Welcome to the January 3, 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."
|
|
AI Early Diagnosis Could Save Heart and Cancer Patients
BBC News Pallab Ghosh January 2, 2018
Researchers at John Radcliffe Hospital in the U.K. have developed Ultromics, an artificial intelligence (AI) system they say can diagnose heart scans with much greater accuracy than conventional methods, picking up details in the scans that physicians may overlook. The system then delivers a recommendation of whether it believes there is a risk of the patient having a heart attack. Ultromics has been tested in clinical trials in six cardiology units, and the data indicates the system greatly outperformed human heart specialists, notes cardiologist Paul Leeson, who developed the AI-based system. In addition, the researchers say the trial results suggest the system could save the National Health Service a substantial amount of money. The researchers note they trained the system using scans of 1,000 patients whom Leeson had treated over the past seven years, along with information about whether they went on to develop heart problems.
|
Getting Physical With Collaborative Robots to Program for Adaptability
Engineering.com Kagan Pittman January 2, 2018
Researchers at Rice University have developed a method for training collaborative robots (cobots) using "gentle feedback" as they perform tasks, physically adjusting their trajectory in real time to simplify the education of robots that work alongside humans. Robots traditionally interpret physical human-robot interaction (pHRI) as a disturbance, resuming their pre-programmed behaviors after the interaction ends, but the Rice algorithm enables the robot to recalculate its path to its goal when interrupted or given new information. The researchers spent several months training a cobot to deliver a coffee cup across a desktop and used pHRI to help it avoid a computer keyboard, while remaining low enough that a drop would not break the cup. "By our replanning the robot's desired trajectory after each new observation, the robot was able to generate behavior that matches the human's preference," says Rice University's Dylan Losey.
|
It's Time to Weave Computational Thinking Into K-12
THE Journal Dian Schaffhauser January 2, 2018
A new study from the nonprofit Digital Promise stresses the need for computational thinking to be integrated across K-12 school curriculums, citing an ACM description of computational thinking as "the thought process behind programming." The report calls K-12 schools' current efforts to teach this principle "highly decentralized" and lacking a "single authority at the national level guiding the purpose and implementation of opportunities to learn about computing." The study's authors say the incorporation of computational thinking into education would enable students to solve complex problems with many interrelated parts, develop processes that are "logical, precise, and repeatable," and comprehend how to work with data to support their problem-solving. Digital Promise notes computational thinking can be integrated into existing topics, and its recommendations for moving computational thinking forward in schools include the undertaking of advocacy campaigns, curriculum and resource development, professional development for educators and school leaders, and further research.
|
How an AI 'Cat-and-Mouse Game' Generates Believable Fake Photos
The New York Times Cade Metz; Keith Collins January 2, 2018
Researchers at NVIDIA say they have developed an artificial intelligence (AI) system that analyzes thousands of celebrity photos, infers common patterns, and generates new images that are closely similar. They note the underlying idea of the project is to expedite and enhance the creation of computer interfaces, games, and other media, eventually enabling software to rapidly generate realistic imagery. The engineers set up two generative adversarial networks--one to produce the images and another to ascertain the images' authenticity. "The computer learns to generate these images by playing a cat-and-mouse game against itself," says NVIDIA's Jaakko Lehtinen. However, Tim Hwang with the Ethics and Governance of Artificial Intelligence Fund sees a negative aspect to this form of technological progress. "The concern is that these techniques will rise to the point where it becomes very difficult to discern truth from falsity," he warns. "You might believe that accelerates problems we already have."
*May Require Free Registration
|
Information Extraction and Visualization From Twitter Considering Spatial Structure
The University of Electro-Communications (Japan) December 28, 2017
Researchers at the University of Electro-Communications, Tokyo in Japan have proposed a framework that considers relationships between data meaning and their spatial structures. The team focused on the distinction between locations of interest (LoI) and locations of activity (LoA), and an evaluation of 60,000 tweets produced good results about the precision and recall of the classification. The researchers note their new method also was successfully applied to extract frequently mentioned locations while classifying them into those which were globally mentioned and those locally mentioned. The results indicate this method could be applied to analyzing relationships between location names and the signified locations. For example, location coordinates attached to the tweet, "Heavy rain in Miura Peninsula" by NHK are not those of the Miura Peninsula, but of Shibuya, where NHK is located. Therefore, the tweet would be flagged by a spatial search of the peninsula but not a location search, because Miura Peninsula is the LoI and Shibuya is the LoA.
|
Personalized Preference-Based Service Selection and Recommendation
CCC Blog Helen Wright January 2, 2018
University of Massachusetts Boston professor Kenneth K. Fletcher has conceived of a method for dealing with data sparsity and cold-start limitations in service recommendation using personalized preferences. His proposal concerns improving on techniques and algorithms that incorporate user preferences for service selection and recommendation. Fletcher notes a gap exists between users' non-functional attribute values and their satisfaction with service selection or recommendations due to personalization and clashing attributes issues. He says his team has developed models that enhance selection and recommendation accuracy overall. "We have developed an intelligent survey curation...application [that] improves respondent engagement and optimizes automatic questionnaire curation," Fletcher notes. "Finally, with fake users in social media on the rise, we look to use our personalized preference-based methods. My team recently began this work and has derived some exciting results that we are going to share with the research community soon."
