SVM Research

[htlin]

Admin: The following post is copied from another thread.

Originally Posted by cambridge11

One last question: What are some of the important outstanding research questions related to SVMs from a theoretical standpoint? Or is this area pretty well understood?

Thanks again!

The original (binary classification) SVM is pretty mature from both the theoretical side and the practical side for mid-scale data. Large-scale learning with SVM remains to be a challenging research direction. In recent years, my observation is that people work more on studying other SVM-related formulations for different machine learning tasks.

Hope this helps.