Unconsciously, we ushered in the first anniversary of TensorFlow.
If you are not familiar with TensorFlow, you must be a machine-learning layman. This second-generation artificial intelligence learning system based on DistBelief has been named "the most popular machine learning open source project" on GitHub.
Tensor refers to the N-dimensional array, Flow refers to the calculation based on the data flow graph, Lei Feng network (public: Lei Feng network) think this and psychologist Mihalyi Csikszentmihalyi proposed "Fu flow" ( Flow concept - mdash; - mdash; complex data structure in the artificial neural network in the process of handling, but also reflects a deep immersive learning experience.
(Editor's note: Fukuoka is a feeling, state and experience, is a person in the spontaneous premise, for a certain activity or things to show strong and strong interest, and to promote their full input into their own The advantage to the extreme, into a completely immersed in the state.
TensorFlow open source in a year ago, no doubt reduce the depth of learning in all walks of life application difficulty, in the eyes of Google Brain, "it can speed up the process of machine learning, technology to help everyone better bring convenience and well-being . & Rdquo;
What did TensorFlow achieve this year? Together with the view of Lei Feng Network:
More than 480 people contributed directly to TensorFlow, including Google employees, external researchers, independent program developers, students, and developers from other companies.
TensorFlow became the most popular project on GitHub;
There are more than 3,000 open source projects related to TensorFlow on GitHub;
Received more than 10,000 Commits, the performance has been greatly improved;
In GitHub, StackOverflow and TensorFlow mailing list answered thousands of questions.
Increased distributed training;
And large data architecture integration - mdash; Currently the latter has been widely used;
Compatible with iOS and raspberry pie;
Increase support for Go, Rust, and Haskkell.
And if you are not too interested in these virtual, Lei Feng first column a few projects it:
DeepMind Launches Machine-Synthesized Speech Level with More Than Half of Human WaveNet
The integrated neural network of Google Machine Translation has made a great breakthrough;
Google 's new natural language processing model SyntaxNet;
Enhance the classification of Inception-ResNet-v2;
The use of artificial intelligence to create art Magenta;
The latest version of the automatic image description system & ldquo; Show and Tell & rdquo ;;
Detectors for the protection of endangered marine life;
Japanese farm cucumber classification machine;
& Hellip; & hellip;
The above content has been covered by Lei Feng network, and TensorFlow product manager Zak Stone in today's Google Research tweets to all concerned TensorFlow users would like to thank:
"Thank you for your interest in TensorFlow. The projects you have done on TensorFlow, whether they are for your high-end products, your fast-growing start-up companies, or your school projects, are what you do to the code base. Contribution to be sincere thanks. We are looking forward to TensorFlow's next year to be even better by working with a variety of organizations that are interested in machine learning. & Rdquo;
Yes, Lei Feng also believe that this network.