Google Translate is rolling out a major upgrade that promises more human-like language translations. Google is bullish on its Neural Machine Translation technology, claiming that it's a bigger upgrade to the service than everything that's been accomplished in the last ten years combined. The company is rolling out the improvements to eight language pairs in Google search, the Translate apps, and the website. You'll find the new technology behind translations between English and French, German, Spanish, Portuguese, Chinese, Japanese, Korean and Turkish. Google says that makes up more than 35 percent of all language queries.

Google is improving its language translation service with a new approach that interprets whole sentences at a time rather than phrases piece-by-piece, the search giant announced Tuesday. The firm says that should make translations from the service, called Google Translate, easier to understand. "It uses this broader context to help it figure out the most relevant translation, which it then arranges and adjusts to be more like a human speaking with proper grammar," Barak Turovsky, product lead for Google Translate, wrote in a new blog post. This new version of Google Translate is powered by neutral machine translation, which is a new method of teaching computers to translate human languages, according to the Association for Computational Linguistics. Google published updates regarding its research into this field in September; it's now integrating the technique directly into Google Translate.

What started with Mandarin Chinese is expanding to English; French; German; Japanese; Korean; Portuguese and Turkish, as Google has increased the languages its Neural Machine Translation (NMT) handle. "These represent the native languages of around one-third of the world's population, covering more than 35 percent of all Google Translate queries," according to The Keyword blog. The promise here is that because NMT uses the context of the entire sentence, rather than translating individual words on their own, the results will be more accurate, especially as time goes on, thanks to machine learning. For a comparison of the two methods, check out the GIF embedded below. Google says that the ultimate goal is to have all 103 languages in Translate using machine learning.

Google says its Translate app now spits back more natural translations. Google said Tuesday that it has vastly improved its Google Translate app, available on phones and the web. The search giant said it's now incorporating "neural machine translation" into the software, which translates whole sentences at a time, instead of breaking the text down to smaller chunks and translating those pieces. That means translations come out more natural, with better syntax and grammar. "It has improved more in one single leap than in 10 years combined," said Barak Turovsky, the product lead for Google Translate, during a press event at Google's San Francisco office. The new translation system is coming to eight of the 103 languages supported by the app.

WIPO Director General Francis Gurry speaks about WIPO's ground-breaking new "artificial intelligence"-based translation tool for patent documents, a new service that hands innovators around the world the highest-quality service yet available for accessing information on new technologies. WIPO Translate now incorporates cutting-edge neural machine translation technology to render highly technical patent documents into a second language in a style and syntax that more closely mirror common usage, out-performing other translation tools built on previous technologies.

Reusability: One of the reasons why AI progress is accelerating is the community is creating more and more reusable components that can be plugged into different domains, frequently attaining performance equivalent to or better than hand-designed algorithms. The 2015 ImageNet challenge was won by a Microsoft system built out of Residual Networks, then in 2016 Microsoft made a speech processing breakthrough via a system that also relied on Residual Networks. Similarly, DeepMind's WaveNet system has been slightly tweaked and re-applied to the domain of neural machine translation (PDF). This kind of re-use is a good thing as it suggests we are beginning to create the right sorts of low-level primitives that general intelligences can be built out of.

Which authors of this paper are endorsers? Disable MathJax (What is MathJax?)

Artificial intelligence is difficult to develop because real intelligence is mysterious. This mystery manifests in language, or "the dress of thought" as the writer Samuel Johnson put it, and language remains a major challenge to the development of artificial intelligence. "There's no way you can have an AI system that's humanlike that doesn't have language at the heart of it," Josh Tenenbaum, a professor of cognitive science and computation at MIT told Technology Review in August. In September, Google announced that its Neural Machine Translation (GNMT) system can now "in some cases" produce translations that are "nearly indistinguishable" from those of humans. "Machine translation is by no means solved. GNMT can still make significant errors that a human translator would never make, like dropping words and mistranslating proper names or rare terms, and translating sentences in isolation rather than considering the context of the paragraph or page."

Have you ever been in a situation where knowing another language would have come in handy? I remember standing on the platform at Tokyo Station watching my train to Nagano -- the last train of the day -- pulling away without me on it. What ensued was a frustrating hour of gestures, confused smiles, and head-shaking as I wandered the station looking for someone who spoke English (my Japanese is unfortunately nonexistent). It would have been really helpful to have a bilingual pal along with me to translate. Bilingual pals can be hard to find, but Google's new translation software may be an equally useful alternative.

Code to train Neural Machine Translation systems as described in our EMNLP paper Effective Approaches to Attention-based Neural Machine Translation. Here, we convert train/valid/test files in text format into integer format that can be handled efficiently in Matlab. This trains a very basic model with all the default settings. We set'isResume' to 0 so that it will train a new model each time you run the command instead of loading existing models. Decode with beamSize 2, collect maximum 10 translations, batchSize 1.