Welcome to the December 1, 2023 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."
|
|
|
Robo-Tentacle Lets People Find Out How It Feels to Have an Octopus Arm
Nature
Bianca Nogrady
November 29, 2023
Researchers at China's Beihang University and the National University of Singapore have developed a soft robotic tentacle inspired by the "bend propagation" movement used by octopuses to catch prey. The robotic tentacle, which can be operated remotely via a finger glove, aims to mimic that movement. The octobot features an electronic network similar to the nervous system in an octopus arm, comprised of five soft silicone segments with embedded "wires" (metal that is liquid at room temperature) and standard silicon chips. The tip of the tentacle features suckers and temperature sensors, and the entire standalone unit can extend 1.5 times its original length. In tests, the octobot successfully captured a plastic toy in both air and water.
|
Digital Emblem for Humanitarian Law in Cyberspace
ETH Zurich (Switzerland)
Samuel Schlaefli
November 29, 2023
In partnership with the International Committee of the Red Cross, computer scientists at Switzerland's ETH Zurich developed a protective emblem that can be integrated easily and cost-effectively into existing digital systems to signify that a digital infrastructure is entitled to protection. The Authentic Digital EMblem (ADEM) is based on the Web Public Key Infrastructure and Certificate Transparency ecosystem. It is cryptographically secured using a digital signature, making it possible to retrieve information about the owner and the IP or domain worthy of protection, as well as the publisher of the emblem. Said ETH's Felix Linker. "Hacker software needs to automatically load and read the emblem, so it can recognize that it is accessing a system belonging to an organization that is protected by international humanitarian law."
|
Developer Conference Axed After Outcry Over Fake Female Profiles
Bloomberg
Natalie Lung; Ella Ceron
November 28, 2023
The online DevTernity developers conference was canceled after several technology executives pulled out of the event following accusations that the organizer made up female speakers’ profiles. Organizer Eduards Sizovs confirmed the cancellation of the event, after admitting in a post on X that he “auto-generated” a fake woman’s profile after a female speaker had dropped out of the conference. Gergely Orosz, who runs a popular tech newsletter, posted on social media last week that he had identified fabricated profiles of women on DevTernity’s speakers list and notified attendees. He also said he found fake women’s profiles on the speakers lists for previous and future events. Some saw the moves as an attempt to boost the event's diversity credentials.
|
Made-to-Order Diagnostic Tests
McGill University Newsroom (Canada)
November 29, 2023
McGill University researchers invented a lab-on-a-chip diagnostic testing system that can be three-dimensionally (3D) printed in 30 minutes. The single-use capillaric chips require no external power source and can be 3D-printed for various tests, including COVID-19 antibody quantification. They function through capillary action, the phenomena by which a spilled liquid is absorbed into a paper towel used to wipe it up. Said McGill's David Juncker, "Our diagnostics are a bit what the cellphone was to traditional desktop computers that required a separate monitor, keyboard, and power supply to operate.”
|
Google DeepMind Researchers Use AI Tool to Find 2 Million New Materials
Financial Times
Michael Peel
November 29, 2023
Researchers at Google DeepMind used the GNoME artificial intelligence tool to identify 2.2 million novel crystal structures, exceeding the number of crystal structures discovered in the history of science by more than 45 times. They plan to make available 381,000 of the most promising structures to other researchers to gauge their viability in solar cells, semiconductors, and other applications. The findings already have been employed by researchers at the University of California, Berkeley, and the Lawrence Berkeley National Laboratory, as they work to develop new materials. The researchers created 41 novel compounds using an autonomous laboratory (A-lab) guided by computation, historical data, and machine learning. Berkley's Gerbrand Ceder said, "While the robotics of the A-lab is cool, the real innovation is the integration of various sources of knowledge and data with A-lab in order to intelligently drive synthesis."
*May Require Paid Registration
|
Seeing Activity Inside a Living Cell
MIT News
Anne Trafton
November 28, 2023
Massachusetts Institute of Technology researchers developed a method to observe several molecules at a time within a cell. The approach makes use of green or red fluorescent molecules that flicker on and off at different rates. By imaging a cell over time and extracting each of the fluorescent signals using a computational algorithm, the amount of each target protein can be tracked as it changes. Using this approach, the researchers were able to label six different molecules involved in the cell division cycle in mammalian cells. This allowed them to identify patterns in how the levels of enzymes called cyclin-dependent kinases change as a cell progresses through its cycle.
|
UK Biobank Releases Half-Million Whole-Genome Sequences for Biomedical Research
Science
Catherine Offord
November 29, 2023
The UK Biobank, one of the largest existing databases of whole genomes, has made available to scientists the full genetic sequences of 491,554 people, more than double the 200,000 whole genomes in a dataset it released in 2021. The latest release was funded by the U.K. government, the Wellcome Trust, and four pharmaceutical companies: Amgen, AstraZeneca, GSK, and Johnson & Johnson. The data will be available to approved researchers through a cloud-based analysis platform. The UK Biobank said more than 30,000 researchers from around 90 countries already have been approved to access the data.
