Hands-On Deep Learning Algorithms with Python

Train the DCGAN

Let's begin the training. Define the number of batches, epochs, and learning rate:

num_batches = int(x_train.shape[0] / batch_size)steps = 0num_epcohs = 500lr = 0.00002

Define a helper function for generating and plotting the generated images:

def generate_new_samples(session, n_images, z_dim):    z = tf.random_normal([1, z_dim], mean=0.0, stddev=1.0)    is_training = tf.placeholder(tf.bool, [], name='training_bool')     samples = session.run(generator(z, z_dim, batch_size, is_training, reuse=True),feed_dict={is_training: True})            img = (samples[0] * 255).astype(np.uint8)    plt.imshow(img)    plt.show()

Start the training:

with tf.Session() as session:

Initialize all variables:

    session.run(tf.global_variables_initializer())

To execute ...