Welcome to the November 23, 2022, edition of ACM TechNews, providing timely information for IT professionals three times a week.

Please note: In observance of the U.S. Thanksgiving holiday, TechNews will not be published on Friday, November 25. Publication will resume Monday, November 28.

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."

Computer monitors show financial data. The World Generates So Much Data, New Unit Measurements Were Created to Keep Up
Ashley Ahn
November 19, 2022

Four new prefixes to the International System of Units were announced by the 27th General Conference on Weights and Measures on Nov. 18, marking the first expansion of the metric system since 1991. The new prefixes are ronna (27 zeroes after the first digit) and quetta (30 zeroes) at the top of the measurement range, and ronto (27 zeroes after the decimal point) and quecto (30 zeroes) at the bottom. Said the U.K.'s National Physical Laboratory (NPL), "The change was largely driven by the growing requirements of data science and digital storage, which is already using prefixes at the top of the existing range (yottabytes and zettabytes, for expressing huge quantities of digital information)." NPL indicated ronto and quecto will be useful in quantum science and particle physics.

Full Article
Researchers Building Robots That Can Build Themselves
Brian Heater
November 22, 2022

Researchers at the Massachusetts Institute of Technology's Center for Bits and Atoms have reportedly made progress in the design of self-assembling robots. The system's core components are voxels, which transmit power, data, and force that can be shared between modules. The voxels grasp and attach additional voxels before moving across the grid for further construction. Challenges include equipping the robots with the intelligence to figure out how and where to build, when to start assembling new robots, and how to avoid colliding with each other. The researchers suggest the robots could save significant time committed to prototyping by determining the optimal build.

Full Article

Nvidia revealed the first benchmark tests for AI training on the new H100 GPU. Records for the Biggest, Smallest AI Computers
IEEE Spectrum
Samuel K. Moore
November 17, 2022

The latest benchmarks released by the machine learning (ML) consortium MLCommons show that Nvidia remains the leader in ML training, with its A100 graphics processing units (GPUs) continuing to dominate the MLPerf list. However, Microsoft's Azure cloud systems were on par with the best A100 on-premises computers in terms of image classification. Nvidia’s Dave Salvator said the A100's average performance is now 2.5 times greater than when it appeared on the MLPerf benchmarks in 2020. In contrast, its upcoming H100 GPU is 6.7 times faster than the A100's initial scores, but only 2.6 times faster than its current scores.

Full Article

The University of Maine Advanced Structures and Composites Center unveiled BioHome3D, the first 3D-printed house made entirely with bio-based materials. First 100% Bio-Based 3D-Printed Home
UMaine News
November 21, 2022

The University of Maine (UMaine) has unveiled what it is calling the first three-dimensionally (3D)-printed house constructed wholly of bio-based materials. UMaine researchers created BioHome3D with partners at the U.S. Department of Energy's Oak Ridge National Laboratory, the Maine Technology Institute, and the MainHousing housing authority. The 600-square-foot prototype house was 3D-printed in four modules, then transferred to a foundation outside UMaine's Advanced Structures and Composites Center, where it was completed in a half-day. BioHome3D's floors, walls, and roof are composed of wood fibers and bio-resins, featuring 100% wood insulation and customizable R-values. Maine’s Governor Janet Mills called the project "a positive step forward" toward addressing the state’s housing crisis and labor shortage.

Full Article
Meet the Smartest Office Building in the World
Mike Moore
November 20, 2022

The 31-story JTC Summit smart office building in Singapore incorporates a network of approximately 60,000 sensors to collect data from various building systems. It uses the Open Digital Platform created by the Singapore government's development arm, GovTech, to merge various technologies, including smart energy, building management, and robot delivery services, into a single platform. The structure’s owners can view data virtually via the building's digital twin to make decisions regarding energy use, remotely unlocking entry gates, and identifying escalator malfunctions, among other things, in real time. Additionally, the building is equipped with robots that can deliver packages, identify maintenance issues, and perform security patrols of its halls.

Full Article
A Navigation System with 10-Centimeter Accuracy
TU Delft (Netherlands)
November 16, 2022

Researchers at the Netherlands’ Delft University of Technology (TU Delft), Vrije Universiteit Amsterdam, and VSL National Metrology Institute have developed a navigation system that is accurate within 10 centimeters (3.9 inches). Designed as part of the SuperGPS project to create an alternative positioning infrastructure that taps a mobile telecommunication network rather than satellites, the system’s network is linked to an atomic clock, in order to transmit messages perfectly timed for positioning. The system also uses radio signals with atypically large bandwidth, which TU Delft's Gerard Janssen said "enables higher positioning accuracy."

Full Article
Nvidia's Magic3D Can Generate 3D Models from Text
Ars Technica
Benj Edwards
November 21, 2022

An artificial intelligence model developed by researchers at Nvidia can transform text descriptions into three-dimensional (3D) models. Magic3D produces a 3D mesh model with colored texture in around 40 minutes, which could enable faster video game and virtual reality development. Magic3D optimizes a coarse model generated in low resolution to a higher resolution, a process that, the researchers said, is twice as fast as Google's DreamFusion text-to-3D model. Magic3D also offers prompt-based editing of 3D meshes, and can preserve a subject through several generations. In their paper, the researchers wrote, "We hope with Magic3D, we can democratize 3D synthesis and open up everyone's creativity in 3D content creation."

Full Article

Final inspection of these 100 AI-equipped vehicles on the morning before their release onto I-24 for the field test. Traffic Experiment Pits ML Against 'Phantom' Jams
Berkeley News
Kara Manke
November 22, 2022

A five-day experiment aimed to assess the ability of artificial intelligence (AI)-outfitted vehicles to mitigate "phantom" traffic jams using machine learning technology developed by the multi-university CIRCLES consortium. Scientists at the University of California, Berkeley (UC Berkeley), Nissan North America, Toyota, General Motors, and the Tennessee Department of Transportation dispatched vehicles with AI-powered cruise control onto a stretch of Nashville's I-24 highway. The cruise control systems automatically adjusted the vehicles’ speeds to local traffic conditions, to improve overall traffic flow and fuel economy. Said UC Berkeley's Maria Laura Delle Monache, "By conducting the experiment at this large of a scale, we hope to show that our results can be reproduced at the societal level."

Full Article
Online Advertising Estimated to Use as Much Energy as a Small Country
New Scientist
Jeremy Hsu
November 18, 2022

Researchers at Spain's UC3M-Santander Big Data Institute and the Norway-based advertising technology company Cavai developed a system that can calculate energy consumption related to online advertising. CarbonTag adds a small piece of code to an advertisement in order to collect data that can be used to determine how much energy is used by the user's device, along with software that measures clicks and other ad performance metrics. An artificial intelligence-powered system then estimates the energy consumption of each ad. The system determined median per-ad energy consumption ranged from 0.3 milliwatt-hours on a Windows laptop to 8 milliwatt-hours on a Windows desktop. Said UC3M-Santander Big Data Institute's Patricia Callejo, "Online advertising consumes more than we think, comparable to a small country" such as Luxembourg.

Full Article
*May Require Paid Registration
Technique Prints Flexible Circuits on Curved Surfaces
NC State University News
Matt Shipman
November 21, 2022

North Carolina State University researchers printed flexible circuits directly onto curved and corrugated surfaces. The process involves generating a template for the relevant application that incorporates a microscale groove pattern, then reproducing it in an elastic polymer film. The researchers affix the film to the relevant substrate, filling the grooves with a liquid solution with silver nanowires. After drying, the nanowires remain in a soft material with the preferred shape and circuit pattern. The researchers have applied the technique to produce "smart" contact lenses, pressure-sensitive latex gloves, and transparent electrodes.

Full Article

This high-tech patrol robot is hitting Seoul’s streets, where it works like a mobile surveillance camera that can inspect the blind spots of the city’s fixed CCTV cameras. Self-Driving Robot Patrols Seoul Streets
Roselyne Min
November 19, 2022

South Korea has launched its first autonomous urban patrol robot, which patrols the streets of Seoul in search of dangerous situations. HL Mando, which developed "Goalie," said the robot is like a "moving surveillance camera," able to go where fixed CCTV cameras cannot. Equipped with satellite navigation and remote sensing technology, Goalie can avoid pedestrians and obstacles. HL Mando's Young-ha Cho said, "When the robot sees a dangerous situation or hears a sound like 'help me’, the control center operates the robot to move there and can check whether it is really a dangerous situation or not." HL Mando livestreams but does not store the footage from the robot, and encrypts its communications with the control center.

Full Article

Phase 2 of the Setonix rollout will be completed this year, with the supercomputer available to researchers from early next year. Space-Age Perth Supercomputer Joins the Ranks of World's Greenest
The Sydney Morning Herald (Australia)
Cameron Myles
November 20, 2022

Beginning early next year, researchers will be able to access the Setonix supercomputer at the Pawsey Supercomputing Centre in Perth, Australia. When fully operational, Setonix will use 217,088 AMD cores and 768 graphics cores to provide raw compute power of 43 petaflops. Pawsey's Mark Stickells attributes Setonix's energy efficiency to the use of AMD MI250X graphics processing units and Milan computer processing units, as well as its cooling system. He explained, "The Pawsey Centre employs a cooling mechanism which leverages a shallow aquifer beneath the center to shed the accumulated heat created by Setonix. This removes the need to use cooling towers as the primary cooling mechanism and thus loses no water to evaporation while having sufficient solar generation to run the required pumps.”

Full Article
Algorithm for Optimal Decision-Making Under Heavy-Tailed Noisy Rewards
Chung-Ang University (South Korea)
November 17, 2022

Researchers at South Korea's Chung-Ang University (CAU) and Ulsan Institute of Science and Technology created an algorithm that supports minimum loss under a maximum-loss scenario (minimax optimality) with minimal prior data. The algorithm addresses sub-optimal performance for heavy-tailed rewards by algorithms designed for stochastic multi-armed bandit (MAB) problems. CAU's Kyungjae Lee said the researchers proposed minimax optimal robust upper confidence bound (MR-UCB) and adaptively perturbed exploration (MR-APE) methods. The team obtained gap-dependent and independent upper bounds of the cumulative regret, then assessed their methods via simulations conducted under Pareto and Fréchet noises. The researchers found MR-UCB outperformed other exploration techniques with stronger robustness and a greater number of actions under heavy-tailed noise; MR-UCB and MR-APE also could solve heavy-tailed synthetic and real-world stochastic MAB problems.

Full Article
Probabilistic and Causal Inference: The Works of Judea Pearl
ACM Discounts and Special Offers Program

Association for Computing Machinery

1601 Broadway, 10th Floor
New York, NY 10019-7434

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]