis, it could identify 90 percent of classified cables.

Why the Software Industry Needs Computing Education Research
The Huffington Post (11/14/16) Andrew J. Ko; Susanne Hambrusch

Job opportunities for software development abound, but schools are producing coders of less-than-adequate skill, according to University of Washington professor Andrew J. Ko and Purdue University professor Susanne Hambrusch. They cite a 2004 study that determined after passing college-level introductory programming courses, most students could not predict the output of even basic computer programs. Research on coding boot camps uncovered similar trends, with students failing to learn or gain employment. "Why are students failing to learn?" Ko and Hambrusch ask. "Our best evidence shows that learning to code is hard, and teaching coding effectively is even harder." In addition, software jobs require other skills besides coding, such as problem solving, design, and communication. "Not only do we not know how to teach these skills well--we don't know how to teach them across K-12, college, or post-college levels," Ko and Hambrusch note. They say greater investment in computing education research is needed to solve this problem. "If we can build a strong evidence base for how to learn and teach computing, we will not only reinvent how students in K-12 schools and colleges learn, but we'll also improve how adults retrain and effectively learn new skills," Ko and Hambrusch say.

Cybercriminals Can Steal All Your Passwords From Thin Air With This $25 Wi-Fi Hack
International Business Times (United Kingdom) (11/14/16) Mary-Ann Russon

Researchers from the University of Massachusetts Boston, the University of South Florida, and China's Shanghai Jaio Tong University have developed WindTalker, a system that can analyze Wi-Fi networks and covertly detect and record passwords by looking at the directions radio waves travel to provide wireless Internet coverage. The antennas of modern Internet routers make small changes to strengthen radio signals going in some directions, while signals going in other directions are canceled out. Since the routers are designed to detect and manage very small changes to the signals to ensure that devices are always receiving the highest-quality signal, this technology can be exploited to track a user's hand as it swipes and types. The researchers deployed a malicious public Wi-Fi access point using $20 antennas, the attacker's laptop, and a $5 networking card. The setup was one meter from a target sitting at a table with a smartphone. After the user connected to the free Wi-Fi, WindTalker extracted sensitive data by analyzing and processing the radio signals to separate the parts of the signal it needed. The researchers found WindTalker accurately monitored and detected passwords commonly used by banks and payment apps with 81.7-percent accuracy.

Defibrillator-Equipped Drones Could Be First on Scene in Cardiac Arrest
CBC News (Canada) (11/15/16) Blair Bigham

Timothy Chan at the University of Toronto in Canada wants to use drone technology to help people who suffer cardiac arrest. Chan used computer models to determine that strategically placed drones with defibrillators could beat ambulances to the scene by several minutes and, in some cases, cut response times in half, increasing the chances of survival. Ambulance response times average 5 to 10 minutes in cities and often more than 20 minutes in rural communities. Chan and his team studied historical ambulance response times to 56,000 cardiac arrests in southern Ontario, Canada, over a nine-year period. They then applied an algorithm to determine where drones would have to be placed to arrive faster than 911 responders. The researchers examined the impact of a network of 81 drone bases with 100 drones in the eight municipalities in the Greater Toronto Area, and found it would cut the time it takes for ambulances to arrive by more than half in 90 percent of cardiac arrests. In addition, rural regions would experience a reduction in average response times from 19 minutes to 9 minutes in 90 percent of cases, and urban centers would see response times drop from more than 10 minutes to less than 4 minutes.

Enabling Wireless Virtual Reality
MIT News (11/14/16) Adam Conner-Simons

Massachusetts Institute of Technology (MIT) researchers have developed MoVR, a prototype wireless virtual reality (VR) system for gamers. MoVR can enable untethered communication at a rate of multiple Gbps (billions of bits per second) using special high-frequency radio signals called "millimeter waves" (mmWaves). "The ability to use a cordless headset really deepens the immersive experience of virtual reality and opens up a range of other applications," says MIT professor Dina Katabi, who received the ACM Grace Murray Hopper Award in 2012. However, to deliver constant connectivity for a VR system, the mmWaves would need to always have a constant line of sight between transmitter and receiver. The researchers overcame this challenge by developing MoVR to act as a programmable mirror that detects the direction of the incoming mmWave signal and reconfigures itself to reflect it toward the receiver on the headset. MoVR can learn the correct signal direction to within two degrees, enabling it to correctly configure its angles. "With MoVR, angles can be specifically programmed so that the mirror receives the signal from the mmWave transmitter and reflects it towards the headset, regardless of its actual direction," says MIT researcher Omid Abari. The researchers presented their findings last week at the ACM Workshop on Hot Topics in Networks (HotNets 2016) in Atlanta, GA.

