0

I am newbie on deep learning and it happens to me to confuse between Keras and tensorflow. knowing that tensorflow is a framework and Keras is a library, what is the difference between using these two deep learning tools.

HISI
  • 4,557
  • 4
  • 35
  • 51

1 Answers1

3

Keras purposes is to use a framework in backend like Tensorflow, Theano or CNTK in an easier way.

For example, create a simple convolutional model under Tensorflow can be hard. While create the same model under keras is very instinctive.

The difference between Tensorflow/Theano/CNTK and Keras is the following : Keras is a framework who use the functions of Tensorflow/Theano/CNTK. So Keras needs one of them to do something.

Tensorflow/Theano/CNTK or other like coffee can do everything by themselves. But, often, it's harder to develop a model with them.

Pusheen_the_dev
  • 2,077
  • 4
  • 17
  • 33
  • You mean Keras is a library! – HISI Mar 07 '18 at 16:35
  • You can look at Panos's answer here : https://stackoverflow.com/questions/148747/what-is-the-difference-between-a-framework-and-a-library I think Keras is more likely a framework according to Panos's answer. – Pusheen_the_dev Mar 07 '18 at 16:51
  • Since we can use keras on the top of Tf Teano... so it shouldn't be a framework – HISI Mar 07 '18 at 17:03