Heterogeneous Knowledge for Augmented Modular Reinforcement Learning

TL;DR

This paper introduces Augmented Modular Reinforcement Learning (AMRL), a framework that integrates heterogeneous knowledge sources like rules and skills into modular RL to enhance performance, generalization, safety, and interpretability.

Contribution

It proposes a novel AMRL framework that combines heterogeneous knowledge modules with a selector mechanism, addressing limitations of homogeneous modular RL architectures.

Findings

Improved performance and efficiency in RL tasks.

Enhanced generalization capabilities.

Insights into safety and interpretability issues.

Abstract

Existing modular Reinforcement Learning (RL) architectures are generally based on reusable components, also allowing for "plug-and-play" integration. However, these modules are homogeneous in nature - in fact, they essentially provide policies obtained via RL through the maximization of individual reward functions. Consequently, such solutions still lack the ability to integrate and process multiple types of information (i.e., heterogeneous knowledge representations), such as rules, sub-goals, and skills from various sources. In this paper, we discuss several practical examples of heterogeneous knowledge and propose Augmented Modular Reinforcement Learning (AMRL) to address these limitations. Our framework uses a selector to combine heterogeneous modules and seamlessly incorporate different types of knowledge representations and processing mechanisms. Our results demonstrate the performance and efficiency improvements, also in terms of generalization, that can be achieved by augmenting traditional modular RL with heterogeneous knowledge sources and processing mechanisms. Finally, we examine the safety, robustness, and interpretability issues stemming from the introduction of knowledge heterogeneity.