Association for Computing Machinery
Welcome to the November 7, 2016 edition of ACM TechNews, providing timely information for IT professionals three times a week.

Updated versions of the ACM TechNews mobile apps are available for Android phones and tablets (click here) and for iPhones (click here) and iPads (click here).


After Mastering Go, These Computers Are Learning to Play StarCraft
The Washington Post (11/05/16) Brian Fung

Researchers from Google's DeepMind lab in the U.K. are following up on the success of their Go-playing machine-learning algorithm earlier this year by training an algorithm to play the military strategy computer game "Starcraft II" in real time. Starcraft is far more complex than Go, requiring players to oversee a constantly fluid digital economy to realize victory. Starting next year, the game will function as a testbed for any artificial intelligence (AI) researcher who wants to use it. "StarCraft is an interesting testing environment for current AI research because it provides a useful bridge to the messiness of the real world," DeepMind says. "The skills required for an agent to progress through the environment and play StarCraft well could ultimately transfer to real-world tasks." The game's open-ended manner of play creates unresolved problems for AI, which has made it very attractive to scientists. A 2013 research paper by one international team suggested "optimizing assembly line operations in factories is akin to performing build-order optimizations" in strategy games. "Troop positioning in military conflicts involves the same spatial and tactical reasoning used in [real-time strategy] games. Robot navigation in unknown environments requires real-time path-finding and decision-making to avoid hitting obstacles."
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Internet-Based and Open Source: How E-Voting Works Around the Globe
Ars Technica (11/03/16) Cyrus Farivar

Although electronic voting in various forms has become more prevalent outside the U.S., its primary difficulties include its inability to verify source code and cost. Brazil's direct recording electronic system sought to enfranchise illiterate voters, and in its nearly 20-year reign it has prevailed thanks to an onscreen display of candidates and vote confirmation printouts, despite a lack of source code verification. Meanwhile, Australia's Australian Capital Territory adopted an open source e-voting software model combining Linux-based PCs, multilingual ballots, barcodes and scanners, and keylogging. Size and hardware expenses have been the chief obstacles for the model's wider expansion throughout the country. Internet-based voting is broadly implemented in Estonia, built atop a digital ID card infrastructure that universally replaces written signatures and is applicable anywhere. However, former ACM president Barbara Simons cites the uncertainty of this system's legitimacy because Estonia's government has never performed post-election auditing. The Spanish startup Scytl is drawing interest with a crypto-voting solution offering secure ballot transmission to individual voters, but its source code has not been released to the public for vetting. The lack of verification is the main reason why many experts continue to oppose e-voting.

Virtual Reality: Hybrid Virtual Environment 3D Comes to the Cinema
University of Montreal (11/03/16) Julie Gazaille

Researchers at Canada's University of Montreal (UdeM) recently conducted a study comparing the virtual reality (VR) experience of VR headsets and with the Hybrid Virtual Environment 3D (Hyve-3D), a three-dimensional (3D) immersive projection system using a concave-spherical screen. The researchers immersed 20 subjects of various ages in both types of virtual environments and noted their reactions and behavior. "Ultimately, the people much preferred the virtual reality without headsets, because they could interact with other viewers and share their impressions in real time," says UdeM professor Tomas Dorta. Viewers using VR headsets must continuously look around to explore the scene, which often hinders the storytelling and the cinema experience, but Hyve-3D viewers do not miss anything and have the same immersive feeling. In addition, Dorta says VR headsets restrict users to an individual experience, limiting non-verbal communication such as facial expressions, gestures, and postures. The researchers believe fans of horror or action films would be better served by a theater equipped with a system such as Hyve-3D. Dorta has worked on a prototype that can be seen at the Hybridlab design research laboratory, and he published his findings last month in the ACM Digital Library.

Why Light Bulbs May Be the Next Hacker Target
The New York Times (11/03/16) John Markoff

The Internet of Things (IoT) could prove highly vulnerable to cyberattackers, according to a new study from researchers at Canada's Dalhousie University and Israel's Weizmann Institute of Science. By focusing on the potential for hackers to hijack a smart Philips light bulb by exploiting a wireless flaw, the researchers say malware could spread across thousands or even hundreds of thousands of Internet-linked devices in close proximity, by infecting a single device. The team found the ZigBee wireless radio standard can be used to generate a malware-proliferating computer worm that targets IoT devices. They say the recent attack against the company Dyn demonstrated hackers have the means to commandeer a range of Internet-connected devices and use them to orchestrate similar attacks, steal information, transmit spam, or execute other malicious activities. "Even the best Internet defense technologies would not stop such an attack," warns cryptographer and study co-author Adi Shamir. The researchers say they used readily available and relatively inexpensive equipment to hack the Philips light bulb, which again demonstrates "how difficult it is to get security right even for a large company that uses standard cryptographic techniques to protect a major product."

