Beyond Neural MT image

Beyond Neural MT

May 29th 2018 07:00 - 07:20

A lot of excitement has surrounded the latest advances in Neural Machine Translation (NMT), but some of the claims need to be qualified and practical implementation of NMT is not without difficulty: typically double the training material is required when compared to Statistical Machine Translation (SMT) and training a new engine can take weeks not days. Most practical translation projects do not have anywhere near enough training data and do not have the luxury of waiting weeks for an engine to be trained if there was enough data. In addition most quality SMT systems require ‘tuning’ to get the best results, something that is not possible with NMT as there is no indication possible of why a given translation is selected. This presentation will look at the practical limitations of NMT and at what is on the horizon beyond NMT that offers answers to many of NMT’s limitations.

Andrzej Zydroń avatar
Andrzej Zydroń

CTO at XTM International:

Andrzej is one of the leading IT experts on Localization and related Open Standards. Zydroń sits/has sat on many Open Standard Technical Committees, amongst them LISA OSCAR, W3C, OASIS, ETSI LIS, etc. Zydroń has been responsible for the architecture o...


This website uses cookies. By continuing to browse you agree to this and Conferize's terms of service.