Welcome to the April 2, 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."
|
|
|
System Using Smartphone or Computer Cameras to Measure Pulse, Respiration Rate Could Help Future Personalized Telehealth
UW News
Sarah McQuate
April 1, 2021
Researchers at the University of Washington (UW) and Microsoft Research have developed a system that uses a person's smartphone or computer camera to read pulse and respiration from video of their face. The system preserves privacy by operating on the device rather than in the cloud, while machine learning (ML) records subtle changes in light reflecting off a person's face, which correlates with changing blood flow. The researchers first trained the system on a dataset of facial videos, and individual pulse and respiration data from standard field instruments; it calculated vital signs using spatial and temporal information from the videos. Said UW’s Xin Liu, “Every person is different, so this system needs to be able to quickly adapt to each person’s unique physiological signature and separate this from other variations, such as what they look like and what environment they are in.”
|
Scientists Create Next Generation of Living Robots
UVM Today
March 31, 2021
Computer scientists at the University of Vermont (UVM), working with Tufts University biologists, followed up on the development of self-healing biological machines from frog cells (Xenobots) by creating a new generation of Xenobots that self-assemble from individual cells, do not use muscle cells for movement, and are capable of recordable memory. The next-generation Xenobots outperformed the previous generation, and also were shown to support molecular memory and self-healing. Tufts' Doug Blackiston said, "This approach is helping us understand how cells communicate as they interact with one another during development, and how we might better control those interactions."
|
MIT Study Finds 'Systematic' Labeling Errors in Popular AI Benchmark Datasets
VentureBeat
Kyle Wiggers
March 28, 2021
An analysis by Massachusetts Institute of Technology (MIT) researchers demonstrated the susceptibility of popular open source artificial intelligence benchmark datasets to labeling errors. The team investigated 10 test sets from datasets, including the ImageNet database, to find an average of 3.4% errors across all datasets. The MIT investigators calculated that the Google-maintained QuickDraw database of 50 million drawings had the most errors in its test set, at 10.12% of all labels. The researchers said these mislabelings make the benchmark results from the test sets unstable. The authors wrote, "Traditionally, machine learning practitioners choose which model to deploy based on test accuracy—our findings advise caution here, proposing that judging models over correctly labeled test sets may be more useful, especially for noisy real-world datasets."
|
VR Brings Joy to People in Assisted-Living Facilities
The Wall Street Journal
Bonnie Miller Rubin
March 26, 2021
Long-term care communities increasingly are using virtual reality (VR) devices and systems to improve residents' wellness and quality of life amid pandemic-related restrictions on visitors and activities. Studies have documented the positive effects of the technology, with a 2018 field study by researchers at the Massachusetts Institute of Technology finding that nearly 39% of assisted-living residents shown VR images related to travel and relaxation reported better perceived overall health. Companies like MyndVR sell VR packages to senior-care facilities, while others like Embodied Labs use VR to train caregivers. MyndVR's Paula Harder said, "Residents in our memory-care neighborhood have been observed to be more oriented to their surroundings...and even more coordinated in their speech and movement."
*May Require Paid Registration
|
Australian Researchers Use ML to Analyze Rock Art
ZDNet
Asha Barbaschow
April 1, 2021
South Australian researchers at Flinders University are analyzing the evolution of rock art via machine learning. The team studied images of art collected during surveys of the Arnhem Land region using previously trained and published convolutional neural network models and dataset combinations each designed for object classification. The Flinders investigators used transfer learning to deploy these networks on the dataset without retraining, and analyzed the models' response or activation on a rock art dataset. Flinders' Daryl Wesley said the computer observed over 1,000 different types of objects, and learned to differentiate them by looking at photos. Flinders' Ian Moffat said transfer learning removed a significant amount of human bias from the analysis, and an especially exciting aspect of this research is that "it is replicating the results of other studies that have used a more traditional approach."
|
Smart Bandage Could Hasten Healing, Might Even Detect Covid
IEEE Spectrum
Kathy Pretz
March 29, 2021
A smart bandage developed by researchers at Scotland's University of Glasgow could reduce the number of in-person doctor's visits required for patients with open wounds. The clear, flexible adhesive patch applies pressure to aid healing and uses sensors to measure the amount of strain on the skin and the patient's temperature. That data is transmitted wirelessly via a near-field communication tag to a smartphone app developed by the researchers, which can send the data to healthcare providers to determine whether the bandage is providing the correct amount of pressure, or whether the patient has a fever that could indicate an infection. The bandage also can monitor the lung function of patients with respiratory conditions or who are on a ventilator, and detect symptoms of Covid-19. Notifications from the app could help speed testing and prevent patients with Covid-19 from infecting others.
|
Getting to the Core of HIV Replication
Texas Advanced Computing Center
Jorge Salazar
March 31, 2021
Supercomputer models supported by the U.S. National Science Foundation-funded Extreme Science and Engineering Discovery Environment have helped to explain how the human immunodeficiency virus-1 (HIV) brings DNA synthesis-driving nucleotides into its core. Researchers at the University of Delaware (UD), the University of Pittsburgh School of Medicine, and the Harvard Medical School used the Texas Advanced Computing Center's Stampede2 supercomputer, and the Pittsburgh Supercomputing Center's Bridges system, to demonstrate that a virus can recruit small molecules from a cellular environment into its core to perform a process critical to its replication. “We are discovering new biology,” said UD's Juan R. Perilla said, "To my knowledge, it's the first piece of work that comprehensively shows an active role of the capsids in regulating a very specific lifecycle of the virus, not only computationally, but also in vitro assays and ultimately in the cells."
