Jessica and two collaborators (Dr. Xin Tong at USC and Dr. Yang Feng at Columbia) published an article on Science Advances about an umbrella algorithm to implement general binary classification algorithms the Neyman-Pearson (NP) paradigm, which aims to control the type I error (or the type II error by symmetry) under a pre-specified threshold with high probability while minimizing the other type of error. This work also introduces a graphical tool, NP-ROC bands, to compare two classification methods under the NP paradigm.

Here is a cartoon video illustrating the basic ideas behind the NP classification: