Welcome to the October 23, 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."
|
|
|
Disjointed Covid-19 Apps Across U.S. Raise Questions About Tech's Role
The Wall Street Journal
David Uberti
October 21, 2020
U.S. states are relying on a patchwork of different technologies to perform Covid-19 contact tracing, in lieu of a coherent national plan. Ten states and Washington DC launched cellphone applications that alert users of exposure to infected people via a development framework built by Google and Apple, while 11 more states are piloting or building such tools. Technology and public health experts say the effectiveness of such apps across state lines is not ensured without a national strategy. Recent upgrades in the Bluetooth-based Google-Apple system have reduced states' opt-in costs and created a central server to store anonymized markers of positive tests for when phones cross state lines. Meanwhile, 27 states have not launched an app or are not developing one, with many citing worries about utility, as well as actual or perceived fears of privacy infringement.
*May Require Paid Registration
|
The Police Can Probably Break Into Your Phone
The New York Times
Jack Nicas
October 21, 2020
In an analysis of public records, Washington DC-based nonprofit Upturn found law enforcement officials regularly break into encrypted smartphones, and police in all U.S. states have phone-hacking tools. The New York Times confirmed authorities have used these tools in a growing range of cases; records suggested hundreds of thousands of phones have been searched over the past five years. Phone-hacking tools typically exploit security flaws to strip a phone's limit on passcode attempts, then enter passcodes until the phone unlocks; police often use tools from Atlanta-based Grayshift and Sun's Israeli unit Cellebrite to crack phones. The spread of these products has encouraged police to search phones even for minor offenses, and Upturn's Logan Koepke worries about the lack of oversight or transparency.
*May Require Paid Registration
|
Programming Language Popularity: JavaScript Leads—5 Million New Developers Since 2017
ZDNet
Liam Tung
October 21, 2020
SlashData's State of the Developer Nation Q3 2020 survey found JavaScript remains the reigning coding language for developers, having added 5 million developers since 2017 for a total of 12.4 million worldwide. SlashData ranked Python the second-most-popular language with 9 million developers, having gained 2.2 million over the past year due to machine learning, data science, and Internet of Things applications. Anaconda CEO Peter Wang said it was "kind of a miracle that Python is even on the radar, much less ranked in the top three languages," considering its low uptake for mobile and app development. Java achieved third place in the ranking with 8.2 million active developers (adding 1.6 million since mid-2017). Oracle says Java is still used by more than 69% of full-time developers globally, with 51 billion active Java Virtual Machines deployed.
|
U.S. Government Agencies to Use AI to Cull, Cut Outdated Regulations
Reuters
David Shepardson
October 16, 2020
The White House Office of Management and Budget (OMB) said federal agencies will use artificial intelligence (AI) to remove outdated, obsolete, and inconsistent requirements across government regulations. A 2019 pilot employing machine learning algorithms and natural-language processing at the U.S. Department of Health and Human Services turned up hundreds of technical errors and outdated mandates in agency rulebooks. The White House said agencies will utilize AI and other software "to comb through thousands and thousands of regulatory code pages to look for places where code can be updated, reconciled, and generally scrubbed of technical mistakes." According to OMB director Russell Vought, the initiative would help agencies "update a regulatory code marked by decades of neglect and lack of reform." Participating agencies include the departments of Transportation, Agriculture, Labor, and the Interior.
|
Probing Fine-Scale Connections in the Brain
Nature
Esther Landhuis
October 19, 2020
A global consortium of researchers aim to map the 70 million neurons in the mouse brain over the next decade. Said Harvard University's Jeff Lichtman, "We're dealing with a dataset that will be on the sale of an exabyte," or 1 billion gigabytes. Lichtman noted that mapping the mouse brain connectome will be "enormous compared to anything that's been done as a single project. Connectomes are just magnificently complicated." Researchers are beginning to map living neural networks and their connections at higher resolutions and scale, thanks to advances in microscopy and artificial intelligence, and crowdsourced help from human gamers. For instance, researchers at Princeton University, Baylor College of Medicine, and the Allen Institute for Brain Science are collaborating on the creation of a new online community, Pyr, to map the mouse brain.
|
Supercomputers Dig Into First Star Fossils
Texas Advanced Computing Center
Jorge Salazar
October 22, 2020
Researchers at the Georgia Institute of Technology (Georgia Tech) used supercomputers to model faint supernovae of metal-free first stars. Georgia Tech's John Wise said, "After you run the simulation, you can watch a short movie of it to see where the metals come from and how the first stars and their supernovae actually affect these fossils that live until the present day." While most of the simulations were run on the Georgia Tech PACE cluster, the researchers also obtained computer allocations on the U.S. National Science Foundation-funded Extreme Science and Engineering Discovery Environment (XSEDE) system. They also utilized Stampede2 at the Texas Advanced Computing Center (TACC), and the San Diego Supercomputer Center's Comet system, to run some of the main sequence radiative transfer simulations via XSEDE allocations.
|
Pufferfish-Inspired Robot Could Improve Drone Safety
CU Boulder Today
Daniel Strain
October 21, 2020
Engineers at the University of Colorado, Boulder (CU Boulder) have designed a pufferfish-inspired robot that could enhance the safety of flying drones. PufferBot is a hovering quadcopter drone with an expanding plastic shield that basically forms an airbag, in order to prevent collisions. The robot's airbag consists of plastic hoops fastened to its top, which can inflate from about 20 inches to 33 inches in diameter. When the drone approaches a potentially dangerous situation, PufferBot expands the hoops over its four spinning rotors, to keep them from striking people or objects. CU Boulder's Hooman Hedayati said the shield "can act as a temporary cage. It also communicates with users to tell them: 'Don't come close to me'."
|
3D Hand Pose Estimation Using Wrist-Worn Camera
Tokyo Institute of Technology (Japan)
October 21, 2020
A team of researchers at Japan's Tokyo Institute of Technology, Carnegie Mellon University, the U.K.'s University of St. Andrews, and Australia's University of New South Wales has developed a wrist-worn camera device that can estimate hand poses in three dimensions. The tool includes a camera supported by a neural network called DorsalNet, which accurately estimates three-dimensional (3D) hand poses by detecting changes to the position of the back of the hand. The researchers said their system outperformed earlier work with an average 20% higher accuracy in identifying dynamic gestures, and 75% accuracy in detecting 11 distinct grasp types.
|
A Radical Technique Lets AI Learn with Practically No Data
MIT Technology Review
Karen Hao
October 16, 2020
Scientists at Canada's University of Waterloo suggest artificial intelligence (AI) models should be capable of “less than one”-shot (LO-shot) learning, in which the system accurately recognizes more objects than those on which it was trained. They demonstrated this concept with the 60,000-image MNIST computer-vision training dataset, based on previous work by Massachusetts Institute of Technology researchers that distilled it into 10 images, engineered and optimized to contain an equivalent amount of data to the full set. The Waterloo team further compressed the dataset by generating images that combine multiple digits and feeding them into an AI model with hybrid, or soft, labels. Said Waterloo’s Ilia Sucholutsky, “The conclusion is depending on what kind of datasets you have, you can probably get massive efficiency gains.”
|
SocialBlock: Technology That Could Improve Data Security in Smart Cities
Universitat Oberta de Catalunya (Spain)
October 13, 2020
A researcher at the Universitat Oberta de Catalunya (UOC) in Spain has developed an architecture for creating user-centric data management applications that takes advantage of blockchain technology. The idea behind SocialBlock is to decentralize data storage and management, returning control over data to those who generate it. This idea could play a significant role as smart cities—which use the Internet of Things, sensors, drones, big data, and other technologies to improve quality of life—shift from small-scale projects to those that cover entire urban regions. Said UOC's Víctor García Font, "In smart cities, it's easier for the bodies that promote data management rules to force providers to follow certain rules on how to process the data they gather."
|
Duke Launches Center to Bring Computational Thinking to All Students
Duke Today
October 12, 2020
Duke University's new Center for Computational Thinking (CCT) plans to incorporate data literacy across the academic spectrum, to prepare students to consider the ethical, legal, and social ramifications of technology. CCT programs will connect learners to diverse accessible learning opportunities through multiple pathways, and also will combine offerings on ethics and policy of emerging technologies. Said Valerie Ashby, dean of Duke’s Trinity College of Arts & Sciences, “We want all Duke students to be computationally literate, regardless of major or minor. Computational thinking is about adopting a specific mindset and approach to problem solving. This is liberal arts for the 21st century.”
|
Tesla Putting 'Self-Driving' in the Hands of Drivers, Amid Criticism the Tech Is Not Ready
The Washington Post
Faiz Siddiqui
October 21, 2020
Electric automaker Tesla has selected a number of owners of its vehicles to have a software update download automatically into those vehicles to enable the cars to steer better and accelerate without human control. Critics are troubled by the absence of LiDAR sensors, a safety feature used by most self-driving car makers, from Tesla's system, which instead uses a suite of cameras and radar linked to an advanced neural network. Tesla CEO Elon Musk said the new software will better capture the exterior view of the vehicle and more seamlessly integrate collected footage to create a multidimensional perspective; safety experts disagree, warning the system cannot always perceive the true shape or depth of obstacles. The Partners for Automated Vehicle Education campaign said, "Public road testing is a serious responsibility and using untrained consumers to validate beta-level software on public roads is dangerous and inconsistent with existing guidance and industry norms."
*May Require Paid Registration
|
Finally: A Usable, Secure Password Policy Backed by Science
Carnegie Mellon University CyLab Security and Privacy Institute
Daniel Tkacik
October 20, 2020
The passwords research group in Carnegie Mellon University (CMU)'s CyLab Security and Privacy Institute has developed a passwords policy with hard scientific backing that balances security and usability. The researchers applied a state-of-the-art neural network-powered password-strength meter to evaluate combinations of minimum-length requirements, character-class requirements, minimum-strength requirements, and password blocklists. CMU's Nicolas Christin said, "We found that a policy requiring both a minimum strength and a minimum length of 12 characters achieved a good balance between security and usability." The researchers said minimum-strength policies can be flexibly designed for a desired security level, and are easier to implement with real-time requirements feedback in high-security environments.
|
|
