2019 Seton Hall 16 Months

Welcome to the June 7, 2019 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."

workplace and thumbs up, illustration Why Should We Care About Technology Ethics? The Updated ACM Code of Ethics
Ben Linders
June 5, 2019

The ACM Code of Ethics and Professional Conduct was updated last year in response to changes in the computing field over the past decades. Catherine Flick of the ACM Committee on Professional Ethics said ethical responsibility is especially critical to artificial intelligence (AI) developers, as uncertainty surrounding machine learning techniques and uses could sow public distrust, especially with their integration into infrastructure. Flick added that the Code of Ethics must be holistically applied to artificial intelligence and machine learning applications. “Put simply,” Flick said, “if you can’t predict what your system will do, and you can’t or won’t monitor it, you shouldn’t deploy it.”

Full Article

icons from keyboard typing, illustration Hackers May Soon Be Able to Tell What You're Typing—Just By Hearing You Type
The Wall Street Journal
Matthew Kassel
June 4, 2019

Researchers at the University of Cambridge in the U.K. and Linköping University in Sweden have found that the microphones built into Android devices can be used to capture the vibrations produced when users touch the virtual keyboard. Recordings of those sound waves can be interpreted to find out where on the screen the user tapped, and which keys were struck. The researchers used this method to recover numerical codes, letters, and whole words with some accuracy. For example, using a machine learning algorithm that classified each vibration, the researchers were able to identify seven out of 27 passwords on a smartphone, and 19 out of 27 passwords on a tablet.

Full Article
*May Require Paid Registration
Robot Baristas are Latest Front in South Korea Automation Push
Associated Press
Jung Yoon Kim
June 6, 2019

A wide range of South Korean businesses is increasingly incorporating robots and other automation technologies. For example, the Dal.komm Coffee franchise has 45 robot-outfitted retail robot cafes, in shopping malls, corporate cafeterias, schools, and an airport. A Dal.komm robot barista can take orders remotely via a mobile app or kiosk cashier, and is able to brew as many as 90 cups an hour or about 300 cups daily on one charge of beans and supplies. Concerns about job attrition have not dampened South Korean food and retail businesses' push to replace cashiers with automated kiosks, with advocates noting such innovations are particularly welcomed by millennial consumers. Suh Yong Gu, dean of the Sookmyung Women's University Business School, said, "This generation tends to not like meeting other people, so they favor...technology that enables people to minimize face-to-face interactions with others."

Full Article

Montana State University’s Lucia Williams, Buwani Manuweera, Indika Kahanda, and Brendan Mumey MSU Computer Scientists Help Expand Horizon of Genetics Research
Montana State University
Marshall Swearingen
June 3, 2019

Researchers at Montana State University and the non-profit National Center for Genome Resources have developed software that can compare multiple genomes and analyze the results. Previously available software struggled with analyzing pangenomes (the entire gene sets of all strains of a species) for relatively primitive organisms such as yeast, whose genome contains only 12 million DNA base pairs (compared to the 3 billion base pairs in the human genome). The researchers developed machine learning models that help the new software adjust its gene-sorting analysis according to input from users. That approach helped identify genes of interest in a yeast pangenome that includes about 100 strains.

Full Article

facial recognition demonstration in Fujian province Microsoft Quietly Deletes Largest Public Face-Recognition Dataset
Financial Times
June 6, 2019

Microsoft has deleted the MS Celeb dataset from the Internet, which had been used since 2016 to train facial recognition systems worldwide. The publicly available dataset contained 10 million facial images of nearly 100,000 individuals. Its removal came days after the Financial Times (FT) reported it was being used by military scientists and Chinese firms. Several of the Chinese companies supply equipment to officials in the Xinjiang region, where minorities of mostly Uighurs and other Muslims are being monitored and confined in internment camps. The FT investigator also found Microsoft's dataset featured images of private individuals, including security journalists and critics of the company. Also taken down following the FT report were Duke University's MTMC surveillance dataset, and Stanford University's Brainwash database.

Full Article
*May Require Paid Registration

Illustration of an interactive tool that lets users control ML systems work. Graphic by Chelsea Turner, MIT. Cracking Open the Black Box of Automated Machine Learning
MIT News
Rob Matheson
May 31, 2019

Researchers at the Massachusetts Institute of Technology, the Hong Kong University of Science and Technology (HKUST), and Zhejiang University in China have developed ATMSeer, an interactive tool that allows users to see and control how automated machine learning (AutoML) systems work. The tool was created to establish confidence in these systems and find ways to improve them. ATMSeer takes as input an AutoML system, a dataset, and some information about the user's task. It then visualizes the search process in a user-friendly interface, which presents in-depth information on the models' performance. During testing, the researchers found that about 85% of AutoML novices who used ATMSeer were confident in the models selected by the system; nearly all said using the tool made them comfortable enough to use AutoML systems in the future. Said HKUST researcher Qianwen Wang, "ATMSeer will mostly benefit machine learning practitioners, regardless of their domain, [who] have a certain level of expertise."

