Assignment about Wasserstein GAN
Need to change it to
1- Run multiple WGANs in parallel
2- Compute the max of value returned by all WGAN per iteration and based on it compute other equation
General idea: The original code run one Wasserstein GAN in multiple iterations(100000), The new changes are:
1- running multiple Wasserstein GAN in parallel in multiple iterations(100000), (the number of Wasserstein GAN is defined by user- I means the user can choose any number of WGAN)- similar to the original code but instead of run one, run many in parallel.
2- For each iteration (100000), Each WGAN will compute the mean of the fake examples vs. the real ones. [And based on all WGANs that run in parallel, critique w's fake examples vs. the real ones by selecting the max score of w based on the computation above. Then, update the generator w based on the max score of Critic w that compute.
Attachment:- Assignment File.rar