Welcome to the November 6, 2020 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."
|
|
How AI Can Help Save Forests
The Wall Street Journal Ted Alcorn November 3, 2020
Scientists are developing artificial intelligence (AI)-driven methods to aid forest conservation by enabling a new level of real-time awareness. Foresters in California are hoping to prevent or mitigate future wildfires via a forest-clearing plan that relies largely on remotely sensed data and machine learning (ML). Meanwhile, the Global Forest Watch is integrating radar data to penetrate clouds that conceal tropical areas. To reduce false positives of deforestation, consumer-goods company Unilever hired Descartes Labs to apply ML techniques to better differentiate between vegetation of forests requiring protection, and palm plantations where deforestation is necessary. In the hope of helping to create a sustainable trade in carbon offsets, forestry data company SilviaTerra employs AI to process satellite imagery of various tree species and time periods, calculating size and species based on factors like when leaves start changing color in fall.
*May Require Paid Registration
|
Python's Popularity: Ahead of Java for 1st Time, Still Trailing C
ZDNet Liam Tung November 4, 2020
Python has overtaken Java to become the second-most popular programming language on the Tiobe index for the first time, although it still lags behind C. Tiobe indicates a 2.27% year-over-year increase in Python's popularity, versus Java's concurrent -4.47% decline. Tiobe’s Paul Jansen credits Python's recent spike in popularity to its simplicity, which allows its use by non-coders. However, Stephen O'Grady at industry analyst firm RedMonk suggests Python has ascended by becoming a "language of first resort" and the "glue" for many small projects, while also seeing wide adoption in areas like data science.
|
World's Fastest Open-Source Intrusion Detection Is Here
Carnegie Mellon University CyLab Security and Privacy Institute Daniel Tkacik November 5, 2020
Researchers in Carnegie Mellon University's CyLab Security and Privacy Institute have developed an open source intrusion detection system that achieves speeds of 100 gigabits per second on a single server. The team programmed a field-programmable gate array (FPGA) for intrusion detection, and crafted algorithms that cannot run on traditional processors. CyLab's Justine Sherry said the server’s five cores are necessary because the FPGA processes an average 95% of data packets when placed in a network, with the remaining 5% shunted to central processing units when the array is overwhelmed. The system consumes 38 times less power than hundreds of processing cores would in executing the same tasks.
|
Building a Quantum Network, One Node at a Time
University of Rochester NewsCenter Bob Marcotte November 3, 2020
A nanoscale quantum node fabricated from magnetic and semiconducting material by researchers at New York State’s University of Rochester and Cornell University could interact with other nodes though the use of laser light to send and receive photons. The node is built from an array of pillars just 120 nanometers high, within a platform containing atomically thin layers of tungsten diselenide and chromium triiodide. Each pillar is a location marker for a quantum state that can engage with photons, and the associated photons can potentially interact with other sites across the device, and with similar arrays at other locations. Rochester's Arunabh Mukherjee said the development “will go a long way in miniaturizing a quantum computer based on single hole spins."
|
Leveraging a 3D Printer 'Defect' to Create a Quasi-Textile
MIT News Becky Ham October 26, 2020
A graduate student at the Massachusetts Institute of Technology, Jack Forman, created a tulle-like textile by controlling a common defect in three-dimensional (3D) printers. DefeXtiles are based on an under-extruding process developed by Forman called "glob-stretch," in which globs of thermoplastic polymer are connected by fine strands, producing a flexible and stretchy textile very similar to a woven fabric. The new textile, created using a standard $250 3D printer, can be sewn, de-pleated, and heat-bonded. Said Forman, "This is exciting because there's a lot of opportunities with 3D printing fabric, but it's really hard for it to be easily disseminated, since a lot of it uses expensive machinery and special software or special commands that are generally specific to a printer."
|
A Next-Generation Computer Chip With Two Heads
EPFL (Switzerland) November 5, 2020
Engineers at the Swiss Federal Institute of Technology Lausanne (EPFL) Laboratory of Nanoscale Electronics and Structures (LANES) have developed a computer chip that combines logic operations and data storage in a single architecture. The chip is the first to use a two-dimensional material, MoS2, which is only three atoms thick, for logic-in-memory architecture. Said LANES' Andras Kis, "This ability for circuits to perform two functions is similar to how the human brain works, where neurons are involved in both storing memories and conducting mental calculations.” The next-generation circuit could pave the way for smaller, faster, more energy-efficient devices, benefiting artificial intelligence systems.
|
Drone Technology Improves Ability to Forecast Volcanic Eruptions
UCL News October 30, 2020
An international team of researchers led by the U.K.'s University College London modified long-range drones that collect data from active volcanoes, in order to better forecast eruptions. The ABOVE project used drone measurements from Papua New Guinea's Manam volcano to demonstrate the ability to gain predictive insights by combining air, Earth, and satellite-based data. Project collaborators co-developed techniques to measure gas emissions from volcanoes via drones, blending in situ aerial measurements with readings from satellites and ground-based remote sensors. Said Alessandro Aiuppa of the University of Palermo, “Ten years ago, you could have only stared and guessed what Manam’s CO2 emissions were.”
|
Robots Help Answer Age-Old Question: Why do Fish School?
