Comments on: Nonparametric inference with generalized likelihood ratio tests

摘要

We discuss the issues raised by Fan and Jiang in the context of high dimensional models and argue that fitting sparse nonparametric models might be preferable to hypothesis testing.