Association for Computing Machinery
Welcome to the February 20, 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


Watch Out, Coders--a Robot May Take Your Job, Too
InfoWorld (02/19/15) Bill Snyder

A new paper from researchers at Boston and Columbia universities suggests the current boom in software and technology that is driving up wages for developers could be laying the foundations for its own collapse and a perpetually stagnant economy. The researchers suggest a world in which the economy enters a new boom-and-bust cycle that accelerates the production of new products and code at such a rapid pace that supply inevitably outstrips demand. "The long run in such a case is no techno-utopia," the researchers write. The cycle begins with huge demand for code and dramatically increasing wages for software developers, a scenario that already may be playing out. However, the researchers suggest sooner or later, some of the code produced during this boom will be directed at making smart machines capable of learning new tasks and improving themselves on their own. Over time, if such code spreads to different sectors of the economy, it whittles away the need for software developers to design new and better systems, as the systems themselves will do that. Eventually, demand for new code stagnates and wages drop across the economy, leading to a broad-based economic stagnation in which few people have the means to afford the amazing products produced by the previous boom or the capital to invest in the next one.


Microsoft, Google Beat Humans at Image Recognition
EE Times (02/18/15) R. Colin Johnson

In the run-up to this year's ImageNet Large Scale Visual Recognition Challenge, which tests the ability of computer systems to recognize and properly identify images, Microsoft announced that for the first time its system has beaten humans at the test. The human benchmark of the ImageNet challenge is an error rate of 5.1 percent, and Microsoft Research Asia says its Visual Computing Group has beaten that with a 4.94-percent error-grabbing neural network. The team that accomplished this feat was led by Jian Sun and included researchers from Jiaotong University and the University of Science and Technology of China. Sun says the group used deep convolutional neural networks with 30 weight layers to achieve its goal. He says the key to their success was their Deep-Learning algorithm, and in particular its initialization method. "In our work, we derive a theoretically sound initialization method, which allows us to freely exploit more powerful--deeper and wider--neural networks," Sun says. Microsoft already has applied the team's results to the Bing image search and OneDrive services. However, their triumph was short-lived; only five days after announcing their results, Google said its own computer-vision system had beaten Microsoft's by 0.04 percent.


Can an LED-Filled "Robot Garden" Make Coding More Accessible?
MIT News (02/18/15) Adam Conner-Simons

Massachusetts Institute of Technology (MIT) researchers have developed a tablet-operated system that illustrates distributed algorithms via robotic sheep, origami flowers that can open and change colors, and robotic ducks that fold into shape by being heated in an oven. The researchers say the system serves as a visual embodiment of their latest work in distributed computing, as well as an aesthetically pleasing way to get more young students, especially girls, interested in programming. "It's meant to be a launchpad for schools to demonstrate basic concepts about algorithms and programming," says MIT researcher Lindsay Sanneman. The system is equipped with 16 tiles that are connected via microcontrollers and programmed with search algorithms that explore the space in different ways. "The garden tests distributed algorithms for over 100 distinct robots, which gives us a very large-scale platform for experimentation," says MIT professor Daniela Rus. The researchers also developed eight distinct varieties of origami flowers that are embedded with printable motors, enabling them to blossom in multiple ways. The robotic sheep were developed with traditional print-and-fold origami techniques, and the magnet-powered ducks started as two-dimensional paper prints that were heated in an oven, causing them to automatically fold into shape. "We're hoping that rapid fabrication techniques will continue to improve to the point that something like this could be easily built in a standard classroom," says MIT researcher Joseph DelPreto.


Privacy by Design Workshop: Concepts and Connections
CCC Blog (02/17/15) Helen Wright

The Computing Community Consortium is sponsoring a series of four workshops over the course of the year to identify a shared research vision for supporting different facets of the practice of Privacy by Design. In the first workshop held earlier this month, a group of more than 40 collaborators from industry, academia, government, and civil society analyzed the applicability of existing privacy frameworks such as the Fair Information Practice Principles, taxonomies of privacy harms and justifications, and new concepts of privacy. The group focused on the "essentially contested" concept of privacy and how different concepts or analytical tools can help identify and address privacy concerns. The workshop also examined reports from the field on those who have implemented privacy programs in the real world. In addition, participants discussed how to engage with the complexity of conceptualizing privacy, and how to utilize expertise from other relevant perspectives. The participants identified a desire to bridge the research work creating new privacy tools and the adoption of those tools in the practice of Privacy by Design. The second workshop, to be held in May, will focus on privacy from the perspective of design, and the third workshop will bring together software engineers at Carnegie Mellon University to discuss their development practices. The fourth workshop will offer a discussion platform for regulators and policymakers on catalyzing Privacy by Design.


