Abstract
A feature extraction method based on decision boundaries has been proposed for neural networks. The method is based on the fact that normal vectors to the decision boundary provide the information necessary for discriminating between classes. However, it is observed that the previous implementation of numerical approximation of the gradient has resulted in some performance loss and a long processing time. In this paper, we propose a new method to calculate normal vectors analytically for neural networks with multiple hidden layers. Experiments showed noticeable improvements in performance and speed.
This work was supported in part by Biometrics Engineering Research Center (KOSEF).
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Richards, J.A.: Remote Sensing Digital Image Analysis. Springer, Heidelberg (1993)
Lee, C., Landgrebe, D.A.: Feature extraction based on decision boundaries. IEEE Trans. on Pattern Analysis and Machine Intelligence 15, 388–400 (1993)
Lee, C., Landgrebe, D.A.: Decision boundary feature extraction for neural networks. IEEE Trans. Neural Networks 8, 75–83 (1997)
Lee, C., Landgrebe, D.A.: Decision Boundary Feature Extraction for Non-Parametric Classification. IEEE Trans. on Systems, Man, and Cybernetics 23(2), 433–444 (1993)
Biehl, L.L., et al.: A crops and soils data base for scene radiation research. In: Proc. Machine Process. of Remotely Sensed Data Symp., West Lafayette, Indiana, pp. 169–177 (1982)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2003 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Go, J., Lee, C. (2003). Analytical Decision Boundary Feature Extraction for Neural Networks with Multiple Hidden Layers. In: Petkov, N., Westenberg, M.A. (eds) Computer Analysis of Images and Patterns. CAIP 2003. Lecture Notes in Computer Science, vol 2756. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-45179-2_71
Download citation
DOI: https://doi.org/10.1007/978-3-540-45179-2_71
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-40730-0
Online ISBN: 978-3-540-45179-2
eBook Packages: Springer Book Archive