Mitigating Data Imbalance and Representation Degeneration in Multilingual Machine Translation
Submission Type: Regular Long Paper
Submission Track: Machine Translation
Keywords: Data Imbalance, Representation Degeneration, Multilingual Machine Translation
TL;DR: We propose a novel approach to mitigate the data imbalance and representation degeneration issues in multilingual machine translation which only requires target-side monolingual data and a bilingual dictionary.
Abstract: Despite advances in multilingual neural machine translation (MNMT), we argue that there are still two major challenges in this area: data imbalance and representation degeneration. The data imbalance problem refers to the imbalance in the amount of parallel corpora for all language pairs, especially for long-tail languages (i.e., very low-resource languages). The representation degeneration problem refers to the problem of encoded tokens tending to appear only in a small subspace of the full space available to the MNMT model. To solve these two issues, we propose Bi-ACL, a framework which only requires target-side monolingual data and a bilingual dictionary to improve the performance of the MNMT model. We define two modules, named bidirectional autoencoder and bidirectional contrastive learning, which we combine with an online constrained beam search and a curriculum learning sampling strategy. Extensive experiments show that our proposed method is more effective than strong baselines both in long-tail languages and in high-resource languages. We also demonstrate that our approach is capable of transferring knowledge between domains and languages in zero-shot scenarios.
Submission Number: 3650
Loading