>>> * if you don't then you should use an sgd-type algorithm >>> * if you have more than two classes, you should use a larank-type >>> algorithm (i think?), but ... >>> >> @mblondel is planning to work on a LaSVM. I wonder if LaRank shares >> some design (I have not re-read the paper recently). >> >> > I might be mis-using terminology, I meant to refer to the multi-class > margin defined by the difference between the correct label and the > best-among-incorrect-labels. > > Ah, alright. I knew that as Crammer-Singer loss, as they proposed this for the use in SVMs. I was wondering why you didn't mention that but I guess there are different names for it. This is implemented in LibSVM btw, but I think not often used as it is slow in the kernelized case and not implemented in LibLinear afaik.
Hope that gets into the scikit soon for SGD, together with multinomial logistic regression. ------------------------------------------------------------------------------ This SF email is sponsosred by: Try Windows Azure free for 90 days Click Here http://p.sf.net/sfu/sfd2d-msazure _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
