Noise-based Local Learning using Stochastic Magnetic Tunnel Junctions

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Koenders, Kees, Schnitzpan, Leo, Kammerbauer, Fabian, Shu, Sinan, Jakob, Gerhard, Kläui, Mathis, Mentink, Johan, Ahmad, Nasir, van Gerven, Marcel
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908430270201856
author Koenders, Kees
Schnitzpan, Leo
Kammerbauer, Fabian
Shu, Sinan
Jakob, Gerhard
Kläui, Mathis
Mentink, Johan
Ahmad, Nasir
van Gerven, Marcel
author_facet Koenders, Kees
Schnitzpan, Leo
Kammerbauer, Fabian
Shu, Sinan
Jakob, Gerhard
Kläui, Mathis
Mentink, Johan
Ahmad, Nasir
van Gerven, Marcel
contents Brain-inspired learning in physical hardware has enormous potential to learn fast at minimal energy expenditure. One of the characteristics of biological learning systems is their ability to learn in the presence of various noise sources. Inspired by this observation, we introduce a novel noise-based learning approach for physical systems implementing multi-layer neural networks. Simulation results show that our approach allows for effective learning whose performance approaches that of the conventional effective yet energy-costly backpropagation algorithm. Using a spintronics hardware implementation, we demonstrate experimentally that learning can be achieved in a small network composed of physical stochastic magnetic tunnel junctions. These results provide a path towards efficient learning in general physical systems which embraces rather than mitigates the noise inherent in physical devices.
format Preprint
id arxiv_https___arxiv_org_abs_2412_12783
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Noise-based Local Learning using Stochastic Magnetic Tunnel Junctions
Koenders, Kees
Schnitzpan, Leo
Kammerbauer, Fabian
Shu, Sinan
Jakob, Gerhard
Kläui, Mathis
Mentink, Johan
Ahmad, Nasir
van Gerven, Marcel
Emerging Technologies
Mesoscale and Nanoscale Physics
Machine Learning
Brain-inspired learning in physical hardware has enormous potential to learn fast at minimal energy expenditure. One of the characteristics of biological learning systems is their ability to learn in the presence of various noise sources. Inspired by this observation, we introduce a novel noise-based learning approach for physical systems implementing multi-layer neural networks. Simulation results show that our approach allows for effective learning whose performance approaches that of the conventional effective yet energy-costly backpropagation algorithm. Using a spintronics hardware implementation, we demonstrate experimentally that learning can be achieved in a small network composed of physical stochastic magnetic tunnel junctions. These results provide a path towards efficient learning in general physical systems which embraces rather than mitigates the noise inherent in physical devices.
title Noise-based Local Learning using Stochastic Magnetic Tunnel Junctions
topic Emerging Technologies
Mesoscale and Nanoscale Physics
Machine Learning
url https://arxiv.org/abs/2412.12783