Welcome to the July 10, 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."
|
|
|
U.S. Military Takes Generative AI Out for a Spin
Bloomberg
Katrina Manson
July 5, 2023
The U.S. Department of Defense is testing five large language models (LLMs) as part of an effort to develop data integration and digital platforms for military use. U.S. Air Force Col. Matthew Strohmeyer said one experiment using a LLM to perform a military task was "highly successful" and "very fast." However, Strohmeyer said, "That doesn't mean it's ready for primetime right now." The exercises involve feeding classified operational information into the LLMs, with the ultimate goal of using AI-enabled data in decision-making, sensors, and firepower. Specifically, the LLMs are being tasked with helping plan a military response to an escalating global crisis, with a focus on the Indo-Pacific region.
*May Require Paid Registration
|
Dashcam Images Reveal Where Police Are Deployed
Cornell Chronicle
Tom Fleischman
July 6, 2023
Scientists at Cornell University and the University of California, Berkeley obtained indications of how police in New York City may be deployed in local neighborhoods by tapping a deep learning computer model and a dataset of dashboard camera (dashcam) images from rideshare drivers. The researchers trained their model on thousands of annotated images from dashcam provider Nexar to identify marked police vehicles. They counted more than 233,000 images including such vehicles recorded throughout all five boroughs of New York City at all times of day, then checked them against various factors, including local census data. Patterns revealed by this analysis included a greater police presence in affluent commercial areas and in low-income neighborhoods with more numbers of Blacks and Latinos.
|
Fluxonium Qubit Retains Information for 1.43 Milliseconds—10x Longer Than Before
ScienceAlert
David Nield
July 6, 2023
University of Maryland researchers extended the longevity of superconducting quantum bits (qubits) 10-fold by assembling a fluxonium qubit that retains information for 1.43 milliseconds. Fluxonium qubits form from operations at key junctions in a superconducting circuit, and the researchers increased the qubit's relaxation time by modifying its operating frequency and circuit parameters. Explained the researchers, "Notably, even in the millisecond range, the coherence time is limited by material absorption and could be further improved with a more rigorous fabrication. Our demonstration may be useful for suppressing errors in the next-generation quantum processors."
|
Smartwatches May Provide Early Parkinson's Diagnosis
BBC News
Annabel Rackham
July 4, 2023
Scientists at the U.K.'s Cardiff University analyzed data from 103,712 smartwatch wearers to predict who would develop Parkinson's disease by monitoring their speed of movement over a week between 2013 and 2016. The researchers tapped data from the U.K. Biobank database encompassing the health data of more than 500,000 people. Cardiff's Kathryn Peall said, "We compared our model across a number of different disorders, including other types of neurodegenerative disorders, individuals with osteoarthritis, and other movement disorders, amongst others, an advantage of being able to work with a dataset such as the U.K. Biobank. The results from individuals diagnosed with Parkinson's disease were distinct." Cardiff's Cynthia Sandor said the results of this analysis could inform the development of a screening tool for the early detection of Parkinson's disease.
|
Algorithmic Breakthrough Unlocks Path to Sustainable Technologies
University of Liverpool (U.K.)
July 5, 2023
Researchers at the U.K.'s University of Liverpool demonstrated an algorithm's ability to predict the structure of any material accurately based on its atomic composition. The algorithm simultaneously assesses entire sets of possible structures to identify the correct makeup faster. This process enables the identification of manufacturable materials, as well as their characteristics. Said Liverpool's Matt Rosseinsky, "With this new tool, we will be able to define how to use those chemical elements that are widely available and begin to create materials to replace those based on scarce or toxic elements, as well as to find materials that outperform those we rely on today, meeting the future challenges of a sustainable society."
|
How Safe Is Your Office Air? There's One Way to Find Out
The New York Times
Emily Anthes
June 29, 2023
Indoor air-quality sensors installed in commercial buildings during the pandemic are now proving useful in areas affected by wildfire smoke. These sensors help optimize energy use while monitoring air quality in real time, tracking particulate matter, volatile organic compounds, and carbon dioxide. The ventilation system responds automatically when the sensors detect a change in air quality. Additionally, if the sensors indicate higher levels of indoor particulate matter, for instance, building managers are alerted to alter system settings to recirculate more indoor air and lower the amount of outdoor air being drawn into the building. Design and engineering firms, among the early adopters of such systems, say the pandemic spurred interest in the technology from clients.
