GitHub - umutisik/gan-experiments: Experiments with generative adversarial networks

Generative Adversarial Network Experiments

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

Name		Name	Last commit message	Last commit date
Latest commit History 7 Commits
outputimages		outputimages
.gitignore		.gitignore
Convolutional_GAN.py		Convolutional_GAN.py
GAN_on_MNIST.ipynb		GAN_on_MNIST.ipynb
GAN_on_NotMNIST.ipynb		GAN_on_NotMNIST.ipynb
Keras_GAN_on_MNIST.ipynb		Keras_GAN_on_MNIST.ipynb
Keras_GAN_on_MNIST_better_model.ipynb		Keras_GAN_on_MNIST_better_model.ipynb
Keras_GAN_on_MNIST_convolutional.ipynb		Keras_GAN_on_MNIST_convolutional.ipynb
README.md		README.md