Scholarlink

We present an adaptation of the \(L^{\#}\) algorithm, originally designed to learn Mealy Machines, for active learning of Deterministic Finite Automata (DFAs). The main changes are an adaptation of the underlying observation tree data structure and the definition of apartness. The top level algorithm remains unchanged, modulo a few details. The algorithm has been implemented in Rust, by adapting the code of the Mealy machine version. Initial experiments suggest that \(L^{\#}\) is competitive with state-of-the-art active learning algorithms for DFAs, as implemented in the LearnLib and AALpy tools, in particular when adaptive membership queries can be used.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

\(L^{\#}\) for DFAs

  • 章节

  • 发表日期：

    2025年

查看全文
  • Frits Vaandrager,
  • Martijn Sanders

摘要

We present an adaptation of the \(L^{\#}\) algorithm, originally designed to learn Mealy Machines, for active learning of Deterministic Finite Automata (DFAs). The main changes are an adaptation of the underlying observation tree data structure and the definition of apartness. The top level algorithm remains unchanged, modulo a few details. The algorithm has been implemented in Rust, by adapting the code of the Mealy machine version. Initial experiments suggest that \(L^{\#}\) is competitive with state-of-the-art active learning algorithms for DFAs, as implemented in the LearnLib and AALpy tools, in particular when adaptive membership queries can be used.