High-Dimensional Non-Linear Variable Selection through Hierarchical Kernel Learning

We consider the problem of high-dimensional non-linear variable selection for supervised learning. Our approach is based on performing linear selection among exponentially many well-defined groups of