← Back to papers
2018 ACL ACL 2018

On the Impact of Various Types of Noise on Neural Machine Translation

Abstract

AbstractWe examine how various types of noise in the parallel training data impact the quality of neural machine translation systems. We create five types of artificial noise and analyze how they degrade performance in neural and statistical machine translation. We find that neural models are generally more harmed by noise than statistical models. For one especially egregious type of noise they learn to just copy the input sentence.

🌉 Interdisciplinary Bridge — Machine Learning and Natural Language Processing
🧭 Keyword Pioneer — parallel training datum
🐣 Hot Topic Early Bird — neural machine translation
🐝 Cross-Pollinator — Artificial Intelligence, Computer Vision, Deep Learning, Interdisciplinary, Machine Learning, Natural Language Processing, Speech & Audio

Authors

Huda Khayrallah , Philipp Koehn

Topics

Machine Learning > Optimization & Theory > Learning Theory Natural Language Processing > Applications > Machine Translation Natural Language Processing > Generation > Machine Translation Machine Learning > Learning Types > Evaluation Deep Learning > Learning Types > Deep Learning

Keywords

model robustness neural machine translation statistical machine translation noise handling translation quality parallel training datum noise impact translation quality degradation
Download PDF

Related papers

Economic Event Detection in Company-Specific News Text 2018
Investigating Effective Parameters for Fine-tuning of Word Embeddings Using Only a Small Corpus 2018
SemAxis: A Lightweight Framework to Characterize Domain-Specific Word Semantics Beyond Sentiment 2018
Fighting Offensive Language on Social Media with Unsupervised Text Style Transfer 2018
Affordances in Grounded Language Learning 2018