CLNov 9, 2024

An $\mathbf{L^*}$ Algorithm for Deterministic Weighted Regular Languages

AI2ETH Zurich
arXiv:2411.06228v21 citationsh-index: 10EMNLP
Originality Synthesis-oriented
AI Analysis

This work provides a method for gaining interpretable insights into complex model behaviors, but it is incremental as it extends an existing algorithm to a weighted setting.

The authors tackled the problem of extracting interpretable finite state automata from black-box models by developing a weighted variant of Angluin's L* algorithm, which exactly learns deterministic weighted FSAs with weights supporting division and directly yields a minimal automaton.

Extracting finite state automata (FSAs) from black-box models offers a powerful approach to gaining interpretable insights into complex model behaviors. To support this pursuit, we present a weighted variant of Angluin's (1987) $\mathbf{L^*}$ algorithm for learning FSAs. We stay faithful to the original algorithm, devising a way to exactly learn deterministic weighted FSAs whose weights support division. Furthermore, we formulate the learning process in a manner that highlights the connection with FSA minimization, showing how $\mathbf{L^*}$ directly learns a minimal automaton for the target language.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes