Reservoir Computing (RC) addresses complex problems by mimicking the way information is processed in the animal’s brain. It relies on a randomly connected network, acting as a repository of information and ultimately leading to more efficient output. Many reservoir materials have been investigated to achieve RC directly in the problem (rather than simulating it on a digital computer). Currently, a team of researchers from Osaka University has designed a sulfonated polyaniline network for RC.
Neural networks in the brain use electrochemical signals carried by ions. Therefore, when choosing an RC material system, the electrochemical approach is the logical choice. Organic Electrochemical Field Effect Transistors (OECFETs) are popular in bioelectronics. However, it is not yet widely used in RC.
The key to reservoir material is that it has abundant (time-dependent) behavior and is chaotic. Polymer material They are a great option as they form a random network on their own.
Polyaniline is a promising polymer for RC applications due to its ease of polymerization, high atmospheric stability, and reversible doping / dedoping behavior, and can alter conduction.
In addition to the benefits of polyaniline, researchers have investigated sulfonated polyaniline (SPAN), which exhibits high water solubility and self-doping behavior. These make SPAN easier to operate and more uniform doping.
“Atmospheric protons are injected directly into the polymer chains of SPAN, which conduct them,” explains Yuki Usami, the lead author of the study. “This conduction can be controlled by adjusting the humidity.”
Researchers used a simple dropcasting method to assemble SPAN on gold electrodes to produce organic electrochemicals. Communication network Device (OEND).
SPAN OEND was tested for RC by checking the waveform and assessing its performance in short-term memory tasks. The results of the test to see how well the voice can be recognized achieved 70% accuracy. This feature of SPAN OEND was equivalent to RC software simulation.
“We have shown that our SPANOEND system can be applied to RC,” said research author Takuya Matsumoto. “Future steps to establish a humidity-independent system offer more practical options, but the success of our SPAN-based system is a positive step in the material base. reservoir Computing is expected to have a significant impact on the next generation of artificial intelligence devices. ”
Yuki Usami et al., Reservoir Computing in Materio in Sulfonized Polyaniline Network, Advanced material (2021). DOI: 10.1002 / adma.202102688
Quote: The intelligence that emerged from the Random Polymer Networks (October 6, 2021) is from https://phys.org/news/2021-10-intelligence-emerging-random-polymer-networks.html, October 6, 2021. Was acquired by
This document is subject to copyright. No part may be reproduced without written permission, except for fair transactions for personal investigation or research purposes. The content is provided for informational purposes only.
Intelligence from a random polymer network
https://phys.org/news/2021-10-intelligence-emerging-random-polymer-networks.html Intelligence from a random polymer network