Welcome to the May 20, 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
Silicon Valley's Tech Workforce Diversity Has Long Way to Go, Say Feds
Computerworld (05/20/16) Patrick Thibodeau
Much progress needs to be made before Silicon Valley's technology industry has a truly diverse workforce, according to an analysis of the top 75 regional tech firms by the U.S. Equal Employment Opportunity Commission (EEOC). The EEOC found 47 percent of those companies' employees are white, while 41 percent are Asian American, 6 percent are Hispanic, and 3 percent are African American. Moreover, the commission estimated only 30 percent of the workforce in the companies is female. In comparison to overall private industry employment, the tech sector nationally employs larger percentages of both whites and Asian Americans, and smaller percentages of African Americans and Hispanics. On a national level, 64 percent of high-tech employees are men, compared to 52 percent in the broader workforce, while women comprise 36 percent of the tech workforce, versus 48 percent of the broader workforce. AARP Foundation attorney Laurie A. McCann says ageism may be a particular shortcoming in the tech sector. "The rampant age discrimination...is perhaps most evident in companies' hiring policies and practices, which are designed to attract and hire younger employees," she notes. "Expanding diversity and inclusion is critical to unlocking the full potential of tomorrow's economy," says EEOC chair Jenny Yang.
Google Isn't Playing Games With New Chip
The Wall Street Journal (05/18/16) Robert McMillan; Don Clark
The Tensor Processing Unit Google unveiled on Wednesday is the culmination of three years of secret development, and the company has been using it for more than a year to speed up artificial intelligence applications as machine learning becomes increasingly vital to its chief businesses. The chip is touted as being 10 times faster than alternatives. Stanford University professor Mark Horowitz sees this and other efforts by rivals as exemplary of the use of new processor designs to enhance general-purpose processors amid the slowing rate of upgrades in the field. According to Google, the Tensor Processing Unit gives it a seven-year advantage over currently available machine-learning processors. The chip employs machine learning in more than 100 programs for apps that include search, voice recognition, and self-driving cars. "In order to make them feasible to roll out, economically, with the required latency for users and all that stuff, we looked around at the existing alternative and we decided that we needed to do our own custom accelerators," says Google's Norman Jouppi, recipient of the 2015 Eckert-Mauchly Award for pioneering contributions to the design and analysis of high-performance processors and memory systems. Jouppi notes Google started using the technology last year to accelerate its StreetView service's reading of street signs, enabling it to process all the text stored in its StreetView image archive in only five days.
STEM Funders Network Expands to 37 U.S. Cities
U.S. News & World Report (05/19/16) Tom Risen
The Science, Technology, Engineering, and Math (STEM) Funders Network on Thursday added 10 new U.S. communities to its nationwide coalition for integrating learning opportunities for millions of students. This brings the group's total Community of Practice community members to 37. The organization's goal is to establish cross-collaborative learning platforms in 100 cities by 2020. A lack of funding for coordination between groups to share best practices by corporate and local donors is a key shortcoming of education programs, according to STEM Next director Ron Ottinger. "If kids don't say they want to work in technology, math, or science by the time they reach eighth grade, it's not going to happen," he says. Teaching Institute for Excellence in STEM president Jan Morrison says the Community of Practice ecosystem seeks to "change the ways kids learn, change the way educators teach, and to improve our communities" by helping them exchange ideas about approaching STEM and computer science. Samueli Foundation executive director Gerald Solomon calls for programs to avoid "donor fatigue" and ensure wise investment in such concepts as how best to attract girls and minorities to STEM-related classes and extracurricular activities. Solomon also says involving local companies in education programs can guarantee diverse interest that will support learning platforms as they evolve.
Largest Study of Online Tracking Proves Google Really Is Watching Us All
Technology Review (05/18/16) Tom Simonite
Researchers at Princeton University say they have conducted the largest study yet on the technology that tracks people's movements around the Web, focusing on the use of tracking code on the Internet's 1 million most popular websites. The five most common tracking tools were all Google-owned, including Google Analytics, whose code was on nearly 70 percent of websites. Although the systems that do the tracking are automated and no human likely looked at the data, consumers might not feel comfortable with the idea of information bouncing around multiple companies and algorithms outside of their control. Research shows how Google's ad-targeting system can use information in ways that might be viewed as discriminatory, such as by targeting men but not women for ads about high-paying jobs. In the study, Princeton professor Arvind Narayanan and graduate student Steven Englehardt surveyed the 1 million websites using OpenWPM software, which automatically visits websites using the Firefox browser and logs any tracking technology it encounters. The researchers found some companies silently send an audio signal to a person's browser, and slight differences in software and hardware can be used to identify a particular computer. However, they note the consolidation of power in Web tracking should make things easier for regulators and citizens to keep tabs on the trackers.
