Welcome to the October 4, 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."
|
|
|
The World Has a Food-Waste Problem. Can Wireless Tech Help Fix It?
Princeton University School of Engineering and Applied Science
Scott Lyon
October 3, 2023
A new tool has the potential to reduce food waste by using high-frequency wireless technology to assess fruit quality by measuring ripeness. Princeton University and Microsoft Research researchers developed the tool, which uses wireless signals in the sub-terahertz band to measure sugar and dry matter content beneath the skin of fruit on a conveyor belt. The wireless pulse's reflection generates a signal with a unique signature; the researchers modeled the physics of these interactions to read this signature and evaluate the ripeness of each individual fruit, after experiments with persimmons, avocados, and apples.
|
More LiDAR, Less Data Crunching
IEEE Spectrum
Michelle Hampson
October 3, 2023
Scientists at New Jersey's Rowan University (RU) formulated a method for simplifying light detection and ranging (LiDAR) signal processing by reducing the data crunching and machine learning workload. The blind source separation (BSS) technique involves two different receivers picking up LiDAR signals from two different angles. Explained RU's James Garofolo, "Once these combinations are received, we can scale them such that the interference is the same amplitude in both channels and then subtract them from one another to almost completely remove the interference while still [picking up] the signal of interest." Experiments showed BSS can remove as much as 40 decibels (dB) of noise, although Garofolo said reductions of 30 dB were more typical.
|
A More Effective Design for Engineering a Cell into a New State
MIT News
Adam Zewe
October 2, 2023
Researchers at the Massachusetts Institute of Technology and Harvard University have developed an algorithmic cellular reprogramming method that utilizes targeted genetic interventions. The technique can detect the best genetic perturbations using fewer experiments than conventional approaches, by applying a causal framework to spotlight the optimal strategy in each series of sequential experiments. The researchers programmed the algorithm to learn only models of the system that factor in causal relationships, then engineered the acquisition function to automatically assess interventions using data on these relationships and prioritize the most likely optimal approaches. They used output weighting to refine the acquisition function and fed the algorithm real biological data that emulated a cellular reprogramming experiment. Interventions identified by the acquisition functions consistently trumped baseline methods.
|
RSA, Other Crypto Systems Vulnerable to Side-Channel Attack
Computer Weekly
Cliff Saran
October 3, 2023
Hubert Kario at open source solutions provider Red Hat found a flaw dating from 1998 that enables a "padding mode" side-channel attack targeting RSA encryption. The exploit cracks the Transport Layer Security (TLS) protocol's confidentiality when used with RSA encryption, and researchers in 2019 highlighted the continued vulnerability of many Internet servers to tweaks of the original attack. Kario said attackers can leverage the flaw to decrypt RSA ciphertexts and forge signatures, and record sessions on a TLS server that defaults to RSA encryption key exchanges for decryption later. He also said hackers can apply the exploit to other interfaces that automatically execute RSA decryption, including Secure/Multipurpose Internet Mail Extensions, JavaScript Object Notation web tokens, and hardware tokens. Said Kario," We have identified the vulnerability in multiple implementations and confirmed fixes in a few of them but believe that most cryptographic implementations are vulnerable in practice."
|
AI Designs New Robot from Scratch in Seconds
Northwestern Now
October 2, 2023
A research team led by Northwestern University scientists created an artificial intelligence (AI) capable of designing robots from scratch almost immediately. The researchers prompted the algorithm to design a robot from a block about the size of a bar of soap, which generated a successful design in 26 seconds. Northwestern's Sam Kriegman said, "We told the AI that we wanted a robot that could walk across land. Then we simply pressed a button and presto!" The algorithm operates on a lightweight personal computer; other AI systems often require power-hungry supercomputers and huge datasets. The researchers fabricated the robot from the AI's blueprint, validating its real-world performance.
|
Data-Driven Regional Ocean Models Essential for Planning
KAUST Discovery (Saudi Arabia)
October 2, 2023
The first precise historical reconstruction of the Red Sea's circulation, developed by researchers at Saudi Arabia's King Abdullah University of Science and Technology (KAUST), combines fine-grained regional data and a high-resolution ocean model. The resulting model produced more data on current circulation, temperature, salinity, and oceanic behavior than analyses based on existing global ocean models that do not consider regional sea characteristics. The researchers generated a regional Red Sea dataset by integrating all available satellite and in situ ocean data and leveraging improved bathymetry, a high-resolution spatial grid, an in-house regional high-resolution atmosphere model, and KAUST's Shaheen supercomputer. Extensive sensitivity experiments were undertaken to fine-tune the ocean model parameters to depict the Red Sea's characteristics accurately.
