November 10, 2015
Google announced yesterday that it would offer its new TensorFlow machine-learning system to outside programmers as open-source code. According to Google, TensorFlow is up to five times faster at training machine-learning models and more flexible than the company’s earlier offering. “Its previous system, DistBelief, developed in 2011, was tailored for building neural networks, the building blocks of deep learning, and for use on Google’s own network of data centers,” explains The New York Times. Google claims that the new TensorFlow software, a more general tool, will have the ability to run on an individual smartphone or across an array of data center computers.