Welcome to the August 17, 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."
|
|
|
AI-Driven Cacophonic Choir Amplifies Voices of Sexual Assault Survivors
CNET
Leslie Katz
August 12, 2020
The 2020 ACM SIGGRAPH conference's online art gallery will showcase an artificial intelligence-powered interactive sound installation designed to amplify the accounts and voices of sexual assault survivors. Cacophonic Choir's creators include Colby College computer science professor Hannah Wolfe, artist and architect Solen Kiratli, and musician Alex Bundy. The installation emits nine individual voices, produced by a machine learning algorithm designed and trained on the anonymous stories of more than 500 victims. The Choir blends real-time computing, proximity sensors, audio processing, and three-dimensional printing for generating sculpted shapes that burst from translucent spherical silicone membranes. To comment on mass media's effect of distorting survivors' experiences, Cacophonic Choir's audio sounds intentionally jumbled from far off, but becomes more coherent and personal closer in.
|
How a 30-Ton Robot Could Help Crops Withstand Climate Change
The Wall Street Journal
Jacob Bunge
August 12, 2020
Data scientists at George Washington University (GWU) and St. Louis University are using a 30-ton, 70-foot-tall robotic field scanner in Arizona to analyze plant genetics, in the hope of developing crops resistant to climate change. The Field Scanalyzer traverses rows in a strip of an irrigated field a few centimeters each second, using wheels on tracks. An array of electronic eyes monitors crops' growth rate, height, development, and hardiness, and transmits up to 10 terabytes of data daily to computers and Illinois and Missouri. Researchers are training those computers to identify links between genes and plant characteristics, in order to help plant breeders recognize traits and genetic markers indicating the most efficient and resilient crop varieties. GWU's Robert Pless added that scientists are using the datasets to pilot artificial intelligence tools, for potential research into human aging or disease progression.
|
UCLA Computer Scientists Set Benchmarks to Optimize Quantum Computer Performance
UCLA Samueli School of Engineering
August 13, 2020
Computer scientists at the University of California, Los Angeles (UCLA) have developed a family of benchmark quantum circuits with known optimal depths or sizes. These benchmarks could be used by quantum computer designers to improve design tools that could be used to determine the best circuit design. They found that improving quantum compilation design could help achieve computing speeds that are 45 times faster than existing speeds. The benchmarks, dubbed QUEKO, are open source and available on the GitHub software repository. Said UCLA's Jason Cong, "[W]e expect these benchmarks to motivate both academia and the industry to develop better layout synthesis tools, which in turn will help drive advances in quantum computing."
|
Developer Jobs: Demand for Programming Language Python Falls Amid Pandemic
ZDNet
Liam Tung
August 12, 2020
Online developer hiring platform HackerRank reports slackening demand for Python coding language skills since March 1 amid the coronavirus pandemic. Job interviews for Python developers and front-end engineers have declined 27% and 34% respectively, while full-stack engineers are seeing 5% fewer interviews. These numbers apparently align with U.S. recruitment website Indeed's estimates, which indicated a 43% year-over-year drop in new postings for data-scientist positions in July. On a positive note, HackerRank found interviews for Web and mobile application developers have climbed 61% since March. HackerRank CEO Vivek Ravisankar cited soaring demand for back-end, JavaScript, and .NET developers as online merchants and gaming companies modify their algorithms and upgrade user interfaces, while legacy industries like banking and healthcare are almost completely starting from scratch.
|
NSA, FBI Expose Russian Intelligence Hacking Tool
Reuters
Christopher Bing
August 13, 2020
The U.S. National Security Agency (NSA) and Federal Bureau of Investigation (FBI) have publicly exposed a Russian hacking tool. Russia's Main Intelligence Directorate apparently used the "Drovorub" malware to penetrate Linux-based computers, which Keppel Wood at NSA's Cybersecurity Directorate said are pervasively employed by National Security Systems, the U.S. Department of Defense, the defense industrial base, and the at-large cybersecurity community. Steve Grobman at cybersecurity company McAfee said, "Drovorub is a 'Swiss Army knife' of capabilities that allows the attacker to perform many different functions, such as stealing files and remote-controlling the victim's computer." The report on Drovorub is the latest in a series of public disclosures by the U.S government targeting Russian hacking operations ahead of the 2020 presidential election.
|
Android Is Now World's Largest Earthquake Detection Network
Ars Technica
Ron Amadeo
August 11, 2020
Google announced "the world's largest earthquake detection network" by adding quake detection features to almost all Google Play Android phones. The Android Earthquake Alerts System taps the smartphones' accelerometers, so any suspicious tremors prompt the handsets to transmit data to Google's earthquake detection server, which blends their information to determine if an earthquake is occurring. Google is collaborating on the California rollout with the coalition behind ShakeAlert, the back-end system that the client of the MyShake earthquake detection application surfaces to users. ShakeAlert integrates smartphone readings with a network of seismometers, and Android phones will add their data to the system and show alerts. Google said, "To start, we'll use this technology to share a fast, accurate view of the impacted area on Google Search. When you look up 'earthquake' or 'earthquake near me,' you'll find relevant results for your area, along with helpful, credible resources on what to do after an earthquake."
|
Tool Created for Discerning Fake News
UT News
August 13, 2020
A study from researchers at the University of Texas at Austin and Indiana University demonstrated novel tools for flagging fake news on social media platforms—a stop sign icon and the statement "Declared Fake by 3rd Party Fact-Checkers." The team first tested the interventions for one second and five seconds. One second is sufficient to induce an automatic gut reaction, and five seconds can record the effect of critical thinking. This was accompanied by a brief explanation of the warnings, while a second test used both warnings together, with users being trained partway through the study. Some participants received awareness training while others did not. The researchers found that the combination of the stop sign, statement, and awareness training had the biggest impact.
|
Peregrine Swoops on Flaws in 3D Printing
The Engineer
August 17, 2020
Researchers at the U.S. Department of Energy's Oak Ridge National Laboratory (ORNL) have developed an artificial intelligence software package for powder bed three-dimensional printers that performs real-time part-quality analysis. Peregrine is a novel convolutional neural network that employs a custom algorithm to process pixel values of images, accounting for composition of edges, lines, corners, and textures. Peregrine alerts operators upon detecting a flaw that may affect part quality, in order that they implement appropriate adjustments. The software generates a common image database that can be transferred to new machines to rapidly train new neural networks, while operating on a single high-powered laptop or desktop. Peregrine is being tested on multiple printers at ORNL, and forms part of the Transformational Challenge Reactor Demonstration Program pursuing the world's first additively manufactured nuclear reactor.
|
How Data Mining Visualizes Story Lines in the Twittersphere
Discover
August 13, 2020
University of Vermont researchers have created a searchable dataset of more than 100 billion tweets in over 150 languages containing more than 1 trillion one-word, two-word, and three-word sequences. The accompanying Storywrangler data visualizer displays the popularity of any words or phrases based on the number of times they have been tweeted and retweeted, and how this fluctuates over time. Storylines also can be measured against other databases to yield finer insights and analysis, facilitating a new form of societal examination, potentially with predictive outcomes. These storylines also possess social and cultural relevance. The researchers said, "Our collective memory lies in our recordings—in our written texts, artworks, photographs, audio, and video—and in our retellings and reinterpretations of that which becomes history."
|
Cutting Vehicle Emissions and Inspections via IoT
Carnegie Mellon University
Dan Carroll
August 10, 2020
With automobiles increasingly being integrated into the Internet of Things, Carnegie Mellon University (CMU) researchers have combined remote data transmission and machine learning to identify over-emitting vehicles, which could boost the effectiveness of emissions inspection and maintenance (I/M) programs. Current I/M technology is likely to correctly detect over-emitting 87% of the time, but has a 50% false pass rate compared to tailpipe emissions testing. CMU's Prithvi Acharya, Scott Matthews, and Paul Fischbeck developed a method that sends data directly from the vehicle to a local state- or county-based cloud server, eliminating regular inspection visits. Machine learning algorithms would process the data to identify trends and codes common within over-emitting vehicles. The team's model for identifying likely over-emitting vehicles is 24% more accurate than current on-board diagnostics systems.
|
Researchers Discovered Significant Vulnerability in Amazon's Alexa
The Hill
Chris Mills Rodrigo
August 13, 2020
Researchers at cybersecurity provider Check Point have issued a report detailing a vulnerability in Amazon's Alexa virtual assistant that would have allowed potential hackers to hijack the devices using malicious Amazon links. The flaw, which was patched in June, would have enabled hackers to install or remove "Skills" – essentially apps – from the devices once those links were clicked. Hackers also would have gained access to the user's voice history and personal information, including banking data and home address. Such vulnerabilities could pose major privacy risks, given that more than 200 million Alexa-enabled devices were sold by the end of last year. An Amazon spokesperson said the company fixed the issue as soon as it became known and is not aware of any instances in which customer information was exposed.
|
Baby-Mounted Cameras Teach AI to 'See' Through Child's Eyes
New Scientist
Chris Stokel-Walker
August 11, 2020
New York University (NYU) researchers used video footage captured by head-mounted cameras worn by children throughout their first three years of life to train an artificial intelligence (AI) neural network to extract meaning from the video. Using the SAYCam dataset, the AI learned to recognize objects that appear repeatedly, often by extending its focus beyond the objects. NYU's Brenden Lake said the algorithm does not identify objects as a child would, but these findings offer "a proof of concept that [visual features] are learnable with enough naturalistic data." He acknowledged that AI requires a significant amount of data and labels in order to perform tasks with child-level adeptness. However, algorithms that can be taught to learn as children do could potentially become more intuitive.
|
Super-Resolution Method Reveals the Fine Details
Texas A&M Today
Vandana Suresh
August 13, 2020
Researchers at Texas A&M University, Virginia Commonwealth University, and the U.S. Air Force Research Laboratory have developed an image processing technique to improve the quality of low-resolution electron micrographs without compromising the integrity of specimen samples. They were able to further enhance details in lower-resolution images by training deep neural networks on pairs of images from the same sample at different physical resolutions. Said Texas A&M's Yu Ding, "[W]ith our image processing techniques, we can super resolve an entire image by using just a few smaller-sized, high-resolution images. This method is less destructive since most parts of the specimen sample needn't be scanned with high-energy electron beams." The researchers found that their algorithm could enhance hard to discern features in low-resolution images by up to 50%, though the technique requires a significant amount of processing power.
|
|
