Online Graduate Programs in Systems Engineering
Welcome to the December 2, 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."
China Issues Rules to Clamp Down on Deepfake Technologies
South China Morning Post (Hong Kong)
Meng Jing
November 30, 2019

China has issued new rules banning online video and audio providers from generating, distributing, and broadcasting fake news as it cracks down on deepfake technologies. This follows California's introduction of legislation to outlaw political deepfakes, and the European Union’s proposed strategy for probing such online disinformation. China's rules are broader, including prohibitions not only on using deep learning and related technologies to produce political news, but the use of virtual reality and deep learning to produce fake news as well. The rules mandate providers and users of online video news and audio information services, among other things, must use technology to determine whether audio and video news content has been manufactured or potentially doctored.

Full Article
Research Reveals Dirty Tricks of Online Shopping
University of Chicago Department of Computer Science
November 26, 2019

University of Chicago (UChicago) and Princeton University researchers found dark patterns, or deceptive practices employed by user interface designs, frequently used on shopping websites by popular online retailers. A Web-crawling tool analyzed over 50,000 product pages from 11,000 shopping sites, and captured text so the researchers could find dark patterns, and quantify how often they appeared. More than 1,800 cases of dark pattern usage turned up on 1,254 sites, which is likely a conservative estimate; a subset of 183 sites exhibited outright deception, including recurring low-stock or high-demand alerts and messages pressuring consumers to buy more expensive products. The researchers analyzed the underlying computer code to unearth third-party services that provide these options to shopping sites, facilitating the spread of dark patterns as plugins. UChicago's Marshini Chetty said, "We wanted to get a sense of: Is this a problem and should we be doing something about it?"

Full Article

Estrada named Emerging Woman Leader in Technical Computing award winner UNM's Estrada Honored at SC19
UNM Newsroom
November 26, 2019

University of New Mexico's Trilce Estrada was awarded the 2019 ACM Special Interest Group on High-Performance Computing (SIGHPC) Emerging Woman Leader in Technical Computing award at the International Conference for High-Performance Computing, Networking, Storage, and Analysis (SC19) in November. SIGHPC recognized Estrada "for her innovative and transformative deployment of machine learning for knowledge discovery in molecular dynamic simulations and in-situ analytics." Estrada's accomplishments include designing scalable machine learning methods applicable to data-intensive scientific challenges; using distributed systems, mobile devices, and crowdsourcing for distributed science and medicine learning; and introducing automatic decision-making processes to distributed computing environments. SIGHPC chair John West said Estrada's “technical contributions increase the ability of researchers to effectively conduct their work using large-scale computing, and her volunteer contributions in our community provide a model of leadership that includes future generations of leaders as well."

Full Article

Job seekers wait in line outside a career fair in Chicago hosted by Inc. Older IT Workers Left Out Despite Tech Talent Shortage
The Wall Street Journal
Angus Loten
November 25, 2019

Older information technology (IT) workers are having difficulty finding jobs despite the talent shortage in the U.S. IT job market. More than 80% of 2,800 employers surveyed by Robert Half International said recruiting tech talent is one of their biggest business challenges. Data compiled by trade association CompTIA found that workers ages 22 to 44 account for 61% of the IT sector, but only 49% of the workforce across all occupations. CompTIA also found workers age 45 and older represent 38% of all IT employees at U.S. companies, while the comparable figure for all occupations is 44%. “In such a candidate-tight market, hiring managers can't afford to overlook any potential talent pool,” said Robert Half’s John Reed.

Full Article
*May Require Paid Registration
Screening Technique Finds 142 Malicious Apps in Apple's App Store
IEEE Spectrum
Michelle Hampson
November 25, 2019

Researchers at Hanyang University in South Korea and Indiana University developed a method of detecting malicious apps that sneak through Apple's App Store screening process and onto some users’ phones. The researchers created a tool called CHAMELEON-HUNTER that analyzes the coding of apps. The team tested the tool against more than 28,000 apps in Apple’s App Store over a six-month period, and discovered 142 malicious apps. CHAMELEON-HUNTER uses two techniques for detection: one that utilizes an app’s multiple user interfaces to detect a Chameleon app; and the other analyzes semantics within the app's code, searching for unusual phrases. Said Indiana University’s Xueqiang Wang, “Our approach is only useful when the hidden user interfaces are already embedded in the app. However, there are other approaches adversaries can utilize to introduce hidden user interfaces.”

Full Article
Technology Makes Internet Memes Accessible for People With Visual Impairments
Carnegie Mellon News
Virginia Alvino Young
November 25, 2019

Carnegie Mellon University (CMU) researchers developed a system that automatically identifies memes and utilizes prewritten templates to add descriptive alternate text for visually-impaired persons. CMU's Cole Gleason said basic computer-vision methods can describe images underlying each meme, while optical character recognition techniques can decipher the overlaid text. For every meme type, only one image-description template is needed, and overlaid text can be added for each meme iteration. However, writing out the meme's intended message requires differing descriptions—explicit or more subtle—depending on the meme, if the humor translates. Said CMU’s Jeff Bigham, "Memes may not seem like the most important problem, but a vital part of accessibility is not choosing for people what deserves their attention. Many people use memes, and so they should be made accessible."

