Issue #45 – Improving Robustness in Real-World Neural Machine Translation

11 Jul19

Author: Dr. John Tinsley, CEO & Co-founder @ Iconic

Next month, the 17th Machine Translation Summit will take place in Dublin, Ireland and the Iconic team will be in attendance. Not only that, we will be presenting our own work – Gupta et al. (2019) – on some of the steps we take to improve the robustness, stability, and quality of the Neural MT engines that we run in production for our clients. In this week’s post, we are staying a little closer to home and will review some of the key topics we covered in the above work.

Getting “Real-World” Ready

In the previous 44 issues of this series, we have reviewed many cutting-edge approaches in Neural MT research and development. However, we are not just paying lip service. We practice what we preach and look to incorporate as many of these as possible into our own software. That being said, many of these approaches are prototypical, or at an early research stage. As a commercial provider of machine translation, we need to rigorously test the implementations to make sure they stand
To finish reading, please visit source site