Google just expanded its free software for creating smart, self-learning computers.
The company added the ability to run this software, known as TensorFlow, across a network of many computers — the same way that Google uses it for its own operations.
This means that anyone with access to a bunch of computer servers can create their own smart computer that can basically program itself. Set the computer program up with whatever it is you want it to learn. Give it a bunch of data to study and then the computer knows how to do things that, until now, only humans could do like talk, recognise pictures, draw, etc.
Every major tech vendor has its own home-grown “machine learning” tech that it is making available to programmers as a service to integrate into their apps.
But programmers have gone nuts for Google’s.
Google first gave away TensorFlow as a free and open source project last November, posted on GitHub, the place where these things are shared.
In the few weeks it was available on GitHub, it became the “most forked” project on GitHub in 2015, according to a website that tracks such things.
That means that more developers downloaded it and made their own versions of it last year than other project on GitHub: 4,355 forks to be exact.
Such sharing is the reason why companies like Google and Facebook give away so much of their software like this. The idea is to create communities of users, with everyone freely modifying the project. Some people will then submit features to back to the original project. In this case, 187 people have contributed to the main project, offering over 3,000 changes.
With the new version of TensorFlow that can run across many computers, it becomes faster and easier to train your computer. The more data the TensorFlow system can consume, the more sophisticated the things it can learn to do.
One example of TensorFlow is a project where a network of computers is “dreaming” up handwritten numbers. This is precursor to teaching a computer how to handwrite.
Machine learning is the foundation of the future. Google’s Eric Schmidt predicts that machine learning startups will become the next crop of Googles, Facebooks and huge companies, starting in the next five years.