Generative Adversarial Networks (GANs)

Source: Path Mind Published: December 2019

Source: Path Mind
Published: December 2019

Generative Adversarial Networks (GANs)

Circulated: February 3, 2020

GANs are deep neural net architectures comprised of two nets, pitting one against the other (thus the “adversarial”). GANs can learn to mimic any distribution of data, such as: images, music, speech, and prose.

  • The generator neural network generates new data. (Think art forger who tries to forge a painting.)

  • The discriminator neural network evaluates the data for authenticity. (Think art investigator who tries to detect imitations.)

To see a GAN in action, generate a human face algorithmically: