Welcome to the January 10, 2018 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).
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."
|
|
|
Japanese Scientists Just Used AI to Read Minds and It's Amazing
CNBC
Catherine Clifford
January 8, 2018
Researchers at Kyoto University in Japan say they have used deep neural networks to decode thoughts; specifically, more "hierarchical" images that have multiple layers of color and structure. They note the technique enables artificial intelligence (AI) to detect objects, and not just binary pixels as engineered in previous methods. "These neural networks or AI model can be used as a proxy for the hierarchical structure of the human brain," says Kyoto University's Yukiyasu Kamitani. The project involved 10 months of showing three subjects natural images, artificial geometric shapes, and alphabetical letters for varying periods while their brain activity was scanned. Afterward, a computer decoded the data to produce visualizations of the subjects' thoughts, and Kamitani notes the team discovered the method could be used to reconstruct visual imagery a person produces by thinking of some recalled images. He says the AI encountered more difficulty rebuilding images when decoding brain signals from a subject remembering those images.
|
Women and Men in STEM Often at Odds Over Workplace Equity
Pew Research Center
Cary Funk; Kim Parker
January 9, 2018
A majority of women employed in science, technology, engineering, and math (STEM) environments where men outnumber women, as well as women working in computer jobs, and women in STEM with postgraduate degrees, have experienced gender discrimination at work, according to a Pew Research Center survey. Half of women in STEM said they experienced workplace discrimination, versus 41 percent of women in non-STEM jobs. The poll also found about 60 percent of blacks in STEM jobs reported experiencing at least one of eight types of racial/ethnic discrimination. Meanwhile, increases in women's share of the STEM profession have been mainly focused among women with advanced degrees, although women are still usually underrepresented among such employees. The survey also found the public perception of STEM jobs includes higher pay and an advantage in drawing young talent versus other industry sectors, while both male and female STEM workers highly value job flexibility to balance their work and family life.
|
KAIST Develops a Technology That Can Reduce Power Consumption of a Smartphone by Half
Etnews
Youngjoon Kim
January 8, 2018
Researchers at the Korea Advanced Institute of Science and Technology (KAIST) have developed a low-power mobile browsing technology by optimizing the number of frames generated by a webpage. The researchers note the amount of power consumed by applications such as Web browsers, YouTube, and Facebook generally ranges from 40 percent to 60 percent of the entire amount of power consumption of a smartphone. However, the researchers say they were able to reduce the amount of power consumption by reducing the number of generated frames of a fixed Web browser. Current Web browsers generate about 30 frames per second (fps) without considering a download of a webpage. Even if there is no change in information due to downloads being delayed, the browser generates excess frames and wastes power. "This technology will be helpful for smartphone manufacturers and Web browser developers that are trying to optimize power consumption," says KAIST professor Shin In-sik.
|
IoT Drives Progress Towards Low-Power Technology
Financial Times
Jessica Twentyman
January 8, 2018
The Towards Zero-Power Electronics project at the Fraunhofer-Gesellschaft Institute in Germany aims to address efficient, low-energy power consumption for the Internet of Things (IoT). "There is still much we can do to minimize power needs in the IoT," says the Fraunhofer Institute for Reliability and Microintegration's Erik Jung. He cites examples of technologies under development that include "wake-up receivers" that use ultra-low currents to monitor wireless sensor networks, and only activate components when they are needed for managing incoming requests or instructions. Another area of concentration is on-chip batteries, in which nanotechnologies etch small cavities on the back of semiconductors or microcontrollers to house a powering electrolyte or capacitor. Also of interest at Fraunhofer are energy-harvesting devices to either extend battery life or make batteries completely redundant. Jung says work is underway at the Institute to explore materials that yield the best power conversion rates for use in energy harvesting.
