Reproduce “Adaptive Stratified Active Statistical Inference” 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
Pinaki Mohanty, Rajiv Khanna
AdaStrat-ASI uses unlabeled predictions to estimate regional difficulty, allocates a fixed human-label budget across regions, then samples actively within each. It estimates population parameters with inverse-probability correction. On Politeness, it reports confidence intervals up to 45.67% narrower than ASI, with near-90% coverage and nearly identical label use. This is a maximum across tested fractions, not a guarantee across settings.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.