Welcome to the January 7, 2015 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).
HEADLINES AT A GLANCE
Heads-Up Technology Puts Data on Car Windshields at CES
The Washington Post (01/06/15) Drew Harwell; Hayley Tsukayama
Automakers and third-party vendors at the Consumer Electronics Show (CES) in Las Vegas this week are demonstrating what many of them hope will be the hot new automotive accessory: windshield-based heads-up displays that project everything from car status information to social media updates. Several automakers already offer windshield heads-up displays in some of their newest models, but the displays being shown at CES are more advanced and offer more functionality. For example, Volvo is demonstrating a system that will flash bright symbols on a driver's windshield if they are about to collide with cyclists, while BMW is demonstrating a system that will let drivers change the radio and display speed warnings if they exceed preset limits. Hyundai is showing the system included in its 2015 Genesis that will project alerts when a crash is imminent. Meanwhile, Jaguar Land Rover is advertising what it calls a "360 Virtual Urban Windscreen," which will fill the windshield with information. Third-party vendors are showing smaller independent displays that can be mounted on a dashboard in any car and provide navigation and various smartphone-like capabilities. Automakers and vendors say such displays should cut down on distracted driving by preventing drivers from checking their phones, but there is little data backing up this assertion and some worry the displays will make distracted driving worse.
New Stanford Course Brings Silicon Valley to the Humanities Classroom
Stanford Report (01/06/15) Ian P. Beacock
Stanford University's new Literature and Social Online Learning course brings computer science and humanities students together in one classroom. The students work in teams to create websites and mobile apps that give readers new ways to experience literature. The interdisciplinary course is part of the university's new CS+X initiative, which enables students to study computer science in parallel with the humanities for a joint major degree. It was developed by Stanford comparative literature lecturer Petra Dierkes-Thrun and Stanford computer science professor (and Udacity CEO) Sebastian Thrun. They say learning to communicate across disciplines is the course's greatest challenge as well as its major payoff. "I want the students to learn to talk to each other and reach levels of achievement that individual disciplines alone never could achieve," says Dierkes-Thrun. The class also has helped graduate students rethink how they teach literature and given them new ideas about how to harness technology in the classroom. In addition, having students develop websites and apps can improve humanities education by encouraging students to learn in public, according to Dierkes-Thrun.
Apps Everywhere, but No Unifying Link
The New York Times (01/05/15) Conor Dougherty
The world of mobile apps today in many ways resembles the Internet of the late 1990s, and giant tech companies and numerous startups are trying to do for mobile apps what Google did for the Internet. The technology in question is called "deep linking" and refers to apps' ability to interact with and link to one another. As it was when the mobile world came into its own with the first iPhone, most mobile apps today exist as their own walled-off space, interacting only to a limited degree with the Internet and other apps. To create deep-linking technologies, startups such as Branch Metrics, URX, and Quixey, as well as tech giants including Google, Facebook, and Twitter, are developing schemes that would make apps more like websites. A standardized way of indexing apps would make it far easier for them to interact with one another and the Web. Other startups, such as Famo.us, are approaching the problem from a different direction: creating an easy way to turn websites into apps that would already have deep-linking capabilities built in. However, all of the competing solutions make the problem more complex. The goal is the creation of a single standardized approach to app deep-linking that all apps can adopt.
In STEM Courses, a Gender Gap in Online Class Discussions
The Chronicle of Higher Education (01/06/15) Rebecca Koenig
Piazza Technologies, maker of a digital class-participation tool, has published a study it says backs up the idea that a "confidence gap" exists between male and female students in computer science and science, technology, engineering, and math (STEM) courses more broadly. The study tracked 420,389 undergraduate and graduate students' use of Piazza's class discussion tool, which can be used by students to ask and answer questions, anonymously if they wish, over four semesters between spring 2012 and fall 2014. The study found female students in computer science classes asked more questions than male students, 2.2 questions on average compared to 1.75 questions for male students, and answered fewer questions, 0.7 on average compared to 1.2. The pattern was similar for STEM classes more generally, with women asking an average of 1.1 questions and men asking an average of 0.9, while women answered an average of 0.49 questions and men answered and average of 0.61. The computer science and STEM classes covered in the study were solidly male-dominated, with women making up slightly less than a third of enrollment in those classes. Piazza has made its own efforts at improving gender parity in STEM classes with its six-week online mentoring programming, Women in Technology Sharing Online.
