Welcome to the September 19, 2018 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."
|
|
Report: Machines to Handle Over Half Workplace Tasks by 2025
Associated Press Jamey Keaten September 18, 2018
Underscoring how rapidly the labor market will change in the near future, the World Economic Forum (WEF) on Monday released a report showing more than half of all workplace tasks will be completed by machines by 2025. The WEF estimates that machines will be responsible for 52% of the division of labor as a share of hours within seven years, marking a significant jump from the current share of 29%. While about 75 million jobs worldwide will be lost by 2022, that number could be offset by the creation of 133 million new jobs, according to the report. However, this will require widespread training and retraining of employees. WEF board member Saadia Zahidi says, "Our research suggests that neither businesses nor governments have fully grasped the size of this key challenge of the Fourth Industrial Revolution."
|
Making Algorithms Less Biased
NextGov.com Dave Nyczepir September 18, 2018
In an effort to help government officials eliminate bias in algorithms that make automated decisions on significant legal and other issues, Johns Hopkins University's Center for Government Excellence (GovEx) on Monday released a toolkit to help assess data science projects. The project evolved from academic and nonprofit discussions about how to address the problem of racially biased algorithms being used for bail determinations and judges' sentences. Working with San Francisco's open data initiative DataSF, Washington D.C.'s Data Community DC, and Harvard University's Civic Analytics Networks, GovEx developed a toolkit that provides officials with questions they can ask of data science projects. GovEx's Andrew Nicklin says the toolkit's questions assess factors such as whether an algorithm makes use of an old dataset that could create a "high risk of historical bias." The toolkit also offers strategies for mitigating risk, which could include a performance monitoring program or large-scale review board, depending on the situation. Joy Bonaguro, former chief data officer for San Francisco, says "Instead of wringing our hands about ethics and AI, our toolkit puts an approachable and feasible solution in the hands of government practitioners—something they can use immediately, without complicated policy or overhead."
|
New Photonic Chip Promises More Robust Quantum Computers
SciMex September 15, 2018
A team led by the Royal Melbourne Institute of Technology's (RMIT) Alberto Peruzzo has developed a topological photonic chip to process quantum information, a breakthrough that could serve as a more robust option for scalable quantum computers. In addition, this research could lead to the development of new materials, next-generation computers, and deeper understandings of fundamental science. The researchers used topological photonics to fabricate a chip with a "beamsplitter," creating a high precision photonic quantum gate. The team used the photonic chip to demonstrate that topological states can undergo high-fidelity quantum interference. Demonstrating high-fidelity quantum interference--a foundational aspect of transmitting accurate data using single photons for quantum communications. Peruzzo says, "Topological photonics have the advantage of not requiring strong magnetic fields, and feature intrinsically high-coherence, room-temperature operation and easy manipulation."
|
One Big Reason Why Women Drop Out of Doctoral STEM Programs
Ohio State News Jeff Grabmeier September 17, 2018
Ohio State University researchers Valerie Bostwick and Bruce Weinberg have found that the fewer women who enter a doctoral program at the same time, the less likely any one of them will graduate within six years. When only one woman is in a new class, she is 12 percentage points less likely to graduate within six years than her male classmates, Bostwick and Weinberg found. For each additional 10% of women in a new class, the on-time graduation rate for women increases by more than 2 percentage points. The work indicates that the "female-friendliness" of doctoral programs may significantly impact the gender gap in science, technology, engineering and mathematics (STEM) fields. The researchers used a new data set, and linked transcript records from all public universities in Ohio to data from the UMETRICS project, which provides information on students supported by federal research grants. The researchers looked at research funding and grades, but did not find differences significant enough to explain why women are not completing their STEM degrees. This suggests the problem is the academic climate for women, the researchers say.
|
New Devices Based on Rust Could Reduce Excess Heat in Computers
Johannes Gutenberg University of Mainz (Germany) September 17, 2018
Researchers at Johannes Gutenberg University Mainz (JGU) in Germany, in cooperation with colleagues at Utrecht University in the Netherlands, and the Center for Quantum Spintronics (QuSpin) at the Norwegian University of Science and Technology (NTNU) in Norway, have demonstrated that antiferromagnetic iron oxide, the main component of rust, is an inexpensive and promising material to transport information with low excess heating at increased speeds. The researchers used platinum wires over the insulating iron oxide to enable an electric current to pass close by. This electric current leads to a transfer of energy from the platinum into the iron oxide, thereby creating magnons; the iron oxide was able to carry information over the large distances needed for computing devices. JGU researcher Romain Lebrun says, "This result demonstrates the suitability of antiferromagnets to replace currently used components. Devices based on fast antiferromagnet insulators are now conceivable."
|
Helping Computers Fill in the Gaps Between Video Frames
MIT News Rob Matheson September 13, 2018
Massachusetts Institute of Technology (MIT) researchers have developed the Temporal Relation Network (TRN), an add-on module that helps convolutional neural networks (CNNs) to close the gaps between video frames to greatly improve the network's activity recognition. TRN learns how objects change in a video at different times by analyzing a few key frames depicting an activity at different stages of the video. Using the same process, the module can recognize the same type of activity in a new video. During testing, TRN outperformed existing models by a wide margin in recognizing hundreds of basic activities. In addition, it more accurately predicted what will happen next in a video, given only a small number of early frames. The researchers trained and tested TRN on three crowdsourced datasets of short videos of various performed activities. MIT researcher Bolei Zhou says, "Because we know a lot of the physics inside these videos, we can train [the] module to learn such physics laws and use those in recognizing new videos."
|
Computer Science Graduates Make More Money Than Anyone Else in the U.K.
TechRepublic Alison DeNisco Rayome September 17, 2018
Computer science graduates in the U.K. earn more than any other undergraduate degree holder, according to the Good University Guide. In addition, computer science graduates earn more than five times as much as those who graduate with arts degrees. For example, computer science graduates from Imperial College London earned a median salary of about $65,720 six months after graduation, while graduates in drama, dance, and cinematics at Liverpool Hope University earned an average of just $11,831 six months after graduation. The guide also shows that only 15 of the 132 institutions included in the report graduate students who earn average starting salaries at or above $32,875, the threshold at which student loan payments begin. Many companies, in an effort to widen their talent pool, are ending the requirement that IT professionals and developers must hold a computer science degree to be hired.
|
Searching Through Noise for Pros and Cons
Technische Universitat Darmstadt September 14, 2018
Researchers at Technische Universitat Darmstadt (TU Darmstadt) in Germany have developed ArgumenText, an argument search engine for Internet texts that can filter concrete pro and con arguments on any topic. The researchers created the program, which is based on Ubiquitous Knowledge Processing, by utilizing various models of argumentation theory and teaching computer systems to use these models. The researchers optimized the algorithms with a powerful computer network that efficiently indexed Internet-based texts. The algorithm sorts statements by how reliably they can serve as arguments. TU Darmstadt researchers Johannes Daxenberger and Christian Stab says that argument-mining—the recognition of linguistic arguments by means of computer science—is becoming increasingly important and visible in the digital humanities. While ArgumenText only works with English documents from 2016, the researchers are working to make the program usable with German-language texts as well.
|
Google Grants €530,000 to Boost Irish Computer Science Curriculum Buzz
Silicon Republic John Kennedy September 18, 2018
Google.org, the charitable arm of Google, has provided a €530,000 grant to Irish NGO Camara Education to double the number of TechSpaces in Irish youth clubs. The TechSpace Network, jointly run by Camara and Foroige, an Irish youth organization, aims to change the lives of young people in Ireland by becoming Ireland's leading creative technology network for outcome-focused youth development. The grant will be used to enhance computer science training and provide equipment and ongoing support to youth workers and project managers in dozens of Foroige youth groups. Youth workers will also be offered a scholarship to complete a Level 8 National University of Ireland certificate in digital creativity in youth settings, which will include a module in computer science education.
