Association for Computing Machinery
Welcome to the October 23, 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).


Sorry, Einstein. Quantum Study Suggests 'Spooky Action' Is Real.
The New York Times (10/22/15) John Markoff

Delft University of Technology scientists report validating quantum theory's fundamental claim of the phenomenon of spooky action, in which objects separated by great distance can instantaneously affect each other. Their experiment, described as a "loophole-free Bell test," eliminates all possible hidden variables by entangling a pair of diamond-entrapped electrons 1.3 kilometers apart and then sharing information between them, using detectors on opposite sides of the Delft campus to ensure no conventional information exchange is possible. Pulses of microwave and laser energy are then applied to entangle the electrons and measure their spin. Massachusetts Institute of Technology physicist David Kaiser says only two out of three major quantum loopholes have been closed by the Delft experiment. He notes the electronic system the researchers used to add randomness to their measurement may actually be subtly predetermined, meaning the outcome also might still be predetermined. The U.S. National Science Foundation is funding work by Kaiser and others to close the final loophole via an experiment to measure light from distant objects on different sides of the galaxy. Such work is seen as a step toward a "quantum Internet" composed of entangled particles that offers absolute security.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Faster Optimization
MIT News (10/23/15) Larry Hardesty

Former and current Massachusetts Institute of Technology (MIT) students won a best student paper award at this week's IEEE Symposium on Foundations of Computer Science (FOCS) for a general-purpose "cutting-plane" algorithm that addresses optimization problems. Cutting-plane techniques converge on the optimal values of a mathematical function by repeatedly cutting out regions of a much bigger range of possibilities. The researchers say the algorithm not only upgrades the running time of its most efficient precursor, but also offers a new method for applying the algorithm to specific problems that delivers order-of-magnitude efficiency gains. "If for many problems, you have one algorithm, then, in practice, we can try to optimize over one algorithm instead of many algorithms, and we may have a better chance to get faster algorithms for many problems," says MIT graduate student Yin-Tat Lee. Algorithm running times are typically measured in the number of required operations, relative to the number of elements being manipulated. Cutting-plane methods make the number of elements the number of variables in the cost function. The researchers say they applied cutting-plane techniques to problems such as submodular minimization, submodular flow, matroid intersection, and semidefinite programming. In many instances, they reported drastic efficiency improvements, from running times that scale with the fifth or sixth power of the number of variables down to the second or third power.

White House National Strategic Computing Initiative Workshop
CCC Blog (10/22/15) Helen Wright

The White House this week held a workshop for its National Strategic Computing Initiative (NSCI). At the meeting, Tom Kalil, deputy director for policy for the White House Office of Science and Technology Policy, described a new nanotechnology-inspired Grand Challenge, and asked the audience about what actions could be taken to better carry out the mission of the NSCI. The three main themes of the workshop were the convergence of data-intensive and numerically intensive computing, potential hardware technology for future high-performance computing (HPC) systems, and improving the productivity of HPC application development and deployment. The workshop's two keynote speakers were the Semiconductor Research Corporation's Thomas Theis and Computing Community Consortium (CCC) council member Kathy Yelick from Lawrence Livermore National Laboratory and the University of California, Berkeley. Theis discussed the importance of investing in exploratory research and stressed the need for more research into post-complementary metal-oxide devices. Yelick gave a presentation titled "More Data, More Science, and Moore's Law." She said increasing amounts of data are poised to change science and noted the NSCI will help to enable this change. Yelick said the change will require capable exascale systems, computer science, statistical machine learning, new mathematical models, and algorithms.

These Researchers Have Discovered the Perfect Password That's Also Easy to Remember
The Washington Post (10/22/15) Ana Swanson

University of Southern California researchers Marjan Ghazvininejad and Kevin Knight say they have developed an approach to creating passwords that are very difficult to solve while also being easy to remember. The passwords take the form of randomly generated poems, inspired by a cartoon that showed a password composed of four random words is much easier for people to recall and more secure. The researchers produce their poems by assigning a distinct code to each word in a 327,868-word dictionary. Software then creates a long random number, breaks it into fragments, and translates the fragments into two short phrases. The program guarantees the two lines end in rhyming words, with the entire phrase rendered in iambic pentameter. Knight calculates cracking these passwords would take approximately 5 million years at current speeds, and he and Ghazvininejad have set up a demonstration site for their online poem generator. People who want their own poetic passwords can send their emails to another address, and the researchers' program will send them a secure password that is then deleted from their server.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Self-Driving Car Makes 1,500-Mile Mexican Road Trip
USA Today (10/21/15) Marco della Cava

