Welcome to the September 13, 2023, 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 Model Speeds Up High-Resolution Computer Vision
MIT News Adam Zewe September 12, 2023
An international team led by researchers at the Massachusetts Institute of Technology (MIT) has developed a computer vision model series able to perform semantic segmentation (categorizing each pixel in a high-resolution image) in real time with limited hardware resources. The new model series, dubbed EfficientViT, can achieve the same or better accuracy and is up to nine times faster than existing state-of-the-art semantic segmentation models due to its linear similarity function and hardware-friendly architecture. By rearranging the order of operations, EfficientViT can reduce total calculations without sacrificing functionality or losing the global receptive field. Said MIT's Song Han, "We can get the best of both worlds and reduce the computing to make it fast enough that we can run it on mobile and cloud devices." The model could help enhance real-time decision-making by autonomous vehicles.
|
China Sows Disinformation About Hawaii Fires Using New Techniques
The New York Times David E. Sanger; Steven Lee Myers September 11, 2023
Researchers at Microsoft, the University of Maryland, and other organizations found China's government is utilizing new methods to promulgate disinformation about last month's wildfires on Maui, claiming they resulted from tests of a secret "weather weapon." Such content includes photos apparently produced by artificial intelligence to add plausibility to Beijing's false narrative. The campaign seems to indicate China has shifted tactics from intensifying state propaganda to actively spreading discord in the U.S. The researchers suggested China was amassing a network of accounts that could be leveraged in future information (or disinformation) campaigns for "amplifying conspiracy theories that are not directly related to some of their interests, like Taiwan," said Brian Liston at cybersecurity company Recorded Future.
*May Require Paid Registration
|
Virtual Drug Quiets Noise in Images of Heart Tissue
Washington University in St. Louis McKelvey School of Engineering Beth Miller September 11, 2023
An algorithm created by Washington University in St. Louis (WashU) researchers can eliminate motion from images of expanding and contracting heart cells and tissues by emulating drug-administered heart stoppage. The algorithm, which virtually replicates the effect of cardiac muscle-inhibiting drug blebbistatin, was developed from a program that estimates displacements for cardiac mapping and a second program that uses the first to plot developing signals back to a stabilized image of the tissue. WashU's Nathaniel Huebsch said the use of virtual blebbistatin enabled monitoring of the coupling of calcium waves, membrane voltage, and mechanical contraction beat by beat in human-induced pluripotent stem cell heart tissue models without disrupting tissue contractility.
|
Researchers Take Step Towards Reliably Processing Quantum Information
Waterloo News (Canada) September 11, 2023
Researchers at Canada's University of Waterloo have developed a new method to control individual qubits comprised of barium ions. They used a waveguide chip to divide a laser beam into 16 light channels, each directed into individual optical fiber-based modulators to control the intensity, frequency, and phase of each laser beam. A series of optical lenses then were used to focus the laser beams 4 microns apart. Waterloo's K. Rajibul Islam said, " Unlike previous methods to create agile controls over individual ions, the fiber-based modulators do not affect each other. This means we can talk to any ion without affecting its neighbors while also retaining the capability to control each individual ion to the maximum possible extent."
|
Small Drones Monitor Stocks of Raw Materials
Swissinfo (Switzerland) Marc-André Miserez September 11, 2023
Nyrstar, a global metals producer and trader, has deployed small autonomous drones in its storage facility in Antwerp, Belgium, to assess inventory quickly and accurately. The inspector drones use a three-dimensional plan of the warehouse to navigate the space and avoid obstacles. The drones were manufactured by France's Parrot SA with software provided by Tinamu Labs, a startup that originated at Switzerland's ETH Zurich. A Nyrstar spokesperson said the automated system eliminates the need for warehouse staff to become drone pilots, and the absence of GPS means the drones can operate indoors. Said Alcherio Martinoli of Switzerland's École Polytechnique Fédérale de Lausanne (EPFL), "The things drones can do, and their capacity to self-operate, just continue to progress."
|
Big Teeth, Bigger Data
Virginia Tech News David Fleming September 8, 2023
Researchers at Virginia Polytechnic Institute and State University (Virginia Tech) and nonprofit marine research institution Beneath the Waves have evaluated shark sanctuaries using a modeling system they are developing that estimates shark catch and mortality rates from publicly accessible fishing data. The researchers based the model on positioning data of fishing vessels from the open-access Global Fishing Watch commercial fishing website, as well as publicly available information from regional fisheries management organizations. The model estimated a catch of 286,820 large sharks within eight sanctuaries of interest in 2019, of which 109,729 sharks died from stress. Virginia Tech's Francesco Ferretti said such research represents a key step toward rethinking marine population studies and how humans impact open ocean species.
|
Robot Fried Chicken: Entrepreneur Seeks to Improve South Korea's Favorite Food
Yahoo! News Claire Lee September 11, 2023
Entrepreneur Kang Ji-young has deployed robotic arms across her Robert Chicken franchise in South Korea to not only cook fried chicken faster, but also to address labor shortages associated with the nation having the lowest birth rate in the world. The flexible robotic arms each can fry 100 chickens in two hours, which normally would take five workers and several deep fryers. The robots handle the entire frying process and perform real-time measurements of oil temperature and oxidation levels to ensure consistency in taste. The South Korean government has announced a "foodtech" fund to invest in startups working on high-tech food-related innovations.
