Welcome to the June 14, 2017 edition of ACM TechNews, providing timely information for IT professionals three times a week.
ACM TechNews mobile apps are available for Android phones and tablets (click here) and for iPhones (click here) and iPads (click here).
|
'Big Data' Resource Raises Possibility of Research Revolution
University of Bristol News
June 12, 2017
Researchers across the U.K. are participating in the Image Data Resource (IDR), a multi-institute collaboration to build a public database aggregating imaging data pertaining to experiments published in scientific journals. IDR automates the collation and storage of these datasets within a free knowledge database. University of Dundee professor Jason Swedlow envisions the resource as having "the potential to speed up research and link datasets so that scientists can look for patterns and commonalities." Swedlow and colleagues in Dundee's Open Microscopy Environment Consortium have used IDR to uncover links between different research projects that have proved elusive to individual scientists. University of Bristol professor Rafael Carazo Salas says his team has demonstrated how "reproducibility and reuse...can enhance research by integrating and cross-validating different imaging studies and by enabling the generation of discoveries, added value, and increased return on investment that could not be obtained from individual studies on their own."
|
Smarter Use of Mobile Data
Swiss National Science Foundation
June 12, 2017
Researchers working on the SwissSenseSynergy projects, a Swiss National Science Foundation (SNSF) effort that is focused on crowdsensing, have discovered ways to improve privacy and localization accuracy and reduce the impact on hardware. The researchers say connecting data from the world's smartphones could provide supercomputing power to all smartphone owners everywhere. Although accessing that processing power would improve the real-time collection and analysis of data, technical hurdles and privacy concerns have held back the technology. The SNSF researchers say their project has addressed these issues and proposed new ways to collect and use such information. The project focuses on crowdsensing, in which access to a smartphone's sensors makes it possible to collect information about a particular area. The researchers developed new approaches to improve crowdsensing technology and established best practices for its application. They are focusing on four key areas--improving location accuracy, boosting security, industry uses, and making data collection more efficient.
|
Women in Tech: Why Bulgaria and Romania Are Leading in Software Engineering
ZDNet
Andrada Fiscutean
June 13, 2017
Bulgaria and Romania have the highest percentages of female software developers across eastern Europe--27.7 percent and 27.2 percent, respectively, according to Eurostat. Bulgarian coder Iva Kaneva says this can be traced to the former Communist regime's push for industrialization in the East Bloc, and its offering of equal work for equal pay to both genders. "This laid the foundation for today's large proportion of girls in tech," she notes. The regime also was heavily influential in teaching women to aim for well-paid professions, and modern Bulgarian and Romanian programmers often earn two or three times their countries' average income, working in outsourcing or research and development for western European or U.S. companies. Misys' Ioana Cicu says both entry-level and managerial positions in those countries have an almost equal male-to-female ratio. "We still have a lot to do to encourage more girls and women to get involved and stay in tech," Kaneva notes.
|
AI Agents Learn to Work Together by Wrangling Virtual Swine
Technology Review
Will Knight
June 13, 2017
Microsoft researchers recently hosted a contest to test the collaboration of artificially intelligent (AI) agents in the execution of complex problems. The competition had AI agents working in the Minecraft-derived Project Malmo environment, in which they were tasked to control and catch virtual pigs, either by themselves or together. Participating teams used machine-learning strategies such as deep learning to train their agents to cooperate, while some participants also employed older, less conventional techniques to instill hard-coded knowledge. "There was no single type of approach that emerged as a clear winner," says Project Malmo researcher Katja Hofmann. The winning team from the University of Oxford in the U.K. resorted to reinforcement learning, in which their agents were rewarded whenever they successfully wrangled a pig via collaboration. Hofmann says hybrid approaches to AI programming such as those demonstrated at the competition likely "will prove particularly promising directions for future research."
