We propose weighted SVM methods with penalization form. By adding weights to loss term, we can build up weighted SVM easily and examine classification algorithm properties under weighted SVM. Through comparing each of test error rates, we conclude that our Weighted SVM with boosting has predominant properties than the standard SVM have, as a whole.
|Author||SungHwan Kim and Soo-Heang Eo|
|Date of publication||2012-10-29 08:59:59|
|Maintainer||SungHwan Kim <firstname.lastname@example.org>|
Error.rate: Calculate Error rate
mixture.example: mixture example
simul.wsvm: Generating simulation data for weighted svm
wsvm: Weighted SVM with boosting algorithm for improving accuracy
wsvm.boost: Weighted SVM using boosting algorithm
wsvm.kernel: Compute kernel K(X, U)
wSVM-package: Weigthed SVM with boosting algorithm for improving accuracy
wsvm.predict: Predict new test set using wsvm function and compute error...