AI Paper Atlas

Research map / Reinforcement learning / MDPs and planning

Optimal Treatment Allocation for Efficient Policy Evaluation in Sequential Decision Making

Ting Li, Chengchun Shi, Jianing Wang, Fan Zhou, Hongtu Zhu

NeurIPS 2023 · Reinforcement learning › MDPs and planning · approach: Average-reward and regularized MDPs · 2 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 the proposed optimal allocation instead of half-half when reward conditional variance difference delta=0proposed allocation design · Binary observations, p_s=0.8, T=50, n=50MSE of ATE estimateHurt
increase delta (variance difference between treatment and control)proposed allocation design · Binary observations, p_s=0.8, T=50, n=50MSE of ATE estimateImproved
vary T in {10,30,50} and p_s in {0.5,0.8}proposed allocation design · Binary observationsMSE of ATE estimateImproved
continuous observations with moderately large deltaproposed allocation design · Continuous observations, delta_s=1, T=50MSE of ATE estimateImproved
increase number of days nall allocation methods · Synthetic dispatchMSE of ATE estimateImproved
proposed design versus Random, Half-half and Greedyproposed allocation design · Synthetic dispatch, n=50MSE of ATE estimateImproved
proposed design versus other allocation designsproposed allocation design · Real-data based dispatch, n=7 and n=10relative MSE of ATE estimateImproved
increase number of days n from 7 to 10most allocation designs · Real-data based dispatchrelative MSE of ATE estimateImproved

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