Skill-Driven Neurosymbolic State Abstractions
Alper Ahmetoglu, Steven James, Cameron Allen, Sam Lobel, David Abel, George Konidaris
Abstract
We consider how to construct state abstractions compatible with a given set of abstract actions, to obtain a well-formed abstract Markov decision process (MDP). We show that the Bellman equation suggests that abstract states should represent distributions over states in the ground MDP; we characterize the conditions under which the resulting process is Markov and approximately model-preserving, derive algorithms for constructing and planning with the abstract MDP, and apply them to a visual maze task. We generalize these results to the factored actions case, characterizing the conditions that result in factored abstract states and apply the resulting algorithm to Montezuma's Revenge. These results provide a powerful and principled framework for constructing neurosymbolic abstract Markov decision processes.
BibTeX
@inproceedings{
ahmetoglu2025skilldriven,
title={Skill-Driven Neurosymbolic State Abstractions},
author={Alper Ahmetoglu and Steven James and Cameron Allen and Sam Lobel and David Abel and George Konidaris},
booktitle={The Thirty-ninth Annual Conference on Neural Information Processing Systems},
year={2025},
url={https://openreview.net/forum?id=ggP2r9S9wY}
}