Keras Deep Learning Cookbook

Average metrics of the GAN

After running the model for 30,000 iterations, the following metrics can be seen:

mean d loss:0.6404680597275333 mean g loss:0.9513815407413333 mean d accuracy:62.71046875

Mean discriminator loss is about 0.64 and generator loss is 0.95. Discriminator accuracy is about 62 percent. We also plotted the metrics as a function of epochs:

Next, let's look at a more advanced GAN called a boundary seeking GAN.