Towards Knowledge-Augmented Agents for Efficient and Interpretable Learning in Sequential Decision Problems

Reem Alansary, Nourhan Ehab

2024

Abstract

The advantages of neurosymbolic systems as solvers of sequential decision problems have captured the attention of reseachers in the field of AI. The combination of perception and cognition allows for constructing learning agents with memory. In this position paper, we argue that the decision-making abilities of such knowledge-augmented solvers transcend those of black-box function approximators alone as the former can generalize through inductive reasoning to behave optimally in unknown states and still remain fully or partially interpretable. We present a novel approach leveraging a knowledge base structured as a layered directed acyclic graph, facilitating reasoned generalization in the absence of complete information.

Paper Citation

in Harvard Style

Alansary R. and Ehab N. (2024). Towards Knowledge-Augmented Agents for Efficient and Interpretable Learning in Sequential Decision Problems. In Proceedings of the 16th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART; ISBN 978-989-758-680-4, SciTePress, pages 1014-1019. DOI: 10.5220/0012430900003636

in Bibtex Style

@conference{icaart24,
author={Reem Alansary and Nourhan Ehab},
title={Towards Knowledge-Augmented Agents for Efficient and Interpretable Learning in Sequential Decision Problems},
booktitle={Proceedings of the 16th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART},
year={2024},
pages={1014-1019},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012430900003636},
isbn={978-989-758-680-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 16th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART
TI - Towards Knowledge-Augmented Agents for Efficient and Interpretable Learning in Sequential Decision Problems
SN - 978-989-758-680-4
AU - Alansary R.
AU - Ehab N.
PY - 2024
SP - 1014
EP - 1019
DO - 10.5220/0012430900003636
PB - SciTePress