2012/2/17 Lars Buitinck <[email protected]>: > 2012/2/13 Olivier Grisel <[email protected]>: >> I don't know for the polynomial kernel part but since C is scale >> according to the number of sample, C=1e4 or more is required for text >> classification. > > I finally had time to try this out. You're absolutely right; C should > be around 1e11 for a quadratic kernel SVM to be competitive on 20 > newsgroups (with feature selection). Thanks!
Interesting, does it bring any improvement over the linear kernel or not? -- Olivier http://twitter.com/ogrisel - http://github.com/ogrisel ------------------------------------------------------------------------------ Virtualization & Cloud Management Using Capacity Planning Cloud computing makes use of virtualization - but cloud computing also focuses on allowing computing to be delivered as a service. http://www.accelacomm.com/jaw/sfnl/114/51521223/ _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
