ACM TechNews

RSS

Timely Topics for IT Professionals

ACM TechNews Home

Recent News

Computer Eyesight Gets a Lot More Accurate

The New York Times

For the second time in its four-year history, the Large Scale Visual Recognition Challenge saw dramatic improvements in the quality of machine-vision technology. The challenge was launched in 2010 by scientists from Stanford, Princeton, and Columbia universities with the goal of advancing the quality of image-recognition technology. The challenge uses the open source Imagenet database of more than 14 million images, which have been tagged and identified by humans. Visual systems this year were tested in six categories based on their ability to detect objects, locate specific items in an object, and classify those images. The winners included Google, Adobe Systems, the National University of Singapore, and the Chinese Academy of Sciences. Olga Russakovsky, the lead organizer of this year's contest, said the accuracy rate of this year's contestants nearly doubled from 22.5 percent to 43.9 percent, while error rates fell by almost half from 11.7 percent to 6.6 percent. Fei-Fei Li, director of Stanford's Artificial Intelligence Laboratory, called this year's results "historic." The improvement is attributed largely to expanded computing power that enabled the use of an approach known as a convolutional neural network by most of the competitors. The technique has existed since the late 1990s, but sufficient computing power to make use of it was not cost effective until recently.

From "Computer Eyesight Gets a Lot More Accurate"
The New York Times (08/18/14) John Markoff
View Full Article - May Require Free Registration

Vehicle-to-Vehicle Networks Could Save Over 1,000 Lives a Year, U.S. Says

IDG News Service

The U.S. National Highway Traffic Safety Administration (NHTSA) has published a research report on vehicle-to-vehicle (V2V) technology, which would enable cars to automatically exchange information and could save more than 1,000 lives a year. V2V technology would enable cars to do some of the work of driving or even accomplish tasks humans cannot, such as virtually seeing into blind intersections before entering them. V2V would operate over wireless networks using the IEEE 802.11p specification, on a band of spectrum between 5.85 GHz and 5.925 GHz, which would help make the technology function between vehicles from different manufacturers. Left Turn Assist and Intersection Movement Assist applications could potentially prevent up to 592,000 crashes and save more than 1,000 lives a year, the NHTSA report notes. Left Turn Assist would warn drivers not to turn left into the path of an oncoming car, while Intersection Movement Assist would warn them not to enter an intersection when there is a high likelihood of colliding with other vehicles there. V2V technology, which NHTSA says could help improve traffic flow and vehicle fuel economy, also may be one step on the path to self-driving cars.

From "Vehicle-to-Vehicle Networks Could Save Over 1,000 Lives a Year, U.S. Says"
IDG News Service (08/18/14) Stephen Lawson
View Full Article

This Android Shield Could Encrypt Apps So Invisibly You Forget It's There

Wired News

Georgia Institute of Technology researchers have developed Mimesis Aegis (M-Aegis), an encryption app they say can enable encrypted communication using cloud-based apps. Mimesis Aegis, Latin for "mimicry shield," is an overlay that mimics the user interface of other apps almost perfectly and enables users to enter messages normally, but immediately encrypts the messages before sending them to the app in question, where they are sent as a normal message and then decrypted at the other end by another M-Aegis user. Lead researcher Wenke Lee says the goal of the project is to make end-to-end decryption "easy as air" for the average user. The researchers plan to present M-Aegis at this week's USENIX security conference and to release the tool as an Android app this fall, although that version will be limited to only a handful of specific apps. Wenke says the team intends to continue developing M-Aegis to enable it to work with a wide range of photo and audio apps. They also want to automate the process by which it mimics other apps' front-ends, which currently has to be done manually. In addition, M-Aegis currently can only be used for communications between M-Aegis users, it only works on Android devices, and it may be subject to undiscovered security bugs.

