You’ve likely used products shaped by Google’s machine learning system. The neural network has helped Google understand the subject in each Google Photos upload and what you’re saying when you talk to your Android phone. It’s even helped improve search results.
Now, that machine learning system is going open source, Google announced today.
This is the second generation of Google’s so-called “deep learning” machine learning system. The first generation, DistBelief, was used internally across Google, powering improvements in speech and image recognition. It was also responsible for those trippy computer-generated interpretations of otherwise boring images through the DeepDream program.
The new system breaks DistBelief free from Google’s internal infrastructure and makes it easier to build into a wide array of applications with less setup time required. TensorFlow lets users compute by creating flow graphs, and uses the same set of tools for both research and building products to help products come to market quicker.
According to the TensorFlow site, Google open sourced the project to help standardize machine learning systems.
“Research in this area is global and growing fast, but lacks standard tools,” the site says. “By sharing what we believe to be one of the best machine learning toolboxes in the world, we hope to create an open standard for exchanging research ideas and putting machine learning in products. Google engineers really do use TensorFlow in user-facing products and services, and our research group intends to share TensorFlow implementations along side many of our research publications.”
Just because TensorFlow is now open source doesn’t mean you’ll be able to build a robot butler for your smart house just by clicking a few buttons, but by letting just about anyone access Google’s toolset, the road to more intelligent computers may be a little bit smoother.