Welcome to the May 23, 2014 edition of ACM TechNews, providing timely information for IT professionals three times a week.
Please note: In observance of the Memorial Day holiday, TechNews will not be published on Monday, May 26. Publication will resume Wednesday, May 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).
HEADLINES AT A GLANCE
Skilled Foreign Workers a Boon to Pay, Study Finds
The Wall Street Journal (05/22/14) Josh Zumbrun; Matt Stiles
Three academic economists recently conducted a study examining wage data and immigration in 219 U.S. metropolitan areas from 1990 to 2010 and found that cities seeing the biggest gains of foreign-born workers in science, technology, engineering, and mathematics (STEM) saw wages climb fastest for the native-born, college-educated population. The study concludes that immigrants can boost the productivity of the overall economy, and job creation leads to more jobs for non-immigrants as well. The economists focused on how wages for college- and noncollege-educated native workers shifted along with immigration. The study found a 1 percent increase in the share of workers in STEM fields could raise wages for college-educated natives by 7 to 8 percent and wages of the noncollege-educated natives by 3 to 4 percent. "A lot of people have the idea there is a fixed number of jobs," says University of California, Davis researcher Giovanni Peri. "It's completely turned around." The study found the areas with the largest gains in foreign STEM workers were Austin, TX; Raleigh-Durham, NC; Huntsville, AL; and Seattle. Those cities had adjusted wage gains of 17 to 28 percent for native college-educated workers, while wages for the college-education population declined in 25 of the 33 cities that had a decline in foreign STEM workers.
Research Teaches Robots to Soften Their Grip
The Engineer (United Kingdom) (05/22/14)
Researchers at the University of Birmingham have developed a new way to teach robots to pick up unfamiliar objects without dropping or breaking them. The team trained a robotic hand by programming it to pick up an object and use information learned in that first grip to grasp and move a whole range of similar objects. The method will enable robots to be used in more flexible ways and in more complex environments, particularly where humans and robots need to be able to work together, according to the researchers. The team used robotic hands that look very similar to human hands, with five jointed fingers, and the robot was able to assess the object and generate about 1,000 different grasp options in five seconds. "That means the robot is able to make choices in real time about the best grasp for the object it has been told to pick up and it doesn't need to be continually retrained each time the object changes," says Birmingham professor Jeremy Wyatt.
MIT News (05/21/14) Larry Hardesty
Massachusetts Institute of Technology researchers have developed an algorithm that can analyze photos of a scene, taken from multiple perspectives, and produce a camouflage covering for an object placed within it. Objects hidden by the algorithm take, on average, more than three seconds to find. The researchers say they developed three algorithms with varying levels of sophistication. The simplest algorithm averages the color values of the patterns produced for each face from each camera angle, while the slightly more complex algorithm picks one angle at random for each face. The most sophisticated algorithm attempts to minimize the difference between the shape perceived by the viewer and the shape patterned on the object. During testing, the researchers found the most complex algorithm fared the best because the topology of the environments they were examining naturally limited the number of angles from which photos could be taken. The researchers created another algorithm that selected a pattern drawn from a single perspective for each face, but performed that selection by choosing the perspective that worked best from as many angles as possible while allowing for smooth transitions between faces.
Despite Data Thefts, the Password Endures
The Wall Street Journal (05/22/14) Danny Yadron; Katherine Rosman
Although passwords are derided by some security experts as being an insufficient security measure, efforts to replace passwords with better protections face an uphill climb. "It's become kind of a nightmare," says 1990 ACM A.M. Turing Award recipient Fernando Corbato, who helped develop the first computer password while at the Massachusetts Institute of Technology in the early 1960s. "I don't think anybody can possibly remember all the passwords." However, researchers say replacing passwords with security measures such as biometrics and USB keys would require competing technology companies to work together to develop standards. In addition, Internet users may be hesitant to give up using passwords in favor of more secure options. Passwords also are inexpensive to use and are already deeply engrained in the design of many websites. Nevertheless, efforts are underway at some companies to move away from using passwords. For example, PayPal, Bank of America, and Google are among the companies that have formed the Fido Alliance, which has released a set of standards for security measures that could replace passwords with other types of online identification measures. Meanwhile, the National Strategy for Trusted Identities in Cyberspace, a task force created by President Barack Obama in 2011 to bolster online security, also is working to develop alternatives to passwords.
Robots Transform Into Furniture at EPFL
Swiss Federal Institute of Technology in Lausanne (05/21/14) Laure-Anne Pessina
Researchers at the Swiss Federal Institute of Technology in Lausanne's Biorobotics Laboratory (BIOROB) have developed Roombots, small robotic modules that can change their shape to create reconfigurable furniture. Roombot pieces, which are connected wirelessly, can be stacked upon each other to create various structures. Each module is equipped with a battery and three motors that enable the module to pivot with three degrees of freedom. In addition, the modules have retractable claws they use to attach to other pieces and form larger structures. "In order to keep costs down and ensure solidity, we have prioritized the construction of hybrid furniture, in other words, objects made up of both robotic modules and solid 'passive' elements," says BIOROB director Auke Ijspeert. The small robots also can climb walls using a special surface with holes adapted to the mechanical claws. The Roombots project currently includes four functional modules with an autonomy of one hour. The researchers currently are testing different user interfaces for the system.
Tracking Lost Luggage Part of AT&T's Vision of the Future
USA Today (05/21/14) Edward C. Baig
AT&T demonstrated its User-Defined Network Cloud and other innovative technologies that could be available soon at a recent company event in New York City. Smart Luggage with embedded Global Positioning System technology would let travelers track a suitcase's whereabouts, receive alerts when it is off the plane, and find it by tapping an app that lights up an LED beacon on the luggage. Users would need to power the Smart Luggage the night before they travel using a USB port built into the suitcase. Meanwhile, AT&T researchers also have developed a SmartMic app that turns mobile devices into a wireless microphone for use in conference halls as well as lecture halls using Web Real Time Communications technology. For example, a professor would connect her laptop to an audio system in the room so everyone can hear, and would select the speaker through the SmartMic website. The researchers also displayed EyeDecrypt, smartphone software that enables users to wave the app in front of a document so spies see only a jumble of meaningless characters. The technology also can decode or mask characters at physical locations such as ATMs, and could potentially work with wearables such as Google Glass.
Carnegie Mellon, Microsoft Research Automate Privacy Compliance for Big Data Systems
Carnegie Mellon News (PA) (05/21/14) Byron Spice
Researchers at Carnegie Mellon University (CMU) and Microsoft Research have developed a prototype automated privacy compliance system that runs on the data analytics pipeline of the Bing search engine. The researchers say their system is the first time automated privacy compliance analysis has been applied to the production code of an Internet-scale system. The researchers showed that a team of just five workers can manage a daily compliance check on millions of lines of code written by several thousand developers. "During our implementation on Bing, we found that more than 20 percent of the code was changing on a daily basis," says CMU Ph.D. student and lead researcher Shayak Sen. At these large scales, automated methods offer the best hope of verifying compliance. As part of the project, the researchers developed a new programming language, called Legalease, that could be easily learned and used by privacy advocates. Legalease utilizes allow-deny rules with exceptions, a structure that is found in many privacy policies and laws, and is expressive enough to capture the real policies of an industrial-scale system. However, privacy policies often cannot be applied to large codebases written by large teams of programmers, so the researchers leveraged a data inventory system called Grok for the backend data analytics over user data.
Employers Want Java Skills More Than Anything Else
Four DARPA Projects That Could Be Bigger Than the Internet
Defense One (05/20/14) Patrick Tucker
The U.S. Defense Advanced Research Projects Agency (DARPA) is working on several projects with the potential to transform society as significantly as the Internet has done. DARPA director Arati Prabhakar, in recent testimony before Congress, discussed four of the agency's projects. DARPA is exploring ways to use atomic physics to improve the Global Positioning System through its chip-scale combinatorial atomic navigation (C-SCAN) and Quantum Assisted Sensing (QuASAR) projects. The research is based on the idea that navigation without a satellite is possible by measuring the Earth's magnetic field acceleration and understanding how position impacts individual atoms. Researchers say the technology could lead to geolocation awareness 1,000 times more accurate than any existing system. Another promising area of research is Terehertz frequency electronics and meta-materials. Using the Terehertz range could enable new devices that do not compete against others for spectrum access. The research also has applications in creating meta-materials that could be used for cloaking equipment such as jets and could possibly enable invisibility. A third area of research is the High Assurance Cyber Military Systems program, which would serve as a cyber-shield for the Internet of Things. Finally, DARPA is working on the Rapid Threat Assessment program, which would significantly accelerate researchers' ability to understand diseases or agents that threaten human life.
