A class of machine learning frameworks where two neural networks, a generator and a discriminator, are trained simultaneously through adversarial processes to generate data that is indistinguishable from real data.

In beauty e-commerce, GANs are used to power virtual try-on features, allowing customers to see how makeup products like lipstick or eyeshadow would look on their face before purchasing. They can also generate realistic product mockups and personalized marketing visuals at scale. For instance, a cosmetics brand might use GANs to let shoppers upload a selfie and instantly preview how different foundation shades appear on their skin.