Association for Computing Machinery
Welcome to the November 19, 2014 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).


Researchers Push Supercomputing's Bleeding Edge With Diverse Applications
IDG News Service (11/18/14) Joab Jackson

One of five finalists will be awarded ACM's Gordon Bell Prize tomorrow at the SC14 high-performance computing conference in New Orleans. This year's finalists include IBM researchers, who developed software and an associated processor capable of executing cognitive computing tasks 100 times more quickly than current techniques; a team from the Netherlands' Leiden Observatory, which used an array of 18,600 graphics-processing units to simulate the long-term evolution of the Milky Way galaxy; DE Shaw Research's new Anton 2 supercomputer design for molecular dynamics simulations; a team of German and U.S. researchers who developed a new technique for modeling earthquakes, and a University of Tokyo researcher whose improvements to earthquake simulation could improve the reliability of urban earthquake response analysis. Georgia Institute of Technology professor Jeffrey Vetter, a previous Gordon Bell Prize recipient who chaired this year's awards committee, says we are living in "a period of exploration in computer architecture, where we're seeing people using heterogeneous computing, new types of memory hierarchies, and different types of programming models to make computers productive and efficient." The $10,000 prize was established by electrical engineer Gordon Bell, who is in attendance at SC14 to receive the Seymour Cray Computer Engineering award in recognition of his years of work in advanced supercomputing.

$45 Million in Grants Fund New Cybersecurity Centers at UC Berkeley, MIT, and Stanford
UC Berkeley NewsCenter (11/18/14) Kathleen Maclay

The William and Flora Hewlett Foundation, the University of California, Berkeley, the Massachusetts Institute of Technology (MIT), and Stanford University on Tuesday jointly announced a trio of new academic initiatives that seek to lay the foundations for new public policy related to the latest cyberthreats. Each of the schools will lead a new initiative focused on a specific cybersecurity issue with the aid of $45 million in grants, $15 million per school, from the Hewlett Foundation. Berkeley will use the funds for its Center for Long-Term Cybersecurity, which will be focused on assessing the possible ways cybersecurity might evolve in the future. Meanwhile, MIT's Cybersecurity Policy Initiative will establish quantitative metrics and qualitative models designed to help inform policymakers about cybersecurity issues. Finally, Stanford's Cyber Initiative will take a multidisciplinary, university-wide approach to the issues of trustworthiness and governance of networks. "Choices we are making today about Internet governance and security have profound implications for the future," says Hewlett Foundation president Larry Kramer. "Having these three universities on board, with their global reach and world-class faculties, is a huge step in addressing one of the defining challenges of our time."

Apple and Others Encrypt Phones, Fueling Government Standoff
The Wall Street Journal (11/19/14) Devlin Barrett; Danny Yadron; Daisuke Wakabayashi

Apple, Google, and other smartphone manufacturers' use of technology to encrypt their devices' communications has provoked a standoff with the U.S. government, which sees such measures as impeding law-enforcement investigations. Federal officials argue the new security measures could undermine efforts to intercept terrorists and other suspected malefactors, while tech companies counter government requests for data stored on phones are intrusive to users' privacy and cost them business. For example, 62 percent of Apple's revenue comes from abroad, and the company stresses encryption is even more crucial to customers overseas, where government spying is a major worry. Compounding the souring relationship between tech companies and the government were former U.S. National Security Agency contractor Edward Snowden's disclosures about federal surveillance and data collection. Apple CEO Tim Cook says law enforcement should go directly to customers with their data requests instead of demanding tech firms supply access. The tech industry also has responded vehemently to law enforcement's warning that encryption on phones would eventually lead to the murder of a child by a criminal as an example of the risk companies run in providing such tools. Industry officials say their products are marketed to ordinary consumers who store personal data on smartphones and are increasingly wary of tech companies.
View Full Article - May Require Paid Subscription | Return to Headlines | Share Facebook  LinkedIn  Twitter 

In Newest Tally, Supercomputing Progress Tapers Off
CNet (11/17/14) Stephen Shankland

The new Top500 supercomputing list indicates supercomputing technology advancements are slowing. In the most recent update to the list, only the 10th ranked system is new to the top 10. "With few new systems at the top of the past few lists, the overall growth rate is now slowing," according to Top500 organizers. Meanwhile, growth at the bottom of the list also is slowing. The slowest member of the Top500 list got 90 percent faster each year on average through 2008. Since then, however, it has increased at only 55 percent each year. China's Tianhe-2 system topped the list for the fourth consecutive time, performing calculations at a speed of 33.86 petaflops. "The greatest barrier to improved effectiveness of [high-performance computing] systems for U.S. industry is software," according to the U.S. Council on Competitiveness. "It seems evident that in both the short term and the long term, there is a need for focused investment in software scalability." The Top500 machines are ranked by how well they can perform complex mathematical calculations using a speed test called Linpack. However, the Linpack benchmark is imperfect and Top500 organizer Jack Dongarra (who received the ACM/IEEE Ken Kennedy Award for 2013) is developing a new benchmark called HPCG. "HPCG is designed to exercise computational and data access patterns that more closely match a broad set of important applications," Dongarra says.

Researchers Announce Advance in Image-Recognition Software
The New York Times (11/17/14) John Markoff

Research teams at Google and Stanford University, working independently, have announced advances in artificial intelligence technology used to recognize and describe the contents of videos and photographs. The researchers from the Stanford Artificial Intelligence Laboratory and the Google Brain project used a similar technique: overlapping neural networks, one trained to recognize images and one to recognize human language. They then trained the neural networks by showing them a selection of images paired with descriptions written by humans. After training on these relatively small sets of images, they then were shown unfamiliar images and asked to identify them, which they did with about double the accuracy of previous computer-vision technologies. Researchers say the most impressive aspect of the two efforts is they were able to teach computers to identify actions in an image, not just the objects being depicted. The technologies could have near-term applications in searching and archiving digital images, and longer-term applications in robot navigation and visual aids for the blind. Although the advances are impressive, experts say they remain extremely remote from human perceptual capabilities. IBM researcher John R. Smith notes computer-vision technology is still a long way off from having a real "understanding" of the content of images.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

NTU Engineers Develop Innovative Process to Print Flexible Electronic Circuits
Nanyang Technological University (Singapore) (11/17/14) Lester Kok

Nanyang Technological University (NTU) researchers say they have developed a technique to print complex electronic circuits using a common T-shirt printer. The electronic circuits are printed using unique materials in layers on top of conventional flexible materials. In addition, the electronic components, such as resistors, transistors, and capacitors, are printed using non-toxic organic materials such as silver nanoparticles, carbon, and plastic. NTU professor Joseph Chang says their unique printing technique has made mass production of inexpensive disposable electronic circuits possible. "This means we can have smarter products, such as a carton that tells you exactly when the milk expires, a bandage that prompts you when it is time for a redressing, and smart patches that can monitor life signals like your heart rate," Chang says. The researchers have used their technique to print a 4-bit digital-to-analog converter and radio-frequency identification tags. The new method is fully additive, and the circuits are printed without the use of toxic chemicals or oxidizing agents. "It is also scalable, as you can print large circuits on many types of materials and most importantly, it is low-cost, as print technology has been available for decades," Chang says.

Researchers Find New Way to Move Atomically Thin Semiconductors for Use in Extremely Flexible Devices
NCSU News (11/13/14) Matt Shipman

North Carolina State University (NCSU) researchers have developed a way to transfer thin semiconductor films of molybdenum sulfide (MoS2) onto arbitrary substrates, which they say paves the way for flexible computing or photonic devices. The researchers say their technique is much faster than existing methods and can perfectly transfer the atomic-scale thin films from one substrate to others, without causing any cracks. "The ultimate goal is to use these atomic-layer semiconducting thin films to create devices that are extremely flexible, but to do that we need to transfer the thin films from the substrate we used to make it to a flexible substrate," says NCSU professor Linyou Cao. The researchers needed a way to move MoS2 films, which are an atom thick and up to 5 centimeters in diameter, without wrinkling or cracking them. The technique takes advantage of the MoS2's physical properties to transfer the thin film using only room-temperature water, a tissue, and a pair of tweezers. "This new transfer technique gets us one step closer to using MoS2 to create flexible computers,” Cao says. "We are currently in the process of developing devices that use this technology."

Software to Automatically Outline Bones in X-Rays
University of Manchester (11/13/14)

Researchers at the University of Manchester are developing software that can automatically pick out the shapes of bones in X-rays. The software already can identify hips, but the team from Manchester's Institute of Population Health intends to adapt the system so it is able to learn to identify other bones and structures within the body. Researchers investigating conditions such as arthritis work with hundreds of images, and then draw conclusions and develop treatments. However, having humans map the outlines of bones from radiographs is not the most efficient way of obtaining the data, says Manchester professor Tim Cootes. "The idea of this software is to take the routine tasks out of human hands, so scientists can focus on drawing conclusions and developing treatments," he says. The software should save thousands of hours on manual work performed by radiographers. The team is building on earlier work that produced Bonefinder, software focusing on hips that has been adopted by several research groups. "Ultimately, we want to get this technology into hospitals where it can save time and resources for the benefit of patients," Cootes says.

