Association for Computing Machinery
Welcome to the August 4, 2014 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).


How Yahoo Research Labs Studies Culture as a Formal Computational Concept
Technology Review (08/01/14)

Yahoo Labs researchers aim to achieve a truly computational understanding of human society by analyzing the links that form on social networks. They employed two data sets from a pair of social networks, with the first comprised of more than 1 million messages sent between 500,000 pairs of users of the aNobii network. The second set consisted of 100,000 anonymized user pairs who commented on each other's photos on Flickr, sending some 2 million messages in total. The messages were analyzed according to the type of information they communicate and are divided into three groups, respectively associated with social status, social support, and knowledge exchange. The researchers then created an algorithm that automatically categorizes the messages sent between individuals based on the content they contain and their similarity to messages of the same type. The last step is evaluation of the algorithm's results by having editors assess a sample of 1,000 randomly chosen messages from each website and label them according to the three categories, with strong agreement between algorithmic and human selections. The study determined that the most frequent aNobii interactions entail status giving, while no domain is predominant on average in Flickr users' communications. The researchers also observed that "status exchange serves to set the foundation for the future relationship, feeding to the interactional background after the tie-formation stage."

Why the Security of USB Is Fundamentally Broken
Wired News (07/31/14) Andy Greenburg

At this week’s Black Hat security conference, SR Labs security researchers Karsten Nohl and Jakob Lell will present a new proof-of-concept malware that exploits a fundamental flaw in the USB format. Called BadUSB, the malware lives in the firmware of a USB device, where it is virtually undetectable and can freely manipulate files, redirect Internet traffic, issue commands as a USB keyboard, and invisibly spread from USB device to computer to USB device. BadUSB exploits the fact that USB firmware does not use code-signing restrictions and a lack of any trust reference USB firmware that potentially infected devices could be compared against, making it almost impossible to detect. The researchers suspect a leaked NSA program to spread malware using USB devices likely worked on similar principles. Such a fundamental flaw raises the question of whether any USB device can be trusted. Nohl says until fundamental changes are made to USB firmware, USB devices should be treated like hypodermic needles: used once and thrown away, never to be shared. Nohl and Lell are presenting their research on BadUSB at the Black Hat conference, but are unsure how much, if any, of the malware they will release publicly. Nohl says he is torn between the need to galvanize manufacturers into making changes to eliminate the vulnerability, and the serious threat the malware could pose in the wild.

PHP Gets Its Own Formal Language Specification
InfoWorld (07/31/14)

A group led by Facebook developers plans to release a formal specification for the PHP scripting language. A draft specification already has been posted on GitHub. "The next major version of PHP--PHP7--is in the works, and in order to ensure full compliance with existing PHP scripts, it's important to know what is expected from the engine," says Facebook's Sara Golemon, who participated in the development of the specification. Facebook developed HHVM, formerly known as HipHop Virtual Machine, with the intention of making PHP run fast. The planned specification will ensure that HHVM is a fully compatible implementation for PHP. "Additionally, with alternative implementations like HHVM coming on the scene, it's important to keep divergence to a minimum, again by understanding what a well-behaved engine looks like," Golemon says. The group will now focus on improving wording, correcting edge cases, and boosting the conformance suite. The language has been around since 1995.

The Curious Evolution of Artificial Life
Technology Review (07/30/14)

Monash University researcher Tim Taylor recently examined the history of artificial life and how it might evolve in the future. The history of Web-based artificial life can be divided into developments taking place before and after the emergence of Web 2.0 in 2005. One of the earliest networked artificial life experiments was based on Tierra, a well-known artificial life system developed in the early 1990s. At about the same time, other researchers launched several other Web-based artificial life projects, including Technosphere, a project in which online users could design creatures, release them into a virtual world, and then receive email updates at key moments in their lives. Second Life emerged at about the same time the Web was transitioning from Web 1.0 to Web 2.0, which enabled more vibrant collaboration and interaction. Taylor's favorite Web-based artificial life project is known as The Wilderness Downtown, which combines artificial life technologies such as flocking with real-time animation superimposed on Google Street View images of any address the user enters.

