Coincidentally I implemented and experimented a lot with RBF kernel PCA on various different datasets and gammas. I used the scikit-learn one as reference and comparison and never had any issues with it as long gamma > 0. Maybe it helps if you could post your code and data (if this is okay to share the data) where the problem ocurred.
Best, Sebastian On Sep 15, 2014, at 10:32 AM, Pagliari, Roberto <[email protected]> wrote: > Actually, you are using nu=.5, which means you are expecting a novelty > detection rate up to 50%. > > You definitely decrease it. With .5 the result will be fairly random . > > Roberto > > > From: Pagliari, Roberto [mailto:[email protected]] > Sent: Monday, September 15, 2014 10:28 AM > To: [email protected] > Subject: Re: [Scikit-learn-general] Bug in one class svm > > Did you try change the value of nu? Perhaps, it’s too large. > > From: Pagliari, Roberto [mailto:[email protected]] > Sent: Monday, September 15, 2014 10:24 AM > To: [email protected] > Subject: Re: [Scikit-learn-general] Bug in one class svm > > I have used it with all kernels and several values of gamma (including the > default) and never had any issue with it, > > Roberto > > > From: Albert Thomas [mailto:[email protected]] > Sent: Monday, September 15, 2014 10:00 AM > To: [email protected] > Subject: Re: [Scikit-learn-general] Bug in one class svm > > When using the rbf kernel, you should try with a gamma > 0. It seems that you > set it to 0. > Albert > 2014-09-15 15:37 GMT+02:00 Luca Puggini <[email protected]>: > Hi, > there is no segmentation fault in the default settings. > Even if according to the original paper it can make sense to use OCSVM also > with not rbf kernel. > Maybe there is a bug in the polynomial kernel, I don't know. > > Despite that also with the RBF kernel I am having some problems with the > frontier. > I have posted a question with my problem here (plot included) > http://stats.stackexchange.com/questions/115481/one-class-svm-strange-decision-boundary > I do not exclude that I am doing some mistakes. So please tell me if I am > wrong. > Thanks a lot, > Luca > > > Hi Luca, > > it segfaults?! Can you confirm that it also segfaults if you use the > default arguments? There is no plot so I cannot say anything about the > strange decision boundaries. > > For my part, I've never used something else than a RBF kernel for a one > class svm; the RBF kernel has the nice property that all data points lie on > the surface of a hypersphere and thus the minimum enclosing ball is just > the hyperplane that separates those points and the origin with the max > distance to the origin. > > 2014-09-15 10:58 GMT+02:00 Luca Puggini <lucapuggio@...>: > > > > > Hi, > > I am having some problems with the OneClassSVM function. > > > > Here you can see my file and the output. > > http://justpaste.it/h3pw > > > > I am sorry but I can not share the used data. > > > > I have experienced also other problems like strange decision boundaries. > > > > Can someone tell me if I am doing something wrong or if there is a problem > > in the function? > > > > Thanks, > > Luca > > > > ------------------------------------------------------------------------------ > Want excitement? > Manually upgrade your production database. > When you want reliability, choose Perforce > Perforce version control. Predictably reliable. > http://pubads.g.doubleclick.net/gampad/clk?id=157508191&iu=/4140/ostg.clktrk > _______________________________________________ > Scikit-learn-general mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general > > > ------------------------------------------------------------------------------ > Want excitement? > Manually upgrade your production database. > When you want reliability, choose Perforce > Perforce version control. Predictably reliable. > http://pubads.g.doubleclick.net/gampad/clk?id=157508191&iu=/4140/ostg.clktrk_______________________________________________ > Scikit-learn-general mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general ------------------------------------------------------------------------------ Want excitement? Manually upgrade your production database. When you want reliability, choose Perforce Perforce version control. Predictably reliable. http://pubads.g.doubleclick.net/gampad/clk?id=157508191&iu=/4140/ostg.clktrk _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
