Encoding linear models as weighted finite-state transducers

26 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Sequence-to-sequence neural net models for grapheme-to-phoneme conversion (2015, methodology): “The Pronlex data contains 83182 words for training, 1000 words for validation, and 4800 words for testing.”
Token-Level Ensemble Distillation for Grapheme-to-Phoneme Conversion (2019, methodology): “Joint sequence n-gram models have been widely used [4, 14, 15] for G2P conversion.”
The SIGMORPHON 2020 Shared Task on Multilingual Grapheme-to-Phoneme Conversion (2020, methodology): “…as a machine learning problem, either using generative models expressed as weighted ﬁnite-state transducers (e.g., Taylor 2005, Bisani and Ney 2008, Wu et al. 2014, Novak et al. 2016) or discriminative models based on conditional random ﬁelds (Lehnen et al. 2013), recurrent neural networks (e.g.,…”
Composition-based on-the-fly rescoring for salient n-gram biasing (2015, methodology): “We note that this WFST encoding is very similar to that for log-linear models in [4], and that paper proved that this representation is minimal for a given set of n-gram features.”