AI Paper Atlas

Research map / Reinforcement learning / Policy gradient and actor-critic

Asymptotic Inference for Multi-Stage Stationary Treatment Policy with Variable Selection

Daiqi Gao, Yufeng Liu, Donglin Zeng

JMLR 2025 · Reinforcement learning › Policy gradient and actor-critic · approach: Confounding-robust policy evaluation · 3 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 Q-function-based nuisance Q(1) instead of Q(0) or Q(2) in the doubly robust estimator (ablation)MSTP · Simulation Scenarios 1 and 2average reward and MADImproved
Q(2) depends on importance sampling weights, used with larger T (ablation)MSTP · Simulation, T up to 10numerical stabilityHurt
estimate a stochastic MSTP policy rather than PEARL's deterministic policyMSTP vs PEARL · Simulation Scenario 1, T=1average rewardHurt
MSTP stochastic policy versus PEARLMSTP vs PEARL · Simulation Scenario 2, T=1average rewardImproved
add one-step correction to the sparse estimator for inference (ablation)MSTP confidence intervals · Simulation, T=10coverage probabilityImproved
increase horizon TMSTP inference · Simulationcoverage probabilityHurt
use Q(1) nuisance versus Q(0), Q(2) and PEARL for one-step estimator (ablation)MSTP one-step estimator · SimulationMAD and coverageImproved
learn policy on OhioT1DM versus behavior policyMSTP · OhioT1DMaverage reward: -6.260 -> -2.367Improved

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