Hands-On Generative Adversarial Networks with Keras
上QQ阅读APP看书,第一时间看更新

The discriminator

In the original GAN framework, the discriminator (detective)  is a function , where  is some input sample in  ,and  is a scalar. When using GANs to produce images, for example, in some formulation  represents an image and  represents ’s predicted probability that  comes from the real distribution.