Figure 5
The generative adversarial network framework contains a Generator A N N and a Discriminator A N N. Random Noise enters the Generator A N N from the left. The generator produces a stack labelled Generated Fake Samples, which is passed to the discriminator. Training Data enters a stack labelled Real Samples, which is also passed to the discriminator. The discriminator receives both real and generated samples and evaluates them. Two output paths from the discriminator lead to Discriminator Loss and Generator Loss on the right. Feedback arrows from the loss paths return towards the discriminator, while a lower feedback path from the discriminator returns to the generator. The labels Real Samples and Generated Fake Samples appear above and below their respective dashed sample groups.Architecture of a GAN