Welcome to the August 12, 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).
HEADLINES AT A GLANCE
France and Spain Team Up to Jumpstart Europe's Exascale Computing Ambitions
ZDNet (08/10/15) Anna Solana
France's Alternative Energies and Atomic Energy Commission (CEA) and Spain's Barcelona Supercomputing Center (BSC) have announced a high-performance computing (HPC) partnership to further the European push toward exascale computing. Both groups entered into an agreement to promote "a globally competitive HPC value chain and flagship industry" aligned with the European Commission's agenda to concentrate on technology, infrastructure, and real-world HPC applications. "BSC and CEA fully support the Commission's strategy of aiming 'to make Europe the world leader in HPC' and ensuring it has 'independent access to this technology,'" says CEA's Jean Gonnord. BSC's Francesc Subirada agrees the initiative intends to meet the Commission's goal to obtain a piece of the worldwide HPC market. Gonnord says the industry's long-term plan is to capture a 30-percent share of the market, and BSC and CEA will work with different collaborators to bolster the HPC sector. He notes both organizations are weighing the use of instruments such as Public Procurement of Innovative Solutions to implement new HPC systems in the European scientific community. Joint CEA/BSC projects include an effort to explore new HPC frameworks based on ARM 64-bit low-power processors, a renewable energies center of excellence, and an initiative to use low-power processors and advanced nanotechnologies.
Researchers Find Security Flaws in Developing-World Money Apps
The Wall Street Journal (08/11/15) Jennifer Valentino-DeVries
A study of seven mobile-money applications in Brazil, India, Indonesia, Thailand, and the Philippines by University of Florida researchers found all but one had severely inadequate security measures. "It was worse than we expected," says University of Florida professor Patrick Traynor. One of the apps, India-based MoneyOnMobile, appeared to use encryption to shield data, but did so by transmitting sensitive data to a server unprotected before encrypting it, thus enabling the theft of the data. A second app, Airtel Money, employed encryption but attempted to protect the data using an easily guessable key composed of the same sequence of eight numbers and letters followed by the person's phone number and account number. The researchers also found the apps' terms of service assign liability for any losses from theft or fraud to users. "It is our belief that these applications create significant financial dangers for their users," they warn. The researchers stress the importance of ensuring the security of such programs, especially given the vulnerability of the populations they serve. The study is being presented at this week's USENIX security conference.
Algorithms and Bias: Q and A With Cynthia Dwork
The New York Times (08/10/15) Claire Cain Miller
In an interview, Microsoft Research scientist Cynthia Dwork describes how algorithms can learn to discriminate because they are programmed by coders who incorporate their biases. In addition, she says they are patterned on human behavior, so they reflect human biases. Dwork defines her research as "finding a mathematically rigorous definition of fairness and developing computational methods--algorithms--that guarantee fairness." She notes a study she co-authored found that "sometimes, in order to be fair, it is important to make use of sensitive information while carrying out the classification task. This may be a little counterintuitive: the instinct might be to hide information that could be the basis of discrimination." Dwork says fairness entails similar people are treated in a similar manner. "A true understanding of who should be considered similar for a particular classification task requires knowledge of sensitive attributes, and removing those attributes from consideration can introduce unfairness and harm utility," she notes. The development of a fairer algorithm would involve serious consideration about who should be treated similarly to whom, according to Dwork. She says the push to train algorithms to protect certain groups from discrimination is relatively young, but the Fairness, Accountability, and Transparency in Machine Learning workshop is a promising research area.
Energy-Efficient Depth-Sensing Camera Gleans 3D Information in Bright Sunlight and Darkness
Carnegie Mellon News (PA) (08/10/15) Byron Spice; Sean Bettam
Researchers at Carnegie Mellon University (CMU) and the University of Toronto have created a mathematical model to help address a major problem of depth-sensing cameras: their inability to work in bright light, especially sunlight. Depth-sensing cameras work by projecting a pattern of lines or dots over a scene and determining depth by interpreting the way the lines or dots deform. However, the low-power projectors used by such cameras can easily be overwhelmed by bright light, which washes out the patterns they are projecting. The researchers' new model involves synchronizing the camera with its projector so it ignores the "noise" of ambient light. "We have a way of choosing the light rays we want to capture and only those rays," says CMU roboticist Srinivasa Narasimhan. One of the prototypes produced by the group synchronizes a laser projector with a common rolling-shutter camera so the camera detects only the light from the points being illuminated by the laser. The camera can see in both bright and diffuse light; it can, for example, see the shape of a lit light bulb and see through smoke. The researchers say their new method has numerous possible applications, from improving gaming devices such as Microsoft's Kinect, to medical imaging, automated cars, and planet-exploring rovers. The researchers presented their work this week at SIGGRAPH 2015 in Los Angeles.
Robots Collaborate to Deliver Meds, Supplies, and Even Drinks
MIT News (08/10/15) Adam Conner-Simons
Scientists at the Massachusetts Institute of Technology's (MIT) Computer Science and Artificial Intelligence Lab (CSAIL) presented a system of collaborative robots at the recent Robotics Science and Systems conference. Three machines work together to deliver items rapidly, accurately, and in unpredictable environments. In the demonstration, the researchers had a PR2 robot serve as a bartender while two wheeled Turtlebots would travel to offices and ask human participants for drink orders. The Turtlebots then reasoned about which orders were needed in the different offices and when other machines may have delivered drinks, so they could efficiently seek new orders and bring them to the rooms. The uncertainties robots must contend with in real-world scenarios are connected to sensors, outcomes, and communications, and they are reflected in the CSAIL team's challenge. "It forced us to work on more complex planning algorithms that allow the robots to engage in higher-level reasoning about their location, status, and behavior," says MIT graduate student Ariel Anders. Programming the machines to perceive tasks much like humans do was key, so they would perform a sequence of "macro-actions" that each include multiple steps in a strategy called MacDec-POMDPs. "The MIT team's approach makes it possible to plan actions at a much higher level, which allows them to apply it to an actual multi-robot setting," says University of Liverpool professor Karl Tuyls.
For 40 Years, Computer Scientists Looked for a Solution That Doesn't Exist
The Boston Globe (08/10/15) Kevin Hartnett
Creating a faster method for performing the "edit distance" calculation--a challenge computer scientists have worked on for four decades--was demonstrated as futile by Massachusetts Institute of Technology (MIT) researchers at the recent Symposium on the Theory of Computing conference. The time-consuming operation involves determining the steps needed to convert one sequence of numbers into another, and researchers use the edit distance calculation for such tasks as genomic comparison. The current edit distance measurement technique is the Wagner-Fischer algorithm, which plots data on a grid and functions in quadratic time. One data string is arranged along the top row of the grid while the other runs down the left-hand column, with the algorithm filling in the grid diagonally and counting changes as it goes. Quadratic time means an increase in the length of the data strings is accompanied by an exponential increase in the number of steps required to compare them. MIT professor Piotr Indyk and graduate student Arturs Backurs have shown the impossibility of creating a faster edit distance measurement methodology, as dictated by mathematical laws. They demonstrated that a quicker edit distance calculation mechanism would rely on the presence of a stronger variant of the P=NP problem, which most scientists agree is nonexistent.
Tech Lady Hackathon: 'A Really Open Community for Women'
Federal Computer Week (08/10/15) Bianca Spinosa
The third annual Tech Lady Hackathon was held Saturday at the Impact Hub co-working space in Washington, D.C. The event attracted more than 150 coders, mostly women in their 20s and 30s, who participated in a day-long slate of collaborative programming projects and training sessions. One project was led by Shannon Turner of Hear Me Code and involved brainstorming ideas for improving her organization's website, while another session took the form of a workshop on data visualization. Other projects on the agenda included learning application program interface programming, and working to clean data and help visualize it for the Rebuilding Re-entry program, which aims to improve outcomes for men and women with criminal records. "It's a really open community for women, especially people like me who are still sort of entering in more junior levels in technology," says Grace Turke-Martinez, who works on a political consulting firm's data and analytics team. The event is the creation of Leah Bannon, a product manager at the U.S. General Services Administration's 18F digital services agency. She is transferring to 18F's San Francisco office this month, so this was her last year running the hackathon. She took time during the event to pass the torch, leading a session on how to run a hackathon.
Web's Random Numbers Are Too Weak, Researchers Warn
BBC News (08/09/15) Mark Ward
The Linux-based Web server software that generates random numbers used to scramble or encrypt data should be stronger, suggests a study presented at the Black Hat security event in Las Vegas. The sources of data that some computers call on to generate random numbers often run dry, according to security analyst Bruce Potter and researcher Sasha Wood. The software generates strings of data used as "seed" for random numbers, and ideally the pool of data would possess a high degree of "entropy." However, Potter and Wood found the entropy of the data streams is often very low because the machines are not generating enough raw information for them. Moreover, the researchers warn the server security software does little to check whether a data stream has high or low entropy. The research exposed unknown aspects of encryption on millions of widely used servers. Potter and Wood describe the finding as "scary," and caution it could mean random numbers are more susceptible to well-known, brute-force attacks that leave personal data vulnerable.
Animal-Eye View of the World Revealed With New Visual Software
University of Exeter (08/06/15)
University of Exeter researchers have developed new camera technology that enables users to see the world through the eyes of animals. Using a camera converted to full-spectrum sensitivity, the software can combine one photograph captured through a visible-pass filter with a second taken through an ultraviolet-pass filter. The software is designed to generate functions to show the image through an animal's eyes. Most birds, reptiles, amphibians, and insects can see in more colors than humans, and many can see into the ultraviolet (UV) range. Flowers often look striking in UV because they are signaling to attract pollinators that can see in that range. The software could be used to analyze colors and patterns, and is particularly useful for studying animal and plant signaling, camouflage, and animal predation. "Until now, it has been surprisingly difficult to use digital photos to make accurate and reliable measurements of color," notes Jolyon Troscianko from Exeter's Center for Ecology and Conservation. "Our software allows us to calibrate images and convert them to animal vision, so that we can measure how the scene might look to humans and non-humans alike. We hope that other scientists will use this open access software to help with their digital image analysis."
