Welcome to the October 26, 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."
|
|
|
Remote Work Not Just for White-Collar Jobs Anymore
The Wall Street Journal
Christopher Mims
October 22, 2020
Technology that allows people to perform physical work remotely is proliferating across a host of occupations. In Tokyo, workers miles away control humanoid robots to stock convenience store shelves via a virtual reality interface; the robots, built by Telexistence, are the world's first commercial application of telepresence. Telexistence's Model T robot is built from relatively inexpensive parts, to ensure the teleoperated machine and remote worker cost less than an equivalent physically present human. The company hopes a system fully puppeteered by humans will collect sufficient training data to teach an artificial intelligence to assume at least some of their tasks. Another form of telepresence lets front-line workers use smart glasses to share what they are seeing with remote experts, who can consult and send images to assist with problems.
*May Require Paid Registration
|
Simple Software Creates Complex Wooden Joints That Interlock with No Nails, Glue, or Tools
SciTechDaily
October 21, 2020
Researchers at Japan's University of Tokyo have created novel three-dimensional (3D) design software for producing sophisticated, interlocking wooden joints that do not require nails, glue, or tools. With the Tsugite software, users with little or no prior experience in woodworking or 3D design can create designs for functional wooden structures in minutes. The tool provides a detailed view of wooden joints represented by voxels, in the form of small cubes that can be manipulated at one end of a component for joining. Changes to the end of one piece automatically adjusts the voxels at the end of the corresponding component so they interlock tightly. By incorporating fabrication limitations into the design process, Tsugite's underlying algorithms have less difficulty considering all possible designs to present to users, as those that are physically impossible are eliminated.
|
Robots Deciding Their Next Move Need Help Prioritizing
University of Illinois at Urbana-Champaign Aerospace Engineering
October 20, 2020
Researchers at the University of Illinois at Urbana-Champaign (UIUC) used a model based on the game Capture the Flag to develop a deep reinforcement learning technique that can help robots weigh their next move. UIUC's Huy Tran said his team noticed the robots needed assistance prioritizing. A hierarchical deep reinforcement learning scheme divides tasks so the model can tackle more complex challenges. Tran said, "We trained a high-level decisionmaker who assigns a sub-task for each agent to focus on," and the hierarchical framework simplifies updating. He added, “This approach has the potential to solve interesting and challenging problems, but there are a lot of issues that we still need to address before we can deploy these systems in real-world situations.”
|
Cybersecurity Company Finds Hacker Selling Info on 186 Million U.S. Voters
NBC News
Ken Dilanian
October 21, 2020
Cybersecurity company Trustwave says it discovered a hacker selling personally identifying information on 245 million Americans, including voter registration data for 186 million. Trustwave found the hacker, calling himself Greenmoon2019, by trawling "dark web" forums for threat information; the firm used fictitious identities to induce Greenmoon2019 to provide more information, including a bitcoin wallet for collecting payment. The hacker also used other stolen data to pair email addresses with voter rolls and sell it as a package. Trustwave has turned these findings over to the U.S. Federal Bureau of Investigation. Trustwave's Ziv Mador said, "In the wrong hands, this voter and consumer data can easily be used for geotargeted disinformation campaigns over social media, email phishing, and text and phone scams before, during, and after the [presidential] election, especially if results are contested."
|
ML Uncovers Potential TB Drugs
MIT News
Anne Trafton
October 15, 2020
Researchers at the Massachusetts Institute of Technology used a new machine learning approach to identify promising drug compounds that target a protein required by the bacteria that causes tuberculosis. The researchers used the Gaussian process to assign uncertainty values to the data used to train the algorithm. The model was trained on a dataset of 72 small molecules and their interactions with more than 400 protein kinases. It then analyzed nearly 11,000 small molecules, many different from those in the training data. The researchers identified molecules with very strong predicted binding affinities for the protein kinases put into the model. About 90% of the molecules assigned the highest certainty by the model proved to be true hits, compared to the hit rate of 30% to 40% for existing machine learning models used for drug screens.
|
MonoEye: A Human Motion-Capture System Using Single Wearable Camera
Tokyo Institute of Technology (Japan)
October 21, 2020
Researchers at Japan's Tokyo Institute of Technology and Carnegie Mellon University have developed a human motion-capture system comprised of an ultra-wide fisheye camera worn on the user's chest. The MonoEye system can capture the user's body motion and their perspective, or "viewport," with a 280-degree field of view. MonoEye incorporates three deep neural networks for real-time calculation of three-dimensional body pose, head pose, and camera pose. The researchers trained the networks on a synthetic dataset of 680,000 renderings of people with a range of body shapes, apparel, actions, background, and lighting conditions, along with 16,000 frames of photorealistic images.
|
2020's Fastest-Rising Tech Jobs? Programming Language PHP Leads the Way
ZDNet
Liam Tung
October 22, 2020
Job search engine Indeed said entry-level PHP developer is the fastest-growing technology job industry-wide, with employment listings for that job title having risen 834% since January. Coder question/answer site Stack Overflow added that PHP remains one of the most commonly utilized scripting languages for building websites and Web applications, even though it also is one of the "most-dreaded" languages to use. Indeed estimated the U.S. national average salary for an entry-level PHP developer with one to five years' experience is $73,334. Junior helpdesk operators and entry-level PC technicians also are in high demand, with Indeed job listings for those job titles growing 501% and 434%, respectively, since January.
|
AI Outperforms Humans in Speech Recognition
Karlsruhe Institute of Technology (Germany)
October 20, 2020
An artificial intelligence (AI) system developed at Karlsruhe Institute of Technology (KIT) in Germany outperformed humans in transcribing conversations containing spontaneous speech, with minimum latency. It traditionally has been more difficult for AI to accurately transcribe conversations that include stops, stutterings, hesitations, laughs, coughs, and words not pronounced clearly. The Lecture Translator system has been used in KIT's lecture halls since 2012 to translate lectures from German or English into the languages spoken by foreign students. KIT's Alex Waibel says highly accuracy and fast speech recognition is essential for further downstream processing of speech, as it allows dialog, translation, and other modules of the AI system to provide better voice-based interaction.
|
Alphabet Trialing Solar-Paneled, Robotic Buggies to Inspect Crops
CNBC
Anmar Frangoul
October 15, 2020
Google parent Alphabet is piloting a project to revolutionize agriculture and food production. The company’s Mineral initiative aims to use solar-powered electric buggies to travel across fields and locate plants using global positioning system software; cameras and other "machine perception tools" then collect crop data. The system combines the robot-acquired data with information on weather and soil health, in order to "help breeders understand and predict how different varieties of plants respond to their environments." Said Alphabet’s Elliot Grant, “Just as the microscope led to a transformation in how diseases are detected and managed, we hope that better tools will enable the agriculture industry to transform how food is grown.”
|
United Airlines Testing Global Health App on U.K.-U.S. Flight in Effort to Reopen Borders
Reuters
Tracy Rucinski
October 21, 2020
United Airlines on Wednesday last week planned to begin testing of a digital health pass application on a flight from Britain to the U.S. The testing was to take place under a global pilot program to set up a common international standard for Covid-19 test results, and ultimately to provide vaccine records that could help reopen borders. The application requires volunteers to upload coronavirus test results from a certified laboratory to their smartphones, and to complete health screening questions, to generate a verified quick response code that United staff and border officials can scan upon passengers’ arrival and before they leave. Paul Meyer with Swiss non-profit The Commons Project said the goal of the trial “is to demonstrate to governments that they can rely on someone getting tested in one country and present their credentials in another country.”
|
Earphone Tracks Facial Expressions, Even with Face Mask
Cornell Chronicle
Melanie Lefkowitz
October 12, 2020
Cornell University researchers have developed an earphone that can continuously track full facial expressions by observing the contour of the wearer’s cheeks, and converting such perceived expressions into emojis or silent speech commands. The C-Face device would allow users to express emotions to online collaborators without cameras in front of their faces. The device features a miniature RGB camera oriented below each ear with headphones or earphones; the cameras record changes in facial contours when facial muscles move. The captured images are reconstructed by computer vision and a deep learning model, using a convolutional neural network. Cornell's Cheng Zhang said C-Face can capture expressions even when users are wearing masks.
|
Scientists Develop ‘Mini-Brains’ to Help Robots Recognize Pain, Self-Repair
Nanyang Technological University (Singapore)
October 15, 2020
Researchers at Nanyang Technological University, Singapore (NTU Singapore) have developed an approach that uses artificial intelligence (AI) to enable robots to recognize “pain” and self-repair minor damage in real time. This involves embedding AI into the network of sensor nodes, which are connected to multiple processing units and act like "mini-brains" on the robot’s skin. The robot can respond to pressure exerted by a physical force, and an ion gel material allows the robot to restore its mechanical functions without human intervention when cut by a sharp object. Said NTU Singapore's Nripan Mathews, “Our team has taken an approach that is off the beaten path by applying new learning materials, devices, and fabrication methods for robots to mimic the human neuro-biological functions.”
|
AI Also Has Illusory Perceptions
RUVID/Network of Valencian Universities for the Promotion of Research, Development, and Innovation
October 16, 2020
Researchers at Spain’s Universitat de València (UV) and Pompeu Fabra University have found that convolutional neural networks (CNN) are affected by visual illusions, much like the human brain. The researchers trained CNNs for simple tasks and found they were susceptible to visual illusions of brightness, although the illusions may not coincide with biological illusory perceptions. Said UV's Jesús Malo, "This is one of the factors that leads us to think that it is not possible to establish analogies between the simple concatenation of artificial neural networks and the much more complex human brain." The researchers warned in a separate study about the use of CNNs to study human vision. Said Malo, "In addition to the intrinsic limitations of these artificial networks to model vision, the non-linear behavior of flexible architectures can be very different from that of the biological visual system."
|
|