University of Nevada-Reno roboticist Raul Rojas recently traveled 1,500 miles from the U.S. border at Nogales to Mexico City in a self-driving vehicle. The Autonomos car utilized onboard computer processing data from seven laser scanners, nine video cameras, seven radars, and a global-positioning system (GPS) unit. "This is a new challenge, a next step to learn and develop systems, to learn ways to solve new problems for driverless cars," Rojas says. "Most of the trip was highway, but there are many different issues such as construction sites, urban areas in between, potholes, and so on." Prior to the trip, the research team digitally plotted out the full 4,000-mile route from Reno to Mexico City, collating GPS data and incorporating speed limits and other variables into the software. Rojas' experiment seeks to assess technology that enables motorists to let the car assume control over certain types of driving situations. He says evaluating driverless cars under real-world conditions is essential to the creation of machines that will behave appropriately no matter what the road or environmental situations is. "One important aspect to be considered is predicting the behavior of other drivers and pedestrians," Rojas notes. "This is especially relevant in cities."

Researchers Warn Computer Clocks Can Be Easily Scrambled
IDG News Service (10/21/15) Jeremy Kirk

The Network Time Protocol (NTP) has exploitable flaws that could undermine encrypted messages, according to Boston University researchers. They cite NTP's rate-limiting mechanism, which can stop a computer from repeatedly checking the time in the event of a technical hitch. The researchers discovered the possibility of hackers spoofing such a packet so it appears to originate from a system in trouble when it actually is not. "We discovered the...vulnerability by just reading the specifications of the [NTP] protocol," notes Boston University professor Sharon Goldberg. The researchers say all a hacker would need to conduct the spoofing attack is one computer that finds NTP clients using network scanners such as nmap and zmap. Goldberg says the attack is partly enabled by the fact that most NTP servers talk to clients without encrypting their communications, due to the lack of a key exchange protocol. Sinister implications of a computer clock being rolled back include accepting an expired SSL/TLS certificate for which the hacker has the encryption key. Other flaws the researchers exposed include one that could allow a denial-of-service attack, and another that permits attackers to shift a computer's clock backwards or forwards on reboot.

Quality Boost for User-Generated Sound
University of Salford Manchester (10/22/15) Gareth Hollyman

New algorithms could help people better understand sound quality on phones, video recorders, and dictaphones, according to researchers at the University of Salford Manchester. A team led by professor Trevor Cox has developed algorithms that can help people control sound quality. The algorithms are capable of automatically assessing the relative impact of sound errors such as microphone handling noise, distortion, wind noise, and a range of other conditions. An app for assessing wind noise is using the algorithms to alert users when there is significant risk that sound will be affected. "We're used to having visual processing improving our photos, such as the camera that spots faces and changes exposure, but we have not had the same tools to do the audio equivalent," Cox says. Salford's three-year Good Recording project is a response to growing demand from consumers and broadcasters who often use amateur footage, which is compromised by sound quality. The researchers say the project also should benefit broadcasters that use amateur footage and need to quickly assess quality.

Introducing MARTY, Stanford's Self-Driving, Electric, Drifting DeLorean
Stanford Report (10/20/15) Bjorn Carey

A team of Stanford University engineers led by professor Chris Gerdes have built an autonomous, drifting DeLorean powered by electricity to research the physical limits of self-driving systems. The Multiple Actuator Research Test bed for Yaw control (MARTY) embodies the Dynamic Design Lab's mission to determine how to leverage all of a car's capabilities to create autonomous driving systems that will control the vehicle more safely in all situations. "We want to design automated vehicles that can take any action necessary to avoid an accident," Gerdes says. "The laws of physics will limit what the car can do, but we think the software should be capable of any possible maneuver within those limits." Stanford graduate student Jonathan Goh says the car will eventually be taught to race around a track using a drifting method to negotiate tight turns around obstacles when needed. MARTY already can self-lock into a continuous circular doughnut at a large drift angle. The car is a joint project between Gerdes' lab, the Revs Program at Stanford, and Renovo Motors. Renovo made available a new platform that delivers 4,000 pound-feet from on-motor gearboxes to the rear wheels in a fraction of a second, enabling precise control of the forces required to drift.

Building a Better Network for Connected Cars
Government Computer News (10/20/15) Patrick Marshall

Connected-car networks face a huge, real-world obstacle in the limited number of vehicles that have the necessary equipment. An integrated network developed by Clemson University researchers could serve as a bridge until all vehicles have dedicated short-range communications (DSRC) devices. The nonprofit tech accelerator U.S. Ignite has provided Clemson professor James Martin with a two-year, $600,000-grant to enhance DSRC, an ad hoc wireless network, with Wi-Fi and LTE cellular services. Martin's team will develop middleware to integrate the network standards and develop apps to run on it, including one for incident detection and another to provide congestion warnings. The researchers will deploy hardware along 10 to 20 miles of South Carolina roadsides. They plan to install devices in South Carolina Department of Transportation vehicles that regularly traverse the trial zone and recruit volunteer drivers. An app also will be developed to interact with the CANBUS microprocessor interface, which has been embedded in all cars manufactured since 2008.

