One of the biggest advantages of using Facebook as your primary social networking site is that you get to use it in your local language as it supports a number of languages. To unify the networking platform for all of its users, Facebook’s inbuilt translation feature is pretty impressive and useful indeed. To bring the level up, Facebook has recently introduced a newer method of machine learning which will provide faster learning as compared to the competitors.
Facebook’s Progress So Far
Facebook has been quite impressive in overcoming the language barrier by introducing timely tweaks in the Translation AI. Features like Auto translate are very helpful in making the service a fully functional one. To reach its goal of making the world more open and connected, Facebook has introduced a translation prototype which is claimed to be 9 times faster than that of its rivals.
Work In Progress
According to the Facebook’s AI engineers Michael Auli and David Grangier, the work is still in progress and there are certain factors that need to be addressed before the newer prototype is made live. Their statement in this regard goes like:
We’re currently talking with a product team to make this work in a Facebook environment. There are differences when moving from academic data to real environments in terms of language. The academic data is news-type data; while conversation on Facebook is much more colloquial.
A Technology Shift
To achieve this much high speed for translation, Facebook instead of relying on the conventional “Recurrent Neural Networks” (RNNs), has brought “Convolutional Neural Networks” (CNNs) to the scene. This makes the translation feature easier and more practical. This kind of Machine Learning can also provide translation for some slang words which Facebook aims to inculcate in their program.
Engineers Michael Auli explains the feature like:
The [CNNs] build a logical structure, a bit like linguistics, on top of the text.
Facebook’s Approach Is Completely New
Taking advantage of what others were lacking, Facebook will now release the newer AI with faster speeds and better results. The world invested all the time in RNNs while completely forgetting the scope of CNN. This gap is now filled by Facebook and it will create a whole new discipline in the translation technology. David Grangier’s statement goes like:
The short answer is that people just hadn’t invested as much time in this, and we came up with some new developments that made it work better.