This site may earn chapter commissions from the links on this page. Terms of apply.

Google (the real Google, within Alphabet) has announced a new, open-source machine learning platform called TensorFlow, i which could greatly advance the pace at which neural networks are taking over the search giant's service portfolio — and its business model. The visitor claims it tin significantly meliorate both the speed and ability of neural network development across a wide variety of platforms and applications. If Google and the community can develop the platform as they conspicuously hope, then TensorFlow could get Google's new organizing principle, and a major driver in the continuing development of the Net.

Google is calling it the company'due south "second generation" machine learning platform, successor to the successful but at present-aging DistBelief platform that has led to many of the company's electric current services. It says that by using TensorFlow, its developers can build and train a motorcar learning algorithms "five times faster" than previously possible.

That's an important accelerate, since Google's machine learning initiatives are some of their most important, at this point.  Auto learning is increasingly how Google sifts the mountains of data we provide for them, how information technology pulls salable signals out of seemingly endless volumes of noise. Automobile learning lets the company couch ever-more than-invasively into people'south lives past providing services too interesting and valuable to laissez passer up, from translation to facial recognition. It's besides the main technology driving the ballsy Now-versus-Siri-versus-Cortana triforce of corporate one-upmanship, which could very well end up determining many users' choice of mobile platforms over the adjacent five years.

Neural NetworkThese solutions are coming to define not but the services Google provides, merely the methods by which it provides them and coordinates their findings. TensorFlow could very quickly become Google's new encephalon, and past extension a meaningful upgrade to the Internet overall.

And then, what the hell is TensorFlow? TensorFlow is a library of pre-built portions of neural network code with easy-to-utilise tools to customize them securely, and add to them with as much flexibility as possible. This is not really a new thought in the context of other open-source machine learning platforms, like Torch — just this is Google, and as such it'south non unreasonable to assume that its standard volition become the standard. And since Google tin can easily attract an army of eager, talented coders from the open up source community, it seems the most likely to progress the almost quickly. They've designed TensorFlow to accept ii of the most widely used programming languages, Python and C++.

neural networks headTensorFlow scales to run on everything from desktop super-crunchers to laptops to smartphones. The program lets developers employ their tablet to dabble with a program's design on the motorcoach, so switch seamlessly to running or grooming that algorithm on a much more than powerful desktop when they get abode.

TensorFlow besides divorces Google'south auto learning workflow from its monolithic company codebase, meaning that it'south at present possible for outsiders to meaningfully contribute to the project. DistBelief was not user-friendly, and its less forward-thinking design made information technology "near impossible" for the company to share its research code externally. With TensorFlow, we could hypothetically see collaboration with the customs pb to an explosion of sophistication in car learning. And knowing Google's arroyo to hiring, such an important contribution to the company might function as a new space for talented coders to distinguish themselves in the company's eyes.