Model Components
Generator
In a GAN, the generator takes random noise as input and outputs data (images, text, audio) that mimics the training distribution. It is trained to maximize the discriminator's probability of misclassifying its outputs as real.
Generator architecture varies by modality: convolutional networks for images, transformers for text, U-Nets for diffusion models. The generator's quality is limited by the discriminator's ability to provide informative feedback.
Authority Links
Related Terms
Model Components
Generative Adversarial Network (GAN)
Framework training two competing networks—a generator and discriminator—to produce realistic synthetic data.
Model Components
Discriminator (in GAN)
GAN component that learns to distinguish real data from fake data generated by the generator.
Model Components
Generative Model
AI model that learns to generate new data instances resembling the training distribution.
Techniques & Methods
Adversarial Training
Training AI models on challenging, adversarially crafted inputs to improve robustness and reliability.