Neural Machine Translation of Mauritian Creole to English Using Transfer Learning
Vazeerudeen Abdul Hameed, Muhammad Ehsan Rana, Neelesh Nursiah · 2023
The discipline of Language Translation to convert from a source language to a target language has been attempted to be automated for quite some time. Neural Machine Translation (NMT) is a field where a source language is automatically converted to a target language without much human intervention using Deep Learning (DL) techniques. Neural Machine Translation has gained appeal due to its swift and reliable accuracy for automatic language translation. However, due to ambiguous syntax and sentence logic structure, it is not a trivial task that can be solved straightforwardly, making this an extant field of research. Till now, a colossal amount of research and funding has been invested in this field. This paper attempts to translate text from Mauritian Creole Language to English using Bi-Directional Long Short-Term Memory (LSTM) with the Bahdanau Attention mechanism by using transferable knowledge (Transfer Learning) of French to English translation, as French and Mauritian Creole are almost similar in language structure. A model to accomplish the translation has been proposed and validated using appropriate sample data to determine the accuracy of the translation.