Welcome to the January 28, 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."
|
|
|
IRS Wants to Scan Your Face
The Washington Post
Drew Harwell
January 27, 2022
By this summer, Americans wanting to access their Internal Revenue Service records online will be required to submit a facial video to private contractor ID.me to confirm their identity. ID.me requires facial scans plus copies of identifying paperwork, then employs facial recognition software to determine whether a person's "video selfie" and official photo match. Privacy advocates are concerned, as no federal law exists regulating such information's use or sharing. Glitches and delays that have kept users from important benefits also plague the system. Researchers contend ID.me has exaggerated the abilities of its face-scanning technology, which could wrongly label people frauds. "We're just skipping right to the use of a technology that has clearly been shown to be dangerous and has issues with accuracy, disproportionate impact, privacy, and civil liberties," said the Electronic Privacy Information Center's Jeramie D. Scott.
*May Require Paid Registration
|
Robot Performs Keyhole Surgery on Pigs
New Scientist
Matthew Sparkes
January 26, 2022
Researchers at Johns Hopkins University programmed a robot to perform minimally invasive keyhole surgery on pigs with little help from human surgeons. The robot performed intestinal anastomosis on four pigs, for a total of 86 stitches; stitches were placed autonomously two-thirds of the time. The results one week after surgery were comparable to those performed by human surgeons. The robot was controlled by custom software that senses depth and maps the changing layout inside the abdomen using images from a three-dimensional (3D) camera on the robot's arm. Despite the trial's success, Johns Hopkins' Justin Opferman said fully autonomous surgery in humans is likely decades away.
*May Require Paid Registration
|
Where Did That Sound Come From?
MIT News
Anne Trafton
January 27, 2022
Massachusetts Institute of Technology (MIT) neuroscientists have developed a computer model that can localize sounds as the human brain does. The team used a supercomputer to train and test roughly 1,500 convolutional neural networks, winnowing them down to 10 that received further training. The models, designed to replicate the same initial sound input as the human ear, were trained on over 400 natural and artificial sounds in a virtual environment with adjustable size and sound reflection parameters. MIT's Josh McDermott said, "The model seems to use timing and level differences between the two ears in the same way that people do, in a way that's frequency-dependent."
|
On the Hunt for Meteorites, Researchers Look to Data-Based 'Treasure Map'
NBC News
Patrick Galey
January 26, 2022
Belgian and Dutch researchers used machine learning to create a "treasure map" of meteorites on Antarctica in the search for clues about the beginnings of our solar system. The researchers determined four key conditions to predict the presence of meteorites in a given area using previous expedition data and satellite imagery. Said Harry Zekollari at Switzerland's ETH Zurich, "We knew that we could get a lot out of satellite data because there are some characteristics you need such, as slow velocities of ice movement, and that the ice needs to be cold. But we also quickly knew that it's not simply that we saw these conditions so there must be meteorites here. There was an interaction, and to get that interaction we really needed machine learning." The results of the machine learning model were correct more than 80% of the time in tests; it estimated 300,000 to 900,000 pieces of space rock remain undiscovered across the continent.
|
Bristol Team Chases Down Advantage in Quantum Race
University of Bristol News (U.K.)
January 26, 2022
Researchers at the U.K.'s University of Bristol, Imperial College London (ICL), and Hewlett Packard Enterprise have shortened the time needed to simulate an optical quantum computer by a factor of around a billion. Scientists at the University of Science and Technology of China previously claimed to have simulated in 200 seconds a Gaussian Boson Sampling (GBS) experiment whose runtime on the world's largest supercomputer would take 600 million years. The Bristol-led team said it had reduced the same simulation’s runtime on a supercomputer to several months. ICL's Bryn Bell said, "As researchers develop larger scale experiments, they will look to make claims of quantum advantage relative to classical simulations. Our results will provide an essential point of comparison by which to establish the computational power of future GBS experiments."
|
How AI Can Identify People Even in Anonymized Datasets
ScienceNews
Nikk Ogasa
January 25, 2022
Artificial intelligence (AI) can identify people in anonymized datasets by studying patterns in their weekly social interactions, according to researchers at the U.K.'s Imperial College London and University of Oxford. The researchers structured mobile phone interaction data on 43,606 anonymous phone service subscribers into web-shaped configurations of nodes representing the user and their contacts, connected with strings threaded with interaction data. When shown the interaction web of a known individual, the AI sifted the anonymized dataset for the most similar-looking web, and was able to correctly identify the target more than half of the time. When supplied the target and contacts' interaction data collected 20 weeks after the anonymous dataset, the AI correctly identified users 24.3% of the time, suggesting social behavior remains identifiable for long durations.
|
ML Model Shows Diamond Melting at High Pressure
Sandia National Laboratories
January 26, 2022
The Spectral Neighbor Analysis Potential (SNAP) machine learning model developed by Sandia National Laboratories researchers predicted the behavior of billions of interacting atoms to simulate diamond melting at high pressures and temperatures. A team of researchers used the Summit supercomputer at the U.S. Department of Energy's Oak Ridge National Laboratory to perform the diamond-specific modeling in a day. SNAP trained a surrogate model of a micron-sized piece of compressed diamond that replicated atomic forces estimated via quantum mechanical calculations, then scaled it up to predict forces and accelerations for systems featuring billions of atoms. By performance-optimizing the software to run on supercomputers like Summit, the researchers accelerated the simulation’s runtime by 97%, said Sandia's Aidan Thompson.