|
Co-learning to Improve Autonomous Driving
Washington University in St. Louis McKelvey School of Engineering
Shawn Ballard
November 27, 2023
A joint learning framework developed by researchers at Washington University in St. Louis (WashU) aims to improve autonomous driving by optimizing stereo matching and optical flow. The framework takes advantage of the inherent similarities in these low-level tasks and addresses them at the same time. Both stereo matching, which produces maps of disparities between two images, and optical flow, which estimates per-pixel motion between video frames, use pixel-wise displacement of images to determine a scene's depth and motion. The model is trained only on ground-truth information from synthetic images. WashU's Nathan Jacobs said, "Since we can obtain a lot of simulated training data, we get more accurate models than training only on the available labeled real-image datasets." The researchers found the model outperforms similar models for estimating stereo matching and optical flow in isolation.
|
Psychology Theories Help Build More ‘Human-like Bots’
The New Indian Express
Puran Choudhary
November 29, 2023
Researchers at the International Institute of Information Technology-Bangalore (IIIT-Bangalore) in India developed a pipeline for modelling engagement in human-robot interactions. Explained IIIT-Bangalore's Arpitha Malavalli, “We wanted to leverage existing psychology theories. We tried to mirror the same in human-robot interaction." The pipeline was trained on human-to-human interactions and human-robot interactions. Said IIIT-Bangalore's Shrisha Rao, "When a human interacts with another human, we analyze their mood, personality, hand gestures, and expressions and alter our responses accordingly. That’s what this pipeline aims to do, and be more cognitive and human-like.”
|
Straining Memory Leads to New Computing Possibilities
University of Rochester News Center
November 30, 2023
University of Rochester (UofR) scientists have developed a new form of computing memory that is fast, dense, and low in power requirements by strategically straining materials that are as thin as a single layer of atoms. The approach couples the best qualities of memristors and phase-change materials, both of which have been explored for their advantages over today’s most prevalent forms of memory. UofR's Stephen M. Wu said memristors tend to suffer from a relative lack of reliability compared to other forms of memory, while phase-change materials, which involve selectively melting a material into either an amorphous state or a crystalline state, require too much power. Said Wu, “We’re making a two-terminal memristor device, which drives one type of crystal to another type of crystal phase. Those two crystal phases have different resistance that you can then store as memory."
|
Neural Network Takes Asia's Air Temperatures
IEEE Spectrum
Rahul Rao
November 28, 2023
A transformer-based neural network developed by researchers at China's Chengdu University of Information Technology and the China Meteorological Administration can generate near-real-time air temperatures using infrared data from a weather satellite. The neural network, TaNET, was trained on infrared surface temperature data from the FengYun-4A (FY-4A) satellite to output a near-surface temperature map corresponding to the European Centre for Medium-Range Weather Forecasts' ERA5, which does not provide real-time data. Tests showed that TaNET outperformed the China Meteorological Administration's CRA and the U.S. National Oceanic and Atmospheric Administration's CFSv2 datasets, as well as a model driven by a U-Net convolutional neural network.
|
Smart Microgrids Can Restore Power More Efficiently, Reliably
UC Santa Cruz Newscenter
Emily Cerf
November 30, 2023
Yu Zhang and Shourya Bose at the University of California, Santa Cruz developed an approach for the smart control of microgrids for power restoration when outages occur. The approach uses deep reinforcement learning to create an efficient framework that includes models of many components of the power system. Said Bose, “We’re modeling a whole bunch of things: solar, wind, small generators, batteries, and we're also modeling when people's electricity demand changes. The novelty is that this specific flavor of reinforcement learning, which we call constrained policy optimization (CPO), is being used for the first time.” CPO takes into account real-time conditions and uses machine learning to find long-term patterns that affect the output of renewable energy sources, unlike traditional systems that use model predictive control (MPC), which bases decisions on the available conditions at the time of optimization.
|
How Do You Make a Robot Smarter? Program It to Know What It Doesn't Know
Princeton University School of Engineering and Applied Science
Molly Sharlach
November 28, 2023
Princeton University and Google researchers developed a technique using large language models (LLMs) to teach robots to realize when they do not know something and to request further instructions. The system involves setting an uncertainty threshold that will prompt the robot to ask for assistance based on the degree of success sought by the user. For instance, the researchers asked a robot to "place the bowl in the microwave," leaving it to choose between metal and plastic bowls. Four actions, each assigned a probability, were generated by the robot's LLM-based planner, and the robot asked which bowl to place in the microwave. Said Princeton's Anirudha Majumdar, "Using the technique of conformal prediction, which quantifies the language model's uncertainty in a more rigorous way than prior methods, allows us to get to a higher level of success."
|
|
