Welcome to the April 19, 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).

Frozen case being removed from pool of liquid nitrogen Microsoft's Cool Quantum Computing Plan Embraces Cryogenic Memory
IDG News Service
Agam Shah
April 18, 2017


Microsoft researchers are collaborating with Rambus to design and build prototype computing systems that employ cryogenic memory subsystems, as part of its quantum computing agenda. Cryogenic memory could potentially address the need for memory performance improvements that are proportional to that of quantum systems. Cryogenic memory is envisioned as a replacement for existing memory products, as it offers far more compactness than dynamic random-access memory arrays, for example. In addition, Microsoft and Rambus say cryogenic memory blocks will be faster and more efficient at lower temperatures in data centers, with supercomputers and discrete cryogenic boxes linked using high-speed interconnects. Microsoft's planned quantum computing system is based on a new topology, exotic materials, and an undiscovered particle, which means the system may not become available for years or even decades. Another drawback is the larger amount of energy needed to run refrigerators for quantum computers and cryogenic memories.

Full Article

A bee in a honeycomb Scientists Develop a Novel Algorithm Inspired by Bee Colonies to Help Dismantling Criminal Social Networks
University of Granada (Spain)
April 17, 2017


Researchers at the University of Granada (UGR) in Spain have developed an algorithm, inspired by bee colony behavior, which is designed to counter both physical and virtual threats to social networks. The software automatically detects and recognizes the most dangerous actors or network nodes and the density of the interlinked relationships between them. The UGR team notes the interrelationships between individuals in a colony supports collective behavior enabling them to efficiently respond to challenges, which they seek to apply to artificial intelligence. The researchers simulated bee colony behavior using in silico "bees" mimicking the data exchange and communication processes they share. "In order to find the most effective way of dismantling a network, it is necessary to develop and put into action an optimization process that analyzes a multitude of situations and selects the best option in the shortest time possible," says UGR's Humberto Trujillo Mendoza.

Full Article

Elderly male fell down the stairs Smart Healthcare Uses Stats to Spot a Stumble
KAUST Discovery
April 16, 2017


Researchers at the King Abdullah University of Science and Technology (KAUST) in Saudi Arabia have developed a statistical scheme that detects when senior citizens or others need help after falling. The system relies on combining data from both wearable sensors and video surveillance. Although accelerometers in most wearable sensors use manually-set thresholds to trigger an alert signal, the KAUST researchers used exponentially weighted moving average (EWMA) charts to dynamically monitor acceleration data over time, enabling any unusual changes to a person's movements to be identified as sharp deviations from the typical dataset. The team then integrated EWMA chart-monitoring into a model "smart home" environment containing multiple surveillance cameras to better recognize significant fall events. They note their method uses computer-vision algorithms to subtract backgrounds and imaging artifacts from the video data to focus only on human shapes.

Full Article
Success in Recognizing Digits and Monosyllables With High Accuracy From Brain Activity Measurement
Toyohasi University of Technology
April 18, 2017


Researchers at the Toyohashi University of Technology in Japan have developed technology that can recognize the numbers zero to nine with 90-percent accuracy using electroencephalogram (EEG) readings while a human subject recites the numbers. In addition, the technology can recognize 18 types of Japanese monosyllables from EEG signals with 60-percent accuracy. The researchers aim to develop a brain-computer interface that recognizes unvoiced speech, or speech imagery. They say the technology could enable people who have lost their vocal ability to speak once again. In addition, the technology could provide others with a more natural computer interface. The researchers also want to develop a device that can be more easily operated with fewer electrodes and connected to smartphones within the next five years.

