Welcome to the August 7, 2017 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."

Heavy traffic on road in Iwo, Tokyo Artificial Intelligence Helps to Keep Tired Drivers Awake
Digital Journal
Tim Sandle
August 6, 2017

Researchers at Panasonic have developed an artificial intelligence (AI) platform to keep tired drivers comfortably alert at all times. The device functions differently based on an assessment for drowsiness, which the platform conducts. The assessment rates driver drowsiness on a five-level scale, ranging from not drowsy to seriously drowsy. The in-car system detects and evaluates drowsiness using a combination of cameras and sensors to analyze vital signs such as the rate of blinking, facial expressions, heat loss from the body, and the degree of illuminance of the skin. More data is recorded about the quality of the driving, enhanced by in-vehicle environmental sensors. The data is then processed via AI and a machine assesses how drowsy the driver is.
Panasonic says the system can identify the onset of drowsiness before the driver is aware of it. The sensors can adjust the environment, such as lowering the temperature inside the vehicle, to keep the driver awake.

Full Article
The Death of Ruby? Developers Should Learn These Languages Instead
Alison DeNisco
August 7, 2017

The Ruby programming language's popularity has plummeted steeply in the past few years, owing to scalability limits, slower application runtimes, and an inability to let computer scientists gain the same types of insight into their data as they can with other languages, according to experts. "[Ruby] might be a good language if somebody wants to start out doing programming, but true computer scientists don't look at it as introducing the true paradigms of computer programming," says Tufts University's Karen Panetta. Many companies have discarded Ruby for languages that offer easier expansion and lower long-term costs, including the MEAN stack, or Python and Java. Coding Dojo's Speros Misirlakis stresses the importance for developers to be agile and conversant in different languages. "Every developer realizes you can't specialize in one language and expect that to be true for 20 or 30 years," Misirlakis notes. "People should be open to learning multiple technologies, languages, and frameworks."

Full Article
21 Technologies Transforming Software Development
Peter Wayner
August 3, 2017

Software development is being transformed by technologies such as continuous integration and smarter languages, while databases are improving and fulfilling a variety of niches. Frameworks are relieving programmers of the burden of writing everything from scratch, and routine libraries also are proving helpful. Meanwhile, the development of mechanisms and rules for enforcing uniform styles is making code easier to understand, and virtual machines are increasingly favored over physical hardware as instruments for running the code. Application programming interfaces have made it largely unnecessary to pack data tightly, and new user interfaces such as smart TVs and smartphones are creating many novel programming opportunities and challenges. In addition, infrastructure as a service and platform as a service are streamlining server and website building, while social media portals are increasingly necessary. Furthermore, performance monitoring is becoming a key necessity as code defects and bottlenecks are no longer restricted to a single machine.

Full Article

Two versions of a Facebook post Facebook Translations Now Rely Entirely on Neural Networks
Eric David
August 3, 2017

Facebook on Wednesday announced its translations are now wholly dependent on state-of-the-art machine-learning neural networks. A team of Facebook researchers says these networks manage more than 2,000 translation directions and 4.5 billion translations daily, generating more accurate translations than Facebook's previous system, which used phrase-based machine translation models. Neural machine translation gauges the complete content of a message together, which is more resource-intensive than phrase-based translation but typically results in a more fluent translation. Facebook also says neural machine translation can handle unknown or misspelled words with greater proficiency, as it can examine contextual clues to determine a word's intended meaning. Facebook thinks convolutional neural networks (CNN) can realize the same accuracy in translation as recurrent neural networks, but significantly faster. In May, the company announced that its CNN-based system was nine times faster than existing networks, and Facebook researchers note CNNs are a better fit for the newest machine-learning hardware.

Full Article
Stanford Bioengineers Encourage Virtual Competitors to Vie for a Different Kind of Athletic Title
Stanford News
Nathan Collins
August 7, 2017

Researchers at Stanford University have built precise models of how individual muscles and limbs move in response to neural control signals, and organized a contest to develop models of the brain's movement control systems. Thus far, 63 teams have submitted 145 concepts to the competition, which is one of five contests developed for the 2017 Neural Information Processing Systems conference. Each team is supplied with computer models of the human body and the world that body must navigate, including stairs and other surfaces. Teams also face internal challenges such as weak or unreliable muscles, and each team is evaluated based on how far its modeled human overcomes those obstacles in a fixed period. The researchers hope some teams will surmount all the various virtual obstacles imposed on them, while a longer-term goal is not only helping children with degenerative diseases, but also aiding the design of better-calibrated devices to assist with walking or carrying loads.

Full Article

An illustration by Christine Daniloff of a person sleeping New AI Algorithm Monitors Sleep With Radio Waves
MIT News
Anne Trafton
August 6, 2017

Researchers at the Massachusetts Institute of Technology (MIT) and Massachusetts General Hospital have developed a sleep monitoring system that uses an artificial intelligence algorithm to analyze radio signals around subjects and translate those measurements into sleep stages. The team previously developed radio-based sensors for remotely measuring vital signs and other health behaviors, built from a wireless device that emits low-power radio frequency signals. The algorithm is based on deep neural networks, and is designed to eliminate unwanted noise while preserving the sleep signals. "Our device allows you not only to remove all of these sensors that you put on the person, and make it a much better experience that can be done at home, it also makes the job of the doctor and the sleep technologist much easier," says MIT professor Dina Katabi. Their approach determined the technique was about 80-percent accurate, which is comparable to the accuracy of ratings ascertained by sleep specialists based on electroencephalogram measurements.

Full Article
Supercomputing Ocean Currents at NASA
Inside HPC
Rich Brueckner
August 5, 2017

Researchers from the National Aeronautics and Space Administration's (NASA) Advanced Supercomputing (NAS) facility at the Ames Research Center in Silicon Valley have developed a new visualization tool for the Estimating the Circulation and Climate of the Ocean project to study the behavior of ocean currents. The new visualization tool provides high-resolution views of the entire globe at the same time, enabling researchers to see new details that have been previously missed. The visualizations are displayed on a 128-screen hyperwall at the NAS facility, and researchers can switch the hyperwall view from one global image to views of single regions. In addition, the hyperwall can display properties such as temperature, surface wind stress, density, and salinity, all of which can be clearly identified with high-contrast colors that can be changed instantly. The visualization project is part of a collaboration between NASA's Ames Research Center, the Jet Propulsion Laboratory, and the Massachusetts Institute of Technology.

Full Article

A modified stop sign Slight Street Sign Modifications Can Completely Fool Machine Learning Algorithms
IEEE Spectrum
Evan Ackerman
August 4, 2017

It is possible to fool machine-learning visual classification algorithms by making slight image modifications, according to a paper by a multi-university research team. In one example, the team fooled a deep neural network-based classifier into thinking it was seeing a speed limit sign, and not a stop sign, all of the time. The novelty of this method is its basis on physical adversarial perturbations, such as altering real-world targets in such a manner as to reliably deceive neural network classifiers from multiple distances and angles while remaining discreet enough to be invisible to casual observers. Techniques the researchers applied to enable this manipulation included subtle fading, camouflage graffiti, and camouflage art. To develop these attacks, the team trained its own road sign classifier in TensorFlow using a publicly available, labeled dataset. The team then fed the image of the target signs as well as the classifier into an attack algorithm that outputs the desired adversarial image.

Full Article

A photo of the principal investigators for the Cornell Neurotechechnology NeuroNex Hub $9M Grant Will Create Neurotech Research Hub at Cornell
Cornell Chronicle
Syl Kacapyr
August 1, 2017

Researchers at Cornell University will receive a $9-million U.S. National Science Foundation grant to develop tools to provide neuroscientists with an unprecedented look into the inner workings of the brain. The grant will establish the Cornell Neurotechnology NeuroNex Hub to research, develop, and disseminate new optical imaging tools for noninvasive recording of neural activity in animals. The grant also will establish the Laboratory for Innovative Neurotechnology, where engineers and biologists will work together to develop and test the tools. The hub aims to overcome three barriers--deep imaging of intact brains, imaging of large and multiple neural regions, and faster imaging for volumetric recording. The hub will integrate the tools within five years, demonstrating the deepest high-resolution, large-scale neural activity recording ever achieved. "At the Cornell NeuroNex Hub, we will create, optimize, and then disseminate the new tools that will enable biologists to attack some of the impossible problems in neuroscience," says Cornell professor Chris Xu.

Full Article
Caltech, Cornell Create Creature-Cataloging Contest for Computers
Emily Velasco
August 2, 2017

Researchers at the California Institute of Technology (CalTech) and Cornell University have created the iNaturalist Challenge, a competition to produce the best machine-learning algorithm for identifying the world's plant and animal species. From April to July, anyone who wanted to compete in the challenge was given access to a database of 650,000 images featuring more than 5,000 species in categories that included, among others, protozoans, fungi, plants, arachnids, and mammals. Participants used that database to develop algorithms for automatic species identification. When the competition closed in July, the organizers had received 32 entries from teams and individuals. The winning algorithm was able to correctly identify species in a test database of 100,000 photos 80 percent of the time when given one guess for each photo, says CalTech researcher Grant Van Horn. He notes when the algorithm was given five guesses for every photo, the accuracy increased to 95 percent.

Full Article
CSU Computer Science Researchers Earn Federal Grant to Develop Cybersecurity Training Tool
Columbus State University News
August 1, 2017

Researchers at Columbus State University (CSU) have received a $174,000 U.S. National Security Agency (NSA) grant to develop an intelligent tool for rapid cybersecurity training and curriculum development. "We are building a tool that people across the nation can use to develop cybersecurity training, which guarantees compliance with government and industry standards for cybersecurity workforce development," says CSU professor Shuangbao Wang. He notes the tool will be cloud-based, with an expert system that can be accessed anywhere in the world. Wang says after the tool is completed, universities, governments, and private-sector organizations can use the solution to quickly develop training and curricula that otherwise would not be possible on account of a lack of experts, knowledge, and skills. The new grant comes in the wake of another NSA grant that CSU was given to develop and implement a course in cybersecurity education specifically designed for middle school students.

Full Article

Student splicing wires as part of diagnostic testing. Pitt Researchers Control Supercomputers in Space
Pittsburgh Post-Gazette
Courtney Linder
July 31, 2017

Researchers at the U.S. National Science Foundation's Center for High-Performance Reconfigurable Computing (CHREC) at the University of Pittsburgh's Oakland campus are conducting high-performance computing operations in space. CHREC's current project includes two space-based processors at the International Space Station that can capture Earth images in 10-second intervals. CHREC director Alan D. George notes these processors are reconfigurable computers, which means they can reshape their hardware to perform different functions once new software has been uploaded. George says this initiative and the following CHREC project will emphasize creating lighter, more energy-efficient computers that are tolerable to space and process data faster. He notes the project's mission to examine space imagery must balance those images' relevance with how much computing power is consumed by the downloading process. Ultimately, George says CHREC seeks to build autonomous space computers that can process collected information, determine which images are important, and automatically export them to Earth-based computers.

Full Article
August 2017 Issue of Communications of the ACM
ACM Insurance

Association for Computing Machinery

2 Penn Plaza, Suite 701
New York, NY 10121-0701

ACM Media Sales

If you are interested in advertising in ACM TechNews or other ACM publications, please contact ACM Media Sales or (212) 626-0686, or visit ACM Media for more information.

To submit feedback about ACM TechNews, contact: technews@hq.acm.org