Generative Neural Machine Translation
This work addresses translation robustness and data efficiency for NLP researchers, though it is incremental as it builds on existing encoder-decoder models.
The authors tackled the problem of improving neural machine translation by introducing a latent variable to model sentence semantics, achieving competitive BLEU scores and superior performance with missing source words, while enabling multilingual translation and semi-supervised learning without extra parameters.
We introduce Generative Neural Machine Translation (GNMT), a latent variable architecture which is designed to model the semantics of the source and target sentences. We modify an encoder-decoder translation model by adding a latent variable as a language agnostic representation which is encouraged to learn the meaning of the sentence. GNMT achieves competitive BLEU scores on pure translation tasks, and is superior when there are missing words in the source sentence. We augment the model to facilitate multilingual translation and semi-supervised learning without adding parameters. This framework significantly reduces overfitting when there is limited paired data available, and is effective for translating between pairs of languages not seen during training.