Welcome to the July 20, 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."
|
|
BCI Startup Implants First Device in U.S. Patient
Bloomberg Ashlee Vance July 18, 2022
Brain-computer interface (BCI) startup Synchron implanted a wire-electrode combination in the brain of a U.S. patient with amyotrophic lateral sclerosis, attempting to enable the patient to perform thought-powered Web surfing, email, and texting. A catheter is used to insert the stentrode implant into the brain through the jugular vein. As the catheter is removed, the stentrode's mesh opens and fuses with the outer edges of a blood vessel in the brain’s motor cortex; the surgeon then wires the implant to a computing device in the patient's chest. The stentrode interprets signals detected by electrodes in the implant when neurons fire in the brain, which the chest device amplifies and transmits to a computer or smartphone via Bluetooth. Signal strength improves over time, and software analyzes and matches patterns of brain data to follow the patient's commands.
*May Require Paid Registration
|
Smart Chip Senses, Stores, Computes, Secures Data in Low-Power Platform
Penn State News Mariah Chuprinski July 19, 2022
Pennsylvania State University (Penn State) scientists have created a smart chip to reduce energy consumption while further securing digital data. Penn State's Saptarshi Das explained current cloud-based encryption is energy-inefficient and prone to data breaches and hacking. The researchers fabricated the cryptographic platform from two-dimensional molybdenum disulfide, incorporating 320 transistors that each feature sensor, storage, and computing units to encrypt data. Machine learning algorithms enabled the team to analyze output patterns and anticipate input information, and Das said the algorithms could not decrypt the data. The researchers also said the energy consumption was lower than that of silicon-based security methods, supporting an all-in-one chip that senses, stores, computes, and communicates information among connected devices.
|
The Never-Ending Quest to Predict Crime Using AI
The Washington Post Pranshu Verma July 15, 2022
Scientists continue to build crime-predicting artificial intelligence, despite a history of the technology disproportionately directing police against communities of color. Algorithms are trained to find patterns in crime reports, arrest records, and license plate images to anticipate where and when a certain type of crime will occur. However, New York University's Vincent Southerland said the software's accuracy reflects historically biased data skewed toward minorities and low-income neighborhoods. University of Chicago (UChicago) researchers recently claimed a new algorithm can forecast crime with "90% accuracy" by identifying locations in major cities with a high likelihood of crimes happening in the next week. UChicago's Ishanu Chattopadhyay said the software was intended to assess bias in policing across neighborhoods in eight major U.S. cities, revealing very different enforcement in affluent and poorer neighborhoods.
*May Require Paid Registration
|
Chinese-Made GPS Tracker Highly Vulnerable
Associated Press Frank Bajak July 19, 2022
Researchers at cybersecurity firm BitSight warn of severe vulnerabilities in a Chinese-made automotive global positioning system (GPS) tracker that is used in 169 nations. The researchers said hackers could exploit flaws in MiCODUS' MV720 GPS tracker to commandeer device-equipped vehicles, and advised users to disable the product until a software patch becomes available. The device has a default password that few users change and a second hard-coded password that works for all devices; vulnerabilities also reside in the software of the Web server used to remotely manage the trackers. BitSight's Pedro Umbelino said malicious actors could remotely sever the fuel line of a moving vehicle, determine the vehicle's location for espionage purposes, or intercept and corrupt location or other data to sabotage operations. The U.S. Cybersecurity and Infrastructure Security Agency said it was unaware of "any active exploitation" of the MV720's vulnerabilities.
|
'Earable' Uses Sonar to Reconstruct Facial Expressions
Cornell University Chronicle Patricia Waldron July 19, 2022
Cornell University researchers have developed EarIO, a wearable earphone device (earable) that can reconstruct the wearer's face using sonar. EarIO sends facial movements to a smartphone. A speaker on either side of the earphone transmits acoustic signals to the sides of the face, and a microphone detects the echoes, which change due to facial movements as wearers talk, smile, or raise their eyebrows. A deep learning algorithm processes and translates that data back into facial expressions via artificial intelligence. The earable can communicate with a smartphone via a wireless Bluetooth connection, maintaining the user’s privacy.
|
Microsoft Launches Simulator to Train Drone Systems
BBC News Chris Vallance July 18, 2022
A new Microsoft flight simulator is designed to train the artificial intelligence systems of autonomous aerial drones. Project AirSim enables companies to conduct millions of simulated test flights in places that would be too dangerous in reality, in just seconds. Project AirSim operates on Microsoft's Azure cloud computing platform. Josh Riedy at drone-based infrastructure inspection company Airtonomy said the simulator "allows us to make mistakes" in training drones, as well as permitting users to formulate "what if" situations too hazardous for real-world testing. Microsoft said the new proprietary platform requires less technical knowledge to use than previous simulators.
