Welcome to the November 4, 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."
|
|
|
Network Pruning Can Skew Deep Learning Models
North Carolina State University News
Matt Shipman
November 2, 2022
Computer science researchers at North Carolina State (NC State), Syracuse, and Carnegie Mellon universities have shown that neural network pruning can undermine the performance of deep learning models at identifying certain groups. The researchers cited disparities in gradient norms across groups, and in Hessian norms linked to inaccuracies of a group's data, as factors impacting performance. This implies network pruning can compound existing accuracy deficiencies. NC State's Jung-Eun Kim said the team has demonstrated a remedial mathematical method "to equalize the groups that the deep learning model is using to categorize data samples." Tests of the mitigation technique found basically restored a deep learning model to pre-pruning levels of accuracy.
|
Smartphones Can Reveal Whether Bridges Are About to Fall Down
New Scientist
Alex Wilkins
November 3, 2022
A multi-institutional team of scientists led by Thomas Matarazzo at the U.S. Military Academy has created a system to monitor bridge health using movement data from smartphones. Using data gathered from phones as they traveled over San Francisco's Golden Gate Bridge and a short concrete bridge in Ciampino, Italy, including global positioning system location data and information from accelerometers, the team was able to quantify the bridges' modal frequencies to within 3% of readings made by static sensors. The researchers estimated the technique could add more than two years of service to old bridges, and nearly 15 years of service to new bridges, by allowing maintenance to take place when it is needed.
|
Security Loophole Allows Attackers to Use Wi-Fi to See Through Walls
University of Waterloo (Canada)
November 3, 2022
A drone-powered device developed by researchers at Canada's University of Waterloo can see through walls by accessing Wi-Fi networks. The Wi-Peep device can fly close to a building and identify all Wi-Fi-enabled devices inside using the building's Wi-Fi network by taking advantage of the "polite Wi-Fi" loophole, in which smart devices automatically respond to contact attempts from any device within range. Comprised of a store-bought drone and $20 of hardware, Wi-Peep can pinpoint the location of a device within one meter by measuring response times to the messages it sends to devices while in flight. Said Waterloo's Ali Abedi, "We need to fix the Polite Wi-Fi loophole so that our devices do not respond to strangers. We hope our work will inform the design of next-generation protocols."
|
Jet-Lagged Plants Enable Researchers to Understand Human Health Problems
Interesting Engineering
Nergis Firtina
November 2, 2022
Researchers at the U.K.'s University of Edinburgh found that plants can suffer from jet lag. The researchers developed a digital plant to study the impact of biological clock changes equivalent to daily flights between New York and the U.K. Further, they created a computer model that could predict the impact of jet lag on plant growth and determine which biochemical pathways are affected. In a study of the plant species Arabidopsis thaliana, the framework revealed that changes in its biological clock slowed its nightly release of sugar from starch, which slowed its growth. Said University of Edinburgh's Andrew Millar, "Not all details of this model will transfer to crop species, but it extends the 'proofs of principle' for informing crop improvement at the molecular level."
|
ML Facilitates 'Turbulence Tracking' in Fusion Reactors
MIT News
Adam Zewe
November 2, 2022
A multi-institutional team of scientists used computer vision models to track turbulent filaments or "blobs" appearing on the fringe of fuel used in nuclear fusion reactors. The researchers trained four blob-tracking computer vision models on a synthetic video dataset of plasma turbulence, which were able to identify such blobs in actual video clips with more than 80% accuracy in some instances, as well as estimating the blobs' sizes and motion. Said the Massachusetts Institute of Technology's Theodore Golfinopoulos. "Now, we have a microscope and the computational power to analyze one event at a time. If we take a step back, what this reveals is the power available from these machine learning techniques, and ways to use these computational resources to make progress."
|
Innovative Shins Turn Quadrupedal Robot Biped
IEEE Spectrum
Evan Ackerman
November 2, 2022
Researchers at Worcester Polytechnic Institute (WPI) and China's ShanghaiTech University demonstrated than an off-the-shelf quadruped robot can be transformed into a biped through the use of software and mechanical modifications. Installing a three-dimensionally printed stick on the shins of a quadruped robot's hind legs provide it sufficient support for standing and walking. After that, WPI's Andre Rosendo said, "We trained the robot in a simulated environment, and the walking gait, after being transferred to the real world, is stable, albeit slow.” Rosendo explained that bipedal robots “usually have more degrees of freedom in their legs to allow a more dynamic and adaptive locomotion, but in our case, we are focusing on the multimodal aspect to reap the benefits from two worlds: stability and speed from quadrupeds, manipulability, and a gain in operational height from bipeds."
|
Researchers Have a Fix for Reproducibility Crisis in Science
USC Viterbi School of Engineering
Julia Cohen
November 2, 2022
University of Southern California (USC) researchers have developed and tested a methodology for overcoming the difficulty of accurately reproducing scientific research. The researchers distilled a knowledge graph from more than 250 million academic studies, combining both micro- and macro-level information. They examined micro-level variables in published articles that are known to affect reproducibility, and incorporated macro-level relationship data between entities to build a network structure. Said USC's Jay Pujara, "We found that if we could put both of these things together, combine some features from the text and some features from the knowledge graph, we could do much better than either of those methods alone."
|
Colleges Go Offbeat for Cybersecurity Training
Inside Higher Ed
Susan D'Agostino
November 3, 2022
As colleges work to guard against cyberattacks, the University of Notre Dame has taken a different approach to cybersecurity awareness training in the form of a cybersecurity festival. Notre Dame's Chas Grundy said the festival was designed "to reach people's hearts and minds in a way that would stick and draw them into it as a counterpart to mandatory training." The festival featured activities like cybersecurity strongman games, a "go phish" activity in which participants were asked to identify the signs of a phishing email, and a lock-picking workshop. Stanford University hosted a similar festival, which featured lock-picking and hacking activities focused on safe cloud computing practices.
|
Algorithm May Provide Insights into Battery Corrosion
Argonne National Laboratory
Jared Sagoff
November 3, 2022
Researchers at the U.S. Department of Energy's Argonne National Laboratory have accelerated the solving of material-structure "jigsaw puzzles" from patterns revealed by X-rays via a new algorithm. The method enables real-time analysis of properties like corrosion or battery charging/discharging. The researchers trained the AutoPhaseNN algorithm on data generated by shining X-ray beams on a material and capturing the diffracted light. "This new algorithm is essentially able to solve what we call an inverse problem, going from the pieces of the puzzle to create the puzzle itself," said Argonne's Mathew Cherukara. "In essence, we're taking a set of observations and trying to identify the conditions that created them. Instead of solving the puzzle by iterating the process of trial and revision based on the prior knowledge, our algorithm assembles the puzzle from the broken pieces in a single step."
|
How Bellevue, WA, Uses Tech to Make Its Roads Safer
Government Technology
Mike Lindblom
November 1, 2022
In Bellevue, WA, smart cameras have been installed above 16 intersections to record the position of vehicles and pedestrians, in order to suggest potential changes to signals or lane layouts to avoid collisions. The system also will be rolled out at four future light-rail crossings. In addition, T-Mobile and Qualcomm are collaborating in Bellevue to deploy in-vehicle notifications of excess speed and to delay signal changes so pedestrians can cross streets safely. Bellevue's city mobility planning manager Franz Loewenherz said the city is looking to "be a national model for other communities."
|
Alzheimer's Drugs Tested Head-to-Head in Virtual Clinical Trial
Penn State News
Adrienne Berard
November 2, 2022
Scientists at Pennsylvania State University (Penn State), the University of Tennessee, Knoxville (UTK), and Duke University have conducted the first-ever virtual clinical trial of multiple Alzheimer's disease drugs. Penn State's Wenrui Hao said, "We're calling this a virtual clinical trial, because we used real, de-identified patient data to simulate health outcomes." The researchers used clinical and biomarker data to build a computational causal model to virtually test low- and high-dose regimens of aducanumab and single-dose regimens of donanemab. The outcomes aligned with the results of real-life human tests. The researchers also applied the model to develop personalized treatment plans for individual virtual patients.
|
Electronic Worker Tracking in NLRB Top Lawyer's Crosshairs
Bloomberg Law
Robert Iafolla
October 31, 2022
National Labor Relations Board (NLRB) General Counsel Jennifer Abruzzo announced her intent to campaign for a new legal framework to protect employees' rights under federal labor law from abusive electronic monitoring. Abruzzo said in a memo that she will urge the NLRB to designate employer surveillance or management practices an unfair labor practice if they interfere with union organizing or other legal activities. "Close, constant surveillance and management through electronic means threaten employees' basic ability to exercise their rights," Abruzzo said. She said NLRB law already criminalizes certain technology-facilitated surveillance/management practices, like the deployment of tracking technology in response to worker actions safeguarded by the National Labor Relations Act.
|
Why Fish Look Down When They Swim
Northwestern University McCormick School of Engineering
Amanda Morris
November 2, 2022
An international collaboration led by Northwestern University researchers simulated a zebrafish's brain and other factors to confirm that fish gaze downward when swimming as a form of adaptive evolutionary behavior. Northwestern's Emma Alexander said the researchers recorded video footage of rivers in India using cameras on robot arms "to model hypothetical scenarios where a simulated fish moved arbitrarily through a realistic environment." The team also followed zebrafish's motions within a sphere of light-emitting diodes (LEDs), and saw the fish looking down to pick up visual cues. The results of the simulations implied that fish look downward to understand environmental movements, then swim to offset them.
|
|
