Welcome to the March 13, 2017 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).
|
Beijing to Release National Artificial Intelligence Development Plan
South China Morning Post (Hong Kong)
Meng Jing
March 12, 2017
The Chinese government is drafting a national development plan on artificial intelligence (AI) and setting up a special fund as part of an effort to advance the technology's economic and national security applications. The plan, which aims to facilitate the adoption of the technology in a wide range of areas, is expected to be released soon after the National People's Conference and Chinese People's Political Consultative Conference, the annual meeting of China's parliament, according to Chinese minister of finance and technology Wan Gang. In addition, research and development and commercialization of AI should be considered long-term investments, according to CFA Institute director of content Larry Cao. China currently ranks second in the world in terms of venture capital funding for AI research at $1.1 billion, trailing the U.S. at $8.1 billion while funding in the rest of the world combined amounts to $3.1 billion.
|
SUTD, NTU join SMU in Deploying Supercomputer for AI Research
Networks Asia
March 13, 2017
Researchers at Singapore University of Technology and Design (SUTD) and Nanyang Technology University (NTU) in Singapore have deployed the NVIDIA DGX-1 (DGX-1) deep-learning supercomputer for their projects on artificial intelligence (AI). The SUTD researchers will use the DGX-1 to advance research in machine reasoning and distributed learning. In addition, the researchers will leverage the power of graphical-processing unit-accelerated neural networks for researching new theories and algorithms for AI. Meanwhile, the NTU researchers will use the DGX-1 system to increase the computing capability of the Rapid-Rich Object Search Lab, which conducts research in intelligent video analytics and visual computing, and also plans to use the additional computing power to launch larger-scale deep-learning projects. Separately, Singapore Management University is using the supercomputer at the Living Analytics Research Center to conduct a range of AI research projects for Singapore's Smart Nation initiative.
|
Cutting Down the Clutter in Online Conversations
MIT News
Rachel Gordon
March 9, 2017
Researchers at the Massachusetts Institute of Technology's (MIT) Computer Science and Artificial Intelligence Laboratory have developed "Wikum," a system that helps users construct concise, expandable summaries that make it easier to navigate disorganized conversations. The researchers tested Wikum on a Google document with tracked changes that aimed to mimic the collaborative editing structure of a wiki. The team found Wikum users completed reading much faster and recalled discussion points more accurately, and that editors made changes 40-percent faster. Wikum bridges the gap between forums and wikis by enabling users to work in short spurts to refine a discussion's main points. The researchers used an automatic highlighting algorithm to pick out important sentences for editors. They say the overall goal is to have a diverse group of contributors organize and moderate discussions. The team presented the research last month at the ACM Conference on Computer-Supported Cooperative Work and Social Computing (CSCW 2017) in Portland, OR.
|
Floods and Hurricanes Predicted With Social Media
University of Warwick
Luke Walton
March 10, 2017
Researchers at the University of Warwick in the U.K. have found social media can be used to warn people about floods, hurricanes, and other extreme weather events before they transpire by focusing on photos and keywords posted online. Warwick's Nataliya Tkachenko and colleagues tracked photos and videos tagged with words such as "river," "water," and "landscape" on the Flickr platform between 2004 and 2014. They found in certain time periods prior to the peak of extreme weather events, these words showed the weather getting worse in the locations where they happened. Tkachenko says this metadata can be used as "social sensors" to provide accurate warnings for weather events, in conjunction with physical sensors. "The opportunities represented by these new data sources are truly exciting as they can help to protect homes, save lives, and design more resilient cities!" she notes.
|
Girls Now Outnumber Boys in High School STEM, but Still Lag in College and Career
EdSource
Carolyn Jones
March 12, 2017
Although girls currently comprise about 50 percent of the enrollment in high school science and math classes, they still lag their male counterparts in college and the workplace in terms of participation in science, technology, engineering, and math (STEM), according to the National Girls Collaborative Project. California Girls in STEM Collaborative director Carol Tang says these figures indicate most girls are STEM-capable but choosing not to pursue STEM careers. Societal shifts, including more progressive mindsets about women's roles at home and at the workplace, have helped boost girls' STEM interest, but these trends have not crystallized across the board. The University of California, Berkeley's Lizzie Hager-Barnard says discouraging factors for girls include a perception of STEM fields as tedious and not impactful when it comes to helping people, and demeaning depictions of females in video games. Tang also notes girls are more likely to pursue STEM careers if they have a mentor.
|
Quantum Computer Learns to 'See' Trees
Science
Jane C. Hu
March 8, 2017
St. Mary's College researchers trained a quantum computer to recognize trees, a breakthrough they say could help scientists use other quantum systems for complicated machine-learning problems such as pattern recognition and computer vision. The researchers fed hundreds of U.S. National Aeronautics and Space Administration satellite images into a D-Wave 2X processor, which contains 1,152 qubits, to determine whether clumps of pixels were trees and not roads, buildings, or rivers. The researchers then told the computer whether its classifications were right or wrong so the system could learn from its mistakes, altering the formula it uses to determine whether something is a tree. Following the training, the system was 90-percent accurate in recognizing trees in aerial photographs of Mill Valley, CA. The results demonstrate how researchers can program quantum computers to analyze images, and could lead to the possibility of using them to solve other complex problems.
