GAN

Description:

All GAN models present here have been taken either from the book by Rowel Atienza "Advanced Deep Learning with Keras" or various web sources on Advanced GAN Models

Models Added:

• gan - Builder Script for Higher Models
• Vanilla GAN - Simple Convolutional GAN for mnist Data
• Deep Convolutional GAN - Deep Convolutional GAN
• Wasserstein GAN - Higher GAN which uses Wasserstein Loss Function
• Conditional GAN - Conditional GAN's where you can give an additional input of label and get the desired result
• Least Squares GAN - Higher GAN, more stable, Uses MSE loss
• Info GAN - Disentagled GAN,used to differentiate and generate between features in the generated images
• Stacked GAN - Hybridized Disentangled GAN,uses Enocders to build features that are used for generation
• Auxiliary Conditional GAN - Auxiliary Conditional GAN, same as cgan with I/O different
• Pix2Pix - Instance Based Cycle GAN, used for conditioning Noise(or other images) onto Specific Images
• Cycle GAN - Non mapping GAN, which can be used to learn representations between two sets of images, (~ Style Transfer)

More Models will be added soon!!!