Deep Neural Machine Translation with Weakly-Recurrent Units
This work addresses computational bottlenecks in machine translation for researchers and practitioners, offering a more efficient alternative to existing RNNs, though it is incremental as it builds on prior recurrent architectures.
The authors tackled the inefficiency of recurrent neural networks (RNNs) in neural machine translation by proposing a new architecture called Simple Recurrent NMT, which uses weakly-recurrent units with layer normalization and multiple attentions. Their model achieved better results than LSTMs on WMT14 English-to-German and WMT16 English-Romanian benchmarks at significantly lower computational cost.
Recurrent neural networks (RNNs) have represented for years the state of the art in neural machine translation. Recently, new architectures have been proposed, which can leverage parallel computation on GPUs better than classical RNNs. Faster training and inference combined with different sequence-to-sequence modeling also lead to performance improvements. While the new models completely depart from the original recurrent architecture, we decided to investigate how to make RNNs more efficient. In this work, we propose a new recurrent NMT architecture, called Simple Recurrent NMT, built on a class of fast and weakly-recurrent units that use layer normalization and multiple attentions. Our experiments on the WMT14 English-to-German and WMT16 English-Romanian benchmarks show that our model represents a valid alternative to LSTMs, as it can achieve better results at a significantly lower computational cost.