Research map / Learning theory and algorithms / Sample complexity
GAAVI: Global Asymptotic Anytime Valid Inference for the Conditional Mean Function
Brian M Cho, Raaz Dwivedi, Nathan Kallus
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use flexible regressors (neural network or k-NN) instead of poorly tuned or restricted ones (ARF, Hoeffding tree, ridge) for tau-hat (ablation) | GAAVI · Synthetic Step/Bump/Sine | rejection probability by T=10,000: rejection with probability 1 by T=10,000 | Improved |
| use adaptive flexible regression instead of linear projection (F-test) | GAAVI vs F-test · Synthetic Bump and Sine | rejection probability by T=10,000: F-test rejection probabilities less than 0.2 by T=10,000 | Improved |
| single weighted test statistic instead of Bonferroni-corrected binning | GAAVI vs Bin(8) · Synthetic Step and Jobs | rejection probability: Bin tests can fail to reject the null completely (Step, Bin(8) in Jobs) | Improved |
| use Bin(8) binning baseline where bins match sign regions | Bin(8) vs GAAVI · Synthetic Sine | rejection probability | Hurt |
| use burn-in t0=250 for anytime-valid error control | all methods · Synthetic null | type I error: error controlled at alpha=0.1 with t0 = 250 | No clear effect |
| vary hyperparameters t0 and gamma over a wide range (ablation) | GAAVI · Synthetic | type I error and stopping probability | No clear effect |
Closest papers by meaning
- Optimal Nuisance Function Tuning for Estimating a Doubly Robust Functional under Proportional Asymptotics (NeurIPS 2025 · similarity 0.85)
- Kernel conditional tests from learning-theoretic bounds (NeurIPS 2025 · similarity 0.84)
- Conditional Testing based on Localized Conformal $p$-values (ICLR 2025 · similarity 0.84)
- Hypothesis Testing for Generalized Thurstone Models (ICML 2025 · similarity 0.84)
- Asymptotically Log-Optimal Bayes-Assisted Confidence Sequences for Bounded Means (NeurIPS 2026 · similarity 0.84)
- Asymptotic Anytime-Valid Inference for U-statistics (NeurIPS 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