From "This Android Shield Could Encrypt Apps So Invisibly You Forget It's There"
Wired News (08/19/14) Andy Greenburg
View Full Article

UAlberta Engineers Take Major Step Toward Photonic Circuits

University of Alberta

University of Alberta researchers are developing nano-optical cables small enough to replace the copper wiring on computer chips. If the researchers are successful, it could result in dramatic increases in computing speed and reduced energy use by electronic devices. "We're already transmitting data from continent to continent using fiber optics, but the killer application is using this inside chips for interconnects--that is the Holy Grail," says Alberta professor Zubin Jacob. "What we've done is come up with a fundamentally new way of confining light to the nano scale." Conventional fiber-optic cables are limited to about one thousandth of a millimeter in diameter, but the University of Alberta-designed cables are 10 times smaller. One of the biggest challenges the researchers have faced is increased temperatures, since metal causes problems beyond a certain point. "If you use metal, a lot of light gets converted to heat. That has been the major stumbling block," the researchers say. "Light gets converted to heat and the information literally burns up--it's lost." The researchers developed a new, non-metallic metamaterial that enables them to "compress" and contain light waves in smaller cables without creating heat, slowing the signal, or losing data.

From "UAlberta Engineers Take Major Step Toward Photonic Circuits"
University of Alberta (08/19/14) Richard Cairney
View Full Article

DARPA Challenges Teams to Predict Virus Spread

Government Computer News

The U.S. Defense Advanced Research Projects Agency (DARPA) has announced a software challenge seeking models that can predict the spread of the Chikungunya virus (CHIKV). The mosquito-borne virus, which causes debilitating illness, is now spreading through the Western Hemisphere. An accurate forecast would help governments and health organizations limit the spread of CHIKV, according to DARPA. Modeling the future spread of infectious diseases is extremely challenging because it is difficult to predict which of the numerous sources of potentially useful, historical data incorporated into a model will be most informative. The data could be more or less predictive based on conditions and regions. "We believe this effort could lead to the creation of tools that work even faster than the speed of an epidemic, giving us the opportunity to act effectively before an infectious disease actually arrives and spreads," says Matthew Hepburn, program manager for the CHIKV Challenge. He notes in addition to public health, a robust and scalable forecasting tool could be useful for emergency response and humanitarian assistance. "The science of forecasting is a work in progress," Hepburn says. "Identifying and acquiring the right data points and figuring out how to link them requires interdisciplinary coordination."

From "DARPA Challenges Teams to Predict Virus Spread"
Government Computer News (08/18/14)
View Full Article

Technology Can Make Lawful Surveillance Both Open and Effective

Technology Review

Yale University professors Bryan Ford and Joan Feigenbaum have developed technology that could enable law enforcement agencies to identify people whose actions justify closer investigation and demonstrate probable cause via an authorized electronic warrant in order to gain access to unencrypted surveillance data or employ secret analysis processes. They say the technology is based on modern cryptography standards, and could enable agencies to find and extract warrant-authorized data about persons of interest, while guarding the secrecy of the investigation and the privacy of innocent users. Ford and Feigenbaum note the system's design ensures that no sensitive data may be decrypted without the use of multiple keys held by independent authorities, such as the law enforcement agency, the authorizing judge, and a legislative oversight body. In addition, they say their approach can target both known and unknown users, as well as unknown associates of known targets. The researchers say this and other cryptographic methods could facilitate the legitimate pursuit of criminals and terrorists while protecting the general public's privacy.

From "Technology Can Make Lawful Surveillance Both Open and Effective"
Technology Review (08/18/14) Bryan Ford; Joan Feigenbaum
View Full Article

New Research Presents an Improved Method to Let Computers Know You Are Human

UAB News

University of Alabama at Birmingham (UAB) researchers are studying the security and usability of a new generation of Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA) services based on basic computer games. The researchers focused on a broad form of game-like CAPTCHAs, known as dynamic cognitive games (DCGs), which challenge the user to perform a game-like cognitive task interacting with a series of dynamic images. The puzzles are designed to be easy for human users to solve but difficult for computer programs. In addition, the researchers say the game-like nature of DCGs could make the process more engaging for the user compared to conventional text-based CAPTCHAs. The researchers investigated the effectiveness of DCGs by creating prototypes to represent a common type of DCG. They then developed a novel, fully automated attack framework to break the DCG challenges. "The attack is based on computer-vision techniques and can automatically solve new game challenges based on knowledge present in a 'dictionary' built from past challenges," says UAB doctoral student Song Gao. The researchers say DCGs appear to be one of the first CAPTCHA schemes that enable the reliable detection of relay attacks.

