ACM TechNews
Association for Computing Machinery
Welcome to the May 22, 2015 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Please note: In observance of the Memorial Day holiday, TechNews will not be published on Monday, May 25. Publication will resume Wednesday, May 27.

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


New 'Deep Learning' Technique Enables Robot Mastery of Skills Via Trial and Error
UC Berkeley NewsCenter (05/21/15) Sarah Yang

University of California, Berkeley researchers have developed deep-learning algorithms that enable robots to learn motor tasks through trial and error. Their technique, a type of reinforcement learning, enables a robot to complete a variety of tasks without being pre-programmed with details about its surroundings. The researchers are working with a Willow Garage Personal Robot 2 they dubbed BRETT, for Berkeley Robot for the Elimination of Tedious Tasks. To test their algorithms, the team presented BRETT with several motor tasks, such as placing differently shaped blocks into matching openings or stacking LEGOs. The algorithms offer BRETT real-time feedback via a system that scores its movements; those that bring it closer to completing the task score higher than those that do not. When BRETT was given the relevant coordinates for the beginning and end of a task, it was able to master a typical assignment in about 10 minutes; without this information, the learning process took about three hours. "What we're reporting on here is a new approach to empowering a robot to learn," says Berkeley professor Pieter Abbeel. "The key is that when a robot is faced with something new, we won't have to reprogram it."


Gauging Materials' Physical Properties From Video
MIT News (05/21/15) Larry Hardesty

Massachusetts Institute of Technology (MIT) researchers have developed an algorithm that can recover intelligible speech from the analysis of the minute vibrations of objects in video captured through soundproof glass. "One of the big contributions of this work is connecting techniques in computer vision to established theory on physical vibrations and to a whole body of work in non-destructive testing in civil engineering," says MIT graduate student Abe Davis. The researchers applied the technique to rods of fiberglass, wood, and metal, and in a separate experiment, to fabrics draped over a line. In the case of the rods, the researchers used a range of frequencies from a nearby loudspeaker to produce vibrations. Since the vibrational frequencies of stiff materials are high, the researchers also used a high-speed camera to capture the video. Meanwhile, the fabrics were flexible enough that the ordinary circulation of air in a closed room was enough to produce detectable vibrations. In addition, the vibration rates were low enough they could be measured using a digital camera. Each object's preferred frequencies, and the varying strength of its vibrations at those frequencies, produce a unique pattern, which an altered version of the visual-microphone algorithm was able to extract. The researchers then used machine learning to find correlations between the vibrational patterns and measurements of the objects' material properties.


South African Scientists Create Cheap Computer
News24 (05/20/15)

University of Witwatersrand researchers are leading a project to create inexpensive computers or tablets to potentially be used by every student in South Africa in the near future. "We are creating this human capacity to solve complicated problems in software, hardware, electronics, computing, and all that," says Witwatersrand professor Bruce Mellado. The goal of the project is to equip the entire educational system of South Africa with low-cost computers or tablets, a mission that would be made easier with locally designed and manufactured devices. "We got a grant from the Technology Innovation Agency to start working with schools and produce the first devices here in South Africa," Mellado says. The researchers are studying the interaction between children and technology so the new devices will help students engage in learning, especially in improving their mathematical skills. In addition, the project will help boost the South African economy by keeping more money inside the country and creating jobs. The researchers are still working on the exact configuration of the final device that will be rolled out in the coming months.


