Google brings neural machine learning for improved translation

Though Google provides one of the most powerful AI assisted language translation, the company still believes they can make it better and are in fact working along the lines. Integrating machine learning into the system, the search engine is creating a model which can translate sentences from one language to another automatically.

Initially, Google used phrase-based machine translation breaking an input sentence into corresponding words and phrases that's translated independently. However, Google Neural Machine Translation system or GNMT takes in the entire sentence as a whole.

"The advantage of this approach is that it requires fewer engineering design choices than previous Phrase-Based translation systems," writes Quoc V. Le and Mike Schuster, researchers on the Google Brain team.

Google Neural Machine Translation system, which utilizes state-of-the-art training techniques for improved translations has today been introduced into one of the most difficult language pair: Chinese to English in the Google Translate mobile and web apps, accounting for around 18 million translations per day.

According to a paper published this week, when the technique was first employed, it was able to match the accuracy of those existing translation systems but over time, it has proved capable of both producing superior results and working at the speed required for Google’s consumer apps and services.

As is clear from the above model, in some cases, GNMT is even bordering human-level translation accuracy, obviously because they are related languages like English to French or English to Spanish. But Google is eager to gather more data for "notoriously difficult" use cases, all of which will help its system learn and improve over time thanks to machine learning techniques.

Google says it's a long road ahead. "GNMT can still make significant errors that a human translator would never make, like dropping words and mistranslating proper names or rare terms," Le and Schuster explain, "and translating sentences in isolation rather than considering the context of the paragraph or page. There is still a lot of work we can do to serve our users better."

But for sure, we can expect a powerful model anytime soon given Google’s capabilities.