Felix Stahlberg

TL;DR: It is concluded that vanilla NMT in its current form requires just the right amount of beam search errors, which, from a modelling perspective, is a highly unsatisfactory conclusion indeed, as the model often prefers an empty translation.

TL;DR: This work traces back the origins of modern NMT architectures to word and sentence embeddings and earlier examples of the encoder-decoder network family and concludes with a survey of recent trends in the field.

TL;DR: Neural network models that treat text normalization for TTS as a sequence-to-sequence problem, in which the input is a text token in context, and the output is the verbalization of that token, are proposed.

TL;DR: The authors trace back the origins of modern NMT architectures to word and sentence embeddings and earlier examples of the encoder-decoder network family and conclude with a survey of recent trends in the field.

TL;DR: This paper investigated the use of hierarchical phrase-based SMT lattices in end-to-end neural machine translation (NMT), and showed that weight pushing transforms the Hiero scores for complete translation hypotheses, with the full translation grammar score and full n-gram language model score, into posteriors compatible with NMT predictive probabilities.