|
A New Scientific Field: Quantum Metamaterials Have Arrived
American Technion Society Kevin Hattori September 13, 2018
American Technion Society researchers have demonstrated that it is possible to apply metamaterials to the field of quantum information and computing, a breakthrough that could lead to numerous practical applications including the development of unbreakable encryptions, and new possibilities for quantum information systems on a chip. The researchers showed that metamaterials can be used to generate and manipulate entanglement, conducting two sets of experiments to generate entanglement between the spin and orbital angular momentum of photons. In the first experiment, the researchers split the photon pairs, directing one through a unique fabricated metasurface and the other to a detector to signal the arrival of the other photon. They then measured the single photon that passed through the metasurface to find that it had acquired orbital angular momentum (OAM) and that the OAM had become entangled with the spin. In the second experiment, the single photon pairs were passed through the metasurface and measured using two detectors to show that they had become entangled.
|
Stanford's JackRabbot 2: The Polite Pedestrian Robot
Stanford News Taylor Kubota September 18, 2018
The Stanford University Vision and Learning Lab has come out with JackRabbot 2, an update of the original JackRabbot, a project conceived to bridge the gap between delivery hubs and individual businesses or homes. The goal of the project is not just to create a robotic system that is capable of navigating an environment, but to create one that is also aware of its surroundings and the social aspects of interactions with humans. The updated robot includes an arm designed to perform actions such as opening a door and lifting a package, as well as to convey intention to humans, such as signaling a person to go ahead or stop. Facial expressions and sounds will also help JackRabbot 2 to communicate with people. The researchers estimate they will need at least 24 hours of data in various environments to teach the artificially intelligent algorithm guiding JackRabbot 2 to navigate autonomously with humanlike etiquette.
|
The New Computer Science GCSE Is Already Being Reviewed
Schools Week Jess Staufenberg September 17, 2018
The British Computer Society (BCS), which developed the computer science GCSE in 2015, has launched a major review of the subject as pupil enrollment slows, especially among girls. The organization will examine whether the subject now needs to be overhauled in favor of a "broader qualification." Recent figures show that nearly 23,000 fewer pupils left school in 2018 without a qualification in a computer-related subject, a drop of 17% compared to 2017. In addition, national statistics show that 57,245 boys took the computer science GCSE this summer, compared to only 14,683 girls. A £2.4 million pilot program to attract more girls to computing will launch next year, running across a range of state-funded schools until 2022. The program will include interventions that can be scaled up if there is evidence of a positive impact.
|
Hate Speech-Detecting AIs Are Fools for 'Love'
Aalto University September 14, 2018
Researchers from the Aalto University Secure Systems research group in Finland and the University of Padua in Italy found weaknesses in many machine learning detectors that identify hate speech. The team tested seven hate speech detectors, all of which failed. Natural language processing techniques can classify text based on individual characters, words, or sentences, but have difficulty with textual data that differs from that used in their training. Aalto University's Tommi Grondahl says, "We inserted typos, changed word boundaries, or added neutral words to the original hate speech. Removing spaces between words was the most powerful attack, and a combination of these methods was effective even against Google's comment-ranking system Perspective." The researchers recommend a greater focus on the quality of data sets used to train machine learning models, and they say existing applications could be improved through character-based detection. The work will be presented at the ACM AISec workshop in Toronto in October.
|
New Center for Mobile Wearable Systems and Augmented Intelligence Opens
Cambridge Network September 14, 2018
Cambridge University in the U.K. and Nokia Bell Labs have jointly opened a new research center for mobile wearable systems and augmented intelligence. The center will advance state-of-the art mobile systems, security, new materials, and artificial intelligence (AI), focusing on helping people improve communication with one another. The research conducted in the center will redefine the way people interact with the digital and physical world. Cambridge University researcher Cecilia Mascolo, who will head up the new lab with Alastair Beresford, says, "The new center provides support for high-quality, long-term research into mobile, wearable and augmented intelligence systems in Cambridge." In addition, the center will be used to establish a strong research community, and support students and postdoctoral researchers over the next five years.
|
|