Semiparametric Double Reinforcement Learning with Applications to Long-Term Causal Inference

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: van der Laan, Lars, Hubbard, David, Tran, Allen, Kallus, Nathan, Bibaut, Aurélien
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908648619376640
author van der Laan, Lars
Hubbard, David
Tran, Allen
Kallus, Nathan
Bibaut, Aurélien
author_facet van der Laan, Lars
Hubbard, David
Tran, Allen
Kallus, Nathan
Bibaut, Aurélien
contents Double Reinforcement Learning (DRL) enables efficient inference for policy values in nonparametric Markov decision processes (MDPs), but existing methods face two major obstacles: (1) they require stringent intertemporal overlap conditions on state trajectories, and (2) they rely on estimating high-dimensional occupancy density ratios. Motivated by problems in long-term causal inference, we extend DRL to a semiparametric setting and develop doubly robust, automatic estimators for general linear functionals of the Q-function in infinite-horizon, time-homogeneous MDPs. By imposing structure on the Q-function, we relax the overlap conditions required by nonparametric methods and obtain efficiency gains. The second obstacle--density-ratio estimation--typically requires computationally expensive and unstable min-max optimization. To address both challenges, we introduce superefficient nonparametric estimators whose limiting variance falls below the generalized Cramer-Rao bound. These estimators treat the Q-function as a one-dimensional summary of the state-action process, reducing high-dimensional overlap requirements to a single-dimensional condition. The procedure is simple to implement: estimate and calibrate the Q-function using fitted Q-iteration, then plug the result into the target functional, thereby avoiding density-ratio estimation altogether.
format Preprint
id arxiv_https___arxiv_org_abs_2501_06926
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Semiparametric Double Reinforcement Learning with Applications to Long-Term Causal Inference
van der Laan, Lars
Hubbard, David
Tran, Allen
Kallus, Nathan
Bibaut, Aurélien
Machine Learning
Methodology
Double Reinforcement Learning (DRL) enables efficient inference for policy values in nonparametric Markov decision processes (MDPs), but existing methods face two major obstacles: (1) they require stringent intertemporal overlap conditions on state trajectories, and (2) they rely on estimating high-dimensional occupancy density ratios. Motivated by problems in long-term causal inference, we extend DRL to a semiparametric setting and develop doubly robust, automatic estimators for general linear functionals of the Q-function in infinite-horizon, time-homogeneous MDPs. By imposing structure on the Q-function, we relax the overlap conditions required by nonparametric methods and obtain efficiency gains. The second obstacle--density-ratio estimation--typically requires computationally expensive and unstable min-max optimization. To address both challenges, we introduce superefficient nonparametric estimators whose limiting variance falls below the generalized Cramer-Rao bound. These estimators treat the Q-function as a one-dimensional summary of the state-action process, reducing high-dimensional overlap requirements to a single-dimensional condition. The procedure is simple to implement: estimate and calibrate the Q-function using fitted Q-iteration, then plug the result into the target functional, thereby avoiding density-ratio estimation altogether.
title Semiparametric Double Reinforcement Learning with Applications to Long-Term Causal Inference
topic Machine Learning
Methodology
url https://arxiv.org/abs/2501.06926