Welcome to the July 10, 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).
HEADLINES AT A GLANCE
IBM Announces Computer Chips More Powerful Than Any in Existence
The New York Times (07/09/15) John Markoff
IBM has announced the production of working samples of ultradense computer chips that overtake the capacity of the most powerful current chips by about a factor of four. The chips contain 7-nanometer transistors, an innovation made possible via the use of silicon-germanium instead of pure silicon in key regions of the switches. The material enables faster transistor switching and less power requirements. The new chips, which were developed by an international consortium led by IBM, suggest the miniaturization of semiconductor technology can be sustained at least through 2018. However, further advances will require new materials and fabrication methods. In addition to the shift to silicon-germanium, the chip industry must consider using extreme ultraviolet (EUV) light in manufacturing. IBM's Mukesh Khare says the consortium envisions a technique to use EUV light in commercial manufacturing operations, with the ultimate goal of shrinking circuits by another 50 percent over the industry's 10-nanometer technology generation slated for release in 2016. IBM notes 7-nanometer transistors would make it possible to build microprocessors with more than 20 billion transistors. "I'm not surprised, because this is exactly what the road map predicted, but this is fantastic," says Subhashish Mitra, director of Stanford University's Robust Systems Group.
How the Audacious Pentagon Agency That Invented the Internet Is Now Trying to Save It
The Washington Post (07/08/15) Christian Davenport
The U.S. Defense Advanced Research Projects Agency (DARPA) is rethinking the underlying mechanisms of cybersecurity to focus on preventive measures instead of simply breach detection. "The computer security industry is basically a bunch of automated detectors set up to let us know when it's time to call the cavalry--those people who can do the job computers can't," says DARPA's Michael Walker. "And when we call in the cavalry, most of the time we've already lost." DARPA's end goal is an automated system that spots and corrects software bugs on its own, and to that end it has invited teams of researchers, hackers, industry players, and others to vie for a $2 million first prize in the Grand Cyber Challenge. The seven finalists chosen for next year's contest include a team from defense contractor Raytheon, which has built a huge cybercenter described by one company executive as a "live-fire cyber range" concentrating on "hard-core systems engineering and hard-core vulnerability assessments." An unhackable system may ultimately be an unachievable goal, but Walker says the Grand Cyber Challenge is an important step forward. "The great thing about trying to kick off an industry revolution is we're trying to make people believe that this is possible and set them on that course," he says.
Cutting Cost and Power Consumption for Big Data
MIT News (07/10/15) Larry Hardesty
Researchers from the Massachusetts Institute of Technology (MIT) are developing a new system that could enable flash-based servers to perform as well as RAM-based servers, but at less cost and using less energy. The researchers presented their system at the International Symposium on Computer Architecture in June. The key to the new system is the use of field-programmable gate arrays (FPGAs), reprogrammable devices that are often used in niche applications where it is too expensive to manufacture purpose-built chips. The researchers built a prototype network of flash-based servers of interconnected FPGAs, each of which controlled two 50GB flash chips. The arrangement created a very fast network that enabled any server to retrieve data from any of the constituent flash drives. The researchers then tested their prototype network against a network using conventional random-access memory (RAM) and found it was just as efficient, but was less expensive and consumed less energy. "This is not a replacement for [dynamic RAM] or anything like that, but there may be many applications that can take advantage of this new style of architecture," says MIT professor Arvind, whose group led the research. "Everybody's experimenting with different aspects of flash. We're just trying to establish another point in the design space."
IBM Researchers Try to Measure Employee Well-Being Using Technology
The Wall Street Journal (07/08/15) Rachael King
Researchers at IBM and the Delft University of Technology are working to develop computer systems that could help foster well-being among employees. The project, known as "Inclusive Enterprise," has been underway for about a year, and currently is focused on creating systems that can help ensure employees are comfortable at work. The effort seeks to use IBM's various platforms, including it's IBM Bluemix cloud development platform and IBM Connections internal social collaboration tool and, eventually, its Watson cognitive computing system, to aggregate and analyze information from social media, mobile apps, and sensors to gain a better understanding of what makes a given worker comfortable. Eventually, the hope is that systems will be able to recommend workspaces optimized for factors such as temperature and noise level. "Eventually, we foresee a system that automatically recommends working conditions that are best for each employee," says Robert-Jan Sips, research lead at IBM's Center for Advanced Studies Benelux in Amsterdam. The research currently is in the exploratory phase, and some of the next steps include finding ways to gather measurements from existing devices, such as smartphones, as well as using gamification techniques to increase employee engagement.
The Rapid Rise of Neural Networks and Why They'll Rule Our World
New Scientist (07/08/15) Hal Hodson
Neural network technology has advanced to the point where it underpins cutting-edge speech, text, and image-recognition systems, and scientists such as the University of Montreal's Yoshua Bengio believe the next frontier is human-computer interaction. Neural networks are unique from other artificial intelligence (AI) systems in that they can learn to perform complex tasks from information generated by humans, without requiring human intervention. However, little is known about how neural networks arrive at answers. Using the Internet or smartphones almost certainly feeds data into a deep-learning system that is likely reliant on neural networks originally trained on human data. Chips customized to run neural networks could be incorporated into next-generation smartphones, making pocket-sized adaptable learning systems a possibility. However, there are serious existential implications of neural networks' evolution, such as whether their advancement could bring about the birth of actual machine consciousness. "Machines will become vastly intelligent, but they're lacking this sense of being in the world," says Sonoma State University's John Sullins. With that sense added, the ethical ramifications of AI become a pressing matter, and Sullins notes the chief threat is of "very capable machines that can get out of control doing what we programmed them to do."
Study Highlights Striking Racial and Gender Gaps in High School STEM
Education Week (07/07/15) Liana Heitin
High school boys are significantly more likely to take engineering and technology classes and to consider pursuing postsecondary science, technology, engineering, and mathematics (STEM) majors, despite the fact that male and female students are earning high school math and science credits at similar rates, according to the National Center for Education Statistics' (NCES) High School Transcript Study. The study also found black and Hispanic students are much less likely than white students to take upper-level STEM courses. The NCES researchers found boys and girls tended to take the same amount of total STEM credits, and although girls earned slightly more credits on average than boys in science and math, boys earned more credits in computer science and engineering. Meanwhile, about 21 percent of males earned credit in engineering and technology, but only 8 percent of females did the same. About 33 percent of male students were considering a STEM major, compared to 14 percent of females, according to the study. In terms of race, nearly 42 percent of Asian students were considering majoring in STEM, compared to 25 percent of white students, 20 percent of Hispanic students, and 15 percent of black students.
Google's Deep Learning Machine Learns to Synthesize Real World Images
Technology Review (07/06/15)
Google's Street View feature makes it possible to view panoramic images of cities and locations from around the world, but to create a seamless experience of exploring specific locations would require far more images than Street View has to offer. One possible solution is to create an algorithm that can interpolate new Street View images, something a team at Google led by John Flynn has been working on. The researchers have turned Google's machine-learning acumen on the problem and created DeepStereo, a system that can be trained to generate new Street View images. The team first trained DeepStereo on 100,000 sequences of images taken from a moving vehicle, then tested it by removing a frame from a Street View sequence and asking it to regenerate the missing frame based on the two adjacent images. The researchers say the results were surprisingly good. "Overall, our model produces plausible outputs that are difficult to immediately distinguish from the original imagery," the researchers say. DeepStereo was particularly good at handling difficult subjects such as trees and grass, as well as moving objects. However, the method is not perfect. Its images lose some resolution, thin foreground structures tend to disappear, and images take 12 minutes to generate on a multicore workstation.
Project Jupyter Gets $6M to Expand Collaborative Data-Science Software
UC Berkeley NewsCenter (07/07/15) Sarah Yang
An open source platform known as Project Jupyter will receive $6 million in grants over three years to strengthen the tool's capabilities for collaborative data science. The Alfred P. Sloan Foundation and the Gordon and Betty Moore Foundation are each providing $1.5 million to the University of California, Berkeley, while the Shelley Charitable Trust is providing $3 million to California Polytechnic University (Cal Poly). The effort expands upon Jupyter Notebook, a Web-based platform developed by an open collaboration co-led by Fernando Perez at the Berkeley Institute for Data Science and Cal Poly professor Brian Granger. The tool enables scientists, researchers, and educators to combine data from multiple formats, including live code, equations, narrative text, and rich media, into a single, interactive document. Perez and Granger estimate more than 1 million people in various fields currently use Jupyter. Applications include analyzing gene-sequencing datasets, images from the Hubble Space Telescope, and models of financial markets. Educators can write instructions in the notebook, include a coding exercise after the instructions, and request an interpretation of the results.
GHOST: Technology That Leaps Out of the Screen
CORDIS News (07/01/15)
Since January 2013, a research project funded by the European Union has been working to develop new tactile interfaces. The Generic, Highly-Organic Shape-Changing Interfaces (GHOST) project seeks to move computer interfaces beyond screens to create displays that can be directly manipulated and shaped. "It's not only about deforming the shape of the screen, but also the digital object you want to manipulate, maybe even in mid-air," says University of Copenhagen professor Kasper Hornbaek, coordinator of the GHOST project. "Through ultrasound levitation technology, for example, we can project the display out of the flat screen. And thanks to deformable screens we can plunge our fingers into it." The effort has so far yielded several promising prototypes, including Emerge, which enables users to pull data from charts and graphs off the screen with their fingertips and then rearrange them as they see fit. Another prototype is a range of "morphees," flexible mobile devices with displays made of lycra or alloy that can bend and stretch; for example, to shield a user's fingers when they input a PIN code. One GHOST technology, ultrasound used to create mid-air tactile feedback, has been spun off into a company, UltraHaptics.
Questioning the Fairness of Targeting Ads Online
CMU News (07/07/15) Byron Spice
A new Carnegie Mellon University (CMU) study found significantly fewer women than men were shown online ads promising them help getting jobs paying more than $200,000. The researchers studied Google ads using AdFisher, a CMU-developed tool that runs experiments with simulated user profiles. The researchers used AdFisher to conduct 21 experiments evaluating Ad Settings, a webpage Google created to give users some control over the ads delivered to them. AdFisher creates hundreds of simulated users, enabling researchers to run browser-based experiments in which they can identify changes in online behavior. "We can't look inside the black box that makes the decisions, but AdFisher can find changes in preferences and changes in the behavior of its virtual users that cause changes in the ads users receive," says CMU Ph.D. alumnus Michael Carl Tschantz, who is now a researcher at the International Computer Science Institute. For the study, the researchers created 1,000 simulated users, half of which were male and the other half female, and had them visit 100 top employment sites. The researchers found the site most strongly associated the male AdFisher profiles was a career coaching service for executive positions paying more than $200,000.
Microsoft Open Sources WorldWide Telescope
eWeek (07/06/15) Darryl K. Taft
Microsoft has open sourced its WorldWide Telescope (WWT) project, a visualization environment that functions as a virtual telescope and provides guided explorations of the universe. WWT is now an independent project as part of the .NET foundation, bringing together imagery from the best ground- and spaced-based telescopes to provide terabytes of images and data. "This year, we decided to make the WorldWide Telescope available under an open source license to allow any individual or organization to adapt and extend the functionality to meet any research or educational need," says Microsoft. Opening the code to the public will facilitate extensions and improvements to the software that will continuously enhance formal and informal learning and astronomical research. The project also includes the WWT software development kit (SDK), which enables developers to build apps that transmit data to be rendered in WWT. The SDK is a set of development tools and sample applications that enable data to be imported and visualized in the WWT. "The SDK simplifies data import and collaboration, lowering the entry barrier so that we can more easily experience and share the visualization and storytelling capabilities of WorldWide Telescope," says Microsoft Research's Rob Fatland.
An Algorithm to Investigate Unwelcome Arabian Sea Plankton
Columbia University (07/01/15) Kim Martineau
Columbia University computer scientists are collaborating with oceanographers to determine what is causing the rise of an unusual plankton-like species in the Arabian Sea, which is threatening to disrupt the food chain. Recent years have seen massive blooms of Noctiluca scintillans, a species native to the tropics that is threatening to overwhelm the tiny diatoms that are the backbone of the Arabian Sea's food chain. Although some theories exist about what is causing the explosion of Noctiluca, including agricultural runoff and more powerful winds tied to climate change, definitive proof is lacking. The goal of the collaboration is to use machine-learning algorithms to tease the problem apart. The researchers will first use satellite data to model currents in the Arabian Sea to determine what effect they have on the development of the Noctiluca blooms. Statistical machine-learning techniques will then be used to infer what physical or biological variables might be fueling the Noctiluca's expansion. “Machine learning will help us find the underlying ecosystem dynamics responsible for the differences between the physical model results and the bloom evolution patterns seen in satellite images," says Columbia professor Ryan Abernathey.
Obama's Top Tech Adviser Takes Fight for Silicon Valley Diversity to Washington
The Washington Post (07/09/15) Cecilia Kang
Former Google technology leader Megan Smith is advocating for more workforce diversity in the U.S. tech sector in her role as President Barack Obama's chief tech policy adviser. Adding urgency to her mission are sobering workforce data disclosures from top U.S. tech firms, estimating women comprise less than 20 percent of employees, while Hispanics account for about 5 percent and African Americans less than 2 percent. In an interview, Smith says the biggest perceived challenge for diversifying workforces are deeply entrenched and unconscious corporate prejudices. She cites as an example research showing if a job has 10 desired qualities, on average women will apply if they possess seven and men will apply if they have three. "We are telling our boys to just try and telling our girls to be more prepared," Smith argues. She also observes tech companies that have made diversity their top priority are making substantial progress. "There are...studies that show the financial performances of companies with diverse management teams are just better," Smith notes. She also reports the most highly progressive educational institutions are those that are deeply probing to find and address the root cause of why more women don't pursue computer science.
Abstract News © Copyright 2015 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
|