The Emerging Science of Human-Data Interaction
Technology Review (01/05/15)
The University of Nottingham's Richard Mortier increasingly believes the balance of power in the growing data ecosystem is tilted in favor of those who collect and aggregate data, while those whose data is being collected are left with little recourse. Together with several colleagues, Mortier has published a manifesto that describes what they call a new science of human-data interaction and seeks to describe the three key themes associated with human-data interaction. The manifesto describes the distinctions between three kinds of data: data that is consciously created, like a Facebook profile; data that is observed, such as shopping behavior; and data that is inferred, such as preferences based on a friend's preferences. The three key themes described in the manifesto are data "legibility," "agency," and "negotiability." The first deals with how easily people can be made aware that data about them and associated analytics exist. The second deals with how easily people can control and interact with data relating to them, such as being able to opt in or out of data collection programs and to correct erroneous data. The final theme involves people's ability to change their data preferences in the future, in line with the recent "right to be forgotten" established in Europe.
Simple Pictures That State-of-the-Art AI Still Can't Recognize
Wired News (01/05/15) Kyle VanHemert
University of Wyoming researchers recently conducted a study to discover if state-of-the-art image-recognizing neural networks were susceptible to false positives. The researchers tested the networks by generating random imagery using evolutionary algorithms. The algorithms would produce an image and then change it slightly. Both the altered copy and the original were shown to a conventional neural network with a data set of 1.3 million images. If the copy was recognized as anything in the algorithm's cache with more certainty than the original, the researchers would keep it and repeat the process. Eventually, the technique produced dozens of images that were recognized by the neural network with more than 99-percent confidence. The researchers also found the artificial intelligence could routinely be fooled by images of pure static. Using a slightly different evolutionary technique, the researchers generated a different set of images that, to humans, all looked like static. However, the neural networks identified them, with upward of 99-percent certainty, as centipedes, cheetahs, and peacocks. The findings suggest neural networks develop a variety of visual cues that help them identify objects, according to University of Wyoming researcher Jeff Clune.
Robots Do Kitchen Duty With Cooking Video Dataset
Phys.Org (01/05/15) Nancy Owano
Researchers from the University of Maryland and National ICT Australia (NICTA) have built a self-learning robot that was able to improve its cooking skills by watching YouTube videos. The group utilized convolutional neural networks to identify the way a hand grasps an item and to recognize specific objects. The system predicts the action involving the object and the hand. The team made use of recent advances in deep neural networks. "The lower level of the system consists of two convolutional neural network-based recognition modules, one for classifying the hand grasp type and the other for object recognition," the researchers note. "The higher level is a probabilistic manipulation action grammar-based parsing module that aims at generating visual sentences for robot manipulation." To train their model, the researchers selected data from YouTube videos of people cooking and then generated commands that a robot could execute. The groups' experiments showed the system learned manipulative actions with high accuracy. The researchers will present their work at the 29th annual conference of the Association for the Advancement of Artificial Intelligence, which takes place Jan. 25-30 in Austin, TX.
Crowd Science Provides Major Boost for Certain Research Projects
Georgia Tech News Center (01/05/15) John Toon
Georgia Institute of Technology (Georgia Tech) researchers recently conducted a study examining common attributes among seven projects hosted on Zooniverse, the most popular crowd science platform. Crowd science projects involve thousands of ordinary people contributing to research projects in a wide range of fields. The study's findings regarding the contributions made by volunteer scientists describe the considerable value of donated time, as well as noting the limitations of nonprofessional research assistance. Assistance from crowd scientists is important because the volume of work involved would otherwise put projects beyond the reach of conventional research teams. "The broad idea is to get people involved who have an interest in science, even if it is a fairly shallow interest," says Georgia Tech professor Henry Sauermann. "Anybody can participate as long as they have a computer and can do the basic tasks required." The researchers examined the records of the seven projects over a 180-day period, following the activities of 100,386 participants who contributed 129,500 hours of unpaid labor, which equals a savings of more than $1.5 million. The researchers found that most volunteers spend relatively little time on the projects they support, with the majority of work done by a small fraction of the volunteers.
Privacy By Design: Protect User Data From 'Get-Go'
InformationWeek (01/05/15) Henry Kenyon
Ann Cavoukian, executive director of Ryerson University's Privacy and Big Data Institute and former Ontario Commissioner for Information and Privacy, is spearheading the Privacy By Design (PBD) project, an international effort to develop a framework for governments and companies to consider privacy protection from the very beginning of a software or mobile device development plan. PBD is being adopted as an international standard and has been translated into 37 languages, according to Cavoukian. She says PBD is vital for maintaining personal privacy in technological applications and notes the U.S. National Institute of Standards and Technology is working to apply privacy engineering and risk models for PBD standards. There are currently nine major applications areas for PBD processes: surveillance cameras in mass transit systems, biometrics used in casinos and gaming facilities, smart meters and smart grids, mobile communications, near-field communications, RFID and sensor technologies, redesigning ID geolocation, remote healthcare, and big data and analytics. Although there also are ongoing projects between the European Union and the U.S. to bridge the gaps in how both regions approach privacy, the U.S. government already is embracing some PBD concepts in its domestic operations.
