Welcome to the September 9, 2022, 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."
|
|
|
Ian Foster to Receive ACM-IEEE CS Ken Kennedy Award
ACM
September 7, 2022
ACM and the IEEE Computer Society have named the University of Chicago's Ian Foster the recipient of this year's ACM-IEEE CS Ken Kennedy Award for his accomplishments in high-performance computing. Foster was cited for his contributions to programming and productivity in computing by creating new programming models and foundational science services. Contributions by Foster and colleagues include high-level task parallelism, including the easing of interacting-task specification, program composition enablement, and large distributed/parallel computer systems scaling. Foster and colleagues also launched an initiative to create what became known as grid computing. Other achievements to which Foster contributed include the development of universal data transfer, trust fabrics, and cloud management services for data-intensive science.
|
Next Generation of Hearing Aids Reads Lips Through Masks
University of Glasgow (U.K.)
September 7, 2022
An international team led by researchers at the U.K.'s University of Glasgow developed a system that can read lips accurately even through face masks using radio-frequency (RF) sensing and artificial intelligence. The researchers had volunteers repeat vowel sounds while masked and unmasked, with RF signals from a dedicated radar sensor and a Wi-Fi transmitter used to scan their faces while speaking and while still. The resulting 3,600 samples of data were used to train machine learning and deep learning algorithms to recognize the lip and mouth movements related to each vowel sound. Because RF signals can pass through masks, the machine learning and deep learning algorithms were able to learn to identify vowel formation by masked speakers.
|
Lensless Camera Creates 3D Images from Single Exposure
Optica
September 7, 2022
Scientists have created a lensless camera that generates three-dimensional (3D) images from a single exposure in real time. "We consider our camera lensless because it replaces the bulk lenses used in conventional cameras with a thin, lightweight microlens array made of flexible polymer," said University of California, Davis' Weijian Yang. The individual microlenses in the array enable the camera to view objects from different angles or perspectives, providing depth information. Yang said the device uses a neural network "based on a physical model of image reconstruction. This makes the learning process much easier and results in high-quality reconstructions."
|
Systematically Finding Optimal Quantum Operation Sequences for Quantum Computers
National Institute of Information and Communications Technology (Japan)
September 2, 2022
A multi-institutional team of researchers in Japan has developed a new method for finding optimal quantum operation sequences for quantum computers. The GRAPE algorithm applies optimal control theory to extrapolate the theoretically optimal sequence from all conceivable quantum operation sequences. The algorithm generates a table of quantum operation sequences and the performance index for each sequence based on the number of quantum bits (qubits) and the number of operations being investigated. The optimal quantum operation sequence is systematically identified according to the accumulated data. The researchers expect the technique to become a practical tool for medium-scale quantum computers, and to enable near-term enhancement of quantum-computer performance.
|
'Predictive-Maintenance' Tech Taking Off as Manufacturers Seek Greater Efficiency
The Wall Street Journal
Angus Loten
September 7, 2022
Manufacturers increasingly are turning to technology that can predict industrial equipment failures as a means of boosting efficiency. PepsiCo, for instance, has deployed predictive-maintenance systems from startup Augury Inc. at four Frito-Lay factories. PepsiCo Labs' Anna Farberov said the systems have helped reduce unexpected breakdowns, interruptions, and replacement part costs, while adding about 4,000 hours per year of manufacturing capacity. Augury's systems uses wireless sensors that capture the sounds made by factory equipment, which are analyzed by software that can identify over 80,000 industrial machinery sounds, detect patterns, and provide real-time feedback. Warren Pruitt of Colgate-Palmolive, which uses Augury's platform, said, "Our predictive-maintenance program also upskills our workforce, giving our employees the bandwidth to look at the big picture and consider how to employ new technologies and initiatives to continuously improve our operation."
*May Require Paid Registration
|
English Bias in Computing: Images to the Rescue
University of Copenhagen (Denmark)
August 31, 2022
An image-based benchmark could overcome cultural bias stemming from machine learning (ML) training datasets being written in English. An international group of researchers led by Denmark's University of Copenhagen (KU) developed the Image-Grounded Language Understanding Evaluation (IGLUE) tool, which can score an ML solution's efficiency in 20 languages. Image labels in ML are typically in English, while IGLUE covers 11 language families, nine scripts, and three geographical macro-areas. IGLUE's images feature culture-specific components supplied by volunteers in geographically diverse countries in their natural language. KU's Emanuele Bugliarello said the researchers hope IGLUE's underlying methodology could improve solutions "which help visually impaired in following the plot of a movie or another type of visual communication."