Google to Roll Out After-School Coding Classes for 100K New York City School Kids
New York Daily News (02/17/15) Ben Chapman; Lisa L. Colangelo

New York City has partnered with Google to give more than 100,000 students access to Google's CS First program that teaches children how to code, as part of the city's continuing push to get children involved in the science, technology, engineering, and mathematics (STEM) fields. "Connecting youth to opportunities that will expose them to hands-on learning and increase their awareness of careers in the technology field is crucial," says New York City Department of Youth and Community Development's Meryl Jones. "Initiatives such as this cultivate curiosity and encourage our youth to inquire, create, and explore." CS First enables students to create their own stories, games, and animations while learning about computer science. CS First will be offered to children enrolled in 857 after-school programs across the five boroughs of New York City starting in September, according to city officials. "At Google, we aim to inspire young people around the world not to just use technology but to create it," says Google New York's William Floyd. The partnership between Google and the city is part of a larger Tech Talent Pipeline initiative led by Mayor Bill de Blasio to boost technology education, training, and job opportunities.


Patterns in Large Data Show How Information Travels
Umea University (Sweden) (02/16/15) Ingrid Söderbergh

Umea University researchers have found an analysis of how people edit content on Wikipedia can reveal what information matters to them and with whom they have most in common. The study's results show people care most about local and regional information related to sports, media, celebrities, or local pages. In addition, people from countries with similar languages or historic backgrounds care about similar information, according to the researchers. "We can project these similarities in a network where countries are nodes and links represent the strengths of similarities," says Umea network scientist Fariba Karimi. She says it is possible to analyze how information, diseases, or financial crises spread over networks by extracting patterns from raw datasets of social and economical interactions. "These social interactions happen in time and, depending on the time, the influence might differ," Karimi says. The researchers found that accounting for time makes the spreading dynamic different compared to static networks. "This is important, because it can help us to better understand the spreading processes in real social systems that are mostly dynamic and change over time," Karimi says.


Out of High School Men Liked Math and Science More Than Women
THE Journal (02/17/15) Dian Schaffhauser

Although 48 percent of male high school graduates agreed or strongly agreed that science was one of their favorite subjects, only 34 percent of female high school graduates said the same, according to an American Institutes of Research and Activate Research study overseen by the National Center for Educational Statistics. The findings were based on data from the U.S. National Assessment of Educational Progress (NAEP), an evaluation of what students know and can do in various subject areas. The most dramatic differences between men and women showed up in advanced biology and computer science, both of which show a gap of at least 10 percent in the number of young men or women who graduate from high school with credits in those courses. In addition, among 2009 high school graduates who had earned credits in specific science, technology, engineering, and mathematics courses, males generally had higher NAEP scale scores than females. For example, in algebra II and calculus courses, the men's average was five and seven points higher, respectively, on the NAEP scale of 300, than the women's average. The data for the study came from a 2009 NAEP high school transcript study, as well as the 2009 NAEP grade 12 math and science student questionnaires.


How a Wedding Engagement Changes Twitter Feeds
Georgia Tech News Center (02/13/15)

A Georgia Institute of Technology (Georgia Tech) researcher has studied engagement on social media, focusing on the phases people experience as they transition from singles to couples in society. Munmun de Choudhury, an assistant professor in Georgia Tech's School of Interactive Computing, followed nearly 1,000 people on Twitter who used "#engaged" to announce their engagements in 2011, examining each person's tweets in the nine months before the engagement and 12 months afterward. She reports tweets with the word "I" or "me" dropped by 69 percent after they got engaged, and were replaced with "we" or "us." The use of familial words such as "future-in-laws" and "children" rose 219 percent after the proposal. Choudhury also notes women tweeted about their significant other by using words tied to emotion, such as they "love" their "wonderful" fiance, while men were more likely to use physical descriptors such as "sexy," "beautiful," and "gorgeous" fiancee. Moreover, the study found tweets using future-tense verbs, which shows a focus on the future, rose 62 percent after engagement. "Twitter can be a powerful tool that can mirror our thoughts and how we're actually feeling," de Choudhury says. "This isn't based on what they told us they did. It's a reliable record--it's what they actually did."


