Neuromorphic design using reward-based STDP learning on event-based reconfigurable cluster architecture
Neuromorphic design using reward-based STDP learning on event-based reconfigurable cluster architecture
Neuromorphic computing systems simulate spiking neural networks that are used for research into how biological neural networks function, as well as for applied engineering such as robotics, pattern recognition, and machine learning. In this paper, we present a neuromorphic system based on an asynchronous event-based hardware platform. We represent three algorithms for implementing spiking networks on our asynchronous hardware platform. We also discuss different trade-offs between synchronisation and messaging costs. A reinforcement learning method known as Reward-modulated STDP is presented as an online learning algorithm in the network. We evaluate the system performance in a single box of our designed architecture using 6000 concurrent hardware threads and demonstrate scaling to networks with up to 2 million neurons and 400 million synapses. The performance of our architecture is also compared to existing neuromorphic platforms, showing a 20 times speed-up over the Brian simulator on an x86 machine, and a 16 times speed-up over a 48-chip SpiNNaker node.
Neuromorphic system, Reconfigurable architecture., Reinforcement Reward-modulated STDP, Spiking neural network simulation
Association for Computing Machinery
Shahsavari, Mahyar
a120fae7-9361-4f5d-ad6a-60f50f84da34
Thomas, David
5701997d-7de3-4e57-a802-ea2bd3e6ab6c
Brown, Andrew
5c19e523-65ec-499b-9e7c-91522017d7e0
Luk, Wayne
ea937a29-564d-4b87-8570-a2c284f956c6
Shahsavari, Mahyar
a120fae7-9361-4f5d-ad6a-60f50f84da34
Thomas, David
5701997d-7de3-4e57-a802-ea2bd3e6ab6c
Brown, Andrew
5c19e523-65ec-499b-9e7c-91522017d7e0
Luk, Wayne
ea937a29-564d-4b87-8570-a2c284f956c6
Shahsavari, Mahyar, Thomas, David, Brown, Andrew and Luk, Wayne
(2021)
Neuromorphic design using reward-based STDP learning on event-based reconfigurable cluster architecture.
In ICONS 2021 - Proceedings of International Conference on Neuromorphic Systems 2021.
Association for Computing Machinery.
8 pp
.
(doi:10.1145/3477145.3477151).
Record type:
Conference or Workshop Item
(Paper)
Abstract
Neuromorphic computing systems simulate spiking neural networks that are used for research into how biological neural networks function, as well as for applied engineering such as robotics, pattern recognition, and machine learning. In this paper, we present a neuromorphic system based on an asynchronous event-based hardware platform. We represent three algorithms for implementing spiking networks on our asynchronous hardware platform. We also discuss different trade-offs between synchronisation and messaging costs. A reinforcement learning method known as Reward-modulated STDP is presented as an online learning algorithm in the network. We evaluate the system performance in a single box of our designed architecture using 6000 concurrent hardware threads and demonstrate scaling to networks with up to 2 million neurons and 400 million synapses. The performance of our architecture is also compared to existing neuromorphic platforms, showing a 20 times speed-up over the Brian simulator on an x86 machine, and a 16 times speed-up over a 48-chip SpiNNaker node.
Text
3477145.3477151
- Version of Record
Restricted to Repository staff only
Request a copy
More information
e-pub ahead of print date: 27 July 2021
Additional Information:
Funding Information:
Thanks to Jonathan Beaumont and Matthew Naylor for their technical supports. This work is sponsored by UK EPSRC grant EP/N031768/1 (POETS project).
Publisher Copyright:
© 2021 ACM.
Venue - Dates:
2021 International Conference on Neuromorphic Systems, ICONS 2021, , Virtual, Onlie, United States, 2021-07-27 - 2021-07-29
Keywords:
Neuromorphic system, Reconfigurable architecture., Reinforcement Reward-modulated STDP, Spiking neural network simulation
Identifiers
Local EPrints ID: 468747
URI: http://eprints.soton.ac.uk/id/eprint/468747
PURE UUID: aa33e35c-2d50-45eb-a5f2-18e5ae9364a5
Catalogue record
Date deposited: 24 Aug 2022 16:39
Last modified: 18 Mar 2024 04:04
Export record
Altmetrics
Contributors
Author:
Mahyar Shahsavari
Author:
David Thomas
Author:
Andrew Brown
Author:
Wayne Luk
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics