Thanks for all the answers. Then the fault is probably due to the overfitting of OCSVM. I was probably mislead by the title of my reference paper "*Estimating *the *support *of a *high*-*dimensional *distribution" <http://www.mitpressjournals.org/doi/abs/10.1162/089976601750264965>
Best, Luca On Wed, Oct 14, 2015 at 3:10 PM Gael Varoquaux < gael.varoqu...@normalesup.org> wrote: > On Wed, Oct 14, 2015 at 01:18:19PM +0000, Luca Puggini wrote: > > I was expecting OCSVM to be not too much influenced by the increasing > number of > > variables even if some of them are irrelevant. > > I am not: it's based on an RBF kernel. These things are not well behaved > in high dimensions. > > Gaël > > > ------------------------------------------------------------------------------ > _______________________________________________ > Scikit-learn-general mailing list > Scikit-learn-general@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general > -- Sent by mobile phone
------------------------------------------------------------------------------
_______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