3D Reconstruction Software for Antique Auto Parts and Pre-Hispanic Objects
Researchers from the University of Guadalajara in Mexico have developed software that could be used to reconstruct objects that are no longer manufactured, such as parts of classic automobiles. The software is designed to perform a three-dimensional (3D) scan of the original object, capture its binary patterns, and process the detected dimensions to produce a 3D model with real measurements. "The reconstruction technique involves the projection and acquisition of binary patterns [stripes of clear and dark lights, deployed vertically and horizontally] using a commercial projector and a digital camera," says Guadalajara professor Jorge Luis Nunez Flores. To acquire the 3D model, first a series of clear and dark lights projected laterally on the object to be scanned, then the 3D camera captures binary patterns or sequence lines to be processed by the software using the detected dimensions to form the model with real measurements. The Guadalajara researchers collaborated with the University of the Republic in Uruguay on the software, which could be used to scan and reconstruct archeological findings. Another potential application is facial recognition. The team also wants to make the models compatible with 3D printing technology to ease the manufacturing process.
NASA to Hack Mars Rover Opportunity to Fix 'Amnesia' Fault
BBC News (12/31/14)
The Mars rover Opportunity is having memory problems that are becoming more severe and causing it to reset itself, and sometimes stop communicating with mission control. The U.S. National Aeronautics and Space Administration (NASA) believes the rover's fits of "amnesia" are caused by a fault in its non-volatile memory, probably related to the hardware's age. "Which is the rover trying to use the flash memory, but it wasn't able to, so instead it uses the RAM...it stores telemetry data in that volatile memory, but when the rover goes to sleep and wakes up again, all [the data] is gone," says NASA project manager John Callas. NASA will try to fix the problem by hacking into Opportunity's software so the rover ignores the faulty part of its flash memory, and instead writes to the working hardware. The work will take a couple of weeks. Callas says the rover could be heading toward the end of its useful life. The initial goal was to spend three months on Mars, but the rover has been on the planet for more than a decade.
What You Tweet When You Go Party Can Be Useful for Improving Urban Planning
Plataforma SINC (Spain) (12/29/14)
Enrique and Vanessa Frias-Martinez, brother and sister computer science researchers at Telefonica Research and the University of Maryland, respectively, suggest using geolocalized tweets for urban planning and land use. "Geolocalized tweets can be a very useful source of information for planning, since it is an activity carried out by a large number of people who provide information on where they are at a specific time and what they are doing," Enrique says. He notes the increased use of smartphones and social networks have made it possible to access and produce information ubiquitously and these networks generate tags with the event's geolocation. For example, using Twitter "you can capture information on urban land use more efficiently and for a much larger number of people than with questionnaires," Enrique says. "Moreover, this type of consultation, traditionally used until now in planning activities, is very costly and can cause problems due to the lack of accuracy of the answers." The researchers developed a technique that automatically determines land uses in urban areas by grouping together geographical regions with similar patterns of Twitter activity. They studied land use in Manhattan, Madrid, and London, and identified areas as residential, business, daytime leisure, nightlife areas, or industrial land uses.
Tougher to Use Bitcoin for Crime?
BankInfoSecurity.com (12/30/14) Mathew J. Schwartz
University of Luxembourg researchers report identifying methods for determining the identity of anonymous Bitcoin users for between 11 and 60 percent of all Bitcoin transactions, "depending on how stealthy [the] attacker wants to be." They say they can deanonymize a Bitcoin user, or tie their pseudonym to the IP address from which they trade the cryptocurrency, with only about $2,000 in equipment. Moreover, the researchers claim they can thwart users who try to conceal themselves with firewalls or network address translation, and unmask as much as 60 percent of Bitcoin users who utilize the Tor anonymizing network by exploiting Bitcoin countermeasures for blocking distributed-denial-of-service attacks. University of Surrey professor Alan Woodward says the deanonymizing technique could give law enforcement agencies a tool for linking transactions to an IP address, and perhaps to a criminal's identity. Still, he cautions it is likely with the emergence of Bitcoin-foiling methods that malefactors will adopt new kinds of virtual currency such as Darkcoin, Dark Wallet, and the forthcoming Zerocoin. Woodward stresses the newest generation of cryptocurrencies "can have no other purpose than just to stay totally anonymous. You look at it and think, now why would anyone want that, except to conduct a criminal activity?"
Abstract News © Copyright 2015 INFORMATION, INC.
To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.