- Implement dcgan with keras and tensorflow
- Implement dcgan with keras only
- Sanity check dcgan keras only with MNIST data
- Sanity check dcgan with keras and tensorflow
- Run dcgan with keras only on dog dataset
- Run dcgan with keras and tensorflow on dog dataset
- Try other variants of GAN
- learning rate used 0.0002 for discriminator and 0.00002 for generator
- batch size 32
- batch normalization
- dropout 0.3 in discriminator
- image size 28x28