Energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse
Neuromorphic computing (NC) is considered a potential solution for energy-efficient artificial intelligence applications. The development of reliable neural network (NN) hardware with low energy and area footprints plays a crucial role in realizing NC. Even though neurons and synapses have already b...
Saved in:
Main Authors: | , , , , , , |
---|---|
Other Authors: | |
Format: | Article |
Language: | English |
Published: |
2024
|
Subjects: | |
Online Access: | https://hdl.handle.net/10356/180493 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
id |
sg-ntu-dr.10356-180493 |
---|---|
record_format |
dspace |
spelling |
sg-ntu-dr.10356-1804932024-10-14T15:35:18Z Energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse Dhull, Seema Mah, William Wai Lum Nisar, Arshid Kumar, Durgesh Rahaman, Hasibur Kaushik, Brajesh Kumar Piramanayagam, S. N. School of Physical and Mathematical Sciences Physics Anisotropy field Domain wall devices Neuromorphic computing (NC) is considered a potential solution for energy-efficient artificial intelligence applications. The development of reliable neural network (NN) hardware with low energy and area footprints plays a crucial role in realizing NC. Even though neurons and synapses have already been investigated using a variety of spintronic devices, the research is still in the primitive stages. Particularly, there is not much experimental research on the self-reset (and leaky) aspect(s) of domain wall (DW) device-based neurons. Here, we have demonstrated an energy-efficient NN using a spintronic DW device-based neuron with self-reset (leaky) and integrate-and-fire functions. An “anisotropy field gradient” provides the self-resetting behavior of auto-leaky, integrate, and fire neurons. The leaky property of the neuron was experimentally demonstrated using a voltage-assisted modification of the anisotropy field. A synapse with a meander wire configuration was used to achieve multiple-resistance states corresponding to the DW position and controlled pinning of the DW. The NN showed an energy efficiency of 0.189 nJ/image/epoch while achieving an accuracy of 92.4%. This study provides a fresh path for developing more energy-efficient DW-based NN systems. National Research Foundation (NRF) Published version The authors gratefully acknowledge the National Research Foundation (NRF), Singapore for the NRF-CRP (No. NRF-CRP21- 2018-0003) grant. 2024-10-09T04:26:05Z 2024-10-09T04:26:05Z 2024 Journal Article Dhull, S., Mah, W. W. L., Nisar, A., Kumar, D., Rahaman, H., Kaushik, B. K. & Piramanayagam, S. N. (2024). Energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse. Applied Physics Letters, 125(1), 012402-. https://dx.doi.org/10.1063/5.0220809 0003-6951 https://hdl.handle.net/10356/180493 10.1063/5.0220809 2-s2.0-85197602880 1 125 012402 en NRF-CRP21-2018-0003 Applied Physics Letters © 2024 Author(s). Published under an exclusive license by AIP Publishing. All rights reserved. This article may be downloaded for personal use only. Any other use requires prior permission of the copyright holder. The Version of Record is available online at http://doi.org/10.1063/5.0220809 application/pdf |
institution |
Nanyang Technological University |
building |
NTU Library |
continent |
Asia |
country |
Singapore Singapore |
content_provider |
NTU Library |
collection |
DR-NTU |
language |
English |
topic |
Physics Anisotropy field Domain wall devices |
spellingShingle |
Physics Anisotropy field Domain wall devices Dhull, Seema Mah, William Wai Lum Nisar, Arshid Kumar, Durgesh Rahaman, Hasibur Kaushik, Brajesh Kumar Piramanayagam, S. N. Energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse |
description |
Neuromorphic computing (NC) is considered a potential solution for energy-efficient artificial intelligence applications. The development of reliable neural network (NN) hardware with low energy and area footprints plays a crucial role in realizing NC. Even though neurons and synapses have already been investigated using a variety of spintronic devices, the research is still in the primitive stages. Particularly, there is not much experimental research on the self-reset (and leaky) aspect(s) of domain wall (DW) device-based neurons. Here, we have demonstrated an energy-efficient NN using a spintronic DW device-based neuron with self-reset (leaky) and integrate-and-fire functions. An “anisotropy field gradient” provides the self-resetting behavior of auto-leaky, integrate, and fire neurons. The leaky property of the neuron was experimentally demonstrated using a voltage-assisted modification of the anisotropy field. A synapse with a meander wire configuration was used to achieve multiple-resistance states corresponding to the DW position and controlled pinning of the DW. The NN showed an energy efficiency of 0.189 nJ/image/epoch while achieving an accuracy of 92.4%. This study provides a fresh path for developing more energy-efficient DW-based NN systems. |
author2 |
School of Physical and Mathematical Sciences |
author_facet |
School of Physical and Mathematical Sciences Dhull, Seema Mah, William Wai Lum Nisar, Arshid Kumar, Durgesh Rahaman, Hasibur Kaushik, Brajesh Kumar Piramanayagam, S. N. |
format |
Article |
author |
Dhull, Seema Mah, William Wai Lum Nisar, Arshid Kumar, Durgesh Rahaman, Hasibur Kaushik, Brajesh Kumar Piramanayagam, S. N. |
author_sort |
Dhull, Seema |
title |
Energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse |
title_short |
Energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse |
title_full |
Energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse |
title_fullStr |
Energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse |
title_full_unstemmed |
Energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse |
title_sort |
energy-efficient neural network using an anisotropy field gradient-based self-resetting neuron and meander synapse |
publishDate |
2024 |
url |
https://hdl.handle.net/10356/180493 |
_version_ |
1814777738539565056 |