Google has created a new neural network, and it’s within striking distance of human-level translation.
Ten years ago Google Translate was released, and Phrase-Based Machine Translation became the key algorithm behind artificial translation. Since then, rapid advances in machine intelligence have improved machine speech recognition and image recognition capabilities.
Google Has A Neural Network
That is Within Striking Distance
of Human-Level Translation
Using human-rated side-by-side comparison as a metric, the new Google Neural Machine Translation (GNMT) system produces translations that are vastly improved. GNMT reduces translation errors by more than 55%-85% on several major language pairs measured on sampled sentences from news websites with the help of bilingual human raters.
In addition to releasing this new research today, Google is also announcing the launch of GNMT on a notoriously difficult language pair: Chinese to English.
The Google Translate mobile and web apps are now using GNMT for 100% of machine translations from Chinese to English—about 18 million translations per day.
Below is the press release with additional info. This tech helps bring the worlds knowledge to everyone everywhere using accurate language translation. What do South Korean scholars think of aerospace engineering? How about African scholars on racism and moral psychology? Now we can find out.