Welcome to the May 31, 2017 edition of ACM TechNews, providing timely information for IT professionals three times a week.
ACM TechNews mobile apps are available for Android phones and tablets (click here) and for iPhones (click here) and iPads (click here).
|
Is China Outsmarting America in AI?
The New York Times
Paul Mozur; John Markoff; Carolyn Zhang
May 27, 2017
Experts think China is catching up to the U.S. in terms of artificial intelligence (AI) technology, and could overtake the West thanks to vast government and private spending while U.S. investment is scaling back. "China seems to think [AI development is] a race and America doesn't," says the Center for Strategic and International Studies' James Lewis. The Association for the Advancement of Artificial Intelligence president Subbarao Kambhampati says whereas the Obama administration wanted to boost basic AI research, the Trump White House is taking the opposite view. Although much of China's AI progress is focused on peaceful applications, the U.S. Department of Defense finds some areas of concentration alarming, such as future weaponry and more advanced facial recognition. Experts warn the continuity of cooperative Chinese-U.S. AI research also remains uncertain. A more positive possibility is China's AI progress will encourage more open sourcing and information sharing, says Tsinghua University professor Wang Shengjin.
|
New Software Key to Early Cancer Detection
CSIRO (Australia)
Sophie Schmidt
May 31, 2017
Researchers at the Commonwealth Scientific and Industrial Research Organization's (CSIRO) Data61 group in Australia and the Shanghai Institute of Applied Physics in China say they are developing software that could significantly improve the detection of the development of new blood vessels, which precedes the growth of cancers. The researchers analyzed 26 high-resolution, three-dimensional, micro-computed tomographic images from 26 mice, depicting their brains and livers at various stages of cancer growth. The team then used the images to develop an algorithm to generate an accurate representation of the vasculature, preserving the length and shape information of the blood vessel and its branches. The researchers note this breakthrough was accomplished using a technique called "end-point constraints." They say the software enables them to measure subtle changes in the proliferation of blood vessels, including the number and length of the blood vessel branches, and to generate significantly clearer skeletons of the vasculature than previously possible.
|
A Network of Crystals for Long-Distance Quantum Communication
University of Geneva
May 29, 2017
Researchers at the University of Geneva (UNIGE) in Switzerland have developed and demonstrated a protocol for long-distance quantum communication using a network of crystals that emit quantum light and store it for arbitrary intervals. UNIGE's Cyril Laplane says the protocol's reliance on single photons entails a significant possibility of particle loss when the photons propagate in conventional communication links, which grows more crucial with distance. For the quantum repeater, "we are using a crystal capable of storing the quantum state of light," says UNIGE's Jean Etesse. "It possesses the advantage of being relatively simple to use with the potential for very long storage times." Etesse says the crystals can absorb light and restore it later, without reading encoded information. In addition, they can produce single photons and store them on demand. The crystal is the source and memory for quantum information, streamlining the protocol for quantum repeaters and establishing the basis for a quantum Internet.
|
Google DeepMind's AlphaGo Is Retiring After Beating the World's Best Human Players
International Business Times
Hyacinth Mascarenhas
May 28, 2017
Google's DeepMind unit on Saturday announced the retirement of its AlphaGo software after it won a three-game match against Chinese Go champion Ke Jie at the Future of Go Summit. With AlphaGo's competitive career concluded, DeepMind's researchers say they will now "throw their energy into the next set of grand challenges." Among the challenges is developing "advanced general algorithms that could one day help scientists as they tackle some of our most complex problems, such as finding new cures for diseases, dramatically reducing energy consumption, or inventing revolutionary new materials," according to the DeepMind team. "If AI systems prove they are able to unearth significant new knowledge and strategies in these domains too, the breakthroughs could be truly remarkable." The researchers also say they will publish one final research paper on AlphaGo and its generalization potential so it can be applied to other problems.
|
Country 2.0: Upgrading Cities With Smart Technologies
Asian Management Insights
Steven Miller
May 22, 2017
Singapore Management University professor Steven Miller recently moderated a panel titled "How Smart Systems Enable More Liveable Cities," in which experts discussed deploying smart technologies for the benefit of cities and their citizens. Google chief Internet evangelist and former ACM president Vint Cerf, who shared the 2004 ACM A.M. Turing Award with Robert E. Kahn, noted integration is a key component of smart systems. He expects five to 10 years to pass before such systems' performance is refined to support infrastructure maintenance, and he also cited the global community's responsibility to ensure smart-system software updates have strong authentication and validation. Meanwhile, Massachusetts Institute of Technology professor and 1992 ACM A.M. Turing Award recipient Butler Lampson pointed to trade-offs between privacy protection and the benefits resulting from wider ranges of data use as people contend with sharing data with artificial intelligence. He also raised the issue of anticipating unexpected scenarios stemming from smart-city upgrades.
|
Here's How to Track the Smartphone Apps That Are Tracking You
FastCompany
Glenn Fleishman
May 30, 2017
Two complementary research projects plan to cooperate further to help smartphone users monitor how their apps are handling their personally identifiable information. A Northeastern University team led by professor Dave Choffnes has developed ReCon, a tool that employs a virtual private network link functioning as a scanning proxy to analyze all data passing between a smartphone and the rest of the Internet by setting up a network profile in iOS or Android. ReCon can examine the contents of the unencrypted links, and it offers a Web-based console enabling users to block or alter the information being transmitted. Meanwhile, researchers at the University of California, Berkeley's International Computer Science Institute are working on the Haystack Project, which is developing an app that captures data at the source, intercepting HTTPS connections and analyzing all traffic between apps and servers. Although the two projects are expected to remain separate, the researchers plan to work together to help each other understand app behavior.
|
Fighting Forgery With Paper Fingerprints
Newcastle University (UK)
May 25, 2017
Researchers at Newcastle University in the U.K. say they can inexpensively validate any paper document's authenticity by taking a photo of it on a standard camera and identifying its "texture" signature by analyzing the paper's translucence patterns using an ordinary light source. Newcastle's Feng Hao says the method can capture the paper's unique random interweaving of wooden particles in only 1.3 seconds, which "can massively reduce the possibility of forgery." Newcastle's Ehsan Toreini says the team developed an algorithm that produces a compact and novel identifier for each sheet of paper, which is then rendered as a quick response code that can be verified efficiently offline by anyone. "Since this identifier is basically representative of that paper texture, any illegal modifications--including copying the contents of the document to another paper sheet--could be identified," Toreini notes. He says the technique involves photographing each sheet in a box, which the algorithm automatically recognizes.
|
How AI Can Keep Accelerating After Moore's Law
Technology Review
Tom Simonite
May 30, 2017
Researchers believe increasingly power-hungry artificial intelligence (AI) systems may be supported by more powerful graphics-processing units (GPUs) as Moore's Law reaches its limit. For example, NVIDIA CEO Jensen Huang says the performance of his company's GPUs has exhibited continued exponential acceleration amid decelerating performance of central-processing units. Similarly, Microsoft NExT engineer Doug Burger notes although conventional software performance "has stopped improving at historical rates," AI performance continues to rapidly accelerate--a trend that is likely to persist. Burger says GPUs are becoming more powerful because they can be engineered for more specialization for graphics or machine-learning tasks. He leads a project to instill more power within AI software using reconfigurable field-programmable gate arrays. Options for meeting longer-term challenges include boosting computer efficiency with chips that do not add accurately, and biologically inspired chip designs such as neuron-like prototype processors from IBM. "We have decades of scaling left in AI," Burger says.
