Thread Subject: svmclassify decision boundary

Hi

I am having some trouble with the decision boundary from the support vector machine. The following example is from the Matlab documentation [1]:

load fisheriris
xdata = meas(51:end,3:4);
group = species(51:end);
svmStruct = svmtrain(xdata,group,'showplot',true);
species = svmclassify(svmStruct,[5 2],'showplot',true)

If you change the point to be classified to something too far away from the training data (e.g. [2 3]) the plot is extended by svmclassify, but the decision boundary is not. Does anybody know a good way of extending the decision boundary to fit the plot?

Thanks,

Robert

[1] http://www.mathworks.com/help/toolbox/bioinfo/ref/svmclassify.html

Just in case someone else is having the same problem: The decision boundary is made with the function svmplotsvs from the biolearning toolbox and the input to this function necessary to (re)plot the decision boundary is part of the structure that you get from svmtrain.