Domain Adaptation of NMT models for English-Hindi Machine Translation Task at AdapMT ICON 2020

TL;DR

This paper explores domain adaptation techniques for English-Hindi neural machine translation, demonstrating effective fine-tuning and mixed-domain training methods that improve translation quality in specific domains.

Contribution

It compares LSTM and Transformer models for domain-specific NMT and applies simple adaptation techniques, achieving top rankings in shared task evaluations.

Findings

Transformer outperforms LSTM in BLEU scores

Fine-tuning improves domain-specific translation quality

Simple adaptation methods are effective for low-resource domains

Abstract

Recent advancements in Neural Machine Translation (NMT) models have proved to produce a state of the art results on machine translation for low resource Indian languages. This paper describes the neural machine translation systems for the English-Hindi language presented in AdapMT Shared Task ICON 2020. The shared task aims to build a translation system for Indian languages in specific domains like Artificial Intelligence (AI) and Chemistry using a small in-domain parallel corpus. We evaluated the effectiveness of two popular NMT models i.e, LSTM, and Transformer architectures for the English-Hindi machine translation task based on BLEU scores. We train these models primarily using the out of domain data and employ simple domain adaptation techniques based on the characteristics of the in-domain dataset. The fine-tuning and mixed-domain data approaches are used for domain adaptation. Our team was ranked first in the chemistry and general domain En-Hi translation task and second in the AI domain En-Hi translation task.