Master's Degree in Electrical and Computer Engineering
Welcome to the April 27, 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."

Seattle-based BRINC Drones has provided unmanned aerial vehicles to Ukraine, including onethat can break through glass. Ukraine Sounds Alarm on Chinese Drones, Opening Skies to U.S. Startups
The Wall Street Journal
Heather Somerville
April 22, 2022

Hundreds of small drones from U.S. startups are searching for survivors and Russian hideouts in Ukraine, after Ukrainian government officials cited Chinese drones as a security risk. The Ukraine officials have called for limits on the deployment of drones made by China's SZ DJI Technology, saying technical glitches may have been intentionally inserted into the drones to undermine the country's defense. Since last month, Seattle-based BRINC Drones has contributed 10 drones to Ukraine and sold roughly 50 more to bolster Ukrainian defense, as well as for search-and-rescue and intelligence-gathering missions. Skydio's Adam Bry said his company gave dozens of drones to Ukraine's Ministry of Defense, and sold hundreds more to Ukraine-supporting government and non-government entities.

Full Article
*May Require Paid Registration
Estimating the Informativeness of Data
MIT News
Rachel Paiste
April 25, 2022

Massachusetts Institute of Technology (MIT) researchers created a scalable method for estimating the likely volume of information contained in any piece of data. The estimators of entropy via inference (EEVI) technique involves applying probabilistic inference algorithms to surmise probable explanations, then using these explanations to formulate high-quality entropy estimates. MIT's Feras Saad cites three reasons why the EEVI method yields useful upper and lower bounds on entropy. "First, the difference between the upper and lower bounds gives a quantitative sense of how confident we should be about the estimates," he says. "Second, by using more computational effort, we can drive the difference between the two bounds to zero, which 'squeezes' the true value with a high degree of accuracy. Third, we can compose these bounds to form estimates of many other quantities that tell us how informative different variables in a model are of one another."

Full Article

A paralyzed person controls a prosthetic arm with their brain activity via a brain-computer interface. Brain-Reading Devices Help Paralyzed People Move, Talk, Touch
Liam Drew
April 20, 2022

Scientists are implanting brain-computer interfaces (BCIs) in paralyzed people to restore their ability to move, communicate, and feel. Approximately 35 people have had a BCI implanted long-term in their brain, despite there being only a dozen or so laboratories conducting such research. Last year, researchers described a BCI subject using a robot arm that could transmit sensory feedback directly to his brain, while another subject rendered mute by a stroke regained the ability to speak when equipped with a prosthetic speech device, and a third person now can communicate at record speeds by imagining himself writing text by hand. Brown University's Leigh Hochberg said the biggest step forward for BCIs has come from machine learning, which has improved the ability to decode neural activity and determine a user's intention.

Full Article
Intel Develops AI to Detect Emotional States of Students
Tom's Hardware
Francisco Pires
April 18, 2022

An artificial intelligence (AI) software solution developed by Intel and Classroom Technologies to identify students' emotional states is generating controversy in the context of ethics and privacy. The technology, incorporated into Classroom Technologies’ Class software product, can classify students' body language and facial expressions whenever digital classes are conducted through Zoom. The software inputs students' video streams into the AI engine alongside contextual, real-time data that enables it to identify students' level of comprehension of subject matter. Intel's Sinem Aslan said the main goal is to improve one-on-one teaching by allowing educators to respond in real time to each student's emotional state. Among the software's caveats is that the act of labeling emotional states into easy-to-grasp categories invites error.

Full Article
Protecting Privacy in Surveillance Video While Mining for Data
IEEE Spectrum
Charles Q. Choi
April 19, 2022

A security system developed by the Massachusetts Institute of Technology's Frank Cangialosi and colleagues may allow analysis of surveillance video while protecting privacy. The Privid system permits analysts to review video for statistical data while shielding personally identifiable information. Privid accepts code from an analyst containing an inquiry that prompts an automatic count of, for example, masked people in a video feed and crowd density. The system segments that footage and runs the code on each chunk, then aggregates and adds noise to the data before sending it back to the analyst. Cangialosi said, "Privid might enable us to actually [make more productive use of] tons of footage from all of the cameras we already have around the world [and do so] in a safe way."

Full Article
A Solution for Moderating Junk Senders on WhatsApp
Rutgers Today
April 25, 2022

An international team of researchers has developed methods to help the WhatsApp mobile messaging application identify junk senders in public groups and automatically screen junk and spam for WhatsApp users. The researchers reviewed 2.6 million messages from 5,051 public-politics-related WhatsApp groups in India. Their analysis revealed nearly 10% of messages posted to these groups were junk, with junk ads for jobs constituting almost 30% of the dataset. URLs and phone numbers are key junk indicators, and the researchers produced a coding model to automatically detect those elements. The researchers said they can help WhatsApp administrators flag and remove these messages, adding that they had created a model that enables users to encode a signal that detects whether a message contains a phone number, a URL, both, or neither.

Full Article

A dollar symbol amid the data. Economic Tool Can Tame Pricing Algorithms
Scientific American
Ethan Wilk
April 26, 2022

A study by researchers at China's Tsinghua University found that price controls could help prevent price-setting algorithms from inadvertently discriminating against minority consumers or colluding to artificially inflate prices. The researchers offered proofs demonstrating how price controls theoretically could balance the surplus between consumers and sellers who use artificial intelligence algorithms. They also used "willingness to pay" (WTP) data to determine that the advantage sellers gained by knowing consumers' WTP would be eliminated by a control on the range of legal prices, although such controls would not impede sellers' profits. Some economists remain concerned price controls encourage collusion among market leaders, but the researchers say that effect is muted by modern pricing algorithms’ limited information sharing.

