AI Paper Atlas

Research map / Reinforcement learning / MDPs and planning

Off-Policy Evaluation for Missingness-Aware Policies in MDPs with Rewards Missing Not at Random

Ziheng Wei, Annie Qu, Rui Miao

ICML 2026 · Reinforcement learning › MDPs and planning · approach: Robust MDP policy gradient · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
use Prox bridge-function estimator instead of IPW-FQEFQE under MNAR rewards · MIMIC-III sepsis, 80% MNARabsolute bias vs oracle FQE: 9.59 -> 2.66Improved
use Prox instead of naive FQEFQE under MNAR rewards · MIMIC-III sepsis, 20% MNARabsolute bias vs oracle FQE: 1.59 -> 0.05Improved
use Prox instead of Impute-FQEFQE under MNAR rewards · MIMIC-III sepsis, 80% MNARabsolute bias vs oracle FQE: 17.70 -> 2.66Improved
increase MNAR missing rate from 20% to 80%Prox · MIMIC-III sepsisabsolute bias vs oracle FQE: 0.05 -> 2.66Hurt
increase MNAR missing rate from 20% to 80%IPW-FQE · MIMIC-III sepsisabsolute bias vs oracle FQE: 1.20 -> 9.59Hurt
replace unobserved rewards with zero in SCOPESCOPE · MIMIC-III sepsisestimated policy value: 10408.2 ± 1991.0Hurt
increase horizon T up to 32Impute-FQE · simulation, ~20% missingnessMSE: MSE exceeding 10^8 at T=32Hurt
use sigmoid instead of linear reward generationProx vs baselines · simulationMSE: gap narrows under linear rewardMixed

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy