Research map / Learning theory and algorithms / Prediction and decision losses
Debiased Machine Learning without Sample-Splitting for Stable Estimators
Qizhao Chen, Vasilis Syrgkanis, Morgane Austern
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 the nuisance estimates without cross-fitting rather than 2-fold cross-fitting | Partially linear treatment-effect estimation with subsampled 1-NN nuisances · Synthetic partially linear model, subsample size m=n^0.49 | confidence-interval coverage | Improved |
| Use the nuisance estimates without cross-fitting rather than 2-fold cross-fitting | Partially linear treatment-effect estimation with subsampled random-forest nuisances · Synthetic partially linear model, subsample size m=n^0.49 | confidence-interval coverage | Improved |
| Use nuisance estimates without cross-fitting | Partially linear treatment-effect estimation with subsampled 1-NN or random-forest nuisances · Synthetic partially linear model, stable subsample sizes | bias | No clear effect |
| Use nuisance estimates without cross-fitting rather than 2-fold cross-fitting | Partially linear treatment-effect estimation with subsampled 1-NN or random-forest nuisances · Synthetic partially linear model, stable subsample sizes | variance | Improved |
| Use nuisance estimates without cross-fitting rather than 2-fold cross-fitting | Standard-error estimation for partially linear treatment effects · Synthetic partially linear model | standard-error estimation accuracy | Improved |
| Use nuisance estimates without cross-fitting | Partially linear treatment-effect estimation with subsampled 1-NN nuisances · Synthetic partially linear model, subsample size m=n | estimator performance | Hurt |
| Use cross-fitting rather than no cross-fitting | Partially linear treatment-effect estimation with subsampled 1-NN nuisances · Synthetic partially linear model, subsample size m=n | estimator performance | Improved |
Closest papers by meaning
- Instability, Computational Efficiency and Statistical Accuracy (JMLR 2025 · similarity 0.85)
- Uncertainty Calibration for Ensemble-Based Debiasing Methods (NeurIPS 2021 · similarity 0.84)
- Orthogonal Machine Learning: Power and Limitations (ICML 2018 · similarity 0.84)
- Stable Learning via Sparse Variable Independence (AAAI 2023 · similarity 0.84)
- Localized Debiased Machine Learning: Efficient Inference on Quantile Treatment Effects and Beyond (JMLR 2024 · similarity 0.84)
- RieszNet and ForestRiesz: Automatic Debiased Machine Learning with Neural Nets and Random Forests (ICML 2022 · 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.
