Offline Policy Comparison with Confidence: Benchmarks and Baselines

TL;DR

This paper introduces benchmarks and baselines for offline policy comparison with confidence estimates, evaluating model-based ensemble methods to improve statistical reliability in offline reinforcement learning.

Contribution

It creates standardized benchmarks for offline policy comparison with confidence, and empirically evaluates ensemble-based model methods for confidence estimation.

Findings

Certain baseline variations show advantages in confidence estimation.

Significant room for improvement remains in current methods.

Benchmarks facilitate future research in offline policy evaluation.

Abstract

Decision makers often wish to use offline historical data to compare sequential-action policies at various world states. Importantly, computational tools should produce confidence values for such offline policy comparison (OPC) to account for statistical variance and limited data coverage. Nevertheless, there is little work that directly evaluates the quality of confidence values for OPC. In this work, we address this issue by creating benchmarks for OPC with Confidence (OPCC), derived by adding sets of policy comparison queries to datasets from offline reinforcement learning. In addition, we present an empirical evaluation of the risk versus coverage trade-off for a class of model-based baselines. In particular, the baselines learn ensembles of dynamics models, which are used in various ways to produce simulations for answering queries with confidence values. While our results suggest advantages for certain baseline variations, there appears to be significant room for improvement in future work.