Reproduce “Integral Probability Boundaries” 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
Bogdan Kulynych, Shahab Asoodeh
The authors introduce integral probability boundaries (IPBs), which bound an expectation under one distribution using its value under another, retaining baseline prevalence that a single integral probability metric (IPM) obscures. In a California Census and CTGAN experiment, IPM lower bounds were vacuous for rare groups while IPB lower bounds remained informative; the smallest-group comparison reached 500×. This empirical result is not a universal guarantee.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.