|
Booby-Trapped Sites Delivered Potent Backdoor Trojan to macOS Users
Ars Technica
Dan Goodin
January 25, 2022
Researchers at Slovak Internet security company ESET have uncovered macOS malware installed by exploits that were almost impossible for most users to detect or halt once the user visits a malicious Website. The DazzleSpy malware is a full-featured backdoor trojan written from scratch to enable hackers to monitor and control infiltrated Macs. ESET's Marc-Etienne M.Léveillé said the malware’s refinement and the apparent absence of a corresponding version for Windows suggests its creators are targeting Macs exclusively. He added that on unpatched systems, DazzleSpy would start running with administrative privileges without the victim realizing. Threat analysis researchers at Google who first discovered DazzleSpy's exploits said the hackers are likely state-financed, "with access to their own software engineering team based on the quality of the payload code." Apple said it has patched the flaws exploited in this attack.
|
Tweets Track Australian Mental Health Swings During COVID Pandemic
University of Arkansas News
January 25, 2022
U.S. and Australian researchers studied approximately 245,000 COVID-19-related geotagged tweets to characterize pandemic-associated mental health shifts across Australia from January 2020 to May 2021. "In this effort, the data collection paradigm we designed aims to intelligently retrieve public emotions and sentiments in an automatic manner," said the University of Arkansas' Xiao Huang. Using machine learning and spatial mapping, the researchers observed a swing from pessimism early on in the pandemic to more optimism in its middle phase, before reverting toward greater pessimism, possibly because of concerns with the vaccine rollout. Siqin Wang at Australia's University of Queensland said, "The clear insights into when and where people are displaying higher levels of pessimistic mental health signals provides important information through which the allocation of finite mental health facilities can be deployed."
|
Software Is Crammed Full of Bugs. This 'Exciting' Project Could Banish Most of Them
ZDNet
Liam Tung
January 25, 2022
Chip designer Arm has released a prototype development board based on the Capability Hardware Enhanced RISC Instructions (CHERI) architecture. The Morello board, developed with researchers at the U.K.’s University of Cambridge and Microsoft, among others, could pave the way for new CPU designs that eliminate memory-related security flaws stemming from code written in programming languages like C and C++. Google's Ben Laurie said CHERI's software compartmentalization is similar to process isolation in software for current operating systems. Said Microsoft's Saar Amar, "There are billions of lines of C and C++ code in widespread use, and CHERI's strong source-level compatibility provides a path to achieving the goals of high-performance memory safety without requiring a ground-up rewrite."
|
Calculating the Best Shapes for Things to Come
University of Michigan News
January 25, 2022
University of Michigan (U-M) and Northeastern University researchers have created an algorithm for designing structures that maximize performance and efficiency. The algorithm applies "nongradient" optimization techniques, and expedites them by first approximating the system via machine learning, then employing a self-directed learning framework. U-M's Changyu Deng said, "This research can dramatically accelerate non-gradient optimizers for topology optimization to make nongradient methods feasible. In this way, more complicated problems can be tackled." When tested in four optimization scenarios, the tool slashed the computational time required to reach the best solution by 100 to 100,000 times over traditional methods, and outperformed other state-of-the-art algorithms.
|
Talent Gaps Stalling Innovation in the Tech Sector, Survey Says
Silicon Republic
Blathnaid O'Dea
January 26, 2022
Digital talent shortages are impeding technological innovation, according to an international poll by online learning platform Udacity and market research company Ipsos. Nearly half (44%) of about 2,000 managers and over 4,000 staff surveyed across the U.K., France, Germany, and the U.S. said employee turnover was stalling their progress in meeting corporate goals, while 59% said a dearth of skilled employees was impacting their business. Half of employers polled in the survey also said scant employee adoption or engagement was hindering their digital transformation projects, while 45% of employees with access to learning and development programs said they felt completely or very satisfied. Said Udacity's Gabe Dalporto, "The talent shortage has reached a crisis pitch. If companies do not invest in talent transformation, they are destined to fail."
|
Researchers Develop Silk-Based Digital Security Device
Gwangju Institute of Science and Technology (South Korea)
January 24, 2022
Researchers at South Korea's Gwangju Institute of Science and Technology (GIST) have utilized natural silk fibers from domesticated silkworms to build an environmentally friendly digital security system. The physical unclonable function (PUF) device leverages the diffraction of light through silk's natural microholes to produce a digital security key. GIST's Young Min Song said an image sensor captures the refracted light, "giving rise to a unique pattern of light" that is converted into a digital format. The researchers incorporated PUF-based tags into a portable lens-free optical PUF (LOP-PUF) module. The researchers said faking the LOP-PUF module's authentication key would take such a long time (5x10 41 years) that they consider the device practically unbreachable.
|
|
