Welcome to the December 12, 2022, 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."
|
|
|
Algorithm Finds Midwest Farms Are Using More Cover Crops
The Washington Post
Erin Blakemore
December 10, 2022
Researchers at the University of Illinois at Urbana-Champaign, the U.S. Department of Agriculture, the Illinois Fertilizer and Chemical Association, and the Indiana State Department of Agriculture applied a machine learning algorithm to satellite imagery to learn the extent to which growers in the midwestern U.S. are using cover crops to nourish soil and decelerate erosion. The algorithm detects satellite images of cover crops by looking for bare fields that turn green during the most common cover crop growing season, then brown once the crops die and decompose. The model can distinguish cover crops from cash crops and weed-filled fields. This solution helped the researchers calculate that cover crop use increased from 1.8% in 2011 to 7.2% in 2021.
*May Require Paid Registration
|
DeepMind's AlphaCode Can Outcompete Human Coders
Gizmodo
Mack DeGeurin
December 8, 2022
DeepMind's AlphaCode model performed well against human coders in a programming competition, with a paper describing its overall performance as similar to that of a "novice programmer" with up to a year of training. AlphaCode achieved "approximately human-level performance" and solved previously unseen natural language problems by forecasting code segments and generating millions of potential solutions. The model then winnowed them down to a maximum of 10 solutions, which the researchers said were produced "without any built-in knowledge about the structure of computer code." Carnegie Mellon University's J. Zico Kolter wrote, "Ultimately, AlphaCode performs remarkably well on previously unseen coding challenges, regardless of the degree to which it 'truly' understands the task."
*May Require Paid Registration
|
Daredevil Drones Find Nearly Extinct Plants Hiding in Cliffs
Reuters
Daisy Chung; Gloria Dickie; Simon Scarr
December 10, 2022
Scientists at Hawaii's nonprofit National Tropical Botanical Garden (NTBG) and the state's Division of Forestry and Wildlife are using drones to find and collect samples of nearly extinct plants in cliffs and other hard-to-reach locations. NTBG's Ben Nyberg and researchers at Canadian aerospace company Outreach Robotics developed the Multi-Use Aerial Manipulator Bidirectionally Actuated (Mamba), which dangles on a cable below a drone. The remote-piloted robot arm can detach from the drone to pluck plant samples in perilous regions as far as 4 meters (13 feet) away. Mamba has so far collected 29 cuttings or seeds from 12 endangered plant species on the island of Kauai. Nyberg suggested the robot "can be the difference between extinction and survival" for such plant species.
|
3D-Printed Decoder, AI-Enabled Image Compression Could Enable Higher-Res Displays
UCLA Newsroom
Wayne Lewis
December 8, 2022
A system developed by researchers at the University of California, Los Angeles can project high-resolution encoded images using only 6% of the pixels in their source images. The system features an artificial intelligence algorithm that compresses the images from a high resolution to a lower resolution, resulting in a pixelated pattern that the human eye cannot read. Then, using an optical decoder that bends and unscrambles the incoming light, the compressed image is translated back to its original resolution. The higher resolution displays potentially could use less power and require less data than current display technologies given that no power is consumed by the decoder, which is comprised of a thin, transparent plastic sheet made using a three-dimensional printer.
|
Deep Learning Gets Boost from Reconfigurable Processor
IEEE Spectrum
Michelle Hampson
December 6, 2022
The reconfigurable ReAAP processor developed by a multi-institutional team of researchers trounced several computing platforms used to support deep neural networks. ReAAP is an integrated software-hardware system featuring a software compiler that assesses and optimizes deep learning workloads. Upon ascertaining the optimal solution for processing data in parallel, the processor instructs the reconfiguration of the hardware coprocessor, which apportions the proper hardware resources. "As an end-to-end system, ReAAP can be deployed to accelerate various deep learning applications just by customizing a Python script in [the] software for each application," said Jianwei Zheng at the Hong Kong University of Science and Technology. ReAAP's compiler performs 1.9 to 5.7 times as fast as the next best complier running on the graphics processing unit and 1.6 to 3.3 times as fast as the same compiler running on the central processing unit.
|
Facial Recognition Technology Scans Your Ear
UGA Today
Olivia Randall
December 8, 2022
University of Georgia (UGA) scientists have created an ear recognition system that can authenticate individuals with up to 97.25% accuracy. The researchers said ear shapes are unique to an individual, and for the most part they remain unchanged by age. UGA's Thirimachos Bourlai said the ear recognition software operates similarly to facial recognition, capturing and saving multiple ear scans for authentication. The software uses an ear recognition algorithm to assess and ascertain if the scans are suitable for automated matching. The researchers tested the software's tolerance for busy images by evaluating several models, using ear images distorted by noise factors like varying blurriness, brightness, and contrast.
