Welcome to the August 2, 2021 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."
|
|
|
_robodog.jpeg)
Robotic Police Dogs: Useful Hounds or Dehumanizing Machines?
Associated Press
Matt O'Brien; Jennifer Sinco Kelleher
July 30, 2021
Police departments claim to use robotic dogs as simply another tool to keep emergency responders out of danger, but privacy advocates say the robots are secretly being deployed without safeguards against aggressive, invasive, or dehumanizing uses. The New York Police Department acquired a Spot robotic canine last year from robotics developer Boston Dynamics, but returned it when videos of the robot in the wild sparked a public outcry. Boston Dynamics' Michael Perry said weaponizing Spot or using it to violate privacy or civil rights laws is prohibited, but rival robot-maker Ghost Robotics has no such restrictions. The Hawaii American Civil Liberties Union's Jongwook Kim said, "There's the potential for these robots to increase the militarization of police departments and use it in ways that are unacceptable."
|
EU Fines Amazon Record $888 Million Over Data Violations
Bloomberg
Stephanie Bodoni
July 30, 2021
Luxembourg's CNPD data protection authority fined Amazon a record $888 million for breaching the EU's General Data Protection Regulation (GDPR). The EU regulator charged the online retailer with processing personal data in violation of GDPR rules, which Amazon denies. The ruling closes an investigation triggered by a 2018 complaint from French privacy rights group La Quadrature du Net. Amazon says it gathers data to augment the customer experience, and its guidelines restrict what employees can do with it; some lawmakers and regulators allege the company exploits this data to gain an unfair competitive advantage. Amazon also is under EU scrutiny concerning its use of data from sellers on its platform, and whether it unfairly champions its own products.
|
AI Can Now Be Recognized as an Inventor
ABC (Australia)
Alexandra Jones
July 31, 2021
Australia's Federal Court has granted artificial intelligence (AI) systems legal recognition as inventors in patent applications, challenging the assumption that invention is a purely human act. The decision recognizes DABUS (device for the autonomous bootstrapping of unified sentience), an AI system whose creators have long argued can autonomously perform the "inventive step" required to qualify for a patent. DABUS is a swarm of disconnected neutral networks that continuously generate "thought processes" and "memories" which independently produce new and inventive outputs. It has "invented" a design for a container based on fractal geometry, and a "device and method for attracting enhanced attention" that makes light flicker in a pattern mimicking human neural activity. Although DABUS is listed as the inventor, its creator Stephen Thaler owns the patent, which means the push for the AI's inventor status is not an attempt to advocate for AI property rights.
|
Scientists Reducing Computational Power Required to Analyze DNA
HPCwire
July 29, 2021
The KAUST Metagenomic Analysis Platform (KMAP) allows scientists worldwide to mine massive DNA datasets for new microbes and proteins far more efficiently. Researchers at Saudi Arabia's King Abdullah University of Science and Technology (KAUST) built KMAP to tap into the university's computational resources and help find proteins and enzymes for industrial use. Said KAUST's Intikhab Alam, "Using KMAP, we were able to analyze and compare 275 million microbial genes in only 13 days using KAUST's Shaheen II supercomputer. In comparison, this would have required 522 years using a single computer [central-processing unit]." KAUST's Takashi Gojobori said KMAP will "eliminate the need for advanced bioinformatics skills in order to explore microbial communities and functions."
|
Unfolding the Hippocampus
Western News (Canada)
Crystal Mackay
July 26, 2021
Understanding disease progression and response to treatment for brain disorders by digitally unfolding the hippocampus is the goal of a new technique developed by researchers at Canada's Western University. The method employs data from magnetic resonance imaging to digitally render the hippocampus' three-dimensional folds in two dimensions. Western's Stefan Köhler said, "It's difficult to pinpoint one part of the hippocampus in one person, and find the corresponding part in another person because of the variability from person to person in this folding. Being able to do that is relevant when dealing with clinical questions because you have to consider what is part of normal variability versus what is unique to clinical abnormalities. It's at that level that this technique will really help in the future."
|
Wearable Devices Could Use Your Breathing Patterns Like a Password
New Scientist
Chris Stokel-Walker
July 31, 2021
Cleveland State University's Jafar Pourbemany and colleagues have developed a protocol that generates a 256-bit encryption key every few seconds based on the way a user breathes; the key can then be sent to a wearable device to keep the two in sync. The protocol employs a respiratory inductance plethysmography sensor to measure the user's breathing, with an accelerometer on the chest providing additional data on the way it moves during each breath. The wearer's unique breathing pattern is translated into an encrypted key that can be used to confirm that a device matches correctly with the wearable. Pourbemany said, "Devices need to have a shared secure key for encryption to ensure that an attacker cannot compromise the process."
|
Like Babies Learning to Walk, Autonomous Vehicles Learn to Drive by Mimicking Others
The Brink (Boston University)
Gina Mantica
July 30, 2021
Engineers at Boston University aim to teach autonomous vehicles to drive safely by having them mimic others, similar to the way babies learn to walk. Their machine learning algorithm estimates the viewpoints and blind spots of other nearby cars to generate a bird's-eye-view of the surrounding environment, in order to help autonomous cars detect obstacles and understand how other vehicles turn, negotiate, and yield without colliding. The self-driving cars learn by translating the surrounding vehicles' actions into their algorithm-powered neural networks. Observations from all of the surrounding vehicles in a scene are a core element in the algorithm's training, so the model encourages data sharing and improves autonomous vehicle safety.
|
Honeypot Security Technique Can Stop Attacks in Natural Language Processing
Penn State News
Jessica Hallman
July 28, 2021
A machine learning framework can proactively counter universal trigger attacks—a phrase or series of words that deceive an indefinite number of inputs—in natural language processing (NLP) applications. Scientists at Pennsylvania State University (Penn State) and South Korea's Yonsei University engineered the DARCY model to catch potential NLP attacks using a honeypot, offering up words and phrases that hackers target in their exploits. DARCY searches and injects multiple trapdoors into a textual neural network to detect and thresh out malicious content produced by universal trigger attacks. When tested on four text classification datasets and used to defend against six different potential attack scenarios, DARCY outperformed five existing adversarial detection algorithms.
|
Model Helps Map the Individual Variations of Mental Illness
YaleNews
Bill Hathaway
July 27, 2021
Two new studies applied a novel computational psychiatry framework developed by Yale University scientists that maps individual variations of mental illness to neural circuits. One study employed advanced statistics to identify sets of symptoms that specify patients more accurately than traditional mental illness diagnoses; these symptom signatures exposed precise neural circuits that more accurately embodied variation across hundreds of patients diagnosed as psychotic. Yale's Alan Anticevic said, "This study shows the promise of computational psychiatry for personalized patient selection and treatment design using human brain imaging technology." The second study mapped the effect of LSD on brain circuits via a new neuroimaging technology that incorporates a computational model featuring data on neural gene-expression patterns. Said Yale's John Murray, "We can develop a mechanistic view of how drugs alter brain function in specific regions and use that information to understand the brains of individual patients."
|
Scientists Invent Information Storage, Processing Device
NYU News
July 29, 2021
A team of scientists at New York University (NYU), the University of California, San Diego, and France's University of Paris-Saclay have created a method for engineering a new type of memory device. NYU's Andrew Kent said the research taps the properties of quantum materials to design "a new structure for computation." The team constructed a nanoconstriction spintronic resonator to manipulate known physical properties, and store and process information similar to the brain's synapses and neurons. The resonator merges quantum materials with those of spintronic magnetic devices. Kent said, "This is a fundamental advance that has applications in computing, particularly in neuromorphic computing, where such resonators can serve as connections among computing components."
|
AI Helps Improve NASA's Eyes on the Sun
NASA
Susannah Darling
July 23, 2021
U.S. National Aeronautics and Space Administration (NASA) scientists are calibrating images of the sun with artificial intelligence to enhance data for solar research. The Atmospheric Imagery Assembly (AIA) on NASA's Solar Dynamics Observatory captures this data, and requires regular calibration via sounding rockets to correct for periodic degradation. The researchers are pursuing constant virtual calibration between sounding rocket flights by first training a machine learning algorithm on AIA data to identify and compare solar structures, then feeding it similar images to determine whether it identifies the correct necessary calibration. The scientists also can employ the algorithm to compare specific structures across wavelengths and improve evaluations. Once the program can identify a solar flare without degradation, it can then calculate how much degradation is affecting AIA's current images, and how much calibration each needs.
|
As Cyberattacks Surge, Security Startups Reap the Rewards
The New York Times
Erin Woo
July 26, 2021
Security startups have seen venture capital flooding in as cyberattacks ramp up. Research firm PitchBook estimates investors have injected over $12.2 billion into startups that offer cloud security, identify verification, and privacy protection so far this year, compared to $10.4 billion during all of 2020. Capital is flowing into companies developing anti-hack measures related to the shift to cloud computing, like identity verification software supplier Qomplx and cloud security provider Netskope. Cloud security startup Lacework, whose products use artificial intelligence to identify threats, got a $525-million funding boost in January, which CEO David Hatfield credits to "the combination of all of these ransomware and nation-state attacks, together with people moving to the cloud so aggressively."
*May Require Paid Registration
|
|
