R help - Mar 2012 - How can I extract the weights (for feature selection) from linear KSVM?

Hello all,
I need some help with extracting the feature-wise weights from the SVM
object that is the output of *ksvm* function in R (from *kernlab* package).
For instance, if I use the same data/example as the help page does, i.e.,

data(spam)

index <- sample(1:dim(spam)[1])

spam.train <- spam[index[1:floor(2*dim(spam)[1]/3)],]

spam.test <- spam[index[((2*ceiling(dim(spam)[1]/3))+1):dim(spam)[1]],]

filter <- ksvm(type ~ ., data = spam.train, kernel = "vanilladot")

Can someone please help me find out how I can extract the weight from the
SVM object *filter*. Your help is much appreciated.
Best,
-- 
Reza Salimi-Khorshidi, D.Phil. (Oxon)
Analysis Group, FMRIB Centre of the University of Oxford
Email: reza@fmrib.ox.ac.uk
Address: FMRIB Centre, John Radcliffe Hospital, Oxford OX3 9DU, UK

	[[alternative HTML version deleted]]