Research map / Reinforcement learning / MDPs and planning
Double Reinforcement Learning for Efficient and Robust Off-Policy Evaluation
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 DRL (MDP) estimator (cross-fit q and marginal density ratio) instead of IS/DM/MIS baselines | OPE in MDPs · Cliff Walking and Mountain Car | RMSE | Helped |
| Marginalized density ratio estimator (MIS) instead of per-step IS | Cliff Walking · Cliff Walking | RMSE: e.g. 18.8 -> 12.8 at n=500 | Helped |
| Marginalized density ratio estimator (MIS) instead of per-step IS | Mountain Car · Mountain Car | RMSE: 6.85 -> 6.82 at n=500 | No clear effect |
| Cross-fit DRL with NMDP nuisances (cumulative density ratio) vs. MDP-structured DRL | DRL estimator · Cliff Walking and Mountain Car | RMSE: DRL(NMDP) worse than DRL(MDP) in every setting | Hurt |
| Direct method (DM) baseline in Cliff Walking | DM · Cliff Walking | RMSE: beats IS/MIS but worse than DRL(MDP) at all sizes | Mixed |
Closest papers by meaning
- More Efficient Off-Policy Evaluation through Regularized Targeted Learning (ICML 2019 · similarity 0.92)
- Intrinsically Efficient, Stable, and Bounded Off-Policy Evaluation for Reinforcement Learning (NeurIPS 2019 · similarity 0.92)
- Statistical Tractability of Off-policy Evaluation of History-dependent Policies in POMDPs (ICLR 2025 · similarity 0.91)
- Doubly Robust Off-policy Value Evaluation for Reinforcement Learning (ICML 2016 · similarity 0.91)
- Model-based Reinforcement Learning for Confounded POMDPs (ICML 2024 · similarity 0.89)
- Demystifying the Paradox of Importance Sampling with an Estimated History-Dependent Behavior Policy in Off-Policy Evaluation (ICML 2025 · similarity 0.89)
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.