project2_midterm

This is a cGANs demo for midterm, we use a simple DNN discriminator and generator with cGAN metedology to generator digital images from MNIST, here we use 50 epoch to train our model. We use a 100 dimensional random vector and a label from 0 to 9 as our input, the label will be embedded into a 10 dimensional vector and combined with the random vector, then generate the following images from 0 to 9.