Welcome to the August 12, 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."
|
|
|
Robot Arms Replacing Shelf Stockers in Japan's Stores
Bloomberg
Vlad Savov; Mia Glass
August 9, 2022
FamilyMart Co. will install a fleet of artificial intelligence (AI)-driven robots from Telexistence Inc. at 300 of its convenience stores in Japan. The robots, called TX SCARA (Selective Compliance Assembly Robot Arm), operate autonomously 98% of the time and will be used to restock shelves. They can be operated remotely if assistance is needed in accessing an item or there is an issue with the AI technology. Telexistence said each robot arm can replace one to three hours of human work daily per store. Said FamilyMart's Tomohiro Kano, "The newly created time can be reallocated to customer service and shop floor enhancement." The robot arms are equipped with Nvidia's Jetson AI platform and Microsoft's Azure cloud infrastructure, which enable them to process information, record sales data, and use the data to optimize restocking tasks.
*May Require Paid Registration
|
Programmable Materials Can Sense Their Own Movements
MIT News
Adam Zewe
August 10, 2022
Massachusetts Institute of Technology (MIT) scientists have created three-dimensionally (3D)-printed materials with programmable mechanical properties that can sense their movements and interaction with the environment. The researchers incorporated networks of air-filled channels into 3D-printed lattices; measuring pressure changes within these channels when the structure is squeezed, bent, or stretched gives engineers feedback on how the material is moving. Said MIT's Lillian Chin, "The idea with this work is that we can take any material that can be 3D-printed and have a simple way to route channels throughout it so we can get sensorization with structure." The researchers 3D-printed a handed shearing auxetics robot, ran it through a sequence of movements, and trained a neural network on the sensor data to accurately predict its motion.
|
The Hacking of Starlink Terminals Has Begun
Wired
Matt Burgess
August 10, 2022
Lennert Wouters at Belgium's Katholieke Universiteit Leuven hacked SpaceX's Starlink network, a web of more than 3,000 small satellites that enables Internet connections to remote locations on Earth. Wouters exploited vulnerabilities in Starlink's satellite dishes to access the network and run custom code. He stripped down a dish and built an attachable printed circuit board from off-the-shelf-parts, through which he could launch a voltage fault injection attack and circumvent signature verification. Wouters alerted Starlink of the flaws last year, and says although SpaceX has released a firmware update that makes the attack harder, the underlying bug can only be corrected if the company produces a new version of the main chip.
|
Quantum Computer Can Simulate Infinitely Many Chaotic Particles
New Scientist
Karmela Padavic-Callaghan
August 4, 2022
An algorithm developed by researchers at the quantum computing company Quantinuum allows quantum computers to simulate infinitely long chains of interacting electron-like particles with few qubits. The researchers used three to 11 qubits made of charged ytterbium atoms, which they programmed to run the algorithm, and set up interactions that would result in chaotic behavior among the particles. The researchers were able to use very few qubits because the algorithm directed the quantum computer to recycle the qubits during the calculation. The quantum computer was able to reset and reuse a qubit without disturbing others involved in the ongoing calculation.
*May Require Paid Registration
|
Nvidia Sees a Metaverse Populated with Lifelike Chatbot Avatars
CNET
Stephen Shankland
August 9, 2022
Nvidia has launched the Avatar Cloud Engine, which combines three-dimensional (3D) graphics and artificial intelligence to build lifelike 3D models of humans to serve as avatars in the metaverse. Nvidia's Rev Lebaredian predicts that it ultimately will be impossible to determine whether these avatars are a human or a bot. Said Nvidia's Jensen Huang, "Avatars will populate virtual worlds to help us create and build things, to be the brand ambassador and customer service agent, help you find something on a website, take your order at a drive-through, or recommend a retirement or insurance plan." Nvidia's Audio2Face tool matches the avatar's expression to the words it speaks, while its Audio2Emotion tool alters the avatar's facial expression based on the feeling of the words.
|
Models Moving 'Washers' That Help DNA Replicate
Rice University News
Mike Williams
August 9, 2022
Rice University scientists have modeled a key mechanism of DNA replication by combining structural experiments with computer simulations. The researchers gained insights about washer-like motor proteins called helicases, which unzip the double strands of DNA into single strands. The simulations add credibility to the notion that DNA-binding loops within the protein's six subunits form a "staircase" that moves down the DNA spine through adenosine triphosphate (ATP) hydrolysis, in which stored chemical energy in ATP molecules is released. The researchers integrated two coarse-grained simulation models—the AWSEM protein-folding prediction model, and the open3SPN2 DNA simulator. The simulations uncovered previously unknown intermediate states, and determined the interactions involved in the helicase's long-distance motion.
|
Can WhatsApp Messages Be Secure and Encrypted, but Traceable?