|
Playing Hide and Seek with Malware
Georgia Tech Research September 8, 2023
Scientists at the Georgia University of Technology (Georgia Tech) and the U.S. Military Academy have identified Web app-engaged (WAE) malware, a new type of malware whose use has risen 226% since 2020. Georgia Tech's Mingxuan Yao said WAE malware is engineered to exploit popular online applications offering services ranging from content delivery to data storage to social networking. WAE malware's approach diverges from that of typical malware by masking its traffic as benign in order to infect targeted systems without detection. The researchers created a tool called Marsea that enabled cybersecurity incident responders and Web app providers to coordinate and purge 79.8% of nearly 1,000 instances of WAE malware discovered in 29 apps. Marsea also discovered attackers are trying to avoid detection by switching their malicious command-and-control servers to these apps.
|
Movement Sensors Detect Disease in Wild Boar
Max Planck Gesellschaft (Germany) September 4, 2023
Researchers at Germany's Max Planck Institute of Animal Behavior found that behavioral sensors can help identify wild boars infected with African Swine Fever. Using lightweight accelerometer sensors, the researchers found that daily activity rates fell up to 20% among wild boars infected with the fatal virus. To validate their findings, the researchers studied wild boars infected with African Swine Fever under controlled conditions, attaching a 30-gram accelerometer tag to each to take ultra-high-resolution measurements of their movements. Their activity patterns were compared to those of healthy boars living in natural conditions. The Max Planck Institute of Animal Behavior's Kevin Morelle said, "This is a game-changer for wildlife disease monitoring. We show that a lightweight behavioral sensor deployed on a wild animal can be a sentinel for potential health threats."
|
System Combines Light, Electrons to Unlock Faster, Greener Computing
MIT News Alex Shipps September 11, 2023
Massachusetts Institute of Technology (MIT) researchers have developed a prototype photonic computing system that can handle machine learning inference requests in real time. The Lightning system is a photonic-electronic reconfigurable SmartNIC (network interface card) the combines the speed of photonics with the dataflow control capabilities of electronic computers. The hybrid system uses a reconfigurable count-action abstraction to act as a unified language between the photonic and electronic computing components. MIT's Manya Ghobadi explained, "Our count-action programming abstraction acts as the muscle memory in Lightning. It seamlessly drives the electrons and photons in the system at runtime." The researchers found that Lightning outperformed standard graphics processing units, data processing units, SmartNICs, and other accelerators in energy-efficiency when completing inference requests.
|
ML Contributes to Better Quantum Error Correction
RIKEN (Japan) September 8, 2023
Researchers at Japan's RIKEN Center for Quantum Computing used a machine learning (ML)-enhanced system for autonomous quantum-computer error correction. The ML component helps search for device overhead-minimizing error correction schemes without impacting performance. The researchers used an artificial environment to eliminate the need for frequent error-detecting measurements, and searched bosonic quantum bit (qubit) encodings for high-performance candidates through reinforcement learning. They determined an approximate qubit encoding reduced device complexity significantly more than other proposed encodings while outperforming peers' error correction. Said RIKEN's Yexiong Zeng, "Our work not only demonstrates the potential for deploying machine learning towards quantum error correction, but it may also bring us a step closer to the successful implementation of quantum error correction in experiments."
|
Wi-Fi Reads Through Walls
The Current (University of California, Santa Barbara) Sonia Fernandez September 11, 2023
A technique developed by University of California, Santa Barbara (UC Santa Barbara) researchers enables Wi-Fi to read through walls. Using the Geometrical Theory of Diffraction and corresponding Keller cones, the technique allows for high-quality imaging of still objects using Wi-Fi signals, with a focus on tracing the objects' edges. The researchers developed a Keller cone-based imaging projection kernel that infers the orientation of the edges using conic footprints. UC Santa Barbara's Anurag Pallaprolu explained, "Once we find the high-confidence edge points via the proposed imaging kernel, we then propagate their information to the rest of the points using Bayesian information propagation." In various tests of the technique, the researchers showed that their Wi-Fi Reader could read the English alphabet through walls. The system also was able to classify the letters.
|
Robotic 'Dog' Surveys Ex-Cold War Site
BBC News Katy Prickett; Alex Dunlop September 9, 2023
The U.K.-based conservation charity National Trust used drones and a mobile robot "dog" from Boston Dynamics to survey otherwise inaccessible buildings or "pagodas" at Orford Ness, a former Cold War-era weapons testing site turned nature reserve in Suffolk. Colin Evison at U.K. civil engineering contractor BAM Nuttall said the goal is to record the "size and geometry" of the decaying buildings "without putting people in harm's way." The pagodas, which the drones and the four-legged robot explored, served as laboratories for environmental experiments on the atomic bomb. The National Trust's Russell Clement said the project aims to record the facilities, which face coastal erosion, for virtual replication so people will be able to explore their interiors.
|
|