Association for Computing Machinery
Welcome to the May 18, 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


Google, NASA Work Together on Disney Show to Inspire Girls Into Sciences
The Washington Post (05/17/15) Cecilia Kang

When it was developing a new series about a family of space adventurers, Disney Junior wanted the show to help bury common media stereotypes about science and programmers, so it turned to the U.S. National Aeronautics and Space Administration (NASA) and Google for advice. A 2014 report by Google found a paucity of media portrayals of women in science has contributed to the low rate of girls pursuing science, technology, engineering, and math careers. Disney Junior visited Google headquarters in Silicon Valley and NASA's Southern California base to speak with technology and space experts about how to authentically portray the Callisto family in its new series, "Miles from Tomorrowland." NASA helped Disney develop the character of Phoebe, the mother and captain of the spaceship, including basing her design on astronaut Yvonne D. Cagle. At Google, the show's creator, Sascha Paladino, met with several female engineers to hone the depiction of Loretta, the main character's sister, who uses programming to solve problems. The Google engineers not only helped influence the Loretta character, but the show's depiction of computer code, helping the creators make it look more realistic. Google is hosting a screening of the show today at its Washington, D.C., office, followed by a panel discussion including Paladino, Disney executives, Cagle, and some of the Google engineers that consulted on the show.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 


Latest Self-Driving Google Car Heading to Public Streets
Associated Press (05/16/15) Dee-Ann Durbin

Google announced plans to debut the latest version of its self-driving car on public roads this summer. The new prototypes look similar to the prototype unveiled last year, but with a more robust feature set. Dimitri Dolgov, head of software for the self-driving car project, says Google's self-driving software has improved in the last year, and is much better at classifying objects and predicting the behavior of pedestrians and other cars. However, the new cars will still have limitations. The small electric cars will have a maximum speed of 25 miles per hour (in part because they lack several federally-required safety features such as air bags), and a maximum range of 80 miles. The cars also will only be able to operate in areas that have been thoroughly mapped by Google. The company says the first group of 25 cars will mostly be tested in the neighborhoods surrounding its Mountain View, CA, headquarters. However, the company plans to eventually produce 50 to 100 of the new prototypes and test them in rainier and hilly areas. Google co-founder Sergei Brin says the company does not aspire to be a car company and wants to partner with automakers to produce the final vehicles.


Code.org Targets High School Computer Science
USA Today (05/14/15) Jessica Guynn

Code.org is collaborating with College Board to work to expand computer science in U.S. high schools and increase the number of female and minority students taking computer science courses. Under the new partnership, high schools in 35 of the U.S.'s largest districts will be encouraged to offer Code.org's computer science course this fall. Targeted school districts are in cities including New York City, Chicago, and Los Angeles. Code.org CEO Hadi Partovi says the nonprofit will provide the curriculum, tools, training, and funding to school districts that qualify, while the College Board will help fund the work if a school district agrees to use the PSAT to identify students who have potential in computer science. Code.org says it wants to build on the track record of its Code Studio, which offers online tutorials in the basics of coding. Partovi says one out of 10 elementary and middle school students nationwide have created accounts with Code Studio. He notes of those students, 43 percent are female, 22 percent are Hispanic, and 15 percent are African American. A recent report from the Level Playing Field Institute found that in California public school students, African-American and Latino students make up 59 percent of the student population but took just 11 percent of the Advanced Placement Computer Science tests in 2014.


Minorities Pack a Major Punch
A*STAR Research (05/13/15)

Researchers from the A*STAR Institute of High Performance Computing say they have developed a new approach to data mining. Datasets can vary widely in the number of features measured and the number of independent observations taken. The team developed an approach for targeted feature selection from datasets with small sample sizes. The class imbalance problem--in which the common class data overwhelm the rare class data--is a significant hurdle in data mining. A*STAR's study shows emphasizing the less-common classes in datasets improves the accuracy of feature selection. Using an incremental approach, researcher Feng Yang and colleagues were able to significantly reduce the computational load related to feature selection from 4,215 seconds to 49 seconds. They started with a common pattern-classification method called linear discriminant analysis, but the dataset had to be regularized to make feature selection tractable. "From the view of sample distribution in the subspace, minority class emphasis will actually 'squeeze' the samples in the minority class to form a compact 'nucleus' in the subspace of selected features, which would be easier to be classified," Yang says.


