Google Open source second generation machine learning system TensorFlow

Deep learning has a profound effect on computer science. It makes it possible for cutting-edge technology to research and develop products that are used by tens of millions of of people everyday.
The study announced the launch of the second-generation machine learning System (TENSORFLOW), which has been strengthened for the previous distbelief, and more importantly,
It's open source and can be used by anyone.

Built in 2011, Google's internal deep learning structure, distbelief, has enabled Google to build larger neural networks and scale training for the thousands of cores of the data center, typically in applications such as enhancing speech recognition in Google Apps and adding image search capabilities to Google search.

But Distbelief has some limitations, such as the more difficult to set up, and the close integration with Google's internal access facilities--which is not a good way to share the research code. So Google study announced the launch of the open-source TensorFlow, which is Google's second-generation machine learning system, for the distbelief of the short board to make a complement. TensorFlow is flexible, mobile, easy-to-use, and fully open source. Based on distbelief speed, scalability, and the features that prepare the product, TensorFlow is doing even better. According to Google, in some benchmarks, TensorFlow's performance was twice times faster than that of distbelief.

The extended support for TensorFlow built-in deep learning extends beyond this-any calculation that can be expressed using a computed flow graph can use TensorFlow. Any gradient-based machine learning algorithm can benefit from TensorFlow's automatic differentiation (auto-differentiation). With a flexible Python interface, it can be easy to express ideas in TensorFlow.

In addition to the research, TensorFlow will be very helpful, for the actual product is also very meaningful. Move your ideas seamlessly from desktop GPU training to mobile phone running. Using Google's sample model architecture, you can quickly start using machine learning technology-Google is planning to publish imagenet computer vision models on top of TensorFlow.

More importantly, TensorFlow is open source, and anyone can use it for free. Google's deep learning researchers are using tensorflow and will continue to use it in machine learning products in the future.

Google Open source second generation machine learning system TensorFlow