Southampton Scientists Light the Way for Future Electronic Devices
University of Southampton (United Kingdom) (11/14/14)

University of Southampton researchers have demonstrated how glass can be manipulated to create electronic devices that will be smaller, faster, and consume less power. The researchers have made prototype devices that use light to bring together different computing functions into one component, using a family of materials called chalcogenides. By doping a chalcogenide glass and forming a multilayer structure with electrical contacts, the researchers demonstrated the same switching behavior seen in silicon transistors. "Non-equilibrium doping within chalcogenide glasses enables a unique information-processing platform within one material system," says Southampton's Behrad Gholipour. "This allows traditional electronic computing, along with memory functionality, which opens up the optical spectrum, from the visible far into the infrared, for next-generation optoelectronic and fully-optical computing applications." The researchers report their work has opened up the range of compositions in which this effect can be achieved and reduced the amount of doping needed more than 100-fold. "We are now in the process of consolidating our chalcogenide glass research into a single interdisciplinary center dedicated to the advancement of these fascinating materials," says Southampton professor Dan Hewak.

Evolution Software Looks Beyond the Branches
Rice University (11/12/14) David Ruth; Mike Williams

Rice University researchers have developed PhyloNet, an open source Java-based program that accounts for both the horizontal and vertical inheritance of genetic material among genomes. The researchers also have developed a "maximum likelihood" method that enables PhyloNet to infer network models that better describe the evolution of certain groups of species than do tree models. Inferring means analyzing genes to determine their evolutionary history with the highest probability of connections between species. The software infers the probability of variations that phylogenetic trees cannot illustrate, such as horizontal gene transfers, which circumvent simple parent-to-offspring evolution and enable genetic variations to move from one species to another by means other than reproduction. "We are the first group to develop a general model that will allow biologists to estimate hybridization while accounting for all these complexities in evolution," says Rice researcher Luay Nakhleh. The researchers used two data sets to test the program. One was a computer-generated set of data that mimics a realistic model of evolution and enabled the researchers to evaluate the accuracy of the program, while the other involved multiple genomes of mice found across Europe and Asia.

'Topological Insulators' Promising for Spintronics, Quantum Computers
Purdue University News (11/13/14) Emil Venere

Purdue University researchers say they have uncovered "smoking-gun" evidence that confirms topological insulators could lead to spintronic devices that make practical quantum computers much more powerful than existing technology. Topological insulators are insulators inside but always conduct electricity via the surface. The researchers say they have conducted the clearest demonstration yet of topological insulators' seemingly paradoxical conducting properties. "This is unambiguous smoking-gun evidence to confirm theoretical predictions for the conduction of electrons in these materials," says Purdue doctoral student Yang Xu. The researchers demonstrated a three-dimensional material with an electrical resistance that is not dependent on the thickness of the material. "This experimental system provides an excellent platform to pursue a plethora of exotic physics and novel device applications predicted for topological insulators," says Purdue professor Yong P. Chen. The researchers also found evidence consistent with the conduction of electrons being topologically protected, meaning its surface is guaranteed to be a robust conductor. "For the thinnest samples, such topological conduction properties were even observed at room temperature, paving the way for practical applications," Xu says.

Using 3D Printers to Print Out Self-Learning Robots
University of Oslo (11/12/14) Yngve Vogt

Researchers at Oslo University are developing self-instructing robots using three-dimensional (3D) printers. "Once the robots have been printed, their real-world functionalities quite often prove to be different from those of the simulated versions," says Oslo professor Mats Hovin. "We are therefore studying how the robots deteriorate from simulation to laboratory stage." When the researchers test the robots, they set up an obstacle course to enable them to teach themselves how to pass hurdles. The researchers hope in the future the robots will be able to give automatic feedback to the simulation program about how well they work, so the computer will be in a position to design an even better robot. "The explanation is that a 3D printer will construct whatever you want it to, layer by layer," Hovin says. "This means that you won't have to bother with molds, and you can produce seemingly impossibly complicated structures as a single piece." Hovin is testing certain technical production limits, such as how thin or thick the legs of the robot can be. He says a key benefit of this approach is the short distance from the ideas stage to the robot-testing stage. "Nevertheless, there are many practical challenges ahead before our robots can be exploited commercially," says Oslo professor Kyrre Glette.

Microsoft Upgrades Its F# Functional Language
InfoWorld (11/14/14) Paul Krill

The pre-release of version 4.0 of Microsoft's object-oriented F# features core runtime and compiler improvements. The F# team at Microsoft says core runtime has been enhanced with normalized collections modules and better async stack traces. Modified compiler settings are said to offer better performance by 10 percent. "Not a language feature, but surely of interest to F# language developers, is a change to the GC mode used by the F# compiler," the team notes. "Fsc.exe now uses GCLatencyMode.Batch, which gives a noticeable improvement in overall throughput, something that any F# developer will welcome." Four years ago and under an open source license, Microsoft released the code for F#, its entry into functional languages, a category that includes Scala, Clojure, and now Java. F# 4.0 also offers language improvements, such as handling constructors as first-class functions, in which they receive the same treatment as other .Net methods. There also is a simplified use of mutable values and support for high-dimensional arrays. The newly announced Visual Studio Community Edition also supports F# and extensions such as Visual F# Power Tools. Microsoft's tooling strategy also features the open-sourcing of the server-side .Net stack and plans for Visual Studio 2015.

Abstract News © Copyright 2014 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