Learning Symbolic Rules over Abstract Meaning Representations for Textual Reinforcement Learning

Text-based reinforcement learning agents have predominantly been neural network-based models with embeddings-based representation, learning uninterpretable policies that often do not generalize well to unseen games. On the other hand, neuro-symbolic methods, specifically those that leverage an inter...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2023-07
Main Authors: Chaudhury, Subhajit, Swaminathan, Sarathkrishna, Kimura, Daiki, Sen, Prithviraj, Murugesan, Keerthiram, Uceda-Sosa, Rosario, Tatsubori, Michiaki, Fokoue, Achille, Kapanipathi, Pavan, Munawar, Asim, Gray, Alexander
Format: Article
Language:English
Subjects:
Deep learning
Games
Machine learning
Neural networks
Policies
Representations
Rule induction
Training
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Text-based reinforcement learning agents have predominantly been neural network-based models with embeddings-based representation, learning uninterpretable policies that often do not generalize well to unseen games. On the other hand, neuro-symbolic methods, specifically those that leverage an intermediate formal representation, are gaining significant attention in language understanding tasks. This is because of their advantages ranging from inherent interpretability, the lesser requirement of training data, and being generalizable in scenarios with unseen data. Therefore, in this paper, we propose a modular, NEuro-Symbolic Textual Agent (NESTA) that combines a generic semantic parser with a rule induction system to learn abstract interpretable rules as policies. Our experiments on established text-based game benchmarks show that the proposed NESTA method outperforms deep reinforcement learning-based techniques by achieving better generalization to unseen test games and learning from fewer training interactions.
ISSN:2331-8422