|
Spreading Awareness About HIV Among Homeless Youth Using Artificial Intelligence
CCC Blog
Helen Wright
May 25, 2017
Amulya Yadav at the University of Southern California (USC) has conceived of an initiative to develop artificial intelligence-based algorithms that can tap the social networks of homeless young people to strategically select participants of HIV intervention camps in order to maximize the proliferation of HIV awareness among homeless youth. Yadav says two distinct algorithms have been devised to address this problem for homeless shelters in Los Angeles. The programs employ sequential planning models such as Partially Observable Markov Decision Processes and game theoretic reasoning to choose the "most influential" individuals to attend interventions. Yadav says when used with 173 homeless youth, the algorithms outperformed cutting-edge methods in current use at homeless shelters by more than 185 percent. He also noted 35 percent more people started getting HIV tests more regularly when interventions were deployed using the USC algorithms. "We hope our algorithms can be deployed nationwide (even worldwide)," Yadav says.
|
Faster, More Nimble Drones on the Horizon
MIT News
Jennifer Chu
May 25, 2017
Researchers at the Massachusetts Institute of Technology have developed an algorithm to tune a Dynamic Vision Sensor (DVS) camera to detect only specific changes in brightness that matter for a particular system. The researchers say this can be applied to any linear system directing a robot to move from point A to point B as a response to high-speed visual data. The results could help boost speeds for more complex systems such as drones and other autonomous robots. The system includes a universal control that can translate DVS data for any linear robotic system by identifying the ideal value for an event-threshold value parameter (H), signifying the minimum change in brightness the system can detect. A system with a low H value would be programmed to perceive and interpret changes in luminosity ranging from very small to relatively large, while a high H value would only "see" and react to large variations in brightness.
|
A New Spin on Electronics
University of Utah News
Lisa Potter
May 25, 2017
Researchers at the University of Utah led a study exploring how organic-inorganic perovskites could transform spintronic devices. The team found perovskites permit easily controllable electron spin and have a sufficiently long spin lifetime to transport information. The researchers formed a thin film from the hybrid perovskite methyl-ammonium lead iodine, positioning it before an ultrafast laser. The laser was divided into two beams, and the first one struck the film to set the electron spin in the desired direction, while the second was bent through a mirror array before hitting the perovskite film at longer intervals to quantify the spin lifetime. The team found the perovskite can sustain the desired spin direction for up to a nanosecond, enabling much data to be stored and manipulated during that time. The researchers rotated the spin more than 10 turns by exposing the electron to different magnetic-field strengths, and they say the material could process data faster and boost random-access memory.
|
Shedding Light on How Humans Walk…With Robots
Harvard University
Lindsay Brownell
May 24, 2017
Researchers at the Wyss Institute for Biologically Inspired Engineering at Harvard University have discovered that humans whose lower limbs are fastened to a typical clinical robot only modify their gait if the forces the robot applies threaten their walking ability. The researchers measured how test subjects' gait changed in response to forces applied by a robotic exoskeleton while on a treadmill. They found the walkers adjusted their stride in response to a change in the length, but not the height, of their step. The researchers note this discrepancy can be explained by the central nervous system's reliance on stability when determining how to adjust to a disruption in normal walking. They say this breakthrough advances the scientific community's understanding of the interactions between robots and patients. "Studying how the human body interacts with robots can not only teach us how to build better clinical rehabilitation machines, but also how our own human bodies work," says Harvard professor Donald Ingber.
|
Combination of Features Produces New Android Vulnerability
Georgia Tech News Center
John Toon
May 22, 2017
Researchers at the Georgia Institute of Technology have identified a new vulnerability affecting Android mobile devices that would enable attackers to silently take control of a mobile device, overlaying the graphical interface with false information to hide its malicious activities. A successful hack would require the user to first install a type of malware that could be hidden in a pirated game or other app. The researchers say the vulnerability does not come from a traditional bug, but from the combination of two legitimate permissions powering desirable and commonly used features in popular apps. They say this combination could support a new class of attacks, called "Cloak and Dagger." The two features used in the Cloak and Dagger attack, known as "BIND_ACCESSIBILITY_SERVICE" and "SYSTEM_ALERT_WINDOW," are used in mapping, chat, or password manager apps, which meaning users will have to sacrifice convenience for security in order to prevent the malicious activity.
|
|
