Association for Computing Machinery
Welcome to the March 16, 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).

HEADLINES AT A GLANCE


Computer Science Education in the Age of CS for All
The Huffington Post (03/15/16) Mark Nelson

Momentum for bringing computer science (CS) to all primary and secondary U.S. schools is building thanks to efforts such as the White House's recently unveiled CS for All initiative, but there are many unresolved issues, writes Computer Science Teachers Association executive director Mark Nelson. He cites the definition of primary/secondary CS education adopted by the U.K. Department of Education, which enables students to use computational thinking and creativity to understand and change the world. The definition also says CS education guarantees the digital literacy of students is appropriate for the future workplace and for participating in a digital world. Nelson says the U.K. definition "balances the theoretical aspects of CS with the applied dimensions," as well as "captures the broader essence of CS as something both creative and integrated with other bodies of knowledge." Once agreement on what CS is and the need for making it available to all students is reached, Nelson notes other challenges must be met. He says those challenges include ensuring there is an abundant supply of CS educators with a diverse combination of skills and backgrounds. To this end, Nelson calls on CS for All's organizers to collaborate with teachers on addressing the CS education conundrum, and to accord U.S. educators a greater level of respect.


Why the Government Can't Actually Stop Terrorists From Using Encryption
The Washington Post (03/15/16) Andrea Peterson

Security experts say the U.S. government's efforts to force technology firms to grant them access to encrypted devices or messaging services will do little to prevent terrorists and other wrongdoers from making use of the technology. They note many encrypted products are developed by people outside the federal government's jurisdiction, either because they are based in foreign countries or are part of open source projects. "Trying to put a mandate on encryption software is really pretty hopeless," says University of Pennsylvania professor Matt Blaze. He stresses a mandate would only impact software covered by U.S. statutes, and will not prevent people from using open source or foreign-made software, even domestically. In 2015, 16 encrypted communications applications developed either abroad or by open source projects were cataloged by researchers at New America's Open Technology institute. Analysts say a federal crackdown on U.S. encryption tech firms may end up spurring consumers to use foreign rivals that do not comply with such constraints. Moreover, a successful government shutdown of an open source project would not eliminate code already accessible via the Internet. Johns Hopkins University professor Matthew Green says law enforcement officials likely hope to restrict the amount of data that is automatically encrypted.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 


Google's Computer Program Beats Lee Se-dol in Go Tournament
The New York Times (03/15/16) Choe Sang-Hun

Google DeepMind's AlphaGo computer program on Tuesday defeated Lee Se-dol, a South Korean master of the ancient board game Go, in the final match of the series, achieving a 4-1 victory in their tournament. Experts say AlphaGo's victory represents the last remaining great hurdle for computer programmers attempting to make software better than humans at board games. Previously, artificial intelligence experts had predicted a computer program needed at least 10 more years of development before it would be able to beat Go masters such as Lee. However, AlphaGo already had surprised the Go community when it beat three-time European Go champion Fan Hui five games to none in October. Following the match against Lee, Hong Seok-hyun, head of the Korean national Go association, awarded the AlphaGo team the certificate of an honorary Go degree of Nine Dan, the highest ranking granted. Playing against Lee exposed many weaknesses of AlphaGo that the Google DeepMind team will try to address, according to DeepMind CEO Demis Hassabis. He notes AlphaGo relies on algorithms that could one day be used in all sorts of problems, from healthcare to science. Lee says AlphaGo was not like any human opponent he has faced. "It remained unfazed psychologically and stayed focused," he says.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 


What AI Can Tell Us About the U.S. Supreme Court
The Conversation (03/11/16) Mohammad Raihanul Islam; K.S.M. Tozammel Hossain; Siddarth Krishnan

Researchers at Virginia Polytechnic Institute and State University's (Virginia Tech) Discovery Analytics Center propose in a new paper using computer-based machine learning to model the U.S. Supreme Court. The researchers have developed a data-driven framework that can infer justices' judicial preferences and voting behavior, as well as answering questions about this behavior. The researchers say their Supreme Court Ideal Point Miner harnesses information on judicial preferences gathered from opinion texts to enhance existing research. The model assumes every case entails a combination of issues or topics on which justices have different views, and it studies the text of opinions and counts the incidence of words related to each issue that factors into the decision. The system then assigns relative importance to each such issue according to its share of relevant words, and it can deduce the strength of each justice's feelings on an issue by aggregating the analyses of multiple opinions. The model also can identify the high court's swing justices by examining the degree of variance of each justice's "ideal point" across multiple issues. The Virginia Tech researchers say knowledge gleaned from the model can describe the decision process for an individual case, as well as predicting justices' decisions in future cases with 79.46-percent accuracy.


To Get Truly Smart, AI Might Need to Play More Video Games
Technology Review (03/16/16) Will Knight

A computer scientist in France is testing the idea of using the life-like worlds of computer games to teach artificial intelligence (AI) algorithms about the real world. Xerox Research Center Europe's Adrien Gaidon is developing realistic three-dimensional environments for training algorithms to recognize real-world objects and scenarios. He says the project is important because cutting-edge AI algorithms need huge quantities of data to learn to perform a task, but most companies do not have access to enormous datasets or the means to generate such data from scratch. "The nice thing about virtual worlds is you can create any kind of scenario," Gaidon notes. His team used the popular game development engine Unity to generate virtual scenes to train deep-learning algorithms. His group also measured the accuracy of the approach by comparing algorithms trained within virtual worlds with those trained using real images annotated by people. The team's AI algorithms will try to find empty parking spaces on the street and also try to learn about medical issues.


DARPA Calls on DIYers for Weaponized Tech
Government Computer News (03/14/16) Mark Pomerleau

The U.S. Defense Advanced Research Projects Agency (DARPA) has launched the Improv program, which will seek ideas on how products assembled from relatively benign technologies might be transformed into security threats. The Improv program aims to quickly surface and prototype devices made from off-the-shelf electronics or open source code that can cost-effectively deliver sophisticated military technologies and capabilities. Improv is designed to address the fact that the U.S. no longer enjoys a near-monopoly on access to the most advanced technologies. DARPA is encouraging participation from a wide range of sectors, including anyone who can reconfigure, reprogram, modify, or recombine commercially available technology. "Improv is being launched in recognition that strategic surprise can also come from more familiar technologies, adapted and applied in novel ways," says Improv program manager John Main. Improv consists of three phases in which DARPA will fund selected proposals, evaluate the prototypes, and advance the relevant capabilities in separate follow-up efforts.


Using Avatars and Robots to Treat Social Disorders
CORDIS News (03/10/16)

The European Union-funded ALTEREGO project has developed a computer architecture and software to help patients diagnosed with social disorders adapt their behavior by interacting with avatars and robots. In the first stage of the project, the researchers worked with about 40 patients, recording their movements with cameras and creating their avatars using virtual-reality techniques. Using a "mirror" game, involving the synchronized handling of colored balls, which is known to increase affiliation, the patients took turns with the avatar to lead, as different variables were introduced in both the game and the movement of the avatar. "Everyone moves in a very personal way and, using variables, we work with this to transform similarity into difference, by morphing the avatar and trying in this way to change the behavior of the patient over time," says University of Montpellier professor Benoit Bardy. In addition, the patients worked with a humanoid robot equipped with 53 motors that move the head, arms, hands, waist, and legs, and it can see, hear, and smile. ALTEREGO includes research in fundamental and clinical neurosciences, interaction modeling, the development of new computer-vision techniques, and human-robot interfaces.


Under the Hood of the NCAA Bracket Builder
TechCrunch (03/14/16) Tishin Donkersley

Microsoft Bing and the National Collegiate Athletic Association (NCAA) are partnering to provide fans a "smarter bracket" based on a machine-learning system that uses the Bing Predictor tool. The Bing Predictor tool incorporates "the wisdom of the crowd," and identifies patterns, correlations, and signals within those data points. The tool consists of machine-learning technology that pulls in consumer-oriented data to find signals that correlate to how people will vote, which Microsoft Bing's Walter Sun says is very valuable in terms of driving signals that might not be easily available to the general public. For the "smarter bracket," there are more than 9 quintillion possible outcomes for all 67 tournament games in the men's division, and the Predictor scours the Internet for fan sentiments and traffic around the teams. Researchers add in layers of historical data to arrive at their predictions and the NCAA has provided Bing with more than 10 years of historical team data. Last year, the smart bracket was 73-percent accurate for the men's tournament and 79-percent accurate for the women's tournament, and finished in the top 30 percent of all brackets created nationwide. "Machine-learning models work best when they can best simulate the game of today from the information they learn," Sun says.


Algorithms and Experiments Make Strange Bedfellows at SXSW
Computerworld (03/14/16) Lamont Wood

Two conference sessions hosted at the South by Southwest (SXSW) Interactive festival last weekend discussed how experimenting with search and social media algorithms can produce unsettling results and provoke equally unsettling user reactions. "We have a crisis of confidence in algorithms," said University of Michigan professor Christian Sandvig, who urged the design of visible instead of invisible algorithms. Intel researcher Dawn Nafus cited a fitness monitor algorithm that berated her for being unhealthy for not taking enough steps at a time when she was confined to a wheelchair, which would have had serious implications for her qualifying for a health insurance discount. She argued the industry should make data downloads routine so users can deviate and "domesticate" their data via personal analysis. The second session highlighted users' backlash to Facebook's experimentation with the algorithm that chose items for their news feeds to see if it would alter their emotions. Stanford University professor Jeff Hancock said the biggest problem with the experiment was likely Facebook's violation of users' expectations. "They thought of it as a platform, and platforms don't experiment on people," he noted. Google's Elizabeth Churchill recommended designers design experiments to support their design decisions as part of the process of producing a user experience using qualitative and quantitative measures that can interrelate to improve overall quality.


