Reproduce “SPACR: Single-Pass Adaptive Training of Uncertainty-Aware Conformal Regressors” 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
Soundouss Messoudi, Sylvain Rousseau, Sébastien Destercke
SPACR jointly predicts values and input-specific uncertainty scales, then uses separate conformal calibration to form intervals at chosen confidence levels. Experiments test 90%, 95%, and 99% coverage. On UTK Face age regression at 90% target coverage, it reports 90.13% marginal coverage and median interval width 21.84. Under exchangeability, this is marginal, not subgroup, coverage.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.