Research map / Probabilistic and scientific modeling / Treatment effects
Decorrelated Local Linear Estimator: Inference for Non-linear Effects in High-dimensional Additive Models
Zijian Guo, Wei Yuan, Cunhui 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 decorrelation bias correction (DLL instead of plug-in) (ablation) | local linear estimator · Simulation Setting 2, p=1500 | bias percentage of RMSE: 0.563 -> 0.133 | Improved |
| add decorrelation bias correction (DLL instead of plug-in) (ablation) | local linear estimator · Simulation Setting 2, p=1500 | coverage error: 7.46% -> 1.06% | Improved |
| add decorrelation bias correction (DLL instead of plug-in) (ablation) | local linear estimator · Simulation Setting 1, p=1500 | RMSE ratio to oracle: 1.248 -> 1.152 | Improved |
| add decorrelation bias correction under dependent covariates (Setting 3) (ablation) | local linear estimator · Simulation Setting 3, p=1500 | coverage error: 1.94% -> 0.77% | Improved |
| add decorrelation bias correction under heavy-tailed t design (Setting 4) (ablation) | local linear estimator · Simulation Setting 4, p=1500 | coverage error: 6.72% -> 1.15% | Improved |
| use DLL instead of ReSmoothing estimator | DLL · Gregory et al. setting, n=100, p=150, r=0.1 | bias: 0.78 -> 0.22 | Improved |
| use DLL instead of ReSmoothing at larger sample size (n=1000) | DLL · Gregory et al. setting, a0=-1 | coverage: OraRS achieves desired coverage at n=1000 | No clear effect |
| replace Lasso treatment model by sparse additive model (DLL-S) under non-linear treatment | DLL · Non-linear treatment model, a0=0.10, n=500 | bias: 0.31 -> 0.23 | Improved |
| replace Lasso treatment model by sparse additive model (DLL-S) under non-linear treatment | DLL · Non-linear treatment model, a0=0.10, n=500 | coverage: 0.84 -> 0.91 | Improved |
| replace Lasso treatment model by sparse additive model (DLL-S) under linear treatment | DLL · Linear treatment model, a0=0.10, n=500 | bias: 0.18 -> 0.22 | No clear effect |
| use DLL instead of SIHR linear-model inference on motif regression | DLL · Motif regression, 666 motifs, n=2587 | detection of non-linear effects: motifs 53, 138, 199, 586 significant under additive model but not under linear SIHR | Improved |
Closest papers by meaning
- High-Dimensional Inference for Generalized Linear Models with Hidden Confounding (JMLR 2023 · similarity 0.86)
- Estimation and Inference for High Dimensional Generalized Linear Models: A Splitting and Smoothing Approach (JMLR 2021 · similarity 0.86)
- Additive Nonlinear Quantile Regression in Ultra-high Dimension (JMLR 2022 · similarity 0.86)
- Inference In High-dimensional Single-Index Models Under Symmetric Designs (JMLR 2021 · similarity 0.86)
- Confidence Intervals and Hypothesis Testing for High-dimensional Quantile Regression: Convolution Smoothing and Debiasing (JMLR 2023 · similarity 0.85)
- Learning Joint Interventional Effects from Single-Variable Interventions in Additive Models (ICML 2025 · 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.
