Association for Computing Machinery
Welcome to the November 23, 2016 edition of ACM TechNews, providing timely information for IT professionals three times a week.

Please note: In observance of the U.S. Thanksgiving holiday, TechNews will not publish Friday, Nov. 25. Publication will resume Monday, Nov. 28.

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).


Neural Network Learns to Identify Criminals by Their Faces
Technology Review (11/22/16)

Researchers from China's Shanghai Jiao Tong University applied a variety of machine-vision algorithms to study faces of criminals and noncriminals. They took still photos of 1,856 Chinese men aged 18 to 55 with no facial hair, half of whom were criminals. Investigators then used 90 percent of these images to train a convolutional neural network to recognize the difference and tested the neural network on the remaining 10 percent of the images. The neural network was able to correctly identify criminals and non criminals with an accuracy of nearly 90 percent. "These highly consistent results are evidences for the validity of automated face-induced inference on criminality, despite the historical controversy surrounding the topic," according to the researchers. The neural network uses three distinct facial features to make its classification: the curvature of the upper lip, which is on average 23 percent larger for criminals than for noncriminals; the distance between the inner corners of the eyes, which is 6 percent shorter in criminals; and the angle between two lines drawn from the tip of the nose to the corners of the mouth, which is 20 percent smaller in criminals. The researchers also found that the data for criminal faces has much greater variance than the data for noncriminals. "In other words, the faces of general law-biding public have a greater degree of resemblance compared with the faces of criminals, or criminals have a higher degree of dissimilarity in facial appearance than normal people," the researchers conclude.

AI Academic Warns on Brain Drain to Tech Groups
Financial Times (11/21/16) Richard Waters

As artificial intelligence (AI) professionals increasingly seek positions within major technology companies, the exodus threatens to stall academic research in the field, according to Yoshua Bengio. The University of Montreal professor helped to pioneer deep learning, a method that models how the human brain functions and has spurred advances in language understanding and image recognition by computers. Bengio helped develop deep learning technology with Geoff Hinton and Yann LeCun, both former academics who have since moved on to work for Google and Facebook, respectively. Another early deep learning expert, Andrew Ng, formerly of Stanford University, now is employed by Chinese Internet search company Baidu. Bengio says he chose to remain in academia to gain broader impact for his work. Despite the brain drain of AI talent away from academia, the city of Montreal has maintained its place as the center of AI research with 1,500 researchers working in the field, thanks to Bengio's work at the University of Montreal and that of researchers at McGill University. Now, Google is planning to set up a deep learning research center in Montreal to capitalize on studies being done in that area. "People don't realize the way deep learning is working right now is capturing very superficial aspects of our world," Bengio says.

Spray Printed Crystals to Move Forward Organic Electronic Applications
University of Surrey (11/22/16) Ashley Lovell

Researchers from the University of Surrey in the U.K. have demonstrated for the first time a low-cost scalable spray-printing process to fabricate high-quality isolated organic single crystals. The researchers say the method could be applied to a wide variety of semiconducting small molecules, which can be dissolved in solvents to make semiconducting inks, and then be deposited on virtually any substrate. The technology combines the advantages of antisolvent crystallization and solution shearing. The crystals' size, shape, and orientation are controlled by the spray angle and distance to the substrate, which govern the spray droplets' impact onto the antisolvent's surface. "This method is a powerful, new approach for manufacturing organic semiconductor single crystals and controlling their shape and dimensions," says University of Surrey researcher Maxim Shkunov. The researchers note their technique is much simpler than other methods, and can be performed entirely at room temperature with an off-the-shelf artist's spray brush. "With a new class of organic semiconductors based on carbon atoms, we can spray-coat organic inks onto anything, and get more or less the right size of crystals for our devices right away," says National Physical Laboratory researcher Grigorios Rigas.

Mapping Migrations by Using Mobile Phone Data
Technical University of Madrid (Spain) (11/21/16)

Researchers from the Complex Systems Group at the Technical University of Madrid (UPM) in Spain have traced migratory patterns within the population of Senegal by using mobile phone data. The study aimed to detect and characterize socioeconomic events through the analysis of mass migration, which could help officials respond to unusual events, such as natural disasters, and better allocate resources to those incidents. The study took into account cultural and geographic context, such as Senegal's predominantly agriculture-based economy and its reliance on seasonal workers. Senegal's majority religion, Islam, also could facilitate travel related to religious holidays and festivals. The UPM researchers collected phone data generated by 9 million users in Senegal and were able to match phone data with the movement of seasonal workers during the country's harvest season. Religious festivals, such as the birth of Muhammad, also reflected significant migration patterns. The researchers say the study could lead to solutions that will enable the real-time monitoring of communication and mobility patterns in order to detect emergencies.

