Sample-Efficient Reinforcement Learning via Counterfactual-Based Data Augmentation
This work is significant for applying reinforcement learning in data-scarce and heterogeneous environments, particularly in domains like healthcare where patient-specific data is limited and responses vary.
This paper addresses data scarcity and mechanism heterogeneity in reinforcement learning (RL) by proposing a data-efficient RL algorithm that uses structural causal models (SCMs) to model state dynamics. The SCMs allow for counterfactual reasoning, avoiding risky real exploration and mitigating bias from limited experiences, ultimately leading to optimal value function convergence.
Reinforcement learning (RL) algorithms usually require a substantial amount of interaction data and perform well only for specific tasks in a fixed environment. In some scenarios such as healthcare, however, usually only few records are available for each patient, and patients may show different responses to the same treatment, impeding the application of current RL algorithms to learn optimal policies. To address the issues of mechanism heterogeneity and related data scarcity, we propose a data-efficient RL algorithm that exploits structural causal models (SCMs) to model the state dynamics, which are estimated by leveraging both commonalities and differences across subjects. The learned SCM enables us to counterfactually reason what would have happened had another treatment been taken. It helps avoid real (possibly risky) exploration and mitigates the issue that limited experiences lead to biased policies. We propose counterfactual RL algorithms to learn both population-level and individual-level policies. We show that counterfactual outcomes are identifiable under mild conditions and that Q- learning on the counterfactual-based augmented data set converges to the optimal value function. Experimental results on synthetic and real-world data demonstrate the efficacy of the proposed approach.