Two-way Deconfounder for Off-policy Evaluation in Causal Reinforcement Learning
This addresses the problem of accurate policy evaluation in reinforcement learning for researchers and practitioners, but it is incremental as it builds on existing deconfounder and panel data methods.
The paper tackles off-policy evaluation in causal reinforcement learning with unmeasured confounders by proposing a two-way deconfounder algorithm, achieving consistent policy value estimation as demonstrated through theoretical results and numerical experiments.
This paper studies off-policy evaluation (OPE) in the presence of unmeasured confounders. Inspired by the two-way fixed effects regression model widely used in the panel data literature, we propose a two-way unmeasured confounding assumption to model the system dynamics in causal reinforcement learning and develop a two-way deconfounder algorithm that devises a neural tensor network to simultaneously learn both the unmeasured confounders and the system dynamics, based on which a model-based estimator can be constructed for consistent policy value estimation. We illustrate the effectiveness of the proposed estimator through theoretical results and numerical experiments.