Welcome to the January 5, 2018 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."
|
|
|
Researchers Discover Two Major Flaws in the World's Computers
The New York Times
Cade Metz; Nicole Perlroth
January 3, 2018
Computer security experts say they have discovered two major security flaws, called Meltdown and Spectre, within the microprocessors in virtually all of the world's computers. They warn these bugs could enable hackers to steal the entire memory contents of computers, including mobile devices, personal computers, and servers operating in cloud computer networks. To exploit the Intel-specific Meltdown flaw, hackers could rent space on a cloud service and then take sensitive information from other customers. The experts note this a serious flaw for cloud services that often share machines among many customers, letting malefactors bypass security measures and protocols designed to separate customers' data. They deem the Spectre flaw more difficult to exploit than Meltdown, but it affects most processors currently in use and there is no known patch. Cryptography Research president Paul Kocher says Spectre is rooted in the push to design faster chips, and a fix for it may not be available until new-generation chips are commercialized.
*May Require Free Registration
|
Sometimes, Computer Programs Seem Too Human for Their Own Good
The Economist
January 4, 2018
Researchers at Chungbuk National University in South Korea say they have demonstrated that increasingly human-like machines can invoke feelings of embarrassment in people, making some users hesitant to use assistive artificial intelligence. One experiment involved almost 200 volunteers who initially believed intelligence to be unchangeable, but who felt more embarrassed and incompetent after tests in which they were presented with 16 sets of three words and attempted to think of a fourth word that linked them, with half of the cohort given hints accompanied by an anthropomorphic computer-shaped icon. A second experiment permitted a different set of participants to ask for help rather than having it forced on them at random, which led to similar results. The researchers concluded some people appear to want to avoid losing face by seeking help from an anthropomorphic icon, suggesting there are situations in which the aggressive pseudo-humanization of machine-human interactions could usefully be reduced.
|
C Completes Comeback in Programming Popularity
InfoWorld
Paul Krill
January 5, 2018
The C programming language won the 2017 Programming Language of the Year designation from the monthly Tiobe Index, recognizing the once-declining language as the biggest gainer in the field. Although the language only grew 1.69 percentage points in its rating year-over-year in the January index, that was more than the next-closest rivals Python and Erlang. However, only five months ago, C had an all-time low rating of 6.477 percent, while this month its rating is at 11.07 percent, putting it in second place behind Java. Tiobe suggests C's resurgence could be attributed to its popularity in manufacturing and industry. The index bases the rankings on a formula assessing searches on languages in popular search engines such as Google, Bing, and Wikipedia. Other languages that saw increases in popularity in 2017 include R, which rose from 16th to eighth place, while Kotlin rose from 89th to 39th place and Erlang went from 44th to 23rd place.
|
A Dead-Simple Algorithm Reveals the True Toll of Voter ID Laws
Wired
Issie Lapowsky
January 4, 2018
Researchers at Tufts and Harvard universities have demonstrated it is possible to match individuals across government databases with nearly 100-percent accuracy, using a few basic identifiers such as a name, birth date, and address. The researchers developed an algorithm that could be used as the basis for a system courts can understand when considering cases concerning allegedly discriminatory voter ID laws. They found only one in 2.7 billion individuals have the same zip code, gender, birth date, and last name, making those combined details a very accurate indicator of identity. However, government records often contain typos, incomplete fields, and other mistakes, so the algorithm scanned Texas' voter rolls and compared it to the federal list of driver licenses, state IDs, and other forms of acceptable identification. The researchers found 98 percent of the records that could be matched using Social Security numbers also could be matched using any three of the key identifiers.
|
CACM Viewpoint Article on the Postdocs Best Practices Program
CCC Blog
Khari Douglas
January 3, 2018
A new Viewpoint article in the January issue of Communications of the ACM by the leads of the Computing Community Consortium's Postdoc Best Practices program focused on the approaches the participating universities/consortia found to be effective at enhancing post-graduate experiences and attempts to boost awareness of the postdoc community's needs in computer science (CS) departments. Issues the leads found necessary to address include the quality of postdoc training, the quality of mentoring, developing new skills, and the participation of postdocs in the community. Among the strategies cited in the article was establishing an Individual Development Plan with specific time-based goals, designed to effect thoughtful development and assessment of postdoc career objectives. The leads' programs also prioritized access to workshops and events for developing and obtaining pertinent skills, as well as dedicated institutional support staff whose goal was to ensure the postdocs gained a sense of community within the CS department.
|
Psychedelic Toasters Fool Image Recognition Tech
BBC News
January 3, 2018
Researchers at Google say they have created psychedelic stickers that can fool image-recognition software into seeing objects that do not exist. In an example, the team produced colorful computer-generated patterns by sampling hundreds of photographs of a toaster; when the patterns were put next to another item--a banana--many neural networks saw the toaster instead. The team says this method could be used to "attack" image-recognition systems, as these patches can be printed, added to a scene, photographed, and presented to image classifiers. Even if the patches are small, they cause the classifiers to ignore other items in the scene and report a chosen target class. The researchers note this works because the computer-generated pattern is more "salient" to image-recognition software than real objects. The team found the pattern consistently tricked software when it comprised at least 10 percent of a scene, while a photo of a real object was less likely to distract the software from another object.
|
Facebook, Phone Records Could Help Stop Disease Outbreak
Agence France-Presse
January 3, 2018
Researchers at the Technical University of Denmark (DTU) conducted a study tracking the digital and physical contacts of more than 500 students and found people who are central to their digital networks also are central in their real-life human networks. The researchers say this breakthrough could be used to identify the best individuals to vaccinate to stop a disease outbreak. "If you are a hub for your friends in the sense that you have many contacts via phone calls or on Facebook, making you a bridge between diverse communities, chances are high that you are also likely to be a bridge to connect those communities in case of an epidemic, such as influenza," says DTU's Enys Mones. The researchers used computer modeling to calculate that vaccinating central individuals would be almost as efficient as the most optimal existing vaccination strategies.
|
The Research Hardware in Your Video-Game System
Nature
Anna Nowogrodzki
January 3, 2018
Motion sensors and other technologies designed for video game systems are being repurposed for sophisticated research applications, with scientists exploiting their affordability, portability, and compatibility with free and easy to use software. One example is the use of Microsoft's Kinect motion sensor, originally designed for the Xbox system, connected to a laptop in order to rapidly scan a dinosaur skull in three dimensions by a person wearing the hardware strapped to their chest. The Kinect also has been used to model glacier beds and the meltwater channels under them at extreme resolution, and as a robot vision system to help machines learn tasks by monitoring people. The Massachusetts Institute of Technology's Anshuman Das, who proposed the portable Kinect scanner for the dinosaur skull project, says such a device may not match the fine resolution of industrial-grade digital scanners, "but since it's so cheap and it's easy to share data, it will encourage collaboration."
|
Combining Experiments, Models Boosts Social Behavior Research
NC State News
Matt Shipman
January 3, 2018
Researchers at North Carolina State (NCSU) and Northwestern universities have outlined a behavioral strategy that combines experimental studies with computer models to gain new insights into organizational and group behavior. The method begins with researchers designing and conducting experiments targeting a behavioral question, with the data from those experiments then fed to a model, enabling the researchers to predict how this behavior would emerge on a larger scale. The outcomes of the model can then be employed to inform future experiments, further validating the model or exposing more insights on the research question. Northwestern professor Ned Smith says the approach "capitalizes on the best aspects of both research techniques to advance our understanding of the behavior of large groups and advance the field." NSCU professor William Rand says this type of research can help answer questions that are pertinent to a broad diversity of issues, ranging from business management to public policy.
|
AI System Sorts News Articles by Whether or Not They Contain Actual Information
Motherboard
Michael Byrne
January 3, 2018
Researchers at the University of Pennsylvania and Google have developed a new machine-learning approach to classifying written journalism according to a formalized idea of "concept density." They say their system achieves about 80-percent accuracy in its ability to classify news stories across a wide spectrum of domains when compared to a ground truth dataset of already correctly classified news articles. The team took articles from an existing New York Times linguistic dataset consisting of original articles combined with metadata and short informative summaries written by researchers. "We have confirmed that the automatic annotation of data captures distinctions in informativeness as perceived by people," the researchers note. In addition, they also say their research demonstrated proof-of-concept experiments that show how the new method can be used to improve single-document summaries of news and the creation of summary pieces in news-browsing applications.
|
DARPA Launches Subterranean Challenge to Improve Underground Ops
Futurism
Chelsea Gohd
January 2, 2018
The U.S. Defense Advanced Research Projects Agency (DARPA) recently announced the Subterranean Challenge, which asks participants to develop systems that could help humans map, traverse, and search underground locations that are normally too difficult and dangerous to enter or explore. "The DARPA Subterranean Challenge aims to provide previously unimaginable situational awareness capabilities for operations underground," says DARPA's Timothy Chung. In addition, he notes the challenge aims to develop tools and methods of understanding what is below ground, enabling emergency responders to better help those in need. Each competing team will have the option of pursuing one of two separate tracks: a Systems Track, in which they would develop hardware-based solutions for an actual underground course, or a Virtual Track, in which the team would develop software for a virtual course. The winner of the Systems Track will receive a $2-million prize, and the Virtual Track winner will receive $750,000.
|
Physicists Take First Step Toward Cell-Sized Robots
Cornell Chronicle
Tom Fleischman
January 2, 2018
Researchers at Cornell University say they have developed a microscale robot exoskeleton that can quickly change its shape upon sensing chemical or thermal changes in the surrounding environment. The machines also are equipped with electronic, photonic, and chemical payloads, and they could become a platform for robots the size of biological microorganisms. "You could put the computational power of the spaceship Voyager onto an object the size of a cell," says Cornell's Itai Cohen. The tiny robots move using a motor called a bimorph comprised of graphene and glass, which bends when driven by a stimulus such as heat, a chemical reaction, or an applied voltage. The researchers say a biomorph is built using atomic layer deposition, and then wet-transferring a single atomic layer of graphene on top of the stack. Cornell's Paul McEuen says the next step is to attempt to develop small-scale muscles for the microscopic robotic exoskeletons.
|
Your Phone Will Know You Better Than Your Friends Do, U of T Researcher Predicts
U of T News
Chris Sorensen
January 4, 2017
In an interview, Richard Zemel at the University of Toronto in Canada discusses artificial intelligence (AI) developments he anticipates for the year ahead. Zemel predicts improved personalization technologies, to the point where digital assistants, and even smartphones, will be able to better understand questions, formulate answers, and become more familiar with users and their behavior. Related to this is the growing field of fairness in machine learning, which Zemel says involves building machine-learning systems to embody ethical and societal fairness principles. He also expects educational innovations such as online learning tools customized for individual students, and is especially interested in transfer learning, in which AIs may be able to learn new tasks without much training data. Another AI area Zemel sees as important is the idea of adding structure to machine-learning systems in the form of capsule networks, and he believes a key issue to resolve concerns the right structure to incorporate.
|
|
