Association for Computing Machinery
Welcome to the August 1, 2014 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).


ScratchJr Teaches the Touchscreen Generation to Code (08/01/14) Marcus Wohlsen

Massachusetts Institute of Technology (MIT) researchers have developed ScratchJr, a new iPad version of the Scratch programming language that will help teach kids how to code. Scratch uses interlocking colored blocks to mimic the logical structures and functions of a typical programming language and has users interact with cartoonish characters called "sprites" to learning coding skills. ScratchJr is designed to bring this dynamic to an even younger audience. "We wanted to make sure young people aren't just using tablet for browsing and consuming," says professor Mitchel Resnick, head of the MIT Media Lab's Lifelong Kindergarten group, which developed ScratchJr along with researchers at Tufts University and the Playful Invention Company. "This is a tool that they can use to get their voice out in the world, not just to consume what other people are doing." ScratchJr, which is targeted at kids aged five to seven, is almost exclusively graphic-driven, which makes it accessible to an age group that might still be learning how to read. For the most part, ScratchJr users can create rudimentary but fun programs by tapping and dragging on the touchscreen interface. However, Resnick says ScratchJr retains the essential computational concepts taught by Scratch, such as events, sequencing, and iteration.

How 'Game of Thrones' Will Predict the Next Bin Laden
Defense One (07/29/14) Patrick Tucker

A common terror-fighting tactic has been using drone strikes to kill the leaders of terrorist organizations in the belief this will hinder or disrupt the group. A research team composed of University of Maryland computer scientists V.S. Subrahmanian, Francesca Spezzano, and Aaron Mannes developed a method of statistical analysis they say can better predict the outcomes of such targeted killings, and potentially avoid situations in which killing one leader leads to the rise of an even more dangerous and brutal one. The researchers' new method, Shaping Terrorist Organization Network Efficiency (STONE), applies network theory to open source data on terrorist groups to identify various traits of different members, such as their position and influence within the group, and uses these factors to determine who is likely to rise to power should the current leader be killed, as well as how that would effect the groups' disposition. Important traits include influence within the group, connectedness to other members, and formal rank within the organization. Subrahmanian says the research team plans to use their STONE methodology to predict the characters most likely to become the next terror mastermind. The researchers published their findings in the August 2014 issue of Communications of the ACM.

U.K. to Allow Driverless Cars on Public Roads in January
BBC News (07/30/14)

The United Kingdom (U.K.) has announced that driverless cars will be allowed on public roads starting in January 2015, and the government has invited cities to compete to host one of three trials of the new technology. Government ministers also have called for a review of the U.K.'s road regulations to provide appropriate guidelines. "Today's announcement will see driverless cars take to our streets in less than six months, putting us at the forefront of this transformational technology and opening up new opportunities for our economy and society," says U.K. Department of Transport business secretary Vince Cable. Although University of Oxford researchers have been experimenting with driverless cars, concerns about legal and insurance issues have restricted the technology to private roads. U.K. cities wanting to host one of the trials, which are expected to run for 18 to 36 months, have until the start of October to declare their interest. The trials will examine how the rules should apply to vehicles in which the driver can take back control at short notice, and how they should apply to vehicles in which there is no driver. Meanwhile, the state governments of California, Nevada, and Florida have all approved tests of driverless cars.

Can Winograd Schemas Replace Turing Test for Defining Human-Level AI?
IEEE Spectrum (07/29/14) Evan Ackerman

Earlier this year a chatbot dubbed Eugene Goostman "beat" a Turing test for artificial intelligence (AI) organized as part of a contest put on by a U.K. university. Many observers and AI experts think the event does more to demonstrate how flawed a tool the Turing test is for measuring artificial intelligence than it does the sophistication of the chatbot, which was able to convince a significant portion of its questioners that it was human by claiming to be a teenager from the Ukraine that spoke English as a second language. This pretense enabled the questioners to excuse the chatbot's errors and tangents as the eccentricities of a young person speaking a foreign language. A new proposed method of testing AI intelligence are so-called Winograd schemas, which present an AI with an ambiguously worded sentence, such as "The trophy doesn't fit in the brown suitcase because it is too big," and asks the AI to identify the ambiguous pronouns referent, something that is intuitive to humans because of their depth of knowledge and understanding about the relationships between the objects or people being described. A new AI contest sponsored by Nuance Communications and is offering a $25,000 prize to an AI that can successfully answer Winograd schemas.

Lack of Coding Skills May Lead to Skills Shortage in Europe (07/30/14) Archana Venkatraman

A lack of basic coding skills could result in Europe facing a shortage of up to 900,000 information and computer technology (ICT) professionals by 2020, according to the European Commission (EC). Coding is today's literacy and important to enable the digital revolution, say EC executives Neelie Kroes and Androulla Vassiliou. They say fundamental coding skills are poised to become critical for many positions in the near future as society transitions to a world in which cloud-based and connected devices are more common. Although the EC says more than 90 percent of professional occupations require some ICT competence, the number of computer science graduates is not keeping pace with this demand. Kroes and Vassiliou have urged European union (EU) education ministers to get children more involved in EU Code Week, which takes place across Europe in October. EU Code Week aims to make coding more visible and motivate children and adults alike to learn new skills. Meanwhile, a poll of U.K. boardrooms found that 94 percent of senior executives consider digital skills to be very important to their business, but 20 percent said the quality of digital skills among graduates is below average.

The Social Laboratory
Foreign Policy (07/29/14) Shane Harris

Singapore's Risk Assessment and Horizon Scanning (RAHS) program has evolved far beyond its initial purpose of deploying mass surveillance and big-data analysis for national defense and future threat deterrence to govern a wide range of activities, including procurement cycle planning, economic forecasting, immigration policy, and educational strategy development. The Asian city-state has become a testbed for how centrally controlled technology might be used to maintain order as well as create a more harmonious society. The RAHS system's platform is a cognitive model that mimics the human thought process to analyze data from virtually any source and generate simulations used to predict potential events. RAHS analysts use an ideas database to produce "narratives" detailing the possible path that threats or opportunities might follow, visualizing a series of potential futures to inform the Singaporean government's strategies. Information from surveillance cameras, publicly available sources such as Twitter messages and blog posts, Internet traffic, and other data is monitored mainly for pornography and racist invective. Singapore's government also is focusing RAHS technology on the challenge of ensuring national unity of purpose to keep the country stable and whole. Singaporeans seem willing to accept a "social contract" between them and the government in which mass surveillance and big-data scanning is tolerated for the sake of fundamental guarantees such as education and security.

What 6.9 Million Clicks Tell Us About How to Fix Online Education
MIT News (07/28/14) Adam Conner-Simons

Researchers at the Massachusetts Institute of Technology's (MIT) Computer Science and Artificial Intelligence Lab (CSAIL) have analyzed data from Harvard University and MIT's shared online learning platform to gain a richer understanding of what makes massively open online courses (MOOCs) succeed and fail. The researchers examined data on the second-by-second viewing habits of about 100,000 learners over 6.9 million video sessions. They found online learners value several things in educational videos: brevity (videos of no more than six minutes); informality, with speakers seated behind a desk rather than standing at a podium; fast talkers, preferably in the range of 254 words per minute; lectures with built-in pauses so complex visuals can be absorbed and processed; and videos crafted specifically for the Web, rather than full-length lectures broken up into shorter videos. The CSAIL researchers used these and other insights to craft the LectureScape tool, which presents MOOC videos in a more intuitive, dynamic, and effective manner. Features include a timeline of the video, searchable interactive transcripts, automatic word clouds and section summaries, and a way to highlight popular content. Juho Kim, one of the lead researchers on the team, says they hope to add other features such as Netflix-like video recommendations and links to other relevant videos.

Unthinking Computers Pull Off Clever Parlor Tricks
Financial Times (07/30/14) Richard Waters

Enthusiasm for a revival in artificial intelligence (AI) research often centers around the concept of deep learning, in which an array of processors could function like a network of neurons and parse information in a brain-like manner once they are fed enough data. Using deep-learning methods to make computers capable of pattern-recognition tasks that are on a par with human capability is impressive, but it carries the risk of making the output of such systems comparable with products of actual human intelligence. However, these are clever parlor tricks that must be carefully evaluated to determine where they should be applied to everyday life, and how much faith to place in them. Deep-learning systems do not utilize the kind of transparent reasoning involved in classical AI, in which computers are fed defined bodies of knowledge and rules about how to interpret them, which makes their output innately mysterious, according to skeptics. AI researcher Oren Etzioni describes a diagnostic system that recommends removing a patient's kidney as an example of what happens when deep learning oversteps its bounds. When controlled by human experts who know when to use it and how to interpret output, deep learning could potentially revolutionize machine-assisted decision-making.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

How to Spot a Social Bot on Twitter
Technology Review (07/28/14)

University of Indiana researchers have developed a method to identify sophisticated social bots and distinguish them from ordinary human users. The researchers chose 15,000 previously known social bots and collected their 200 most recent tweets as well as the 100 most recent tweets mentioning them. This process produced a data set of about 2.6 million tweets, and the researchers added a similar dataset for 16,000 human users consisting of more than three million tweets. The researchers then created Bot or Not?, an algorithm that mines the data looking for significant differences between the properties of human users and social bots. The algorithm examined more than 1,000 features associated with the accounts and found significant differences between human accounts and bot accounts. For example, bots tend to retweet more often than humans and they also have longer usernames and younger accounts. In addition, humans receive more replies, mentions, and retweets. The researchers say these factors can be used to identify bots.

NICTA Pushes seL4 Microkernel Beyond Drones
Techworld Australia (07/29/14) Rebecca Merrett

NICTA, previously known as National ICT Australia, recently released the seL4 microkernel operating system as open source, including all of the kernel's source code, the mathematical proofs, as well as other code and proofs for building highly secure systems. SeL4, developed jointly with General Dynamics, has full "functional correctness proof," which means the implementation adheres to its specification. "[Defense] is pretty much what [General Dynamics] are mostly interested in, so they are less interested in other use cases, particularly when it's really small initial products that you don't know if they will turn into something really big," says NICTA professor Gernot Heiser. The operating system so far largely has been used for military systems, but the researchers hope the seL4 open source code will be employed for medical devices and industrial automation, among other uses. "SeL4 gives you guaranteed isolation between the critical and less critical parts of the system," Heiser notes. In addition, anyone who builds a project using the operating system under the GNU GPLv2 license has to distribute it under the same open source license or a newer version.

Engineers Tap Gaming Technology to Improve Design
University of New South Wales (07/28/14) Ry Crozier

University of New South Wales (UNSW) researchers are using a virtual reality headset and open source software to create immersive worlds for engineers. "We're very interested in immersion--entering the virtual image or space generated by a computer--and how it might help in engineering understanding," says UNSW researcher John Page. He notes the technology could provide a more cost-effective way for engineers to experience the physics of environments for which they design. "It's not really practical to send your engineers into space, for example, to learn how things function in that environment," Page says. "We're trying to create very rich environments using computer game theory in which we're able to expose engineers to real physics of environments such as space." Page also thinks immersive technology could enable operators to perform their jobs in a virtual world first while processes are still being developed, providing valuable feedback to engineers. Engineers also could fill a virtual factory with humanoid robots that work with the processes and develop skills based on learning and training. The researchers are using MakeHuman, free open source software, to create the humanoid agents.

Out in the Open: Sandstorm Makes It Easy to Control Your Apps in the Cloud
Wired News (07/28/14) Klint Finley

Former Google engineer Kenton Varda and neuroscientist Jade Wang have developed Sandstorm, an open source project that gives users the same amount of control over cloud apps as there is for private servers. Sandstorm enables users to sign in with a Google or GitHub account, find the desired app, and install it; Sandstorm manages all the prerequisites, updates, and maintenance. Sandstorm is designed to make it easy for Web hosts to run almost any Linux application, regardless of what language it is written in or what Web server software it requires. Although Sandstorm will offer its own application hosting service, the software platform is open source, which means any hosting company can run it. Varda and Wang say their approach will be a major improvement over the way things are currently done because users will be able to choose between multiple hosts. "In this environment, hosts will be competing on trustworthiness rather than on features," Varda says. Sandstorm's long-term success will rely on convincing other developers to port apps to the program, and Varda and Wang's company is planning an app marketplace where developers will be able to sell both open source and non-open source apps.

USC Researcher Gets Major Commitment to Study Computer Science
USC News (07/24/14) Robert Perkins

The Simons Foundation has selected Shang-Hua Teng, the University of Southern California's Seeley G. Mudd Professor of Computer Science, for its class of 2014 Simons Investigators. Teng is just one of two computer scientists chosen nationwide for the program, and he will receive a five-year, $500,000-grant that will help fund research into theoretical computer science. The foundation also gives its investigators the opportunity to renew the award for an additional five years, representing a significant and unusually lengthy commitment to funding in this area of science. Teng's work focuses on fundamental questions at the foundation of computing, including optimization problems, networking analysis, and game theory. He says the grant enables him to work with a long-term mindset, funding graduate students from both the computer science and mathematics departments to execute research that "focuses on the long game, rather than pursuing the low-hanging fruit." Teng also says watching his two-year-old daughter learn to speak has sparked an intense interest in understanding language acquisition and the process of learning. "This is a very exciting period of time for computing, particularly theoretical computing, as its frontier is rapidly expanding," he says.

Abstract News © Copyright 2014 INFORMATION, INC.
Powered by Information, Inc.

To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe