Dynamic probabilistic logic models for effective abstractions in RL

TL;DR

This paper introduces a dynamic probabilistic logic framework within RePReL to improve state abstraction, leading to more sample-efficient learning and better generalization in reinforcement learning tasks.

Contribution

It presents a novel use of dynamic probabilistic logic models for state abstraction in hierarchical RL, enhancing performance and transferability.

Findings

RePReL achieves superior task performance.

It demonstrates improved generalization to unseen tasks.

The framework enables more efficient learning.

Abstract

State abstraction enables sample-efficient learning and better task transfer in complex reinforcement learning environments. Recently, we proposed RePReL (Kokel et al. 2021), a hierarchical framework that leverages a relational planner to provide useful state abstractions for learning. We present a brief overview of this framework and the use of a dynamic probabilistic logic model to design these state abstractions. Our experiments show that RePReL not only achieves better performance and efficient learning on the task at hand but also demonstrates better generalization to unseen tasks.