Association for Computing Machinery
Welcome to the October 21, 2016 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


Stephen Hawking Opens British Artificial Intelligence Hub
Agence France-Presse (10/19/16)

Scientist Stephen Hawking on Wednesday opened an artificial research (AI) center at the U.K.'s Cambridge University. Funded by a $12.3-million grant from the Leverhulme Trust, the Leverhulme Centre for the Future of Intelligence (CFI) will bring together researchers, industry representatives, and policymakers to make sure AI technology is used to benefit humanity. The ethics of AI is a core concern for Hawking, who has warned the technology's misuse could pose serious risks to civilization. "It will bring disruption to our economy," Hawking says. "And in the future, AI could develop a will of its own--a will that is in conflict with ours." Researchers will be tasked with developing systems that have goals aligned with human values and are sufficiently trustworthy. The center also will pursue projects ranging from the regulation of autonomous weapons to the impact of AI on democracy. "We don't need to see AI as replacing us, but can see it as enhancing us: we will be able to make better decisions, on the basis of better evidence and better insights," says Stephen Cave, the center's director. "AI will help us to learn about ourselves and our environment--and could, if managed well, be liberating."


Combating Cybercrime When There's Plenty of Phish in the Sea
University of Cambridge (10/21/16) Sarah Collins

Computer scientists, criminologists, and legal academics in 2015 combined their expertise to form the Cambridge Cybercrime Center, with the goal of helping governments, businesses, and users to construct better defenses against cyberattacks. The Cambridge Cybercrime Center wants to make it easier for cybercrime researchers from around the world to get access to data and share their results with each other. To accomplish their goals, the Cambridge Cybercrime Center researchers will leverage their existing relationships to collect and store cybercrime datasets. Other researchers then can get a license from the Center to study the data without the hassle of identifying and approaching the data holders themselves. "More people will be able to do research, and by allowing others to work on the same datasets, more people will be able to do reproducible research and compare techniques, which is done extremely rarely at the moment," says Cambridge Cybercrime Center director Richard Clayton. The researchers also are studying issues surrounding what motivates someone to commit cybercrime, and what makes them stop. "Our Cybercrime Center will not only provide detailed technical information about what's going on, so that firms can construct better defenses," says University of Cambridge professor Ross Anderson. "It will also provide strategic information, as a basis for making better policy."


NSF Funds New Projects to Advance Energy-Efficient Computing
National Science Foundation (10/18/16) Sarah Bates

The U.S. National Science Foundation (NSF) and Semiconductor Research Corporation (SRC) have jointly awarded $21.6 million for nine new projects to find revolutionary solutions that will enable more energy-efficient computing. The goal is to enable researchers to create a new type of computer that can proactively interpret and learn from data, solve unfamiliar problems using its new found knowledge, and operate with the energy efficiency of the human brain, as outlined by the White House's Nanotechnology-inspired Grand Challenge for Future Computing. "Only disruptive breakthroughs can enable computers to perform as the human brain does, in terms of problem-solving capability and lower power, which, for the human brain, is less than a light bulb's worth of consumption," says the NSF's Dimitris Pavlidis. The three-year projects, which are part of NSF's Energy-Efficient Computing: from Devices to Architectures (E2CDA) program, consider simultaneously novel approaches while inventing new computer architectures to process, store, and communicate data. NSF's Sankar Basu says the projects align with NSF's efforts to advance the human-technology frontier. "This research aims to spark the interdisciplinary science and engineering needed to shape the future of computing," Basu says. "This effort aims to create the fundamental base required to later tackle the bigger problems."


Mobile Phone Camera Used for Complex 3D Modeling
The Engineer (United Kingdom) (10/18/16) Helen Knight

New technology developed at Oxford University in the U.K. can reconstruct large spaces on mobile devices with better accuracy than existing systems. InfiniTAM enables a handheld camera to scan a complex environment and instantly build a three-dimensional (3D) model. The developers say InfiniTAM carries out 3D reconstructions with photo-like accuracy. The system makes use of cameras that produce depth information, such as stereo cameras used by Microsoft's Kinect. InfiniTAM integrates real-time depth information with tracking data on the position of the camera itself, enabling it to determine its own location and update the 3D map as it moves around. The system minimizes the amount of processing power needed to produce reconstructions by only allocating memory to those surfaces that are currently visible in the scene before it. The researchers say InfiniTAM could be used in virtual and augmented reality games and for industrial applications. "This [system] opens up the ability to reconstruct large spaces very quickly on your mobile device," says Victor Prisacariu at Oxford's Active Vision Lab. "One possibility would be to take one onto a submarine and map the ocean floor, for example."


