Putting Humans Continually In the AI Loop

In learning contexts such as acquiring a new skill, having a human in the loop refers to an optimization process through which a steep learning curve for the first newbie is progressively flattened for each succeeding-generation newbie by incorporating the lessons learned in each generation into improved teaching methods. Thus, the learning time is progressively shortened, until a fully optimized procedure is achieved. FULL TEXT

Exascale Supercomputers Tracking Down Dark Matter, Dark Energy

Detailed cosmological simulations by exascale supercomputers are playing a role in solving the mysteries of dark matter and dark energy. FULL TEXT

Explaining AI Explainability

Having an AI system explain how it reaches its conclusions, legally required in some cases, is still a challenge. FULL TEXT

AI Upgrades the Internet of Things

Adding AI to IoT helps process streams of data for intelligent decision-making and new applications. FULL TEXT

Dark Patterned Voices Manipulate Users

Untrustworthy conversational agents present a significant threat for manipulation of the user FULL TEXT

CONVERSATION AS THE COMMAND

By combining large language models (LLMs) with vision-language models (VLMs, which label images), it is possible for autonomous robots to achieve their tasks by conversing with humans in natural language. This feat makes “conversation the command,” instead of requiring operators to issue robots explicit commands in their own structured language. FULL TEXT

Edge AI Devices Eye Lifetime Learning

Helping artificial intelligence to adapt and learn, now and later.

Above added neurons (gold) to the original set (black) learn new capabilities in-the-field. FULL TEXT

How to Find a Needle in a Haystack

National Lab Works to Cap, Reduce Carbon

Ford getting boost from exascale supercomputers at Argonne National Laboratory to improve manufacturing efficiency and to explore new materials for clean energy applications.

FULL TEXT

Cybersecurity Protects Agriculture Food Chain

Explained Mary Lancaster at the Pacific Northwest National Laboratory, the Food and Agriculture Risk Modeling (FARM) project will identify potential vulnerabilities within today's digital agricultural software and equipment. Credit: Uvation

FULL TEXT

Earbuds to Monitor Brain, Body

The diagnostic goals are to monitor interactions between the earbud readouts of brainwaves, vital signs, and chemical biomarkers in real-world settings by correlating electrical brainwave patterns with the body's chemical biomarkers. The therapeutic goal is to foster improved brain function by monitoring the brain/body followed with appropriate theraputic feedback through the earbuds to improve the patient's condition.

Fabrication process of the prototype earbud showing the printed sensors (middle) bonded with the flexible circuit board (top), and assembled to the outside of earbuds in the optimal sensor assembly area (bottom) with double-sided medical tape. FULL TEXT

Department of Energy Accelerates Emerging Technologies

The Pacific Northwest National Laboratory aims to use computational chemistry expertise to advance the next generation of molecular modeling capabilities. FULL TEXT

Nanophotonics Flattens Smartphone Camera Bump

FULL TEXT

Optical Vectors Beam Multi-Bits

Optical communications—the backbone of the internet—use modulated lasers to transmit over a hundred different colored channels (using wavelength-division multiplexing) each with a hundred gigabits per second of speed. Multi-bit encoding modes—instead of just 1s and 0s—have been proven in the lab to increase throughput even further, but all suffer from errors introduced by noise in the channels, especially over free-space channels such as to satellites, submarines or even through living cells, according to researchers at the University of the Witwatersrand (Johannesburg, South Africa). Now Wits’ researchers have demonstrated a mode-division multiplexing (MDM) scheme using a quantitative measure for a vector (polarized) beam’s purity, its vector-quality factor (technically, its polarization separability) as the information carrier that can encode not just 1s and 0s, but countless modes of vector light that are virtually unaffected by channel noise.

Automated Evolution Tackles Tough Tasks

Deep neural networks (DNNs) using reinforcement learning are getting harder to come by, due to sensitivity to the initial hyper-parameters chosen (such as the width and depth of the DNN, as well as other application-specific initial conditions). However, these limitations have recently been overcome by combining RL with evolutionary computation (EC), which maintains a population of learning agents, each with unique initial conditions, that together "evolve" an optimal solution. FULL TEXT

Keeping Hackers Off the Electrical Grid

FULL TEXT

MXenes Shield Against Radar, RF, Stealth

FULL TEXT

Holy Grail of Batteries Obsolete Lithium Ion

Rechargeable batteries have become the lifeblood of electronics, enabling the mobile revolution. Unfortunately, today's rechargeable batteries incorporate flammable liquid cores. That could change soon, however by switching to rechargeable batteries that have solid cores with nothing to spill, nothing to catch on fire, nothing to potentially explode. FULL TEXT

Bluetooth Hacked Yet Again

FULL TEXT

Artificial intelligence (AI)—in particular, machine learning (ML)— recently began to solve problems for which quantum computers are targeted, according to researchers at the California Institute of Technology (CalTech), the Flatiron Institute (New York City), and IBM (Yorktown Heights, NY) FULL TEXT