Leveraging Demonstrations to Improve Online Learning: Quality Matters
This work addresses the problem of enhancing online learning efficiency for practitioners by leveraging demonstration data, though it is incremental as it builds on existing Thompson sampling methods.
The paper investigates how offline demonstration data improves online learning, showing that improvement depends on data quality and increases with expert competence, with experiments demonstrating substantial regret reduction.
We investigate the extent to which offline demonstration data can improve online learning. It is natural to expect some improvement, but the question is how, and by how much? We show that the degree of improvement must depend on the quality of the demonstration data. To generate portable insights, we focus on Thompson sampling (TS) applied to a multi-armed bandit as a prototypical online learning algorithm and model. The demonstration data is generated by an expert with a given competence level, a notion we introduce. We propose an informed TS algorithm that utilizes the demonstration data in a coherent way through Bayes' rule and derive a prior-dependent Bayesian regret bound. This offers insight into how pretraining can greatly improve online performance and how the degree of improvement increases with the expert's competence level. We also develop a practical, approximate informed TS algorithm through Bayesian bootstrapping and show substantial empirical regret reduction through experiments.