Tech's Gender Gap Is Getting Worse, Not Better, Report Says
TechRepublic (10/20/16) Alison DeNisco

Unless technology companies and educators start reaching out to young women and girls, the number of women in the computer science field will drop from 24 percent to 22 percent by 2025, according to a report from Accenture and Girls Who Code. Accenture's Paul Daugherty believes teaching computer science the same way to girls and boys has reinforced girls' perceptions that computer science is a male-centric field. "Instead, more efforts must be made to tailor engagement with girls to suit the changing influences on their attitudes and preferences as they proceed through their education," Daugherty says. A study by Gallup and Google found girls are less likely than boys to be aware of computer science learning opportunities. Many girls who were engaged in computing in middle school later lost interest in high school, due in part to a lack of friends taking computer science courses. Middle school is a critical period to attract girls to the field; 74 percent of women in computing careers say they were first exposed to computer science in middle school. Research also shows a lack of role models and awareness of career options are prominent factors discouraging girls from considering tech careers; 53 percent of girls say having more information about tech career options would encourage them to consider entering the field.


The Robot Eyes Have It: Cutting-Edge Tool for Koala Conservation
QUT News (10/19/16) Kate Haggman

Queensland University of Technology (QUT) researchers are developing technologies they say could provide less expensive and more accurate koala-tracking methods. The researchers plan to use drones equipped with artificial intelligence and statistical-analysis software to protect vulnerable koala populations. "Using small drones to take images is becoming more common but we know of no others combining this with cutting-edge analytical technologies that draw meaning from those images," says QUT professor Felipe Gonzalez. The researchers found thermal imaging can detect even well-camouflaged koalas, and the counting and tracking algorithms enable them to differentiate the shape of a koala from other animals. Although the project focuses on koala populations, the technology could be adapted for other species, according to the researchers. The technology also will help monitor koalas' movements and population fluctuations over time. "Understanding the abundance of a species in an area is fundamental to the management of that species--and the more regularly and accurately you can monitor the health of the population, the better," says QUT researcher Grant Hamilton. "This combination of technology can provide councils with a wealth of rich data a human cannot, such as exact [global-positioning system] locations and high-resolution imaging."


Think Tank: U.S. Elections Are Far From Hack-Proof
Federal Computer Week (10/20/16) Sean D. Carberry

Panelists this week at a discussion sponsored by the Institute for Critical Infrastructure Technology (ICIT) said voting systems in the U.S. are not immune from hacking or manipulation. Even machines that are not connected to any network can be corrupted via removable media such as USB devices, and the black-box nature of many voting systems means there is no way for election officials to run diagnostics before or after votes are cast, according to the panelists. ICIT recently released its "Hacking Elections is Easy" report, which outlined a range of vulnerabilities, infiltration points, and tactics that could be used to undermine credibility in an election or even manipulate results. "In 2016, 43 states relied on voting machines that were at least 10 years old and that relied on antiquated proprietary operating systems such as Windows XP, Windows 2000, unsupported versions of Linux, and others," and the vulnerabilities for these operating systems are widely available for free download on Deepnet, according to the report. In addition, when devices are connected to networks to tabulate or transmit results, those results can be intercepted or manipulated. An organization such as the U.S. National Institute of Standards and Technology should issue mandatory security standards for electronic voting systems to improve election security, according to ICIT fellow Tony Cole.


Does My Eye Deceive Me? Not With These Digital Forensics Tools
New York University (10/17/16) Kathleen Hamilton

New York University (NYU) researchers, working with faculty from the University of Siena in Italy, Politecnico di Milano in Italy, the University of Campinas in Brazil, Purdue University, the University of Notre Dame, and the University of Southern California are designing new digital forensics tools. They say the tools will be able to catch the subtlest manipulations of still images and video, discerning whether, and how, media has been tampered with. The work is being funded by a $10.4-million U.S. Defense Advanced Research Projects Agency award. The researchers are using a data-driven approach based on machine-learning techniques, combining their collective expertise in a range of computer-related fields to produce extremely sensitive tools that can find details that are impossible to detect via conventional methods. The researchers will produce a system that can analyze a large volume of images and video very quickly. The tools also will be able to identify localized manipulations within an image rather than just whole-image tweaks. The researchers also aim to bring video analytics tools into closer parity with the more advanced tools for analyzing still images. A second research team is working to produce automated "integrity scores" for digital images and video, fusing detailed forensic data into a metric that reflects the degree of authenticity of a video or image.


Pittsburgh's AI Traffic Signals Will Make Driving Less Boring
IEEE Spectrum (10/17/16) Prachi Patel

Carnegie Mellon University (CMU) professor Stephen Smith is developing traffic signals that use artificial intelligence (AI) to adapt to changing traffic conditions in real time. In pilot tests in Pittsburgh, PA, the smart traffic management system reduced travel time 25 percent and idling time by more than 40 percent. The researchers also estimate the system could cut emissions by up to 21 percent, and save cities the cost of road-widening or eliminating street parking by boosting traffic throughput. The system relies on computerized traffic lights that coordinate closely with each other, and each light is equipped with radar sensors and cameras that detect traffic. AI-based algorithms use the data to build a timing plan "that moves all the vehicles it knows about through the intersection in the most efficient way possible," Smith says. In addition, the system sends the data to traffic intersections down the road so they can plan ahead. Unlike other smart traffic management systems, the CMU system is decentralized, enabling each signal to make its own timing decisions. The next phase of the research involves having the traffic signals communicate with cars, and the researchers have installed short-range radios at 24 intersections, potentially letting drivers know of upcoming traffic conditions.