Machine-Learning Algorithm Calculates Fair Distance for a Race Between Usain Bolt and Long-Distance Runner Mo Farah
Technology Review (05/15/15)

Researchers at the Humboldt University of Berlin and University College London have developed a model that accounts for the different kinds of athletic performance required for short, middle, and long-distance running. The researchers used the model to predict an athlete's performance at one distance given their performance at others. It has long been known that a small increase in average speed dramatically reduces the distance at which a world record is possible. When researchers plot world-record speeds against distance, it produces a power-law curve with a strange kink in its shape, as if one power law governs running speeds over distances shorter than a mile while another governs running speeds at longer distances. This kink means that conventional predictive models cannot determine how fast sprinters will be able to run long distances and vice versa. The researchers developed a new model using a database of athletic performance since 1954 in Britain, containing almost 1.5 million individual performances by both genders ranging from the amateur to the elite, both young and old. The researchers then used a machine-learning algorithm to find an equation that best fits the data in a way that predicts an individual's performance at one distance, given their performance at other distances.


Microsoft Hyperlapse Turns Long Video Into Distilled Entertainment
Phys.org (05/15/15) Nancy Owano

Microsoft Research is developing Microsoft Hyperlapse, technology that can convert an otherwise tedious video into a short, distilled version. The technology is designed to glide quickly over times when not much is happening, according to Microsoft's Allison Linn. Microsoft Hyperlapse consists of a set of products designed for smooth, stabilized time lapses from first-person videos. The first is Mobile, which shares videos with friends and family. In addition, there is Pro, which can create a hyperlapse using a Windows computer from video shot on a camera or other device. Finally, developers can use Hyperlapse for Azure Media Services to integrate hyperlapse options into websites and apps. The technology searches for entire frames with the most overlap with each other, which means it can disregard a sudden jerk and go instead to the video's most compatible segments. "The researchers came up with an algorithm that first creates an approximate [three-dimensional] model of the landscape being filmed, and then identifies the dominant path that the camera took through that landscape," Linn says. The algorithm then stitches together parts of different frames to create a smooth, stable hyperlapse that showcases the essence of the original video.


Say Hello to Machines That Read Your Emotions to Make You Happy
New Scientist (05/14/15) Sally Adee

EmoSPARK is an artificial intelligence-based (AI) device that can gauge a person's mood based on what they say and how they say it. The device's brain consists of a 90-millimeter Bluetooth and Wi-Fi-enabled cube. It senses its world through an Internet connection, a microphone, a webcam, and the owner's smartphone.  The cube is programmed to use these technologies to help make the owner happy, for example by responding to commands to play a song in a digital library, making posts on Facebook, checking friends' updates, streaming a Netflix film, answering questions by tapping information from Wikipedia, and making conversation.  EmoSPARK's creators, Brian Fitzpatrick and Patrick Levy Rosenthal, say the device is dedicated to the owner's happiness. To fulfill that, it attempts to take the owner's emotional pulse, adapt its personality to them, and always attempt to understand what makes the owner happy and unhappy.  The drive to give AI an emotional dimension is a result of necessity, says Rana el Kaliouby, founder of Affectiva, which creates emotion-sensing algorithms. She notes as devices ranging from phones to refrigerators increasingly get connected to the Internet, people are seeking ways to impart a human aspect to machine logic.


Artbot Engineers the Discovery of Art
MIT News (05/13/15) Kathryn O'Neill

Researchers in the Comparative Media Studies/Writing's HyperStudio research group at the Massachusetts Institute of Technology (MIT) have developed Artbot, a mobile website app that mines both user preferences and event tags to provide connections to the local art scene. Artbot enables users to select their interests from a wide-ranging list, and collects data from museum websites to find artists, movements, and themes that link events to each other in various ways. Artbot then cross-references the data collected with the user's preferences to generate event recommendations. "While digital media can provide ways to discover art online, our project aims to put people directly in front of works of art," according to the researchers. The project is designed to encourage a greater understanding and appreciation of art. Artbot was designed using open source software and is itself an open source application, which should enable other museums and cities to easily modify it. "Our mission is to come up with very innovative approaches and development models that can be applied in different domains," says HyperStudio executive director Kurt Fendt. "This is typical of what we do at HyperStudio. We make our work open source so we can contribute to work going on in the field and to digital humanities at large."


