Performing Complex Calculations Using Simple Liquids Like Water


In reservoir computing, the propagation of information is like a rippling wave on the surface of a body of water. Therefore, the term “reservoir” is used. The underwater electrode in the photo is the actual multi-terminal electrode used in this study.Credit: Megumi Akai-Kasaya and others

Researchers led by Osaka University have clarified the excellent information processing capability of physical reservoirs based on the electrochemical reaction of faradic current, and presented a simple system for computational systems using electrochemical ion reactions. increase.

After decades of incredible development, advances in semiconductor-based computing have begun to slow as transistors reach their physical limits in size and speed. However, computing requirements continue to grow, especially in artificial intelligence, where neural networks have millions of parameters.One solution to this problem is Reservoir Computing and a team of researchers led by Osaka University and the University of Tokyo. Hokkaido UniversityAre developing a simple system based on the electrochemical reaction of faradic currents, and they believe that development in this area can be started quickly.

Reservoir computing is a relatively recent idea in computing. Instead of traditional binary programs running on semiconductor chips, the reactions of nonlinear dynamic systems (reservoir) are used to perform much of the computation. Various nonlinear dynamic systems, from quantum processes to optical laser components, have been regarded as reservoirs. In this study, researchers investigated the ionic conductance of electrochemical solutions.

Physical reservoir computing

Calculation of physical reservoirs and construction of molecular-based reservoirs. (A) Structure of conventional reservoir computing. (B) The concept of a physical reservoir computing system.Credit: © 2022 Megumi Akai-Kasaya and other advanced science

“Our simple test equipment consists of 90 pairs of planar electrodes with an ionic solution dripping on the surface,” explains Professor Akemi Akai, the lead author of this study. “The response voltage to the input voltage is used as the response of the reservoir.” This voltage response is due to both the ionic and electrochemical currents passing through the solution. This input / output relationship is non-linear and reproducible, making it suitable for use in reservoir computing. A unique multi-directional data acquisition system on the device controls the read node and enables parallel testing.

Researchers used this device to evaluate two liquids, a polyoxometallate molecule in solution and deionized water. The system displayed a “feedforward connection” between the nodes, regardless of the sample used. However, there was a difference. “Polyoxometallate solutions increase the variety of response currents and are good at predicting periodic signals,” says Professor Akai Kasaya. “But we found that deionized water was the best solution for second-order nonlinear problems.” The superior performance of these solutions is more than handwritten font recognition, isolated word recognition, and other classification tasks. It shows the potential for complex tasks.

Polyoxometallate (POM) molecular structure

(A) Structure of the polyoxometallate (POM) molecule. (B) Schematic diagram of an electrochemical reaction-based reservoir. (C) POM solution (left) and deionized water (right) responses to sinusoidal signals, and their predictive performance at 4x sinusoidal (QDW) target signals. (D) POM solution and water prediction performance for non-linear target signals.Credit: © 2022 Megumi Akai-Kasaya and other advanced science

Researchers believe that the transfer of protons or ions with minimal short-term electrochemical reactions could be developed as a low-cost, energy-efficient, and more computationally powerful computing system. I am. The simplicity of the proposed system opens up exciting new opportunities for developing computing systems based on electrochemical ion reactions.

Reference: December 29, 2021, “Physical Implementation of Reservoir Computing by Electrochemical Reaction” by Shaohua Kan, Kohei Nakajima, Tetsuya Asai, Megumi Akai-Kasaya Advanced science..
DOI: 10.1002 / advs.202104076

(function(d, s, id){
var js, fjs = d.getElementsByTagName(s)[0];
if (d.getElementById(id)) return;
js = d.createElement(s); = id;
js.src = “//”;
fjs.parentNode.insertBefore(js, fjs);
}(document, ‘script’, ‘facebook-jssdk’));

Performing Complex Calculations Using Simple Liquids Like Water Source link Performing Complex Calculations Using Simple Liquids Like Water

The post Performing Complex Calculations Using Simple Liquids Like Water appeared first on California News Times.

Source link


Please enter your comment!
Please enter your name here

Share post:


More like this

Executive Director of Oxfam Brags to Audience at World Economic Forum

Oxfam International at the World Economic Forum called for...

Live news: West should lift sanctions to ease global food crisis, Putin tells Draghi

Southwest-based Dallas said it expects second-quarter revenue to rise...

Thursday is the 16th Straight Day of All-Time Record High Gas Prices

Gas prices hit another record high on Wednesday Thursday...

After buying Bungie, Sony goes all-in on live service games – TechCrunch

After buying Bungie earlier this year, Sony is quickly...