Full Article

Amazon’s Jeff Wilke demos the Prime Air delivery drone. Amazon Unveils Hybrid Drone for Deliveries
Alan Levin
June 5, 2019

Amazon has unveiled a drone that combines aspects of helicopters and airplanes, capable of vertical takeoffs and horizontal flight. To be used for test deliveries of household products in the next few months, Amazon Prime Air's drone is equipped with propellers, artificial intelligence, and a battery of sensors, which together allow it to fly autonomously without endangering conventional aircraft or people on the ground. The device is clad in a shroud to shield people from the propellers; the shroud also functions as a high-efficiency wing for horizontal flying. The drone is likely to challenge regulators, because standards do not yet exist for its robotic features.

Full Article
*May Require Free Registration
Stanford Engineers Make Editing Video Easy as Editing Text
Stanford News
Andrew Myers
June 5, 2019

Researchers at Stanford University, Germany's Max Planck Institute for Informatics, Princeton University, and Adobe Research have developed an algorithm to edit interview videos as easily as text editing. The algorithm extracts speech movement from video clips, then employs machine learning to render those clips as a video with seamless lip-sync. If an actor or performer misspeaks or flubs a word, the user can edit the transcript, and the app will stitch the correct word from words or word segments uttered elsewhere. The algorithm adds naturalism by applying intelligent smoothing to motion parameters, generating a three-dimensional animated version, while a machine learning-based Neural Rendering technique converts the low-fidelity digital simulation into a photorealistic video with precise lip-sync. The app's developers suggest concerns of misuse could be partly addressed with solutions like opt-in watermarking, to detect edited content and furnish an edit ledger.

Full Article

A lock sitting atop a keyboard. Only One in 20 Vulnerabilities are Exploited in the Wild, Claim Researchers
Dev Kundaliya
June 5, 2019

Only 5.5% of all vulnerabilities present in computing systems are exploited in the wild, according to a study by researchers at the Virginia Polytechnic Institute and State University (Virginia Tech), Cyentia, the RAND Corporation, and Kenna Security. The researchers analyzed 76,000 security flaws discovered between 2009 and 2018, and found that just 4,183 of them were exploited by hackers. In addition, the researchers said they did not find any correlation between published proof-of-concept exploit code on websites and the launch of attempted exploits. Of the more than 4,000 vulnerabilities that were exploited, exploit code was available on public websites for roughly half, suggesting attackers are willing to exploit some specific flaws and can also develop their own exploits, if necessary. The researchers wrote, "We believe this work has significant implications for decision makers when assessing cyber security risk, to include firms, federal agencies, and national security policy makers."

Full Article

Photo of a swift bird, the inspiration for the new logo for Apple’s SwiftUI tool. Apple Unveils SwiftUI for UI Development in Swift
Paul Krill
June 3, 2019

Apple has released a new set of tools and application programming interfaces (APIs) for building user interfaces (UIs) for any Apple device. The SwiftUI framework provides a way to build interfaces across all Apple platforms with one set of tools and APIs. The framework works with Xcode design tools to synchronize coding and designing, and incorporates support for capabilities such as dynamic type, localization, and accessibility. SwiftUI also includes a declarative syntax for stating what a UI should do; design tools from Xcode 11 that provide for drag-and-drop UI building; and the ability to allow developers to build one or many previews of SwiftUI views to get sample data and configure capabilities for users.

Full Article
Tempted to Cheat on a Written Exam? AI is 90% Certain to Nab You
University of Copenhagen
May 29, 2019

Researchers at the University of Copenhagen in Denmark have integrated big data with artificial intelligence to detect whether students wrote an assignment on their own, or had it ghostwritten or plagiarized it, with nearly 90% accuracy. The university’s Stephan Lorenzen said, "Our program identifies discrepancies in writing styles by comparing recently submitted writing against a student's previously submitted work. Among other variables, the program looks at word length, sentence structure, and how words are used." The Ghostwriter program, built around machine learning and neural networks, analyzed a dataset of 130,000 written assignments from 10,000 high school students. Said Lorenzen, “I think that it is realistic to expect that high schools will begin using it at some point. But before they do, there needs to be an ethical discussion of how the technology ought to be applied.”

Full Article
Black Hole Mystery Solved With Most Detailed Simulation Ever
Jackson Ryan
June 5, 2019

An international team of researchers used a supercomputer and custom-written code to generate the "most detailed" black hole simulation yet, to prove a long-held theory about the Bardeen-Petterson effect. The theory said the innermost region of a spinning black hole eventually would line up with its equatorial plane. The team utilized graphical processing units to develop the code for the simulation, then employed the U.S. National Science Foundation's 15-petabyte Blue Waters supercomputer at the University of Illinois at Urbana-Champaign to run it. The experiment demonstrated that scientists could simulate a more realistic black hole, accounting for magnetic fields within the accretion disk. Rebecca Nealon at the U.K.'s University of Leicester said, "The unique aspect of these simulations is their treatment of the magnetic fields, general relativistic effects, and a cooling function at the same time."

Full Article
ACM Career & Job Center
ACM Publications

Association for Computing Machinery

2 Penn Plaza, Suite 701
New York, NY 10121-0701

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]