|
UC Model Predicts How Racial Makeup of Neighborhoods Will Change
UC News Michael Miller July 14, 2022
Researchers at the University of Cincinnati (UC) and Poland's Adam Mickiewicz University (UAM) have created a map that can accurately predict the changing racial composition of neighborhoods. UC's Tomasz Stepinski developed a machine learning algorithm to forecast neighborhoods' segregation patterns over the next decade using U.S. Census data. The researchers mapped the data by racial makeup in 300-meter squares called cells, then trained the algorithm to read data collected for two successive Censuses a decade apart. Stepinski compared the algorithm's forecasts to actual data from the 2010 and 2020 Censuses, and found it was as much as 86% accurate. "The ability to predict demographic changes is essential from a scientific point of view and for policymakers, city development, etc.," said UAM's Anna Dmowska.
|
Air-Gap Attack Uses SATA Cable as an Antenna to Transfer Radio Signals
The Hacker News Ravie Lakshmanan July 19, 2022
Mordechai Guri of Israel's Ben Gurion University of the Negev identified a new air-gap attack that leverages Serial Advanced Technology Attachment (SATA) cables as wireless antennas to transmit radio signals in the 6GHz frequency band. The SATAn attack aims to use SATA cables to transfer a small amount of sensitive information wirelessly from air-gapped computers, which are highly secured and physically isolated from other networks, to a receiver over a meter away. Said Guri, "The receiver monitors the 6GHz spectrum for a potential transmission, demodulates the data, decodes it, and sends it to the attacker." To detect the potential for such an attack, external radio frequency monitoring systems could be used to identify anomalous transmissions from the air-gapped system in the 6GHz frequency band.
|
Robot Dog Learns to Walk in One Hour
Max Planck Institute for Intelligent Systems (Germany) July 18, 2022
Researchers at Germany's Max Planck Institute for Intelligent Systems (MPI-IS) built a four-legged, dog-sized robot to study how newborn animals learn to walk. The robot, named Morti, learned to walk in one hour with the help of a Bayesian optimization algorithm that adapts the control parameters of a Central Pattern Generator (GPG), a lightweight computer that acts as a virtual spinal cord. The algorithm can change the length and speed of each leg’s swing and how long the leg is on the ground in response to any stumbling. Said MPI-IS's Felix Ruppert, "Our robot is practically 'born' knowing nothing about its leg anatomy or how they work." Ruppert added, "Changing the CPG output while keeping reflexes active and monitoring the robot stumbling is a core part of the learning process."
|
Data Scientists Identify Lakes, Reservoirs Around the World
University of Minnesota College of Science & Engineering July 19, 2022
Data scientists at the University of Minnesota Twin Cities led an interdisciplinary team that developed a first-of-its kind comprehensive global dataset of the Earth's lakes and reservoirs. The Reservoir and Lake Surface Area Timeseries (ReaLSAT) dataset shows changes in lakes and reservoirs over time, including land and fresh water use and the impact of humans and climate change. The dataset features the location and surface area variations of 681,137 lakes and reservoirs, with monthly data on each from 1984 to 2015. The dataset was developed using machine learning algorithms that combine information on the physical dynamics of bodies of water with satellite imagery. Said the University of Wisconsin-Madison's Paul C. Hanson, "Because ReaLSAT shows changes in lakes and their boundaries, rather than just water pixels across the landscape, we can now connect ecosystem process about water quality with hundreds of thousands of lakes around the world."
|
Open Source Platform Enables Research on Privacy-Preserving ML
University of Michigan News Zachary Champion July 19, 2022
University of Michigan (U-M) researchers have open-sourced the largest benchmarking dataset for a privacy-shielding machine learning (ML) method to date. Federated learning trains ML models on end-user devices, rather than transferring private data to central servers. "By training in-situ on data where it is generated, we can train on larger real-world data," said U-M's Fan Lai. "This also allows us to mitigate privacy risks and high communication and storage costs associated with collecting the raw data from end-user devices into the cloud." The FedScale platform can model the behavior of millions of user devices using a few graphic processing units and central processing units, allowing ML model developers to evaluate model performance without large-scale deployments.
|
Tracking Daily Movement Patterns May Help Predict Dementia
Johns Hopkins Bloomberg School of Public Health July 19, 2022
Researchers at the Johns Hopkins Bloomberg School of Public Health (BSPH) suggest wearable movement-tracking devices could help predict dementia among older adults. The researchers reviewed data from ActiGraph activity monitors to detect differences in movement patterns between those with normal cognition and those with mild cognitive impairment or Alzheimer's disease. The latter cohort had significantly lower measures of activity than the normal cohort in the mornings (6 a.m. to noon), and especially in the afternoons (noon to 6 p.m.), while the fragmentation of activity into smaller intervals was 3.4% higher for mild cognitive impairment/Alzheimer's participants in the afternoon. BSPH's Amal Wanigatunga said, "This study reminds us that cognitive decline may in turn slow physical activity—and we might someday be able to monitor and detect such changes for earlier and more efficient testing to delay and maybe prevent cognitive impairment that leads to Alzheimer's."
|
|