English

Options of Interest: Temporal Abstraction with Interest Functions

Machine Learning 2020-01-03 v1 Artificial Intelligence Machine Learning

Abstract

Temporal abstraction refers to the ability of an agent to use behaviours of controllers which act for a limited, variable amount of time. The options framework describes such behaviours as consisting of a subset of states in which they can initiate, an internal policy and a stochastic termination condition. However, much of the subsequent work on option discovery has ignored the initiation set, because of difficulty in learning it from data. We provide a generalization of initiation sets suitable for general function approximation, by defining an interest function associated with an option. We derive a gradient-based learning algorithm for interest functions, leading to a new interest-option-critic architecture. We investigate how interest functions can be leveraged to learn interpretable and reusable temporal abstractions. We demonstrate the efficacy of the proposed approach through quantitative and qualitative results, in both discrete and continuous environments.

Keywords

Cite

@article{arxiv.2001.00271,
  title  = {Options of Interest: Temporal Abstraction with Interest Functions},
  author = {Khimya Khetarpal and Martin Klissarov and Maxime Chevalier-Boisvert and Pierre-Luc Bacon and Doina Precup},
  journal= {arXiv preprint arXiv:2001.00271},
  year   = {2020}
}

Comments

To appear in Proceedings of the Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI-20)

R2 v1 2026-06-23T13:00:56.137Z