Faster Programs, Easier Programming
MIT News (11/07/16) Larry Hardesty

Researchers at the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory and Stony Brook University in New York have developed Bellmania, a system enabling users to describe what they want their computer programs to do in generalized terms. It then automatically generates versions of those programs that are optimized to operate on multicore chips, while also ensuring they will return precisely the same results the single-core versions would, at a much faster rate of speed. The researchers utilized Bellmania to "parallelize" dynamic-programming algorithms so they would run on multicore processors, and the resulting programs were between three and 11 times faster than those produced by earlier automatic parallelization methods while generally as efficient as those manually parallelized by scientists. Bellmania applies the recursive divide-and-conquer parallelization strategy, an approach in which a problem is divided into smaller subproblems. At every recursion level, a Bellmania-generated program will usually conduct an operation on some segment of the matrix and outsource the rest to parallelized subroutines, which will repeat the process. Bellmania determines the amount of data to be processed at each level and which subroutines should manage the remainder. Bellmania was presented last week at the ACM Systems, Programming, Languages, and Applications: Software for Humanity (SPLASH 2016) conference in Amsterdam.

Make America Tweet Again
UNews (UT) (11/02/16) Vincent Horiuchi

Software developed by a team of researchers at the University of Utah can identify an individual's feelings toward current events by analyzing political tweets. The sentiment analysis software was fed more than 1.6 million geotagged tweets from June to October to determine whether states were leaning toward the Republicans or Democrats. Each tweet was analyzed and assigned a score from 0 to 1, with 0 being negative sentiment, 1 being positive, and 0.5 being neutral. The scores were then compiled in a database that calculates a state or county's political affiliation in real time. To measure the software's accuracy, researchers compared its results to the New York Times Upshot election forecast website and found the analyses to be very similar. Based on the number of positive tweets posted since June, the software predicts Hillary Clinton will win the presidential election. The researchers believe their software also could be used to determine the public's feelings toward certain companies by analyzing reviews about products and services. "With sentiment analysis, it will try to predict the emotions behind every human being when he or she is talking or writing something," says University of Utah doctoral student Debjyoti Paul.

Robots With Warm Skin Know What They're Touching
IEEE Spectrum (11/02/16) Evan Ackerman

Georgia Institute of Technology (Georgia Tech) researchers have developed a new kind of robotic skin that incorporates active heating. The system, when combined with traditional force sensing, results in a multimodal touch sensor that helps to identify the composition of objects. The researchers found the combination of force and active thermal sensing works significantly better than force sensing alone. The robotic skin is made of an array of "taxels," each of which consists of resistive fabric sandwiched between two layers of conductive fabric, two passive thermistors, and two active thermistors placed on top of a carbon fiber resistive heating strip. The researchers used all three of these sensing modalities to validate each other, and were able to identify wood and aluminum by touch up to 96 percent of the time while pressing on it, or 84 percent of the time with a sliding touch. "Knowing the haptic properties of the objects that a robot touches could help in devising intelligent manipulation strategies, [for example] a robot could push a soft object more than say a hard object," says Georgia Tech researcher Tapomayukh Bhattacharjee.

NSF Announces New Program to Support Institutes for Theoretical Foundations of Data Science
CCC Blog (11/01/17) Tracy Kimbrel; Gera Jochum

The U.S. National Science Foundation (NSF) in April hosted a workshop, Theoretical Foundations of Data Science (TFoDS): Algorithmic, Mathematical, and Statistical, which determined "theoretical foundations are necessary in all aspects of data science, from the generation and collection of data to the analysis and decision-making processes." Following up on the workshop, NSF's Directorate for Computer Science and Engineering and the Directorate for Mathematical and Physical Sciences recently issued a new program solicitation, Transdisciplinary Research in Principles of Data Science (TRIPODS). The program would support collaborative institutes in combining the theoretical computer science, statistics, and mathematics communities to influence the theoretical bases of data science. NSF says TRIPODS would do this "through integrated research and training activities. Phase I...will support the development of small collaborative institutes. Phase II...will support a smaller number of larger institutes, selected from the Phase I institutes via a second competitive proposal process." Proposals for the first phase may request a maximum of $500,000 a year for three years, while letters of intent are due Jan. 19, 2017, and proposals are due March 15, 2017.

