Welcome to the April 14, 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."
|
|
|
People May Trust Computers More Than Humans
UGA Today
J. Merritt Melancon
April 13, 2021
New research by data scientists at the University of Georgia (UGA) indicates that people may be more willing to trust computer programs than fellow humans, especially for tasks that become too difficult. The researchers asked 1,500 volunteers to count the number of people in photos of a crowd, then provided suggestions from other people and from an algorithm. UGA's Aaron Schecter said as the number of people depicted in the photos grew, volunteers became more likely to follow the algorithm's suggestion rather than count themselves, or follow the "wisdom of the crowd." Said Ph.D. student Eric Bogert, “It seems like there’s a bias towards leaning more heavily on algorithms as a task gets harder and that effect is stronger than the bias towards relying on advice from other people.”
|
Vibrations From a Smartphone Can Help Spot Unsafe Drinking Water
New Scientist
Matthew Sparkes
April 14, 2021
Vibrations from a smartphone's ringtone can measure a liquid's viscosity, according to researchers at China's Shenzhen University. Shenzhen's Yandao Huang and colleagues designed a three-dimensionally-printed drinking cup with an external mount for an iPhone 7, and used the handset's vibrating motor to agitate liquids within the cup; the handset’s built-in motion sensor quantified the friction between the liquid molecules by detecting reflected motion waves. The team could differentiate between 30 types of liquid with more than 95% average accuracy. The phone could distinguish between liquids containing bacteria, dirt, or minerals through changes to viscosity, and differentiated between tap water, rain water, puddle water, and water with prolonged exposure to air, with an error rate of just 2.5%. Huang said the study’s results could lead to a simple test for measuring the safety of drinking water.
|
Is 'Femtech' the Next Big Thing in Healthcare?
The New York Times
Farah Nayeri
April 7, 2021
Numerous apps and technology companies are emerging in the "femtech" industry to address women's biological needs. In addition to period- and fertility-tracking apps, some tech companies have rolled out wearable breast pumps and pelvic exercise apps. Others focus on "menotech," or technologies that help women going through menopause, or on cervical, breast, and other cancers that affect women. Israeli startup MobileODT has developed a smart imaging device that leverages smartphones and artificial intelligence to screen for cervical cancer, delivering diagnoses in about a minute. French startup Lattice Medical has created a three-dimensionally (3D) printed hollow breast implant, which permits the regeneration of tissue and eventually is absorbed by the body. With women spending about $500 billion annually on medical expenses according to PitchBook, there is significant potential in this market.
*May Require Paid Registration
|
Computational Tool Could Help Design Futuristic Jet Engine Turbines
Texas A&M Today
Vandana Suresh
April 12, 2021
An algorithm developed by Texas A&M University (TAMU) researchers can forecast how novel materials will behave at extremely high temperatures, with potential application in the design of gas turbines for jet engines and electrical power generators. TAMU's Raymundo Arróyave said, "We have used an innovative and interdisciplinary approach for screening materials that is a million times faster than traditional techniques." The team calculated fundamental properties of MAX phases (high-temperature-tolerant ceramics) at zero kelvins with density functional theory, then fed those calculations to a machine learning model. The researchers employed computational thermodynamics to ascertain the most stable compounds for a given temperature and a certain MAX phase composition. Arróyave said, "The research will help in rapidly ruling out those elements that might form unstable oxides at the material design phase."
|
Some FDA-Approved AI Medical Devices Are Not 'Adequately' Evaluated, Stanford Study Says
VentureBeat
Kyle Wiggers
April 12, 2021
Certain artificial intelligence (AI)-powered medical devices approved by the U.S. Food and Drug Administration (FDA) are susceptible to data shifts and bias against underrepresented patients, according to a study by Stanford University researchers. The researchers compiled a database of FDA-approved medical AI devices, and analyzed how each was evaluated before approval. They found 126 of 130 devices approved between January 2015 and December 2020 underwent only retrospective studies at submission, and none of the 54 approved high-risk devices were assessed via prospective review. The researchers contend prospective studies are needed especially for AI medical devices, given that field applications of the devices can deviate from their intended uses. They also said data about the number of sites used in an evaluation must be "consistently reported," in order for doctors, researchers, and patients to make informed decisions about the reliability of a AI-powered medical device.
|
Low-Cost NIST Demo Links Public Safety Radios to Broadband Wireless Network
NIST News
April 12, 2021
A low-cost prototype computer system developed by U.S. National Institute of Standards and Technology (NIST) engineers can connect older analog public safety radios to modern broadband networks. The prototype links analog Land Mobile Radio (LMR) handsets and towers to a Long-Term Evolution (LTE) server, allowing LMR radio users and LTE network users to communicate as though they both are on the same push-to-talk network. The system integrates software-defined radio, an open-source software environment for managing software radio, and a user interface for LTE handsets. NIST's Jordan O'Dell said, "There isn't a commercial option that compares to what we are developing. The goal here is to create a prototype and accelerate technology development in industry that will fill a significant gap."
|
KAUST Collaboration With Intel, Microsoft, University of Washington Accelerates Training in ML Models
HPCwire
April 12, 2021
Researchers at Saudi Arabia's King Abdullah University of Science and Technology (KAUST), Intel, Microsoft, and University of Washington have achieved a more than five-fold increase in the speed of machine learning on parallelized computing systems. Their "in-network aggregation" technology involved inserting lightweight optimization code in high-speed network devices. The researchers used new programmable dataplane networking hardware developed by Intel's Barefoot Networks to offload part of the computational load during distributed machine learning training. The new SwitchML platform enables the network hardware to perform data aggregation at each synchronization step during the model update phase. KAUST's Marco Canini said, "Our solution had to be simple enough for the hardware and yet flexible enough to solve challenges such as limited onboard memory capacity."
