Translate at Hyper Speed with Facebook

I wish that the language barrier did not exist, but since that will never go away I wish a quick, accurate translation device existed.  We live in a time when such a device is not science-fiction, but only decades, possibly only years away.  Companies are experimenting with new algorithms, AI, and other processes to speed up translation and accuracy.  Bitext relies on breakthrough Deep Linguistic Analysis, while Digital Trends reports that Facebook is working on speed in “Facebook IS Using AI To Make Language Translation Nine Times Faster.”

Facebook reports that their artificial intelligence translates foreign languages nine times faster than the traditional language software.  What is even more astonishing is that the source code is open source, so anyone can download and use it.  Facebook uses convolutional neural networks (CNN) to translate.  CNNs are not new, but this is the first time they have been used for translation.  How does it work?

The report highlights the use of convolutional neural networks (CNN) as opposed to recurrent neural networks (RNN), which translate sentences one word at a time in a linear order. The new architecture, however, can take words further down in the sentence into consideration during the translation process, which helps make the translation far more accurate. This actually marks the first time a CNN has managed to outperform an RNN in language translation, and Facebook now hopes to expand it to to cover more languages.

I think the open source aspect is the most important part.  Language translation software relies on a lot of data in order to have decent comprehension skills.  Open source users tend to share and share alike, so we can rely on them feeding huge data piles to the code.

Whitney Grace, May 30, 2017

 

Comments

comments