Welcome to the August 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."
|
|
|
For Quick Coronavirus Testing, Israel Turns to Clever Algorithm
The New York Times
David M. Halbfinger
August 21, 2020
Israeli scientists have created a coronavirus testing procedure that they claim is faster and more efficient than any currently in use, testing samples in pools of up to 48 people simultaneously. The Pooling-Based Efficient SARS-CoV-2 Testing (P-Best) method requires only a single round of testing, based on an algorithm created by the Open University of Israel's Noam Shental. The P-Best algorithm optimizes pool design based on the expected prevalence of the virus, enabling all positive individuals in a batch to be localized, provided the total number of positives does not sharply surpass the expected number. Although the technique is less effective the higher a community's positivity rate is, it offers dramatically greater efficiency when rates are lower. P-Best accurately screened 1,115 healthcare workers with just 144 tests.
*May Require Paid Registration
|
White House Announces AI, Quantum Research Institutes
VentureBeat
Kyle Wiggers
August 26, 2020
The White House announced the creation of 12 new artificial intelligence (AI) and quantum information science research institutes, to be funded by federal agencies. The Trump Administration said the U.S. National Science Foundation will invest $100 million in five AI institutes over five years, in partnership with the Department of Agriculture (USDA)'s National Institute of Food and Agriculture, the Department of Homeland Security's Security Science and Technology Directorate, and the Department of Transportation's Federal Highway Administration. USDA will separately fund two institutes of its own, with focus areas including "user-driven trustworthy AI" for weather, climate, and coastal hazards applications, and theoretical challenges like neural architecture optimization. Meanwhile, the Department of Energy will invest $625 million in five quantum information science research centers, whose objectives will include surmounting obstacles in quantum state resilience, controllability, and scalability.
|
U.S. Army Robo-Teammate Can Detect, Share 3D Changes in Real Time
U.S. Army Research Laboratory
August 24, 2020
Researchers at the U.S. Army Combat Capabilities Development Command's Army Research Laboratory (ARL) have shown that robots equipped with LiDAR can detect physical changes in a three-dimensional, real-world environment, and share that information with a person in real time via augmented reality eyewear. The human observer then can evaluate the information and determine how to respond. Said ARL's Christopher Reardon, "This could let robots inform their soldier teammates of changes in the environment that might be overlooked by or not perceptible to the soldier, giving them increased situational awareness and offset from potential adversaries.” Robots equipped with LiDAR, Reardon said, “could detect anything from camouflaged enemy soldiers to IEDs.”
|
Do Cloth Masks Work? Supercomputer Fugaku Says Yes
Nikkei Asian Review
Yuki Misumi
August 25, 2020
Japan's Riken Institute said the Fugaku supercomputer, recently crowned the world's fastest, developed a model that showed nonwoven fabric masks block virus-laden respiratory droplets more effectively than cotton or polyester masks (although all three types were deemed effective at slowing the spread of the coronavirus). The system simulated the performance of the three types of fabric masks in blocking the spray of virus-carrying respiratory droplets from coughing by the wearer, demonstrating that all three types stopped at least about 80% of spray. The team also simulated a virus spreading through a 2,000-seat auditorium, and found little danger of proliferation if visitors are masked and sitting spaced apart.
|
Elephants vs Trains: AI Helps Ensure They Don't Collide
ZDNet
Anna Solana
August 17, 2020
Researchers at Spain’s Polytechnic University of Catalonia (UPC)-BarcelonaTech developed a "smart ear" to help prevent fatal collisions involving elephants and trains in India, where trains on the Siliguri-Jalapaiguri railway line have struck and killed more than 200 elephants over the last decade. The researchers placed recorders and cameras on train tracks in West Bengal, then analyzed the recorded sounds using machine learning techniques to differentiate elephant sounds from others. As a result, the “smart ear” was able to identify elephant sounds from distances of 1 kilometer, and images of elephants from 250 meters, during daylight hours. When an elephant is detected, the system can send real-time alerts to train drivers’ phones. Said UPC's Michel André, "We should listen to wildlife to detect their presence and avoid confrontation to achieve a better coexistence between man and nature. Today's bioacoustics can meet that challenge."
|
New Technique to Prevent Medical Imaging Cyberthreats
Ben-Gurion University of the Negev (Israel)
August 25, 2020
Researchers at Israel's Ben-Gurion University of the Negev (BGU) have developed a new artificial intelligence method to shield medical imaging devices from cyberattacks, and from human error involving anomalous instructions sent from a host computer. BGU's Tom Mahler and colleagues designed a dual-layer architecture to protect devices from both context-free (CF) and context-sensitive (CS) anomalous instructions, and tested the framework on computed tomography instructions. The team assessed the CF layer using 14 unsupervised anomaly detection algorithms, then evaluated the CS layer for four types of clinical objective contexts, using five supervised classification algorithms for each context. The researchers found adding the CS layer increased anomaly detection performance from an F1 score of 71.6% to between 82% and 99%.
|
Predicting the Lifespan of an App
IEEE Spectrum
Michelle Hampson
August 24, 2020
Northwestern Polytechnical University researchers have developed a model for predicting the lifespan of an app. They used a Multi-Task Learning approach to develop the model, analyzing factors like download history, ratings, and reviews at different time intervals. The resulting model, AppLife, predicts the likelihood of an app being removed from app stores in the next year or two. The researchers evaluated AppLife using a dataset of more than 35,000 apps from the Apple Store released in 2015 and available in 2016. Said the university's Bin Guo, "Experiments show that our approach outperforms seven state-of-the-art methods in app survival prediction."
|
Tool Transforms World Landmark Photos Into 4D Experiences
Cornell Chronicle
Melanie Lefkowitz
August 25, 2020
Cornell University researchers used publicly available tourist photos of world landmarks to develop a method of generating maneuverable three-dimensional (3D) images that reflect changes in appearance over time. The technique utilizes deep learning to absorb and synthesize thousands of mainly untagged and undated images, to address the challenge of giving the models photorealism and plenoptic function (showing how something looks from every possible viewpoint in space and time). The researchers devised Deep Multiplane Images scene representation to interpolate appearance in four dimensions (three physical dimensions over time), and demonstrated the model could be taught to create such a four-dimensional scene from roughly 50,000 publicly available images from sites like Flickr and Instagram. Said Cornell Tech’s Noah Snavely, “We use the same idea invented for creating 3D effects in 2D animation to create 3D effects in real-world scenes, to create this deep multilayer image by fitting it to all these disparate measurements from the tourists’ photos.”
|
UBCO Researcher Uses Computer Modeling to Predict Reef Health
UBC Okanagan News (Canada)
August 25, 2020
Bruno Carturan of Canada's University of British Columbia Okanagan partnered with Australia's Flinders University and research firm Nova Blue Environment to develop a computer modeling technique for predicting coral reef health. Agent-based models based on decades' worth of collected data enabled scientists to manipulate the initial diversity of coral in a virtual environment, and to determine how the virtual communities respond to different threats. Flinders' Corey Bradshaw said running simulations repeatedly allows the model to identify trait combinations offering optimized resilience, which will help ecologists design reef management and restoration strategies from those predictions. Said Bradshaw, "This high-resolution coral 'video game' allows us to peek into the future to make the best possible decisions and avoid catastrophes."
|
Sussex Study Enables Predicting Computational Power of Early Quantum Computers
University of Sussex (U.K.)
Anna Ford
August 24, 2020
Quantum physicists at the U.K.'s University of Sussex have developed an algorithm which they say can help early quantum computers perform calculations most efficiently. The routing algorithm manages traffic within the computer similar to how city traffic is regulated, avoiding "traffic jams" in the transport of quantum bits (qubits). The team used this model to predict the computational power of early quantum systems, and applied the Quantum Volume benchmark to compare its trapped-ion framework against an architecture for superconducting qubits. Their model's performance consistently outperformed superconducting qubits because the routing algorithm enables qubits to directly interact with many more qubits, yielding higher expected computational power. Said Sussex's Mark Webber, "Our study indicates a fundamental advantage for trapped-ion devices, and the new routing algorithm will allow us to maximize the performance of early quantum computers."
|
ML Reveals Role of Culture in Shaping Meanings of Words
Princeton Engineering News
Rachel Nuwer
August 14, 2020
Researchers from Princeton University, the University of Wisconsin, Madison, and the U.K.'s University of Bristol conducted a machine learning study of dozens of languages, and found that culture, history, and geography wield a sizable influence on word meanings. Machine learning models analyzed more than 1,000 words in 41 languages, and an algorithm examined neural networks trained on various languages to compare millions of semantic associations. Another algorithm compared similarities of the languages' root cultures based on an anthropological dataset, and could predict the ease of translation between languages according to those similarities. Said Princeton’s William Thompson, “One way to look at what we’ve done is a data-driven way of quantifying which words are most translatable.”
|
3D-Printed Device Demonstrates Enhanced Capture of CO2 Emissions
Oak Ridge National Laboratory
August 19, 2020
Researchers at the U.S. Department of Energy's Oak Ridge National Laboratory (ORNL) used additive manufacturing (three-dimensional/3D printing) to create an aluminum device that improves the process of capturing carbon dioxide emissions from factories and industrial plants by removing excess heat. The circular device, made with a 3D printer, integrates a heat exchanger with a mass-exchanging contactor inside an absorption column featuring seven commercial stainless-steel packing elements. The device was installed between the packing elements in the top half of the column. Said ORNL's Xin Sun, "With 3D printing, the mass exchanger and heat exchanger can co-exist within a single multifunctional, intensified device."
|
|