|
Smart Jacket and VR Headset Let You Pilot a Drone With Your Body
New Scientist
Timothy Revell
June 12, 2017
Researchers at the Swiss Federal Institute of Technology in Lausanne (EPFL) in Switzerland say they have created a virtual reality (VR) headset and jacket system that enables operators to control a drone using only their body movement. For example, when the user wearing the jacket moves their body to the left, the drone moves to the left, and if the user lunges forward, so does the drone. The researchers say the headset enables the user to see what the drone sees, and built-in headphones provide the sounds surrounding the drone. "The goal is to make people fly without ever leaving the ground," says EPFL researcher Dario Floreano. He notes the jacket gives drone pilots a more intuitive way to control their aircraft, so they can focus on other tasks at the same time. Initial testing has shown users learned to control a drone more quickly with the jacket than with standard joystick controls.
|
Cognitive-Related Neural Pattern to Activate Machines
Autonomous University of Barcelona
June 13, 2017
Researchers at the Autonomous University of Barcelona (UAB) in Spain have identified a functional brain pattern linked to cognitive behavior that can activate an iPad's touchscreen, which they say could be useful in brain-machine interfaces. The researchers worked with electrical brain signals that enabled the activation of the presentation of visual stimuli in the iPad's touchscreen. In addition, experimental animals had to touch those stimuli presented on the iPad in order to receive a reward, which completed the task. The researchers note the rats learned to increase the frequency of the selected neural pattern throughout successive experimental sessions. The experiment also proved that the selected pattern is connected to cognitive processes and not to motor or behavioral activity, which represents important progress in the design of brain-machine interfaces.
|
A School District Is Building a DIY Broadband Network
The Hechinger Report
Chris Berdik
June 12, 2017
The Albemarle County, VA, school district is addressing a lack of student broadband access by building its own network, using the Educational Broadband Service. Walton Middle School principal Josh Walton says the broadband shortage defeats the Internet's promise to enable rural and lower-income students to enjoy academic and cultural resources. "These kids are now disadvantaged two ways and that opportunity gap grows even more," he says. The Albemarle district is deploying four mountaintop base stations linking to broadband Internet beamed from school roofs, facilitating what will be a countywide network within two years. The stations will use communications towers employed by emergency services, which in turn can use the district's broadband. Buried fiber-optic cable will route data between school-based transmitters, receivers, servers, and regional data hubs, while outdoor "client unit" routers will pick up the Internet and direct it to school-issued computers at no charge. One router will be distributed to each student household in the district.
|
HKBU Invention Detects and Alerts Sleepy Drivers
Hong Kong Baptist University
June 12, 2017
Researchers at Hong Kong Baptist University (HKBU) in China have developed a system that detects sleepy drivers and alerts them using a smartphone. The researchers say the system uses a smartphone's real-time video to track and analyze the facial features of a driver, especially changes in eyelid and head position, which are prominent fatigue indicators. The user has to position the smartphone near the steering wheel with the camera facing the driver. The researchers note if the camera captures features such as drooping eyelids, drowsiness, or even nodding off, an alarm is triggered and can only be turned off manually or by voice command. HKBU professor Cheung Yiu-ming says the system is suitable for all drivers, but could be especially advantageous for professional drivers and machinery workers who have long working hours. Cheung also notes only a smartphone is needed, without additional devices or sensors.
|
Why It's Important for the Tech Industry to Get More Young Girls Interested in STEM
TechRepublic
Alison DeNisco
June 12, 2017
Jewelbots CEO Sara Chipps says women are significantly underrepresented in technology, composing only 18 percent of computer science majors and 19 percent of Advanced Placement computer science test takers. Chipps notes the de-socialization of girls from science, technology, engineering, and math (STEM) disciplines starts early. She believes companies can close the STEM gender gap by hiring more senior women managers. "If you get diversity in management and people making hiring decisions, it makes for a much more diverse team," Chipps says. Her Jewelbots firm makes programmable bracelets and other open source products for communication and connection that can be fun as well as educational for girls. Chipps also advises tech leaders to "recognize that bias you might have of thinking as women inherently as junior, and start looking at them as people who can be owners of their work, and people who learned a lot over a long period of time."
|
Neural Networks Take on Quantum Entanglement
Joint Quantum Institute
Chris Cesare
June 12, 2017
Researchers at the Joint Quantum Institute (JQI) and the University of Maryland's Condensed Matter Theory Center (CMTC) have demonstrated that certain machine-learning neural networks can describe quantum states. The researchers say the networks can succinctly represent quantum systems with significant entanglement, which JQI's Sankar Das Sarma says "is a new way of solving intractable, interacting quantum many-body problems that uses machine-learning tools to find exact solutions." The team studied a network of visible neurons that represents real quantum particles, while a network of hidden neurons also was employed to account for particle interactions. The numerical framework helped the researchers confirm some mathematical facts about the families of quantum states represented by neural networks, including the area law limiting the total entanglement of networks with only short-range interactions. "If we want to numerically tackle some quantum problem, we first need to find an efficient representation," says the CMTC's Dongling Deng.
|
Mania and the Machine
UC Magazine
Michael Miller
June 9, 2017
Researchers at the University of Cincinnati (UC) have conducted a study on using artificial intelligence (AI) developed for aerospace simulation to anticipate the outcomes of bipolar disorder therapies. The team used "genetic fuzzy trees" to generate a 100-percent accurate predictive model of patient response to lithium, while even the best common models were only 75-percent accurate in comparison. The genetic fuzzy AI continuously refines its answer to problems, rejecting lesser choices in a manner similar to Darwinian natural selection. AI developer Psibernetix produced a genetic fuzzy AI to specifically generate AIs of its type, and it created the LITHium Intelligent Agent for the UC study. The algorithm used brain-scan analysis to predict patients' positive and negative responses to lithium, as well as reductions of symptoms at eight weeks of treatment. "This is a huge first step and ultimately something that will be very important to psychiatry and across medicine," says UC's Caleb Adler.
|
Taxonomy Goes Digital: Getting a Handle on Social Bots
IEEE Spectrum
Lucas Laursen
June 9, 2017
The U.S. Defense Advanced Research Projects Agency two years ago challenged researchers to identify "influence bots," and the agency now is funding further research on social networks. As part of that effort, Indiana University researcher Gregory Maus will present one of a growing number of socialbot taxonomies this month at the ACM Web Science (WebSci'17) conference in New York. Maus says the taxonomy seeks to expand on earlier taxonomies focused on identifying the different types of botnets and categorizing malicious socialbots that flood a Twitter hashtag used to organize political protests. Maus thinks the new taxonomy will be a more "broad, flexible framework useful for researchers" seeking both to understand and interact with bots. His research creates categories based on the degree to which a bot tries to pretend to be human, who its owner is, how the bot interacts with other bots, whether it hides its connection to its owner, and its mission.
|
Erik Lindahl on Bio-Research Advances, the March to Mixed-Precision, and AI-HPC Synergies
HPC Wire
Tiffany Trader
June 8, 2017
In an interview, Stockholm University professor Erik Lindahl discusses trends impacting high-performance computing. He says computer speed upgrades in the last five years are making it possible to model molecular-scale structures and processes as they occur. "What's changed the last few years is the simulations have become so powerful that not just we but even experimentalists tend to trust them," Lindahl says. He also notes single-precision calculation is likely to become more generally favored than double-precision, since "most chip design is driven by the mass market and the consumer market doesn't really need double-precision that much." Lindahl believes artificial intelligence will yield significant advantages in analyzing all simulation-generated data, but this will force scientists to completely rethink modeling and ways of mining experimental data. "Suddenly you will realize that computers today...might be a factor of 100 or 1,000 faster if you accept to do this in new ways," Lindahl says.
|
|