Full Article

The PsiQuantum Wafer, a silicon wafer containing thousands of quantum devices, including single-photon detectors. Chip Startups Using Light Instead of Wires Gain Speed, Investments
Jane Lanhee Lee
April 26, 2022

Momentum and capital are building for startups developing chips that process data via light rather than wires. Ayar Labs, which is developing silicon photonics technology that harnesses photons in chips, said it had raised $130 million from investors, including chip behemoth Nvidia. Other startups using silicon photonics to construct quantum computers, supercomputers, and chips for driverless vehicles also are attracting major investment. "What the Ayar Labs guys do so they solved the data interconnect problem for traditional high-performance [computing]," said Peter Barrett at venture capital firm Playground Global. "But it's going to be a while before we have pure digital photonic compute for non-quantum systems."

Full Article

Antibodies (green, aqua, pink) attack different parts of the SARS-CoV-2 viral particle (yellow/orange sphere). ML Model Can Distinguish Antibody Targets
University of Illinois News Bureau
Diana Yates
April 21, 2022

Researchers at the University of Illinois Urbana-Champaign (UIUC) and Scripps Research have developed a proof-of-concept machine learning model that can distinguish antibodies targeting influenza from those targeting the COVID-19 coronavirus. The researchers trained the model on antibody data from 88 published studies and 13 patents, teaching it to base its predictions of which virus an antibody will attack on each antibody's genetic sequencing. UIUC's Yiquan Wang said the model was nearly 85% accurate. UIUC's Nicholas Wu thinks that with sufficient data, scientists should be able to anticipate the virus an antibody will target, and the pathogenic features to which the antibody binds.

Full Article
Using AI to Detect Cancer from Patient Data Securely
University of Leeds Faculty of Medicine and Health (U.K.)
April 26, 2022

U.K. and German scientists have developed an artificial intelligence (AI) algorithm to predict the likelihood of cancer from patient data, while protecting personal information. "We were able to show that AI models trained with swarm learning can predict clinically relevant genetic changes directly from images of tissue from colon tumors," said Jakob Nikolas Kather at Germany's University Hospital RWTH Aachen. The team trained the algorithm on data from patient cohorts from Northern Ireland, Germany, and the U.S. Tests on image datasets generated at the U.K.'s University of Leeds validated the algorithm’s ability to predict the presence of different cancer subtypes in the images.

Full Article
LightPC Presents Resilient System Using Non-Volatile Memory
KAIST (South Korea)
April 25, 2022

Researchers at South Korea's Korea Advanced Institute of Science and Technology (KAIST) have developed hardware and software that guarantees data and execution resiliency. The Lightweight Persistence Centric System (LightPC) is not affected by power outages, since it uses only non-volatile memory. The tool validated execution while powering up and down mid-execution, demonstrating up to eightfold more memory, 4.3 times faster application execution, and 73% less power consumption versus traditional systems. The researchers said LightPC matches the performance of DRAM memory by minimizing internal volatile memory elements from non-volatile memory, exposing the non-volatile memory media to the host, and boosting parallelism to execute on-the-spot requests as quickly as possible.

Full Article

The virtual reality (VR) experience educates children about bushfires and helps them learn how to be safer in a bushfire incident. Immersive VR: Empowering Kids to Survive in Fire, Flood, War
University of South Australia
April 26, 2022

University of South Australia (UniSA) researchers have developed a virtual reality (VR) experience to teach children aged 10 to 12 about bushfires and making safety decisions during such emergencies. Through the VR experience, children are asked to care for a friend's dog prior to a bushfire event; they must complete problem-solving activities to ensure they and the dog remain safe. In an initial survey, 91% of study participants said they lacked knowledge of fires, and 67% felt too young to make safety decisions in a fire; after being educated by the VR experience, over 80% of study participants felt more confident about making safety decisions during bushfires. UniSA's Delene Weber said immersive VR could easily be applied to educate users about floods, wars, and other disasters.

Full Article
Improved Approach to the 'Traveling Salesperson Problem' Could Boost Logistics, Transport Sectors
University of Cambridge (U.K.)
April 26, 2022

Researchers at the U.K.'s University of Cambridge have developed an enhanced approach to the Traveling Salesperson Problem that yields high-quality solutions at a faster rate than other cutting-edge tools. The challenge involves finding the shortest possible delivery route for visiting multiple destinations in a single trip. The researchers' solution integrates a machine learning model supplying information about the previous best routes with a "metaheuristic" tool that draws the new route from this data. Said Cambridge's Ben Hudson, "Our goal with this research is to improve such methods so that they produce better solutions—solutions that result in lower distances being traveled and therefore lower carbon emissions and reduced impact on the environment."

Full Article
ACM Survey on Research Publishing in AI and Machine Learning
2022 CACM Website Survey Notice

Association for Computing Machinery

1601 Broadway, 10th Floor
New York, NY 10019-7434

ACM Media Sales

If you are interested in advertising in ACM TechNews or other ACM publications, please contact ACM Media Sales or (212) 626-0686, or visit ACM Media for more information.

To submit feedback about ACM TechNews, contact: [email protected]