Welcome to the December 18, 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."
|
|
|
NCI Scores Raijin Replacement and Aims for Top 25 Supercomputer
ZDNet
Chris Duckett
December 18, 2017
The Australian National University's (ANU) National Computer Infrastructure (NCI) has been awarded a two-year government grant to replace its Raijin supercomputer, which NCI describes as "rapidly nearing the end of its service life." Raijin currently is rated at 1.67 petaflops, and it was augmented with four IBM Power System servers procured last December. NCI's Muhammad Atif says using Power Systems with Raijin's x86 architecture yielded an upgrade in performance of two to threefold when running on Power architecture, versus the most recent x86 architecture installed at NCI. NCI wants Raijin's replacement to score as one of the top 25 global supercomputers when it is commissioned in early 2019, with ANU's Brian Schmidt expecting the upgrade to "be a valuable tool for Australian researchers and industry, and will be central to scientific developments in medical research, climate and weather, engineering, and all fields that require analysis of so-called big data."
|
Error-Free Into the Quantum Computer Age
Swansea University
Mari Hooson
December 15, 2017
Researchers at Swansea University in Wales, U.K., led an international team in demonstrating that conventional ion-trap technologies are suitable for building large-scale quantum computers. Modern quantum computers still fail in running complex computations because of environmental noise and errors. "By using quantum error correction, we can respond to this challenge better," says Rainer Blatt of the University of Innsbruck in Austria. The new technique exploits quantum mechanical properties for error detection and correction, and the researchers predict if they can keep the noise below a certain threshold, they can build a quantum computer that can perform quantum computations of arbitrary complexity by increasing the number of entangled quantum bits accordingly. However, in order to achieve that goal, the capabilities of the technological platforms have to be optimally exploited. The researchers introduced new variants of fault-tolerant protocols and found that a new generation of segmented ion traps offers ideal conditions for the process.
|
AUS Students Develop Device to Reduce Car Mishaps
The Gulf Today (United Arab Emirates)
December 18, 2017
Researchers at the American University of Sharjah (AUS) in the United Arab Emirates have designed and developed a smart system to lower the rate of car accidents. The system helps drivers detect the driving activity of surrounding vehicles, and notifies drivers of irregular driving behavior around their vehicles at varying distances. "Our project is meant to be installed in all vehicles as a standard safety mechanism, which is why we made sure to keep it inexpensive and universal to most cars nowadays," says AUS' Saud AlQasimi. The device connects to the car's dashboard to access data on speed, brakes, and engine readings, among other factors. The device then communicates with devices in other cars via a high-level communications protocol used to create personal area networks with small, low-power digital radios, without needing an Internet link. Each device connects to other cars equipped with a similar device through this network, thus sharing the driving data.
|
Complete Design of a Silicon Quantum Computer Chip Unveiled
UNSW Newsroom
Wilson Da Silva
December 16, 2017
Researchers at the University of New South Wales (UNSW) in Australia say they have reinvented silicon microprocessors to create a complete design for a quantum computer chip that can be fabricated mainly from standard industry processes and components. The team says the chip's architecture enables quantum calculations to be conducted using complementary metal-oxide semiconductors. "Our design incorporates conventional silicon transistor switches to 'turn on' operations between [quantum bits] qubits in a vast two-dimensional array, using a grid-based 'word' and 'bit' select protocol similar to that used to select bits in a conventional computer memory chip," notes former UNSW researcher Menno Veldhorst. He also says the selection of electrodes between the qubits facilitates executing two-qubit logic calculations between qubits. UNSW's Andrew Dzurak notes the chip design also "incorporates a new type of error-correcting code designed specifically for spin qubits, and involves a sophisticated protocol of operations across the millions of qubits."
|
The Hard Math Behind Bitcoin's Global Warming Problem
Wired
Adam Rogers
December 15, 2017
A recent Digiconomics report calculating that the global electricity consumption of bitcoin mining would exceed that of the U.S. by July 2019, and of the entire world by November 2020, suggests the cryptocurrency's use discharges the equivalent of 17.7 million tons of carbon dioxide annually. At the root of this problem is the blockchain technology, the secure ledger of all bitcoin transactions, which relies on solving the SHA-256 hashing algorithm to add a block. The SHA-256 algorithm is by design so difficult that it cannot be solved without brute-force computing, which means the computer must be on continuously. With power consumption rising, bitcoin miners are building ASIC clouds in places where electricity is affordable. "Increasing the energy efficiency of bitcoin SHA-256 mining hardware helps only sublinearly, as improving energy efficiency simply means people can deploy more miners at the same operating cost," notes the University of Washington's Michael Taylor.
|
From Algorithms to Institutions
Nieman Journalism Lab (MA)
Lucas Graves
December 17, 2017
The push to automate aspects of fact-checking is expected to spur an honest dialogue between journalists, policymakers, and platform companies about strengthening the fact-building institutions supporting public truth claims. Collaboration between fact-checkers and computer scientists aims to make automation complementary, and not a replacement, for journalism. One example is the University of Texas at Arlington's ClaimBuster, which sifts through transcripts to identify and rank factual statements that fact-checkers might want to probe. Meanwhile, Trends, a U.K.-based project, provides fact-checkers with a strategic perspective of misinformation for more effective corrections. Thus far the most potent automation technology is the ClaimReview tagging scheme, which makes fact-checkers' work legible to algorithms. These projects are bringing a real-time end-to-end fact-checking engine closer to reality with the hope that such efforts will "make sure the kinds of resources fact-checkers can rely on are available as structured data that computers can use," according to a report from Full Fact.
