Association for Computing Machinery
Welcome to the January 6, 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).


The Father of Online Anonymity Has a Plan to End the Crypto War
Wired (01/06/16) Andy Greenberg

David Chaum, who has invented many cryptographic protocols, today will present PrivaTegrity, a new encryption scheme designed to allow fully secret, anonymous communications that no eavesdropper can crack. PrivaTegrity is meant to be both more secure than existing online anonymity systems and also more efficient. In future versions, Chaum and his collaborators at Purdue, Radboud, and Birmingham universities plan to add features such as larger file sharing for photos and video, the ability to follow Twitter-like feeds, and financial transactions. "It's a way to create a separate online reality in which all the various things we now know people like to do online can be done in a lightweight manner under a completely different and new and very attractive privacy and security model," Chaum says. He notes PrivaTegrity also features a carefully controlled backdoor that allows anyone doing something "generally recognized as evil" to have their anonymity and privacy stripped completely. The power to decide who counts as "evil" is too great for any single company or government, so Chaum gave the task to a council system of nine server administrators from nine different countries who would all need to work together to trace criminals within the network and decrypt their communications.

Why STEM's Future Rests in the Hands of 12-Year-Old Girls
TechCrunch (01/05/16) Erin Sawyer

Girls have a lower level of confidence in their math abilities and are more anxious when conducting math-related tasks than boys, according to a recent Organization for Economic Cooperation and Development (OECD) report. Some studies found girls' liking for math and science starts to decline around age 12, and they expect to perform less well in these subjects and blame a lack of ability for their failures. These and other findings indicate girls in the 9-to-12 age bracket tend to opt out of higher-level math and science courses by high school, lessening their chances of pursuing science, technology, engineering, and math (STEM) majors in college, as well as related careers. Three core causes that prevent girls from entering STEM fields must be addressed, according to the OECD report, including reforming their perception that such fields involve little interaction and teamwork. The report says this can be done via positive female role models who highlight STEM employers' hunger for employees with "soft skills." Another challenge to be met involves encouraging a "growth mindset" among girls about their learning potential, as opposed to a "fixed" mindset. The third requirement is to broaden girls' exposure to practical STEM applications via in-class and extracurricular activities.

How Drones May Avoid Collisions by Sharing Knowledge
Technology Review (01/04/16) Signe Brewster

The U.S. National Aeronautics and Space Administration is working with more then 130 research teams to solve how to manage drone traffic. The drone traffic management system, which will be under development for the next several years, will help drones communicate with each other and avoid potential collisions. One of the teams is from the Stanford Intelligent Systems Laboratory, which is led by Mykel Kochenderfer. The Stanford team has developed a quick decision process the traffic management system can use to reroute drones and avoid collisions. The researchers ran more than 1 million simulations for conflict situations for anywhere between two and 10 drones. The drones were given varying levels of information about the other drones in the system and then were tested on their response time and how often they ran into conflict. The Stanford researchers found drones could make the quickest decisions when they were paired with the closest other drone, and the two solely considered each other's behavior. Although decision time increased as more drones entered the simulation, the system was always able to make a decision on rerouting a drone within 50 milliseconds. The researchers found drones feeding their data into a central decision-making system came to the slowest decisions, but they were less likely to encounter conflict, making them safer.

Erica, the 'Most Beautiful and Intelligent' Android, Leads Japan's Robot Revolution
The Guardian (12/31/15) Justin McCurry

Researchers at Osaka and Kyoto universities and the Advanced Telecommunications Research Institute International (ATR) are calling "Erica" their most advanced humanoid. Erica is unable to walk independently, but she possesses improved speech and an ability to understand and respond to questions with an uncanny human-like change in facial expression. The researchers say she should appeal to everyone because images of the facial features of beautiful women were used to develop her face. Team leader Hiroshi Ishiguro, a professor at Osaka University's Intelligent Robotics Laboratory, calls Erica the most beautiful and intelligent android in the world. He says Erica can engage in a flawless chat on a limited number of subjects, and believes free-flowing exchanges could be only a few years away. Ishiguro says developers will need to build robots with a more human-like presence, which will help people overcome their phobias regarding robots. "They will have to be able to guess a human's intentions and desires, then refer to an internal system in order to partly or wholly match those intentions and desires in their response," he says.

The Machine Knows How You Feel: How AI Can Detect Emotion
TechRepublic (01/04/16) Hope Reese

One of the ways artificial intelligence (AI) will improve is in reading emotions, according to Andrew Moore, dean of Carnegie Mellon University's (CMU) School of Computer Science. Computer systems can be more useful if they can understand the emotional state of the person with whom they are communicating, and CMU professor Justine Casselle is studying how children learn through their interactions with computers. Casselle examined children using a computer simulation of another child and found there is a huge improvement in learning outcomes when the student is engaged with a simulated child that reacts realistically to their own emotional state. Meanwhile, University of Pittsburgh researcher Jeff Cohn is studying how people respond to treatments for depression by having a computer monitor facial cues. High-resolution cameras can detect enough skin movement to see what facial muscles are being used at any given time. The system uses a facial-action coding system to map slight facial movements, both deliberate and subconscious, with the emotional state. Finally, CMU professor LP Morency has demonstrated how emotion-understanding robots can elicit better outcomes in terms of getting information that will help with obtaining accurate diagnoses, when compared to traditional robots.

Internet of Things Brings New Era of Weather Forecasting
Computerworld (01/04/16) Patrick Thibodeau

As the Internet of Things (IoT) continues to expand and develop, weather systems will collect data from other vehicles on the road and wirelessly transmit road condition and weather data. These IoT sensors include those that monitor temperature, pressure, moisture, light, and motion. For example, remote-monitoring stations aimed at agriculture, which measure atmospheric and ground conditions, produce data that is combined with government and private sources of weather data to help develop forecasts that update every hour. However, weather researchers now want good metadata along with the sensor data, which will help researchers know which instruments were used to gather the data and their accuracy. There are potential life-saving benefits to using vehicle sensor data to improve weather information and alert systems for drivers. "On average, nearly 6,000 people are killed and over 445,000 people are injured in weather-related crashes each year," notes the U.S. Federal Highway Administration. The data from urban environments, lakes, rivers, streams, and other conditions all influence microclimates, according to the U.S. National Center for Atmospheric Research's William Mahoney. He notes forecasters currently are not using cellphone or vehicle data routinely, "but that's coming." Mahoney predicts over the next few years, research will lead to methods and techniques to take advantage of those data sources.

Computer Model Matches Humans at Predicting How Objects Move
MIT News (01/04/16) Adam Conner-Simons

Researchers at the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory (CSAIL) have developed Galileo, a system that uses a computational model of the human brain to deduce physical properties of an object via a three-dimensional (3D) physics engine. Galileo can anticipate how objects move with accuracy equal to that of humans by training itself on videos while intuiting movement from the physics engine. "From a ramp scenario, for example, Galileo can infer the density of an object and then predict if it can float," says CSAIL postdoctoral researcher Ilker Yildirim. "This is just the first step in imbuing computers with a deeper understanding of dynamic scenes as they unfold." Galileo initially was trained on a series of 150 videos depicting physical events involving objects of various materials, which generated a dataset of objects and their physical properties. The model information then was fed to the Bullet 3D physics engine to physically simulate a given scene forward in time. Deep-learning algorithms enabled the model to educate itself to further improve its predictions. Carnegie Mellon University professor Abhinav Gupta praises CSAIL's breakthrough for solving the problem of computers interacting with the world without training data, "by combining deep-learning convolutional networks with classical [artificial intelligence] ideas like simulation engines."

The Unsung Heroes of Scientific Software
Nature (01/04/16) Dalmeet Singh Chawla

A new website could provide a transparent and meaningful way to track the impact of software built by academics. Launched in November, the free service measures the impact of software by counting citations and informal mentions in research papers, tracking how others reuse code, and gathering download statistics on code packages. The platform is called Depsy, and the name originates from "dependency network," an overarching term for a map of factors that depend on each other. Impactstory, a nonprofit firm based in Vancouver, Canada, developed the technology behind Depsy. Depsy apportions fractional credit to each participant who has contributed to a software program by counting the percentage of code they have contributed or edited. Signatures of each commit are saved in the code, making it possible to track down the originator; but not every edit has the same impact, and Depsy currently cannot differentiate between valuable contributions and trivial ones. The tool may be adapted to attempt this distinction by tracking the influence of individual commits in the future, says Impactstory co-founder Jason Priem. Other websites do some of this as well, but Depsy is unconventional in that it focuses on research software. Depsy's creators hope to add the ability to track software in coding languages other than R and Python, as well as add a social-influence metric.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Privacy-Preserving Inference of Social Relationships From Location Data
CCC Blog (01/04/16) Helen Wright

Researchers from the University of Southern California, Emory University, and Utah State University have conceived of an extensible framework labeled Privacy-Preserving Location Analytics and Computation Environment (Private-PLACE). Private-PLACE is designed to facilitate social relationship inference studies by analyzing individually generated location data. The framework uses an untrusted server and computes several building blocks to enable various social relationship studies, while keeping location data anonymous. When implemented, Private-PLACE promises to enable the acquisition of private location data from individual devices, due to the strong privacy guarantees. The framework also will support a broad spectrum of applications, including Reachability in epidemiology to examine disease proliferation through human contacts, Social-Strength in criminology to identify new or unknown members of a criminal gang or a terrorist cell, and Spatial-Influence in policy to affect local influence in electing a tribal representative. The Private-PLACE use cases employ four privacy-preserving modules, which feature encryption and differential privacy primitives. A paper on Private-PLACE was one of the winners of the Blue Sky Ideas Track Competition at ACM SIGSPATIAL 2015 in Seattle, WA.

