Boundary-Aware Deformable Spiking Neural Network for Hyperspectral Image Classification
A few spiking neural network (SNN)-based classifiers have been proposed for hyperspectral images (HSI) classification to alleviate the higher computational energy cost problem. Nevertheless, due to the lack of ability to distinguish boundaries, the existing SNN-based HSI classification methods are v...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-10-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/20/5020 |
| _version_ | 1797572373561475072 |
|---|---|
| author | Shuo Wang Yuanxi Peng Lei Wang Teng Li |
| author_facet | Shuo Wang Yuanxi Peng Lei Wang Teng Li |
| author_sort | Shuo Wang |
| collection | DOAJ |
| description | A few spiking neural network (SNN)-based classifiers have been proposed for hyperspectral images (HSI) classification to alleviate the higher computational energy cost problem. Nevertheless, due to the lack of ability to distinguish boundaries, the existing SNN-based HSI classification methods are very prone to falling into the Hughes phenomenon. The confusion of the classifier at the class boundary is particularly obvious. To remedy these issues, we propose a boundary-aware deformable spiking residual neural network (BDSNN) for HSI classification. A deformable convolution neural network plays the most important role in realizing the boundary-awareness of the proposed model. To the best of our knowledge, this is the first attempt to combine the deformable convolutional mechanism and the SNN-based model. Additionally, spike-element-wise ResNet is used as a fundamental framework for going deeper. A temporal channel joint attention mechanism is introduced to filter out which channels and times are critical. We evaluate the proposed model on four benchmark hyperspectral data sets—the IP, PU, SV, and HU data sets. The experimental results demonstrate that the proposed model can obtain a comparable classification accuracy with state-of-the-art methods in terms of overall accuracy (OA), average accuracy (AA), and statistical kappa (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>) coefficient. The ablation study results prove the effectiveness of the introduction of the deformable convolutional mechanism for BDSNN’s boundary-aware characteristic. |
| first_indexed | 2024-03-10T20:55:21Z |
| format | Article |
| id | doaj.art-171804e4a73940d689aa5d419b8400ae |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T20:55:21Z |
| publishDate | 2023-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-171804e4a73940d689aa5d419b8400ae2023-11-19T17:59:43ZengMDPI AGRemote Sensing2072-42922023-10-011520502010.3390/rs15205020Boundary-Aware Deformable Spiking Neural Network for Hyperspectral Image ClassificationShuo Wang0Yuanxi Peng1Lei Wang2Teng Li3State Key Laboratory of High-Performance Computing, College of Computer Science, National University of Defense Technology, Changsha 410073, ChinaState Key Laboratory of High-Performance Computing, College of Computer Science, National University of Defense Technology, Changsha 410073, ChinaCollege of Computer, National University of Defense Technology, Changsha 410073, ChinaCollege of Advanced Interdisciplinary Studies, National University of Defense Technology, Changsha 410073, ChinaA few spiking neural network (SNN)-based classifiers have been proposed for hyperspectral images (HSI) classification to alleviate the higher computational energy cost problem. Nevertheless, due to the lack of ability to distinguish boundaries, the existing SNN-based HSI classification methods are very prone to falling into the Hughes phenomenon. The confusion of the classifier at the class boundary is particularly obvious. To remedy these issues, we propose a boundary-aware deformable spiking residual neural network (BDSNN) for HSI classification. A deformable convolution neural network plays the most important role in realizing the boundary-awareness of the proposed model. To the best of our knowledge, this is the first attempt to combine the deformable convolutional mechanism and the SNN-based model. Additionally, spike-element-wise ResNet is used as a fundamental framework for going deeper. A temporal channel joint attention mechanism is introduced to filter out which channels and times are critical. We evaluate the proposed model on four benchmark hyperspectral data sets—the IP, PU, SV, and HU data sets. The experimental results demonstrate that the proposed model can obtain a comparable classification accuracy with state-of-the-art methods in terms of overall accuracy (OA), average accuracy (AA), and statistical kappa (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>) coefficient. The ablation study results prove the effectiveness of the introduction of the deformable convolutional mechanism for BDSNN’s boundary-aware characteristic.https://www.mdpi.com/2072-4292/15/20/5020deformable convolution neural networkstemporal channel joint attention (TCJA)spiking neural networks (SNNs)hyperspectral image (HSI) classificationresidual network (ResNet) |
| spellingShingle | Shuo Wang Yuanxi Peng Lei Wang Teng Li Boundary-Aware Deformable Spiking Neural Network for Hyperspectral Image Classification Remote Sensing deformable convolution neural networks temporal channel joint attention (TCJA) spiking neural networks (SNNs) hyperspectral image (HSI) classification residual network (ResNet) |
| title | Boundary-Aware Deformable Spiking Neural Network for Hyperspectral Image Classification |
| title_full | Boundary-Aware Deformable Spiking Neural Network for Hyperspectral Image Classification |
| title_fullStr | Boundary-Aware Deformable Spiking Neural Network for Hyperspectral Image Classification |
| title_full_unstemmed | Boundary-Aware Deformable Spiking Neural Network for Hyperspectral Image Classification |
| title_short | Boundary-Aware Deformable Spiking Neural Network for Hyperspectral Image Classification |
| title_sort | boundary aware deformable spiking neural network for hyperspectral image classification |
| topic | deformable convolution neural networks temporal channel joint attention (TCJA) spiking neural networks (SNNs) hyperspectral image (HSI) classification residual network (ResNet) |
| url | https://www.mdpi.com/2072-4292/15/20/5020 |
| work_keys_str_mv | AT shuowang boundaryawaredeformablespikingneuralnetworkforhyperspectralimageclassification AT yuanxipeng boundaryawaredeformablespikingneuralnetworkforhyperspectralimageclassification AT leiwang boundaryawaredeformablespikingneuralnetworkforhyperspectralimageclassification AT tengli boundaryawaredeformablespikingneuralnetworkforhyperspectralimageclassification |