|
GM's Cruise Recalls, Updates Software in Robotaxis
TechCrunch
Kirsten Korosec
September 1, 2022
A crash in June prompted General Motors' Cruise autonomous vehicle (AV) unit to recall 80 robotaxis and update their software. In a regulatory filing with the National Highway Traffic Safety Administration, Cruise attributed the June collision to a "rare circumstance" in which the automated driving system caused the unmanned robotaxi to brake hard while making an unprotected left turn. "The report explains how the Cruise AV responded to an oncoming vehicle speeding in the wrong lane, and how through our normal course of continuous improvements, Cruise AVs are even better equipped to prevent this singular, exceptional event," the unit said. Cruise explained the automated driving system had chosen the risk scenario with the least potential for a serious crash, before the oncoming vehicle suddenly shifted direction.
|
AI Can Better Monitor Maine's Forests
UMaine News
September 1, 2022
A study by researchers at the universities of Maine (UMaine), New Hampshire, and Vermont applied artificial intelligence (AI) and machine learning to monitor soil moisture more efficiently. The researchers designed a wireless sensor network that uses AI to learn to monitor soil moisture and process the data with greater power efficiency. The software learns over time how to best employ available network resources, which helps generate power-efficient systems at reduced cost for large-scale monitoring. "AI can learn from the environment, predict the wireless link quality and incoming solar energy to efficiently use limited energy, and make a robust low-cost network run longer and more reliably," said UMaine's Ali Abedi.
|
'Virtual Rapper's' Firing Raises Questions About Art, Tech
The New York Times
Marc Tracy
September 6, 2022
Capitol Records' "firing" of rapper avatar FN Meka for allegedly promoting stereotypes and cultural appropriation has raised questions about whether artificial characters based on real people are appropriate. The virtual rapper's songs were written and voiced by humans, while its look, persona, and lyrics were inspired by real-life artists. Critics said FN Meka's partial debt to artificial intelligence and digital existence absolved its creators of accountability, while adding to the controversy is young people's increasing comfort with digital avatars as cultural mediators. The Massachusetts Institute of Technology's Ziv Epstein said, "These emerging technologies require new legal frameworks and research to understand how we reason about them."
*May Require Paid Registration
|
A Novel Approach to Creating Tailored Odors, Fragrances Using Machine Learning
Tokyo Tech News (Japan)
September 5, 2022
Researchers at Japan's Tokyo Institute of Technology (Tokyo Tech) used machine learning (ML) to predict the sensing data of odor mixtures, and to design customized fragrances. The technique taps standard mass spectrum data and ML models to forecast molecular features based on odor impression. As an example, Tokyo Tech's Takamichi Nakamoto explained the approach was able to identify molecules that emit the mass spectrum of apple flavor with enhanced "fruit" and "sweet" impressions. "Combinations of either 59 or 60 molecules give the same mass spectrum as the one obtained from the specified odor impression," Nakamoto said. "With this information, and the correct mixing ratio needed for a certain impression, we could theoretically prepare the desired scent."
|
Apps Used as Alternatives to Prison in U.S. Found to Have Privacy Flaws
New Scientist
Jeremy Hsu
September 3, 2022
University of Washington researchers have discovered privacy flaws in smartphone monitoring apps used in the U.S. to track people waiting for immigration court dates, those in juvenile detention systems, and those on parole or probation. Of the 16 Android monitoring apps studied, the researchers found that seven either did not link to a privacy policy or linked to generic privacy policies, in violation of the Google Play Store's user data policies. One app used by U.S. Immigration and Customs Enforcement, BI SmartLINK, required "dangerous permissions" (to access the device’s camera, obtain its precise location, make telephone calls without user permission, and record audio), but did not disclose that it in its privacy policy.
*May Require Paid Registration
|
Chrome Patches Sixth High-Severity Zero-Day This Year
Ars Technica
Dan Goodin
September 6, 2022
Google engineers have published an emergency update for the Chrome browser to correct a high-severity zero-day flaw that can be exploited with available code. Google said the vulnerability (CVE-2022-3075) stems from "insufficient data validation in Mojo," a Chrome component for messaging across inter- and intra-process boundaries between the browser and the operating system. "Google is aware of reports that an exploit for CVE-2022-3075 exists in the wild," the company explained, without disclosing whether hackers are exploiting the vulnerability or just possess exploit code. Engineers also updated Microsoft's Edge browser, based on the same Chromium engine as Chrome, to fix the same bug. The Mojo exploit marks the sixth zero-day vulnerability Chrome has encountered this year.
|
City Digital Twins Help Train Deep Learning Models to Separate Building Facades
Research at Osaka University (Japan)
September 5, 2022
Researchers at Japan's Osaka University used images generated automatically by digital city twins to train a deep learning model that can accurately separate out the building facades in an image. The researchers used a three-dimensional city model from the PLATEAU platform to create the synthetic digital city twin data. They input the model into the Unity game engine and drove a virtual car with a camera setup around the city to gather virtual images in various lighting and weather conditions. Real street-level images were incorporated from Google Maps API. The researchers found segmentation accuracy was improved by adding synthetic data to the real dataset, and in terms of performance, the model can compete with a deep convolutional neural network trained on 100% real data. Said Osaka University's Tomohiro Fukuda, "These results reveal that our proposed synthetic dataset could potentially replace all the real images in the training set."
|
|