Artificial Brains Learn to Adapt
National Science Foundation (05/15/14)
Duke University researchers are studying a new type of spiking neural network that more closely mimics the behavioral learning processes of mammalian brains. Behavioral learning relies on sensory feedback to improve motor performance and enable people to quickly adapt to their changing environment. Spiking neural networks model brain dynamics, with neurons signaling to other neurons with a rapid spike in cell voltage. "Although existing engineering systems are very effective at controlling dynamics, they are not yet capable of handling unpredicted damages and failures handled by biological brains," says Duke professor Silvia Ferrari, who is leading the research. The team is applying the spiking neural network model to complex engineering systems, such as aircraft and power plants. To accomplish this, the researchers wrote an algorithm that tells the spiking neural networks which information is relevant and ranks the importance of each factor to the main goal. The team now wants to test the algorithm biologically, using an optogenetics technique in which lab-grown brain cells are genetically altered to respond to certain types of light, enabling researchers to control how nerve cells communicate. The research could advance the development of prosthetic devices that fulfill motor, sensory, and cognitive functions.
Campaigns Emerge to Attract More Women to Careers in IT
TechTarget (05/15/14) Gina Narcisi
Nonprofit associations, academic organizations, and major corporations are launching efforts to attract more women to the information technology (IT) field. According to the Computing Technology Industry Association (CompTIA), 95 percent of girls say they like technology, but only 9 percent are interested in pursuing an IT career. Despite the IT field's high salaries and job satisfaction, as well as low unemployment rates, women represent only 28 percent of core IT occupations, CompTIA reports. Through its recently launched Dream IT program, CompTIA is providing educational resources and sending speakers to schools and community programs to talk to girls about IT opportunities. Major IT companies also working to draw women to the field. For example, Cisco participates in the annual Girls in Information and Communication Technologies (ICT) Day, organized by the International Telecommunication Union. For ICT Day, Cisco offers girls between the ages of 13 to 18 the chance to visit 18 offices across 18 countries. In addition to tours, participants have the opportunity to meet with women working at Cisco and discuss career opportunities. Hewlett-Packard's Margaret Dawson says the lack of women in IT is not related to capability. "It's more about willingness to try something different, and taking advantage of opportunities to grow."
Europe Wants a Supercomputer Made From Smartphones
IEEE Spectrum (05/20/14) Mark Anderson
Mont-Blanc, a European public-private consortium, aims to make exaflop supercomputers based on the central-processing units (CPUs) used in smartphones and tablet computers. The Mont-Blanc project, which was launched at the Barcelona Supercomputing Center in 2011, currently has 14 partners and is set to run through September 2016. The researchers recently unveiled a prototype blade server that could lead to a full exascale system in the future. The new system would consume between one-fifteenth and one-thirtieth as much energy per processor as conventional high-performance computing systems. The system relies on ARM cores, which are designed to run on small smartphone and tablet batteries, and could yield more speed while consuming less power. Although ARM-based supercomputers are a good investment for the future, Intel X86 CPUs, combined with accelerators such as graphics-processing units, will continue to dominate the Top500 list of the world's fastest computers, says European Technology Platform for High Performance Computing president Jean-Francois Lavignon. Meanwhile, University of Tennessee professor Jack Dongarra notes ARM cores are not a universally agreed-upon path to exascale supercomputer architecture. For example, he says, "the Japanese exascale system, which will use commodity processors with an accelerator, will draw about 30 to 40 megawatts of power."
Three Questions With the Man Leading Baidu's New AI Effort
Technology Review (05/21/14) Tom Simonite
In an interview, Adam Coates, who is leading artificial neural network research at Baidu's new Silicon Valley laboratory, discusses the potential impact of deep learning on creating software that approaches human performance at some tasks. Coates, who worked on the Google Brain project, will help Baidu develop software that learns without human input, in an approach called unsupervised learning. Using unsupervised learning, Google Brain did not reach human performance at recognizing human faces, as it was able to detect human faces just 81 percent of the time. However, better results are possible with supervised learning, in which software is given hand-labeled data to facilitate learning. "If you give me a lot of examples of what you want to predict then I can train software to get that right," says Coates. "The challenge is how to succeed when you don’t have a lot of examples. Human beings do not have to see a million cats to understand what one is." Coates says the appropriate mix of unsupervised and supervised learning will be critical to Baidu's efforts. He notes making a neural network larger will not necessarily make it smarter. Coates says the Baidu researchers want to "build a framework to run big enough experiments to test out all variations in algorithms that might universally improve performance."
Abstract News © Copyright 2014 INFORMATION, INC.
To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.