Welcome to the July 21, 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."
|
|
|
Computer Science Professor Wins 'Test of Time' Award for Influential Paper
USC Viterbi News
Caitlin Dawson
July 20, 2021
ACM's Special Interest Group on Algorithms and Computation Theory named the University of Southern California Viterbi School of Engineering Seely G. Mudd Professor of Computer Science and Mathematics Shang-Hua Teng, and his collaborator, Yale University professor of applied mathematics and computer science Daniel A. Spielman, recipients of the Symposium on Theory of Computing Test of Time Award. Teng and Spielman authored a paper on smoothed analysis of algorithms, which offers a more realistic comprehension of algorithmic performance. The authors determined that algorithms, especially the simplex algorithm for linear programming, function as long as the input has noise, because real-world data typically contains noise. Said Teng, "Our theory demonstrates that these properties can in fact be helpful to algorithms in practice, because under these conditions, the worst-case scenarios are harder to arise." These findings have been applied in a wide range of practical algorithms, including faster Internet communications, deep learning, data mining, game theory, and personalized recommendation systems.
|
China Spy Agency Blamed by U.S., Others of Using Contract Hackers
Voice of America News
Steve Herman
July 19, 2021
U.S. President Joe Biden said he expected to receive a report Tuesday detailing how China's Ministry of State Security has employed contract hackers to hold U.S. businesses hostage with ransomware. This follows the Biden administration’s public accusation that Beijing is conducting unsanctioned cyber operations worldwide. An international coalition claims China launched a zero-day hack in March that impacted tens of thousands of organizations through Microsoft Exchange servers. In a jointly issued advisory Monday, the U.S. National Security Agency, Cybersecurity and Infrastructure Security Agency, and the Federal Bureau of Investigation said they “have observed increasingly sophisticated Chinese state-sponsored cyber activity targeting U.S. political, economic, military, educational, and CI (critical infrastructure)?personnel and organizations.”
|
How AI Is Fighting Wildfires
The New York Times
Jill Cowan
July 15, 2021
The University of California, San Diego's Ilkay Altintas and colleagues have spent eight years helping fire managers and scientists combat wildfires through the WIFIRE Lab. Such efforts involve combining vast troves of information with artificial intelligence to rapidly predict fires' progression, and help plan countermeasures. Examples include a fire map with an intuitive interface, which Altintas said was used to help fight every major fire in California; and the BurnPro 3D product, which displays three-dimensional images of vegetation down to a one-meter resolution. Altintas said one of lab's primary goals is to make data and data analysis accessible to a wide range of collaborators nationwide, since that is the only way researchers and officials can keep pace with evolving environmental crises.
*May Require Paid Registration
|
Google Demonstrates Vital Step Towards Large-Scale Quantum Computers
New Scientist
Matthew Sparkes
July 14, 2021
Google scientists have demonstrated a key step toward large-scale quantum computing through the Sycamore quantum computer's ability to spot and correct computational errors. Sycamore has logical quantum bits (qubits) ranging in size from five to 21 physical qubits, with each additional physical qubit entailing an exponential decline in logical qubit error rates. The Google AI Quantum researchers could measure the extra qubits that did not collapse their state, while still returning sufficient data to deduce the occurrence of errors. The odds of a logical qubit encountering an error climb as the number of qubits within it increases, and Google's approach has not yet reached the threshold where error-correction features correct more errors than they generate. Sycamore currently features 54 physical qubits, and the team believes mature quantum computers will need 1,000 qubits for each logical qubit.
|
Contact-Aware Robot Design
MIT News
Rachel Gordon
July 19, 2021
A system developed by Massachusetts Institute of Technology (MIT) researchers uses software to optimize the shape and control of a robotic manipulator for a specific task. After manipulating the design and simulating the robot manipulator performing a task, the system assigns an optimization score to assess its design and control. The researchers used "cage-based deformation" to change the geometry of a shape in real time, in order to create more involved manipulators. This involves putting a cage-like structure around a robotic finger, with the algorithm altering the cage dimensions automatically to create more sophisticated, natural shapes. MIT's Jie Xu said, "We not only find better solutions, but also find them faster. As a result, we can quickly score the design, thus significantly shortening the design cycle."
|
Algorithm May Help Scientists Demystify Complex Networks
Penn State News
Matt Swayne
July 16, 2021
A new algorithm capable of analyzing models of biological systems can lead to greater understanding of their underlying decision-making mechanisms, with implications for studying how complex behaviors are rooted in relatively simple actions. Pennsylvania State University (Penn State)'s Jordan Rozum said the modeling framework includes Boolean networks. Said Penn State's Reka Albert, "Boolean models describe how information propagates through the network," and the nodes' on/off states eventually slip into repeating patterns that correspond to the system's stable long-term behaviors. Complexity can scale up dramatically as the system incorporates more nodes, particularly when events in the system are asynchronous. The researchers used parity and time-reversal transformations to boost the efficiency of the Boolean network analysis.
|
Platform Allows Autonomous Vehicles to Safely Drive at Small Distances
Eindhoven University of Technology (Netherlands)
July 12, 2021
A Ph.D. student at the Eindhoven University of Technology in the Netherlands, Robbin van Hoek, has integrated the benefits of cooperative and autonomous vehicles into a single platform. Cooperative vehicles exchange motion data via vehicle-to-vehicle or vehicle-to-infrastructure communication, but generally can perform only a single task, like following a preceding vehicle. Autonomous vehicles use radar, LiDAR, and computer vision systems to detect the road, other traffic participants, and other relevant features or obstacles and can plan a vehicle trajectory. Combining the two enables the vehicle to take advantage of motion data communicated by other vehicles by following them at close distances, while preventing traffic jams and autonomously making decisions like, for instance, overtaking a vehicle that is driving too slowly.
|
Rats Took Over This Pacific Island. Now Drones Are Leading the Fightback
Wired
Tom Ward
July 16, 2021
The conservation group Island Conservation plans to use drones to try to eliminate rats on Tetiaroa atoll and two other islands in French Polynesia, beginning in August. Hexacopter (six-rotor) drones from New Zealand's Envico Technologies will be used to drop 30 tons of rat poison on the islands over a two-week period. A trial drone initiative undertaken in 2019 on Seymour Norte in the Galapagos resulted in that island being declared 100% rat-free two years later. Island Conservation's David Will said, "We've been watching drone technology for a number of years with the idea that it can dramatically reduce cost and also democratize island restoration by allowing local experts to be able to fly them using precision automating processes."
|
Tool to Explore Billions of Social Media Messages Could Predict Political, Financial Turmoil
University of Vermont
Joshua Brown
July 15, 2021
Researchers at the University of Vermont (UVM), Charles River Analytics, and MassMutual Data Science (the data science unit of the Massachusetts Mutual Life Insurance Company) have developed an online tool that uncovers the stories within the billions of Twitter posts made since 2008. The Storywrangler, powered by UVM's supercomputer at the Vermont Advanced Computing Core, breaks tweets into one-, two-, and three-word phrases across 150 languages and determines the frequencies with which more than a trillion words, hashtags, handles, symbols, and emoji appear. The data can be used to analyze the rising and falling popularity of words, ideas, and stories around the globe. Said UVM's Peter Dodds, "This tool can enable new approaches in journalism, powerful ways to look at natural language processing, and the development of computational history."
|
GPUs Can Now Analyze a Billion Complex Vectors in Record Time
IEEE Spectrum
Michelle Hampson
July 16, 2021
Researchers at Facebook AI Research have developed an approach to leverage graphical processing units (GPUs) for the analysis of high-dimensional media. The researchers developed an algorithm that enables GPUs to host and analyze a library of vectors, and employed a technique that compresses vector databases so the GPUs can analyze them more easily. Facebook AI's Jeff Johnson said, "By keeping computations purely on a GPU, we can take advantage of the much faster memory available on the accelerator, instead of dealing with the slower memories of [central processing unit-based] servers and even slower machine-to-machine network interconnects within a traditional supercomputer cluster." With the new approach, four GPUs were able to analyze more than 95 million high-dimensional images in 35 minutes, 8.5 times faster than previous techniques.
|
Brain Injury Computer Models Map Brain Blood Vessels in Highest Resolution Yet
Imperial College London (U.K.)
Caroline Brogan
July 15, 2021
A new computer model of traumatic brain injury (TBI) can map a rat's brain vasculature with the highest resolution to date. Scientists at Imperial College London (ICL) in the U.K. designed the model to incorporate blood vessels in the rat's brain just 10 microns in diameter. The team determined adjacent blood vessels maintain significantly distinct levels of stress, depending on their alignment with neighboring ones. The digital models enabled the accurate prediction of stress distribution in small blood vessels, and allow time to be slowed to examine the details of TBI more closely. ICL's Siamak Khosroshahi said their approach "explains the unrecognized role of the vascular anatomy and shear stresses in how large forces cascade through the brain."
|
Air-Powered Computer Memory Helps Soft Robot Control Movements
UC Riverside News
Holly Ober
July 16, 2021
A new air-powered computer memory can be utilized to control soft robots, thanks to engineers at the University of California, Riverside (UC Riverside). The researchers designed an 8-bit pneumatic random-access memory (RAM) chip that substituted microfluidic valves for electronic transistors. The valves stay sealed against a pressure differential even when detached from an air supply line, generating trapped pressure differentials that serve as memories and maintain the states of a robot's actuators. Dense valve arrays can conduct sophisticated operations and streamline the bulky, power-intense hardware typical of pneumatic robot controls. The UC Riverside team incorporated the pneumatic RAM chip into a pair of three-dimensionally-printed rubber hands, and induced a robot to use them to play notes, chords, and an entire song on a piano by varying the mixture of atmospheric pressure and vacuum within the channels on the chip.
|
Scientists Decipher Activity of the Brain in Deep Sleep
News-Medical Life Sciences
Emily Henderson
July 16, 2021
A new system can decode the brain's activity during sleep to analyze memory consolidation. Scientists at Switzerland's University of Geneva (UNIGE) blended artificial intelligence, magnetic resonance imaging (MRI), and electroencephalography to examine the sleeping brain's ability to assess memories in order to store only the most useful ones. The team placed participants in an MRI in the early evening and had them play a face-recognition game and solve a three-dimensional maze; participants were unaware only one game could be won. The subjects then slept in the MRI for one or two hours as their brain activity was again recorded, and UNIGE's Virginie Sterpenich said patterns observed in deep sleep indicated their brains were reliving only the game won by reactivating regions used during wakefulness. Tests showed that when participants performed better, more brain regions related to the game were activated during their subsequent sleep.
|
|
