A burst-dependent algorithm for neuromorphic on-chip learning of spiking neural networks
The field of neuromorphic engineering addresses the high energy demands of neural networks through brain-inspired hardware for efficient neural network computing. For on-chip learning with spiking neural networks, neuromorphic hardware requires a local learning algorithm able to solve complex tasks....
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IOP Publishing
2025-01-01
|
| Series: | Neuromorphic Computing and Engineering |
| Subjects: | |
| Online Access: | https://doi.org/10.1088/2634-4386/adb511 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850192756384727040 |
|---|---|
| author | Michael Stuck Xingyun Wang Richard Naud |
| author_facet | Michael Stuck Xingyun Wang Richard Naud |
| author_sort | Michael Stuck |
| collection | DOAJ |
| description | The field of neuromorphic engineering addresses the high energy demands of neural networks through brain-inspired hardware for efficient neural network computing. For on-chip learning with spiking neural networks, neuromorphic hardware requires a local learning algorithm able to solve complex tasks. Approaches based on burst-dependent plasticity have been proposed to address this requirement, but their ability to learn complex tasks has remained unproven. Specifically, previous burst-dependent learning was demonstrated on a spiking version of the ‘exclusive or’ problem (XOR) using a network of thousands of neurons. Here, we extend burst-dependent learning, termed ‘Burstprop’, to address more complex tasks with hundreds of neurons. We evaluate Burstprop on a rate-encoded spiking version of the MNIST dataset, achieving low test classification errors, comparable to those obtained using backpropagation through time on the same architecture. Going further, we develop another burst-dependent algorithm based on the communication of two types of error-encoding events for the communication of positive and negative errors. We find that this new algorithm performs better on the image classification benchmark. We also tested our algorithms under various types of feedback connectivity, establishing that the capabilities of fixed random feedback connectivity is preserved in spiking neural networks. Lastly, we tested the robustness of the algorithm to weight discretization. Together, these results suggest that spiking Burstprop can scale to more complex learning tasks and is therefore likely to be considered for self-supervised algorithms while maintaining efficiency, potentially providing a viable method for learning with neuromorphic hardware. |
| format | Article |
| id | doaj-art-6c9a3d0b14b84bf58d5b4d4ed9572e24 |
| institution | OA Journals |
| issn | 2634-4386 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IOP Publishing |
| record_format | Article |
| series | Neuromorphic Computing and Engineering |
| spelling | doaj-art-6c9a3d0b14b84bf58d5b4d4ed9572e242025-08-20T02:14:26ZengIOP PublishingNeuromorphic Computing and Engineering2634-43862025-01-015101401010.1088/2634-4386/adb511A burst-dependent algorithm for neuromorphic on-chip learning of spiking neural networksMichael Stuck0https://orcid.org/0009-0007-2064-7097Xingyun Wang1https://orcid.org/0000-0002-6192-4456Richard Naud2https://orcid.org/0000-0001-7383-3095Department of Physics, University of Ottawa , 150 Louis Pasteur Pvt, Ottawa, ON, K1N 6N5, Canada; Centre for Neural Dynamics and Artificial Intelligence, University of Ottawa , 150 Louis Pasteur Pvt, Ottawa, ON, K1N 6N5, CanadaDepartment of Cellular and Molecular Medicine, University of Ottawa , 451 Smyth Rd, Ottawa, ON K1H 8M5, Canada; Centre for Neural Dynamics and Artificial Intelligence, University of Ottawa , 150 Louis Pasteur Pvt, Ottawa, ON, K1N 6N5, Canada; University of Ottawa Brain and Mind Research Institute University of Ottawa , 451 Smyth Rd, Ottawa, ON K1H 8M5, CanadaDepartment of Physics, University of Ottawa , 150 Louis Pasteur Pvt, Ottawa, ON, K1N 6N5, Canada; Department of Cellular and Molecular Medicine, University of Ottawa , 451 Smyth Rd, Ottawa, ON K1H 8M5, Canada; Centre for Neural Dynamics and Artificial Intelligence, University of Ottawa , 150 Louis Pasteur Pvt, Ottawa, ON, K1N 6N5, Canada; University of Ottawa Brain and Mind Research Institute University of Ottawa , 451 Smyth Rd, Ottawa, ON K1H 8M5, CanadaThe field of neuromorphic engineering addresses the high energy demands of neural networks through brain-inspired hardware for efficient neural network computing. For on-chip learning with spiking neural networks, neuromorphic hardware requires a local learning algorithm able to solve complex tasks. Approaches based on burst-dependent plasticity have been proposed to address this requirement, but their ability to learn complex tasks has remained unproven. Specifically, previous burst-dependent learning was demonstrated on a spiking version of the ‘exclusive or’ problem (XOR) using a network of thousands of neurons. Here, we extend burst-dependent learning, termed ‘Burstprop’, to address more complex tasks with hundreds of neurons. We evaluate Burstprop on a rate-encoded spiking version of the MNIST dataset, achieving low test classification errors, comparable to those obtained using backpropagation through time on the same architecture. Going further, we develop another burst-dependent algorithm based on the communication of two types of error-encoding events for the communication of positive and negative errors. We find that this new algorithm performs better on the image classification benchmark. We also tested our algorithms under various types of feedback connectivity, establishing that the capabilities of fixed random feedback connectivity is preserved in spiking neural networks. Lastly, we tested the robustness of the algorithm to weight discretization. Together, these results suggest that spiking Burstprop can scale to more complex learning tasks and is therefore likely to be considered for self-supervised algorithms while maintaining efficiency, potentially providing a viable method for learning with neuromorphic hardware.https://doi.org/10.1088/2634-4386/adb511spiking neural networksneuromorphic learningsupervised learningsynaptic plasticitylocal learning algorithmsburst-dependent learning |
| spellingShingle | Michael Stuck Xingyun Wang Richard Naud A burst-dependent algorithm for neuromorphic on-chip learning of spiking neural networks Neuromorphic Computing and Engineering spiking neural networks neuromorphic learning supervised learning synaptic plasticity local learning algorithms burst-dependent learning |
| title | A burst-dependent algorithm for neuromorphic on-chip learning of spiking neural networks |
| title_full | A burst-dependent algorithm for neuromorphic on-chip learning of spiking neural networks |
| title_fullStr | A burst-dependent algorithm for neuromorphic on-chip learning of spiking neural networks |
| title_full_unstemmed | A burst-dependent algorithm for neuromorphic on-chip learning of spiking neural networks |
| title_short | A burst-dependent algorithm for neuromorphic on-chip learning of spiking neural networks |
| title_sort | burst dependent algorithm for neuromorphic on chip learning of spiking neural networks |
| topic | spiking neural networks neuromorphic learning supervised learning synaptic plasticity local learning algorithms burst-dependent learning |
| url | https://doi.org/10.1088/2634-4386/adb511 |
| work_keys_str_mv | AT michaelstuck aburstdependentalgorithmforneuromorphiconchiplearningofspikingneuralnetworks AT xingyunwang aburstdependentalgorithmforneuromorphiconchiplearningofspikingneuralnetworks AT richardnaud aburstdependentalgorithmforneuromorphiconchiplearningofspikingneuralnetworks AT michaelstuck burstdependentalgorithmforneuromorphiconchiplearningofspikingneuralnetworks AT xingyunwang burstdependentalgorithmforneuromorphiconchiplearningofspikingneuralnetworks AT richardnaud burstdependentalgorithmforneuromorphiconchiplearningofspikingneuralnetworks |