Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/33511
Title: Localized generalization error of Gaussian-based classifiers and visualization of decision boundaries
Authors: Ng, WWY
Yeung, DS
Wang, D
Tsang, ECC
Wang, XZ
Keywords: Decision boundary visualization
Generalization error
Most sensitive vector
Radial basis function neural network
Stochastic sensitivity measure
Support vector
Support vector machine with gaussian kernel
Issue Date: 2007
Publisher: Springer
Source: Soft computing, 2007, v. 11, no. 4, p. 375-381 How to cite?
Journal: Soft computing 
Abstract: In pattern classification problem, one trains a classifier to recognize future unseen samples using a training dataset. Practically, one should not expect the trained classifier could correctly recognize samples dissimilar to the training dataset. Therefore, finding the generalization capability of a classifier for those unseen samples may not help in improving the classifiers accuracy. The localized generalization error model was proposed to bound above the generalization mean square error for those unseen samples similar to the training dataset only. This error model is derived based on the stochastic sensitivity measure(ST-SM)of the classifiers. We present the ST-SMS for various Gaussian based classifiers: radial basis function neural networks and support vector machine in this paper. At the end of this work, we compare the decision boundaries visualization using the training samples yielding the largest sensitivity measures and the one using support vectors in the input space.
URI: http://hdl.handle.net/10397/33511
ISSN: 1432-7643
DOI: 10.1007/s00500-006-0092-4
Appears in Collections:Journal/Magazine Article

Access
View full-text via PolyU eLinks SFX Query
Show full item record

SCOPUSTM   
Citations

6
Last Week
0
Last month
0
Citations as of Aug 17, 2017

WEB OF SCIENCETM
Citations

4
Last Week
0
Last month
0
Citations as of Aug 13, 2017

Page view(s)

33
Last Week
1
Last month
Checked on Aug 13, 2017

Google ScholarTM

Check

Altmetric



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.