Offline Policy Optimization with Eligible Actions

TL;DR

This paper addresses overfitting in offline policy optimization by introducing a normalization constraint, demonstrating improved performance and reduced overfitting in healthcare and control tasks.

Contribution

It proposes a novel per-state-neighborhood normalization algorithm to mitigate overfitting in importance-weighted offline policy optimization, with theoretical and empirical validation.

Findings

Reduced overfitting in policy learning

Improved test performance over existing methods

Effective in healthcare and control environments

Abstract

Offline policy optimization could have a large impact on many real-world decision-making problems, as online learning may be infeasible in many applications. Importance sampling and its variants are a commonly used type of estimator in offline policy evaluation, and such estimators typically do not require assumptions on the properties and representational capabilities of value function or decision process model function classes. In this paper, we identify an important overfitting phenomenon in optimizing the importance weighted return, in which it may be possible for the learned policy to essentially avoid making aligned decisions for part of the initial state space. We propose an algorithm to avoid this overfitting through a new per-state-neighborhood normalization constraint, and provide a theoretical justification of the proposed algorithm. We also show the limitations of previous attempts to this approach. We test our algorithm in a healthcare-inspired simulator, a logged dataset collected from real hospitals and continuous control tasks. These experiments show the proposed method yields less overfitting and better test performance compared to state-of-the-art batch reinforcement learning algorithms.