Applications of artificial intelligence are growing significantly and are expected to become mainstream technologies in the near future. However, these applications run on conventional computer hardware and are extremely power hungry.
This creates opportunities for the development of new energy-efficient hardware solutions inspired by nature, for example brain-based computing. Some of the best-known examples are neuromorphic computing and neural networks that mimic the functioning of the human brain. One possible realization of such neural networks is to use an artificial spin ice (ASI) network. The UK National Physical Laboratory and partners investigated the impact of introducing hexagonal magnetic defects into such an ASI structure.
The research is published in the journal Communications equipment.
Through interdisciplinary research, the international team succeeded in demonstrating a mechanism to adapt the behavior of the ASI system by introducing the designed magnetic defects causing stochastic topological excitations in the system and controlling the dynamics of the neural networks based on the ‘ASI. The implications of this discovery are expected to be used in applications such as magnetic memory devices and spin-based logic applications.
The results of this study provide insight into collective and stochastically controlled behavior in artificial neural networks realized via the ASI magnetic network and pave the way for future research into emerging applications such as reconfigurable spin waveguides. and the hardware achievements of future low-power computing. systems.
NPL Fellow Olga Kazakova said: “This work marks a very important milestone for us: being able to controllably create topological states associated with ASI faults and demonstrate stochastic but statistically predictable behaviors within the ASI network. The results bring us closer to realizing energy-efficient neuromorphic computing. It was the result of a large international collaboration with major research facilities in the UK, Germany and France.
More information:
Robert Puttock et al, Stochastic hexagonal injectors in artificial spin ice, Communications equipment (2024). DOI: 10.1038/s43246-024-00614-0
Provided by the National Physical Laboratory
Quote: Hexagonal magnetic defects could lead to energy-efficient neuromorphic computing (October 3, 2024) retrieved October 3, 2024 from
This document is subject to copyright. Except for fair use for private study or research purposes, no part may be reproduced without written permission. The content is provided for informational purposes only.