Google shares its machine learning software, TensorFlow, with the world
For Google itself, machine learning has become core to many of its services, and it's become integral to many features in its search, language, photo and other products.
Now with TensorFlow, the company is giving access to independent developers to build their own software using AI technology.
According to the blog post from Jeff Dean, senior google fellow, and Rajat Monga, technical lead, TensorFlow is an improvement on the old neural network system the company employed called DistBelief. The new technology improves on some of the shortcomings of that system, including being more "general, flexible, portable, easy-to-use, and completely open source," they wrote.
Further, the company said it originally built TensorFlow as a research tool, but the technology is now ready for real digital products.
"Our engineers use it to infuse Google Search with signals derived from deep neural networks, and to power the magic features of tomorrow," read the blog post. "We'll continue to use TensorFlow to serve machine learning in products, and our research team is committed to sharing TensorFlow implementations of our published ideas."
Included on the website for the TensorFlow project is the software download, tutorials with lessons for beginners and experts, mechanics to use with the system, documentation on how the system runs and other assorted resources.
"TensorFlow was built from the ground up to be fast, portable, and ready for production service," read the blog post. "You can move your idea seamlessly from training on your desktop GPU to running on your mobile phone."
As Bloomberg noted, Google has worked for the past few years to create an ecosystem of AI developers across academic and private sector spaces. The publication suggests that Google will see clear benefits to open sourcing this tool. Allowing the curious tinkerers of the software world open access to the AI tools Google itself uses will only work to bring fresh ideas to the company.
For the average enterprise, any headway that Google makes with machine learning that it then open-sources can only be a good thing.
Though it might seem like a far way off, many believe this to be the next step for software. With tech giants like Google carrying the water for the rest of the industry – and forcing rivals like Apple and Facebook to compete or be shut out – it will build to a future with richer AI-based technologies that much quicker.
- read the article from Bloomberg
- read the blog post from Google
- visit the TensorFlow website
Machine learning key to 'rethinking everything' at Google, says CEO Pichai
How 'deep learning' works is still a mystery, Microsoft Research head said
Artificial intelligence will replace traditional software approaches, Alphabet's Eric Schmidt says