*May Require Paid Registration
|
Building Models to Predict Interactions in Plant Microbiomes
ETH Zurich (Switzerland)
Michael Keller
July 6, 2023
Scientists in Switzerland explored the formation of bacterial communities in plant microbiomes using computer models to predict interaction based on nutrient preferences and individual metabolic processes. The researchers tested the growth ability of more than 200 representative strains of bacteria from Arabidopsis thaliana using 45 different carbon sources, and found broad convergence between their food niches, indicating intense rivalry. They then assembled metabolic models for all the strains based on these carbon profiles, and modeled interactions between more than 17,500 bacterial pairs. Simulations showed negative interactions, in which at least one strain's population declines with competition, tended to dominate, although positive interactions also could be predicted reliably. ETH Zurich's Julia Vorholt said the models "can be used to identify mechanisms that trigger certain interaction patterns," which "paves the way for targeted microbiome design."
|
Single-Photon Raman LiDAR Is Practical for Underwater Applications
Optica
June 29, 2023
A single-photon Raman LiDAR system developed by researchers at China’s Xiamen University can operate up to 1 kilometer underwater to identify various substances in the water and their distribution characteristics. For instance, the system can detect oil thickness underwater up to 12 meters away, which means it could be used to monitor underwater oil pipelines for leaks. Said Xiamen University's Mingjia Shangguan, " The use of single-photon detection technology made this work possible by improving detection sensitivity to the level of single photons."
|
Teaching AI to Write Better Chart Captions
MIT News
Adam Zewe
June 30, 2023
A dataset developed by Massachusetts Institute of Technology (MIT) researchers aims to improve automatic captioning systems by training machine learning models to customize the complexity and content in chart captions based on users' needs. The VisText dataset features over 12,000 charts, each represented as a data table, image, and scene graph. The researchers found machine learning models trained with scene graphs performed as well as or better than models trained with data tables. MIT's Benny J. Tang said, "A scene graph is like the best of both worlds — it contains almost all the information present in an image while being easier to extract from images than data tables. As it's also text, we can leverage advances in modern large language models for captioning."
|
Redox-Based Transistor as a Reservoir System for Neuromorphic Computing
Tokyo University of Science (Japan)
July 3, 2023
Researchers at Japan's Tokyo University of Science (TUS) and National Institute for Materials Science invented a redox reaction-based ion-gating reservoir transistor with applications for neuromorphic computing. The researchers deposited a 200-nanometer-thick film of lithium cobalt oxide onto a 0.15-millimeter-thick lithium ion-conducting glass ceramic (LICGC) substrate. The LICGC acts as an electrolyte that doubles the number of reservoir states via the drain current and an additional gate current. The drain current's delayed response relative to the gate current also facilitates short-term memory capabilities. Said TUS' Tohru Higuchi, "The developed system has the potential to become a general-purpose technology that will be implemented in a wide range of electronic devices including computers and cellphones in the future."
|
Researchers Encourage Brain Data Reuse with CAROT
YaleNews
Mallory Locklear
July 5, 2023
A tool developed by Yale University researchers will allow scientists to compare and reuse connectomes, maps of the connections between different regions of the brain comprised of imaging data superimposed on atlases detailing the locations and borders of various brain regions. The publicly available tool, Cross Atlas Remapping via Optimal Transport (CAROT), allows connectomes based on one atlas to be reconfigured into a connectome based on a different atlas using the mathematical theory of optimal transport to determine the best way to redistribute the data. Said Yale's Dustin Scheinost, "Our research shows CAROT can reconstruct connectomes on different brain atlases without altering the makeup of that connectome and without the need for raw data. In essence, it means we can do more with data we already have and gain a better understanding of what all of that data can collectively tell us about the brain."
|
|
