English

Maximally Permissive Reward Machines

Machine Learning 2024-08-16 v1 Artificial Intelligence

Abstract

Reward machines allow the definition of rewards for temporally extended tasks and behaviors. Specifying "informative" reward machines can be challenging. One way to address this is to generate reward machines from a high-level abstract description of the learning environment, using techniques such as AI planning. However, previous planning-based approaches generate a reward machine based on a single (sequential or partial-order) plan, and do not allow maximum flexibility to the learning agent. In this paper we propose a new approach to synthesising reward machines which is based on the set of partial order plans for a goal. We prove that learning using such "maximally permissive" reward machines results in higher rewards than learning using RMs based on a single plan. We present experimental results which support our theoretical claims by showing that our approach obtains higher rewards than the single-plan approach in practice.

Keywords

Cite

@article{arxiv.2408.08059,
  title  = {Maximally Permissive Reward Machines},
  author = {Giovanni Varricchione and Natasha Alechina and Mehdi Dastani and Brian Logan},
  journal= {arXiv preprint arXiv:2408.08059},
  year   = {2024}
}

Comments

Paper accepted for publication at the European Conference on Artificial Intelligence (ECAI) 2024

R2 v1 2026-06-28T18:13:37.937Z