With One False Tweet, Computer-Based Hack Crash Led to Real Panic
UB NewsCenter (05/20/15) Bert Gambini

A false tweet sent from a hacked account owned by the Associated Press in 2013 caused the Dow Jones Industrial Average to fall by 143.5 points and the Standard & Poor's 500 Index to lose more than $136 billion of its value in seconds. When it was discovered the tweet was fake, the markets corrected themselves nearly as quickly. The event is known as Hack Crash, and demonstrates the need to better understand how social media data is linked to decision-making in the private and public sector, says State University of New York at Buffalo professor Tero Karppi. Based on its speed, Hack Crash was identified as a computer-based event initiated by sophisticated algorithms designed to identify online content with the potential to influence markets. Those algorithms launched a panicked trading spree resulting in thousands of trades per second. Karppi says, "to know exactly what particular financial algorithms do is almost impossible because of their proprietary nature," but he adds the incident is an example of algorithms working according to design. Karppi says the only way to understand Hack Crash is to continue exploring the relationship between social media, the market, and its algorithms. He worked with Kate Crawford of Microsoft Research and the Massachusetts Institute of Technology Center for Civic Media to analyze the Hack Crash in an upcoming issue of Theory, Culture & Society.


Oculus Rift Hack Transfers Your Facial Expressions Onto Your Avatar
Technology Review (05/20/15) Tom Simonite

Researchers at the University of Southern California (USC) and Facebook have developed a system to track the facial expressions of users wearing a virtual-reality headset and transfer them to a virtual avatar. The system tracks the motion of a user's mouth using a three-dimensional (3D) camera attached to the headset with a short boom. Meanwhile, movements of the upper part of the face are measured using strain gauges added to the foam padding that fits the headset to the face. The two data sources then are combined to create an accurate 3D representation of the user's facial movements that can be used to animate a virtual character. "This is the first facial tracking that has been demonstrated through a head-mounted display," says USC professor Hao Li. The system is based on software that can combine data from the sensors tracking the upper and lower parts of the face and match the result onto a 3D model of the face. The software requires a user to go through a brief calibration process the first time the system is used. This step collects data that helps the software correctly match the streams of data from the upper and lower parts of the face. However, Li is working on eliminating this step by feeding the software more data on different faces.


Rutgers Students Win with App for Diabetes Management
My Central Jersey (05/20/15) Cheryl Makin

Team "Copernicus Health" won the Nicholson Foundation's Rutgers Healthcare Delivery Challenge for its smartphone app that engages and motivates underserved populations to better manage their type 2 diabetes. The Rutgers Healthcare Delivery Challenge encouraged teams to develop ready-to-implement service delivery or technology innovations that can improve the quality and contain the costs of healthcare for underserved populations. The Copernicus Health app uses gamification techniques to enable patients to receive points for taking their medicine, self-educating about their disease through the apps' embedded learning tools, and monitoring important clinical metrics at the doctor's office. After users receive a predetermined number of points, they receive rewards in the form of direct cash deposits to their reloadable Copernicus debit cards or discounts to use at healthy lifestyle businesses. The Rutgers teams' "innovative ideas, tenacious problem-solving skills, and commitment to reaching at-risk populations will help change the future of healthcare and the lives of patients in New Jersey," says the Nicholson Foundation's Joan Randell. The Challenge also honored Team "Save A Neck" for its BreatheNVS, an app that directs patients to educate themselves and share information with their physicians on noninvasive management of their respiratory systems, and Team "MAP Training," which developed an app that to help women overcome severe stress and trauma caused by homelessness, sexual or physical abuse, and mental illness.


Researchers Help Video Gamers Play in the Cloud Without Guzzling Gigabytes
Duke University News (05/20/15)

Researchers at Duke University and Microsoft Research have developed Kahawai, a tool they say offers graphics and game play on par with conventional cloud-gaming platforms for a fraction of the bandwidth. Similar to video-streaming services, cloud-gaming lets users stream high-end video games from the Internet. However, transmitting state-of-the-art games with high-resolution graphics and audio fast enough for smooth game play can quickly use a lot of data. Kahawai utilizes a technique called "collaborative rendering" to reduce the amount of data that remote servers have to send during a game. Collaborative rendering reduces the stress on remote servers by letting the mobile device's graphics-processing unit do some of the work. Although the task of quickly generating specific details is still left to the remote server, collaborative rendering lets the mobile device generate a rough sketch of each frame, or a few high-detail sketches of select frames, while the remote server fills in the gaps. The researchers say Kahawai provides the same visual quality as conventional cloud-gaming systems while using one-sixth of the bandwidth. "Games are a natural place to start understanding how collaborative rendering can work, but any graphics-intensive application could potentially benefit from Kahawai, from [three-dimensional] medical imaging to computer-aided design software used by architects and engineers," says Duke professor Landon Cox.