HP to Award Big Money to Winners of Pwn2Own Browser Hacking Challenge
eWeek (02/16/15) Sean Michael Kerner

Hewlett-Packard's (HP) Zero Day Initiative (ZDI) has announced the prize structure for its 2015 Pwn2Own browser hacking challenge. HP will pay security researchers a $75,000 award if they are able to exploit Google's Chrome browser on Windows. There will be a $65,000 prize for a Microsoft Internet Explorer (IE) 11 exploit, and a $30,000 prize for a Mozilla Firefox exploit. Researchers who can exploit Adobe Reader or Adobe Flash running in IE will receive $60,000 awards. HP also is awarding $50,000 for an Apple Safari exploit on Mac OS X. "Our program is seeing an increased number of quality use-after-free vulnerabilities and sandbox escapes in the Pwn2Own targets," says HP Security Research's Brian Gorenc. He notes among the rule changes this year, all attacks must work when Microsoft's Enhanced Mitigation Experience Toolkit mitigation protections compatible with the target are enabled. "To encourage contestants to bring stealthy attacks, we are also forcing all attacks to require no user interaction beyond the action required to browse to the malicious content," Gorenc says. HP ZDI also is offering researchers a $25,000 bonus for system-level code execution exploits, which Gorenc says illustrate another way to escape the sandbox of the target browser, but to do so they will need to exploit a vulnerability in Windows.


A 'Flickr-ing' View of the World, in 4-D
Cornell Chronicle (02/12/15) Bill Steele

Cornell University researchers have developed Scene Chronology, a system that enables an observer to navigate a virtual three-dimensional space while using a slider control to move forward and backward in time. The software works with flat surfaces in the image and treats them as patches that are stitched together to create the total scene. The display currently shows only the planes, because the algorithm does not recognize the flat gray and brown surfaces of walls and pavement, according to Cornell researcher Kevin Matzen. The program compiles lists of positive and negative observations of a given feature and computes a time span over which the feature exists. In creating the entire scene at a particular moment, the program selects the features whose time span includes that moment. After a four-dimensional model is created, a new photo can be time-stamped by comparing it with the model. The major hurdle the researchers face is getting accurate time stamping, as amateur photographers do not always correctly set the clocks in their cameras, and in early experiments, human observers found anomalies in some of the images.


A Coast-to-Coast Picture of America's Cacophony of Sounds
Science News (02/16/15) Susan Milius

The U.S. National Park Service (NPS) used algorithms to predict the loudness of a typical summer day from coast to coast. NPS' Kurt Fristrup, Colorado State University's Daniel Mennitt, and other researchers fed data from about 1.5 million hours of acoustical monitoring into a machine-learning program. For each location, the scientists included details such as average summer precipitation and weekly plane overflights. The program discerned patterns in the geography data and predicted where the noise is. The team also created a map to display the summer soundscape. The map shows the eastern half of the U.S. is louder than the West, and cities and loud highways are clearly visible. The researchers also predicted the loudness of a summer's day without people, and again the East is louder, with the Mississippi River standing out as well as south Florida. The researchers say the findings also could benefit urban planners and biologists.


UNL Professor Develops Big Data Tool for Bridge Inspections
Lincoln Journal Star (NE) (02/11/15) Chris Dunker

University of Nebraska-Lincoln professor Zhigang Shen wants to create a bid data tool that would enable inspectors to quickly and easily access information about the condition of a bridge. Shen, a former construction manager with a degree in computer science, and colleagues have developed a program that displays the condition of individual bridge elements rather than rating different bridge assemblies--the deck, truss, beams--as a whole. Individual elements shown in the color green are in good condition, while those in yellow must be watched, and those in red require immediate repair or replacement. The program produces photos of buckling, corrosion, or other damage that could create problems, when clicking on any of the elements. "If you can see these problems in a [three-dimensional] model, you can connect the dots immediately," Shen says. He notes it is important to clearly represent the ratings of the various parts of bridges and provide an easy way to access past and current inspection records for structures. Shen also says bridges could be cataloged in the system in a week or even a few days.


Net Neutrality's Technical Troubles
IEEE Spectrum (02/12/15) Jeff Hecht

The U.S. Federal Communications Commission (FCC) is set to vote on new net neutrality rules on Feb. 26, which could lead to a regulatory regime that requires Internet service providers (ISPs) to treat all data the same. However, there is a technical dimension to the net neutrality debate that rarely gets much attention and could result in dramatically degraded service quality in some areas if net neutrality is implemented too stringently. One of the key mantra's of the net neutrality movement is that all data should be treated the same, but one of the fundamental tenets of network management is that to ensure the best service, some data needs to be treated differently due to the way the Internet transmits data. This is most notable when it comes to voice service, where prioritizing the transmission of specific packets, known as packet scheduling, is key to improving the quality of voice data sent over the Internet. Some net neutrality proposals would disallow packet scheduling, which could lead to significantly degraded voice-over-Internet service. Engineers and others say truly effective net neutrality rules should avoid such extremism and leave room for reasonable packet and network management on the part of ISPs.


Abstract News © Copyright 2015 INFORMATION, INC.
Powered by Information, Inc.


To submit feedback about ACM TechNews, contact: technews@hq.acm.org
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe