Research map / Learning theory and algorithms / Kernels and regression theory
Composite Goodness-of-fit Tests with Kernels
Oscar Key, Arthur Gretton, François-Xavier Briol, Tamara Fernandez
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 parametric instead of wild bootstrap for MMD test (ablation) | MMD composite test · Gaussian model, d=1 (type I error and power) | type I error / power | Improved |
| use parametric instead of wild bootstrap (ablation) | MMD composite test · Gaussian model, d=1, n=100 | time per test (ms): 1ms -> 5ms | Hurt |
| use KSD instead of MMD | composite goodness-of-fit test · Gaussian model, d=1, Student-t alternative | power | Improved |
| increase dimension d | MMD composite test · Gaussian vs multivariate t | power | Improved |
| increase dimension d | KSD composite test · Gaussian vs multivariate t | power | Hurt |
| use proposed MMD tests instead of non-asymptotic MMD test | MMD composite test · 1D Gaussian, known variance | power | Improved |
| use tilted kernel in KSD test (ablation) | KSD composite test · Huber contamination, d=1 | robustness to corruption: not robust -> robust to >6\% corruption | Improved |
| use specialised Gaussianity tests instead of MMD composite test | Gaussianity testing · Student-t alternative, varying nu | power | Improved |
| choose lengthscale l=2 by grid search instead of median heuristic (ablation) | KSD composite test · Gaussianity test vs Student-t | power | Improved |
| increase p of kernel exponential family (ablation) | density model on galaxies data · Galaxies density fit | test rejection: rejects p=1,2,3; not p=4,5,25 | Improved |
Closest papers by meaning
- A Kernel Test of Goodness of Fit (ICML 2016 · similarity 0.90)
- Kernel conditional tests from learning-theoretic bounds (NeurIPS 2025 · similarity 0.86)
- Kernel Selection is Model Selection: A Unified Complexity-Penalised Approach for MMD Two-Sample Tests (NeurIPS 2026 · similarity 0.86)
- Anchor-based Maximum Discrepancy for Relative Similarity Testing (NeurIPS 2025 · similarity 0.85)
- Sequential Kernel Goodness-of-fit Testing (ICML 2024 · similarity 0.85)
- DUAL: Learning Diverse Kernels for Aggregated Two-sample and Independence Testing (NeurIPS 2025 · similarity 0.85)
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.
