Choosing the kernel parameters for support vector machines by the inter-cluster distance in the feature space
Journal
Pattern Recognition
Journal Volume
42
Journal Issue
5
Pages
710-717
Date Issued
2009-05
Author(s)
Kuo-Ping Wu
Abstract
Determining the kernel and error penalty parameters for support vector machines (SVMs) is very problem-dependent in practice. A popular method to deciding the kernel parameters is the grid search method. In the training process, classifiers are trained with different kernel parameters, and only one of the classifiers is required for the testing process. This makes the training process time-consuming. In this paper we propose using the inter-cluster distances in the feature spaces to choose the kernel parameters. Calculating such distance costs much less computation time than training the corresponding SVM classifiers; thus the proper kernel parameters can be chosen much faster. Experiment results show that the inter-cluster distance can choose proper kernel parameters with which the testing accuracy of trained SVMs is competitive to the standard ones, and the training time can be significantly shortened. © 2008 Elsevier Ltd. All rights reserved.
Subjects
Inter-cluster distances; Kernel parameters; Support vector machines; SVM
Other Subjects
Classifiers; Image retrieval; Learning systems; Vectors; Computation times; Feature spaces; Grid searches; Inter-cluster distances; Kernel parameters; Penalty parameters; SVM; Svm classifiers; Testing accuracies; Testing processes; Training processes; Training times; Support vector machines
Type
journal article
File(s)![Thumbnail Image]()
Loading...
Name
38.pdf
Size
228.55 KB
Format
Adobe PDF
Checksum
(MD5):0bbde972d6ee3d6b7654fa3f7184c906