Welcome to the January 12, 2018 edition of ACM TechNews, providing timely information for IT professionals three times a week.
Please note: In observance of the U.S. Martin Luther King, Jr. Day holiday, TechNews will not be published on Monday, Jan. 15. Publication will resume on Wednesday, Jan. 17. ACM TechNews mobile apps are available for Android phones and tablets (click here) and for iPhones (click here) and iPads (click here).
To view "Headlines At A Glance," hit the link labeled "Click here to view this online" found at the top of the page in the html version.
The online version now has a button at the top labeled "Show Headlines."
|
|
|
EU Launches Project to Build Fastest Supercomputer in the World by 2023
Science|Business
Eanna Kelly
January 11, 2018
The European Union (EU) has launched a $1.2-billion project to construct the world's fastest supercomputer by 2023, an exascale system capable of 1 quintillion calculations each second. The project is seen as a way to challenge the supercomputing dominance of China and the U.S., which collectively own the highest-performing systems on the Top 500 list, despite Europe being home to 105 supercomputers. "The question is not if the pieces of the [exascale] puzzle come from China or the U.S.," says EU research commissioner Carlos Moedas. "The question is who is able to do it and get the best machine. The world is a marathon, we're all running, and we have to run a little faster." Researchers expect exascale computers to provide more definitive answers on critical questions such as the dangers posed by climate change, with a key obstruction to their development being the need to boost the power efficiency of components.
|
A Clever Radio Trick Can Tell If a Drone Is Watching You
Wired
Andy Greenberg
January 12, 2018
Researchers at Ben Gurion University (BGU) in Israel say they have built a proof-of-concept system for counter-surveillance against aerial drones. Their technique involves generating an identifiable pattern on whatever object a person might want to guard from potential surveillance, and then remotely intercepting a drone's radio transmissions to seek that pattern in the streaming video the drone sends to its operator. The technique exploits "delta frames," in which video is compressed into a series of changes from the previous image, so when a streaming video displays a stationary object, it sends fewer bytes than when it shows a moving object. Research has demonstrated this compression feature can reveal key information about the video content to someone intercepting the streaming data, even when that data is encrypted. "If you can control the stimulus and intercept the traffic as well, you can fully understand whether a specific object is being streamed," says BGU's Ben Nassi.
|
Momentum Builds for U.S. Exascale
HPCwire
Alex R. Larzelere
January 9, 2018
U.S. exascale computing pursuits should ramp up in 2018 given last year's developments helping to set a solid foundation. In terms of hardware, the initial installation of the SC Summit system at Oak Ridge National Laboratory and the NNSA Sierra system at Lawrence Livermore National Laboratory were significant, while clarification of the status of the Argonne National Laboratory's Aurora supercomputer, which will use unspecified "novel" processors, also was newsworthy. Meanwhile, the Exascale Computing Initiative continues to create the middleware to develop end users' applications to guarantee the productive use of exascale. For example, the Exascale Computing Project (ECP) has adopted standard software development kits to support the construction of a comprehensive, coherent software stack so application developers can productively write highly parallel applications targeting diverse exascale architectures. ECP also is progressing in developing applications software, including the deployment of innovative strategies that take advantage of machine learning to use the graphics-processing units that will be part of future exascale computers.
|
A Major Step Forward in Organic Electronics
Linkoping University
Monica Westman Svenselius
January 11, 2018
Researchers at Linkoping University (LiU) in Sweden say they have created the first complementary electrochemical logic circuits that function stably for long periods of time in water. Organic electrochemical transistors previously have been made from a p-type material in which the charge carriers are holes. The LiU team used an n-type conducting material in which the ladder-type structure of the polymer backbone favors ambient stability and high current when doped. The researchers developed a method to create thick films of the material, because the thicker the film, the greater its conductivity. The researchers note their method also can be used together with printed electronics across large surfaces. The researchers also showed the new circuits function for long periods, both in the presence of oxygen and water. "With an n-type material in our toolbox, we can produce complementary circuits that occupy the available space much more efficiently, since resistors are no longer required in the logical circuits," says LiU professor Magnus Berggren.
|
Computer Science Is Fastest-Growing Subject for Undergraduates
ComputerWeekly.com
Brian McKenna
January 12, 2018
Computer science (CS) is the fastest-growing subject for undergraduate students at U.K. universities, experiencing a spike of 4 percent growth in undergraduate enrollment numbers between the 2015-2016 and 2016-2017 academic school years, according to the most recent higher education student statistics survey. The poll found although more than 80 percent of CS and engineering and technology students were male, the number of female CS students saw an 8.4-percent increase over the past three years. In addition, the percentage of female students studying science subjects overall has continued to rise, from 39 percent in 2012-2013 to 42 percent in 2016-2017. Meanwhile, the number of students of both genders who entered the General Certificate of Secondary Education computing exam rose by 9 percent, from 63,650 in 2016 to 69,350 in 2017. At the graduate level, an increase of 885 full-time, first-year computer science students also was recorded.
|
Developing a Secure, Un-Hackable Net
UCL News
Bex Caygill
January 11, 2018
A team led by researchers at University College London in the U.K. say they have developed a technique of securely communicating between three or more quantum devices, regardless of who made them. "Our approach works for a general network where you don't need to trust the manufacturer of the device or network for secrecy to be guaranteed," notes the University of Oxford's Matty Hoban. "Our method works by using the network's structure to limit what an eavesdropper can learn." The researchers say their method uses machine learning and causal inference to assess the devices' security before engaging in communications with the entire network by checking if the correlations between devices are innately quantum and cannot have been generated by another means. The team uses the correlations to set up secret keys that can be used to encrypt any desired communication, and which cannot be intercepted because quantum mechanics ensures their secrecy can be tested and guaranteed.
|
Researchers Implement 3-Qubit Grover Search on a Quantum Computer
Phys.org
Lisa Zyga
January 11, 2018
Researchers at the University of Maryland (UMD) and the U.S. National Science Foundation say they have implemented Grover's search algorithm with trapped atomic ions, using three quantum bits (qubits) on a scalable quantum computing system. "Additionally, this is the first implementation of the algorithm using Boolean oracles, which can be directly compared with a classical search," notes UMD's Caroline Figgatt. The team says the reported implementation of Grover's algorithm had a success rate lower than the theoretical value but still notably higher than the classical success rate. In testing the algorithm on databases with two correct solutions, the researchers found the theoretical success rates were 47 percent and 100 percent for classical and quantum computers, respectively. The team also thinks this implementation of Grover's algorithm can be scaled up to larger databases in the future. "We plan to continue developing systems with improved control over more qubits," Figgatt says.
|
These States Embraced Computer Science Education in 2017
EdScoop
Corinne Letsch
January 9, 2018
Schools across the U.S. started to embrace computer science (CS) in 2017, with several states now moving ahead with legislation to make CS education mandatory. "I think the field has obviously progressed [in the] last year," says the CSforAll Consortium's Ruthe Farmer. "Last year, we had a call to action, and this year we have action." Among the notable efforts is Ohio Gov. John Kasich's signing of a bill requiring the state education department to adopt standards and a curriculum for CS in grades K-12. Meanwhile, Virginia in November established mandatory standards for schools to incorporate CS content into their curricula. Idaho's House Education Committee in early 2017 passed CS standards, urging the state's STEM Action Center board, the state board of education, and the state education department to collaborate on their adoption, provide training to teachers, distribute grants to schools, and establish an online portal with instructional materials.
|
Artificial Intelligence Can 'Evolve' to Solve Problems
Science
Matthew Hutson
January 11, 2017
A series of papers from Uber researchers demonstrate how "evolving" artificially intelligent neural networks has enabled them to play video games, solve mazes, and even make a simulated robot walk. The reports detail a new approach to neuroevolution, in which the best networks are mutated and selected, without relying on programming tricks to simplify problems. "That means complex problems requiring a large network are now accessible to neuroevolution, vastly expanding its potential scope of application," says Uber researcher Kenneth Stanley. The most novel paper tries multiple concurrent solutions, in which a large collection of randomly programmed neural networks is tested on a video game and the best are copied, with slight random mutations, supplanting the previous generation. This strategy outperformed two popular methods for training neural networks on five of 13 games while also managing to teach a virtual humanoid robot to walk, developing a neural network 100 times larger than any previously developed via neuroevolution to control the robot.