Supercomputer Benchmark Gains Adherents
Sandia National Laboratories (12/23/15) Neal Singer

Sandia National Laboratories (SNL) researchers have developed the High-Performance Conjugate Gradients (HPCG) benchmark, a program that ranks supercomputers based on their ability to solve complex problems instead of on raw speed, and which continues to gain traction in the high-performance computing community. "HPCG is designed to complement the traditional High-Performance Linpack (HPL) benchmark used as the official metric for ranking the top 500 systems," says SNL researcher Mike Heroux. The HPCG list contains the same entries as many of the top 50 systems on Linpack's TOP500, which is based on the HPL benchmark, but significantly shuffles HPL rankings, indicating HPCG prioritizes different characteristics. The differences can be attributed to the varying measures provided by HPCG and HPL, which act as bookends on the performance spectrum of a given system. "While HPL tests supercomputer speed in solving relatively straightforward problems, HPCG's more complex criteria test characteristics such as high-performance interconnects, memory systems, and fine-grain cooperative threading that are important to a different and broader set of applications," Heroux says. He also notes "all major vendor computing companies have invested heavily in optimizing our benchmark. All participating system owners have dedicated machine time to make runs. These investments are the strongest confirmation that we have developed something useful."

Two Steps Closer to a Quantum Internet
IEEE Spectrum (12/30/15) Alexander Hellemans

A team led by Anton Zeilinger, a pioneer in the field of quantum mechanics, has taken the process of teleporting photons two important steps further. The University of Vienna physicist and his team realized the first teleportation of photons in 1997. In November, the researchers reported they teleported a photon's usual properties but also its entanglement, doing so over a record distance of 143 kilometers (89 miles), which is significant because it is nearly as far as the boundary of low Earth orbit. At a shorter distance, the team pulled off a similar feat using twisted light, the kind featuring photons having a property called orbital angular momentum. Photonics experts believe this property has the potential to significantly increase the bandwidth of optical telecommunications networks. Teleportation is key to perhaps the most unassailable version of quantum communications. Teleportation of entanglement will be an important component of future secure quantum links with satellites, notes Thomas Scheidl, a member of Zeilinger's research group.

Project Underway to Preserve Holocaust Experience in Virtual Form
University of Huddersfield (12/23/15)

University of Huddersfield researchers are working on the Interact project, which aims to preserve the first-hand accounts of Nazi persecution survivors. The project involves Holocaust survivors providing testimony and responding to questions from future generations. The researchers developed technology called Embodied Conversational Agent (ECA), which projects a three-dimensional (3D) image of the speaker that can interact with the audience by responding to a wide variety of questions. The researchers conducted 3D high-definition stereoscopic video recording sessions with survivors, including their responses to hundreds of questions deemed most likely to be asked. The computer can instantly detect and analyze the spoken question, so the virtual survivor provides the appropriate response. The technology, known as "mixed reality," ensures high levels of realism. "It means that in the future, when all the survivors have passed away, we can keep this experience and help future generations better understand the history," says University of Huddersfield professor Minhua Ma. So far, six Holocaust survivors have taken part in week-long interview sessions, during which they begin by giving personal testimonies and then provide their responses to questions from a pre-determined list.

What Powers Facebook and Google's AI--and How Computers Could Mimic Brains
The Conversation (01/05/16) Thomas Nowotny

Graphics processing units (GPUs) are powering the computer servers used by Google and Facebook to host their artificial intelligence (AI) systems, writes University of Sussex professor Thomas Nowotny. Originally designed as co-processors that operated alongside a computer's main central-processing unit in order to offload demanding computational graphics tasks, Nowotny says GPU hardware has been put to radically different use by manufacturers due to parallelism. GPUs can be turned into specialized processors that can run any parallelized code. GPU hardware is suitable for AI because, for example, in deep learning it enables the neural network training process to be parallelized so it can be accelerated. Greater parallelism is now seen as the only credible route to greater processing speeds, and GPUs have a head start. However, Nowotny says the even more radically different neuromorphic computer, in which the chip is the neural network, may be the future of AI. Proponents believe these systems will help developers scale up neural networks to the size and complexity of the human brain, enabling AI to rival human intelligence.

Abstract News © Copyright 2016 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