Generative Adversarial Networks (GANs) — Roadmap Step & Resources
A generator and discriminator trained adversarially to produce realistic data
- Level: advanced
- Estimated time: 2 days
- Roadmap: Deep Learning & Neural Networks
Before this step
Mini projects
- Train a simple GAN to generate handwritten digit images
Interview questions
- Why can GAN training be unstable, and what techniques help stabilize it?
Study resources
- Generative Adversarial Networks (GANs) (Video) — A clear explanation of the generator/discriminator adversarial training setup.
- Generative Adversarial Networks (Article) — The original GAN paper by Ian Goodfellow introducing the architecture.
Part of
- Generative Models — section
- Deep Learning & Neural Networks roadmap — the full learning path