Memory-enhanced Decoder for Neural Machine Translation
This work addresses translation quality for machine translation systems, showing incremental gains with a novel memory mechanism.
The paper tackled the problem of improving neural machine translation by enhancing the RNN decoder with external memory, resulting in a 4.8 BLEU improvement over Groundhog and 5.3 BLEU over Moses on Chinese-English translation.
We propose to enhance the RNN decoder in a neural machine translator (NMT) with external memory, as a natural but powerful extension to the state in the decoding RNN. This memory-enhanced RNN decoder is called \textsc{MemDec}. At each time during decoding, \textsc{MemDec} will read from this memory and write to this memory once, both with content-based addressing. Unlike the unbounded memory in previous work\cite{RNNsearch} to store the representation of source sentence, the memory in \textsc{MemDec} is a matrix with pre-determined size designed to better capture the information important for the decoding process at each time step. Our empirical study on Chinese-English translation shows that it can improve by $4.8$ BLEU upon Groundhog and $5.3$ BLEU upon on Moses, yielding the best performance achieved with the same training set.