|
Researchers Develop Virtual Molecular Library of Thousands of 'Command Sentences' for Cells
News-Medical Life Sciences
Emily Henderson
December 8, 2022
Researchers at the University of California, San Francisco (UCSF) and IBM Research have compiled a virtual molecular database of thousands of cellular "command sentences." The researchers based the sentences on combinations of molecular "words" that directed engineered CAR-T immune cells to hunt down and destroy cancer cells. USCF's Wendell Lim and Kyle Daniels concentrated on the part of a receptor within the cell containing amino acid strings, each of which functions as a command "word." Stringing the words into a "sentence" influences the commands the cell will follow. Daniels and the IBM researchers applied novel machine learning to the data to produce new receptor sentences to boost the CART-T cells' effectiveness.
|
Leased Robots to Help Logistics Firms Handle Holiday Crush
The Wall Street Journal
Angus Loten
December 6, 2022
Logistics companies are leasing package-handling robots to supplement fleets of warehouse, distribution, and fulfillment center robots to help manage the holiday surge in orders. Sally Miller at German logistics company's DHL's North American supply-chain business said she intends to implement "a good number" of temporary Locus Robotics "surge bots" for picking up orders and fulfilling them at several retail-focused facilities. At least a third of Locus' more than 90 corporate clients are ordering temporary robots for the holiday crush, up from about 25% last year. Locus CEO Rick Faulk attributed the logistics sector's growing demand for leased robots to labor shortages, supply-chain disruptions, and pandemic-fueled traction in online shopping.
*May Require Paid Registration
|
Deepfake Detector Spots Fake Videos of Ukraine's President Zelensky
New Scientist
Jeremy Hsu
December 7, 2022
A deepfake detector can accurately identify fraudulent videos of Ukraine's president Volodymyr Zelensky, and can be trained to flag deepfakes of other prominent figures. Researchers at the University of California, Berkeley and the Czech Republic's Johannes Kepler Gymnasium trained a computer model on more than eight hours of publicly posted videos featuring Zelensky. The detector vets 10-second clips from a single video, analyzing up to 780 behavioral characteristics. Flagging multiple clips from the same video as false indicates human analysts should look closer. The University at Buffalo in New York's Siwei Lyu said the deepfake detector's holistic head-and-upper-body analysis is uniquely suited to identifying doctored videos.
|
North Korean Hackers Again Exploit Internet Explorer's Leftover Bits
Ars Technica
Kevin Purdy
December 8, 2022
Google's Threat Analysis Group (TAG) claims a North Korean government-backed hacker group continues to exploit persisting flaws in Microsoft's Internet Explorer (IE) browser. TAG analysts said the APT37 group targeted visitors to the South Korean website Daily NK, focusing on the Halloween crowd crush in Seoul's Itaewon district. The group allegedly circulated a potentially malware-bearing Microsoft Word .docx document related to the tragedy that exploited a long-known flaw in Office and WordPad rooted in IE's JavaScript (JScript) engine. APT37 has previously issued exploits that activated BLUELIGHT, ROKRAT, and DOLPHIN malware aimed at North Korean political and economic interests. Microsoft patched the exploit in the JScript engine, but the persistence of remote-code Word doc attacks suggests such hacks will linger.
|
Researchers Use Quantum Computing to Observe Entanglement
MIT News
Julia C. Keller
December 1, 2022
A team that included researchers at the Massachusetts Institute of Technology and the California Institute of Technology (Caltech) demonstrated that qubits can travel from one system of entangled particles to another in a model of gravity. The researchers sent quantum information in a method akin to traversing a wormhole using Google's Sycamore 53-qubit quantum processor. They used machine learning techniques to generate examples of 10-qubit systems that exhibit behavior consistent with quantum gravity. After inserting a qubit into one system and applying an energy shockwave across the processor, the researchers viewed the information on the other quantum system on the processor. Caltech's Maria Spiropulu said, "We showed that if the wormhole is propped open for a long-enough time by the negative energy shockwaves, a causal path is established between the two quantum systems. The qubit inserted into one system is indeed the same that appears on the other system."
|
Smallest Robotic Arm Is Controlled by AI
Aalto University (Finland)
December 7, 2022
Researchers at Finland's Aalto University and the Finnish Center for Artificial Intelligence have manipulated silver atoms into a lattice configuration via deep reinforcement learning, a critical advance for nanodevice construction. The approach rewards the algorithm for correct guidance and outputs. "It took the algorithm on the order of one day to learn and then about one hour to build the lattice," said Aalto's I-Ju Chen. She also said such atomic guidance can be used for testing how and whether nanodevices operate at their absolute limit, as well as for exposing properties related to superconductivity or quantum states.
|
|
