hi, an intuition is that the scaling depends if you penalize by a norm (L1 case) or a squared norm (L2 case).
Thanks Gael for summarizing our intense conversations of last week. I'll take care of the scale_C removal later today Alex On Sun, Apr 29, 2012 at 11:11 PM, Andreas Mueller <[email protected]> wrote: > Hi Gael. > First of all, THANK YOU SO MUCH (and of course Jaques and Alex) > for working on this issue. > > If you have a reference for the L2 case, I would be really interested. > I think you didn't include this picture in your mail: > http://www.flickr.com/photos/79425020@N04/6963277866/in/photostream > which I find quite interesting :) > > Also: I'm ok with removing scale_C and going back to the old ways. > > I think releasing in the foreseeable future would be good. > > Cheers, > Andy > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Scikit-learn-general mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
