Welcome to the August 20, 2021 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."
|
|
|
10 Girls on Afghanistan's Robotics Team Rescued
NBC News
Kait Hanson
August 19, 2021
Ten members of Afghanistan's girls robotics team have been evacuated safely from their homeland amid the Taliban's takeover. The team's parent organization, the Digital Citizen Fund (DCF), said the girls, ages 16-18, arrived in Doha, Qatar, from the city of Kabul. DCF's Elizabeth Schaeffer Brown said, "Ultimately the girls 'rescued' themselves. If it were not for their hard work and courage to pursue an education, which brought them in contact with the world, they would still be trapped." Several team members are still in Afghanistan, and the DCF has partnered with Qatar to arrange transportation for them and their aides. Said Brown, "We appreciate [the rescue efforts] and hope it translates to long-term commitment to girls' education."
|
Data Mining Tools Combat COVID-19 Misinformation, Identify Symptoms
UC Riverside News
Holly Ober
August 19, 2021
Researchers at the University of California, Riverside (UCR) and the University of Texas Rio Grande Valley developed an algorithm that identified three unique COVID-19 symptoms that are distinct from those of influenza. The algorithm mined Google Trends data for 2019 and 2020 using nonnegative discriminative analysis to root out terms unique to one dataset compared to the other. UCR's Jia Chen said, "We assumed that symptom searches in 2019 would lead to influenza or other respiratory ailments, while searches for the same symptoms in 2020 could be either." UCR’s Jia Chen described the algorithm as potentially easy to incorporate into a tool that might help scientists investigating other diseases learn more about symptoms.
|
AI Gave Val Kilmer His Voice Back. Critics Worry the Technology Could Be Misused.
The Washington Post
Dalvin Brown
August 18, 2021
U.K. software company Sonantic helped actor Val Kilmer speak again after surgery robbed him of his natural voice, but critics worry about potential misuse of such artificial intelligence-based voice-cloning software. Sonantic reproduced Kilmer's voice using samples from old footage cleaned of background noise; engineers linked the audio in "short chunks" and processed the data through "voice engine" algorithms, which learn to speak from the recordings. The end product was a desktop-based text-to-speech program that Sonantic claims can emulate Kilmer's projection levels and emotion when reading lines of text aloud. The technology concerns people like voice actors, who imagine unauthorized use of the technology endangering their livelihood, as well as making statements they themselves would not make.
*May Require Paid Registration
|
At Japanese Robot Cafe, Staff Serve Customers While Working From Home
South China Morning Post
August 20, 2021
The Dawn Cafe in Tokyo's Nihonbashi district is staffed by robot waiters serving diners under the remote direction of physically and mentally impaired staff based in Japan and overseas. The cafe is part of an experiment in inclusive employment, with the remote-controlled robots offering job opportunities to people who find working outside the home difficult. The establishment features about 20 OriHime robots built by the Ory Laboratory company equipped with cameras, microphones, and speakers so operators can communicate with customers. Three larger, humanoid robots serve drinks or greet customers, while a barista robot prepares coffee. Ory Laboratory co-founder Kentaro Yoshifuji said, "Customers here are not exactly coming to this location just to meet OriHime. There are people operating OriHime behind the scenes, and customers will come back to see them again."
|
Israel 'Prints' World's First 3D Living Malignant Brain Tumor
The Jerusalem Post (Israel)
Maayan Jaffe-Hoffman
August 19, 2021
Researchers at Israel's Tel Aviv University (TAU) have produced the first-ever three-dimensionally (3D)-printed viable malignant brain tumor from surgically excised tissue. The 3D glioblastoma replicates the brain's microenviroment, with blood vessels linked to a tubing network through which researchers can pump blood cells and different drugs to the model, resulting in better predictions of effective agents. TAU's Ronit Satchi-Fainaro said the 3D-bioprinted tumor is superior to petri dish-grown tumors in many respects, and she envisions using it to find "novel druggable target proteins and genes in cancer cells."
|
Aurora Releases Tool to Gauge Safety of Self-Driving Systems
Reuters
Paul Lienert
August 18, 2021
Silicon Valley-based self-driving startup Aurora has unveiled what it describes as the industry's first tool for assessing the relative safety of autonomous vehicles. Aurora's Chris Urmson said the Safety Case Framework provides a "structured approach" to assessing the safety of autonomous vehicles on actual streets, featuring four levels of claims associated with the safe development, testing, and evaluation of the company's self-driving systems, as well as required supporting evidence. The framework supports a systematic approach to assessing the vehicles’ safety, as well as metrics for measuring progress across their full development cycle.
|
Drones, Autonomous Robots Could Be Deployed in Coal Mines to Improve Safety
ABC News (Australia)
Melissa Maddison; Angel Parsons
August 13, 2021
Technology developed by scientists at Australia's Commonwealth Scientific and Industrial Research Organization (CSIRO) enables drones to map and inspect subterranean coal mines, to avoid putting people into potentially hazardous situations. Aerospace engineer Luke Horwood said the autonomous HoverMap scanner uses fixed points to generate a network of coordinates, emitting 300,000 laser points a second to map the surroundings; if a drone loses contact with its operator, it returns to a waypoint to find its way back. Holstein Wong at drone company Emesent said operators also can mount the scanner on a backpack, ground robot, or vehicle. The technology hopefully will let mine operators make faster decisions on changing environments. Said Horwood, "If someone sees deterioration, we can scan the area multiple times and compare them and do change detections."
|
Humanitarian Aid Guided by Satellite Data May Harm Marginalized Groups
New Scientist
Edd Gent
August 16, 2021
Lukas Kondmann and Xiao Xiang Zhu at the German Aerospace Center warn of potential bias in satellite data that could harm marginalized groups when policymakers tap such information to guide humanitarian efforts. The researchers trained a model to forecast poverty levels and electricity rates in Indian villages using socioeconomic data, satellite images of nighttime lights, and coordinates from 386,000 villages. The model consistently erred in its findings for villages with large numbers of historically discriminated scheduled castes and scheduled tribal communities. Kondmann said the bias could stem from the unfair apportionment of public investment, or the close proximity of affluent and poor neighborhoods. He said the results imply that features of the data reflect historical prejudices, which could perpetrate discrimination if used to guide policy.
|
Millions of Web Camera, Baby Monitor Feeds Exposed
Wired
Lily Hay Newman
August 17, 2021
Researchers at cybersecurity firm Mandiant have identified a vulnerability in a software development kit (SDK) affecting more than 83 million smart devices. The flaw in ThroughTek Kalay, an SDK that facilitates the connection between a device and mobile apps, could enable hackers to access live video and audio streams over the Internet, assume full control of devices remotely, launch denial of service attacks, or install malicious firmware. If a hacker obtains the device's unique identifier (UID) through a social engineering attack or by searching for a manufacturer's Web vulnerabilities, they could reregister the UID and hijack the connection when a user next accesses the device. The researchers, who said they have seen no evidence of real-world exploitation of the vulnerability, said they hope to raise awareness about the problem without telling potential attackers how to exploit it.
|
Magnets Could Offer Better Control of Prosthetic Limbs
MIT News
Anne Trafton
August 18, 2021
Using magnetic beads rather than electromyography could give amputees better control over prosthetic limbs. Researchers at the Massachusetts Institute of Technology (MIT) and Brown University used muscle-implanted beads to measure muscle length during contraction, then fed that data to a bionic prosthesis. This magnetomicrometry approach works with an algorithm that greatly reduces the time needed for sensors to pinpoint the positions of small magnets within the body. MIT's Hugh Herr said, "Through mathematical modeling of the entire limb, we can compute target positions and speeds of the prosthetic joints to be controlled, and then a simple robotic controller can control those joints."
|
Common Solar Tech Can Power Smart Devices Indoors, NIST Study Finds
NIST
August 19, 2021
Researchers at the National Institute of Standards and Technology (NIST) studying the indoor charging capability of small modular photovoltaic (PV) devices determined that a mini modular PV device made of silicon, absorbing only light from an LED, supplied more power to a wireless temperature sensor than the sensor consumed in operation. They said this indicates it is possible to power Internet of Things devices with PV modules. The researchers found the silicon module converted 9.3% of the LED light into electrical power, compared to power conversion efficiency rates of 23.1% for gallium indium phosphide (GaInP) and 14.1% for gallium arsenide (GaAs) modules. NIST's Andrew Shore said, "Even with a less-efficient mini module, we found that we could still supply more power than the wireless sensor consumed."
|
Can AI Make a Better Fusion Reactor?
IEEE Spectrum
Rebecca Sohn
August 13, 2021
Researchers at Portugal’s University of Lisbon have trained artificial intelligence, machine learning, and deep learning models for use in nuclear fusion research. The models were trained using diagnostic data from 48 sensors connected to the U.K.'s Joint European Torus reactor. In a study at the University of Washington, researchers used a single graphics processing unit (GPU) to control a fusion experiment that previously required multiple computers. At the KTH Royal Institute of Technology in Sweden, researchers have developed a deep learning model that calculates the electric fields of plasma more quickly and accurately than previous methods. Said KTH’s Stefano Markidis, “At the end of the day, it's going to be our minds that [will] solve the fusion problem. It's just a matter of what tools we use, and AI and machine learning will be a key tool."
|
|
