Contextual Parameter Generation for Universal Neural Machine Translation
This addresses the need for efficient and adaptable translation systems across languages, though it is an incremental improvement over existing NMT models.
The authors tackled the problem of building a single neural machine translation model for multiple languages by introducing a contextual parameter generator that creates language-specific parameters, enabling the use of monolingual data and zero-shot translation. They achieved state-of-the-art performance on IWSLT-15 and IWSLT-17 datasets.
We propose a simple modification to existing neural machine translation (NMT) models that enables using a single universal model to translate between multiple languages while allowing for language specific parameterization, and that can also be used for domain adaptation. Our approach requires no changes to the model architecture of a standard NMT system, but instead introduces a new component, the contextual parameter generator (CPG), that generates the parameters of the system (e.g., weights in a neural network). This parameter generator accepts source and target language embeddings as input, and generates the parameters for the encoder and the decoder, respectively. The rest of the model remains unchanged and is shared across all languages. We show how this simple modification enables the system to use monolingual data for training and also perform zero-shot translation. We further show it is able to surpass state-of-the-art performance for both the IWSLT-15 and IWSLT-17 datasets and that the learned language embeddings are able to uncover interesting relationships between languages.