Artificial Intelligence By Example

Running the session

The last step is to run the architecture that has just been designed. The thinking was done in the previous steps.

To run the program, the minimum code is for opening a session and running iterations, as shown in the following code snippet:

#II.dataXOR_X = [[0,0],[0,1],[1,0],[1,1]]XOR_Y = [[0],[1],[1],[0]]#III.data flow graph computationinit = tf.global_variables_initializer()sess = tf.Session()sess.run(init)for epoch in range(50000): sess.run(train_step, feed_dict={x_: XOR_X, y_: XOR_Y})

Feeding the data into the data flow graph is done with feed_dict. Feed is a keyword in FNN. This is an important feature. It is possible to feed parts of a dataset, not all of the data in stochastic (picking only parts of the dataset ...