Association for Computing Machinery
Welcome to the October 30, 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).


EU Research Project 'ExaHyPE': Software for Exascale-Class Supercomputers
Technical University of Munich (Germany) (10/28/15)

A new international project funded by the European Commission plans to develop software that will help take full advantage of the exascale supercomputing resources that many believe are only a few years away. The European Union is already planning to develop an exascale supercomputer by 2020, and the new Exascale Hyperbolic PDE Engine (ExaHyPE) project will ensure there is scientific software that will be ready for it. The project is the work of an international team of researchers from seven institutions in Germany, Italy, the United Kingdom, and Russia, coordinated at the Technical University of Munich. The team hopes to have established the algorithmic foundations for exascale supercomputers in the next four years. They will start by focusing on developing software for creating simulations that will help researchers better understand earthquakes and cosmic phenomena known as gamma-ray bursts. Key qualities this new software will need to have include speed and efficiency, so it can keep up with exascale computing. Although the ExaHyPE researchers will focus on these two specific applications, they will try to keep the new algorithms as general as possible so they can be applied to other disciplines.

Dog Robot Copes with Rough Terrain
Reuters (10/28/15) Jim Drury

Researchers at the Swiss Federal Institute of Technology in Zurich (ETH) have designed and built a new robot "dog" that can traverse rough terrain either autonomously or while controlled by a remote operator. About as large as a medium-sized dog, StarlETH is battery-powered and uses an array of nearly 50 on-board sensors to perceive its environment and 12 actuators to enable it to move around. "It's meant to be a robot that can climb over obstacles, so being very versatile," says lead researcher Marco Hutter. It also is small enough that it can be deployed by a single person. Hutter says one of StarlETH's more interesting features is integrated springs that help it store and release energy with each bound. He says this was inspired by the mechanics of the human leg, which allows humans to recuperate much of their energy with each stride when they run. Hutter notes there could be any number of possible applications for the technology used to build StarlETH. Successor robots could be used for industrial inspections, bomb disposal, construction work, search and rescue, and demining operations. They also could fill some human roles, such as security guard, waiter, or healthcare assistant.

Researchers Develop Algorithm to 3D Print Vibrational Sounds
Columbia University (10/28/15) Holly Evarts

Researchers at Columbia and Harvard universities and the Massachusetts Institute of Technology have demonstrated that sound can be controlled by three-dimensional (3D) printing shapes. The researchers designed an optimization algorithm and used computational methods and digital fabrication to control acoustic properties by altering the shape of two-dimensional (2D) and 3D objects. The researchers focused on simplifying the process of designing musical instruments that produce sounds through vibrations in the instrument itself, known as idiophones. They demonstrated the new technique by building a "zoolophone," a metallophone with playful animal shapes. The algorithm optimized and 3D-printed the instrument's keys in the shape of colorful animals, modeling the geometry to achieve the desired pitch and amplitude. "By automatically optimizing the shape of 2D and 3D objects through deformation and perforation, we were able to produce such professional sounds that our technique will enable even novices to design metallophones with unique sound and appearance," says Columbia professor Changxi Zheng. In order to increase the chances of finding the most optimal shape, the researchers developed Latin Complement Sampling, a fast stochastic optimization method, which outperformed all other alternative optimizations and can be applied to a variety of other problems.

New Traffic App and Disaster Prevention Technology Road Tested
CORDIS News (10/28/15)

A new app that enables smartphone users to provide feedback on traffic accidents is now available for download. Feedback from the CrowdAlert app, which was tested in Dublin, Ireland, enabled traffic management centers to respond more quickly to accidents around the city. The European Union-funded Insight project developed the app, along with a citywide real-time traffic-monitoring tool called the Insight System. The system, which was tested in the Dublin City control room and with nationwide disaster-monitoring technologies, provided early warnings to experts at situation centers, enabling them to monitor developments in real time. New algorithmic techniques and methodologies for traffic modeling, active learning, crowdsourcing, and event monitoring were integrated into the system. The researchers say the project shows how smartphones and social networks can be used to improve public services and safety. They say the results should be of interest to public services, which until now have lacked the necessary infrastructure to exploit miscellaneous data streams. "The keys to success of this project were focusing on incorporating the user's perspective, developing robust systems, and evaluating these under realistic conditions," says project coordinator Dimitrios Gunopulos. "The project has shown that bringing in new technologies can improve traffic and emergency management and provide a successful model for developing such complex systems."

