Abstract
Neural networks that are used in the classification of very-high-dimensional remotely sensed data are discussed in comparison to statistical classification methods. Both two-layer and three-layer iterative neural networks are used in experiments together with a parallel hierarchical neural network. The statistical methods applied include the maximum likelihood method, the minimum Euclidian distance, and two pooling methods (statistical multisource classification and the linear pool). The data used in experiments are simulated High-Resolution Imaging Spectrometer (HIRIS) data. All the methods are compared based on classification performance with different numbers of features, different numbers of training samples, speed (CPU time), and classification accuracy for training and test data. Statistical methods are shown to yield performance superior to that of the neural network methods.
Original language | English |
---|---|
Pages | 1269-1272 |
Number of pages | 4 |
Publication status | Published - 1990 Dec 1 |
Event | 10th Annual International Geoscience and Remote Sensing Symposium - IGARSS '90 - College Park, MD, USA Duration: 1990 May 20 → 1990 May 20 |
Other
Other | 10th Annual International Geoscience and Remote Sensing Symposium - IGARSS '90 |
---|---|
City | College Park, MD, USA |
Period | 90/5/20 → 90/5/20 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Computer Science Applications
- Earth and Planetary Sciences(all)
Cite this
}
Classification of very high dimensional data using neural networks. / Benediktsson, J. A.; Swain, P. H.; Ersoy, O. K.; Hong, D.
1990. 1269-1272 Paper presented at 10th Annual International Geoscience and Remote Sensing Symposium - IGARSS '90, College Park, MD, USA, .Research output: Contribution to conference › Paper
TY - CONF
T1 - Classification of very high dimensional data using neural networks
AU - Benediktsson, J. A.
AU - Swain, P. H.
AU - Ersoy, O. K.
AU - Hong, D.
PY - 1990/12/1
Y1 - 1990/12/1
N2 - Neural networks that are used in the classification of very-high-dimensional remotely sensed data are discussed in comparison to statistical classification methods. Both two-layer and three-layer iterative neural networks are used in experiments together with a parallel hierarchical neural network. The statistical methods applied include the maximum likelihood method, the minimum Euclidian distance, and two pooling methods (statistical multisource classification and the linear pool). The data used in experiments are simulated High-Resolution Imaging Spectrometer (HIRIS) data. All the methods are compared based on classification performance with different numbers of features, different numbers of training samples, speed (CPU time), and classification accuracy for training and test data. Statistical methods are shown to yield performance superior to that of the neural network methods.
AB - Neural networks that are used in the classification of very-high-dimensional remotely sensed data are discussed in comparison to statistical classification methods. Both two-layer and three-layer iterative neural networks are used in experiments together with a parallel hierarchical neural network. The statistical methods applied include the maximum likelihood method, the minimum Euclidian distance, and two pooling methods (statistical multisource classification and the linear pool). The data used in experiments are simulated High-Resolution Imaging Spectrometer (HIRIS) data. All the methods are compared based on classification performance with different numbers of features, different numbers of training samples, speed (CPU time), and classification accuracy for training and test data. Statistical methods are shown to yield performance superior to that of the neural network methods.
UR - http://www.scopus.com/inward/record.url?scp=0025530719&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0025530719&partnerID=8YFLogxK
M3 - Paper
AN - SCOPUS:0025530719
SP - 1269
EP - 1272
ER -