Notebooks experimenting with Generative Adversarial Networks. Using Keras and Tensorflow.
The convolutional model (Convolutional_GAN.py) worked best.
Following:
- Goodfellow et. al, Generative Adversarial Nets
- Radford et. al., Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
Random generator outputs:
Discriminator Favorites:
Transitions in the generator input space:
MIT