|
Computer Systems Predict Objects' Responses to Physical Forces
MIT News
Larry Hardesty
December 13, 2017
Massachusetts Institute of Technology professor Josh Tenenbaum and student Jiajun Wu are investigating the use of computer systems to explore basic cognitive abilities that an intelligent agent needs to navigate, such as discerning distinct objects and deducing their response to physical forces. "We're starting to be able to build machines that capture more and more of people's basic understanding of the physical world," Tenenbaum says. Their projects involve systems that are trained to infer a physical model of the world using synthetic data, but then work backward, using the model to re-create the input data, with performance assessed on how well the resynthesized data matches the original data. In one project, they trained a system to analyze audio recordings of an object being dropped to deduce properties such as its shape, composition, and the height from which it fell.
|
Carnegie Mellon Reveals Inner Workings of Victorious AI
CMU News
Byron Spice
December 17, 2017
Researchers at Carnegie Mellon University have detailed the operation of their Libratus artificial intelligence (AI) system, which beat professional poker players in no-limit Texas Hold'em using a three-pronged strategy. The team says Libratus was designed to deconstruct the game into computationally manageable segments and, based on opponents' gameplay, correct potential weaknesses during the tournament. "The techniques in Libratus do not use expert domain knowledge or human data and are not specific to poker," note the researchers. Libratus initially abstracts the game in a smaller and more easily solvable form by considering all possible decision points, then generates a detailed strategy for the early rounds of Texas Hold'em and a coarse strategy for the later rounds. In the final rounds the AI builds finer-grained abstraction based on the state of play, and computes in real time a strategy for this subgame that balances strategies across different subgames, using the coarse strategy as a guide.
|
Surrey Reveals Cutting-Edge Motion Capture Technology That Could Transform Creative Industries
University of Surrey
December 14, 2017
Researchers at the University of Surrey in the U.K. have developed TotalCapture, a sophisticated but low-cost full-body motion-capture (mocap) system which they say could revolutionize the film and video game industry. TotalCapture uses standard video cameras and inertial measurement units typically found in mobile phones, requires no optical markers or specialized infrared cameras, and can be used anywhere. Surrey professor Adrian Hilton says the technology gives "filmmakers the freedom to imagine a world where they are not restricted to a sound stage with expensive cameras to accurately capture a performance for a video game or [computer-generated] character in a film." Hilton says TotalCapture makes it easy to capture the performance of a subject in a variety of locations using a simple and low-cost setup. "Our system has modest hardware requirements, produces a quality image, and can be used in outdoor or indoor environments," says Surrey's Charles Malleson.
|
Researchers Use WWII Code-Breaking Techniques to Interpret Brain Data
University of Pennsylvania
Evan Lerner
December 13, 2017
Researchers at the University of Pennsylvania, the Georgia Institute of Technology, and Northwestern University in Illinois have applied cryptanalysis methods dating from the second World War to decode motor neuron activity, and interpret brain data to predict the direction in which monkeys will move their arms. As the monkeys performed tasks in which they had to reach to a target that appeared at different points around a central starting point, electrodes recorded elevations of electrical activity corresponding with the movement of their arms. The researchers sought the statistical structures of movements to consistently and mathematically map those patterns to the monkeys' arm movements, using insights about language structure used to decrypt the Enigma code. "The algorithm tries a range of possible decoders until we get something where the output looks like typical movements," says University of Pennsylvania professor Konrad Kording. The researchers see potential for using this cryptanalysis technique to enable brain-computer interfaces to achieve literal mind-reading.
|
Engineers Program Tiny Robots to Move, Think Like Insects
Cornell Chronicle
Syl Kacapyr
December 13, 2017
Researchers at Cornell University are studying a new type of programming that mimics an insect's brain functions. The amount of computing power required for a robot to sense a gust of wind, adjust its flight, and plan a path to make a landing would require it to carry a desktop-size computer, but the Cornell researchers want to use neuromorphic computer chips to shrink the payload. Neuromorphic chips process spikes of electrical current that fire in complex combinations, similar to how neurons fire inside a brain. The researchers are developing a new class of "event-based" sensing and control algorithms that mimic neural activity and can be implemented on neuromorphic chips. The chips would require less power than traditional processors, letting developers pack more computation into the same payload. "This network is capable of learning in real time to account for irregularities in the robot introduced during manufacturing, which make the robot significantly more challenging to control," says Cornell's Taylor Clawson.
|
The Great AI Paradox
Technology Review
Brian Bergstein
December 15, 2017
Concerns about artificial intelligence (AI) making jobs or even humans obsolete distract from the need to assume more responsibility for the impact of current automation and addressing the concentration of power in the technology industry. The most notable AI breakthroughs stem from advancements in adaptive machine learning, although experts contend common sense is still beyond AI, which limits the likelihood that machines one day actually will think. Other scientists believe the arrival of truly intelligent machines is inevitable, given computers' tendency to improve exponentially. Despite these predictions, far more pressing are issues of things going awry with current AI and computers, such as automated cyberattacks and computerized decision-making that is based on biased and discriminatory data. Technology publisher Tim O'Reilly makes the argument that companies increasingly automate simply to save money and maximize returns for investors, with resultant savings not poured back into job creation, which goes against long-term corporate interests.
|
|
