I found an implementation at http://code.google.com/p/cascadesvm/ . However, there are no specifications about this. Has anyone tried this? Or where can I find an alternative implementation of SVM on Hadoop? Thanks a lot ~
It seems like someone did this in the Mahout project, not sure if it was integrated into the trunk, but it looks like a good place to run:
https://issues.apache.org/jira/browse/MAHOUT-232
https://code.google.com/p/cascadesvm/ Matlab. https://code.google.com/p/cascadesvm/wiki/CascadeSVMMatlabVersion