Prev: uibuttongroup SelectionChangeFcn
Next: simulated ARMA model output
From: Harshavardhan Agashe on 11 Dec 2009 20:14
parameter C (relative importance of error penalization) is the same as boxconstraint (Bishop, C.M. "Pattern recognition and Machine learning", look up 'boxconstraints' in the index). In fact, the boxconstraint option in svmtrain allows one to provide a relative importance for each error individually.
From: Novae on 14 Dec 2009 08:44
Place the C value in the configuration string for your SVM.

Example for a one vs. one SVM with RBF kernel, Gamma = 1 and Cost = 10

Cvalue = 10;
ConfigStr= sprintf('-s 0 -t 2 -g 1 -c %d'. Cvalue);

Good luck

Igor

"DIA D." <poovich(a)mavs.com> wrote in message <g6rjs5$c0p$1(a)fred.mathworks.com>...
> Does anyone know how to change C parameter of svmtrain from
> Matlab Bioinfo toolbox?
 | 
Pages: 1
Prev: uibuttongroup SelectionChangeFcn
Next: simulated ARMA model output