Generative Adversarial Network (GAN) Training System

A Generative Adversarial Network (GAN) Training System is an unsupervised generative learning system that implements a GAN training algorithm to solve a GAN training task.

• Context:
  • It is based on a Generative Adversarial Network (GAN) Model.
  • It requires a generator NNet and a discriminator NNet such that the discriminator cannot discriminate between real data and generated data).
• Example(s):
  • DiscoGAN System,
  • LeakGAN System,
  • …
• Counter-Example(s):
  • an Adversarial ML Algorithm (which assumes the existence of an adversarial opponent to foil its effectiveness).
  • Deep Belief Networks System.
• See: Backpropagation, Adversarial Learning, Deep Fake.

---

---

References

2014

• (Goodfellow et al., 2014) ⇒ Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. (2014). “Generative Adversarial Nets.” In: Advances in Neural Information Processing Systems.

---