From "New Research Presents an Improved Method to Let Computers Know You Are Human"
UAB News (08/18/14) Katherine Shonesy
View Full Article

Wireless Sensors and Flying Robots: A Way to Monitor Deteriorating Bridges

Tufts Now

Tufts University researchers are developing wireless sensors and flying robots that could help authorities monitor the conditions of bridges in real time. The smart sensors could be permanently attached to bridge beams and joints to continuously record vibrations and process the recorded signal. Changes in the vibration response could signify damage, says Tufts professor Babak Moaveni. He is collaborating with fellow professor Usman Khan to develop a wireless system that would use autonomous flying robots, or quad-copters, to hover near the sensors and collect data while taking visual images of bridge conditions. The drone-like robots then would transmit data to a central location point for analysis. The research is supported by a $400,000 U.S. National Science Foundation grant, which is needed because there are still significant navigational and communications hurdles to overcome before the system can function as a reliable tool. Moaveni says a major goal of the research is to develop algorithms that can automatically detect damage in a bridge from the changes in its vibration measurements. Once installed, the sensors would supply data about the condition of bridges that cannot be acquired by visual inspection alone and would enable authorities to identify and focus on bridges that require immediate attention.

From "Wireless Sensors and Flying Robots: A Way to Monitor Deteriorating Bridges"
Tufts Now (08/14/14) Alexander Reid
View Full Article

New Algorithm Gives Credit Where Credit Is Due

Northeastern University News

An algorithm developed at Northeastern University's Center for Complex Network Research (CCNR) could help determine how to properly allocate credit for science papers that have multiple authors. Chinese Academy of Sciences professor Hua-Wei Shen developed the algorithm and a Northeastern team used it to build a credit allocation system that focuses on how often a paper is cited with the other papers published by a paper's co-authors, capturing an author's additional contributions to the field. "The idea behind this is that based on an author's previous line of work, people have a perception of where the credit lies," says CCNR director Albert-Laszlo Barabasi. "And the algorithm's goal is simply to extract that perception." The team examined 63 Nobel prize-winning papers related to physics, chemistry, and medicine, and the system found the authors deserving the most credit corresponded to the Nobel laureate in 81 percent of the papers. The algorithm also showed that physicist Tom Kibble, who wrote a research paper on the Higgs boson theory in 1964, should have received the same amount of credit as Nobel prize winners Peter Higgs and Francois Englert.

From "New Algorithm Gives Credit Where Credit Is Due"
Northeastern University News (08/14/14) Joe O'Connell
View Full Article

NASA Selects Proposals to Increase STEM Education at Community and Technical Colleges

NASA News

The U.S. National Aeronautics and Space Administration's (NASA) Office of Education will award 35 grants totaling $17.3 million through the National Space Grant and Fellowship Program to increase student and faculty engagement in science, technology, engineering, and mathematics (STEM) at U.S. community colleges and technical schools. Each award has a two-year performance period and a maximum value of $500,000. The winning proposals focused on ways to attract and retain more students from community and technical colleges in STEM curricula, develop stronger collaborations to increase student access to NASA's STEM education content, and increase the number of students who advance from an associate to a bachelor's degree. For example, the California Space Grant Consortium proposes to enhance STEM preparation at 12 state community colleges and improve opportunities for approximately 300 students to transfer to either the University of California or the California State University system. Meanwhile, the Colorado Space Grant Consortium proposes to add four new community college campuses as affiliates to the consortium. NASA notes its grant program continues the agency's tradition of investing in the U.S. education infrastructure with the goal of developing STEM skills and capabilities.

