Research map / Reinforcement learning / Policy gradient and actor-critic
Deeply-Debiased Off-Policy Interval Estimation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 deeply-debiased TR confidence intervals instead of DRL intervals | Off-policy value estimation · CartPole and Diabetes | coverage probability | Helped |
| Use deeply-debiased TR confidence intervals instead of DRL intervals | Off-policy value estimation · CartPole and Diabetes | CI length | Hurt |
| Use CoinDICE confidence intervals on weakly dependent trajectories | Off-policy interval estimation · CartPole and Diabetes | coverage probability | Hurt |
| Use stepwise IS confidence intervals instead of TR intervals | Off-policy interval estimation · CartPole and Diabetes | CI length | Hurt |
| Use bootstrap intervals for the stepwise IS estimator on weakly dependent data | Off-policy interval estimation · CartPole and Diabetes | coverage probability | Hurt |
| Use TR instead of DRL for point estimation | Off-policy value estimation · CartPole and Diabetes | bias | Helped |
| Use TR instead of FQE for point estimation | Off-policy value estimation · CartPole and Diabetes | bias | Helped |
| Use TR instead of FQE for point estimation | Off-policy value estimation · CartPole and Diabetes | RMSE | Helped |
| Add conditional-density-ratio debiasing to the initial Q-estimator (ablation) | TR value estimation · CartPole and Diabetes | bias | Helped |
| Increase the debias iteration order from m=2 to m=3 (ablation) | TR · CartPole and Diabetes | coverage probability | No clear effect |
| Increase the debias iteration order (ablation) | TR · Off-policy value estimation | bias | Helped |
| Increase the debias iteration order (ablation) | TR · Off-policy value estimation | variance | Hurt |
Closest papers by meaning
- CoinDICE: Off-Policy Confidence Interval Estimation (NeurIPS 2020 · similarity 0.88)
- Off-Policy Interval Estimation with Lipschitz Value Iteration (NeurIPS 2020 · similarity 0.88)
- Minimax Value Interval for Off-Policy Evaluation and Policy Optimization (NeurIPS 2020 · similarity 0.87)
- Off-policy Evaluation Beyond Overlap: Sharp Partial Identification Under Smoothness (ICML 2024 · similarity 0.87)
- Cross-Validated Off-Policy Evaluation (AAAI 2025 · similarity 0.85)
- Off-Policy Evaluation under Nonignorable Missing Data (ICML 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.