A Secure, Anonymous, Easy Way to Pay for Online Content
University of Luxembourg (05/13/15)

Researchers at the University of Luxembourg have developed a new way for readers, viewers, and gamers to pay for online content without making cash payments. Every time users like the content, they can choose to donate a small amount of their PC's spare computing power. Bitcoin and other virtual currencies can use the spare processing power to perform the billions of calculations they need to build and maintain their virtual ledgers. The donation would generate virtual currency payments to the content providers. The virtual currency could then be converted into standard fiat currencies to remunerate authors, artists, and other content generators. To address the issues of security and privacy, a verifiable proof of work certificate is generated and sent to the content provider using an anonymous network when the computation is completed. The content provider is paid in virtual currency for the amount of computation performed by the user. "Each transaction would only be a micro-payment of a fraction of a cent, but this could become an important source of revenue for very popular content providers," says Luxembourg professor Alex Biryukov.


FAU Receives NSF Rapid Response Grant to Develop Innovative Computer Model for Ebola Spread
Florida Atlantic University (05/13/15) Gisele Galoustian

Florida Atlantic University (FAU) researchers have received a U.S. National Science Foundation (NSF) Rapid Response Grant to develop an innovative model of how Ebola spreads by using big data analytics techniques and tools. The program will feed massive amounts of data from various sources, including Twitter, Facebook, and Google, into a decision-support system that will model the spread pattern of the Ebola virus and create dynamic graphs and predictive diffusion models on the outcome and impact of the disease. "Our system can be a proactive approach to reasonably reduce the risk of exposure of Ebola spread within a community or a geographic location," says FAU professor Borko Furht. The model will utilize FAU's cloud-computing system, as well as LexisNexis' open source platform for big data analysis and processing for large volumes of data. "Our program is being quickly developed to identify and visualize families and tightly connected social groups who have had some contact with an Ebola patient," says FAU professor Borko Furht, director of the NSF Center for Advanced Knowledge Enablement at FAU. The researchers also are developing a mobile interface that enables the system to analyze the movement of a user and possible contacts in the areas affected by Ebola.


UW Study Examines Gender Bias in Stock Images
The Daily of the University of Washington (05/13/15) Kate Clark

Researchers from the universities of Washington (UW) and Maryland analyzed gender bias in online image results, and their study found a systematic underrepresentation of women. In occupations that have the same number of women and men, the researchers report women only account for 45 percent of the search images. Moreover, women in the images sometimes appear highly sexualized. When study participants were asked to identify which images showed a more professional and appropriate-looking person for a given occupation, they did not tend to exhibit open sexism in their choices. One surprising finding was that participants more often clung to stereotypes, which meant when an image matched the stereotype of the profession, it was more likely to be chosen by the participant. Since launching the project, Getty Images, in partnership with Facebook CEO Sheryl Sandberg, has released a set of images that more accurately portrays women. The researchers hope their work will lead search engines to consider how their algorithms are promoting negative stereotypes and shaping perceptions about occupations. “A lot of groups have the ability to be part of the solution and so can we when we choose which images we select," says UW professor Sean Munson. "The algorithms take into account what we choose when they choose what images to show the next person."


Autonomous Car Prototype Folds, Shrinks, Drives Sideways
Scientific Computing (05/12/15) Suzanne Tracy

A group of engineers, software developers, and designers in Germany have developed a design for a new type of electric micro car. The prototype, now in its second phase, is able to convert from conventional driving to driving sideways in just seconds. Each wheel is powered by its own motor, which enables the two-seater car to shrink from eight feet to less than five feet in length. The car shifts its rear axle to the front and slides on a set of rails to raise the interior upwards, all while the occupants are comfortably seated. The Innovative Technologies Electromobility (ITEM) initiative is an undertaking of the German Federal Ministry of Transport, Building & Urban Development's New Mobility in Rural Areas project. Under ITEM, the DFKI Robotics Innovation Center is designing and implementing the vehicle as a technology demonstrator for "(electro)mobility of the future." Project partners include H2O e-mobile and Fraunhofer IFAM. The prototype has a top speed of 40 miles per hour, and can travel 30 to 44 miles on a single four-hour battery charge. A foldable docking interface fits into the body of the car for charging, and when not charging the vehicle, it can be used to connect range extenders and passenger or cargo modules.


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