Active Automata Learning with Advice

2025-08-14

AI Summary

To address the excessive query complexity and strong dependence on the teacher in active automata learning, this paper proposes a suggestion-driven learning framework integrated with deductive reasoning. Without altering the structural design of Angluin-style algorithms, the method employs a string rewriting system as an external knowledge representation to formally derive answers to membership and equivalence queries, thereby reducing explicit teacher interactions. The key contribution lies in encoding domain knowledge into computable rewriting rules, enabling symbolic inference of query answers while preserving language equivalence guarantees. Experimental evaluation across multiple benchmark models demonstrates substantial reductions in both membership and equivalence query counts; average query complexity decreases by 30%–65% compared to conventional active learning approaches, consistently outperforming them in efficiency without compromising correctness.

Technology Category

• Machine Learning: Active Learning
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

We present an extended automata learning framework that combines active automata learning with deductive inference. The learning algorithm asks membership and equivalence queries as in the original framework, but it is also given advice, which is used to infer answers to queries when possible and reduce the burden on the teacher. We consider advice given via string rewriting systems, which specify equivalence of words w.r.t. the target languages. The main motivation for the proposed framework is to reduce the number of queries. We show how to adapt Angluin-style learning algorithms to this framework with low overhead. Finally, we present empirical evaluation of our approach and observe substantial improvement in query complexity.

Problem

• Combines active automata learning with deductive inference
• Reduces query burden via string rewriting advice
• Improves query complexity in automata learning

Innovation

• Combines active learning with deductive inference
• Uses string rewriting systems as advice
• Reduces query complexity significantly