Welcome to the June 22, 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."
|
|
|
Technology Helps Self-Driving Cars Learn From Own Memories
Cornell University Chronicle
Tom Fleischman
June 21, 2022
Technology developed by Cornell University researchers aims to imbue autonomous vehicles with the ability to generate "memories" of previous experiences for use in navigation. The researchers amassed the Ithaca365 traversal dataset by driving a car with LiDAR (Light Detection and Ranging) sensors along a 15-kilometer (9.3-mile) loop 40 times over 18 months in various environments, weather conditions, and times of day. The approach applies neural networks to compute descriptors of objects the car passes, then compresses and stores these Spatial-Quantized Sparse History (SQuaSH) features on a virtual map. When the car next traverses the same location, it can query the local SQuaSH database of all LiDAR points along the route and recall that knowledge.
|
Sensor Imperfections Perfect for Forensic Camera Analysis
University of Groningen (Netherlands)
June 21, 2022
Computer scientists from the University of Groningen (UG) in the Netherlands and Spain's University of León have developed a system that analyzes the noise produced by individual cameras in order to link video or images to specific cameras. UG's George Azzopardi said cameras contain imperfections in their embedded sensors that "manifest themselves as image noise in all frames but are invisible to the naked eye." UG's Guru Bennabhaktula designed a computational model to extract camera noise from video frames shot by 28 different cameras, and fed it to a convolutional neural network. The system identified frames made by the same camera with 72% accuracy; it also could classify 18 camera models with 99% accuracy using publicly available images.
|
Google Wants to Challenge AI with 200 Tasks, to Replace the Turing Test
New Scientist
Matthew Sparkes
June 14, 2022
Artificial intelligence (AI) researchers at 132 institutions, including Google, have proposed an update to the Turing test for rating the capabilities of AI technology. The researchers suggested using the Beyond the Imitation Game benchmark (BIG-bench), comprised of 204 diverse tasks designed not to be fully solved by state-of-the-art AI models. The researchers observed that AI models performed poorly on most of the tasks compared with humans, although they showed improvement as computing power increased. Adrian Hilton of the U.K.'s University of Surrey said, "I can see that setting up a set of benchmarks is one way of comparing one machine-learning algorithm with another, one AI with another. But I don't think that necessarily answers the question of intelligence though."
*May Require Paid Registration
|
Meta Agrees to Alter Ad Technology in Settlement with U.S.
The New York Times
Mike Isaac
June 21, 2022
Facebook parent company Meta has agreed to change its advertising technology and pay a $115,054 penalty as part of a settlement with the U.S. Justice Department over allegations it had discriminated against Facebook users. The U.S. Attorney's Office for the Southern District of New York found Facebook's algorithms direct ads for housing in Black and white neighborhoods mainly to users of those racial groups, and "actually and predictably reinforce or perpetuate segregated housing patterns because of race." Meta pledged to employ a still-in-development computer-assisted system to verify whether persons who are targeted and eligible to receive ads for housing, employment, and credit actually view them.
|
Seeing the Whole from Some of the Parts
MIT News
Steve Nadis
June 17, 2022
Researchers at the Massachusetts Institute of Technology, the University of Illinois, and Canada's University of Toronto developed a new approach to three-dimensional scene reconstruction that eliminates the need for overlapping images. Called "virtual correspondence," the approach works with images taken from different views of a scene that lack the same features. Said MIT's Wei-Chiu Ma, "By looking through the object and coming out the other end, this technique provides points in common to work with that weren't initially available." Ma added, "The prior knowledge and common sense that is built into our algorithms is first captured and encoded by neural networks."
|
Chicago Has a 124-Mile Quantum Network. This Is What It's For
Popular Science
Charlotte Hu
June 20, 2022
Earlier this month, Chicago-based research laboratories launched a 124-mile quantum network in order to enable experimentation toward the creation of a preliminary quantum Internet. The network includes a 35-mile-long stretch of optical fiber mated to an 89-mile quantum loop launched in 2020 by the U.S. Department of Energy's Argonne National Laboratory, connecting labs from the Chicago Quantum Exchange and the University of Chicago. Japan's Toshiba is using the network to test its distributed quantum encryption keys in an environment prone to noise, weather, and temperature fluctuations in order to measure their robustness, and any potential issues that may arise. Scientists will use the system to explore the quantum field's ambiguous nature, given that the nodes at the Chicago labs are all testing different approaches.
|
DARPA Report Finds Vulnerabilities in Blockchain Tech, Non-Secure Crypto Transactions
Nextgov
Alexandra Kelley
June 21, 2022
Researchers at cybersecurity consulting firm Trail of Bits vetted the security of blockchain software's distributed ledger technologies for the U.S. Defense Advanced Research Projects Agency (DARPA). The researchers found some technologies can be altered, jeopardizing the data stored within the proof-of-work blockchain. The team explained the system's "immutability can be broken not by exploiting cryptographic vulnerabilities, but instead by subverting the properties of a blockchain's implementations, networking, and consensus protocol." For example, if just one node in the blockchain ledger network lacks proper security protocols or is operated by a bad actor, the data routed through the blockchain can be hacked or changed. Meanwhile, the corruption of a third party within the network route between nodes means actors can potentially disrupt ledger transactions, because all bitcoin protocol traffic is unencrypted.
|
Computer Hardware Mimics Brain Functions
Argonne National Laboratory
Joseph E. Harmon
June 21, 2022
A multi-institutional team of researchers has developed a material for building microelectronic devices that permits the devices to reprogram themselves by mimicking human brain functions. The material is made from a hydrogen-infused perovskite nickelate, to which the researchers attached electrodes to allow the application of electrical pulses at different voltages. Researchers at the U.S. Department of Energy's Argonne National Laboratory found changing the voltage controls the movement of hydrogen ions in the nickelate, generating behavior that is alternately neuron- and synapse-like. Other voltages yield the on-off currents of computer chips.
|
Hybrid Machine Learning Forecasts Lake Ecosystem Responses to Climate Change
UC San Diego Scripps Institution of Oceanography
Steve Koppes
June 20, 2022
An international team of researchers applied a hybrid empirical dynamic modeling (EDM) methodology to predict the effects of climate change and phosphorus pollution on Switzerland's Lake Geneva. George Sugihara at the University of California, San Diego's Scripps Institution of Oceanography explained that EDM is a form of supervised machine learning that can help model the mechanisms of interconnected and independent ecosystems. The hybrid model implies that increasing air temperature by 3 degrees Celsius (5.4 degrees Fahrenheit) would impact Lake Geneva's water quality as much as the phosphorus contamination of the past 100 years has.
|
Optical Microphone Sees Sound Like Never Before
Carnegie Mellon University School of Computer Science
Aaron Aupperle
June 21, 2022
Carnegie Mellon University researchers have built an optical microphone that can see sound vibrations to an unprecedented degree, enabling the reconstruction of music from a single instrument. The device combines two cameras and a laser to sense high-speed, low-amplitude surface vibrations beyond the human eye's ability to perceive. The system analyzes differences in laser-induced speckle patterns from images captured with a rolling shutter and a global shutter; an algorithm measures the difference in the video streams' patterns, and renders them as vibrations to reconstruct the sound. The researchers demonstrated the system's ability to record isolated audio of separate guitars playing in concert, and of individual speakers playing different music concurrently. They also analyzed tuning fork vibrations, and tapped the vibrations of a bag of Doritos near a speaker to capture the speaker's emitted sounds.
|
SeqScreen Can Reveal 'Concerning' DNA
Rice University News
Mike Williams
June 21, 2022
Researchers at Rice University, the University of Maryland, and the laboratory of consulting firm Signature Science developed the open-source SeqScreen software platform to enhance the identification and monitoring of pathogenic DNA sequences. SeqScreen is designed to screen short DNA sequences, or oligonucleotides, to highlight what Rice's Todd Treangen called functions of sequences of concern (FunSoCs) that could be harmful. Treangen said SeqScreen's algorithm was trained on Signature Science's curated database of thousands of gene sequences representing 32 types of virulence functions. He explained, "SeqScreen really homes in on the presence or absence of functions that represent virulence factors," adding that the platform also can help to detect emerging pathogens in the environment.
|
What Quantum Information, Snowflakes Have in Common, and What We Can Do About It
CU Boulder Today
Daniel Strain
June 15, 2022
Researchers from the University of Colorado Boulder (CU Boulder) and the U.S. National Institute of Standards and Technology demonstrated the ability to read out signals from a superconducting quantum bit (qubit) using laser light, without destroying it. The fragility of qubits is comparable to that of snowflakes; the smallest disturbance can collapse their superposition. The researchers bypassed this weakness by striking an electro-optic transducer, a thin wafer of silicon and nitrogen, with laser light to convert the superconducting qubit's output into visible light. CU Boulder’s Cindy Regal said, “Researchers have done experiments to extract optical light from a qubit, but not disrupting the qubit in the process is a challenge.”
|
|