Automating Big-Data Analysis
MIT News (10/21/16) Larry Hardesty

Massachusetts Institute of Technology (MIT) researchers have developed an approach to automating most of the process of big data analysis. The researchers say their new system could perform tasks that normally take months, in just a matter of days. "The goal of all this is to present the interesting stuff to the data scientists so that they can more quickly address all these new data sets that are coming in," says MIT researcher Max Kanter. The researchers have written two papers on the topic, both focusing on time-varying data, which reflects observations made over time. The first paper describes a general framework for analyzing time-varying data, which splits the analytic process into three stages. The second paper describes a new language for describing data-analysis problems and a set of algorithms that automatically recombine data in different ways, to determine the types of prediction problems the data might be useful for solving. "Probably the biggest thing here is that it's a big step toward enabling us to represent prediction problems in a standard way so that you could share that with other analysts in an abstraction from the problem specifics," says Kiri Wagstaff, a senior researcher in artificial intelligence and machine learning at the U.S. National Aeronautics and Space Administration.


Lego-Like Wall Produces Acoustic Holograms
Duke University News (10/14/16) Ken Kingery

Researchers at Duke University and North Carolina State University have demonstrated a system that can create and control three-dimensional (3D) acoustic holograms. Like visual holograms, which manipulate light to make it appear as though an object has materialized in empty space, sound waves can be formed into 3D patterns. "Anybody can tell the difference between a single stereo speaker and a live string quartet playing behind them," says Duke doctoral student Yangbo Xie. "Part of the reason why is that the sound waves carry spatial information as well as notes and volume." Researchers can create 3D sound patterns by using synthetic metamaterials resembling a wall of Lego blocks, with each block composed of 3D-printed plastic and containing a spiral; the tightness of the spiral affects the speed of the sound waves traveling through it. If one side of the sound wave is slower than the other, the sound will be bent toward the slow side. Tests showed the device could manipulate incoming sound waves into certain shapes and concentrate sound waves into several loud spots of sound. The researchers say the technology could revolutionize applications ranging from home stereo systems to medical ultrasound devices. "It's like an acoustic virtual reality display," says Duke professor Steve Cummer.


UTA Researchers Use Artificial Intelligence to Assess, Enhance Cognitive Abilities in School-aged Children
UT Arlington News Center (10/13/16) Herb Booth

University of Texas at Arlington (UTA) researchers are using the latest methods in computer vision, machine learning, and data mining to assist experts in their effort to assess learning difficulties in children very early in their lives. UTA professors Fillia Makedon and Vassilis Athitsos are using a computer-vision and machine-learning system to collect data while children perform certain physical and computer exercises designed to produce executive function skills, all of which involve attention, decision-making, and managing emotions. Makedon says the data collected will be analyzed to recognize patterns of inattention, hyperactivity, or acting impulsively. Monitoring and analyzing how children are behaving during game-like exercises can be used to build a knowledge base that will enable healthcare professionals to apply predictive techniques and make recommendations for effective intervention. "The goal is to design a low-cost, easy-to-use systems that can be implemented in special education practices worldwide," Makedon says. The U.S. National Science Foundation has awarded a $2.7-million grant for the project because of its potential to enhance the cognitive abilities of children.


What Happens When You Give an AI a Working Memory?
Technology Review (10/12/16) Will Knight

Researchers at Google DeepMind in the U.K. have developed a "differentiable neural computer" that uses backpropagation to learn how to perform relatively complex tasks by determining what data to store in its memory. "Like a conventional computer, it can use its memory to represent and manipulate complex data structures but, like a neural network, it can learn to do so from data," the researchers say. The new system, for example, can work out the best route between stations on London's Underground network by studying diagrams of other kinds of networks and learning their most important features. Carnegie Mellon University professor Ruslan Salakhutdinov says a more advanced version might one day be capable of useful tasks, such as crawling Wikipedia to learn which significant concepts it should memorize. However, he notes scaling up a differentiable neural computer to a greater level of complexity could be daunting, as it must perform a complex calculation querying every stored piece of data to access its memory. The research is seen as attempting to span the gap between symbolic information representation and self-training neural networks. Still, New York University's Brenden Lake says the system must be trained on scores of examples of each task, whereas humans can learn new tasks quickly with far fewer examples.


Abstract News © Copyright 2016 INFORMATION, INC.
Powered by Information, Inc.


To submit feedback about ACM TechNews, contact: [email protected]

Unsubscribe