|
Threatened by Amazon, Albertsons Partners With Google to Digitalize the Grocery Shopping Experience
Forbes
Johan Moreno
March 31, 2021
A multi-year partnership between Google and Albertsons Companies, the No. 2 grocer in the U.S. by store count, aims to digitalize the grocery shopping experience at a time when competitor Amazon is expanding its brick-and-mortar grocery footprint via Amazon Fresh. The partnership would integrate Google's Cloud AI technologies into Albertsons' operations to create what Google called "the world's most predictive grocery engine." The move could help Google as it works to make its cloud business profitable and help Albertsons take on Amazon Fresh. Albertsons has launched Google's Business Messages in a limited capacity to provide shoppers with Covid-19 vaccine information, and plans to release the shoppable maps and predictive grocery-list-building capability soon.
|
Novel Disaster Communication System Empowers Communities and Can Save Lives
TU Delft (Netherlands)
March 30, 2021
An energy-efficient communication system developed by researchers at the Delft University of Technology (TU Delft) in the Netherlands relies on smartphones, so more people can communicate longer during a disaster. The system, Self-Organization for Survival (SOS), ensures that even people whose phone batteries are low can communicate during the first crucial 72 hours after a disaster. TU Delft's Indushree Banerjee said, "SOS is designed in such a way that phones choose to connect only with one other phone, the one with the highest battery charge in their transmission range.” Banerjee explained that by limiting contacts and switching connections, “The phone battery will last a lot longer and prevent any single phone from being unnecessarily overused" during an emergency.
|
Early Warning System for Self-Driving Cars
Technical University of Munich (Germany)
March 30, 2021
An early warning system for self-driving vehicles developed by researchers at Germany's Technical University of Munich (TUM) leverages artificial intelligence (AI) to learn from real traffic situations. During tests on public roads, the researchers identified about 2,500 situations that required driver intervention; they found the system issued warnings about potentially critical situations seven seconds in advance with more than 85% accuracy. Using a recurrent neural network, the system recognizes patterns in the data collected by sensors and cameras and will warn drivers if it identifies a situation the control system has had difficulty handling in the past. TUM's Eckehard Steinbach said, "We limit ourselves to the data based on what actually happens and look for patterns. In this way, the AI discovers potentially critical situations that models may not be capable of recognizing, or have yet to discover."
|
Computational Models to Understand Colon Cancer
University of Illinois Institute for Genomic Biology
Ananya Sen
March 29, 2021
Researchers at the University of Illinois Urbana-Champaign (UIUC) and Mayo Clinic, Minnesota facilitated a new understanding of colon cancer through the development of computational models of tumor dynamics. UIUC's Saba Ghaffari said, "The advantage of the model is that it can integrate different types of experimental data, which is not an easy task. It gave us a list of transcription factors, ranked based on their relevance to colorectal cancer aggressiveness." The team used the model to integrate data from multiple studies to identify signaling pathways relevant to metastasis-associated processes. The researchers tested the model's predictions using human cancer cell lines, and determined the transcription factors played a role in boosting the aggressiveness of colorectal cancer cells.
|
Flagging Coronavirus Misinformation Tweets Changes User Behaviors, Research Shows
The University of Alabama in Huntsville
Jim Steele
March 29, 2021
University of Alabama in Huntsville (UAH) researchers found flagging tweets containing misinformation related to the coronavirus impacts their credibility among most Twitter users. UAH's Candice Lanius, William MacKenzie, and Ryan Weber surveyed 299 respondents using Amazon’s Mechanical Turk on whether they felt Covid-19 numbers were underreported, overreported, accurate, or had no opinion; participants convinced of underreporting or overreporting were influenced by tweets claiming those respective views. When presented successive flags that tweets were either from a suspected bot account or contained misinformation, participants' skepticism increased. Said MacKenzie, "Our research would suggest that individuals want to consume social media that is factual, and if mechanisms are in place to allow them to disregard false information, they will ignore it."
|
Researchers Develop 'Explainable' Algorithm
University of Toronto (Canada)
Matthew Tierney
March 31, 2021
An "explainable" artificial intelligence (XAI) algorithm developed by researchers at Canada's University of Toronto (U of T) and LG AI Research was designed to find and fix defects in display screens. XAI addresses issues with the "black box" approach of machine learning strategies, in which the artificial intelligence makes decisions entirely on its own. With XAI's "glass box" approach, XAI algorithms are run simultaneously with traditional algorithms to audit the validity and level of their learning performance, perform debugging, and identify training efficiencies. U of T's Mahesh Sudhakar said LG "had an existing [machine learning] model that identified defective parts in LG products with displays, and our task was to improve the accuracy of the high-resolution heat maps of possible defects while maintaining an acceptable run time." The new XAI algorithm, Semantic Input Sampling for Explanation (SISE), outperformed comparable approaches on industry benchmarks.
|
|