Innovative Trust Model to Help Journalists Verify Social Media Content
CORDIS News (05/18/16)
The European Union-funded REVEAL project is developing solutions to help journalists authenticate useful social media information and distinguish false rumors from facts. REVEAL's "trust model" partly automates the filtering of information on social media using trusted sources, enabling journalists to maintain a list of sources and link new content to authors. The model also is designed to help journalists rapidly gather new eyewitness content, such as identifying relevant images or video within five minutes of publication that is probably not yet verified. The REVEAL team used last November's Paris terror attacks as a case study, and crawled through social media platforms using natural-language processing to identify named entities in English and French and cited URLs. The data was imported to the trust model, which already had a sample list of trusted and untrusted sources. REVEAL enabled the retrieval of all content written by, mentioning, or attributed to a specific source, and then five pictures--two fake, three genuine--posted on the night of the attacks were chosen. The team identified URLs for copies of each posted image that might have been shared instead of the original image URL, and queried its database every 10 minutes during the first hour after each image was published to see how often it was shared. An analysis of eyewitness content showed untrusted sources share images earlier than trusted sources.
Scan Your Doodles to Find the Perfect Matching Photo Online
New Scientist (05/18/16) Aviva Rutkin
Georgia Institute of Technology (Georgia Tech) researchers say they have developed software that can scan hand-drawn sketches and search for a photograph that looks just like them. Georgia Tech professor James Hays says the program is an exciting step toward a search engine based on drawings. The researchers recruited 664 workers on Amazon Mechanical Turk to draw sketches. A photo was randomly selected from a stack of thousands and then shown to a worker for two seconds; each picture fell into one of 125 categories of recognizable objects. The worker then drew what they had seen from memory, with the whole group spending nearly 4,000 hours sketching. The program matched the sketches to the original photographs using two neural networks. One of the networks analyzed the sketches, while the other evaluated the photos. The two algorithms then determined which pairs were most similar. During testing, the program correctly matched the sketch to the photograph 37 percent of the time. Meanwhile, humans completed the same task correctly about 54 percent of the time, which is not an insurmountable goal for artificial intelligence systems, according to Hays. The researchers will present their work in July at the ACM SIGGRAPH 2016 conference in Anaheim, CA.
Shape-Shifting Modular Interactive Device Unveiled
University of Bristol News (05/17/16)
University of Bristol researchers, in collaboration with colleagues from Purdue University, the University of Lancaster, and the University of Sussex, have developed Cubimorph, an interactive mobile device that can change shape on demand. Cubimorph has touchscreens on each of its six module faces, and uses a hinge-mounted turntable mechanism to self-reconfigure in the user's hands. The researchers say the device could be used for a variety of applications, such as a mobile phone that can transform into a console when a user launches a game. The device consists of a chain of cubes, and contributes toward the vision of programmable matter, in which interactive devices change their shape to fit the functionalities required by end users. At the International Conference on Robotics and Automation (ICRA 2016) this week in Stockholm, the researchers are presenting a design rationale demonstrating user requirements to consider when designing homogeneous modular interactive devices. They will show the Cubimorph mechanical design, and three prototypes demonstrating key aspects--turntable hinges, embedded touchscreens, and miniaturization--and an adaptation of the probabilistic roadmap algorithm for the reconfiguration. "We hope our work will create discussion between the human-computer interaction and robotics communities that could be of benefit to one another," says Bristol researcher Anne Roudaut.
Mobile Apps and Games Are Also Energy Thieves
Linkoping University (05/18/16) Monica Westman Svenselius
Linkoping University's Ekhiotz Jon Vergara developed EnergyBox, a tool for measuring how much power mobile devices use due to data communication, for his doctoral thesis on comparing the energy consumption of various apps, computer games, and chat services. He found consumption relies not only on the amount of data transmitted, but also on how it is sent. Vergara says the length and energy efficiency of the "handshake" between systems is a key determinant of the amount of energy consumed. Among the energy-efficient solutions Vergara proposes in this instance is queueing the message for a second, because "if the application can queue what we're writing and then send everything at once, we can save up to 43 percent of the energy." In his tests of mobile games, Vergara found single-player games can run equally well while consuming significantly less energy when an Internet connection is absent. Because there currently are no clear incentives for software developers to cut energy consumption, Vergara has assessed various strategies for distributing the total energy use of a system among the consuming entities as fairly as possible. "In my thesis, I have identified some possible methods and provided guidelines to choose among the alternatives," he notes.
Animal Training Techniques Teach Robots New Tricks
WSU News (05/17/16) Michelle Frederickson
A team led by researchers at Washington State University (WSU) is making it easier for people to give robots instructions to perform tasks such as cleaning the house or cooking. The group has designed a program that lets people teach a virtual robot that looks like a computerized dog. Non-computer programmers trained the robot in WSU's Intelligent Robot Learning Laboratory, teaching the virtual dog by either reinforcing good behavior or punishing incorrect behavior. The researchers say their algorithm enabled it to become more adept at predicting the correct course of action. "As it receives more feedback and becomes more confident in what to do, it speeds up," says WSU doctoral student Bei Peng. The algorithm enabled the virtual dog to understand the meanings underlying implicit feedback. The team is now working with physical robots as well as virtual ones, and also believes the program could help people become more effective animal trainers. The researchers presented their work last week at the Autonomous Agents and MultiAgent Systems (AAMAS) 2016 international conference in Singapore.
