Neural Induction of Finite-State Transducers
This addresses the difficulty of manually building FSTs for applications requiring efficiency, though it is incremental as it builds on neural network insights.
The paper tackles the problem of automatically constructing Finite-State Transducers (FSTs) for string-to-string rewriting tasks, achieving up to 87% higher accuracy than classical methods on real-world datasets like morphological inflection.
Finite-State Transducers (FSTs) are effective models for string-to-string rewriting tasks, often providing the efficiency necessary for high-performance applications, but constructing transducers by hand is difficult. In this work, we propose a novel method for automatically constructing unweighted FSTs following the hidden state geometry learned by a recurrent neural network. We evaluate our methods on real-world datasets for morphological inflection, grapheme-to-phoneme prediction, and historical normalization, showing that the constructed FSTs are highly accurate and robust for many datasets, substantially outperforming classical transducer learning algorithms by up to 87% accuracy on held-out test sets.