Full Article

A pedal bin turned into a robot Turn Any Object Into a Robot Using This Program and a 3D Printer
New Scientist
Donna Lu
November 26, 2019

Researchers at the University of California, Los Angeles created a program to render household items into hand-activated robots using a three-dimensional (3D) printer. The Robiot program can make object-robots perform simple tasks like turning on water taps in response to hand gestures, or shutting windows when temperatures fall. Users first shoot a video of themselves moving an object; Robiot analyzes the movement to generate a 3D model that can duplicate the motion, then isolates the object from its surroundings. Robiot can determine if the object is moving in a linear or rotational manner, and which parts of the object must remain stationary. The program matches the movements to a 3D model repository, selects the most similar model, then 3D-prints it with motion sensors that users can attach to the object.

Full Article

Brain scan templates of various ages Producing Better Guides for Medical-Image Analysis
MIT News
Rob Matheson
November 26, 2019

Researchers at the Massachusetts Institute of Technology (MIT) and Cornell University developed a technique to expedite creation and customization of templates used in medical-image analysis. The researchers' automated machine learning model produces conditional atlases based on patient characteristics such as age, gender, and disease. The model also leverages shared information from a complete dataset, to synthesize atlases from patient subpopulations for which the data may not account. The process combines a neural network that automatically learns an atlas at each iteration with another network that aligns that atlas to images within a dataset; the final product is a function that has learned how specific patient attributes correlate to structural variations across all images in a dataset. Said former MIT researcher Adrian Dalca, “The world needs more atlases. Atlases are central to many medical image analyses. This method can build a lot more of them and build conditional ones as well.”

Full Article
Stevens Team Completes First Census of Wikipedia Bots and Their Interactions With Humans
Stevens Institute of Technology
November 20, 2019

Stevens Institute of Technology researchers have analyzed all 1,601 software agents (bots) used by the collaboratively-written Wikipedia digital knowledge repository to edit content and persuade contributors to improve. The researchers employed algorithms to classify the bots and amass knowledge on how bot-human collaboration improves and expands the archive. The algorithms mapped each bot function as part of a network, examining areas where functions clustered, to classify bots as: fixers that repair broken content or delete vandalism; connectors that link pages to resources; protectors that rein in bad behavior, or advisers to suggest new activities and helpful tips. Bots fulfill nine primary roles, comprising about 10% of all Wikipedia activity and up to 88% of activity on sub-sections like the Wikidata platform.

Full Article
Mozilla Doubles Payouts as It Tries to Attract Software Vulnerability Hunters
Steve Ranger
November 20, 2019

Mozilla has increased payouts for its bug bounty program, while adding new websites and services to the list in order to draw more interest from bug hunters. The company said it has doubled all Web payouts for finding critical, core, and other bugs on Mozilla sites, and tripled payouts for identifying remote code execution exploits on critical sites. New sites and services Mozilla has added to the bug bounty program include Autograph, a cryptographic signature service that signs Mozilla products; the Lando automatic code-landing service; the Phabricator code manager for reviewing Firefox code changes, and the Taskcluster task execution framework.

Full Article
Wearable Sweat Sensor Detects Gout-Causing Compounds
Caltech News
November 25, 2019

California Institute of Technology (Caltech) scientists have created a sensor that analyzes the wearer's perspiration to determine metabolite and nutrient levels in their bloodstream. The sweat sensor has greater sensitivity than current devices and is easier to mass-produce, according to the researchers. Caltech’s Wei Gao suggested such a device could enable continuous monitoring of patients with cardiovascular disease, diabetes, or kidney disease, and make blood sampling with hypodermic needles unnecessary. The sensor is based on graphene, with microfluidics channels engraved in graphene sheets via laser. The prototype sensor measures uric acid, which is associated with gout when it reaches high levels; tests proved the sensor's accurate detection of higher uric acid levels in gout patients.

Full Article

Xue Lin, an assistant professor at Northeastern Can a Stealth T-Shirt Foil Digital Spies?
The Boston Globe
Hiawatha Bray
November 25, 2019

Researchers at Northeastern University, the Massachusetts Institute of Technology, and IBM have designed stealth apparel that renders the wearer invisible to spying computers. The adversarial T-shirt sports patterns that confuse artificial intelligence (AI) systems so they cannot identify the wearer as a human being. The shirt exploits AI vision systems' perception of pixels, the addition of which can fool the systems. Northeastern's Xue Lin and Kaidi Xu engineered the shirt to trick the Yolo object-recognition algorithm, which did not identify the clothing's wearers as people 63% of the time. Said Lin, “We try to explore the vulnerability of these neural networks, and hopefully, we can fix this problem.”

Full Article
*May Require Free Registration
December 2019 Issue of Communications of the ACM
Master's in Information Technology

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]