From "NASA Selects Proposals to Increase STEM Education at Community and Technical Colleges"
NASA News (08/13/14) Ann Marie Trotta
View Full Article

Brain-Computer Interface Project Gives Hope to Disabled

Engineering and Technology Magazine

The recently concluded European Union-funded Tools for Brain-Computer Interaction (TOBI) project could help the severely disabled regain some of their lost functionality. For example, one participant, who suffered a stroke that left him completely paralyzed and unable to speak, controlled a computer with his thoughts and communicated by typing email messages. For three experiments, subjects--patients with severe physical disabilities--were instructed to control objects by thinking about the desired effect, as electrical signals were transmitted from their brains through electrodes attached to a cap into a computer. In the first experiment, participants gained the ability to take full advantage of the Internet and computer technology. In the second experiment, participants took small remote-controlled telepresence robots on virtual walks or sent them to meet with other people. For the third experiment, participants used their own thoughts to control electrodes attached to their limbs and generate movement. The TOBI experiments did not entail invasive surgical procedures such as brain implants on the tested subjects. The researchers say the study participants became an essential element of the research team. "We listened to the feedback of all the patients to correct design mistakes and made any changes right away," says project coordinator Jose del R. Millan.

From "Brain-Computer Interface Project Gives Hope to Disabled"
Engineering and Technology Magazine (08/13/14) Tereza Pultarova
View Full Article - May Require Paid Subscription

Heartbleed Software Flaw Exposes Weaknesses in Hardware Design

IDG News Service

The Heartbleed bug exposed not only a flaw in OpenSSL, but the underlying hardware vulnerability that it exploited to enable attackers to extract data from memory and cache, said Princeton University professor Ruby Lee at the recent Hot Chips conference. Presenting the results of a project funded by the U.S. Department of Homeland Security, Lee said Heartbleed acted as a side-channel attack that was able to exploit the fact that in modern memory chips, memory addresses in the cache are fixed. The design flaw means attackers can reconstruct information such as a public key even if they are only able to capture very small bits of data at a time, because they will know where to look for the rest of the data they want. To counter this, Lee and the Princeton team created a new memory architecture, Newcache, that instead uses dynamic and randomized cache mapping that would make it much harder for attackers to map the cache and extract data from it. Newcache is ready to implement, and Lee says it even offers some slight performance enhancements, while many chip-level security features often cause a drop in performance. However, Lee says it will likely be several years before new features such as Newcache make their way into new chips and systems.

From "Heartbleed Software Flaw Exposes Weaknesses in Hardware Design"
IDG News Service (08/15/14) Agam Shah
View Full Article

Why Teaching Grandmothers to Code Isn't a Crazy Idea

The Washington Post

The same efforts being directed at teaching youngsters the fundamentals of coding and entrepreneurship also should be extended to older workers and retired people, who may have the perspective, experience, and knowledge necessary to craft technological solutions to the world's problems, writes Stanford University fellow Vivek Wadhwa. He notes there is a well-known bias for youthfulness in startups, especially in the tech sector and among venture capitalists that finance many startups. However, these same venture capitalists often generate returns below that of the stock market, bringing into question the validity of this preference for the young. Wadhwa says educational programs promoting entrepreneurship and intensive day- and week-long programs designed to teach children coding skills could be extended to older workers, and initiatives to fund and incubate startups should be more welcoming of those headed and even staffed by older workers. He also suggests a modified version of billionaire Peter Thiel's initiative to pay students to drop out of college to pursue business, which would instead pay workers to quit their jobs to pursue a startup idea. Wadhwa says support for such initiatives is not a zero-sum game, as both older workers and the young can be supported and mentored simultaneously.

From "Why Teaching Grandmothers to Code Isn't a Crazy Idea"
The Washington Post (08/15/14) Vivek Wadhwa
View Full Article - May Require Free Registration