Research map / Learning theory and algorithms / Sample complexity
Linear Hypothesis Testing in High-Dimensional Expected Shortfall Regression with Heavy-Tailed Errors
Gaoyu Wu, Jelena Bradic, Kean Ming Tan, Wen-Xin Zhou
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 Huber-loss robust second step instead of least squares (ablation) | l1-penalized ES regression · t2.5 noise, C1, alpha=0.05 | relative estimation error on support (x100): 33.151 -> 15.622 | Improved |
| use Huber-loss robust second step instead of least squares (ablation) | l1-penalized ES regression · t2.5 noise, C2, alpha=0.05 | relative estimation error on support (x100): 36.914 -> 18.706 | Improved |
| use Huber-loss robust second step instead of least squares (ablation) | l1-penalized ES regression · N(0,1) noise, C1, alpha=0.05 | relative estimation error on support (x100): 4.684 -> 6.950 | Hurt |
| use Huber-loss robust second step instead of least squares (ablation) | l1-penalized ES regression · N(0,1) noise, C2, alpha=0.05 | relative estimation error on support (x100): 5.671 -> 10.099 | Hurt |
| use robust second step (ablation) | l1-penalized ES regression · t2.5 noise, C1, alpha=0.05 | false-positive relative error (x100): 11.842 -> 7.263 | Improved |
| use robust second step (ablation) | l1-penalized ES regression · t2.5 noise, C2, alpha=0.05 | false-positive relative error (x100): 9.591 -> 9.488 | No clear effect |
| use robust second step (ablation) | l1-penalized ES regression · N(0,1) noise, C1, alpha=0.05 | false-positive relative error (x100): 1.998 -> 0.215 | Improved |
| use robust debiased estimator instead of non-robust (ablation) | Debiased ES estimator · t2.5 noise, a4, alpha=0.05 | 95% CI coverage (%): 82.8 -> 92.6 | Improved |
| use robust debiased estimator instead of non-robust (ablation) | Debiased ES estimator · t2.5 noise, a1, alpha=0.05 | 95% CI coverage (%): 85.2 -> 90.0 | Improved |
| use robust debiased estimator instead of non-robust (ablation) | Debiased ES estimator · t2.5 noise, a1, alpha=0.05 | estimation error and CI width (x100): 19.46 -> 14.41 error; 64.33 -> 58.38 width | Improved |
| use robust debiased estimator under lighter-tailed t3.1 noise (ablation) | Debiased ES estimator · t3.1 noise, a2, alpha=0.05 | 95% CI coverage (%): 88.6 -> 93.2 | Improved |
| use robust method on NHANES sTRP data | Debiased ES estimator · NHANES overweight Asian vs non-overweight White | estimated coefficient: 0.51 -> 3.59 | Improved |
Closest papers by meaning
- Deep Neural Expected Shortfall Regression with Tail-Robustness (JMLR 2026 · similarity 0.89)
- $\ell_1$-regression with Heavy-tailed Distributions (NeurIPS 2018 · similarity 0.86)
- Sub-Gaussian Confidence Intervals for Heavy-Tailed Data: Characterizing the Limits of Inference (NeurIPS 2026 · similarity 0.84)
- Robust Sparse Regression with Non-Isotropic Designs (NeurIPS 2024 · similarity 0.83)
- Robust group and simultaneous inferences for high-dimensional single index model (NeurIPS 2024 · similarity 0.83)
- Risk and cross validation in ridge regression with correlated samples (ICML 2025 · similarity 0.82)
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.
