Welcome to the April 12, 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."

Signing in to the Facebook service. Facebook Algorithm Shows Gender Bias in Job Ads, Study Finds
The Wall Street Journal
Jeff Horwitz
April 9, 2021


Facebook disproportionately shows certain types of job ads to men and women, raising questions about efforts to eliminate bias in its algorithms, according to University of Southern California (USC) researchers. Facebook systems were more likely to present ads to users if their gender identity mirrored the concentration of that gender in a specific position or industry. This bias extended through all tiers of jobs, suggesting "a platform whose algorithm learns and perpetuates the existing difference in employee demographics." The study highlights Facebook's problems in understanding and addressing the societal effects of its content-recommendation systems, and USC's Aleksandra Korolova said she was surprised at the company's failure to remedy the situation, because "they've known about this for years."

Full Article
*May Require Paid Registration
AI Could 'Crack the Language of Cancer, Alzheimer's'
University of Cambridge (U.K.)
April 8, 2021


A study by researchers at St. John's College, University of Cambridge in the U.K. found that the "biological language" of cancer, Alzheimer's, and other neurodegenerative diseases can be predicted by machine learning. The researchers used algorithms similar to those employed by Netflix, Facebook, and voice assistants like Alexa and Siri to train a neural network-based language model to study biomolecular condensates. St. John's Tuomas Knowles said, "Any defects connected with these protein droplets can lead to diseases such as cancer. This is why bringing natural language processing technology into research into the molecular origins of protein malfunction is vital if we want to be able to correct the grammatical mistakes inside cells that cause disease."

Full Article

KTH researcher Hongyi Liu tests a robot arm by placing his hand in its path. Robots Can Be More Aware of Human Co-Workers, with System That Provides Context
KTH Royal Institute of Technology (Sweden)
April 7, 2021


Researchers at Sweden's KTH Royal Institute of Technology have designed a context-aware system that could enable robots to work alongside humans on assembly lines with greater efficiency, while foregoing needless interruptions. KTH's Hongyi Liu said the human-robot collaboration system can identify each co-worker and their skeleton model, which is an abstract of their body volume. This lets the system identify the worker's pose and even anticipate the next pose, providing a context for the robot to be cognizant of while interacting. Liu said the system uses artificial intelligence that is less computationally intensive and requires smaller datasets than traditional machine learning, via its reliance on transfer learning. Experiments demonstrated that increasing their contextual awareness allows robots to function more safely and efficiently without slowing production.

Full Article

A sticker that absorbs sweat for analysis. Sticker Absorbs Sweat—and Might Diagnose Cystic Fibrosis
Wired
Max G. Levy
April 8, 2021


University of Hawaii at Manoa, Northwestern University, and other researchers collaborated on the design of a sticker that changes color as it absorbs progressively higher amounts of salt, which can indicate the presence of cystic fibrosis. The circular sensors lie flat on the skin, absorbing perspiration through its center and into capillaries that extend to its edge. A clinician uses a low electric current to drive a sweat-gland-activating gel into the skin, and when the sticker is applied five minutes later, sweat slips into the sticker’s capillaries, where it blends with silver chlorinalite. This chemical changes color in contact with chloride ions, and an application processes a smartphone-captured photo of the sticker. Said Northwestern's John Rogers, "The idea of just being able to sense what's going on in yourself, simply by looking down at your smartphone or a sensor—it's just amazing.”

Full Article
Bitcoin Mining Emissions in China Will Hit 130 Million Tons by 2024
New Scientist
Donna Lu
April 6, 2021


Researchers at China's Tsinghua University forecast that Chinese bitcoin mining will emit roughly 130 million metric tons of carbon into the atmosphere by 2024 and account for about 5.4% of carbon emissions from China's electricity generation. The team predicted the 2024 emissions peak by calculating when the investment in computing equipment and electricity costs associated with bitcoin mining will overtake the financial benefits. The researchers used financial projections and carbon emissions analysis to model the emissions footprint, and factored in location. Tshinghua’s Guan Dabo said regulations will be needed to reduce carbon emissions from bitcoin mining and future emergent sectors, in light of China’s commitment to a 2060 net-zero carbon goal.

Full Article

New Toyota vehicles. Toyota Unveils Models in Advanced Driver-Assist Technology Push
Reuters
Eimi Yamamitsu
April 8, 2021


Japanese automaker Toyota Motor has unveiled the newest models of its Lexus and Mirai vehicles in Japan, outfitted with advanced driver assistance systems (ADAS). Toyota's Advanced Drive solution features a level 2 autonomous system that helps with driving, including keeping the vehicle in its lane, maintaining its distance from other vehicles, and changing lanes safely. Toyota's James Kuffner said the new models are the first the company has brought to market that provide over-the-air updates to its software, and which employ deep learning artificial intelligence. Future cars, Kuffner added, will be “upgradable” and “more customizable.”

Full Article
DNA Breakthrough Could Finally Make Tape Storage Obsolete
TechRadar
Joel Khalili
April 10, 2021


Researchers at the U.S. Department of Energy's Los Alamos National Laboratory (LANL) have brought DNA data storage one step closer to realization by converting digital binary files into a genetic alphabet. LANL's Latchesar Ionkov said, "Our software, the Adaptive DNA Storage Codec [ADS Codex], translates data from what a computer understands into what biology understands. It's like translating English to Chinese, only harder." To combat natural errors that occur when additional values are added or accidentally erased from the letters composing a DNA sequence, when the data is converted back to binary, the codec looks for anomalies and, if one is spotted, adds and subtracts letters from the chain until the data can be verified.

Full Article
Technological Leap for Holograms: Real 'Doodles of Light' in Real Time
SciTechDaily
April 9, 2021


An algorithm developed by researchers at Tokyo Metropolitan University (TMU) in Japan can convert freely drawn lines into holograms on a standard desktop central processing unit. The team connected a tablet to a standard desktop PC with no graphics processing unit, and conventional hologram generation hardware. The algorithm was shown to be sufficiently fast to allow handwriting on the tablet to be translated into images in the air in near-real time, while the use of the PC potentially expands the interactive system's scope of implementation. The TMU researchers are particularly interested in deploying the technology in heads-up displays of navigation instructions in helmets and cars.

Full Article

A soft robotic starfish that can swim autonomously with real fish. Robo-Starfish Aims to Enable Closer Study of Aquatic Life
MIT Computer Science and Artificial Intelligence Laboratory
April 8, 2021


Researchers at the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Lab (CSAIL) have developed a tool to simulate and fabricate a functional soft robot, which they used to create a soft robotic starfish to study aquatic life. The tool aims to address the challenges of designing an effective underwater robot given the movement of water, and speed up the process for producing one. The tool uses a machine learning model to perform an initial simulation and design of control mechanisms and quickly fabricate the robot, after which real-world experiments are conducted to generate more data to optimize the design. The process generally requires that the robot be re-fabricated once. CSAIL's Josephine Hughes said the simulator used control strategies that humans would not have considered.

Full Article
UCLA Researchers Develop AI to Analyze Cells Without Killing Them
Daily Bruin
Madison Pfau; Anna Novoselov
April 9, 2021


An artificial intelligence (AI) model developed from images of stem cells by University of California, Los Angeles (UCLA) researchers enables the analysis of a cell's appearance and protein content without killing it. The model features two AI networks; UCLA's Sara Imboden explained the first is fed colorful immunofluorescent images and black-and-white microscope images from the same field of view, learning by detecting the inputs' relationships. Imboden said after studying these images, the generator tries to output an image as similar to the colorful image as possible. The second network compares the predicted image to the AI-generated image and guesses which is fake, while highlighting weaknesses produced by the first network. UCLA's Cho-Jui Hsieh said, "No matter what kind of input you give, the AI will always try to predict something. Measuring the faithfulness of the AI prediction is very important future work."

Full Article
Computer Model Fosters Potential Improvements to 'Bionic Eye' Technology
Keck School of Medicine of the University of Southern California
Wayne Lewis
April 9, 2021


Researchers at the Keck School of Medicine of the University of Southern California used an advanced computer model to mimic the human retina, in order to improve prosthetic eye technology. The model replicates the shapes and positions of millions of nerve cells in the eye, along with their associated physical and networking characteristics. The Keck team focused on nerve cells that send visual information from the eye to the brain, and identified potential ways to boost clarity and grant color vision to future retinal prostheses. Said Keck's Gianluca Lazzi, "There’s a long road, but we’re walking in the right direction. We can gift these prosthetics with intelligence, and with knowledge comes power."

Full Article
Master of Science in Biomedical Informatics
 
AI-Curated Custom Feeds by Subject
 

Association for Computing Machinery

1601 Broadway, 10th Floor
New York, NY 10019-7434
1-800-342-6626
(U.S./Canada)



ACM Media Sales

If you are interested in advertising in ACM TechNews or other ACM publications, please contact ACM Media Sales or (212) 626-0686, or visit ACM Media for more information.

To submit feedback about ACM TechNews, contact: [email protected]