Welcome to the February 7, 2020 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."
|
|
|
TextFooler Generates Adversarial Text to Strengthen Natural Language Models
VentureBeat
Kyle Wiggers
February 7, 2020
Researchers at the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory (CSAIL), the University of Hong Kong, and Singapore’s Agency for Science, Technology have developed a framework for creating adversarial text examples that can deceive natural language processing (NLP) systems. MIT's Di Jin said the TextFooler framework demonstrates potentially disastrous scenarios for compromising critical online systems based on text classification models—including voice assistants and email spam filters. Jin acknowledged that TextFooler can be used to attack any NLP model to gauge its robustness, adding that in addition, "the generated adversaries can be used to improve the robustness and generalization of deep learning models via adversarial training, which is a critical direction of this work."
|
Apps Could Take Up Less Space on Your Phone, Thanks to 'Streaming' Software
Purdue University News
Kayla Wiles
February 6, 2020
Purdue University researchers have developed software that reduces the amount of space apps take up on a smartphone. The AppStreamer software "streams" data and code resources to an app from a cloud server, so the app only uses the space it needs on the phone when accessed. Testing showed AppStreamer slashed storage requirements by at least 85% for popular gaming apps on the Android platform. The middleware automatically predicts when to retrieve data from the cloud server. AppStreamer is designed also for use in edge computing, where it could facilitate instant downloads, faster runtime, and less space consumption when used on a 5G network.
|
Israel's Sheba Unveils Coronavirus Telemedicine Program
The Jerusalem Post
Maayan Jaffe-Hoffman
February 5, 2020
The Sheba Medical Center in Israel has introduced the world's first known coronavirus telemedicine program, applying a remote-controlled robot and an app to the treatment of patients suspected of being infected with the pathogen. The hospital is deploying a Vici bot designed by virtual healthcare firm Intouch Health to monitor patients’ vital signs and minimize direct contact between patients and staff. Sheba's telemedicine app was designed by oncology patient management platform Datos Health, to monitor patients from their homes via video. Said Sheba's Galia Barkai, "Although we don't have any [coronavirus]-positive patients in Israel, we are always dealing with suspected patients and preparing for the worst-case scenario. So, we are creating all these systems to help us deal with the occasion when we might have to deal with many patients."
|
Wearable Tech Tells Drowsy Truckers It's Time to Pull Over
The New York Times
Julie Weed
February 6, 2020
Trucking companies are equipping drivers with wearable monitors that check for signs of fatigue, and notify them when it’s time to pull over and rest. Optalert makes glasses that measure eye-blinks with a light-emitting diode monitor, and displays those measurements in real time on a dash-mounted device with alarms and alerts. Meanwhile, a headset by the software firm Maven Machines detects the angle of the driver's gaze, mirror checks, head bobs and jerks, and other signs of drowsiness. The device uses software that pulls data from internal accelerometers, onboard sensors, and global-positioning system data from cellphone towers. Some products attempt to predict drowsiness, and fatigue management technology developer Fatigue Science sells software that analyzes sleep data from wearables and helps to boost drivers' awareness of fatigue risk ahead of time.
*May Require Paid Registration
|
World's First Classical Chinese Programming Language
IEEE Spectrum
Charles Q. Choi
January 31, 2020
Carnegie Mellon University graduate Lingdong Huang has created the first programming language based on classical Chinese, which includes a renderer that displays programs in a manner similar to pages from ancient Chinese texts. Huang and others have written dozens of programs with his wenyan-lang language, many of which are renditions of mathematical algorithms in ancient Chinese volumes—including a fortune-telling algorithm derived from the I-Ching. Wenyan-lang employs both traditional Chinese characters and classical Chinese grammar, and its design had to incorporate how classical Chinese does not use spaces to split sentences into words. Huang said coding languages only possess a set amount of keywords, and wenyan-lang progresses from the longest keyword to the shortest keyword in each program to decipher the programmer's intentions. Huang envisions wenyan-lang as contributing to classical Chinese's survival, by forming a community of users.
|
Improving AI's Ability to Identify Students Who Need Help
NC State News
Matt Shipman
February 5, 2020
Researchers at North Carolina State University's Center for Educational Informatics have developed an artificial intelligence (AI) model to predict students' absorption of knowledge via educational gameplay. The model utilizes multi-task learning, in which it is asked to execute multiple tasks to forecast whether students would answer each question on a test correctly, based on their game behavior. The AI was assigned to learn 17 tasks correlating with the test's 17 questions. The model studies each student's gameplay and question-answering pattern on the test's first question, and identifies common behaviors of students who answered the question correctly or incorrectly to ascertain how new students would answer; it simultaneously performs this function for all questions. The multi-task model is about 10% more accurate than models dependent on conventional AI training, and the researchers think the AI could help flag students who may need additional instruction.
|
Researcher Creates Drone System to Give Early Warning on Natural Disasters
Queen's University Belfast
February 5, 2020
Trung Duong of Queen's University Belfast in the U.K. developed a drone system to warn of natural disasters and function as a Wi-Fi hotspot when telecommunications are disrupted by extreme weather. The drones support an inexpensive telecom network as they fly over large surface areas and provide real-time data on weather conditions. The Catastrophe-Tolerant Telecommunications Network (CTTN) can supply connectivity in a crisis situation if networks are destroyed or compromised. Said Duong, "The research could make a real difference to people living in areas exposed to extreme weather and it will certainly make the work of emergency services much easier." Vietnam’s Disaster Management Authority has already adopted the system.
|
Online Game Has Transnational Impact as 'Vaccine' Against Fake News
Uppsala University
Linda Koffmar
February 4, 2020
A study by researchers at Sweden's Uppsala University and the U.K.'s University of Cambridge found an online game can train players to better identify fake news and misinformation. The Bad News game casts players as online misinformation producers, with the goal of accumulating a maximum number of followers without losing too much credibility via tactics like conspiracy, discrediting, and trolling. The researchers worked with colleagues at the U.K.’s University of Cambridge and the Dutch media platform DROG to develop Bad News, and more than 5,000 players have tried the game in Swedish, Greek, Polish, or German. The players assessed fake and real Twitter posts before and after play sessions, and the game measurably improved their ability to spot fabricated content while upholding their trust in authentic news.
|
Where Computing Is Headed—Beyond Quantum
The Wall Street Journal
Sara Castellanos
February 5, 2020
Dozens of startups are using light, quantum physics, molecular biology, and new design methods to build computer chips and data-storage techniques for future computing. The advent of quantum computing has paved the way for other experimental techniques, startup executives say. The market for new computing technology comes as advances in conventional chip manufacturing are hitting a physical limit under Moore's Law. Some startups are creating chips focused on specific software tasks, while others are developing processing and storage solutions using new materials, including synthetic DNA. Said David Moehring of venture-capital firm Cambium Capital Partners, “If you don't come up with more exotic ways of computing, computing itself [won't] progress like it has for the last 50 years.”
*May Require Paid Registration
|
Researchers Create 3D-Printed, Sweating Robot Muscle
Cornell Chronicle (NY)
David Nutt
January 29, 2020
Cornell University researchers have engineered a soft robot muscle that perspires to regulate its internal temperature. The researchers produced nanopolymer materials for sweating using multi-material stereolithography, a three-dimensional printing technique that cures resin into predesigned configurations with light. They fabricated fingerlike actuators from hydrogel materials that retain water and respond to temperature, then incorporated the actuator fingers into a robot hand that could grab and lift objects. Said Cornell's Rob Shepherd, “I think that the future of making these more biologically analogous materials and robots is going to rely on the material composition. This brings up a point [about the importance of] multidisciplinary research in this area, where really no one group has all the answers."
|
U.S. Lets Autonomous Vehicle Bypass Safety Rules
Associated Press
Tom Krisher
February 6, 2020
The U.S. National Highway Traffic Safety Administration (NHTSA) has granted temporary approval to Nuro, a robotics company, to run low-speed autonomous delivery vehicles designed with no accommodations for human drivers. The vehicles lack mirrors, windshield wipers, steering wheels, and brake pedals. The approval marks the first time the federal agency has approved a company's request to deploy self-driving vehicles that do not meet the same safety standards for cars and trucks driven by humans. Under the temporary approval, Nuro must submit real-time safety reports to NHTSA, and will have to hold regular meetings with the agency and keep in contact with the community where the vehicles will travel. NHTSA will use its enforcement powers if it finds any evidence of an unreasonable risk to safety.
|
What if Half of Switzerland's Rooftops Produced Electricity?
EPFL (Switzerland)
Sandy Evangelista
February 3, 2020
Researchers at Ecole Polytechnique Fédérale de Lausanne (EPFL) in Switzerland are using machine learning algorithms, geographical information systems, and physical models to estimate the overall photovoltaic (PV) potential of Swiss rooftops to help the country transition to a low-carbon energy system. The algorithms account for parameters such as the size of each roof, its orientation, and whether the building is in a city center or a more isolated location. The researchers found solar PV panels could be fitted to 55% of Switzerland's total rooftop area, but even if panels were installed only on mainly south-facing rooftops, they still could provide more than 40% of the country’s electricity demand.
|
Tailor-Made Vaccines Could Almost Halve Rates of Serious Bacterial Disease
Imperial College London
Hayley Dunning
February 3, 2020
In collaboration with colleagues at Simon Fraser University (SFU) in Canada, researchers at the U.K.'s Imperial College London and Wellcome Sanger Institute integrated genomic data, bacterial evolution simulations, and predictive modeling to determine whether vaccines could be optimized for specific age groups, geographic regions, and bacterial communities. The researchers hypothesized tailoring vaccines to strains circulating in a particular area, or repurposing them for use with specific age cohorts, could reduce the occurrence of invasive diseases significantly. They used models of vaccine performance over time to evaluate the risk of vaccine-targeted strains being replaced by other potentially dangerous strains, helping to identify new vaccine designs that can lower overall disease rates. SFU's Caroline Colijn said, "Such an approach also enables public health policymakers to assess the likely effectiveness of an existing vaccine for a local population based on genomic surveillance data."
|
|
