Google unleashes TensorFlow 0.8 for your distributed machine learning needs
Distributed intelligence across the Earth? Let's call it 'Skynet'
Version 0.8 of Google's open source machine learning library, TensorFlow, has been released, adding support for distributed computing so you can get your whole Beowulf cluster infected with its Skynet.
TensorFlow is an open source software library for numerical computation using data flow graphs, and was released under the Apache 2.0 open source licence last year.
At the time the library was described as the successor to Google's DistBelief system, which had been used to separate spam in Gmail, scan YouTube videos and enable various forms of recognition on unstructured audio and video data. Its applications have extended to the Chocolate Factory's trippy DeepDream images.
The distributed TensorFlow is powered by the gRPC library, and is intended to complement the recent announcement of Google Cloud Machine Learning, "which enables you to train and serve your TensorFlow models using the power of the Google Cloud Platform."
Research conducted last year (PDF) found TensorFlow to be "a very flexible framework," though noted "its performance is currently not competitive compared to the other studied frameworks."
"The current version of distributed computing support in TensorFlow is just the start," the Chocolate Factory's release statement proclaimed. "We are continuing to research ways of improving the performance of distributed training—both through engineering and algorithmic improvements—and will share these improvements with the community on GitHub." ®