Uncertainty-Aware Balancing for Multilingual and Multi-Domain Neural Machine Translation Training
This work addresses data imbalance issues in multilingual and multi-domain machine translation, offering a practical solution for improving model convergence and performance in real-world applications, though it is incremental relative to existing balancing methods.
The paper tackles the challenge of training multilingual and multi-domain neural machine translation models with heterogeneous and imbalanced data by proposing MultiUAT, which dynamically adjusts training data usage based on model uncertainty on trusted clean data. The approach substantially outperforms baselines in experiments across 16 languages and multiple domains, showing improved performance over static and dynamic strategies.
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.