New Technology Could Fundamentally Change Future Wireless Communications
University of Bristol News (05/19/15)

University of Bristol researchers say they have developed a technique that can estimate and cancel out the interference from a transmission, enabling a radio device to simultaneously transmit and receive data on the same channel. The technology requires only one channel for two-way communication, using half as much spectrum compared to conventional technology, a breakthrough the researchers say could fundamentally change radio design and increase data rates and network capacity, reduce power consumption, create less-expensive devices, and enable global roaming. The full-duplex transceiver architecture combines electrical balance isolation and active radio frequency cancellation to suppress interference by a factor of more than 100 million. In addition, the system uses low-cost, small-form-factor technologies, making it well suited for use on mobile devices. "Since the radio spectrum is a limited resource, and with network operators paying billions of pounds to access the spectrum, solving this problem would bring us one step closer to the faster, cheaper, and greener devices of our connected future," says Bristol researcher Leo Laughlin. The technology could be applied to Wi-Fi systems, cellular systems, and radio circuitry.


Real-Time Data Sharing Can Make Airports Greener
University of Lincoln (05/19/15) Marie Daniels

University of Lincoln researchers Michal Weiszer, Jun Chen, and Giorgio Locatelli have created a new framework designed to reduce delays, improve efficiency, and curb pollution at major international airports. The framework, which is based on the creation of a multi-objective algorithm designed to make air travel more sustainable, uses data on the scheduling and routing of aircraft, four-dimensional trajectory (4DT) optimization (the integration of time into the 3D aircraft trajectory), and runway and airport bus scheduling. The researchers tested the framework on real-world data at Doha International Airport, which closed in 2014 to make way for a new airport. "In this study, we tested a new concept, which is the cooperation and sharing of real-time data between airports, aircraft operators, ground workers, and air traffic control in order to reduce delays, improve the predictability of events, and optimize resources," Weiszer says. "By 2030, it is forecasted that the number of airline passengers globally will double to about 6 billion per year. Without action, 12 percent of flights could not be accommodated because of lack of airport capacity."


NASA Selects UNM to Manage Collaborative Agreement for 'Swarmathon' Competition
UNM Newsroom (05/18/15) Karen Wentworth

The University of New Mexico (UNM) has been selected by the U.S. National Aeronautics and Space Administration (NASA) for a new cooperative agreement with NASA's Minority University Research and Education Project (MUREP). UNM will receive a three-year, $1.8-million grant to manage the agency's "Swarmathon" challenge, which seeks to engage 1,000 students at 50 Minority Service Institutions (MSIs) nationwide in cutting-edge computer science and engineering research. "Swarmathon will harness student creativity to solve difficult and complex problems," says UNM professor Melanie Moses, whose lab has developed ant-inspired search algorithms.  The first-ever competition will be held in 2016 at the NASA Kennedy Space Center (KSC) and will challenge students from MSIs to accelerate the progress of cooperative swarm robotics and its eventual use in resource exploration on Mars, asteroids, and celestial objects. During the competition, robots will work cooperatively to autonomously search for and retrieve resources in unmapped environments.  The tiered competitions will engage undergraduates in developing search algorithms and implementing them in physical robots. Teams will compete virtually and on-site at NASA KSC. College participants also will be selected for paid research internships and will involve local high school students in a parallel virtual competition.  Swarmathon builds on findings from Moses' lab, evolutionary simulations, an iAnt swarm of cooperative robots, and NASA KSC Swamp Works' physical robots and hardware innovations.