Study: Interest in STEM Fuels Growth in Number of International Students in U.S.
U.S. News & World Report (11/14/16) Kelly Mae Ross

The number of international students in the U.S. topped 1 million for the first time in a single academic year, according to the 2016 Open Doors Report on International Educational Exchange. In 2015-2016, the number of international students pursuing undergraduate and graduate degrees, non-degree studies, and practical training rose 7.1 percent to surpass that mark, according to the annual survey from the Institute of International Education in partnership with the U.S. Department of State's Bureau of Educational and Cultural Affairs. Subjects related to science, technology, engineering, and mathematics (STEM) were popular areas of study for international students. More than a third were studying engineering, math, and computer science in 2015-2016. There were 216,932 students studying and training in engineering, up 10.3 percent from the previous year, and 141,651 students in math and computer science fields, a 25.4-percent year-over-year increase. Engineering overtook business management to become the most popular field of study. China led the way with students in STEM fields, followed by India, Saudi Arabia, South Korea, and Canada. Over the last decade, the overall number of international students in the U.S. has grown almost 85 percent.
View Full Article - May Require Paid Subscription | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Quantum Computers Can Talk to Each Other via a Photon Translator
New Scientist (11/15/16) Sophia Chen

Researchers from Germany's University of Paderborn have developed a system that can convert photons from one spread of frequencies to another, while still preserving their delicate quantum state. The new system includes a converter that "translates" photons emitted from one component into the infrared region of the spectrum. The infrared photon is transmitted over a fiber-optic cable connected to a second component, and then translated into another frequency the receiving component can read. Although the researchers have only converted infrared photons into visible wavelengths--while leaving their quantum state intact--with a success rate of about 75 percent, the technique could be adapted to build the full system, says Paderborn's Christine Silberhorn. After the researchers develop the full system, the next step would be to determine how to fit the device on a chip that could be manufactured in large quantities. The science works, but "scalability is the biggest problem," says University of Washington professor Arka Majumdar. "Making the same device 1,000 times is extremely difficult."

Mathematical Algorithms Calculate Social Behavior
Technical University of Munich (Germany) (11/11/16)

Predicting and influencing human behavior is well on its way to becoming reality, according to researchers at the Technical University of Munich (TU Munich) in Germany. They have proved mathematical statements that demonstrate how surprisingly easy it is to automatically generate precise models for specific, relatively simple group interactions based on observed dynamics data. Using computer simulations, the researchers can describe potential collective behavioral patterns of a large number of individuals who mutually influence each other in a given situation. The team is now developing appropriate tools that could be used to predict the behavior of groups and influence them. The researchers say the tools could be used to simulate and improve security at major events or increase the efficiency of evacuating people in emergency situations. They note forecasts work best with groups that show generalized patterns of behavior, which helps protect against the abuse of such mathematical modeling. "The good news in this context is that we have also proven that behavior is not so easy to predict or control for all kinds of dynamics and situations," says Massimo Fornasier, chair of TU Munich's Department of Applied Numerical Analysis.

New 'Bottlebrush' Electroactive Polymers Make Dielectric Elastomers Increasingly Viable for Use in Devices
NCSU News (11/09/16)

A new electroactive polymer material capable of changing its size and shape when exposed to a relatively small electric field has been developed by a multi-institutional team of researchers. The new material could lead the way to a wide range of new applications, from microrobotics to designer haptic, optic, microfluidic, and wearable technologies. The new material, a "bottlebrush" silicone elastomer, overcomes hurdles that have prevented the wider use of electroactive polymers in commercial devices. Previous dielectric elastomers required large electric fields to trigger movement, whereas the new material has a much lower actuation level. The material's polymers are prepared by grafting long polymeric side chains to a polymer backbone, forming molecules in thick, flexible filaments that enable the material to be stretched. Molecules prepared with different degrees of polymerization and grafting densities create materials with varying mechanical properties. "This architectural control of mechanical properties has reduced the limit of stiffness in dry polymer materials by 1,000 times, demonstrated extensibility of up to eight times, and opened up new applications not available to stiffer materials or materials with liquid fractions," says University of North Carolina professor Sergei S. Sheiko. "We're at the earliest stages of identifying all the potential ways in which we could use this new class of material."