|
DeepMB: Deep Learning Framework for High-Quality Optoacoustic Imaging in Real Time
Helmholtz Munich (Germany)
October 2, 2023
Researchers at Germany's Helmholtz Munich and the Technical University of Munich have developed a deep learning framework that permits the creation of high-quality optoacoustic images in real time, allowing for noninvasive assessments of a wide range of diseases. Clinical use of optoacoustic imaging had been limited by long image processing times. The DeepMB framework can reconstruct high-quality multispectral optoacoustic tomography (MSOT) images around 1,000 times faster than existing algorithms that produce high-quality images, without loss of image quality. It also can reconstruct all scans of a patient, no matter the region of the body or the disease being assessed.
|
Simulations Reveal Atomic-Scale Story of Qubits
University of Chicago Pritzker School of Molecular Engineering
Katherine J. McAlpine
October 2, 2023
University of Chicago researchers led a team that combined multiple computational techniques and algorithms to predict the formation of specific spin defects in silicon carbide. These spin defects (known as "divacancies"), which involve removing a silicon and a carbon atom in close proximity in a silicon carbide solid, could be valuable in quantum sensing. The researchers combined the first-principles molecular dynamics code Qbox and other advanced techniques to computationally simulate the atoms' movements and charges during defect formation. Said the University of Chicago's Giulia Galli, "Our combined techniques and multiple simulations revealed to us the specific conditions under which divacancy spin defects can be efficiently and controllably formed in silicon carbide. In our calculations, we are letting the fundamental physics equations tell us what is happening inside the crystal structure when defects form."
|
Raspberry Pi Tracks Unseen Drones Using Sound
Tom's Hardware
Ash Hill
October 2, 2023
Researchers from the universities of Texas and Tennessee used Raspberry Pi computers to track unseen drones with sound. The DroneChase implementation involves programming the Pi to enclose a detected drone within a red box, using audio to zero in on its location and keep tabs on it even when it is hidden by intervening objects. A three-dimensionally-printed housing contains the tracker, which could be mounted on a vehicle. The researchers trained the artificial intelligence model to locate a drone if it is within the camera's visual scope. DroneChase operates as part of a custom-designed system that uses six microphones to triangulate the drone's position.
|
Scientists Simulate Ant Warfare Using Age of Empires Computer Game
ABC News (Australia)
Megan Hughes
October 2, 2023
Scientists at Australia's Commonwealth Scientific and Industrial Research Organization and the University of Western Australia (UWA) used the computer game "Age of Empires" to simulate ant ‘warfare’, in efforts to eradicate an invasive ant species. The researchers modeled armies, battlefields, and clashes between large, native Australian meat ants and smaller, invasive Argentine ants, and mapped the results. Mathematical models of warfare emerged as the ants fought in a simple, predictable, quantifiable manner. The researchers compared the models to real-world examples "to identify points of commonality and difference between the simple and more complex systems," said UWA's Samuel Lymbery. He said this research could inform new habitat management strategies to give native ants a competitive advantage over non-native ants.
|
Algorithm Helps Non-Experts Fine-Tune Personal Bots
EE Times
Ilene Wolff
September 29, 2023
A machine learning algorithm developed by researchers at the Stevens Institute of Technology, Massachusetts Institute of Technology (MIT), and University of California, Berkeley, can improve personal robot performance with reduced input. The researchers used the algorithm to ask a collaborative robot's owner for information after the bot failed a task, which determined what knowledge the bot was missing, and filled in the gap using a visual motor attention agent simulation application developed by Stanford University. The algorithm generates "counterfactual" explanations identifying what the bot must do to complete a task, then combines them with human feedback to refine its knowledge.
|
Unique Voice Print in Parrots Proven
Max Planck Gesellschaft (Germany)
October 4, 2023
A study by scientists at Germany's Max Planck Institute of Animal Behavior and Spain's Museu de Ciències Naturals de Barcelona found monk parakeets each possess a unique tone of voice or voice print. The researchers recorded more than 5,000 vocalizations from the calls of hundreds of individual birds in Barcelona, re-recording the same individuals over two years to evaluate the calls' stability. They trained models to recognize individuals within each of the five core monk parakeet call types, revealing their "contact call" varied widely, contradicting long-held beliefs. The researchers trained a timbre-based machine learning model used in human voice recognition to identify "tonal"-classed calls of individual parrots, then used it to detect the same individuals from calls classed as "growling." The model performed this task three times better than expected at random.
|
Open Source Software Could Get Microscopes Speaking the Same Language
Nature
Michael Brooks
October 2, 2023
Researchers in the U.K., the U.S., and Germany are working to standardize image formats for biological microscopy data through an open source specification. The project melds the development of the Open Microscopy Environment (OME) with the generic Zarr method for storing and accessing large data arrays in the cloud. Zarr splits the arrays into chunks that can be compressed to retain all the information while still enabling fast reading of, and writing to, files. The format stores neighboring pixels in the same chunk so they all arrive together when downloaded. This year saw the launch of the OME-Zarr data storage platform, and researchers at Seattle's Allen Institute for Neural Dynamics are using it to power the expansion-assisted selective plane illumination microscopy system.
|
|