Malware that Turns PCs into Eavesdropping Devices Demonstrated by Ben-Gurion U.
Ben-Gurion University of the Negev (Israel) (11/22/16) Lavin Andrew

Researchers at Ben-Gurion University of the Negev (BGU) in Israel have demonstrated SPEAKE(a)R, malware that can turn computers into perpetual eavesdropping machines. Using SPEAKE(a)R, malware can secretly transform headphones into a pair of microphones, according to the researchers. "The fact that headphones, earphones and speakers are physically built like microphones and that an audio port's role in the PC can be reprogrammed from output to input creates a vulnerability that can be abused by hackers," says BGU professor Yuval Elovici. The malware can covertly reconfigure the headphone jack from a line-out jack to a microphone jack, making the connected headphones function as a pair of recording microphones and transforming the computer into an eavesdropping device. This technique works even when the computer does not have a connected microphone. The researchers studied several attack scenarios to evaluate the signal quality of simple off-the-shelf headphones. "We demonstrated it is possible to acquire intelligible audio through earphones up to several meters away," says BGU researcher Yosef Solewicz. Software-based countermeasures could include completely disabling audio hardware, using an HD audio driver to alert users when microphones are being accessed, or developing and enforcing a strict rejacking policy within the industry, according to the researchers. Anti-malware and intrusion detection systems could also be developed to monitor and detect unauthorized speaker-to-microphone retasking operations and block them.

Single Photon Converter - A Key Component of Quantum Internet
University of Warsaw (11/22/16)

Physicists from the University of Warsaw in Poland and the University of Oxford in the U.K., have constructed an electro-optical device capable of modifying the quantum properties of individual photons. In recent years, physicists have developed ways to generate light pulses with a specific wavelength or polarization to control the quantum bits used in quantum computing. Existing methods of modifying the properties of individual photons utilize nonlinear optical techniques, exposing photons to a strong optical pump beam. Whether the individual photons were modified remained a matter of chance. In constructing the new device, researchers used the electro-optic effect occurring in certain crystals, which alters the intensity of an external magnetic force. The researchers were able to achieve a six-fold lengthening of the duration of a single-photon pulse while preserving a high level of conversion efficiency. Previous converters were only able to modify one among tens of photons, whereas the new device has demonstrated efficiency up to 200 times better. "In essence we process every photon entering the crystal," says Michal Jachura, a PhD student at the University of Warsaw. "The efficiency is less than 100% not because of the physics of the phenomenon, but on account of hard-to-avoid losses of a purely technical nature, appearing for instance when light enters or exits optical fibers." In addition to being efficient, the device is also stable, compact, and easy to install in an optical fiber system. Scientists say the new converter should help facilitate the development of hybrid quantum computers.

U.S. Sets Plan to Build Two Exascale Supercomputers
Computerworld (11/21/16) Patrick Thibodeau

U.S. Department of Energy (DoE) officials say they expect to seek vendor proposals to build two exascale supercomputers by 2019. The systems, which will cost about $200 million to $300 million each, will be built at the same time and will be ready for use by 2023, according to the DoE. However, the officials caution that President-elect Donald Trump, who did not address the topic of supercomputing during the campaign, could change directions. Development work has begun on the software and other technologies that will utilize hundreds of millions of simultaneous parallel events. The systems will likely need their own power plant--the U.S. wants an exascale system that can operate on 20 megawatts but no more than 30 megawatts. China could have an exascale system ready by 2020, but the real test will be their usefulness. The U.S. approach is to develop an exascale eco-system involving vendors, universities, and the government, with the expectation that a wide range of relatively easy-to-program applications will be ready so the systems can be put to immediate and broad use.

A New Standard in Robotics
YaleNews (11/17/16) William Weir

A team of researchers are distributing kits of objects and tasks to laboratories specializing in robotic manipulation to standardize manipulation tests and establish universal benchmarks. As increasingly sophisticated robots are developed for a range of applications, these systems require the work of multiple disciplines. Robotics laboratories previously worked by their own standards and protocols, making collaboration between laboratories difficult. The standardized kits, known as the Yale-Carnegie Mellon-Berkeley (YCB) Object and Model Sets, contain common, household items that are inexpensive and durable, such as a hammer, a water pitcher, and a can of Spam. Some objects have simple shapes that are easy for a robot to grasp, while other objects provide more of a challenge for robotic hands. Researchers can use these objects to measure the grasping and manipulation abilities of their robotic arms and share their results with other researchers. "When we have a new idea for a new component or hand, we want to test it out and see how well it works," says Yale professor Aaron Dollar. "With quantitative evaluation, we can see how things stack up compared to other ideas." The YCB program also provides examples of manipulation tasks and benchmarks and allows other laboratories to contribute their own protocols.

More Than Animation: Software Supports Animated Storytelling
EurekAlert (11/17/16) Jennifer Liu

Disney Research has developed CANVAS, a computer-assisted tool for creating narratives, and Story World Builder, a graphical platform in which people can create "story worlds" populated with characters and props. The new tools are designed to eliminate the distracting details encountered when telling animated stories, which can hamper creativity. With CANVAS, users can synthesize a three-dimensional animation to get feedback on the narrative at any point, and it also can work in the background to fill holes in the plot to ensure the story makes sense. Researchers presented CANVAS this summer at the ACM SIGGRAPH/Eurographics Symposium on Computer Animation in Zurich, Switzerland. The researchers say Story World Builder simplifies the task of creating story worlds by leading the user through the process of defining characters and objects and of describing how events within those worlds unfold. Professionals as well as casual users could take advantage of the tools. "Our aim is to produce technologies to enhance the creative process," says Disney researcher Steven Poulakos. "We embrace several principles, including iterative design, reuse to minimize the cost of content production, and computer assistance to reduce complexity for creating story worlds and authoring stories."

