A pre-trained Transformer MT system outperforms an LSTM baseline on noisy text and gains further robustness from fine-tuning on noisy data without harming clean-text performance.
Robust Neural Machine Translation with Joint Textual and Phonetic Embedding
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Neural machine translation (NMT) is notoriously sensitive to noises, but noises are almost inevitable in practice. One special kind of noise is the homophone noise, where words are replaced by other words with similar pronunciations. We propose to improve the robustness of NMT to homophone noises by 1) jointly embedding both textual and phonetic information of source sentences, and 2) augmenting the training dataset with homophone noises. Interestingly, to achieve better translation quality and more robustness, we found that most (though not all) weights should be put on the phonetic rather than textual information. Experiments show that our method not only significantly improves the robustness of NMT to homophone noises, but also surprisingly improves the translation quality on some clean test sets.
fields
cs.CL 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
CUNI System for the WMT19 Robustness Task
A pre-trained Transformer MT system outperforms an LSTM baseline on noisy text and gains further robustness from fine-tuning on noisy data without harming clean-text performance.