|
'Holostream' Allows High-Quality Wireless 3D Video Communications
Purdue University News
Emil Venere
January 9, 2018
Researchers at Purdue University say they have developed Holostream, a platform enabling high-quality three-dimensional (3D) video communications on mobile devices. Purdue professor Song Zhang says Holostream reduces the data size of 3D video without substantially sacrificing data quality, permitting the transmission of video files within the bandwidths provided by existing wireless networks. Zhang notes the breakthrough could improve teleconferencing and telepresence, which uses virtual reality and other interactive technologies to enable users to feel or appear as if they were present in a remote location. "This technology also could enable emerging applications that may require high-resolution, high-accuracy 3D video data delivery, such as remote robotic surgery and telemedicine," he says. Holostream converts 3D video into a two-dimensional (2D) format, so users can leverage 2D communication platforms for low-bandwidth 3D video communications. When applied to the Holostream platform, 3D objects are represented by a mesh of intersecting lines forming triangles, atop of which a "texture" of features is placed.
|
Making the Internet of Things Possible With a New Breed of 'Memristors'
Aalto University
January 10, 2018
Researchers at Aalto University in Finland say they have designed and fabricated the building blocks of future components for neuromorphic computers. Aalto's Sayani Majumdar says the goal is to achieve the extreme energy efficiency of a human brain and mimic the way neural networks process information via electric impulses. The researchers developed "ferroelectric tunnel junctions," which are nanometers-thick ferroelectric thin films sandwiched between two electrodes, with abilities beyond existing technologies that could lead to energy efficient and stable neuromorphic computing. The junctions rely on various electrode materials and less than five volts of electricity, and they also can retain data for more than 10 years without power. The researchers say ferroelectric thin-film components are well suited for neuromorphic computers because they can switch between binary states and a large number of intermediate states, enabling them to store information for a long period of time with small amounts of energy.
|
New Discovery Could Improve Brain-Like Memory and Computing
University of Minnesota News
Lacey Nygard
January 9, 2018
Researchers at the University of Minnesota (UM) and Pennsylvania State University say they have demonstrated the existence of a new type of magnetoresistance involving topological insulators that could potentially improve computing and computer storage. "Our discovery is one missing piece of the puzzle to improve the future of low-power computing and memory for the semiconductor industry, including brain-like computing and chips for robots and 3D (three-dimensional) magnetic memory," says UM professor Jian-Ping Wang. He says the magnetoresistance was discovered in topological insulator-ferromagnet bilayers, and its presence validated the supposition that adoption of topological insulators doubles the magnetoresistance performance at 150 degrees Kelvin, compared to heavy metals. The researchers note from an application viewpoint, this discovery provides a key missing ingredient for the creation of a proposed 3D and cross-bar-type computing and memory device that involves topological insulators by adding the previously missing or very inconvenient read functionality.
|
USC ISI to Develop Translation and Information-Retrieval System for Uncommon Languages
USC Viterbi News
Caitlin Dawson
January 8, 2018
Researchers at the Information Sciences Institute of the University of Southern California (USC) Viterbi School of Engineering have received a $16.7-million grant from the U.S. Intelligence Advanced Research Projects Activity (IARPA) to develop Summarization and domain-Adaptive Retrieval (SARAL), an automated information translation and summarization tool to quickly translate obscure languages. The USC researchers are leading a team of about 30 colleagues from other institutions, and the SARAL project includes experts in machine translation, speech recognition, morphology, information retrieval, representation, and summarization. "The overall objective is to provide a Google-like capability, except the queries are in English but the retrieved documents are in a low-resource foreign language," says USC's Scott Miller. The researchers note they aim to retrieve relevant foreign-language documents and to provide English summaries explaining how each document is relevant to the English query. They will test their system using the low-resource languages Tagalog and Swahili.
|
|