Now You Can Use Emojis to Search for Cute Cat Videos
Technology Review (10/27/15) Rachel Metz

Researchers at the University of Amsterdam and Qualcomm Research have built a prototype search engine that uses emojis to conduct searches for video content. Visitors to the Emoji2Video site can click on one or more icons from a curated list of 385, which are used to search a data set of 45,000 YouTube videos for relevant matches. "It's a way to drill down very quickly to the very specific thing you're looking for," notes University of Amsterdam graduate student Spencer Cappallo. The researchers applied deep-learning methods to generate emoji labels for videos that appear to represent their content, and to determine how likely it is what is being searched for is in a given frame. Cappallo reports about one out of every 50 frames was analyzed, and the emojis selected to represent those are averaged to produce one short emoji list, ordered in decreasing confidence, for that specific video. Outside of video applications, Cappallo thinks this research could be useful for the generation of small, emoji-driven summaries of photo albums, or for obtaining information from small displays that may make typing difficult but could support information access via image tapping. Linguist Tyler Schnoebelen believes Emoji2Video could help locate videos made by people who speak different languages.

WPI Researchers Show How Side-Channel Attacks Can Be Used to Steal Encryption Keys
Worcester Polytechnic Institute (10/27/15)

Worcester Polytechnic Institute (WPI) researchers have demonstrated that RSA encryption keys can be obtained using a side-channel attack. The researchers used a combination of techniques to create a virtual machine on the same server as a target machine. They then used the co-located machine to spy on the target. The researchers observed how the target accessed information in memory, enabling them to determine when it was retrieving its RSA key. The researchers also were able to determine the key's numeric sequence by charting the timing of the memory access. "We believe this is the first report in the cryptography literature to describe a successful RSA key recovery attack in a commercial cloud environment," says WPI professor Berk Sunar. The researchers focused on a type of cloud service called infrastructure-as-a-service, in which users run computer applications on a virtual machine. After co-locating with a target machine, the researchers launched a prime-and-probe attack, which involves filling a portion of the cache with data and then observing how the target responds. "Our research has shown that covert channels still exist and may be exploited by side-channel attacks," says WPI professor Thomas Eisenbarth. "Crypto keys are safe if users follow security best practices and stick to well-maintained and fully patched crypto libraries."

Facebook's AI Can Caption Photos for the Blind on Its Own
Wired (10/27/15) Cade Metz

Facebook's Accessibility Team says it is developing an artificial intelligence tool that will be able to automatically describe photos posted on the social network to blind users. Matt King, a member of the team who lost his sight in college, says the tool already offers several capabilities. He notes it can describe the objects in a photo, whether it is set inside or outdoors, and to some extent what the people in the scene are doing; for example, whether or not they are smiling. Although it is still limited, King says the current progress is promising. The tool is built on deep-learning techniques such as those used by Facebook to identify people in pictures, and is part of the Accessibility Team's broader goal of making Facebook easier to use for those with disabilities. The lab also is working on closed-caption technologies and tools such as mouth-controlled joysticks, which enable people who cannot use their hands to use Facebook. Facebook researchers also are trying to open up the development process to those with disabilities, and they recently modified Facebook's open source app development tool React to work with text-to-speech readers and other software that aids the disabled.

Seeing Sound
California Institute of Technology (10/26/15) Lori Dajose

California Institute of Technology (Caltech) researchers have developed and tested an assistive device to translate images into sound, using neural connections known as crossmodal mappings. Caltech professor Shinsuke Shimojo and postdoctoral student Noelle Stiles have used crossmodal mappings to stimulate the visual cortex with auditory transmissions that encode environmental data. Their vOICe sensory substitution device combines a small computer linked to a camera attached to darkened glasses, enabling it to "see" what a human eye would. An algorithm scans each camera image, and for each column of pixels it produces an associated sound with a frequency and volume reliant on the vertical location and brightness of the pixels. A visually impaired person wearing the camera on a pair of glasses could then associate different sounds with features of their surroundings. The researchers say both blind and sighted people on whom the vOICe device was tested demonstrated an intuitive ability to identify textures and images from their associated sounds. "We found that using this device to look at textures--patterns of light and dark--illustrated 'intuitive' neural connections between textures and sounds, implying that there is some preexisting crossmodality," Shimojo notes. He says this led to the discovery that these crossmodal connections can be used to make sensory substitution intuitive without any instruction or training.

How Sensorimotor Intelligence May Develop
Institute of Science and Technology Austria (10/27/2015)

The emergence of self-directed behavior in robots can be grounded in the synaptic plasticity of their nervous systems, according to a study by researchers at the Institute of Science and Technology in Austria. The study's authors, Max Planck Institute professor Ralf Der, and Georg Martius, a fellow at the Institute for Science and Technology in Austria, have demonstrated the emergence of sensorimotor intelligence in robots based on their proposed learning rule. Der and Martius used bio-inspired robots consisting of a humanoid and a hexapod robot in physically realistic computer simulations. The robots received sensory input from their bodies but were not given any form of instruction or task. The tight coupling of environment, body, and an artificial neural network enabled the robots to obtain feedback from their situation and adapt quickly. The researchers say the proposed synaptic plasticity is a coupling mechanism that enables a neural network to generate constructive movements for almost any given body. They note the concept could potentially lead to a new understanding of the early stages of sensorimotor development in the natural world.

