The discriminator loss is given as follows:
First, we will implement the first term, that is, :
D_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=D_logits_real,
labels=tf.ones_like(D_logits_real)))
Now we will implement the second term, :
D_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=D_logits_fake,
labels=tf.zeros_like(D_logits_fake)))
The final loss can be written as follows:
D_loss = D_loss_real + D_loss_fake