# Sample Efficient Policy Search for Optimal Stopping Domains

## TL;DR

This paper introduces GFSE, a model-free policy search method for optimal stopping problems that improves sample efficiency and guarantees convergence, outperforming existing methods across multiple domains.

## Contribution

The paper presents GFSE, a novel, flexible policy search approach that reuses data effectively and provides theoretical guarantees with logarithmic horizon dependence.

## Key findings

- GFSE outperforms model-based and model-free baselines in experiments.
- Sample complexity bounds are tightened with logarithmic horizon dependence.
- The method is applicable across diverse optimal stopping domains.

## Abstract

Optimal stopping problems consider the question of deciding when to stop an observation-generating process in order to maximize a return. We examine the problem of simultaneously learning and planning in such domains, when data is collected directly from the environment. We propose GFSE, a simple and flexible model-free policy search method that reuses data for sample efficiency by leveraging problem structure. We bound the sample complexity of our approach to guarantee uniform convergence of policy value estimates, tightening existing PAC bounds to achieve logarithmic dependence on horizon length for our setting. We also examine the benefit of our method against prevalent model-based and model-free approaches on 3 domains taken from diverse fields.