The advent of big data has proven that current computational architectures are inadequate. The difficulty of reducing the size of transistors, high power consumption, and limited operating speeds make neuromorphic computing a promising alternative.
Neuromorphic computing is a new computational paradigm inspired by the brain that uses artificial neural networks to recreate the activity of biological synapses. Because such devices act as a system of switches, the on position corresponds to holding or “learning” information, and the off position corresponds to deleting or “forgetting” information.
Recent publications include the Autonomous University of Barcelona (UAB), CNR-SPIN (Italy), Catalonia Nanoscience and Nanotechnology Institute (ICN2), Micro and Nanotechnology Institute (IMN-CNM-CSIC), and ALBA Synchrotron. Investigated the emulation of artificial synapses using a new advanced material device. The project is part of Sofia Martins Ph.D, led by Serra Húnter Fellow Enric Menéndez of the UAB Faculty of Physics and Jordi Sort, a researcher at ICREA. paper.
A new approach to mimic synaptic function
To date, most systems used for this purpose were ultimately controlled by current and were associated with significant energy losses due to heat dissipation. Here, the researcher’s suggestion was to use magnetic ions, which are the non-volatile control of the magnetic properties of the material by voltage-driven ion transfer. This significantly reduces power consumption and improves the energy efficiency of data storage.
Nevertheless Heat dissipation Reduced by the ion transfer effect, the magnetic ion motion of oxygen at room temperature is usually slow Industrial application, It takes a few seconds or minutes to switch the magnetic state. To solve this problem, the team investigated the use of target materials that already contained ions that were transported into the crystal structure. Such magnetic ion targets are completely reversible from a non-ferromagnetic (switch-off) state to a ferromagnetic (switch-on) state, or vice versa, by voltage-driven oxygen motion from the target to the reservoir (ON). You can receive the conversion. The reverse is also true (off).
Given their crystal structure, cobalt oxide was the material of choice for the production of films in the 5 nm to 230 nm thickness range. Researchers investigated the role of thickness in the resulting magnetic ion behavior and found that the thinner the membrane, the faster the magnetization is generated.
X-ray absorption spectra (XAS) of the sample were performed on the BOREAS beamline of the ALBA synchrotron. XAS room temperature, Elemental composition and oxidation state of cobalt oxide film. This resulted in different results for thinner and thickest membranes. These discoveries were important for understanding the differences in the magnetic ion motion of oxygen between films.
The speed of operation achieved in this task was similar to that used in neuromorphic computing, so we further investigated the thinnest cobalt oxide film. In particular, the effects associated with learning neuromorphic abilities were elicited, and the results provided evidence that the magnetic ion system could be emulated.learn“And” forgetting “function.
In addition to Neuromorphic computingOther practical applications such as magnetic memory and spintronics will benefit from the results of this study. The combination of magnetic memory and energy-efficient magnetic ions may be a possible way to reduce the operating energy of next-generation data storage media, but the magnetic ion mechanism that controls the antiferromagnetic layer is currently , Is a promising candidate for the development of spintronic devices.
Sofia Martins et al, Dynamic Electric Field-Induced Magnetic Effects on Cobalt Oxide Thin Films: Towards Magnetic Ion Synapses, Nanoscale (2021). DOI: 10.1039 / D1NR06210G
Barcelona Autonomous University
Quote: Researchers used a magnetic system to acquire brain learning functions from https: //phys.org/news/2022-02-magnetic-artificially-functions-brain.html on February 21, 2022. And the oblivion function (February 21, 2022) is artificially reproduced.
This document is subject to copyright. No part may be reproduced without written permission, except for fair transactions for personal investigation or research purposes. Content is provided for informational purposes only.
Researchers use magnetic systems to artificially reproduce the learning and forgetting functions of the brain
Source link Researchers use magnetic systems to artificially reproduce the learning and forgetting functions of the brain