Unstable Archaeological Dig Sites Could be Better Accessed by Robots
The Eyeopener (11/01/16) Noella Ovid

Researchers at Canada's Ryerson University are developing robotic systems that could help access unstable archaeological dig sites. Ryerson professor Alexander Ferworn's Human Robotic Interaction class designed six remote-operated devices with the goal of exploring underground tunnels in el-Hibeh, Egypt. The researchers tested the robots in a simulated underground cave, and they currently are working on developing prototypes that can be brought to el-Hibeh to field test this winter. "This is one of these issues of how do we...use the resources we have here...use the innovation, the technology, the interdisciplinary [elements] come up with a solution for a real-world problem that also helps disseminate knowledge, that helps highlight the crisis of cultural heritage that's going on not just in Egypt but all over the world," says Ryerson professor Jean Li. The project, known as the BUSA dig project, involves having each of the six prototype robots drive around and find artifacts inside the simulated underground cave without setting off booby traps that could damage the robots. "I provided very little guidance apart from 'here's the tunnel, here's the chamber,' it's going to be full of stuff and you have to find everything in it but then you have to get your robot out of it," Ferworn says.

Machine-Vision Algorithm Learns to Judge People by Their Faces
Technology Review (11/01/16)

University of Notre Dame researcher Mel McCurrie and colleagues tested whether a computer can study a human face and make the same first-impression judgments that people make by training a machine-learning algorithm. They used the website to get participants to rate 6,300 black-and-white facial images according to trustworthiness, dominance, IQ, and age. The team then used 6,000 images to educate the algorithm, and an additional 200 images to refine the machine-vision parameters. The researchers tested the algorithm with the last 100 images, and they say its ratings for trustworthiness, dominance, IQ, and age are comparable to a human's. By masking different parts of a face and asking the system to make its judgment, the team can determine which facial areas the algorithm focuses on. From these experiments McCurrie found the parts of the face the algorithm is most reliant on for its ratings also correlate for people. "These observations indicate that our models have learned to look in the same places that humans do, replicating the way we judge high-level attributes in each other," the researchers note. In addition, by applying the algorithm to each frame in a film, the researchers say they can measure how perceptual judgments shift over time.

NYU Is Building a Real-Life Holodeck Brooklyn (11/01/16) April Joyner

New York University (NYU) recently received a $2.9-million U.S. National Science Foundation grant to develop an immersive virtual reality experience in which participants can freely interact with their surroundings. NYU is partnering with 18 other universities, including the Massachusetts Institute of Technology and the University of California, Berkeley, as well as private companies, to develop the real-life version of the holodeck from "Star Trek." The researchers expect an initial prototype will be developed by the end of next year. The NYU Holodeck project is a combination of many technologies, including virtual and augmented reality, wearable devices, brain-computing interaction, three-dimensional printing, robots, advanced sound systems, and game controllers, with IBM's Watson underpinning the whole system. The NYU Holodeck's platform will be distributed among several institutions, which will enable the system to become "smarter" by having more simulations and scenarios to draw upon as it undergoes testing. In addition, universities and other organizations will be able to collaborate more efficiently by using the same shared resource. "We very much believe in this as a large-scale open innovation endeavor," says Winslow Burleson with the NYU-X Lab. "We hope it will pave the way for many other Holodeck nodes to be built."

Social Media Photos Priceless for Natural Resources Research
NCSU News (10/31/16) D'Lyn Ford

North Carolina State University (NCSU) researchers found geotagged photos can provide data for predictive models to help guide land use policy, conservation planning, and development decisions worldwide. "We found that Panoramio, Instagram, and Flickr provided comparable data that was reliable as an indicator of which landscapes visitors value most on a continental scale," says NCSU postdoctoral researcher Derek van Berkel. The researchers developed algorithms to filter data on Instagram, Panoramio, and Flickr to map the geographic distribution of images in Europe and ranked sites into four quartiles, from most- to least-visited locations. The researchers analyzed visitor patterns and found the most-valued landscapes included mountainous areas, locations near rivers and lakes, and areas near population centers. "Using social media to uncover and quantify people's interest in ecosystem services is an exciting new approach to understanding the important connection between natural resources and human health and well-being," says Ross Meentemeyer, director of NCSU's Center for Geospatial Analytics. The researchers plan to model what will happen in different scenarios by using a computer model driven by real data about real places.

Abstract News © Copyright 2016 INFORMATION, INC.
Powered by Information, Inc.

To submit feedback about ACM TechNews, contact: