The TensorFlow graph
One of the more important and powerful features of TensorFlow is its graph. When you define one of the three types of TensorFlow data structures previously described, you automatically add a node and an edge to your graph. Nodes represent operations and edges represent tensors, so if we were to do basic multiplication such as the preceding example, const1 and const2 would represent edges in the graph, tf.multiply would represent a node, and product would represent an outgoing edge from that node. TensorFlow's graph is static, which means we cannot change it at runtime.
Remember, an ANN performs hundreds of computations; computing and interpreting at each step would be extremely compute-intensive. The TensorFlow graph ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access