The Brink (Boston University)
Andrew Thurston
August 10, 2022
Boston University (BU) researchers have created Hecate, an algorithm that can strengthen a secure messaging application's confidentiality and allow moderators to rein in abuse. The app moderator uses Hecate to generate a unique batch of electronic signatures or tokens for each user, which accompany each message the user sends. If the recipient reports that message, the moderator can confirm the sender's token and take action, a process called asymmetric message franking. BU's Mayank Varia said deniability is ensured because the token is encrypted and only useful to the moderator, so "even if the moderator goes rogue, they can't show and convince the rest of the world—they have no digital proof." Varia calls Hecate "the first message franking scheme that simultaneously achieves fast execution on a phone and for the moderator server, support for message forwarding, and compatibility with anonymous communication networks like Signal's sealed sender."
|
AI May Come to the Rescue of Future Firefighters
NIST News
August 10, 2022
A new artificial intelligence model from researchers at the U.S. National Institute of Standards and Technology and Hong Kong Polytechnic University could warn firefighters of imminent combustion or flashovers. The Flashover Prediction Neural Network (FlashNet) model can predict flashovers as early as 30 seconds before eruption, with up to 92.1% accuracy. The researchers used graph neural networks to strengthen FlashNet and trained it on nearly 25,000 fire cases, then used another 16,000 for refinement and final testing. In digital simulations of over 41,000 fires in 17 types of buildings representing common U.S. residential floorplans, FlashNet outperformed five other machine learning-based models.
|
Supercomputer Simulates First Stars, Black Holes
Texas Advanced Computing Center
Jorge Salazar
August 11, 2022
University of Texas, Austin (UT Austin) scientists investigating how primordial black holes may have impacted the formation of the first stars in the universe made use of simulations on the Texas Advanced Computing Center's Stampede2 supercomputer. The researchers modeled numerical frameworks of the gravity hydrodynamics, chemistry, and cooling in structure and early star formation using cosmological hydrodynamic zoom-in simulations. They added a sub-grid model for black hole accretion and feedback, which estimates how a black hole conglomerates gas and heats its surroundings at each timestep. UT Austin's Boyuan Liu said the models showed that primordial black holes have little effect on star formation, because their abilities to heat gas and form star-seeding accretion disks nearly neutralize each other.
|
APIC Fail: Intel 'Sunny Cove' Chips with SGX Spill Secrets
The Register (UK)
Thomas Claburn
August 9, 2022
An international group of computer scientists discovered an architectural error in certain Intel central processing units (CPUs) affecting the memory-mapped registers of the local Advanced Programmable Interrupt Controller (APIC). This error could be used to expose private encryption keys and other SGX (Software Guard Extensions) enclave data. The researchers said the ÆPIC Leak is "the first architectural CPU bug that leaks stale data from the microarchitecture without using a side channel." An Intel spokesperson said the company "recommends that operating systems and virtual machine monitors enable x2APIC mode, which disables the xAPIC MMIO page and instead exposes APIC registers through model specific registers, which mitigates this issue in affected products.”
|
Amazon Expands Palm-Scanning Payment Tech to 65 More Whole Foods Locations
Tech Crunch
Lauren Forristal
August 10, 2022
Amazon plans to expand its Amazon One palm-scanner payment technology to more than 65 Whole Foods stores in California. A part of the company’s mission to use “contactless” technology that makes it faster to pay, the technology employs machine learning to create a palm signature for each customer and linking it to a payment card at the kiosk or point-of-sale station, allowing them to complete transactions by holding their hand over a scanner. The images captured by the kiosk are encrypted and sent to a server to match to the customer, but are not stored.
|
Busting Anti-Queer Bias in Text Prediction
USC Viterbi School of Engineering
Lillian Goodwin
August 11, 2022
A system developed by researchers at the University of Southern California Viterbi School of Engineering (USC Viterbi) can measure and correct anti-queer bias in text prediction. The researchers applied their system to a large language model (LLM) called BERT, which exhibited homophobic prejudice. USC Viterbi's Katy Felkner established a benchmark that quantifies bias, comparing the probability that the LLM predicts heteronormative sentences versus sentences including a queer relationship. Felkner refined BERT by feeding it tweets and news articles containing LGBT+ keywords from her own QueerTwitter and QueerNews databases, preferring hashtags being used mainly by queer and trans people. The inclusion of this data helped to reduce BERT's bias, with QueerTwitter's tweets nearly halving its heteronormative text predictions.
|
Ordinary Computers Can Beat Google's Quantum Computer After All
Science
Adrian Cho
August 2, 2022
Scientists at China's Chinese Academy of Sciences (CAS) said they could beat Google's Sycamore quantum computer in performing an esoteric calculation using ordinary processors paired with a supercomputer. In 2019, Google researchers set all quantum bits (qubits) to 0, applied to single qubits and pairs a random yet fixed set of logical operations (gates) over 20 cycles, then read out the qubits; they said Sycamore could compute in 200 seconds what would take a supercomputer 10,000 years to complete. The CAS team said classical processors can be used to achieve the same result in just a few hours. The researchers reimagined the calculation as a three-dimensional mathematical array called a tensor network, in which CAS' Pan Zhang said, "we can use many GPUs [graphics processing units] to do the computations in parallel."
|
|
