Uncertainty-Aware Balancing for Multilingual and Multi-Domain Neural Machine Translation Training

TL;DR

This paper introduces MultiUAT, a dynamic data balancing method for multilingual and multi-domain neural machine translation that adjusts training based on model uncertainty, improving performance over static methods.

Contribution

The paper proposes MultiUAT, a novel uncertainty-based dynamic balancing approach for NMT training across multiple languages and domains, addressing limitations of existing static and similarity-based methods.

Findings

MultiUAT outperforms static and dynamic baselines in multilingual translation tasks.

Uncertainty measures effectively guide data balancing in multi-corpus NMT.

Static and similarity-based methods show deficiencies in cross-domain transfer.

Abstract

Learning multilingual and multi-domain translation model is challenging as the heterogeneous and imbalanced data make the model converge inconsistently over different corpora in real world. One common practice is to adjust the share of each corpus in the training, so that the learning process is balanced and low-resource cases can benefit from the high resource ones. However, automatic balancing methods usually depend on the intra- and inter-dataset characteristics, which is usually agnostic or requires human priors. In this work, we propose an approach, MultiUAT, that dynamically adjusts the training data usage based on the model's uncertainty on a small set of trusted clean data for multi-corpus machine translation. We experiments with two classes of uncertainty measures on multilingual (16 languages with 4 settings) and multi-domain settings (4 for in-domain and 2 for out-of-domain on English-German translation) and demonstrate our approach MultiUAT substantially outperforms its baselines, including both static and dynamic strategies. We analyze the cross-domain transfer and show the deficiency of static and similarity based methods.