ResearchSpace

Towards understanding the influence of SVM hyperparameters

Show simple item record

dc.contributor.author Van Heerden, CJ
dc.contributor.author Barnard, E
dc.date.accessioned 2010-12-14T13:38:42Z
dc.date.available 2010-12-14T13:38:42Z
dc.date.issued 2010-11
dc.identifier.citation Van Heerden, CJ and Barnard, E. 2010. Towards understanding the influence of SVM hyperparameters. 21st Annual Symposium of the Pattern Recognition Association of South Africa (PRASA), Stellenbosch, South Africa, 22-23 November 2010, pp 69-74 en
dc.identifier.isbn 978-0-7992-2470-2
dc.identifier.uri http://hdl.handle.net/10204/4675
dc.description 21st Annual Symposium of the Pattern Recognition Association of South Africa (PRASA), Stellenbosch, South Africa, 22-23 November 2010 en
dc.description.abstract In this article, the authors investigate the relationship between SVM hyperparameters for linear and RBF kernels and classification accuracy. The process of finding SVM hyperparameters usually involves a gridsearch, which is both time-consuming and resource-intensive. On large datasets, 10-fold cross-validation grid searches can become intractable without supercomputers or high performance computing clusters. They present theoretical and empirical arguments as to how SVM hyperparameters scale with N, the amount of learning data. By using these arguments, the authors present a simple algorithm for finding approximate hyperparameters on a reduced dataset, followed by a focused line search on the full dataset. Using this algorithm gives comparable results to performing a grid search on complete datasets. en
dc.language.iso en en
dc.publisher PRASA 2010 en
dc.relation.ispartofseries Conference Paper en
dc.subject Support vector machine en
dc.subject Hyperparameters en
dc.subject PRASA 2010 en
dc.title Towards understanding the influence of SVM hyperparameters en
dc.type Conference Presentation en


Files in this item

This item appears in the following Collection(s)

Show simple item record