Domain Adaptation and Multi-Domain Adaptation for Neural Machine Translation: A Survey

TL;DR

This survey reviews various domain and multi-domain adaptation techniques for Neural Machine Translation, emphasizing methods that improve translation quality across different domains while addressing challenges like overfitting and catastrophic forgetting.

Contribution

It categorizes and analyzes existing adaptation methods, highlighting their benefits and challenges in multi-domain NMT scenarios.

Findings

Data selection and generation improve domain adaptation.

Model architecture modifications enhance robustness.

Multi-domain adaptation benefits other NMT research areas.

Abstract

The development of deep learning techniques has allowed Neural Machine Translation (NMT) models to become extremely powerful, given sufficient training data and training time. However, systems struggle when translating text from a new domain with a distinct style or vocabulary. Fine-tuning on in-domain data allows good domain adaptation, but requires sufficient relevant bilingual data. Even if this is available, simple fine-tuning can cause overfitting to new data and `catastrophic forgetting' of previously learned behaviour. We concentrate on robust approaches to domain adaptation for NMT, particularly where a system may need to translate across multiple domains. We divide techniques into those revolving around data selection or generation, model architecture, parameter adaptation procedure, and inference procedure. We finally highlight the benefits of domain adaptation and multi-domain adaptation techniques to other lines of NMT research.