What question did this study set out to answer?

The aim is to improve the efficiency of controller synthesis methods using a novel approach that leverages graph structures.

February 26, 2026IEICE Transactions on Information and Systems0 citationsOpen Access

Graph Contextual Reinforcement Learning for Efficient Directed Controller Synthesis

TUToshihide UBUKATA EMEnhong MU TYTakuto Yamauchi

Key Points

The aim is to improve the efficiency of controller synthesis methods using a novel approach that leverages graph structures.
Introduced GCRL, integrating Graph Neural Networks with traditional reinforcement learning.
Encoded the history of Labeled Transition System exploration into graph structures.
Conducted comparative experiments against existing synthesis methods.
GCRL demonstrated improved learning efficiency in most benchmark domains.
Outperformed state-of-the-art methods in four out of five tested scenarios.
Experienced limitations only in a domain with high symmetry and local interactions.

Abstract

Controller synthesis is a formal method approach for automatically generating Labeled Transition System (LTS) controllers that satisfy specified properties. The efficiency of the synthesis process, however, is critically dependent on exploration policies. These policies often rely on fixed rules or strategies learned through reinforcement learning (RL) that consider only a limited set of current features. To address this limitation, this paper introduces GCRL, an approach that enhances RL-based methods by integrating Graph Neural Networks (GNNs). GCRL encodes the history of LTS exploration into a graph structure, allowing it to capture a broader, non-current-based context. In a comparative experiment against state-of-the-art methods, GCRL exhibited superior learning efficiency and generalization across four out of five benchmark domains, except one particular domain characterized by high symmetry and strictly local interactions.

AI से पूछें

Bookmark

View Full Paper