Generative Adversarial Networks (GANs)


Hello guys,
This topic is not yet discussed in AI Saturdays. But eager to ask a question whose solution I have been searching. I have trained a GAN and trying to change parameters and digging information about GANs. How does the number of layers in generator and discriminator affect the accuracy of GAN? How does the parameter tuning affect the accuracy of GAN? If there are any resources regarding in depth analysis of working of GAN and comparison based resources, please share

Thank you.


We will be covering this via the CS231n course.

Here’s the lecture video if you wish to watch it earlier.

And here’s a good place for an overview:


hi … @AbhilashKR this are some basic hacks you can test to get better accuracy.

Some resources to start with different GAN models.
for Keras implementation :
Tensorflow :