Social Therapeutic and Robotic Systems (STARS) Lab: A Computing Research in Action Showcase
CCC Blog (07/31/14) Ann Drobnis

Mississippi State University students participating in the Social Therapeutic and Robotic Systems (STARS) Lab are exploring human-robot interaction via various projects. For example, STARS researchers are working with the Starkville (MS) Police Department SWAT Team to probe the uses of a remotely-operated robot and a robot that has supervised autonomy capabilities in coordination with a SWAT team. The robot features that are most useful to a SWAT team are being examined, and the effort is expected to save lives by helping law enforcement officers manage dangerous situations more safely. Meanwhile, the U.S. National Science Foundation (NSF)-funded Therabot team is developing a robotic therapy support system. Their aim is to provide a robotic dog to people undergoing therapy, especially those with Post-Traumatic Stress Disorder. Patients will use the robot to perform therapy exercises at home, improving the therapy experience and supporting less frequent supervised therapy sessions. NSF also is funding START's Interview team, which is investigating the use of robots as intermediaries for gathering sensitive information from kids. This research examines whether there will be distinctions in the quantity and validity of information based on whether the information is acquired by a human interviewer, a robot interviewer, or a paper survey.

Robot 'Learns to Keep Going With Broken Leg'
BBC News (07/30/14)

When a human or animal breaks or injures a leg, they are quickly able to find alternative ways to continue moving, but ambulatory robots lack such a capability. Roboticists Antoine Cully and Jean-Baptiste Mouret of the Sorbonne in Paris, and Jeff Clune of the University of Wyoming are seeking to correct this deficiency. The scientists used a trial-and-error methodology modeled on the behaviors of injured animals to help a six-legged robot relearn how to walk again once one or more of its legs were damaged. The method enables the robot to learn which leg is damaged and then, through trial and error, determine the best alternative method of locomotion. Fumiya Iida of the University of Cambridge's Machine Intelligence Laboratory says the ability to continue walking on damaged limbs addresses part of the larger problem of creating "robots that can adapt to uncertain and unstructured environments." Iida says the technology has applications in everything from space exploration to the military to disaster recovery and response.

Robots Helped Inspire Deep Learning and Might Become Its Killer App (07/29/14) Derrick Harris

As researchers use more and better data to train their artificial intelligence (AI) models and generate algorithms, the smarter their robots become. Deep learning is a focus of AI research because the advantage of self-teaching systems is the removal of a great deal of manual labor. Baidu researcher Andrew Ng says deep learning is the optimal technique for ingesting and analyzing large data volumes, and much of his research at Stanford University has involved the application of machine learning to robots so they can walk, fly, and see better. Ng says deep learning is very adept at learning features from labeled datasets, but it also is becoming proficient in unsupervised learning, in which systems learn concepts as they process large amounts of unlabeled data. Such capabilities could be helpful as we attempt to build machines that can better perceive their surroundings. Ng says as the nexus of deep learning gravitates toward unsupervised learning, its usefulness to roboticists is likely to expand. "For a lot of applications, we're starting to run out of labeled data," he notes. For example, a robot trained to recognize 50,000 coffee mugs is a minor achievement compared to the need to improve its accuracy, which requires scaling the training datasets to millions, Ng says.

Computational Biologists From Saarbrucken Simplify Diagnosis for Hereditary Diseases
Saarland University (07/29/14)

A program developed by bioinformatics experts from Saarbrucken can help in the rapid diagnosis of hereditary disorders by comparing different patterns of genetic diseases from an extensive online database and gauging them by their probability of occurrence. The Phenomizer app uses the Human Phenotype Ontology database developed at the Charité clinical center in Berlin, which lists more than 10,000 disease traits structurally and assigns them to 7,500 diseases, notes Max Planck Institute for Informatics researcher Marcel Schulz. The app scans, compares, and weights the data related to symptoms supplied by users, and then assigns the characteristics to certain diseases. A list of the most likely results is generated in a matter of seconds. "Doctors no longer have to research in databases or books for several hours," Schulz says. "The list supports them in detecting the disease more quickly." The Phenomizer app was recently released online as an Android version for smartphones and tablets. "We developed the app together with six different computer scientists from Saarbrucken," Schulz says. The app was created by Saarland University students within the context of a software engineering course.

Study: Software Developers Starting to Write for Internet of Things
The VAR Guy (07/29/14) DH Kass

About 40 percent of 1,400 software developers worldwide are writing applications for Internet of Things (IoT)-connected devices or intend to do so within the next six months, according to an Evans Data study. The study found the Asia-Pacific region leading the trend, with 20 percent of its developers currently working on IoT projects and another 30 percent expecting to begin initiatives shortly. Only 16 percent of developers in North America currently are involved with IoT projects, according to the study. "The needed technologies are now converging with cloud, big data, embedded stems, real-time event processing, even cognitive computing combining to change the face of the technological landscape we live in, and developers are leading the way," says Evans CEO Janel Garvin. The study also found that 31 percent of developers associate the IoT with cloud computing, followed by real-time event processing at 26 percent, big data at 17 percent, and machine-to-machine at 15 percent. Major corporations increasingly are focusing on the IoT through partnerships such as the Industrial Internet Consortium (IIC) and the AllSeen Alliance. The IIC consists of about 60 companies and organizations aiming to collaborate with one another, academia, and government to tackle issues related to the IoT.

A Smart Wristband for Nocturnal Cyclists
Swiss Federal Institute of Technology in Lausanne (07/29/14) Laure-Anne Pessina

The "Intelligent Blinker," a smart illuminating wristband developed by five doctoral students at the Swiss Federal Institute of Technology in Lausanne, could make it safer for cyclists to ride at night in big cities. The bracelet is designed to flash when the rider reaches out to indicate a lane change. The device is equipped with a light-emitting diode (LED) and other electronics, including an accelerometer and a magnetometer, a type of compass that can detect the position of the user's arm. When the rider reaches out laterally, the sensors send data to the microcontroller, which directs the LED to engage. "According to the habits of the rider, its possible to adjust the angle at which the LED starts flashing," says student team member Pietro Buccella. Small solar panels have been connected to a battery to enable the device to run on solar energy, but it also includes a USB port for recharging with a computer. The team now wants to reduce the device's size and energy consumption and add more sensors.

Can Computers Stop Suicides?
SBS World News (Australia) (07/29/14) Anne Lin

Richard Harvey, a clinical psychiatrist from Deakin University in Victoria, Australia, is collaborating with computer scientist Svetha Venkatesh to develop a computer system that automatically analyzes hospital records to identify patients that are likely to be a suicide risk. Harvey says the goal of the system is to more effectively shape intervention efforts when patients present in a suicidal state. The system examines data gathered by clinicians during inpatient admissions and emergency department visits to find patterns of traits and behaviors that signal more serious suicidal behavior, and quickly and efficiently present that information to clinicians so they can act on it. "It might take the clinician several hours to go through thousands of pages of medical records and summarize it all," Harvey says. "So what the system does, it very quickly picks out and presents visually a map of previous presentations and what's happened to this person." The system already is being piloted in several Melbourne hospitals and the researchers say so far it is two-thirds better than clinicians alone at predicting the likelihood of a serious event.

New Protection Scheme Makes Weak Passwords Virtually Uncrackable
Security Week (07/29/14) Brian Prince

New York University Polytechnic (NYU-Poly) researchers have developed PolyPasswordHasher, an open source password protection scheme that could help organizations better protect passwords. Most passwords are stored in databases using a salted hash, a one-way encryption technique that offers protection in the event a database is hacked. However, if hackers can get privileged access to a running system, they can intercept an administrator's password information before that protection is in place. PolyPasswordHasher never stores password information directly in the database. Instead, the information is used to encode a cryptographic "store" that cannot be validated unless a certain number of passwords are entered. "PolyPasswordHasher divides secret information--in this case, password hashes--into shares, and just like a puzzle that is meaningless unless the pieces are assembled, no individual password can be validated unless a certain number of them are known and entered," says NYU-Poly professor Justin Cappos. In the event an attacker was able to enter the system, all remaining password data would remain under the same protections offered by conventional salted hashing schemes. "Even if the password file and all other data on disk is obtained by a malicious party, the attacker cannot crack any individual password without simultaneously guessing a large number of them correctly," Cappos says.

Abstract News © Copyright 2014 INFORMATION, INC.
Powered by Information, Inc.

To submit feedback about ACM TechNews, contact:
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe