Last year a computer built by Google’s engineers team beat one of the world’s best players (the Korean grandmaster Lee Sedol) at the ancient game Go. Now the same team has brought their efforts together and implemented this into Google Translate.

When translating speech or text, it seems pretty obvious that one should read a whole sentence before figuring out what it means. But until now that hasn’t been easy for computers to do so, mainly because of the resources that it would require.

Last Tuesday, Google announced that they figured out a way to circumvent that problem, and to integrate their new solution into Google Translate. The new technology is called GNMT (Google Neural Machine Translation system) and can work on whole sentences (even analyze the context) before translating it! this improves the accuracy of translation on average by 60%.

To summarize and without jumping into a long romantic technical discussion, using deep neural network artificial intelligence allows Google Translate not only to translate but also to learn from the users input. Where the previous system took chunks of sentences and broke them up into individual words and phrases without considering the context, the new system are  capable of doing so, and even derive and analyze a sentence based on previous submits as well as book, articles and blog posts.

At Volo we are very proud to have a team with members from all over the world, while our headquarters are located in China fact remains that most of us are expats from countries all around the world. With most of our team members now speaking fluent Chinese fact remains that we all used (and still do) translations applications over our stay here in China! We salute Google for their efforts and hoping that future development within Artificial Intelligence and Deep Neural Networks can make it easier for people all over the world to understand each other.

You can read more over on top topic over at Google Research blog
https://research.googleblog.com/2016/09/a-neural-network-for-machine.html

Building Brands That Matter 👅