On the robustness of kernel-based pairwise learning

Patrick Gensler, Andreas Christmann

It is shown that many results on the statistical robustness of kernel-based pairwise learning can be derived under basically no assumptions on the input and output spaces. In parti…