Researchers Prove Connected Cars Can Be Tracked
IEEE Spectrum (10/21/15) Mark Harris

Researchers in the Netherlands have demonstrated that vehicle-to-vehicle (V2V) and vehicle to infrastructure (V2I) communications, together known as V2X, can be used to track individual vehicles. V2X involves vehicles broadcasting speed and position data over part of the Wi-Fi spectrum; other vehicles can detect this data and use it to avoid collisions. Although the data broadcast by V2X systems does not contain personally identifiable information, it is digitally signed. The researchers designed their attack to track these digital signatures. They equipped a campus security vehicle with a V2X system and set up sniffing stations at two intersections at the University of Twente. The V2X system broadcast more than 2.7 million messages over 16 days of normal operation and the sniffers were able to detect just 40,000 of the messages. However, the researchers were still able to use these messages to place the vehicle in one of two areas of the campus with 78-percent accuracy and were able to locate the vehicle 40 percent of the time. The researchers note even taking steps to randomize the digital signatures used by the V2X systems only increases the number of sniffer stations an attacker would need to deploy. In addition, one of the researchers notes attackers likely could inexpensively build their own sniffers from off-the-shelf parts.

Researchers Aim to Make Privacy Second Nature for Software Developers
New York University (10/20/15)

A New York University researcher and colleagues are working to make user privacy an integral part of the software development process. Professor Sameer Patil is developing "privacy ideation cards" as a way to educate software developers on user privacy regulatory requirements. Patil says the idea is to make U.S. data-protection laws and regulations understandable to software developers and students so they can take them into account at every step of the development process. Privacy matters are often treated as an afterthought, according to Patil, who has been awarded a $175,000 Early-concept Grant for Exploratory Research by the U.S. National Science Foundation. Patil suggests the language the government uses to regulate privacy in technology is not accessible to most software professionals. Moreover, he says developers often lack formal training in the sociotechnical aspects of privacy. Patil's team also plans to promote the "Privacy by Design" approach, which holds that privacy must be the default mode of operation for organizations. The privacy ideation cards, which Patil thinks should be made freely available online, will enable the design, development, and deployment of systems that take into account relevant privacy laws and regulations at every stage of the system-building process.

Yale Quantum Institute to Launch Oct. 23
Yale News (10/19/15) Jim Shelton

Yale University today will open its Quantum Institute, a state-of-the-art research hub aimed at revolutionizing the way information is stored, processed, and safeguarded. "Even in an era of technological leaps, the possibilities for quantum science inspire awe," says Yale president Peter Salovey. The institute will unite more than 120 Yale researchers, whose work already has produced major advances in the study of quantum information science, increasing the understanding of the quantum world, and making Yale a leader in the field, according to Yale Quantum Institute director Robert Schoelkopf. "In order to do cutting-edge quantum information science, you need new kinds of collaboration among engineers, physicists, computer scientists, materials scientists, and a host of other disciplines," Schoelkopf says. Yale has a long history of success in quantum information science, including the development and understanding of quantum networks that use light particles; the demonstration of new, topological states of matter; and the invention of a novel, "speckle-free" laser for medical imaging. "A wonderful feature of this institute is the opportunity to welcome a mix of physicists, engineers, and mathematicians from all over the world for an extended period of time," says Yale professor Michel Devoret.

Settling the Controversy Over Photo of Lee Harvey Oswald
Dartmouth Now (10/19/15) John Cramer

Dartmouth College researchers have used three-dimensional (3D) modeling to confirm the authenticity of the backyard photo of Lee Harvey Oswald holding the same type of rifle used to assassinate President John F. Kennedy. In particular, a team led by professor Hany Farid addressed claims Oswald's pose was physically implausible, as it appears as if he is standing off balance. The team conducted a 3D stability analysis, which involved building a physiologically plausible 3D model of Oswald, posing it to match his appearance in the photo, and adding appropriate mass to each part of the model. The analysis revealed the pose is stable. Moreover, it shows the lighting and shadows in the photo are physically plausible and the length of the rifle is consistent with the one used to kill the president. "Our analysis refutes purported evidence of manipulation in the Oswald photo, but more generally we believe that the type of detailed 3D modeling performed here can be a powerful forensic tool in reasoning about the physical plausibility of an image," Farid says. "With a simple adjustment to the height and weight, the 3D human model that we created can be used to forensically analyze the pose, stability, and shadows in any image of people."

Abstract News © Copyright 2015 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