Welcome to the October 13, 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).
To view "Headlines At A Glance," hit the link labeled "Click here to view this online" found at the top of the page in the html version.
The online version now has a button at the top labeled "Show Headlines."
|
|
|
Alibaba Aims to 'Master the Laws' of AI and Put Virtual Helpers Everywhere
Technology Review
Yiting Sun
October 12, 2017
Alibaba executive chairman Jack Ma on Wednesday announced the establishment of the Alibaba Discovery, Adventure, Momentum, and Outlook (DAMO) Academy in China to research artificial intelligence (AI), quantum computing, and other emerging technologies. In an interview, DAMO Academy director Jeff Zhang says the academy wants to draw top talent in various fields, with the overall goal of forming "a deeper connection with our core business." Zhang says developing a series of solid theories of deep learning is currently the most critical issue in AI, and the academy will focus on "mastering the laws" of AI. Zhang notes much of Alibaba's concentration now involves creating "a perfect match between consumers and manufacturers, and this perfect match relies on data." Zhang says other research areas for the academy will include converting the voice-based AI technology that powers Alibaba's TMall Genie smart speaker into a standardized system that can be incorporated into other devices, such as children's toys.
*May Require Free Registration
|
Securing Encryption Data Against the Threat of Future Quantum Computers
CORDIS News
October 11, 2017
A European Union-funded project seeks to lay the groundwork for quantum-attack-resistant encryption, and the most obvious approach is to focus on new encryption protocols that avoid the use of current techniques that could be cracked in the future. The Post-Quantum CRYPTOgraphy for long-term security (PQCRYPTO) team has outlined a dual strategy that includes not only the development of new encryption techniques, but also mathematical operations which quantum computers cannot perform efficiently. The researchers emphasize several limitations for the techniques that will need to be created, and say these should be inexpensive to apply, without excessively taxing computing systems. The avoidance of using keys of inordinate length also is recommended by the PQCRYPTO team. The researchers note one critical aspect for the production of post-quantum encryption systems will be an early focus on standardization, since all parties will be required to use the same cryptographic systems.
|
No Dark Side to Using LED Lights to Supplement Wi-Fi, Research Reveals
Engineering and Physical Sciences Research Council
October 11, 2017
Researchers at the University of Edinburgh in the U.K. have found energy-saving light-emitting diodes (LEDs) could help meet the demand for wireless communications without affecting their quality of light or environmental benefits. The researchers found transmitting digital data via LEDs while they are being used to generate light does not affect the brightness of the light or its color, or make the LED consume more energy. The team focused on LEDs that produce "warm white" and "cool white" light, and examined two different data transmission techniques--on-off keying and continuous signaling. The researchers say neither technique was found to significantly reduce the bulbs' brightness or their life expectancy, or to cause any significant change in the color of the light. "Plugging a key knowledge gap, our results are very encouraging for the future of light-based communications that could help realize the full economic and social potential of a wireless future," says the University of Edinburgh's Wasiu Popoola.
|
NSF Announces Awards to Support Fundamental Research to Advance Local Cities and Communities
National Science Foundation
Kim L. Silverman
October 12, 2017
The U.S. National Science Foundation's (NSF) Smart & Connected Communities (S&CC) program has announced its first round of awards totaling about $19.5 million, supporting 38 projects involving 34 institutions across the country. "The collaborative research undertaken by these groups will address challenges faced by our cities and communities, helping to transform communities and improve people's lives," says NSF's Jim Kurose. Many S&CC projects will focus on building capacity for long-term research innovation. NSF says the awards will support planning and coordinating activities to grow interdisciplinary and cross-sector teams that can facilitate new lines of research with meaningful community engagement. Some of the awards seek to pursue visionary and integrative research agendas by engaging with local community stakeholders to advance the understanding, development, and implementation of S&CC solutions from technological and social perspectives. The projects are being led by teams at Purdue University, the University of Michigan, and the University of Minnesota, among others.
|
The Female Code Breakers Who Helped Defeat the Nazis
Politico
Liza Mundy
October 10, 2017
More than 10,000 female "cryptoanalysts" were enlisted by the U.S. Army and Navy to help crack Nazi codes and ensure the Allies' victory in World War II, but until now they have been mostly overlooked by history. Women's entry into the emergent field of what is now called information security was eased somewhat by the nascent state of the discipline at the time, but they were sworn to secrecy. Women also were considered better suited than men for code-breaking because of the tedious and repetitive nature of the task, which was a reflection of the attitudes and sexism of the times. The female code breakers also laid the foundations of modern cybersecurity, and they pioneered work that would eventually give birth to today's computing industry. Their contributions are now starting to come to light thanks to several years' of scholarly labor in archives, declassification requests, and interviews with surviving participants.
|
Driverless Cars Learn From Humans in Greenwich Project
The Engineer (United Kingdom)
Jon Excell
October 11, 2017
Tests that could expedite the development of safer driverless vehicles are underway in the London borough of Greenwich in the U.K. under the government-funded Move_UK project. The borough is operating five conventionally driven but heavily equipped Land Rovers that have completed more than 30,000 miles of driving on public roads around Greenwich. The cars are gathering data on drivers' reactions to different events, and driver-assistance systems enable engineers to perform real-time comparisons of human drivers and autonomous systems' behavior and decision-making. The Move_UK project is focusing on specific types of data and employs connectivity technology to transfer data in real time via either Wi-Fi or 3G to a cloud-based system run by transport consultancy and research service the Transport Research Laboratory. Bosch engineer Simon Morley says the system is programmed to be activated by particular events, such as harsh braking by the driver, or situations in which the on-board system thinks there should be braking but the driver has opted not to do so.
|
Facebook Thinks the Most Useful Digital Assistant Is the One That Can Read Minds
Quartz
W. Harry Fortuna
October 9, 2017
Cheryl Dugan, head of Facebook's Building 8 moonshot division, says the company wants to use existing brain-computer interface (BCI) technology to create a product that will enable people to type with their minds. Building 8 researcher Mark Chevillet notes Facebook set out on this project because users are hesitant to ask personal digital assistants questions in public. Facebook wants to develop a digital assistant that can privately listen to the user's thoughts. Chevillet highlights, as a proof of concept, the work of Christian Herff at the University of Bremen in Germany showing limited but promising results when it comes to translating thoughts to typing. In Herff's study, subjects were limited to only 10 words, and were told to speak them out loud. The computer was able to read and translate the brain signals accurately about 75 percent of the time. However, Facebook's goal is to achieve 100 words per minute, relaying words silently directly from a user's brain.
|
New Theory Cracks Open the Black Box of Deep Neural Networks
Wired
Natalie Wolchover
October 8, 2017
Researchers at the Hebrew University of Jerusalem in Israel say they have demonstrated a hypothesis that deep-learning networks learn via an "information bottleneck" process. Hebrew University's Naftali Tishby theorized a network eliminates extraneous data input by compressing the information until only the data most pertinent to general concepts is retained. The team's experiments involved small networks trained to label input data with a 1 or 0, then assigned random initial strengths to 282 neural connections. They monitored what occurred as the networks engaged in deep learning with 3,000 sample input datasets. The researchers found the networks converged to the information bottleneck theoretical limit, at which point input compression was maximized without reducing accurate label prediction. Google researcher Alex Alemi says the bottleneck could function "not only as a theoretical tool for understanding why our neural networks work as well as they do currently, but also as a tool for constructing new objectives and architectures of networks."
|
Could Virtual Reality Replace Therapy?
The Guardian
Simon Hatterstone
October 7, 2017
Psychologists are testing virtual reality (VR) systems as a therapeutic tool for phobias and disorders. University of Oxford professor Daniel Freeman in the U.K. says there are few conditions VR cannot treat, "because...every mental health problem is about dealing with a problem in the real world, and VR can produce that troubling situation for you." University College London professor Mel Slater cites VR's advantage of enabling users to transcend physical or discomfiting situations for therapeutic purposes. People developing VR for therapy increasingly are dependent on the private sector to support their research, often starting their own businesses directly from their work. Slater thinks VR therapy has a better chance of catching on now that technologies such as the Oculus Rift headset are becoming more affordable. Meanwhile, Kate Anthony with the British Association for Counseling & Psychotherapy believes VR treatments should be prescribed only, as wide commercial availability carries the risk of compounding patients' conditions.
|
Mathematical Modeling for Better Designs and Simulations
Dublin City University
October 6, 2017
In an interview, Dublin City University professor Patricia Moore in Ireland discusses her work, which focuses on improving the three-dimensional (3D) computer modeling and simulation of objects. "I try to make it easier for computers to solve simulation problems as efficiently and accurately as possible," Moore says. She notes a lot of the research's practical applications involve deconstructing the task at hand. "My work looks at preserving the smooth geometry of the original design and using as much knowledge of the geometry and material behavior as possible to improve accuracy and reduce the workload in performing simulations," Moore says. She notes improved simulation algorithms enhance the design process in turn, "and we can have designs that are safer, more efficient, and less expensive to produce." Moore believes modeling and simulation is a particularly exciting field "with the increased computational resources in today's processors and recent trends in things like 3D printing."
|
Sensitivity to Time Improves Performance at Remotely Controlling Devices
NC State News
Doug Gillan; Federico Scholcover; Matt Shipman
October 6, 2017
Researchers at North Carolina State University (NCSU) have found that people more sensitive to the passage of time are better at accounting for the time lag inherent in remotely controlling robots or other tools. As part of the study, the researchers asked 22 volunteers to perform two tasks, with one designed to test people's sensitivity to changes in time. The second task involved maneuvering a remotely controlled car through a predetermined course. The second task was performed eight times, with volunteers dealing with four different lag times ranging from 400 milliseconds to one second. The study found people with greater time sensitivity were no faster than their less sensitive counterparts at completing the course, but they made far fewer mistakes. "The most sensitive participants would make an average of 3.5 errors per minute, whereas the less sensitive participants made an average of 6.1 errors per minute," says NCSU's Federico Scholcover.
|
UMass Lowell Professor Steers Ethical Debate on Self-Driving Cars
UMass Lowell News
Christine Gillette; Nancy Cicco
October 5, 2017
University of Massachusetts, Lowell professor Nicholas Evans studies the ethical dilemmas presented by autonomous vehicles, and he is using a U.S. National Science Foundation grant to address these questions by developing decision-making algorithms for self-driving autos. Evans will then test the algorithms' impact on public health under different risk situations via computer simulation. "The first [ethical] question is, 'How do we value, and how should we value, lives?'" Evans says. He thinks understanding how computers make decisions by sifting through numerous possible scenarios according to programmed rules and then quickly rejecting 99.99 percent of them to arrive at a solution can help create better algorithms that maintain fairness while also providing a high level of self-protection. Evans' team also will examine insurance companies' role in algorithm design and issues of how many self-driving vehicles have to be on the road before they lower the overall number of accidents and improve safety.
|
The 7 Deadly Sins of AI Predictions
Technology Review
Rodney Brooks
October 6, 2017
Misconceived notions of how artificial intelligence (AI) will take shape can be attributed to seven common reasons, writes Massachusetts Institute of Technology Panasonic Professor of Robotics (emeritus) Rodney Brooks. He cites Amara's Law that people tend to overestimate a technology's short-term effect and underestimate its long-term effect as one reason. Another is people's inability to imagine a nonexistent technology's limitations, while a third factor is their often-mistaken assumption that performance of a task by AI equals competence. Brooks also says people's penchant to parallelize AI progress in learning a certain task to the same process in humans is misleading. In addition, Brooks says people should not expect AI to continue to progress steadily on an exponential performance path but rather in fits and starts, and they should not believe media-promulgated visions of unexpected AI scenarios. His final reason is the expectation of accelerated AI deployments, when gradual deployments are far more likely.
*May Require Free Registration
|
|
