Is it possible to compute Cross entropy for support vector machine?

Hallo

Is it possible to compute cross entropy (CE) for support vector machine (SVM)? For example for this gerneral problem:

load fisheriris
X = meas(:,1:2);
y = ones(size(X,1),1);

% Train an SVM classifier:
SVMModel = fitcsvm(X,y,'KernelScale','auto','Standardize',true,...
    'KernelFunction','rbf','BoxConstraint',1);

% Predict classification: http://se.mathworks.com/help/stats/compactclassificationdiscriminant.predict.html
[label,score,cost] = predict(SVMModel, y);

Thank you

Answer 1

[the cyclist]

I did a bit of searching, and I found cross entropy calculations in MATLAB for neural nets, but not for SVM. (Disclaimer: I am not an expert in these techniques.)

[Lily]

Yes, that is what I found out yesterday as well. I thought maybe someone had written or MATLAB and an inbuild cross entropy calculations similar to computing the mean of e.g. vector.

[Greg Heath]

MATLAB14a doesn't reveal in documentation or source code which version (exclusive targets (e.g., man/woman))or (non-exclusive targets (e.g., tall/dark/handsome)) of crossentropy that they are using.

In 2013 or 2014 I compared the output of crossentropy(t,y) with the two versions. It was the nonexclusive version. I alerted MATLAB and assume that it has been changed (instead of offering the option!)

Answer 2

[Greg Heath]

I don't remember if my post was in the NEWSGROUP or ANSWERS.

Therefore, search both using

 greg crossentropy

Hope this helps.

Thank you for formally accepting my answer

Greg