Welcome to the February 3, 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 Model Makes Strides in Search for Covid-19 Treatments
Ohio State News
Emily Caldwell
February 1, 2021
A deep learning computer model developed by Ohio State University (OSU) computer scientists to predict human genes' interaction with drugs has identified at least 10 compounds as potential Covid-19 treatments. The DeepCE model taps the L1000 repository of human cell-line data concerning how gene expression changes in response to medications, and the DrugBank dataset on chemical structures and other details of roughly 11,000 approved and investigational drugs. The investigators applied DeepCE's gene expression prediction matrix to the genetic data in early Covid-19 papers and additional government information. Said OSU's Ping Zhang, "We can predict gene expression values for new chemicals not from one cell to one cell, but automatically predict the role of a drug on different cell lines and different genes."
|
How Google Searches Reveal the Hidden Cost of Lockdown
University of Warwick (U.K.)
January 27, 2021
Researchers at the U.K.'s University of Warwick, Canada's University of Ottawa, and France's Paris School of Economics and Aix-Marseille University found that Google Trends data from 10 countries across Europe and the U.S. between January 2019 and April 2020 demonstrated the impact of pandemic lockdowns on mental health. The researchers observed a sharp increase in the number of people searching on Google for terms related to boredom, loneliness, and worry at the beginning of the first lockdown. Said the University of Warwick’s Nick Powdthavee, "Our findings indicate that people’s mental health may have been severely affected by the pandemic and lockdown.” Powdthavee added, “It may be necessary to make sure support is provided to help those struggling most with lockdown.”
|
F-35's Buggy Software Prompts Pentagon to Call in Universities
Bloomberg
Anthony Capaccio
February 2, 2021
The Pentagon is consulting with U.S. universities to evaluate software on aerospace company Lockheed Martin's F-35 fighter jet, in the hope of correcting the buggy system. The F-35 program's Laura Seal said software experts at the Johns Hopkins University Applied Physics Laboratory, the Carnegie Mellon University Software Engineering Institute, and the Georgia Institute of Technology Research Institute are conducting an independent technical assessment. The $398-billion F-35 program involves Lockheed fighter jets equipped with more than 8 million lines of code each. Seal said the program office will analyze the assessment as part of "a broad range of information," then announce dates for program milestones, including simulated combat testing to rate the F-35's performance against the latest Russian and Chinese aircraft and air defenses.
|
Ford to Use Google's Android System in Most Cars
The Wall Street Journal
Mike Colias
February 1, 2021
Ford Motor announced a six-year agreement to use Google's Android operating system to power in-vehicle multimedia displays beginning in 2023, incorporating Google applications into all Ford models outside of China. The automaker also plans to utilize Google's cloud services to help develop in-car features, manage data streaming from vehicles, and organize production. Ford said the cloud services will support everything from labeling potential new service offerings to managing equipment in plants to streamlining supply chains. Sam Abuelsamid at consultancy Guidehouse Insights said Ford, GM, and other automakers are working with Google to provide Android as built-in software, which allows vehicle owners to download apps directly to the cars' displays.
*May Require Paid Registration
|
'Liquid' ML System Adapts to Changing Conditions
MIT News
Daniel Ackerman
January 28, 2021
A team of researchers from the Massachusetts Institute of Technology (MIT), the Institute of Science and Technology Austria, and the Vienna University of Technology in Austria has developed flexible algorithms, also known as "liquid" networks, that continually alter underlying equations to adapt to new data inputs. Unlike most neural networks, whose behaviors are fixed after training, a liquid network can adapt to the variability of real-world systems and is more resilient to unexpected or noisy data. Said MIT's Ramin Hasani, "Just changing the representation of a neuron, you can really explore some degrees of complexity you couldn't explore otherwise. The model itself is richer in terms of expressivity." The network outperformed other state-of-the-art time series algorithms by a few percentage points in predicting future values in datasets. Said Hasani, "Everyone talks about scaling up their network. We want to scale down, to have fewer but richer nodes."
|
High-Performance Computers Under Siege by Newly Discovered Backdoor
Ars Technica
Dan Goodin
February 2, 2021
Researchers at Slovak security firm Eset said a newly discovered backdoor allows hackers to remotely execute arbitrary commands on some high-performance computer networks. The Kobalos backdoor operates on Linux, FreeBSD, and Solaris, and code artifacts imply it may have previously run on AIX and the Windows 3.11 and Windows 95 platforms. Once installed, Kobalos infiltrates the file system of the target network and facilitates access to a remote terminal that allows intruders to run commands; infected systems also can become proxies connecting to other compromised servers, which can be linked to compromise a final target. Kobalos was released no later than 2019, and the group behind it was active throughout 2020. Eset researchers wrote that the backdoor's features and network evasion methods indicate those behind Kobalos “are much more knowledgeable than the typical malware author targeting Linux and other non-Windows systems."
