Research map / Learning theory and algorithms / Matrix and tensor methods
Non-Asymptotic Guarantees for Robust Statistical Learning under Infinite Variance Assumption
Lihu Xu, Fang Yao, Qiuran Yao, Huiming Zhang
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 |
|---|---|---|---|
| add log-truncation to DNN LAD under Gaussian pixel noise (20%) (ablation) | DNN LAD · MNIST | accuracy (%): 75.71 -> 86.85 | Improved |
| add log-truncation to DNN LAD under Uniform pixel noise (80%) (ablation) | DNN LAD · MNIST | accuracy (%): 91.65 -> 93.45 | Improved |
| add log-truncation to DNN LAD under Pareto pixel noise (80%) (ablation) | DNN LAD · MNIST | accuracy (%): 60.96 -> 65.12 | Improved |
| add log-truncation to standard LAD regression (ablation) | l2-regularized LAD · Boston housing | MAE: 6.4843 -> 6.2448 | Improved |
| add log-truncation to deep LAD regression (ablation) | deep LAD · Boston housing | MAE: 6.3186 -> 6.0628 | Improved |
| add high-order log-truncation to DNN logistic regression (ablation) | DNN logistic regression · Synthetic (d=100, n=1000, Pareto tau=4.01) | accuracy (%): 80.13 -> 93.36 | Improved |
| add high-order log-truncation to elastic-net logistic regression (ablation) | logistic regression · Synthetic (d=100, n=200, Pareto tau=1.60, varsigma=0) | l2-estimation error: 3.5344 -> 2.9463 | Improved |
| use high-order log-truncation instead of Cauchy 1-order truncation (ablation) | log-truncated logistic regression · Synthetic (d=200, n=500, Uniform noise pi=0.3, varsigma=0) | l2-estimation error: 4.9339 -> 4.3418 | Improved |
| use high-order log-truncation instead of Cauchy truncation (ablation) | log-truncated logistic regression · Synthetic (d=100, n=200, Pareto tau=1.60, varsigma=0) | l2-estimation error: 3.1587 -> 2.9463 | Improved |
| vary truncation index beta from 1.1 to 2.0 (best at 1.8) (ablation) | truncated deep LAD · Boston housing | MAE: 6.2822 -> 6.0628 | Mixed |
Closest papers by meaning
- Outlier Robust Mean Estimation with Subgaussian Rates via Stability (NeurIPS 2020 · similarity 0.87)
- Regularized least squares learning with heavy-tailed noise is minimax optimal (NeurIPS 2025 · similarity 0.86)
- Robust Sparse Regression with Non-Isotropic Designs (NeurIPS 2024 · similarity 0.86)
- Adversarial Robustness of Nonparametric Regression (NeurIPS 2025 · similarity 0.86)
- Robust Statistical Estimators with Bounded Empirical Sensitivity (NeurIPS 2026 · similarity 0.85)
- Risk Bounds For Distributional Regression (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.
