I've been reading/looking around for literature on support vector regressions that are relatively robust to outliers. I understand that standard SVRs can be significantly influenced by a few large outliers. From what I've read (and I'm no academic to be sure), there appears to be a number of different approaches. My questions are two: 1) can anyone explain generally how people approach building a robust SVR? And 2) is there any open source code that has implemented some of these methods yet? It would be interesting and educational to dig through them I think.