TensorFlow is an open source software that provides libraries for numerical computation used for Machine Learning (ML) research. Google released TensorFlow on Nov. 10, 2015 (Mon) and many people use it along with other frameworks such as Caffe, MxNet, CNTK, Chainer, PyTorch, etc.

It is designed to be used easily. Python is the scripting language that interfaces the core part of the libraries written in C++. A complex ML code can be written with several lines in Python to run experiments while one can write codes in C++ to extend the existing functionalities of TensorFlow.


White paper:TensorFlow: Large-scale machine learning on heterogeneous systems