Full Article

A melting glacier breaking off into the water Machine Learning Storms Into Climate Research
The Next Platform
Jeffrey Burt
April 18, 2017


Climate change researchers are aggressively pushing for the incorporation of machine learning within their field, as showcased at the recent Conference on Artificial and Computational Intelligence and Its Applications to the Environmental Sciences. The various projects on display were linked by the common goal of applying machine-learning tools to identify patterns in vast volumes of climate data in order to save lives, time, and money in a rapidly changing environment. For one project, scientists at the Goddard Space Flight Center combined a machine-learning algorithm and crowdsourcing methods to better classify marine low clouds using satellite images from the U.S. National Aeronautics and Space Administration. Goddard's Cynthia Rosenzweig says climate researchers can provide decision-makers with the information they need to make more informed decisions on responses to the effects of climate change. "We need to utilize our science to understand how effective the interventions are--what's working and what's not," she says.

Full Article
Maryland College Wins 2017 National Collegiate Cyber Defense Championship
Fifth Domain
April 17, 2017


A team from the University of Maryland, Baltimore County (UMBC) defeated nine other finalist teams to win the 2017 National Collegiate Cyber Defense Competition (NCCDC). The contest featured the top 10 college and university teams from across the U.S. facing off to protect a network against constant cyberattacks. More than 230 college and university teams participated in regional competitions before the national championship round. NCCDC was the first national cybersecurity competition to test how well college students operate and manage a network infrastructure similar to those found in the commercial sector. At this year's National Championship, the competing teams worked to secure a multi-site retail corporation with 160 employees, point-of-sale systems, inventory systems, and other systems found in a typical retail company. The UMBC team earned the highest score, followed by teams from the University of Tulsa and Brigham Young University.

Full Article
Researchers Working Toward Indoor Location Detection
Rice University
David Ruth
April 17, 2017


Researchers at Rice University are developing a machine-learning interior navigation location detection methodology using the sensors in mobile devices. Six months ago, the team detailed its CaPSuLe indoor mobile positioning system, which depends on machine learning. Both CaPSuLe and the new solution accelerate calculations and lower energy expenditure compared with existing location technologies, but the former relies on image-matching techniques and uploaded data. The original model tapped gyroscope and accelerometer data for indoor-location detection, and the addition of mapping data has improved performance substantially, says Rice's Chen Luo. The researchers also incorporated studies of standard human movements so the algorithm could generate accurate estimations. "The kernel of our current location detection solution may also have wide applications for daily use across a variety of fields, including marketing, healthcare, and pet care, among others," says Rice's Jose Gonzalez Espana.

Full Article
Meet the Man Who Makes Facebook's Computers Think
CNBC
Alex Kantrowitz
April 17, 2017


Yann LeCun leads the Facebook Artificial Intelligence Research (FAIR) unit, which is making progress based on his once-marginalized hypotheses about AI. He envisions the transformation of human-computer interaction via advances led by FAIR, which include improvements to machine vision, listening, and communication. LeCun's theories of neural-networking AI were out of favor for a long time, but their resurgence came primarily thanks to his efforts to demonstrate neural nets could be modified for multiple tasks. Helping this resurgence immeasurably were exponential boosts in computing power and available data. LeCun says FAIR is teaching computers to predict outcomes by feeding AIs related videos, similar to computer-vision improvements enabled by training neural networks on vast numbers of images. LeCun also sees potential for addressing prediction and uncertainty in AI via adversarial training.

Full Article
The Art of Algorithms: How Automation Is Affecting Creativity
VentureBeat
Paul Sawers
April 15, 2016


There is increasing cross-fertilization between art and computers, with machine intelligence and automation likely to disrupt the creative fields. For example, the Prisma application employs deep-learning neural networks to render smartphone photos in an artistic style selected by the user. Automation also is making inroads into the design field, with an artificial intelligence (AI)-based do-it-yourself Web design service from Wix one of many products. Music is another area in which machine intelligence is progressing, with an emphasis on AI-augmented music writing that could blur the line between amateur and veteran composers and musicians. Constantly improving algorithms also are composing readable journalistic articles, with one scientist anticipating a machine will eventually win a Pulitzer while others expect assistive AI tools for human journalists. Language translation could benefit from algorithms, with 100-percent automation used in some cases and human-computer collaboration in others. Ultimately, artistically-inclined computers could be excellent teachers or mentors to human artists.

Full Article
Can a Supercomputing Algorithm Kill Gerrymandering?
TOP500.org
Michael Feldman
April 14, 2017


Researchers at the University of Illinois at Urbana-Champaign have developed a supercomputing algorithm that can determine whether state legislative districts have been unfairly drawn, with the potential of reshaping U.S. electoral politics. The application, when run on the university's Blue Waters supercomputer, can produce 1 billion possible maps using only the criteria required by state law and traditional districting principals. The maps are inherently nonpartisan because no political demographics are used to create them. Algorithm co-developer Wendy Tam Cho says the vast volume of produced maps would provide the court with a statistically relevant dataset from which to infer partisan intent. "If a billion different maps are very different from the map the court is evaluating, then the Supreme Court has some evidence that partisanship was part of the motivation behind the alleged partisan gerrymandering," Cho notes. The algorithm could end up effectively outlawing gerrymandering across the U.S.

Full Article

A giraffe standing in front of a sunset  Deep Learning Tells Giraffes From Gazelles in the Serengeti
New Scientist
Matt Reynolds
April 11, 2017


Researchers at the University of Wyoming used a database of 3.2 million photos taken by hidden camera traps in the Serengeti National Park in Tanzania to train a deep-learning system to distinguish between 48 animal species, including elephants, giraffes, and gazelles. During testing, the system correctly identified the species present in an image 92 percent of the time. The researchers say the system is better at identifying the most common animals in the dataset, and it has greater difficulty identifying rarer species. The team notes the system could be used to classify most of the photos in the database, and it could be further trained on hand-labeled images to improve its performance at recognizing rarer species. The researchers also plan to test whether the system can identify animal behavior in images.

Full Article
Internet Atlas Maps the Physical Internet to Enhance Security
University of Wisconsin-Madison News
Jennifer Smith
April 10, 2017


Researchers at the University of Wisconsin-Madison (UW-Madison) have developed Internet Atlas, which they say is the first-ever detailed worldwide map of the Internet. The map, which was developed using a grant from the U.S. Department of Homeland Security, outlines the Internet's physical infrastructure, including out-of-sight undersea cables, using data that includes publicly available sources such as what Internet service providers publish on their websites. The UW-Madison researchers intend to augment Internet Atlas and share their work so it can be implemented by others to boost network performance and security. "We'll complement the static maps with the ability to actually examine the status of the network in real time," says UW-Madison professor Paul Barford. "We've built certain capabilities that allow exactly that to be done, and one of the important focuses going forward is to enhance that capability, basically putting the maps in motion."

Full Article
Larry Smarr Talks Machine Intelligence at Jackson State
Jackson State Newsroom
Rachel James-Terry
April 12, 2017


Larry Smarr, founding director of the California Institute of Telecommunications and Information Technology (Calit2), is a major advocate of machine intelligence, much of which he says is rooted in pattern recognition. He notes Calit2 uses specialized processors such as hearing aids or voice authentication technology to develop machine learning capable of identifying patterns and gathering this data into a lab that faculty and students can access to conduct experiments to understand how big data can be analyzed more efficiently. Smarr also is a proponent of lifelogging, in which computers help maintain an ongoing study of his own body, based on the question of "can we figure out unintended consequences of some of our activities?" Smarr says lifelogging technology could empower people to recognize early signs of incipient diseases and make appropriate lifestyle changes before they develop. He also is researching the use of three-dimensional imagery to enhance healthcare.

Full Article
MIT Advances in CAD for Manufacturing
 
ACM Online Books and Courses
 

Association for Computing Machinery

2 Penn Plaza, Suite 701
New York, NY 10121-0701
1-800-342-6626
(U.S./Canada)



ACM Media Sales

If you are interested in advertising in ACM TechNews or other ACM publications, please contact ACM Media Sales or (212) 626-0686, or visit ACM Media for more information.

To submit feedback about ACM TechNews, contact: technews@hq.acm.org