Incremental Decoding and Training Methods for Simultaneous Translation\n in Neural Machine Translation

We address the problem of simultaneous translation by modifying the Neural MT\ndecoder to operate with dynamically built encoder and attention. We propose a\ntunable agent which decides the best segmentation strategy for a user-defined\nBLEU loss and Average Proportion (AP) constraint. Our agent outperforms\npreviously proposed Wait-if-diff and Wait-if-worse agents (Cho and Esipova,\n2016) on BLEU with a lower latency. Secondly we proposed data-driven changes to\nNeural MT training to better match the incremental decoding framework.\n