|
Researchers Identify Indicators for Audience Measurement Model for Streaming Platforms
Universitat Oberta de Catalunya (Spain)
Pablo Ramos
April 8, 2021
A study by researchers at Spain's Universitat Oberta de Catalunya (UOC) analyzed Netflix's audience behavior and measurement systems as the potential basis for a more-reliable audience measurement model. UOC's Elena Neira said, "At present, viewers can decide how, where and when to watch a series or a film, and therefore the traditional audience measurement models are not capable of covering the new consumer reality fully." Neira said the number of people who have shifted from watching traditional television to online content is unknown. The researchers said an improved audience measurement model should take into consideration such things as audience fragmentation; viewing intensity, or binge watching; and external factors like social media usage, numbers of downloads, and numbers of Google searches.
|
Millions of Devices at Risk From NAME:WRECK DNS Bugs
Computer Weekly
Alex Scroxton
April 13, 2021
Researchers at cybersecurity provider Forescout Research Labs and Israeli cybersecurity consultancy JSOF discovered nine new Domain Name System (DNS) vulnerabilities that could imperil more than 100 million connected Internet of Things (IoT) devices, at least a third of them located in the U.K. Collectively designated NAME:WRECK, the bugs affect four popular Transmission Control Protocol/Internet Protocol (TCP/IP) stacks: FreeBSD, IPnet, Nucleus NET, and NetX. Malefactors who exploit the vulnerabilities in a denial of service or remote code execution attack could disrupt or hijack targeted networks. Forescout's Daniel dos Santos said, "Complete protection against NAME:WRECK requires patching devices running the vulnerable versions of the IP stacks, and so we encourage all organizations to make sure they have the most up-to-date patches for any devices running across these affected IP stacks."
|
China's Factories Automate as Worker Shortage Looms
CNBC
Evelyn Cheng
April 9, 2021
China's working-age population has dropped by more than 5 million in the last decade, and factories are responding to the resulting labor shortage with automation. Home appliance giant Midea, for instance, has rolled out a three-year plan to outfit its 34 factories with more technology. Midea's Shirley Zhou said its two factories that have implemented sensors and robots have seen an almost 30% jump in assembly efficiency. Datacenter operator Equinix's Jeremy Deutsch said technology to track and analyze global production is of particular interest, and factory digitalization is fueling demand for datacenters. Said Victor Du at consulting firm Alvarez & Marsal Asia, “As a society, the concern should (be) achieving the same level of manufacturing output, or even higher quality, higher output, with a lower population after 20, 30 years. If you look at this point, digitalization or upgrading of technology will be very necessary.”
|
Speeding Up Sequence Alignment Across the Tree of Life
Max Planck Gessellschaft (Germany)
April 7, 2021
Scientists at the Max Planck Institutes of Developmental Biology and the Max Planck Computing and Data Facility (MPCDF) in Germany have developed a sequence search engine that compares the biochemical composition of different species with unmatched accuracy and speed. The tool extends the DIAMOND search algorithm to retain its speed while gaining the sensitivity of the basic local alignment search tool (BLAST) algorithm used for comparing primary biological sequence information. The enhanced DIAMOND will let researchers conduct comparative genomics research with BLAST-level accuracy, 80 to 360 times faster. MPCDF's Klaus Reuter said, "In addition, DIAMOND enables researchers to perform alignments with BLAST-like sensitivity on a supercomputer, a high-performance computing cluster, or the cloud in a truly massively parallel fashion, making extremely large-scale sequence alignments possible in tractable time."
|
French Army Testing Boston Dynamics' Robot Dog Spot in Combat Scenarios
The Verge
James Vincent
April 7, 2021
France's military used robotics developer Boston Dynamics' four-legged Spot robot in training exercises, raising questions about future applications. The French military school École Spéciale Militaire de Saint-Cyr described the tests as "raising students' awareness of the challenges of tomorrow," including the "robotization of the battlefield." A more detailed account by French newspaper Ouest-France said Spot was one of a number of robots tested by students at the École Militaire Interarmes (Combined Arms School), apparently to evaluate their reconnaissance utility. Spot's terms and conditions prohibit its use "to harm or intimidate any person or animal, as a weapon, or to enable any weapon," and Boston Dynamics' Michael Perry said he is adamantly against the robot's weaponization.
|
Using AI to Diagnose Neurological Diseases Based on Motor Impairment
Heidelberg University (Germany)
April 7, 2021
Researchers at Germany's Heidelberg University, working with collaborators in Switzerland, have developed a machine learning (ML) technique for recognizing motor impairments, in order to diagnose neurological diseases. The team's unsupervised behavior analysis and magnification using deep learning (uBAM) method features an ML-based algorithm that utilizes artificial neural networks, and identifies independently and fully automatically characteristic behavior and pathological deviations. The uBAM interface's underlying convolutional neural network was trained to identify similar movement behavior in the case of different subjects, despite differences in outward appearance. Heidelberg's Björn Ommer said, "As compared to conventional methods, the approach based on artificial intelligence delivers more detailed results with significantly less effort."
|
|
