Translation Paves the Way for Omnipresent Machine Learning

The zest with which Google has recently put its new, and in every sense of the word ground-breaking, Neural Machine Translation (GNMT) system into the focus of its massive TensorFlow ambitions, itself designed to cement the company’s lead as a cloud provider, marked a not only timely, but symbolic move: few things we do with data represents people’s desire to share and communicate more than interlingual exchange.

If you’re wondering how this shift will affect you and your business, you’re likely on to something: the trend which became globally known last November, now seems to be only the beginning.

 

The Brain Team

The Google Brain Team, true to their name, have just posted their machine translation manifesto vowing to teach the public how to build a competitive MT platform, thus arguably passing on their coveted skills and resources to any third party developer.

Surely, experts will observe that Google make parts of their research platform open source only once they have secured unsurmountable advantage over their competitors – not to mention the amount of data they gain access to via the activities of smaller challengers.

Whatever the cause, anyone interested can now partake in the Neural Machine Translation (NMT) tutorial for TensorFlow on GitHub. Not a bad idea for your team when it comes to a professional development workshop!

Connecting minds everywhere

The tech giant’s bold bid to prioritise machine learning in general, and machine translation as central, foreshadows a near future of 5 (then 6, 7…) G mobile empowered, hyper-fast, hyper-communicative, hyper-quantity data supplies and users.

Services and providers will likely need a lot of ingenuity and creativity to keep up – as their users will surely want to.

Lessons in data

This isn’t the first time Google would pave the way for a revolution in how we use technology worldwide, and analysts agree that they merely use machine translation -i.e., a generalisation of Google Translate- as a vanguard: this is a service of theirs used by everybody regardless of vertical, a platform that isn’t entangled with advertising, as well as a distinguished field in artificial intelligence that nevertheless no one before Google could successfully tackle. Moreover, Google Translate was the first tangible proof that the neural network approach to AI really works better than all others so far.

 

Neural Networks Demystified Part 1

 

As eloquently chronicled by The New York Times, this indeed looks like the ideal machine learning big data experiment, used by and useful to anybody with access to the internet.

But it would be a mistake to start worrying about enormous requirements or security red flags surrounding ML. Rather, we would urge you to look at the pattern of Google’s new emphasis on its cloud service, and within that a machine translation platform available to all.

While machine learning, a data-heavy field that’s just now becoming a real and useful possibility, is still in many ways uncharted territory, you have just received an invitation to help roam it.