Making Robots Talk to Each Other
Technology Review (08/04/15) Julia Sklar
Carnegie Mellon University (CMU) researchers have enabled two types of robots with very different capabilities to collaborate in order to fulfill people's requests. One of the robots, Baxter, is a stationary robot equipped with two arms that can delicately manipulate objects, while the other robot, CoBot, has no arms but is adept at navigating indoor spaces and can deliver objects using its front-end basket. The robots communicate wirelessly using a common domain language to convey events as they occur and to provide feedback to each other. When the robots communicate they must decide on one of three options--one robot can tell the other robot to wait until a certain movement to act, one can instruct the other to repetitively carry out the same activity until a particular moment, or one robot can ask the other what to do. "The robots can adjust to each other and optimize their work," says CMU professor Manuela Veloso. She notes the individual robots work independently until they must interact to complete a task, leaving fewer opportunities for mistakes and providing more flexibility. Veloso says the robot teams work best when one specialized robot is the nexus of a group of simpler machines.
Terahertz Optical Transistors Beat Silicon
EE Times (08/04/15) R. Colin Johnson
Purdue University researchers have demonstrated a complementary metal-oxide semiconductor (CMOS)-compatible all-optical transistor that is potentially 1,000 times faster than silicon transistors, boosting switching by nearly 4 terahertz. The aluminum-doped zinc oxide material from which the nano-photonic transistors are manufactured has a tunable dielectric permittivity compatible with all telecommunications infrared standards. "Electrical transistors are limited by the RC delay time while the limiting mechanism for our 'all optical transistor' is recombination time," notes Purdue doctoral candidate Nathaniel Kinsey. "These are entirely different mechanisms and the latter could enable much more freedom in engineering performance and responses to reach faster switching speeds than the electrical counterpart." The researchers note the transparent conducting oxides that make up the photonic transistors can be processed at temperatures low enough for back-end-of-line manufacture, and they are well suited for fabricating optical transistors atop CMOS chips. Their electron-hole recombination time for emitting photons previously topped 100 picoseconds, but the researchers have reduced that time to less than 1 picosecond, fast enough for optical transistors to outperform silicon. "We would like to develop an all-optical plasmonic circuit where we can take light from off-chip [a laser] and feed it efficiently on-chip where it can be modulated to carry data," the researchers say.
Computer Algorithm Could Aid in Early Detection of Life-Threatening Sepsis
Hub (08/05/15) Arthur Hirsch
Johns Hopkins University (JHU) researchers say they have developed a computer-based method that correctly predicts septic shock in 85-percent of cases without increasing the false-positive rate from conventional screening methods. The research's major advance is to detect septic shock patients early enough so medical professionals have time to intervene, says JHU professor Suchi Saria. The research promises significant progress in treating a condition that is estimated to kill about 200,000 Americans a year, notes study co-author Peter J. Pronovost. The study is based on electronic health records of more than 16,000 patients admitted to intensive care units in Boston between 2001 and 2007. The researchers created an algorithm that combines 27 factors into a Targeted Real-time Early Warning Score (TREWScore) for measuring the risk of septic shock. The TREWScore method differs from previous attempts to predict septic shock because it is based on a larger data pool, takes into account more health indicators, and factors in several elements that could have distorted the results. The TREWScore algorithm could be programmed into an electronic health records system to alert medical staff about a patient at risk of septic shock, notes study co-author David Hager.
Testing Trust in Autonomous Vehicles by Fooling Human Passengers
IEEE Spectrum (08/10/15) Evan Ackerman
Researchers have few options when it comes to studying how people react to being driven by autonomous cars. They can use simulations, but subjects know they are not in any real danger, which prevents accurate testing of their reactions. Researchers also can put subjects in real autonomous cars on public roads, but there could be legal issues as well as safety issues in testing specific types of situations to assess reactions and trust. However, a team at Stanford University's Center for Design Research has come up with a third option called the Real Road Autonomous Driving Simulation (RRADS). The approach works by fooling human participants into thinking they are in an autonomous car when they are really not, without lying to them. RRADS is based on a regular car, driven by a human researcher, with a partition that prevents the passenger or subject from seeing the driver. Although the subjects signed a consent form stating the vehicle would be operated by a licensed driver at all times, many assumed it was driving itself. "This provides a lens onto the attitudes and concerns that people in real-world autonomous vehicles might have, and also points to ways that a protocol that deliberately used misdirection could gain ecologically valid reactions from study participants," the researchers note.
Abstract News © Copyright 2015 INFORMATION, INC.
To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.