Research map / Reinforcement learning / Policy gradient and actor-critic
Intrinsically Efficient, Stable, and Bounded Off-Policy Evaluation for Reinforcement Learning
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 |
|---|---|---|---|
| Choose control-variate parameters by REG rather than using DR directly | Contextual-bandit OPE · SatImage; behavior policy 0.7πd+0.3πu | RMSE×1000: 3.0 -> 2.8 | Helped |
| Choose control-variate parameters by EMP rather than using DR directly | Contextual-bandit OPE · Pageblock; behavior policy 0.4πd+0.6πu | RMSE×1000: 2.7 -> 2.4 | Helped |
| Choose control-variate parameters by EMP rather than using DR directly | Contextual-bandit OPE · PenDigits; behavior policy 0.4πd+0.6πu | RMSE×1000: 2.2 -> 2.0 | Helped |
| Use REG with learned control-variate parameters rather than DR | Windy GridWorld OPE · Windy GridWorld; size 250 | RMSE: 0.17 -> 0.09 | Helped |
| Use EMP with learned control-variate parameters rather than MDR | Cliff Walking OPE · Cliff Walking; size 2000 | RMSE: 2.2 -> 1.5 | Helped |
| Use REG with learned control-variate parameters rather than DR | Mountain Car OPE · Mountain Car; size 3000 | RMSE: 1.4 -> 1.0 | Helped |
| Use EMP instead of REG | OPE estimators · Windy GridWorld; size 250 | RMSE: 0.09 -> 0.09 | No clear effect |
| Optimize too many parameters in MDR rather than using the REG/EMP parameterization | MDR · Contextual-bandit experiments | RMSE | Hurt |
Closest papers by meaning
- Doubly Robust Off-policy Value Evaluation for Reinforcement Learning (ICML 2016 · similarity 0.93)
- More Robust Doubly Robust Off-policy Evaluation (ICML 2018 · similarity 0.92)
- More Efficient Off-Policy Evaluation through Regularized Targeted Learning (ICML 2019 · similarity 0.92)
- Double Reinforcement Learning for Efficient and Robust Off-Policy Evaluation (ICML 2020 · similarity 0.92)
- Demystifying the Paradox of Importance Sampling with an Estimated History-Dependent Behavior Policy in Off-Policy Evaluation (ICML 2025 · similarity 0.90)
- Model Selection for Off-policy Evaluation: New Algorithms and Experimental Protocol (NeurIPS 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.