This section focuses on Keras. We are using Keras because it simplifies the TensorFlow interface into general abstractions. In the backend, the computations are still performed in TensorFlow—and the graph is still built using TensorFlow components—but the interface is much simpler. We spend less time worrying about individual components, such as variables and operations, and spend more time building the network as a computational unit. Keras makes it easy to experiment with different architectures and hyperparameters, moving more quickly towards a performant solution.
As of TensorFlow 1.4.0 (November 2017), Keras is now officially distributed with TensorFlow as tf.keras. This suggests that Keras is now tightly integrated with TensorFlow and that it will likely continue to be developed as an open source tool for a long period...