New Approach to Sorting Cells
MIT News (05/16/16) Anne Trafton
A team from the Massachusetts Institute of Technology (MIT) has developed a microfluidic device that can sort cells based on their acoustic properties, or how they are affected by sound waves. The acoustic properties rely on cell content and structure, and are independent of size, so the method can be used to separate cell types of similar size. Moreover, the approach does not require altering the cells in any way with chemical labels. The new device consists of a microfluidic channel that vibrates at a very low frequency. As cells flow through the channel, they are pushed to a certain position depending on how they interact with the acoustic forces generated by the vibration. The technique could potentially be used to develop a handheld device that would make it easier and faster to perform a complete blood count test. The technology also could be used to isolate tumor cells from a blood sample, perhaps for monitoring the progression of cancer. "If we make the liquid super-dense in the middle and less dense at the edges, then the particles or cells will move until their acoustic properties match whatever the local environment is," says MIT professor Joel Voldman.
IBM Storage Breakthrough May Speed Up Smartphones, Apps
InformationWeek (05/18/16) Thomas Claburn
IBM on Tuesday announced advances in phase-change memory (PCM) that could be a step toward universal memory technology. The technology stores three bits of data per cell in a 64K-cell array at elevated temperatures and following 1 million endurance cycles, which translates into a potential tripling of PCM chip capacity, according to IBM Research's Haris Pozidis. "Reaching three bits per cell is a significant milestone, because at this density the cost of PCM will be significantly less than [dynamic random-access memory] and closer to flash," he says. Such innovations are unfolding amid industry worries about the ability of storage technology to accommodate expanding volumes of digital content. Stanford University's David S.H. Rosenthal predicts storage will be less free than previously, due to factors such as the slowing of Carnegie Mellon University professor Mark Kryder's growth projections for magnetic disk drives' areal density. Kryder notes chipmakers have been making cheats to keep up with Moore's Law by boosting physical chip size when transistor density fails to deliver desired performance increases. He also says hard disk companies are installing more disks or more disk heads in drive enclosures to compensate for lower areal density gains. "I don't think we're going to be lacking for ways to continue to drive the [storage] cost down," Kryder says.
Stanford Computer Scientists Show Telephone Metadata Can Reveal Surprisingly Sensitive Personal Information
Stanford Report (05/16/16) Bjorn Carey
An analysis of telephone metadata by Stanford University researchers found that data can clue people into a person's private information, while tracking "hops" from a single individual's communications can involve thousands of others. The analysis entailed information collected by the U.S. National Security Agency (NSA), which can obtain metadata such as the numbers dialed and call length without a warrant. The researchers constructed a smartphone application that retrieved the previous call and text message metadata from more than 800 volunteers' smartphone logs. They then combined automated and manual processes to show how many people would be involved in a scan of a single person, and the level of sensitive information that can be inferred about each user. A small sample of users enabled the researchers to surmise, for example, that a person who placed calls to a cardiologist, a local drugstore, and a cardiac arrhythmia-monitoring device hotline likely suffers from cardiac arrhythmia. The researchers also estimated via extrapolation of participant data that the NSA's current authority could permit surveillance on about 25,000 people, if not more, starting from a single phone user metadata scan. The researchers say their study contradicts the government's rationale for tapping metadata on the assumption it does not constitute sensitive information.
New Method of Producing Random Numbers Could Improve Cybersecurity
UT News (05/16/16) Marc Airhart
Researchers at the University of Texas at Austin (UT Austin) say they have developed a new method for producing truly random numbers. The method takes two weakly random sequences of numbers, which harbor predictable patterns, and turns them into one sequence of truly random numbers. Previous versions of randomness extractors either required that one of the two source sequences be truly random or both source sequences be close to truly random, but the new method sidesteps the restrictions that made them less practical. The researchers say their approach creates truly random numbers with less computational effort than other methods, and could make everything from consumer credit card transactions to military communications much more secure. "One common way that encryption is misused is by not using high-quality randomness," says UT Austin professor David Zuckerman. "So in that sense, by making it easier to get high-quality randomness, our methods could improve security." He says the method could be used to encrypt data, make electronic voting more secure, conduct statistically significant polls, and more accurately simulate complex systems such as Earth's climate. The researchers will present a paper about their method next month at the ACM Symposium on Theory of Computing (STOC 2016) conference in Cambridge, MA.
Abstract News © Copyright 2016 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
|
