MachineLearning/gan library

🎭 Generative Adversarial Network (GAN)

Compact GAN scaffolding: generator and discriminator primitives with a clear training loop contract. Designed for clarity and testability with instructions on how to extend to real generative models. Strong docstrings describe the expected input/output shapes and training behavior.

Contract:

• Input: noise vectors for generator, real samples for discriminator.
• Output: trained generator/discriminator; generator.predict returns samples.
• Errors: throws ArgumentError for invalid shapes.

Classes

GAN

Simple GAN scaffolding using ANN heads for generator and discriminator.