May 2018
Beginner
490 pages
13h 16m
English
TensorFlow relies on graph representations of data flow. The graph computes dependencies between operations.
When using Keras, as in Chapter 9, Getting Your Neurons to Work, the details of graphs and sessions do not appear. However, to compare our brain with deep learning, some details need to be examined using Google_Tensor_Classifier.py, which classifies MNIST handwritten images and saves the summaries (information about the model).
The graph of the program is as shown in the following diagram:

The nodes (labels with lines around them) and edges (lines connecting the labels) ...
Read now
Unlock full access