Tech Would Use Drones and Insect Biobots to Map Disaster Areas
NCSU News (11/17/16) Matt Shipman

Unmanned aerial vehicles (UAVs) and insect cyborgs, or biobots, could use a combination of software and hardware developed by researchers at North Carolina State University (NCSU) to map large, unfamiliar areas, such as collapsed buildings, after a disaster. The researchers plan to restrict movement of the biobots to a defined area using remote-control technology. Custom software would use an algorithm to translate the biobot sensor data into a rough map of the unknown environment. Once the program receives enough data to map the defined area, the UAV moves forward to hover over an adjacent, unexplored section. The biobots move with it, and the technology repeats the mapping process. The software stitches the new map to the previous one; the process can be repeated until the entire region or structure has been mapped. "It would be of much more practical use for helping to locate survivors after a disaster, finding a safe way to reach survivors, or for helping responders determine how structurally safe a building may be," says NCSU's Edgar Lobaton, who has authored two papers on the research. "The next step is to replicate these experiments using biobots, which we're excited about."

Researchers Analyze Gender Gaps in STEM Employment
The Daily Texan (11/18/2016) Freya Preimesberger

Researchers at the University of Texas, Cornell University, and Syracuse University recently analyzed a U.S. Department of Commerce report that says less than 25 percent of science, technology, engineering, and math (STEM) jobs are held by women. The study found that although many women study STEM fields, they disproportionately study fields such as life sciences, which provide fewer jobs than fields such as computer science and engineering. The researchers also analyzed data from a survey by the U.S. Bureau of Labor Statistics that focused on men and women born between 1957 and 1964, which includes women who came of age during a period when they graduated from college at a higher rate than men for the first time. The study found 53 percent of men who received a bachelor's degree in STEM found a job in a STEM field within two years of graduation, compared to 41 percent of women. The bulk of this gap can be attributed to the underrepresentation of women in fields that most often lead to jobs in STEM. "There's all sorts of things as to why that could be the case--there's talk of tipping points where, once women make up a certain threshold, it signals to other women that it's a major they can also succeed in," says Syracuse University professor Katherine Michelmore.

Miniature Wi-Fi Device Developed by Stanford Engineers Supplies Missing Link for the Internet of Things
Stanford News (11/16/16) Andrew Myers

Stanford University researchers have developed HitchHike, a tiny, ultra-low-energy wireless radio that enables data transmission using just micro-watts of energy. HitchHike "can be used as-is with existing Wi-Fi without modification or additional equipment," and consumers can use it today with a cell phone and an off-the-shelf Wi-Fi router, according to Stanford researcher Pengyu Zheng. HitchHike requires so little power a small battery could drive it for a decade or more, and it has the potential to harvest energy from existing radio waves and use that electromagnetic energy to power itself, potentially indefinitely. "HitchHike could lead to widespread adoption in the Internet of Things," says Stanford professor Sachin Katti. HitchHike is a variation on a backscatter radio. The system bounces Wi-Fi signals back into the atmosphere, a signal known as backscatter. In order to function as a true radio, HitchHike must produce its own messages, rather than reflect existing messages. To do that, the Stanford researchers developed "code word translation." HitchHike shifts its new signal to another Wi-Fi channel, thus avoiding the radio interference between the original signal and the new data stream. "HitchHike opens the doors for widespread deployment of low-power Wi-Fi communication using widely available Wi-Fi infrastructure and, for the first time, truly empower the Internet of Things," Zheng says.

Researchers Have a Better Way to Predict Flight Delays
Inside Binghamton University (11/15/16)

Binghamton University researchers say they have developed a computer model that can more accurately predict airline delays faster than other conventional systems. "Our proposed method is better suited to analyze datasets with categorical variables (qualitative variables such as weather or security risks instead of numerical ones) related to flight delays," says Binghamton researcher Sina Khanmohammadi. Flight delays currently are predicted by artificial neural network (ANN) computer models that are backfilled with delay data from previous flights. The Binghamton researchers introduced a new multilevel input layer ANN to handle categorical variables with a simple structure to help airlines more easily see the relationships between input variables and outputs. The research will help airlines inform travelers more quickly and more accurately about problems. "Air traffic controllers at a busy airport can also use this information as a supplement to improve the management the of airport traffic," Khanmohammadi says. The researchers trained the new model to identify 14 different variables that affected arrival times for 1,099 flights from 53 different airports to John F. Kennedy International Airport in New York City. The new model predicted the length of delays with about 20 percent more accuracy than traditional models and required about 40 percent less time to come to those conclusions.
Return to Headlines | Share Facebook  LinkedIn  Twitter 

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

To submit feedback about ACM TechNews, contact: [email protected]