Efficient online learning with Kernels for adversarial large scale problems

We are interested in a framework of online learning with kernels for low-dimensional, but large-scale and potentially adversarial datasets.