|
Toyota Tests Backseat-Driver Software That Could Take Control in Dangerous Moments
Technology Review
Tom Simonite
March 7, 2017
Toyota is testing Guardian, a new software system that judges whether a human driver is about to make a dangerous mistake. Radar and other sensors on the outside of the car monitor what is happening around the vehicle, while cameras inside the car track the driver's head movements and gaze. Guardian uses the sensor data to estimate when a driver needs help spotting or avoiding a hazardous situation. Toyota researchers are testing the ability of the software to understand the hazards around the car and whether the driver has spotted them, and they want to make Guardian capable of taking action if a driver does not look ready to respond in time. Unlike completely self-driving cars, Guardian won't rely on hyper-detailed maps and could be integrated into a conventional vehicle sold to consumers, according to Toyota Research's Ryan Eustice.
|
Researchers Create 'Time Crystals' Envisioned by Princeton Scientists
Princeton University
Catherine Zandonella
March 8, 2017
Researchers at Harvard University and the University of Maryland (UMD) have successfully generated time crystals based on the theoretical work of Princeton University scientists. Princeton professor Shivaji Sondhi says the team at Harvard discovered the basic physics of how time crystals function, which builds on breakthroughs at Princeton focusing on understanding complex systems that are in and out of equilibrium. The researchers believe their work could lead to concepts about how to safeguard information in quantum computers, which can be disrupted by interference by the outside world. The crystals are formed from atoms and molecules configured across space and time, with periodic atomic movement. Harvard's researchers realized non-equilibrium systems can be produced by periodically prodding a crystal by beaming a laser on its atoms. They created time crystals by generating an artificial lattice in a synthetic diamond, while the UMD team employed a chain of ytterbium ions.
|
Machine Learning Reveals Lack of Female Screen Time in Top Films
New Scientist
Matt Reynolds
March 8, 2017
University of Southern California (USC) researchers have developed machine-learning software that automatically detects how often men and women appear on screen in recent popular films, and found men have almost double the screen time compared to women. The software uses face- and voice-recognition algorithms that have been trained on annotated video to identify whether a character is male or female. The software also can measure how long they are on screen to a fraction of a second. The researchers analyzed the 100 highest-grossing live-action films from each of the past three years, and found women appear on average for just 36 percent of the total time that characters are on screen. In addition, Oscar-winning films are even less representative, with women receiving only 32 percent of screen time and 27 percent of speaking time in films that won an Academy Award.
|
Developing Tools for Research Reproducibility
University of Notre Dame
Brandi Klingerman
March 7, 2017
Researchers at the University of Notre Dame are developing several tools that can be used to save data, analytical methods, and processes so these elements can be shared among the research community. For example, the researchers are working on the Data and Software Preservation for Open Science project, a collective effort among several universities to explore the realization of a viable data, software, and computation preservation architecture for high-energy physics. In addition, the Whole Tale project includes researchers from across the U.S. working together to develop building blocks for a data-sharing infrastructure. The project aims to establish a repository that will enable scientists to publish their papers along with the data and methods used in their experiments. Separately, the Notre Dame researchers are creating and maintaining a repository that supports several research groups. Finally, the Circuit Realization at Faster Timescales project is pursuing the development of affordable low-energy circuits.
|
Opening Paths to Progress With Programmable Materials
ASU Now
Joe Kullman
March 6, 2017
Researchers at Arizona State University (ASU), the Georgia Institute of Technology, and the University of Paderborn in Germany are developing new computer science theories and distributed algorithms that will be used to explore and broaden the possibilities of using programmable matter. Distributed algorithms are designed to run concurrently on several independent, interconnected computational devices that cooperate via local interactions. The researchers developed a theory for a system of self-organizing particles that can perform "universal coating" and "compression." Universal coating involves layering an object with material as evenly as possible, and compression refers to gathering the material together as tightly as possible. The findings were based on University of Paderborn research to develop a framework to describe programmable matter systems, which they call the "amoebot model." The ASU researchers used this model to address the problem of coating surfaces of any shape with the universal coating algorithm for programmable material.
|
Tim Berners-Lee: I Invented the Web. Here Are Three Things We Need to Change to Save It
The Guardian
Tim Berners-Lee
March 11, 2017
World Wide Web pioneer Sir Tim Berners-Lee warns of three trends to be overcome in order to sustain the Web as beneficial for everyone. He writes of a loss of individuals' control over their personal data, encouraged by many websites' current business models. Berners-Lee also says mass online surveillance and other insidious practices as a result of this trend can stifle free speech. The second trend he notes is the proliferation of misinformation by bad actors for financial or political profit, while the third trend is a lack of transparency in political advertising online. Berners-Lee's solutions to these challenges include scientists cooperating with Web companies to put data ownership back into users' hands, a pushback against misinformation by encouraging continued opposition by Web gatekeepers, and "more algorithmic transparency to understand how important decisions that affect our lives are being made, and perhaps a set of common principles to be followed."
|
|
