Efficient Sampling for Learning Sparse Additive Models in High Dimensions

Neural Information Processing Systems 

We consider the problem of learning sparse additive models, i.e., functions of the form: f(x) =