Semiconductor-Free Microelectronics Are Now Possible, Thanks to Metamaterials
University of California, San Diego (11/07/16) Liezel Labios

University of California, San Diego (UCSD) researchers say they have developed the first semiconductor-free, optically controlled microelectronic device built from metamaterials. The device exhibits a 1,000-percent increase in conductivity when activated by low voltage and a low power laser. The researchers sought to remove hindrances associated with existing microelectronic devices by replacing semiconductors with free electrons in space, but liberating electrons from materials requires either applying high voltages, high power lasers, or extremely high temperatures, which are impractical in micro- and nanoscale electronic devices. The researchers say they fabricated a microscale device that can release electrons from a material without such extreme requirements. The device consists of a metasurface on top of a silicon wafer, with a layer of silicon dioxide in between. The metasurface is comprised of an array of gold nanostructures on an array of parallel gold strips, so when low DC voltage and a low power infrared laser are both applied, the metasurface generates "hot spots" that provide enough energy to pull electrons out from the metal and liberate them into space. "This...may be the best approach for certain specialty applications, such as very high frequencies or high-power devices," says UCSD professor Dan Sievenpiper.

Unlocking Big Genetic Datasets
Columbia University (11/07/16) Kim Martineau

Algorithms used to personalize movie recommendations and extract topics from vast volumes of text could bring doctors closer to diagnosing, treating, and preventing disease on the basis of an individual's unique genetic profile, according to a new study. Researchers from Columbia and Princeton universities describe a new machine-learning algorithm that scans massive genetic datasets to infer an individual's ancestral makeup, which is key to identifying disease-carrying genetic mutations. Called TeraStructure, the new algorithm builds on the widely used STRUCTURE algorithm, which refines its model after repeated passes through a dataset. However, TeraStructure is designed to update the model as it goes. "You don't have to painstakingly go through all the points each time to update your model," says Columbia professor David Blei, recipient of the ACM-Infosys Foundation Award in the Computing Sciences (now the ACM Prize in Computing) in 2013. When the team ran TeraStructure on a simulated data set of 10,000 genomes, it was more accurate and two to three times faster than current state-of-the-art algorithms. Blei estimates TeraStructure could analyze 1 million individuals, which he says is orders of magnitude beyond modern software capabilities. "I think these results will motivate future applications of this kind of algorithm in challenging inferences problems," says Matthew Stephens, a genetics researcher at the University of Chicago who helped develop the STRUCTURE algorithm.

High-Performance Computing: A Look at What's Next With TACC's Dan Stanzione
ZDNet (11/14/16) Larry Dignan

In an interview, Dan Stanzione, executive director of the Texas Advanced Computing Center (TACC) at the University of Texas at Austin, discusses the state of high-performance computing (HPC), sustainability, consumption models, and use cases as HPC becomes mainstream. Stanzione notes TACC's Hikari project aims to improve data center sustainability by eliminating AC-to-DC power conversions via native DC generation. Enterprise penetration of such equipment depends on more server vendors offering DC systems and agreeing on a standard. Stanzione also says time to solution has been a key metric and the basic driver of HPC--and it has gone mainstream. "With hardware and high-end chips it's getting harder to measure peak performance," he says. "What you're really trying to get to is the capability." In terms of HPC market trends, Stanzione notes the biggest shift is a move away from traditional industrial users--such as oil and gas, aerospace, and automotive firms--to more specialized sectors such as pharmaceuticals, life sciences, and corporate analytics departments that need more number-crunching capability. Stanzione says HPC consumption should range from cloud use for "big collections for small problems" to specialized providers and national labs for larger initiatives.

Abstract News © Copyright 2016 INFORMATION, INC.
Powered by Information, Inc.

To submit feedback about ACM TechNews, contact: