Shared-Private Bilingual Word Embeddings for Neural Machine Translation
This work addresses parameter efficiency and performance in neural machine translation, offering an incremental improvement over existing methods.
The paper tackles the problem of isolated source and target word embeddings in neural machine translation by proposing shared-private bilingual word embeddings, which reduce model parameters and improve performance, achieving significant boosts across 5 language pairs with dramatically fewer parameters.
Word embedding is central to neural machine translation (NMT), which has attracted intensive research interest in recent years. In NMT, the source embedding plays the role of the entrance while the target embedding acts as the terminal. These layers occupy most of the model parameters for representation learning. Furthermore, they indirectly interface via a soft-attention mechanism, which makes them comparatively isolated. In this paper, we propose shared-private bilingual word embeddings, which give a closer relationship between the source and target embeddings, and which also reduce the number of model parameters. For similar source and target words, their embeddings tend to share a part of the features and they cooperatively learn these common representation units. Experiments on 5 language pairs belonging to 6 different language families and written in 5 different alphabets demonstrate that the proposed model provides a significant performance boost over the strong baselines with dramatically fewer model parameters.