Abstraction for Offline Goal-Conditioned Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wibault, Clarisse, Goldie, Alexander, Villares, Antonio, Osborne, Maike, Foerster, Jakob
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911705077907456
author Wibault, Clarisse
Goldie, Alexander
Villares, Antonio
Osborne, Maike
Foerster, Jakob
author_facet Wibault, Clarisse
Goldie, Alexander
Villares, Antonio
Osborne, Maike
Foerster, Jakob
contents Markov Decision Processes (MDPs) often exhibit significant redundancy due to symmetries and shared structure across state-goal pairs in real-world Goal-Conditioned Reinforcement Learning (GCRL). While hierarchical policies have been motivated for horizon reduction via temporal abstraction in offline GCRL, we demonstrate that hierarchy also enables absolute abstraction. By introducing relativised options as well as distinct representations for different levels of the hierarchy, we demonstrate how an agent can reuse experience across similar contexts of the state-space. Based on this framework, we introduce two simple algorithms for learning relativised options and abstracting from the absolute frame of reference. Our experiments show that such inductive biases significantly improve performance in offline GCRL.
format Preprint
id arxiv_https___arxiv_org_abs_2605_22711
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Abstraction for Offline Goal-Conditioned Reinforcement Learning
Wibault, Clarisse
Goldie, Alexander
Villares, Antonio
Osborne, Maike
Foerster, Jakob
Machine Learning
Artificial Intelligence
Markov Decision Processes (MDPs) often exhibit significant redundancy due to symmetries and shared structure across state-goal pairs in real-world Goal-Conditioned Reinforcement Learning (GCRL). While hierarchical policies have been motivated for horizon reduction via temporal abstraction in offline GCRL, we demonstrate that hierarchy also enables absolute abstraction. By introducing relativised options as well as distinct representations for different levels of the hierarchy, we demonstrate how an agent can reuse experience across similar contexts of the state-space. Based on this framework, we introduce two simple algorithms for learning relativised options and abstracting from the absolute frame of reference. Our experiments show that such inductive biases significantly improve performance in offline GCRL.
title Abstraction for Offline Goal-Conditioned Reinforcement Learning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2605.22711