*May Require Free Registration
|
Scanning an Ancient Biblical Text That Humans Fear to Open
The New York Times
Nicholas Wade
January 5, 2018
Researchers hope to make an ancient Coptic codex, which had been deemed too fragile and damaged for scholars to open without destroying it, readable by first scanning it with computerized tomography (CT) and then using software to extract legible text. Developed by University of Kentucky professor W. Brent Seales, the software simulates the surface of a misshapen piece of papyrus or parchment from x-ray data and then assigns letters to their proper surface so the text is readable. Seales says his method "can turn things thought to be of no value into precious objects." The CT scans of the charred codex parchment, which dates to between 400 and 600 A.D., were completed in December, and the research team plans to start producing readable pages later this month. Project collaborator Paul C. Dilley at the University of Iowa believes such techniques could potentially be applied to many more damaged codices throughout the world.
*May Require Free Registration
|
A New AI That Detects 'Deception' May Bring an End to Lying as We Know It
Futurism
Dom Galeon
January 9, 2018
Researchers at the University of Maryland (UMD) have developed the Deception Analysis and Reasoning Engine (DARE), which uses artificial intelligence (AI) to autonomously detect deception in courtroom trial videos. The team trained DARE to seek out and classify human micro-expressions, such as "lips protruded" or "eyebrows frown," and analyze audio frequency for vocal patterns that signal whether a person is lying or not. DARE then was tested with a training set of videos in which actors were told to either lie or be honest. UMD's Bharat Singh says DARE outperformed the average person in detecting lies, and notes "a remarkable observation was that the visual AI system was significantly better than common people at predicting deception." Singh estimates DARE scored an area under the curve (AUC) of 0.877, which rose to 0.922 when combined with human annotations of micro-expressions, while ordinary people score an AUC of 0.58.
|
Expansion of AP Computer Science Courses Draws More Girls and Minorities
The Washington Post
Nick Anderson
January 8, 2018
The College Board's Advanced Placement (AP) computer science classes are being expanded to attract more girls and underrepresented minorities in the U.S. College Board president David Coleman stresses "this is about...computer science (CS) breaking out of its narrow role." The 2016 national launch of the College Board's AP Computer Science Principles course is seen as key to this growth, since it is designed to appeal to more diverse students. Approximately 27 percent of about 100,000 AP computer science test-takers in spring 2017 were girls, while African Americans comprised 5 percent of those tested and Latino students 15 percent, which is well below their overall school enrollment averages. "We're trying to diversify a field that for whatever reason has remained not so for generations," Coleman says. "There's more work to do." Universities are invested in the success of the AP computer science program because they have long struggled to broaden the demographic base of CS students.
*May Require Free Registration
|
Google and Others Are Building AI Systems That Doubt Themselves
Technology Review
Will Knight
January 9, 2018
Researchers at Uber and Google say they are modifying two popular deep-learning frameworks so they can handle probability, and note the change will provide a way for the smartest artificial intelligence (AI) programs to know when they should doubt themselves by measuring their confidence in a prediction or a decision. The team says the new approach could be useful in critical situations involving self-driving cars and other autonomous machines. University of Cambridge professor Zoubin Ghahramani notes the research reflects the realization that uncertainty is a key aspect of human reasoning and intelligence, and adding it to AI programs could make them smarter and less error-prone. Uber recently released Pyro, a programming language that merges deep learning with probabilistic programming. Stanford University professor Noah Goodman says giving deep learning the ability to handle probability makes it smarter by helping programs recognize things, with a reasonable degree of certainty, from only a few examples instead of many thousands.
|
Manchester Scientists Develop Graphene Sensors That Could Revolutionize the Internet of Things
University of Manchester
Charlotte Powell
January 8, 2018
Researchers at the University of Manchester in the U.K have developed radio-frequency identification (RFID) tags embedded with graphene sensors, technology they say could revolutionize the Internet of Things. The researchers developed humidity sensors for remote sensing that can connect to any wireless network by layering graphene-oxide over graphene to create a flexible heterostructure. The technology can enable various applications such as battery-free smart wireless monitoring for manufacturing processes. "The excitement does not end with this new application here, but leads to the future possibilities of integrations of this technique with other 2D (two-dimensional) materials to open up a new horizon of wireless sensing applications," says Manchester researcher Zhirun Hu. In addition, the technique has the potential to simplify how information is gathered through its wireless system. Manchester professor Kostya Novoselov notes the research is the first example of printable technology in which several 2D materials come together to create a functional device that can be immediately used for industrial applications.
|
With the Summit Supercomputer, U.S. Could Retake Computing's Top Spot
IEEE Spectrum
David Schneider
January 6, 2018
Scientists at Oak Ridge National Laboratory (ORNL) are building Summit, a new supercomputer they say could help the U.S. reclaim the top spot on the semiannual Top500 rankings of supercomputers by capacity from China's Sunway TaihuLight system. Sunway TaihuLight has a peak performance of 125 petaflops and helped Chinese researchers capture the 2016 Gordon Bell Prize for their work modeling atmospheric dynamics. After falling from the top spot in the rankings in 2013, U.S. engineers drafted proposals for a new generation of supercomputers, the first of which is Summit. The new supercomputer's peak performance will be about 200 petaflops when it comes online in a few months, possibly making it the world's most powerful supercomputer. In addition, Summit should be able to run simulations five to 10 times as fast as Titan, the last U.S. supercomputer to lead the Top500 rankings, while using just twice as much power as Titan.
|
The Key to a Scalable, Manageable, and Adaptable Internet
IMDEA Networks
January 8, 2018
Researchers at IMDEA Networks in Spain have completed the HyperAdapt project, which examined the impact of intensive use of software-defined networking (SDN) in the scalability, adaptability, and management of the Internet for industrial purposes. The researchers note they evaluated the potential effect of SDN concepts at a fundamental level, working with idealized models and dealing with the inherent volatility of computer networks. The team proposed a collection of mechanisms and techniques to deal with this dynamism, and provided mathematical proofs of performance limits. In addition, the researchers explored problems in the interconnection between Internet service providers and how these problems can be addressed with SDN. The researchers designed and evaluated techniques and algorithms that adapt to changing network conditions with the goal of improving performance along several dimensions, and also focused on issues affecting wireless access networks. Meanwhile, the team substantiated the need for providing mobile users with new technologies such as device-to-device communication, visible light communication, and millimeter waves.
|
Team's Lifesaving App Could Help Farmworkers Beat the Heat
USC News
Daniel Druhora
January 8, 2018
Researchers at the University of Southern California's (USC) Viterbi School of Engineering have developed Calor, a smartphone app designed to send automatic heat warnings to farmworkers in the field. The researchers note they had to overcome several obstacles in developing Calor. For example, the team faced a lack of good data services on farms, which makes it difficult to get accurate locations for retrieving and delivering weather information. In addition, the researchers had to keep data usage as low as possible for workers who do not have data plans, and had to consider the anonymity of workers through a registration system. "We wanted to tailor the app to their specific needs without overburdening the problem," says USC researcher Faith Florez. In addition to the heat warnings, the researchers also added an educational component to the app that provides media content in English and Spanish to inform workers about their rights and how to protect their health.
|
Driverless Cars Might Follow the Rules of the Road, but What About the Language of Driving?
The Conversation
Abdesalam Soudi
January 7, 2018
Navigating the complexity of unspoken rules and social and linguistic cues to communicate with drivers, pedestrians, and others sharing the road is a challenge for developers of driverless vehicles, writes the University of Pittsburgh's Abdesalam Soudi. He says these rules and protocols are all the more complicated given their variance by region or country. For example, autonomous automobiles programmed to recognize hand signals could still commit potentially disastrous errors because such signals vary widely from region to region and even person to person. "It remains to be seen if the engineers working on driverless cars will be able to program these subtle--but important--differences into these vehicles as more and more appear on the roads," Soudi says. He also notes these and similar issues raise the question of how much knowledge about human societal and linguistic values should be embedded within driverless cars. "How can driverless cars learn to interpret hand and auditory signals?" Soudi asks.
|
|
