How we train the discriminator actually? It wants 0 and we pass both real and fake images together, the real image will closer to 1 so the loss will high. So does it mean high error means real image and low error means a fake image? That's how the discriminator determines fake or real images?
question from:https://stackoverflow.com/questions/65546699/how-discriminator-determine-whice-one-is-real-image-and-whice-one-is-fake