Research map / Reinforcement learning / MDPs and planning
A Minimax Learning Approach to Off-Policy Evaluation in Confounded Partially Observable Markov Decision Processes
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 the proposed bridge-based estimator instead of the naive estimator that substitutes observations for latent states | Toy POMDP · Toy POMDP ε=0.25 | policy value estimation: 7.0 vs true value 4.94; 5.01 vs true value 4.94 | Helped |
| Use the proposed bridge-based estimator instead of the naive estimator that substitutes observations for latent states | Toy POMDP · Toy POMDP ε=0.5 | policy value estimation: 5.008 vs true value 4.99; 5.004 vs true value 4.99 | No clear effect |
| Use the proposed bridge-based estimator instead of the naive estimator that substitutes observations for latent states | Toy POMDP · Toy POMDP ε=0.75 | policy value estimation: 3.008 vs true value 5.03; 5.07 vs true value 5.03 | Helped |
| Use bridge-based estimation rather than standard LSTDQ that assumes no unmeasured confounders | One-Dimensional Dynamic Process · One-Dimensional Dynamic Process, σO≥1 | relative mean squared error | Helped |
| Use bridge-based estimation rather than standard LSTDQ that assumes no unmeasured confounders | One-Dimensional Dynamic Process · One-Dimensional Dynamic Process, σO=0.5 | relative mean squared error | No clear effect |
| Use bridge-based estimators rather than MQL, MWL, and DR baselines that do not handle unmeasured confounding | CartPole · CartPole, partially observable settings | estimation error | Helped |
Closest papers by meaning
- Model-based Reinforcement Learning for Confounded POMDPs (ICML 2024 · similarity 0.92)
- A Policy Gradient Method for Confounded POMDPs (ICLR 2024 · similarity 0.92)
- Statistical Tractability of Off-policy Evaluation of History-dependent Policies in POMDPs (ICLR 2025 · similarity 0.91)
- Efficient and Sharp Off-Policy Evaluation in Robust Markov Decision Processes (NeurIPS 2024 · similarity 0.90)
- On the Curses of Future and History in Future-dependent Value Functions for Off-policy Evaluation (NeurIPS 2024 · similarity 0.89)
- RL in Latent MDPs is Tractable: Online Guarantees via Off-Policy Evaluation (NeurIPS 2024 · 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.