Spark 1.6的多类SVM？

Question

I ran a multi-class Logistic Regression with Spark but I would like to use SVM to cross validate results. It looks like Spark 1.6 only supports SVM binary classifications. Should I use other tools to do this? H20 for example?

Answer 1

After some research, I found this branch which was not integrated in Spark 1.6 that allowed me to run the SVM on a multi class classification problem.

Big thanks to Bekbolatov.

The commit can be ofund here: https://github.com/Bekbolatov/spark/commit/463d73323d5f08669d5ae85dc9791b036637c966