Researchers to Create "CyberHeart" Platform for Advanced Medical Device Development
Stony Brook University (05/18/15)

The U.S. National Science Foundation (NSF) is providing $4.2 million in funding over five years to Stony Brook University researchers for the CyberHeart project, which aims to improve and accelerate medical-device development and testing. The CyberHeart project is part of NSF's center-scale Cyber-Physical Systems initiative to develop advanced engineered systems that are built from, and depend upon, the seamless integration of computation and physical components. The CyberHeart platform can be used to test and validate medical devices faster and at a much lower cost than existing methods. In addition, CyberHeart can be used to design optimal, patient-specific device therapies, lowering the risk to the patient. "We believe that our coordinated, multi-disciplinary approach, which balances theoretical, experimental, and practical concerns, will yield transformational results in medical-device design and foundations of cyber-physical system verification," says Stony Brook professor Scott Smolka. He says their approach combines patient-specific computational models of heart dynamics with advanced mathematical techniques for analyzing how these models interact with medical devices. The analytical techniques can be used for a range of applications, including detecting potential flaws in device behavior, and in clinics to optimize device settings on a patient-by-patient basis before devices are implanted.


Could a Computer Predict the Next Pandemic?
Science (05/18/15) David Shultz

In a new study, researchers used machine learning to make accurate forecasts of whether animals carry dangerous pathogens. The researchers say the predictions could help experts improve how they prevent and respond to disease outbreaks. The researchers developed software to analyze a large database of mammalian habits and habitats, including the geographic range and reproductive strategies for hundreds of species. The program evaluated 86 variables, including body size, life span, and population density, to identify patterns common among animals known to carry zoonotic diseases. Team leader Barbara Han at the Cary Institute of Ecosystem Studies and her colleagues restricted their analysis to rodents, which carry a disproportionately high number of zoonotic diseases. Han and her team first used the program to identify lifestyle patterns common to rodents harboring diseases such as black plague, rabies, and hanta virus, and found their model had an accuracy rate of 90 percent. After the machine "learned" the telltale signs, the researchers searched for new rodents that fit the profile but were not previously thought to be carriers. The researchers say the program already has found more than 150 new animal species that could harbor zoonotic diseases. The program also predicted 58 new infections in rodents that were already known to carry one zoonotic disease.


9 Programming Languages and the Women Who Created Them
Computerworld (05/18/15) Phil Johnson

Despite software development's reputation as a male-dominated field, women have made many important and lasting contributions to the field, including the development of programming languages. One of the first was the ARC assembly language created by Kathleen Booth for the ARC (Automatic Relay Calculator) computer in 1950. In 1955, Soviet physicist and mathematician Kateryna Yushchenko developed the Address programming language, which was widely used in the Soviet Union for more than 20 years. A few years later, the U.S. Navy's Grace Hopper was one of the technical advisers for the committee that created COBOL. In 1962, IBM's Jean Sammet developed the FORMAC programming language, which was widely used for symbolic mathematical computation. Later that decade, the Massachusetts Institute of Technology's Cynthia Solomon helped develop Logo, which would go on to influence educational programming languages such as Scratch. In 1974, Barbara Liskov, the first woman awarded a computer science Ph.D. in the U.S., led the team that developed CLU, which would influence later languages such as Java and Python. Xerox PARC researcher Adele Goldberg was part of the team that developed the Smalltalk programming language released in 1980. British computer scientist Sophie Wilson developed BBC BASIC, a programming language used to teach people about programming. Finally, in 1991, Christine Paulin-Mohring created Coq, a new implementation based on the Calculus of Inductive Constructions language.


Abstract News © Copyright 2015 INFORMATION, INC.
Powered by Information, Inc.


To submit feedback about ACM TechNews, contact: technews@hq.acm.org
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe