Reproduce “Beyond Fixed False Discovery Rates: Post-Hoc Conformal Selection with E-Variables” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Meiyi Zhu, Osvaldo Simeone
PH-CS offers a data-derived path of shortlists with estimated false discovery proportions, allowing users to choose a utility-preferred point after observing data. It uses level-independent conformal e-variables and e-BH. Under exchangeability and score monotonicity, the expected ratio of realized to estimated false discovery proportion is at most one. In constrained-size experiments, it met size requirements every run, though FDP was higher on Musk.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.