Forget the Robots--Here Come the Geminoids!
USA Today (03/14/16) Rick Jervis

Last weekend's South by Southwest (SXSW) Interactive Festival was a showcase for many types of robots, and a highlight was Japanese roboticist Hiroshi Ishiguro's demonstration of an android of himself that held unassisted conversations in a very human-like manner. Ishiguro also discussed the potential of another robot called a geminoid, which enables users to remotely manipulate it and communicate through it via a wearable brain-machine interface. Noting his androids are already employed in Japan as TV hosts, shopkeepers, and theater actors, Ishiguro predicted "we'll have a robot society in the very near future, in maybe three or five years." Also showcased at SXSW was Aldebaran Robotics' Pepper, a "companion robot" designed to live and interact with humans in an assistive function. Aldebaran's Rodolphe Gelin said Pepper is intentionally designed to have a non-human-like form. "We don't want to cheat people: if it's a robot, it's a robot," he said. "If you ever have to question [whether] that's a human being or a robot, that's no good." Another notable demonstration was of Sougwen Chung's Drawing Operations Unit: Generation-1, a robotic arm that draws in real-time symmetry with the artist by following her strokes while generating its own movements.


After Moore's Law: Double, Double, Toil and Trouble
The Economist (03/12/16) Tim Cross

Moore's law is approaching its physical limits after 50 years of exponentially increasing the density of electronic components and computing power on chips. For some time, shrinking transistors has no longer resulted in a boost in energy efficiency, which means the operating speed of high-end processors has been flat since the mid-2000s. Moreover, the costs of chip fabrication have risen as the chip components near atomic scale. "From an economic standpoint, Moore's law is over," says analyst Linley Gwennap. With Moore's law losing traction, scientists are hoping to bypass its limitations with alternative approaches to computing. "The end of Moore's law could be an inflection point," notes Microsoft Research's Peter Lee. "It's full of challenges--but it's also a chance to strike out in different directions, and to really shake things up." One concept involves tapping quantum mechanics to execute calculations at a much faster speed than a classical computer, while another is to mimic the biological brain to realize faster processing with low energy consumption. A third concept entails a diffusion of computer power instead of a concentration, spreading the ability to communicate and calculate across an ever-wider spectrum of everyday objects in the Internet of Things.


Puzzle Game Launched to Help Program Quantum Computers
New Scientist (03/10/16) Jacob Aron

Researchers at the RIKEN Center for Emergent Matter Science have turned the problem of programming a quantum computer into a game called meQuanics. The game is based on topological error correction, which many research groups are using to attempt to create large-scale quantum computers. Topological error correction works by carving out circuits in a three-dimensional (3D) grid of qubits; the larger the circuit, the more qubits are needed and the more challenging it is to build a computer. It is the general structure of the circuits, and not their specific shape, that determines their function. Topologically rearranging a quantum circuit can shrink its overall volume, reducing the number of qubits required. "We need to teach a classical computer how to optimize and compile quantum circuits, and to do that we need to give it a huge amount of examples to learn from," says RIKEN researcher Simon Devitt. In the game, circuits appear as 3D puzzles, and players have tools to manipulate and shrink the puzzles without breaking rules that relate to the underlying scientific theory. "If meQuanics becomes as popular as professionally designed games, we can make a huge impact," Devitt says.


Amputee Feels Texture With a Bionic Fingertip
Ecole Polytechnique Federale de Lausanne (03/08/16) Hillary Sanctuary

Technology developed by researchers at the Swiss Federal Institute of Technology in Lausanne and Scuola Superiore Sant'Anna (SSSA) has enabled an amputee to feel rough or smooth texture in real time. The arm of amputee Dennis Aabo Sorenson was wired to an artificial fingertip equipped with sensors. An electrical signal generated by the sensors was translated into a series of electrical spikes, imitating the language of the nervous system, then delivered to the nerves. Sorenson says he was able to distinguish between rough and smooth surfaces 96 percent of the time. The team conducted the same experiment on non-amputees, without the need of surgery, and they were able to distinguish roughness in textures 77 percent of the time. Comparing the brain-wave activity of the non-amputees, the team reports the brain scans revealed activated regions analogous to the feeling of touch from a real finger. The technology could be "translated to other applications, such as artificial touch in robotics for surgery, rescue, and manufacturing," says Calogero Oddo of the BioRobotics Institute of SSSA.


Abstract News © Copyright 2016 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