One of the most fundamental problems in learning theory is to view input data as random samples from an unknown distribution and then to make statistical inferences about the underlying distribution. In this talk, we focus ...