|
South African Government Releases Its Own Browser Just to Re-enable Flash Support
ZDNet
Catalin Cimpanu
January 26, 2021
The South African Revenue Service (SARS) has released a customized Web browser exclusively for the purpose of re-enabling support for Adobe Flash Player, rather than port its current website from using Flash to HTML-based Web forms. Adobe's official support for Flash Player expired Dec. 31, 2020; the software company began blocking Flash content from playing inside the application starting Jan. 12, to prevent continued real-world use. SARS tweeted that starting that day it could not receive any tax filings through its Web portal, where upload forms were engineered as Flash widgets. The SARS eFiling Browser is a simplified version of the Chromium browser that both re-enables Flash support and allows users to access the SARS eFiling site. However, the browser is only available for Windows users, so those using other operating systems cannot file their taxes.
|
CCNY Researchers Demonstrate How to Measure Student Attention During Remote Learning
City College of New York
January 29, 2021
Jens Madsen and Lucas C. Parra at City College of New York (CCNY) have demonstrated an eye-tracking method for the measurement of attention during remote learning using standard Web cameras. The researchers were able to predict student performance on quizzes based on online educational videos, testing the theory that attentive learners' eye gazes are similar. The CCNY team proved inter-subject correlation of eye movement during educational video presentation is higher for attentive students, and that synchronized eye movements can predict individual test scores on video content. Said Parra, "With video content increasing online, remote sensing of attention to video at scale may have applications beyond education, including entertainment, advertising, or politics. The applications are limitless."
|
Roche to Use Quantum Computing for Drug Discovery
Fortune
Jeremy Kahn
January 28, 2021
Swiss pharmaceutical company Roche has partnered with the U.K.'s Cambridge Quantum Computing to identify new drug treatments for diseases, including Alzheimer's, via quantum computing. Cambridge Quantum Computing, which helps businesses use specialized algorithms that run on quantum computers, said Roche will start using a software platform that models quantum-level chemical interactions. Those algorithms can operate on a quantum system or, in certain cases, be employed to simulate quantum effects on a standard computer. Cambridge Quantum Computing's Ilyas Khan said, "For many years quantum computing has held out great promise for discovering new therapeutics that aid humanity in fighting some of the most devastating and damaging diseases. We are pleased that due to the careful and pioneering efforts of our research teams, some of this promise is starting to come to fruition."
|
Singapore Faces Talent Crunch as Tech Giants Scale Up
Reuters
Chen Lin; Aradhana Aravindan
January 27, 2021
Singapore's ambition to be a regional technology hub faces a talent crunch as major companies like China's Tencent conglomerate and U.S.-based Zoom Video Communications expand in the city-state, according to Reuters interviews with over a dozen recruiters, firms, and employees. Singapore's government has been retraining thousands on tech skills, while enrollment for information technology courses at Singapore colleges has climbed 17% over the past three years to about 7,600 for the 2020 academic year. The American Chamber of Commerce's Lei Hsien-Hsien said, "Certain member companies have been expanding their operations ... and looking to hire more data scientists, more coders." Added Lee, “The demand is very strong but the supply is relatively weak, which then slows down some of the expansion plans.”
|
A NEAT Reduction of Complex Neuronal Models Accelerates Brain Research
University of Bern (Switzerland)
January 27, 2021
Neuroscientists at Switzerland's University of Bern have crafted a computational method to simplify complex dendrite models, which could help artificial intelligence (AI) accelerate brain research. Bern's Willem Wybo said, "With these simpler models, neural responses can more easily be characterized and simulation of large networks of neurons with dendrites can be conducted." The technique leverages a mathematical relation between responses of simplified and complex dendrite models, rendering a linear model of the optimized objective within the parameters of the simplified model. The methodology has been compiled into the open source NEAT (NEural Analysis Toolkit) software solution for automating the simplification process. Said the university’s Walter Senn, ”With these simpler models, neural responses can more easily be characterized and simulation of large networks of neurons with dendrites can be conducted."
|
Thread-Based Sensors Can Measure Head Movements in Real Time
News-Medical Life Sciences
January 29, 2021
Flexible thread-based sensors developed by Tufts University researchers can measure neck movements, including the direction, angle of rotation, and degree of displacement of the head. The sensors, which could be woven into textiles, detect motion when the threads bend and create strain that alters the way they conduct electricity. In testing the sensors, the signals they generated were sent to a small Bluetooth module and wirelessly transmitted to a computer or smartphone, then analyzed using machine learning and translated into quantitative head movements in real time; results were found to be 93% accurate. Tufts' Yiwen Jiang said, "This is a promising demonstration of how we could make sensors that monitor our health, performance, and environment in a non-intrusive way."
|
AI Art Critic Can Predict Which Emotions a Painting Will Evoke
New Scientist
Edd Gent
January 29, 2021
Stanford University researchers have developed an artificial intelligence (AI) art critic that can predict emotional reactions to famous paintings and even provide explanations. The team created a massive dataset of human reactions using a survey of more than 6,000 people who were asked to choose the dominant emotion evoked by 81,000 paintings and write a caption describing their decision. After being trained with this data, the AI was tasked with predicting emotions for paintings it had never seen before and providing captions. In a test using human evaluators, captions produced by the AI “passed” as being written by a human 50% of the time.
*May Require Paid Registration
|
|
