Pattern Selection for Support Vector Classifiers
2002
Conference Paper
ei
SVMs tend to take a very long time to train with a large data set. If "redundant" patterns are identified and deleted in pre-processing, the training time could be reduced significantly. We propose a k-nearest neighbors(k-NN) based pattern selection method. The method tries to select the patterns that are near the decision boundary and that are correctly labeled. The simulations over synthetic data sets showed promising results: (1) By converting a non-separable problem to a separable one, the search for an optimal error tolerance parameter became unnecessary. (2) SVM training time decreased by two orders of magnitude without any loss of accuracy. (3) The redundant SVs were substantially reduced.
Author(s): | Shin, H. and Cho, S. |
Book Title: | Ideal 2002 |
Journal: | Intelligent Data Engineering and Automated Learning (IDEAL 2002) |
Pages: | 97-103 |
Year: | 2002 |
Month: | January |
Day: | 0 |
Editors: | Yin, H. , N. Allinson, R. Freeman, J. Keane, S. Hubbard |
Publisher: | Springer |
Department(s): | Empirische Inferenz |
Bibtex Type: | Conference Paper (inproceedings) |
DOI: | 10.1007/3-540-45675-9_70 |
Event Name: | Third International Conference on Intelligent Data Engineering and Automated Learning |
Event Place: | Manchester, United Kingdom |
Address: | Berlin, Germany |
Digital: | 0 |
Institution: | Seoul National University, Seoul, Korea |
Language: | en |
Organization: | Max-Planck-Gesellschaft |
School: | Biologische Kybernetik |
Links: |
PDF
Web |
BibTex @inproceedings{2692, title = {Pattern Selection for Support Vector Classifiers}, author = {Shin, H. and Cho, S.}, journal = {Intelligent Data Engineering and Automated Learning (IDEAL 2002)}, booktitle = {Ideal 2002}, pages = {97-103}, editors = {Yin, H. , N. Allinson, R. Freeman, J. Keane, S. Hubbard}, publisher = {Springer}, organization = {Max-Planck-Gesellschaft}, institution = {Seoul National University, Seoul, Korea}, school = {Biologische Kybernetik}, address = {Berlin, Germany}, month = jan, year = {2002}, doi = {10.1007/3-540-45675-9_70}, month_numeric = {1} } |