A New GPU Implementation of Support Vector Machines for Fast Hyperspectral Image Classification
The storage and processing of remotely sensed hyperspectral images (HSIs) is facing important challenges due to the computational requirements involved in the analysis of these images, characterized by continuous and narrow spectral channels. Although HSIs offer many opportunities for accurately mod...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-04-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/12/8/1257 |
| _version_ | 1827718531655925760 |
|---|---|
| author | Mercedes E. Paoletti Juan M. Haut Xuanwen Tao Javier Plaza Miguel Antonio Plaza |
| author_facet | Mercedes E. Paoletti Juan M. Haut Xuanwen Tao Javier Plaza Miguel Antonio Plaza |
| author_sort | Mercedes E. Paoletti |
| collection | DOAJ |
| description | The storage and processing of remotely sensed hyperspectral images (HSIs) is facing important challenges due to the computational requirements involved in the analysis of these images, characterized by continuous and narrow spectral channels. Although HSIs offer many opportunities for accurately modeling and mapping the surface of the Earth in a wide range of applications, they comprise massive data cubes. These huge amounts of data impose important requirements from the storage and processing points of view. The support vector machine (SVM) has been one of the most powerful machine learning classifiers, able to process HSI data without applying previous feature extraction steps, exhibiting a robust behaviour with high dimensional data and obtaining high classification accuracies. Nevertheless, the training and prediction stages of this supervised classifier are very time-consuming, especially for large and complex problems that require an intensive use of memory and computational resources. This paper develops a new, highly efficient implementation of SVMs that exploits the high computational power of graphics processing units (GPUs) to reduce the execution time by massively parallelizing the operations of the algorithm while performing efficient memory management during data-reading and writing instructions. Our experiments, conducted over different HSI benchmarks, demonstrate the efficiency of our GPU implementation. |
| first_indexed | 2024-03-10T20:25:03Z |
| format | Article |
| id | doaj.art-53436f6834df499fb07671d018131d28 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T20:25:03Z |
| publishDate | 2020-04-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-53436f6834df499fb07671d018131d282023-11-19T21:50:22ZengMDPI AGRemote Sensing2072-42922020-04-01128125710.3390/rs12081257A New GPU Implementation of Support Vector Machines for Fast Hyperspectral Image ClassificationMercedes E. Paoletti0Juan M. Haut1Xuanwen Tao2Javier Plaza Miguel3Antonio Plaza4Hyperspectral Computing Laboratory (HyperComp), Department of Computer Technology and Communications. Escuela Politecnica de Caceres, University of Extremadura, Avenida de la Universidad sn, E-10002 Caceres, SpainHyperspectral Computing Laboratory (HyperComp), Department of Computer Technology and Communications. Escuela Politecnica de Caceres, University of Extremadura, Avenida de la Universidad sn, E-10002 Caceres, SpainHyperspectral Computing Laboratory (HyperComp), Department of Computer Technology and Communications. Escuela Politecnica de Caceres, University of Extremadura, Avenida de la Universidad sn, E-10002 Caceres, SpainHyperspectral Computing Laboratory (HyperComp), Department of Computer Technology and Communications. Escuela Politecnica de Caceres, University of Extremadura, Avenida de la Universidad sn, E-10002 Caceres, SpainHyperspectral Computing Laboratory (HyperComp), Department of Computer Technology and Communications. Escuela Politecnica de Caceres, University of Extremadura, Avenida de la Universidad sn, E-10002 Caceres, SpainThe storage and processing of remotely sensed hyperspectral images (HSIs) is facing important challenges due to the computational requirements involved in the analysis of these images, characterized by continuous and narrow spectral channels. Although HSIs offer many opportunities for accurately modeling and mapping the surface of the Earth in a wide range of applications, they comprise massive data cubes. These huge amounts of data impose important requirements from the storage and processing points of view. The support vector machine (SVM) has been one of the most powerful machine learning classifiers, able to process HSI data without applying previous feature extraction steps, exhibiting a robust behaviour with high dimensional data and obtaining high classification accuracies. Nevertheless, the training and prediction stages of this supervised classifier are very time-consuming, especially for large and complex problems that require an intensive use of memory and computational resources. This paper develops a new, highly efficient implementation of SVMs that exploits the high computational power of graphics processing units (GPUs) to reduce the execution time by massively parallelizing the operations of the algorithm while performing efficient memory management during data-reading and writing instructions. Our experiments, conducted over different HSI benchmarks, demonstrate the efficiency of our GPU implementation.https://www.mdpi.com/2072-4292/12/8/1257hyperspectral images (HSIs)support vector machines (SVMs)graphics processing units (GPUs)hardware parallelization |
| spellingShingle | Mercedes E. Paoletti Juan M. Haut Xuanwen Tao Javier Plaza Miguel Antonio Plaza A New GPU Implementation of Support Vector Machines for Fast Hyperspectral Image Classification Remote Sensing hyperspectral images (HSIs) support vector machines (SVMs) graphics processing units (GPUs) hardware parallelization |
| title | A New GPU Implementation of Support Vector Machines for Fast Hyperspectral Image Classification |
| title_full | A New GPU Implementation of Support Vector Machines for Fast Hyperspectral Image Classification |
| title_fullStr | A New GPU Implementation of Support Vector Machines for Fast Hyperspectral Image Classification |
| title_full_unstemmed | A New GPU Implementation of Support Vector Machines for Fast Hyperspectral Image Classification |
| title_short | A New GPU Implementation of Support Vector Machines for Fast Hyperspectral Image Classification |
| title_sort | new gpu implementation of support vector machines for fast hyperspectral image classification |
| topic | hyperspectral images (HSIs) support vector machines (SVMs) graphics processing units (GPUs) hardware parallelization |
| url | https://www.mdpi.com/2072-4292/12/8/1257 |
| work_keys_str_mv | AT mercedesepaoletti anewgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification AT juanmhaut anewgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification AT xuanwentao anewgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification AT javierplazamiguel anewgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification AT antonioplaza anewgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification AT mercedesepaoletti newgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification AT juanmhaut newgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification AT xuanwentao newgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification AT javierplazamiguel newgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification AT antonioplaza newgpuimplementationofsupportvectormachinesforfasthyperspectralimageclassification |