Welcome to the January 13, 2020 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."
|
|
Paralyzed Man Breaks World Record for Finishing a Marathon in an Exoskeleton Suit
CNN Alaa Elassar January 13, 2020
A man paralyzed from the waist down has broken the world record for finishing a marathon in an exoskeleton suit. Competing in the 2020 Charleston (SC) Marathon, Adam Gorlitsky finished the race with a time of 33 hours, 50 minutes, and 23 seconds, unofficially beating a British man who completed the 2018 London Marathon in 36 hours and 46 minutes, according to Guinness World Records. Gorlitsky was in a car accident that left him paralyzed from a severe spinal cord injury in December 2005. Ten years later, he was able to stand and walk using a ReWalk Robotic Exoskeleton. In 2016, Gorlitsky became the first paralyzed man to walk the Cooper River Bridge Run in Charleston using a robotic exoskeleton, and founded the non-profit I GOT LEGS to improve the lives of the disabled.
|
From Surgery Simulators to Medical Mishaps in Space, Video Game Tech Is Helping Doctors at Work
The Washington Post Elise Favis January 9, 2020
Video game developers are creating technologies to support doctors performing surgery and other medical procedures, with the Osso VR company offering surgeons on-demand information and training through virtual reality (VR) headsets. Osso VR allows trainees to refine their surgical skills by operating on virtual patients, and a multiplayer component lets surgeons collaborate in VR; the program also features embedded analytics to objectively measure a user's surgical performance. Meanwhile, game studio Schell Games is developing a mobile game to train emergency room doctors, with the University of Pittsburgh's Deepika Mohan. Developer Level Ex offers free hyper-realistic simulations for on-the-go engagement with doctors, including one designed to train for medical procedures in space by modeling the body's physiological and anatomical changes in low gravity.
*May Require Paid Registration
|
Microsoft Chat Scanner to Detect Child Sex Predators
Bloomberg Dina Bass January 9, 2020
Microsoft says it will release a tool that can scan online text chats to identify child sex predators. Microsoft's Courtney Gregoire said Project Artemis, which the company uses on its Xbox gaming service, sifts through historical messages for relevant patterns and traits, then assigns a probability rating. Organizations can use this rating to decide which conversations on their platforms should be more closely scrutinized by a human moderator. The tool emerged from a 2018 hackathon, and Microsoft has been developing Project Artemis in partnership with the developers of the online video game Roblox and the Kik messenger app, social meeting app owner The Meet Group, and the Thorn anti-child-sex-abuse nonprofit. Dartmouth College's Hany Farid guided the tool's development.
*May Require Free Registration
|
Algorithms Help Find Minimum Energy Paths, Saddle Points More Effectively
Aalto University January 8, 2020
A researcher at Finland's Aalto University developed and tested machine learning algorithms based on Gaussian process regression to improve searches for minimum energy paths and saddle points, an arduous challenge in theoretical chemistry. Accurately assessing energy and forces for each atomic configuration consumes much computing power; Aalto's Olli-Pekka Koistinen designed algorithms to reduce the number of observation points and costly energy evaluations required to compute energy paths and saddle points. The process traditionally involves iterative techniques that follow along an energy surface, but Koistinen's method allows use of earlier observations to model the surface and reach objectives with fewer iterations. Said Koistinen, “This is another example of a research topic in which machine learning methods can be helpful.”
|
Bosch Deploys AI to Prevent Attacks on Car Electronics
The Wall Street Journal James Rundle; John McCormick January 6, 2020
Engineering company Robert Bosch is deploying artificial intelligence (AI) to fortify cars' electronic systems against hackers who attempt to feed the systems intentionally incorrect road-sign information. Road-sign standardization makes traffic-sign recognition technology well-suited to machine learning and deep learning image-identification algorithms, but malefactors can deceive the algorithms by defacing the signs. Bosch’s Michael Bolle said the company has unveiled a computer-vision-based AI process designed to analyze and compare an object from two different perspectives. The findings of deep learning algorithms that identify road signs are checked by computer-vision algorithms, and discrepancies between the readings could indicate spoofing. Said Bolle, "In the area of machine learning and AI, products and machines learn from data, and so the data itself can be part of the attack surface."
*May Require Paid Registration
|
'Holy Grail' Digital Effects Rewinding the Clock for Actors
Associated Press Matt Kemp January 12, 2020
Recent movies have made extensive use of digital visual effects to make performers appear younger. "The Irishman," for example, is notable for avoiding use of tracking markers, the dots painted on actors' faces to let computers to track and replicate their facial movements. Pablo Helman at Industrial Light and Magic enabled this innovation using a rig with a central director camera and two lateral witness cameras to capture infrared footage. This allowed the elimination of shadows, which could interfere with the geometric facial shapes constructed by de-aging software, from on-set lighting. Said Helman, “You're not interrupting the director's thread of thinking. You're not changing the light on set, but the computer can see in a different spectrum.”
|
Using a Robot to Deploy Robots in Remote Oceans
University of East Anglia, U.K. January 10, 2020
Researchers at the University of East Anglia in the U.K.'s Centre for Ocean and Atmospheric Sciences designed an unmanned surface vessel adapted to deploy research equipment in remote and inaccessible ocean locations. The five-meter-long, sensor-equipped AutoNaut is powered by solar panels and can withstand heavy seas. It has been adapted to carry and release the underwater Seaglider, which is equipped with sensors to collect data for research on ocean processes important for climate. Seagliders can reach depths of 1,000 meters and travel thousands of kilometers over a period of months. Said East Anglia’s Karen Heywood, "Having sensors on both the AutoNaut and the Seaglider means we'll get simultaneous meteorological and oceanographic measurements, giving us a more detailed picture of conditions at the study site and helping to further our understanding of factors affecting our climate."
|
Hundreds of Millions of Cable Modems Vulnerable to Cable Haunt Vulnerability
ZDNet Catalin Cimpanu January 10, 2020
Danish researchers reported a security vulnerability affecting cable modems that use Broadcom chips, which is believed to impact about 200 million cable modems in Europe alone. The Cable Haunt flaw is within the chips' spectrum analyzer, a hardware/software component that shields the modem from signal surges and cable-originating interruptions. Internet service providers (ISPs) frequently use the analyzer in debugging connection quality, and the researchers warned the component is not protected against Domain Name System rebinding attacks; the analyzer also employs default credentials and hosts a firmware-based coding error. The researchers said hackers can use a browser to communicate exploits to and execute commands on the analyzer by deceiving users into accessing a malicious page. The researchers have established a dedicated Cable Haunt website to encourage ISPs to test their devices and issue firmware updates to patch the attack vector.
|
Panasonic Cloud Analytics Will Give Cars a Guardian Angel
IEEE Spectrum Tekla S. Perry January 6, 2020
Panasonic engineers, in conjunction with the departments of Transportation in Colorado and Utah, are working on a digital seatbelt that uses technology enabling cars to send information, like speed and direction, to transportation infrastructure, which would be able to inform drivers about construction zones, traffic jams, and other impending hazards. Panasonic's cloud-based Cirrus system, which the company hopes to launch later this year, would go further by using more data and processing it centrally, regardless of where it originates. Panasonic will feed data collected from its implementations in Colorado and Utah into machine-learning programs to improve the algorithms' ability to predict changing or hazardous road and traffic conditions. The company said it will make the system available to app developers. Said Panasonic’s Chris Armstrong, "We are building a central nervous system for connected cars."
|
OSU Study Shows How to Reduce Physical Risk in VR
Oregon State University News Molly Rosbach January 7, 2020
A study by Oregon State University (OSU) and Northern Illinois University researchers evaluated the risks of musculoskeletal injuries in virtual reality (VR) in order to develop guidelines to reduce those risks by redesigning interfaces. The researchers recorded movements and electromyography to measure muscular electrical activity as study participants performed common VR gestures. Tests with visuals positioned at, above, and below eye level showed extending the arm straight out quickly induces shoulder strain, which could lead to problems like rotator cuff injuries with prolonged use—while a hefty VR headset could worsen neck strain. OSU's Jay Kim said, "Objects that are being interacted with more often should be closer to the body. And objects should be located at eye level, rather than up and down."
|
Machine Learning Shapes Microwaves for Computer's Eyes
Duke Pratt School of Engineering Ken Kingery January 9, 2020
Engineers at Duke University and France’s Institut de Physique de Nice have developed a more accurate, efficient machine learning technique for identifying objects via microwaves. A metamaterial antenna sculpts a microwave wave front into various configurations; the pattern reflects off the object to be identified and migrates back to a similar antenna, and the computer analyzes the transmission and tries to recognize the object. The method directly analyzes pure data, and ascertains optimal hardware settings that reveal the most valuable data while concurrently learning what that data is. Nice's Philipp del Hougne said, "Microwaves are ideal for applications like concealed threat detection, identifying objects on the road for driverless cars, or monitoring for emergencies in assisted-living facilities. When you think about all of these applications, you need the sensing to be as quick as possible."
|
|