Cybersecurity Careers: Where Are the Women?
Network World (10/27/15) Michael Cooney

The cybersecurity industry suffers from a widening gender gap in terms of career professionals, a trend exacerbated by millennials' general disinterest in the field, according to a new global study from Raytheon and the National Cyber Security Alliance. The report found 74 percent of young women and 57 percent of young men said schools did not offer the skills needed to pursue a computer science degree. In the U.S., 67 percent of men and 77 percent of women said no high school or secondary school teacher, guidance, or career counselor ever mentioned the possibility of a cybersecurity career. Worldwide, 62 percent of men and 75 percent of women reported no secondary or high school computer classes offered them the skills to help them pursue a cybersecurity career. "It's just woeful that we don't have anywhere close to the number of women we need in the cyber workforce," says Raytheon U.K.'s Paul Crichard. "Cybersecurity today is masculine, and defense is as well. We want to drive to change that." Raytheon's Valecia Maclin stresses the importance of public-private partnerships to encourage girls to nurture an interest in science, technology, engineering, and math in order to prepare more women to enter and diversify the workforce.

Dartmouth Study Illustrates How Game Design Can Reduce Stereotypes and Social Biases
Dartmouth College (10/26/15)

A new Dartmouth College study highlights how games can have beneficial societal effects, using the approach of "embedded game design" to show how games can alter players' prejudices, reduce social stereotypes and biases, and bring about a more complex view of diversity. Dartmouth's Tiltfactor Lab conducted the research, which used embedded game design to incorporate an intended persuasive message within the game's content, mechanics, or context of play, instead of making the message plain to players. The researchers tested the strategies of "intermixing" and "obfuscating." The first approach blends "on-topic" and off-topic" game content to make themes less apparent, and the second uses game genres or framing devices to steer players' concentration away from the true objectives of the game. Two-party card games were employed, with their objectives being to challenge gender stereotypes and implicit bias in science, technology, engineering, and math (STEM). "Our work reveals that strategically embedding psychological techniques in a game's design both enhances the game's impact and provides a transformative player experience," says Tiltfactor founding director Mary Flanagan. One game used the intermixing approach by interspersing cards that address situations involving bias against girls in STEM or a lack of gender equity with cards that do not address these situations. The other game applied obfuscation, challenging players to name a real or fictional person who matches the adjective-noun pairing revealed as two cards are flipped over.

Is the Field of Artificial Intelligence Sexist?
Quartz (10/26/15) Sarah Todd

With concerns about the future dangers of artificial intelligence (AI) on the rise, some AI researchers say addressing a lack of gender diversity in the field could help mitigate those risks. Artificial intelligence is an overwhelmingly white and male field, and some women in the field say this has caused its focus to be somewhat myopic. University of Maryland, Baltimore County professor Marie desJardins notes men tend to be focused narrowly on the technical aspects of AI, while women tend to focus more broadly on how the technologies they develop will ultimately be used to benefit their communities. Tessa Lau, co-founder of robotics company Savioke, says because of AI's homogeneity researchers are often not exposed to perspectives different than their own, which means the systems they design tend not to take into account the needs and desires of people different from them. Other AI researchers say schools need to emphasize the humanistic applications of artificial intelligence. Olga Russakovsky, a postdoctoral research fellow at the Carnegie Mellon Robotics Institute, this year helped launch the world's first AI summer camp for girls. "The camp aims to teach AI from a top-down perspective, and show kids, 'here are some of the ways it can really be useful,'" Russakovsky says.

A Powerful Look at the Future of AI, From Its Epicenter at Carnegie Mellon
TechRepublic (10/27/15) Hope Reese

In an interview, dean of Carnegie Mellon University's (CMU) School of Computer Science Andrew Moore cites a shortage of people skilled in the underlying sciences of artificial intelligence (AI), even though demand for such expertise is growing. "The thing that keeps me up at night is whether we will find enough middle-schoolers and high-schoolers who want to come into this area," Moore says. He cites advances in computer vision and adaptive robotics as points of innovation at CMU, and he says researchers have set themselves a five-year goal of developing a dexterous robot arm that can easily grasp and manipulate objects to help disabled people. Moore also says CMU faculty are working on the ethical ramifications of sociable robots, and the pressing need for all stakeholders in autonomous vehicles and other AI systems to consider how they operate when presented with moral dilemmas. He says he feels an urgent need to develop AI systems because of their potential to save lives by reducing unnecessary deaths and suffering from disease, accidents, and other calamities. Moore is excited by the increasing inclusion of computer science and robotics in educational curricula in the U.S. and elsewhere, but he says addressing the gender gap in AI education and high-tech leadership requires industrial-academic-governmental partnerships.

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

To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe