DTS-SNN: Spiking Neural Networks With Dynamic Time-Surfaces
Convolution helps spiking neural networks (SNNs) capture the spatio-temporal structures of neuromorphic (event) data as evident in the convolution-based SNNs (C-SNNs) with the state-of-the-art classification-accuracies on various datasets. However, the efficacy aside, the efficiency of C-SNN is ques...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9903429/ |
| _version_ | 1811205663434473472 |
|---|---|
| author | Donghyung Yoo Doo Seok Jeong |
| author_facet | Donghyung Yoo Doo Seok Jeong |
| author_sort | Donghyung Yoo |
| collection | DOAJ |
| description | Convolution helps spiking neural networks (SNNs) capture the spatio-temporal structures of neuromorphic (event) data as evident in the convolution-based SNNs (C-SNNs) with the state-of-the-art classification-accuracies on various datasets. However, the efficacy aside, the efficiency of C-SNN is questionable. In this regard, we propose SNNs with novel trainable dynamic time-surfaces (DTS-SNNs) as efficient alternatives to convolution. The novel dynamic time-surface proposed in this work features its high responsiveness to moving objects given the use of the zero-sum temporal kernel that is motivated by the simple cells’ receptive fields in the early stage visual pathway. We evaluated the performance and computational complexity of our DTS-SNNs on three real-world event-based datasets (DVS128 Gesture, Spiking Heidelberg dataset, N-Cars). The results highlight high classification accuracies and significant improvements in computational efficiency, e.g., merely 1.51% behind of the state-of-the-art result on DVS128 Gesture but a <inline-formula> <tex-math notation="LaTeX">$\times 18$ </tex-math></inline-formula> improvement in efficiency. The code is available online (<uri>https://github.com/dooseokjeong/DTS-SNN</uri>). |
| first_indexed | 2024-04-12T03:34:48Z |
| format | Article |
| id | doaj.art-c1665df696ce4070bf5abcd5be5fb837 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-12T03:34:48Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-c1665df696ce4070bf5abcd5be5fb8372022-12-22T03:49:27ZengIEEEIEEE Access2169-35362022-01-011010265910266810.1109/ACCESS.2022.32096719903429DTS-SNN: Spiking Neural Networks With Dynamic Time-SurfacesDonghyung Yoo0https://orcid.org/0000-0003-0857-292XDoo Seok Jeong1https://orcid.org/0000-0001-7954-2213Division of Materials Science and Engineering, Hanyang University, Seoul, South KoreaDivision of Materials Science and Engineering, Hanyang University, Seoul, South KoreaConvolution helps spiking neural networks (SNNs) capture the spatio-temporal structures of neuromorphic (event) data as evident in the convolution-based SNNs (C-SNNs) with the state-of-the-art classification-accuracies on various datasets. However, the efficacy aside, the efficiency of C-SNN is questionable. In this regard, we propose SNNs with novel trainable dynamic time-surfaces (DTS-SNNs) as efficient alternatives to convolution. The novel dynamic time-surface proposed in this work features its high responsiveness to moving objects given the use of the zero-sum temporal kernel that is motivated by the simple cells’ receptive fields in the early stage visual pathway. We evaluated the performance and computational complexity of our DTS-SNNs on three real-world event-based datasets (DVS128 Gesture, Spiking Heidelberg dataset, N-Cars). The results highlight high classification accuracies and significant improvements in computational efficiency, e.g., merely 1.51% behind of the state-of-the-art result on DVS128 Gesture but a <inline-formula> <tex-math notation="LaTeX">$\times 18$ </tex-math></inline-formula> improvement in efficiency. The code is available online (<uri>https://github.com/dooseokjeong/DTS-SNN</uri>).https://ieeexplore.ieee.org/document/9903429/Lightweight spiking neural networkspiking neural networkdynamic time-surfacesevent-based data |
| spellingShingle | Donghyung Yoo Doo Seok Jeong DTS-SNN: Spiking Neural Networks With Dynamic Time-Surfaces IEEE Access Lightweight spiking neural network spiking neural network dynamic time-surfaces event-based data |
| title | DTS-SNN: Spiking Neural Networks With Dynamic Time-Surfaces |
| title_full | DTS-SNN: Spiking Neural Networks With Dynamic Time-Surfaces |
| title_fullStr | DTS-SNN: Spiking Neural Networks With Dynamic Time-Surfaces |
| title_full_unstemmed | DTS-SNN: Spiking Neural Networks With Dynamic Time-Surfaces |
| title_short | DTS-SNN: Spiking Neural Networks With Dynamic Time-Surfaces |
| title_sort | dts snn spiking neural networks with dynamic time surfaces |
| topic | Lightweight spiking neural network spiking neural network dynamic time-surfaces event-based data |
| url | https://ieeexplore.ieee.org/document/9903429/ |
| work_keys_str_mv | AT donghyungyoo dtssnnspikingneuralnetworkswithdynamictimesurfaces AT dooseokjeong dtssnnspikingneuralnetworkswithdynamictimesurfaces |