|
'Quantum Radio' May Aid Communications and Mapping Indoors, Underground, and Underwater
NIST News Ben P. Stein January 2, 2018
Researchers at the U.S. National Institute of Standards and Technology (NIST) say they have demonstrated that quantum physics could enable communications and mapping in locations where global-positioning systems and standard cellphones and radios do not work reliably. The team notes the new technology involves low-frequency magnetic radio, very low frequency (VLF) digitally modulated magnetic signals that travel farther through building materials, water, and soil than conventional electromagnetic communications signals at higher frequencies. VLF electromagnetic fields already are used underwater in submarine communications, but does not provide enough data-carry capacity for audio or video files. NIST's Dave Howe says the increased magnetic field sensitivity via quantum sensors theoretically leads to longer communications range and higher bandwidth communications. In an effort to reach that goal, the NIST researchers have demonstrated the ability to detect digitally modulated magnetic signals by a magnetic-field sensor that relies on the quantum properties of rubidium atoms.
|
Single Metalens Focuses All Colors of the Rainbow in One Point
Harvard University Leah Burrows January 1, 2018
Researchers at the Harvard University School of Engineering and Applied Sciences say they have developed the first single lens (a metalens) that can focus the entire visible spectrum of light in the same spot and in high resolution, which could lead to the development of advanced optical devices that can be used in virtual and augmented reality. Harvard professor Federico Capasso says metalenses are thin, easy to fabricate, and cost-effective, and their research extends those attributes across the whole visible range of light. The new metalenses use arrays of titanium dioxide nanofins to equally focus wavelengths of light and eliminate chromatic aberration. The researchers created units of paired nanofins that control the speed of different wavelengths of light simultaneously, controlling the refractive index on the metasurface. The nanofins also are tuned to produce different time delays for the light passing through different fins, which ensures all wavelengths reach the focal point at the same time.
|
Berkeley Scientists Are Building a Quantum Computer
California Magazine Kali Persall December 27, 2017
Researchers at the University of California, Berkeley are using a $3-million U.S. Department of Energy grant to construct a quantum computer and develop its operating software. The hardware team will build the computer and study its properties and applications over five years, while the software team will design algorithms, or possibly even new kinds of math, over the next three years. The hardware team, led by Berkeley researcher Irfan Saddiqi, already has produced an eight-quantum-bit (qubit) system, with a goal of 64 qubits. Saddiqi says if the team can achieve that goal, it would mark the first time U.S. researchers have passed the quantum-classical boundary, creating a quantum processor that surpasses any digital supercomputer. Although quantum computers are unlikely to reach mainstream consumers anytime soon, Lawrence Berkeley National Laboratory researcher Bert de Jong predicts in the next five to 10 years quantum systems could become commonplace in research facilities and national laboratories.
|
This Autonomous Quadrotor Swarm Doesn't Need GPS
IEEE Spectrum Evan Ackerman December 27, 2017
Researchers at the University of Pennsylvania (UPenn) have trained a swarm of 12 quadrotor drones to fly in close formation without motion capture or a global-positioning system. Each drone operates using only one or two cores of processing power, an onboard inertial measurement unit, and a downward-looking video graphics array camera with a 160-degree field of view. Each drone employs visual inertial odometry to calculate how far and in what direction it is traveling from its starting point, which produces a good approximation of its relative position; they send updates to a ground station that transmits back commands to change formation. Each drone estimates its own trajectory, and the swarm performs well under low-light conditions. UPenn researcher Giuseppe Loianno cites the use of a scalable and extensible architecture for controlling multiple vision-based quadrotors as a step up from previous experiments, with plans to use a localization algorithm that identifies each vehicle relative to its neighbors.
|
Wisdom of the Crowd Accurately Predicts Supreme Court Decisions
Technology Review December 26, 2017
Researchers at the Chicago Kent College of Law say they have analyzed how well crowdsourcing can predict decisions made by the U.S. Supreme Court by crunching data from the online FantasySCOTUS league. Using a hypothetical contest to predict high court decisions from 2011 through 2017 using crowdsourcing data as a rationale, the researchers generated about 250,000 models to test on data produced by FantasySCOTUS. The team notes it first analyzed the models' aggregated performance and then compared it to a "null model," or a rule of thumb attorneys use to guess the outcome of Supreme Court decisions, which was found to be 60-percent accurate. The researchers say many crowdsourced models consistently outperformed the null model, with the best models predicting decisions with 80-percent accuracy. They now are planning to apply this same technique to predict the outcome of other legislative processes, as well as elections.
*May Require Free Registration
|
CHS Student Wins Congressional App Challenge for California's 52nd District
Coronado Times Andres de la Lama December 23, 2017
Thousands of U.S. students over the last four months have programmed original applications as part of the district-wide competitions that made up the 2017 Congressional App Challenge, which aimed to engage students in coding and computer science. The competition included 190 Congressional districts across 42 states, and saw 4,100 students participate in the 14-week regional competition, submitting more than 1,270 original apps. The winning app was developed by Padraig MacGabann of California's 52nd Congressional District in Coronado, CA. His winning app makes the text-to-911 system more efficient, and provides proofs-of-concept for automated emergency dispatch and location tracking during Internet failures. "Time is precious in a disaster, so automating parts of the process would save many lives," MacGabann says. The Congressional App Challenge will invite MacGabann and the other winners to showcase their apps to members of Congress and the technology community at #HouseOfCode, a reception on Capitol Hill to be held in April.
|
Does Your Phone Know If You're Sick?
Deutsche Welle (Germany) January 3, 2018
In an interview, Notre Dame University professor Kevin W. Bowyer suggests within the next two years there will be significant progress made incorporating facial-recognition software into smartphones to assess a person's physical and mental wellness. However, he also warns this has the potential to raise controversial ethical and privacy issues. "There are groups...monitoring your psychological well-being as well as your physical well-being from the time sequence of face images," Bowyer says. "You might also get the person's pulse rate, or their respiration rate, from a face video." Although Bowyer sees such data being of use to health authorities, the question remains as to whether people will be willing to have this data disclosed to such entities. "We're potentially crossing a line when you begin to report semi-anonymous health information to another party," he notes.
|
|