University of Konstanz October 26, 2020
Researchers at Germany's Max Planck Institute of Animal Behavior (MPI-AB) and the University of Konstanz, working with colleagues at China's Peking University, used biomimetic fish-like robots to demonstrate that fish save energy by swimming in schools without having to keep fixed distances from each other. The robotic fish allowed the researchers to directly measure the power consumption of fish swimming together, versus on their own. Said MPI-AB's Liang Li, "If we then have multiple robots interacting, we gain an efficient way to ask how different strategies of swimming together impact the costs of locomotion." In more than 10,000 trials, the researchers found that fish in the front of the school influence the hydrodynamics of fish behind them, and to save energy, follower-fish match their tail beat to the leader's with a specific time lag based on spatial positioning, in a strategy called vortex phase matching.
|
2 Motivational Artificial Beings Better Than 1 for Enhancing Learning
University of Tsukuba (Japan) November 5, 2020
A study by researchers at the University of Tsukuba in Japan found that praise delivered by robots and virtual graphics-based agents has an impact similar to that of praise delivered by humans, when it comes to improving offline learning. Study participants learned a finger-tapping task under different conditions over a two-day period, with variations in timing and frequency of praise, the number of agents, and whether the agents were physically present or on a screen. Said the university's Takamasa Iio, "We found that praise led to a measurable increase in task performance, indicating increased offline consolidation of the task. Further, two agents led to significantly greater participant performance than one agent, even when the amount of praise was identical."
|
Graphene-Based Memory Resistors Show Promise for Brain-Based Computing
Penn State News Walt Mills October 29, 2020
Researchers at Pennsylvania State University (Penn State) are building an artificial neural network that aims to emulate the energy and area efficiencies of the brain. This artificial neural network can be reconfigured by applying a brief electric field to a sheet of graphene. The researchers were able to show at least 16 possible memory states with graphene-based memory resistors, compared with two in most oxide-based memory resistors. Said Penn State's Saptarshi Das, "What we have shown is that we can control a large number of memory states with precision using simple graphene field effect transistors."
|
Know When to Unfold 'Em: Study Applies Error-Reducing Methods From Particle Physics to Quantum Computing
Lawrence Berkeley National Laboratory Glenn Roberts Jr. November 5, 2020
A team of physicists and computer scientists at the U.S. Department of Energy's Lawrence Berkeley National Laboratory (Berkeley Lab) has applied a common particle physics error-mitigation (unfolding) technique to quantum computing. The researchers simulated a quantum computer to compare the performance of three distinct unfolding methods. The iterative Bayesian unfolding (IBU) technique exhibited greater robustness in a noisy, error-prone environment, and outperformed the other two when more common noise patterns were present. The team used the quantum-computing environment model in more than 1,000 pseudo-experiments, and learned that the IBU technique's results were the closest to predictions. Berkeley Lab's Ben Nachman said the simulated and actual quantum computers used ranged from five to 20 quantum bits, and the method should scale to larger systems.
|
Researchers Develop Sentence Rewriting Technique to Fool Text Classifiers
VentureBeat Kyle Wiggers October 27, 2020
Researchers at the Massachusetts Institute of Technology (MIT) created a framework to rewrite sentences specifically to attack text classifiers and trigger misclassification. Attacks on text classifiers could hurt industries like home lending, which relies on artificial intelligence (AI) for document processing. The conditional BERT sampling (CBS) framework, which feeds sentences from an AI language model to RewritingSampler, has a higher attack success rate than existing word-level methods. The CBS framework and RewritingSampler iteratively sample and replace words in a seed sentence for a given number of times, using the sum of word embeddings to maintain the sentence's literal meaning. The system could be misused for attacks, but also may be used to test the robustness of models and to improve their generalization via adversarial training.
|
Students Develop Tool to Predict the Carbon Footprint of Algorithms
University of Copenhagen November 3, 2020
Researchers at Denmark's University of Copenhagen (UCPH) have developed a tool that can calculate the carbon footprint that results from developing deep learning algorithms. The free Carbontracker program can estimate and forecast the energy consumption and carbon dioxide (CO2) emissions of training a deep learning model. The latter metric considers the CO2 resulting from the production of energy in whichever region the training is occurring. The researchers recommend users consider when their model training will take place, as power is not typically environmentally sustainable over an entire 24-hour period. Said UCPH’s Lasse F. Wolff Anthony, "Some require less compute, and thereby less energy, to achieve similar results. If one can tune these types of parameters, things can change considerably."
|
|